Vol. 12 No.1 / 2021 ISSN: 1847-9375

Vol. 12 No.1 / 2021ISSN: 1847-9375

Business Systems Research | Vol. 12 No. 1 | 2021

i

Impressum

Focus and Scope

Business Systems Research Journal (BSR) is an international scientific journal focused on

improving the competitiveness of businesses and economic systems. BSR examines a wide

variety of decisions, processes, and activities within the actual business setting and the systems

approach framework. Theoretical and empirical advances in business systems research are

evaluated regularly. Special attention is paid to educational, social, legal, and managerial

aspects of business systems research. In this respect, the BSR journal fosters the exchange of

ideas, experience, and knowledge between regions with different technological and cultural

traditions, in particular in transition countries. Papers submitted for publication should be original

theoretical and practical papers. The journal also publishes case studies describing innovative

applications and critical reviews of theory.

Abstracted/indexed in: Baidu Scholar, Cabell's Whitelist, CEJSH (The Central European Journal of Social Sciences and

Humanities), CNKI Scholar, CNPIEC – cnpLINKer, Dimensions, DOAJ, EBSCO (relevant

databases), EBSCO Discovery Service, EconBiz, Engineering Village, ERIH PLUS (European

Reference Index for the Humanities and Social Sciences), Genamics JournalSeek, Google

Scholar, Hrcak, Inspec, J-Gate, JournalGuide, JournalTOCs, Microsoft Academic, Naver

Academic, Naviga (Softweco), Norwegian Register for Scientific Journals, Series and Publishers,

Primo Central (ExLibris), ProQuest (relevant databases), Publons, QOAM (Quality Open Access

Market), ReadCube, Research Papers in Economics (RePEc), SCImago (SJR), SCOPUS,

Semantic Scholar, Sherpa/RoMEO, Summon (ProQuest), TDNet, TEMA Technik und

Management, Ulrich's Periodicals Directory/ulrichsweb, WanFang Data, Web of Science -

Emerging Sources Citation Index, WorldCat (OCLC)

Editor-in-Chief Mirjana Pejić Bach, University of Zagreb, Faculty of Economics & Business, Department of

Informatics, Croatia

Associate Editors João Varajão, Universidade do Minho, Portugal

Josip Stepanić, University of Zagreb, Faculty of Mechanical Engineering and Naval

Architecture, Department of Non-destructive Testing, Croatia

Nataša Šarlija, University of Osijek, Faculty of Economics in Osijek, Croatia, Croatia

Advisory Board Sarunas Abramavicius, ISM University of Management and Economics, Lithuania

David Al-Dabass, Nottingham Trent University, School of Computing & Informatics, United

Kingdom

Jakov Crnkovic, University at Albany, School of Business, USA

Martin Fieder, University of Vienna, Rector's Office, Austria

Anita Lee Post, University of Kentucky, School of Management, Decision Science and

Information Systems Area, United States

Gyula Mester, University of Szeged, Hungary

Matjaž Mulej, International Academy of Cybernetics and Systems, Austria, University of

Maribor and IRDO Institute for development of social responsibility, Slovenia

Olivia Par-Rudd, OLIVIAGroup, United States

Ada Scupola, Department of Communication, Business and Information Technologies,

Roskilde University, Denmark

Tadas Šarapovas, ISM University of Management and Economics, Lithuania

Ajay Vinze, Arizona State University, WP Carey School of Business, United States

Business Systems Research | Vol. 12 No. 1 | 2021

ii

Editorial Board Nahed A. Azab, School of Business, American University in Cairo, Egypt

Sheryl Buckley, University of South Africa, School of Computing, South Africa

Terence Clifford-Amos, Université Catholique de Lille, France

Josef Basl, University of Economics, Prague, Czech Republic

Nijaz Bajgorić, University of Sarajevo, School of Economics and Business, Bosnia and

Herzegovina

Anton Florijan Barisic, Centre for Excellence in Development - CEDEX, Croatia

Rajeev Dwivedi, Institute of Management Technology, India

Inês Dutra, Universidade do Porto, Portugal

Francisco Garcia, Universidad de Salamanca, Spain

B. B. Gupta, National Institute of Technology Kurukshetra, India

Mojca Indihar Štemberger, Faculty of Economics, University of Ljubljana, Slovenia

Božidar Jaković, University of Zagreb, Faculty of Economics & Business, Department of


Mira Krpan, University of Zagreb, Faculty of Economics & Business, Department of Economic

Theory, Croatia

Helmut Leitner, Graz University of Technology. Institute for Information Systems and Computer

Media (IICM), Austria

Sonja Sibila Lebe, Faculty of Economics and Business, Maribor, Slovenia

In Lee, School of Computer Sciences, Western Illinois University, USA

Olivera Marjanović, University of Sydney, Faculty of Economics & Business, Department of

Business Information Systems, Australia

Irena Palić, University of Zagreb, Faculty of Economics & Business, Department of Statistics,

Croatia

Sanja Pekovic, University Paris-Dauphine, France

Giuseppe Psaila, University of Bergamo, Italy

Lei Ping, Shanghai University of International Business and Economics, China

Markus Schatten, University of Zagreb, Faculty of Organization and Informatics, Croatia

Nikola Vlahović, University of Zagreb, Faculty of Economics & Business, Department of


Ilko Vrankić, University of Zagreb, Faculty of Economics & Business - Zagreb, Croatia, Croatia

Jusuf Zeqiri, South East European University, Faculty of Business and Economics, Macedonia

Jovana Zoroja, University of Zagreb, Faculty of Economics & Business, Department of


Zhang Wei-Bin, Ritsumeikan Asia Pacific University, Japan

Berislav Žmuk, University of Zagreb, Faculty of Economics & Business, Department of Statistics,

Croatia

Language Editors Abstract Editing: Andrea-Beata Jelić, Poliglossa Language Centre, Croatia

Managing Editors Ljubica Milanović Glavan, University of Zagreb, Faculty of Economics & Business, Croatia

Jasmina Pivar, University of Zagreb, Faculty of Economics & Business, Croatia

Mihovil Braim, University of Zagreb, Faculty of Economics & Business, Croatia, Student Assistant

Publisher IRENET, Society for Advancing Innovation and Research in Economy

ISSN Business systems research (Online) = ISSN 1847-9375

Editorial Office e-mail: [email protected]

Web:

http://www.bsrjournal.org; http://hrcak.srce.hr/bsr; http://www.degruyter.com/view/j/bsrj

Business Systems Research | Vol. 12 No. 1| 2021

iii

Business Systems Research

A Systems View across Technology & Economics

Economic and Business Systems Research Articles

Privatization in Croatia: Standpoint of Croatian Citizens in 1998 and 2018

Jan Horaček, Helena Nikolić ……………………………………………….............................. 1

Performance Measurement of Vietnamese Publishing Firms by the Integration of the

GM (1,1) Model and the Malmquist Model

Xuan-Huynh Nguyen, Quoc Chien Luu……………...……………………….….................... 17

Critical Success Factors of New Product Development: Evidence from Select Cases

Rajeev Dwivedi, Fatma Jaffar Karim, Berislava Starešinić................................................ 34

Differences in Slovenian NUTS 3 Regions and Functional Regions by Gender

Samo Drobne …….…….…................................................................................................... 45

The Effect of External Knowledge Sources on Organizational Innovation in Small and

Medium Enterprises in Germany

Shoaib Abdul Basit.…….….................................................................................................. 60

Gender Disparity in Students’ Choices of Information Technology Majors

Yu Zhang, Tristen Gros, En Mao............................................................................................. 80

The Effect of Auditor Rotation on the Relationship between Financial Manipulation

and Auditor’s Opinion

Ivica Filipović, Toni Šušak, Andrea Lijić…............................................................................. 96

Freelancing in Croatia: Differences among Regions, Company Sizes, Industries and

Markets

Ana Globočnik Žunac, Sanja Zlatić, Krešimir Buntak......................................................... 109

Familiarity with Mission and Vision: Impact on Organizational Commitment and Job

Satisfaction

Dunja Dobrinić, Robert Fabac............................................................................................ 124

Collaborative Strategic View in Corporate Social Responsibility – Construction

Industry Case

Lana Lovrenčić Butković, Dina Tomšić, Simona Kaselj...................................................... 144

Cultural Tourism and Community Engagement: Insight from Montenegro

Ilija Moric, Sanja Pekovic, Jovana Janinovic, Đurđica Perović, Michaela Griesbeck... 164

Position and Role of Social Supermarkets in Food Supply Chains

Blaženka Knežević, Petra Škrobot, Berislav Žmuk………..................................................... 179

Business Systems Research | Vol. 12 No. 1| 2021

iv

Information Systems Research Articles

Enterprise Digital Divide: Website e-Commerce Functionalities among European

Union Enterprises

Božidar Jaković, Tamara Ćurlin, Ivan Miloloža……............................................................ 197

Does the “Like” Habit of Social Networking Services Lower the Psychological Barriers

to Recommendation Intention in Surveys?

Takumi Kato………………………………………...................................................................... 216

The Proportion for Splitting Data into Training and Test Set for the Bootstrap in

Classification Problems

Borislava Vrigazova.............................................................................................................. 228

1

Business Systems Research | Vol. 12 No. 1 |2021

Privatization in Croatia: Standpoint of

Croatian Citizens in 1998 and 2018

Jan Horaček

Electrolux d.o.o.

Helena Nikolić

Faculty of Economics and Business – Zagreb, University of Zagreb, Croatia

Abstract

Background: The break-up of Yugoslavia has led to a transition from planned to the

market economy. The main task of transition is privatization, which implies transferring

most of the former social ownership to private individuals. The privatization process has

marked the end of the twentieth century in Croatia and still carries many unanswered

questions that have arisen because of the persistent need for privatization in the

former, unconsolidated state. Objectives: The main objective of the paper is to make

a comparison of respondents’ perception of Croatian privatization in 2018 compared

to 1998. The aim is to investigate the similarities and changes in the attitudes of the

Croats regarding the privatization processes that Croatia has engulfed in several

stages. Methods/Approach: The survey was conducted on a sample of one hundred

Croatian citizens about their perception of the privatization process in Croatia in 2018.

Results of the survey in 1998 and 2018 were compared using the chi-square test.

Results: The respondents in 2018 are convinced that the main goals of privatization

have not been realized. Citizens' distrust towards the system and institutions

conducting the privatization process is greater in 2018 compared to 1998.

Conclusions: Respondents perception of privatization has not changed significantly

concerning the 20-year gap. Dissatisfaction due to the unfulfilled fundamental goals

is still present, as is the need for revision of privatization.

Keywords: privatization; Croatia; Croatian economic system

JEL classification: L33, N14

Paper type: Research article

Received: Mar 28 2020

Accepted: Jun 08 2020

Citation: Horaček, J., Nikolić, H. (2021), “Privatization in Croatia: Standpoint of Croatian

Citizens in 1998 and 2018”, Business Systems Research, Vol. 12, No. 1, pp. 1-16.

DOI: https://doi.org/10.2478/bsrj-2021-0001

2


Introduction Privatization of the economy is the process of transforming a state-controlled and

central planning system into a market system, firmly and consistently based on the

principles of private ownership (Bjørnskov and Potrafke, 2011; Čengić, 1995).

Privatization is also defined as the necessary and desirable transfer of the entire or

greater part of the ownership of the public (or social) sector to private individuals

(Kalogjera, 1993). The main objective of privatization is to achieve more efficient and

competitive business enterprises because the state has been proved to be a bad

manager (Bonneau and Shoven, 2011; Bennett et al., 2007; Njavro,1993).

The peculiarities of the Croatian transition process stem from the ownership

structure, which was characterized by the social form of ownership instead of the usual

state ownership that dominated the planned economies (Arsov and Naumoski, 2014).

Despite the great influence of the state and political structures on the economy, the

majority of SMEs had a great right to freely decide on the size and structure of

production, considering market requirements (Vukšić, 2016). There was also a great

deal of autonomy related to income distribution, consumption, and savings, which

created a stronger identification of employees with the company (Pejić Bach et al.,

2018). Large companies, on the other hand, remained heavily influenced by state and

political factors. Given the volatility of the development function, it was impossible to

achieve the mobility of capital and direct it towards the most profitable investments

(Rogić, 1998).

The Croatian economy has developed significantly in the last 20 years, but the

relatively rapid privatization in Croatia was marked by numerous controversies

regarding the efficacy, abuse and ultimate benefits to the national economy

(Josipović, 2018; Jelić et al., 2006,). At the beginning of the privatization process, the

Croatian citizens had a significant trust regarding the transition. Due to the

abovementioned problems with privatization, it is likely that this trust has decreased,

which is reflected in the media and various public events (Vušković, 2020; Gatarić,

2019; Ožanić, 2016; Protulipac, 2014; Iveković, 2012)

To shed some light on this issue, the paper aims to acquaint the current perception

of Croatian citizens about the privatization process, the goals, the winners and losers

of privatization in Croatia, and to compare these results with the results of research

conducted by Aleksandar Štulhofer on the perception of the privatization process in

Croatia in 1998 and 1996 (Štulhofer, 1999).

The organization of the paper is as follows. Section 2 presents a brief overview of

the chronology of Croatian privatization through a comparison of its features. Section

3 presents the used methodology and data while Section 4 shows the results. Section

5 summarizes the paper with the concluding remarks.

Privatization in Croatia Conversion into private property took place in four stages (Nikić, 2004). Privatization in

Croatia began in 1989 with liberalization as a response to the actual crisis.

Transformation of ownership and adoption of the Law which regulated the

implementation represented the first phase of Croatian privatization (1991-1994). The

emphasis was on protecting strategically important state-owned enterprises from

privatization. Other companies were subject to sale, and employees were given

priority in buying shares on preferential terms. However, the question arose as to whom

the ownership of individual companies belonged after the collapse of the old system.

In his work, Njavro (1993) cites examples of Hungary and Poland where the ownership

problem was solved relatively easily. Companies were assumed to be state-owned.

3


On the other hand, in Yugoslavia was not known who the real owners of the capital

were. The funds were perceived as the social property that was placed on the use of

workers. The disadvantage of the first phase was the subjective valorisation of the

value of enterprises due to the prompt sale and war.

The second phase was marked by economic stabilization which affected the

competitiveness of the companies. During this period, the companies that avoided

the first privatization wave entered the privatization process. The Croatian Privatization

Fund and the Pension Fund disposed of their assets (Franičević, 2002). The third wave

of privatization was political in nature, characterized by mass coupon privatization

and the distribution of shares of questionable quality to the general public.

Specifically, those companies facing bankruptcy were for sale. The fourth stage of

privatization continued the sale of bankrupt companies. However, an initiative to

revise conversion and privatization emerged on suspicion of numerous frauds and

criminal acts (Bendeković, 2000).

In the case of privatization, the state took over the entire management mechanism

on all the essential items related to the company. This form of asset centralization was

described by Gregurek (2001) as "original state-level accumulation". The lack of

strategic goals, the neglect of the market as a reference in determining the value of

the company, and the frequent changes in institutional rules and legal frameworks

had affected the uncertainty of potential prospective investors and led to numerous

frauds not characteristic of Central European countries.

The goals of privatization have changed and adjusted depending on the

economic and political situation in the country. The political elite sought to pursue

privatization as socially just and at the same time economically efficient. However, it

became clear that the privatization model and its implementation were entering

projects of high social, economic and political risk (Čengić, 2000). The goals of the

transformation of social enterprises were: protection of national wealth, the

introduction of fresh capital into the economy, start-up of entrepreneurship and the

wide dispersion of ownership (Kalogjera, 1993). The normative goals that were

proclaimed were largely not met and privatization was largely reduced to filling the

state budget and developing a primitive type of capitalism. The result was an

insufficient number of new jobs and neglect of entrepreneurship as well as high

administrative barriers that prevented many business initiatives. There was a lack of

managers and early retirement indirectly encouraged by the ruling party (Šokčević,

2007).

There was also no benefit to consumers in the form of lower prices due to market

liberalization (Čučković, 2002). Privatization had not fulfilled its purpose given the large

number of enterprises that had outdated technology and a large number of

employees. The demand for this type of enterprise was very low. However,

privatization reduced the public debt, but not successfully enough because the

majority of public and state-owned enterprises were the main generators of budget

alimony losses (Grgurek, 2001). The negative effects of privatization are reflected in

the fact that a large number of owners did not have any development concept and

that they were buying companies for trading purposes only. Converting debt to equity

led to the fact that a large number of companies were privatized at low prices (Lasić,

2000). This resulted in minimal liability of the owner towards the company,

fragmentation of large enterprises and sale of assets. Manufacturing and maintaining

a business were not the primary task of new owners; so many people remained

unemployed and became a burden to the state as a social category.

Table 1 presents the characteristics of the Croatian economy in 1990. Privatization

in Croatia began in the 1980s when business executives disregarded control over

4


companies to turn them into private property (Stojcic, 2012). This was achieved by

diverting money into private companies and selling the assets of the state to

enterprises (Olgić Draženović and Kusanović, 2016; Haramija and Njavro, 2016). The

liberalization led to the fact that in 1990 the number of private companies was 6.785,

while the number of socially owned enterprises was 3.637. However, despite the

numerical superiority of private enterprises, state-owned companies employed 97.6%

of the workforce and their strength manifested itself in owning capital of 57.6 billion

German marks.

Table 1

Croatian economy in 1990

Form of ownership Firms

Number %

Labour employed

Number %

Social capital

Mil. DM %

Social firms 3.637 35,5 1.105.873 97,6 57.609,3 100,0

Public social firms 98 2,7 123.097 11,1 18.089,3 31,4

Private firms

6.785 62,5 19.602 1,7 - -

Cooperative firms 284 2,6 5.290 0,5 - -

Mixed firms 153 1,4 2.001 0,2 - -

Total 10.859 100,0 1.132.766 100,0 57.609,3 100,0

Source: Agency for Restructuring and Development (1992)

Table 2 presents the characteristics of the Croatian economy in 2018. A significant

increase in the total number of active business entities can be seen if a comparison is

made with the situation in 1990. The number of mixed firms increased as did the

number of cooperative companies. But, in particular, the total number of private

companies increased markedly. On the other hand, the number of state-owned

enterprises decreased. This is supported by a series of privatization processes over the

years. Also, the percentage of legal entities whose ownership is not monitored is

evident.

Table 2

Croatian economy in 2018

Form of ownership Number of firms %

State-owned firms 1.238 0,8

Private firms 125.300 83,6

Cooperative firms 936 0,6

Mixed firms 897 0,6

No ownership firms 21.602 14,3

Total 149.973 100,0

Source: Central Bureau of Statistics (2019)

However, privatization also had positive effects: the transition from a system in

which self-management was declared null and void to a system of known owners,

harmonization with European standards and thereby facilitating participation in the

international market. Gregurek (2001) cites as positive effects the direct appropriation

of formal legal responsibility and economic risks in the decision-making process, as

well as the formation of a market-based macroeconomic system.

5


Methodology Research in 1996 and 1998 The study, which served as a foundation and landmark for this research was

conducted in 1998 under the name of the “Privatization in the eyes of the Croatian

public”. It was conducted on a representative sample of 1001 national subjects.

Štulhofer occasionally compares data from his research with data from a study called

"Sociocultural Aspects of Transition", which was conducted on a sample of 1056

respondents (Štulhofer, 1998). Considering many indicators that influence the

perception of this topic, (i.e. age, tradition, education...) there was a need to research

a more representative sample. Therefore, the basic assembly, made by the citizens of

Croatia, is divided into strata geographically.

In his research, Štulhofer (1999) speaks about the inherited obstacles that affect the

perception of privatization, thus distinguishing economic traditionalism that is

expressed in older generations which is prone to state paternalism, and one of the

characteristics of Croats is the escape from politics because of the perception that it

is a dishonest occupation. Situation ally induced barriers have led to an increase in

opportunism, especially among younger generations, which can be interpreted by

the rapid increase in social inequalities characteristic of all transition countries.

The indicators of Štulhofer (1999) are age (the older the respondent is, the longer he

or she is socialized in the former system and shows sympathy and sentiment towards

the social system of the time), a tradition whose main characteristic is a tendency for

traditional authority and civility that signifies individualistic values and belief in legal

and rational authority.

Research in 2018 To compare the results of the Štulhofer (1999) with the current attitudes of Croatian

citizens after 20 years, the survey was conducted in 2018. Respondents were randomly

selected from the telephone directory depending on the county they live in and a

total of 100 respondents participated in the survey. The research was conducted from

May 21st to June 25th, 2018. In its latest census for 2011, the Central Bureau of Statistics

(2013) states that 4,284,889 people live in Croatia. Due to differences in the number of

inhabitants per county, each county makes a single stratum, depending on the ratio

of its population to the number of residents in Croatia.

The survey collected data on the respondents' specific behaviour, attitudes,

opinions, desires, and expectations regarding privatization. It provides insight into the

indicators that influence attitudes towards privatization, as well as the relevance of

the media and the wider social environment to developing attitudes on the topic of

research. The survey questionnaire consists of a total of three questions that are part

of sociodemographic data and eleven questions related to the research topic.

On two questions, related to the main goals of privatization and typical features of

a Croatian entrepreneur, the respondents were entitled to multiple choices of

answers. On four questions, concerning the biggest winners and losers of privatization,

opinion on the need for privatization audit and privatization of public companies, they

had to opt for one of the offered answers. The remaining five issues (trust in core

institutions, satisfaction with privatization and negative effects of privatization, the

success of privatized companies and achievement of main goals) statements were

made that the respondents were expected to agree with. Processing and analysing

the data an image of the problem being researched is created and compared with

data from a previous survey.

6


In this research, sociodemographic data (gender, age, and education level) were

used to determine differences in respondents' perceptions of privatization. The

research showed that individuals with a graduate degree have the most negative

attitude towards privatization compared to respondents with other levels of

education. A total of 96% of respondents with completed primary or secondary

education consider job preservation as the main unfulfilled goal of privatization.

The results of both pieces of research are compared using the chi-square test,

which is utilized for investigating group differences based on the frequencies.

However, the data collected for the 2018 research allowed us to conduct additional

comparisons that are not presented in the Štulhofer (1999).

Results Geographically, the majority of respondents (53%) from four Dalmatian counties

consider socially-owned enterprises more successful than privatized ones, while

respondents from the City of Zagreb largely (50%) agree that privatized enterprises are

more successful. The most striking result of the research is the fact that respondents

from 18 to 35 years have the same negative attitude about the entrepreneur's

characteristics as those over 66 years so that only every seventh respondent between

the ages of 18 and 35 chose one positive characteristic of a Croatian entrepreneur.

Also, interestingly, it is the fact that in 92% of cases women provided answers that were

neutral in value (“I do not know, I cannot evaluate, nor do I agree or disagree”), unlike

men who gave more exact answers.

Generally, the media, which represented one of the two main sources of

information on the topic, had an important role in creating a perception of

privatization. However, interpersonal communication in the family and close social

groups was considered as much more valuable and reliable communication than the

“media campaign”. More precisely, salary non-payment, cancellations and small

shareholders blackmailing raised doubts about the truth about privatization. Namely,

based on the life experience of citizens as employees, the lack of confidence in the

official versions of the privatization results was justified (Šokčević, 2007). It was very

often in the service of the government that transmitting strictly controlled information

resulted in the suspicion of citizens.

The results of the research according to the areas of interest are presented below.

The confidence of Croatian citizens in basic institutions Confidence in the legal system of the Republic of Croatia and the Government has

fallen sharply. Namely, while in 1998 more than 80% of the respondents had a certain

level of trust in the Legal System and the Government (Table 3), in 2018 the percentage

of trust in the judiciary was only 5%, while the confidence in the Government of the

Republic of Croatia was 17% (Table 4).

Table3

The confidence of Croatian citizens in basic institutions (1996 and 1998)

Category The degree of agreement with the statement

Complete confidence Partial

confidence

Total distrust

1996. 1998. 1996. 1998. 1996. 1998.

Legal system 22% 9% 69% 76% 10% 15%

Croatian Government 20% 9% 69% 71% 11% 20%

Source: Štulhofer, A. (1999)

7


Table 4

The confidence of Croatian citizens in basic institutions (2018)

Statement

The degree of agreement with the statement

Complete

confidence

Partial

confidence Total distrust

I believe in the judiciary of the

Republic of Croatia. 5% 29% 66%

I trust the Government of the Republic

of Croatia. 17% 25% 58%

Source: Author's research (2018)

Note: Sample for 2018 is 100

Mistrust of both institutions of the system has increased in 2018. This can be explained

by the fact that in 1998 there was still fresh experience gained from the war and there

was optimism that the institutions would perform their tasks better in peacetime.

Additionally, there was an opinion that positive effects could only be observed over a

long period.

Chi-square test indicates that the differences between the confidence of Croatian

citizens in the legal system are statistically significant at 1% between 2018 and 1996

(χ2=68.290, p-value=0.000) and 1998 (χ2=54.292, p-value=0.000). Besides, the chi-

square test indicates that the differences between the confidence of Croatian citizens

in the Croatian government are statistically significant at 1% between 2018 and 1996

(χ2=52.853, p-value=0.000) and 1998 (χ2=43.016, p-value=0.000).

Overall satisfaction with the privatization Data related to the satisfaction of the privatization process were expected due to the

frequent topics of Croatian media on privatization fraud and a whole series of articles

on irregularities and abuses during the privatization of the company. The results

between 1998 and 2018 are not too different as can be seen in Table 5. In 2018, 10%

of respondents were somehow satisfied with the process of privatization so far, while

69% said they were dissatisfied. In 1998, the ratios were as follows: 69% of respondents

were dissatisfied, while 18% of respondents were satisfied with privatization so far. The

negative notion of privatization is evident, and it is obvious that it is a "solid popular

consensus" (Štulhofer, 1999). However, the chi-square test indicated that the

differences are not statistically significant (χ2=4.652; p-value=0.324).

Table 5

Satisfaction with the current process of privatization (1998 and 2018)

Year I completely

disagree

I do not

agree

I do not

know

I agree I completely

agree

Chi-square

2018 50% 19% 21% 9% 1% χ2=4.652

p-value=0.324

1998 47% 22% 13% 17% 1%

Source: Author's research (2018) and Štulhofer, A. (1999)

Note: Sample for 2018 is 100

The dissatisfaction with privatization in the last twenty years has led to a negative

perception of the respondents about further privatization and privatization of public

companies. Results from the surveys conducted in 1998 and 2018 can be seen in Table

6.

8


Table 6

Further privatization of public companies (1998 and 2018)

Statement 1998 2018 Chi-square

Public companies do not have to be privatized. 64% 72% χ2=7.012

p-value=0.030**

Public enterprises should certainly be privatized. 15% 20%

I do not know. 21% 8%

Source: Štulhofer (1999) and Author's research (2018)

Note: Sample for 2018 is 100; Sample for 1998 is 1001; ** statistically significant at 5%.

The percentage of respondents who oppose the privatization of public enterprises

has increased from 64% to 72%, although Croatia is among the countries with a large

share of state-owned enterprises, and the contribution to the budget of public

enterprises is relatively small. Interestingly, even though wages in public companies are

on average lower than wages in private companies, most respondents showed a

preference for work in a public company. The reason lies in the certainty of the

workplace and the impression of a job without stress and overwork. The fact that a

large number of jobs have been created through kinship relationships should not be

overlooked. The observed differences are statistically significant at 5% (χ2=7.012; p-

value=0.030).

Privatization effects A high percentage of respondents agree on the negative effects of privatization.

Opinions did not change significantly during all these years, as can be seen from Table

7.

Table 7

Negative effects of privatization (1996, 1998, 2018) - % of agreement

Statement

1996

1998

2018

Chi-square

(1996-2018)

Chi-square

(1998-2018)

The ultimate effect of

privatization is to

deepen the gap

between the rich and

poor.

67% 67% 66% χ2=0.029

p-value=0.864

χ2=0.029

p-value=0.864

People from powerful

parties have mostly

benefited.

71% 83% 91% χ2=12.995

p-value=0.000***

χ2=2.829

p-value=0.092*

Workers suffered the

most damage in

privatization.

82% 73% 81% χ2=0.033

p-value=0.856

χ2=1.807

p-value=0.179

Several families have

created “business

empires” suspiciously.

77% 86% 81% χ2=0.482

p-value=0.488

χ2=1.907

p-value=0.341

Source: Author's research (2018); Štulhofer (1999)

Note: Sample for 2018 is 100; *** statistically significant at 1%; * 10%

All respondents were generally (~67%) of the opinion that the ultimate effect of

privatization was definitely to widen the gap between rich and poor. Namely,

privatization brought the most benefits to the people from the parties in power, while

the workers of privatized enterprises were damaged mostly. The proportion of like-

minded people has grown progressively since 1996. Finally, in 2018, 91% of the

9


respondents classify powerful parties as privatization beneficiaries and 81% of them

consider workers as an affected party. About 80% of all respondents (regardless of the

survey year) also think that several families have created "business empires" in a very

dubious way. Chi-square indicates that there are no significant differences for most of

the statements. However, the attitudes of the respondents regarding the benefits of

the people from powerful parties are statistically different between the citizens in 2018

and 1996 at 1% (χ2=12.995; p-value=0.000) and between the citizens in 2018 and 1998

at 10% (χ2=2.829; p-value=0.092).

Table 8 presents the attitudes of the respondents regarding the competitiveness of

privatized companies in 1998. There are considerable doubt and divergence of

opinion regarding the increased competitiveness of privatized companies. As can be

seen at the bottom of Table 7, a total of 33% of respondents (in 1998, only 15% of

respondents according to the Table 8) believe that privatized companies are more

successful than former companies. This represents a major obstacle to the

development of entrepreneurship and demonstrates a tendency for state paternalism

that adversely affects the economic situation in the country. On the other hand,

almost 30% of respondents (in both surveys) were unable to assess the dependence

of business success on privatization. The reason for these results is reflected in the fact

that the population is ageing, the "Baby Boomer" generation is retiring, the negative

natural population growth is continuously increasing and some of the answers related

to the former socially-owned enterprises stem from sentiment towards youth and

subjective reasons.

Table 8

Competitiveness of privatized companies (1998)

Statement %

Social enterprises were more successful. 35

Privatized and socially-owned enterprises are equally successful. 21

Privatized companies are more successful. 15

I do not know; I cannot rate. 29

Source: Štulhofer (1999)

Objectives of privatization Table 9 presents the response of the citizens in 1998 regarding the achievement of the

main goals of the privatization.

Table 9

To what extent have the main goals of privatization been achieved? – Perception in

1998

Goals Not

achieved

I cannot

evaluate

Fully

achieved

Job preservation 80% 17% 3%

Employee welfare 83% 14% 3%

Business efficiency magnification 55% 36% 10%

Equitable distribution of social property 81% 16% 3%

The arrival of capable people in leadership positions 54% 36% 10%

Source: Štulhofer (1999)

Štulhofer (1999) emphasizes that Croatian citizens are characterized by economic

traditionalism, which is strongly reluctant to take private initiatives and, by relying on

state paternalism, tries to “freeze” the existing situation. Comparing the regulatory

10


objectives that the Government emphasized as the reasons for privatization and the

goals that citizens linked to the notion of well-being, a discrepancy is detected.

Preserving jobs remained recognized as a goal that was supposed to be primary in the

process of privatization, but 80% of respondents believe that job stability was not

preserved. Furthermore, one of the main goals of privatization should have been the

fair distribution of social property according to the criterion that it belongs to those

who created it. However, the equitable distribution of social property hasn’t been

fulfilled in the opinion of 81% of the respondents. The third objective was to achieve

the well-being of employees. However, this goal has not been achieved either

according to 83% of the respondents. Through privatization, it was also essential to

increase business efficiency and enable the entry of capable individuals into

management positions within the company. Though, slightly more than 50% of

respondents believe that privatization has failed in these two respects as well.

Table 10

To what extent have the main goals of privatization been achieved? – Perception in

2018

Goals Have not been

realized

I cannot

evaluate

They were

realized

Chi-square

(1998-2018)

Job preservation 84% 6% 10% χ2=9.128

p-value=0.010**

Employee welfare 77% 8% 15% χ2=9.861

p-value=0.007***

Business efficiency

magnification

45% 23% 32% χ2=15.384

p-value=0.000***

Equitable distribution of

social property

84% 3% 13% χ2=15.199

p-value=0.000***

The arrival of capable

people in leadership

positions

50% 19% 31% χ2=16.1164

p-value=0.000***

Source: Author's research (2018)

Note: Sample for 2018 is 100; *** statistically significant at 1%; ** 5%

Table 10 presents the response of the citizens in 2018 regarding the achievement of

the main goals of the privatization. Respondents' attitudes did not change significantly

in comparison with the ones that have been formed in 1998. Still, the goals considered

by the Croatian public to be of paramount importance in carrying out the privatization

process have largely not been met. About 80% of respondents believe that the main

objective hasn’t been accomplished, as well as the goals related to employee well-

being and fair distribution of social assets. On the other hand, opinions are divided

when it comes to increasing business efficiency and the presence of competent

leadership. Numerical, 45% of respondents believe that privatization has

underperformed in terms of business efficiency magnification while 32% consider it to

be successful on this point. The rest of the respondents (23%) remained restrained. The

arrival of capable people in leadership positions is rated as follows: 50% consider the

goal achieved, 31% disagree and 19% cannot evaluate the success of realization.

Often, respondents' reasoning about the (in)ability of people in managerial positions

at the time of social ownership is related to the planned economy and the lack of a

market element in economic life, as well as frequent perceptions of executives as

individuals without the responsibilities inherent in market leadership. Chi-square

indicates that the observed differences between the attitudes of respondents in 2018

and 1998 are statistically for all the observed statements.

11


The biggest winners vs. the biggest losses

Comparing data related to the winners and losers of the privatization process, a

significant change in public perception is noticeable (Table 11). Unlike the 1998 survey,

where, inter alia, members of only one ruling party were perceived as the biggest

winners according to 89% of respondents, in 2018, 52% of respondents perceived the

politicians as major winners regardless of their affiliation with the ruling or opposition

bloc, reflecting the public's view on the saturation of politics and perception described

by the syntagmatic "they are all the same". However, 39% generalizes - they believe

that political leaders have come out of privatization as greatest winners. Withal, only

9% cite the managers of private companies as the top winners while none of the

respondents holds the managers of state-owned enterprises as utmost privatization

winners. In addition to the members of one leading party, Štulhofer's respondents

estimated that the managers of the company, regardless of ownership, had made a

visible profit in the privatization process. More specifically, 87% consider managers of

private companies as winners and 78% managers of state-owned enterprises. The chi-

square test indicates that all of the observed differences between 1998 and 2018

attitudes regarding the biggest winners are statistically significant at 1%.

Table 11

The biggest winners in the privatization process so far - % of agreement

1998 2018 Chi-square

(1998-2018)

Politicians 80% 52% χ2=17.469

p-value=0.000***

Members of the ruling parties 89% 39% χ2=54.253

p-value=0.000***

Managers of privatized enterprises 87% 9% χ2=121.875

p-value=0.000***

Managers of state-owned companies 78% 0% χ2=127.846

p-value=0.000***

Source: Štulhofer (1999); Author's research (2018)

Note: Sample for 2018 is 100; *** statistically significant at 1%

The workers of privatized enterprises are considered the biggest losers according to

58% of the respondents in 2018. They also point out exporters as losers (35% of

respondents). In the 1998 survey, most respondents appear to have considered the

overall economic dynamics, not just those aspects that are a direct consequence of

the privatization process. So, 73% of the respondents put the peasants in the place of

the biggest losers (Table 12).

Table 12

The biggest losers in the privatization process so far - % of agreement


(1998-2018)

The workers of privatized companies 62 % 58 % χ2=0.333; p-value=0.564

Experts 49 % 35 % χ2=4.023; p-value=0.045**

Peasants 73 % 7 % χ2=90.750

p-value=0.000***

The workers of state-owned enterprises 34 % 0 % χ2=40.964

p-value=0.000***


Note: Sample for 2018 is 100; *** statistically significant at 1%; ** 5%

12


Privatization audit Privatization audit is often debated, especially after the enactment of the Law on Non-

Aging War Profit Offenses and Offenses of Transformation and Privatization. The

economic effects of a privatization audit would be doubtful because more than 20

years have passed since the largest privatization wave and economic trends and

technology have changed drastically. However, the results of this survey are consistent

with the 1998 survey but with one exception - respondents' perception that audit is

required solely in cases of large enterprises has increased dramatically (almost 7

times). This can be explained by the high resonance in the public through the media

about privatization frauds exclusively in large companies, while controversial cases of

privatization in small companies do not reach the general public (Table 13). Chi-

square test has indicated that the differences regarding the privatization audit issue

between 1998 and 2018 are statistically significant at 1% (χ2=35.061; p-value=0.000).

Table 13

Privatization audit issue - % of agreement


(1998-2018)

A revision of the entire privatization is required 65 % 67 % χ2=35.061

p-value=0.000***

The audit is required only in cases of large

enterprises

4 % 27 %

The audit is only required in known cases of breach

of law

19 % 5 %

No revision is required 4 % 1 %

I do not know 8 % 0 %


Note: Sample for 2018 is 100; *** statistically significant at 1%

Features of a Croatian entrepreneur Understandably, the perception of entrepreneurs in 1998 was negative because of

the recent negative effects of privatization and resentment of the failure to achieve

the set goals that citizens considered important to achieve well-being. However, there

is no major change in the perception of the Croatian entrepreneur today (Table 14).

Negative traits were considered to be typical traits of the average Croatian

entrepreneur, who is perceived as a tycoon credited with political eligibility,

immorality, and violation of the law (Štulhofer, 1999).

Table 14

Typical features of Croatian managers

Performance 1996 (rank) 1998 (rank) 2018 (rank)

Dishonesty 3 3 1

Exploitations of others 2 2 2

Political connections 1 1 3

Industriousness 4 5 4

Intelligence 5 4 5

Improving the economy 6 6 6


Note: Sample for 2018 is 100; 1 – the most relevant trait of a Croatian manager; 6 – the least

relevant feature of a Croatian manager

In addition to these attributes, entrepreneurs are considered highly dishonest,

exploitative and prone to politics to profit. They are neither distinguished by excessive

13


workmanship nor intelligence. Respondents throughout all these years consider that

Croatian managers contribute the least to the improvement of the economic

activities. Respondents' perceptions are also negative due to lack of encouragement

of entrepreneurship by institutions (example of the Government and administrative

barriers complained of by entrepreneurs or education that does not sufficiently

encourage the development of entrepreneurial mindset).

Conclusion Privatization was expected to bring fresh capital, a more efficient management

system, and investments into the impoverished companies (Petrović and Šonje, 2016).

To compare the changes in attitudes of Croatian citizens regarding the privatization,

the survey was conducted in 2018 and its results are compared to the research of

Štulhofer et al. (1999).

Observing Croatia's privatization with a time lag of 20 years, it is noticeable that

many of the privatization goals have not been accomplished. The main objectives of

privatization were supposed to include: job preservation, employee welfare, business

efficiency magnification, equitable distribution of social property and competent

leadership. But, considering that the main goals of privatization in the opinion of the

respondents were not achieved, there is a negative perception of the concept of

private property and a generally negative connotation to privatization in Croatian

society. Citizens' distrust of the system and institutions that conducted the privatization

process (which should lead the privatization process in the future) is high and is

increasing as time goes on. Mistrust is evident through every aspect of the research.

The majority does not support privatization in this form and considers it unnecessary.

Indeed, they believe that privatization caused significantly more negative effects than

positive ones. Most notable is diversity among the population in terms of wealth, the

evident privileged status and the associated benefits of powerful politicians and

deprivation of the working class. On the other hand, there has been very little benefit

from the increased competitiveness of transformed companies. Moreover, many

believe that state-owned enterprises were more successful. Particularly worrying is the

negative perception of the Croatian entrepreneur. Namely, the Croatian

entrepreneur is associated with dishonesty, exploitation of the environment, well-being

conditioned by political cohesion, laziness, lack of intelligence and a poor effect on

economic development.

Moreover, in many cases, there has been an exploitation of positions of power and

the permanent destruction of the acquired enterprise. Borrowing from credit

institutions was necessary since the privatization actors had virtually no sufficient

capital. The modus operandi consisted of borrowing new owners from banks since

they did not have any equity to buy shares in the companies. The accumulation of

funds that may have existed in certain companies for research and development,

new products or technological renewal spores is mostly inappropriately, and the

constant borrowing and buying up of new businesses maintained a vicious cycle of

survival (Gupta, 2005, Kraft and Jankov, 2005).

The limitations of this research that need to be taken into account are the following.

The most evident limitation appears in the form of sample size and the time limit at

which the survey was conducted. One of the constraints is certainly the fact that the

survey did not include the same respondents. In other words, the paper compared

results that did not include identical participants. Also, the subjectivity of the

respondents and the real familiarity with privatization is questionable for both surveys.

Namely, a lot of information that respondents possess is the result of media reporting

and expertise is left out. Therefore, the main recommendations for future research on

14


this topic could include the following: engaging more respondents and conducting

the survey over a longer period. Also, the survey questionnaire could be more detailed

and designed on a non-general basis. In other words, privatization could be observed

on a case-by-case basis and should involve impartial experts.

Ultimately, to conclude, respondents' perceptions have not changed significantly

since the 20-year gap, entrepreneurial thinking has not evolved, and the mentality of

the average Croat progresses slowly from a desire for state paternalism to market

conditions for an economic match. In the light of all the above, it is logical that most

respondents advocate for full privatization audit to correct numerous injustices and

damage inflicted to the biggest losers in privatization – workers and professionals.

References 1. Agency for Restructuring and Development (1992), “Annual report”, Agencija za

restrukturiranje i razvoj, Zagreb.

2. Arsov, S., Naumoski, A. (2014), “Transparency and Disclosure Practices in Selected Post

Transition Economies”. European Journal of Business and Management, Vol. 6, No. 16,

pp.146-154.

3. Bendeković, J., (2000), “Privatization in Croatia”, Economic review, Vol. 51, No. 1, pp. 55-90.

4. Bennett, J., Estrin, S., Urga, G. (2007), “Methods of privatization and economic growth in

transition economies 1” Economics of Transition, Vol. 15, No. 4, pp. 661-683.

5. Bjørnskov, C., Potrafke, N. (2011), “Politics and privatization in Central and Eastern Europe:

A panel data analysis 1”, Economics of Transition, Vol. 19, No. 2, pp. 201-230.

6. Bonneau, A. (2011) The effect of privatization method on inequality in Eastern

European transition economies. MSc thesis, Stanford University, California. 7. Central Bureau of Statistics (2013) Census 2011 Zagreb available at:

https://www.dzs.hr/Hrv/censuses/census2011/censuslogo.htm, (05, November, 2019)

8. Central Bureau of Statistics (2019), First release, Number and structure of business entities,

december 2018, available at: https://www.dzs.hr/Hrv_Eng/publication/2018/11-01-

01_04_2018.htm / (18 March, 2020)

9. Čengić, D., (1995), “Managers and privatization. Sociological aspects of enterprise take-

overs” Alinea, Zagreb.

10. Čengić, D., (2000), “The Ruling Elite and the Delegitimation of the Privatization Project”,

Društvena istraživanja - Časopis za opća društvena pitanja, Vol. 9, No. 4-5, pp. 497-525.

11. Čučković, N. (2002), “Siva ekonomija i proces privatizacije u Hrvatskoj 1997-

2001” Financijska teorija i praksa, Vol. 26, No. 1, pp. 245-271.

12. Franičević. V., (2002), “Political and moral economy in the first decade of the transition in

Croatia”, Politička misao , Vol. 39, No. 1, pp. 3-34.

13. Gatarić, Lj. (2019) “Tko je opljačkao Hrvatsku: komunistički direktori ili rodijački kapitalizam

HDZ-a?”, Večernji list, available at: https://www.vecernji.hr/vijesti/tko-je-opljackao-

hrvatsku-komunisticki-direktori-ili-rodijacki-kapitalizam-hdz-a-1364645, (16 March 2020)

14. Gregurek, M., (2001), “Level and effects of privatization in Croatia”, Economic review, Vol.

52, No. 1-2, pp. 155-188.

15. Gupta, N. (2005), Partial privatization and firm performance. The Journal of Finance, Vol.

60, No. 2, pp. 987-1015.

16. Haramija, P., Njavro, Đ., (2016), “The Transition and its Results – Why the Transition from the

Communist to the Democratic System of a Market Economy did not Meet Expectations”,

Obnovljeni život: časopis za filozofiju i religijske znanosti, Vol. 71, No. 4, pp. 515-527.

17. Iveković, M. (2012), “Privatizacija u Hrvatskoj: jučer, danas, sutra”, Zarez, 05.12.2012.

available at: http://www.zarez.hr/clanci/privatizacija-u-hrvatskoj-jucer-danas-sutra, (16

March 2020)

18. Jelić, N. Plišić, A., Vresnik, V., (2006) “Kraj hajke na privatizaciju” Jutarnji list, available at:

https://www.jutarnji.hr/arhiva/kraj-hajke-na-privatizaciju/3754959/ (16 March, 2020)

15


19. Josipović, I. (2018), “Zastara u predmetima ratnog profiterstva i kriminala u pretvorbi i

privatizaciji i kontroverze nakon odluke Ustavnog suda u predmetu Hypo”, Hrvatski ljetopis

za kaznene znanosti i praksu, Vol. 25, No. 2, pp. 197-259.

20. Kalogjera, N., (1993),"Privatization in the stabilization and development of Croatian

economy" Društvena istraživanja-Časopis za opća društvena pitanja, Vol. 2, No. 1, pp. 51-

86.

21. Kraft, E., Jankov, L. (2005), “Does speed kill? Lending booms and their consequences in

Croatia”. Journal of Banking & Finance, Vol. 29, No. 1, pp. 105-121.

22. Lasić, V. (2000), “Koncept modela privatizacije u Hrvatskoj” Ekonomski pregled, Vol. 51, No.

1-2, pp. 107-117.

23. Nikić G., (2004), “Economic aspects of Croatia's accession to the European union - passed

ten and the forthcoming ten years”, Economic review, Vol. 55, No. 3-4, pp. 197-226.

24. Njavro, Đ., (1993),“Privatization”, Journal for general social issues, Vol. 2, No. 1, pp. 87-105.

25. Olgić Draženović, B., Kusanović, T. (2016), “Determinants of capital market in the new

member EU countries.” Economic research - Ekonomska istraživanja, Vol. 29, No. 1, pp. 758-

769.

26. Ožanić, M. (2016) “Pretvorba i privatizacija (3) – Prtljaga s kojom smo ušli u kapitalizam i

privatizaciju” Sve o poduzetništvu, 5.08.2016., available at:

http://www.sveopoduzetnistvu.com/index.php?main=clanak&id=78 /(16 March 2020)

27. Pejić Bach, M., Dumičić, K., Jaković, B., Nikolić, H., Žmuk, B. (2018), “Exploring impact of

economic cost of violence on internationalization: Cluster analysis approach” International

journal of engineering business management, Vol. 10, No. 1, 1.

28. Petrović, S., Šonje, V. (2016) “Je li privatizacija državnih poduzeća nužan uvjet njihovog

uspješnog poslovanja u Hrvatskoj?”. Privredna kretanja i ekonomska politika, Vol. 25, No. 1,

pp. 37-88.

29. Protulipac, A. (2014), “Privatizacija u Hrvatskoj – tijek, posljedice, mitovi” Slobodni filozofski,

available at: http://slobodnifilozofski.com/2014/05/armin-protulipac-privatizacija-u.html

/(16 March, 2020)

30. Rogić, I., Zeman, Z., (1998), “Privatizacija i modernizacija”, Institut društvenih znanosti lvo

Pilar, Zagreb.

31. Stojčić, N. (2012), “Two decades of Croatian transition: a retrospective analysis.” South East

European Journal of Economics and Business, Vol. 7, No. 2, pp. 63-76.

32. Šokčević, Š., (2007), “Privatization of State-Owned Property - (Un)Successful Proces of

Tranzition? Social - etical judgement?”, Diaconvensia: teološki prilozi, Vol. 15, No. 1, pp.103-

154.

33. Štulhofer, A., (1998), "Krivudava staza hrvatske privatizacije", in Rogić, I., Zeman, Z. (Ed.)

Privatizacija i modernizacija, Institut društvenih znanosti lvo Pilar, Zagreb, pp. 163-177.

34. Štulhofer, A., (1999), “Proces privatizacije i hrvatska javnost 1996.-1998.”, in Čengić, D. (Ed.)

Privatizacija i javnost, Institut društvenih znanosti lvo Pilar, Zagreb, pp. 1-31.

35. Vukšić, G. (2016), “Effects of private ownership, trade, and foreign direct investment on

labour productivity growth in transition economies: evidence from the Croatian

manufacturing industry” Emerging Markets Finance and Trade, Vol. 52, No 2, pp. 322-335.

36. Vušković, P. (2020), “Petar Vušković o privatizaciji: Nitko od stranaca nam neće ništa uzeti

– ni vode, ni ceste ni šume. Bojimo se bez veze.”, available at: https://lider.media/poslovna-

scena/hrvatska/petar-vukusic-o-privatizaciji-nitko-od-stranaca-nam-nece-nista-uzeti-ni-

vode-ni-ceste-ni-sume-bojimo-se-bez-veze-129836 / (16 March, 2020)

16


About the author

Jan Horaček he finished elementary school and gymnasium in Zagreb. After

completing the second year at the Faculty of Political Science, he switched to the

Professional study of business economics - Trade Business Operations. Since 2015, he

has been working in Electrolux. First as Brand ambassador, afterword he became

Event and promoter coordinator and since 2018 he works as Sell out Specialist for

continental Croatia. The author can be contacted at [email protected].

Helena Nikolić graduated from the Faculty of Economics and Business of the University

of Zagreb, where she also received her PhD degree on “Determinants of export

activities of Croatian companies in the Eastern Europe countries” in 2015. Helena holds

a position of Assistant Professor at the Faculty of Economics and Business Zagreb,

Department of Trade and International Business. Previously, she had been working in

Croatian Bank for Reconstruction and Development, in the Export Credit Insurance

Department, for two years. The author can be contacted at [email protected].

17


Performance Measurement of Vietnamese

Publishing Firms by the Integration of the GM

(1,1) Model and the Malmquist Model

Xuan-Huynh Nguyen

Vietnam National University, Vietnam

Quoc Chien Luu

Thanh Dong University, Vietnam

Abstract

Background: In the new technology context, the publishing industry cannot continue

to maintain its business operations and to develop relying solely on traditional product

offerings, such as books, magazines, and newspapers. There needs to be an

expansion into innovative products, such as e-books, micro-publishing, and websites.

Objectives: The paper addresses the factors influencing financial reports of

Vietnamese publishing firms using two methodological approaches, namely the Grey

first-order one variables (GM,1,1) model in the Grey theory and the Malmquist model

in the data envelopment analysis (DEA). Methods/Approach: The GM(1,1) model

predicts the future period of 2020–2023 based on the historical time series analysis. The

Malmquist model presents catch-up, frontier-shift, and Malmquist Productivity Index

(MPI) in whole terms. Results: The analysis provides an overview of the publishing

industry in Vietnam. The final empirical results show that twelve companies reached a

production efficiency higher than 1 and fourteen companies are expected to attain

a productivity score higher than 1. Conclusions: Only a few firms do not need to

change significantly; however, the remaining firms must re-evaluate their current

operations.

Keywords: Vietnamese publishing firms; GM(1,1) model; Malmquist model; production

efficiency

JEL classification: G17, N25, P34


Received: 15 Nov 2020

Accepted: 21 Mar 2021

Citation: Nguyen, X-H., Luu, Q.C. (2021). “Performance Measurement of Vietnamese

Publishing Firms by the Integration of the GM (1,1) Model and the Malmquist Model”,

Business Systems Research, Vol. 12, No. 1, pp. 17-33.


The acknowledgments: The authors would like to thank the editors and reviewers for

their constructive comments related to this article.

18


Introduction Industry 4.0 affects directly and deeply the publishing industry because it is an

effective support tool for the rapid transfer of information. Hence, the number of

traditional publications has been reduced and replaced by electronic publications.

As the publishing industry applies Industry 4.0, it can bring a high degree of

effectiveness around the world. Nowadays, Vietnam has adopted new globally

achieved techniques to catch up with the growth of electronic publishing and e-

books on the internet.

The development process of the publishing industry has always met with difficulties

in the process of change, regarding innovative technology implementation and

economic growth (Lacy, 1979). In recent years, digital publishing has had a great

effect on publications’ market shares (Lin, Chiou & Huang, 2013; Sara & Markus, 2014),

and publishing firms need to have a production plan, schedule and control, inventory

management, and reverse logistics (Meysam, Mohammad, Ebrahim & Ali, 2013).

Besides, they need to investigate the consumer’s requirements and optimize excess

product offerings (Anat, John & Pat, 2003; Edoardo, Antonello & Laura, 2008). Any

nation applying digital techniques to the publishing industry also builds up its growth

strategies (McCready & Molls, 2018; Edelmann & Schoßböck, 2020). Operational

strategies are a key factor for the maintainability and sustainability development

process to occur so that the publishing company can receive good revenues. A

publishing firm of educational books (Lee & Liang, 2018), media content (Alexander

& Thomas, 2007), software (Matthew, 2008) makes an effort to overcome specific

challenges and reach full effectiveness. These previous studies have explored the

development of the global publishing industry via many different methods. In this

study, we utilize GM(1,1) model in the Grey theory system and the Malmquist model in

DEA.

GM(1,1) model is a forecasting tool that can deal with the minimum historical time

series as four terms. The Grey theory supports solving characteristics of poor and

insufficient information (Wu, Liu, Fang & Xu, 2015), thus it is useful to deal with the lack

of available data. Previous research has utilized the GM(1,1) model for predictive

purposes. For instance, Liu, Peng, Bai, Zhu & Liao (2014) forecasted future values of the

factors affecting the tourist flow by GM(1,1) model. Maciej & Czeslaw (2015) used a

set of GM(1,1) models to predict values of vibration symptoms of fan mills in a

combined heat and power generation plant. Qian & Wang (2020) approached the

GM(1,1) model to predict wind power generation in China based on the historical

data from 2013 to 2019. Nguyen (2020) utilized the GM(1,1) model to forecast the

operational efficiency of Vietnamese construction companies. Nguyen, Le, Ngo &

Hoang (2020) investigate the business efficiency of global electric cars in 14 countries

around the world.

Data envelopment analysis (DEA) is a statistical analysis method that can be

applied to various industries, such as economics (Gunes & Guldal, 2019), education

(Montoneri, Lin, Lee & Huang, 2012), manufacturing (Ehsan & Hadi, 2014), banking

(Bošković & Krstić, 2020), and others. Therefore, DEA is a useful method allowing the

efficiency of a specific decision-making unit (DMU) to be measured by the ratio

between virtual output and virtual input. The relative efficiency of a homogenous set

of a specific DMU is computed by the presence of multiple inputs and outputs. DEA

has many different models so that each model may have a separate functionality. For

example, Liu & Wang (2008) evaluated the efficiency of Taiwanese semiconductor

companies when using the Malmquist model with three components, including

technical change, frontier forward shift, and frontier backward shift. Chen, Tzeremes

& Tzeremes (2018) discovered the productivity levels of the Chinese airline market

19


through the Malmquist model. Mavi, et al., (2019) found out the efficiency of freight

transportation in Iran and determined a declining trend in terms of eco-innovation

and environmental efficiency when observed by the Malmquist model approach.

Wang, Tibo, Nguyen & Duong (2020) measured the relative performance of New

Zealand universities by use of a Malmquist productivity approach.

This study uses the GM(1,1) model to forecast the main factors of financial reporting,

including current assets, non-current assets, fixed assets, liabilities, owner’s equity, net

revenue, gross profit, and net profit- after-tax in Vietnamese publishing firms. With the

actual and forecasted data, the DEA has been applied. Whereas to have a good

economic comparison of productivity efficiency among Vietnamese publishing

companies, the Malmquist model in DEA is used to compute the score in every term,

and the average score in whole terms. The estimated values throughout 2020–2023

are predicted utilizing the GM(1,1) model based on historical data obtained from 2015

to 2019, and then the productivity efficiencies in both previous and future terms are

conducted by the Malmquist model. Combining the two above models, namely

GM(1,1) model, and the Malmquist model, we can figure out the overall picture of

Vietnamese publishing firms from the past to the near future. The empirical results

reveal effective variations and identify the operational trends. Research results may

serve as references that can help publishing firms in emerging economies foresee their

operational processes and make a suitable plan for future development. Besides, the

investors may choose the best partners in future terms for better profitability based on

the presented analysis.

The paper is arranged as follows. The introduction gives an overview of the

publishing industry, the background of GM(1,1), and the Malmquist model. The

second chapter sets up the conceptual research, materials, and methods. The third

chapter presents the results of the empirical results. The fourth chapter discusses the

main analysis results, while the last chapter summarizes the key empirical results, and

it indicates the limitations and future research directions.

Materials and methods Research process The research process is carried on a step-by-step basis, as shown in Figure 1.

Step 1. The study determined the objective research, and then it collected all

related data including inputs and outputs. All selected data must be positive values,

and they must be removed and reselected if they are positive values.

Step 2. The estimated data from 2020 to 2023 are calculated by a GM(1,1) model

based on the primary data. These predicted values must check the accuracy level

through the Mean Absolute Percentage Error (MAPE). All forecasted values owning

unsuitable MAPE are removed and used with another forecast model.

Step 3. The Malmquist model in DEA is used for conducting the efficiency score and

determining the position of the publishing firms under consideration. Before the actual

and forecasted data are utilized to conduct the scores, they must test the Pearson

correlation between variables. Any unappreciated values must be returned and re-

selected as other values. The appreciated data is applied to computing the

productivity efficiency from past to future time.

Step 4. Major analysis results are given and discussed, which can be used to define

the extent of efficient and inefficient cases.

20


Figure1

Research framework

Source: Author’s illustration

Materials The global publishing industry includes newspapers, periodicals, books, directories,

and software. The growth of companies in the Vietnamese publishing industry is

analyzed based on the actual data published on Vietstock (2020).

The purpose of this study is to measure the financial performance of publishing firms

in Vietnam throughout 2016–2023. The input and output variables of nineteen

Vietnamese publishing firms are taken and then collected from 2016 to 2019. The

names of nineteen companies are given, as shown in Table 1.

21


Table 1

List of publishing firms

No. Firm code Name of publishing firms

1 ADC Art Design And Communication JSC

2 BDB Binh Dinh Book & Equipment Joint Stock Company

3 BED Danang Books & School Equipment JSC

4 BST Binh Thuan Book And Equiptment JSC

5 DAD Da Nang Education Development & Investment JSC

6 DAE Educational Book JSC in Da Nang City

7 EBS Educational Book JSC in Hanoi City

8 EID Education Cartography And Illustration JSC

9 HAP Hanoi Education Development & Investment JSC

10 HBE Ha Tinh Book And Equipment Education JSC

11 KBE Higher Education And Vocational Book JSC

12 LBE Long An School Book & Equipment JSC

13 NBE North Books and Educational Equipment Joint Stock Company

14 QST Quang Ninh Book & Educational Equipment JSC

15 SED Phuong Nam Education Investment & Development JSC

16 SGD Educational Book JSC in Ho Chi Minh City

17 SMN South books and Educational Equipment JSC

18 STC Book & Education Equipment JSC Of HCMC

19 TPH Ha Noi Textbooks Printing Joint Stock Company

Source: Vietstock (2020)

The following input and output factors have been taken into the consideration,

which are the key parts of a financial statement that can help to give a deeper

identification of an enterprise’s operational enterprise. These indexes equip to

calculate the performance in operating progress that displays upward or downward

trends in each term. Each of the highest and lowest efficiency points is identified to

observe the worst or best business status in every term. All historical data are gathered

from Vietstock and summarized as shown in Table 2. Table 2 indicates that the

minimum value of CA, NA, FA, LS, OE, NR, GP, and PT is 8623; 1463; 621; 1342; 12698;

16619; 3911; 693, respectively. Therefore, all historical data realize positive values, they

are appreciable to use for forecasting future data via the LTS (A, A, A) model and to

measure the efficiency via the Malmquist model as well as the GM(1,1) model.

Input factors:

o Current assets (CA): Cash, cash equivalents, accounts, stock inventory,

marketable securities, and other liquid assets.

o Non-current assets (NA): The long-term investments made by a specific

publishing firm.

o Fixed assets (FA): The long-term tangible assets that are comprised of property,

facilities, and equipment.

o Liabilities (LS): Loans, mortgages, deferred revenues, bonds, warranties, and

accrued expenses.

o Owner’s equity (OE): The owner’s investment in the operational business of the

enterprise.

Output factors:

o Net revenue (NR): Total sales of a company are received from selling the goods.

o Gross profit (GP): The profit of a publishing firm after deduction of the charges

for making and selling products.

22


o Net profit-after-tax (NT): The net income of a publishing firm after deducting

taxes.

Table 2

Historical data of publishing firms

Code CA NA FA LS OE NR GP PT

2016

Max 283150 140198 41922 190209 233138 516773 142396 36504

Min 8623 1872 940 1764 12959 16619 3911 1072

Ave 64480 29805 10271 32599 61687 180175 38930 8512

SD 72056 30062 10126 43442 57503 154352 38439 10030

2017

Max 306786 121173 40172 181606 246353 577062 159499 36223

Min 10561 1590 621 2164 12779 19203 4107 760

Ave 70959 28218 10285 33976 65201 194884 43002 8637

SD 85262 26327 11104 45775 63944 165895 43333 10016

2018

Max 340056 106057 38331 183902 262211 599103 163408 40947

Min 11084 1463 864 1487 12698 23200 4430 707

Ave 76252 26824 11585 34043 69034 206116 45731 10609

SD 92511 23979 10775 48671 67717 176569 46316 11864

2019

Max 372315 95769 38590 185068 283016 652590 174643 45363

Min 12405 1484 747 1342 12993 26322 5464 693

Ave 77590 28906 12777 36628 69868 225307 49314 9702

SD 99761 24275 11549 53582 71268 197027 51390 11845

Note: Ave: average; SD: Standard deviation

Source: Vietstock (2020)

GM(1,1) model The GM(1,1) model is a forecasting tool based on the Grey theory system utilizing the

previous time series (Deng, 1989). This model only calculates future data when the

historical time series maintains positive data and is computed in the following order:

From the primary data(0) (0) (0) (0)( (1), (2),..., ( ))A A A A n= , the consequence

(1)A is

calculated by:

1 (0)( ) (1) / ( 0,1,..., )k

i

A h A h n= = (1)

where,

1 0

1 0 0

1 0 0 0

0, (1) (1)

1, (2) (1) (2)

, ( ) (1) (2) ... ( )

h A A

h A A A

h n A n A A A n

= =

= = +

= = + + +

when having(1)A series, the mean equation

(1)Z is built up:

(1) (1) (1)1( ) ( ( ) ( 1)) / ( 1,2,..., )

2A h A h A h h n= + − = (2)

where,

23


(1) (1) (1)

(1) (1) (1)

11, (1) ( (1) (0))

2

12, (2) ( (1) (2))

2

= = +

= = +

h A A A

h A A A

(1) (1) (1) (1)1, (2) ( (1) (2) ( ))

2= = + +h n A A A A n

The mathematical equation for a and b is determined by:

(1) (1)( ) ( ) / ( 2,3,..., )A h a Z h b h n+ = = (3)

where a and b are coefficients.

The linear equation of a matrix is presented by:

(0) (0)

(0) (0)

(0)(0)

1

(2) (2) 1

(3) (3) 1, / ( 1, 2,... )

( ) 1( )

( )T T

A Z

A ZE D h n

Z hA h

aand D D D E

b −

−

− = = =

−

=

(4)

The whitening equation is formed:

(1)(1)dA

a A bdt

+ = (5)

Set up the estimated values (0) (0) (0) (0)

(1), (2),..., ( ) ( 0,1,2,..., )A a a a n n n= = , the

predicted equation is built up:

(1)(0)( 1) (1) / ( 1,2,.., )ahb b

A h A e h na a

− + = − + =

(6)

The forecasted values must examine the accuracy level via the mean absolute

error percentage (MAPE):

(0)(0)

(0)1

100 ( ) ( )/ ( 1, 2,.., )

( )

n

t

A h A hMAPE h n

n A h=

−= = (7)

According to Lewis (1982), the MAPE indicator may be used to distinguish four

groups, such as the excellent group (smaller than 10%); good group (10-20%);

reasonable group (20-50%); and, poor group (higher than 50%). The unsuitable tested

values require the usage of another forecast model or reselection of the primary data.

Malmquist model The Malmquist model defines the efficiency change of a DMU between two

consecutive periods (Tone, 2004). This study uses the Malmquist-radial model for

measuring the efficiency of publishing firms in Vietnam over the period from 2016 to

24


2023. The distance functions of inputs and outputs at t and 1+t is 0 0 0( , )tDF x y , and

1

0 0 0( , )+tDF x y , respectively. The catch-up effect is estimated as follows:

1 1 1

0 0 0

0 0 0

( , )

( , )

t t t

t t t

DF x yCUI

DF x y

+ + +

= (8)

The frontier-shift effect is computed:

1/21 1

0 0 0 0 0 0

1 1 1 1

0 0 0 0 0 0

( , ) ( , )

( , ) ( , )

t t t t t t

t t t t t t

DF x y DF x yFSI

DF x y DF x y

+ +

+ + + +

=

(9)

The Malmquist Productivity Index (MPI) has four terms including0 0 0( , )t t tDF y ,

1

0 0 0( , )t t tDF y+ , 1 1

0 0 0( , )t t tDF y+ + and 1 1 1

0 0 0( , )t t tDF y+ + + , the mathematical equation of MPI is

calculated:

1/21 1 1 1 1

0 0 0 0 0 0

1

0 0 0 0 0 0

( , ) ( , )

( , ) ( , )

t t t t t t

t t t t t t

DF x y DF x yMPI

DF x y DF x y

+ + + + +

+

=

(10)

Let input and output matrices at the period ( )p as 1( , )p p p

nX x x= and 1( , )p p p

nY y y=

respectively. The input-oriented radial MPI is presented by the scores ( ) that are given

by the linear programs as follows:

, 0 0min ( , )p p pDF x y = (11)

where,

0 0, , 0p p p px X y Y

Results Based on the collected data of nineteen Vietnamese publishing companies, we

predict the future values. When having appreciated actual and forecasted data, the

technical efficiency scores are computed by the Malmquist model.

Estimated values The primary data of nineteen publishing companies in Vietnam is used for predicting

the forecasted values utilizing the GM(1,1) model. We utilize the input variable (CA) of

the ADC firm to illustrate the predictive process.

Let the primary time series(0)A .

(0) 80,055;88,350;96,533)(66,306;A = (12)

Calculate time series(1)A .

(1) 66,306;1 46,361; 234,711; 331,244)(A = (13)

Count the mean sequence(1)Z .

(1) 1 06,334;1 90,536; 282,978)(Z = (14)

Formulate a and b.

25


80,055 1 06,334

88,350 1 90,536

96,533 282,978

b a

b a

b a

= +

= +

= +

(15)

Compute the linear equation.

80,055 106334106334 19

,0536 282978

88,350 190536

96,533 2

1

, 11 1 1

182978

E D ET

= = =

− −

−

−−

− (16)

Determine a and b values.

1 0(

5

.093)

70297

1

.5

2T TaD D D E

b − = =

−

(17)

Formulate the whitening equation.

66,3060.09321 66,306 70297.55

d

dt =−+ (18)

Estimate the forecasted value.

(0)

(0)

(0)

(0)

(0)

(0)

(0)

(0)

66,306

2, 80,155

3, 87,985

4, 96,580

5, 1 06,015

6, 1 16,372

7, 1 27,740

8, 1 40,218

1, (1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

h A

A

A

A

A

A

h

A

A

h

h

h

h

h

h

= =

=

=

=

=

=

=

=

=

=

=

=

=

=

=

(19)

Upon application of these steps, the forecasted data are formulated, these

forecasted data are summarized in Table 3; however, all of them must check the

MAPE indicator to ensure the accuracy standard is adhered to, as shown in Table 4.

Table 3

Forecasted data of publishing firms from 2020 to 2023

Year /

Statistics

CA NA FA LS OE NR GP PT

2020

Max 410263 102874 79266 213281 302679 689545 187637 50790

Min 12416 1408 665 964 13040 30861 5691 590

Ave 82486 29624 16680 38760 72943 242247 52938 10804

SD 108044 25973 19309 58584 75373 214358 56121 13024

2021

Max 451747 135352 191741 249647 324517 733992 215770 56788

Min 10247 1358 588 744 13150 36003 6157 464

Ave 87235 31663 24782 41963 75682 262469 57114 11524

SD 117201 30937 42719 65139 79809 235549 61993 14273

26


Table 3

Forecasted data of publishing firms from 2020 to 2023 (continued)

2022

Max 497425 178083 463814 292213 347930 781304 248122 63495

Min 8456 1311 380 574 13260 42001 6619 365

Ave 92597 34661 42853 46007 78602 285084 61796 12340

SD 127279 39210 102511 73305 84600 259455 68815 15699

2023

Max 547723 234305 1121946 342037 373032 831665 285323 70993

Min 6979 1265 216 299 13372 48519 6223 287

Ave 98616 38867 83851 51063 81714 310374 67054 13267

SD 138380 51210 248416 83524 89773 286417 76732 17327

Source: Author’s calculation

MAPE indicators in Table 4 show that the minimum and maximum values are

0.30504% and 12.13654%, respectively. According to Lewis (1982), we could conclude

that the highest value of MAD (12.1365 for firm BAD) is still acceptable. Other firms

obtain an excellent level because their MAPEs are lower than 10%. Besides, the

average MAPE for all observations is 3.00176%. Therefore, the estimated data

summarized in Table 3 have a good forecasting performance. These values are

appreciated to apply to the Malmquist model in DEA.

Table 4

MAPE indicators (%)

DMU codes MAPE

ADC 1.08478

BDB 1.78103

BED 12.1365

BST 3.46064

DAD 3.12158

DAE 2.62495

EBS 0.66254

EID 0.36527

HAP 2.12362

HBE 0.96593

KBE 2.40343

LBE 3.1183

NBE 6.17556

QST 0.61211

SED 1.77095

SGD 0.83829

SMN 8.7351

STC 0.30504

TPH 4.74778

Average 3.00176


27


Performance measurement With the purpose of efficiency measurement, all actual and predicted data are used

in the Malmquist model. These variables need to contain not only positive data but

also demonstrate a significant correlation; thus, they must be checked using Pearson’s

correlation. The relationship between input and input, input and output, output and

output must be isotonic, with the correlation coefficient limitation demonstrated from

-1 to +1. The relationship is defined as good relation when it is close to 1 . In this study,

the correlation ranges from -0.2286 to 1, which is acceptable for the Malmquist model.

As shown in Table 5, there are six DMUs including ADC, BED, BST, NBE, SED, and SMN

that indicate no significant progress of the technical efficiency in the entire previous

period, as well in the future terms. HAP reached the efficiency from 2016 to 2017, but

it declines and maintains a performance score value of 1 from 2018-to-2023. LBE keeps

a stable state but predicted as an inefficient score. Although the score of TPH

fluctuates sharply, it always obtains efficiency in the whole term. Other companies

exhibited minor downward and upward trending in every term; therefore, they did not

attain adequate performance in the whole term. Whereas, BDB, DAD, EID, KBE, and

QST are expected to extend the scores and obtain the efficiency scores in all of the

future terms. This result implies that these firms will achieve good conditions. DAE, EBS,

HBE, LBE, SGD, and STC are seen to exhibit a downward trend in the future. The

empirical result denotes that their operational process can be greatly affected by

fluctuations occurring in the previous term.

Table 5

Catch-up performance

DMU

code

2016

=>2017

2017

=>2018

2018

=>2019

2019

=>2020

2020

=>2021

2021

=>2022

2022

=>2023

ADC 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

BDB 1.0018 0.6581 1.0986 0.9512 1.1013 1.1121 1.1111

BED 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

BST 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

DAD 1.0000 0.7901 1.2656 0.9070 1.0776 1.0231 1.0000

DAE 1.0000 1.0000 1.0000 1.0000 0.9568 0.8884 0.8119

EBS 1.2355 0.9988 1.0012 0.9682 0.8870 0.8751 0.8726

EID 0.9613 1.0571 1.0629 1.0000 1.0000 1.0000 1.0000

HAP 1.2384 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

HBE 0.9572 0.6463 0.9967 0.8148 1.0398 0.9653 0.9804

KBE 1.0000 0.9705 1.0304 1.0000 1.0000 1.0000 1.0000

LBE 1.0000 1.0000 1.0000 1.0000 1.0000 0.9992 0.9825

NBE 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

QST 1.0000 1.0000 1.0000 0.9815 1.0189 1.0000 1.0000

SED 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

SGD 1.1810 0.8910 1.0023 0.9367 0.9301 0.9569 0.9652

SMN 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

STC 1.0294 0.8160 1.0887 0.8823 0.9800 0.9797 0.9827

TPH 1.0679 1.0147 1.6431 1.2340 1.3690 1.3171 1.0000


Operational progress corresponding to the effect of technology innovation, raw

material and product price, etc., makes the financial indicators of publishing firms the

efficiency scores of DMUs change. They are measured by the frontier-shift index, as

shown in Table 6. There are five DMUs including BDB, EBS, EID, HAP, and SMN, which

always gain efficiency, as indicated by the scores above 1 from 2016-to-2023. The

28


remaining firms have a high degree of changing scores between 0.5637 and 2.0519.

These values describe an unstable movement.

Table 6

Frontier-shift efficiency score

DMU

code

2016

=>2017

2017

=>2018

2018

=>2019

2019

=>2020

2020

=>2021

2021

=>2022

2022

=>2023

ADC 1.2351 0.7331 0.9892 0.9032 0.9198 0.9256 0.9437

BDB 1.0111 1.4231 1.0397 1.1510 1.0465 1.0418 1.0483

BED 0.9722 2.0519 0.5637 1.4850 1.0666 1.0724 1.0744

BST 1.1478 0.9141 1.1399 0.9972 1.0510 1.0481 1.0448

DAD 0.9553 1.4672 0.7055 1.2807 1.0608 1.1029 1.1227

DAE 1.0548 1.0955 0.8127 0.9986 1.0179 1.0911 1.1867

EBS 1.0944 1.0423 1.0032 1.0791 1.1526 1.1714 1.1843

EID 1.0152 1.0667 1.0179 1.3091 1.4229 1.4206 1.4137

HAP 1.0981 1.1891 1.2766 1.1755 1.2109 1.2368 1.2476

HBE 0.9769 1.4967 1.1211 1.267 0.9890 1.0632 1.0562

KBE 1.0826 1.0900 0.9938 1.0573 1.0393 1.0417 1.0457

LBE 1.0114 1.0232 0.8319 1.0115 0.8654 0.9067 0.9555

NBE 1.2593 0.9673 1.1038 1.6175 1.2226 1.1427 1.0904

QST 0.8738 1.0276 0.9710 1.0809 1.0443 1.0727 1.0774

SED 1.1222 1.3987 0.8772 1.0888 1.1050 1.1106 1.1146

SGD 1.0221 1.1774 0.9169 1.0845 1.0690 1.0435 1.0415

SMN 1.0064 1.0364 1.2014 1.1663 1.1913 1.1926 1.1933

STC 1.0859 1.1961 0.8906 1.1891 1.0830 1.0741 1.0603

TPH 0.9103 1.3879 0.8651 1.1153 1.0286 0.9985 1.2482


After the catch-up and frontier-shift scores are provided, the Malmquist

productivity index (MPI) is calculated through catch-up and frontier-shift techniques,

as seen in Table 7. Four DMUs, including EBS, HAP, KBE, and SMN, attain the

performance in the whole term and have a slight change of their efficiency score

between terms from 1.0044 to 1.36. The remaining DMUs have a variation of scores

that both increase and decrease continuously and slightly. In addition, they own both

efficient and inefficient terms over the period 2016–2023.

ADC and BDB experience the same situation by showing an upward and

downward trend in every year in the previous time; and, they are expected to extend

consecutively in the future time. ADC demonstrated efficiency with 1.2351 in 2016–

2017. BDB obtained adequate performance in most terms except during 2017–2018.

The historical scores of BED, BST, DAD, EID, HBE, NBE, QST, SED, STC, and TPH

increased and decreased slightly, with both efficient and inefficient terms being

present. All of the firms are expected to hold their technical efficiency in the future

period of 2020–2023. The score of SGD exhibits a dramatic consecutive change in

which efficient and inefficient scores occur in long term.

Although DAE, ADC, and LBE generate a marked change in progress during the

past, with both efficient and inefficient terms being present, they reveal that they will

not likely reach adequate performance demonstrated by scores under 1 in the future.

These publishing firms will have the lowest operational progress and require definite

and immediate improvement.

29


Table 7

Malmquist productivity efficiency score

DMUs 2016=>

2017

2017=>

2018

2018=>

2019

2019=>

2020

2020=>

2021

2021=>

2022

2022=>

2023

ADC 1.2351 0.7331 0.9892 0.9032 0.9198 0.9256 0.9437

BDB 1.0129 0.9366 1.1422 1.0948 1.1525 1.1586 1.1647

BED 0.9722 2.0519 0.5637 1.485 1.0666 1.0724 1.0744

BST 1.1478 0.9141 1.1399 0.9972 1.0510 1.0481 1.0448

DAD 0.9553 1.1593 0.8929 1.1615 1.1432 1.1284 1.1227

DAE 1.0548 1.0955 0.8127 0.9986 0.9739 0.9694 0.9635

EBS 1.3522 1.0411 1.0044 1.0448 1.0224 1.0251 1.0334

EID 0.9759 1.1276 1.082 1.3091 1.4229 1.4206 1.4137

HAP 1.36 1.1891 1.2766 1.1755 1.2109 1.2368 1.2476

HBE 0.9351 0.9674 1.1174 1.0324 1.0283 1.0263 1.0355

KBE 1.0826 1.0578 1.0241 1.0573 1.0393 1.0417 1.0457

LBE 1.0114 1.0232 0.8319 1.0115 0.8654 0.9059 0.9387

NBE 1.2593 0.9673 1.1038 1.6175 1.2226 1.1427 1.0904

QST 0.8738 1.0276 0.9710 1.0609 1.0640 1.0727 1.0774

SED 1.1222 1.3987 0.8772 1.0888 1.1050 1.1106 1.1146

SGD 1.2071 1.0490 0.9190 1.0159 0.9943 0.9986 1.0052

SMN 1.0064 1.0364 1.2014 1.1663 1.1913 1.1926 1.1933

STC 1.1178 0.9759 0.9696 1.0492 1.0614 1.0522 1.0420

TPH 0.9721 1.4084 1.4214 1.3763 1.4081 1.3152 1.2482


Discussion Results evident in Tables 5–7 indicate that the catch-up shows “no-change”, and the

frontier-shift and MPI are going to either extend or decrease. The overall observation

of operating progress from past to future is drawn by the average score of each index,

as shown in Table 8 which gives a comparison of the average scores of catch-up,

frontier-shift, and MPI.

Table 8

Average scores of catch-up, frontier-shift and MPI DMUs Catch-up Frontier-shift MPI

ADC 1.0000 0.9500 0.9500

BDB 1.0049 1.1088 1.0946

BED 1.0000 1.1837 1.1837

BST 1.0000 1.0490 1.0490

DAD 1.0091 1.0993 1.0805

DAE 0.9510 1.0367 0.9812

EBS 0.9769 1.1039 1.0748

EID 1.0116 1.2380 1.2503

HAP 1.0341 1.2049 1.2423

HBE 0.9144 1.1386 1.0203

KBE 1.0001 1.0501 1.0498

LBE 0.9974 0.9436 0.9412

NBE 1.0000 1.2005 1.2005

QST 1.0000 1.0211 1.0211

SED 1.0000 1.1167 1.1167

SGD 0.9804 1.0507 1.0270

SMN 1.0000 1.1411 1.1411

STC 0.9655 1.0827 1.0383

TPH 1.2351 1.0791 1.3071

Average 0.9845 1.0947 1.0931


30


As seen in Table 8, the total average of catch-up is lower than 1 with 0.9845, the

average of frontier-shift and MPI is higher than 1 as 1.0947 and 1.0931, respectively.

The average score of all publishing companies among frontier-shift and MPI has a

smaller difference than that of frontier-shift, which is higher than MPI, at 0.0016. The

efficiency indexes point out progress, regression, and no-change. The score of catch-

up divides into three classifications. The first group has seven companies, including

ADC, BED, BST, NBE, QST, SED, and SMN with an average score of “no change”, shown

as 1. The second group has six companies, including BDB, DAD, EID, HAP, KBE, and TPH,

with average scores from 1.0001 to 1.2351; and, the third group has six companies,

including DAE, EBS, HBE, LBE, SGD, and STC with average scores under 1. The frontier-

shift and MPI only have two classifications. For the frontier-shift, the first group has

seventeen companies, including BDB, BED, BST, DAD, DAE, EBS, EID, HAP, HBE, KBE, NBE,

QST, SED, SGD, SMN, STC, and TPH with an efficiency score above 1; and, two

companies, including ADC and LBE, have an average score under 1. For the MPI, three

companies possess average scores under 1, including ADC, DAE, and LBE; the

remaining companies have an average score above 1.

According to the equation of the Malmquist model, MPI is determined based on

the catch-up and frontier shift. The final empirical result of productivity efficiency in

Table 7 indicates a soft fluctuation of operations for each publishing firm every year.

Four excellent publishing firms reached good progress with their efficiency score in

both previous and future periods. The integration of the GM(1,1) model and Malmquist

model explores the productivity efficiency for all Vietnamese publishing firms from 2015

through 2023. The analysis results exhibit the best and worst firms in each term or whole

term as well. The average MPI in the whole term revealed the three worst firms, as their

average scores are less than one number. Based on the score of production

efficiency, these firms should make plans to improve their productive efficiency in the

future by increasing the values of their output variables and reducing the values of

their input variables immediately.

Implications for practice From a practical point, two models, namely GM(1,1) model in Grey theory and the

Malmquist model in DEA were applied in this research. The Malmquist model in DEA

can measure the technical efficiency of each period based on a consecutive

timeframe; thus, this model calculates the productivity efficiency of the publishing

firms in each year from past to future. The ratio between the sum of outputs and the

sum of inputs evaluates the operational process of firms; moreover, the average score

exhibits an overview of development in the entire time. The notion of technical

efficiency helps these publishing firms make an effective and profitable operational

plan for the future. The described procedure can be used to predict future values in

the publishing industry in other countries, as well as in other industries, such as with

pharma (Wang, Wei, Sun & Li, 2016), logistics (Wang, Day & Nguyen, 2018), or energy

(Qian & Wang, 2020).

Contributions to the literature This paper contributes to the literature as follows. Firstly, previous papers have given

an evaluation of the publishing industry by qualitative methods (Maxim, 2012; Lin,

Chiou & Huang, 2014), while our paper analyses the quantitative data from the

financial statements, which measures the operational process of a firm. Secondly, Lee

and Liang (2018) only forecasted manufacturers’ printing methods to reduce the

manufacturer’s inventory in the educational publishing industry. Our paper has an

31


overall picture of the publishing industry. Thirdly, our paper not only uses the actual

data and forecasts the future data but also measures the performance of publishing

firms by using the Malmquist model approach. We built a set of variables that are

related to the operational process, and the findings evaluate the efficiency level of

these firms and recommend a strategy to develop and improve their efficiency score

in the future.

Conclusion This research integrates GM(1,1) model and the Malmquist model to observe the

publishing firms in Vietnam. The GM(1,1) model estimates the high degree of

accuracy, as indicated by the average MAPE of 3.00176%. The operational progress

of nineteen publishing companies in Vietnam is measured by the technical efficiency

(catch-up), technological change (frontier-shift), and MPI when applying the

Malmquist model in DEA.

An analysis of actual and predicted data by the use of the Malmquist model

describes a picture of the Vietnamese publishing industry. The catch-up efficiencies

of publishing firms indicate that six firms are in no need of substantial change;

however, remaining firms must re-evaluate current operations since these firms can

improve their performance by employing increased values of revenue and

profitability while reducing the values of assets, equity, and liability.

On the other hand, frontier-shift results describe only a slight change for all

publishing firms. Similarly, MPI also denotes that in recent years, publishing companies

have experienced considerable fluctuation. Nine publishing companies, including

BDB, BED, DAD, EID, HBE, NBE, QST, SED, and STC did not reach the desired efficiency

level in previous years, but they are expected to have a good result in the future when

their forecasted efficiency scores will be higher than one number. There are four

publishing companies including EBS, HAP, KBE, and SMN that always attain adequate

performance in every term where they experienced progress.

Although the study estimates the future values and computes the performance of

publishing firms, it still has some limitations. First, the research is focused on only one

country, while the publishing industry formally exists, and is developed in all countries

around the world. The next study can expand the scope of research to point out how

the different national characteristics of publishing industries between Vietnam and

others countries around the world differ. Second, the study measures the technical

efficiency in each term, but it does not determine the relative position for all publishing

firms. Further research should be conducted using more models, such as a super

slacked-based measurement model, a super slacked-based measurement max

model, etc., to afford appropriate ranking.

References 1. Alexander, B. L., Thomas, H. (2007), “A contingency model for the allocation of media

content in publishing companies”, Information & Management, Vol. 44 No. 5, pp. 492–502.

2. Anat, B. N., John, M. G.; Pat, A. (2003), “Business process digitization, strategy, and the

impact of firm age and size: the case of the magazine publishing industry”, Journal of

Business Venturing, Vol. 18 No. 6, pp. 789–814.

3. Bošković, A., Krstić, A. (2020), “The Combined Use of Balanced Scorecard and Data

Envelopment Analysis in the Banking Industry”, Business Systems Research, Vol. 11 No. 1, pp.

4–15.

4. Chen, Z. F., Tzeremes, P., Tzeremes, N. G. (2018), “Convergence in the Chinese airline

industry: A Malmquist productivity analysis”, Journal of Air Transport Management, Vol. 73,

pp. 77–86.

https://www.sciencedirect.com/science/journal/03787206

32


5. Deng, J. L. (1989), “Introduction to Grey System Theory”, The Journal of Grey System, Vol. 1,

pp. 1–24.

6. Edelmann, N., Schoßböck, J. (2020), “Open Access Perceptions, Strategies, and Digital

Literacies: A Case Study of a Scholarly-Led Journal”, Publications, Vol. 8 No. 3, pp. 1–22.

7. Edoardo, G., Antonello, E. S., Laura, V. (2008), “Demand distribution dynamics in creative

industries: The market for books in Italy”, Information Economics and Policy, Vol. 20 No. 3,

pp. 257–268.

8. Ehsan, P., Hadi, S. (2014), “A data envelopment analysis approach for measuring the

efficiency in continuous manufacturing lines: a case study”, International Journal of

Services and Operations Management, Vol. 18 No. 2, pp. 142–158.

9. Gunes, E., Guldal H.T. (2019), “Determination of economic efficiency of agricultural

enterprises in Turkey: a DEA approach”, New Medit, Vol. 18 No. 4, pp. 105–116.

10. Lacy, D. (1979), “The publishing industry: The dynamics and ramifications of change”,

Library acquisitions: Practice & Theory, Vol. 3 No. 1, pp. 39–46.

11. Lee, C. Y., Liang, C. L. (2018), “Manufacturer’s printing forecast, reprinting decision, and

contract design in the educational publishing industry”, Computers &. Industrial

Engineering, Vol. 125, pp. 678–687.

12. Lewis, A. (1982), “The social psychology of taxation”, British Journal of Social Psychology,

Vol. 21 No. 2, pp. 151–158.

13. Lin, C. C., Chiou, W. C., Huang, S. S. (2013), “The Challenges Facing E-book Publishing

Industry in Taiwan”, Procedia Computer Science, Vol. 17, pp. 282–289.

14. Liu, F. H. F., Wang, P. H. (2008), “DEA Malmquist productivity measure: Taiwanese

semiconductor companies”, International Journal of Production Economics, Vol. 112 No. 1,

pp. 367–379.

15. Liu, X. Y., Peng, H. Q., Bai, Y., Zhu, Y. J., Liao, L. L. (2014), “Tourism Flows Prediction based on

an Improved Grey GM(1,1) Model”, Procedia – Social and Behavioral Science, Vol. 138,

pp. 767–775.

16. Maciej, T., Czeslaw, C. (2015), “Using a set of GM(1,1) models to predict values of diagnostic

symptoms”, Mechanical Systems and Signal Processing, Vol. 52–53, pp. 416–425.

17. Matthew. L. F. (2008), “Job hopping, earnings dynamics, and industrial agglomeration in

the software publishing industry”, Journal of Urban Economics, Vol. 64 No. 3, pp. 590–600.

18. Mavi, R. K., Fathi, A., Sean, R. F., Mavi, N. K. (2019), “Eco-innovation in transportation industry:

A double frontier common weights analysis with ideal point method”, Vol. 147, pp. 39–48.

19. Maxim, A. (2012), “The Role of e-books in Reshaping the Publishing Industry”, Procedia -

Social and Behavioral Sciences, Vol. 62, pp. 1046–1050.

20. McCready, K., Molls, E. (2018), “Developing a Business Plan for a Library Publishing

Program”, Publications, Vol. 6 No. 4, pp. 1–14.

21. Meysam, S., Mohammad, R. H., Ebrahim, E., Ali, A. A. T. (2013), “Developing Sustainable

SCM Evaluation Model Using Fuzzy AHP in Publishing Industry”, Procedia Computer Science,

Vol. 17, pp. 340–349.

22. Montoneri, B.; Lin, T. T.; Lee, C. C.; Huang, S. L. (2012), “Application of data envelopment

analysis on the indicators contributing to learning and teaching performance”, Teaching

and Teacher Education, Vol. 28 No. 3, pp. 382–395.

23. Nguyen, H. K. (2020), “Combining DEA and ARIMA Models for Partner Selection in the

Supply Chain of Vietnam’s Construction Industry”, Mathematics, Vol. 8 No. 6, pp. 1–20.

24. Nguyen, T. K. L., Le, H. N., Ngo, V. H., Hoang, B. A. (2020), “CRITIC Method and Grey System

Theory in the Study of Global Electric Cars”, World Electric Vehicle Journal, Vol. 11 No. 4,

pp. 1–15.

25. Qian, W. Y., Wang, J. (2020), “An improved seasonal GM(1,1) model based on the HP filter

for forecasting wind power generation in China”, Energy, Vol. 209, pp. 1–15.

26. Sara, O., Markus, M. B. (2014), “Digitisation of publishing: Exploration based on existing

business models”, Technological Forecasting and Social Change, Vol. 83, pp. 54–65.

27. Tone, K. (2004) Malmquist Productivity Index. In: Cooper W.W., Seiford L.M., Zhu J. (eds)

Handbook on Data Envelopment Analysis. International Series in Operations Research &

Management Science, Vol 71. Springer, Boston, MA. pp. 203–227.

28. Vietstock. Available online: https://finance.vietstock.vn/ (02 September 2020).

33


29. Wang, C. N., Day, J. D., Nguyen, T. K. L. (2018), "Applying EBM Model and Grey Forecasting

to Assess Efficiency of Third-Party Logistics Providers", Journal of Advanced Transportation,

Vol. 2018, Article ID 1212873.

30. Wang, C. N., Tibo, H., Nguyen, V.T., Duong, D.H. (2020), “Effects of the Performance-Based

Research Fund and Other Factors on the Efficiency of New Zealand Universities: A

Malmquist Productivity Approach”, Sustainability, Vol. 12 No. 15, pp. 1–18.

31. Wang, Y. N., Wei, F. F., Sun, C. Q., Li, Q. Z. (2016), "The Research of Improved Grey GM (1, 1)

Model to Predict the Postprandial Glucose in Type 2 Diabetes”, BioMed Research

International, Vol. 2016, Article ID 6837052.

32. Wu, L. F., Liu, S. F., Fang, Z. G., Xu, H. Y. (2015), “Properties of the GM(1,1) with fractional

order accumulation”, Applied Mathematics and Computation, Vol. 252, pp. 287–293.

About the authors Xuan-Huynh Nguyen is a lecturer at Hanoi School of Business and Management,

Vietnam National University. He obtained his Ph.D. in International Business at the

National Kaohsiung University of Science and Technology in Taiwan. His research areas

include business administration, business development, and business strategy. The

author can be contacted at [email protected].

Quoc-Chien Luu is a lecturer of the Department of State Management, Thanh Dong

University in Vietnam. He obtained his Ph.D. in Industrial and engineering management

at the National Kaohsiung University of Science and Technology, Taiwan. His research

areas include business administration, business development. The author can be

contacted at [email protected].

34


Critical Success Factors of New Product

Development: Evidence from Select Cases

Rajeev Dwivedi

Eastern Washington University, United States

Fatma Jaffar Karim

Salma’s Collections Boutique, Dubai, United Arab Emirates

Berislava Starešinić

Privredna banka d.d., Croatia

Abstract Background: The unique, yet complex, new product development (NPD) process

represents one of firms’ most significant operations that impose high weightage onto

its profitability margins and market reputation. Objectives: The object of the research

is to identify critical success factors (CSFs) of a new product development in Dubai

firms. Methods/Approach: The paper uses literature as a basis for identifying critical

success factors for a new product development, which is supported by a semi-

structured interview of senior management-level executives in Dubai. Results: To

pinpoint a set of the most influential CSFs, 12 factors for the NPD process are

highlighted, based on their reoccurrence patterns in the literature and semi-

structured interviews. Impact levels of 12 CSFs on the NPD process are expressed

through a presentation from the highest to the lowest recurrent factor. Conclusions:

Each CSF’s role in driving the NPD process to success has also been justified using

real-time evidence, depicted throughout 4 case studies from different industries.

Keywords: new product development; critical success factors; business success and

qualitative study


JEL classification: M11

Received: 20 Dec 2020


Citation: Dwivedi, R., Karim. F.J., Starešinić, B. (2021). “Critical Success Factors of New

Product Development: Evidence from Select Cases”, Vol. 12 No. 1, pp. 34-44.


Acknowledgments: Authors acknowledge all 15 executives who participated in the

semi-structured interview and shared insight about the NPD process.

35


Introduction A decade ago, the world witnessed the beginning of a rivalry amongst two tech

giants: Apple and Samsung. History records that only one month after Samsung

introduced its first-generation Galaxy S smartphone in June 2010, Apple backfired

with the iPhone 4 (Bouwmeester, 2016). From that point onwards, competition

between the two in terms of features, innovation, and market share has been fierce,

with one company’s launch strongly countering the other’s (Bouwmeester, 2016).

Hallstedt et al. (2020) justify that such competitive behavior represents an

organization’s eagerness to succeed and grow, which can only be possible through

introducing attractive products into the marketplace. To do so profitably entails the

organization to invest in the process of new product development (NPD).

Defined as a collection of related activities that begins with recognizing a market

opportunity, and proceeds with converting it into a new product (Hallstedt et al.,

2020), NPD is considered often as a source of competitive advantage (Owens, 2007).

NPD is a process, whose inputs are idea generation, idea screening and feasibility

studies (Kazimierska et al., 2017), while outputs are the manufacturing (Kazimierska et

al., 2017), commercialization, and pricing of new product.

The significance of the NPD process emerges from a product’s risk of failure

(Owens, 2007). Realizing that all new products carry an inherent possibility of the loss

of the new product fails, which urges organizations to spend maximum efforts to

prevent this outcome (Owens, 2007). Due to the NPD complexity and ambiguity,

organizations are usually drawn to dedicate huge resources to it (Lester, 1998). On

the other hand, the success of a newly introduced product contributes heavily to

the organization’s reputation and direct sales. For example, Apple’s positive market

image is mainly driven by their business shrewdness in introducing a transformed

telecommunications device that fulfils the customer needs. Their success is primarily

driven by their innovative ideas, accurate market search, and timely market launch

(Tariq et al., 2011). Hence, even with changing customer requirements,

technological progressions, and challenges, the yearn for market domination tempts

businesses to take upon the wavering risk of product failure (Hallstedt et al., 2020).

In the course of doing so, businesses ought to manage risks by ensuring the

availability of certain factors that increase NPD success rates, namely, critical

success factors (CSFs). Business Dictionary (2019) defines CSFs as a set of conditions

that has a direct influence on the effectiveness, efficiency, and practicality of the

subject matter (in this case, the process). According to Jeston (2018), some CSFs are

common amongst organizations and play equal roles in driving all types of business

processes to success. However, considering the elevated relevance of the NPD

process to businesses, there emerges a rising need to define precise CSFs concerning

this process specifically, considering its unique activities, developmental phases, and

diversified outputs. As a response to this requirement, this paper is dedicated to

studying the most reoccurring CSFs for the NPD process based on literature. To

validate their role and express their criticality to the process, real examples from a set

of industries will be presented.

Since the beginning of the industrial revolution, the global economy is

continuously expanding and innovation, research &development, and new product

development are buzz words. Nowadays, products are developed daily based on

the information technology and smart technology. For example, Sony alone

develops almost 11000 products every year and launch approximately 1100 in the

market. How many products become successful? This is a big question. Even, out of

1100 products how many products become familiar to consumers? The success of

the new product development process depends on various factors which are critical

36


in the success of the NPD process. Companies that can rapidly innovate need to

understand the CSFs of NPD to gain and maintain market share and remain

competitive.

Caught amid a struggle to effectively drive the complex NPD process and

produce successful products, organizations must pinpoint the CFFs that can at least

help ensure this success. Literature communicates a large set of CSFs for the NPD

process, many of which are classified based on the process’s phases and the

industry it is applied within. After reviewing various industries, it was deduced that

certain CSFs are given a higher value than others, due to their heftier role in

influencing NPD success rates. The independent critical success factor for NPD,s can

be top management support, cross-functional teamwork, NPD process, NPD

strategies, and market research activities (Aziz et. al, 2014).

There have been multiple studies conducted on critical success factors of new

product development but with limited scope and limited setting (e.g.; engineering

equipment development). However, there is little or very limited research is available

in the Middle East context. Therefore, this research becomes an important step in the

direction. The objective of the paper is to explore and understand the concerns of

the area in new product development, which raise the questions: (i) RQ1: What are

the critical success factors for new product development?; (ii) RQ2: Are Critical

success factors reflected in the real-world class organization?

After the introduction, methodology is presented in the second chapter, while

findings and discussion are organized around the most relevant CSFs. Evidence

based on the experience of leading markets and companies is presented in the

fourth chapter. Final chapter provides the concluding remarks.

Methodology The research is based on literature and supported in-depth semi-structured interviews

with product/services/project development managers who played vital roles in the

success of the products/services/project in their respected organizations. We have

identified 15 such individuals based on convenience sampling where we have

personal contacts from different organizations and they were interviewed as a part

of this study. Characteristics of participants are presented in the Appendix 1.

We approached 15 organizations as per convenience sampling in the top 100

organizations in the United Arab Emirates (UAE). We gather the responses from each

15 of the organizations (Appendix). Out of 15 respondents, 9 of them were female

managers level and 6 of them were male. Some of the designations were up to vice

president-level people. The profile of respondents includes; vice president, chief

information officer, general manager, project officer, head of sales and marketing,

and head of the product and promotion. The average age of the respondents was

42 years and the average experience was approximately 16 years ranging from 13

to 21 years. The average experience in the related field of product management or

new product development was approximately 8 years ranging from 6 years to 17.

The respondents are having multi-industry experience that includes; manufacturing,

oil, gas and energy, utilities, aviation, government, healthcare, and finance and

banking. The study could be carried out with different industries and could cover

many respondents but due to time and funding constraints, we limit our study to 15

respondents in UAE only.

With each participant the semi-structured interview was conducted with the goal

to collect the information about their perceptions on the most relevant critical

success factors.

37


Findings Based on semi-structured interviews with the participating managers, the study

findings have been summarized in 12 critical success factors. The critical success

factors demonstrate the complete spectrum of the new product development

process. The list of factors that came out prominently from the depth of semi-

structured interviews are given below:

o Top Management Commitment

o Presence of Clear Goals & Milestone Measurement

o User/Customer Involvement (i.e. Market Research)

o Involvement of Cross-Functional Teams

o Placement of Structured NPD Process

o Talented Team Members with Relevant Experience to NPD Process & Activities

o Establishment of An Entrepreneurial Culture

o Effective Communication Amongst Team Members & With Management

o Alignment of NPD Process Activities with Strategy

o Focusing on Innovation & Out-Of-The-Box Ideas

o Availability of Financial Requirements

o NDP Process Speed

Top Management Commitment As the highest-ranked CSF amongst all studied research works, senior management’s

commitment to the NPD process is represented in defining the organization’s vision,

mission, and strategy (Lester, 1998). These factors communicate a futuristic

perception of the organization in the minds of its people, who in return drive

organization-wide initiatives to pursue these goals. The NPD process is merely but one

of these initiatives, that must be directed towards the business’s target market, the

products it wishes to manufacture, and its business orientation (Lester, 1998).

Furthermore, senior managers must act as process sponsors to approve, allocate

and drive the flow of the process (Holland, Gaston & Gomes, 2000). Besides, the

severity of the NPD process occasionally pauses its team members towards a fork,

where top management’s intervention is required to make the decisions that the

venture team is unauthorized to make (Cengiz et al., 2005).

Presence of Clear Goals & Milestone Measurement Once the strategy of an organization is set, there emerges a need for an NPD

process strategy (Cooper et al., 1995). Questions as “What are the goals of the

process?” and “What kind of products is the organization expecting out of the NPD

team?” must be clearly defined for the team upon establishing it. Lester (1998) clubs

this CSF with the need for healthy project management, and denotes the

importance of setting a tactical plan to follow, starting with feasibility studies to

enable reaching the final product as soon as possible. As for milestone

measurement, he proposes defining strategic constraints, identifying milestones,

defining their requirements and the tasks they incorporate as well as setting a

realistic timeline as to when each of them will be achieved (Lester, 1998).

User/Customer Involvement (i.e. Market Research) One of the direct reasons behind NPD failure is producing the wrong product

(Cengiz et al., 2005). Classified as a task of the organization’s marketing team, the

business must listen to its users’ inputs (Cooper & Kleinschmidt, 1995). Furthermore,

Cengiz et al. (2005) research indicates that paying special attention to the market’s

38


new requirements can enable businesses of the ‘first-mover’ advantage, which

carries high product success rates considering the weak competition.

Involvement of Cross-Functional Teams Cooper et al. (1995) state the need of having team members from across different

departments within the venture team. Holland, Gaston & Gomes (2000) warns of

constituting teams that are either too large or too small for the scope of the process

and its associated activities. Lester (1998) adds that cross-sectional venture teams

seem to operate as an organization on their own simply because they possess a

combination of entrepreneurial traits that complement one another to boost the

process’s performance and results. Cengiz et al. (2005) believes that the diversity

within cross-sectional teams produces innovation.

Placement of Structured NPD Process Determining the NPD process structure, policies and guidelines fall under

management’s responsibilities towards the venture team. Such activity enables

team members of understanding what is expected out of them and how to

approach the NDP process in the first place (Lester, 1998). Cooper et al. (1995) add

that NPD processes ought to highlight quality throughout the deployment. Processes

also must exhibit flexibility in combining steps, performing them in parallel, or skipping

them after careful consideration (Cooper et al., 1995, Holland et al., 2000).

Talented Team Members with Relevant Experience To NPD Process

& Activities In general, all team members must have experience in project management, player

roles, and responsibilities (Florén et. al., 2018). The appointed team leader must be

task-aware and emotionally intelligent in understanding the team members’ work

mannerisms, strengths, and weaknesses. This factor allows him or her to create

synergy amongst all members and influence them to unveil their best collaborative

efforts. (Holland et al., 2000). Most importantly, the leader must not be burdened with

more than one project at a time, to strengthen focus and enable efficiency in one

direction (Cooper et al., 1995).

Establishment of an Entrepreneurial Culture Both Lester (1998) and Cooper et al. (1995) emphasize the need for an innovation-

fostering culture within the host organization, only because worthy-of-investment

ideas mainly generate in the minds of creative, unstressed employees. Both studies

believe that to establish this culture, organizations must allow their employees

enough time to get creative. Cooper et al. (1995) adds that firms must even allocate

budgets to build unofficial prototypes in teams. Moreover, acts of rewarding creative

thinking efforts represent tokens of appreciation and further encouragement to all

(Lester 1998, Cooper et al., 1995).

Effective Communication Amongst Team Members & With

Management Holland et al. (2000) research on 289 projects concluded that healthy

communication amongst team members exhibits a strong correlation with success.

Transparency established as a result of sharing all types of information through

weekly meetings, phone calls, or any other communication method, is crucial to

ensuring that all members stand on the same page (Holland et al., 2000).

39


Communication with management must include project progress, critical issues

faced, possible solutions, and lessons learned (Lester, 1998). Communicating with the

organization’s staff external to the venture team can also help in promoting the new

product, receiving feedback, and evaluating progress from a peripheral point of

view (Lester, 1998).

Alignment of NPD Process Activities with Strategy Before approving the launch of the NPD process, top management much ensure

strategic alignment between the process’s outcomes with the organization’s short

and long-term goals (Florén et. al., 2018). Cooper et al. (1995) emphasizes this

alignment by indicating that the NPD process’s goals must fit into the organization’s,

considering that driving the process to success translates into the partial (or total)

achievement of the organization’s objectives. Hence, top management must

always be able to validate how achieving process success would contribute to the

organization’s ambitions (Cooper et al., 1995).

Focusing on Innovation & Out-Of-The-Box Ideas According to Cengiz et al. (2005), technological evolvements introduce fresh

product potentials to NPD. However, the generation of new-to-the-organization

ideas, as a result of this evolvement proves difficult. Nevertheless, tapping into

technology’s latest developments represents a very important factor in the

successful development of a noble product (Lester, 1998). Not only should ideas

introduce new paradigms, but to sell, product depictions of these ideas must

genuinely add value to customers (Lester, 1998).

Availability of Financial Requirements Budget allocation to any project at the organization represents empowerment.

Hence, if an organization wishes to introduce successful products, it needs to boost

the confidence of its venture team by financially investing in the purpose (Holland et

al., 2000). Senior management must view the financial allocation of resources for an

NPD process as the budget allocated for achieving the organization’s strategic

objectives, which is precisely what the venture team aims to accomplish (Cooper et

al., 1995), considering synergy between the organization’s and NPD process’s

strategy.

NDP Process Speed The faster an NDP process is, the quicker its organization would be able to introduce

new products into the market and win a competitive advantage (Cengiz et al,,

2005). Moreover, with rapid technology transformations dominating the current

marketplace, speed has become an economic requirement of the NDP process.

Profit figures prove that delaying a product introduction can affect the sales of the

product up to 35%, which justifies why most managers are more willing to increase

resources by 50% than delay a new product launch (Cengiz et al., 2005).

Discussion The Table 1 exhibits the reoccurrence count of the above factors based supported

by author credentials, organized from most to least reoccurring.

40


Table 1

Distribution of CFSs across relevant research papers CSF Le

ster (1

998

)

Ce

ng

iz et a

l.

(2005)

Flo

rén

et. a

l.

(2018)

Co

op

er e

t al.

(1995)

Fe

rna

nd

es e

t al.

(2017)

Su

n e

t al. (2

004)

Su

wa

nn

ap

orn

et

al. (2

010)

Ho

llan

d e

t al.

(2000)

Co

nn

ell e

t. al.

(2001)

Tota

l

Top

Management

Commitment

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 9

Presence of

Clear Goals &

Milestone

Measurement

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 8

User/Customer

Involvement

(i.e. Market

Research)

✓ ✓ ✓ ✓ ✓ ✓ ✓ 7

Involvement of

Cross-

Functional

Teams

✓ ✓ ✓ ✓ ✓ ✓ ✓ 7

Placement of

Structured NPD

Process

✓ ✓ ✓ ✓ ✓ ✓ 6

Talented Team

Members with

Relevant

Experience to

NPD Process &

Activities

✓ ✓ ✓ ✓ ✓ ✓ 6

Establishment of

an

Entrepreneurial

Culture

✓ ✓ ✓ ✓ ✓ 5

Effective

Communication

Amongst Team

Members &

Management

✓ ✓ ✓ ✓ ✓ 5

Alignment of

NPD Process

Activities with

Strategy

✓ ✓ ✓ ✓ 4

Focusing on

Innovation &

Out-Of-The-Box

Ideas

✓ ✓ ✓ ✓ 3

Availability of

Financial

Requirements

✓ ✓ ✓ 3

NDP Process

Speed

✓ ✓ 2

Note: Critical Success Factors have been identified in the literature

Having listed a set of CSFs for the NPD process based on reoccurrence in literature

is not enough to prove their role in driving the process to success. The latter can only

be justified through real examples emerging from various industries to validate the

41


mentioned CSF's real potentials in ensuring the universal win. Below are a series of

case studies captured from real-time situations that exhibit the contribution of each

factor in the undertaken NPD process. Generally, cases exhibit more than one factor

simultaneously, depending on the industry and local conditions.

In their article on Apple’s NPD process, Tariq et al. (2011) reveals the company’s

secret behind the phenomenal success of its iPod and iPhone products. It all begins

with an innovative culture and exploration of an innovative technology that the

market can readily absorb. User involvement and external research would then

determine if the proposed product (still an idea) will be accepted by potential

buyers, considering the perception that it should fit into customers’ current use

patterns. In other words, the device must smoothly renovate the way people

operate their daily affairs without reinforcing too much change over a short period.

Bringing the devices to reality with a flexible workflow, the organization was able to

exploit the ‘first-mover advantage’ of introducing noble, unparalleled products

(Tariq et al., 2011).

AT&T is an American telecommunications company. According to Connell et. al.

(2001), its College Market sector announced a strategic partnership with Student

Advantage to launch a calling student card clubbed with a loyalty card. The

product’s strategy was closely aligned with the organization’s objective of increasing

phone call usage by directly billed cardholders, as well as expanding the

organization’s student market share (Connell et. al., 2001). The product enjoyed

huge success mainly because it was developed by an effective leader and diligent,

cross-sectional team members, who had the right expertise, knowledge, and know-

how of implementing the project (Connell et. al., 2001). The result was a huge market

jump for the organization in terms of sales margins and reputation (Connell et. al.,

2001).

The Hong Kong toy industry was examined by Sun et al. (2004) due to potential

takeover threats imposed by its neighbouring Chinese competitors. CSFs were

studied relative to the NPD process phases (4 phases) and identified key success

factors were classified into 4 categories, based on implementation and relevance

(i.e. biblical model). It was realized that top management commitment and

availability of financial resources represented two of the highly implemented (but

not important) CSFs in phase I of the new toy development process (Sun et al., 2004).

Top management commitment continues as one of the highly implemented CSFs in

the product’s development phase II as well. Factors as the speed of the NPD

process, timely launch, and on-time delivery appear as high importance – high

implementation in the final stage of the NPD process, emphasizing how they directly

influence the sales of the product once launched (Sun et al., 2004).

After studying a set of management groups working in medium to large Thai food

companies, Suwannaporn et al. (2010) concluded that success rates of NDP

processes in the food industry mainly rely on user involvement, effective

communication with parties internal and external to the organization as well as the

necessity of having a clear NPD process strategy and tactical planning. Although

these factors do not match with what respondents perceive are food industry NPD’s

CSFs, they have nevertheless been deduced as a result of qualitative research

carried out by the scholars with the return of 114 questionnaires from the targeted

industry (Suwannaporn et al., 2010).

Innovation and entrepreneurship are in Sony’s culture. Sony corporation thrives on

innovation and entrepreneurship over 7 decades. Sony innovates almost 10,000

products every year and launches successfully around 10% of it every year. This

cannot be possible without the commitment of top leadership and support from

42


senior management (Fortune et al., 2006). Sony retains its market position and the

main reason for this is the continuous new product development.

Concluding Remarks Recognizing the direct and indirect presence of all 12 denoted CSFs in universal NPD

projects from diverse industries justifies the validity of each of them, highlighting their

importance to the process in general, and its success in specific. It is important to

note that different industries require a varied mix of CSFs, depending on region,

nature of the product, top management practices, and culture. Moreover, it can

also be concluded that certain CSFs gain rising importance throughout a limited

phase of the NPD process. In reality, the scopes of CSFs seem to overlap one

another, where the availability of one indirectly leads to the presence of another

(i.e. cross-sectional teams often present different experiences in terms of project

management, which enables constructive progress towards end products).

Realistically, ensuring the application of crucial-to-phase CSFs throughout the

tedious NPD process advances the project’s rate of success substantially.

References 1. Aziz, F.A., Jaffar, N.N., Surya S. (2014), “Critical Success Factors of New Product

Development and Impact on Malaysian Automobile Industry”, Advanced Materials

Research, Vol. 903, pp. 431-437.

2. Bouwmeester, J. (2016) iPhone vs Galaxy S: A timeline comparison of innovation.

Available from: https://techaeris.com/2016/09/03/iphone-vs-galaxy-s-timeline-

comparison-innovation/ (June 22, 2019).

3. Business Dictionary (2019) Critical Success Factors (CSF). Available from:

http://www.businessdictionary.com/definition/critical-success-factors-CSF.html

/(Accessed: June 23, 2019).

4. Cengiz, E., Ayyildiz, H., & Kirkbir, F. (2005), “Critical Success Factors in New Product

Development”, Atatürk Üniversitesi Sosyal Bilimler Enstitüsü Dergisi, Vol. 6 No. 2, pp. 405-

420, 2005.

5. Connell, J., Edgar, G. C., Olex, B., Scholl, R., Shulman, T., Tietjen, R. (2001), “Troubling

successes and good failures”, Engineering Management Journal, Vol. 12 No. 4, pp. 35-39.

6. Cooper R. G., Kleinschmidt E. J. (1995), “Benchmarking Firm’s Critical Success Factors in

New Product Development”, Journal of Product Innovation Management, Vol. 12 No. 5,

pp. 374-391.

7. Hallstedt, S.I., Isaksson, O. and Öhrwall Rönnbäck, A., (2020), “The Need for New Product

Development Capabilities from Digitalization, Sustainability, and Servitization Trends”,

Sustainability, Vol. 12 No. 3, p.10222.

8. Fernandes, R. Q. K., Cavalcanti, V., de Andrade, A. M. (2017), “Design management in

the toy industry: Case studies on design insertion for the development process in Brazilian

toy companies”, Strategic Design Research Journal, Vol. 10 No. 3, pp. 230-240.

9. Florén, H., Frishammar, J., Parida, V. & Wincent, J. (2018), “Critical success factors in early

new product development: a review and conceptual model”, International Enterprise

Management Journal, Vol. 14, pp. 411-427.

10. Fortune, J., White, D. (2006), “Framing of project critical success factors by systems

model”, International Journal of Project Management, Vol. 24, pp. 53-65,

11. Holland, S., Gaston, K., Gomes, J. (2000), “Critical success factors for cross-functional

teamwork in new product development”, International Journal of Management Reviews,

Vol. 2, No. 3, pp. 231-259.

12. Jeston, J. (2018), Business Process Management: Practical Guidelines to Successful

Implementations, New York: Routledge.

https://techaeris.com/2016/09/03/iphone-vs-galaxy-s-timeline-comparison-innovation/

https://techaeris.com/2016/09/03/iphone-vs-galaxy-s-timeline-comparison-innovation/

http://www.businessdictionary.com/definition/critical-success-factors-CSF.html

43


13. Kazimierska, M., Grebosz-Krawczyk, M. (2017), “New Product Development Process – An

Example of Industrial Sector”, Management Systems in Production Engineering, Vol. 25

No. 4, pp. 246-250.

14. Lester, D. H. (1998), “Critical Success Factors for New Product Development”, Research-

Technology Management, Vol. 41 No. 1, pp. 36-43.

15. Owens, J. D. (2007), “Why do some UK SMEs still find the implementation of a new product

development process problematical?”, Management Decision, Vol. 45 No. 2, pp. 235-

251.

16. Sun, H., Wing, W. C. (2004), “Critical success factors for new product development in the

Hong Kong toy industry”, Technovation, Vol. 25 No. 3, pp. 293-303.

17. Suwannaporn, P. Speece, M. W. (2010), “Assessing new product development success

factors in the Thai food industry”, British Food Journal, Vol. 112, No. 4, pp. 364-386.

18. Tariq, M.; Ishrat, R.; and Khan, H. A case study of Apple’s success with iconic iPod and

iPhone. Interdisciplinary Journal of Contemporary Research in Business, 3, 1 (2011), 158–

168.

About the authors

Rajeev Dwivedi is a visiting Associate Professor of Business Systems and Analytics at

College of Business, Eastern Washington University, WA, USA Before joining EWU, he

served as an Affiliate Professor of Information Systems at Howe School of Technology

Management, Stevens Institute of Technology, USA. He was associated with New

Jersey City University and Farleigh Dickinson University in New York Metropolitan area

as Adjunct Professor. He received his Doctorate from the Indian Institute of

Technology (I.I.T.) Delhi on e-Business Strategy. He has over sixteen years of teaching,

research, and consultancy experience in the field of IT Strategy and E-Business

Strategy, and involved actively in various consultancy assignments of

government/industry in the USA, Kenya, Middle East, as well in India. The author can

be contacted at [email protected]

Fatma Jaffar Karim is an entrepreneur and co-founder of Salma’s Collections, an

online-based boutique in the UAE. She has earned her undergraduate degree from

the American University of Sharjah and Master’s degree in Innovation & Change

Management from Hamdan Bin Mohamed Smart University, Dubai. She was on the

dean’s list in both universities. Karim is a Ph.D. aspirant and very involved in research,

especially related to her venture. She believes her boutique’s success and prosperity

solely rely on hands-on experience and in-depth research of fashion trends, content

marketing, and customer behaviour. The author can be contacted at

[email protected].

Berislava Starešinić, Ph.D. is Director for Affluent Segment Development, Function of

Affluent Clients, Retail group, Privredna Banka Zagreb. Her research interests include

the banking sector, organizational behaviour, banking reputation, and direct and

digital channels. The author can be contacted at [email protected].

44


Appendix 1 Table A1

Respondents Profiles

No. Organization Gender Experience NPD

Experience

Industry/Product

1 Etisalat Female 16 8 Telecom

2 Du Male 15 7 Telecom

3 Julphar Female 16 8 Healthcare/Pharmaceutical

4 Nakeel

Property

Male 14 7 Real State

5 Dubal Female 14 6 Aluminium (manufacturing)

6 Souq.com Female 15 7 E-Commerce

7 Liwa

chemicals

Male 16 6 Chemicals and Petroleum

8 Jumairah

Group

Male 19 8 Hospitality

9 Al Dahra

Agriculture

company

Male 13 6.9 Farming, Dairy, and

Agriculture

10 Noor takaful Male 20 13 Insurance

11 EMIRATES NBD

Bank

Female 21 17 Banking and Finance

12 EMIRATES Female 16 8 Aviation

13 Dubai Govt Female 18 8 Government

14 Adnoc Female 13 6 Oil Gas and Energy

15 Dewa Female 15 6 Utilities

Source: Authors’ work

45


Differences in Slovenian NUTS 3 Regions and

Functional Regions by Gender

Samo Drobne

Faculty of Civil and Geodetic Engineering, University of Ljubljana, Slovenia

Abstract

Background: Regions at the level of NUTS 3, which is a system used in the EU for various

analyses and statistical reports, can be defined as functional regions in terms of labour

markets, education areas, and supply markets. Objectives: This study analyses the

functional regions of Slovenia, differentiated by gender, and their correspondence

with the statistical regions at the level of NUTS 3. Methods/Approach: Functional

regions are analysed as labour market areas, which are modelled according to the

CURDS method, and evaluated using the fuzzy set approach. Results: The analysis of

functional regions resulted in ten regions for male commuters and fourteen regions for

female commuters. Only four of the twelve functional regions for commuters relate to

the corresponding statistical regions. Functional region Ljubljana is much larger than

the corresponding statistical region, mainly at the expense of neighbouring regions. In

recent decades, two new functional regions have been created which are becoming

candidates for inclusion in the system of NUTS 3 regions. Conclusions: A detailed

analysis showed that functional region Velenje is becoming an important local labour

market and should be included in the system of NUTS 3 regions of Slovenia, while the

Central Sava Statistical Region should be removed from it.

Keywords: NUTS 3 regions; functional regions; gender; Slovenia

JEL classification: C02, C44, J16, J20, R23




Citation: Drobne, S. (2021), “Differences in Slovenian NUTS 3 Regions and Functional

Regions by Gender”, Business Systems Research, Vol. 12, No. 1, pp. 45-59.


Acknowledgments: The author acknowledges the financial support from the

Slovenian Research Agency (research core funding P2-0406 Earth observation and

geoinformatics and research projects J6-9396 Development of Social Infrastructure

and Services for Community Based Long-Term Care and J5-1784 Creating Social Value

with Age-Friendly Housing Stock Management in Lifetime Neighbourhoods).

46


Introduction The Nomenclature of Territorial Units for Statistics (NUTS) is a system for the collection,

compilation and dissemination of European statistics at different territorial levels of the

European Union (EU). The nomenclature NUTS divides the economic territory of the

Member States hierarchically into territorial units. It divides each Member State into

NUTS territorial units at level 1, each of which is subdivided into NUTS territorial units at

level 2, which are in turn subdivided into NUTS territorial units at level 3 (EC, 2019). The

current NUTS 2016 classification is valid from 1 January 2018 to 31 December 2020 and

lists 104 regions at NUTS 1, 281 regions at NUTS 2, and 1348 regions at NUTS 3 (Eurostat,

2020a). The NUTS 2021 classification, which will be valid for data transmissions to

Eurostat from 1 January 2021, lists 104 regions at NUTS 1, 283 regions at NUTS 2, and

1345 regions at NUTS 3 (Eurostat, 2020b). According to the NUTS regulation (EC, 2019),

changes in the scope and number of NUTS regions are only possible every three years.

The changes are usually based on changes in the territorial structure of a Member

State.

For statistical purposes at local level, Eurostat maintains a system of Local

Administrative Units (LAU) that is compatible with NUTS. LAU units, like municipalities

and communes, are the building blocks of NUTS regions (Eurostat, 2020c).

According to the criterion for the number and size of these regions, each region at

the NUTS 3 level must have between 150,000 and 800,000 inhabitants, based on the

average population (EC, 2019). In practice, this means that Slovenia can have a

maximum of thirteen statistical regions at this level.

Slovenia includes twelve regions at the NUTS 3 level that are mainly used for

statistical reporting, which is why they are also called “statistical regions”. Twelve

Slovenian statistical regions are as follows (SORS, 2020a; Wikipedia, 2020): SI031 Mura

Statistical Region, SI032 Drava Statistical Region, SI033 Carinthia Statistical Region,

SI034 Savinja Statistical Region, SI035 Central Sava Statistical Region, SI036 Lower Sava

Statistical Region, SI037 Southeast Slovenia Statistical Region, SI038 Littoral–Inner

Carniola Statistical Region, SI041 Central Slovenia Statistical Region, SI042 Upper

Carniola Statistical Region, SI043 Gorizia Statistical Region, SI044 Coastal–Karst

Statistical Region. Figure 1 shows twelve regions at the NUTS 3 level and 212

municipalities at the LAU level in Slovenia in 2020.

Regions at the NUTS 3 level can also be defined as functionally connected areas,

i.e. functional regions, in terms of labour markets, education areas, and supply markets

(Drobne, 2016). In this study, we analysed functional regions, differentiated by gender,

at the level of twelve statistical regions (NUTS 3) in Slovenia. We tested the hypothesis

that there are differences in gender-specific functional regions of Slovenia. We

discussed also the differences between NUTS 3 regions and the corresponding

functional regions (FRs) and between men and women within FRs.

47


Figure 1

Twelve regions at the NUTS 3 level and 212 municipalities at the LAU level in Slovenia

in 2020

Source: Author’s work, SMARS (2020)

Literature review The first version of the classification of Slovenian statistical regions was prepared in the

mid-1970s. It was based on a detailed gravity analysis of labour markets, educational

areas and supply markets in twelve regional and sub-regional centres (Vrišer, 1974,

1978; Rebec, 1983, 1984; Vrišer & Rebernik, 1993). From this time onwards, statistical

regions were used for regional planning and cooperation in various fields. This is why

Slovenian regions at the NUTS 3 level have been very consistently stable (Drobne,

2016). However, labour and supply markets are changing all the time, especially

during economic and/or financial crises, like the one that arose in 2008. For that

reason, OECD (2002) and Eurostat (Coombes et al., 2012) suggested the more

frequent analysis of labour market areas. However, labour market areas are good

approximations of so-called functional regions (OECD, 2002) that could be the basis

for regions at the NUTS 3 level in the EU.

There are many definitions of functional regions (FRs). One of the first definitions

describes a FR as an area surrounding a strong economic centre that attracts residents

from the near and far hinterland (Berry & Garrison, 1958), whereas the centre of FR is

understood as a location as defined in Christaller's theory of central places (Christaller,

1933), the size of which depends on the supply of goods and services to the residents.

Today, a FR is the most often understood as a territorial area characterised by the high

frequency of intra-regional economic interaction, such as intra-regional trade in

goods and services, labour commuting, and household shopping. Even though it is

characterised by its agglomeration of activities and by its intra-regional transport

infrastructure, according to many researchers (e.g., Vanhove & Klaassen, 1987;

Karlsson & Olsson, 2006; Cörvers et al., 2009; Coombes et al., 2012), the fundamental

characteristic of FR is the integrated labour market, where commuting and job search

and demand are much more intense than their counterparts outside of the region.

This is why FRs are most frequently analysed by flows of commuters. From this point of

view, FRs are understood also as areas delimitated by generalized patterns of

commuting flows (Drobne et al., 2020).

48


Commuting to work is not just a daily or weekly movement in space, but also a

personal experience that men and women experience differently (Prashker et al.,

2008). Commuting is influenced by various factors that have different effects on each

gender. Studies by White (1977) and Fanning-Madden (1981) have shown that gender

and occupation form the basis for differences in wages, working hours, places of work,

and household duties, which also leads to differences in the distance and time spent

commuting among genders. On the other hand, many authors (e.g. White, 1977;

Green et al., 1986; Tkocz & Kristensen, 1994; Sang, 2008; Prashker et al., 2008; Roberts

et al., 2011; Nafilyan, 2019) note that the commuting distance and time to work

increases for both genders. However, men have always commuted on average

longer distances to work and spent more time commuting, which the researchers cite

as a consequence of cultural standards regarding restrictions on female domestic

work and childcare. Roberts et al. (2011) explicitly pointed out that childcare is one of

the strongest factors that significantly influence the choice of time and distance to

commute among women.

A gender-specific comparison of the distance and time spent commuting by age

group showed that women in almost all age groups were less inclined to commute for

longer periods, except during the first years of employment (Nafilyan, 2019). The

willingness to commute longer to work increases and is very similar for both sexes up

to the age of 25. For women, it remains constant until the age of 35, when it begins to

decline. For men, on the other hand, the willingness to commute to work for longer

periods increases until the age of 35 and remains at a similar level until the age of 45,

after which it begins to decrease.

The choice of place of residence is influenced by many different factors, such as

socio-economic characteristics, life cycle, place of work, and other important factors

such as school, family, friends, shopping centres, property value, and characteristics

of the working and living environment. Some choose to live by their place of work,

and some choose to work by their place of residence, or both at the same time

(Prashker et al., 2008). It is also known that, as incomes and specialisation increase, so

does the average distance and travel time to work (ibid.). However, Roberts et al.

(2011) showed that daily commuting to work affects mental health and leads to stress

and greater social isolation. With benefits such as higher income, better living

environment, and more favourable working conditions, daily commuting has an

impact on the mental health of women, while it has almost no effect on men (ibid.).

Despite lower wages, women tend to choose to replace more remote jobs with

nearby ones (Sang, 2008; Nafilyan, 2019).

However, analysis of the spatial distribution of occupations by gender in the urban

environment revealed a higher proportion of male jobs in more developed urban

areas, while women are employed equally throughout the urban area (Blumen, 1994).

Methodology The basic data source for the study of functional regions (FRs) included the average

annual flows of commuters between 212 municipalities in Slovenia over the three years

of 2016-2018, obtained from the SI - Stat Data Portal and taken from Statistical Register

of Employment (SORS, 2020b). These resources provided the place of residence and

work and the gender of the worker. The flows of commuters were considered in a

quadratic matrix of interactions of the dimension 𝑛 × 𝑛, 𝐹,

𝐹 = [𝑓𝑖𝑗], 𝑛 × 𝑛 matrix, 𝑛 = 212, (1)

49


where 𝑓𝑖𝑗 ≥ 0 is the value in the 𝑖-th row and the 𝑗-th column, i.e. the flow from the

municipality of origin 𝑖 to the municipality of destination 𝑗. Spatial data on municipalities and statistical regions at the NUTS 3 level in Slovenia

were obtained from the "Free Access Database" of the Surveying and Mapping

Authority of the Republic of Slovenia (SMARS, 2020).

We modelled FRs with the use of the CURDS method, which comes from the Centre

for Urban and Regional Development Studies, from Newcastle University, UK. The

method was first introduced in the mid-1980s by Coombes et al. (1986) and was later

improved several times. We used the third version of the method, which was presented

by Coombes and Bond (2008). The method is also called the EURO method because

it has been tested by EUROSTAT and several research groups in Europe (Coombes et

al., 2012).

FR was modelled according to an iterative procedure of the third version of the

method CURDS (Coombes & Bond, 2008) in the R software tool with the library

LabourMarketAreas 3.0 (Franconi et al., 2016a, 2016b, 2017). When modelling FR, we

follow the principle of maximizing internal flows (flows within FR) and minimizing

external flows (flows across the boundaries of FR). For commuter flows, we monitor

these two principles with FR self-sufficiency, which is treated as supply-side self-

containment (𝑆𝑆𝑆𝐶) and demand-side self-containment. 𝐷𝑆𝑆𝐶). 𝑓ℎ𝑘 is the flow of

commuters from the group of municipalities ℎ to the group of municipalities 𝑘 or 𝑓ℎ𝑘 is

the number of workers living in origin ℎ and working in destination 𝑘.

𝑆𝑆𝑆𝐶 =𝑅𝑊𝑖

𝑅𝑖 is supply-side self-containment (2)

and

𝐷𝑆𝑆𝐶 =𝑅𝑊𝑖

𝑊𝑖 is demand-side self-containment, (3)

where 𝑅𝑖 = ∑ 𝑓𝑖𝑘𝑘 is the number of workers living in 𝑖, 𝑊𝑖 = ∑ 𝑓ℎ𝑖ℎ is the number of workers

working in 𝑖, and 𝑅𝑊𝑖 = 𝑓𝑖𝑖 is the number of workers living and working in 𝑖. Supply-side self-sufficiency (𝑆𝑆𝑆𝐶) indicates the extent of employment opportunities

for the local population. The high level of 𝑆𝑆𝑆𝐶 indicates a relatively closed FR (a large

part of the local population finds employment in FR). Conversely, a low 𝑆𝑆𝑆𝐶 rate

indicates a relatively open FR (a large part of the local population works in other FRs).

Demand-side self-sufficiency (𝐷𝑆𝑆𝐶) provides a range of housing options for FR

employees. The high 𝐷𝑆𝑆𝐶 rate, therefore, means that a large proportion of FR

employees have found accommodations there, and at the same time, this may also

indicate a lack of jobs in FR (Drobne, 2016). Van der Laan and Schalke (2001) therefore

suggest that, when assessing FRs, 𝑆𝑆𝑆𝐶 should always be confronted with 𝐷𝑆𝑆𝐶. In

addition to self-containment, an important criterion in the evaluation or modelling of

FRs according to the CURDS method is also the number of workers or employed active

population (𝐸𝐴𝑃). Before performing an iterative procedure of the CURDS method, we

must therefore define the four parameters with which we model the FRs; these are the

minimum number of 𝐸𝐴𝑃 in FR (𝑚𝑖𝑛𝐸𝐴𝑃), the target number of 𝐸𝐴𝑃 in FR (𝑡𝑎𝑟𝐸𝐴𝑃), the

minimum self-sufficiency of FR (𝑚𝑖𝑛𝑆𝐶) and the target self-sufficiency of FR (𝑡𝑎𝑟𝑆𝐶); we

consider self-sufficiency to be the smaller of the two self-sufficiencies considered:

𝑆𝐶 = 𝑚𝑖𝑛(𝑆𝑆𝑆𝐶, 𝐷𝑆𝑆𝐶). (4)

The CURDS algorithm groups the basic spatial units (BSUs), in our case municipalities,

step by step into the FRs. The algorithm treats each municipality as a FR. The algorithm

50


checks the validity of the FR in the aggregation process using the defined parameters

(𝑚𝑖𝑛𝐸𝐴𝑃, 𝑡𝑎𝑟𝐸𝐴𝑃, 𝑚𝑖𝑛𝑆𝐶, and 𝑡𝑎𝑟𝑆𝐶) that define the criteria function 𝑓𝑣:

𝑓𝑣(𝐸𝐴𝑃, 𝑆𝐶) = (1 − (1 −𝑚𝑖𝑛𝑆𝐶

𝑡𝑎𝑟𝑆𝐶)𝑚𝑎𝑥 (

𝑡𝑎𝑟𝐸𝐴𝑃−𝐸𝐴𝑃

𝑡𝑎𝑟𝐸𝐴𝑃−𝑚𝑖𝑛𝐸𝐴𝑃, 0))

𝑚𝑖𝑛(𝑆𝐶,𝑡𝑎𝑟𝑆𝐶)

𝑡𝑎𝑟𝑆𝐶. (5)

A group of municipalities becomes an FR if the condition is met:

𝑓𝑣(𝑊𝑃, 𝑆𝐶) ≥𝑚𝑖𝑛𝑆𝐶

𝑡𝑎𝑟𝑆𝐶. (6)

The validity condition of FR is checked after each merge step. Namely, the algorithm

merges step-by-step municipalities (groups of municipalities), between which the

strongest link, 𝐿ℎ𝑘, is defined by labour mobility flows:

𝐿ℎ𝑘 =𝑓ℎ𝑘2

𝑅ℎ𝑊𝑘+

𝑓𝑘ℎ2

𝑅𝑘𝑊ℎ, (7)

where 𝑓ℎ𝑘 is the number of active population living in a municipality or group of

municipalities ℎ and working in a municipality or group of municipalities 𝑘; 𝑓𝑘ℎ is the

number of active population living in a municipality or group of municipalities 𝑘 and

working in a municipality or group of municipalities ℎ; 𝑅ℎ is the number employed

active population in a municipality or group of municipalities ℎ; and 𝑊𝑘 is the number

of jobs in a municipality or group of municipalities, 𝑘. The algorithm of the third version

of the method CURDS, implemented in the library LabourMarketAreas 3.0 for use in the

software tool R, is described in detail in Franconi et al. (2016a).

A specific feature of the CURDS method is the possibility of disaggregating FR into

BSUs, in our case municipalities, if FR does not meet the validity condition (6), and we

include them on the reserve list with the possibility of reuse in the merging process. The

final result of modelling FRs using the CURDS method is determined by the parameters

𝑚𝑖𝑛𝐸𝐴𝑃, 𝑡𝑎𝑟𝐸𝐴𝑃, 𝑚𝑖𝑛𝑆𝐶, and 𝑡𝑎𝑟𝑆𝐶, but they depend mainly on the size (population)

of the area under consideration and the size of the labour market in that area.

Recommendations for the above parameters can be found in the literature (e.g.

Coombes and Bond, 2008; Franconi et al., 2016a, 2016b), but they generally apply to

modelling FR at the micro and mezzo levels. However, for all levels of treatment, the

target of self-sufficiency should be greater than 0.65 (𝑡𝑎𝑟𝑆𝐶 ≥ 0.65), and the minimum

of self-sufficiency should be greater than 0.60 (𝑚𝑖𝑛𝑆𝐶 ≥ 0.60), while the target (𝑡𝑎𝑟𝐸𝐴𝑃)

and the minimum number of the employed active population (𝑚𝑖𝑛𝐸𝐴𝑃) in FR depend

on the characteristics of BSUs, in our case municipalities, the flows of labour mobility,

and other characteristics of the area under consideration, especially population

density. The parameter 𝑚𝑖𝑛𝐸𝐴𝑃 has a significant impact on the size of the modelled

FRs. Coombes and Bond (2008) recommend at least a generalised knowledge of FR

at a selected level of an analysed area.

By changing the parameters 𝑚𝑖𝑛𝐸𝐴𝑃, 𝑡𝑎𝑟𝐸𝐴𝑃, 𝑚𝑖𝑛𝑆𝐶, and 𝑡𝑎𝑟𝑆𝐶, we initially

modeled twelve FRs for all commuters in Slovenia. In the separate analyses of labour

mobility flows by gender, we used the same parameters 𝑚𝑖𝑛𝑆𝐶 and 𝑡𝑎𝑟𝑆𝐶 as in the

analysis for all commuters, while 𝑚𝑖𝑛𝐸𝐴𝑃 and 𝑡𝑎𝑟𝐸𝐴𝑃 were calculated concerning the

number of women and men in the analysed population and rounded for hundreds,

as proposed by Arnuš (2020).

Three final sets of functional regions, i.e. twelve FRs for commuters together, ten FRs

for male commuters, and fourteen FRs for women, were evaluated using the Fuzzy Set

Theory (FST) approach proposed by Feng (2009) and Watts (2009, 2013) and improved

51


by Drobne (2020) and Drobne et al. (2020). The membership function values of each

municipality were calculated as the geometric mean of the membership function

values of municipality 𝑖 concerning fuzzy residential functional region 𝑚, 𝑀′𝑖𝑚 =∑ 𝑓𝑗𝑖 𝑓∙𝑖⁄𝑗∈(𝑔)𝑚 , and to fuzzy local employment functional region 𝑚, 𝑀′′𝑖𝑚 = ∑ 𝑓𝑖𝑗 𝑓𝑖∙⁄𝑗∈(𝑔)𝑚 :

𝑀𝑖𝑚 = √𝑀′𝑖𝑚 ∙ 𝑀′′𝑖𝑚 . (8)

To evaluate the entire sets of FRs, geometric mean membership values were

calculated for each FR and also for the whole system of FRs. The calculation of

membership function values was performed in Mathematica 11.3 with the programme

code developed by Drobne and Lakner (2016) and Drobne (2020).

In addition to the modelled FRs, we also calculated some interesting statistics. Of

particular interest is the weighted average distance between home and work

municipalities obtained by multiplying the commuter flow by the Euclidean distance

between the municipal centres.

Results Between 2016 and 2018, an average of 829,626 people was employed in Slovenia, of

which 448,976 (54.1%) were men and 380,366 (45.9%) women. Of those in

employment, slightly less than half (398,258; 48%) worked in the municipality of

residence and the rest (431,368; 52%) in another municipality. Comparison by gender

shows a slightly higher proportion of women who found a job in their home

municipality (48.3%) than men (47.8%); see Table 1.

Of the 44,944 interactions between 212 municipalities, less than a third (13,915; 31%)

were non-empty interactions in matrix 𝐹. A comparison between the sexes shows that

employed men commuted to work in more different municipalities (11,582; 25.8% of

interactions) than employed women (9,363; 20.8%). This statement is also confirmed

by the fact that the non-empty interactions include interactions 𝑖𝑖, (i.e. interactions of

the municipality with itself) and that a higher proportion of women than men worked

in their home municipality.

The capital of Slovenia Ljubljana (code 61 in Figures 2) is the country's most

important centre of employment, providing more than a quarter (220,779; 26.6%) of

jobs. Of all jobs in the country by gender, relatively more women (27.9%) than men

(25.5%) found employment in the municipality of Ljubljana. Also, out of a total of 99,681

persons in employment, more women (12.9%) than men (11.3%) were employed in

their home municipality of Ljubljana compared to Slovenia. The most intensive

interaction in daily labour mobility is between Ljubljana and the neighbouring

Domžale (code 23 in Figures 2), where just over 7,000 workers from the municipality of

Domžale come to work in the municipality of Ljubljana every day. Again, this region

includes relatively more women (1% of all employed women) than men (0.8% of all

employed men in Slovenia).

The above data shows that, regarding the number of jobs by gender in Slovenia,

relatively more women than men stay in their home municipality or commute to work

mainly in neighbouring municipalities. This statement is also confirmed by the results of

the analysis of the weighted average distance to work, according to which

commuters travelled an average of 16 km to work every day from 2016 through 2018,

16.8 km for men and 14.9 km for women (see Table 2).

Analysis of the functional regions performed by the CURDS method and parameters

listed in Table 2 revealed twelve FRs for all commuters, ten for men, and fourteen for

women.

52


Table 1

Employed active population and labour commuting interactions for 2016-2018 in

Slovenia Together Men Women

Employed active population (𝑬𝑨𝑷) 829,626

(100%)

448,976

(100%)

380,366

(100%)

Work in residential municipality 398,258

(48.0%)

214,584

(47.8%)

183,673

(48.3%)

Work in another municipality 431,368

(52.0%)

234,392

(52.2%)

196,693

(51.7%)

Number of full interactions 13,915

(31%)

11,582

(25.8%)

9,363

(20.8%)

Number of empty interactions 31,029

(69%)

33,362

(74.2%)

35,581

(79.2%)

Maximum number of working places in the

municipality (Ljubljana)

220,779

(26.6%)

114,484

(25.5%)

106,300

(27.9%)

Maximum number of employed active population

in the residential municipality (Ljubljana)

99,681

(12%)

50,553

(11.3%)

49,129

(12.9%)

Maximum volume of interaction between

municipalities (Domžale-Ljubljana)

7,083

(0.9%)

3,415

(0.8%)

3,668

(1%)

Source: Authors’ work.

Note: The proportions in the table are calculated according to the starting point of the column.

Table 2

Population, weighted mean commuting distance, and parameters for modelling

functional regions for 2016-2018 in Slovenia Together Men Women

Employed active population (𝑬𝑨𝑷) 829,626

(100%)

448,976

(54.1%)

380,366

(45.9%)

Weighted mean commuting distance [km] 16.0 16.8 14.9

Minimum number of the employed active

population in a functional region (𝒎𝒊𝒏𝑬𝑨𝑷)

10,000 5,400 4,600

Target number of employed active population in

functional region (𝒕𝒂𝒓𝑬𝑨𝑷)

50,000 27,000 23,000

Minimum self-sufficiency of a functional region

(𝒎𝒊𝒏𝑺𝑪)

0.6 0.6 0.6

Target self-sufficiency of a functional region

(𝒕𝒂𝒓𝑺𝑪)

0.7 0.7 0.7


Note: The values for the employed active population in the table are calculated according to

the starting point of the row.

A comparison of twelve FRs and twelve statistical regions at the NUTS 3 level (see Figure

2a) shows that only four FRs are fully consistent with the statistical regions. These are

the FRs of Murska Sobota, Slovenj Gradec, Krško, and Kranj. These are naturally

delimited and historically known regions and/or important employment areas in

Slovenia. FR of Ljubljana is much larger than an adequate statistical region; it also

contains the whole Central Sava Statistical Region and parts of three other

neighbouring statistical regions (Southeast Slovenia Statistical Region, Coastal–Karst

Statistical Region, and Gorizia Statistical Region). Other FRs are much smaller than

adequate statistical regions. In the western part of Slovenia, FRs occur at the expense

of FR Ljubljana, or they simply do not exist compared to the statistical regions. Much

53


smaller are the FRs of Maribor and Ptuj and the FRs of Celje and Velenje, which cover

the territory of two statistical regions (two FRs for each NUTS 3 region) and FR Novo

mesto and FR Nova Gorica, which are smaller at the expense of FR Ljubljana.

Surprisingly, there are no FRs of Trbovlje and Postojna in Slovenia, which has been listed

as statistical regions at the NUTS 3 level for 40 years. These are mainly covered by the

FR in Ljubljana and that of Koper.

The modelling of gender-specific functional regions at the NUTS 3 level revealed ten

male and fourteen female FRs in Slovenia (see Figures 2b and 2c). In general, this

indicates that men commute longer distances and women shorter distances.

Figure 2

Functional regions at the NUTS 3 level in Slovenia in 2020: (a) twelve functional regions

for common commuter flows, (b) ten functional regions for male commuting flows,

and (c) fourteen functional regions for women commuting flows

Figure 2a

Figure 2b

54


Source: Author’s work, SORS (2020a,b), SMARS (2020)

A comparison of twelve FRs for all commuters and ten FRs for men shows the

specificities of labour mobility for men in the eastern part of Slovenia, while the regions

in the western part correspond exactly to the general FRs. The male commuters form

a single FR of Maribor (excluding FR of Ptuj for all commuters), which corresponds to a

particular statistical region. Similarly, a single male FR for Novo mesto is formed,

including the already-mentioned FR of Krško for all commuters. In southeastern

Slovenia, more precisely in the Lower Sava Statistical and Southeast Slovenia Statistical

Regions, the areas of labour mobility for men thus differ significantly from the

corresponding regions at the NUTS 3 level. As with the general FR, other areas of labour

mobility for male commuters are considered more important than the areas around

Trbovlje and Postojna, where statistical regions have been nominally defined.

As previously mentioned, fourteen FRs for women appeared at the level of twelve

NUTS 3 regions in which ten FRs for male commuters were found, which indicates that

women generally commute shorter distances than men. Three FRs for women are fully

aligned with the statistical regions: those of Murska Sobota, Slovenj Gradec, and Krško.

However, in contrast to the general FRs, women commuters form two FRs in the Upper

Carniola Statistical Region, i.e. FR of Kranj and FR of Jesenice. In addition to this

specificity, another specificity for women is found in the FR of Postojna, which covers

the entire territory of the adequate statistical region and most of the Coastal-Karst

Statistical Region. At the expense of the FR of Postojna, two other FRs (the FR of

Ljubljana and FR of Koper) are much smaller than the adequate FRs in the system of

twelve FRs for all commuters together.

Functional regions should be delimited so that as only few labour mobility flows as

possible cross their borders. This basic principle of delimiting FRs is already built into the

method CURDS itself, which we used to model FRs. Nevertheless, in the study, we

analyzed the final sets of FRs using an FST approach. We calculated the membership

value of each municipality belonging to FR, and at the same time, we calculated the

average membership value of all municipalities in FR (i.e. the average membership

value of FR), and the average membership value of all municipalities in the system of

all FRs (i.e. the average membership value of FR systems in the country). It is

understood that a higher number of FRs will result in lower average fuzzy membership

values and vice versa. Nevertheless, some common features can be drawn from the

results in Table 3.

Of all three FRs systems, ten FRs system for men have the highest average fuzzy

membership value of the municipalities belonging to FRs, i.e. 0.84, while the system for

Figure 2c

55


women and the system for all commuters have the same average membership

values, i.e. 0.82. In all three compositions of FRs, the FR of Ljubljana has the highest

average membership value, i.e. 0.9, and the FR of Krško has the lowest. In the case of

the common FR system, the average fuzzy membership value of belonging to FR of

Krško is low, i.e. 0.67, and higher in the case of female FRs, i.e. 0.73. The latter indicates

that many more women than men have found work in the home FR of Krško. A similar

situation can be observed in the case of FR of Ptuj, where the average membership

value of municipalities to FR is 0.73 for all labour commuters, while it is 0.77 for women.

Table 3

Mean membership values of the functional regionalization in 2016-2018 in Slovenia

ID Functional region / Slovenia Together Men Women

Slovenia 0.82 0.84 0.82

11 Celje 0.79 0.79 0.80

41 Jesenice N.A. N.A. 0.78

50 Koper 0.80 0.78 0.85

52 Kranj 0.81 0.81 0.71

54 Krško 0.67 N.A. 0.73

61 Ljubljana 0.90 0.90 0.89

70 Maribor 0.82 0.87 0.84

80 Murska Sobota 0.87 0.86 0.88

84 Nova Gorica 0.87 0.88 0.89

85 Novo mesto 0.81 0.82 0.82

94 Postojna N.A. N.A. 0.72

96 Ptuj 0.73 N.A. 0.77

112 Slovenj Gradec 0.83 0.82 0.85

133 Velenje 0.80 0.77 0.80


Note: N.A. means that it does not apply to a particular case in question (there is no FR).

Discussion and conclusions

The study examined whether the statistical regions, i.e. the regions at the level NUTS 3

in Slovenia, still correspond to the gravitational zones around the more important

regional centres as defined more than 40 years ago (Vrišer, 1974, 1978; Rebec, 1983,

1984; Vrišer & Rebernik, 1993). We analysed the gravitational areas of all labour

commuters together and separately for men and women. Separately for both sexes,

we also estimated the average distance between home and work. The results of our

study are in line with the literature (White, 1977; Green et al., 1986; Tkocz & Kristensen,

1994; Sang, 2008; Prashker et al., 2008; Roberts et al., 2011; Nafilyan, 2019), since in

Slovenia, the average distance between home and work is also shorter for women

(14.9 km) than for men (16.8 km).

The number and extent of gravitational areas at the NUTS 3 level were analysed

based on areas of functional urban regions, i.e. functional regions around the main

employment centres in Slovenia. FRs were modelled using the CURDS method

(Coombes & Bond, 2008), whereby generalized areas of commuter flows were

identified at the level considered, which was defined by the population and self-

sufficiency of the regions.

A comparison of the statistical and functional regions of all commuters showed that

2/3 of FRs do not correspond to the statistical regions, that FR Ljubljana is much larger

56


and FR of Novo mesto much smaller than the corresponding statistical regions, and

that two new and important employment centres have emerged in eastern Slovenia,

i.e. Ptuj and Velenje. At the level of twelve regions in Slovenia, these employment

centres replace the previously defined centres (and their gravitational areas) of

Postojna and Trbovlje.

The FRs were modelled also separately for both sexes using the corresponding

parameters of employed men or women and regional self-sufficiency. Using such an

FR modelling approach resulted in ten FRs for males and fourteen FRs for female

commuters. The generalisation of commuter flows for men thus leads to a smaller

number of larger labour market areas for men (ten male FRs), while for women, a

larger number of smaller labour market areas (fourteen female FRs) are formed for

Slovenia. Given the shorter commuting distances for women, such a result can be

expected.

A comparison of general FRs and FRs separately for both sexes reveals some

interesting facts. While the new FR Velenje is shaped by flows of labour mobility of men

and women, the FR around Ptuj is mainly created because of the flows of labour

mobility of women. In comparison to men, women form two additional FRs, i.e. FR

Postojna and FR Kranj, which are less important if all flows are generalized at the level

of NUTS 3 regions. All three maps of FRs show a decline in the importance of Trbovlje

as a regional employment centre in Slovenia at the level in question. Based on this

result, it would be useful to reconsider the inclusion of the Central Sava Statistical

Region (SI035) in the system of NUTS 3 regions in Slovenia.

In our study, we also investigated the self-sufficiency of FRs using a fuzzy set

approach. In the system of twelve FRs for all commuters together, the average value

of the municipality's membership of FR is 0.82. FRs with lower values are regions from

which on average relatively more workers commute to another FR than from others.

These FRs are FR Celje, FR Koper, FR Kranj, FR Novo Mesto, and FR Krško. These results

are consistent with the findings of Bole (2011), who found that, from 2000 through 2009,

commuting flows from the municipalities of the Littoral–Inner Carniola Statistical Region

(SI038) and of Coastal–Karst Statistical Region, and also from the municipalities from

Southeast Slovenia Statistical Region to Ljubljana increased by more than 100%. In this

period until 2009, important sections of the motorway in all directions were completed,

making Ljubljana, Slovenia's main employment centre, more accessible.

The inclusion of FR in the system of statistical regions can also be based on the

average membership values of Slovenia's functional regionalisation. In this case, it is

useful to consider the possibility of excluding Southeast Slovenia Statistical Region from

the system of NUTS 3 regions, as the average value of municipalities in FR Krško, which

is fully consistent with the Southeast Slovenia Statistical Region, is relatively low (0.67).

A similar consideration is offered in the case of adding a new region to the system. In

this case, FR Ptuj has a low average value of municipality membership (0.73), while FR

Velenje has a much higher value (0.8); this means that FR Velenje is more suitable for

inclusion in the NUTS 3 system than FR Ptuj.

We consider the lack of analysis of the gravitational regions in the educational

systems and the regional supply systems to be an important shortcoming of our study.

Therefore, it makes sense to investigate additional FRs in the future that has a

significant impact on the creation of functional urban regions around regional

centres. These are mainly FRs in secondary and higher education, as primary

education is usually provided in the home municipality and FRs of regional supply

systems.

Functional regions, as discussed in this article, are formed based on labour mobility

of the working population. Europe, and Slovenia in particular, is faced with a rapidly

57


ageing population. According to Eurostat population projections, the age structure of

the Slovenian population is expected to change very significantly in the coming

decades, but not in the structure by gender (Eurostat, 2019). Therefore, we expect

that the change in the demographic composition of the population will not

significantly affect the formation of FRs in the coming years, but new jobs related to

the care of the elderly population will have an impact on the formation of FRs in the

coming decades.

References 1. Arnuš, D. (2020), “Analiza funkcionalnih regij v Sloveniji po spolu delovno aktivnega

prebivalstva« (Analysis of functional regions in Slovenia by gender of persons in

employment), BSc thesis University of Ljubljana, available at: https://repozitorij.uni-

lj.si/IzpisGradiva.php?lang=slv&id=117357 (15 November 2020).

2. Berry, B. J. L., Garrison, W. L. (1958), “The Functional Bases of the Central Place Hierarchy”,

Economic Geography, Vol. 34 No. 2, pp. 145–154.

3. Blumen, O. (1994), “Gender Differences in the Journey to Work”, Urban Geography, Vol.

15, No. 3, pp. 223–245.

4. Bole, D. (2011), “Changes in employee commuting: a comparative analysis of employee

commuting to major Slovenian employment centers from 2000 to 2009”, Acta geographica

Slovenica, Vol. 51, No. 1, pp. 93–108.

5. Christaller, W. (1933). Die zentralen Orte in Süddeutschland. Gustav Fischer, Jena, pp. 327-

331.

6. Coombes, M. G., Green, A. E., Openshaw, S. (1986), “An efficient algorithm to generate

official statistical reporting areas: the case of the 1984 travel-to-work-areas revision in

Britain”, Journal of the Operational Research Society, Vol. 37 No. 10, pp. 943-953.

7. Coombes, M., Bond, S. (2008), “Travel-to-Work Areas: the 2007 review Office for National

Statistics”, London, available at:

https://www.ncl.ac.uk/media/wwwnclacuk/curds/files/TTWA%20report.pdf (15 August

2020).

8. Coombes, M., Casado-Díaz, J. M., Martínez-Bernabeu, L., Carausu, F. (2012), “Study on

comparable labour market areas - Final research report”, available at:

http://www.istat.it/it/files/2014/12/Final-Report_LMA-v1-0-17102012.pdf (15 August 2020).

9. Cörvers, F., Hensen, M., Bongaerts, D. (2009), “Delimitation and coherence of functional

and administrative regions”, Regional Studies, Vol. 43 No. 1, pp. 19–31.

10. Drobne, S. (2016), “Model vrednotenja števila in območij funkcionalnih regij” (A model

evaluating the number and areas of functional regions), PhD thesis, University of Ljubljana,

available at: https://repozitorij.uni-lj.si/Dokument.php?id=97829&lang=slv (7 November

2019).

11. Drobne, S. (2020), “Kakovost funkcionalne regionalizacije po metodah CURDS in Intramax

na makro ravni: študija primera za Slovenijo” (The quality of functional regionalisation by

using CURDS and Intramax methods at the macro level: a case study for Slovenia),

Geodetski vestnik, Vol. 64 No. 1, pp. 13–32.

12. Drobne, S., Garre, A., Hontoria, E., Konjar, M. (2020), “Comparison of Two Network-Theory-

Based Methods for Detecting Functional Regions”, Business Systems Research, Vol. 11 No.

2, pp. 15-29.

13. Drobne, A., Lakner, M. (2016), “Use of constraints in the hierarchical aggregation procedure

Intramax”, Business Systems Research, Vol. 7 No. 2, pp. 5-22.

14. EC (2019), “Regulation (EC) No 1059/2003 of the European Parliament and of the Council

of 26 May 2003 on the establishment of a common classification of territorial units for

statistics (NUTS), Consolidated text 13/11/2019”, European Commission (EC), Brussels,

available at: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A02003R1059-

20191113 (15 October 2020).

15. Eurostat (2019), “Ageing Europe, Looking at the Lives of Older People in the EU, 2019

Edition”, available at: https://ec.europa.eu/eurostat/documents/3217494/10166544/KS-

02-19%E2%80%91681-EN-N.pdf/c701972f-6b4e-b432-57d2-91898ca94893 (19 March 2021).

58


16. Eurostat (2020a), “NUTS – Nomenclature of territorial units for statistics. History of NUTS”,

available at: https://ec.europa.eu/eurostat/web/nuts/history (15 October 2020).

17. Eurostat (2020b), “NUTS – Nomenclature of territorial units for statistics. Background”,

available at: https://ec.europa.eu/eurostat/web/nuts/background (12 October 2020).

18. Eurostat (2020c), “NUTS – Nomenclature of territorial units for statistics. Local Administrative

Units (LAU)”, available at: https://ec.europa.eu/eurostat/web/nuts/local-administrative-

units (11 October 2020).

19. Fanning-Madden, J. (1981), “Why Women Work Closer to Home”, Urban Studies, Vol. 18 No.

2, pp. 181-194.

20. Franconi, L., D'Alò, M., Ichim, D. (2016a), “Istat Implementation of the Algorithm to Develop

Labour Market Areas, Technical Report”, Istat, available at:

https://www.istat.it/en/files/2016/03/Description-of-the-LabourMarketAreas-algorithm.pdf,

(15 September 2020).

21. Franconi, L., Ichim, D., D'Alò, M., Cruciani, S. (2016b), “Guidelines for Labour Market Area

Delineation Process: From Definition to Dissemination”, Istat, available at:

https://ec.europa.eu/eurostat/cros/system/files/guidelines_for_lmas_production08082017

_rev300817.pdf (15 September 2020).

22. Franconi, L., Ichim, D'Alò, M. (2017). Labour Market Areas for Territorial Policies: Tools for a

European Approach. Istat, available at:

https://content.iospress.com/download/statistical-journal-of-the-

iaos/sji160343?id=statistical-journal-of-the-iaos%2Fsji160343 (15 September 2020).

23. Feng, Z. (2009), “Fuzziness of travel to work areas”, Regional Studies, Vol. 43 No. 5, pp. 707-

720.

24. Green, A. E., Coombes, M. G., Owen, D. W. (1986), “Gender-Specific Local Labour Market

Areas in England and Wales”, Geoforum, Vol. 17 No. 3, pp. 339–351.

25. Karlsson, C., Olsson, M. (2006), “The identification of functional regions: theory, methods,

and applications”, The Annals of Regional Science, Vol. 40 No. 1, pp. 1-18.

26. Van der Laan, L., Schalke, R. (2001), “Reality versus policy: the delineation and testing of

local labour market and spatial policy areas”, European Planning Studies, Vol. 9 No. 2, pp.

201-221.

27. Nafilyan V. (2019), “Gender Differences in Commute Time and Pay”. Office for National

Statistics, London, United Kingdom, available at

https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkin

ghours/articles/genderdifferencesincommutetimeandpay/2019-09-04#toc (9 September

2020).

28. OECD (2002), Redefining territories - The functional regions. OECD Publishing, Paris.

29. Prashker, J., Shiftan, Y., Hershkovitch-Sarusi, P. (2008), “Residential Choice Location, Gender

and the Commute Trip to Work in Tel Aviv”, Transport geography, Vol. 16 No. 5, pp. 332–

341.

30. Rebec, J. (1983), Razvoj naselij mestnega značaja z več kot 2000 prebivalci v SR Sloveniji,

1971–1981. Prikazi in študije, Ljubljana. No. 28.

31. Rebec, J. (1984). Funkcijska klasifikacija mestnih naselij z več kot 2000 prebivalci v SR

Sloveniji, 1981. Prikazi in študije, No. 29.

32. Roberts, J., Hodgson, R., Dolan, P. (2011), “It’s Driving her Mad: Gender Differences in the

Effects of Commuting on Psychological Health”, Journal of Health Economics, Vol. 30 No.

5, pp. 1064–1076.

33. Sang, S. M. A. (2008), “Examining Commuting Patterns and Spatial Mismatch by

Occupation and Gender: Disaggregate Journey-to-Work Model”. PhD thesis, Ohio State

University, Columbus.

34. SMARS (2020), “Spatial data on municipalities, 2019”, the Surveying and Mapping Authority

of the Republic of Slovenia, available at: https://egp.gu.gov.si/egp/ (10 August 2020).

35. SORS (2020a), “Slovene Statistical Regions and Municipalities in Figures”, Statistical Office

of the Republic of Slovenia, Ljubljana, available at https://www.stat.si/obcine/en (12

October 2020).

36. SORS (2020b), “SI-Stat data portal”, available at:

http://pxweb.stat.si/pxweb/dialog/statfile1.asp (25 August 2020).

http://pxweb.stat.si/pxweb/dialog/statfile1.asp

59


37. Tkocz, Z., Kristensen, G. (1994), “Commuting Distances and Gender: A Spatial Urban

Model”, Geographical analysis, Vol. 26 No. 1, pp. 1–14.

38. Vanhove, N., Klaassen, L. H. (1987). Regional Policy: A European Approach, 2 Edition.

Gower Publishing Company Limited, Aldershot.

39. Vrišer, I. (1974), “Mesta in urbano omrežje v SR Sloveniji. Značilnosti njihovega razvoja in

družbeno gospodarskega pomena s posebnim ozirom na mala mesta”, Geografski zbornik,

Vol. 14 No. 3, pp. 179–337.

40. Vrišer, I. (1978), Regionalno planiranje. Mladinska knjiga, Ljubljana.

41. Vrišer, I., Rebernik, D. (1993), »Družbenogospodarska in dejavnostna usmeritev slovenskih

mest«, Geografski zbornik, No. 33, pp. 9–40.

42. Watts, M. (2009), “Rules versus hierarchy: an application of fuzzy set theory to the

assessment of spatial grouping techniques”, in Kolehmainen, M., Toivanen, P., Beliczynski,

B. (eds), 9th International Conference on Adaptive and Natural Computing Algorithms, 23-

25 April, Springer-Verlag, Berlin Heidelberg, pp. 517-526.

43. Watts, M. (2013), “Different spatial grouping algorithms: an application to the design of

Australia’s new statistical geography”, Spatial Economic Analysis, Vol. 8 No. 1, pp. 92-112.

44. White, M. J. (1977), “A Model of Residential Location Choice and Commuting by Men and

Women Workers”, Journal of Regional Science, Vol. 17 No. 1, pp. 41–52.

45. Wikipedia (2020), “Statistical Regions of Slovenia”, Wikipedia Free Encyclopedia, available

at https://en.wikipedia.org/wiki/Statistical_regions_of_Slovenia (12 October 2020).

About the author

Samo Drobne is an Assistant Professor and the vice dean of educational affairs within

the Faculty of Civil and Geodetic Engineering, University of Ljubljana (Slovenia). He

teaches courses on statistics, geographical information systems (GIS), and spatial

analyses in GIS. His main research fields include regional development and planning,

spatial interaction models, functional regions, commuting, migration, spatial analysis

in GIS, and operational research in spatial systems. Currently, he is a member of the

narrow working group for concepts and legislative bases for establishing provinces in

Slovenia where he helps develop the concepts of functional regions as bases for

provinces. He is actively involved in several international and national research

projects. Further information is available at http://fgg-web.fgg.uni-lj.si/~/sdrobne/. The

author can be contacted at [email protected]

http://fgg-web.fgg.uni-lj.si/~/sdrobne/

60


The Effect of External Knowledge Sources on

Organizational Innovation in Small and

Medium Enterprises in Germany

Shoaib Abdul Basit

Faculty of Economics and Business Administration, Chemnitz University of

Technology, Germany

Abstract

Background: Firms increasingly depend on external actors for the process of

generating innovation. Interaction with these actors might occur through an official

collaboration agreement or via external actors as the source of information.

Objectives: Although open innovation has received more attention, still less is known

about its effect on organizational innovation. To fill this gap, this study investigates the

impact of various external knowledge sources on the willingness of small and medium-

sized enterprises to introduce organizational innovation. Methods/Approach: To

achieve the proposed objective, the German Community Innovation Survey

conducted in 2017 is used for the econometric analysis. Results: Different external

sources of knowledge are relevant for the introduction of organizational innovation in

small firms (customers in the private sector, competitors, conferences, and

crowdsourcing) compared to medium-sized firms (customers in the private sector and

industry associations). Conclusions: External knowledge sources are more important

for small firms compared to medium firms, and those small firms are more likely to use

various sets of external knowledge.

Keywords: organizational innovation; workplace; external knowledge sources; small

and medium-sized firms

JEL classification: O31, O36


Received: Jan 13 2020

Accepted: Jun 25 2020

Citation: Abdul Basit, S. (2021), “The Effect of External Knowledge Sources on

Organizational Innovation in Small and Medium Enterprises in Germany”, Business

Systems Research, Vol. 12, No. 1, pp. 60-79.


61


Introduction Innovation is considered an important factor for firms’ longevity in the marketplace. In

literature, many scholars confirm that innovation is the key element for growth

(Schumpeter, 1934; Volberda et al., 2014). The broad definition of innovation is

described as the implementation of significantly improved or new product and service

as well as process, and the introduction of a new marketing method or a new

organizational system at the firms’ workplace or the adoption of new organizational

procedures in occupational practices (OECD, 2005).

Chesbrough presents the concept of open innovation and highlights the

significance of external sources of information for the innovation process and further

posits that internal R&D is no longer any more strategic assets for the firms, once it was

(Chesbrough, 2003). Open Innovation provides valuable knowledge for better

innovation performance and its widely acknowledged as a key factor for innovation

management practices (Chesbrough et al., 2014).

The role of external sources of knowledge as an important factor of innovation has

received significant attention in the literature. Numerous empirical studies claim that

increasing openness towards external knowledge enhances firms’ innovation

performance (Van de Vrande et al., 2009; Leiponen and Helfat, 2011; Gómez, et al.,

2016). Previous studies identify the variety of useful external sources of knowledge and

their effect on innovation. For instance, knowledge sources are considered as the

information transfer channels through informal networks such as competitors, suppliers,

universities and customers (Gassmann and Enkel, 2004). Despite a growing literature on the effect of various sources of external knowledge

on different types of innovation, prior studies mostly focus on a general effect of

external knowledge from the customers without delineating into the public and

private sector. It is also notable that sourced knowledge from competitors, suppliers,

and public research institutes is limited to product innovation (Tsai, 2009; Spithoven et

al., 2013; Köhler et al., 2012). Also, most scholars focused on external sources impact

in general on technological innovation i.e., radical or incremental innovation (Wang

and Xu, 2018; Zouaghi et al., 2018) and technology innovation performance (Kang

and Kang, 2009) and new product development (Santoro et al., 2018a; Iglesias-

Sánchez et al., 2019) and product and process innovation (Dotzel and Faggian, 2019;

Criscuolo et al., 2018).

Some previous studies argue that various types of external sources of knowledge

differ significantly for firms’ innovation performance (Kang and Kang, 2009; Köhler et

al. 2012). In addition, Knoben and Oerlemans (2010) reveal that the effect of various

external sources of knowledge on innovation output differs significantly and further

suggest that it is beneficial to differentiate the various types of external knowledge

sources as well as the diverse level of novelty to the innovation outcomes. Therefore,

to develop a suitable strategy for the type of knowledge search, it is strongly needed

to understand the effect of the various external knowledge search on firm innovation

activities.

Also, West and Bogers (2014) provide the literature review of research on open

innovation and suggest that further research is required on individuals as a source of

innovation. However, the scholars examine the impact of sources of knowledge on

innovation output but without considering the various types of external knowledge

sources for innovation activities in small and medium firms. Although scholars

acknowledge that SMEs play an important role in innovation (Chesbrough et al., 2006),

however, researchers have scarcely explored how small and medium enterprises use

62


various sources of ideas and knowledge for innovation (Brunswicker and

Vanhaverbeke, 2015). Previous research on open innovation mostly focused on large

firms ( Hinteregger et al., 2019; Bogers et al., 2017; van de Vrande et al., 2009), although

studies claim that SMEs play a vital role in economic growth (Muller et al., 2015;

Gassmann, et al., 2010). Further, some scholars repeatedly stress the importance of

SMEs and suggest that there is a need to explore in-depth open innovation in SMEs

(Wynarczyk, 2013; Spithoven et al., 2013; van de Vrande et al., 2009) and also open

innovation in SMEs receive less attention from scholars (Hossain, 2013). Consequently,

the present study considers small and medium-sized firms to offer a deeper

understanding of how SMEs involve in the sources of open innovation.

Therefore, the present paper recognizes the key research gap in the existing

literature on how the relationship with external partners can bring the information

needed to adopt an extensive range of internal organizational innovation practices,

which might enhance firm performance at the workplace. As observed in the

literature, scholars emphasis on the impact of external knowledge sources from

customers in general (without delineating customers from the public and private

sector), competitors, suppliers, public research institute and its impact is limited to

product innovation (Tsai, 2009; Fitjar and Rodríguez-Pose, 2013; Köhler et al., 2012) and

on internal and external R&D activities (Pejić Bach et al., 2015). Research conducted

on various external sources of knowledge and their effect on organizational

innovation is scarce. Hence, the present study aims to shed more light on this particular

topic by addressing the following research question: Do various external sources of

knowledge from conferences, customer private, customer public, industry

associations, competitors and from crowdsourcing have an effect on organizational

innovation in small and medium enterprises? This study contributes to the current

literature by considering the specific external knowledge sources (i.e conferences,

industry associations, customers from the public and private sector, and

crowdsourcing: ideas from the general public) and its impact on organizational

innovation in terms of decision making and the adoption of new methods of work

organizing responsibilities at the working place in SMEs. The present research is mainly

triggered by West and Bogers (2014), who reviews the literature on open innovation

and suggest examining the effects of individual external knowledge sources on

innovation. However, the scholars mainly examine the combined effect of the sources

of external knowledge on innovation outcomes but without considering the various

types as an individual source of external knowledge for innovation activities in small

and medium firms. The identification and suitability of the various sorts of external

partners available to share their information and knowledge freely may decrease

costs that are linked with formal forms of collaboration to enhance organizational

innovation in the firm’s workplace. Additionally, this study advances research by

considering the different modes of external knowledge sources impact on

organizational innovation in small and medium-sized enterprises separately, as the

researchers have scarcely explored how small and medium enterprises use various

sources of ideas and knowledge to enhance innovation performance (Brunswicker

and Vanhaverbeke, 2015). Thus, recognizing the certain type of external knowledge

source could be beneficial for the firms to know whether the specific information

sources affect the introduction of organizational innovation in small and medium-sized

firms according to their needs.

The rest of the paper is organised as follows. First, the theoretical background of the

study and hypothesis development are presented. Then, the data source and

statistical methods are described. Third, the details of descriptive statistics and the

63


empirical findings are discussed. Finally, the paper provides the discussion,

conclusions, and policy implications.

Theoretical background Organizational innovation Literature shows that organisational knowledge plays a significant role in competitive

benefits in the market, as it helps the organization to create identical and valuable

ways to compete (Hall, 1992; Reed and DeFillippi, 1990). Similarly, Tsoukas (1996)

highlight that those firms which contain the organizational knowledge about how to

utilize the resources can attain a high degree of efficiency and effectiveness in the

organization. In addition, organizational changes are linked with better performance

(Greenan and Mairesse, 2003), and the introduction of new methods, management

tools, as well as practices that enable organizational changes and eventually

advance the organizational competitiveness (Damanpour, 2014).

Mostly prior research acknowledged the importance of traditional innovation types

(namely, product and process innovation) and their economic impact (Hervas-Oliver

et al., 2015; Brettel and Cleven, 2011). Likewise, Keupp et al. (2012) conduct a

systematic literature review on strategic management of innovation view and find

that out of 342 studies, 246 studies include the product and process innovation while

only 23 studies consider the non-technological innovations (i.e. marketing and

organizational innovation) in their analyses. Focusing mostly on technological

innovation (i.e. product and process innovation) signifies a research gap and in the

future, it would not be sufficient to ensure the success of innovation (Eurostat, 2016). In

this aspect, many scholars argue that technological innovations need to be

downsized and more attention should be given to the importance of non-

technological innovation (Volberda et al., 2014; Damanpour, 2014).

In the acknowledgement of organizational innovation, previous studies shed lights

on the importance of organizational innovation and support that there is a positive

effect of organizational innovation on performance outcomes (Anzola-Román et al.,

2018; Chen et al., 2019). In addition, others stress this link even more and state that

organizational innovation could support long term competitive benefit to the firms, as

it is a resource that is firm-specific, unique, valuable and difficult to replicate

(Damanpour, 2014; Mol and Birkinshaw, 2009).

Moreover, organizational innovations are frequently intended to improve

workplace satisfaction, enhance the exchange of valuable knowledge and boost a

firm’s capability to learn from the environment and use new knowledge and reduces

administrative and transaction costs (OECD, 2005). As the concept of innovation has

changed from a technological innovation into a broader viewpoint with the OECD

inclusion of non-technological innovation and particularly organization innovation.

Hence, this change needs a comprehensive analysis of how external knowledge

sources influence organizational innovation.

Knowledge sources and organizational innovation Open innovation is acknowledged as an important ingredient for innovation

management (Gassmann and Enkel, 2004). The idea of blending sources of external

knowledge (instead of depending on internal sources only) for the innovation process

has been emphasized frequently in the previous studies on innovation. Through

different approaches, previous studies highlight the importance of taking benefit from

64


external sources of knowledge to improve organizational innovation (Ferraris et al.,

2017; Cohen, and Levinthal, 1990).

Many scholars highlight the variety of useful external sources of knowledge such as

customers, competitors, suppliers, universities, consultants, professional and industrial

associations (Gassmann and Enkel 2004; Chesbrough, 2003). Innovation studies show

that knowledge source from various partners is considered as a significant factor for

innovation success (Mol and Birkinshaw, 2014; Gellynck and Vermeire, 2009). Further,

different types of external knowledge source i.e. suppliers, clients, and universities are

linked with a series of advantages. These external actors give access to the external

source of knowledge (Teece, 1986) and especially this is relevant in the case of

knowledge transfer which is tacit and not easily modifiable (Hippel, 1988).

In this sense, firms do not operate in the market alone but also get benefit from the

external environment such as external sources of knowledge. Hence, in the purpose

of the present study, it is essential to know the effect of external sources of knowledge

on organizational innovation. This study considers six types of external knowledge

sources called as: i) customers from public sector ii) customers from private sector iii)

conferences iv) industry associations v) competitors vi) crowdsourcing i.e. ideas from

the general public. These six sources are selected for two main reasons. First, the role

of this external knowledge on the innovation process is well established in the existing

literature (Tidd et al., 2005). Secondly, these sources are certain organizations that

show the source of knowledge that they include.

Hypothesis development Knowledge sources from customers are considered an important factor for

organizational innovation (Tether and Tajar, 2008b). The new ways of organizing firms’

customers' interaction can enhance firms’ organizational innovation performance. As

customers know their needs and expectations, so they provide valuable knowledge

to the firms (Santoro et al., 2018b; Tether, 2002) and therefore it encourages the firms

to adopt the innovation practices at the firm workplace (Guler et al., 2002). Firms can

get benefit from the external knowledge sources for innovation capabilities that some

firms do not possess and such external knowledge sources might grant access to

innovation (West and Bogers, 2014). Previous studies argue that information sources

from the customers, suppliers, competitors, and consultants make it possible for the

firms to bring new ideas for the innovation through combining these external

information sources with their internal existing knowledge (Tether and Tajar 2008a;

Lefebvre et al., 2015).

Moreover, Birkinshaw et al. (2008) argue that external partners (such as suppliers as

well as clients, form a common and cooperative group of partners) share the

management knowledge which encourages the adoption of the firm’s organizational

innovation. Further, Mol and Birkinshaw (2009) point out that the uses of greater

breadth of external knowledge search by the firms lead to the greater introduction of

new management practices. Feedbacks from the customers are very important, as

the customers provide first-hand user experiences and inform their sensitivity to the

market trend and customers also evaluate the firm’s new product concept (Chang

and Taylor, 2016). Thus, this present study contends that firms may adopt new

organizational innovation at the firms’ workplace when ideas are offered by

customers from the public and private sector and crowdsourcing (i.e. ideas from the

general public). This leads to the following hypothesis:

65


o H1. Knowledge source through the customers in the private sector positively

influences firms’ propensity to undertake organizational innovation at the firms’

workplace.

o H2. Knowledge source through the customers in the public sector positively

influences firms’ propensity to undertake organizational innovation at the firms’

workplace.

o H3. Knowledge source from crowdsourcing such as ideas from the general

public positively influences firms’ propensity to undertake organizational

innovation at the firms’ workplace.

Information source from competitors is significant to firms as rivals mostly require

similar needs for their innovation process (Lhuillery and Pfister, 2009). Also, competing

firms face similar technological issues, so market collaboration with competitors and

customers support firms to obtain new technological knowledge as well as to practice

and access other information sources (Gnyawali and Park, 2011). Further, cooperation

with competitors also provides opportunities to seek a successful organizational

structure from the rivals (Pippel, 2014). It is also possible when all face the same issues

in the market which might be outside of the competition area, for instance, the

creation of regulatory structure in the operating market (Tether, 2002). Further, market

partners provide operational knowledge which is related to the focal firms for

improvement of the organization process (Al-Laham et al., 2010). Therefore, in this

way, it might encourage the development of organizational structure at the firms’

workplace. Based on the presented literature, the following hypothesis is formulated:

o H4. Knowledge source from the competitors positively influences firms’

propensity to undertake organizational innovation at the firms’ workplace.

Some scholars argue that information sources from the specialist knowledge

providers (such as consultants, trade associations, private research organizations, and

universities) are more likely to be used by firms that tend to complement their internal

innovation activities (Lefebvre et al., 2015; Tether and Tajar, 2008a). Further, external

knowledge sources from conferences and trade fairs are presumed to serve as an

instrument where firms can make connections with various potential knowledge

suppliers and then firms can obtain knowledge from these sources (Sofka and Grimpe,

2010). Additionally, knowledge sources from scientific and industry publications, trade

fairs and conferences are easily accessible and almost no barriers exist to access

knowledge from these sources. Moreover, Sofka and Grimpe (2010) argue that

externally available information sources from science-driven search strategy such as

knowledge source from public research centres and universities and supply-driven

search strategy including conferences, trade fairs, and suppliers enhance innovation

performance. Based on the aforementioned discussion, the following hypothesis is

proposed:

o H5. Knowledge source from conferences positively influences firms’ propensity

to undertake organizational innovation at the firms’ workplace.

Firms that are associated with business groups take advantage of intragroup

network resources which are different from those sources acquired from its externally

inter-firm networks (Yiu et al., 2005). Additionally, some studies argue that business

groups offer group-level resources different from those provided by the external

networks and further business group-level resources are very important for innovation,

especially when the market infrastructures are in a developing stage (Choi et al., 2011;

Chang et al., 2006).

The share of tacit management knowledge happens when firms operate in a similar

market and firms possess similar competencies, resources, and skills, thus such similarity

66


results in collaboration with others for the implementation of new knowledge (Mowery

et al., 1996). Moreover, Turulja and Bajgorić (2018) claim that for organizational

learning shared values and openness play a positively significant role in the

knowledge management competencies of the firms. Hence, the present study argues

that industry associations should enhance firms’ organizational innovation at the firms’

workplace through knowledge transfer and industry complementarity. This leads to

the following hypothesis:

o H6. Knowledge source from the industry associations positively influences firms’

propensity to undertake organizational innovation at the firms’ workplace.

In the next section, we discuss the dataset, characteristics of perspective estimation

variables, and the statistical method used for the analysis.

Methodology Data The present study examines the Mannheim Innovation Panel (MIP) database. The MIP

database is financed by the German Federal Ministry of Education and Research and

MIP is the German part of the Community Innovation Survey. Since 1993, the Center

for European Economic Research (ZEW) has been conducting the annual survey (in

Germany) on innovation activities with the firms having at least 5 employees. The MIP

data is collected by sending the questionnaires via email. The survey methodology is

based on the recommendations of Eurostat and OECD Oslo Manual on innovation

statistics. In the survey, the managers are asked about their firms´ process for

generating innovation. Therefore, the MIP database provides an extensive variety of

general information on innovation activities i.e. firm size, sector of activity,

geographical markets, product, process, marketing and organizational innovations,

external knowledge sources for innovation etc.

This study employs the data from MIP by using the survey wave conducted in the

year 2017 that is referred to as Community Innovation Survey (CIS 2017). CIS (2017)

covers the years from 2014 – 2016 and includes information about the sector of

business group activities, external information sources, geographical market, firm size,

and innovation activities. This sample considers the German manufacturing and

service firms and provides information regarding the introduction of new products,

services, and innovation process (such as product, process, marketing, and

organizational) within firms.

Research instrument The first part contains the information on organizational innovation (the dependent

variable) measured as the introduction of new methods of organizing work

responsibilities and decision making at the firm workplace. It is a binary variable that

corresponds to the measurement of organisational innovation at the workplace (see

Abdul Basit et al., 2018). This organizational innovation measurement is proposed by

the OECD (2005) and it is widely used in the literature. The third part measures the

innovations in logistics as a digital innovation supply chain management

(Diginnospmana) for robustness check. The dummy variable “Diginnospmana” takes

the value of “1” if the firm introduced innovations in logistics as a digital supply chain

management and “0” otherwise.

67


Table 1

Dependent variables Variable code Variable name Type Description

Organizational

inno

Organizational

innovation

Dummy 1 if the firm introduced new methods of

organizational innovation (i.e new methods of

organizing work responsibilities and decision

making) activities in the period of 2014-2016 and 0

otherwise

Diginnospmana Digital supply

chain

management

Dummy 1 if firm introduced innovations in logistics as a

digital supply chain management (i.e. including

planning, organization, management, paperless,

transparent supply chain transactions) from 2014

to 2016 and 0 otherwise

Source: Author work

The second part measures the source of information for new ideas for innovation

projects. These information sources include six distinctive external knowledge sources,

namely knowledge sources from the customers in the public and private sector,

conferences, competitors, industry associations and crowdsourcing to measure the

independent variables. Community Innovation Survey (CIS) 2017 asks the respondents

about the degree of importance of various external knowledge sources usage for

innovation activities. CIS 2017 measures the importance of these information sources

in firms’ innovative activities as high, medium, low and not used. Thus, for the aim of

this paper, the variables are re-scaled as binary: “0” not perceived by the firm as a

type of information source; “1” perceived by the firm as a type of information source.

These knowledge sources are measured following previous studies (Tether and Tajar,

2008b, Tsai, 2009).

Table 2

Independent variables – Knowledge sources for innovation Variable code Variable name Type Description

Customer_private Customers from

the private sector

as the source of

knowledge

Dummy 1 if the firm receives new ideas from the

customers from a private sector between

2014 and 2016 and 0 otherwise

Customer_public Customers from

the public sector

as the source of

knowledge

Dummy 1 if the firm receives new ideas from

customers from the public sector between

2014 and 2016 and 0 otherwise

Competitors Competitors as the

source of

knowledge


competitors between 2014 and 2016 and 0

otherwise

Conferences Conferences as

the source of

knowledge


conferences, trade fairs between 2014 and

2016 and 0 otherwise

Crowd_sourcing Crowdsourcing as

the source of

knowledge


crowdsourcing, ideas or inputs from the

general public between 2014 and 2016 and 0

otherwise

Indus_association Industrial

associations as the

source of

knowledge


professional and industry associations

between 2014 and 2016 and 0 otherwise

Source: Author work

In the third part, control variables (i.e. industry characteristics, graduate employees

and R&D intensity) are included that might affect the firm’s organizational decision to

implement open innovation practices. Moreover, following Castellacci (2008)

68


classification, we classify the industries in eight innovation trajectories. The detailed

information and measurement of the variables and industry classification are

presented in Tables 1, 2 and 3.

Table 3

Control variables and industry dummies Variable code Variable name Type Description

Graduate

employees

Graduate employees Centred The total number of employees holding a university

degree in survey year (2017), due to continuous

variable, mean is calculated and then centred it.

National_market National market Dummy 1 if the firm operates in the national market of

Germany and 0 otherwise

R&D Intensity R&D Intensity Continuous R&D expenditures as a share of turnover

MPG-SB Mass production

goods: science-based

manufacturing

Dummy MPG-SB =1 if firms are classified in mass production

goods: science-based manufacturing (electrical

equipment, media service, chemicals, office

machinery, and computers) and 0 otherwise

MPG-SI Mass production

goods: scale intensive

manufacturing

Dummy MPG-SI =1 if firms are classified in mass production

goods: scale intensive manufacturing (mining, plastics,

metals, other non-metallic mineral products, motor

vehicles) and 0 otherwise

PGS-M Personal goods and

services: supplier

dominated

manufacturing

Dummy PGS-M =1 if firms are classified in personal goods and

services: supplier dominated manufacturing (food,

tobacco, textiles, wood, paper, furniture, and toys)

and 0 otherwise

AKP-M Advanced

knowledge providers:

specialized supplier

manufacturing

Dummy AKP-M =1 if firms are classified in advanced

knowledge providers: specialized supplier

manufacturing (glass, ceramics, machinery and

equipment, precision and optical instrument) and 0

otherwise

AKP-S Advanced

knowledge providers:

knowledge-intensive

business services

Dummy AKP-S =1 if firms are classified in advanced knowledge

providers: knowledge-intensive business services (IT,

telecommunication or computer and related

activities, technical services and R&D services,

consulting advertisement or other business activities)

and 0 otherwise

PGS-S Personal goods and

services: supplier

dominated services

Dummy PGS-S =1 if firms are classified in personal goods and

services: supplier dominated services (automobile,

retail or sales, maintenance and repair of motor

vehicles) and 0 otherwise

SIS-N Supporting

infrastructure services:

network infrastructure

Dummy SIS-N =1 if firms are classified in supporting infrastructure

services: network infrastructure (banking, insurance

and pension funding, financial intermediation, post,

and telecommunication) and 0 otherwise

SIS-P Supporting

infrastructure services:

physical infrastructure

Dummy SIS-P =1 if firms are classified in supporting infrastructure

services: physical infrastructure (wholesale, energy,

land, water, supporting and auxiliary transport

activities) and 0 otherwise

Note: The industry groups are introduced based on Castellacci, (2008) classification.

Source: Author work

Statistical methods The empirical analysis is divided into two steps. The first step provides the descriptive

analysis among small and medium-sized firms. Second, to test the hypothesis, a logit

regression is used for the main analysis and the robustness checks as well. Since the

dependent variable is a binary, which takes the value of 1 if firms introduce the new

methods of work responsibility and decision-making at the firm workplace and 0

otherwise, hence, logit estimation seems to be a suitable technique for the analysis

(Papke and Wooldridge, 1996). The previous study suggests that if the dependent and

independent variables are in a binary or dichotomic nature, a logit regression might

be the appropriate technique (see Hair et al., 2010). Also, the logit regression

69


technique is used in earlier studies with similar data structures (see Spithoven et al.,

2013; Damanpour et al., 2018). To investigate the impact of external knowledge

sources on a firm’s innovativeness, we use the logistic regression models. For the main

analysis, organizational innovation is the dependent variable.

𝑂𝑟𝑔𝑎𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝐼𝑛𝑛𝑜= 𝐵0 + 𝛽1(Customer_private) + 𝛽2(Customer_public) + 𝛽3(Competitors) + 𝛽4(Conferences) + 𝛽5(Crowd_sourcing) + 𝛽6(𝐼𝑛𝑑𝑢𝑠_𝑎𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑖𝑜𝑛)+ 𝛽7(Graduate employees) + 𝛽8(National_market) + 𝛽9(R&D Intensity) + 𝛽10(Industries dummies) (1)

For the robustness check, digital innovation supply chain management

(Diginnospmana) is the dependent variable. Diginnospmana

= 𝐵0 + 𝛽1(Customer_private) + 𝛽2(Customer_public) + 𝛽3(Competitors) + 𝛽4(Conferences) + 𝛽5(Crowd_sourcing) + 𝛽6(𝐼𝑛𝑑𝑢𝑠_𝑎𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑖𝑜𝑛)+ 𝛽7(Graduate employees) + 𝛽8(National_market) + 𝛽9(R&D Intensity) + 𝛽10(Industries dummies) (2)

Results Descriptive statistics by firm size To analyze the effect of various external knowledge sources according to the

characteristics of the firm, the sample is separated by firm size (i.e. small and medium-

sized firms). Table 4 confirms on average that the use of external knowledge sources

is higher in medium-sized firms than in small firms.

Table 4

Descriptive statistics by small and medium-sized firms Small firms

(less than 50 employees)

Medium firms

(50 to 249 employees)

Variables Firms % Mean SD Firms % Mean SD

Organizational innovation 18.27 0.183 0.386 28.69 0.287 0.452

Diginnospmana 4.41 0.044 0.205 9.62 0.096 0.295

Customer_ private 33.82 0.338 0.473 46.21 0.462 0.499

Customer_ public 22.28 0.223 0.416 28.38 0.284 0.451

Ccompetitors 34.04 0.340 0.474 48.79 0.488 0.500

Conferences 32.35 0.323 0.468 47.77 0.478 0.499

Crowd_sourcing 11.45 0.114 0.318 14.93 0.149 0.356

Indus_association 24.25 0.242 0.429 38.94 0.389 0.488

Graduate employees 66.15 0.661 0.473 87.96 0.879 0.325

National_market 62.20 0.622 0.485 71.46 0.714 0.452

R&D Intensity - 0.010 0.033 - 0.010 0.027

MPG-SB 13.83 0.138 0.345 14.15 0.141 0.349

MPG-SI 13.08 0.131 0.337 18.30 0.183 0.387

AKP-S 20.31 0.203 0.402 9.93 0.099 0.299

AKP-M 4.47 0.045 0.207 6.80 0.068 0.252

PGS-S 1.66 0.016 0.128 2.74 0.027 0.163

SIS-N 3.25 0.032 0.177 2.81 0.028 0.165

PGS-M 17.71 0.177 0.382 20.09 0.201 0.401

SIS-P 25.69 0.257 0.437 25.18 0.252 0.434

Obs. 3196 1279

Source: Author work

About 29 % of the firms in this dataset introduce organizational innovation in

medium firms while 18 % of the firms introduce organizational innovation in the

70


workplace for small firms. On the other hand, in medium firms, 9 % of the firms are

involved in logistic innovation as digital supply chain management “Diginnospmana”

and 4% of the firms are involved in digital supply chain management in small

enterprises. Further, about 66% in small firms and 88 % in medium firms’ employees hold

a university degree. Approximately, 71 % of the medium firms and 62 % of the small

innovative firms actively seeking business in the national market.

Effect of external knowledge sources on organizational innovation

in small and medium-sized firms Table 5 presents the results of logit regression for the effect of external knowledge

sources on organizational innovation in the workplace by firm size.

Table 5

Results of logit regression for the effect of external knowledge sources on

organizational innovation in the workplace by firm size Small firms

(less than 50 employees)

Medium firms

(50 to 249 employees)

Variables Organizational

innovation

dy/dx

(Marg Eff)

Organizational

innovation

dy/dx

(Marg Eff)

External Knowledge sources

Customer_private 0.639*** (0.160) 0.086*** (0.021) 0.763*** (0.208) 0.143*** (0.038)

Customer_public 0.133 (0.132) 0.018 (0.018) -0.324* (0.169) -0.061* (0.032)

Competitors 0.566*** (0.170) 0.076*** (0.023) 0.111 (0.228) 0.021 (0.043)

Conferences 0.280* (0.165) 0.038* (0.022) 0.102 (0.235) 0.019 (0.044)

Crowd_souring 0.260* (0.141) 0.035* (0.019) 0.260 (0.185) 0.049 (0.035)

Indus_association 0.098 (0.139) 0.013 (0.019) 0.472** (0.187) 0.089** (0.035)

Control variables and industry dummies

Graduate employees 0.035* (0.020) 0.005* (0.003) 0.019 (0.036) 0.004 (0.007)

National_market 0.310*** (0.111) 0.042*** (0.015) 0.057 (0.159) 0.011 (0.030)

R&D Intensity -3.609** (1.432) -0.484** (0.191) -1.914 (2.526) -0.360 (0.475)

MPG-SI 0.095 (0.193) 0.013 (0.026) 0.035 (0.234) 0.007 (0.044)

AKP-S 0.309* (0.165) 0.041* (0.022) 0.376 (0.266) 0.071 (0.050)

AKP-M -0.021 (0.264) -0.003 (0.035) -0.006 (0.293) -0.001 (0.055)

PGS-S -0.301 (0.402) -0.040 (0.054) -0.208 (0.434) -0.039 (0.082)

SIS-N 0.176 (0.287) 0.024 (0.038) 0.537 (0.404) 0.101 (0.076)

PGS-M 0.056 (0.180) 0.008 (0.024) -0.131 (0.229) -0.025 (0.043)

SIS-P 0.321* (0.170) 0.043* (0.023) 0.018 (0.236) 0.003 (0.044)

Constant -2.600*** (0.169) -1.628*** (0.233)

LR chi2(16) 305.98 102.93

Prob > chi2 0.0000 0.0000

Pseudo R2 0.1007 0.0671

Log likelihood -1366.728 -715.1577

Observations 3,196 3,196 1,279 1,279

Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1

Source: Author work

: - Note in the estimation (in Table 5), organizational innovation is the dependent variable. The

reference category in the case of sectoral industry dummies is the mass production goods—

science-based manufacturing (MPG-SB).

In Table 5, the results show that external knowledge gained from the customers in

the private sector positively and significantly affects firms´ likelihood of doing

organizational innovation in SMEs, thus supporting H1. The marginal effects show that

small firms and medium firms` using private customers as an external knowledge

source has an 8.6% and 14.3% higher probability of doing organizational innovation,

71


respectively. This indicates that the effect of using private sector customers as an

external knowledge source on organizational innovation is more pronounced in

medium-sized firms than small firms. This finding is in line with the study of Chesbrough

(2011) who shows that innovation in services is closely related to the customers.

Similarly, Tether (2005) argues that for organizational orientation in innovation

activities, service firms are more likely to collaborate with customers and suppliers.

Also, this finding is consistent with the study of Faems et al., (2005) who affirm that

customer collaboration positively associates with product innovation. Moreover,

external knowledge source from the customer in the public sector has no significant

effect in small firms’ ability to introduce organisational innovation, while in medium

firms, it has a negative significant effect on the introduction of organizational

innovation. Hence, H2 is not supported. The finding is in line with the study of Stuermer

et al. (2009), who find that when a firm relies on external sources of innovation; it could

bring unexpected costs associated with control and communication. Another

possible explanation could be that too much openness towards external information

sources of innovation might hamper the search efficiency. In a similar vein, Laursen

and Salter (2006) argue that firms that go beyond the optimum level of search

strategies and heavily rely on various external knowledge sources of innovation results

in a decline in innovation performance.

Furthermore, the small firms using competitors as external knowledge source have

a 7.6% higher probability of doing organizational innovation, this validates H4.

Leiponen (2005) finds the same for Finnish business services firms in case of new service

introductions. Similarly, Hipp (2000) states that using competitors as external

knowledge source enhances new ideas for innovation in knowledge-intensive

business services (KIBS). However, the present study does not find any significant

relationship between organizational innovation and competitors as an external

knowledge source in the case of medium-sized firms.

In addition, knowledge sourced from external conferences relates significantly

positive to organizational innovation in small firms and no effect is observed for the

medium-sized firms, thus, H5 is confirmed for small firms only. Marginal effects show that

the small firms sourcing knowledge from external conferences have a 3.8 % higher

probability of introducing an organizational innovation than the firms who do not

consider conferences as a knowledge source.

Further, knowledge sources from crowdsourcing relate significantly positive with the

adoption of organizational innovation in small firms only, hence supporting H3 for small

firms only. This suggests that small firms are more likely to gain from sourced information

emanating from crowdsourcing for the introduction of organizational innovation at

the workplace.

Additionally, external knowledge sources from the industry association show a

positive significant association with the adoption of new methods of organizing work

responsibility and decision making in medium-sized firms. Consequently, H6 is

supported for medium-sized firms and not for small firms. This finding indicates that to

achieve organizational innovation performance, medium-sized firms could leverage

on association with other firms and institution. Love and Mansury (2007) also find that

external knowledge sourced from the strategic alliances enhances firms’ innovation

performance, specifically in terms of the introduction of new services. Further, Pullen

et al. (2012) also show that a consistent, closed, and focused network strategy

enhance firms’ innovation performance.

Concerning control variables, the findings show that employees with higher

education have a significant effect on organizational innovation in small firms, in line

72


with the study of Mol and Birkinshaw (2009). Furthermore, the small firms targeting the

national markets (National_market) have a 4.2% higher probability of introducing

organizational innovation than the small firms not targeting the national market. R&D

intensity shows a negatively significant impact on the introduction of new methods of

organizational innovation in small firms. It could be argued that the dearth of resources

by small firms could make it challenging for them to invest in R&D to support

organizational innovation. Similarly, Spithoven et al. (2013) analyse the open

innovation practices in SMEs and find that R&D intensity (as a control variable) has no

significant effect on the introduction of new product/service development, which is

in line with the present study in the case of medium-sized firms.

Lastly, the industries dummies are included as a control since firms’ specific

characteristics could have a significant effect on innovation. In small firms, the

knowledge-intensive business services (AKP-S) have a 4.1 % higher likelihood of the

introduction of organizational innovation than small firms active in mass production

goods—science-based manufacturing (MPG-SB). Also, in small firms, supporting

infrastructure service industries (SIS-P) have a 4.3 % higher likelihood of the introduction

of organizational innovation than small firms active in mass production goods-science-

based manufacturing (MPG-SB). However, there is no significant effect of industry

dummies in medium-sized firms.

Robustness check To ascertain the stability of the main results in Table 5, a robustness estimation is

performed by using a logit model where the dependent variable is digital innovation

supply chain management (Diginnospmana). The robustness results illustrate that the

findings are robust to using a digital innovation supply chain management

“Diginnospmana” as a dependent variable. The coefficients of the estimation

variables almost retain their sign and significance in the robustness regression. The

results are available upon request from the author.

Discussion Over the last years, the innovation concept emerges differently from a technical

method to a broader perspective for innovation activities including organizational

innovation. Notably, in existing literature, scholars have less focused on the different

knowledge sources as a determinant of such type of innovation. Therefore, this study

investigates the impact of various external knowledge sources (including customers

from the public and private sector, competitors, conferences, crowdsourcing and

industry associations) on organizational innovation especially in terms of the adoption

of new methods of work organizing responsibility and decision making at firms’

workplace.

The findings of this present study lead to the conclusion that source of knowledge from

the customers in the private sector, and crowdsourcing have a significant positive

effect on organizational innovation in small firms (see Table 5). This indicates that

interaction with customers from the private sector and communication with

crowdsourcing play an important role in the introduction of organizational innovation

in small firms. As the customers and crowdsourcing ideas are relevant to the firms for

the adoption of new methods in workplace organization. So, their necessities and

desires provide useful knowledge that gives support to the firms to innovate in work

organizing tasks as well as in decision making. Further, small firms gaining knowledge

from competitors have a higher probability of innovating new organizational methods.

73


This suggests that small firm seek information through competitors for the introduction

of organizational innovation because competitors also face similar challenges that

are linked to cultural and organizational issues. In addition, it also shows that firm

interaction and communication with the competitors might support firms to obtain

new knowledge that might other firms don’t possess for the adoption of new methods

of organizational innovation.

Next, conferences as an external information source have a significant positive

influence on the performance of organizational innovation at the firm workplace in

small firms only. It indicates that small firms generally do not have sufficient resources

to conduct their R&D for the improvement of organizational level. So, small firms take

advantage of conferences and trade fairs to obtain new ideas for the development

of organizational innovation and improvement in decision-making strategy.

For medium enterprises, the effect of external knowledge source from the private

sector has a highly significant impact on the introduction of organizational innovation

at the firm workplace. Based on the marginal effect, medium enterprises relatively

take more advantage of the use of the information sources through the customer in

the private sector than small firms. This suggests that such a knowledge source is useful

in enabling them to implement the organizational innovation. Concerning that

knowledge obtained through the customer in the public sector has a negative

significant effect on the organizational innovation in medium-sized firms; this in effect,

indicates that the firms might not require this form of knowledge to support their

organizational performance. Another plausible explanation could be the dearth of

resources at the disposal of the firm. So, management would have to re-examine such

knowledge adoption and integration in their performance strategy.

Furthermore, the medium-sized firms gaining knowledge from industry associations

have a higher probability of innovating new organizational methods. The findings

suggest that knowledge sourced from the industry associations encourages firms to

innovate for new methods of organizing work responsibilities as well as improve

decision making skills. Similarly, the business group could provide scientific networks

with foreign firms in progressive markets and increase the knowledge sharing process

due to their close linkages and knowledge exchange or internal personnel transfer. In

addition, a business group comprises a corporation that contains officially

autonomous firms that are related to other firms through operating in a common

administrative network and financial management (Khanna and Rivkin, 2001).

Conclusion Overall results of the present study illustrate that the utilization of external sources of

information is a very important aspect that allows firms to adopt organizational

innovation at the firm workplace. In general, we can draw a conclusion that through

interaction and communication with the external partners (i.e. customers from the

private sector, competitors, conferences and crowdsourcing and industry

associations) firms can increase their knowledge and reduce the uncertainty about

the complex environment in which they operate. Also, the closest association with the

informal external knowledge sources supports the organizational innovation at the firm

workplace. We can conclude that based on our estimation, external knowledge

sources play an essential role in the introduction of organizational innovation in small

firms, but not undermining their relevance to medium-sized firms. In effect, small firms

are more likely to use various sets of external knowledge sources for the introduction

of organizational innovation than medium firms.

74


Policy implications for management A few practical implications are inferred from this study that is beneficial to managers

and the management team as they decide on the suitability of knowledge mix to

enhance performance. The decisions of managers and management can mar and

at the same time spur the success of a firm. External source of knowledge could

stimulate the adoption of new management practices in small and medium

enterprise. The present study provides consistent results with the external knowledge

source literature on technological innovation, which claims that new ideas and skills

for innovation usually come from outside of the working environment of firms.

Concerning the identification of the various sets of external partners available to

share their information and knowledge freely may decrease costs that are linked with

formal forms of collaboration to enhance organizational innovation in the workplace.

The closest association with the informal external knowledge sources may support the

internal and external organizational innovation at the firm level. The results of our

findings suggest that managers should establish a better connection with the

customers, competitors, industry association, crowdsourcing and trade fair exhibitions

to benefit from the mix of external information for the potential development of

organizational innovation.

Limitation and future research This study has some limitations that further research with available dataset could

address. This study uses a cross-sectional dataset and reveals external sources of

knowledge as an essential determinant of organizational innovation performance in

small and medium-sized firms, respectively. As a result, this study could not estimate

the long-term impact of external knowledge sources on organizational innovation.

However, it does not becloud the relevant findings that are quintessential to the

internal performance of firms in the way things are organized. As it is difficult to observe

the causality problems in a cross-sectional dataset, future research might consider the

causality issues when the appropriate data is available. Further, future research might

provide a cross-country comparison with a longer period. Finally, for future research,

the following might be considered; how various external knowledge sources support

the enterprise strategies or goals (e.g. reducing in house costs of operation, reducing

costs of purchased material or services and increase the quality of existing goods and

services).

References 1. Abdul Basit, S., Kuhn, T., Ahmed, M. (2018), “The Effect of Government Subsidy on Non-

Technological Innovation and Firm Performance in the Service Sector: Evidence from

Germany”, Business Systems Research, Vol. 9 No. 1, pp. 118-137.

2. Al‐Laham, A., Amburgey, T. L., Baden‐Fuller, C. (2010), “Who is my partner and how do

we dance? Technological collaboration and patenting speed in US biotechnology”,

British Journal of Management, Vol. 21 No. 3, pp. 789-807.

3. Anzola-Román, P., Bayona-Sáez, C., García-Marco, T. (2018), “Organizational

innovation, internal R&D and externally sourced innovation practices: Effects on

technological innovation outcomes”, Journal of Business Research, Vol. 91, pp. 233-247.

4. Bach, M. P., Lojpur, A., Peković, S., Stanovčić, T. (2015), “The influence of different

information sources on innovation performance: Evidence from France, The

Netherlands and Croatia”,. South East European Journal of Economics and Business, Vol.

10 No. 2, pp. 89-101.

75


5. Birkinshaw, J., Hamel, G., Mol, M. J. (2008), “Management innovation”, Academy of

management Review, Vol. 33 No.4, pp. 825-845.

6. Bogers, M., Zobel, A.-K., Afuah, A., Almirall, E., Brunswicker, S., Dahlander, L., Frederiksen,

L., Gawer, A., Gruber, M., Haefliger, S. (2017), “The open innovation research landscape:

established perspectives and emerging themes across different levels of analysis”,

Industry and Innovation, Vol. 24 No.1, pp. 8-40.

7. Brettel, M., Cleven, N. J. (2011), “Innovation culture, collaboration with external partners

and NPD performance”, Creativity and Innovation Management, Vol.20 No.4, pp. 253-

272.

8. Brunswicker, S., Vanhaverbeke, W. (2015), “Open innovation in small and medium‐sized

enterprises (SMEs): External knowledge sourcing strategies and internal organizational

facilitators”, Journal of Small Business Management, Vol. 53 No.4, pp. 1241-1263.

9. Castellacci, F. (2008), “Technological paradigms, regimes and trajectories:

Manufacturing and service industries in a new taxonomy of sectoral patterns of

innovation”, Research Policy, Vol. 37 No.6-7, pp. 978-994.

10. Chang, S. J., Chung, C. N., Mahmood, I. P. (2006), “When and how does business group

affiliation promote firm innovation? A tale of two emerging economies”, Organization

Science, Vol.17 No. 5, pp. 637–656.

11. Chang, W., Taylor, S. A. (2016), “The effectiveness of customer participation in new

product development: A meta-analysis”, Journal of Marketing, Vol. 80 No.1, pp. 47-64.

12. Chen, Q., Wang, C. H., Huang, S. Z. (2019), “Effects of organizational innovation and

technological innovation capabilities on firm performance: evidence from firms in

China’s Pearl River Delta”, Asia Pacific Business Review, Vol. 26 No.1, pp. 1-25.

13. Chesbrough, H. (2003), “Open Innovation: The New Imperative for Creating and Profiting

From Technology”, Boston, MA, Harvard Business School Press.

14. Chesbrough, H. (2011), “Open services innovation: Rethinking your business to grow and

compete in a new era”, San Francisco, CA, Jossey Bass A Wiley Imprint.

15. Chesbrough, H., Vanhaverbeke, W., West, J., Eds., (2006), “Open innovation:

Researching a new paradigm”, New York, Oxford University Press.

16. Chesbrough, H., Vanhaverbeke, W., West, J., (2014), “New frontiers in open innovation”,

Oxford, Oxford University Press.

17. Choi, S. B., Lee, S. H., Williams, C. (2011), “Ownership and firm innovation in a transition

economy: Evidence from China”, Research Policy, Vol. 40No.3, pp. 441–452.

18. Cohen, W. M., Levinthal, D. A. (1990), “Absorptive capacity: A new perspective on

learning and innovation”, Administrative science quarterly, Vol. 35 No.1, pp. 128-152.

19. Criscuolo, P., Laursen, K., Reichstein, T., Salter, A. (2018), “Winning combinations: search

strategies and innovativeness in the UK”, Industry and Innovation, Vol. 25No.2, pp. 115-

143.

20. Damanpour, F. (2014), “Footnotes to research on management

innovation”,. Organization Studies, Vol. 35 No. 9, pp. 1265-1285.

21. Damanpour, F., Sanchez‐Henriquez, F., Chiu, H. H. (2018), “Internal and external sources

and the adoption of innovations in organizations”, British Journal of Management, Vol.

29 No.4, pp. 712-730.

22. Dotzel, K. R., Faggian, A. (2019), “The impact of external knowledge sourcing on

innovation outcomes in rural and urban businesses in the US”, Growth and Change, Vol.

50 No.2, pp. 515-547.

23. Faems, D., Van Looy, B., Debackere, K. (2005), “Interorganizational collaboration and

innovation: Toward a portfolio approach”, Journal of product innovation

management, Vol. 22 No.3, pp. 238-250.

24. Ferraris, A., Santoro, G., Dezi, L. (2017), “How MNC’s subsidiaries may improve their

innovative performance? The role of external sources and knowledge management

capabilities”, Journal of Knowledge Management, Vol. 21 No. 3, pp. 540-552.

25. Fitjar, R. D., Rodríguez-Pose, A. (2013), “Firm collaboration and modes of innovation in

Norway”, Research Policy, Vol. 42 No.1, pp. 128-138.

76


26. Gassmann, O., Enkel, E. (2004), “Towards a theory of open innovation: three core

process archetypes”, In Proceedings of the R&D Management Conference, Lisbon, July

6–9, RADMA, Portugal, pp.1-18.

27. Gassmann, O., Enkel, E., Chesbrough, H. (2010), “The future of open innovation”, R&D

Management, Vol.40 No.3, pp. 213-221.

28. Gellynck, X., Vermeire, B. (2009), “The contribution of regional networks to innovation

and challenges for regional policy”, International Journal of Urban and Regional

Research, Vol. 33 No.3, pp. 719-737.

29. Gnyawali, D.R., Park, B.J., (2011), “Co-opetition between giants: collaboration with

competitors for technological innovation”, Research Policy. Vol. 40 No.5, pp. 650–663.

30. Gómez, J., Salazar, I., Vargas, P. (2016), “Sources of information as determinants of

product and process innovation”, PLoS One, Vol. 11No.4, e0152743.

31. Greenan, N., Mairesse, J. (2003), “How do new organizational practices shape

production jobs”, working paper No. 28, Centre d’études de l’emploi et du travail,

France, September 2003.

32. Gulerwe, I., Guillén, M. F., Macpherson, J. M. (2002), “Global competition, institutions,

and the diffusion of organizational practices: The international spread of ISO 9000 quality

certificates”, Administrative science quarterly, Vol. 47 No. 2, pp. 207-232.

33. Hair, J., Black, W. C., Babin, B. J., Anderson, R. E. (2010), “Multivariate data analysis”, (7th

ed.), Pearson Prentice Hall, available at: http://library.wur.nl/WebQuery/clc/1809603 (12

November 2019).

34. Hall, R. (1992), “The strategic analysis of intangible resources”, Strategic management

journal, Vol. 13 No.2, pp. 135-144.

35. Hervas-Oliver, J. L., Sempere-Ripoll, F., Boronat-Moll, C., Rojas, R. (2015), “Technological

innovation without R&D: unfolding the extra gains of management innovations on

technological performance”, Technology Analysis & Strategic Management, Vol. 27

No.1, pp. 19-38.

36. Hinteregger, C., Durst, S., Temel, S., Yesilay, R. B. (2019), “The impact of openness on

innovation in SMEs”, International Journal of Innovation Management, Vol. 23 No.01,

1950003.

37. Hipp, C. (2000), “Information flows and knowledge creation in knowledge-intensive

business services: scheme for a conceptualization”, in Metcalfe J.S., Miles I.

(Ed.), Innovation Systems in the Service Economy, Boston, MA, Springer, pp. 149-167.

38. Hippel, E.V. (1988), “The Sources of Innovation”, New York, Oxford University Press.

39. Hossain, M. (2013), “Open innovation: so far and a way forward”, World Journal of

Science, Technology and Sustainable Development, Vol.10 No.1, pp. 30-41.

40. Iglesias-Sánchez, P. P., Correia, M. B.,Jambrino-Maldonado, C. (2019), “Challenges of

Open Innovation in the Tourism Sector”, Tourism Planning & Development, Vol.16 No.1,

pp. 22-42.

41. Kang, K. H., Kang, J. (2009), “How do firms source external knowledge for innovation?

Analysing the effects of different knowledge sourcing methods”, International Journal

of Innovation Management, Vol.13 No.01, pp. 1-17.

42. Keupp, M. M., Palmié, M., Gassmann, O. (2012), “The strategic management of

innovation: A systematic review and paths for future research”, International Journal of

Management Reviews, Vol.14 No.4, pp. 367-390.

43. Khanna, T., Rivkin, J. W. (2001), “Estimating the performance effects of business groups

in emerging markets”, Strategic Management Journal, Vol. 22 No.1, pp. 45–74.

44. Knoben, J., Oerlemans, L. A. G. (2010), “The importance of external knowledge sources

for the newness of innovations of South African firms”, International Journal of Innovation

and Regional Development, Vol. 2 No.3, pp. 165-181.

45. Köhler, C., Sofka, W., Grimpe, C. (2012), “Selective search, sectoral patterns, and the

impact on product innovation performance”, Research Policy, Vol.41 No.8, pp. 1344-

1356.

http://library.wur.nl/WebQuery/clc/1809603

77


46. Laursen, K., Salter, A. (2006), “Open for innovation: the role of openness in explaining

innovation performance among UK manufacturing firms”, Strategic management

journal, Vol.27 No.2, pp. 131-150.

47. Lefebvre, V. M., De Steur, H., Gellynck, X. (2015), “External sources for innovation in food

SMEs”, British Food Journal, Vol. 117 No.1, pp. 412-430.

48. Leiponen, A. (2005), “Organization of knowledge and innovation: the case of Finnish

business services”, Industry & Innovation, Vol.12 No.2, pp. 185-203.

49. Leiponen, A., Helfat, C. E. (2011), “Location, decentralization, and knowledge sources

for innovation”, Organization Science, Vol. 22 No.3, pp. 641-658.

50. Lhuillery, S.,Pfister, E. (2009), “R&D cooperation and failures in innovation projects:

Empirical evidence from French CIS data”, Research policy, Vol.38 No.1, pp. 45-57.

51. Love, J. H., Mansury, M. A. (2007), “External linkages, R&D and innovation performance

in US business services”, Industry and Innovation, Vol.14 No.5, pp. 477-496.

52. Mol, M. J.,Birkinshaw, J. (2009), “The sources of management innovation: When firms

introduce new management practices”, Journal of business research, Vol. 62 No.12, pp.

1269-1280.

53. Mol, M. J.,Birkinshaw, J. (2014), “The role of external involvement in the creation of

management innovations”, Organization Studies, Vol. 35 No.9, pp. 1287-1312.

54. Mowery, D. C., Oxley, J. E., Silverman, B. S. (1996), “Strategic alliances and interfirm

knowledge transfer”, Strategic Management Journal, Vol.17 No. S2 - Winter Special

Issue, pp. 77–91.

55. Muller, P., Caliandro, C., Peycheva, V., Gagliardi, D., Marzocchi, C., Ramlogan, R., Cox,

D. (2015), “Annual report on European SMEs”, European Commission, available at:

https://op.europa.eu/en/publication-detail/-/publication/7c9fbfe0-e044-11e5-8fea-

01aa75ed71a1/language-en/format-PDF/source-208087183, (15 November 2019).

56. OECD. (2005), “The measurement of scientific and technological activities”, in Oslo

Manual, Guidelines for collecting and interpreting innovation data (3rd ed.), OECD

EUROSTAT, available at:

https://ec.europa.eu/eurostat/documents/3859598/5889925/OSLO-EN.PDF/60a5a2f5-

577a-4091-9e09-9fa9e741dcf1?version=1.0, (10 November 2019).

57. Papke, L. E., Wooldridge, J. M. (1996), “Econometric methods for fractional response

variables with an application to 401 (k) plan participation rates”, Journal of applied

econometrics, Vol. 11 No.6, pp. 619-632.

58. Pippel, G. (2014), “R&D cooperation for non-technological innovations”, Economics of

Innovation and New Technology, Vol. 23 No.7, pp. 611-630.

59. Pullen, A. J., de Weerd‐Nederhof, P. C., Groen, A. J., Fisscher, O. A. (2012), “Open

innovation in practice: goal complementarity and closed NPD networks to explain

differences in innovation performance for SMEs in the medical devices sector”, Journal

of product innovation management, Vol. 29 No.6, pp. 917-934.

60. Reed, R., DeFillippi, R. J. (1990), “Causal ambiguity, barriers to imitation, and sustainable

competitive advantage”, Academy of management review, Vol.15 No.1, pp. 88-102.

61. Santoro, G., Ferraris, A., Giacosa, E., Giovando, G. (2018a), “How SMEs Engage in Open

Innovation: a Survey”, Journal of Knowledge Economy, Vol.9 No.2, SI, pp. 561–574.

62. Santoro, G., Vrontis, D., Thrassou, A., Dezi, L. (2018b), “The internet of things: building a

knowledge management system for open innovation and knowledge management

capacity”, Technological Forecasting and Social Change, Vol.136, pp. 347-354.

63. Schumpeter, J. A. (1934), “The theory of economic development”, Cambridge, MA,

Harvard University Press,

64. Sofka, W., Grimpe, C. (2010), “Specialized search and innovation performance–

evidence across Europe”, R&D Management, Vol.40 No.3, pp. 310-323.

65. Spithoven, A., Vanhaverbeke, W. Roijakkers, N. (2013), “Open innovation practices in

SMEs and large enterprises”, Small Business Economics, Vol.41 No.3, pp. 537-562.

66. Stuermer, M., Spaeth, S., Von Krogh, G. (2009), “Extending private‐collective innovation:

a case study”, R&D Management, Vol.39 No.2, pp. 170-191.

78


67. Teece, D. J. (1986), “Profiting from technological innovation: Implications for integration,

collaboration, licensing and public policy”, Research policy, Vol.15 No.6, pp. 285-305.

68. Tether, B. S. (2002), “Who co-operates for innovation, and why: an empirical

analysis”, Research policy, Vol. 31No.6, pp. 947-967.

69. Tether, B. S. (2005), “Do services innovate (differently)? Insights from the European

innobarometer survey”, Industry & Innovation, Vol.12 No.2, pp. 153-184.

70. Tether, B. S., Tajar, A. (2008 a), “Beyond industry–university links: Sourcing knowledge for

innovation from consultants, private research organisations and the public science-

base”, Research Policy, Vol.37 No.6-7, pp. 1079-1095.

71. Tether, B. S., Tajar, A. (2008b), “The organisational-cooperation mode of innovation and

its prominence amongst European service firms”, Research policy, Vol. 37 No.4, pp. 720-

739.

72. Tidd, J., Bessant, J., Pavitt, K. (2005), “Managing Innovation: Integrating Technological,

Market, and Organization Change”, 3rd ed., Chichester, UK, John Wiley and Sons.

73. Tsai, K. H. (2009), “Collaborative networks and product innovation performance: Toward

a contingency perspective”, Research Policy, Vol.38 No.5, pp. 765-778.

74. Tsoukas, H. (1996), “The firm as a distributed knowledge system: A constructionist

approach”, Strategic management journal, Vol.17 No. S2, pp. 11-25.

75. Turulja, L., Bajgorić, N. (2018), “Knowing Means Existing: Organizational Learning

Dimensions and Knowledge Management Capability”, Business Systems Research, Vol.9

No.1, pp. 1-18.

76. Van de Vrande, V., De Jong, J. P., Vanhaverbeke, W., De Rochemont, M. (2009), “Open

innovation in SMEs: Trends, motives and management challenges”, Technovation, Vol.

29 No.6-7, pp. 423-437.

77. Volberda, H. W., Van Den Bosch, F. A., Mihalache, O. R. (2014), “Advancing

management innovation: Synthesizing processes, levels of analysis, and change

agents” Organization Studies, Vol.35 No.9, pp. 1245-1264.

78. Wang, X., Xu, M. (2018), “Examining the linkage among open innovation, customer

knowledge management and radical innovation: The multiple mediating effects of

organizational learning ability”, Baltic Journal of Management, Vol.13 No.3, pp. 368-389.

79. West, J., Bogers, M. (2014), “Leveraging external sources of innovation: a review of

research on open innovation”, Journal of Product Innovation Management, Vol.

31No.4, pp. 814-831.

80. Wynarczyk, P. (2013), “Open innovation in SMEs: a dynamic approach to modern

entrepreneurship in the twenty-first century”, Journal of Small Business and Enterprise

Development, Vol. 20 No.2, pp. 258-278.

81. Yiu, D. W., Bruton, G. D., Lu, Y. (2005), “Understanding business group performance in an

emerging economy: Acquiring resources and capabilities in order to prosper”, Journal

of Management Studies, Vol. 42 No.1, pp. 183–206.

82. Zouaghika, F., Sánchez, M., Martínez, M. G. (2018), “Did the global financial crisis impact

firms' innovation performance? The role of internal and external knowledge capabilities

in high- and low-tech industries”, Technological Forecasting and Social Change, Vol.

132, pp. 92-104.

79


About the author Shoaib Abdul Basit, M.Sc. is a PhD student at the Faculty of Economics and Business

Administration, TU Chemnitz, Germany. He participated in the 17th International

Joseph A. Schumpeter Society (ISS) conference on “Innovation, Catch-up, and

sustainable development in Seoul, South Korea. He has published papers in good

journals such as the European Journal of Innovation Management and Journal of

Innovation & Knowledge. His main research interests lie in the economics of

innovation, knowledge exploitation, firm strategy, innovation and market

environment. The author can be contacted at [email protected].

80


Gender Disparity in Students’ Choices of

Information Technology Majors

Yu Zhang

Mount St. Joseph University, Cincinnati, Ohio, United States

Tristen Gros, En Mao

Nicholls State University, Thibodaux, Louisiana, United States

Abstract

Background: The gender disparity in the Information Technology (IT) field has persisted

over the years. In 2018, only 27.2% of IT workers were women. Once hired, women face

more challenges, and they are leaving the field twice as fast as men are. The

misconception that women are weak in tech is one of the root causes of gender

disparity issues in IT. Objectives: We examine the gender disparity in students’ choices

of IT majors, as well as the decision process of Computer Information Systems (CIS)

graduates. Methods/Approach: We use the United States public universities’ student

data from 2010 to 2018. Both the Pooled and the Satterthwaite t-test are used to

investigate the gender disparity issue among the students. Results: Our results support

our hypothesis that female students are statistically less likely to choose CIS than their

male peers are. An additional analysis of students’ grades in CIS courses shows that

female students perform equally well as male students do. We did not find any

evidence that it takes longer for female students to get the CIS degree; however,

female students did change their majors more often. Conclusions: Female students

tend to avoid IT majors; they often think they may not do well in the courses; however,

such an assumption is not true. Our findings provide strategies for university and high

school administration to be more proactive in developing recruiting strategies to

attract and retain female CIS students.

Keywords: information technology; computer information systems; gender disparity;

information technology education

JEL classification: I23, I24, J16


Received: Jan 08 2020

Accepted: Nov 03 2020

Citation: Zhang, Y., Gros, T., Mao, E. (2021), “Gender Disparity in Students’ Choices of

Information Technology Majors”, Business Systems Research, Vol. 12, No. 1, pp. 80-95.


81


Introduction The gender disparity in the Information Technology (IT) field has persisted over the

years. In fact, it is getting worse. According to the U.S. Bureau of Labor Statistics, in

2018, based on over 3.3 million employed in IT-related fields, only 27.2% were women

(United States Department of Labor, 2019). Women are leaving the field twice as fast

as men (Mundy, 2017).

Once hired, women face more challenges. The key reasons for women leaving the

field have nothing to do with the nature of the job. Women enjoy tech jobs. The main

reasons include “workplace conditions, a lack of access to key creative roles, a sense

of feeling stalled in one’s career”, and “Undermining behavior from managers”

(Mundy, 2017). In general, women have not been nurtured in their career path in the

technology field.

While women have outnumbered men in college nowadays, the percent of

women in technology peaked at 37% in 1984 (Mundy, 2017). The issue is complicated.

From an early age, we associate tech toys with boys rather than girls. Observing local

area grade school robotics club photos posted on social media, the majority of the

members are boys. Most women became involved in technology post-high school;

therefore, they feel excluded in the hiring process when the interviewers and recruiters

made male-centric references (Schoenberger, 2018). A male Google employee

stated, “I’m simply stating that the distribution of preferences and abilities of men and

women differ in part due to biological causes and that these differences may explain

why we don’t see equal representation of women in tech and leadership” (Clifford,

2017). Such misconceptions persist in our society and are one of the root causes of the

gender disparity issues in IT.

Since 2014, some tech giants released data on women and minorities employed

and have been implementing strategies to improve the representation of women and

minorities. However, the changes are slow. To solve the issue of gender disparity,

companies have been using a variety of methods such as unconscious bias training,

objective skill assessment and standard questionnaires in hiring practices, and set

explicit hiring goals (e.g., Intel set a goal of 45% of new hires to be women and

minorities) (Mundy, 2017).

Prior research of the gender disparity in the IT field mainly used survey methods and

mostly examined the differences in computer attitudes (e.g., Young, 2000; Fedorowicz

et al., 2010; Carter, 2006; Hunsinger et al., 2009; Beyer, 2008). Although those studies

have found the attitudinal root causes of the gender disparity issue, they are

subjective and perception-based.

There is a very limited number of studies on the gender disparity issue in students’

actual choice of IT-related majors using objective assessments. To fill the gap, we

examined the gender differences in choosing the Computer Information System (CIS)

majors or minors and the gender difference in course performances using a U.S. public

university’s archival data from 2010 to 2018.

Our analysis studied graduation statistics – out of 10,254 graduates, 1.68% earned

their degree in CIS, 0.66% of those students being women. To get a closer insight into

the enrolment issue, the major decision process of CIS graduates in terms of the

number of times they took to decide on the CIS major is examined. The university

tracked each declared major during each graduate’s college career. We also

analyzed those who declared CIS as their major but later changed. Overall, 24% of

the CIS graduates from this university were female.

Moreover, we used a t-test to check if the percentages of female and male

students choosing CIS as a major or a minor are significantly different. Our results

support the hypothesis that female students are statistically less likely to choose CIS

82


than their male peers. Additional analysis of students’ grades on CIS courses shows

that female students perform just as well as male students. We didn’t find any

evidence that it takes longer for female students to get the CIS degree. However,

female students did change their majors more often.

Although the technology industry is rapidly growing, needing more people with IT

backgrounds, the number of students enrolled in this field has not significantly

increased over the years. It is now not only a problem where women are

underrepresented – It is a problem where more IT talents are needed. Female students

tend to avoid IT majors or think they may not do well in the courses. The findings of this

research will provide strategies for university and high school administrations to be

proactive in developing recruiting and development strategies to attract and retain

female students in the IT field. The results will also inspire more students, women

especially, to consider information technology as a career option.

This paper contributes to the literature on gender disparity in the IT field in the

following ways.

First, this study adds to the existing gender disparity in IT literature. Gender disparities

are an often-cited concern of the IT workforce (DiSabatino, 2000; Patel and

Parmentier, 2005; Trauth et al., 2008; Langer et al., 2020). It has been documented that

working women face severe challenges in the IT workforce. For example, there is a

pay gap in hourly compensation of 22% in favor of men. Moreover, women are most

underrepresented in the IT occupations (DiSabatino, 2000). Women’s participation is

still based on a continuation of traditional gender roles, which places women on the

periphery of an IT organization (Patel and Parmentier, 2005). Cultural attitudes about

maternity, childcare, parental care, and working outside the home seriously affect a

woman’s choice of an IT career (Trauth et al., 2008). Women realize less benefit from

performance gains than men and less benefit from tenure within IT firms (Langer et al.,

2020).

Our paper differs from the studies mentioned above in that it shows challenges for

women are not only in the workplace but also in education. We find that there’re

significantly fewer female students choosing CIS as their majors, and female CIS

students seem to change their majors more frequently.

Second, the paper also contributes to the existing gender disparity in education

literature. Gender disparity has attracted considerable attention in today’s

educational research and practice. However, very few studies examine the gender

differences in students’ choices of CIS majors in higher education.

Early studies have revealed that gender discrimination in schools exists in the areas

of science and mathematics. Girls are not receiving the same quality or even quantity

of education as their male classmates (Emfinger, 2002; Tindall and Hamil, 2004). In

terms of IT, topics mainly focus on attitudes toward the usage of IT-related tools and

applications among students (Wong and Hanafi, 2007) or computer-related

competence (Basavaraja and Kumar, 2017). Studies find that not only female

students but also female instructors had lower confidence and less experience in the

use of computers (Zhou and Xu, 2007). Instead of testing attitudes or competence, we

investigate how students choose the CIS major and their performance in the major.

Third, previous research has found that developing countries have a monopoly on

gender inequality. It is particularly true in the area of IT (Akubue, 2001; Patel and

Parmentier, 2005; Geldof, 2011; Bhattacharyya and Ghosh, 2012).

In Indian, women’s participation failed to occur at the same speed as IT expansion

(Patel and Parmentier, 2005; Bhattacharyya and Ghosh, 2012). In Ethiopia and

Malawi, existing gender norms in terms of domestic responsibilities gave women less

time to interact with IT and restricted their mobility. Due to this limited time and

83


mobility, they had less exposure to IT beyond the vicinity of their homes (Geldof, 2011).

The power of the socialization process in inhibiting women's education in science,

engineering, mathematics, and technology education is often underestimated and

has not received the attention it deserves in Third World (Akubue, 2001).

However, relatively few studies examine the gender disparity in IT in developed

countries. This study tries to fill this research gap by examining a case in a typical

developed country, the United States. We show that even in a developed country,

gender disparity still exists and should not be neglected.

Finally, unlike most of the prior research on the gender disparity, which uses a survey

as the primary method of investigation (Wong and Hanafi, 2007; Zhou and Xu, 2007;

Johnson et al., 2008), this paper used archival data of a U.S. public university. The data

are more objective. To the best of our knowledge, it is the first paper to do such an

analysis.

The remainder of the paper proceeds as follows. The next section provides literature

review and develops our hypotheses. Section three presents the research methods.

Section four describes the results. Section five discusses the findings and concludes.

Literature review and hypotheses development Prior research documents that gender differences regarding IT can be found as early

as in middle schools and high schools, and the effects of such differences is long-

lasting throughout college and career advancement (e.g., Young, 2000; Fedorowicz

et al., 2010; Carter, 2006; Hunsinger et al., 2009; Malgwi et al.,2005; Beyer, 2008).

At an early age, girls in general have less confidence with technology and perceive

the field as male dominant. Young (2000) used 462 middle and high school students

as subjects to investigate gender differences in computer attitudes. Five aspects have

been examined: confidence, perception of computers as male domain, positive

teacher attitudes, negative teacher attitudes, and perceived usefulness of

computers. Survey results show that the main gender differences are greater

confidence among boys, and the perception of computers as a male domain is

prevalent among boys.

One of the causes of negative perceptions of the technology field could be the

amount of exposure and access to technology. Fedorowicz et al. (2010) surveyed

teenagers in middle and high schools. They found that there were differences in the

amount of time boys and girls spend using technology. Boys owned more technology

and had more access to technology in their homes. However, these differences did

not become pronounced until high school.

The girls’ lack of exposure to technology manifests even further in college. Several

studies investigated the factors that affect college students’ choices of their majors.

Carter (2006) reported on a study in which 836 high school calculus and pre-calculus

students were surveyed to try to determine why students did not pursue a major in

Computer Science. They found that female students have more reasons to reject a

Computer Science major because they have an incorrect or no perception of what

the field is. This was echoed in Hunsinger et al. (2009), where they interviewed and

surveyed female college students to better understand why they choose to major (or

not major) in CIS. They found that female students lacked knowledge about the CIS

major and believed that it was a challenging major.

Malgwi et al. (2005) found that besides interest, female students were more likely

influenced by the aptitude in the subject. However, male students were significantly

more influenced by the major’s potential for career advancement and job

opportunities and the level of compensation in the field.

84


Beyer (2008) distributed surveys to 159 Business majors enrolled in Management

Information Systems (MIS) classes at the University of Wisconsin-Parkside. She found

that female high school computer teachers and role models are very important for

female students to choose MIS as their majors.

Another reason that prevents female students from choosing IT-related majors is the

lack of confidence in their abilities, which is documented by Shashaani (1997), Beyer

et al. (2003), Lee (2003), and Lee and Huang (2014).

The gender disparity is apparent in choosing the IT major and career in young

women. Unfortunately, such difference has translated into job insecurity and other

work attitudes for women in IT fields. Truman et al. (1994) examined the extent to which

gender discrimination was a force affecting the senior managerial ranks of the

information systems (IS) occupation. They analyzed data gathered by the Society for

Information Management (SIM) and found that women received lower salaries than

men even when job level, age, education, and work experience are controlled. Using

the data from 159 African Americans and 98 Anglo Americans, Johnson et al. (2008)

found the ethnic and gender differences in IT fields. Specifically, Anglo American

women reported lower levels of IT self-efficacy than did members of all other groups.

Based on the discussion above, we see that the prior literature has depicted a clear

picture of the attitudinal root causes of the gender disparity issue in the IT field. The

perception of the field of being male dominant has prevented female students from

choosing to study and work in this field. The lack of exposure to technology by girls

from a young age has resulted in an erroneous understanding of the field and, in

general, less confidence in their ability in technology. The prior gender disparity

literature, however, is mainly subjective and perception-based. There is a very limited

number of studies on the gender disparity issue in students’ actual choice of IT-related

majors using objective assessments. Therefore, we advance our first hypothesis:

H1: There are significantly fewer female graduates than male graduates in the CIS

major.

Students take minors that appeal to their personal interests and motivate them the

most (STOCK, P., and STOCK, E., 2018). As we discussed previously, women are less

confident in their abilities to use computers. Moreover, parents tend to emphasize the

importance of math, physics, and computer science for their sons and literature and

reading for their daughters (Shashaani and Khalili, 2001). To provide further evidence

on the gender disparity in students’ choices of CIS, we hypothesize that:

H2: There are significantly fewer female graduates than male graduates in the CIS

minor.

Gender has been documented as a factor that affects students’ academic

performance (Kruck and Lending, 2003; Nyikahadzoi et al., 2013; John et al., 2018). In

the study of Nyikahadzoi et al. (2013), Male students seem to have a better chance

of achieving higher grades. However, data provided by John et al. (2018) do not show

that females underperform relative to their male counterparts. Since prior research

provided mixed results, we make the following hypothesis:

H3: There’s no significant difference between male and female CIS students’

academic performance.

Most students do not receive enough assistance or advice in their decision of an

academic major. This might explain why the majority of college students change their

major at least once during their college years (STOCK, P., and STOCK, E., 2018). Allen,

H. (1973) Found that the major reasons given by the seniors for changing their majors

are as follows: (1) had a change of interest, (2) had greater success in another field,

(3) discovered he had unrealistic goals in terms of ability, and (4) felt he had received

85


inadequate counseling. There’s no evidence showing that CIS particularly affect

students to change their majors. Therefore, we have the following hypothesis:

H4: There’s no significant difference in the frequency of changing majors between

CIS and other major graduates.

Changing an academic major can add one or more semesters to the student’s

total time in college. Besides additional time in college and delayed graduation, a

change of major incurs additional cost for tuition and fees. No studies have found any

gender differences in the frequency of changing majors. We hypothesize that:

H5: There’s no significant difference between male and female CIS students in terms

of the frequency of changing majors.

College students who select an academic major matching their interests are more

likely to finish their degree plans on time. However, most students are not choosing

academic majors that match their interests or skills (Sheehy, 2013). These arguments

lead to the following hypothesis:

H6: There’s no significant difference in the length of time to complete a degree

between CIS and other majors.

Sheridan and Pyke (1994) used a multiple regression procedure to predict the time

taken to complete a degree. Selected demographic (e.g., sex, age, marital status,

registration status, citizenship), academic (e.g., GPA, discipline, type of program), and

financial support were used as independent variables. Results indicate that full-time

registration, increased financial support, and a higher GPA significantly decrease time

to completion. In their study, sex is not a significant factor that affects the time to

complete a degree. Thus our last hypothesis is as follows:

H7: There’s no significant difference in the length of time to complete the degree

between male and female CIS students.

Methodology For this study, we utilized two research methods.

First, we analyzed student and course data ranged from 2010 to 2018 from a public

university in the United States. This university is located in the South, with over 6,000

students enrolled. For our analyses, we requested gender, age, enrollment date,

graduation date, graduating major, past majors, minors, department, GPA, and ACT

scores from the university.

We used the T-test to examine if there are any significant differences between the

male and female student groups for all the hypotheses. Both Pooled and Satterthwaite

methods were used to address the issue of equality of variance in standard deviation

and unequal sample sizes for the t-test.

As our second study method, a combination of CIS graduates in the workforce and

current students were interviewed about their opinions and experiences with the

gender disparity issue in IT. This method was primarily used to shed light on some

possible solutions towards the gender disparity issue. The interviews also revealed

attitudes toward this issue from both genders.

Results Table 1 reports the summary statistics of our data. During the sample period of 2010 –

2018, 10,254 students graduated. The average age of graduates is 25.70. The average

length to complete a degree is 5.4 years, while the median is 4 years. CIS major is

coded to be a binary variable with a value of 1 if the student is a CIS major and 0

otherwise. The data shows that out of 10,254 graduates, 1.68% of students graduated

with a CIS degree. Gender is also a binary variable, which is coded as 1 for female

86


and 0 for male. About 62% of the graduates are female and only 38% are male. On

average, the graduates changed their majors 1.73 times before they got their

degrees. The average college GPA, high school GPA and ACT score for all the

graduates are 2.97, 3.32, and 21.72, respectively.

Table 1

Descriptive Statistics

Variable N Mean Median Std Dev Lower

Quartile

Upper

Quartile

Age 10254 25.70 23.00 6.22 22.00 26.00

Years to complete a

degree

10254 5.40 4.00 4.61 3.00 5.00

CIS major 10254 0.02 0.00 0.13 0.00 0.00

Gender 10254 0.62 1.00 0.48 0 1.00

Frequency of changing

majors

10254 1.73 1.00 1.53 1.00 2.00

College GPA 10254 2.97 2.98 0.51 2.60 3.35

High school GPA 7560 3.32 3.37 1.53 2.98 3.69

ACT score 8277 21.72 21.00 3.46 20.00 24.00

Note: Age is calculated as graduation year minus birth year. Years to complete a degree is

calculated as graduation year minus first year.

Figures 1 and 2 depict the number of female and male graduates/ CIS graduates per

year from 2010-2018. We can see that over the years, there were more female

graduates than male graduates. However, the situation is just the opposite for the CIS

major, in which there have been more male graduates than females. Female CIS

graduates are always less than 10 for the years under investigation.

Figure 1

Number of Female and Male Graduates Per Year

87


Figure 2

Number of Female and Male CIS Graduates Per Year

Table 2 shows the results of our H1 test. From 2010 to 2018, the university had 6,400

female graduates and 3,854 male graduates. Among those, only 0.0066 or 0.66% of

the female students chose CIS as their majors, while 0.0337 or 3.37 % of the male

students are CIS majors. The difference is -0.0271 or -2.71%. Female students are 2.71%

less likely to choose CIS as their majors compared to their male peers. Both the Pooled

and Satterthwaite methods of the t-test show that this difference is statistically different

at the 1% level (P <.0001). Therefore, our H1 is supported by the results.

Table 2

The t-test of Students Choosing the CIS Major

Variable: CIS major

Gender N Mean Std Dev Std Err

Female 6400 0.0066 0.0807 0.00101

Male 3854 0.0337 0.1806 0.00291

Diff (Female - Male) -0.0271 0.1278 0.0026

Method Variances df t-value Pr > |t|

Pooled Equal 10252 -10.43 <.0001***

Satterthwaite Unequal 4795.1 -8.82 <.0001***

Note: CIS major is a binary variable with a value of 1 if the student is a CIS major and

0 otherwise.

Table 3 presents the results of the H2 test. The means show that there are 0.16% of

the female students chose CIS as their minors, while 0.21% of the male students did so.

We can see that the male students are still more likely than the female students to

choose the CIS minor with a 0.05% higher rate. However, this difference is not

statistically significant. Both Pooled and Satterthwaite methods of the t-test don’t show

a significant P-value. This might be due to the fact that the sample size is very small for

the CIS minor data. Only 18 observations were pulled up from the university’s

databases.

88


Table 3

The t-test of Students Choosing the CIS Minor

Variable: CIS Minor


Female 6400 0.0016 0.0395 0.000494

Male 3854 0.0021 0.0455 0.000733

Diff (Female - Male) -0.0005 0.0419 0.000854


Pooled Equal 10252 -0.6 0.5476

Satterthwaite Unequal 7243 -0.58 0.5615

Note: CIS minor is a binary variable with a value of 1 if the student is a CIS minor and 0 otherwise.

There seems to be a misconception that female students don’t do well in IT-related

courses compared to their male counterparts. Table 4 shows the results of Hypothesis

3. We collected the course grades for all the CIS courses up to the year 2018. There

are 5,219 course grade observations for female students and 7,593 course grade

observations for male students. The average CIS grade is 2.78 and 2.81 for females

and males, respectively. The grade is 0.03 lower for female students; however, this

difference is not statistically significant based on the Pooled and Satterthwaite

methods’ P-Values. Both P-Values exceed 0.1. In addition, we tested the final GPA of

the CIS graduates and found no difference between male (2.92) and female (2.98)

students. The results are shown in Table 5. Thus, H3 is supported.

Table 4

The t-test of CIS Course Grades

Variable: Grade


Female 5219 2.78 1.053 0.015

Male 7593 2.81 1.043 0.012

Diff (Female - Male) -0.03 1.047 0.019


Pooled Equal 12810 -1.61 0.1064

Satterthwaite Unequal 11145 -1.61 0.107

Note: Grade is a numerical variable. It is coded as 4 for A; 3 for B; 2 for C; 1 for D and

0 for F.

Table 5

The t-test of College GPA between Male and Female CIS Graduates

Variable: College GPA


Female 42 2.98 0.5536 0.0854

Male 130 2.92 0.5008 0.0439

Diff (Female - Male) 0.06 0.5140 0.0912


Pooled Equal 170 0.57 0.5690

Satterthwaite Unequal 64.121 0.54 0.5897

To investigate the patterns of how students choose majors, we counted the number

of times they switched majors before they finally completed the degrees. We first

compared the switching patterns between CIS and other major graduates. The results

are shown in Table 6. There are 172 CIS graduates and 10,082 other graduates during

our sample period. On average, CIS graduates switched their majors 1.1221 times

while other graduates switched 1.7395 times before they completed their degrees.

89


The difference shows that CIS graduates are 0.6174 times less likely to change their

majors, compared to other graduates. Both Pooled and Satterthwaite methods show

that this difference is statistically significant at 1% level (P-Value<.0001). The results do

not support H4 that there’s no difference between CIS and other major graduates in

terms of how they make their choices. There is a significant difference. CIS graduates

are much less likely to change their majors once they made their decisions.

Table 6

The t-test of Changing Majors between CIS and Other Major Graduates

Variable: frequency of changing majors

Majors N Mean Std Dev Std Err

Other majors 10082 1.7395 1.5301 0.0152

CIS 172 1.1221 1.0663 0.0813

Diff (Other majors -CIS) 0.6174 1.5235 0.1172


Pooled Equal 10252 5.27 <.0001

Satterthwaite Unequal 183.22 7.46 <.0001

We then compared male and female CIS graduates’ switching patterns. The results

are tabulated in Table 7. As we discussed previously, overall, 62% of the graduates are

female and 38% are male. It is just the opposite for the CIS major. Out of 172 students

that graduated with a CIS degree, 24% were female, and 76% were male. Over the

eight years, only 42 females graduated with a CIS degree.

In Table 7, the mean of the frequency of changing majors for females is 1.381, which

means female CIS graduates changed their majors 1.381 times before they

graduated. Male CIS graduates only changed 1.0385 times on average. The

difference is 0.3425, which is marginally significant at the 10% level (P-value <0.1).

Although it seems that female CIS graduates are more likely to change their majors

than their male counterparts, if we compare the number with other majors’ in table 6,

female CIS graduates are still 0.3585 (1.7395-1.381) times less.

Overall, CIS students are more stable in their major decision, but the female CIS

students may still have some problems in determining their majors. Interestingly, we

noticed that for non-CIS majors, the male graduates changed their majors more

frequently than females.

Table 7

The t-test of Changing Majors between Male and Female CIS Graduates

Variable: frequency of changing majors


Female 42 1.381 1.1884 0.1834

Male 130 1.0385 1.0147 0.089

Diff (Female - Male) 0.3425 1.0592 0.188


Pooled Equal 170 1.82 0.0702


We also investigated how long it takes the graduates to complete their degrees.

Same as the frequency of changing major tests, we first compared the CIS and other

major graduates and then compared the male and female students within the CIS

major. Results are demonstrated in tables 8 and 9. It takes CIS graduates 5.1977 years

while other majors require, on average, 5.4044 years. The difference is 0.2067 years,

which is not statistically significant. The results support our H6 that CIS students spend

90


as much time completing their degrees as others. Table 9 shows the results of a further

comparison between male and female CIS students. The females completed the

degree 0.3875 years faster than males. However, the difference is not statistically

significant (P-value > 0.1). Thus, H7 holds.

Table 8

The t-test of Length of Time to Complete A Degree between CIS and Other Major

Graduates

Variable: length of time to complete a degree

Majors N Mean Std Dev Std Err

Other majors 10082 5.4044 4.6236 0.046

CIS 172 5.1977 4.0604 0.3096

Diff (Other majors -CIS) 0.2067 4.6148 0.3549


Pooled Equal 10252 0.58 0.5602


Table 9

The t-test of Length of Time to Complete A Degree between Male and Female CIS

Graduates

Variable: length of time to complete a degree


Female 42 4.9048 4.4051 0.6797

Male 130 5.2923 3.956 0.347

Diff (Female - Male) -0.3875 4.0689 0.7222


Pooled Equal 170 -0.54 0.5922

Satterthwaite Unequal 63.773 -0.51 0.6133

We also interviewed both CIS graduates in the workforce and current students

about their opinions and experiences with the gender disparity issue in IT. The

responses are listed in Appendix B. The interviews revealed attitudes and some

possible solutions towards the gender disparity issue. The female students who are not

enrolled in CIS but are within the College of Business stated that they enjoyed the CIS

courses that they have taken, but that they are not confident that they could succeed

in this field. Interviewed students expressed that they have noticed the gender

disparity issue within the CIS department. When asked if they have experienced any

sexism or disrespect, 3 out of 5 female students said that they had experienced this

within the CIS department. A CIS graduate that is now in the workforce admitted that

she had experienced sexism and disrespect within the workplace. She also mentioned

that a big reason why women aren’t in IT is likely due to cultural norms and societal

expectations. The most common reasons that prevent females from choosing CIS

mentioned by the interviewees are self and societal expectations. Women tend to

think they can’t handle the subject, and their credibility is questioned.

Discussion and conclusion This paper examines the gender disparity in students’ choices of information

technology majors. We used a U.S. public university’s archival data from 2010 to 2018

to test our hypotheses. Our findings show that there are significantly fewer female

graduates than male graduates in the CIS major. However, we failed to find similar

91


results in the minor test. Contrary to popular belief, our results revealed that female

students are not weak in IT-related courses. Even though the female students did as

well as the male students, we found that females may still have some problems in

determining their majors. Specifically, female CIS students changed their majors more

frequently than males. In terms of years to complete the CIS degree, we didn’t find

any significant differences between male and female students.

This study makes the following contributions. First, it adds to the prior literature that

examines the gender disparity issue in IT. Specifically, we conduct an inter-temporal

analysis to determine whether female students are less likely to choose CIS majors or

minors. The analysis provides direct evidence to show the gender disparity. Second,

our study also contributes to the CIS education literature by giving another example.

Third, the gender difference in IT has been studied in-depth in developing countries.

However, relatively few studies examine the issue in developed countries. This study

fills the gap. Finally, we didn’t use survey methods, which have been extensively used

in this type of research. Instead, we used archival data. The data are more objective.

Although the gender disparity issue is being recognized at the national level, little

attention is being paid to the issue at the university level. Many students are

uncomfortable discussing the issue. University administrations seem to have not looked

into this, as there are few programs to encourage gender inclusion. Our research

shows that female and male CIS students are equal in their ability and performance,

even though many people believe that men in this field are more capable than

women. Female students tend to avoid IT majors or think they may not do well in the

courses - there is a general lack of confidence in their technical ability.

A Women in Tech club or a scholarship for female CIS majors would be a good

starting point. Setting an enrollment goal for female students would be another game-

changer. Many students don’t think of CIS as an option due to the lack of promotion

for the CIS department at the university level. This issue could be addressed by simply

promoting the CIS department in the same way that other departments in the College

of Business are promoted. The University of Washington produces double the national

average of graduates in technology. This is not an accident - this university holds

programs like Women’s Research Day to promote their department and get more

students involved. Overall, the solution has to be systemic starting early on in life -

raising awareness to local K-12 schools. Starting programs and mentorships to

encourage and include young girls in technology-oriented activities would allow girls

to be educated about the tech industry. By getting girls involved at a young age and

systematically improving female entry and retention in IT, we may see a future where

gender disparity in IT is simply in the past.

The limitation of our paper is that the study is based on one university only. Our

sample is relatively small and may not be a perfect representative of the population.

Future studies can include more universities. In addition, we didn’t identify other

factors, such as age, marital status, citizenship and financial support, that may affect

students’ choices and performance. Multiple regression/correlation analyses to

include other factors should be examined in future research. However, the results still

shed light on the fact that female students are under-represented in IT-related majors.

The education authorities need to find proactive ways of attracting and retaining

female students in technology.

92


References 1. Akubue, A. I. (2001), “Gender Disparity in Third World Technological, Social, and

Economic Development”. The Journal of Technology Studies, Vol. 27 No. 2, pp.1-10.

2. Emfinger, K. (2002). “Gender equity and education” in Aldridge, J., Goldman (Ed.),

Current issues and trends in education, Boston, Pearson, pp. 163-170.

3. Allen, H. (1973), “A comparison study of why college students make changes in their

majors”, University of the Pacific Theses and Dissertations. Available at:

https://scholarlycommons.pacific.edu/uop_etds/3030 / (20 October 2016).

4. Basavaraja, M. T., Kumar, B. S. (2017), “Gender disparities in the use of ICT: a survey of

students in urban schools”, Journal of Information Science Theory and Practice, Vol. 5

No. 4, pp. 39-48.

5. Beyer, S. (2008), “Gender Differences and Intra-Gender Differences amongst

Management Information Systems Students”, Journal of Information Systems Education,

Vol. 19 No. 3, pp. 301-310.

6. Beyer, S., Rynes, K., Perrault, J., Hay, K., Haller, S. (2003), “Gender differences in

computer science students”, ACM SIGCSE Bulletin, Vol. 35 No. 1, pp. 49-53.

7. Bhattacharyya, A., Ghosh, B. N. (2012), “Women in Indian information technology (IT)

sector: A sociological analysis”, IOSR Journal of Humanities and Social Science, Vol. 3

No. 6, pp. 45-52.

8. Carson, E. (2016), “In US, Getting Women into Tech Is Everyone's Business -- Study.”

Available at www.cnet.com/news/women-in-tech-girls-who-code-report-cracking-

the-gender-code-computer-science-majors/ (20 October 2016).

9. Carter, L. (2006), “Why students with an apparent aptitude for computer science don't

choose to major in computer science”, ACM SIGCSE Bulletin, Vol. 38 No. 1, pp. 27-31.

10. Cheng, R. (2015), “Women in Tech: The Numbers Don't Add Up.” CNET, CNET, 6 May

2015, Available at www.cnet.com/news/women-in-tech-the-numbers-dont-add-up/

(6 May 2015).

11. Clifford, C. (2017), “12 Highlights from the Viral Anti-Diversity Google Manifesto.”

Available at www.cnbc.com/2017/08/07/shocking-quotes-from-the-viral-google-

manifesto.html (9 August 2017).

12. DiSabatino, J. (2000), “Glass Ceiling for Women in IT Persists”, Computerworld, Vol. 34

No. 20, pp. 12.

13. Fedorowicz, J., Vilvovsky, S.G., Golibersuch, A.J. (2010), “Gender Differences in

Teenagers' Elective Use of Computer Technology”, CAIS, Vol. 27 No. 1, pp. 3.

14. Geldof, M. (2011), “Earphones are not for women: Gendered ICT use among youths in

Ethiopia and Malawi”, Information Technologies & International Development, Vol. 7

No. 4, pp. 69-80.

15. Hempel, J. (2018), “Melinda Gates Has a New Mission: Women in Tech” available at

www.wired.com/2016/09/melinda-gates-has-a-new-mission-women-in-tech/ (10

January 2018).

16. Hunsinger, D.S., Holt, A.E., Knight, M.B. (2009), “Factors Influencing Females Whether to

Become Computer Information Systems Majors”, Information Systems Education

Journal, Vol. 7 No. 1. Available at https://www.isedj.org/7/1/ISEDJ.7(1).Hunsinger.pdf /

(10 January 2018).

17. John, T. M., Badejo, J. A., Popoola, S. I., Omole, D. O., Odukoya, J. A., Ajayi, P. O.,

Atayero, A. A. (2018), “The role of gender on academic performance in STEM-related

disciplines: Data from a tertiary institution”, Data in brief, Vol. 18, pp. 360-374.

18. Johnson, R.D., Stone, D.L., Phillips, T.N. (2008), “Relations among ethnicity, gender,

beliefs, attitudes, and intention to pursue a career in information technology”, Journal

of Applied Social Psychology, Vol. 38 No. 4, pp. 999-1022.

19. Kruck, S. K. S., Lending, D. L. D. (2003), “Predicting Academic Performance in an

Introductory College Introductory College-Level IS Course Level IS Course”, Information

Technology, Learning, and Performance Journal, Vol. 21 No. 2, pp. 9-15.

20. Langer, N., Gopal, R. D., Bapna, R. (2020), “Onward and Upward? An Empirical

Investigation of Gender and Promotions in Information Technology Services”,

Information Systems Research, Vol. 31 No. 2., pp. 1-16.

93


21. Lee, A.C.K. (2003), “Undergraduate students' gender differences in IT skills and

attitudes”, Journal of computer assisted learning, Vol. 19 No. 4, pp. 488-500.

22. Lee, C.L., Huang, M.K. (2014), “The influence of computer literacy and computer

anxiety on computer self-efficacy: the moderating effect of gender”,

Cyberpsychology, Behavior, and Social Networking, Vol. 17 No. 3, pp. 172-180.

23. Malgwi, C.A., Howe, M.A., Burnaby, P.A. (2005), “Influences on students' choice of

college major”, Journal of Education for Business, Vol. 80 No. 5, pp. 275-282.

24. McKinsey & Company (2016), “Women in the Workplace 2016”, Available at

www.mckinsey.com/business-functions/organization/our-insights/women-in-the-

workplace-2016 / (September. 2016).

25. Mundy, L. (2017), “Why Is Silicon Valley So Awful to Women?” Available at

www.theatlantic.com/magazine/archive/2017/04/why-is-silicon-valley-so-awful-to-

women/517788/ (6 December 2017).

26. Nyikahadzoi, L., Matamande, W., Taderera, E., Mandimika, E. (2013), “Determinants of

Students' Academic Performance in Four Selected Accounting Courses at University of

Zimbabwe”, Research in Higher Education Journal, Vol. 21, pp.1-9.

27. Patel, R., Parmentier, M. J. C. (2005), “The persistence of traditional gender roles in the

information technology sector: A study of female engineers in India’, Information

Technologies & International Development, Vol. 2 No. 3, pp. 29-46.

28. Shashaani, L. (1997), “Gender differences in computer attitudes and use among

college students”, Journal of Educational Computing Research, Vol. 16 No. 1, pp. 37-

51.

29. Shashaani, L., Khalili, A.(2001), “Gender and Computers: Similarities and Differences in

Iranian College Students’ Attitudes toward Computers”, Computers and Educations,

Vol. 37 No. 3-4, pp. 363-375.

30. Sheehy, K. (2013). “Study: High school grads choosing wrong college majors”, US News

& World Report, pp. 281-99.

31. Sheridan, P. M., Pyke, S. W. (1994), “Predictors of time to completion of graduate

degrees”, Canadian Journal of Higher Education, Vol. 24 No. 2, pp. 68-88.

32. Stock, P., Stock, E. (2018), “Factors that Influence a College Student's Choice of an

Academic Major and Minor”, Vol. 9 No.1, pp. 56-78.

33. Tindall, T., Hamil, B. (2004), “Gender Disparity in Science Education: The Causes,

Consequences, and Solutions”, Education, Vol. 125 No. 2, pp. 282-295.

34. Trauth, E. M., Quesenberry, J. L., Huang, H. (2008), “A multicultural analysis of factors

influencing career choice for women in the information technology workforce”, Journal

of Global Information Management, Vol. 16 No. 4, pp. 1-23.

35. Truman, G.E., Baroudi, J.J. (1994), “Gender differences in the information systems

managerial ranks: An assessment of potential discriminatory practices”, MIS quarterly,

Vol. 18 No. 2, pp.129-142.

36. United States Department of Labor (2019), “Employed Persons by Detailed Industry, Sex,

Race, and Hispanic or Latino Ethnicity”, Available at www.bls.gov/cps/cpsaat18.htm /

Accessed (18 January 2019).

37. Wong, S. L., Anafi, A. (2007), “Gender differences in attitudes towards information

technology among Malaysian student teachers: A case study at University Putra

Malaysia”, Journal of Educational Technology & Society, Vol. 10 No. 2, pp. 158-169.

38. Young, B.J. (2000), “Gender differences in student attitudes toward computers”,

Journal of research on computing in education, Vol. 33 No. 2, pp. 204-216.

39. Zhou, G., Xu, J. (2007), “Adoption of Educational Technology: How Does Gender

Matter?”, International Journal of Teaching and learning in higher education, Vol. 19

No. 2, pp. 140-153.

94


About the authors Yu Zhang, Ph.D., is an Assistant Professor at the School of Business, Mount St. Joseph

University. She received her Ph.D. in Accounting from the University of Texas at

Arlington. Before coming to the United States, Dr. Zhang worked in a Chinese state-

owned company and a German family firm. Her main research interests are capital

market research in accounting, financial reporting, social and environmental

accounting. She has published in Review of Accounting and Finance, International

Journal of Accounting, Auditing and Performance Evaluation, and Journal of

Accounting and Taxation. The author can be contacted at [email protected].

Tristen Gros is a recent graduate of Nicholls State University with her Bachelor of

Science degree in Computer Information Systems. Tristen is a dedicated individual

with a passion for education and research. She has been awarded multiple times for

her academic achievements, including her research on women in the Information

Technology field. Tristen has the honour of being the first student at Nicholls State

University’s College of Business to be awarded the first-place position in the

Undergraduate Student Research Competition. The author can be contacted at

[email protected]

En Mao, Ph.D., is Professor of Computer Information Systems and Candies 500

Endowed Professor at the College of Business Administration, Nicholls State University.

She has published in journals such as the Information and Management, Data Base,

Communications of the AIS, and Journal of Consumer Research. She is a technology

expert and has a wide range of knowledge in technology implementation. The author

can be contacted at [email protected]

95


Appendix A - 2019 University Enrolment Data

o Out of 5103 students currently enrolled, 64% are female and 36% are male.

o 3% of students are CIS majors

o Out of 153 CIS majors, 15% are female and 85% are male.

o An average ACT composite score, as well as an average GPA, was

calculated for female and male current CIS students – no significant

difference was found.

o Average female ACT score: 22.31

o Average male ACT score: 22.29

o Average female GPA: 2.92

o Average male GPA: 2.9

Appendix B - Interviews with students and employees

Female students that are not enrolled in CIS, but are within the College of Business

stated that they enjoyed the CIS courses that they have taken, but that they are not

confident that they could succeed in this field.

o “I like thinking I could be in IT, but in reality, I don’t think I could handle the

deeper understanding of what goes into the background of IT.”

Female Business major

o “I took an intro CIS course as a prerequisite to another course. I liked it a lot so I

changed my major from general business to CIS.”

Female CIS major

Interviewed students expressed that they have noticed the gender disparity issue in

the CIS department. When asked if they have experienced any sexism or disrespect,

3 out of 5 female students said that they had experienced this within our department.

A former IBM employee admitted that she has experienced sexism and disrespect

within the workplace. A lot of men don’t trust women’s expertise.

o “It’s kind of like a question of credibility against women.”

Former IBM employee

The former IBM employee also mentioned that a big reason why women aren’t in IT is

likely due to cultural norms and societal expectations – it is typically expected, from a

young age, that females become nurses, teachers, etc.

o “We need to make people aware of how much IT matters. Doctors, lawyers,

etc. cannot do what they do without the applications that we program to allow

them to do their job. We are an internal part…If we can show people how

important IT is and how big of an impact IT is, I think more people would want

to make a difference… if people can understand that we can do that in IT,

they might be more inclined to pursue it.”

Former IBM employee

96


The Effect of Auditor Rotation on the

Relationship between Financial

Manipulation and Auditor’s Opinion

Ivica Filipović, Toni Šušak, Andrea Lijić

University of Split, University Department of Forensic Sciences, Croatia

Abstract

Background: Since external auditors possess the expertise necessary for detecting

manipulations in financial statements, they should also take into account earnings

management that could lead to it. In that context, auditor’s independence, which

can be affected by auditor’s rotation, is of utmost importance. Objectives: This

paper aims to examine the moderating effect of auditor rotation on the relationship

between the extent of financial manipulation and the type of auditor’s opinion for

companies listed on the Zagreb Stock Exchange in the Republic of Croatia.

Methods/Approach: A panel analysis with logistic regression is conducted to test the

research hypothesis. The sample consists of 210 observations during the three years

from 2015 to 2017. Results: Results show a significant positive relationship between

auditor rotation in a current financial year and auditor’s opinion. Furthermore, there

is a negative, but the statistically insignificant moderating effect of auditor rotation in

a current financial year on the relationship between financial manipulation and

auditor’s opinion, as well as the statistically insignificant moderating effect of auditor

rotation frequency over five years on the relationship between financial

manipulation and auditor’s opinion. Conclusions: It is not likely that auditors take

earnings management into account when generating their opinion on financial

statements, and auditor rotation is not proven to be an adequate stimulus in that

context.

Keywords: auditor’s opinion; auditor rotation; earnings management; financial

manipulation

JEL classification: M41, M42


Received: 4.12.2020.

Accepted: 11.3.2021.

Citation: Filipović, I., Šušak, T., Lijić, S. (2021), “The Effect of Auditor Rotation on the

Relationship between Financial Manipulation and Auditor’s Opinion”, Business



97


Introduction The beginning of the current century was extremely challenging for the corporate

sector because of numerous accounting scandals (Bajra et al., 2018) which have

drawn attention to earnings management practices worldwide (Idris et al., 2018).

Accounting irregularities were discovered in companies such as Enron (the U. S.),

Olympus (Japan), Tesco (the U. K.), One Tel (Australia) (Enomoto et al., 2018),

evidencing their geographical comprehensiveness. The cardinal issue in those cases

was “a discrepancy between real earnings and reported earnings in financial

statements” (Nezami, 2011, in Moazedi et al., 2016, pp. 113). Legislative responses

were timely and included “many stringent regulations to strengthen financial

disclosures and improve corporate governance practices” (Gounopoulos et al.,

2018, pp. 13).

Given that external auditors can be classified among the fundamental

mechanisms for alleviating information asymmetry between management and

investors (Rusmanto et al., 2014), it is presumable that they will successfully detect

and report earnings management practices (Butler et al., 2004). The importance of

auditor’s consideration of earnings management activities was also emphasized in

the International Standard on Auditing 240: The Auditor's Responsibilities Relating to

Fraud in an Audit of Financial Statements, which has been effective since 15th

December 2009. This standard stipulates that “discussion among the engagement

team may include a consideration of circumstances that might be indicative of

earnings management and the practices that might be followed by management

to manage earnings that could lead to fraudulent financial reporting” (IAASB, 2009,

pp. A11).

Unlike theoretical assumptions, the situation in corporate practice indicates that

“it is not obvious that earnings management will typically lead to a modified audit

opinion” (Butler et al., 2004, pp. 140). Among the reasons used to explain suboptimal

audit reporting, auditor’s independence may be indicated as of utmost importance.

In support of this notion, the external audit was seriously questioned after the

collapse of Arthur Andersen (Jamal, 2006) whose close relationship with the Enron

Corporation was highlighted as a possible reason for their compromised objectivity

(Herrick et al., 2002, in Kerler et al., 2009) and lack of professional skepticism

(Johnstone et al., 2001:5, in Kerler et al., 2009).

Thus, auditor’s independence was one of the issues affected by the

aforementioned legislative actions which could be improved in a variety of ways.

One of them is shortening auditor tenure “so that the engagement can be viewed

with fresh and skeptical eyes” (SOX, 2002, in Anis, 2014, pp. 105). There are also

certain negative effects on auditor’s competence which can be caused by more

frequent auditor rotation.

In light of those considerations, the purpose of this paper was to examine the

moderating effect of auditor rotation on the relationship between the extent of

financial manipulation and the type of auditor’s opinion. Results of this research add

evidence to intensive scientific and professional discussions on the usefulness of

auditor rotation but from the standpoint of auditor’s reporting decisions regarding

earnings management.

As far as the authors are aware, previous research (e.g. Herbohn et al., 2008;

Garcia Blandon et al., 2013; Omid, 2015; and Alhadab, 2016) hadn’t addressed that

question and, therefore, it is the scientific contribution of this paper. Prior studies have

been focused on the relationship between earnings management and independent

auditor opinion. Having regard to the long-standing debate about the effects of

98


auditor rotation, which makes the anticipation of its impact on the relationship

between financial manipulation and auditor’s opinion very complex, the statistical

significance of mentioned variable’s moderating effect was tested in this research.

Research results provide important practical implications and may be insightful for

a variety of stakeholders. Contrary to the expectations, research results indicated

that auditors tend to not consider earnings management activities and that could

be due to the absence of legal coercion to do so. Auditor rotation, as a way to

ensure greater auditors’ effort regarding the inclusion of earnings management

information, did not prove to be an effective mechanism implying that there is no

need to shorten the auditor’s tenure.

The remainder of the paper is organized as follows – in the second section results

of relevant previous research were provided. The third section presents statistical

methodology and models applied for hypothesis testing, the fourth section

comprises results of analysis, the fifth section provides theoretical and practical

implications of the research, and the sixth section contains the explanation of results

as well as final remarks.

Literature Review and Hypothesis Development As indicated in the previous section, auditors should have an important role

regarding earnings management in the financial statements of their clients. Francis

et al. (1999) founded that U. S. companies with higher accruals were more likely to

be provided with a modified audit report, as well as companies with income-

increasing accruals. Based on the results of their research, they highlighted the

“potentially important role played by accounting accruals in audit report formation

process” (Francis et al., 1999, pp. 135). In that context, Herbohn et al. (2008, pp. 576)

noticed that “prior research has considered the possibility that auditors modify their

opinions to communicate information about potential earnings management by

firms with large accruals”. Alhadab (2016) determined a positive association of

qualified audit opinion with both real and accrual earnings management, while

Omid (2015), who also analyzed both of these activities, founded a relationship

exclusively with accrual earnings management which doesn’t affect cash flows.

Not all researches have proven the relationship between earnings management

and auditor’s opinion. For instance, Garcia Blandon et al. (2013, pp. 36), besides

reporting failure to find a significant relationship between audit report qualification

and earnings management, stated that previous research on the relationship

between aforementioned variables was “scarce and almost limited” and “lacks

agreement on whether external auditors are aware of earnings management when

issuing a report”. A similar conclusion was reached by Butler et al. (2004, pp. 162) for

companies with modified audit opinions, who did not found “evidence that auditors

use their opinions to alert financial statement users of either excessive earnings

management or the consequences of high positive accruals” and Bradshaw et al.

(2001, pp. 45) whose research did not result with “evidence that auditors signal the

future earnings problems associated with high accruals through either their audit

opinions or through auditor changes”.

This was explained by the remark that “earnings quality issues are beyond the

scope of the audit” because auditors “are not required to share this information by

investors through their audit opinions” (Bartov, 2001, in Omid, 2015, pp. 49).

Tsipouridou et al. (2014) have investigated the same relationship for companies listed

on the capital market in Greece and founded no relationship between qualified

audit opinions and earnings management. They have divided audit reports into

categories considering reasons for their qualification and analyzed the basic

99


relationship in an environment with strong incentives to manage earnings, i.e.

financial distress.

Herbohn et al. (2008, pp. 575) have extended the previous research by

considering the possibility that “managers adjust accruals to report earnings that

better predict future firm performance, which has the side-effect of placing them in

conflict with their auditors” and concluded that there is “no evidence of earnings

management leading to an audit opinion modification” among listed companies in

Australia.

One of the variables with a potentially significant effect on the relationship

between auditor’s opinion and extent of earnings management is the auditor

rotation which could be an indication of both audit efficiency, because “long audit

tenures gain value and knowledge about the client since an audit firm can better

evaluate the risk of material misstatements, gain more experience and have better

insights into client’s operations and business strategies as well as internal controls

over financial reporting” (Yet et al., 2013, in Alvarado et al., 2019, pp. 15), or auditor

independence which is considered “one of the key factors in increasing the quality

of audited accounting statements” (Kim et al., 2015, in Silvestre et al., 2018, pp. 412)

and goes in line with the statement that “when a company practices earnings

management it does not necessarily mean they are likely to receive a qualified

opinion from their auditors” (Rusmanto et al., 2014, pp. 1).

This has resulted in a permanent conflict of opinions between supporters and

opponents of mandatory audit rotation (Silvestre et al., 2018). Despite the recent

implementation of mandatory audit rotation in European legislation (Silvestre et al.,

2018), some countries, which had done so earlier, abandoned it after identifying an

absence of expected benefits (Raiborn et al., 2006, in Ryken et al., 2007). The Big

Four audit companies Ernst & Young and PricewaterhouseCoopers are opponents of

mandatory auditor rotation and believe that “costs of mandatory audit firm rotation

would outweigh the perceived benefits of a required "fresh look" at the financial

statements by a new audit firm” (PricewaterhouseCoopers, 2012, in Bamahros et al.,

2015:146) and that mandatory audit partner rotation is more effective alternative

(Ernst & Young, 2013, in Bamahros et al., 2015). In line with that idea are results of

research conducted by Firth et al. (2011), what cannot be stated for Gates et al.

(2007, pp. 5) who founded that “even in an environment of strong controls for

corporate governance, audit firm rotation incrementally influenced individuals’

confidence in financial statements” and “audit partner rotation did not have a

similar effect”.

Taking into account the discussion in this chapter, research hypothesis can be

formulated as follows:

H1: There is a statistically significant moderating effect of auditor rotation on the

relationship between financial manipulation and auditor’s opinion.

Methodology The research sample comprises large nonfinancial companies listed on the Zagreb

stock exchange in the Republic of Croatia (210 observations during the three years

from 2015 to 2017). Financial companies were excluded because of their

accounting and legal specificities such as “the differences in the accrual process”

(Johl et al., 2007, pp. 713), tax rules (Kang et al., 2019), auditing process (Desender et

al., 2013) and the strict oversight by Croatian National Bank. The data was collected

from financial statements and independent auditors’ reports publicly available for

companies whose shares were listed on the Zagreb Stock Exchange. The extent of

financial manipulation was estimated using the Dechow & Dichev model which

100


quantifies earnings management with discretionary accruals (Peni et al., 2010). The

model is specified as follows (Kallapur et al., 2008):

Δ WCit = β0 + β1*CFOi,t−1 + β2*CFOi,t + β3*CFOi,t+1 + εi,t (1)

where:

Δ WC = change in the value of net working capital

CFO = operating cash flows

ε = residual value (Kallapur et al., 2008).

The standard deviation of three-year period residuals from Dechow & Dichev

model was utilized as a measure of accruals quality (Kallapur et al., 2008). Content

analysis was applied to determine the information contained in independent

auditors’ reports, such as auditor’s opinion, auditor’s size, and auditor rotation, while

financial data was gathered from annual financial statements. Furthermore,

correlational analysis and panel analysis with logistic regression were applied for

statistical analysis which was conducted in Stata 13.1. (StataCorp, 2013). Hausman

test was conducted “to determine which model is best suited to … data (the fixed

effects … or random effects)” (Hausman, 1978, in Saenz Gonzalez et al., 2014, pp.

427-429). The mentioned test indicates random effects as more appropriate for both

models. Also, panel data was used because it is “appropriate for treating the

unobserved heterogeneity problem that often appears in the cross-sectional data

analysis” (Yasser et al., 2017, pp. 186) and it “may offer a solution to the problem of

bias caused by unobserved heterogeneity and reveal dynamics that are difficult to

analyze with cross-sectional data” (Cerqueira et al., 2013, pp. 42-43). Breusch-Pagan

Lagrange multiplier (Wooldridge, 2009, in Flores et al., 2016, pp. 191) “indicated the

presence of unobserved heterogeneity” and thus random effects were used (Flores

et al., 2016, pp. 191). Given that two research variables were used for testing the

established hypothesis, two statistical models were specified:

IAOi,t = β0 + β1*MNPi,t + β2*ROTi,t + β3*MNP x ROTi,t + β4*BIG4i,t + β5*ROAi,t + β6*LBLi,t +

β7*SIZi,t + β8*INDi,t + ui,t (2)

IAOi,t = β0 + β1*MNPi,t + β2*ROT_5i,t + β3*MNP x ROT_5i,t + β4*BIG4i,t + β5*ROAi,t + β6*LBLi,t +

β7*SIZi,t + β8*INDi,t + ui,t (3)

where:

Dependent variable:

IAO = independent auditor’s opinion (1 = positive auditor’s opinion, 0 = modified

auditor’s opinion)

Test variables:

MNP = financial manipulation estimated with earnings management measure –

Dechov and Dichev model

ROT = auditor rotation in a current financial year (1 = different audit company

appointed in comparison to previous financial year, 0 = the same audit company

appointed as the previous year)

ROT_5 = number of auditor rotations over five years

MNP x ROT = interaction between financial manipulation and auditor rotation in a

current financial year

MNP x ROT_5 = interaction between financial manipulation and number of auditor

rotations in five years

Control variables:

101


BIG4 = type of auditor (1 = Big Four, 0 = not Big Four)

ROA = return on assets

LBL = total liabilities to total assets

SIZ = natural logarithm of total assets

IND = industry

u = model error.

The dependent variable is the independent auditor’s opinion (IAO) which denotes if

an auditor has given a positive or modified opinion. Test variables are financial

manipulation (MNP), auditor rotation in a current financial year (ROT), number of

auditor rotations in five years (ROT_5), and the interaction between those variables

(MNP x ROT and MNP x ROT_5). The latter variables are the most important in the

context of this research for hypothesis testing. To provide more accurate results,

several control variables defined in previous researches were included in the models.

These include the size of a company (SIZ) (Rusmanto et al., 2014), financial health

variables (Tsipourodou et al., 2014) such as return on assets (ROA), and total liabilities

to total assets (LBL), as well as auditor characteristics (BIG4).

The effect of company size is complex to predict because, on the one side, there

could be a higher propensity to issue a qualified opinion because of higher litigation

costs for larger companies (Lys & Watts, 1994; Shu, 2000, in Garcia Blandon et al.,

2013), but on the other side “auditors could be less independent when auditing large

clients and, therefore, less willing to issue a qualified report to large than small

clients” (DeAngelo, 1981, in Garcia Blandon et al., 2013:42). A positive relationship is

expected between the qualification of an audit report and the ratio of total liabilities

and total assets because “high levels of debt increase the probability of bankruptcy,

and consequently increase litigation risk” (Garcia Blandon et al., 2013:42). Profitability

measure is included because losses could “indicate … poor financial health”

(Monroe et al., 1993, in Johl et al., 2007:695), while variable denoting audit

company’s affiliation to the Big Four was included because “these auditors are

expected to qualify more frequently” (Johl et al., 2007:695).

Results Correlation analysis (Table 1) was applied to test the existence of the multicollinearity

issue.

Table 1

Correlation Coefficients between Research Variables

MNP ROT ROT_5 BIG4 ROA LBL SIZ IND

MNP 1

ROT -0.06 1

ROT_5 -0.10 0.5* 1

BIG4 -0.04 0.08 0.07 1

ROA -0.20* -0.01 -0.04 -0.14* 1

LBL 0.02 -0.01 -0.01 0.09 -0.4* 1

SIZ 0.01 -0.01 -0.13 0.41* -0.03 0.15* 1

IND 0.05 0.12 0.09 -0.08 0.11 -0.18* -0.06 1

Note: * Correlation coefficient is statistically significant (5 percent threshold).

Source: Authors’ analysis using data available at the official website of the Zagreb Stock

Exchange and the Stata software – StataCorp (2013). Stata Statistical Software: Release 13.

College Station, TX: StataCorp LP.

102


A relevant correlation was detected only between auditor rotation in the current

financial year and the number of auditor rotations in five years. The remaining

coefficients in Table 1 prove the absence of a strong correlation between

explanatory variables, i.e. absence of the multicollinearity issue. Table 2 presents the

descriptive statistics of relevant variables.

Table 2

Descriptive statistics

Variable Mean Std. Dev. Min Max

MNP 0,067 0,090 0,003 0,749

ROT 0,152 0,360 0,000 1,000

ROT_5 0,652 0,835 0,000 4,000

BIG4 0,500 0,501 0,000 1,000

ROA -0,026 0,232 -2.221.928 0,226

LBL 0,448 0,274 0,040 2.366.989

SIZ 8.757.801 0,514 7.730.895 1.030.925




As it is apparent from Table 3, the average values of the financial manipulation

variable did not vary significantly from 2015 to 2017. Average values decreased in

2016, but in 2017 their extent was similar to in 2015. Logistic regression analysis was

used to estimate test variables’ coefficients (auditor rotation in current financial year

and number of auditor rotations in five years). For each of these variables, a

separate model was created.

Table 3

Descriptive statistics – variable financial manipulation (MNP)

Year Mean Std. Dev. Min Max

2015 0.073 0.100 0.003 0.736

2016 0.053 0.099 0.003 0.749

2017 0.073 0.066 0.006 0.329




In Model 1 (Table 4), which was used to test the first hypothesis, the coefficient of

auditor rotation in a current financial year is positive and statistically significant at 10

percent threshold denoting its positive association with the opinion given by the

independent auditor, indicating that the negative effect of reduced auditor’s

knowledge on their clients’ operations (Yet et al., 2013, in Alvarado et al., 2019) may

prevail over the positive effect of increased auditor’s independence.

Despite the negative relationship between auditor rotation in a current financial

year and the relationship between financial manipulation and auditor’s opinion,

which indicates the strengthening of an initial effect, there was no statistical

significance.

103


Table 4

Coefficients of Variables included in Logistic Regression Model – Auditor Rotation in

Current Financial Year (Model 1)

Coefficient Std. Err. Z P > z

MNP -6.148 5.702 -1.080 0.281

ROT 3.038 1.798 1.690 0.091

MNP x ROT -32.708 23.340 -1.400 0.161

BIG4 -0.129 0.894 -0.140 0.885

ROA 1.404 1.608 0.870 0.383

LBL -5.540 2.247 -2.470 0.014**

SIZ 4.182 1.372 3.050 0.002***

IND 0.172 0.286 0.600 0.547

Constant -31.917 11.172 -2.860 0.004***




Note: ** statistically significant at 5%; *** 1%

Similar to Model 1, the results of Model 2 (Table 5), which were used to test the

second hypothesis, show an insignificant negative relationship between auditor

rotation frequency during the considered period and the relationship between

financial manipulation and auditor’s opinion. Also, the coefficients of control

variables LBL and SIZ have been proven to be statistically significant in both Model 1

and Model 2. The value of variable SIZ is positive and in line with the prediction that

“auditors could be less independent when auditing large clients and, therefore, less

willing to issue a qualified report to large than small clients” (DeAngelo, 1981, in

Garcia Blandon et al., 2013:42). The coefficient of LBL was negative, indicating an

inverse relationship between total liabilities to total assets and issuing positive

auditor’s opinion, i.e. a greater likelihood of providing highly leveraged companies

with qualified auditor’s opinion according to the notion that “high levels of debt

increase the probability of bankruptcy, and consequently increase litigation risk”

(Garcia Blandon et al., 2013:42).

Table 5

Coefficients of Variables included in Logistic Regression Model – Number of Auditor

Rotations in Period of Five Financial Years (Model 2)

Coefficient Std. Err. Z P > z

MNP -6.672 6.415 -1.040 0.298

ROT_5 0.289 0.680 0.430 0.670

MNP x ROT_5 -2.090 10.116 -0.210 0.836

BIG4 0.416 0.826 0.500 0.615

ROA 1.343 1.648 0.810 0.415

LBL -5.287 2.097 -2.520 0.012**

SIZ 3.931 1.277 3.080 0.002***

IND 0.205 0.271 0.760 0.450

Constant -30.239 10.493 -2.880 0.004***

Source: Authors’ analysis using data available at the official website of Zagreb Stock

Exchange and Stata software – StataCorp (2013). Stata Statistical Software: Release 13.



104


Discussion Theoretical Contributions Financial statement manipulation is one of the most important issues in a

contemporary business environment that undermines stakeholders’ trust and,

consequentially, hinders investment activities. There is a significant number of papers

that analyze different forms of this phenomenon in various situations. The primary aim

of these efforts is directed towards finding the most efficient solutions to deter or

decrease the occurrence of manipulative activities.

External auditing can be classified among mechanisms with the greatest

untapped potential to do so. An external auditor usually possesses expert knowledge

and skill set relevant for detecting manipulative activities in their clients’ financial

statements. In that context, their most efficient tool for enhancing financial reporting

quality is an independent auditor’s report which they use as a medium for disclosing

their opinion.

As stated in the section Literature Review and Hypothesis Development, previous

research conducted in other countries yielded mixed results in an analysis of the

relationship between earnings management and auditor’s opinion. This research

paper gave an insight into Croatian business practice regarding the inclusion of

earnings management information in independent auditors’ reports. Research results

did not confirm that auditors took into consideration earnings management activities

in their clients’ financial statements most probably due to insufficient regulatory

pressure.

Although prior studies have been focused on the relationship between earnings

management and independent auditor opinion, research efforts should be focused

on discovering factors that could potentially strengthen that association. Auditor

rotation may result in increased auditors’ independence and their higher objectivity.

Subsequently, the primary theoretical contribution of this paper was expanding the

previous knowledge on the basic relationship by constructing the model for analysis

of the moderating effect that auditor rotation has on the relationship between

financial manipulation and auditor’s opinion.

Practical Implications Research results provide important practical implications and may be insightful for a

variety of stakeholders, such as regulatory bodies, audit companies, auditees,

forensic accountants and other financial investigators, investors, creditors, and

academics.

The role of regulatory bodies is crucial given that they provide a framework for

corporate activities. As stated in the previous subchapter and contrary to the

expectations, research results indicated that auditors tend to not consider earnings

management activities and that could be due to the absence of legal coercion to

do so. The identification of earnings management practices requires additional time

and effort, which are scarce resources in a competitive business environment, and

auditors predominantly opt not to do facultative activities.

Financial fraud related to tax evasion is incriminated with legal provisions of the

Croatian Criminal Code (Official Gazette, 2011) but earnings management, due to

its lower intensity, vagueness, and lower-level repercussions for the national budget,

is regulated on the level of a principle. The important issue is that, as stated in the

International Standard on Auditing 240: The Auditor's Responsibilities Relating to

Fraud in an Audit of Financial Statements, “… earnings management … could lead

to fraudulent financial reporting” (IAASB, 2009, pp. A11). These potential threats

105


indicate that earnings management, although less harmful to society than

fraudulent activities, shouldn’t be neglected.

Subsequently, mentioned bodies should revise existing regulations and

incorporate certain provisions in a legal system which will be more binding than

mere recommendations and stimulate auditors to increase their professional effort.

This is also important because it is presumable that auditees adjust their accounting

activities and decisions according to expectations based on previous audit

engagements. If auditors do not have a rigorous approach regarding earnings

management, the level of financial manipulation will likely be high. Thus, auditors

should be stimulated to take earnings management into account.

Given that the large companies listed on the national stock exchange were the

subject of research, additional legal efforts towards ensuring the consideration of

earnings management could contribute to restoring investors’ trust indispensable to

the proper functioning of financial markets and increasing the quality of their

decision making. Besides investors, creditors are also interested in that information

when deciding on the creditworthiness of a client, because they seek to secure

repayment of loan installments and steady cash flows in the future what could be

jeopardized if a company did not disclose reliable financial information.

On the other side, in certain cases, these results could also be attributed to the

lack of specialized forensic accounting knowledge relevant for identifying earnings

management. This issue can be effectively addressed by providing professional

training on forensic accounting techniques.

Given the insufficient inclusion of earnings management by auditors in the

Republic of Croatia while generating their reports, audit companies that do so could

gain a competitive advantage over other audit companies. Accordingly, they

would provide more reliable information on the auditee’s financial reporting

informativeness and quality and, consequently, establish their reputation among

potential clients.

Besides that, several directions could ensure greater auditors’ effort regarding the

inclusion of earnings management information while deciding on their opinion. One

of them is auditor rotation which was analyzed in this paper, but the results imply that

it did not prove to be an effective mechanism for improving auditor’s independence

and, consequentially, the level of reporting on earnings management.

Correspondingly, there is no need to act in the direction of shortening the

auditor’s tenure. Also, research results reduce the importance of auditor reports as a

reliable information source in forensic investigations conducted by forensic

accountants and other financial investigators such as tax professionals and police

inspectors who could use them for preliminary screening of manipulation extent in

financial statements of a company that is subject of their investigation.

Conclusion Relations to previous findings Auditors are perceived as the essential mechanism for alleviating the agency

problem between managers and investors. Their efforts in reducing information

asymmetry are, among others, focused on analyzing and communicating the

presence of financial manipulation in their client’s financial statements. Results did

not indicate that auditors in the Republic of Croatia take into account the

recommendation provided in International Standard on Auditing 240 on

“consideration of circumstances that might be indicative of earnings management”

(IAASB, 2009:A11), so they could not be of use for financial statement users in this

106


regard. This is also in line with the findings of Bradshaw et al. (2001), Butler et al.

(2004), and Garcia Blandon et al. (2013).

Auditor rotation has been a controversial issue that started the long-standing

debate on the cost-effectiveness of its implementation in an auditing system. This

study analyzed the moderating effect of auditor rotation on the relationship

between financial manipulation and auditor’s opinion. Since results had not been

statistically significant, the research hypothesis was not accepted corroborating

remarks by opponents of mandatory auditor rotation quoted in previous sections of

the paper (PricewaterhouseCoopers, 2012, in Bamahros et al., 2015 and Ernst &

Young, 2013, in Bamahros et al., 2015).

Results of the research have shown a positive effect of auditor rotation in a

current financial year on auditor’s opinion which could be attributed to decreased

knowledge about business operations of a client (Yet et al., 2013, in Alvarado et al.,

2019). Presumably, this detrimental effect is much stronger than increased auditor’s

independence resulting from the appointment of a new auditor.

Finally, it can be concluded that it is not likely that auditors will incorporate

information on earnings management in their reports, probably because

International Standard on Auditing only recommends consideration of earnings

management, and thus auditors are not obliged to do so. This is in line with remarks

stated by Bartov (2001, in Omid, 2015). As far as the authors are aware, this paper is

the first empirical attempt to analyze the moderating effect of auditor rotation on

the relationship between financial manipulation and auditor’s opinion. Results of this

research provide a basis for regulative actions regarding auditor rotation and they

could also be useful for investors, auditors, and other stakeholders.

Research Limitations and Future Research Possible limitations of the study also have to be considered – the Dechow & Dichev

model, as all other earnings management measures, has potential shortcomings,

and the fact that the research sample is focused exclusively on large companies in

the Republic of Croatia implies that results may not be generalizable. Also,

researching other countries could provide an opportunity for testing those

relationships on larger samples given that the Croatian stock market is relatively small

and inefficient. Despite the authors’ efforts to increase the precision of model

estimation, there is always a possibility that not all variables with a significant impact

were included in the model. Future research should consider factors that could

stimulate the inclusion of earnings management information in independent

auditor’s reports. The aforementioned research limitations are also features that

could be improved in future studies.

References 1. Alhadab, M. (2016), “Auditor report and earnings management: Evidence from FTSE 350

companies in the UK”, Risk Governance and Control: Financial Markets & Institutions, Vol. 6

No. 4, pp. 334-344.

2. Alvarado, N. R., de Fuentes, P., Laffarga, J. (2019), “Do auditors mitigate earnings

management during economic crisis?”, Revista de Contabilidad-Spanish Accounting

Review, Vol. 22 No. 1, pp. 6-20.

3. Anis, A. (2014), “Auditors' perceptions of audit firm rotation impact on audit quality in

Egypt”, Accounting & Taxation, Vol. 6 No. 1, pp. 105-120.

4. Bajra, U., Cadez, S. (2018), “The impact of corporate governance quality on earnings

management: Evidence from European companies cross‐listed in the US”, Australian

Accounting Review, Vol. 28 No. 2, pp. 152-166.

107


5. Bamahros, H. M., Wan-Hussin, W. N., Abdullah, A. (2015), “Non-audit services, audit firm

tenure and earnings management in Malaysia”, Asian Academy of Management Journal

of Accounting and Finance, Vol. 11 No. 1, pp. 145-168.

6. Bradshaw, M. T., Richardson, S. A., Sloan, R. G. (2001), “Do analysts and auditors use

information in accruals?”, Journal of Accounting research, Vol. 39 No. 1, pp. 45-74.

7. Butler, M., Leone, A. J., Willenborg, M. (2004), “An empirical analysis of auditor reporting

and its association with abnormal accruals”, Journal of Accounting and Economics, Vol.

37 No. 2, pp. 139-165.

8. Cerqueira, A., Pereira, C. (2013), “Financial reporting quality and information asymmetry in

Europe”. In Polouček, S., Stavárek, D. (Ed.), Financial regulation and supervision in the

after-crisis period, 14th International Conference on Finance and Banking, Ostrava, Czech

Republic, 16-17 October 2013, Silesian University, School of Business Administration, pp. 32-

51.

9. Desender, K. A., Aguilera, R. V., Crespi, R., GarcÍa‐cestona, M. (2013). “When does

ownership matter? Board characteristics and behaviour”, Strategic Management Journal,

Vol. 34 No. 7, pp. 823-842.

10. Enomoto, M., Kimura, F., Yamaguchi, T. (2018), “A cross‐country study on the relationship

between financial development and earnings management”, Journal of International

Financial Management & Accounting, Vol. 29 No. 2, pp. 166-194.

11. Firth, M., Rui, O. M., Wu, X. (2011), “How do various forms of auditor rotation affect audit

quality? Evidence from China”, available at https://ssrn.com/abstract=1830741 (8

February 2021).

12. Flores, E., Weffort, E. F. J., da Silva, A. F., Carvalho, L. N. G. (2016), “Earnings management

and macroeconomic crises: Evidences from Brazil and USA capital markets”, Journal of

Accounting in Emerging Economies, Vol. 6 No. 2, pp. 179-202.

13. Francis, J. R., Krishnan, J. (1999), “Accounting accruals and auditor reporting

conservatism”, Contemporary accounting research, Vol. 16 No. 1, pp. 135-165.

14. Garcia Blandon, J., Argiles Bosch, J. M., Martinez Blasco, M. (2013), “Earnings

management and audit qualifications: A non-matched sample approach”, Research in

Finance, Vol. 29, pp. 29-62.

15. Gates, S. K., Jordan Lowe, D., Reckers, P. M. J. (2007), “Restoring public confidence in

capital markets through auditor rotation”, Managerial Auditing Journal, Vol. 22 No. 1, pp.

5-17.

16. Gounopoulos, D., Pham, H. (2018), “Financial expert CEOs and earnings management

around initial public offerings, available at https://ssrn.com/abstract=3177760 (8 February

2021).

17. Herbohn, K., Ragunathan, V. (2008), “Auditor reporting and earnings management: Some

additional evidence”, Accounting & Finance, Vol. 8 No. 4, pp. 575-601.

18. Idris, M., Siam, Y. A., Nassar, M. (2018), “Board independence, earnings management and

the moderating effect of family ownership in Jordan”, Management & Marketing.

Challenges for the Knowledge Society, Vol. 13 No. 2, pp. 985-994.

19. International Auditing and Assurance Standards Board – IAASB (2009), “The Auditor's

Responsibilities Relating to Fraud in an Audit of Financial Statements”, International

Standard on Auditing 240.

20. Jamal, K. (2006), “Mandatory audit of financial reporting: A failed strategy for dealing with

fraud”, Available at https://ssrn.com/abstract=914156 / (8 February 2021).

21. Johl, S., Jubb, C. A. and Houghton, K. A. (2007), “Earnings management and the audit

opinion: evidence from Malaysia”, Managerial Auditing Journal, Vol. 22 No. 7, pp. 688-715.

22. Kallapur, S., Sankaraguruswamy, S., Zang, Y. (2008), “Audit Market Concentration and

Audit Quality”, available at http://citeseerx.ist.psu.edu/viewdoc/download?

doi=10.1.1.628.5104&rep=rep1&type=pdf (8 February 2021).

23. Kang, S. M., Hwang, I. T., & Hur, K. S. (2019), “Non‐audit services and auditor

independence depending on client performance”, Australian Accounting Review, Vol. 29

No. 3, pp. 485-501.

108


24. Kerler, W. A., Killough, L. N. (2009), “The effects of satisfaction with a client’s management

during a prior audit engagement, trust, and moral reasoning on auditors’ perceived risk of

management fraud”, Journal of Business Ethics, Vol. 85 No. 2, pp. 109-136.

25. Moazedi, E., Khansalar, E. (2016), “Earnings management and audit opinion”, International

Journal of Economics and Finance, Vol. 8 No. 4, pp. 113-122.

26. Official Gazette of the Republic of Croatia (2011), Criminal Code, Zagreb: Narodne

Novine, d. d., 125/11, 144/12, 56/15, 61/15, 101/17, 118/18, 126/19.

27. Omid, A. M. (2015), “Qualified audit opinion, accounting earnings management and real

earnings management: Evidence from Iran”, Asian Economic and Financial Review, Vol. 5

No. 1, pp. 46-57.

28. Peni, E., Vahamaa, S. (2010), “Female executives and earnings management”,

Managerial Finance, Vol. 36 No. 7, pp. 629-645.

29. Rusmanto, T., Djamil, A. B., Salim, Y. (2014), “The effect of earnings management to

issuance of audit qualification: Evidence from Indonesia”, Journal of Business Studies

Quarterly, Vol. 6 No. 1, pp. 1-28.

30. Ryken, K., Radich, R., Fargher, N. L. (2007), “Audit partner rotation: Evidence of changes in

audit partner tenure as the result of mandatory regulation in Australia”, Current Issues in

Auditing, Vol. 1 No. 1, pp. A28-A35.

31. Saenz Gonzalez, J., Garcia-Meca, E. (2014), “Does corporate governance influence

earnings management in Latin American markets?”, Journal of business ethics, Vol. 121

No. 3, pp. 419-440.

32. Silvestre, A. O., Costa, C. M., Kronbauer, C. A. (2018), “Audit Rotation And Earnings Quality:

An Analysis Using Discretionary Accruals”, BBR. Brazilian Business Review, Vol. 15 No. 5, pp.

410-426.

33. StataCorp (2013), “Stata Statistical Software: Release 13”, College Station, TX: StataCorp

LP.

34. Tsipouridou, M., Spathis, C. (2014), “Audit opinion and earnings management: Evidence

from Greece”, Accounting Forum, Vol. 38 No. 1, pp. 38-54.

35. Yasser, Q. R., Al Mamun, A., Hook, M. (2017), “The impact of ownership structure on

financial reporting quality in the east”, International Journal of Organizational Analysis, Vol.

25 No. 2, pp. 178-197.

About the authors

Ivica Filipović, Ph.D. is a Full Professor at the University of Split, University Department

of Forensic Sciences. He received a Ph.D. in Economics at the Faculty of Economics

Split. His main research interests are accounting and audit. The author can be


Toni Šušak, M. Econ., M. Law is a Research and Teaching Assistant at the University of

Split, University Department of Forensic Sciences. He received his Master’s degree in

Economics from the Faculty of Economics Split, as well as a Master’s degree in Law

from the Faculty of Law Split and is currently a Ph.D. candidate at Faculty of

Economics Split. His main research interests are forensic accounting and audit. The

author can be contacted at [email protected].

Andrea Lijić, M. Forens. received her Master’s degree in Forensics from the University

of Split, University Department of Forensic Sciences. Her previous work experience

was in private companies and includes areas such as accounting and finance. Her

main research interests are accounting, audit, forensic accounting, and forensic

audit. The author can be contacted at [email protected].

109


Freelancing in Croatia: Differences among

Regions, Company Sizes, Industries and

Markets

Ana Globočnik Žunac, Sanja Zlatić, Krešimir Buntak

University North, Croatia

Abstract

Background: Freelancers have a significant impact on economic growth due to their

specific skills that are nowadays often used as complements to the regular full-time

workforce of a company, not as their competition. Objectives: The study aims to

investigate employers’ attitudes towards the employment of freelancers in Croatia,

taking into account the place of establishment, the operational market, the size,

and the industry of the organization hiring freelancers. Methods/Approach:

Differences among organizations according to their attitudes towards freelancers

are analyzed by multiple 2xc Fisher’s exact tests with the Monte Carlo Simulation,

and binomial logistic regression analysis. Results: Significant differences are found in

terms of the operational market and the industry in which the company operates.

Besides, the binomial logistic regression analysis identified the following independent

constructs as significant predictors of hiring freelancers: the region of the company’s

seat, company size, and area of operation. Conclusions: The national legislation

should complement the developmental policies to encourage employment and

especially self-employment of freelancers.

Keywords: freelancing; outsourcing; business externalization; hiring; logistic regression

JEL classification: J24


Received: 17 Sep 2020


Citation: Globočnik Žunac, A., Zlatić, S., Buntak, K. (2021), “Freelancing in Croatia:

Differences among Regions, Company Sizes, Industries and Markets”, Business



110


Introduction Modern employment in the context of resource management in business is taking a

new form. Outsourcing has been a common form of choice for an efficient and

profitable business for a long time, and Marshall et al. (2015) talk about it as the

organizational policy that has a direct impact on the organizational structure. The

term which is frequently changed in use by synonyms to externalize or subcontract

refers to the business practice of hiring a person (or a company) to perform business

activities, which were usually done by a company’s employees. Kavčić et al. (2015a)

think that process of outsourcing fill gaps for the organizations have due to the lack

of needed resources. As well, they see this process as the key for a developmental

policy as the outsourcing decision is the base for a long-term competitive and

successful business. Outsourcing can be used as a measure to reduce costs, but also

as a solution when a highly skilled worker or professional is needed when the

company does not have them within their workforce (Cieslik, 2015).

Davis-Blake and Uzi (1993) implicate four factors are influencing the use of

outsourcing as a form of employment. These are the cost of employment, the

external environment, the size of the organization and the degree of bureaucracy,

and the skills required to perform work tasks. All mentioned highlights the importance

and significance of managing the relationship with this other, external side, of the

company where Piltan and Sowlati (2016) talk about issues springing from the need

of joint decision making, information sharing, taking joint risks, and developing trust

and commitment. These issues are the possible reasons for an organization not being

keen on externalizing its activities. Being bigger and used to sharing all mentioned

can be one of the possible factors of influence when deciding to externalize

activities. In favor of that, Kavčić et al. (2015b) say that companies are complex

social, economic, and technical systems and the very complexity is the reason to

choose the outsourcing model. They think that each participant in the outsourcing

relationship needs two features. First, they need to be specialized in their core

competencies but as well, they need to have the creativity and the ability to

cooperate.

This form of employment brings up in focus the freelancers. The term freelancer is

used for someone who is self-employed and works independently of a company,

works alone or with partners in occupations that are non-manual and require

specific knowledge and skills, e.g. software development, web design, translating

services, creative writing, business consultants, managers, and others (Kitching, 2015;

Kozica et al., 2014; Malaga, 2016; Kazia et al., 2014). However, in the very term

‘freelancer’ there is freedom engraved according to Henning (2011) these have

everything but freedom, as they are characterized by the constant pressure of doing

the present job activity and thinking about the ahead, projects and in the same time

having in focus the constant need of being networked. Freelancers are perceived as

high-skilled workers who can respond to the requirements of a company (Burke,

2015). Hoedemaekers (2020) talks about Stanworth and Stanworth's typology of

three reasons why someone becomes a freelancer: a person is unable to find in-

house jobs; a person is freelance by choice or due to some circumstances outside of

work. The same author states that according to Fraser and Gold (2001) some start

freelancing out of necessity but remain there by choice. This reason shows that the

root is not only in the characteristics of the person but as well, it might be in

environmental circumstances. In any case, the concept of freelancing is related to

the concept of independent professional, and it brings new features to the human

resource theory. Volberda (1998) sees a challenge for management in the need to

create a balance between job interchanges and retention of previously known,

111


taking into account the dynamics of operations, the results of intelligence gathering,

and the multi-directional causal connections between the employees and the

organization. But on the other hand, as being such an important resource, many

professionals would find the interest and embark on a venture of freelancing and

face a big challenge in the shape of the need to develop their market. To do that

they need to have adequate characteristics and knowledge which is not (mostly)

connected to their expertise.

Freelancers’ market does not exist before it is developed by the very employee.

The question of the market for freelancers arose as well as the need to find out

whether the employers recognized the opportunity. Therefore, the research

presented in this paper had the aim to get insight into the side of employers and find

out what impacts their choice to hire freelancers. The following factors of possible

influence are taken: region of where is the seat of the company, the size of the

organization, the market in which the company operates, and activities it is

engaged in. The research was conducted with the help of the Croatian Chamber of

Commerce and through their network all; the registered business entities were

targeted. Unfortunately, not all targeted responded which influenced the size of the

sample. The research instrument was developed and possible differences of the

independent variables in terms of hiring freelancers were analyzed, achieved by

multiple 2xc Fisher’s exact tests with Monte Carlo Simulation.

The paper introduces the topic and the review of relevant literature at the

beginning, which is followed by a detailed description of the research methodology

and statistical methods used. Results of the research according to the discussion of

one block of four hypotheses are presented in the second part of the paper. The

conclusion gives research limits and guidelines for further investigation of the topic.

Literature review In Croatia, a topic frequently mentioned in the media and expert roundtables is the

position of freelancers, i.e. not recognizing this form as one of the possible forms of

employment as well as the other legislative frameworks and support for freelancers.

Freelancers Union in the USA (since 1995) or the Association of Independent

Professionals and the Self-Employed (IPSE) in the UK (since 1999) have been there to

provide any help, advice, health insurance, education, or anything else a freelancer

or an independent professional might need. In Croatia, the Croatian Independent

Professionals Association (CIPA) has been providing Croatian freelancers with the

same support since 2013.

The main feature of freelancers is independence. Kozica et al. (2014) simply

define freelancers as individuals who do some business as independent contractors

while Shepard (2018) thinks that this independence enables independent

professionals to work on more than one project at the same time, but they are free

to decide on the number of companies they are selling their knowledge and skills to,

on the time they will spend on their work, and on the place where they will do their

job, which is often in their home or a co-working space. The question is whether there

is a market for multitudinous projects in Croatia for any type of freelancers’ activity

and in any region of Croatia or is this opportunity reserved for only those operating in

the big business developed zones or large cities. Not recognizing freelancing by

economic policymakers as a possibly good option leads to self-employment in forms

such as one-man-band, or starting a company with only one employee. Often these

are Home-Based Businesses (HBBs) which Mason et al. (2010) named insignificant

businesses which as such is often undervalued. Kitching and Smallbone (2012) think

that freelancing as a form of employment is neglected by researchers of small

112


business enterprises and they agree that it should be considered differently not as

part of SME. Freelancers as a large subset of the small business population are an

interesting target group and attention to them should be paid. Mason et. al (2010)

consider that this form of work fosters distinctive geography of the area as rural and

non-metropolitan parts.

Home becomes a central location for practicing labor and according to the

Labor Force Survey, 3.1 million people are working mainly from home in the United

Kingdom which is 11% of the workforce (Ruiz and Walling, 2005). This type of

employment prescribes a different approach to business and entails entrepreneurial

assumptions, which often complicates the life and business of individuals who could

solve their problems as freelancers. New technologies that create networks of young

people and professionals, but also a high level of specialization in certain areas, on

the other hand, affect the new nature of employment. Some authors see

freelancing as a future trend especially with the development of these technologies

that enable work done from any distant place. According to Gheorghe (2015),

Croatia has a 0.20-market share of freelancing and is at 23rd position on the list were

the first in India with 34.19 market share or the USA which is with 14.15 at the second

place. Countries that are closer to Croatia as Macedonia and Bulgaria have a

similar share while Serbia has a market share of 1.08. Beck (2000) examines how work

has become unstable in the modern world and presents a new vision for the future.

He states that there are no longer safe job paths and predicts the end of traditional

working practices. According to him, new models include forms of self-organizing,

which is necessary to provide equal access to comprehensive social protection.

These changes in access to employment will lead to freeing individuals to become

authors of their own 'do-it-yourself' biographies as work is 'chopped up by time and

by contract.

A complete insight into the situation on the labor market in Croatia from the

aspect of self-employment shows factors that influence freelancers’ opportunities. As

such, it speaks of the possibilities concerning their activities and provides significant

prerequisites for business planning of freelancers, but also guidelines for the future

definition of economic employment policy and the prerequisites for smooth

operation and development of freelance activities. Bolarić Škare (2012) presented

self-employment benefits that vary in percentage in different counties in Croatia and

according to her data; the highest percentage is in the big urban and industrial

towns, which could be the standpoint to the conclusion that the place where the

organization is situated can be of influence when subcontracting the activity

performed by a freelancer. The market for activities of freelancers in Croatia

according to the research presented in this paper is not limited by the size of the

company nor by where it is situated in Croatia, which speaks of equal opportunities

throughout the country. A significant influence on the freelancers’ opportunity has

the market in which the company operates and the activity the company is

engaged which is following the results of a survey conducted by Kavčič et al.

(2015a) in Slovenia. They say that the importance of particular services varies from

organization to organization considering the particular segment to which the

organizations belong.

Kitching and Smallbone (2012) discuss the term of freelancing and agree with

many other researchers (Storey et al., 2005; Holgate and McKay, 2009; Moeran, 2009,

according to Kitching and Smallbone, 2012) who use it as for independent workers in

creative and media occupations, though some authors mostly see them in the field

of software development, IT and engineering, while Leighton and Brown (2013) unite

the concept and talk mostly about journalists, designers, IT experts, and different

113


consultants. Academic researchers typically use the term freelance to refer to own-

account workers in creative and media occupations, including journalism television

and radio, film, publishing, public relations, translation services, and artists (ibid), and

while so Kitching and Smallbone (2012) discuss own-account workers in managerial,

professional, scientific and technical occupations that should also be treated as part

of the freelance workforce. Solicitors working for personal clients are excluded from

this but those serving both organizational and personal end-users might be defined

as freelance. At the beginning of investigating freelancers, the researchers focused

on a narrow set of occupations and sectors. Expansion of employment

externalization brought in other jobs and professions. The reason for this expansion

Beardwell et al. (2004) sees in the competitive labor market. Johnson and Ashforth

(2008) call it the 'paradox of externalization' as they argue the problem of relying on

external employees in the process of connecting with customers and building

customer-oriented services while Beardwell et al. (2004) talks about the new feature

of employment as 'not providing commitment'. Externalization of employment

through outsourcing was the research topic of Harland, et al. (2005) who view it as a

term not only used for support services but also activities 'closer to core' ones.

Cheesley (1997 according to Harland et al., 2005) that in the UK hospital trusts and

health authorities outsourced clinical support services, e.g. provision of sterile supplies

and patient appliances, and parts of occupational health, pathology, radiology,

and pharmacy services. This is evidence that outsourcing is equally relevant for

“core” as well as “non-core” activities.

Methodology Research instrument Results presented in this paper are part of the much broader survey on freelancers

for which a special research instrument in form of a questionnaire was developed

(Table 1). The main purpose of this research is to explore the differences in the

decision-making of freelancer’s employment. Research questions examined in this

paper are shown in the table below. Dependent variable measures if the company

hires freelances.

Table 1

Research instrument description

Item Modalities

Have you ever hired a freelancer in

your company?

Yes, no

Which county is the location of

headquarter of your company?

Country codes

In which category does your

company belong (according to the

Accounting Act (NN 78/15, 134/15))?

Micro, medium, and large

In what market does your company

operate?

Locally, regionally (more than two counties), Croatia,

countries around Croatia, European Union, or worldwide.

What does your company do?

Individual and small series production, process industry,

service activities, and logistics, public administration

services, services of utility and public companies,

independent profession (lawyers, dentists, freelance artists),

agriculture and fisheries, education, or other (specify).

Source: Author’s work

114


Data Data collection is achieved by distributing a questionnaire, conducted online

through the survey tool esurveycreator.com. The target population consists of small,

medium, and large companies based in Croatia. An e-mail with the link to the web-

based questionnaire was sent to the companies in Croatia by the Croatian

Chamber of Commerce. During the time interval between the 4th and the 25th of

June 2018, the survey was completed by 158 respondents.

Statistical methods Research questions are evaluated with the construction of a binary logistic

regression model, where the independent variables are the region where the

company is seated, the size of the organization, the market on which it operates,

and the type of its activity. The dependent, binary outcome variable of the model is,

whether the given company has hired a freelancer (yes/no). Outliers are searched

iteratively and those cases, in which standardized residuals display greater than 2

standard deviations, are excluded from the model. Model fit is interpreted with chi-

square goodness-of-fit statistic and the Hosmer and Lemeshow goodness-of-fit test

(Hosmer Jr., Lemeshow, & Sturdivant, 2013). For the interpretation of the model’s

explained variance, Nagelkerke’s R2 will be evaluated (Nagelkerke, 1991), which can

reach a theoretical maximum of 1, contrary to Cox and Snell’s R2 (Field, 2018).

In the conducted logistic regression model, the paper used indicator dummy

coding for the categorical predictors with the first category as the reference. In this

manner, the City of Zagreb served as the reference category of region, “micro” held

the place of reference by company size, while local firm became the reference

category of market operation and individual and small series production were

present as the reference category of operational type.

The statistical significance of the model’s independent variables is interpreted with

their Wald test results (z2) to determine, whether their addition significantly improves

the model’s predictive ability (Field, 2018). Since the B coefficients of the predictors

only show the log odds of occurrence for a one-unit change in the respective

independent variable, if all other independent ones are kept constant, the paper

also notes the exponential B coefficients (eB) with their 95% confidence intervals for

the change of odds (Field, 2018).

Validity The discriminative ability of the model is presented with a classification table

including the model’s sensitivity (i.e., the percentage of cases where the positive

outcome was classified correctly) and its specificity (i.e., the percentage of cases

where the negative outcome, in this case not hiring a freelancer, was classified

correctly)(Steyerberg, 2009). Additionally, the model’s positive and negative

predictive values are calculated and discussed.

The accuracy of the constructed binary logistic regression model is interpreted

with the Receiver Operating Characteristics (ROC) curve to assess the model’s ability

to correctly classify cases (i.e., its discriminating ability) in the determination of hiring

a freelancer (Steyerberg, 2009). The model’s sensitivity and specificity are discussed

and interpreted, followed by the summary of the constructed ROC curve with the

area under the curve (AUC), which is equivalent to its concordance probability

(Gönen, 2007). The AUC, which represents the ability of the model to discriminate

between the two outcome possibilities of hiring a freelancer, ranges from 0.5 and 1.0

and is interpreted according to the general guideline of Hosmer and colleagues

(Hosmer Jr. et al., 2013, p. 177).

115


Results From the 158 respondents of the study, 44.3% were male (n = 70) and 48.7% female

(M = 77), while further 11 respondents did not indicate their gender. Only 1

respondent (0.6% of the sample) stated age between 18 and 25 years, further 20.3%

of the respondents (n=32) marked the group of 26-35 years of age. The age groups

36-45 and 46-55 were both marked by 29.1% of the sample, 46 respondents each.

The age group of 55 years or older was stated by 20.9% of the sample (n=33

participants).

The highest number of companies were sampled in the City of Zagreb (n=51),

while the least participants were from the region of Central Croatia (n=15). The

percentage proportion of hiring freelancers compared to all respondents in the

respective region was the highest in the City of Zagreb (n=36, 70.6% of the sampled

companies hired one already), while the lowest proportional percentage was found

in Northern Adriatic and Lika region (n=10, 41.7%). The following table (Table 2)

represents the regional comparisons of hiring freelancers in Croatia.

Table 2

Number of companies in different Croatian regions according to freelancers’ hiring


Table 3 depicts the number of freelancers hired in companies of different sizes. The

majority of the sampled respondents were employees in micro firms (n=111, 71.6% of

the sample), while the least amount of respondents worked in medium (n=11, 7.1%)

and large companies (n=11, 7.1%). The proportional percentage of hired freelancers

was the highest in medium firms (n=7, 63.6%), while the lowest percentage was found

in the case of little companies (n=10, 45.5%).

Table 3

Number of companies in the different company according to freelancers’ hiring


Companies that hire

freelancers

Companies that do not hire

freelancers

Region n % n %

Central and Southern Adriatic 11 64.7% 6 35.3%

Northern Adriatic and Lika 10 41.7% 14 58.3%

Eastern Croatia 12 63.2% 7 36.8%

Northwestern Croatia 14 46.7% 16 53.3%

Central Croatia 9 60% 6 40%

City of Zagreb 36 70.6% 15 29.4%

Total 92 59% 64 41%

Fisher test p = 0.145

Companies that hire

freelancers


freelancers

Company size n % n %

Micro 68 61.3% 43 38.7%

Little 10 45.5% 12 54.5%

Medium 7 63.6% 4 36.4%

Large 6 54.5% 5 45.5%

Total 91 58.7% 64 41.3%

Fisher test p = 0.535

116


The number of freelancers hired is analyzed according to different operational

areas as well (Table 4). The majority of the respondents stemmed from firms that

operate within Croatia (n=41, 26.3%), while the least percentage of respondents are

employed in companies that operate worldwide. The highest proportional

percentage of hired freelancers was found in worldwide companies (n=15, 93.8% of

respondents from worldwide companies hired a freelancer already), while the least

percentage of firms who hired freelancers was found in companies that operate

locally (n=6, 24%).

Table 4

Number of companies in different operational markets areas according to

freelancers’ hiring


There are significant differences in the employment of freelancers in different

operational types according to the presentation in Table 5. The majority of

respondents were employed in companies engaged in the business area of service

activities and logistics (n=89, 58.9%), whilst the least amount of participants stemmed

from public companies and those that specialized themselves for services of utilities

(n=2, 1.3%). The proportional percentage of hired freelancers was found in the case

of independent professions such as lawyers, dentists, and free artists (n=8, 88.9%),

while no freelancers were employed in the sampled companies of services of utility

and public companies as well as agriculture and fisheries.

Table 5

Number of companies in different types of activity according to freelancers’ hiring


Companies that hire

freelancers


freelancers

Operational markets n % n %

Locally 6 24% 19 76%

Regionally (more than two

countries)

8 40% 12 60%

Croatia 25 61% 16 39%

Countries around Croatia 13 72.2% 5 27.8%

European Union 25 69.4% 11 30.6%

Worldwide 15 93.8% 1 6.3%

Total 92 59% 64 41%

Fisher test p ˂ 0.001

Companies that hire

freelancers


freelancers

Type of activity n % n %

Individual and small series

production

23 71.9% 9 28.1%

Process industry 6 66.7% 3 33.3%

Service activities and logistics 46 51.7% 43 48.3%

Services of utility and public

companies

0 0% 2 100%

Independent profession

(lawyers, dentists, free artists)

8 88.9% 1 11.1%

Agriculture and fisheries 0 0% 3 100%

Education 5 71.4% 2 28.6%

Total 88 58.3% 63 41.7%

Fisher test p ˂ 0.001

117


2 x c Fisher’s exact tests with Monte Carlo simulation (2000 replicates) were used

to determine whether the proportions between the groups of the independent

variables (region of the company’s seat, size, area of operation, and operation type)

are significantly different in terms of hiring a freelancer. Since Fisher’s exact tests are

omnibus, pairwise group comparisons will be made during post hoc tests with

Benjamini-Hochberg adjustment.

In the case of the different regions where the seat of the company is, there were

no significant proportional differences in terms of hiring a freelancer (p = 0.145). The

analysis of different company sizes (micro, little, medium, large) and the difference in

terms of their hiring a freelancer was not proportionally significant either (p = 0.535).

However, significant differences in the categories of company operation area were

observable in terms of hiring a freelancer (p < 0.001). The post hoc test revealed that

there were significant proportional differences in terms of hiring a freelancer

between locally based firms and Croatia-wide companies (p = 0.015), those that

reside in the countries around Croatia (p = 0.009), the European Union (p = 0.005)

and the whole world (p < 0.001). Furthermore, regional firms differed also significantly

from worldwide companies.

Significant differences were also observable in the proportions of types of activity

in terms of hiring freelancers (p < 0.001). The post hoc test revealed that there were

two significantly different pairs in this variable: companies with the profile of

agriculture, fisheries, and public sector differed significantly from firms in the industrial

and small series production (p = 0.035) and companies in independent professions,

such as lawyers, dentists and free artists (p = 0.035).

After the analysis or Fisher’s exact tests, the binary logistic regression was

conducted. Casewise outliers of the model were detected iteratively in three stages.

In the first stage, 5 cases were excluded, followed by additional 4 cases in the

second stage and 2 cases in the third stage.

During the baseline analysis, the model without independent variables, only

including the constant was evaluated. Without the inclusion of the independent

variables, the model assumes that all firms hired freelancers, thereby correctly

classifying 58.8% of the cases.

The logistic regression model was statistically significant, determined by omnibus

Chi-square statistics, 𝜒2(19) = 120.002, 𝑝 < 0.001. Furthermore, the Hosmer and

Lemeshow goodness of fit test was used to determine the adequacy of the model in

terms of how well it performs in predicting the categorical outcome of hiring a

freelancer. The Hosmer and Lemeshow test was not statistically significant with a

result of 𝜒2(8) = 4.292, 𝑝 = 0.830, indicating a good fitting model. Explained variation

was evaluated through Nagelkerke R2, whereby the model explained 77.5% of the

variance.

Table 6

Classification table of the effectiveness of the predicted classification against the

practical one


Have you ever hired a freelancer in your

company?

Observed No Yes % Correct

Have you ever hired a freelancer in your

company?

No 41 22 65.1

Yes 17 70 80.5

Overall Percentage 74.0

118


The effectiveness of the predicted classification against the practical classification

in terms of the data on hand (i.e., observed classification) was assessed to evaluate

the prediction and its accuracy including the independent, predictor variables as

well. With the addition of the independent variables, 74.0% of the cases could be

classified correctly. In comparison with the classification table, containing only the

constant of the model, where 57.8% of the cases could be classified correctly, it can

be concluded that with added independent predictors, the model improved in the

overall prediction of cases by 16.2%.

Sensitivity was present with 80.5% accuracy, while specificity was present with

65.1% accuracy. The positive predictive value of the model is 100 ∗ (70 ÷ (22 + 70)) =

76.09%, which means, that from all cases where it was predicted that a freelancer

would be hired, 76.09% were correct, while the negative predictive value is 100 ∗

(41 ÷ (17 + 41)) = 70.69%, representing the correctly predicted percentage of all

cases where no hired freelancer was predicted.

The event of interest, which in the case of the present paper is hiring a freelancer,

was determined as the positive actual state of the ROC curve, indicating that it was

stated correctly. The area under the ROC curve was 0.780 (95% CI, 0.706 to 0.855),

which can be determined as good discrimination, based on the general rules of

thumb of Hosmer and colleagues (Hosmer Jr. et al., 2013, p. 177).

Figure 1

Representation of the conducted ROC Curve


Table 7 presents the results of the binary logistic regression model. Table 7 presents

the predictors of the constructed binary logistic regression model predicting the

likelihood of hiring a freelancer based on the region of the company’s seat,

company size, area, and type of operation

119


Table 7

Binary logistic model

B S.E. Wald df p

Region 13.717 5 0.018**

Central Croatia 5.581 2.305 5.861 1 0.015**

Northwestern Croatia -1.757 0.863 4.146 1 0.042**

East Croatia 2.368 1.402 2.853 1 0.091*

North Adriatic and Lika -1.713 1.066 2.582 1 0.108

Middle and South Adriatic 4.566 2.188 4.356 1 0.037**

Company size 11.584 3 0.009***

Little -6.601 1.974 11.177 1 0.001***

Medium 0.516 1.167 0.196 1 0.658

Large -25.290 6860.678 0.000 1 0.997

Area of operation 14.733 5 0.012**

Regionally (more than two counties) 7.157 2.843 6.339 1 0.012**

Croatia 10.030 3.306 9.207 1 0.002***

Countries around Croatia 12.196 3.502 12.128 1 0.000***

European Union 11.158 3.338 11.177 1 0.001***

Whole world 54.232 9577.221 0.000 1 0.995

Type of operation 10.844 6 0.093*

Process industry -6.768 2.701 6.281 1 0.012**

Service activities and logistics -11.582 3.865 8.982 1 0.003***

Services of utility and public companies -11.170 3.560 9.847 1 0.002***

Independent professions (lawyers, dentists, etc.) -24.090 26710.717 0.000 1 0.999

Agriculture and fisheries -1.959 2.025 0.936 1 0.333

Education -35.313 19837.403 0.000 1 0.999

Constant 1.979 1.464 1.826 1 0.177

Note: The dependent variable of the model is whether the given company have

hired a freelancer (yes/no); *** statistically significant at 1%; ** 5%; *10%


Based on the constructed model, three of the hereby-used independent

constructs were found to be significant predictors of hiring freelancers: the region of

the company’s seat, company size, and area of operation.

The region of the company’s seat has a significant influence on hiring a freelancer

(p = 0.015). The City of Zagreb was set as the reference category of this variable,

thus, the odds of hiring freelancers in other regions were compared to companies

residing in the City of Zagreb. The results show that the odds of hiring a freelancer

was significantly higher in the regions “Central Croatia” (B = 5.581, p = 0.015) and

“Middle and South Adriatic” (B = 4.566, p = 0.037) than in Zagreb City, however, the

likelihood of hiring a freelancer was significantly lower in the region of “Northwestern

Croatia” (B = -1.757, p = 0.042).

The size of the company is a significant predictor of hiring a freelancer according

to the results of the performed binary logistic regression model (p = 0.009). Micro

companies were appointed as reference category of this predictor and the analysis

revealed that little companies have significantly fewer odds of hiring freelancers

than micro ones (B = -6.601, p = 0.001).

Table 7 interprets the odds ratios for the predictors as well (see column Exp(B)),

which represent the exponentiation of the analyzed coefficients. These estimates tell

about the amount of increase/decrease in the odds of the dependent variable (i.e.,

hiring freelancers) when increasing the respective predictor with 1 unit, holding the

other ones constant.

120


As the model shows, the operational area of companies is also a significant predictor

of hiring a freelancer (p = 0.012). The reference category of this variable was local

firms and as the model shows, the odds of hiring a freelancer is significantly higher in

all wider operational areas: regional firms operating in more than two counties (B =

7.157, Exp(B) = 1283.336, p = 0.012), Croatia-wide companies (B = 10.030, Exp(B) =

22694.988, p = 0.002), firms that operate in countries around Croatia as well (B =

12.196, Exp(B) = 198086.893, p < 0.001) and those that operate EU-wide (B = 11.158,

Exp(B) = 70138.516, p = 0.001). The highest odds compared to local firms for hiring

freelancers were thereby shown to be companies that operate also in countries

around Croatia.

The fourth construct of the model, the operation type, did not have a significant

influence on hiring a freelancer (p = 0.093).

Conclusion The present study aimed to discover key company-related characteristics of hiring

freelancers, considering factors such as the region of the sampled companies’ seats,

the size of the firms, their operational areas, and operation types. This paper aimed

to explore factors that might influence hiring a freelancer; research questions were

proposed accordingly, each of them aiming to discover the influence of a specific

factor (i.e., region, where the seat of the company is, the size of the organization, the

market on which the company operates, and the type of company activity) on the

employment of freelancers. To meet this goal, the binomial logistic regression

analysis was conducted.

The present paper used primary data acquired from a population of small,

medium, and large companies based in Croatia. An online questionnaire was used

as a research instrument, and an e-mail with the link containing this survey was sent

to respondents by the Croatian Chamber of Commerce. As a result, n=158

respondents were evaluated in the present study. Based on the considerations

about the operationalization of this research, the data collection was limited to

companies residing in Croatia.

Elements of the research questions proposed that the discussed independent

variables are significant predictors of hiring a freelancer. RQ1 proposed that the

region where the company’s seat is, has a significant influence on hiring a

freelancer, which was accepted (p = 0.015), so was RQ2 stating that the size of the

company is a significant predictor in this relationship (p = 0.009) and RQ3 proposing

that the area of operation is a significant predictor in this model (p = 0.012). RQ4,

setting the type of operation as a significant predictor was, however, rejected (p =

0.093). The authors should note the limitations of the study.

Research with its conclusions provides the standpoint for the freelancers for

evaluation and judgment on their business possibilities. These results may give not a

significant value to the companies but can open the question of considering that

workforce and encourage the sustainability of their business model. In the end, there

is the third cornerstone that needs to be taken into account, the national legislative

part that can use the conclusions when designing developmental policies to

encourage employment and especially self-employment as this is the often-

discussed topic. Here the policy designers try to create benefits for those who self-

employ but as well, the benefits for the companies who subcontract freelancers can

be taken into consideration. The scientific presentation given in this paper provides

guidelines for practitioners.

The presently discussed results regarding influential factors in terms of freelancer

employment need to be generalized first at the country level, in Croatia. Future

121


research should therefore aim to collect data, which meets the requirements of

representatively in Croatia, and verify, whether the hereby interpreted effects truly

reflect freelancer employment on a national level. Further research is required in

terms of country comparison as well, to provide indications for possible international

generalisability, or explore the existence of possible differences in this regard.

References 1. Arsov, S., and Naumoski, A. (2014). “Transparency and Disclosure Practices in Selected

Post Transition Economies”. European Journal of Business and Management, Vol. 6, No. 16,

pp.146-154.

Beck, U. (2000), The Brave New World of Work, Cambridge, Polity Press.

Beardwell, I., Holden, L., Claydon, T. (2004), Humane Resource Management – A

Contemporary Approach, 4th Ed., Prentice Hall FT, Pearson Education, Harlow, UK.

Bolarić Škare, M. (2012), “Entrepreneurship and Self-employment - Examples of Good

Practice”, Učenje za poduzetništvo, Vol. 2 No. 2, pp. 303-310.

Burke, A. (2015), ”Introduction: A Freelancing and Self-Employment Research

Agenda”, in Burke A. (Ed.) The Handbook of Research on Freelancing and Self-

Employment, Senate Hall Academic Publishing, Dublin, pp. 3-8.

Cieślik, J. (2015), ”Capturing Statistically the “Intermediate Zone” Between the

Employee and Employer Firm Owner”, in The Handbook of Research on Freelancing

and Self-Employment, A. Burke, Ed., Dublin, Senate Hall Academic Publishing, pp. 29-

38.

Davis-Blake, A., Uzzi, B. (1993), ”Determinants of Employment Externalization: A Study

of Temporary Workers and Independent Contractors”, Administrative Science

Quarterly, Vol. 38 No. 2, pp. 195-223.

Field, A. (2018), Discovering Statistics using IBM SPSS Statistics (5th ed.), SAGE

Publications Ltd., London, UK.

Gheorghe, M. (2015), State of Freelancing in IT and Future Trends, World Academy of

Science, Engineering and Technology International Journal of Social, Behavioral,

Educational, Economic, Business and Industrial Engineering Vol. 9 No. 5., p. 31252.

Gönen, M. (2007), Analyzing Receiver Operating Characteristic Curves With SAS (1st

ed.), Cary, NC, SAS Publishing.

Harland, C., Knight, L., Lamming, R., Walker, H. (2005), “Outsourcing: assessing the risks

and benefits for organisations, sectors and nations”, International Journal of

Operations & Production Management, Vol. 25 No. 9, pp. 831-850.

Henning, G. K. (2011), “Corporation and Polis”, Journal of Business Ethics, Vol. 103 No.

2, pp. 289–303.

Hoedemaekers, C. (2020), “Selling themselves: conceptualising key features of

freelance work experience”, Culture and Organization,

https://doi.org/10.1080/14759551.2020.1833206.

Hosmer Jr., D. W., Lemeshow, S., & Sturdivant, R. X. (2013), Applied Logistic Regression

(3rd ed.), Hoboken, NJ, USA: John Wiley & Sons.

Johnson, S.A., Ashforth, B.E. (2008), “Externalization of Employment in a Service

Environment: The Role of Organizational and Customer Identification”, Journal of

Organizational Behavior, Vol. 29 No. 3, pp. 287-309.

Kazia, A. G., Yusoffb, R. M., Khana, A., Kazib, S. (2014), “The Freelancer: A

Conceptual Review”, Sains Humanika, Vol. 2 No. 3, pp. 1-7.

122


Kavčić, K., Bončina, K., Suklan, J., Milost, F. (2015a), “Development design of

outsourcing strategies: The case of Slovenia”, Ekonomski pregled, Vol. 66 No. 5, pp.

455-476.

Kavčić, K., Gošnik, D., Beker, I., Suklan, J. (2015b), “How does Logistics Outsourcing

Influence Organisation Performance?”, International Journal of Industrial Engineering

and Management (IJIEM), Vol. 6 No. 3, pp. 101-107.

Kitching, J., Smallbone, D. (2012), “Are Freelancers A Neglected Form Of Small

Business?”, Journal Of Small Business and Enterprise Developmnet, Vol.19 No. 1, pp.

74-91.

Kitching, J. (2015), “Tracking UK Freelance Workforce Trends 1992-2015”, in Burke A.

(Ed.) The Handbook of Research on Freelancing and Self-Employment, Dublin,

Senate Hall Academic Publishing, pp. 15-28.

Kozica A., Bonss, U., Kaiser, S. (2014), “Freelancers and the absorption of external

knowledge: practical implications and theoretical contributions”, Knowledge

Management Research and Practice, Vol. 12 No. 4, pp. 421-431.

Leighton, P., Brown, D. (2013), “Future working: The rise of Europe’s independent

professionals (iPros), European Forum of Independent Professionals. Available at

https://apo.org.au/node/201721 / (15 Mar 2021)

Malaga, R. A. (2016) , “Digital Micro Entrepreneurship a Preliminary Analysis - Case

studies from the gig economy”, Journal of Business and Economics, Vol. 7, No. 9, pp.

1469-1476.

Marshall, D., Ambrose, E., McIvor, R., Lamming, R. (2015), “Self-interest or the greater

good: How political and rational dynamics influence the outsourcing process”,

International Journal of Operations and Production Management, Vol. 35 No. 4, pp.

547-576.

Mason, C., Carter, S., Tagg, S. (2010), “Insignificant Businesses: the characteristics of

home-based businesses in the United Kingdom, Regional Studies, Vol. 45 No. 5, pp.

625-639,

Nagelkerke, N. J. D. (1991), “A Note on a General Definition of the Coefficient of

Determination”, Biometrika, Vol. 78 No. 3, pp. 691–692.

Piltan, M., Sowlati, T. (2016) , “Multi-criteria assessment of partnership components”,

Expert Systems with Applications, Vol. 64, pp. 605-617.

Ruiz, Y., Walling, A. (2005), “Home-based working using communication technologies

Recent trends in teleworking, using data from the Labour Force Survey”, Labour

market trends October, pp. 417–426.

Shepard, J. M. (2018), “Understanding Co-Working with The Growth and

Development of Freelancers”, Management and Organizational Studies, Vol. 5, No.

2, pp. 1-6.

Steyerberg, E. W. (2009), “Clinical Prediction Models”. International Statistical Review.

New York, NY: Springer New York.

Volberda, H.W. (1998), Building the Flexible Firm: How to Remain Competitive, Oxford

University Press, New York.

123


About the authors

Ana Globočnik Žunac is an Assistant professor at University North with Ph. D in

information-communication sciences. She was executive director of HEI, and later

assistant of vice-rector for scientific work and international affairs at University North.

She was a visiting lecturer at University in Prague and University in Tirana. She took

part in many conferences and has numerous publications. She was twice awarded

for the best paper and was praised for the exceptional contribution to the

development of the Department of Communication and PR. The author can be

contacted at [email protected]

Sanja Zlatić graduated from the Faculty of Natural Sciences and Mathematics in

Zagreb in the field of financial and business mathematics. She is a lecturer of

mathematics and operational research at University North in Croatia. She is a Ph.D.

student at the International Joint Cross-Border Ph.D. Programme in International

Economic Relations organized and administered by the Consortium of Universities in

Bratislava, Sopron, Pula, Varaždin, Mostar, and Prague and collaboration with the

University of Applied Sciences Burgenland (UAS), Eisenstadt, Austria. She was

awarded the best speaker at the SGEM Conference on Social Sciences and Arts

2018 - Albena, Bulgaria. The author can be contacted at [email protected]

Krešimir Buntak is a professor at University North. After finishing the University of

Engineering and Shipbuilding in Zagreb, he gained his Ph.D. in economics. During his

education, he worked over 20 years in management positions from the public and

local governments, through the private sector to large export organizations in the

automotive industry. Since 2004, he has taught at various national and international

schools and colleges about managing, management, organization, business

economics, BPM, quality management systems, and business excellence. He is the

author of more than 100 scientific and technical papers and one polytechnic

textbook. The author can be contacted at [email protected]

mailto:[email protected]



124


Familiarity with Mission and Vision: Impact

on Organizational Commitment and Job

Satisfaction

Dunja Dobrinić, Robert Fabac

University of Zagreb, Faculty of Organization and Information, Croatia

Abstract

Background: The relationship between organizational mission and vision statements,

organizational commitment, and job satisfaction has been discussed vastly in previous

research, both in the domain of public sector organizations and in profit organizations.

Objectives: The goal is to investigate if there are differences in organizational

commitment and job satisfaction between employees who are familiar with the

mission and vision of their organization, compared to those who are not familiar with

them. Methods/Approach: A survey research has been conducted on a sample of

114 employees in private and public sector organizations in the Republic of Croatia.

Data were analysed using a t-test to determine the differences between two groups

of respondents, i.e. those who are familiar with the visions and mission of their

organisation, and those who are not. Results: There are differences in job satisfaction

levels between employees who are familiar with the mission and vision of the

organization in which they are employed and those who are not. Furthermore,

differences are particularly evident in the group of public sector employees.

Conclusions: The presence of awareness of the organizational mission and vision

among employees has a positive effect on their job satisfaction. This is possibly an

indicator of the organization's culture, which fosters positive values embedded in the

organizational vision and mission.

Keywords: job satisfaction; organizational commitment; mission; vision; public sector;

private sector; Republic of Croatia

JEL classification: L20


Received: 05 May 2020


Citation: Dobrinić, D., Fabac, R. (2021), “Familiarity with Mission and Vision: Impact on

Organizational Commitment and Job Satisfaction”, Business Systems Research, Vol.

12, No. 1, pp. 124-143.


125


Introduction The mission represents a brief statement by which an organization explains the reason

for its existence. Based on the organization’s mission statement, its strategic goals are

formulated in practice. Mission statements have been seen as a critical strategic

management tool in recent years, while the mission itself is perceived as an asset for

public sector organizations (Mullane, 2002; Wright and Pandey, 2011).

On the other hand, the vision can be conceived of as a notion, that is, a projection

of a future state or event, or as a long-term result which is actualized by the

identification of problems on the part of employees and the resolution thereof (Buble,

2000). The vision is an idea of a certain ideal future of the organization. The employees’

familiarity with the organization’s mission and vision can have an impact on the

employees’ attitude towards the organization. That attitude will be under the

influence of job satisfaction and organizational commitment. Job satisfaction denotes

the degree of individuals’ satisfaction with the job they are currently performing within

the organization, while organizational commitment refers to individuals’ commitment

to the company that they work for.

The central theme in this paper is the relationship between the variable of

employee familiarity with organizational mission and vision with job satisfaction (JS)

and organizational commitment (OC) variables. Most of the commonly cited

highlighted previous research on the relationship between the mission and the JS

and/or OC variables was focused on the public sector. The mission itself has most

widely been examined in terms of defining the mission statement construct. One rare

exception is the doctoral thesis by Clark (2006) which aimed to establish the

correlation between mission statement familiarity and job satisfaction, and some other

variables of organizational behaviour. The author successfully proved the thesis about

a significant connection between mission statement familiarity and job satisfaction.

One of the studies that focus on the concept of the mission statement was

conducted in the Canadian public sector by Bart (2004) to determine the

management's awareness of the mission statement. The results indicated a positive

relationship between mission awareness and employee commitment. Krueger et al.

(2002) researched JS level predictors and concluded that this multidimensional

construct is determined by some predictors which are, however, organization and

context-specific. In that respect, the statement "Believes the organization carries out

its Mission Statement" was the highest-ranked predictor.

Furthermore, in most recent research on the relationship between organizational

mission and OC and/or JS variables, OC and JS have been seen through their

"mediating role”. As a result, these studies do not focus on the topic of knowledge or

ignorance of the mission among a certain portion of respondents, nor is familiarity an

important research issue. Instead, they are based on structural equation models,

where mission statements are presented with multiple indicators as an exogenous

variable (see, e.g. Kwong and Wong, 2013). The assumptions of the impact of mission

statements on organizational performance are then tested, wherein the broader

model also addresses the issue of the relationship between mission and JS and/or OC

variables (Figure 1).

In the paper by Yazhou and Jian (2011), Chinese non-profit organizations were

investigated using structural equation modelling, whereby one of the hypotheses (i.e.

"mission statements are positively related to job satisfaction") was successfully proven.

Furthermore, in their research within the Portuguese nonprofit sector, Macedo et al.

(2016) set up a model that assumes a mediating role of organizational commitment in

126


explaining the relationship between mission and organizational performance. One of

their main hypotheses that were confirmed is about a significant and positive

relationship between mission statements and organizational commitment (Figure 1).

Figure 1

Structural models for analysing relations between mission statements and

organizational performance

MISSION STATEMENTS

JOB SATISFACTION

ORGANIZATIONAL PERFORMANCE

MISSION STATEMENTS

ORGANIZATIONAL COMMITMENT

ORGANIZATIONAL PERFORMANCE

Source: Based on models presented in Yazhou and Jian (2011) and Macedo et al. (2016).

In the mentioned structural equation models, the employee's knowledge about the

mission or familiarity with the mission is not questioned. In contrast, Glassdoor – one of

the world's largest job and recruiting sites – published a study of Glassdoor's Mission &

Culture Survey (2019) based on a survey conducted online among adults from the

United States, UK, France, and Germany. This study points to the importance of

organizational culture and mission statements in employee recruitment and retention

processes. When the answers by the employed respondents from the sample are

concerned, one of the key conclusions is that "mission is one of the main reasons that

64 percent of employees stay in their job" (Glassdoor, Inc., 2019, p. 2).

Furthermore, a large body of research has focused on the comparison between

the public and the private sector in terms of the connection between mission and JS

and/or OC. Wright and Pandey (2011) proved the importance of "mission valence" by

illustrating its effect on two important human resource outcomes – job satisfaction and

absenteeism. Regarding mission valence, if a certain mission is viewed as valuable,

interesting, and attractive by a large number of persons, it can be expected that

owing to such mission, the organization (in the state sector) will lure quality and gifted

individuals into its ranks (Pandey et al., 2008). Organizations in the public or state sector

are focused on meeting the needs and interests of citizens, and on serving the

population in a certain community, city, or country. It can be argued that a mission

that sets out to accomplish such goals and tasks is harmonized with personal values

that individuals – the organization’s employees – strive for (Rainey and Steinbauer,

1999; Wright, 2007).

The fact that the mission statement has a more important role in non-profit

organizations than for-profit organizations was also established by Bart (2007).

Furthermore, a significant difference between the private and the public sector

regarding OC and JS was addressed in research into Greek public sector

127


organizations (Markovits et al., 2010). Considering the expected levels of OC and JS,

the authors pointed out the difference between periods of economic prosperity and

recession in a particular country. In prosperity conditions, the private sector is a better

employer and “private-sector employees are more extrinsically satisfied than civil

servants and more organizationally committed" (Markovits et al., 2010, p. 9). However,

in periods of economic recession, the expected results are opposite – civil servants

become more extrinsically satisfied and more committed than employees in the profit

sector organizations (Goulet and Frank, 2002).

The research gap we identified in our review of previous research refers to the lack

of studies that would specifically address the impact of employees’ familiarity with

organizational mission and vision on job satisfaction and organizational commitment.

The next valuable feature of our research refers to the geographical, social,

economic, and cultural context in which it is conducted, considering that the relations

among organizational variables established in research conducted in Greece, USA,

China, Portugal, Canada, etc., may not apply to the Croatian context.

A further contribution of our work, bearing in mind Glassdoor's research, would lie in

determining the extent of knowledge of the mission and vision, or the degree of

familiarity with the mission, among Croatian employees. The value of results presented

in this paper regarding the differences between public sector employees and those

in for-profit organizations in terms of the impact of employee mission knowledge on

the variables of their JS and OC levels should also be highlighted. Finally, we point out

that in our research two demographic characteristics of the respondents (gender and

age) were also briefly considered as a variable, wherein certain regularity was

established – a relationship that is not completely trivial.

In the following sections of the paper, we present a literature review and the

research questions we set out to test, after which we describe the statistical processing

of the collected data, obtained results, discussion and conclusion.

Literature review Mission and vision Employees – with their knowledge, skills, competencies, and attitudes – are the key to

achieving organizational goals. An organization without engaged employees cannot

thrive, as suggested by the words of Mary Barra, the current CEO of GM: “If we win the

hearts and minds of employees, we're going to have better business success”

(Katzenbach et al., 2019). If the goals of the organization to a greater or lesser extent

coincide with the employees’ personal goals, their internal satisfaction and motivation

will be empowered. Naturally, this will occur only if the employees are aware of the

goals, mission, and vision of their organization. A clear and well-put organizational

vision and mission can greatly facilitate the formulation and realization of the

organization’s goals (Buble, 2000). Miller and Dess define vision as an aspiration

towards the future that inspires and motivates employees (Bart and Baetz, 1998). While

the vision defines a desired future of the organization, the mission describes the

rationale for the organization, that is, what the organization is about. The mission

communicates the values, aspirations, and the reason for the existence of the

organization, and, being the expression of organizational purpose, is of great

importance for the processes of formulation of strategies and strategic goals (Bart and

Baetz, 1998; David, 1989; Kaplan & Norton, 1996).

When it comes to the representation of vision in research models, the foundation

for defining the construct of vision was laid by Larwood et al., (1993) while Collins and

128


Porras (2008) proposed a framework for creating the vision by defining its key

components.

The construct of the mission could include indicators (attributes) such as Macedo

et al. (2016): providing a common purpose and orientation; allowing the CEO to exert

control over the organization; creating standards of performance for the organization;

promoting shared values among organizational members; promoting the interests of

external stakeholders; providing a sound basis for the allocation of organizational

resources. An original consideration of organizational mission, mission statements, and

vision can be found in the papers by (Campbell and Yeung, 1991; Lucas, 1998; Bart et

al., 2004; Erol et al., 2014; Allison, 2019).

Although some contemporary authors would argue that the mission and the vision

often represent abstract and somewhat archaic documents, in research in this field

there is evidence that indicates their practical usefulness in the day-to-day operation

of organizations (Mullane, 2002; Darbi, 2012).

Interrelationships between the mission statement and employee performance have

been particularly extensively studied in the context of public sector organizations. In

that respect, certain cause-and-effect relationships have been established and

valuable concepts devised. One such concept, termed “mission valence” (by

analogy with the concept of valence in chemistry), was formulated by Rainey and

Steinbauer (1999). As a concept, mission valence draws on formulations from

expectancy theory Vroom (1964) and roughly denotes a positive or negative

attractiveness of organizational mission.

Job satisfaction and organizational commitment Job satisfaction represents the level to which employees like or dislike their job. It refers

to individuals' level of satisfaction with the job they are currently performing within the

organization. According to (Weiss, 2002; Breaugh et al., 2018) job satisfaction is

defined “… as the positive attitudes, judgments, and feelings a person has for work

tasks, work experiences, and appraisals of one’s job“.

Organizational commitment, on the other hand, denotes the employees'

commitment to the organization for which they work. In other words, it represents the

strength of the individuals’ identification with the organization and their engagement

therewith. According to Greenberg and Baron (2003) organizational commitment

relates to the intensity of an employee’s dedication to an organization.

Organizational commitment is characterized by three factors: firm belief and

acceptance of the organization’s goals and values, willingness to invest effort in the

organization, and a strong desire to stay with the organization (Brčić et al., 2018;

Mowday et al., 1979; Porter et al., 1974). Three components of OC are considered in

the literature – affective (emotional connection to the organization), continuance

(consideration of costs and losses by eventual departure from the organization), and

normative (feeling of obligation to continue with work in the organization)

commitment (Allen and Meyer, 1990).

The difference between job satisfaction and organizational commitment lies in the

fact that the latter concept is more stable over time and refers to individuals’ general

commitment to the organization, as well as to their identification with the goal and

values of the organization over a certain time. On the other hand, job satisfaction

refers to a specific job that is currently performed by an individual in the organization,

and their satisfaction with it, which can vary over time and is affected by everyday

situations in the work environment (Mowday, et al., 1979).

129


In their paper, Markovits et al. (2010) considered the differences in the relationship

between job satisfaction and individual components of commitment between the

public sector and private sector employees.

Mission, vision, job satisfaction, and organizational commitment In their study, Kirkpatrick and Locke (1992) confirmed that the employees’ positive

attitude to the vision of the organization that they work for increases their commitment

and quality of performance (Testa, 1999; Kirkpatrick and Locke, 1992). A positive

correlation between satisfaction with the formulated vision of the organization on the

one hand and job satisfaction on the other was established by Testa (1999).

The report SHRM (2016) provides an indicator that directly illuminates the impact of

the mission on JS, which is: “Meaningfulness of the job (understanding how your job

contributes to the organization’s mission).” Positive relations between mission (or

mission statements) and JS or OC have been established in research (Buelow et al.,

1999; Wright and Pandey, 2011; Yazhou and Jian, 2011; Macedo et al. 2016).

For a broader coverage of issues and more ample consideration of the relationship

between relevant variables, we identified the following research foci in the work by

other authors: the relationship between the (mission and vision) formulation and

organizational success, measurement of organizational commitment, and job

satisfaction in different industries as well as their relationship with employee

performance, their emotional burnout and abandonment of the organization

(Prentice and Thaichon, 2019; Rodrigo et al., 2019; Hong Lu et al., 2019; Vidić, 2010;

Bart and Baetz, 1998).

Various authors have confirmed the interrelationship between the mission and

vision formulation on the one hand and the organization’s success or performance on

the other (Alavi and Karami, 2009; Bart and Hupfer, 2004; Erol and Kanbur, 2014; Price,

2012; Sheaffer et al., 2008).

Gender and age and organizational commitment Several authors have dealt with the impact of gender and age on organizational

commitment and job satisfaction. For example, (Singh et al., 2004; Marchiori and

Henkin, 2004) have found a higher level of organizational commitment among female

employees, while some other researchers established greater organizational

commitment among male employees (Dixon et al., 2005; Marchiori and Henkin, 2004;

Savery and Syme, 1996; Singh et al., 2004). The results of the impact of gender on

organizational commitment are varied although gender constitutes one of the most

frequently used demographic variables in organizational commitment research

(Anari, 2012). In the existing studies, no differences in job satisfaction with regards to

the employees’ gender have been identified (Anari, 2012; Arani, 2003).

Considering the impact of age on organizational commitment and job satisfaction,

Anari (2012) stated that many authors who looked into the relationship between job

satisfaction and age found that higher satisfaction was established among older

employees when compared to their younger colleagues (Anari, 2012; Warr, 1992;

Glenn et al, 1977). Kacmar et al. (1999) reported a positive relationship between age

and organizational commitment. Contrary to those results, Anari (2012) found no

significant differences between job satisfaction and organizational commitment

regarding the age differences among participants

Public sector In comparison with the private sector, organizations in the public sector usually have

a more all-encompassing mission which also has a more profound (societal) impact

(Baldwin, 1984). At the level of the employee, a contribution that is accomplished by

130


an individual that is actualizing the mission of a public organization represents a

certain intrinsic reward to that person. The feeling of being rewarded manifests itself

as internal satisfaction that is experienced by an employee performing a certain job

while accomplishing one's own goals through organizational tasks.

A large body of research has confirmed that employees in the public sector are

not as focused on the expectations of financial rewards as their counterparts in the

private sector (Wright, 2007,). According to Herzberg’s motivation theory (also known

as “the two-factor theory”), extrinsic factors such as external sources (for instance,

financial rewards) eliminate employees’ work dissatisfaction and ensure their work

eagerness, whereas intrinsic factors determine the quality of their work (Herzberg,

2005; Robbins, 2009).

With regards to job satisfaction, a large number of surveys have found that public

sector employees primarily value interesting work, while employees in for-profit

organizations most highly value good wages, as reported in papers by (Karl and

Sutton, 1998; Naff and Crum, 1999).

Regarding organizational commitment, the results of research published in the

article by Markovits et al. (2007) show that in Greece organizational commitment in

the public sector is significantly higher than in their private sector. This is contrary to

evidence from some other countries.

Furthermore, characteristics of JS and OC association with other distinctive

organizational variables in public sector analysis (mission valence, employee

performance, public service motivation) were discussed in papers by (Frey and Jegen,

2001; Harrison et al., 2006; Moè et al., 2010; Cerasoli, et al., 2014; Caillier, 2014;

Potipiroon and Ford, 2017). The evidence which is of particular interest for the topic of

our research was presented in Wright and Pandey (2011), where a substantial, direct

effect of mission valence on employee job satisfaction was established.

Methodology Research questions Based on previous considerations, this research departs from the following conjecture:

If the vision is the element that makes an organization more successful and the mission

helps the organization to realize success that often encompasses broader societal

goals, then employees that are familiar with the mission and vision will be more

satisfied with the job and more committed to the organization. Accordingly, the

following research questions are defined:

o RQ1. Are there differences in organizational commitment and job satisfaction

between employees who are familiar with the mission and vision of the

organization they are employed in and those who are not?



organization they are employed in and those who are not considering the

employees’ gender?

o RQ3. Are there differences between the public and the private sector

considering the impact of familiarity with the mission and vision on

organizational commitment and job satisfaction?



organization they are employed in and those who are not considering the

employees’ age differences?

131


Data collection and sample description To respond to the research questions above we researched employees in the

Republic of Croatia. In Croatia, according to the data of March 31, 2019, there are

154,184 active companies.

The data were collected via the survey questionnaire that was sent to the e-mail

addresses of 750 of those companies, 114 of which fully completed the questionnaires

and returned them, amounting to the respondent rate of 15%.

The sample comprised of 114 employees in private and public organizations.

Among the 114 respondents, 66 were employed in the public or state sector, 48 in the

private or predominantly private sector. Out of the 114 respondents, 58.5% were

female and 41.2% male, the majority of whom (57%) were up to 40 years of age. The

sample structure is presented in Table 1.

Table 1

Sample structure

Sample characteristics Number Percentage

Gender

Male 47 41.2%

Female 67 58.8%

Age

up to 40 years of age 65 57.0%

more than 40 years of age 49 43.0%

Sector

State or public 66 57.9%

Private or predominantly private 48 42.1%


Research instrument To measure organizational commitment we used 13 statements from the

Organizational Commitment Questionnaire (OCQ) created by Mowday et al.

(Mowday et al., 1979). For the measurement of job satisfaction, we used 17 statements

from the Job Satisfaction Survey (JSS) by Spector (Spector, n.d.). A Likert scale was

used to measure the attitudes of the respondents. The agreement with the statements

used in the questionnaire was measured using a 5-point ordinal scale (1 completely

disagree, 5 completely agree). The research instrument is presented in Table 2 below.

Statistical methods The collected data were analyzed through several statistical methods utilizing the IBM

SPSS Statistics 23 statistical package. First, the instrument validity was checked and the

internal reliability of the measurement model was confirmed by the Cronbach alpha

coefficient (Field, 2013).

To determine whether any statistically significant differences existed between the

employees in terms of their familiarity with the mission and vision considering the

gender, age, and considering the sector that they belong to a t-test has been

applied. We investigated whether there were statistically significant differences in job

satisfaction and organizational commitment between employees who are familiar

with the mission and vision of the organization they are employed in and those who

are not in terms of gender, age, and considering the sector that they belong to.

132


Table 2

Research instrument

Construct Code Statements

Job satisfaction JSS1 I feel satisfied with my chances for salary increases.

JSS2 I feel unappreciated by the organization when I think

about what they pay me.

JSS3 The benefits package we have is equitable.

JSS4 I don’t feel my efforts are rewarded the way they should

be.

JSS5 There are few rewards for those who work here.

JSS6 Raises are too few and far between.

JSS7 Those who do well on the job stand a fair chance of

being promoted.

JSS8 I feel I am being paid a fair amount for the work I do.

JSS9 I am satisfied with my chances for promotion.

JSS10 I am not satisfied with the benefits I receive at my job.

JSS11 There are benefits we do not have what we should have.

JSS12 When I do a good job, I receive the recognition for it that

I should receive.

JSS13 The benefits we receive are as good as most other

organizations offer.

JSS14 There is too little chance for promotion in this organization.

JSS15 I do not feel that the work I do is appreciated.

JSS16 My supervisor shows too little interest in the feelings of

subordinates.

JSS17 People get ahead as fast here as they do in other places.

Organizational

commitment

OCQ1 I am proud to tell others that I am part of this organization.

OCQ2 I am extremely glad that I chose this organization to work

for over others I was considering at the time I joined.

OCQ3 For me, this is the best of all possible organizations for

which to work.

OCQ4 I talk up this organization to my friends as a great

organization to work for.

OCQ5 Deciding to work for this organization was the best

decision I could make.

OCQ6 This organization inspires the very best in me in the way of

job performance.

OCQ7 I find that my values and the organization’s values are

very similar.

OCQ8 I feel loyalty to the organization I work for.

OCQ9 It would take very little change in my present

circumstances to cause me to leave this organization.

OCQ10 There’s not too much to be gained by sticking with this

organization indefinitely.

OCQ11 I care about the fate of this organization.

OCQ12 I am willing to put in a great deal of effort beyond that

normally expected to help this organization be successful.

OCQ13 Often, I find it difficult to agree with this organization’s

policies on important matters relating to its employees.


133


Validity analysis The instrument validity was verified by conducting factor analysis (principal

components analysis). The goal was to determine the existence of two factors, one

for the job satisfaction and one for the organizational commitment constructs,

respectively, and confirm the unidimensionality of each of the two constructs

The sampling adequacy for conducting a factor analysis was confirmed by the Kaiser-

Meyer-Olkin measure (KMO=0,893) (Field, 2013). Tables 3 represent the results of the

exploratory factor analysis (principal components analysis). Concerning the t-test

application, we checked and confirmed that the samples fulfill certain preconditions,

such as normality, approximately equal variance, and independence (Kim, 2015).

Reliability analysis The instrument reliability was measured by using the Cronbach alpha. Table 3

represents the Cronbach alpha coefficients, mean values, and standard deviations.

The values of the Cronbach alpha are above the recommended value (0.7) (Field,

2013).

Table 3

Factor analysis (by principal components method, Varimax rotation method) of

statements

Item Factor

1 2

Job satisfaction JSS1 0.620

JSS2 0.587

JSS3 0.653

JSS4 0.521

JSS5 0.726

JSS6 0.661

JSS7 0.586

JSS8 0.503

JSS9 0.789

JSS10 0.502

JSS11 0.477

JSS12 0.755

JSS13 0.690

JSS14 0.746

JSS15 0.591

JSS16 0.738

JSS17 0.609

Organizational commitment OCQ1 0.580

OCQ2 0.796

OCQ3 0.809

OCQ4 0.605

OCQ5 0.753

OCQ6 0.869

OCQ7 0.741

OCQ8 0.589

OCQ9 0.813

OCQ10 0.530

OCQ11 0.679

OCQ12 0.731

OCQ13 0.704


134


Table 4

Descriptive statistics

Statement Mean Std.

Deviation

Cronbach

alpha

Job satisfaction OCQ1 4.3770 0.7688 0.931

OCQ2 3.6750 1.0684

OCQ3 4.2020 0.9041

OCQ4 2.6320 1.2569

OCQ5 3.3160 1.0159

OCQ6 3.7460 1.1352

OCQ7 3.2190 1.1426

OCQ8 3.6754 1.2726

OCQ9 3.7980 1.1224

OCQ10 3.5526 1.1757

OCQ11 4.4210 0.8508

OCQ12 3.5090 1.1071

OCQ13 3.6750 1.1170

Organizational

commitment

JSS1 3.5350 1.0491 0.918

JSS2 2.9825 1.3236

JSS3 3.2632 1.2341

JSS4 3.0790 1.2276

JSS5 2.4474 1.2127

JSS6 3.0350 1.2404

JSS7 2.7280 1.0749

JSS8 3.3333 1.2214

JSS9 3.3070 1.1983

JSS10 2.6490 1.0558

JSS11 3.2456 1.3073

JSS12 3.0180 1.1673

JSS13 2.4123 1.1735

JSS14 2.9470 1.2540

JSS15 2.9561 1.2718

JSS16 2.8070 1.2111

JSS17 3.2370 1.1545


Results In the observed sample of 114 respondents, 95 of them (58 female and 37 male

respondents) are familiar with the mission of the organization in which they work, while

19 (9 female and 10 male respondents) of them are not familiar with it. Furthermore,

94 respondents (58 female and 36 male) are familiar with the vision of their

organization, while 20 (11 female and 9 male) of them are not familiar with it.

These results indicate that approximately 78% of employees from the Croatian

organizations included in our research are aware of the mission and vision of their

organizations. Our findings can be considered in the light of the study by Glassdoor,

(2019) in which the mission was found to be one of the main reasons why 64% of

employees stay at their jobs.

From table 5 it is evident that there is a statistically significant difference in job

satisfaction between employees who are familiar with the organization’s mission and

those who are not (with level p< 0.05). When it comes to the difference in job

satisfaction between employees familiar with the organization’s vision and those who

are not, statistically significant differences were obtained at level (p≤ 0.10). Likewise,

in terms of organizational commitment, there are no statistically significant differences

135


between employees who are familiar with the mission and vision of the organization

in which they work and those who are not (p> 0.1).

Concerning the notion of statistical significance, attention should be drawn to the

understanding associated with the p-value. Namely, with the assumption of the null

hypothesis (that there is no difference between the two samples), the negation of this

hypothesis (H0) is realized based on the calculation of the (Fisher's) p-value. The most

common levels of statistical significance are related to the following p-values: 0.01,

0.05 and 0.10. For the obtained value 0.05 ≤ p ≥ 0.01, it can be said that it is a moderate

evidence of denial H0, while at 0.1 ≤ p ≥ 0.05 there is weak evidence, and also at 0.01

≤ p ≥ 0.001 it is strong evidence, as reported in (Bland, 2015; Held and Ott, 2018).

According to Wasserstein and Lazar (2016) “… the widespread use of statistical

significance, generally interpreted as p ≤ 0.05, as a license for making a claim of a

scientific - leads to considerable distortion of the scientific process.” Therefore, due to

misapplications and incorrect interpretations of the p-value, one part of the

researchers insists on supplementing the p-values considerations with other

appropriate approaches (confidence, Bayesian methods, false discovery rates, etc.).

Table 5

Differences between employees regarding familiarity with mission/vision

Familiarity with the

mission/ vision

Mean Std. Deviation t-test p

Job satisfaction

(mission)

Familiar 3,0644 0,7561 2,005 0,047**

Not familiar 2,6718 0,8898

Job satisfaction

(vision)

Familiar 3,0638 0,7554 1,924 0,057*


Organizational

commitment

(mission)

Familiar 3,7321 0,7485 1,206 0,230


Organizational

commitment

(vision)

Familiar 3,7418 0,7521 1,459 0,147


Note: ** statistically significant at 5%; *10%


From the results in table 6, it is evident that there is a statistically significant difference

in job satisfaction between male employees who are familiar with the organization’s

mission and those who are not (level p≤ 0.10). Significant differences exist between

male employees who are familiar with the mission (p≤ 0.10) and vision (p≤ 0.10) and

those who are not regarding their organizational commitment.

Table 6

Differences between employees regarding gender


mission/vision

Mean Std.

Deviation

t-test p

Job

satisfaction

(mission)

Male Familiar 3,1463 0,7327 1,764 0,085*


Female Familiar 3,0122 0,7725 1,124 0,265


Job

satisfaction

(vision)

Male Familiar 3,1471 0,7299 1,665 0,103


Female Familiar 3,0122 0,7725 1,124 0,265


136


Table 6

Differences between employees regarding gender (Continued)


mission/vision

Mean Std.

Deviation

t-test p

Organizational

commitment

(mission)

Male Familiar 3,7887 0,7690 1,685 0,099*


Female Familiar 3,6959 0,7396 -0,117 0,907


Organizational

commitment

(vision)

Male Familiar 3,8157 0,7764 2,005 0,051*


Female Familiar 3,6959 0,7396 -0,117 0,907


Note: * statistically significant at 10%


The analysis which was aimed to compare groups of respondents by age yielded

the following results shown in table 7. There is a statistically significant difference in job

satisfaction between the employees in the more than 40 years old age group who are

familiar with the organization’s mission and vision and those who are not (p< 0.01). In

terms of organizational commitment, there are no statistically significant differences

regarding the age differences between employees who are familiar with the mission

and vision of the organization in which they work and those who are not.

Table 7

Differences between employees regarding the age

Familiarity

with the

mission/vision

Mean Std.

Deviation

t-test p

Job

satisfaction

(mission)

< 40 years Familiar 3,0430 0,8059 0,706 0,483


40 years + Familiar 3,0903 0,6998 2,749 0,008***


Job

satisfaction

(vision)

< 40 years Familiar 3,0288 0,8196 0,412 0,681


40 years + Familiar 3,1053 0,6786 3,243 0,002***


Organizational

commitment

(mission)

< 40 years Familiar 3,6626 0,7628 0,385 0,702


40 years + Familiar 3,8161 0,7309 1,434 0,158


Organizational

commitment

(vision)

< 40 years Familiar 3,6774 0,7697 0,651 0,518


40 years + Familiar 3,8182 0,7321 1,488 0,143


Note: *** statistically significant at 1%


The analysis which was aimed to compare groups of respondents from the public and

the private sector yielded the following results shown in table 8. It is evident that there

is a statistically significant difference in job satisfaction between employees in the

state or public sector who are familiar with the organization’s mission and those who

are not (p< 0.05), and there is a statistically significant difference in job satisfaction

between employees in state or public sector who are familiar with the organization’s

vision and those who are not (p< 0.01).

137


Table 8

Differences between employees regarding the sector Familiarity

with the

mission/vision

Mean Std.

Deviation

t-test p

Job

satisfaction

(mission)

State or public YES 3,0368 0,7487 2,537 0,014**

NO 2,3941 0,6679

Private or

predominantly

private

YES 3,1053 0,7852 0,403 0,689

NO 2,9804 1,0374

Job

satisfaction

(vision)

State or public YES 3,0545 0,7397 2,871 0,006***

NO 2,3636 0,6649

Private or

predominantly

private

YES 3,0774 0,7970 -0,067 0,947

NO 3,0980 1,0017

Organizational

commitment

(mission)

State or public YES 3,6526 0,7298 0,458 0,649

NO 3,5364 0,7977

Private or

predominantly

private

YES 3,8541 0,7785 1,276 0,208

NO 3,4444 1,1883

Organizational

commitment

(vision)

State or public YES 3,6612 0,7359 0,644 0,522

NO 3,5041 0,7523

Private or

predominantly

private

YES 3,8636 0,7780 1,439 0,157

NO 3,4040 1,1731



Discussion The goals of the research in this paper were to examine whether there are statistically

significant differences in organizational commitment and job satisfaction between

employees who are familiar with the mission and vision of the organization they are

employed in and those who are not, as well as investigate the impact of the

employees’ gender and age on the two aforementioned constructs. Furthermore, we

intended to determine whether there are statistically significant differences in

organizational commitment and job satisfaction, respectively, within the group of

public sector employees considering their familiarity with the mission and vision of the

organization they are employed in. Finally, we wanted to verify whether such

differences apply to the group of employees in the private sector.

In our research, the statistically significant difference in job satisfaction was

obtained between employees familiar with the mission (p≤ 0.05) and vision (p≤ 0.1),

and those who are not. On the other hand, no significant differences were confirmed

between employees who are familiar with the mission and vision and those who are

not considering their organizational commitment.

The identified positive impact of familiarity of employees with the mission of the

organization on the JS level is in agreement with the results of the aforementioned

research by (Krueger et al., 2002; Clark, 2006; Jizhou and Jia, 2011).

No statistically significant differences were established in job satisfaction and

organizational commitment among female employees, in terms of familiarity (or lack

thereof) with the mission and vision of the organization, they work in. The statistically

significant difference in job satisfaction and organizational commitment was obtained

between male employees familiar with the mission (p≤ 0.1) and those who are not.

Furthermore, between male employees familiar with the vision and those who are not

138


a statistically significant difference was also found (p< 0.1) for organizational

commitment. The results are in line with previous studies by Anari (2012), and contrary

to the results from Kirkpatrick and Locke (1992).

In our study, the lack of difference in organizational commitment (Table 5) among

employees that are familiar with the mission and vision of their organization can be

accounted for by the structure of the sample, wherein 57% of respondents were aged

up to 40 years old. In the context of Croatia, fairly younger employees have adopted

global trends about increasingly more frequent changes in the work environment,

which explains why their familiarity with the mission and vision does not influence their

attitude toward organizational commitment.

We also found a statistically significant difference in job satisfaction between the

employees in the more than 40 years old age group that are familiar with their

organization’s mission and vision and those that are not (p <0.01). No statistically

significant differences were established in organizational commitment among

employees regarding the age differences in terms of familiarity with the mission and

vision of the organization they work in (table 7). These results are in line with some

previous studies where it was found that older employees have higher satisfaction

than their younger colleagues (Anari, 2012; Warr, 1992; Glenn et al, 1977).

Furthermore, in our research, we established that in public sector organizations the

employees’ familiarity with the mission and vision has an impact on job satisfaction (p<

0.5 and p< 0.1). On the other hand, such interrelationship does not apply to the sample

of private-sector employees (table 8). It has to be noted that the comparable results

and non-contradictory to those obtained for the Croatian public sector employees in

our research were previously reported in studies by (Krueger et al., 2002; Clark, 2006;

Bart, 2007; Wright and Pandey, 2011; Jizhou and Jian, 2011).

When explaining the results for the public and the private sector, we should also

mention the conclusion of a study by Gazioglu and Tansel (2006) conducted in Great

Britain, which stated that, among other things, "... employees who feel that their job is

secure exhibit higher levels of job satisfaction.” In interpreting the results of our study,

a potential limitation of the research is, firstly, the sample size and, secondly, the

distribution (within the sample) of employees that are familiar with the mission and

vision of the organization and those that are not. Another limitation is a lack of in-depth

verification of the trustworthiness of the respondents’ statements regarding their

familiarity with the mission and vision of the organization they work in.

Conclusion Formulating the mission and vision has become a commonplace contemporary

business practice. A good and well-put vision statement communicates the desired

future of the organization, the description and purpose of which are contained in the

mission statement.

Employees’ familiarity with organizational mission and goals and identification of

their own goals with the goals of their organization can contribute to retaining the

existing employees and attracting new ones. One of the key findings from Glassdoor’s

survey is that “79 percent of adults would consider a company’s mission and purpose

before applying for a job there” (Glassdoor, 2019).

The research in this paper was aimed to determine the impact of employees’

familiarity with the mission and vision of the organization in which they are employed

on their job satisfaction and organizational commitment. In our research, statistically

significant differences in job satisfaction were obtained between employees that are

familiar with the mission and those that are not. On the other hand, the employees’

139


familiarity with the vision influences their job satisfaction, with a milder claim of

statistical significance (p< 0.1).

In terms of the sector that the respondents belong to, statistically, significant

differences were particularly apparent among employees in the Croatian public

sector regarding their familiarity with the mission and vision on the one hand and their

job satisfaction on the other. Upon analysing the private-sector employees as a

separate sample, we concluded that for this group there was no statistically significant

difference regarding the relationship between their familiarity with the mission and

vision on the one hand and their job satisfaction on the other. When the organizational

commitment variable is concerned, no differences were established between

employees that are familiar with the mission and the vision of the organization they

work for and those that are not.

Taking into consideration the employees’ gender, differences in job satisfaction

and organizational commitment were established concerning the familiarity with the

mission and vision, primarily in groups of male respondents. Considering the age of the

employees, statistically, significant differences in job satisfaction were obtained in the

more than 40 years old age group in terms of their familiarity with the vision and the

mission of the organization they work for, respectively. Furthermore, no differences in

organizational commitment were established between employees of all age groups

in terms of their familiarity with the mission and vision of their organization. The above

findings indicate that employees’ familiarity with the mission and vision does

contribute to job satisfaction, particularly to older employees, whose familiarity with

the vision also promotes their organizational commitment. Finally, the impact of

employees’ familiarity on the mission and vision is visible in the case of the job

satisfaction construct when employees in the public sector in Croatia are concerned.

References 1. Alavi, T. M., Karami, A. (2009), “Managers of small and medium enterprises: Mission

statement and enhanced organizational performance”, Journal of Management

Development, Vol. 8 No. 6, pp. 555-562.

2. Allen, N.J., Meyer, J.P. (1990), “The measurement and antecedents of affective,

continuance and normative commitment to the organization”, Journal of Occupational

Psychology, Vol. 63 No. 1, pp. 1-18.

3. Allison, J. (2019), “Mission statements and vision statements: examining the relationship

toward performance outcomes”, Global Journal of Management and Marketing, Vol. 3

No. 1, pp. 1-21.

4. Anari, N. N. (2012), “Teachers: emotional intelligence, job satisfaction, and organizational

commitment”, Journal of Workplace Learning, Vol. 24 No. 4, pp. 256-269.

5. Arani, A. (2003), “A comparative study of secondary school teachers’ job satisfaction in

relation to their value orientation and school organizational climate in Iran and India”,

Humanities Education, Vol. 4 No.1, pp. 200-225.

6. Baldwin, J. N. (1984), “Are We Really Lazy?”, Review of Public Personnel Administration, Vol.

4 No. 2, pp. 80-89.

7. Bart, C. K., Baetz, M. C. (1998), “The relationship between mission statements and firm

performance: An exploratory study”, Journal of Management Studies, Vol. 35 No. 6, pp.

823-853.

8. Bart, C. K., Hupfer, M. (2004) “Mission statements in Canadian hospitals”, Journal of Health

Organization and Management, Vol. 2 No. 3, pp. 92-110.

9. Bart, C. (2007), “Comparative Analysis of Mission Statement Content in Secular and Faith-

Based Hospitals”, Journal of Intellectual Capital, 2007, Vol. 8 No. 4, pp. 682–694.

10. Bland, M. (2015), An Introduction to Medical Statistics, 4th ed., Oxford University Press,

Oxford.

140


11. Breaugh, J., Ritz, A., Alfes, K. (2018), “Work motivation and public service motivation:

disentangling varieties of motivation and job satisfaction”, Public Management Review,

Vol. 20 No. 10, pp. 1423-1443.

12. Brčić, R., Dulčić, Ž., Černe, M., Hernaus, T., Malbašić, I., Matić, I., Pološki Vokić, N., Rey, C.

(2018), Koraci uspješnog organiziranja, Školska knjiga, Zagreb.

13. Buble, M. (2000), Management, Ekonomski fakultet Split, Split.

14. Buelow, J. R., Winburn, K., Hutcherson, J. (1999), “Job Satisfaction of Home Care Assistants

Related to Managerial Practices”, Home Health Care Services Quarterly, Vol. 17 No. 4, pp.

59-71,

15. Caillier, J. G. (2014), “Toward a Better Understanding of the Relationship Between

Transformational Leadership, Public Service Motivation, Mission Valence, and Employee

Performance: A Preliminary Study”, Public Personnel Management, Vol. 43 No. 2, pp. 218-

239.

16. Clark, R. H., Jr. (2006), “A study of the relationship between job satisfaction, goal

internalization, organization citizenship behavior, and mission statement familiarity”, The

University of Alabama, ProQuest Information and Learning Company, Michigan, Ann Arbor,

available at https://search.proquest.com/docview/305350115?pq-origsite=gscholar (11

March 2021)

17. Campbell, A., Yeung, S. (1991), “Creating a sense of mission”, Long Range Planning, Vol.

24 No. 4, pp. 10-20.

18. Cerasoli, C., Nicklin, J., Ford, M. (2014), “Intrinsic Motivation and Extrinsic Incentives Jointly

Predict Performance: A 40-Year Meta-Analysis", Psychological bulletin, Vol. 140 No. 4, pp.

980-1008.

19. Collins, J. C., Porras, J. I. (2008), “Organizational Vision and Visionary Organizations”,

California Management Review, Vol. 50 No. 2, pp. 117-137.

20. Darbi, W. P. K. (2012), “Of mission and vision statements and their potential impact on

employee behaviour and attitudes: The case of a public but profit-oriented tertiary

institution”, International Journal of Business and Social Science, Vol. 3 No. 14, pp. 95-109.

21. David, F. (1989), “How Companies Define Their Mission”, Long Range Planning, Vol. 22 No.

1, pp. 90-97.

22. Dixon, M., Cunningham, G., Sagas, M., Turner, B., Kent, A. (2005), “Challenge Is Key: An

Investigation of Affective Organizational Commitment in Undergraduate Interns”, The

Journal of Education for Business, Vol. 80 No. 3, pp. 172-180.

23. Erol, Y., Kanbur, E. (2014), “Entrepreneurial characteristics of Turkey’s top 100 industrial

enterprises according to their mission and vision statements”, Business and Economics

Research Journal, Vol. 5 No. 3, pp. 149-165.

24. Field, A. (2013), Discovering statistics using IBM SPSS Statistics, SAGE Publications Ltd.,

London.

25. Frey, B., Jegen, R. (2001), “Motivation crowding theory”, Journal of Economic Surveys, Vol.

15 No.5, pp. 589-561.

26. Gazioglu, S., Tansel A (2006), “Job satisfaction in Britain: individual and job related factors”,

Applied Economics, Vol. 38 No. 10, pp. 1163-1171.

27. Glassdoor (2019), “Glassdoor’s Mission & Culture Survey”, available at

https://www.glassdoor.com/about-us/app/uploads/sites/2/2019/07/Mission-Culture-

Survey-Supplement.pdf (11 March 2021)

28. Glenn, N., Taylor, P., Weaver, C. (1977), “Age and job satisfaction among males and

females: a multivariate, multisurvey study”, Journal of Applied Psychology, Vol. 62, pp. 189-

93.

29. Goulet, L. R., Frank M. L. (2002), “Organizational commitment across three sectors: public,

non-profit, and for-profit”, Public Personnel Management, Vol. 31 No. 2, pp. 201-210.

30. Greenberg, J., Baron, R. A. (2003), Behaviour in Organizations, 8th ed., Prentice Hall, New

Jersey.

31. Harrison, D. A., Newman, D. A., Roth, P. L. (2006), “How important are job attitudes?

Metaanalytic comparisons of integrative behavioral outcomes and time sequences”,

Academy of Management Journal, Vol. 49 No 2, pp. 305-325.

141


32. Held, L., Ott, M. (2018), “On p-values and Bayes factors”, Annual Review of Statistics and Its

Application, Vol. 5 No. 1, pp. 393–419.

33. Herzberg, F. (2005), "Motivation-hygiene theory", in Miner, J. B. (Ed.) Organizational behavior

I: Essential theories of motivation and leadership, M.E. Sharpe Inc, New York, pp. 61-74.

34. Hong Lu, L., Yang, Z., Alison, W. (2019), “Job satisfaction among hospital nurses: A literature

review”, International Journal of Nursing Studies, Vol. 94, pp. 21-31.

35. Kacmar, K. M., Carlson, D. S., Brymer, R.A. (1999), “Antecedents and consequences of

organizational commitment: a comparison of two scales”, Educational and Psychological

Measurement, Vol. 59 No. 6, pp. 976-994.

36. Kaplan, R., Norton, D. (1996), “Using the Balanced Scorecard as a Strategic Management

System”, Harvard Business Review, Vol. 74 No. 1, pp. 75-85.

37. Karl, K. A., Sutton, C. L. (1998), “Job values in today’s workforce: A comparison of public

and private sector employees”, Public Personnel Management, Vol. 27 No. 4, pp. 515-527.

38. Katzenbach, J. R., Thomas, J., Anderson, G. (2019), The Critical Few - Energize Your

Company's Culture By Choosing What Really Matters, Berrett-Koehler Publishers, Oakland.

39. Kim, T. K. (2015), “T test as a parametric statistic", Korean Journal of Anesthesiology, Vol. 68

No. 6, pp. 540-546.

40. Kirkpatrick, S. A., Locke, E. A. (1992), “Direct and indirect effects of tree core charismatic

leadership components on performance and attitudes", Journal of Applied Psychology,

Vol. 81 No. 1, pp. 36-51.

41. Krueger, P., Brazil, K., Lohfeld, L., Edward, H. G., Lewis, D., Tjam, E. (2002), “Organization

specific predictors of job satisfaction: Findings from a Canadian multi-site quality of work

life cross-sectional survey”, BMC Health Services Research, Vol. 2 No. 1. Available at

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC102756/ (11 March 2021)

42. Kwong, K., Wong, K. (2013), “Partial least squares structural equation modeling (PLS-SEM)

techniques using SmartPLS”, Marketing Bulletin, Vol. 24 No. 1, pp. 1-32.

43. Larwood, L., Kriger, M. P., Falbe, C. M. (1993), “Organizational Vision: An Investigation of the

Vision Construct-In-Use of AACSB Business School Deans”, Group & Organization

Management, Vol. 18 No. 2, pp. 214-236.

44. Lucas, J. (1998), “Anatomy of a vision statement”, Management Review, Vol. 87 No. 2, pp.

22-26.

45. Macedo, I. M., Pinho, J. C., Silva, A. M. (2016), “Revisiting the link between mission

statements and organizational performance in the non-profit sector: The mediating effect

of organizational commitment”, European Management Journal, Vol. 34 No. 1, pp. 36–46.

46. Marchiori, D. M., Henkin, A. B. (2004), “Organizational commitment of a health profession

faculty: dimensions correlated and conditions”, Medical Teacher, Vol. 26 No. 4, pp. 353-

358.

47. Markovits, Y., Davis, A. J., Dick, R. (2007), “Organizational Commitment Profiles and Job

Satisfaction among Greek Private and Public Sector Employees”, International Journal of

Cross Cultural Management, Vol 7, No.1, pp. 77-99.

48. Markovits, Y., Davis, A. J., Fay, D., van Dick, R. (2010), “The Link Between Job Satisfaction

and Organizational Commitment: Differences Between Public and Private Sector

Employees”, International Public Management Journal, Vol. 13 No. 2, pp.177-196.

49. Mowday, R. T., Steers, R. M., Porter, L. W. (1979), “The measurement of organizational

commitment", Journal of Vocational Behavior, Vol. 14 No. 2, pp. 224-247.

50. Moè, A., Pazzaglia, F., Ronconi, L. (2010), “When being able is not enough. The combined

value of positive affect and self-efficacy for job satisfaction in teaching”, Teaching and

Teacher Education, Vol. 26 No. 5, pp. 1145-1153.

51. Mullane, J. V. (2002), "The mission statement is a strategic tool: when used properly",

Management Decision, Vol. 40 No. 5, pp. 448-455.

52. Naff, K.C., Crum, J. (1999), “Working for America: Does public service motivation make a

difference?”, Review of Public Personnel Administration, Vol. 19 No. 4, pp 5-16.

53. Pandey, S., Wright, B., Moynihan, D. (2008), “Public Service Motivation and Interpersonal

Citizenship Behavior: Testing a Preliminary Model", International Public Management

Journal, Vol. 11 No.1, pp. 89-108.

142


54. Porter, L. W., Steers, R. M., Mowday, R. T., Boulian, P. V. (1974), “Organizational commitment,

job satisfaction, and turnover among psychiatric technicians”, Journal of Applied

Psychology, Vol. 59 No. 5, pp. 603-609.

55. Potipiroon, W., Ford, M. T. (2017), “Does public service motivation always lead to

organizational commitment? Examining the moderating roles of intrinsic motivation and

ethical leadership”, Public Personnel Management, Vol. 46 No. 3, pp. 211–238.

56. Prentice, C., Thaichon, P. (2019), “Revisiting the job performance - burnout relationship”,

Journal of Hospitality Marketing & Management, Vol. 28 No. 7, pp. 807-832.

57. Price, H. W. (2012), “Vision statement impact on organization strategic roles”, Business

Journal for Entrepreneurs, 2012(1), pp. 27-42.

58. Rainey, H., Steinbauer, P. (1999), “Galloping Elephants: Developing Elements of a Theory of

Effective Government Organizations”, Journal of Public Administration Research and

Theory, Vol. 9 No.1, pp. 1-32.

59. Robbins, S. P. (2009), Organizational Behaviour, International Version, 13th Ed., Pearson

Higher Education.

60. Rodrigo, P., Aqueveque, C., Duran, I. J. (2019), “Do employees value strategic CSR? A tale

of affective organizational commitment and its underlying mechanisms", Business Ethics-A

European Review, Vol. 28 No. 4, pp. 459-475.

61. Savery, L. K., Syme, P. D. (1996), “Organizational commitment and hospital pharmacists”,

Journal of Management Development, Vol. 15 No. 1, pp. 14-22.

62. Sheaffer, Z., Landau, D., Drori, I. (2008), "Mission statement and performance: An evidence

of “Coming of Age”", Organization Development Journal, Vol. 26 No. 2, pp. 49-62.

63. SHRM (2016), “Employee Job Satisfaction and Engagement - Revitalizing a Changing

Workforce”, A research Report by the Society for Human Resource Management, April

2016, USA, available at www.shrm.org (11 March 2021)

64. Singh, P., Finn, D., Goulet, L. (2004), “Gender and job attitudes: a re-examination and

extension”, Women in Management Review, Vol. 19 No. 7, pp. 345-355.

65. Spector, P. E., n.d. Professor Paul E. Spector, Phd., available at:

http://paulspector.com/scales/our-assessments/job-satisfaction-survey-jss/ (20 December

2019)

66. Testa, M. R. (1999), “Satisfaction with organizational vision, job satisfaction and service

efforts: an empirical investigation”, Leadership & Organization Development Journal, Vol.

20 No. 3, pp. 154-161.

67. Vidić, T. (2010), “Organizacijska odanost učitelja u osnovnoj školi", Napredak: časopis za

pedagogijsku teoriju i praksu, Vol. 151 No. 1, pp. 33-44.

68. Vroom, V. H. (1964), Work and motivation, Wiley. New York.

69. Warr, P. (1992), “Age and occupational well-being”, Psychology and Aging, Vol. 7 No. 1,

pp. 37-45.

70. Weiss, H. (2002), “Deconstructing Job Satisfaction. Separating Evaluations, Beliefs and

Affective Experiences”, Human Resource Management Review, Vol. 12 No. 2, pp. 173–194.

71. Wasserstein, R., Lazar, N. (2016), “The ASA’s statement on P-values: Context, process and

purpose”, The American Statistician, Vol. 70 No. 2, 129–133.

72. Wright, B. E. (2007), “Public service and motivation: Does mission matter?”, Public

Administration Review, Vol. 67 No. 1, pp. 54-64.

73. Wright, B., Pandey, S. (2011), “Public Organizations and Mission Valence: When Does

Mission Matter ?”, Administration & Society, Vol. 43 No. 1, pp. 22-44.

74. Yazhou, W., Jian, L. (2011), “Empirical Research on Influence of Mission Statements on the

Performance of Nonprofit Organization”, Procedia Environmental Sciences, Vol. 11, pp.

328–333.

http://www.shrm.org/

http://paulspector.com/scales/our-assessments/job-satisfaction-survey-jss/

143


About the authors

Dunja Dobrinić is currently a Ph.D. student at the Faculty of Organization and

Informatics, University of Zagreb. She received her Master’s degree in Economics from

the Faculty of Economics Zagreb, University of Zagreb. The author can be contacted

at [email protected].

Robert Fabac is a Professor of Organizational Design at the Faculty of Organization

and Informatics, University of Zagreb. He studied interdisciplinary postgraduate studies

for the needs of the Croatian army and received his Master’s degree from the

University of Zagreb. After that, he received his Ph.D. from the Faculty of Organization

and Informatics, University of Zagreb. His research interests include organizational

theory and design. The author can be contacted at [email protected].

144


Collaborative Strategic View in Corporate

Social Responsibility – Construction Industry

Case

Lana Lovrenčić Butković

University of Zagreb, Faculty of Civil Engineering, Croatia

Dina Tomšić

Zagreb Fair Ltd., Croatia

Simona Kaselj

HŽ Infrastruktura d.o.o., Croatia

Abstract

Background: The incorporation of corporate social responsibility (CSR) into the

business strategy of construction firms boosts their corporate reputation, while at the

same time reduces the risk and the external pressure for minimizing a negative societal

footprint. Objectives: This study aims to determine the current state of CSR in the

Croatian construction industry, in terms of knowing and practicing, and to offer a

collaborative strategic view as a viable CSR approach. Methods/Approach: A survey

research among large Croatian construction companies regarding CSR in the context

of collaboration with stakeholders was carried out and the results were analyzed using

the multidimensional unfolding procedure. Results: Results show that for the Croatian

construction companies CSR activities are important, but they are not widely seen as

a benefit to overall business strategies yet. Conclusions: Results of the research could

be helpful to construction firms in the efficient and effective stakeholder engagement,

as well as in the development of the calibrated CSR strategy.

Keywords: construction industry; corporate social responsibility; stakeholder

engagement; collaboration; non-financial reporting

JEL classification: M

Paper type: Research paper

Received: 19 Aug 2020

Accepted: 8 Sep 2020

Citation: Lovrenčić Butković, L., Tomšić, D., Kaselj, S. (2021), “Collaborative Strategic

View in Corporate Social Responsibility – Construction Industry Case”, Business Systems

Research, Vol. 12, No. 1, pp. 144-163.


145


Introduction A company's interactions and interdependencies with society are numerous and

complex. The contemporary complexity of the economic, social, environmental, and

governmental reality in which the firms operate brings the concept of corporate social

responsibility (hereafter CSR) to the center of the strategic concerns (Vishwanathan

et al., 2020; Atli et al., 2018). Furthermore, in disruptive times and the fast-changing

world, many of the existing business models have proven to be effective in the past,

but are no longer fully relevant. As explained in Keys et al. (2009), when it comes to

CSR, there are no easy answers on what to do or how to do it. In their view, CSR is

about doing good business and creatively addressing important issues facing business

and society. While supporting the authors' arguments, we would also like to emphasize

that the process of CSR implementation is dynamic. The most salient topics are

activism and movements in a wide range of activities that provoke public stand

against, in terms of all variety of non-responsible firm behavior that the construction

industry is exposed to.

The standpoint behind the dynamism of CSR strategy is the process of socially

constructed change anchored in the complexities of business reality, where firms

should be able to deal with myriads of difficulties in performing, complying, delivering

value, while simultaneously assuring sustainability. Stainer and Stainer (1998) explained

the fundamental principles of sustainable business as operating ethically, responsibly,

and profitably. Each of these three postulates tackles different aspects, so

implementing sustainable development at the organizational level requires a

governance model focused on long-term performance. As corporate goals in the

business domain are multidimensional and frequently contradictory, such a model is

likely to suffer from goal paradoxes (e.g., Bondy, 2008). Hence, to be aligned with

stakeholders’ expectations and the operating ecosystem, a firm’s behavior is

subjected to change. The magnitude and outcomes of the change are more certain

to be effective as well as efficient if the firm collaborates and engages its stakeholders’

web, and so to achieve the consensus of change content.

The integration of CSR activities into business strategy enables contribution to the

overall community and the environment, particularly in the sectors “where the

business activities generate substantial stakeholder interests” (Xia et al., 2018, p. 341).

Undoubtedly, the construction industry generates a large impact on the ecosystem

and community. Hence, it is important to observe the possibilities of implementing CSR

principles and activities into the operating mode to assure and sustain industry

responsibility and the requirements needed for its legitimate and sustainable

performance. Moreover, we find that the integration of CSR strategy and business

strategy of firms in the construction sector, among other things, suits favorably to

corporate disclosure and reputation, while at the same time reduces risk and external

pressure for minimizing negative society and environmental footprint.

Usually, firms disclose their CSR activities through non-financial reporting, by showing

the relevant information mainly covering its impact on society and the environment.

Non-financial reporting has become widely adopted by business enterprises globally

as stakeholders seek greater transparency on environmental, social, and governance

issues (Buallay, 2019). From 2018, non-financial statements have to be included in

companies’ annual reports. All major corporations are required to publish particular

information regarding social and environmental issues under the non-financial

reporting directive – NFRD (EU, 2014). In practice, there are a few international

standards such as reports, the most commonly used are ISO26000: Guidance on Social

Responsibility or The Global Reporting Initiative. In the construction sector, other

146


commonly used guidelines are the ECS 2000 standard, Social Accountability 8000, or

Global Compact Initiative.

Aside from those global standards, we deem that the industry-specific CSR

activities, as well as collaboration with stakeholders in their development and

implementation, are of high importance. The goal of this study is to investigate to what

extent Croatian construction companies are informed about CSR and non-financial

reporting; along with how socially responsible principles of business in the Croatian

construction sector are being applied. In total, 37 key areas of CSR activities have

been investigated, which are adapted from Jiang et al. (2016), and have been tested

within the Croatian context. Research findings specify 11 CSR activities of particular

interest for the construction industry, which we use as the basis to develop a

collaborative CSR strategy model.

The paper is composed as follows. After the introduction, in the second part, we

describe the theoretical background in the underlying topic of CSR and its

interconnectedness with corporate sustainability, collaborative strategy’s view, and

stakeholder engagement. Thirdly, we discuss the conceptual model and

methodology used in the study. In the fourth part, the research results and findings are

being used to sculpt a new conceptual model for shaping the industry-specific

collaborative CSR strategy. The model is dynamic and is aimed at sustaining CSR and

business performance, thus offering a new collaborative strategic CSR approach that

is not linked only to firm financial performance. The conclusion brings some potentially

useful managerial implications and highlights the trajectory for future research.

Theoretical background Though being rooted in the literature on business and society (Andriof et al., 2002),

nowadays, CSR has become a widely embraced phenomenon, being not just an

optional activity, but also a strategic driver of business (Atli et al., 2018; Aguinis et al.,

2012). There is no universally agreed-upon definition of CSR, leading to much confusion

about what constitutes a CSR activity (Keys et al., 2009). The concept is considered

multidimensional and normative (Carroll, 1991), which constitutes a CSR activity where

companies incorporate social and environmental concerns voluntarily into their

business operations and interactions with stakeholders. The UN Global Compact

defines CSR as a “firm’s delivery of long-term value in financial, environmental, societal

and ethical terms” (UNGC, 2016, p.9). The UN Global Compact identifies five

characteristics of CSR: principled business, community empowerment, executive

engagement, progress reporting, and local action. In addition, CSR is susceptible to

different interpretations in different contexts (Liyanage et al., 2016).

In sum, CSR today represents the entire range of activities and relationships that

arise between firms and the community. Therefore, CSR could be viewed as an

umbrella term encompassing “environmental sustainability, business ethics,

governance, public relations, stakeholder analysis, and corporate relations issues”

(Barthorpe, 2010, p.4). CSR can also be analyzed from a soft law perspective, where

a firm adopts self-regulation activities to assure its legitimacy, reputation, and

stakeholder alignment.

CSR activities are often equated with the broader terms of corporate responsibility,

stakeholder relations, sustainability, and sustainable development (Steurer et al.,

2005). According to Hillenbrand et al. (2007), corporate responsibility is used as a

broader term to describe the topics related to corporate responsibility, “that are

fundamental to all actions, decisions, behaviors, and impact of business” (Waddock,

2003, p.115). In contrast, CSR is often associated with the study of stakeholder

relationships. While CSR focuses on the relationship between companies and the

147


larger society, stakeholder relations management is about actually strategically

addressing the relationships between companies and the ecosystem (Tomsic, 2013a).

Furthermore, CSR is considered relevant for achieving sustainable development goals.

While sustainable development and corporate sustainability principles are heavily

influenced by “a society's interpretation, CSR is a voluntary management approach

in which the stakeholders of a company and their respective demands play an

important role” (Staurer et al., 2005, p.263).

Because construction has significant environmental and social impacts, the

association of the construction industry with sustainable development has blossomed.

(Sev, 2009). Besides, when the construction firm sets up sustainability as the firm goal,

it often sets up CSR policies for implementing necessary procedures (Shen et al., 2010).

The construction industry impacts and features The interdependence of the construction industry and society is immense. Since the

construction industry generates a large impact on every national economy and plays

a significant role in the development of every country, its relationship with society and

the environment is indisputable. Construction activities generate many “negative

impacts on the physical environment and society, such as dust and carbon emissions,

noise, waste, or air pollution” (Tam et al., 2007, p.1471). For example, according to

European Construction Industry Federation (2019), the total construction output in the

EU in 2018 was 1,427 billion €, which represents about 10.4% of EU GDP. Moreover, there

are 3.3 million entities active in the EU 28 construction sector, with 95% of firms with less

than 20 employees. The EU construction sector employs around 14.8 million workers,

accounting for 6.4% of total employment in the EU. According to that, health and

safety are also topics of great concern due to its labor-intensive feature, high exposure

to accident risks, and injury inclination.

According to Sev (2009, p.161), “sustainable construction principles can be

distinguished based on the three dimensions of sustainable development, which are

environmental, social, and economic”. The most comprehensive review of CSR

research in the construction industry is one of Xia et al. (2018). Their findings revealed

four research areas in the construction industry: “CSR perception, CSR dimensions CSR

implementation and CSR performance” (Xia et al., 2018, p.340). As pointed out by

Jiang et al. (2016), the construction industry devotes itself to societal progress in its

ways. Construction companies' efforts should be directed toward improving the

quality of life and relationships among communities, the built environment, the natural

environment, and living beings. Subject to social concerns, the challenge for the

construction sector in sustainability is to respond appropriately to increasing demands

from governments and the public. By doing so, we argue that the construction industry

can reach an aligned and collaborative relationship with the ecosystem it operates

in.

Linking CSR to a broader strategic and governance domain Previous research has suggested several approaches to exploring CSR issues. Some

approaches that can be found in the literature are the stakeholder approach, “a

social issues management approach, a social value chain approach” (Jiang et al.,

2016), or a social contract theory (Liyanage et al., 2016). Barraket et al. (2016) propose

a straightforward model that summarizes the four major CSR areas: “instrumental,

political, integrative, and ethical”. Vishwanathan et al. (2020) advanced the field by

discussing a causal strategic approach. Through a meta-analysis of the available

empirical evidence on the relationship between CSR and corporate financial

performance, they propose the idea of strategic CSR. They document four empirical

148


mechanisms that explain how CSR positively affects financial performance, namely

by improving corporate reputation, empowering stakeholder reciprocity, mitigating

corporate risk, and enhancing innovation capacity. Therefore, they have proposed

those mechanisms as distinguishing attributes between strategic CSR and traditional

CSR.

In terms of effectiveness, Jiang et al. (2016) pointed out that an effective CSR

strategy should be further defined to boost the corporate social performance of the

construction industry. Consequently, an effective paradigm is still missing. We find this

existing gap a challenging starting point. This is in line with Whittington's (2012) “big

strategy and small strategy view”, as well as with the view seeking to link strategy

research to broader societal issues (e.g., King et al., 2014; Vaara et al., 2012). Having

in mind firm strategy in general, Whittington (2012, p. 264) stated almost a decade

ago that: “This focus on financial outcomes is not a close fit with what a good deal of

contemporary business is about… Big strategy starts with impact: the firm strategies

that matter the most are those with the greatest repercussion, of all kinds. Big strategy

values big effects over large sample sizes.”. In brief, the designing paradigm for the

dynamic conceptual framework will be explained below. The framework is considered

to be well structured for construction industry-specific CSR activities, but not the kind

of one size fits all. On the contrary, the collaborative CSR model is aimed to be

idiosyncratic, dynamic, and firm-specific.

Conceptualization To design and sustain a dynamic and collaborative model, we build on the Pettigrew

triangle, in particular, his “what, why, and how of translating executive intentions into

realized change” (Pettigrew, 1987, p.649), as well on the contextualism, which raises

the issue of management practice (Sminia et al., 2012). We also use Hamel et al. (2014)

idea of building a change platform. For a firm to shape its CSR strategy, the former is

important for the implementation of strategy content research, as well as for

humanization of strategic management (Sminia et al., 2012), while the latter allows us

to shape dynamic, socially constructed, and change sensitive approach for an

effective CSR strategy, which is idiosyncratic, and collaborative.

The Pettigrew triangle As stated by Sminia et al. (2012), the problem that Pettigrew (1987) addressed aimed

at elucidating how exactly leaders' intentions can be translated into real

organizational changes. He has posited that integral to the organizational change

process is “the management of meaning” (Pettigrew, 1987, p. 659). In his triangle of

“context, content, and process, the context was subdivided into inner context, the

structure, corporate culture, and political context within the firm and the outer

context, the social, economic, political, and competitive environment”. Content

“refers to the particular areas of transformation under examination”, while process

“refers to the actions, reactions, and interactions from the various interested parties as

they seek to move the firm from its present to its future state” (Sminia et al., 2012, p.

1331 from Pettigrew, 1987, p. 657–658).

Thus, we propose implementing the strategic construction CSR framework into the

Pettigrew triangle of content, context, and process issue, especially his “why, what

and how questions of management practice”, because the firm is considered as a

nexus of ties and contracts in today's business world. This perspective stresses the

construction firm's connectivity with and embeddedness in the surrounding and

global ecology. Moreover, CSR vivid activities and practices indicate that their scope

is streaming beyond the classical social dimension towards a newly emerging CSR

149


practice dimension (Xia et al., 2018), sustained by a strong activism movement (King

et al, 2014). Consequently, contractual relations transform towards relational

contracting. This brings the relational assets of the firm to the fore, particularly

corporate reputation (Tomsic, 2013b).

Even though CSR literature generally draws on stakeholder management and

states that firms should have good knowledge about their stakeholders (e.g., Atli et

al., 2018; Surroca et al., 2010; Choi et al., 2009), it's critical to distinguish between two

major strategies to stakeholder management. The traditional strategy focuses on

screening stakeholders, but the proactive method focuses on interactions between

stakeholders (Harrison et al., 1996). Recent studies (e.g., McKinsey Quarterly, 2020b)

more strongly than previous research (e.g., Wu et al., 2008; Andriof et al., 2002)

emphasize the proactive approach and advocate the term "stakeholder

engagement" instead of "stakeholder management" to underline the value of

collaboration and partnership between the company and its stakeholders.

Within this article, the stakeholder engagement view is followed, and the market

and non-market arena of firm CSR performance are considered. We go beyond a

generic CSR paradigm and understand collaborative CSR strategies as a broader

term that encompasses a firm’s activities in an enlarged uncertain context,

embedded in the notion of the business ecosystem (Teece, 2007). By outlining and

integrating internal and external stakeholder engagement, and the firm’s purpose to

serve as a force for good for business and society, we aim not to cover the whole

scale and scope of CSR activities and performance. Instead, collaborative CSR

strategy is understood as being a core of firms’ integrative business strategy,

composed of both, competitive and collaborative dimensions (Tipuric et al., 2016).

Consequently, a collaborative CSR strategy protects a firm’s sustainability for creating

business and social value in a long term. Moreover, we find this process of value-

creating as being dynamic. The standpoint behind this newly spotted dynamism of

CSR strategy is the fact that we are dealing with a socially constructed change

process. Socially constructed need for change resides within a market arena, while

unspotted and unsatisfied needs, as well as future trends, have their contours in the

non-market arena. Hence, a good strategy should include both perspectives, even

though they sometimes compete and sometimes interact to generate firm

performance. Important is that they are balanced (Tipuric et al., 2016), so to operate

in a manner that “the combined whole is the greater than the sum of the parts”

(Makadok, 2003, p.1044). So, we find it suitable to follow Hamel et al. (2014) call for

firms to be designed as a change platform, not a change program.

Collaborative strategic view According to King et al. (2014), achieving strategic advantage relies heavily on the

ability of stakeholders to influence other people's perceptions. Consequently, much

of the strategic action that takes place as a result of these competitions are aimed at

shaping the perceptions of key target groups. Nonetheless, issues such as how firms

deal with stakeholder claims and conflicts are usually side-lined as secondary types of

strategy, i.e., political or non-market strategy (Mahon, 2002).

As explained in Tipuric et al. (2016), the traditional view of competitive, market-

based strategies has neglected the broader, so-called secondary strategies view,

aimed at the non-market arena for a long time. The narrow focus on competitive

advantage has clouded other possibilities to be evaluated as the sources of firm

sustainability, in particular those stemming from a collaborative advantage.

Collaborative advantage encapsulates the argument for synergy (Huxham et al.,

1992; Huxham, 2003). While this concept has provided useful insights for collaboration,

150


collaborative advantage, as defined more recently, is the advantage gained by a

firm as a result of its contribution to the ecosystem. By taking leadership in prominence

or addressing the issues that can reduce the gap between stakeholders’ expectations

and the corporate goals, it is unlikely to gain a direct effect on a company's financial

objectives or business model.

Tipuric et al. (2016) advance the idea that companies need a balanced

combination of perspectives and views of their strategy, so to become able to

compete in both market and non-market arenas. Activities taken in the market arena

are not isolated from non-market influences. So, companies need a balanced

strategy for achieving their goals in both markets and non-market arenas, viewing

them as equally important environments of operation. This strategic approach has

been labeled as a collaborative strategy. Consequently, we believe it is critical to

highlight that “the notion of advantage can also work in non-marketplace arenas,

notable areas such as public opinion, political and regulatory action, and social areas

in what has been termed the ‘marketplace of ideas” (Mahon, 2002, p.420). To

summarize, to be efficient and effective, the paradigm that is being followed for the

construction firm to design its good CSR strategic frame relies on the integration of its

competitive and collaborative strategy perspectives. As a result, Figure 1 depicts the

design of an integrative strategy-CSR-performance framework.

Figure 1

The Collaborative CSR strategy - performance framework

Source: Author’s illustration, adapted from Tomsic, D. (2013b), p.80

CHANGE PLATFORM

CSR GOALS AND STRATEGY

THE MANAGEMENT OF MEANING CORPORATE PRUPOSE, GOALS, MISSION, VISION AND VALUES

COLLABORATIVE CORPORATE SOCIAL RESPONSIBILITY DYNAMIC PROCESS

CORPORATE REPUTATION ECONOMIC-SOCIAL-ENVIRONMENTAL

CSR PERFORMANCE AND DISCLOSURE

ESTABLISH AND

ENFORCE

CONDUCT

CSR STRATEGY

IMPLEMENTATION

FIRM'S ACTIONS,

BEHAVIOR AND

COMMUNICATION

BUILDS OR

DESTROYS

BUSINESS BENEFITS

SO

CIE

TY

BE

NE

FIT

S

CSR PERCEPTION

SELF-CONTROL

SELF-CORRECTION SIGNALS

RISK EVALUATION

CO

NT

INU

UM

OF

ST

AK

EH

OLD

ER

IN

TE

RA

CT

ION

S

INTEN

TIO

NAL

AN

D E

XTERN

AL C

ORRECTIO

N

151


Research methodology Research instrument This study aims to determine whether Croatian construction firms are familiar with the

concept of CSR and whether this concept is being applied in the construction sector.

Therefore, a survey about key CSR activities specific to the construction industry has

been carried out. CSR activities are measured using the research instrument adapted

from Jiang et al. (2016) (Table 1), that elicits the most important CSR activities, grouped

into six key areas: F1: Environment preservation, F2: Construction quality and safety, F3:

Well-being of local community, F4: Employees’ interests, F5: Clients’ interests, and F6:

CSR institutional arrangements. For this study, we group those activities into three main

aspects: economic, environmental, and social (Figure 3), according to the GRI

standards. This approach also fits into Sev’s (2009) work where he defined sustainable

construction among the same three dimensions.

A questionnaire consists of three parts. In the first part, respondents were asked to

provide general information about themselves and the company (see Table 2). In the

second section, respondents were questioned about the strategic position of their

company and whether they were familiar with sustainability reporting (Table 3).

The third part contains the respondents’ judgment on CSR activities specific to the

construction firm. The CSR activities were adopted by Jiang et al. (2016) are shown in

Table 1. Responses were given using a five-point Likert scale (ranking from 1-negligible

to 5-extremely important) to indicate the significance of every CSR activity.

Table 1

CSR activities for construction firms

Code Activities

Act-1 Applying/optimizing a confidentiality system for customers' information

Act-2 Applying/optimizing a customer satisfaction management system in

responding to customers' claims

Act-3 Applying/optimizing an environmental impact assessment and precaution

system before construction

Act-4 Applying/optimizing precaution mechanism for safety management

Act-5 Applying/optimizing a training and education system for occupational skills

Act-6 Applying environmental technology and green energy to promote energy

saving and emission reduction

Act-7 Applying evaluation mechanism for collaborators to implement CSR

Act-8 Applying pollution emission control systems (e.g., gas, dust, noise, sewage,

and waste)

Act-9 Applying a post-construction service system and providing customers with

proper post-construction services

Act-10 Applying quality management certification

Act-11 Applying a saving and recycling system for resources and energy utilization

Act-12 Applying a selection, management, and supervision system for sub-

contractors

Act-13 Applying a strict quality inspection system for material and equipment

procurement

Act-14 Conducting green office

Act-15 Conducting research development and technological innovation to

improve quality and safety management level

Act-16 Establishing/improving employer-employee communication and negation

mechanism (e.g., Labour union)

Act-17 Establishing effective communication channels with the local community

Act-18 Establishing regular and effective communication mechanism with

customers

152


Table 1

CSR activities for construction firms (continued)

Code Activities

Act-19 Formulating/implementing CSR crisis precautionary and response

mechanisms

Act-20 Formulating/implementing a CSR training scheme

Act-21 Giving priority to the procurement of local products and services

Act-22 Guiding employees in career development and establishing an employee

promotion mechanism

Act-23 Implementing/optimizing environmental training scheme to improve

employees' environmental awareness and skills

Act-24 Implementing/optimizing a quality management system to strictly prevent

quality accidents

Act-25 Implementing/optimizing quality training to improve employees' quality

awareness and skills

Act-26 Implementing/optimizing a safety management system to prevent safety

accidents

Act-27 Implementing disaster prevention/relief activities for the society and local

community

Act-28 Implementing emergency mechanism and scheme for environmental

pollution accidents

Act-29 Organizing/supporting occupational skills training programs for the local

community

Act-30 Protecting biological diversity and ecological systems

Act-31 Respecting and protecting cultural tradition and heritage of the

community

Act-32 Setting up special division(s) for CSR management

Act-33 Setting up special units or/and positions to conduct daily environmental

management

Act-34 Strengthening communication with collaborators and improving

collaboration space and efficiency

Act-35 Supporting the development of infrastructure and public services of the

local community

Act-36 Taking care of employees and their families and help employees to achieve

work-life balance

Act-37 Taking care of low-income groups (e.g., Build-transferring low-income

housing without charge)

Source: Jiang and Wong (2016)

Sample and data This research identifies and analyzes the importance of CSR activities in the

construction sector. According to Croatian law and EU regulations, companies with

more than 500 employees are required to submit non-financial (sustainability) reports

on an annual basis, in which they list the CSR activities they perform. Given that, in

2018 there were only 7 Croatian construction companies that met this criterion. In this

research, we include all large construction companies that in 2018 had more than 250

employees (22 companies). Although large companies, which have less than 500

employees, are not required to publish non-financial reports, by looking at their

websites and business strategies, we noticed that they’ve been carried out certain

activities in the field of CSR. Furthermore, all of the large construction firms in Croatia

make major contributions to the development of the construction sector and also

have a significant share in total income.

153


As mentioned earlier, a total of 22 questionnaires were emailed to top managers

and 12 valid questionnaires were received, resulting in a 54.5 percent response rate.

This response rate exceeded the norm for construction survey research (Edum-Fotwe

et al., 2000). Despite a satisfactory rate of return, the sample is small, so this is a

significant limitation of this study.

Table 2 summarizes the final sample structure included in the research and Table 3

shows characteristics of the firms that participated in the research.

Table 2

Profile of respondents

Items Construction

Number of employees

251–500 6 50.0%

>500 6 50.0%

Basic activity

Building construction 3 25.0%

Civil engineering 5 41.7%

Both 2 16.7%

Both plus traffic handling, billing, maintenance 1 8.3%

Energy sector 1 8.3%

Position

General manager 5 41.7%

Technical manager 1 8.3%

Contracting director 1 8.3%

Manager of corporate communications 1 8.3%

EU fonds coordinator 1 8.3%

Project manager 1 8.3%

Quality manager 1 8.3%

Estimator 1 8.3%

Experience

Less than 5 years 0 0.0%

5 to below 10 years 1 8.3%

10 to below 20 years 6 50.0%

More than 20 years 5 41.7%

Statistical analysis First, the One-sample Wilcoxon-signed rank test was conducted, to determine if the

median sample values were significantly different from the test median of 3 - the

midpoint of the five-point Likert scale.

The Multidimensional Unfolding (MDU) procedure was conducted, attempting to

find a common quantitative scale that allows visual examining the relationships

between two sets of objects (IBM, 2020). MDU represents “data as distances among

points in a geometric space of low dimensionality” (Borg et al., 2005). MDU takes into

account two groups of objects - in this case, CSR activities & respondents and maps

both objects to the map.

154


Table 3

Firms’ characteristics

Items Construction

Defined business strategy

Yes 12 100.0%

No 0 0.0%

In preparation 0 0.0%

Mission & vision statement

Yes 12 100.0%

No 0 0.0%

Defined CSR strategy

Yes – in general strategy 4 33.3%

Yes – in mission statement 0 0.0%

Yes – in the vision statement 3 25.0%

Yes – in a separate statement 3 25.0%

No 2

Sustainable reporting

Yes 6 50.0%

No 6 50.0%

Sustainable reporting lead to a firm’s success

Yes 8 66.7%

No 4 33.3%

Familiar with CSR standards

GRI 2 16.7%

ISO 26000 3 25.0%

Both GRI & ISO 26000 3 25.0%

ISO 9001 & ISO 14001 2 16.7%

None 2 16.7%

Results Table 4 shows some differences compared to previous research on CSR activities

important for the construction sector. Our findings show that for large Croatian

construction companies CSR activities are important, but they have not widely seen

as beneficial to overall business strategies yet (at most activities have a mean of

around 4).

From these results, we could not conclude which of the three dimensions of

sustainable construction is the most important. The eleven most important activities

cover all three dimensions. For example, Act-11, Act-12, or Act-20 gained the same

score and the Act-11 belongs to the environmental dimension of sustainable

construction, Act-12 belongs to the social dimension and Act-20 belongs to the

economic dimension.

As we mentioned earlier, we have a small number of firms in the sample and did

not get any coherent results from factor analysis, so we tried with the Multidimensional

Unfolding methodology. The MDU method was performed in 768 iterations, included

98.5% variance in the data, and explains 38.9% of the variations in the respondents7

responses. A measure of the overall correlation, the Spearman Rho coefficient, is

almost 0.6 (Table 5).

155


Table 4

Respondents’ perspective on the importance of CSR activities Activity

code

Intensity in % One-sample Wilcoxon-signed rank

test median value = 4,0

5-

extremely

important

4-

important

3-

neutral

2-not

important

Mean Median Test

statistic

Sig

Act-01 41.7 50.0 8.3 0.0 4.3 4.0 57.00 0.026

Act-02 41.7 50.0 8.3 0.0 4.3 4.0 36.00 0.009

Act-03 33.3 50.0 16.7 0.0 4.2 4.0 55.00 0.004

Act-04 333 50.0 16.7 0.0 4.2 4.0 55.00 0.004

Act-05 58.3 33.3 8.3 0.0 4.5 5.0 66.00 0.002

Act-06 25.0 58.3 16.7 0.0 4.1 4.0 55.00 0.004

Act-07 0.0 58.3 41.7 0.0 3.6 4.0 28.00 0.008

Act-08 50.0 41.7 8.3 0.0 4.4 4.5 66.00 0.003

Act-09 41.7 50.0 8.3 0.0 4.3 4.0 66.00 0.003

Act-10 33.3 58.3 8.3 0.0 4.3 4.0 66.00 0.002

Act-11 58.3 41.7 0.0 0.0 4.6 5.0 78.00 0.002

Act-12 58.3 41.7 0.0 0.0 4.6 5.0 78.00 0.002

Act-13 25.0 66.7 0.0 8.3 4.1 4.0 73.00 0.005

Act-14 16.7 66.7 16.7 0.0 4.0 4.0 55.00 0.003

Act-15 25.0 41.7 33.3 0.0 3.9 4.0 36.00 0.009

Act-16 41.7 33.3 25.0 0.0 4.2 4.0 36.00 0.009

Act-17 33.3 58.3 8.3 0.0 4.3 4.0 66.00 0.002

Act-18 58.3 33.3 8.3 0.0 4.5 5.0 66.00 0.002

Act-19 16.7 66.7 16.7 0.0 4.0 4.0 55.00 0.003

Act-20 58.3 41.7 0.0 0.0 4.6 4.0 28.00 0.008

Act-21 16.7 66.7 16.7 0.0 4.0 4.0 55.00 0.003

Act-22 50.0 50.0 0.0 0.0 4.5 4.5 78.00 0.002

Act-23 66.7 25.0 8.3 0.0 4.6 5.0 66.00 0.002

Act-24 50.0 50.0 0.0 0.0 4.5 4.5 78.00 0.002

Act-25 50,0 41.7 8.3 0.0 4.4 4.5 55.00 0.004

Act-26 58.3 41.7 0.0 0.0 4.6 5.0 78.00 0.002

Act-27 33.3 25.0 33.3 8.3 3.8 4.0 33.50 0.026

Act-28 50.0 33.3 16.7 0.0 4.3 4.5 55.00 0.004

Act-29 8.3 41.7 41.7 8.3 3.5 3.5 24.50 0.058

Act-30 25.0 66.7 8.3 0.0 4.2 4.0 55.00 0.004

Act-31 16.7 50.0 16.7 0.0 3.3 4.0 45.00 0.006

Act-32 8.3 16.7 50.0 25.0 3.1 3.0 12.00 0.739

Act-33 16.7 33.3 41.7 8.3 3.6 3.5 25.00 0.053

Act-34 41.7 50.0 8.3 0.0 4.3 4.0 66.00 0.003

Act-35 16.7 58.3 25.0 0.0 3.9 4.0 45.00 0.005

Act-36 58.3 33.3 8.3 0.0 4.5 4.0 66.00 0.002

Act-37 0.0 83.3 16.7 0.0 3.8 4.0 55.00 0.002

Note: None of the respondents rated any CSR activity as 1-negligible so this column is not

shown in the table

156


Table 5

Summary of MDU method

Iterations 768

Goodness of Fit Dispersion Accounted For 0.985

Variance Accounted For 0.389

Spearman's Rho 0.599

For this research, we conducted a 2-dimensional MDU where the distance between

variables and respondents can be seen on a joint plot in a two-dimensional distance

space concerning the importance of the CSR activities (Figure 2).

Figure 2

Dissimilarity map

Source: Author’s illustration

From the answer distance map shown in Figure 3, it is noticeable that there are two

groups of statements. Eleven statements were grouped on one side of the dimension

and the other statements on the other side of the dimension. The statements that

stand out on the right side of dimension 1 are: Act-04, Act-05, Act-06, Act-12, Act-16,

Act-25, Act-27, Act-29, Act-32, Act-33, Act-34 and they gained the best scores.

According to the distribution of objects representing activities on the map, it can

be also seen that they do not follow the above grouping of activities into three groups,

at least not given the similarities of responding to the activities. Moreover, it is important

to consider the small number of participants due to which it was not possible to

replicate the analysis, i.e., factor analysis, but a completely different method of

analysis was performed, suitable for the data obtained in the study. On the other

hand, it is also visible how the respondents can be visually divided into two groups. In

one smaller group, there are three respondents and in the other, there are 9

respondents. If we look at the average score on all activities in total, it can be seen

157


that the three respondents who stand out, have an overall average score on all

activities slightly lower than they have the other respondents (Table 6) on all activities.

Table 6

Results for respondents using the MDU method

Respondents The average score of all activities

1 4.00

2 4.38

4 4.05

5 4.43

8 3.89

9 4.03

10 4.43

11 4.14

12 4.68

Group average 4.23

3 3.76

6 3.73

7 3.54

Group average 3.68

Average of all activities 4.09

The results presented in Table 6 show that out of 3 companies that have lower

ratings of the importance of CSR activities, there're 2 companies with 500 or more

employees. Those firms are required by Croatian law to compose non-financial

reports. In this sense, knowledge of CSR activities should be an important segment of

their overall business strategy. Finally, the local action is focused on the different needs

and sensibilities that different countries and areas are expressing, a firm that is acting

in an international scenario must consider that the different stakeholders will make

different claims, for example, if they are both workers associations but representing

employees working for the firm in two different countries (UNGC, 2016).

Discussion The aforementioned activities, which are concerned with business's environmental,

social, and economic issues, are viewed as a solid foundation for the development of

construction firms' CSR strategy. As presented in the model, by shaping multilateral

relations among the firm and its stakeholders, as well as by connecting self-correction

mechanisms of firm’s behavior, we have constructed the change platform as a

mechanism enabling us to deal with both of the CSR perspectives, prescriptive aimed

to sustain firm’s legitimacy, and descriptive aimed to sustain firm’s sustainable

performance. Moreover, the model is designed to permanently attract management

attention to a new source of change initiatives, still not visible at present, but significant

for the future overall strategic posture of the firm. Furthermore, the model shifted

stakeholder issues from relationship management to collaborative relationship

building appropriate for the confidential exchange of ideas, requirements, skills, and

information, bringing a fresh dynamic perspective to the strategic CSR approach.

New ideas, knowledge, and information for sustaining a construction firm's

strategies and performance are available both inside and outside the firm's

boundaries. Furthermore, values serve as the guiding principles for the lives of

individuals (Anderson et al., 2014). The change in pace, as well as behaviors that are

strongly opposed to the values of stakeholders, would be unwelcome, and the desired

outcome would not be realized. Therefore, managerial and organizational activities

158


in the construction firm are to be aligned, which is a demanding task. Thus, we have

proposed Pettigrew’s (1987) management of meanings as a comprehensive

approach for management to delve into the firm’s stakeholders' web expectations,

and to define the firm’s purpose (McKinsey Quarterly, 2020a), to calibrate with more

accuracy, the firm’s CSR activities. Having in mind Augier et al. (2009, p.411) saying

that “a firm excellent at making the wrong things will fail”, it seems noteworthy for the

construction firm to sculpt CSR strategy within Pettigrew’s triangle. The context and

action-based framework could objectify the decision-making situation, so that

construction firm’s managers can bring tamely decisions and undertake purposeful

CSR actions.

The firm is viewed as an integrative system of stakeholder relationships in the model.

The initial research perspective underlying the dynamism of CSR strategy was that we

are dealing with a socially constructed change process. Hence, the construction firm

should become capable of making the “deep dives” in its stakeholders’ web, to gain

an understanding of their salient claims and often conflicting goals. The problem

solution possibly comes from Mitchell et al. (1997, p.853) “theory of stakeholder

identification and salience”, which is commonly used to resolve goal paradoxes.

Consequently, the construction firm's management should create a typology of

stakeholders and identify their CSR-related salient claims at a given point in time.

Within the dynamic view, context and content analysis, as well as self-regulating

mechanisms interpolated in the model, might be supportive to align construction firm

stakeholders’ expectations, without significant gaps.

The model is also intended to serve as a driver for the content of stakeholder

engagement activities. The interconnection of CSR strategy with the firm's goals,

mission, vision, and values, as well as the underlying processes, enables measuring the

firm's economic, financial, environmental, and social accomplishments, as well as the

firm's overall business, financial, environmental, or social reputation. Though the

chosen approach may be evaluated as over-compliance, the construction firm's

behaviors, as well as its reputation, are undoubtedly its strategic relational assets (e.g.,

Dyer et al., 1998 for review). The important perspective for the construction firms is the

mode of interacting. “Corporate interactions take place both in the marketplace of

goods and services (where strategy is central) and in the marketplace of ideas (where

corporate social performance and political strategy research are central) (Mahon,

2002, p.417)”. Therefore, it seems wise to shape the collaborative CSR strategy and

make the most of it.

A collaborative CSR strategy could protect the construction firm from unpleasant

occurrences in contemporary business reality, notably from the public's increasing

concern for the environment and the free flow of information made possible by the

Internet and social networks. The issues that the public considers important should be

considered by the construction company management, to avoid being labeled as

socially irresponsible and risking the relationship with all of its stakeholders. By taking a

collaborative strategic approach to CSR, due to its dynamic nature, a construction

firm’s management can determine purposeful CSR activities and calibrate the

required resources to deploy, and so devotes its behavior towards. Thus, the firm

assures to be regarded as socially responsible, while at the same time sustains its

competitive and collaborative advantage. Balancing the creation of the economic

value with that of the societal value is a challenging activity, but a firm successful in

engaging stakeholders can gain a lot from those multilateral relationships, especially

in terms of reducing risks and seizing promising opportunities.

A construction firm should pay special attention to disclosure. When delivering non-

financial statements, by highlighting the firm’s collaborative CSR activities, the

159


management can gain and sustain favorable social, environmental, or overall

business reputations and wins stakeholders’ and public’ hearts and minds. To be able

to, we find Morsing et al. (2006) work valuable for a revisit, and recommend the

stakeholder involvement strategy, which assumes a continuous stakeholder dialog

upon CSR activities, and ongoing and systematic interaction, instead of imposing a

specific CSR initiative. In particular, these authors proposed three approaches to CSR

communication strategy, namely: “stakeholder information strategy, stakeholder

response strategy, and stakeholder involvement strategy”. Though all three CSR

communication strategy approaches can be applied interchangeably, in today’s

digital reality, online communication surpasses yearly publishing report obligation,

therefore endorsement of CSR activities should fit the business situation and be

delivered timely. To be able to develop sustainable business practices, we find it

important for a construction firm to empower its ability to enact strategically its

productive relationships. Collaborative CSR strategy is aimed to enable and sustain

such a commitment process.

Conclusion In this paper, we looked into the situation of CSR in Croatian construction large-size

firms at the moment. We have adopted and combined Jiang et al.’s (2016) 37 CSR

activities of particular interest for the construction industry and broader strategic issues

to develop our collaborative CSR strategy model. The model is industry-specific,

aimed at sustaining CSR and business performance, and is dynamic, thus offering a

fresh view of strategic CSR that is not linked only to firm financial performance.

The research results unveil many industries and country-specific CSR principles and

practice for the construction sector. Our findings indicate that CSR activities are

important, but managers do not see them as beneficial to overall business strategies

yet. The conceptualized model might be of help for the management of the

construction firms to reconsider the perspective, envision the advantages of

collaborative CSR strategy, and implement it in the business strategy. Although the

distribution of objects representing activities on the distance map shows that results

do not follow three dimensions of sustainable construction, defined in previous works,

at least not given the similarities of responding to the activities, such output could

occur due to the small sample or due to the lack of CSR knowledge.

Since a construction firm should pay special attention to disclosure, by highlighting

the firm’s collaborative CSR activities, the management can gain and sustain

favorable social, environmental, or overall business reputations. The model with a self-

regulating mechanism points too. Another benefit offered by collaborative CSR

strategy is the better calibration of CSR activities design, due to the dynamism of the

feedback loop included in the process. As stakeholders’ expectations are subject to

change, so is the operating ecosystem. By following the stakeholder engagement

approach, managers could delve into socially constructed perceptions of

stakeholders regarding the firm CSR performance and behavior. Hence, the

magnitude and outcomes of the change are more certain to be aligned, effective

and efficient, due to a consensus of change content. By using the dynamic

collaborative approach to CSR strategy, CSR activities can be adjusted and

calibrated over time, so to fit the requirements for a firm’s responsible behavior.

The study contributes to CSR field development empirically and conceptually. The

scientific contribution of our work is reflected in the fact that this type of research is

being carried out for the first time within the Croatian construction industry, and that a

new collaborative approach to the CSR strategy was created. There are several

potentially useful managerial implications of our work. They are reflected in the

160


research findings, which identify the most important activities that construction

companies could engage in to help develop their CSR strategy. The question of why

Croatian construction firms are still unaware of the significance of CSR and its

application in defining overall business strategy remains unanswered.

The limitations of this work refer to a small sample of research units, as well as to only

one country involved. However, in Croatia, the total number of large construction

companies is small (22 in total), and even fewer are required to publish non-financial

reports (only seven). Second, using managers of large construction companies as the

only stakeholder group gives a one-dimensional view of the situation. Expanding the

study and including other key stakeholders (such as material manufacturers, suppliers,

or the government) would provide a more comprehensive view of the research

context. The direction for further research might be to include middle and small firms

and find out which activities, if any, are being carried out, and amongst find out the

most common activities pointing to a Croatian construction industry practice. Another

research trajectory could be to investigate the effectiveness of CSR communication

activities and channels in the construction industry in the digital era, as well as to revisit

the mechanisms and drivers of CSR strategy implementations.

References 1. Aguinis, H., Glavas, A. (2012), “What we know and don’t know about corporate social

responsibility: A review and research agenda”, Journal of Management, Vol. 38 No. 4, pp.

932–968.

2. Anderson, N., Potočnik, K., Zhou, J. (2014), “Innovation and Creativity in Organizations: A

State-of-the Science Review, Prospective Commentary, and Guiding Framework”, Journal

of Management, Vol. 40 No. 5, pp. 1297-1333.

3. Andriof, J., Waddock, S. (2002), “Unfolding stakeholder Engagement”, in Sutherland

Rahman, S., Waddock, S., Andriof, J., Husted, B. (Ed.), Unfolding Stakeholder Thinking,

Greenleaf, London, UK.

4. Atli, D., Vidovic, M., Omazic, M. A. (2018), “Communicating Corporate Social Responsibility

on the Web: A Content Analysis of Sustainability Reports in World’s Best Multinational

Workplaces”, International Journal of Sustainable Entrepreneurship and Corporate Social

Responsibility, Vol. 3 No. 2, pp. 1-17.

5. Augier, M., Teece, D. J. (2009), “Dynamic Capabilities and the Role of Managers in Business

Strategy and Economic Performance”, Organization Science, Vol. 20 No. 2, pp. 410-421.

6. Barraket, J., Keast, R., Furneaux, C. (2016), Social procurement and new public

governance, Routledge, London.

7. Barthorpe, S. (2010), “Implementing corporate social responsibility in the UK construction

industry”, Property Management, Vol. 28 No. 1, pp. 4-17.

8. Bondy, K. (2008), “The paradox of power in CSR: A case study on implementation”, Journal

of Business Ethics, Vol. 82 No. 2, pp. 307-323.

9. Borg, I., Groenen, P. J. F. (2005), Modern Multidimensional Scaling: Theory and Applications,

2nd Ed., Springer, New York, USA.

10. Buallay, A. (2019), “Is sustainability reporting (ESG) associated with performance? Evidence

from the European banking sector”, Management of Environmental Quality: An

International Journal, Vol. 30 No. 1, pp. 98-115.

11. Carroll, A. B. (1991), “The Pyramid of Corporate Social Responsibility: Toward the Moral

Management of Organizational Stakeholders”, Business Horizons, Vol. 34 No. 4, pp. 39–48.

12. Choi, J., Wang, H. (2009), “Stakeholder relations and the persistence of corporate financial

performance”, Strategic Management Journal, Vol. 30 No. 8, pp. 895-907.

13. Dyer, J.H., Singh, H. (1998), “The Relational View: Cooperative Strategy and Sources of

Interorganizational Competitive Advantage”, Academy of Management Review, Vol. 23

No. 4, pp. 660-679.

161


14. Edum-Fotwe, F. T., McCaffer, R. (2000), “Developing project management competency:

perspectives from the construction industry”, International Journal of Project Management,

Vol. 18 No. 2, pp. 111-124.

15. EU (2014), “Directive 2014/95/EU” available at https://eur-lex.europa.eu/legal-

content/EN/TXT/PDF/?uri=CELEX:32014L0095&from=EN (18 June 2020).

16. European Construction Industry Federation (2019), “Key figures activity in 2018”, FIEC

17. Hamel, G., Zanini, M. (2014), “Build a change platform, not a change program”, Mckinsey

& Company, available at http://www.mckinsey.com/Insights/Organization, (11 November

2014).

18. Harrison, J. S., St. John, C. H. (1996), “Managing and Partnering with External Stakeholders”,

The Academy of Management Executive, Vol. 10 No. 2, pp. 46-60.

19. Hillenbrand, C., Money, K. (2007), “Corporate Responsibility and Corporate Reputation:

Two Separate Concepts or Two Sides of the Same Coin?”, Corporate Reputation Review,

Vol. 10, pp. 261–277.

20. Huxham, C. (2003), “Theorizing Collaboration Practice, Public Management Review, Vol. 5

No. 3, pp. 401-423.

21. Huxham, C., Macdonald, D. (1992), “Introducing collaborative advantage: Achieving

interorganizational effectiveness through meta-strategy”, Management Decision, Vol. 30

No. 3, pp. 50–56.

22. IBM (2020), “Multidimensional Unfolding (PREFSCAL), available at

https://www.ibm.com/support/knowledgecenter/en/SSLVMB_25.0.0/statistics_mainhelp_

ddita/spss/categories/idh_prfs.html /(18 June 2020).

23. Jiang, W., Wong, J. (2016), “Key activity areas of corporate social responsibility (CSR) in the

construction industry: A study of China”, Journal of Cleaner Production, Vol. 113, pp. 850-

860.

24. Keys, T., Malnight, T. W., van der Graaf, K. (2009), “Making the most of corporate social

responsibility”, McKinsey Quarterly, pp. 1-8.

25. King, B. G., Walker, E. T. (2014), “Winning Hearts and Minds: Field Theory and the Three

Dimensions of Strategy”, Strategic Organization, Vol. 12 No. 2, pp. 134-141.

26. Liyanage, C., Ulutas¸ D. D., Giritli, H., McDermott, P. (2016), “Corporate social responsibility

in construction industry: a comparative study between UK and Turkey”, Built Environment

Project and Asset Management, Vol. 6 No. 2, pp. 218-231.

27. Mahon, J. F. (2002), “Corporate reputation: Research agenda using strategy and

stakeholder literature”, Business and Society, Vol. 41 No. 4, pp. 415-445.

28. Makadok, R. (2003), “Doing the right thing and knowing the right thing to do: why the whole

is greater than the sum of the parts”, Strategic Management Journal, Vol. 24 No. 10, pp.

1043-1055.

29. McKinsey Quarterly (2020a), “Purpose: Shifting from why to how”, April, 2020, available at:

https://www.mckinsey.com/~/media/mckinsey/business%20functions/organization/our%2

0insights/purpose%20shifting%20from%20why%20to%20how/purpose-shifting-from-why-to-

how-v3.pdf?shouldIndex=false (10 May 2021)

30. McKinsey Quarterly (2020b), “The pivotal factor for effective external engagement”, May,

2020, available at:

https://www.mckinsey.com/~/media/mckinsey/business%20functions/strategy%20and%2

0corporate%20finance/our%20insights/the%20pivotal%20factors%20for%20effective%20ex

ternal%20engagement/the-pivotal-factors-for-effective-external-

engagement.pdf?shouldIndex=false (10 May 2021)

31. Mitchell, R. K., Agle, B. R., Wood, D. J. (1997), “Toward a theory of stakeholder identification

and salience: Defining the principle of who and what really counts”, Academy of

Management Review, Vol. 22 No. 4, pp. 853–886.

32. Morsing, M., Schultz, M. (2006), “Corporate social responsibility communication:

stakeholder information, response and involvement strategies”, Business Ethics: A European

Review, Vol. 15 No. 4, pp. 323-338.

33. Pettigrew A. M. (1987), “Context and Action in the Transformation of the Firm”, Journal of

Management Studies, Vol. 24, pp. 649–670.

162


34. Sev, A. (2009), “How can the construction industry contribute to sustainable development?

A conceptual framework”, Sustainable Development, Vol. 17 No. 3, pp. 161-173.

35. Shen, L.-Y., Tam, V. W., Tam, L. Tam, Ji, Y.-B. (2010), “Project feasibility study: the key to

successful implementation of sustainable and socially responsible construction

management practice”, Journal of Cleaner Production, Vol. 18 No. 3, pp. 254-259.

36. Sminia, H., de Rond, M. (2012), “Context and Action in the Transformation of Strategy

Scholarship”, Journal of Management Studies, Vol. 49 No. 7, pp. 1329–1349.

37. Stainer, A., Stainer, L. (1998), “Business Performance – a Stakeholder Approach”,

International Journal of Business Performance Management, Vol. 1 No. 1, pp. 2-12.

38. Staurer, R., Langer, M. E., Konrad, A., Martinuzzi, A. (2005), “Corporations, Stakeholders and

Sustainable Development I: A Theoretical Exploration of Business-Society Relations”, Journal

of Business Ethics, Vol. 61, pp. 263-281.

39. Surroca, J., Tribó, J. A., Waddock, S. (2010), “Corporate responsibility and financial

performance: the role of intangible resources”, Strategic Management Journal, Vol. 31 No.

5, pp. 463-490.

40. Tam., V. W. Y., Shen, L. Y., Yau, R. M. Y., Tam, C. M. (2007), “Assessing the levels of material

wastage affected by sub-contracting relationships and projects types with their

correlations”, Building Environment, Vol. 42 No. 3, pp. 1471-1477.

41. Teece, D. J. (2007), “Explicating dynamic capabilities: The nature and microfoundations of

(sustainable) enterprise performance”, Strategic Management Journal, Vol. 28 No. 13, pp.

1319-1350.

42. Tipuric, D., Tomsic, D. (2016), “Collaborative strategies to heal the competitive paradigm

shortcuts”, Tipuric, D., Kovac, I. (Ed.), Proceedings of the 4th International OFEL Conference

on Governance, Management and Entrepreneurship: New Governance for Value

Creation, Towards Stakeholding and Participation Zagreb: CIRU - Governance research

and Development Centre, pp. 1247-1260.

43. Tomsic, D. (2013a), “The Role of Firm Reputation in Corporate Governance Processes”,

Tipurić, D., Podrug, N. (Ed.), Proceedings of the 1st International OFEL Conference on

Corporate Governance, Zagreb: CIRU - Governance research and Development Centre,

pp. 845-862.

44. Tomsic, D. (2013b), “Towards Relational View of Corporate Governance”, Montenegrin

Journal of Economics, Vol. 9 No. 2, pp. 71-87.

45. United Nations Global Compact (2016), “Guide to Corporate Sustainability, Shaping a

sustainable future”, available at

https://d306pr3pise04h.cloudfront.net/docs/publications%2FUN_Global_Compact_Guide

_to_Corporate_Sustainability.pdf (30 January 2020)

46. Vaara E., Durand, R. (2012), 2How to make strategy research connect with broader issues

that matter?”, Strategic Organization, Vol. 10 No. 3, pp. 248-255.

47. Vishwanathan, P., van Oosterhout, H. J., Heugens, P. P. M. A. R., Duran, P., van Essen, M.

(2020), “Strategic CSR: A Concept Building Meta‐Analysis”, Journal of Management

Studies, Vol. 57 No. 2, pp. 314-350.

48. Waddock, S. (2003), “Stakeholder performance implications of corporate responsibility”,

International Journal of Performance Management, Vol. 5 No. 2/3, pp. 114–124.

49. Whittington, R. (2012), “Big Strategy/Small Strategy”, Strategic Organization, Vol. 10 No. 3,

pp. 263-268.

50. Wu, M., Eweje, G. (2008), “Stakeholder engagement as a facilitator of organizational

learning”, Department of Management and International Business Research Working

Paper Series 2008, no. 1. Auckland, NZ: Massey University

51. Xia, B., Olanipekun, A., Chen, Q., Xie, L., Liu, Y. (2018), “Conceptualising the state of the art

of corporate social responsibility (CSR) in the construction industry and its nexus to

sustainable development”, Journal of Cleaner Production, Vol. 195, pp. 340-353.

163


About the authors

Lana Lovrenčić Butković, Ph.D. is an Assistant Professor at the University of Zagreb,

Faculty of Civil Engineering, the Department of Construction Management. She

received a Ph.D. in Business Economics at the Faculty of Economics and Business

Zagreb with the dissertation thesis “Influence of key success factors for construction

companies when choosing business strategy performances on foreign markets”. She

teaches economic courses at the Faculty of Civil Engineering Zagreb and she is a

secretary of Interdisciplinary Postgraduate Study MBA in Construction at the University

of Zagreb. Her main research interest is focused on the field of strategic management

of construction companies, their internationalization, and the strategy of the socially

responsible business of construction companies. The author can be contacted at

[email protected].

Dina Tomšić, Ph.D. is CEO at the Zagreb Fair Ltd. and a Senior Lecturer at three

University Colleges in Zagreb. She received a Ph.D. in Strategic Management and

Corporate Governance at the Faculty of Economics and Business with the dissertation

thesis „The Role of Corporate Reputation in building Dynamic Capabilities of Firm“. She

was also educated at the Diplomatic Academy in Zagreb in the field of Diplomacy,

and at the University of Nevada in Las Vegas in the field of Event Management. Her

main research interests are strategic management and corporate governance

disciplines. The author can be contacted at [email protected].

Simona Kaselj, mag.ing.aedif. is Chief engineer for railway structures at the HZ

Infrastruktura d.o.o. She graduated from the Faculty of Civil Engineering, University of

Zagreb, with the thesis “Corporate Social Responsibility in Croatian Construction

Companies”. Her main interest is focused on the development of Croatian

infrastructure companies. The author can be contacted at [email protected].

164


Cultural Tourism and Community

Engagement: Insight from Montenegro

Ilija Moric, Sanja Pekovic, Jovana Janinovic

University of Montenegro, Montenegro

Đurđica Perovic

University of Montenegro, Faculty of Tourism and Hotel

Management, Montenegro

Michaela Griesbeck

University of Vienna, Austria

Abstract

Background: Cultural tourism in Montenegro is growing, mostly due to the integral

growth and development of tourism products. However, an in-depth insight into the

relationship between cultural tourism and community engagement is missing.

Objectives: The paper aims to examine the relationship between cultural tourism

development and community engagement in Montenegro. Methods/Approach:

Using the extensive literature, available secondary data, and an analysis of relevant

policies, the paper explores new possibilities for diversifying tourism offer at heritage

sites, by engaging volunteers, enhancing understanding of the socio-historical

background, promoting the usage of digital tools, partnering with relevant

stakeholders, introducing innovative funding tools and schemes. Results: Several

management issues associated with heritage tourism and community participation

are acknowledged. Conclusions: Key findings indicate the need for a systemic,

dynamic, and innovative framework for sustainable and highly impactful heritage

tourism in Montenegro, which policymakers, heritage ventures, and other stakeholders

might use to strengthen community engagement and development at the heritage

sites.

Keywords: cultural tourism, heritage, community engagement, Montenegro

JEL classification: M21


Received: Mar 28, 2020

Accepted: Jun 08, 2020

Citation: Moric, I., Pekovic, S., Vukčević, J., Perović, Đ., Grisbeck, M. (2021), “Cultural

Tourism and Community Engagement: Insight from Montenegro”, Business Systems



165


Introduction According to UNWTO, cultural tourism attracts visitors, which are motivated to learn,

discover, experience, and consume the tangible and intangible cultural

attractions/products in a tourism destination (UNWTO, 2019). Interestingly, key features

of cultural tourism are quite diverse, including tangible heritage (art, architecture,

historical, cultural, etc.) and intangible components (e.g., music, lifestyles, value

systems, beliefs, traditions, etc.). Worldwide, cultural tourism covers at least 20% of

arrivals in towns in the sense that culture is the main attraction to visit, learn, discover,

and experience (MSDT, 2019). This percentage is even higher if leisure tourists are taken

into account due to their interest in the local culture, especially food, music, events.

Additionally, cultural tourists are well known for being more respectful towards cultural

and natural resources, with higher education and higher expenditure (Hughes, 1987).

On the other side, the local community is playing important role in the general

economic and social development of a certain geographical area. Several authors

(Cooper et al., 2005) stress the significance of community engagement in sustainable

tourism development, in economic (e.g., generation of income), socio-cultural (e.g.,

provision of employment and reduction of poverty), environmental (e.g., nature

protection) and sense of tourist satisfaction (e.g. added value to total experience).

Therefore, successful sustainable tourism development has to respond to the

expectations, needs of hosts, and guests alike (UNWTO, 2018). Consequently, the

valorization of cultural resources cannot be planned and managed without a host

community and its active engagement.

From a visitor's point of view, authentic cultural contact and experience need the

presence of local and professional intermediaries and interpreters (Salazar, 2012).

However, in reality, this role is not always played efficiently due to several reasons.

Chirikure et al. (2010) show that participatory management in the heritage industry

could have various effects, pointing out that success depends on the local situation,

type of cultural heritage, the general context of development, etc. Crooke (2010)

emphasizes the importance of different determinants such as motivations, issues of

authority, and the value of community‐heritage engagement in the function of

management and development of cultural tourism. These findings are important to

understand the complexity of the community participation process in the context of

tourist experience delivery. Due to this, community engagement represents a critical

factor for the sustainable development of cultural tourism and their link is necessary to

explore and explain. In addition, attention has to be focused on the importance of

innovation and technology that can improve governance, profits, and the wellbeing

of residents as well as empower local populations and communities, especially in

terms of retaining their authenticity (UNWTO, 2018). Key benefits could be obtained in

terms of better preservation of intangible and tangible heritage resources and

consequently higher quality of the tourist experience. Apart from technology and

innovation, new trends in visitor management have to be examined to amplify cultural

tourism’s attractiveness while managing appropriately numerous visitors

(UNWTO/UNESCO, 2018). This could engender various benefits to visitors and host

communities while preserving cultural values.

Inevitably, cultural tourism strategies must consider the interests and expectations

of the local community (UNWTO/UNESCO, 2018). In other words, community

engagement is often seen as a means of implementing sustainable tourism (Okazaki,

2008). According to conclusions of the UNESCO Convention on the safeguarding of

the intangible cultural heritage, it is quite important to foster roles of local cultural

communities, especially their participation in safeguarding, where this term refers to

different activities that include identification, research, preservation, promotion,

166


enhancement and the revitalization of the various aspects of intangible cultural

heritage (UNESCO, 2003). In a practical sense, this could be achieved via different

frameworks and mechanisms. UNWTO and UNESCO (2018) promote the necessity for

a stronger role of the local community in heritage conservation and safeguarding in

a way that tourism revenue has to be redirected towards these two challenges. But

direct financial support is often not enough. So, merging creativity and technological

innovation among the host community, as well as protecting heritage, are also seen

as essential for promoting responsible and sustainable tourism (UNWTO/UNESCO,

2018). All these recommended issues related to community engagement have to be

carefully examined and included in the new, dynamic, flexible, and systemic

framework that can empower the local community, culture, and tourism

development.

Several authors examine processes and activities in tourism from the system

approach. Perovic et al. (2018) investigate the influence of tangible and intangible

dimensions of tourism offer on tourist intention to return to Montenegro and the tourist

overall satisfaction using a systemic approach. Moric (2013b) proposes a systemic

cluster approach in developing rural tourism. A similar systems approach to tourism as

an example of a complex system could be found in other reteaches (Lazanski et al.,

2006; Gmür et al., 2010). In addition, Pejic-Bach et al. (2014) suggest the usage of a

system approach in the area of tourism organization management with a focus on

socially responsible behavior. There is a need for a new framework that turns tourism

and host community into a tool to protect tangible and intangible cultural heritage

(UNWTO/UNESCO, 2018). This refers to systemic and innovative solutions that could

help to overpass existing challenges in different aspects of cultural tourism

development. For example, communities and cultural heritage in rural areas could

face different obstacles for their additional sustainable initiatives compering to those

in urban destinations, mostly due to contemporary global environment of competition

and changing consumer behavior on one side and specific ecosystem present in the

rural less developed area (Moric, 2013a). In addition, similar problems of community

non-engagement are unfortunately often present in transition economies and

appropriate stimulation by government bodies and other stakeholders (e.g., foreign

development agencies) is necessary, as well as new and encouraging legal and

strategic frameworks. This necessity is recognized in the Cultural tourism development

program of Montenegro (MSDT, 2019) where the support of the local community

together with experts in cultural and tourism sectors represent two key factors of

success. Anyway, Montenegrin experiences in the area of community-based tourism

are not examined.

Therefore, this paper aims to provide an insight from Montenegro that will contribute

to a better understanding of the necessity for more systemic frameworks and

mechanisms for sustainable development of cultural tourism, as well as provide a

platform for future research and strategic consideration. Existing literature on

community-based tourism is extensive, but the experiences are quite different due to

the great number of social, political, economic, and other factors that determine the

success of community engagement. Hence, this study enriches previous research in

two important ways. First, findings will include the influence of the transitional

character of the economy on community participation. Second, this paper

contributes to human resource education and training by examining the guidelines

for more professional and innovative development among the locals.

The remainder of this paper is organized as follows. Section 2 reviews the literature

related to the conceptual basics of tourism, culture, community, and their

interrelation. Section 3 presents the data and methodology. Results and discussion

167


are provided in section 4, while conclusions and suggestions for future development

are given in section 5.

Literature Review Community role There is a vast number of studies related to the conceptual basics of community role,

engagement, and importance for tourism development (Salazar, 2012; Goodwin and

Santilli, 2009; MacDonald and Jolliffe, 2003). Its complexity is widely recognized, but still

not entirely examined, mostly due to various groups of factors that could generate a

more dynamic and competitive ambient for locals and their behavior. Goodwin and

Santilli (2009) examine the success factors of tourism initiatives in the context of

community-based tourism and point out following ten key determinants: social capital

and empowerment, environment conservation, improved standard of living, local

economic development and commercial viability, education, sense of place, tourism

(e.g., new tourist experiences), collective benefits (e.g., infrastructure) and other (e.g.

funding). For example, it is interesting that social capital and empowerment (e.g.,

local community management or ownership) are ranked as primary, and collective

benefits (e.g., infrastructure, social services) are behind commercial viability and

environment conservation.

Despite the mentioned complexity of motives related to community involvement,

local-level participation is seen as the key factor of successful sustainable

development, but it is not always practical, desired, and possible (Salazar, 2012). For

example, total involvement and control of the local community over cultural

resources should not be always understood as a sustainable and responsible

development mechanism, mostly due to possibilities of misuse, degradation, and/or

over-tourism. On the other side, it is understandable that non-material cultural tourism

requires higher involvement of local populations to present local specifics, way of life,

activities, and traditions than more material or tangible forms (e.g., museums) of

experience (Briassoulis, 2002). Consequently, community engagement challenges

and opportunities are different in various segments of cultural tourism such as religious,

educational, festival, folklore, heritage, and non-material cultural tourism.

Also, theory tends to treat the local community as a homogenous social unit, but in

practice, it is rarely seen (Blackstock, 2005). In reality, there are several groups in the

community with often opposite approaches and ideas about heritage tourism

development, especially in sense of the following issues: modes of commercialization,

cultural integrity, and identity, sustainability, etc. For example, the process of

destination image formation may conflict with residents’ concepts of self-identity

(Hughes and Allen, 2005). Indeed, this is the case when certain heritage aspects are

more emphasized compared to others, to create an image that will have better

commercial effects and attract more tourists. Also, there has been expressed much

concern about locally distinctive products becoming indistinct due to the lack of

possibilities to protect them from invasive global systems of value (Richards and Wilson,

2006). In such circumstances, the local community tends to oppose and protect its

identity, even via provoking conflicts with tourists.

Local community active involvement

Despite the mentioned strategic and technical challenges, the local community has

to take (pro)active in the provision of cultural experiences rather than being just

passive elements in indistinctive products (Richards and Wilson, 2006). This opportunity

could be obtained by creating a flexible framework that will allow different

stakeholders to be cooperative and creative. In line with mentioned, the most efficient

168


mechanism for local community involvement lies in the hand of representatives of

local authorities (MSDT, 2019). Namely, their role is to technically and financially

support and stimulate local initiatives, so they are usually quite informed about the

needs and ideas of locals. On the other side, they could provide necessary support

from government bodies and foreign organizations (e.g., development agencies).

However, similar influence could have tourism and hotel industry (e.g., resorts, tour

operators) in destination, mostly due to its financial potentials and promotional effects

that could be induced on the wider market.

Local community and authenticity To understand the size and structure of this issue, especially negative and disturbing

consequences on tourist satisfaction, Cetin and Bilgihan (2016) point out the lack of

authenticity in cultural tourist destinations where there is a decreasing number of locals

living in the area, followed by expansion of hospitality and shopping facilities for tourists

(e.g., cafes, souvenir shops). In this way, cultural resources become front stages for

commercial purposes, while their authentic meaning is adapted to simple tourist

demand. On the other side, there are several immaterial resources such as norms,

habits, behavior, “smell-scape”, “soundscape” and the place’s genius loci or feeling

of the place, that could be consumed by tourists (Briassoulis, 2002). Most of these

elements are created and maintained by the local community. Consequently, they

could be modified or controlled only if the flexible and inclusive framework is obtained,

which naturally has to include locals. It offers new market possibilities for managers

and policymakers in the sense of branding, mostly because these intangible cultural

attributes are authentic, strategically quite effective for long-term differentiation, and

almost impossible to imitate by competition. Not surprisingly, this approach reduces

the costs of experienced production but requires vast investments in the social and

creative capital of local people (Richards and Wilson, 2006). Moreover, research

shows that kindness and politeness of local people, communication, politeness with

children, and other similar intangible elements, have a stronger impact on tourist

satisfaction than tangible ones (Perovic et al., 2018). Also, intangible components

generated mostly by the local community could be communicated via storytelling,

advice, and recommendations related to sightseeing, gratis services (e.g., transport,

animation, gastronomy). Due to this challenge, locals are important for the generation

of social interaction during the visit where intangible elements of this experience

include local hospitality, politeness, friendliness, etc. (Cetin and Bilgihan, 2016).

Such a strategic and innovative approach that includes locals and their

spontaneous and authentic positive attitude and behavior should be systemically

supported to provide a higher quality of tourist cultural experience. Although most of

these processes are out of direct managerial control, policymakers have to generate

a wider, integrative and destination-level approach. This opens two key issues. First,

this shows the importance of planning general creative human resource development

programs in a destination. Second, activities such as continuous awareness programs,

informative and interactive workshops, and educational campaigns represent the

strategic means for more effective and efficient local engagement into development

processes, as well as higher community satisfaction.

Local community and education Professional education and training are pointed out as important tools to improve

hospitality and general communication competencies (MSDT, 2019). Otherwise,

insufficient and/or wrong education and training could provoke barriers in the

provision of authentic cultural experience as well as social-cultural exchange. In this

way, successful local professionals in cultural tourism represent efficient intermediaries

169


among the rest of the community, especially in the context of cultural habits, respect,

and confidence expansion. For example, the role of local, professional, and well-

trained tour guides is widely recognized as important due to their communication

effect and ‘insider’ impact on tourists (Salazar, 2012). On the other side, these skilled

and educated individuals are an important link between the local community and

the cultural/tourism industry.

Local community and networking Clusters and networks are recognized as important tools for local community

involvement in the area of cultural tourism development because they could help

development, preservation, and promotion, especially in isolated areas where there

is a lack of other cooperative opportunities (MacDonald and Jolliffe, 2003). Also, to

promote the profitability of tourism among local rural community, Moric (2013b)

proposes clustering micro and small and medium enterprises (SMEs) that stimulates the

flow of financial resources within local economies and communities, preventing

economic leakage within the destination. In this way, local entrepreneurs could help

long-term economic sustainability on one side, and more diversified and developed

products and services on the other. Also, their ability to generate employment and

income has to be supported by continuous education opportunities, available

consulting services, and a flexible business ecosystem. It could be concluded that

profitability and training opportunities could retain local professionals and provide

possibilities for their further improvement and specialization. This opens opportunities

for the rest of the locals to participate and support local development.

Methodology Using the significant literature resources, existing secondary data from Montenegrin

government bodies, NGOs, and foreign development agencies (e.g., Ministry of

Culture, Ministry of Sustainable Development and Tourism), and relevant international

organizations (UNWTO, UNESCO), the research is focused on community engagement

in cultural tourism as an important and critical factor of success. Using these secondary

data, combined with data obtained by fieldwork conducted during 2019 at heritage

sites in Montenegro (e.g., Kotor, Cetinje, and Podgorica), the paper investigates the

key challenges (e.g., status quo, business opportunities, future perspectives, etc.)

associated with the role of locals in development processes. Based on qualitative

analysis, these elements are examined to suggest guidelines for future development.

Moreover, the paper explores new possibilities for diversifying tourism offer at heritage

sites in Montenegro, by engaging volunteers, enhancing understanding of the socio-

historical background, promoting the usage of digital tools, partnering with relevant

stakeholders, introducing innovative funding tools and schemes.

Results According to the Cultural Tourism Development Programme of Montenegro with

Action Plan 2019 – 2021, as a key national strategic document for cultural tourism

development, followed by other strategies and plans (e.g., Montenegro tourism

development strategy to 2020, National Strategy for Sustainable Development of

Montenegro by 2030), an insight of important issues related to local community

engagement is created and commented. Also, data from fieldwork at heritage sites

in Montenegro are added to gain more in-depth results. Such general results are

further carefully analyzed in the context of community involvement seen from

strategic as well as technical and operational levels. According to mentioned

secondary data sources and methodological approach, several key features could

be pointed out.

170


General tourism context Initial findings obtained during field research are linked to general conditions present

in Montenegrin typical transitional economy. Most of these issues are linked with

financial challenges and lack of funding when culture is in question. Apart from the

financial aspect, there is an evident challenge in the context of general tourism

sustainability. For example, as the most of tourist overnights in Montenegro are

generated in the south region (over 90%), more balanced regional development

should be considered. In addition, the focus should be shifted to the development of

alternative forms of tourism that could generate more visits in central and northern

parts of the country (Monstat, 2019; Moric, 2013b). In addition, there is a wide number

of other issues generated by mass tourism, which could create an obstacle for

alternative cultural tourism development (e.g., limited funding for cultural initiatives,

over-tourism, etc.).

Cultural tourism context Montenegro has a rich base of tangible and intangible cultural resources. But, a

significant percentage of locals are not fully familiar with existing cultural heritage,

primarily due to a lack of knowledge and awareness about its historical, artistic, socio-

cultural, and economic values. This could be identified as a significant obstacle for

more (pro)active local community involvement. Otherwise, further cultural

degradation is inevitable, especially in terms of the lack of opportunities for the next

generations in the area of sustainable cultural tourism development.

Apart from the conventional cultural heritage, abandoned industrial structures in

Montenegro represent valuable cultural tourism resources, especially in terms of their

spatial capacities (e.g., factories, military bases), where the most important idea is

MACCOC - Marina Abramović Community Centre Obod Cetinje (Cetinje

Municipality, 2019). New cultural ideas and products could be also linked with other

activities and heritage, especially in the area of agriculture (e.g., Katun), intangible

heritage (e.g., skills and knowledge), handcraft, etc.

Participation in international culture routes is recognized as an important potential

for international visibility, especially since Europe has 33 such cultural routes. However,

at the local level, there are no developed schemes or approaches to involve locals.

In addition, UNESCO included certain cultural resources (e.g., Kotor) on its list and in

that way raised international awareness on the cultural heritage of Montenegro, but

the general cultural image is weak in sense of global visibility, manifested as a low level

of knowledge amongst tourists about the cultural heritage of Montenegro. This finding

requires the need for implementation of a strategic approach in cultural tourism

development. In an operational sense, this means the involvement of all stakeholders,

especially tourism organizations, government and local authorities, media, etc.

In line with global aspects of development, new marketing opportunities are

identified in distant markets and consequently regional cooperation with benefits for

local communities in terms of cultural identity promotion, potential profitability, and

contact with very different cultures and visitors (e.g., China).

Local community context Locals could benefit from active involvement in the organization of festivals and other

cultural initiatives, mostly from training courses, workshops, preservation and usage of

traditional values, access to assets, local production of food and beverage, etc. These

business opportunities are partly recognized by locals but require a more systemic and

sustainable approach for higher benefits. In addition, cultural tourism in Montenegro

171


is seen as an important instrument that can contribute to image creation, tourist

season extension, cultural heritage protection, and foster sustainable development at

the macro/destination level. Unfortunately, these benefits are not clearly understood

at the micro and local levels.

According to the general tourism policy in Montenegro, the local community is

identified as an important factor and includes the following key stakeholders: non-

government organizations, owners of construction buildings, and lessees. In addition,

the local community, which is not involved in the process of planning the

development of cultural tourism, could create resistance towards different

stakeholders. There is a wide recognition that involvement of the local community has

to be stimulated via measures such as awareness stimulation on the importance of

tourism; entrepreneurship encouragement and enhancement of tourism links with

other sectors of the local economy. Unfortunately, the practical application of

mentioned statements and programs is not enough and requires further improvement.

Interesting weak sides related to local community capacities are: lack of knowledge

of foreign languages (e.g., German), insufficient number of cultural/tourism

professionals, locally produced food and beverage is insufficiently present in

hospitality offer, etc.

On the other side, in sense of (e.g., foreign) expert engagement, a key challenge

is the fact that they are not always familiar with local opinions, attitudes, and

specificities. In addition, the lack of their specific competencies could shift the

attention from crucial to marginal and unimportant issues. The lack of specialized local

experts is also clearly a noticeable challenge.

To keep this analysis inside the local community context, the following results are

most important for further analysis. First, cultural tourism is developing and still facing a

range of barriers, challenges, and issues, which are the consequence of the general

business environment in Montenegro (e.g., lack of funding for cultural ideas and

initiatives). Second, the dominant focus is on sea & sun tourism, followed by a lack of

initiatives in alternative tourism (e.g., cultural). On one side, most entrepreneurial

capacities are involved in mass tourism, whereas on the other side, there is an evident

lack of entrepreneurial abilities among cultural institutions and professionals. This

imbalance requires structural reform. Third, cultural heritage (e.g., museums) is still seen

mostly as a complementary product (e.g., excursion or a short visit for mass tourists). In

addition, cultural tourism products are old-fashioned often lacking the application of

modern digital technology. Moreover, heritage sites of secondary or tertiary

significance, due to limited potential to attract a critical number of visitors, are quite

often experiencing a certain form of devastation or irresponsible usage, while primary

attraction is sometimes “administratively overprotected” resulting in its isolation from

commercial purposes and consequent disappearance from tourism market. Finally,

there is an evident gap between community and heritage, especially those placed

in museums and galleries. This fact is further problematized by the lack of systemically

organized schemes or approaches for the local community to take initiatives

efficiently and effectively. This complex problem is followed by limited (e.g., public)

funding, lack of IT application, and occasionally unclear procedures when valorization

of cultural heritage is in question.

Discussion Related to previously mentioned key opportunities, barriers, challenges, and specifics

linked with the local community and cultural tourism, several management and

marketing issues have to be discussed (see Table 1).

172


Table 1

Key management and marketing issues related to cultural tourism development in

Montenegro

Management/marketing

issues

Strategic aspects Operational/technical aspects

New cultural tourism

products

Innovative and

internationally visible

products;

Cooperative approach;

IT opportunities.

“Live” cultural products;

Commercially important products;

Thematically conceptualized tours;

Combinations with health and

MICE offers;

Daily excursions;

Panoramic routes and viewpoints.

Cooperation and

networks

Diversification of

products.

Integration into international

networks and organizations;

Cultural routes.

Sustainable and

responsible

development

Cultural products

harmonized with the

economic,

environmental, and

social context.

Responsible commercialization;

Efficient preservation;

Opportunities for new ideas;

Socio-cultural importance and

identity.

Support for the local

community

Continual education and

training;

Raising awareness

programs.

Financial and non-financial support

(e.g., grants, workshops, training).

Source: Authors

New cultural tourism products Although Montenegro could provide a vast variety of cultural material and immaterial

heritage, there is an obvious lack of precisely and operationally defined and

internationally visible, independently organized, and commercially important cultural

products. Innovative and professional offerings are key to success. Great potential lays

in immaterial cultural heritage, which is not sufficiently valorized mostly due to lack of

knowledge, skill, and technological opportunities. Although, it probably represents the

most innovative segment of cultural tourism thanks to its ‘live’, rich, creative, and

dynamic character. Indeed, ‘live’ and real cultural products are quite necessary

because they are based on real-life and individuals as intermediates in the process of

cultural exchange and communication. In addition, there is a need for a more

cooperative approach between different stakeholders (e.g., private, public sector) to

activate primary, secondary and tertiary cultural attractions. In line with this, mixed

and combined programs are necessary due to relatively weaker possibilities of

Montenegrin heritage to attract enough visitors. These combinations could include

the following options: thematically conceptualized tours, combinations with health

and MICE, daily excursions, panoramic routes, etc.

It is well known in theory and practice that minor heritage resources could quite

often have a major role, especially in sense of cultural identity of local communities

(Cuccia and Rizzo, 2011). Hence, local entrepreneurs should be informed, educated,

and trained to select and develop those experiences that are more suitable for

tourists, but also local social context and tourism industry in Montenegro. This could

help to avoid leakage of scarce resources and not to lose focus on crucial and

strategic issues and resources.

This leads to another quite common challenge, such as the insufficient number of

professionals with necessary interdisciplinary and entrepreneurial skills. This is not the

173


only problem in Montenegro, but also often seen in other more tourism developed

destinations in Europe (Salazar, 2012). This could be overpassed with constant support

(e.g., financial, educational, and technical) to different groups of experts and locals

interested in the development of real, market-oriented, long-term sustainable tourism

businesses. Significant professionals are local tour guides that could have a better

effect on the quality of experience as long as their professional involvement is followed

by adequate awareness and understanding of the importance of the image-building

process. In addition, limited human and financial resources could be overpassed via

micro-clusters and networks in Montenegrin rural and cultural tourism. In this vein,

community involvement should be considered as a strategic tool in the sustainable

planning of cultural tourism development in Montenegro.

Cooperation and networks Cultural tourism, due to its complementary character, surely could enrich the integral

tourism product of Montenegro, creating strategic opportunities for mixing and

diversifying existing services as well as generation of new innovative products (e.g.,

culture and health, culture and MICE). On the other side, hotels need to develop more

intensive cooperation with locals, providing in that way their guests with the

opportunity to participate in local cultural tradition and life. Also, local food and drinks

have to find their steady place in menus of the local hospitality industry. Hence, it is of

strategic importance to develop closer relations between cultural and other more

popular and/or profitable forms of tourism, as often suggested in theory and practice

(Jovicic, 2016). This could provide diversified products and better opportunities for

more sustainable commercialization. As previously, mentioned, economic leakage is

seen as a quite disturbing concern in typical “3S” destinations and transition

economies. Namely, the “sun and sea” are still the most dominant products with a

significant share of over 90% in Montenegro. Provision of new services and products in

cultural aspects by the local community and economy could be suggested to reduce

the leakages. Interestingly, integration into international networks, organizations,

cultural routes, and other forms of cooperation could provide a great number of

benefits. There is a wide number of good experiences and scholars indicate positive

experiences of theme routes/trails as a tool that can further improved tourism growth

(e.g., Meyer-Czech, 2003; Soteriadeset al., 2009; Demonja and Ružić, 2012; Hall and

Mitchell, 2008; Bruwer, 2003; Telfer and Hashimoto, 2005).

Sustainable and responsible development Cultural products have to be harmonized with the economic, environmental, and

social context. However, such a general statement requires an effective local

implementation approach to be sustainable and more important accepted by locals.

This implies the necessity to invest in local innovative abilities as well as communicate

with them every aspect of the development process (e.g., image creation, new

product, pricing). There is a growing necessity for further investment and other actions

in the area of education and awareness about the benefits of cultural tourism,

especially in sense of responsible commercialization, efficient preservation,

opportunities for new ideas, socio-cultural importance, etc. For example, the

introduction of modern technological opportunities could attract locals, especially

talented and young entrepreneurs into the local cultural industry. The development

of cultural tourism has to have the support of the local population. To achieve this,

locals need to be involved in the process from the beginning, planning phase,

implementation, and finally control or monitoring. In line with this, the dynamic and

flexible framework should be developed and offered to locals to provide their

174


engagement. Unfortunately, with more static and fixed approaches is not possible to

obtain the necessary support. In other words, poor profitability could be neutralized

by abandoning the “ad-hoc” approach in planning cultural tourism.

Support for the local community Another important issue is the potential lack of funding that could provoke a slowdown

in the creation of complex and innovative products. In this situation, support has to be

obtained from other stakeholders such as the hotel industry, local tourism

organizations, local authorities, etc. Also, lack of knowledge of foreign languages

(e.g., German, French) creates an obstacle for development. Closely related to

foreign languages are the storytelling competencies that could help as a mean of

presentation of local immaterial cultural heritage (e.g., myths, legends, customs),

tangible components (e.g., food), the interpretative technique in museums, visitor

centers, farms, eco-villages and effective marketing tool to distribute memories of

tourists to other tourists, operators, and media. That is why constant training has to be

obtained for all age groups of potentially interested individuals (e.g., students,

volunteers, professionals, local authorities, rural community) for different skills and

competencies. Additionally, raising awareness of locals about preservation,

revitalization, valorization, and promotion of cultural heritage has to be the priority that

is constantly fostered via financial and non-financial measures and actions (e.g.,

workshops for school children, study trips for entrepreneurs, educative events).

Conclusion Summary of the research This paper discusses the most relevant issues related to cultural tourism and community

engagement in Montenegro. Key findings indicate the need for a systemic, dynamic,

and innovative framework for sustainable and highly impactful heritage tourism in

Montenegro, which policymakers, heritage ventures, and other stakeholders might

use to strengthen community engagement and development at the heritage sites.

The local community has to be encouraged as the driver of tourism entrepreneurship

while key elements of destination competitiveness could be found in process of

diversification of the offer with a focus on typical local distinctive cultural styles and

resources that are worth visiting year-round. But this theoretical concept needs

effective technical operationalization where tourism revenue has to be channeled

into conservation, promotion, education, and development of local community and

heritage. Followed by other non-financial support (e.g., training), this systemic

framework can maximize benefits for culture, tourism, and the local community.

Practical implications As cultural tourism in Montenegro continues to grow, destination marketers and

managers will face the strategic challenge of developing a more systemic and

professional, but flexible at the same time, approach towards the host community.

Based on this principle, several practical implications should be pointed out.

First, a further increase in demand for authentic cultural experiences will require

more professionals to develop and manage such products. Consequently, more

training opportunities need to be provided in the future. Locals have to be supported

and educated on how to commercialized resources as final independent products or

as a segment of wider “visitors’ experience” (e.g., excursion). In addition, training

could help the host community to select those experiences that are more suitable for

tourists. Priority areas of training should be a specialist in gastronomy, rural tourism

experts, event managers, storytelling specialists, tour/route guides, etc.

175


Second, the public sector has to encourage the host population to develop its

forms of tourism products. Local themes and local cultural heritage (e.g., intangible

heritage) are an important source of authenticity and competitiveness in tourism.

Development of such local and special products is a more competitive approach

than the imitation of already existing “mass” cultural events or programs with

questionable value. Based on previous research, valorization and commercialization

have to be undertaken by the private sector, with appropriate support from the public

sector. In line with this, special programs and plans have to be designed to protect

authentic attributes, especially those (e.g., handicrafts, agricultural practices) facing

disappearance, distortion, or trivialization.

Third, competencies related to narrative and interpretation technics are necessary

to commercialize existing tangible and intangible heritage. Since the local community

has to play the role of originator and interpreter of cultural narratives/stories, it is crucial

to invest in and support human resource training together with marketing, innovation,

and technological support.

Fourth, the management mechanism in destination has to be improved to maintain

lasting growth in this sector. For example, to obtain continuous communication with

the host community as an important stakeholder, there is a need to identify a leader

that is willing to promote the idea of participation and cooperation, or can organize

the community and help to reach a consensus within the locals about the

development goals. Besides, an efficient mechanism for community participation is

seen in a cluster approach that could benefit from the fragmented tourism and

cultural sector. The product they produce is the main cohesive factor. Accordingly,

networks and clusters are seen as an efficient tools for community participation, both

in urban and in less developed, rural areas.

Limitations and further research This paper offers insight from Montenegro about the role of the host community in

cultural tourism development. We hope that this insight could help basic

understanding in this area and could be useful for the design of public-support policies

in this sector. The key limitation is seen in the fact that this paper is partially based on

the results obtained through the HERTOUR project “Strengthening heritage tourism and

community development in Austria and Montenegro”. As such, this research

represents just a pilot study into the vast field of cultural tourism. The general insight

given in this paper requires further research and data related to complex issues of

local community engagement. In line with this, further research of this topic is

recommended, especially in the area of technology implementation and innovation

generation related to more effective community engagement and heritage

safeguarding. In addition, research should empirically analyze the potentials of locals

in sense of creating new value and the impact of different measures on this strategic

goal. Key limitations are related to more in-depth analysis that is necessary for a

deeper understanding of this complex issue.

References 1. Blackstock, K. (2005), “A Critical Look at Community Based Tourism”, Community

Development Journal, Vol. 40 No. 1, pp. 39–49.

2. Briassoulis, H. (2002), “Sustainable Tourism and the Question of the Commons”, Annals

of Tourism Research, Vol. 29 No. 4, pp. 1065-1085.

3. Bruwer, J. (2003), „South African Wine Routes: Some Perspectives on the Wine Tourism

Industry’s structural Dimensions and Wine Tourism Product“, Tourism Management, Vol.

24 No. 4, pp. 423–435.

176


4. Cetin, G. and Bilgihan, A. (2016), “Components of Cultural Tourists’ Experiences in

Destinations”, Current Issues in Tourism, Vol. 19 No. 2, pp. 137-154.

5. Cetinje Municipality (2019), “Project: Marina Abramović Community Center Obod

Cetinje“, available at:

http://www.cetinje.me/cetinje/site_mne/public/index.php/index/artikli?id=189#, (25

December 2019).

6. Chirikure, S., Manyanga, M., Ndoro, W. and Pwiti, G. (2010), “Unfulfilled Promises?

Heritage Management and Community Participation at some of Africa's Cultural

Heritage Sites”, International Journal of Heritage Studies, Vol. 16 No. 1-2, pp. 30-44.

7. Cooper, C., Fletcher, J., Fyall, A., Gilbert, D., Wanhill, S. (2005), Tourism: Principles and

Practice, Pearson Education, Essex, UK.

8. Crooke, E. (2010), “The Politics of Community Heritage: Motivations, Authority and

Control”, International Journal of Heritage Studies, Vol. 16 No. 1-2, pp. 16-29.

9. Cuccia, T. and Rizzo, I. (2011), “Tourism Seasonality in Cultural Destinations: Empirical

Evidence from Sicily”, Tourism Management, Vol. 32 No. 3, pp. 589-595.

10. Demonja, D., Ružić, P. (2011), “Rural Tourism in Croatia with Croatian Case Studies of

Good Practice and European Experiences”, Meridians.

11. Gmür, B., Bartelt, A., and Kissling, R. (2010), “Organization from a Systemic Perspective:

Application of the Viable System Model to the Swiss Youth Hostel Association,

Kybernetes, Vol. 39 No. 9/10, pp. 1627-1644.

12. Goodwin, H. and Santilli, R. (2009), “Community-Based Tourism: A success?”, ICRT

Occasional Paper 11, GTZ.

13. Hall, C. M. and Mitchell, R. (2008), “Wine Marketing: A Practical Guide”, Butterworth-

Heinemann.

14. Hughes, H. and Allen, D. (2005), “Cultural Tourism in Central and Eastern Europe: The

Views of ‘Induced Image Formation Agents”, Tourism Management, Vol. 26 No. 2, pp.

173–183.

15. Jovicic, D. (2016), Cultural Tourism in the Context of Relations between Mass and

Alternative Tourism, Current Issues in Tourism, Vol. 19 No. 6, pp. 605-612.

16. Lazanski, T. and Kljajić, M. (2006), “Systems approach to complex systems modelling

with special regards to tourism”, Kybernetes, Vol. 35 No. 7/8, pp. 1048-1058.

17. MacDonald, R. and Jolliffe, L. (2003), “Cultural Rural Tourism: Evidence from Canada”,

Annals of Tourism Research, Vol. 30 No. 2, pp. 307–322.

18. Meyer-Cech, K. (2003), “Food Trails in Austria”, in Hall, C.M, Sharples, L., Mitchell, R.,

Macionis, N. and Cambourne, B. (Ed.) Food Tourism Around the World: Development,

management and Markets, Oxford, Butterworth-Heinemann, pp. 149-157.

19. Monstat (2019), Tourist Arrivals and Overnights in Montenegro in 2018, Montenegro

Statistical Office – Monstat, Podgorica.

20. Moric, I. (2013a), “The Role and Challenges of Rural Tourism Development in Transition

Countries: Montenegro Experiences”, Turizam, Vol. 17, No. 2, pp. 84-95.

21. Moric, I. (2013b). “Clusters as a Factor of Rural Tourism Competitiveness: Montenegro

Experiences”, Business Systems Research, Vol. 4 No. 2, pp. 94-107.

22. MSDT (2019), Cultural Tourism Development Programme of Montenegro with Action

Plan 2019 – 2021, Ministry of Sustainable Development and Tourism, CAU-Centre for

Architecture and Urbanism, Podgorica.

23. Okazaki, E. (2008), “A Community-Based Tourism Model: Its Conception and Use”,

Journal of Sustainable Tourism, Vol. 16 No. 5, pp. 511-529.

24. Pejic Bach, M., Zoroja, J., and Merkac-Skok, M. (2014), “Social Responsibility in Tourism:

System Archetypes Approach”, Kybernetes, Vol. 43 No. 3/4, pp. 587-600.

25. Perovic, Đ., Moric, I., Pekovic, S., Stanovcic, T., Roblek, V. and Pejic Bach, M. (2018), "The

Antecedents of Tourist Repeat Visit Intention: Systemic approach", Kybernetes, Vol. 47

No. 9, pp. 1857-1871.

26. Richards, G. and Wilson, J. (2006), “Developing Creativity in Tourist Experiences: A

Solution to the Serial Reproduction of Culture?”, Tourism Management, Vol. 27 No. 6,

pp. 1209–1223.

177


27. Salazar, N. B. (2012), “Community-based Cultural Tourism: Issues, Threats and

Opportunities”, Journal of Sustainable Tourism, Vol. 20 No. 1, pp. 9–22.

28. Soteriades, M.D., Tyrogala, E.D. and Varvaressos, S.I. (2009), “Contribution of Networking

and Clustering in Rural Tourism Business”, Tourismos: An International Multidisciplinary

Journal of Tourism, Vol. 4 No. 4, pp. 35-56.

29. Telfer, J. D., Hashimoto, A. (2003), “Food Tourism in the Niagara Region: The

Development of a Nouvelle Cuisine”, in Hall, C.M, Sharples, L., Mitchell, R., Macionis, N.

and Cambourne, B. (Ed.) Food Tourism Around the World: Development, management

and Markets, Oxford, Butterworth-Heinemann, pp. 158-177.

30. UNDP, (2004), Strategic Framework for Development and Sustainable Tourism in

Northern and Central Montenegro, Podgorica, UNDP.

31. UNESCO, (2003) Convention for the Safeguarding of the Intangible Cultural Heritage,

available at: https://ich.unesco.org/en/convention, (27 December 2019).

32. UNWTO (2018), International Seminar on Harnessing Cultural Tourism through Innovation

and Technology, 12-14 November 2018, Hamedan, Islamic Republic of Iran, available

at: https://www.unwto.org/global/event/40th-unwto-affiliate-members-plenary-

session-international-seminar-harnessing-cultural-tourism, (27 December 2019).

33. UNWTO (2019), “Tourism and Culture”, available at: https://www.unwto.org/tourism-

and-culture, (27 December 2019).

34. UNWTO/UNESCO (2018), “Third conference: Cultural tourism sustains communities and

living heritage”, 3-5 December 2016, Istanbul, Turkey, available at:

http://tourismandculture.cvent.com/events/third-unwto-unesco-world-conference-

on-tourism-and-culture/event-summary-c85f6e964b5747b99ee96fe666d40912.aspx,

(27 December 2019).

178


About the authors Ilija Moric holds a Ph.D. in Economics from the Faculty of Economics, University of

Montenegro, obtained in 2015. He is Assistant Professor at the Faculty of Tourism and

Hotel Management in Kotor. His main research interests are within the field of

marketing, marketing communications, understanding consumer behavior in tourism,

rural development, and rural tourism. The author can be contacted at

[email protected].

Sanja Pekovic holds a Ph.D. in Economics from the University Paris-EST. She is the

Assistant Professor at the Faculty of Tourism and Hotel Management. Since October

2017, she was Director of the Centre for Studies and Quality Assurance of UoM.

Currently, she is Vice-rector for internationalisation. Between 2006 and 2011, she was

the Researcher at the Center for Labor Studies (Centre d’Eudes de l’Emploi) and

Lecturer at the University Paris-EST. Her research interests are within the field of quality

and environmental economics, the economics of innovation, applied econometrics,

and on this topic, she has presented studies at national and international scientific

congresses, which have been published in international journals. Ph.D. Pekovic was

visiting scholar at the INRASupArgo (Montpellier), at the University of Montenegro

(Podgorica), at the Laboratoire CNRS UMI 2615 Franco-Russe PONCELET (Moscow)

and the Institute of Environment, UCLA (Los Angeles), etc. The author can be


Djurdjica Perovic, Ph.D., is Associate Professor at the Faculty of Tourism and Hotel

Management of the University of Montenegro. She defended her thesis „The state and

directions of tourism development of the Montenegrin coast in the function of a

successful market appearance” at the Faculty of Science and Mathematics of the

University of Novi Sad. She was the Vice-Rector at the University of Montenegro till

December 2020. Currently, she is the Dean of the Faculty of Tourism and Hospitality.

Her research fields are Tourism, Tourist Regions, Cultural Tourism, and Selective Forms

of Tourism. She has published more than 30 papers and has participated in more than

20 science conferences. She is the author and co-author of two monographs. She has

experience in Heric, IPA, Erasmus+, and bilateral projects. She was a visiting professor

at “MESI” University and Russian New University from Moscow. The author can be


Jovana Janinovic is the research and teaching assistant at the Faculty of Tourism and

Hotel Management, University of Montenegro. Previously, she was engaged as a

researcher at Leipzig Graduate School Global and area studies, the start-up scholar

at Bielefeld University, and ZEIT Stiftung's pre-doctoral fellow. She obtained her Erasmus

Mundus Master Degree from the Charles University of Prague and EHESS Paris in the

field of European studies. She also holds BA in Economics from the University of

Montenegro and an M.A. in Management from the University of Nice. She presented

at academic conferences in Berlin, Dublin, Paris, Florence, Warsaw, and participated

in several European training and programs. The author can be contacted at

[email protected].

Michaela Griesbeck holds a Ph.D. in communication science and semiotics and works

currently as a senior researcher at the Franz Vranitzky Chair for European Studies,

University of Vienna, and since 2002 as a lecturer at the Department of

Communication at the University of Vienna. Her fields of research are intercultural

communication, social semiotics, and the so-called Generation In-between, the

Children of the Balkan Wars, which led her lately to research trips throughout South-

Eastern Europe. The author can be contacted at [email protected].

179


Position and Role of Social Supermarkets in

Food Supply Chains

Blaženka Knežević, Petra Škrobot, Berislav Žmuk

University of Zagreb, Faculty of Economics and Business, Zagreb, Croatia

Abstract

Background: Social supermarkets were developed in Europe after the economic crisis

2008-2014. Their purpose is to decrease food waste that occurs in traditional food

supply chains and to ensure access to food to socially endangered citizens.

Objectives: This paper analyses the general perception of consumers regarding the

mission and purpose of social supermarkets in four Central Eastern European (CEE)

countries: Croatia, Poland, Lithuania, and Serbia. Methods/Approach: The paper

brings the results of the survey research conducted in the observed CEE countries

measuring attitudes towards the relevance and the role of social supermarkets.

Results: There is a positive attitude regarding the existence of social supermarkets in all

the analysed CEE countries. Less than 10% of respondents claim that there is no need

for such organizations. In Croatia, Lithuania, and Poland examinees claim that

reduction of food waste rather than reduction of poverty should be emphasized as a

mission of social supermarkets. Conclusions: Social supermarkets require improvement

of a legal framework, welfare system integration, and implementation of state

monitoring. Moreover, larger involvement of religious communities, national and local

governments, as supporting institutions is observed as a necessity in all the countries.

Keywords: supply chain; social supermarkets; food distribution; CEE

JEL classification: M3, I3


Received: 12 Oct 2020

Accepted: 06 Apr 2021

Citation: Knežević, B., Škrobot, P., Žmuk, B. (2021), “Position and Role of Social

Supermarkets in Food Supply Chains”, Business Systems Research, Vol. 12, No. 1, pp.

179-196.


Acknowledgements: The paper is a result of the project funded by the Croatian

Science Foundation “Potentials and Obstacles of Social Supermarkets Development

in Central and Eastern Europe” (UIP-2014-09-4057).

180


Introduction As a new type of organization, social supermarkets emerged as the answer to the

recent economic crisis across Europe (2008-2014) when the number of people living

at risk of poverty or social exclusion increased rapidly (EU, 2014a; EU, 2014b; EU, 2014c).

They are focused on those groups of customers who have low income or who are in

severe material deprivation. There are numerous examples of social supermarkets

across Europe, but their level of development and type of operational activity varies

from country to country. As they are an emerging type of organization, there is no

common definition of social supermarkets because it should be broad enough to

integrate all the variations, which are developed and existing in different markets.

Moreover, social supermarkets are not sufficiently analyzed in the literature nor

explored in primary research, but we can find a lot of different definitions and

determinations of the term social supermarkets.

A social supermarket is defined as “a small, non-profit oriented retailing operation

offering a limited assortment of products at symbolic prices primary in a self-service

manner. Authorized for shopping are needy people only. The products are donated

by food production and retail companies free of charge, as they are edible but not

marketable due to small blemishes. Achieved profit is reinvested into social projects”

(Leinbacher et al, 2011). Holweg and Lienbacher (2011) define social supermarkets as

food-oriented retailers selling food to a restricted group of people living in or at risk of

poverty. By definition given in Holweg and Lienbacher (2011), social supermarkets are

nonprofit organizations that base their activity on volunteerism and charity and if they

generate any profits they use them for charitable activities. According to Maric and

Knezevic (2014), a social supermarket is a new retail format that fosters positive social

change by fulfilling the material needs of the socially disadvantaged groups and

allowing them to preserve their dignity in an environment where they can choose

various kinds of goods at extremely low prices. Some social supermarkets are offering

goods free of charge as explained by Knezevic and Skrobot (2018).

In addition, (Schneider et al., 2015) emphasize three types of benefits of social

supermarkets: (1) social benefits, (2) environmental benefits, and (3) economic

benefits. Social benefits are observed through: reduction of food insecurity and life

quality improvement of socially endangered citizens, improvement of their social

inclusion, growth of self-confidence in communication with others, and fostering a

feeling of belonging to a certain community by treating their users as clients rather

than charity users, what strengthens their sense of dignity. On the other hand,

environmental benefits are related to food waste reduction throughout the

distribution of food surplus from companies and individuals to final users. Finally,

economic benefits are concerned with better reallocation of scarce household

budget because users can purchase products at lower prices in social supermarkets,

while companies that donate surpluses improve their cost efficiency by decreasing

handling and warehousing costs for goods with low stock turnover ratios.

Maric et al. (2015) state that “social supermarkets represent a specific form of social

entrepreneurship because they are a voluntary non-profit organization and a special

form of retail which supply socially vulnerable individuals with necessities” and they

claim that social supermarkets should be observed as a specific form of social

innovation. Because they promote the strengthening of social capital, social cohesion

and develop social responsibility among all stakeholders involved in the distribution of

food to socially endangered citizens. Moreover, building on the definition of social

enterprise given in Dees (1998) and Dees and Anderson (2003), Maric and Knezevic

(2014) argue that social supermarkets are the subset of social enterprises because

181


they strive to make positive social changes and create social value throughout social

innovation.

Klindzic et al (2016) analyze and define the role of social stakeholders, which enable

daily function and support the development of social supermarkets. They isolate the

following types of stakeholders: (1) individuals (social entrepreneurs, volunteers, users,

and donors), (2) organizations (in non-profit and in profit sector), (3) society

(government and local community). For each group of stakeholders, they explain the

position, role, and responsibilities regarding social supermarkets’ operation.

Besides, when defining social supermarkets there is a discussion on their role and

characteristics as a new retail format (see Lienbacher, 2012; Bogetic et al, 2018). This

discussion is taking into account elements of the retail mix such as assortment, prices,

location, service, and promotion, and establishing distinction towards other retail

formats, especially towards convenience stores, hard discounters, and traditional

supermarkets.

However, all definitions of social supermarkets emphasized selling or distributing

goods to people in severe material deprivation. Moreover, according to analyzed

definitions, social supermarkets can be viewed as a new type of intermediaries within

the food distribution chain because they have been developed to transfer surpluses

of food or products to people in need. Therefore, we can conclude that the purpose

of social supermarkets is twofold: The purpose of a social supermarket is twofold: (1)

the poverty reduction through the distribution of food to people in need and (2)

reduction of inefficiency in traditional (dominantly food) supply chains trough removal

of surpluses of produced goods.

Therefore, the paper is structured as follows. Firstly, social supermarkets are

explained as a new intermediator in food supply chains. Secondly, the research

sample and methodology are explained. The primary research took place during April

and May 2018 in four Central and Eastern European (CEE) countries: Croatia, Serbia,

Poland, and Lithuania. Thirdly, the research results on attitudes of consumers towards

social supermarkets as a new type of organization within the food supply chains are

discussed and elaborated. Finally, conclusions, implications, and limitations of the

research are given. Due to its geographical scope of interest, the paper represents a

valuable contribution to understanding this new phenomenon from the perspectives

of the under-researched European region.

Social supermarkets as new intermediates in food supply

chains

Christopher and Ryals (1999) gave one of the commonly cited definitions of the supply

chain. They define the supply chain as „the network of organizations that are involved,

through upstream and downstream linkages, in the different processes and activities

that produce value in the form of products and services in the hands of the ultimate

consumer". Schroeder and Meyer Goldstein (2018, p. 5) define supply chain as a

“network of manufacturing and service operations (often multiple organizations) that

supply one another from raw materials through production to the ultimate consumer.”

As a managerial discipline, supply chain management aims to improve the

coordination of goods, information, and financial flows within individual companies

(internally) and between companies that are participants in a certain supply chain

(externally) (see Lysons and Gillingham, 2003; Emmet and Crocker, 2009; Van Wheele,

2015). Similarly, Monczka et al. (2015) emphasize that supply chains are composed of

interrelated activities that are internal and external to a company. Booth (2014) also

distinguishes internal and external perspectives of supply chain claiming that supply

182


chain is a series of activities that deliver an outcome to internal (a colleague) or

external recipient (a customer).

According to Bailey et al. (2008) and Hughes et al. (1999), successful companies

seek to establish an integrated supply chain by applying, the so-called, helicopter

perspective to their supply chain (see Bailey et al. 2008). Thus, by applying the concept

of integrated supply chain management, business strategy is developed upon the

complete picture of related suppliers and customers to reduce costs and increase

value to the consumer at the level of the entire chain rather than at the level of

individual companies. Moreover, Monczka et al. (2015) claim that integrated supply

chain management developed in the twenty-first century and relies on a cooperative

approach in supplier relationship, on strategic purchasing orientation, and on

intensive usage of information technology (integrated Internet linkages, shared

databases, enterprise-wide systems, cloud computing, intensive use of mobile

devices, etc.).

Pullman and Zhaoui, (2012) argue that the food supply chain is formed of

interconnected companies on the way of food from farm to the table of the

consumer. The structure of the food supply chain is specific concerning the supply

chains of other types of products (eg. cars, shoes, clothes, or electronic products). In

such a supply chain participants are agro-producers: farmers, gardeners, herders,

fishermen, growers of various fruits and vegetables. They can sell their products (see

Pullman and Zhaoui, 2012): (1) directly to consumers, (2) an intermediary organization

(wholesale, retail, HORECA i.e. hotel, restaurant, catering, etc.) or (3) as a raw material

to the manufacturing industry that will turn it into finished food products and distribute

to the market. In addition, the manufacturing industry can sell its products to

consumers directly or can use one or more intermediaries to reach the final consumer.

In Figure 1, see arrows depicting the flow of goods in traditional supply chains. Besides,

Zeljko and Prester (2012) emphasize that supply chains should include other

organizations that are, either directly or indirectly, related in receiving and fulfilling

requests of consumers and/or facilitation of goods, money, or information flows.

Examples of those organizations are transporters, warehouses, banks, IT companies.

Similarly, in sequential approach, Lipinski et al. (2013) explain the 5 basic processes

in the food supply chain: (1) agricultural cultures are sown then, animal husbandry or

harvesting is done, in advance, (2) the produced food is stored and distributed to the

market, or goes further to processing. Then (3) in the processing phase, raw materials

are transformed into the finished food products that are packaged and stored and.

Finally, (4) through a market distribution system are delivered to the final consumer

who is going to (5) consume them.

Food loss and food waste can occur at any company involved in food supply and

at any stage of the food supply chain. Lipinski et al. (2013) claim that food loss occurs

in the stages of production, storage, processing, and physical distribution as an

unintended consequence of business processes or technical limitations in storage,

transport infrastructure, packaging, or marketing activities. While, usually, food waste

occurs in retail or at the stage of consumption (at the point of the final consumer),

and it is the result of negligence or a conscious decision to throw food away (Lipinski

et al., 2013).

Additionally, Lipinski et al. (2013) and FAO (2011) elaborate that there are significant

differences between developed and developing countries at stages of the supply

chain in which food losses and food waste occur In Europe, more than half of the food

is wasted at the stage of consumption (52%) and about a quarter in the production

stage (23%). Moreover, Principato et al. (2015) are adding that food is wasted in the

early stages of the food supply chain due to limits in technical, financial, and

183


managerial resources. On the other hand, in the final stages, food waste appears due

to adverse storage methods, poor planning before buying, impulsive purchase of

large quantities of food, food spoilage, inadequate quantity of preparation, etc. More

data on food waste occurrence can be seen in several studies (WRAP, 2007; FAO,

2011; Stefan et al., 2013; Principato et al., 2015; Parfitt et al., 2010, Koivupuro et al.,

2011.

Figure 1

The flow of goods in the food supply chain


The social and economic consequences of food waste are reflected in the uneven

distribution of food between the developed and developing parts of the world, but

also in the uneven distribution of food between members of society within a certain

country. Therefore, new organizations are emerging to deal with this problem of the

modern economy and society.

One type of such organization is social supermarkets. Within the context of the food

supply chain, they position themselves as an intermediator between traditional

members of food supply chains and consumers who are in material deprivation (see

Figure 1 – the flow of donated goods). Therefore, we can say that social supermarket

serves as leverage trying to establish an equilibrium between the appearance of food

surpluses and food waste in the traditional supply chains and the appearance of food

poverty among the population in a given area on the other side (Knezevic et al, 2017).

Data and methodology Research instrument and data

For the social supermarkets' study, a web survey was conducted. In primary research

there are several questions to be answered when approaching social supermarkets

as a new type of organization within food supply chains in CEE countries:

• (RQ1) Is there a need for such type of organization in the CEE region?

• (RQ2) What should be emphasized as a mission of such an organization?

184


• (RQ3) Is the attitude towards the concept positive or negative as we are

dealing with the concept implemented in post-communist countries?

• (RQ4) What is the perception towards existing legal frameworks and state

support regarding this type of organization?

• (RQ5) Is there a difference in perceptions regarding the country where the

survey was taken?

Due to a specific and sensitive topic, it has been decided to apply a non-

probabilistic survey approach to selecting respondents. In that way here snowball

sampling was used (Kish, 1995). In the first phase, the hyperlink to the web

questionnaire was sent to overall 20 scientists who work at universities in Zagreb, Split,

Belgrade, Niš, Cracow, Katowice, Poznan, Gdansk, and Vilnius. So, respondents from

four different European countries were observed: Croatia, Lithuania, Poland, and

Serbia. In the following step, those scientists have shared the web questionnaire

hyperlink to their colleagues and/or students, which have shared the hyperlink further

and so on. The survey answers collection period was from April to May 2018. At the

end of the survey overall 419 completed questionnaires were collected.

The web questionnaire consists of a brief description of social supermarkets as a

new form of organization within the food supply chain: “Social Supermarkets are non-

profit organizations that aim to distribute surpluses of produced food to people who

are in material deprivation. Primarily, social supermarkets raise donations in groceries

and organize their distribution to poor citizens (people in need) offering them the

possibility to choose needed stuff from a social supermarket's assortment. They

distribute groceries at extremely low prices or free of charge. The collected money is

further used to finance the everyday operation of social supermarkets (e.g. to pay

rent for used space or to buy necessary equipment) or to replenish assortment by

buying the new food at low prices from suppliers. Now, when we briefly inform you

about the scope of activities of social supermarkets, please answer a few questions

regarding this form of organizations.” (Research questionnaire, 2018). Then followed

questions were divided into several groups. The first question group consisted of some

demographic questions like gender; age and working status of respondents (see Table

3). After that are followed general questions about the social supermarkets (see Table

5 and Table 6). In that group, it can be found 11 questions from which 10 questions are

given on Likert scale from Joshi et al. (2015). In the following group of questions, it can

be found 5 questions, all given in Likert scale form, related to social supermarket

managers, but those questions are not in the focus of this paper. The final group of

questions emphasizes the role of frameworks and institutions in the social supermarket

area and it consists of eight questions all given in Likert scale form (Table 8).

Statistical methods

Due to using a non-probabilistic selection method and relatively small sample size, all

conclusions are limited to the observed respondents only. In addition, the design of

questionnaire questions vastly limited the possibility of using different statistical

approaches to the analysis of collected data. Therefore, to inspect differences

between the respondents in the four observed countries, the main emphasis in the

analysis will be given to descriptive statistics methods. The vast majority of questions

are given in Likert scale form. Because of that non-parametric chi-square tests for

equality of three or more proportions will be applied as well (Bolboacă et al., 2011).

185


Validity

Before the differences between the countries will be examined, the internal

consistency of the 10 general social supermarket variables is observed. The internal

consistency is inspected by using Cronbach's alpha and by observing respondents on

a country level and overall.

Table 1

Reliability analysis of generally on social supermarkets variables

Country No of

responses

Cronbach's

alpha

Standardized

alpha

Average inter-item

correlation

Croatia 117 0.5729 0.5768 0.1246

Lithuania 71 0.6181 0.6216 0.1460

Poland 123 0.6194 0.6217 0.1454

Serbia 108 0.5057 0.4934 0.0901

Overall 419 0.5869 0.5875 0.1269

Note: number of variables = 10


The results from Table 1 have shown that the internal consistency here is poor to

questionable (George and Mallery, 2003). For example, in the Croatia case, about

57% of the variability in the sum score is true score variability between respondents

concerning the concept common in all items. In other words, the used variables

turned out not to be so good and consistent in measuring the concept of social

markets in general. These results are speaking in favor of observing and analyzing

each variable separately rather than all together.

Table 2

Reliability analysis of framework and institutions variables

Country No of

responses

Cronbach's

alpha

Standardized

alpha

Average inter-item

correlation

Croatia 117 0.6266 0.6451 0.1977

Lithuania 71 0.5717 0.5578 0.1415

Poland 123 0.5733 0.5577 0.1431

Serbia 108 0.6586 0.6560 0.2032

Overall 419 0.6345 0.6286 0.1842

Note: number of variables = 8


The main results of conducted reliability analysis, where eight variables related to

framework and institutions were included, are given in Table 2. The resulting

Cronbach's alpha is ranging from 0.5717 to 0.6560. In this case, the conclusion about

poor to questionable internal consistency can be made (George and Mallery, 2003).

Analysis and discussion Analysis of respondents’ main characteristics

In the conducted web survey participated overall 419 respondents from which 117

being from Croatia, 71 from Lithuania, 123 from Poland, and 108 from Serbia. Table 3

shows distributions of respondents by country of their origin and according to their

main demographic characteristics.

According to Table 3 majority of respondents in all four observed countries were

females. The highest share of females in the total number of respondents in a country

186


was achieved in Croatia (74%) whereas the lowest share of females was registered in

Poland (58%). When the age structure of respondents is observed it can be concluded

that in Serbia (81%), Poland (66%), and Croatia (54%) the majority of respondents were

younger than 25 years. On the other hand, the majority of respondents in Lithuania

(52%) were aged from 25 to 40 years. The distribution of respondents’ age is explained

if respondents’ working status is closely observed. Namely, the majority of respondents

in Serbia (72%), Poland (59%), and Croatia (56%) are students who are just studying or

who occasionally work. The vast majority of respondents in Lithuania are not students

but respondents who are employed (70%).

Table 3

Main demographic characteristics of respondents

Country Variable Characteristics No of

respondents

% of

respondents

Croatia Gender Female 87 74

Male 30 26

Age Less than 25 63 54

25-40 43 37

More than 40 11 9

Working

status

A student who is just studying 21 18

A student who occasionally works 45 38

Employed 47 40

Unemployed and retired 4 3

Lithuania Gender Female 51 72

Male 20 28


25-40 37 52

More than 40 16 23

Working

status



Employed 50 70


Poland Gender Female 71 58

Male 52 42


25-40 26 21

More than 40 16 13

Working

status



Employed 50 41


Serbia Gender Female 74 69

Male 34 31


25-40 18 17

More than 40 2 2

Working

status



Employed 20 19



187


Respondents’ rating of their economic situation in relation to the average of their

country is given in Table 4. In all four countries, most respondents rated their economic

situation as an average one in comparison to the average of their country. However,

whereas in Lithuania (56%), Poland (54%), and Serbia (56%) the majority of respondents

have selected the “average” option. In Croatia, 47% of respondents stated that their

economic situation is average concerning the average of their country. If the

respondent’s distributions about a perceived economic situation are observed, it can

be concluded that the distribution tends to be negatively skewed because more

respondents have chosen the answers from the right side of the scale (“above

average” and “significantly above the average”) than those from the left side

(“significantly below the average” and “below the average”).

Table 4

Respondents’ rating of their economic situation concerning the average of their

country Perceived economic situation Country

Croatia Lithuania Poland Serbia

Significantly below the average 2% 3% 1% 3%

Below the average 14% 15% 13% 16%

Average 47% 56% 54% 56%

Above average 32% 24% 27% 25%

Significantly above the average 5% 1% 6% 1%


Figure 2

Social supermarkets presence


When it comes to the question about social supermarkets presence in their city,

most respondents from Lithuania (59%) confirmed that there is a social supermarket in

their city. On the other hand, only 8% of respondents from Serbia know that there is a

social supermarket in their city. However, 86% of respondents from Serbia have

emphasized that there is a need for a social supermarket. Figure 2 are shown the

distributions of answers in more detail.

188


Analysis of social supermarkets in general After the basic demographic questions, the respondents were asked general

questions about social supermarkets. In that way, they were asked about the top

priority of the social supermarket mission. The distribution of answers is shown in Figure

3.

According to Figure 3, the highest share of respondents who think that reduction of

food waste is the top priority of social supermarket mission can be found in Poland

(65%). On the other hand, the highest share of respondents who think that reduction

of poverty is the top priority of social supermarket mission can be found in Serbia (49%).

The share of respondents, who cannot decide what the top priority of social

supermarket mission is, seems to be quite similar in all four observed countries. In

addition, Croatia and Lithuania have almost the same distribution of respondents’

answers regarding this question.

Figure 3

The top priority of social supermarket mission


Except for the question about the top priority of social supermarket mission, in this

part of the questionnaire respondents had to answer 10 more questions about social

markets in general. All questions were given in the Likert scale form. The scale has

consisted of five items where item 1 means completely disagreeing with the statement

whereas item 5 means that the respondent is completely agreed with the given

statement. Table 5 shows the main descriptive statistics results of generally on social

supermarkets variables on the country levels and overall. Due to the nature of the

Likert scale, those results are given just to get a sense of answers distributions and for

comparison with other countries and overall.

189


Table 5

Main descriptive statistics results of generally on social supermarkets variables

Variable Country No of

respondents Mean

Standard

deviation

Social Supermarkets have a noble

purpose and mission because they

return dignity to poor people.

Croatia 117 3.87 1.02

Lithuania 71 3.42 1.04

Poland 123 3.61 1.01

Serbia 108 3.83 0.97

Overall 419 3.71 1.02

I often see volunteers doing

fundraising or collecting food for

poor people.

Croatia 117 2.81 1.16


Poland 123 2.74 1.11

Serbia 108 2.38 1.21

Overall 419 2.80 1.20

If I had a social supermarket

nearby, I'd like to volunteer there.

Croatia 117 3.11 1.14


Poland 123 2.71 1.16

Serbia 108 3.45 1.20

Overall 419 2.97 1.23

When foodstuff is collected at a

local school, at a university, or a

shopping mall, I usually donate.

Croatia 117 3.54 1.13


Poland 123 3.39 1.19

Serbia 108 3.42 1.33

Overall 419 3.36 1.26

In addition to the distribution or

sales of groceries, there is a large

scope for expanding the services

of social supermarkets through the

organization of education,

workshops, etc.

Croatia 117 3.67 0.96


Poland 123 3.34 1.01

Serbia 108 3.58 1.13

Overall 419 3.50 1.03

Social supermarkets may harm

ordinary retailers because if

people get food for free, the

amount of food bought in the

classic stores diminishes.

Croatia 117 2.52 1.30


Poland 123 2.70 1.14

Serbia 108 2.62 1.23

Overall 419 2.60 1.21

Sometimes users (citizens in need)

misuse the goodwill of others. They

are just expecting free stuff without

any effort.

Croatia 117 3.65 1.07


Poland 123 3.60 1.10

Serbia 108 3.59 1.08

Overall 419 3.60 1.07

I have seen or heard in the mass

media about some examples of

social supermarkets and the

reportage was very encouraging, I

liked the concept very much.

Croatia 117 3.58 1.21


Poland 123 2.69 1.40

Serbia 108 2.56 1.36

Overall 419 2.92 1.36

I fear to donate money or

foodstuffs because some affairs

with humanitarian actions

occurred recently, I'm afraid that

my donation will not end in the

right hands.

Croatia 117 3.68 1.14


Poland 123 2.98 1.14

Serbia 108 3.38 1.19

Overall 419 3.31 1.20

The social supermarket should be

extremely active in the usage of

social media and Internet

communication in general.

Croatia 117 4.32 0.78


Poland 123 3.84 1.01

Serbia 108 4.27 0.87

Overall 419 4.10 0.93


In Table 6 the results of conducted chi-square tests for equality of three or more

proportions of generally on social supermarkets variables are shown. Due to fact that

the five-item Likert scale was used and that the sample size is relatively small, to fulfill

190


the prerequisite of the chi-square test used, in the analysis of equality of proportions

responses “agree” and “completely agree” are merged and observed together.

Table 6

Chi-square tests for equality of three or more proportions of generally on social

supermarkets variables, responses “agree” and “completely agree” observed

together

Variable

% of responses Emp.

Chi-

square

p-value Croatia Lithuania Poland Serbia

Social Supermarkets have a noble

purpose and mission because they

return dignity to poor people.

65% 49% 57% 69% 8.2752 0.0407*

I often see volunteers doing

fundraising or collecting food for

poor people.

30% 48% 26% 19% 18.6865 0.0003**

If I had a social supermarket nearby,

I'd like to volunteer there.

38% 18% 24% 51% 27.6233 <0.0001**

When foodstuff is collected at a local

school, at a university, or a shopping

mall, I usually donate.

54% 35% 50% 55% 7.8166 0.0500**

In addition to the distribution or sales

of groceries, there is a large scope

for expanding the services of social

supermarkets through the

organization of education,

workshops, social events, etc.

57% 41% 42% 53% 7.9290 0.0475*

Social supermarkets may harm

ordinary retailers because if people

get food for free, the amount of food

bought in the classic stores

diminishes.

26% 18% 26% 25% 1.7225 0.6319

Sometimes users (citizens in need)

misuse the goodwill of others. They

are just expecting free stuff without

any effort.

59% 54% 61% 56% 1.1795 0.7579

I have seen or heard in the mass

media about some examples of

social supermarkets and the

reportage was very encouraging, I

liked the concept very much.

58% 24% 32% 31% 30.7845 <0.0001**

I fear to donate money or foodstuffs

because some affairs with

humanitarian actions occurred

recently, I'm afraid that my donation

will not end in the right hands.

62% 45% 35% 49% 17.2619 0.0006**

The social supermarket should be

extremely active in the usage of

social media and Internet

communication in general.

85% 68% 65% 81% 18.0006 0.0004**

Note: Sample size according to countries : Croatia=117, Lithuania=71, Poland=123, Serbia=108;

** statistically significant at 1%; * 5%


The results from Table 6 are showing that, at a significance level of 5%, the null

hypothesis can be rejected in eight, from 10, cases. In other words, at eight variables

the structure of respondents who are agreed with the given statements in a country is

different than in other countries. Only at variables “social supermarkets may harm

ordinary retailers because if people get food for free, the amount of food bought in

191


the classic stores diminishes” and “sometimes users (citizens in need) misuse the

goodwill of others – they are just expecting free stuff without any effort” seems not to

be statistically significant differences in proportions of respondents who are agreeing

with the statements between the observed countries.

Frameworks and institutions In the last part of the questionnaire framework and institutions related to social

supermarkets are investigated. To do that eight questions defined in Likert scale form

are used. Again, five-item Likert scale forms were used with items ranging from

completely disagree (code 1) to completely agree (code 5).

Table 7

Main descriptive statistics results of framework and institutions variables

Variable Country No of

respondents Mean

Standard

deviation

Laws in the field of food waste in my

country are good enough.

Croatia 117 2.21 1.04


Poland 123 2.44 1.06

Serbia 108 2.06 0.92

Overall 419 2.32 1.00

Social supermarkets should be

controlled by state bodies as they

contact a very vulnerable group of

citizens.

Croatia 117 3.42 1.23


Poland 123 3.05 1.15

Serbia 108 3.28 1.11

Overall 419 3.26 1.15

There should be a significant

improvement in the legal framework

related to the operation of social

supermarkets.

Croatia 117 4.15 0.75


Poland 123 3.73 0.92

Serbia 108 3.99 0.96

Overall 419 3.89 0.90


integrated into the social welfare

system.

Croatia 117 4.05 0.98


Poland 123 3.50 1.10

Serbia 108 3.91 1.09

Overall 419 3.82 1.05


partially financed from local

government budgets.

Croatia 117 4.08 0.97


Poland 123 3.21 1.22

Serbia 108 3.97 1.10

Overall 419 3.64 1.18

The local government or national

government should provide facilities

for social supermarkets.

Croatia 117 4.13 0.91


Poland 123 3.49 1.08

Serbia 108 4.12 0.98

Overall 419 3.83 1.03

EU funds are too complex and too

demanding and inflexible in

supporting this kind of activity.

Croatia 117 3.39 1.11


Poland 123 3.26 0.96

Serbia 108 3.17 1.00

Overall 419 3.21 1.02

Religious communities should support

the work of social supermarkets.

Croatia 117 4.26 0.98


Poland 123 3.68 1.13

Serbia 108 3.92 1.24

Overall 419 3.84 1.17


192


In Table 7 main descriptive statistics results of framework and institutions variables

are provided. The given results can be used for describing distributions of respondents’

answers and for comparison between the countries and with overall level.

Table 8

Chi-square tests for equality of three or more proportions of framework and institutions

variables, responses “agree” and “completely agree”

Variable

% of responses Emp.

Chi-

square

p-value Croatia Lithuania Poland Serbia

Laws in the field of food waste in my

country are good enough.

11% 13% 13% 5% 5.2684 0.1532


controlled by state bodies as they

contact a very vulnerable group of

citizens.

54% 34% 39% 45% 8.9203 0.0304*

There should be a significant

improvement in the legal framework

related to the operation of social

supermarkets.

80% 56% 63% 68% 14.2226 0.0026**


integrated into the social welfare

system.

79% 69% 61% 68% 8.8736 0.0310**


partially financed from local

government budgets.

76% 37% 46% 69% 42.7853 <0.0001**

Local government or national

government should provide facilities

for social supermarkets.

81% 45% 55% 78% 39.1397 <0.0001**

EU funds are too complex and too

demanding and inflexible in

supporting this kind of activity.

43% 20% 34% 30% 11.3181 0.0101*

Religious communities should

support the work of social

supermarkets.

84% 42% 61% 69% 36.0318 <0.0001**

Note: Sample size according to countries: Croatia=117, Lithuania=71, Poland=123,

Serbia=108** statistically significant at 1%; * 5%


The results of conducted chi-square tests for equality of three or more proportions for

framework and institutions variables are presented in Table 8. To obey the chi-square

test demands, responses “agree” and “completely agree” were observed together

here as well. According to the results, at a significance level of 5%, the null hypothesis

can be rejected in seven, from eight, cases. So, at seven variables the structure of

respondents who agreed or completely agreed with the given statements in a country

is different than in other countries. Only at variable “laws in the field of food waste in

my country are good enough” differences in proportions of respondents who are

agreeing with the statement between the observed countries seem not to be

statistically significant.

Conclusions Social supermarkets emerged across Europe during the economic crisis 2008-2014 as

a solution to the increasing problem of poverty. As a new type of organization, they

distribute food to consumers in need and they find their position at the end of the food

supply chain. Also, they positively contribute to a reduction of food waste that occurs

193


in traditional food supply chains. In the primary research, we addressed and discussed

several research questions.

The first research question is: Is there a need for such type of organization in the CEE

region? Findings show that there is a strong positive attitude regarding the existence

of social supermarkets in all analyzed CEE countries (Croatia, Lithuania, Poland, and

Serbia).

The second research question is: What should be emphasized as a mission of such

an organization? In all given countries, examinees claim that reduction of food waste

should be emphasized as a mission of social supermarket rather than reduction of

poverty. Only in Serbia, examinees claim that poverty should be prioritized in the

mission of social supermarkets.

The third research question is: Is the attitude towards the concept positive or

negative as we are dealing with the concept implemented in post-communist

countries? Generally, there is a positive attitude regarding the concept of social

supermarkets. In all given countries, examinees agree and strongly agree with the

claim “Social supermarkets have a noble purpose and mission because they return

dignity to the poor people”. Moreover, in all countries, (except Lithuania) majority of

examinees agree or strongly agree with the claim “When foodstuff is collected in a

local school, at a university or a shopping mall, I usually donate”. However, there is a

certain concern regarding the misuse of goodwill of others from social supermarket

users and fear that those donations will not end in the right hands.

The fourth research question is: What is the perception towards existing legal

frameworks and state support regarding this type of organization? Examinees agree

that there is a wide space for improvement of a legal framework, welfare system

integration, implementation of state monitoring and control systems when we deal

with social supermarkets as a new type of intermediary organization. Moreover, in all

countries, examinees seek for larger involvement of national governments, local

governments, and religious communities as supporting institutions for social

supermarkets. Regarding EU financing, only in Lithuania majority of respondents do not

agree with the claim that “EU funds are too complex and too demanding and

inflexible in supporting this kind of activity”.

The fifth research question is: Is there a difference in perceptions regarding the

country where the survey was taken? There are differences in perceptions regarding

countries. There are only a few claims where differences in proportions of respondents

who are agreeing with the statements are not statistically significant. From Table 6 we

can observe that out of 10 statements there are only 2 where the p-value is higher

than 0.05. In addition, From Table 8 we can observe that out of 8 statements there is

only one statement where the p-value higher than 0.05. Therefore, we can conclude

that the level of agreement with given statements on social supermarkets differs

regarding the country where the survey was taken.

The research results can be useful for social supermarket managers, policymakers,

and traditional food supply chain managers as the study deal with an emerging form

of intermediary organizations at the end of the food supply chains. The study can be

useful for managers in traditional supply chains when consider implementing

sustainable practices regarding food waste, then for managers in social supermarkets

when considering improvements of their existing strategies and operations. Besides,

findings can be used as a basis for future scientific research within the fields of food

supply chains, social entrepreneurship, corporate social responsibility, and sustainable

development in CEE countries.

There are certain limitations of the study. First of all, it deals only with four CEE

countries (Poland, Lithuania, Croatia, and Serbia), so in the future, more countries

194


should be involved to observe more data and to enable further comparisons in the

region. In addition, there are rather small samples per country. Therefore, further

research should be broadened by involving a larger number of respondents from

various age and socio-demographic groups. Finally, the more complex statistical

methods could be applied to discover more complex relationships and causalities

within the collected data.

References 1. Bogetic, Z., Petkovic, G., Knezevic, B. (2018), “Retail mix and its Specifics in Social

Supermarkets”, in Knezevic, B. (Ed.), Social supermarkets as entrepreneurial ventures in

the socially responsible economy, University of Zagreb, Faculty of Economics and

Business, Zagreb, pp. 73-88.

2. Bolboacă, S. D., Jäntschi, L., Sestraş, A. F., Sestraş, R. E., Pamfil, D. C. (2011), “Pearson-

Fisher Chi-Square Statistic Revisited”, Information, Vol. 2, No. 3, pp. 528-545.

3. Booth, C. (2014), Strategic Procurement. Kogan Page, London, Philadelphia, New Delhi.

4. Christopher, M., Ryals, L. (1999), “Supply Chain Strategy: its impact on shareholder

value”, International Journal of Logistics Management, Vol. 10, No. 1, pp. 1-10.

5. Dees, J. G. (1998), “Enterprising Nonprofits”, Harvard Business Review, January-February,

pp. 55-67.

6. Dees, J. G., Anderson, B. B. (2003), “Sector Bending: Blurring the Lines between Nonprofit

and For-Profit”, Society, Vol. 40, No. 4, pp. 16-27.

7. Emmet, S., Crocker, B. (2009), Excellence in Supplier Management, Cambridge

Academic, Cambridge.

8. EU (2014a), “Poverty and Social Exclusion”, available at:

http://ec.europa.eu/social/main.jsp?catId=751 (18 March 2019)

9. EU (2014b), “Eurostat News release Euroindicators”, available at:

http://epp.eurostat.ec.europa.eu/cache/ITY_PUBLIC/3-31012014-AP/EN/3-31012014-

AP-EN.PDF (18 March 2019)

10. EU (2014c), “EU Employment and Social Situation”, available at:

http://ec.europa.eu/social/main.jsp?catId=737&langId=hr&pubId=7699 (20 March

2019)

11. FAO (2011), “Global food losses and food waste – extent causes and prevention”,

available at: http://www.fao.org/docrep/014/mb060e/mb060e00.pdf (8 December

2015)

12. George, D., Mallery, P. (2003), SPSS for Windows step by step: A simple guide and

reference, Allyn & Bacon, Boston.

13. Holweg, C., Lienbacher, E. (2011), “Social Marketing Innovation: New Thinking in

Retailing”, Journal of Nonprofit & Public Sector Marketing, Vol. 23, No. 4, pp.307-326.

14. Hughes, J., Ralf, M., Michels, B. (1999), Transform your Supply Chain, International

Thomson Press, London.

15. Joshi, A., Kale, S., Chandel, S., Pal, D. K. (2015), “Likert Scale: Explored and Explained”,

British Journal of Applied Science & Technology, Vol. 7, No. 4, pp. 396-403.

16. Kish, L. (1995), Survey Sampling, John Wiley & Sons, New York.

17. Klindzic, M., Knezevic, B., Maric, I. (2016), “Stakeholder Analysis of Social Supermarkets”,

Business Excellence, Vol. 10, No. 1, pp. 151-165.

18. Knezevic, B., Maric, I., Sucur, Z. (2017), “Cross-Sector Cooperation in the Field of Food

Distribution as a Response to Problems of Poverty and Material Deprivation”, Revija za

socijalnu politiku, Vol. 24, No. 2, pp. 143-167.

19. Knezevic, B., Skrobot, P (2018), “Social supermarkets’ profiles in Croatia”, in Knezevic, B.

(Ed.), Social supermarkets as entrepreneurial ventures in socially responsible economy,

University of Zagreb, Faculty of Economics and Business, Zagreb, pp. 147-158.

20. Koivupuro, H-K., Hartikainen, H., Kirsi Silvennoinen, K., Katajajuuri, J-M., Heikintalo, N.,

Reinikainen, A., Jalkanen, L. (2012), “Influence of socio-demographical, behavioural

http://ec.europa.eu/social/main.jsp?catId=737&langId=hr&pubId=7699

195


and attitudinal factors on the amount of avoidable food waste generated in Finnish

households”, International Journal of Consumer Studies, Vol. 36, No. 2, pp. 183-191.

21. Linebacher, E. (2012), Corporate Social Responsibility im Handel, Springer Gabler.

22. Lienbacher, E., Holweg, C. (2011), “Strukturanalyse Sozialmärkte in Österreich”, Eigenverl.

Inst. für Handel und Marketing: Wien, Austria.

23. Lipinski, B., Hanson, C., Lomax, J., Kitinoja, L., Waite, R., Searchinger, T. (2013), “Reducing

food loss and waste”, available at:

http://www.wri.org/sites/default/files/reducing_food_loss_and_waste.pdf (15

December 2015)

24. Lysons, K., Gillingham, M. (2003), Purchasing and Supply Management, Prentice Hall

Financial Times.

25. Maric, I., Knezevic, B. (2014), “Social Supermarkets as a New Retail Format Inspired by

Social Needs and Philantrophy - Case of Croatia”, in K., Demetri (Ed.), Global Business &

Economics Anthology, Business & Economics Society International, Danvers, pp. 278-286.

26. Maric, I., Knezevic, B., Dzambo, D. (2015), “Social Supermarket Rijeka as a Social

Innovation in Food Distribution”, in Knego, N., Renko, S., Knežević, B. (Eds.), Trade

Perspectives 2015: Innovations in Food Retailing, Faculty of Economics Zagreb, Zagreb,

pp. 235-245.

27. Monczka, R.M., Hanfield, R. B., Guinpero, L.C., Patterson, J.L. (2015), Purchasing and

Supply Chain Management, Cengage Learning, Boston.

28. Parfitt, H., Barthel, M., Macnaughton, S. (2010), “Food waste within food supply chains:

quantification and potential for change to 2050”, Philosophical Transactions of the Royal

Society, Vol. 365, pp. 3065-3081.

29. Principato, L., Secondi, L., Pratesi, C. A. (2015), “Reducing food waste: an investigation

on behaviour of Italian youths”, British Food Journal, Vol. 117, No. 2, pp. 731-748.

30. Pullman M., Zhaoui, W. (2012), Food Supply Chain Management: Economic, Social and

Environmental Perspective, Taylor and Francis, Routledge.

31. Schneider, F., Scherhaufer, S., Montoux, H., Gheoldus, M., O’Connor, C., Derain, A.

(2015), “Implementing social supermarkets in Europe”, available at: https://www.eu-

fusions.org/phocadownload/feasibility-

studies/Supermarkets/Advancing%20social%20supermarkets%20report.pdf (20 March

2019)

32. Schroeder, R., Meyer Goldstein, S. (2018), Operations Management in the Supply Chain,

McGraw Hill Education, New York.

33. Stefan, V., Van Herpen, E., Tudoran, A. A., Lähteenmäki, L. (2013), “Avoiding food waste

by Romanian consumers: The importance of planning and shopping routines”, Food

Quality and Preference, Vol. 28, No. 1, pp. 375-381.

34. Van Wheele, A. (2015), Purchasing and Supply Chain Management, Cengage Learning

EMEA, Hampshire.

35. WRAP (2007), “Understanding Food Waste”, available at:

http://www.wrap.org.uk/sites/files/wrap/FoodWasteResearchSummaryFINALADP29_3_

_07.pdf (28 December 2015),

36. Zeljko, D., Prester, J. (2012), “Upravljanje lancem dobave i učinkovit odgovor na potrebe

potrošača u maloprodajnom sektoru”. Ekonomski pregled, Vol. 63, No. 5-6, pp. 322-351.

196


About the authors

Blaženka Knežević, Ph.D. is a Full professor at the Faculty of Economics and Business,

University of Zagreb, Croatia. She teaches courses: Retail information systems;

Economics of electronic commerce; Trade and trade policy; Procurement

management; Supplier relationship management. She participated in various

scientific research projects and published more than 40 papers in conference

proceedings, books, and academic journals. She is a member of the editorial board

of the Business Excellence Journal (BEJ) and the advisory board of Entrepreneurial

Business and Economics Review (EBER). She is the regular reviewer of several

international scientific journals. She can be contacted at [email protected].

Petra Škrobot, Ph.D. candidate and research and teaching assistant at Faculty of

Economics and Business, University of Zagreb, Croatia. Graduated at Trade

department at Faculty of Economics and Business, University of Zagreb. Currently, she

is enrolled in the doctoral study program at the Faculty of Economics and Business,

University of Zagreb. She is doing Ph.D. research in the field of e-commerce and

corporate social responsibility. She is involved in teaching in seminars for courses: Retail

information systems; Economics of electronic commerce; Trade and Trade Policy;

Procurement Management; Supplier relationship management. She can be


Berislav Žmuk graduated with a major in Accounting, post-graduated Statistical

Methods for Economic Analysis and Forecasting, and gained his Ph.D. degree in

Business Economics at the Faculty of Economics and Business, University of Zagreb.

Currently, he is an Assistant Professor at the Department of Statistics, Faculty of

Economics and Business, the University of Zagreb where he teaches the following

subjects: Statistics, Business Statistics, Business Forecasting, and Introduction to

economic statistics. His main research fields include applications of statistics in business

and economy, survey methodology, and statistical quality control. He can be




197


Enterprise Digital Divide: Website e-

Commerce Functionalities among European

Union Enterprises

Božidar Jaković, Tamara Ćurlin

The University of Zagreb, Faculty of Economics and Business, Zagreb, Croatia

Ivan Miloloža

The University of Osijek, Faculty for Dental Medicine & Health, Osijek, Croatia

Abstract

Background: Information and communication technologies (ICTs) gained prevalent

organizational and structural value in the modern economy. E-commerce is one of

the sectors directly influenced by technological change. However, not all countries

have the same opportunities to develop e-commerce growth; there are significant

discrepancies in ICT utilization worldwide, known as the digital divide. Objectives: The

purpose of this paper is to explore the level of difference among European countries

regarding the e-commerce functionalities in their enterprises using a cluster analysis.

Methods/Approach: To accomplish the paper goal, the k-means cluster analysis was

conducted on the Eurostat data from 2019. Enterprises from 28 European countries

were taken into consideration. The Kruskal-Wallis test is used to explore if the

differences among clusters regarding the digital development, measured by the

Digital Economy and Society Index are significant. Results: The investigation confirmed

that there are significant differences among European countries regarding the

development of e-commerce. However, a similar level of e-commerce is not related

to economic and digital development. Conclusions: Since the relationship between

economic development and e-commerce development in European countries is not

linear, country-level policies are likely to be significant factors driving e-commerce

development, which leads to the need for further investigation of this issue.

Keywords: e-commerce; website functionalities; digital divide; European Union

JEL classification: O33


Received: Jan 27, 2021

Accepted: Mar 17, 2021

Citation: Jaković, B., Ćurlin, T., Miloloža, I. (2021). Enterprise Digital Divide: Website e-

Commerce Functionalities among European Union Enterprises. Business Systems

Research, Vol. 12 No. 1, pp. 197-215.


198


Introduction ICT gained prevalent organizational and structural value in the modern economy

(Pazaitis et al., 2017). It became a significant backbone for economic growth and

social development (Latif et al., 2018). Disruptive technologies like robotics, AR, VR,

artificial intelligence, and the Internet of things are now used in various business sectors

daily (Bongomin et al., 2020). New mechanisms, organizations, relations, and

management are being developed by the ICT growth and it will do more in the future

(Neirotti et al., 2018).

E-commerce is an industry that lies in ICT development (Cui et al., 2017). It became

relevant in the 1990s with Internet expansion (Yue et al., 2020). E-commerce provides

various benefits such as overcoming geographical barriers by the ICT utilization, it

consolidates dispersed markets which results in a more immense supply of products

and services offered by the c-commerce enterprises. E-commerce has evolved with

the technological change and it became decisive to understand different dimensions

of it from all perspectives. Nowadays, e-commerce sales have reached 4,13 trillion

dollars, and it’s expected to grow even more extensive, because of mobile

commerce which is expected to take a market share of e-commerce of nearly 80%

(E-commerce, 2020)

Various authors seek to address questions about factors that affect e-commerce

utilization. For instance, Rodríguez-Ardura et al. (2008) stress that Internet security is

pivotal for e-commerce enterprises activities, Alnemr (2010) emphasized trust as an

essential factor for gaining competitive advantage in the e-commerce industry, which

was confirmed by Nica (2015) who also added reputation as a significant element of

e-commerce advancement. However, none of the factors could be considered

competitive if technology adoption is low in the enterprise, which depends on

technology utilization in the country where the enterprise obtains its activity.

The purpose of this paper was to explore if there was a difference among European

countries' e-commerce functionalities. Furthermore, this analysis explored if there are

similarities between the European countries' e-commerce enterprises, which could

divide the countries into homogeneous groups. To accomplish the paper goal, the k

means cluster analysis was conducted on the Eurostat data from 2019. Enterprises from

28 European countries were taken into consideration. Kruskal-Wallis test was used to

explore tests whether the identified clusters are significant. The e-Commerce

functionalities in the European countries were observed thru three dimensions: Website

e-Commerce functionalities, CRM indicator variables, and DESI connectivity

dimension. Every dimension consists of the indicator variables.

The paper is organized as follows: After the Introduction, the Literature review

section presents an overview of the enterprise digital divide, website functionalities,

and CRM as support to e-commerce. The methodology section describes the data

and the methodology used to fulfil the research goal. The results section describes the

descriptive statistic and the cluster analysis results, followed by the discussion where all

research results are displayed. The article’s final section is the conclusion.

Literature review Enterprise digital divide Digital technologies transformed the way people live in the past decade dramatically

and they will continue to do it in the future. From the way of communication to training

and education, disruptive technologies transformed the mechanisms and relations

(Shen et al., 2020). There are numerous positive dimensions of the change, and

enterprises intensively invest their efforts to stay up to date with technologies and gain

199


an advantage from technology utilization (Grover et al., 2018). If the company fails to

stay up to date and take advantage of ICT (information and communication

technologies) solutions, they are at risk of becoming irrelevant and therefore less

competitive. The divergence in the ICT utilization between enterprises resulted in the

exclusion, known as the digital divide (Szeles et al., 2018). The concept of the digital

divide was initially proposed at the beginning of 21 century to describe the

disproportion of the people who have and do not have Internet access (Blank et al.,

2018). Over time, the concept became broader, and started to cover diverse aspects

of ICT, and began to be considered globally (Chen et al., 2004).

The term digital divide implies the social consequences of the phenomena, and it

could be distinguished as (i) global, (ii) social, and (ii) democratic digital divide (Norris,

2001). The global digital divide suggests the discrepancies of Internet access between

high and low developed countries, the social digital divide refers to the information

gap between highly and poorly developed countries, and the democratic digital

divide concerns the differences between countries that use or do not use ICT

resources in the public life.

Numerous initiatives seek to reduce the digital divide between countries, for

instance, the European information society had a few initiatives where the focus was

on government actions that could enhance ICT adoption in European countries (Ojo

et al., 2017). The World Economic Forum, OECD, and G20 are the organizations that

also pay special attention to the topic and commence different international

initiatives and recommendations for countries (Ojo et al., 2017).

The most common barriers to disruptive technologies adoption were split into two

distinguished categories: macroeconomic and microeconomic (Giua, 2020).

Macroeconomic aspects included problems such as lack of innovation culture, lack

of flexibility of the production environment, inoperability, and lack of investments. The

microeconomic barriers are concentrated on the lack of customer demand in low-

development countries, the lack of adaptation of the education system, and the lack

of digital content solutions.

The early investigations on the topic concentrated on the socio-economic

dimensions of the digital divide. Newer investigations focused on the digital divide

measurements approaches, such as Brynjolfsson et al. (2019) who investigated

difficulties in comprehending the value of the digital products and services. Bukht et

al (2017) obtained statistical investigations on the topic, and Ahmad et al. (2019)

proposed analytical frameworks for measuring the digital divide. The more recent

investigations conduct their studies on the large sets of variables where they perform

cross-sectional analysis or time-series analysis (Gunn et al., 2019).

Websites functionalities E-commerce enterprises use their online presence for different reasons such as

marketing, employee recruitment, communication with their partners, etc. (Kremez et

al., 2019). Enterprises nowadays are aware of the significant impact of electronic Word

of Mouth (e-WOM) on their reputation. Internet became the main channel for

enterprises to communicate with their audience, reach their potential customers and,

finally, sell their product (Tsimonis, 2014). Information about customers provided by the

Internet enables e-commerce to create personalized and individual-oriented

products. Digital platforms have become a new phenomenon; today they are one of

the key components of global economic exchange, creating new market

mechanisms (Richardson, 2020). Enterprises often have a presence on social media,

but their website remains the primary focus where customers inform, communicate,

and buy their products or services (Kim et al., 2018). Websites became a distribution

200


channel, which is complementary to the physical stores where customers can

purchase products and services online (Pénard et al., 2017).

There are close to 2 billion digital buyers worldwide and it climbs every year (Statista,

2021). Since the websites became significant revenue drivers, every e-commerce

enterprise should have functional, aesthetic, and relevant websites to stay

competitive (Di Fatta et al., 2018). To check if the website is functional, and the

evaluation process is critical, the e-commerce enterprises' website evaluation can

help the enterprise to modernize their services.

Authors have agreed that some of the website attributes, which should be

evaluated, are precise and complete information, loading speed, website aesthetics

colors, photo, graphics), interactivity, and accessibility (Kwak et al., 2019). Various

techniques are used to evaluate e-commerce website functionalities.

For instance, the authors Albuquerque et al. (2002) created a framework that

evaluates e-commerce website functionalities based on several features such as

usability and reliability. Akhter et al. (2009) use a fuzzy logic system as an instrument

that determines website functionality. Moreover, the authors (Cebi et al, 2013) discuss

the characteristics of e-commerce websites where they investigated the functionality

factor. The paper where authors Al-Qaisi et al. (2015) investigated e-commerce

website functionalities by Mamdani fuzzy system utilization concluded that accuracy

and flexibility are the most important dimensions in website functionality. Furthermore,

the same investigation confirmed website functionality as a key driver of overall

website quality and customer satisfaction.

CRM as support to e-commerce Customer Relationship Management CRM is considered as a business strategy to

decide and manage potential customers and clients to optimize long-term value

(Chen et al., 2003). It aims to recognize and predict the needs of both enterprises and

potential customers. ICT transformed CRM mechanisms in e-commerce, and without

the Internet, we would not know the CRM we know today (Agnihotri, 2021). It provided

new marketing techniques and strategies which enable enterprises to attract and

retain customers online which enforced them to develop new skills and transform their

CRM capabilities online (Li et al., 2020).

Online enterprises need to attract traffic to their website and web stores by the

usage of various online marketing tools such as social media marketing, e-mail

marketing, SEO optimization, and search engine marketing. Customer relationship

management successfully adopted new operations and established personalization

and customization as key activities that secure customer loyalty and recurrent

shopping of their online products or services (Grover et al., 2020). They use

personalization tools, online analysis systems, recommendation platforms, and

feedback tools to establish long-term customer e-loyalty (Oumar et al., 2017). The

customer data processing evolution is a result of the ICT and online strategies

development (Chen, 2017). CRM built its strategies on data analytics results to develop

competitive strategies (Harrison, 2019).

CRM in the context of e-commerce is continuously evolving. There are two main

impacts of CRM in e-commerce: impact on customers and the impact on suppliers

(Thaichon et al.,2020). There are two main CRM strategies on the impact on customers:

pushing information that impact and collective behavior and investigating in terms of

better customer control over configuration and prices of goods and services and the

wider array of options. CRM's impact on supplies considers creating new demand

chains, effectively communicating, and technology adopting. CRM has an impact

201


on overall e-commerce enterprises; it shapes organizational culture, sales and

marketing functions, marketing strategies, and support functions (Lin et al., 2010).

Because of the interconnection with digital technologies, the success of CRM

depends on ICT adoption. Today, CRM is no longer a competitive strategy, it is a

necessity, it is essential for e-commerce enterprises to stay competitive and attract

online customers. Therefore, for e-commerce enterprises, being up to date and

investing in the technologies is decisive for success.

Methodology Research variables The observed research variables were obtained from the Eurostat database for the

year 2019. European Union member countries (28 countries) at the given period were

taken into consideration. We were concentrated on the EU members only, so we did

not include other European countries in the research because of the lack of data and

variables of our focus.

The e-Commerce functionalities in the European countries were observed thru

three dimensions: Website e-Commerce functionalities, CRM indicator variables, and

DESI connectivity dimension.

E-commerce website functionalities were measured for the European countries

enterprises, with ten or more employed persons, with the financial sector excluded:

o WEB_PRICE - website describing goods or services, price-lists (% of enterprises)

o WEB_SHOP - website with online ordering or reservation or booking, e.g. the

shopping cart (% of enterprises)

o WEB_TRACK - website with order tracking available online (% of enterprises)

o WEB_SOCIAL- website with links or references to the social media (% of

enterprises)

o WEB_BOT – website with the chat-bot (% of enterprises)

o WEB_BUY_BOT – website with the chat-bot supporting buying process (% of

enterprises)

The CRM indicator variables were measured also for the European enterprise with

ten or more employees and without the financial sector included:

o CRM- Enterprises using software solutions like Customer Relationship

Management (% of enterprises)

o CRM_ANALYSIS -Enterprises using Customer Relationship Management to

analyze information about clients for marketing purposes (% of enterprises)

o CRM_STORE-Enterprises using Customer Relationship Management to capture,

store and make available client’s information to other business functions (% of

enterprises)

DESI index represents a summary of the relevant indicators on Europe’s digital

performance and tracks the evolution of EU Member States in digital competitiveness

(EU, 2020). For this investigation, five indicator variables were considered:

o DESI_1_CONN – Connectivity

o DESI_2_HC – Human capital

o DESI_3_UI - Use of Internet Services

o DESI_4_IDT - Integration of Digital Technology

o DESI_5_DPS - Digital Public Services

DESI Connectivity Dimension is measured as the weighted average of the four sub-

dimensions: (i) Fixed Broadband take-up, (ii) Fixed broadband coverage, (iii) Mobile

broadband, and (iv) Broadband price index.

202


DESI Human Capital Dimension is calculated as the weighted average of the two

sub-dimensions: (i) Internet User Skills, and (ii) Advanced Skills and Development.

DESI Use of Internet Dimension is calculated as the weighted average of the three

sub-dimensions: (i) Internet Use (ii), Activities Online, and (iii) Transactions.

DESI Integration of Digital Technology Dimension is calculated as the weighted

average of the two sub-dimensions: (i) Business digitization and (ii) e-Commerce.

DESI Digital Public Services Dimension is calculated by taking the score for e-

Government.

K-means cluster analysis Cluster analysis is a knowledge discovery technique that is utilized for the identification

and classification of similar groups of statistical indicator variables. The variables are

homogenous within the group and heterogeneous among the other groups.

Cluster analysis is a form of unsupervised learning, and the goal of cluster analysis is

to explore hidden patterns or to identify groups of objects with similar traits. Partition

clustering and hierarchical clustering are two prevalent groups of cluster analysis

(Govender et al, 2020). The analysis starts with the research item identification,

followed by the clustering procedure selection. For this investigation, the

nonhierarchical cluster analysis with the K-means algorithm was calculated to

systemize variables into comprehensive groups.

K-means falls under the partition clustering method, where data cluster groups

have no overlapping. K-means technique is utilized to divide n observations into k non-

overlapping groups where each observation belongs to one cluster with the nearest

mean. K-means is often used to process a large of data to be representative data,

called cluster centers (San et al., 2004). For this investigation, the K-means algorithm

was used to group the indicator variables into nested groups, starting with all statistical

units in one group, after which it divides them using the top-down method. A V-fold

approach with 10 folds was used to test the validity of the solution. Euclidean distance

was used to distribute iteratively research data to the cluster with the closest centroid.

The process resulted in the selected three clusters.

Furthermore, the statistical difference between clusters regarding the DESI

indicators was investigated by the usage of the Kruskal-Wallis test, which tests whether

the identified clusters.

Results Descriptive statistics analysis Descriptive statistics are presented in Table 1. For the sample of the enterprises without

financial sector (10 persons employed or more) of 28 European countries and 3

encompassed dimensions: Web site e-commerce functionalities, Customer

relationship management, and DESI indicators (The Digital Economy and Society

Index, 2020).

The majority of the European countries' e-Commerce websites had the highest

average grades for the dimension Web site e-Commerce functionalities, especially for

the variable WEB_BOT which was found on 62.11% of websites. The second research

item with the high average grades was the WEB_PRICE item, which indicates that

61.04% of e-Commerce websites in European countries have a price list or descriptions

of the goods and services. On the other hand, the variables DESI_3_UI and DESI_4_ITD

for the European countries had the lowest average grade (8.25%), which suggests that

the integration of digital technologies and digital public services are utilized the least

in e-Commerce between European countries. As for the Customer relationship

203


management, the CRM_ANALYSIS variable has notably lower results than CRM and

CRM_STORE, which puts forward that the e-Commerce enterprises in European

countries should broadly implement systems for CRM analysis to enhance client

relationships.

Table 1

Descriptive statistics of e-Commerce usage indicators and obstacles for selected

European countries N Minimum Maximum Mean Std. Dev. Skewness Kurtosis

Web site e-commerce functionalities

WEB_PRICE 28 34.00 96.00 61.04 16.585 -0.104 -0.764

WEB_SHOP 28 9.00 34.00 20.50 7.351 0.452 -0.820

WEB_TRACK 28 3.00 14.00 8.36 2.313 0.013 0.858

WEB_SOCIAL 28 15.00 68.00 39.54 13.686 0.286 -0.481

WEB_BOT 28 36.00 86.00 62.11 15.586 -0.295 -1.156

WEB_BUY_BOT 28 9.00 34.00 19.50 6.995 0.595 -0.455

Customer relationship management

CRM 28 12.00 56.00 29.93 10.360 0.437 -0.003

CRM_ANALYSIS 28 7.00 26.00 18.18 5.464 -0.195 -0.945

CRM_STORE 28 11.00 55.00 28.57 10.609 0.428 -0.058

DESI indicators

DESI_1_CONN 28 7.37 15.01 11.66 1.922 -0.029 -0.206

DESI_2_HC 28 7.13 19.38 12.07 3.139 0.428 -0.323

DESI_3_UI 28 5.24 11.29 8.25 1.617 0.336 -0.254

DESI_4_ITD 28 3.38 13.82 8.25 2.775 0.301 -0.671

DESI_5_DPS 28 6.75 12.74 10.12 1.807 -0.306 -1.081

Source: Authors work (Eurostat, 2019)

Cluster analysis Graph of cost sequence, which displays the error function for the different cluster

numbers, was produced to investigate the best number of clusters for the sample data

presented. The error function could be interpreted as the average distance of

observations of the explored dataset from the cluster centroids to which the

observations were assigned (Sugar et al., 2003).

The objective is to minimize the cluster cost to the preferable value (Amaro et al,

2016). Various methods could be used to identify the preferable number of the cluster.

For this investigation, the Elbow method was chosen as the decision indicator. Figure

1 shows “the elbow” point on the number of the three clusters. The decrease of the

error function is considered to be large to the point of the three clusters, after which it

decreases slightly. Decrease between 3 and 4 number of clusters is less than 5%.

Therefore, the number of clusters selected is the optimal solution and three clusters will

be observed in further analysis.

204


Figure 1

Graph of cost sequence

Source: Authors work

Moreover, the Anova analysis was undertaken for the three selected clusters and

the dimension Website e-Commerce functionalities. The table demonstrates six

variables that exemplify the given dimension. Furthermore, the null hypnotizes was

proposed. All the variables came out as statistically significant at 1% and the null-

hypnotizes were rejected which suggests that the means between the variables

observed statistically differ. Table 2 confirms that the selected number of three clusters

included in the investigation is justified.

Table 2

The Anova analysis Between SS df Within SS df F p-value

WEB_PRICE 5851.254 2 1575.710 25 46.418 0.000***

WEB_SHOP 792.444 2 666.556 25 14.861 0.000***

WEB_TRACK 53.305 2 91.123 25 7.312 0.003***

WEB_SOCIAL 3314.990 2 1741.974 25 23.788 0.000***

WEB_BOT 5301.468 2 1257.210 25 52.711 0.000***

WEB_BUY_BOT 904.162 2 416.838 25 27.114 0.000***

Source: Authors work (Eurostat, 2019) **Note: statistically significant at 1%

The descriptive statistics of the e-Commerce variables of the dimension: Website e-

commerce functionalities was conducted. Cluster 1 includes 7 European countries

and has the highest means and the standard deviations for all the variables included

which suggests that the countries included in Cluster 1 have the highest developed e-

Commerce functionalities among European Union enterprises. As for Cluster 2 and 3,

Cluster 2 which includes the most European countries, a total of 11 have the higher

standard deviation and cluster mean in all observed values, except WEB_TRACK

where cluster three outperform Cluster 2. Cluster 3 consists of a total of ten European

countries. Table 3 shows the cluster means and standard deviations.

205


Table 3

Cluster means and standard deviations Cluster 1 Cluster 2 Cluster 3

WEB_PRICE 77.29

(9.673)

67.73

(7.431)

42.30

(7.166

WEB_SHOP 29.71

(3.773)

17.45

(5.087)

17.40

(5.985)

WEB_TRACK 10.43

(1.813)

6.91

(2.256)

8.50

(1.509)

WEB_SOCIAL 57.71

(6.576)

36.64

(8.869)

30.00

(8.794)

WEB_BOT 77.29

(6.525)

68.73

(7.377)

44.20

(7.131)

WEB_BUY_BOT 29.29

(3.729)

16.91

(4.460)

15.50

(3.866)

Number of cases 7 11 10

Percentage (%) 25.00 39.29 35.71

Source: Authors work (Eurostat 2019)

Note: Standard deviations in parenthesis

Figure 2 displays the distribution of variables across clusters for the dimension Website

e-commerce functionalities. The distributions can be used to get an insight into how

many variables in a cluster differ according to the observed variable. The narrower

distribution is the smaller difference among the variables across clusters. Additionally,

the taller the distribution is, the differences between variables are larger.

The variable WEB_PRICE that suggests that the e-Commerce websites provided a

description of the goods or services or the price lists, shows the normal distribution for

all three clusters, with similar values with slightly higher probability density peaks in

Cluster 2 and 3 than Cluster 1. The Cluster 2 and 3 distribution are narrower than Cluster

1 distribution. The variable WEB_SHOP shows the European enterprises' website are

provided with online ordering or reservation or booking, e.g. shopping cart. Cluster 1

shows the highest distribution peak and the furthest from the graph origin. Cluster 3

shows the lowest probability density peak and the widest distribution of all three

clusters.

The variable WEB_TRACK portrays the enterprises where the website provided order

tracking available online. The values were the highest for Cluster 3, which correlates

with the results from the Cluster means and the standard deviation table following

Cluster 1. Cluster 2 showed the widest distribution for the given variable with the lowest

peak closest to the origin. The WEB_SOCIAL presents data of the European enterprises

where the website had links or references to the enterprise's social media profiles.

206


Figure 2

Distribution of variables across clusters


Discussion Cluster membership across countries As mentioned in the previous section, Cluster 1 contains seven countries: Belgium,

Denmark, Ireland, Malta; Netherlands; Finland, and Sweden. All countries in Cluster 1

are highly developed. Cluster 2 is the largest; it consists of eleven European countries:

Germany, Estonia, France, Cyprus, Latvia, Luxembourg, Austria, Poland, Slovenia,

Slovakia, and United Kingdom. The countries' structure for the Cluster 2 is diverse, it

consists of both some of the highly developed countries such as the UK, Germany, and

France, and some of the countries, which struggled hardly thru economic crises, but

recovered, such as Slovakia and Cyprus. Cluster 3 consists of the 10 European

207


countries: Bulgaria, Czechia, Greece, Spain, Croatia, Italy, Lithuania, Hungary, Poland,

and Romania, which are mostly developing countries (e.g. Croatia) and the countries,

struggling with the economic crisis (e.g. Greece).

Table 4

Countries across clusters

Cluster Country

Cluster 1 Belgium, Denmark, Ireland, Malta, Netherlands, Finland, Sweden

Cluster 2 Germany, Estonia, France, Cyprus, Latvia, Luxembourg, Austria, Poland,

Slovenia, Slovakia, United Kingdom

Cluster 3 Bulgaria, Czechia, Greece, Spain, Croatia, Italy, Lithuania, Hungary,

Portugal, Romania


Figure 3 presents the European map according to the three specified clusters. There

can be identified similar socio-economic development similarities as well as the close

geographic position within clusters. All Scandinavian countries are grouped in Cluster

1 (Denmark, Finland, and Sweden) which are considered highly developed. Alongside

Scandinavian countries, Cluster 1 consists of two of three Benelux countries (Belgium

and Netherlands), and, Ireland, and Malta which are all highly developed. Therefore,

Cluster 1 could be considered as highly developed. Cluster 2 consists of most of the

Central European countries, which are presented, in the figure. The countries included

are developed, some are highly developed, and the cluster overall could be defined

as developed. Cluster 3 consists of developing countries and countries, which are

recovering from the economic crises. According to the map, the countries included

in the Cluster three are mostly eastern European countries. Therefore, Cluster 3 could

be defined as developing.

Figure 3

European map according to countries grouped into specific clusters

Source: Authors work using mapchart.net

The k-means analysis shows data for a sample of 28 European countries for indicators

of the dimensions of the webshop e-commerce functionalities (Figure 4). It represents

the mean value for the six observed variables from the given dimension: WEB_PRICE,

WEB_SHOP, WEB_TRACK, WEB_SOCIAL, WEB_BOT, and WEB_BUY_BOT. The mean values

208


were observed within three identified clusters. By comparing result means, interesting

conclusions have emerged, such as knowledge that e-Commerce webshops from

highly developed countries from Cluster 1 outperform the other two clusters when it

comes to all variables (not only webshop). Cluster 1. Also has high means regarding

web_social, web_bot, and web_buy_bot. The Cluster 2 countries focus on building

customer relationships including innovative digital marketing strategies in their

websites. The Cluster 1 values are the furthest from the origin so they show the most

significant values.

Figure 4

Cluster means of indicators of webshop e-commerce functionalities


European countries grouped in Cluster 1 (highly developed) outperform Cluster 2

and Cluster 3 for all the selected variables from the dimension webshop e-Commerce

functionalities. The average means of the observed variables are higher than any

variable from Cluster 2 and Cluster 3. This knowledge correlates to the findings of the

countries selected for Cluster 1, which are highly developed countries such as

Scandinavian countries. Most enterprises in Cluster 1 countries have webshops

integrated into their e-Commerce websites, and the least of them have the web track

variable.

The Cluster 2 countries (developed) have the highest average means value for the

variable WEB_BOT. It suggests digital marketing strategies and customer relationship is

essential for the countries from Cluster 2. The average mean for the variable web price

is also high, opposite to the values from the other clusters for the given variable, which

also could be the strategy to intensify customer relationship. The Cluster 2 European

countries have the lowest average means for the variables WEB_SHOP that represents

online shop or booking and WEB_BUY_BOT. This could be explained as the Cluster 2

countries' e-Commerce websites focus their website on digital marketing and building

customer loyalty.

209


The Cluster 3 countries (developing) have the highest mean values from the

webshop e-Commerce functionalities dimension for the variable WEB_TRACK. This

could be explained as the some of the countries from Cluster 3 are the newer

members of the EU or they do not have some of the most developed and reliable

shipping companies as UPS and DHL so the shipment could last longer than in

developed and highly developed countries, so the tracking is essential. The

WEB_BUY_BOT and the WEB_PRICE variables have the lowest mean values, which

mean that Cluster 3 countries do not yet concentrate on disruptive technology

implementation and customer relationships.

The difference between European countries is presented by the cluster analysis. The

difference between cluster countries. There are few indicators of diversity between

clusters identifies (i) development which explains the outperformance of the highly

developed countries related to other clusters, the countries technological structure,

and countries orientation to technology development as the online shopping habits.

Relationship between cluster membership and CRM

implementation

Table 5 displays the comparison of cluster members according to CRM indicator

variables. Descriptive statistic was used to and the Kruskal-Wallis test was used to

explore the statistical differences of the standard deviations. The Kruskal-Wallis test

confirms that all the clusters are statistically significant at the 1% for the CRM and

CRM_ANALYSIS variable, and at the 5% for the CRM_STORE indicator variable.

Table 5

Comparison of cluster members according to CRM indicator variables – Descriptive

analysis and Kruskal-Wallis test CRM CRM_ANALYSIS CRM_STORE

Cluster 1 Mean 39.86 24.43 37.86 N 7 7 7 Std. Deviation 8.80 1.81 9.67



Total Mean 29.93 18.18 28.57 N 28 28 28 Std. Deviation 10.36 5.46 10.61

Kruskal-Wallis test Kruskal-Wallis H 10.118 13.504 9.151 df 2 2 2 Asymp. Sig. 0.006*** 0.001*** 0.010**


Note: *** statistically significant at 1%; ** 5%

Figure 5 represent error bars of means of CRM indicator variables across clusters.

Standard deviations are the lowest for the CRM_analysis variable indicator and are

similar for variables CRM and CRM_STORE. Error bars are the largest for the Cluster 1

values and the smallest for the Cluster 3 variables.

210


Figure 5

Error bars (95%) of means of CRM indicator variables across clusters


Relationship between cluster membership and DESI index Table 6 present the comparison of cluster members according to DESI index variables.

Descriptive statistic was used to and the Kruskal-Wallis test was used to explore the

statistical differences of the standard deviations. The Kruskal-Wallis test confirms that

all the differences among the variables included are statistically significant.

DESI_1_CONN and CRM_ANALYSIS variables are significant at 1%, and the CRM_STORE

variable is significant at 5%.

Table 6

Comparison of cluster members according to DESI index variables – Descriptive

analysis and Kruskal-Wallis test DESI_1_

CONN

DESI_2_

HC

DESI_3_

UI

DESI_4_

ITD

DESI_5_

DPS

Cluster 1 Mean 12.52 15.40 10.06 12.06 11.60 N 7 7 7 7 7 Std. Dev. 2.05 2.49 1.33 1.17 0.86



Total Mean 11.66 12.07 8.25 8.25 10.12 N 28 28 28 28 28 Std. Dev. 1.92 3.14 1.62 2.77 1.81

Kruskal-Wallis test Kruskal-

Wallis H

1.668 14.226 13.575 14.830 8.102

df 2 2 2 2 2 p-value 0.434 0.001*** 0.001*** 0.001*** 0.017**


Note: *** statistically significant at 1%; ** 5%

211


Figure 6

Error bars (95%) of means of DESI index variables across clusters


The DESI connectivity dimensions mean values are the highest for the European

countries from Cluster 1, which means that the highly developed countries are the

most connected between three Clusters. The Cluster 1 counties have the highest

mean for the human capital connectivity dimension, so the Internet user skills and

advanced skills and development are developed the highest among the given

variables.

Oppositely, the Cluster 1 countries have the lowest mean for the DESI indicator Use

of Internet which indicates that the Internet usage, activities online, and transaction

perform the lowest for the Cluster 1 countries. Cluster 2 have also the highest mean for

the variable Human capital with a mean value of 12.30. Cluster 2 performs the lowest

at the DESI 4 variable: Digital Public Services Dimension, which is the Government

connectivity score. The DESI 4 variable is also the lowest variable for the Cluster 3

European countries, with a mean of 6.70. The Cluster 4 countries perform the best at

the Connectivity indicator with a mean of 11.03.

Figure 6 represent error bars of means of indicator variables DESI index variables

across clusters. Standard deviations are the lowest for the DESI_3_UI variable indicator.

Error bars are the largest for the DESI_2_HC values, which means that human skills are

highly developed in the Cluster 1 countries.

Conclusion E-Commerce is one of the industries where disruptive technologies and internet

development had the strongest impact on mechanisms and operation. E-Commerce

has transformed during the past decade, and it is now positioned as one of the most

competitive industries.

The digital divide is the consequence of the discrepancy between ICT usages in

various countries. There is a gap between countries, which are ICT high adopters and

low adopters, which affect the global economy

This paper investigated the digital divide between European Union countries

enterprises by performing k means cluster analysis and the Kruskal-Wallace test.

Interesting knowledge useful for both practitioners and academics has emerged. The

212


existence of similarities between the groups of European countries has been

confirmed.

The k-means cluster analysis divided the European country's enterprises into three

separated homogenous groups. The countries in the same clusters showed similarities

between the two factors: geographical position and the e-commerce development

state.

Therefore, the three identified clusters could be determined as highly developed,

developed, and developing. The highly developed cluster countries outperformed

other cluster countries in all observed dimensions: website e-commerce functionalities,

CRM, and DESI index.

The Highly developed cluster countries are mostly Scandinavian and Benelux

countries, which have good technological and internet infrastructure, and high

standards. The developed cluster countries are mostly central European countries,

which mostly perform higher than the developing countries, and lower than the highly

developed cluster countries. The developing countries are mostly eastern European

countries as well as the countries who struggled recently with some kind of economic

crisis such as Greece and Portugal. The Developing countries cluster mostly performs

the lowest on all observed functionalities.

The e-commerce functionalities development across countries are often

interrelated with overall economic development, which indicated that the counties

which show lower economic growth have also lower e-commerce functionalities.

The European e-commerce enterprise countries clustering could be very useful for

further investigations on the topic. The knowledge that e-commerce functionalities

are not evenly distributed across European countries could be useful for both

investigators and practitioners. The clusters could be analyzed separately or the highly

developed cluster practice could serve as a benchmark for other countries. This could

be useful for practitioners as well.

Furthermore, this study is not without limitations, further investigations could

concentrate on the particular sector to get more comprehensive results. Additionally,

more variables could be included, such as more website functionalities, or more CRM

dimensions. Future investigations should also consider newer trends in e-commerce

such as social media and mobile shopping.

References 1. Agnihotri, R. (2021), “From sales force automation to digital transformation: how social

media, social CRM, and artificial intelligence technologies are influencing the sales

process”, In Jaramillo, F., Mulki, J.P. (Ed.), A Research Agenda for Sales, Cheltenham,

Edward Elgar Publishing, pp. 21-47.

2. Ahmad, N. A., Ayub, A. F. M., Khambari, M. N. (2019), “Gender digital divide: digital skills

among Malaysian secondary school“, International Journal of Academic Research in

Progressive Education and Development, Vol. 8 No. 4, pp. 668-687.

3. Akhter, F., Buzzi, M. C., Buzzi, M., Leporini, B. (2009, September), „Conceptual framework:

How to engineer online trust for disabled users“, In 2009 IEEE/WIC/ACM International

Joint Conference on Web Intelligence and Intelligent Agent Technology, Vol. 3, pp.

614-617

4. Al-Qaisi, L., Mesleh, A., Al-Qaisi, A., Sharadqh, A., Zahran, B., Alkasassbeh, J., Olaimat,

M. (2015), „Evaluation of e-commerce website functionality using a Mamdani fuzzy

system“, Engineering, Technology & Applied Science Research, Vol. 5 No. 5, pp. 860-

863.

5. Albuquerque, A. B., Belchior, A. D. (2002), „E-commerce websites: a qualitative

evaluation“, WWW 2002 Poster Session.

213


6. Alnemr, R., König, S., Eymann, T., Meinel, C. (2010, „Enabling usage control through

reputation objects: A discussion on e-commerce and the internet of services

environments“, Journal of theoretical and applied electronic commerce research,

Vol. 5 No. 2, pp.59-76.

7. Amaro, S., Duarte, P., Henriques, C. (2016), “Travelers’ use of social media: A clustering

approach”, Annals of Tourism Research, Vol. 59 No. C, pp. 1-15.

8. Blank, G., Lutz, C. (2018), „Benefits and harms from Internet use: A differentiated analysis

of Great Britain“, New media & society, Vol. 20 No. 2, pp. 618-640.

9. Bongomin, O., Gilibrays Ocen, G., Oyondi Nganyi, E., Musinguzi, A., Omara, T. (2020),

„Exponential disruptive technologies and the required skills of industry 4.0.“, Journal of

Engineering, 2020.

10. Brynjolfsson, E., Collis, A. (2019), „How should we measure the digital economy“,

Harvard Business Review, Vol. 97 No.6, pp. 140-148.

11. Bukht, R., Heeks, R. (2017), „Defining, conceptualising and measuring the digital

economy“, Development Informatics working paper, Vol. 68., Available at

SSRN: https://ssrn.com/abstract=3431732 / (20 December 2020)

12. Cebi, S. (2013), „Determining importance degrees of website design parameters based

on interactions and types of websites“, Decision Support Systems, Vol. 54 No. 2, pp.

1030-1043.

13. Chen, I. J., Popovich, K. (2003), „Understanding customer relationship management

(CRM)“, Business process management journal Vol. 9 No. 5, pp. 672-688.

14. Chen, W., Wellman, B. (2004), „The global digital divide–within and between

countries“, IT & Society, Vol.1 No. 7, pp. 39-45.

15. Chen, Z., Li, Y., Wu, Y., Luo, J. (2017), „The transition from traditional banking to mobile

internet finance: an organizational innovation perspective-a comparative study of

Citibank and ICBC“, Financial Innovation, Vol. 3 No.1, pp. 1-16.

16. Cui, M., Pan, S. L., Newell, S., Cui, L. (2017), „Strategy, resource orchestration and e-

commerce enabled social innovation in Rural China“, The Journal of Strategic

Information Systems, Vol. 26 No. 1, pp. 3-21.

17. Di Fatta, D., Patton, D., Viglia, G. (2018), „The determinants of conversion rates in SME

e-commerce websites“, Journal of Retailing and Consumer Services, Vol. 41, pp. 161-

168.

18. E-commerce (2020), „The e-commerce guide“, available at:

https://ecommerceguide.com/ecommerce-statistics/ (20 December 2020)

19. Giua, C., Materia, V. C., Camanzi, L. (2020), “Management information system

adoption at the farm level: evidence from the literature”, British Food Journal

20. Govender, P., & Sivakumar, V. (2020), “Application of k-means and hierarchical

clustering techniques for analysis of air pollution: a review (1980–2019)”, Atmospheric

Pollution Research, Vol. 11 No. 1, pp. 40-56.

21. Grover, P., Kar, A. K., Dwivedi, Y. K. (2020), „Understanding artificial intelligence

adoption in operations management: insights from the review of academic literature

and social media discussions“ Annals of Operations Research, pp.1-37.

22. Grover, V., Chiang, R. H., Liang, T. P., Zhang, D. (2018), „Creating strategic business

value from big data analytics: A research framework“, Journal of Management

Information Systems, Vol. 2 No. 35, pp. 388-423.

23. Gunn, V., Muntaner, C., Ng, E., Villeneuve, M., Gea-Sanchez, M., Chung, H. (2019),

„Gender equality policies, nursing professionalization, and the nursing workforce: a

cross-sectional, time-series analysis of 22 countries, 2000–2015“, International journal of

nursing studies, pp. 99, 103388.

24. Harrison, D. E., Ajjan, H. (2019), „Customer relationship management technology:

bridging the gap between marketing education and practice“, Journal of Marketing

Analytics, 7(4), pp. 205-219.

25. Kim, N., Kim, W. (2018), “Do your social media lead you to make social deal purchases?

Consumer-generated social referrals for sales via social commerce”, International

Journal of Information Management, Vol. 39 pp. 38-48.

https://ssrn.com/abstract=3431732

214


26. Kremez, Z., Frazer, L., Thaichon, P. (2019), „The effects of e-commerce on franchising:

Practical implications and models“, Australasian Marketing Journal (AMJ), Vol. 27 No.

3, pp.158-168.

27. Kwak, D. H., Ramamurthy, K. R., & Nazareth, D. L. (2019), “Beautiful is Good and Good

is Reputable: Multiple-Attribute Charity Website Evaluation and Initial Perceptions of

Reputation Under the Halo Effect”, Journal of the Association for Information Systems,

Vol. 20 No.11, pp.3.

28. Latif, Z., Latif, S., Ximei, L., Pathan, Z. H., Salam, S., Jianqiu, Z. (2018), „The dynamics of

ICT, foreign direct investment, globalization and economic growth: Panel estimation

robust to heterogeneity and cross-sectional dependence“, Telematics and

Informatics, Vol. 35 No. 2, pp. 318-328.

29. Li, F., Larimo, J., Leonidou, L. C. (2020), „Social media marketing strategy: definition,

conceptualization, taxonomy, validation, and future agenda“, Journal of the

Academy of Marketing Science, Vol. 49, pp. 1-20.

30. Lin, R. J., Chen, R. H., Chiu, K. K. S. (2010), „Customer relationship management and

innovation capability: an empirical study“, Industrial Management & Data Systems. Vol.

110 No. 1, pp. 111-133.

31. Neirotti, P., Raguseo, E., Paolucci, E. (2018), „How SMEs develop ICT-based capabilities

in response to their environment“, Journal of Enterprise Information Management, 31(2)

32. Nica, E. (2015), „Positive drivers of consumer trust in e-commerce.“, Journal of Self-

Governance and Management Economics, Vol. 3 No. 1, pp. 60-65.

33. Norris, P. (2001), „Digital divide: Civic engagement, information poverty, and the

Internet worldwide“, Cambridge, Cambridge University Press.

34. Ojo, A., Adebayo, S. (2017), „Blockchain as a next generation government information

infrastructure: A review of initiatives in D5 countries“, Government 3.0–Next Generation

Government Technology Infrastructure and Services, pp. 283-298.

35. Oumar, T. K., Mang’Unyi, E. E., Govender, K. K., Rajkaran, S. (2017), „Exploring the e-

CRM–e-customer-e-loyalty nexus: a Kenyan commercial bank case study“,

Management & Marketing. Challenges for the Knowledge Society, Vol. 12 No. 4, pp.

674-696.

36. Pazaitis, A., De Filippi, P., Kostakis, V. (2017), „Blockchain and value systems in the

sharing economy: The illustrative case of Backfeed“, Technological Forecasting and

Social Change, Vol. 125, pp. 105-115.

37. Pénard, T., Perrigot, R. (2017) „Online search–Online purchase in franchising: An

empirical analysis of franchisor website functionality“, Journal of Retailing and

Consumer Services, Vol. 39, pp. 164-172.

38. Richardson, L. (2020), „Coordinating office space: Digital technologies and the

platformization of work“, Environment and Planning D: Society and Space.

doi:10.1177/0263775820959677

39. Rodríguez-Ardura, I., Meseguer-Artola, A., Vilaseca-Requena, J. (2008), „Factors

influencing the evolution of electronic commerce: an empirical analysis in a developed

market economy“, Journal of theoretical and applied electronic commerce

research, Vol. 3 No. 2, pp.18-29.

40. San, O. M., Huynh, V. N., Nakamori, Y. (2004), “An alternative extension of the k-means

algorithm for clustering categorical data”, International Journal of applied

mathematics and computer science, Vol. 14 No. 2, pp. 241-247.

41. Shen, C. W., Ho, J. T. (2020), “Technology-enhanced learning in higher education: A

bibliometric analysis with latent semantic approach”, Computers in Human Behavior,

Vol. 104, pp. 106-177.

42. Statista, (2021), „E-commerce worldwide - statistics & facts“, available at:

https://www.statista.com/topics/871/online-shopping/ (08 January 2021)

43. Sugar, C. A., James, G. M. (2003), “Finding the number of clusters in a dataset: An

information-theoretic approach”, Journal of the American Statistical Association, Vol.

98 No. 463, pp. 750-763.

44. Szeles, M. R. (2018), „New insights from a multilevel approach to the regional digital

divide in the European Union“, Telecommunications Policy, Vol. 42 No. 6, pp.452-463.

https://doi.org/10.1177/0263775820959677

215


45. Thaichon, P., Surachartkumtonkun, J., Quach, S., Weaven, S., Palmatier, R. W. (2018),

“Hybrid sales structures in the age of e-commerce”, Journal of Personal Selling & Sales

Management, Vol. 38 No. 3, pp. 277-302.

46. The Digital Economy and Society Index (2020), “DESI index 2020”, available at:

https://ec.europa.eu/digital-single-market/en/digital-economy-and-society-index-

desi (08 January 2021)

47. Tsimonis, G., Dimitriadis, S. (2014), „Brand strategies in social media“, Marketing

Intelligence & Planning. Vol. 32 No. 3, pp. 328-344.

48. Yue, T. (2020), „The Strategic Management of Travel E-Commerce Platform (OTA) in the

“Post-Epidemic” Period.“ In 2020 4th International Seminar on Education, Management

and Social Sciences (ISEMSS 2020), pp. 1018-1020.

About the authors Božidar Jaković, Ph.D., is an Associate Professor and currently Vice Dean at the Faculty

of Economics & Business, University of Zagreb, Croatia. He received his Ph.D., MSc, and

BSc degrees from the Faculty of Economics and Business, University of Zagreb. In

addition, he is an author of numerous articles in journals on the topic of e-commerce.

His current research interests include e-commerce, web services, mobile technologies

and applications, document management, and e-learning, Knowledge

management, and Information management. He is actively engaged in several

scientific projects. The author can be contacted at [email protected].

Tamara Ćurlin is a Teaching Assistant and a Ph.D. student at the Faculty of Economics

and Business, University of Zagreb, Department of Informatics. She received her BSc

and MSc degrees from the Faculty of Economics and Business, University of Zagreb.

She is teaching Informatics and Enterprise Information Systems courses exercises. Her

current research interests include Information Technologies in Tourism, Mobile

Technologies, Knowledge management, and Information management. The author

can be contacted at [email protected].

Ivan Miloloža, Ph.D. graduated from the Faculty of Economics and Business in Zagreb

and received a Ph.D. at the Faculty of Economics in Osijek in 2015. He lived and

worked abroad in the period from 1983 to 1986 (Argentina and the Netherlands). Since

1986, he has been employed by Munja, the only Croatian battery manufacturer,

where he has performed virtually all management functions and is currently the CEO

of the Board (since 1999). He is Assistant Professor at the Department of Dental

Medicine and Health, Dean for Institutional Cooperation and Development, and

Chair of the Department of History of Medicine and Social Sciences. He has performed

many social functions in various state bodies, associations, and banks, and was a

participant and guest lecturer at numerous domestic and foreign faculties and

international conferences. The author can be contacted at [email protected].


216


Does the “Like” Habit of Social Networking

Services Lower the Psychological Barriers to

Recommendation Intention in Surveys?

Takumi Kato

Graduate School of Humanities and Social Sciences, Saitama University, Japan

Abstract

Background: Companies often measure their customers’ recommendation intention

using the loyalty index based on the idea that a customer who has high loyalty and is

committed to a brand has the confidence to recommend it to others. The

psychological barrier is higher for recommendation intention, which may influence the

behavior of others than for satisfaction on an individual level. However, the action of

recommending has become commonplace due to the spread of social networking

services (SNS). Pushing the “like” button for posts by family, friends, and co-workers has

become an ingrained practice for consumers. Therefore, it is thought that “like” habits

in SNS may lower the psychological barriers to the recommendation. Objectives: In

this study, it was hypothesized that the more people habitually like posts on SNS, the

higher the score for their recommendation intention in a customer survey.

Methods/Approach: Propensity score matching was used to investigate a causal

effect between the likes and the recommendation intention in a customer survey.

Results: Based on the results of an online survey of chocolate brands in Japan, the

causal effect was verified by propensity score matching. Conclusions: The results

suggest that not only in companies but also in academic research, a valid concern is

that the causal effect cannot be accurately evaluated unless a survey design is

performed in consideration of the effects.

Keywords: customer relationship management; loyalty; customer survey; social

networking services

JEL classification: M10, M31




Citation: Kato, T. (2021), “Does the “Like” Habit of Social Networking Services Lower the

Psychological Barriers to Recommendation Intention in Surveys?”, Business Systems



217


Introduction Customer relationship management (CRM) is a corporate activity that has long been

emphasized as important. Its purpose is to build long-term loyalty and increase profits

efficiently (Rigby et al., 2002). In other words, increasing loyalty contributes significantly

to company profits. Two main factors influence loyalty to profits. One is the increase in

the repurchase rate. It is said that the retention of existing customers as repeaters is

several times more efficient than acquiring new customers (Reichheld and Sasser,

1990). The other is for customers to act as “sales personnel” and recommend brands

to their acquaintances. Loyal customers are passionate about the brand, understand

the product well, and act as evangelists for the brand (Aaker and Joachimsthaler,

2000).

Therefore, managing customer loyalty in CRM is extremely important for

companies. Typical indicators used in marketing research to measure loyalty are

repurchase intention and recommendation intention. Repurchase intention is

frequently used as a loyalty index. However, for durable consumer goods with a long

replacement period, it is difficult for consumers to indicate this intention unless they

are aware of the next purchase (Kato, 2019). On the other hand, recommendation

intention can be indicated without being affected by the length of the replacement

period. Customers with high loyalty and commitment to the brand are confident

enough to recommend it to others (Aaker, 1991). This means that the psychological

barrier to recommendation intention to influence the behavior of others is higher than

that to satisfaction through individual emotions. Beyond that barrier, having a

recommendation intention is a strong indication of interest in the brand. Based on this

idea, recommendation intentions have been generally used in marketing surveys that

measure loyalty in both industry and research.

However, recommending has become common due to the spread of social

networking services (SNS). Consumers are in the habit of pushing the “like” button for

posts by the people they follow. Although it is thought that like habits in SNS may lower

the psychological barriers to the recommendation in a customer survey, few examples

quantitatively show this effect. Accordingly, the purpose of this study is to clarify the

effect of like habit in SNS to reduce the psychological barrier of favorable reaction of

recommendation intention in marketing research. An online survey was conducted

on Japanese chocolate brands, and the effect was evaluated by propensity score

matching. This study provides implications for the design of customer surveys that are

regularly conducted in business.

The literature review section describes previous research on the penetration status

of SNS and like motive. The methodology section describes the survey and data

analysis methods. Then, the results and discussion section describe the results and

implications for practice. Finally, the conclusion section describes the summary,

limitations, and future research tasks.

Penetration status of SNS and like motive Currently, 511,200 tweets are posted every minute on Twitter worldwide, 55,140 photos

are posted on Instagram every minute (Domo, 2019), and Facebook posts are liked

more than 3 billion times a day (Moffat, 2019). In this way, SNS is rooted in the daily life

of consumers. In the United States, 86% of people use social media at least once a

day (Herhold, 2018), and Facebook, which is the most used SNS, is used by about 70%

of people (Perrin and Anderson, 2019). The most used SNS in Japan is Line, which is

used by 82% of people (Mori and Nitto, 2020). It should be noted that Line is a means

of contact in a closed community with family and friends, so its purpose is different

218


from Facebook, Twitter, and Instagram, which are used to share information widely

and publicly. Therefore, Line was excluded from this study.

“Likes” on SNS were first introduced by the video site Vimeo in 2005 but have spread

globally, being adopted by Facebook in 2009 (Moffat, 2019). When Facebook was

first developed, it was considered an "awesome" button instead of "like." However,

there was concern that users would find the button annoying (Khrais, 2018). Ultimately,

the function to show support for family members and friends easily and make

recommendations to others has resulted in an environment that encourages

consumers to post.

Likes have a substantial impact on business, so companies attach great

importance to them (Lipsman et al., 2012; Trattner and Kappe, 2013). The like button

was installed within a few months of Facebook's appearance on more than 350,000

websites, including BestBuy.com (Gelles, 2010). A 1% increase in the number of likes of

pre-released movies was shown to increase box office revenue in the first week by

about 0.2% (Ding et al., 2017). Accumulated likes are also used in academic studies

because they are sources of information that express user preferences. Like data show

that detailed personal attributes such as ethnicity and political views can be inferred

(Kosinski et al., 2013; Youyou et al., 2015).

As mentioned above, likes can recommend content to family members and

acquaintances with one click. However, likes’ motives are not so simple. They are used

for various purposes including maintaining social ties with acquaintances and making

dating efforts (Chin et al., 2015; Eranti and Lonkila, 2015). Hence, consumers may have

a habit of blindly pushing the like button regardless of whether the content is good. As

a result, research has focused on the possibility that SNS reduces the psychological

barrier to the act of recommendation. Thus, in this study, it was hypothesized that "the

more people that habitually press like on SNS, the higher the score for their

recommendation intention in a customer survey." There are few examples of

scientifically verifying this hypothesis.

Methodology Method of survey In this study, an online survey was conducted in Japan from May 22 to 26, 2020.

Chocolate was set as the target product because it was considered a product that

people would easily recommend to others. For example, items like health insurance,

which is associated with sensitive information such as medical history, and luxury

watches, for which people often have strong preferences, would be inappropriate for

this study. Four major chocolate product brands in Japan were chosen. All brands are

by different manufacturers. There are two conditions for the respondents: they should

(a) be aged 20 to 60, and (b) have a habit of purchasing chocolate at least once a

week. The sample size was set at 5 generations x 2 genders = 10 categories, with a

total of 1,000 people, with 100 in each category. The survey was distributed by email

until the target sample size was achieved. The number of emails sent was 72,743, the

number of participants was 9,012, the number of respondents who answered

completely was 7,134, and the number of those who met the respondent conditions

was 1,325. Then, the sample of 1,000 people was randomly selected so that

generation and gender were even. This sample was used for verification.

The online survey method has been criticized for a minimization of efforts called

"satisficing" that reduces the reliability of answers (Krosnick, 1991). Therefore, reducing

the number of questions in the survey is effective in increasing reliability and lessening

the burden on the respondents. Based on this, the following 10 items were asked

219


about: (1) gender, (2) age, (3) living area, (4) annual household income, (5) frequency

of SNS usage (Facebook, Twitter, Instagram, TikTok), (6) average monthly frequency of

likes on others’ posts via the four SNS, (7) average monthly frequency of posting on the

four SNS, (8) frequency of chocolate purchase, (9) most purchased brands (four

brands were presented as options), and (10) recommendation intention for the most

purchased brand.

The target of this study is the relationship between likes and recommendation

intention, but the relationship between posting and recommendation intention was

also confirmed. It is ideal to use the data tracking usage status to accurately grasp

the frequency of likes and posts on SNS. However, this method has a high risk of privacy

infringement. Therefore, in (6) and (7), the method of asking participants to share the

average number of times they used the like function or posted per month was

selected. Considering how limited memory can be, it was difficult for consumers to

answer the number of likes individually for each SNS. Therefore, the total number of

likes given in the four chosen SNS (Facebook, Twitter, Instagram, and TikTok) were

collected.

As shown in Table 1, the SNS with the highest usage rate is Twitter (60.4%), followed

by Facebook (53.9%), Instagram (52.3%), and TikTok (28.2%).

Table 1

Frequency of SNS usage Frequency of use Facebook Twitter Instagram TikTok Total

Freq Ratio Freq Ratio Freq Ratio Freq Ratio Freq Ratio

Every day 237 23.7% 337 33.7% 281 28.1% 118 11.8% 973 24.3%

Four to six times

a week

81 8.1% 83 8.3% 78 7.8% 68 6.8% 310 7.8%

Two to three

times a week

81 8.1% 90 9.0% 85 8.5% 47 4.7% 303 7.6%

Less than once

a week

140 14.0% 94 9.4% 79 7.9% 49 4.9% 362 9.1%

Never use 461 46.1% 396 39.6% 477 47.7% 718 71.8% 2,052 51.3%

Total 1,000 100.0% 1,000 100.0% 1,000 100.0% 1,000 100.0% 4,000 100.0%

Utilization ratio − 53.9% − 60.4% − 52.3% − 28.2% − 48.7%


As shown in Table 2, the usage rate of Twitter by generation shows that the usage

rate is higher among younger participants.

Table 2

Frequency of Twitter usage by generation Frequency of use The 20s The 30s The 40s The 50s The 60s

Freq Ratio Freq Ratio Freq Ratio Freq Ratio Freq Ratio

Every day 122 61.0% 86 43.0% 67 33.5% 41 20.5% 21 10.5%

Four to six times

a week

28 14.0% 18 9.0% 10 5.0% 15 7.5% 12 6.0%

Two to three

times a week

9 4.5% 22 11.0% 28 14.0% 17 8.5% 14 7.0%

Less than once

a week

12 6.0% 16 8.0% 19 9.5% 25 12.5% 22 11.0%

Never use 29 14.5% 58 29.0% 76 38.0% 102 51.0% 131 65.5%

Total 200 100.0% 200 100.0% 200 100.0% 200 100.0% 200 100.0%

Utilization ratio − 85.5% − 71.0% − 62.0% − 49.0% − 34.5%


220


Table 3 shows that the average number of likes per month is 10.33 It is noteworthy

that 452 respondents out of 1,000 selected 0 times. For posting, the mean is 3.62 times,

which is smaller than the likes, and the number of respondents who answered 0 times

reached 556, which is the majority. These people are referred to as “read-only

members (ROM).”

Here, it is necessary to define groups based on “liking and posting” frequency.

Based on the observed distribution, Group 1 is defined as 0 times (ROM), Group 2 as

1–5 times, Group 3 as 6–30 times, and Group 4 as 31 times or more and who like posts

once or more every day. Thus, it was verified that the tendency of responding to the

recommendation intention was significantly higher in the treatment groups, Group 2–

4 (people who have liking/posting habits), than in the control group, Group 1 (people

who do not have liking/posting habits).

The recommendation intention in (10) was: “How much would you recommend the

brand that you selected as your most purchased to friends and acquaintances?” The

answer options ranged from "1: Not recommend at all" to "10: Highly recommend." The

mean value was 7.097 and the standard deviation was 1.751.

Table 3

Likes and Posts per month on SNS Descriptive statistics Group Total

Mean Median Min Max Group 1

(0)

Group 2

(1-5)

Group 3

(6-30)

Group 4

(31-)

Like 10.326 1 0 500 452 285 196 67 1000

Post 3.62 0 0 200 556 317 105 22 1000


Method of verification When the random assignment is possible, it is appropriate to perform the most reliable

method of randomized controlled trial on a scientific basis (Torgerson and Torgerson,

2008). However, it is difficult to instruct a randomly selected person to communicate

a certain period of their performance by specifying the frequency of likes on SNS. In

addition, it is more appropriate, for this study, to gain information about a habit rather

than forcibly encouraging participants to press the like button.

A propensity score proposed by Rosenbaum and Rubin (1983) is a typical method

for estimating the causal effect when random assignment is difficult. Covariates are

adjusted by aggregating multiple covariates into one variable called the propensity

score. The characteristics of consumers tend to be biased between those who have

a large number of likes on SNS and those who do not. Therefore, the causal effect is

estimated by matching respondents with close propensity scores to each other and

homogenizing both groups.

Since the true value of the propensity score of each respondent is unknown, it is

common to estimate it from the data using a logistic regression model. As shown in

Table 4, the acquired attribute variables were made into dummy variables and put

into the explanatory variables of the model. The dummy variable criterion is not used

in the model. The objective variable was 0/1, indicating either the control group or the

treatment group. That is, the propensity score represents the probability that each

respondent belongs to the treatment group. Since there are many explanatory

variables, the stepwise method was used to select variables. In this study, since

multiple comparisons were performed, the propensity score was estimated for each

of the two target groups: for likes, Group 1 vs. Group 2 (Test 1), vs. Group 3 (Test 2), vs.

Group 4 (Test 3), and for posts, vs. Group 2’ (Test 1'), vs. Group 3’ (Test 2'), vs. Group 4’

(Test 3').

221


Table 4

Attribute variable used for propensity score matching

No Variable Breakdown Mean SD

Gender

1 Female Female 0.500 0.500

Age

2 Age_20s 20s 0.200 0.400

3 Age_30s 30s 0.200 0.400

4 Age_40s 40s 0.200 0.400

5 Age_50s 50s 0.200 0.400

6 Age_60s 60s 0.200 0.400

Residential area

7 Area_01_Hokkaido Hokkaido 0.043 0.203

8 Area_02_Tohoku Tohoku 0.101 0.301

9 Area_03_Kanto Kanto 0.458 0.498

10 Area_04_Tokai Chubu 0.100 0.300

11 Area_05_Kinki Kansai 0.143 0.350

12 Area_06_Chugoku Chugoku 0.041 0.198

13 Area_07_Shikoku Shikoku 0.019 0.137

14 Area_08_Kyusyu Kyusyu 0.095 0.293

Household income

15 Income_199 Less than two million yen 0.073 0.260

16 Income_200_399 Two to four million yen 0.214 0.410

17 Income_400_599 Four to six million yen 0.223 0.416

18 Income_600_799 Six to eight million yen 0.181 0.385

19 Income_800_999 Eight to 10 million yen 0.126 0.332

20 Income_1000_1499 10 million to 15 million yen 0.117 0.322

21 Income_1500 15 million yen or more 0.066 0.248

Most purchased brand

22 Brand_A Brand A 0.331 0.344

23 Brand_B Brand B 0.257 0.232

24 Brand_C Brand C 0.142 0.349

25 Brand_D Brand D 0.270 0.255


The difference in the distribution of recommendation intention was verified by

Fisher's exact test for the two groups matched using the propensity score. The reason

the chi-square test was not applied is that there are numbers less than 10 in the cell.

The null hypothesis is that “there is no difference in the distribution of recommendation

intention between the two groups.” The null hypothesis is rejected and a significant

difference is confirmed when the p-value becomes smaller than 0.05.

To ensure that the test is rigorous, the following two procedures were performed.

First (see Table 5), the recommendation intention was converted from 10 levels to 4

levels; Low: 1–2, Lower-middle: 3–5, Upper-middle: 6–8, and High: 9–10. This is because

if the distribution is made finer than necessary, even slight differences that have

essentially no meaning are detected. Second, the sample size was adjusted

appropriately. A sample size that is too large will detect even meaningless differences.

Therefore, the appropriate sample size was confirmed by power analysis. It was

calculated that 121.13 by significance level was 5%, the power of the test was 80%,

the effect size was medium at 0.3 (Cohen, 1992), and the degree of freedom was 3.

Therefore, 60 people in each group, 120 people in total, were randomly sampled for

each test.

222


Table 5

Distribution of recommendation intentions

Recommendation

intention

Number of

people

Low (1-2) 13

Lower-middle (3-5) 164

Upper-middle (6-8) 626

High (9-10) 197

Total 1000


Statistical analysis software R was used. The stepwise method was the stepAIC

function of the MASS package, the propensity score matching was done using the

Match function of the Matching package, and the power analysis was conducted

through the pwr.chisq.test function of the pwr package.

Results and discussion First, the propensity score was calculated by the logistic regression model. Table 6

shows the results of Test 1–Test 3 for likes. The odds ratio of Age_40s, Age_50s, and

Age_60s was well below 1 in all models. This means that if consumers are in their 40s or

older, fewer people commonly press the like button. From the SNS usage rate

according to age in Table 2, it can be seen that the relationship between SNS and

age is strong. The validity of the model was confirmed because the c-statistics were

0.7 or more in all models. The same was done for Test 1'–Test 3' for posts.

Table 6

Estimated result of logistic regression model Variable Test1 (Group 1 vs Group 2) Test1 (Group 1 vs Group 3) Test3 (Group 1 vs Group 4)

Odds

ratio

SE p-value Odds

ratio

SE p-value Odds

ratio

SE p-value

(Intercept) 2.822 0.234 0.000 *** 2.884 0.247 0.000 *** 0.399 0.197 0.000 ***

Female 0.578 0.165 0.001 ** 0.628 0.190 0.014 *

Age_30s 0.460 0.274 0.005 ** 0.307 0.299 0.000 ***

Age_40s 0.232 0.277 0.000 *** 0.241 0.286 0.000 *** 0.227 0.378 0.000 ***

Age_50s 0.110 0.288 0.000 *** 0.096 0.307 0.000 *** 0.138 0.399 0.000 ***

Age_60s 0.172 0.272 0.000 *** 0.066 0.336 0.000 *** 0.049 0.617 0.000 ***

Area_01_Hokkaido

0.190 0.804 0.039 *

Area_02_Tohoku 0.463 0.353 0.029 *

Income_1000_1499

1.921 0.388 0.092

Income_1500 3.529 0.475 0.008 **

c-statistics 0.723 0.755 0.785

Note: ***p<0.001; **p<0.01; *p<0.05. SE: standard error


Using the estimated propensity score, the control group and the treatment group

were homogenized. As shown on the left side of Figure 1, the distribution of propensity

scores of Test 1 can be understood to be completely different. As a result of propensity

score matching, 224 respondents were extracted from each group, and as shown on

the right side of Figure 1, the propensity scores are homogenized. As shown in Table 7,

even when compared for each variable, the values are almost the same. Similarly,

163 respondents in each group were extracted in Test 2, 64 respondents in Test 3, 273

respondents in Test 1', and 105 respondents in Test 2'. However, Test 3' was excluded

from verification because the test only had 21 respondents, below the standard of 60

223


participants. As shown in Table 3, the reason for this is that posting is less frequent than

liking.

Figure 1

Distribution of propensity scores for Test 1(left: before matching, right: after matching)


Table 7

Results of propensity score matching Variable Test1 Test2 Test3

Group 1 Group 2 SMD Group 1 Group 3 SMD Group 1 Group 4 SMD

Female 0.433 0.433 0.000 0.436 0.454 0.037 0.578 0.531 0.094

Age_30s 0.263 0.263 0.000 0.270 0.252 0.042 0.406 0.328 0.161

Age_40s 0.219 0.219 0.000 0.258 0.258 0.000 0.172 0.172 0.000

Age_50s 0.152 0.152 0.000 0.166 0.166 0.000 0.141 0.141 0.000

Age_60s 0.219 0.219 0.000 0.110 0.110 0.000 0.047 0.047 0.000

Area_01_Hokkaido 0.040 0.094 0.215 0.012 0.012 0.000 0.000 0.000 0.000

Area_02_Tohoku 0.125 0.112 0.041 0.055 0.074 0.075 0.125 0.094 0.099

Area_04_Tokai 0.125 0.103 0.070 0.117 0.061 0.194 0.141 0.109 0.094

Area_05_Kinki 0.134 0.138 0.013 0.110 0.147 0.110 0.141 0.203 0.165

Area_06_Chugoku 0.022 0.067 0.217 0.018 0.037 0.112 0.031 0.031 0.000

Area_07_Shikoku 0.000 0.022 0.213 0.006 0.061 0.308 0.000 0.000 0.000

Area_08_Kyusyu 0.103 0.080 0.077 0.153 0.086 0.208 0.109 0.078 0.107

Income_200_399 0.228 0.183 0.110 0.184 0.147 0.099 0.188 0.203 0.039

Income_400_599 0.228 0.250 0.052 0.270 0.233 0.085 0.203 0.156 0.121

Income_600_799 0.165 0.165 0.000 0.141 0.215 0.193 0.078 0.219 0.400

Income_800_999 0.156 0.138 0.050 0.129 0.129 0.000 0.109 0.094 0.051

Income_1000_1499 0.116 0.116 0.000 0.135 0.092 0.135 0.188 0.188 0.000

Income_1500 0.058 0.067 0.037 0.074 0.080 0.023 0.094 0.094 0.000

Brand_B 0.041 0.009 0.204 0.043 0.074 0.131 0.047 0.063 0.068

Brand_C 0.136 0.177 0.112 0.135 0.160 0.069 0.156 0.172 0.042

Brand_D 0.059 0.036 0.107 0.049 0.080 0.125 0.063 0.063 0.000

Sample size 224 224 − 163 163 − 64 64 −


As shown in Table 8, as a result of applying Fisher’s exact test in Test 1 (p-

value=0.270), the null hypothesis was not rejected. However, in Tests 2 and 3, the p-

value was smaller than 0.05, and a significant difference was confirmed. Cramer's V

represents the effect size and is generally judged as 0.1=small, 0.3=medium, and

224


0.5=large (Khalilzadeh and Tasci, 2017; Grant et al., 2012). Since both Tests 2 and 3 are

0.3 or more, the effect size is medium. Next, looking at the post results shown in Table

9, as with the likes, there was no significant difference in Test 1′, but in Test 2′, a

significant difference in effect size similar to that for likes was confirmed.

Table 8

The effect of the frequency of Likes on the response to recommendation intention Test Like Recommendation intention Total p-value Cramer's Va

Low Lower-

middle

Upper-

middle

High

1 Group 1 1 16 35 8 60 0.270 0.177

Group 2 0 12 33 15 60

2 Group 1 2 14 40 4 60 0.008** 0.302

Group 3 1 5 39 15 60

3 Group 1 0 11 45 4 60 0.001*** 0.335

Group 4 0 9 31 20 60

Note: ***p<0.001; **p<0.01; *p<0.05. a effect size 0.1=small, 0.3=medium, 0.5=large


Table 9

The effect of the frequency of Posts on the response to recommendation intention Test Post Recommendation intention Total p-value Cramer's Va

Low Lower-

middle

Upper-

middle

High

1' Group 1 1 10 41 8 60 0.166 0.196

Group 2 0 13 32 15 60

2' Group 1 2 17 33 8 60 0.000*** 0.407

Group 3 0 2 37 21 60

Note: ***p<0.001; **p<0.01; *p<0.05. a effect size 0.1=small, 0.3=medium, 0.5=large


large

From the above, the hypothesis "the more people that habitually press like on SNS,

the higher the score for their recommendation intention in a customer survey" was

supported. As expected, consumers with stronger like habits are more likely to have

lower psychological barriers to responding positively to the survey's recommendation

intention. In addition, the conditions under which this effect occurred were also

clarified. Here, there was no significant difference in the group with a mean monthly

rate of 1–5 times compared to those who do not use the like function. Similar results

were obtained for posts. This is an expected result because there is a correlation

between liking and posting. The correlation coefficient based on the data of 1,000

people was 0.686, and the result of Pearson's product-moment correlation was p-

value=0.000, confirming a correlation between both variables.

One of the most important features in a regular customer survey in a company is to

survey the same item at the same time for the same sample. However, since it is

difficult to continue surveying the same individuals, a homogeneous sample is

extracted each time. If there are variations in these values, biases will affect the data,

and it will not be possible to obtain true values. Therefore, in companies, this is

managed as a matter of course. However, the conditions of the survey respondents

are often limited regarding age, gender, and residential area. There are few cases

where the usage status of SNS is included. If the usage status of SNS is different for each

survey sample, there is a concern that the recommendation intention increases due

to the effects of SNS habits rather than because of products and services. As

225


technology progresses rapidly and the usage of SNS changes from moment to

moment, this should be taken into consideration when designing a survey.

Conclusion In managing loyalty, companies regularly measure recommendation intention.

Customers with high loyalty and commitment to the brand are confident enough to

recommend it to others. In other words, the psychological barrier is higher for

recommendation intention, which may influence the behavior of others, than for

emotions of satisfaction experienced by the individual. However, the action of

recommendation has become commonplace due to the popularity of SNS. It has

become a daily practice for consumers to like posts by family members and

acquaintances. Therefore, it was thought that like habits in SNS may lower the

psychological barrier of recommendation intention.

Thus, in this study, the hypothesis that "the more people that habitually like posts on

SNS, the higher their score for recommendation intention in a customer survey" was

verified by using propensity score matching for the data observed in the online survey.

As a result, a significant difference was confirmed between the group that does not

habitually “like” posts and the group that does. Thus, the hypothesis was supported for

chocolate brands in Japan. According to Cramer's V criteria, the effect size is

medium. Further, it is considered undesirable to ignore this effect. However, it was also

revealed that there was no significant difference between those who used the like

function on average 1–5 times monthly compared to those who did not use like at all.

In a company's CRM activities, the results of regular customer surveys are used as

material for decision-making. At that time, if the respondents’ usage of SNS fluctuates,

the result may be biased, and there is a concern that decision-making may be

mistaken. Since the recommendation intention is frequently used in academic

research as well, it may be necessary to consider this effect for precise verification.

A limitation of this study is that the number of likes/posts used in the data is based

on a self-reported attitude survey rather than recorded behavior. If a behavioral

record could be used without violating the participants’ privacy, not only could

detailed conditions of the effect occurrence be understood, but the effect of each

SNS could be verified. In addition, this study did not consider the motives for likes. By

considering motives in addition to the number of times, it may be possible to

understand the conditions that influence the response of the recommendation

intention. This is something that should be investigated in future studies.

References 1. Aaker, D.A. (1991), Managing Brand Equity. New York, The Free Press.

2. Aaker, D.A., Joachimsthaler, E. (2000), Brand Leadership. New York, The Free Press.

3. Chin, C.Y., Lu, H.P., Wu, C.M. (2015), “Facebook users' motivation for clicking the “like”

button”, Social Behavior and Personality: An International Journal, Vol. 43, No. 4, pp. 579–

592.

4. Cohen, J. (1992), “A power primer”, Psychological Bulletin, Vol. 112, No. 1, pp. 155–159.

5. Ding, C., Cheng, H.K., Duan, Y., Jin, Y. (2017), “The power of the “like” button: The impact

of social media on box office”, Decision Support Systems, Vol. 94, pp. 77–84.

6. Domo (2019), Data never sleeps 7.0. Available at https://www.domo.com/learn/data-

never-sleeps-7 / (15 November 2020).

7. Eranti, V., Lonkila, M. (2015), “The social significance of the Facebook Like button”, First

Monday, Vol. 20, No. 6, pp. 1–14.

8. Gelles, D. (2010), E-commerce takes instant liking to Facebook button. Financial Times, 21

September. Available at https://www.ft.com/content/1599be2e-c5a9-11df-ab48-

00144feab49a / (accessed 15 November 2020).

226


9. Grant, J.E., Chamberlain, S.R., Schreiber, L.R., Odlaug, B.L. (2012), “Gender-related clinical

and neurocognitive differences in individuals seeking treatment for pathological

gambling“, Journal of Psychiatric Research, Vol. 46, No.9, pp. 1206–1211.

10. Herhold, K. (2018), “How people use social media in 2018.“, Available at

https://themanifest.com/social-media/how-people-use-social-media-2018 (accessed 15

November 2020).

11. Kato, T. (2019), “Loyalty management in durable consumer goods: Trends in the influence

of recommendation intention on repurchase intention by time after purchase“, Journal of

Marketing Analytics, Vol. 7, No. 2, pp. 76–83.

12. Khalilzadeh, J., Tasci, A.D. (2017), “Large sample size, significance level, and the effect size:

Solutions to perils of using big data for academic research“, Tourism Management, Vol. 62,

pp. 89–96.

13. Khrais, R. (2018), “The most powerful tool in social media.“, Available at

https://www.marketplace.org/2018/04/11/it-was-known-button-it-was-awesome-button/

(accessed 15 November 2020).

14. Kosinski, M., Stillwell, D., Graepel, T. (2013), “Private traits and attributes are predictable from

digital records of human behavior“, Proceedings of the National Academy of Sciences,

Vol. 110, No. 15, pp. 5802–5805.

15. Krosnick, J.A. (1991), “Response strategies for coping with the cognitive demands of

attitude measures in surveys“, Applied Cognitive Psychology, Vol. 5, No. 3, Vol., pp. 213–

236.

16. Lipsman, A., Mudd, G., Rich, M., Bruich, S. (2012), “The power of “like”: How brands reach

(and influence), fans through social-media marketing“, Journal of Advertising Research,

Vol. 52, No. 1, pp. 40–52.

17. Moffat, B. (2019), “The power of likes on social media: Friend or foe? “. Available at

https://www.the-future-of-commerce.com/2019/10/07/the-power-of-likes-on-social-

media/ (accessed 15 November 2020).

18. Mori, T., Nitto, H. (2020), “Personal value of social networking services”, Available at

https://www.nri.com/en/knowledge/report/lst/2020/scs/digital/0129 / (accessed 15

November 2020).

19. Perrin, A., Anderson, M. (2019), “Share of U.S. adults using social media, including Facebook,

is mostly unchanged since 2018“, Available at https://www.pewresearch.org/fact-

tank/2019/04/10/share-of-u-s-adults-using-social-media-including-facebook-is-mostly-

unchanged-since-2018/ (accessed 15 November 2020).

20. Reichheld, F.F., Sasser, W.E. (1990), “Zero defections: Quality comes to services“, Harvard

Business Review, Vol. 68, No. 5, pp. 105–111.

21. Rigby, D.K., Reichheld, F.F., Schefter, P. (2002), “Avoid the four perils of CRM“, Harvard

Business Review, Vol. 80, No. 2, pp. 101–109.

22. Rosenbaum, P.R., Rubin, D.B. (1983), “The central role of the propensity score in

observational studies for causal effects“, Biometrika, Vol. 70, No. 1, pp. 41–55.

23. Torgerson, D., Torgerson, C. (2008), Designing randomised trials in health, education and

the social sciences: An introduction. New York, Palgrave Macmillan.

24. Trattner, C., Kappe, F. (2013), “Social stream marketing on Facebook: A case study“,

International Journal of Social and Humanistic Computing, Vol. 2, No. 1–2, pp. 86–103.

25. Youyou, W., Kosinski, M., Stillwell, D. (2015), “Computer-based personality judgments are

more accurate than those made by humans“, Proceedings of the National Academy of

Sciences, Vol. 112, No. 4, pp. 1036–1040.

https://www.nri.com/en/knowledge/report/lst/2020/scs/digital/0129%20/

227


About the author

Takumi Kato is currently an Assistant Professor at the Graduate School of Humanities

and Social Sciences, Saitama University, Japan. He obtained his Ph.D. in Business

Administration and Master of Business Administration from Graduate School of Business

Sciences from University of Tsukuba, and Bachelor of Science Degree from Keio

University, Tokyo, Japan. He joined Mitsubishi Electric Corporation in 2012. In 2014, he

joined Honda Motor Co Ltd. and was a chief analyst of the Business Analytics Division.

His role was product planning and brand management. In 2015, he joined Saitama

University. His research interests include marketing, marketing research, consumer

behavior, and brand management. The author can be contacted at

[email protected]

228


The Proportion for Splitting Data into Training

and Test Set for the Bootstrap in

Classification Problems

Borislava Vrigazova

Sofia University, Faculty of Economics and Business Administration, Bulgaria

Abstract

Background: The bootstrap can be alternative to cross-validation as a training/test set

splitting method since it minimizes the computing time in classification problems in

comparison to the tenfold cross-validation. Objectives: Тhis research investigates what

proportion should be used to split the dataset into the training and the testing set so

that the bootstrap might be competitive in terms of accuracy to other resampling

methods. Methods/Approach: Different train/test split proportions are used with the

following resampling methods: the bootstrap, the leave-one-out cross-validation, the

tenfold cross-validation, and the random repeated train/test split to test their

performance on several classification methods. The classification methods used

include the logistic regression, the decision tree, and the k-nearest neighbours. Results:

The findings suggest that using a different structure of the test set (e.g. 30/70, 20/80)

can further optimize the performance of the bootstrap when applied to the logistic

regression and the decision tree. For the k-nearest neighbour, the tenfold cross-

validation with a 70/30 train/test splitting ratio is recommended. Conclusions:

Depending on the characteristics and the preliminary transformations of the variables,

the bootstrap can improve the accuracy of the classification problem.

Keywords: the bootstrap; classification; cross-validation; repeated train/test splitting

JEL classification: C38, C52, C55


Received: 12 Aug 2020


Citation: Vrigazova, B. (2021), “The Proportion for Splitting Data into Training and Test

Set for the Bootstrap in Classification Problems”, Business Systems Research, Vol. 12,

No. 1, pp. 228-242.


Acknowledgments: The author expresses her gratitude to prof. Ivan Ivanov for his

recommendations. Some of the results presented in this paper were also presented at

the ENTRENOVA 2020 conference. This research contains extended experiments and

results. The author expresses her gratitude to the reviewers for their valuable

comments.

229


Introduction Long computational time is a problem that often occurs in big datasets. Slow

computation can occur due to many reasons. On the one hand, computationally

exhaustive methods like the mixed linear integer approach can be used with

classification methods (Maldonado et. al., 2014). On the other hand, the input data

may be used in their original version and the differences among their units can slow

down the computation. A third reason can be the presence of too many independent

variables. To avoid those problems and reduce computational time in classification,

some authors suggest improved versions of existing computationally exhaustive

methods for classification (Maldonado et. al., 2014), standardization of independent

variables to unify the input variables (James et. al.), and variable selection (Velliangiri

et. al., 2019). These approaches can reduce the time for splitting the dataset into

training and test set to evaluate the performance of the classification model.

Some evidence suggests that computing time in machine learning algorithms for

classification can also depend on the resampling method used for splitting the data

into training and test set (James et. al., 2013). For instance, the leave-one-out cross-

validation produces the training and test sets slower than the tenfold cross-validation,

and the prediction time increases (James et. al., 2013). This paper shows that the

tenfold bootstrap procedure introduced in (Vrigazova and Ivanov, 2020a) can

decrease the overall time for prediction in classification problems. The paper

compares the behaviour of the tenfold bootstrap to other resampling methods like

the tenfold cross-validation (James et. al., 2013), the leave-one-out (LOO) cross-

validation (James et. al., 2013), and the repeated random train/test split procedure

available in Python (Pedregosa et. al., 2011). They are applied to several classification

methods like the logistic regression, decision tree classifier, and the k-nearest

neighbours. The aims of this paper are first to check if the tenfold bootstrap has the

computational advantage as a training/test splitting method in classification

methods. Similar research was conducted for the Support Vector Machines, so this

paper can be considered as an extension of (Vrigazova and Ivanov, 2020b).

Secondly, to propose train/test split proportion for the bootstrap procedure to reduce

computational time and preserve high accuracy of the classification model.

The next section reviews existing academic literature, section 3 presents the

methodology, and sections 4 and 5 comment on the results and discuss the

advantages and disadvantages of the proposed methodology. Section 6 concludes.

Literature review The bootstrap was first introduced in 1979 by Efron (Efron, 1979). It has wide

applications in various fields. For example, it can be used for inferring the unknown

distribution of data, thus allowing confidence intervals to be built. One thousand

iterations of the bootstrap can make data’s distribution closer to the Gaussian

distribution. As a result, the bootstrap is widely used in Monte Carlo simulations

MacKinnon (2002). The bootstrap is also used in the random forest classifier and for

pruning decision trees (Breiman, 1996). In 1992, Breiman (1992) devised the little

bootstrap procedure for applications as a resampling method in small datasets. Later,

in 1995, he showed that the little bootstrap procedure can be used as a resampling

method in data with fixed regressors (Breiman, 1995). He recommended cross-

validation as a resampling technique in datasets with random regressors. In 2018

Vrigazova (2018) showed that the little bootstrap procedure (Breiman, 1992) can

successfully be used for feature selection in panel data with fixed effects.

230


The bootstrap procedure has widely been used for estimating unknown

distributions. Its properties as a resampling method have started to be more thoroughly

researched lately. In 1997, Efron and Tibshirani (Efron et. al., 1997) tested the

performance of the 0.632 + bootstrap procedure in machine learning methods for

classification (k-nearest neighbour, logistic regression, and decision tree) suggesting

that the bootstrap can be an alternative to cross-validation. Since then few

experiments have been made in this direction. The standard resampling procedure

for splitting the dataset into training and test set in classification problems has been

cross-validation. Repeated random training/test split is also used as an alternative to

cross-validation.

Based on the research of Efron and Tibshirani (Efron et. al., 1997), the question of

the bootstrap procedure can be used as a technique for splitting into training and test

set and be a reliable alternative to cross-validation has been raised. Recent research

(Vrigazova and Ivanov, 2020a and b) has shown that the bootstrap procedure can

be a reliable resampling procedure in the logistic regression, decision tree, k-nearest

neighbour, and the support vector machines when using 70/30 proportion for train/test

split. However, more experiments need to be conducted to conclude whether

bootstrap is an appropriate training/test set splitting technique for various types of

datasets. Also, it is subject to further experiments whether the 70/30 training/set

proportion is appropriate in most cases to preserve high accuracy. This paper aims to

fill these gaps in the academic literature.

Methodology This research compares the performance of the decision tree classifier (James et. al.,

2013), the logistic regression, and the k-nearest neighbour (Pampel, 2000) in terms of

time, accuracy, and error rate. Logistic regression (Pampel, 2002) is a method for

binary or multiclass classification based on the probability that one observation

belongs to a particular class. It is relatively easier for interpretation than the decision

tree classifier and the k-nearest neighbour. The decision tree classifier (James et.al.,

2013) is not a computationally expensive method but it provides the predicted classes

as a tree with possible outcomes leading to each class. Each branch of the tree is a

particular variable. Therefore, it may be harder for interpretation particularly in the

case of multiclass classification. Unlike the logistic regression and the decision tree

classifier, the k-nearest neighbour splits the observations into classes based on how

close they are to one another. It assumes that observations belonging to the same

class will be close to one another. Typically, the three classification methods use

tenfold cross-validation to split the dataset into training and test set and make

predictions (James et. al., 2013). This paper investigates whether the tenfold bootstrap

can be used instead of the tenfold cross-validation to split the dataset into training

and test set so that the time for the prediction can be reduced.

To perform the research three fully available datasets were used. These are the

Monica1, the Food2 , and the Adult3 datasets. The Monica dataset is the smallest one,

containing 6,367 observations and 11 independent variables. The dependent variable

is called ‘outcome”. The Food dataset contains 23,971 observations and 5

independent variables, with the ‘sex’ variable being the dependent one. The last

dataset is the Adult dataset with 45,222 observations and 11 independent variables.

The dependent variable is ‘income’. All datasets are increasing in size so that the

1 Available at https://www.kaggle.com/ukveteran/who-monica-data/tasks

2 Available at https://vincentarelbundock.github.io/Rdatasets/doc/Ecdat/BudgetFood.html

3 Available at https://archive.ics.uci.edu/ml/datasets/adult

https://www.kaggle.com/ukveteran/who-monica-data/tasks%20/

https://vincentarelbundock.github.io/Rdatasets/doc/Ecdat/BudgetFood.html

https://archive.ics.uci.edu/ml/datasets/adult

231


performance of the resampling methods in large datasets can be observed. The

author did not apply preliminary transformations to the input variables.

All experiments were conducted in Python 3.6 using a computer with a processor

Intel Core i7, 2.80 GHz., Windows 10. Time is measured in seconds, while accuracy and

error rate are shown in equations 1 and 2.

𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑜𝑟𝑟𝑒𝑐𝑡 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑖𝑜𝑛𝑠

𝑇𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑖𝑜𝑛𝑠 (1)

𝐸𝑟𝑟𝑜𝑟 𝑟𝑎𝑡𝑒 = 1 − 𝑎𝑐𝑐𝑢𝑟𝑎𝑐𝑦 (2)

The first type of experiment is to split each dataset into training and test set using

the tenfold cross-validation (Hoerl et. al., 1997) and 70/30, 50/50, 30/70, and 20/80 as

train/test split proportions. The author then fitted each classification method and

calculated time, accuracy, and error rate. The author used the Python 3.6 function

model_selection.cross_val_score() with the parameter cv fixed to 10 to perform the

tenfold cross-validation.

The leave-one-out (LOO) cross-validation was also used (Wong, 2015) as an

alternative to tenfold cross-validation. The author used the same train/test split

proportions as in the tenfold cross-validation. To run the leave-one-out (LOO) cross-

validation, the function model_selection.LeavePOut(p=1) in Python was used with the

parameter p set to 1. Then the leave-one-out cross-validation was applied to the three

classification methods.

As a third resampling alternative, the repeated random train/test split (Krstajic et.

al., 2015) was applied to the logistic regression, decision tree classifier, and the k-

nearest neighbour. The function ShuffleSplit() can be used to randomly and

repeatedly divide the dataset into training and test set. The author fixed the

parameter n_splits to 10 and the random_state parameter to 7 to be able to replicate

the results.

The author also ran the tenfold bootstrap (Vrigazova and Ivanov, 2019) procedure

as an alternative to the three resampling methods. The bootstrap procedure for

classification problems that this research follows was introduced in (Vrigazova and

Ivanov, 2019). This paper shows that for some datasets the standard splitting

proportion of 70/30 is not enough to optimize the performance of the bootstrap

procedure. Other proportions may preserve accuracy, while further reduce

computational time. Figure 1 summarizes the standard approach and the novel

approach in this study.

232


Figure 1

Standard vs proposed resampling methods

Source: Author’s presentation

To compare the performance of each model, the author uses time, accuracy, and

error rate. The next section presents the results.

Results Logistic regression Table 1 presents the results from the resampling methods applied to the logistic

regression.

Table 1 shows that the slowest resampling method is the leave-one-out cross-

validation (LOO). Regardless of the size of the dataset and the splitting proportion, the

leave-one-out cross-validation was between 18 and 6440 times slower than the rest of

the resampling methods. Despite this, it produced an accuracy and error rate similar

to the tenfold cross-validation. Its computational disadvantage makes it rarely used in

large datasets. The tenfold cross-validation is faster than the leave-one-out cross-

validation but slower than the random train/test split and the tenfold bootstrap.

The tenfold bootstrap proved to be the fastest resampling method for the logistic

regression. Its computational advantage was significant. For instance, the adult

dataset (70/30) was classified by the LOO in 6440 seconds, while the bootstrap did

that in 0.23 seconds. The tenfold cross-validation led to the output from the logistic

regression in 1.78 seconds, while the random train/test split produced results similar to

the bootstrap. The two produced an accuracy of 79.1%, while the cross-validation –

79.8%. However, the accuracy of the bootstrap is stable regardless of the splitting

proportion, similarly to the random train/test split. Unlike them, the tenfold cross-

validation’s accuracy fell from 79.8% to 79.2%. So, possible overfitting can be present

in the cross-validation.

Standard Classification Methods

Training/ test set division via tenfold cross-validation, leave-one-out cross-validation and repeated random train/test

splitting

Fitting logistic regression/decision tree classifier using

using 70/30 train/test split proportion

Prediction and evaluation of each classification model's performance chosen in the

previous step

Result: high accuracy, no overfitting

Proposed modifications

New: Training/test set split using tenfold bootstrap

Fitting logistic regression/decision tree

classifier using 70/30 train/test split proportion

Prediction and evaluation of each classification model's performance chosen in the

previous step

Result: no overfitting, similar accuracy and accelerated time compared to standard methods

233


Table 1

Logistic regression results

Dataset Train/test ratio Resampling method Accuracy Error rate Time (s)

Monica 70/30 10-fold cross-validation 87.8 12.2 1.84

LOO 87.8 12.2 105.56

Random train/test split 87.9 12.1 0.05

10-fold bootstrap 87.8 12.2 0.02

50/50 10-fold cross-validation 87.7 12.3 0.14

LOO 87.7 12.3 44.70




LOO 87.9 12.1 18.32




LOO 88.0 12.0 7.68



Food 70/30 10-fold cross-validation 86.2 13.8 0.83

LOO 86.2 13.8 306.52




LOO 86.2 13.8 145.48




LOO 86.3 13.7 55.77




LOO 86.1 13.9 28.24



Adult 70/30 10-fold cross-validation 79.8 20.2 1.78

LOO 79.7 20.3 6440.27




LOO 79.7 20.3 3029.14




LOO 79.6 20.4 659.80




LOO 79.2 20.8 273.80



Source: Author’s calculations

The accuracy did not change so drastically with reducing the training set. All

resampling methods provided an error rate between 13.6% and 14%. The bootstrap

234


resulted in the highest accuracy of 86.1% (70/30), while the tenfold cross-validation –

86.2% (70/30). The random train/test split resulted in an accuracy of 86.4% (70/30).

However, when the train/test random split was applied with a 50/50 splitting

proportion, its accuracy dropped to 85.8%. The 30/70 proportion led to increased

accuracy (86.3%) resulting from the tenfold cross-validation. Changing the splitting

proportion did not lead to significant changes in the logistic regression’s error rate but

significantly accelerated the computing time. It accelerated the logistic regression to

be 306 times faster than the leave-one-out cross-validation and 27 times faster than

the tenfold cross-validation.

Splitting the dataset into 70/30 proportion led to 87.8% accuracy from the cross-

validation and the bootstrap. The exception was the leave-one-out cross-validation

that produced an accuracy of 87.9%. When using a smaller training set, the random

train/test split resulted in 88% accuracy, while the other methods had a slight increase.

However, the bootstrap procedure was the fastest. Using splitting proportions of 50/50,

30/80, and 20/80 did not cause the bootstrap to reduce accuracy significantly.

However, computational time decreased compared to the 70/30 proportion. The

bootstrap procedure in the logistic regression had relatively stable performance in

terms of accuracy regardless of the train/test split proportion. However, other

resampling methods lost accuracy as the training set decreased.

The author considers the bootstrap procedure as suitable for train/test set split for

the logistic regression in a large dataset as it provided similar results to the tenfold

cross-validation that did not change much with the decreasing of the size of the

training set. Therefore, the author recommends using the 30/70, 20/80, and 50/50

proportions to preserve accuracy, while further decreasing computational time.

The Decision Tree Classifier Similar observations can be made for the decision tree classifier. Table 2 summarizes

its performance.

The bootstrap optimizes the performance of the decision tree classifier as well. The

bootstrap produced the output from the decision tree classifier (70/30) in 0.17 seconds

on the adult dataset, while the tenfold cross-validation in 1.65 seconds. As table 2

shows the bootstrap resulted in an accuracy and error rate, similar to those from the

other resampling methods. However, the computational time was much faster. In

some cases, the bootstrap decreased the error rate of the model.

Like the logistic regression, the accuracy of the decision tree classifier started to

increase as a result of the cross-validation and the random train/test split when the

size of the training set decreased. With the decrease of the size of the training set,

cross-validation and the random train/test set tendеd to overfit, which increased the

model’s accuracy. However, the performance of the bootstrap remained relatively

unchanged with the decrease of the training set size and close to the accuracy from

the standard cross-validated 70/30 version of the decision tree.

The author believes the reason behind this result is that the bootstrap can reduce

overfitting even when the training set is smaller than the test set. It is important to be

noted that the datasets did not have any preliminary transformations. In previous

research, Vrigazova and Ivanov (2020a) showed that if the input data have been

standardized and variable selection is performed, the bootstrap produces higher

accuracy than other resampling methods.

235


Table 2

Resampling methods for the Decision Tree Classifier

Dataset Train/test ratio Resampling method Accuracy Error rate Time


LOO 80.8 19.2 40.92




LOO 81.7 18.3 20.40




LOO 82.1 17.9 8.11




LOO 80.1 19.9 0.02




LOO 83.6 16.4 1383.83




LOO 83.5 16.5 635.69




LOO 83.2 16.8 211.96




LOO 83.6 16.4 100.17




LOO 80.6 19.4 4815.19




LOO 80.5 19.5 2566.74




LOO 80.7 19.3 858.83




LOO 79.8 20.2 339.88




The bootstrap produced similar accuracy to that of the tenfold cross-validation

regardless of the splitting proportion. The cross-validation methods and the random

236


train/test split varied in accuracy depending on the splitting ratio. Therefore, the

bootstrap can also be applied with other splitting proportions like the ones presented

in this research. The bootstrap procedure can avoid overfitting not only by using a

smaller training set but also by using nontransformed data as tables 1 and 2 suggest.

Some research (Vrigazova and Ivanov, 2020a) suggests that the bootstrap applied

with a 30/70 splitting proportion can also preserve accuracy while decreasing

computing time. The authors there show that the support vector machines classifier

with tenfold bootstrap and 30/70 splitting ratio can produce similar accuracy to that

produced from the tenfold cross-validation with a ratio of 70/30. The advantage is the

computing time. As tables 1 and 2 show, this paper confirmed this finding for the

logistic regression and the decision tree classifier as well. However, when applied to

untransformed data without variable selection to the logistic regression and the

decision tree classifier, the bootstrap can be used with a 50/50 splitting ratio instead

of 30/70. Depending on the characteristics of the dataset, other proportions can also

be suitable as tables 1 and 2 show.

This is an important finding as the bootstrap can additionally decrease computing

time by applying a smaller size of the training set but preserve the accuracy of the

model. The other resampling methods suffer from fluctuations, so changing the

splitting ratio affects the error rate and may cause overfitting. As the tables show the

computing time decreased but the accuracy either fell, either increased. The

bootstrapped classification is affected by non-transformed data the least while

reducing further computational time.

The K-nearest Neighbour Table 3 presents the results for the k-nearest neighbour. As the table shows, the

bootstrap procedure used with a 70/30 splitting proportion was faster than the tenfold

cross-validation with a 70/30 split. However, the bootstrap’s performance in the k-

nearest neighbour was not so good compared to the logistic regression and the

decision tree. The bootstrap with a 70/30 split proportion resulted in about 2

percentage points higher error rate than the tenfold cross-validation. This finding is in

line with (Vrigazova and Ivanov, 2020a).

However, increasing the size of the test set did not lead to significant improvement

of the accuracy from the bootstrap. The leave-one-out cross-validation and the

repeated training/test split produced better accuracy than the bootstrap. Despite

this, the bootstrap procedure was the fastest. This result is not surprising as previous

research (Vrigazova and Ivanov, 2020a) suggested that the bootstrap procedure may

not be suitable for the k-nearest neighbour as a resampling method using the 70/30

train/test split proportion. We extended the research of these authors by confirming,

firstly, that changing the training/test split proportion cannot increase the accuracy of

the bootstrapped k-nearest neighbour.

Secondly, the bootstrap may not be a suitable resampling method for the k-nearest

neighbour. Our experiments suggest that the most suitable resampling method for the

k-nearest neighbour is the tenfold cross-validation with a train/test splitting proportion

of 70/30. Although the tenfold cross-validation was slower than the bootstrap as table

3 shows, it resulted in high accuracy and was relatively faster than the leave-one-out

cross-validation. Although the repeated train/test split was faster than the tenfold

cross-validation, it produced lower accuracy. Thus, we recommend using the tenfold

cross-validation with 70/30 splitting ratio as a resampling method in the k-nearest

neighbours.

237


Table 3

Resampling methods for the K-nearest Neighbour

Dataset Train/test ratio Resampling method Accuracy Error rate Time


LOO 80.1 19.9 26.26




LOO 79.3 20.7 13.41




LOO 80.1 19.9 26.07




LOO 80.8 19.2 33.05




LOO 84.9 15.1 486.95




LOO 84.9 15.1 201.47




LOO 84.8 15.2 62.31




LOO 85.0 15.0 28.56




LOO 77.3 22.7 6373.29




LOO 77.3 22.7 2624.45




LOO 76.3 23.7 572.23




LOO 75.9 24.1 196.10




238


Discussion This paper proposes a new approach to accelerate computational time in

classification models. The issue of slow computational time becomes severe in large

datasets, where classification can take days and months. Existing literature uses

various approaches to solve this issue. Most of them include changing the equation of

the classification model, using a different type of model, dimensionality reduction, or

data transformation.

For instance, the mixed linear integer approach (Iannarilli & Rubin, 2003) is a

mathematical method that is known to be computationally exhaustive. However,

mathematical methods can be modified to be applied in combination with machine

learning classification, while reducing computational time. The mixed linear integer

approach in classification models (Iannarilli & Rubin, 2003) was modified by

Maldonado (2014) so that it might be used in classification models but perform faster

prediction than the version of Iannarilli & Rubin (2003).

Despite the adaptation of the mixed linear integer approach to classification

models, it still performs slower predictions than traditional machine learning methods

(Vrigazova & Ivanov, 2020b). Therefore, improving one class of methods does not

guarantee the fastest classification. Another approach for reducing computational

time in classification problems that academic literature recommends is changing the

type of model. For example, both the logistic regression and the decision tree classifier

can be appropriate for a particular dataset but the decision tree classifiers can be

faster as they are not a computationally expensive method (Grubinger et. al., 2014).

The logistic regression, however, can be interpreted more easily. Depending on the

aim of the research, the researcher needs to decide whether he/she will use a

computationally inexpensive method.

Another approach to reducing computational time in classification is by using

dimensionality reduction techniques. They can be built-in in the classification model

(Kim & Shin, 2019) or used as a preprocessing step (Yeturu, 2020). Dimensionality

reduction techniques may include feature selection, feature ranking, and principal

component analysis (Yeturu, 2020). These methods aim to choose the features that

carry the most important information for the prediction. Therefore, a subset of the

independent variables is produced that is later used in classification. With the

reduction of features, the classification model becomes less computationally

expensive (Yeturu, 2020). However, the focus of this approach is not to reduce

computational time but rather to improve the classification metrics like accuracy.

Preliminary transformations of data like standardization can also reduce

computational time by limiting high fluctuations in data and transforming the features

to have small values that do not require computationally expensive calculations

(James et. al., 2013). However, this approach is not widely used for reducing

computational time as it has become a standard step in the building of a classification

model (Yeturu, 2020). As Wong (2015) and machine learning textbooks (Yeturu, 2020),

(James et. al., 2013) stated, the velocity for making a prediction depends also on the

resampling procedure used for splitting the dataset into training and test set. For

instance, the leave-one-out cross-validation (Wong, 2015), (James et. al., 2013),

(Yenturu, 2020) is computationally expensive, which slows down predictions. This result

is confirmed in this research (tables 1-3). Their work raises the question of whether

another resampling method can reduce computational time without loss of

accuracy. This paper provided an answer to this question.

The results in this paper extend existing academic literature (Wong, 2015), (James

et. al., 2013), (Yenturu, 2020) by proposing a new practical application of bootstrap

as a training/test splitting method that reduces computational time in classification.

239


The paper shows that changing the resampling method can be another approach to

solve the issue with long computation in classification problems. This result has

important implications in a large dataset as the bootstrap can lead to a much faster

result than the cross-validation and the repeated random train/test split. The paper

shows that the random repeated train/test split method is faster than the tenfold cross-

validation and the leave-one-out cross-validation but slower than the bootstrap. The

random repeated train/test split algorithm leads to loss of accuracy in some cases,

while the bootstrap resulted in similar accuracy to that from the tenfold cross-

validation. Another important recommendation from this paper is using the bootstrap

as a resampling method with a 70/30 train/test split proportion to achieve the best

results. To the author’s best knowledge, this research has been the most detailed one

concerning the applications of bootstrap in machine learning classification. A very

important finding from the research is that the bootstrap is suitable for the logistic

regression and the decision tree classifier but it causes loss of accuracy in the k-nearest

neighbours. With this, the paper recognizes not only the advantages of the bootstrap

in classification problems but for the first time, it outlines a case, where it may not be

suitable for use.

Several limitations of the approach in this paper should be noted, however. The first

one is that input data were not transformed. When standardized, for example, the

accuracy resulting from the four resampling methods may change. The bootstrap

may result in better accuracy than that achieved by the rest of the resampling

method. Although the author has a reason to believe that can be the case, this

hypothesis should be checked. Therefore, a further direction of this research would be

to check what happens with time and accuracy when data are standardized.

Second, standardization of data combined with the bootstrap can also affect the

outcome from the k-nearest neighbour. Standardized data and the bootstrap may

preserve or increase the accuracy of the k-nearest neighbour. This hypothesis should

also be checked.

Also, this paper proposes the use of ten iterations of the bootstrap. It should be

noted that ten iterations are enough to preserve the accuracy of the model.

Increasing the number of iterations can result in computationally exhaustive

classification. For example, running 100 iterations of the bootstrap may result in similar

or slightly better average accuracy than that from the tenfold bootstrap but the time

would increase. The author chose ten iterations of the bootstrap to be comparable to

the tenfold cross-validation in terms of the number of iterations.

It is also possible that the proposed approach may not be suitable for some

datasets despite using the logistic regression and the decision tree classifier. A future

extension of this research would be to examine how the bootstrap would affect the

outcome of the decision tree when it is pruned. Also, are those findings valid in the

case of multiclass classification? The paper proposes using 70/30 proportion to split the

dataset into training and test set. However, depending on the characteristics of the

data and their preliminary transformations, this proportion can differ from dataset to

dataset. This is hardly a limitation of this paper as machine learning textbooks (James

et. al., 2013) do not provide a rule for selecting the training/test splitting ratio.

Therefore, despite providing good results on the datasets used in this research, the

70/30 ratio may not be suitable for all kinds of datasets.

Despite the limitations of this research, it has very important practical implications.

As tables, 1-3 show the bootstrap can reduce computational time several times

compared to the cross-validation and the repeated random train/test split while

preserving accuracy high. This finding is important as the tenfold bootstrap can

perform classification in large datasets without variable selection much faster than the

240


tenfold and leave-one-out cross-validation. This allows the proposed methodology to

be used either as a way to quickly acquaint with the data, for model specification

(e.g. the logistic regression/ decision tree classifier) or as a predictive model with

reduced computing time. All these advantages of the bootstrap procedure allow it

to be a powerful tool in performing machine learning classification models.

Conclusion In this paper, it is shown that using a smaller training set with the bootstrap can preserve

high accuracy and further decrease the computational time of the classification

model. The advantages of using 50/50, 30/70, and 20/80 ratio as training/test set

splitting proportions with the bootstrap procedure, however, are valid only for the

logistic regression and the decision tree classifier. Using the bootstrap procedure as a

resampling method in the k-nearest neighbour is not recommended due to loss of

accuracy. Instead, this research recommends that the k-nearest neighbour might be

fitted by using the tenfold cross-validation with a train/test splitting ratio of 70/30.

Using a 20/80 training/test ratio differs from academic literature and machine

learning textbooks as the number of training instances have to be large enough to

make correct predictions of the test data. A small number of training observations

may fail to capture all important characteristics of the data and make incorrect

predictions. However, the bootstrap procedure allows for correct predictions even

when the training set contains 20% of the dataset (in the case of the logistic regression

and the decision tree classifier). Also, the academic literature suggests improvements

of existing versions of the logistic regression and the decision tree classifier to reduce

computational time but they usually do not involve a change of the resampling

method.

The k-fold cross-validation has become the standard resampling method used in

both the classic versions of the logistic regression and the decision tree classifier and

their modifications. The reason for this is that the k-fold cross-validation provides a

reasonable balance between accuracy and computational time. The experiments

conducted in this research show that the tenfold bootstrap has similar advantages in

the case of the logistic regression and the decision tree classifier. On one hand, the

bootstrap resulted in similar accuracy as the tenfold cross-validation, while performing

faster classification than other resampling methods, including the tenfold cross-

validation. This advantage of the bootstrap can be observed using various

training/test split proportions, e.g. 20/80. These findings have important practical

implications in large datasets as the bootstrap complements existing academic

literature by extending the ways for accelerating fitting and making predictions with

the logistic regression and the decision tree classifier.

Despite the practical advantages of the tenfold bootstrap as a resampling

method, several disadvantages should be considered. Depending on the

characteristics of the dataset, the 20/80 splitting proportion may not always

guarantee high accuracy. So, the rule for a larger training set than the test set may

be valid using the tenfold bootstrap as well. The best training/test set splitting

proportion via the bootstrap can differ in each dataset. Also, preliminary

transformations of data may affect the accuracy of the model. Thus, it is possible that

if independent variables are standardized, the accuracy of the classification may be

increased even in the case of the k-nearest neighbours. A further step of this research

would be to check if standardization of data can increase the accuracy of the

bootstrap procedure. If so, the advantages of the tenfold bootstrap as a resampling

method in classification problems can be further extended.

241


References 1. Breiman L., (1995), “Better Subset Regression Using the Nonnegative Garrote”,

Technometrics, Vol. 37 No 4, pp. 373 – 384.

2. Breiman L., (1992), “The Little Bootstrap and Other Methods for Dimensionality Selection in

Regression: X-fixed Prediction Error”, Journal of American Statistical Association, Vol. 87 No.

419, pp. 738 -754.

3. Breiman, L. (1996), "Bagging predictors", Machine Learning. 24 (2), pp. 123–140.

4. Grubinger, T., Zeileis, A. and Pfeiffer, K., 2014. Evtree: Evolutionary learning of globally

optimal classification and regression trees in R. J. Stat. Software 61 (1), pp. 1-29.

5. Efron B., (1979), “Bootstrap Methods: Another Look at the Jackknife”, the Annals of Statistics,

Vol. 17, pp. 1–26.

6. Efron B., Tibshirani R., (1997), “Improvements on Cross-Validation: The .632+ Bootstrap

Method”, Journal of the American Statistical Association, vol. 92, pp. 548–560.

7. Hoerl E., Kennard W., (1970), “Ridge Regression. Applications to nonorthogonal Problems”,

Technometrics, Vol. 12 No. 1, pp. 69-82. Iz 2012

8. Iannarilli F., Rubin P., (2003), Feature selection for multiclass discrimination via mixed-integer

linear programming, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.

25 No. 6, pp. 779-783.

9. James G., D. W., Hastie T., Tibshirani R., (2013), An Introduction to Statistical Learning.

Springer, STS Vol. 103.

10. Kim B., Shin S., (2019), “Principal weighted logistic regression for sufficient dimension

reduction in binary classification”, Journal of the Korean Statistical Society, Vol. 48 No. 2,

pp. 194-206.

11. Krstajic D., Buturovic J., Leahy E., Thomas S., (2014), “Cross-validation pitfalls when selecting

and assessing regression and classification models”, Cheminformatics, Vol. 6 Article No. 10.

12. MacKinnon J., (2002), “Bootstrap Inference in Econometrics”, The Canadian Journal of

Economics, Vol. 35 No. 4, pp. 615—645.

13. Maldonado S., Pérez J., Weber R., Labbé M., (2014), Feature Selection for Support Vector

Machines via Mixed Integer Linear Programming, Information Sciences, Vol. 279, pp. 163–

175.

14. Pampel F., (2000), Logistic regression: A primer. Sage University Papers Series on

Quantitative Applications in the Social Sciences, 07-132. Sage Publications, Thousand Oaks,

CA.

15. Pedregosa et al., (2011), Scikit-learn: Machine Learning in Python, Journal of Machine

Learning Research 12, pp. 2825-2830.

16. Velliangiri S., Alagumuthukrishnan S., Joseph S., (2019), A Review of Dimensionality

Reduction Techniques for Efficient Computation, Procedia Computer Science, Vol. 165, pp.

104-111.

17. Vrigazova B., (2018), “Nonnegative Garrote as a Variable Selection Method in Panel Data”,

International Journal of Computer Science and Information Security, Vol. 16 No. 1.

18. Vrigazova B., Ivanov I., (2019), “Optimization of the ANOVA Procedure for Support Vector

Machines”, International Journal of Recent Technology and Engineering, Vol. 8 No. 4.

19. Vrigazova B., Ivanov I., (2020a), “The bootstrap procedure in classification problems”,

International Journal of Data Mining, Modelling and Management, Vol. 12 No. 4.

20. Vrigazova, B.& Ivanov, I., (2020b), “Tenfold bootstrap procedure for support vector

machines”, Computer Science, Vo. 21 No. 2, pp. 241-257. 10.7494/csci.2020.21.2.3634.

21. Wong T., (2015), “Performance evaluation of classification algorithms by k-fold and leave-

one-out cross-validation”, Pattern Recognition, Vol. 48 No. 9, pp. 2839–2846.

22. Yeturu К., (2020), Chapter 3 - Machine learning algorithms, applications, and practices in

data science, Editor(s): Arni S.R. Srinivasa Rao, C.R. Rao, Handbook of Statistics, Elsevier,

Vol. 43, pp. 81-206.

242


About the author

Borislava Vrgazova is a data science practitioner. Her research areas include practical

applications of machine learning algorithms for prediction and how their performance

can be boosted. Also, applications of big data techniques to small datasets in the

field of economics as an alternative to traditional econometrics theory. She

challenges traditional econometric modelling techniques used to find connections

among variables from institutional economics by combining feature selection

methods and big data prediction models. As a result, new applications of machine

learning techniques to economic data appear. The author can be contacted at

[email protected]

Vol. 12 No.1 / 2021 ISSN: 1847-9375

Documents

Transcript of Vol. 12 No.1 / 2021 ISSN: 1847-9375