The effect of IFRS on earnings management in Brazilian non-financial public companies

Emerging Markets Review 21 (2014) 42–66

Contents lists available at ScienceDirect

Emerging Markets Review

j ourna l homepage : www.e lsev ie r .com/ locate /emr

The effect of IFRS on earnings management inBrazilian non-financial public companies

Marta Cristina Pelucio-Grecco a,⁎, Cecília Moraes Santostaso Geron b,1,Gerson Begas Grecco a,2, João Paulo Cavalcante Lima b,3

a Mackenzie Presbyterian University, Rua Dr. Gabriel dos Santos, 794 – apto 111, São Paulo, 01231-010, Brazilb Mackenzie Presbyterian University, Rua São Bento, 545 – 5SL, São Paulo, 01011-904, Brazil

a r t i c l e i n f o

⁎ Corresponding author. Tel.: +55 1139274010.E-mail addresses: [email protected]

[email protected] (G.B. Grecco), joao.ca1 Tel.: +55 1139274011.2 Tel.: +55 1139274020.3 Tel.: +55 1123089628.

http://dx.doi.org/10.1016/j.ememar.2014.07.0011566-0141 © 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

Article history:Received 30 April 2014Received in revised form 20 June 2014Accepted 22 July 2014Available online 25 July 2014

This article evaluatedwhether changes in accounting practices brought areduction in earnings management (EM) in listed Brazilian non-financialcompanies through discretionary accruals. We developed a model toobserve the effect of the International Financial Reporting Standards(IFRS) on firms' EM as well as the restrictive effects of the audit, thecorporate governance and the regulatory environment. We find that theones with the most limiting effects is the regulatory environment. Wealso find that the transition to IFRS had a restrictive effect on EM in Brazilafter its complete implementation.

© 2014 Elsevier B.V. All rights reserved.

JEL classification:G34G38M10M41

Keywords:Earnings managementIFRSBrazilEconomic regulated sectors

1. Introduction

Financial statements are part of the set ofinformation that firms make available to investorsand contribute with the equilibrium of informa-tion between main (investors) and agent(manager).

(M.C. Pelucio-valcante2013@

The process of elaboration for financial state-ments involves management making a series ofestimations and judgments based on interpretingthe operation and choosing which accountingpractices to adopt.

This process of choosing and judging directlyinfluences the firm's accounting value, as shown

Grecco), [email protected] (C.M.S. Geron),gvmail.br (J.P.C. Lima).

http://crossmark.crossref.org/dialog/?doi=10.1016/j.ememar.2014.07.001&domain=pdf

http://dx.doi.org/10.1016/j.ememar.2014.07.001

mailto:[email protected]




Unlabelled image

http://dx.doi.org/10.1016/j.ememar.2014.07.001

Unlabelled image

http://www.sciencedirect.com/science/journal/15660141

43M.C. Pelucio-Grecco et al. / Emerging Markets Review 21 (2014) 42–66

in its financial statements. The difficulty of the situation is that if, on the one hand, it is necessary formanagement to make choices, on the other hand, there is the worry on the part of investorsthat management will not properly use its power of choice. This could happen, for example, ifmanagement manipulates the accounting results, a situation which would be known by earningsmanagement.

It is important to highlight that earnings management is not based on fraud; it bases its decisions onspecifically formulated alterations that have been defined by the accounting regulation (Martinez, 2001).

There have been many studies done about earnings management, especially in the last two decades;for instance, Jones (1991), Dechow et al. (1995), Kang and Sivaramakrishnan (1995), Teoh et al. (1998),Dechow and Dichev (2002), Kothari et al. (2005) and Ball and Shivakumar (2008).

Earnings management can be restricted by accounting regulations, particularly those issued by theregulatory agencies of capital markets that aim to ensure quality, comparability and transparency ofinformation as well as the disclosure of the firm's patrimonial position and performance. The moreeffective the regulation, the lower the possibility for the manager to opportunistically manipulate thefinancial statements during the elaboration process and, as a result, the better the quality of theaccounting information that is produced. It is expected that when accounting regulations are modified,there is an improvement in the quality of the accounting information and a reduced possibility that themanager will manipulate the results.

It is pointed out that apart from the regulatory agencies of capital markets, there are other regulatoryagencies depending on the activity sector that the entity is in. Also, in addition to regulating andsupervising economic activity, Brazilian regulatory agencies regulate accounting too.

Besides the restrictive role that accounting regulation plays for earnings management, the importanceof independent auditing is emphasized. The auditor has the responsibility to verify if the financial andpatrimonial position disclosed by an entity is represented reliably. According to Santos and Grateron(2003), ‘In the accounting information external users and even the clients' point of view, the auditor'sacting is a synonym of trust and credibility.’

In terms of the restricting earnings management and ensuring the quality of accounting information,the good practices of corporate governance stand out. According to Almeida dos Santos et al. (2011: 61),‘Firms with a better corporative governance level may have some bigger issues when managing theirresults considering that the information asymmetry can be a counter-incentive for the use of thismechanism.’

In Brazil, as in several other countries, a significant alteration in accounting regulations has beenoccurring due to the convergence process to International Financial Reporting Standards (IFRS), which hasbeen motivating academics and market analysts to search for quality information on the evidence in thisnew era.

The accounting standards have the objective of enhancing the quality of accounting information. Thegreater the effectiveness of the said standards the better the quality of the accounting information.Consequently, the informational asymmetry between agent and the owner will be lower for theunderstanding of the firm.

Several studies have analyzed earnings management in the IFRS era (e.g. Aubert and Grudnitski, 2012;Barth et al., 2012; Fernandes, 2011; G. Iatridis, 2012; Jeanjean and Stolowy, 2008; Leventis et al., 2011;Shelton et al., 2011; Tsipouridou and Spathis, 2012; Wang and Campbell, 2012; Zéghal et al., 2011).However, there has not been any research done about earnings management in Brazil after the beginningof the convergence process.

Because of this gap in research, the objective of this study is to evaluate if the changes in the accountingpractices resulting from IFRS led to reduction of earnings management in the Brazilian non-financialpublic firms causing an improvement in the quality of accounting information.

It is our hope that this research will contribute findings relevant to the market analysis and to societyas a whole about the quality of information in financial statements after the application of IFRS in Brazil;findings relevant to the emission-standards agencies (the same ones who regulate businesses in Brazil)about the effectiveness of the accounting standards on the restrictions of earnings management; andfindings relevant to the academy concerning the development of research that targets the quality ofaccounting information and its reflexes in the area of finance.

44 M.C. Pelucio-Grecco et al. / Emerging Markets Review 21 (2014) 42–66

In the past decade, there has been a sharp increase in direct foreign investments in emergingcountries; chief among them the BRICS (Brazil, Russia, India, China and South Africa), justifying theimportance of studies seeking to measure the quality of the accounting information in thesecountries. In 2012, 20% of foreign investments were directed to the BRICS, and of this amount, 25%went to Brazil (United Nations Conference on Trade and Development, 2013).

Some aspects of Brazil environment causes opportunities for earnings management (Cavalier-Rosa andTiras, 2013), but the IFRS implementation for replacing the tax compliance system represents animportant accounting evolution in terms of accounting quality. Notably, IFRS adoption represented areduction of earnings management. This advance improves investors' trust in Brazil and could cause moreinflux of foreign investments. We would also note that enforcement for regulatory agency is an importantfactor to restrict earnings management at least in legislative law countries such as Brazil.

This paper is structured as follows. In the second section, the current relevance of earningsmanagement and the process adoption of IFRS in Brazil are described and we also develop theresearch hypothesis. The third section describes the sample, variables used and methodologicalprocedure employed. The fourth section shows the empirical findings obtained and its comparisonwith academic references. Finally, we present main conclusions in the fifth section.

2. Background considerations and development of the research hypothesis

2.1. Earnings management

According to Degeorge et al. (1999), ‘Executives have both the incentive and ability to manageearnings. It is hardly surprising that the popular press frequently describes companies as engaged inearnings management—sometimes referred to as manipulation.’

The intentional manipulation of accounting information occurs when managers look to trick investors(Watts and Zimmerman, 1986). According to Christie and Zimmerman (1994), managers' choices can bedone in an efficient way, where the firm's value is maximized, or intentionally privilege their own interestsover those of the investor.

Matsumoto and Parreira (2007) identified that the factors that lead managers to manage earnings isthe lack of a range of standards for all possible situations and the existence of economic and financialincentives that they can obtain.

The earnings management by accounting accruals is directly linked to the difference between cashand accrual methods. According to Healy (1985), non-discretionary accruals refer to those accrualsrequired by the accounting standards in relation to the application of the accrual basis, likeaccounting for fixed assets and its systematic depreciation basis. Discretionary accruals, on the otherhand, refer to the adjustments made intentionally by managers.

Cupertino and Martinez (2008) showed that the level of accruals can be used as a measurement ofearnings manipulation and consequently as a sign of the potential auditing orientation. Additionally,Almeida et al. (2009) showed the possible existence of heterogeneous practices of earnings managementamong firms in the same sector, suggesting the analysis by strategic groups.

There has been a substantial effort among academics to find empirical evidence of the relationshipbetween earnings management practices and various factors, such as in the form of corporate governance(e.g. G.E. Iatridis, 2012; Price et al., 2011; A. Chen et al., 2010); monitoring, shareholding structure and therotation between CEOs and CFOs (e.g. Jouber and Fakhfakh, 2012; Kang et al., 2011; Ayers et al., 2011;Hazarika et al., 2012); management structure and the existence of an audit committee (e.g. Ghosh et al.,2010); the cost of equity and debt capital (e.g. Rodríguez-Pérez and Van Hemmen, 2010; Coelho andLopes, 2007; Nardi et al., 2009); tax planning (e.g. Formigoni et al., 2009; Rezende and Nakao, 2012) andthe identity of the auditing firm (e.g. Martinez and Reis, 2010; Silva and Bezerra, 2010).

2.2. International and Brazilian accounting practices

Many efforts have been made to develop an accounting standard on an international scale based onIFRS, with many of them highlighting the European Union's 2005 adoption of the IFRS for the companieslisted on their stock exchanges.


Among the advantages of the adoption of the convergent accounting practices are the high-qualitystandards of the information supply and the decreased investment risk and capital cost enabled bytransparency and comparability, as shown by Martins and Brasil (2008).

Studies have analyzed IFRS implementation in Europe using the earnings-management models andshown a reduction in earnings management (e.g. Aubert and Grudnitski, 2012; Barth et al., 2012; Leventiset al., 2011; Zéghal et al., 2011) or indifference (e.g. Fernandes, 2011; Wang and Campbell, 2012; Jeanjeanand Stolowy, 2008). Studies have also pointed out that firms based in countries with customary lawstructure tend to show less manipulation of results than those governed by the legislative law, even afterthe IFRS implementation (e.g. Jeanjean and Stolowy, 2008; Shelton et al., 2011).

According to the Brazilian convergence process to IFRS and considering the already-found evidence ofits effect on earnings management, we formulated the following hypothesis:

Hypothesis 1. There is a decrease in earnings management in the financial statements, as will beelaborated in the following section, in the light of IFRS in comparison to the valid accounting practices thattook place in Brazil before the transition to the standards (until 2007), as evidenced in other countries(Aubert and Grudnitski, 2012; Barth et al, 2012; Leventis et al., 2011; Zéghal et al., 2011).

2.2.1. Brazilian convergence to IFRSIn 2005, in Brazil, the Accounting Pronouncements Committee (CPC; Comitê de Pronunciamentos

Contábeis) was created to emit Brazilian accounting standards in consonance with IFRS. Since 2010, allpublic companies, including the financial institutions, according to a commitment made by both theBrazilian Securities Commission (CVM; Comissão de Valores Mobiliários) and the Brazilian Central Bank(BACEN; Banco Central do Brasil), must present their financial statements in accordance to IFRS.

The Brazilian convergence process to the IFRS was divided in two stages: The first, in 2008, was madeup of the alterations brought by Law No. 11.638/07 and the documents issued by the CPC in 2008. Thesecond stage, which took place in 2010, composed by pronouncements issued by the CPC, valid since 2010,but with comparative effect to the year of 2009.

IFRS influence in Brazil can be classified according to three periods: The Pre-IFRS Period, the accountingpractices of which were valid until 2007; the Hybrid Period, the accounting practices of which werepartially convergent to the IFRS, taking place during 2008 and 2009; and the Full-IFRS Period, in whichthere was a complete adoption of the IFRS, taking place in 2010 and with balances represented in 2009 forcomparative effect.

The alterations made in the Brazilian accounting practices brought forth a series of transformations in therecognition, measuring and accounting criteria of assets, liabilities, revenues and expenses in the country.

Alves Oliveira and Lemes (2011) analyzed the Brazilian financial statements from 2008 of the HybridPeriod to evaluate the adaptation level to IFRS of the information disclosed in the Brazilian andNorth-American markets. The authors detected a higher convergence level between USGAAP (GenerallyAccepted Accounting Principles in the United States) and IFRS than between BRGAAP and IFRS, which suggeststhat the Brazilian financial statements tend to substantially could be modified, increasing the disclosure level.

Barbosa Neto et al. (2009) did a comparative analysis looking at the economic and financial indicatorsbetween the Pre-IFRS and Hybrid periods and found that they were not affected in a statistically significantway. In a similar study, Martins and Paulo (2010) made the contrary observation that the calculatedindicators between the Brazilian and the international accounting standard were reduced during thepartial adoption of the IFRS in the Hybrid Period.

Zonatto et al. (2011) analyzed public and private companies in the electric energy sector and showedthat size, financing needs, total indebtedness and return on equity are the factors that best explain thecompanies' adherence to the international accounting standards.

In the process of altering accounting practices that took place in Brazil, divided into two stages, wehope that a gradual improvement in the quality of accounting occurred. On this basis, we formulated thefollowing hypothesis of our research:

Hypothesis 2. There is a reduction of earnings management in the financial statements elaborated in theHybrid Period in comparison to the Brazilian accounting practices in place before the beginning of thetransition period (until 2007).


2.3. Restrictive factors for earnings management

As seen before, we are hopeful that that accounting fulfils its role of reducing informational asymmetryand that the implementation of IFRS in Brazil has brought on an improvement in the quality of accountinginformation. However, regardless of the set of valid standards in a particular country, it is worth pointingout the importance of the quality of auditing and the use of good practices for corporate governance asrestrictive factors in earnings management.

Law No. 6.404/76 establishes that independent auditors must audit the financial statements ofBrazilian public companies. This mandatory requirement seeks to ensure that these demonstrations reflectthe financial and patrimonial position of these entities in a reliable way.

In a study about earnings management conducted by accounting choices and operational decisions,Martinez (2009) found that Big Four auditing (PricewaterhouseCoopers, Deloitte Touche Tohmatsu, KPMGand Ernst & Young) does not ensure less use of earnings management, except in special cases, such as inmanagement made by operational decisions. However, concerning earnings management by accountingchoices, the acquired results in the study suggest that there is a restriction of earningsmanagementwhen BigFour auditing takes place. The sample was composed by Brazilian public companies between 1998 and 2005.

In a similar study, Almeida and Almeida (2009) found that the companies that underwent Big Fourauditing have a lower degree of discretionary accruals (DA) in relation to other companies and, therefore,the power to restrict earnings management by the Big Four.

Silva and Bezerra (2010), looking at a sample of 25 companies for the period from 2000 to 2008, did notfind evidence that a change in a firm's auditing would be related to a reduction of its earningsmanagement, which was the hypothesis of their research.

In a similar vein, Martinez and Reis (2010) explored whether the rotation procedure of auditing firmswould have an effect on earnings management. They analyzed Brazilian public companies for the periodfrom 1997 to 2007 and found evidence that with or without the auditing rotation, there were nosignificant changes on earnings management.

Tsipouridou and Spathis (2012) analyzed Greek public companies between 2005 and 2009 in order toverify if audits encouraged opportunist behaviors to use earnings management after the implementationof IFRS. They did not find any evidence that the Big Four auditing firms presented bigger restrictions onearnings management than the rest.

Ghosh et al. (2010) and Jouber and Fakhfakh (2012) found that the auditing of financial statements bya Big Four auditing firm restricts a company's earnings management.

We assumed that auditing would be a restrictive factor for earnings management. Looking at thedevelopment of the role of accounting to ensure the reliability of accounting information for the externaluser and based on the belief that the Big Four offer better quality than the rest of the auditing firms, wetraced the following hypothesis:

Hypothesis 3. Entities audited by the Big Four present lower earnings management by DA than others.

Another restrictive factor to earnings management that must be emphasized is corporate governance.A growing search for the entities to improve the controls and monitoring of business has led to demand inthe business world for mechanisms that make this monitoring and control possible. Corporate governanceis an answer to this need. The better the business monitoring, the lower will be the possibility of earningsmanagement.

In Brazil, the New Market (NM) is the highest standard of corporate governance from the StockExchange and Mercantile & Futures Exchanges (BM&FBovespa; Bolsa de Valores, Mercadorias e Futuros),and the companies listed on the New Market can only issue shares with voting rights; in other words,common shares. Level 1 (N1; Nível 1) requires that the companies adopt practices that favor transparencyand investors' access to the information. Level 2 (N2; Nível 2) requires listed companies to fulfil theobligations of the New Market with exceptions, like the right to maintain preferred shares.

Martinez (2009) searched for evidence that the New Market would be associated with the propensitytoward earnings management in the form of accounting choices and operational decisions. The evidenceobtained by the author suggests that there is a reduction in earnings management in the accountingchoices made in the New Market. Regarding earnings management made by operational decisions, the


New Market does not ensure lower earnings management except in special cases. It has used data fromBrazilian companies between 1998 and 2005.

Almeida dos Santos et al. (2011) analyzed 14 Brazilian steel mill and metallurgy industries between2005 and 2009, and found that the corporate governance is a counter-incentive to earnings management.In the study, the authors also looked at operational earnings management involving operational decisionsrelative to the costs of sales, general and administrative.

G.E. Iatridis (2012), in a comparative study looking at South Africa and Brazil, found evidence ofeffective mechanisms of corporate governance that reduce the agency cost in South Africa, whereas hefound evidence of significant DA in Brazil. Price et al. (2011) in an analysis of Mexican firms did not find anassociation between corporate governance and earnings management.

A. Chen et al. (2010), observing a sample of firms from Taiwan, demonstrated that good corporategovernance practices are associated with the increase of investors' evaluation of accounting results.

Thus, considering corporate governance to be a restrictive factor to earnings management, we came upwith the following hypothesis:

Hypothesis 4. Entities with a higher corporate governance level show a lower earnings management levelthan others.

Another factor that can be restrictive to earnings management, in its standardization and checking of thefulfilment of the established standards, is the influence of regulatory agencies from certain economy sectors.The regulatory agencies in Brazil are created by laws and have the power to check and regulate the activities ofthe public services executed by the firms enabled to perform these services.

Some examples of regulatory agencies in Brazil are the Telecom National Agency, or ANATEL; the ElectricEnergy National Agency, or ANEEL; and the Petroleum, Natural Gas and Biofuels National Agency, or ANP.

Almeida et al. (2009) suggested earnings management analysis by strategic groups, because in theirresearch they found evidence of the possibility of existence of heterogeneous practices by earningsmanagement.

Extending the suggestion by Almeida et al. (2009), and based on the regulatory agencies having thepower to standardize and check the accounting practices, we formulated the next hypothesis:

Hypothesis 5. Entities in regulated economy sectors show lower earnings management levels than others.

3. Research design

The population of the study is composed of Brazilian public companies, and the non-probabilisticsample is composed of all the Brazilian non-financial public companies listed on BM&FBOVESPA as ofOctober 2, 2012 (361 companies). Since the earnings-management models are based on accountvariations between periods, for 2006 modeling it was necessary to obtain the data of 2005. The data werecollected from the BM&FBOVESPA Web site.

Those entities showing atypical observations (outliers) in the observed period were excluded from theinitial sample, along with those with incomplete data. The final sample is composed of the results of theobservation of 317 companies.

3.1. Empirical models for determining earnings management by DA

The term accrual refers to the difference in how revenues and costs are recognized between the cashand accrual methods. Thus, a firm's accounting earnings can be represented as in Eq. (1) (Rayburn, 1986):

AE ¼ CF þ TA: ð1Þ

Here,

AE accounting earnings (according to accrual basis)CF cash flow (according to cash basis)TA total accruals.


In order to verify whether manipulation of accruals has taken place, it is necessary to split the totalaccruals between the normal ones (non-discretionary) and the abnormal ones (discretionary).

Discretionary accruals (DA) are not observable and can only be obtained by finding the differencebetween the estimation of the total accruals and the estimation of the non-discretionary accruals (NA).The first step is to perform an estimation of the firm's total accruals.

The total accruals can be obtained directly by determining the difference between profit andoperational cash flow observed in the issued financial statements (DeAngelo, 1986: 410). Wheninformation on the operational cash flow is not available, a model must be applied to estimate the totalaccruals.

In Brazil the demonstration of cash flows was not mandatory until 2007, so we estimated the totalaccruals. These can be estimated by using data from a company's balance sheet and the profit and lossstatements, as in Eq. (2) (Jones, 1991: 211; Dechow et al., 1995: 203):

Table 1Studies

Jonesmodi

Jones

JonesDec

JonesKot

JonesTeo

TAt ¼ ΔCAt−ΔCashtð Þ− ΔCLt−ΔSTDtð Þ−Dept½ �=At−1: ð2Þ

In this equation:

TAt the firm's total accruals in period t

ΔCAt the variation of the firm's current assets from the end of period t − 1 until the end of period tΔCasht the variation of the firm's cash and equivalent amount of cash from the end of period t− 1 until

the end of period tΔCLt the variation of the firm's current liability from the end of period t − 1 until the end of period tΔSTDt the variation of the firm's financing and short-term loans from the end of period t − 1 until the

end of period tDept the firm's depreciation and amortization expenses during period tAt − 1 the firm's total assets in the end of period t − 1.

3.1.1. Models for the estimation of DAIn the studies that analyse DA, once the total number of accruals in a determined period are observed

or estimated, the second step is to estimate which of these accruals are non-discretionary and which arediscretionary.

published in international journals using the Jones Model and its modifications between 2008 and 2012.

Model and itsfications

International studies using the model

Model (1991) Baxter and Cotter (2009), Bozec (2008), Cahan et al. (2008), Caramanis and Lennox (2008), Chan etal. (2010), Cohen and Zarowin (2010), Daniel et al. (2008), Herbohn and Ragunathan (2008), G.E.Iatridis (2012), Nguyen and Xu (2010), Niskanen et al. (2011)

Model modified byhow et al. (1995)

Ahn and Choi (2009), Ayers et al. (2011), Balachandran et al. (2008), Botsari and Meeks (2008), A.Chen et al. (2010), Chen et al. (2012), Chi and Gupta (2009), Chung et al. (2009), Cornett et al.(2008), Gargouri et al. (2010), Ghosh et al. (2010), Hadani et al. (2011), Hutton et al. (2009), Jiang etal. (2010), Jorion et al. (2009), Kang et al. (2011), Kim and Yi (2009), Labelle et al. (2010), McInnisand Collins (2011), Mora and Sabater (2008), Sawicki and Shrestha (2008), Wang and Yung (2011),Yu (2008), Zhao and Chen (2008)

Model modified byhari et al. (2005)

Bona‐Sánchez et al. (2011), Bozec (2008), Cheng et al. (2010), Choi and Pae (2011), Cohen andZarowin (2010), Ettredge et al. (2010), Guthrie and Sokolowsky (2010), Hazarika et al. (2012),Ibrahim (2009), Jaggi et al. (2009), Jouber and Fakhfakh (2012), Kang et al. (2011), Lee and Masulis(2011), Prior et al. (2008), Rodríguez-Pérez and Van Hemmen (2010), Sun et al. (2011), Wilson andWang (2010), Wilson and Wu (2011)

Model modified byh et al. (1998)

Chang et al. (2010), A.S. Chen et al., 2010; Fischer and Louis (2008), Gong et al. (2008)


According to Healy (1985), NA reflects the average amount of the firm's total accruals in the studiedperiod. DeAngelo (1986) estimated the variations in NAs in accordance with the variations in the totalaccruals between the studied period and the previous period, instead of using the overall average, asdescribed by Healy (1985).

As observed by Jones (1991) and Dechow et al. (1995), only if NAs are constant across time will thesemodels not show errors; if they are not constant, the models will show estimation defects.

Jones (1991) proposed using more flexible model when dealing with non-discretionary variations intime. As opposed to the models of Healy (1985) and DeAngelo (1986), in Jones's model the estimation ofaccruals is controlled by changes in the entity's economic circumstances. In addition, the Jones Model firstapplies the regression shown in Eq. (3) to obtain NA, as follows:

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVit=Ait−1ð Þ þ β2i ΔPPEit=Ait−1ð Þ þ εit : ð3Þ

In this equation:

ΔREVit the revenues in period t minus the revenues in period t − 1, divided by the assets in period t −1, by firm i

ΔPPEit property, plant and equipment in period t, divided by the assets in period t − 1, by firm iAit − 1 the total assets in period t − 1, by firm iεit the regression error in period t by firm i. The amount of NA can be obtained by the estimated

residuals.

Dechow et al. (1995) suggested that the Jones Model (Jones, 1991) does not consider that the revenueis an element that is likely to be manipulated and therefore should be considered as a discretionaryelement as well. The authors proposed making an alteration to the model, which became known inacademic literature as the Modified Jones Model. The model increases the variation of receiving accounts(ΔRECit), as shown in Eq. (4):

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVit−ΔRECitð Þ=Ait−1½ � þ β2i ΔPPEit=Ait−1ð Þ þ εit : ð4Þ

Teoh et al. (1998) used the Modified Jones Model; however, they proposed that the property, plant andequipment effect be withdrawn; for, as found in Guenther (1994), the long-term accruals are less likely tobe manipulated by the managers, as in Eq. (5):

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVit−ΔRECitð Þ=Ait−1½ � þ εit : ð5Þ

Working with the observations of Dechow et al. (1995), Kothari et al. (2005) emphasized that theproposed Modified Jones Model was likely to show an increasing of DA when the firm was growing. Thus,they offered another model that also modified the Jones Model, which includes a return on assets (ROA). Itis shown in Eq. (6):

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVit−ΔRECitð Þ=Ait−1½ � þ β2i ΔPPEit=Ait−1ð Þ þ β3i ROAitð Þ þ εit : ð6Þ

A wide utilization of the Jones Model and its modifications takes place in the international literature,where we find the use of the Jones Model (Jones, 1991), the use of the Modified Jones Model by Dechow etal. (1995), the Modified Jones Model as put forth by Kothari et al. (2005) and the Modified Jones Model byTeoh et al. (1998). Table 1 shows some research conducted between 2008 and 2012 that is published ininternational periodicals that uses these models.

Kang and Sivaramakrishnan (1995) developed another model to detect earnings management, knownin academic literature as the KS Model, represented in Eq. (7):

TAit=Ait−1ð Þ ¼ θ0 þ θ1 δ1 REVit=Ait−1ð Þ½ � þ θ2 δ2 EXPit=Ait−1ð Þ½ � þ θ3 δ3 GPPEit=Ait−1ð Þ½ � þ εit : ð7Þ


Here:

δ1 RECit − 1/REVit − 1

δ2 (INVit − 1 + OCAit − 1 - CLit − 1)/EXPit − 1

INVit − 1 firm i's inventory balance in period t − 1OCAit − 1 firm i's balance of prepaid expenses in period t − 1CLit − 1 firm i's balance of receivables in the current liability in period t − 1EXPit − 1 firm i's expenses in period t, excluding the ones with depreciation and amortization at the end of

period t 1δ3 DEPit − 1/GPPEit − 1

DEPit − 1 firm i's amount of depreciation and amortization expenses in period t − 1GPPEit − 1 firm i's balance of gross fixed assets in period t − 1θ0, θ1, θ2, θ3 estimated regression coefficientεit regression error.

According to Martinez (2001: 48), ‘There is no perfect model, but the KS Model seems to be, in thecurrent state of art, the one providing the best results.’ There have not been found any studies using the KSModel in the international literature, according to a bibliographic research conducted in CAPES databasebetween 2008 and 2012.

The majority of the Brazilian studies about earnings management explore earnings management thatuses DA, giving priority to the Jones Model and its modifications as well as the KS Model. Table 2 shows themain Brazilian studies looking at management analysis and the use of DA.

To measure earnings management that uses DA, the following models have been applied in this study:(1) the Jones Model (Eq. (3)); (2) the Jones Model modified by Dechow et al. (1995) (Eq. (4)); (3) theJones Model modified by Teoh et al. (1998) (Eq. (5)); (4) the Jones Model modified by Kothari et al. (2005)(Eq. (6)); and (5) the KS Model (Eq. (7)).

We have also applied the Jones Model modified by Teoh et al. disregarding the variation adjustmentof receivables, as shown in Eq. (8), to measure earnings management that uses DA. This modification wassubsequently used by Fischer and Louis (2008). Teoh et al. (1998) had already observed that theadjustment could have been omitted.

Table 2Brazilia

Modediscre

JonesJonesDec

KS M

JonesTeo

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVitð Þ=Ait−1½ � þ εit ð8Þ

Combining the suggestion by Teoh et al. (1998) that the fixed effect should be removed since it was lesslikely to have been manipulated with the suggestion by Kothari et al. (2005) that ROA should be includedin regard to the firm's growing, we have also applied Eqs. (9) and (10), to measure earnings managementthat uses DA.

n studies using discretionary accruals of various models.

l featuringtionary accrual

Study utilizing the model

Model (1991) Cupertino and Martinez (2008), Vasconcelos et al. (2008)Model modified byhow et al. (1995)

Coelho and Lopes (2007), Paulo et al. (2007), Cupertino and Martinez (2008), Almeida et al.(2009), Nardi et al. (2009), Paulo and Leme (2009)

odel (1995) Paulo et al. (2007), Martinez (2008), Vasconcelos et al. (2008), Almeida and Almeida (2009),Almeida et al. (2009), Martinez (2009), Nardi and Nakao (2009), Nardi et al. (2009), Paulo andLeme (2009), Silva and Bezerra (2010), Martinez (2011), Rezende and Nakao (2012)

Model modified byh et al. (1998)

Nardi et al. (2009)


In Eq. (9), we kept the variation in adjustment of receivables, as suggested by Dechow et al. (1995):

Table 3Main ge

Indep

ΔREVPPE/AConstR2

AdjusF statProbAkaikSchwHannDW s

*Signifi(TAit/Ait

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVit−ΔRECitð Þ=Ait−1½ � þ β2i ROAitð Þ þ εit : ð9Þ

In Eq. (10), we omitted the variation adjustment of receivables:

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVitð Þ=Ait−1½ � þ β2i ROAitð Þ þ εit : ð10Þ

We also used the Jones Model modified by Kothari et al. (2005), omitting the receivables variation, asshown in Eq. (11), in order to measure earnings management that uses DA:

TAit=Ait−1ð Þ ¼ αit 1=Ait−1ð Þ þ β1i ΔREVitð Þ=Ait−1½ � þ β2i ΔPPEit=Ait−1ð Þ þ β3i ROAitð Þ þ εit : ð11Þ

We made the choice to use the Jones Model and apply its modifications because they have had agreater acceptance in the international academic literature, as found in empirical studies analyzing theimplementation of IFRS internationally (Fernandes, 2011; G.E. Iatridis, 2012; Tsipouridou and Spathis,2012; Zéghal et al., 2011). We calculated the ROA amount applied in Eqs. (6), (9), (10) and (11) bydividing the net profit by the total assets in the reference period, following Kothari et al. (2005). We chosethe KS Model because of its wide use in Brazilian academic research and because, as evidenced byMartinez (2001), we thought that it would most accurately represent Brazilian reality.

In Eq. (7), showing the application of the KS Model, we used the net fixed value, replacing the grossfixed asset, according to the available data in the Standard Financial Statements by CVM.

3.1.2. Model to detect the effects of IFRS on earnings managementIn order to analyze the effect of IFRS on earnings management, we used the absolute value of the

estimated DA as a dependent variable. According to Jouber and Fakhfakh (2012), the absolute value of theestimated DA measures the transition from one period to another, without sensitivity to the significanttime when earnings are increased or reduced.

In order to verify if there was an alteration in earnings management after IFRS was implemented inBrazil, we used categorical dependent variables to indicate the influence of IFRS in the elaboration offinancial statements in the Hybrid Period (HIB) and in the Full-IFRS Period (IFRS).

As stated in Hypotheses 1 and 2, we expect that the coefficient signs of the HIB and IFRS will benegative, for the expectation is that a reduction in earnings management in the presence of internationalaccounting standards occurs.

As previous empirical evidence shows, there are several factors that can be related to earningsmanagement or its restriction. Thus, in the model we applied to detect the effect of IFRS on earningsmanagement, there will be some independent variables based on the pointed evidence, as describedhereafter.

neral regression statistics of Eq. (3) for the Jones Model.

endent variable 2006 2007 2008 2009 2010 2011

/A 0.05196 0.10679 0.13570** 0.04524 0.07370* 0.10386**−0.15813** −0.23590** −0.13442** −0.02195 −0.09592** −0.07768**

ant 0.04406 0.07465* 0.00654 −0.02519 0.02824 0.002870.07003 0.12924 0.09923 0.01043 0.06595 0.09115

ted R2 0.06068 0.12091 0.09097 0.00267 0.05684 0.08429istic 7.49217 1.55102 1.20077 1.34373 7.23707 1.32887(F statistic) 0.00073** 0.00000** 0.00001** 0.26271 0.00092** 0.00000**e info criterion −0.67708 −0.41617 −0.81043 −1.41334 −1.30703 −1.64375arz criterion −0.62795 −0.36867 −0.76430 −1.37202 −1.25890 −1.60355an–Quinn criterion −0.65720 −0.39697 −0.79181 −1.39673 −1.28757 −1.62761tatistic 1.82517 2.17628 1.93124 1.83235 1.87492 1.94874

cant at 5%. **Significant at 1%.− 1) = αit (1/Ait − 1) + β1i (ΔREVit/Ait − 1) + β2i (ΔPPEit/Ait − 1) + εit.

Table 4Main general regression statistics of Eq. (3) for the Polynomial Jones Model.

Independent variable 2006 2007 2008 2009 2010 2011

ΔREV/A 0.13643 0.15289* 0.29495* 0.08402 0.20864** 0.20351**ΔREV/A2 0.13839 0.15002 0.10610 0.34760** −0.07823 −0.20693ΔREV/A3 −0.19681 −0.02454 −0.62510 −0.31371** −0.11435* 0.00642ΔREV/A4 0.03674 −0.03699 0.30995 −0.31611** 0.04861* 0.04353PPE/A −1.77847 −1.62636* −0.93647 −0.68297** −0.41079 0.01160PPE/A2 5.51399 3.07958 2.43245 1.63992** 0.90066 −0.32723PPE/A3 −6.43956* −2.31992 −2.61704 −1.42610** −0.84577 0.33925PPE/A4 2.39619* 0.59977 0.92848 0.35624** 0.26103 −0.08723Constant 0.13918 0.20400** 0.04596 0.02937 0.03574 −0.00189R2 0.13010 0.25829 0.14239 0.15927 0.09751 0.11112Adjusted R2 0.09405 0.22906 0.11003 0.13226 0.06123 0.08366F statistic 3.60816 8.83666 4.39973 5.89636 2.68771 4.04704Prob (F statistic) 0.00062** 0.00000** 0.00006** 0.00000** 0.00793** 0.00015**Akaike criterion −0.68446 −0.51998 −0.80523 −1.52983 −1.28372 −1.62119Schwarz criterion −0.53706 −0.37748 −0.66684 −1.40589 −1.13930 −1.50059Hannan–Quinn criterion −0.62482 −0.46238 −0.74935 −1.47999 −1.22532 −1.57275DW statistic 1.83786 2.04052 1.89111 1.77518 1.90614 1.88965

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i (ΔREVit/Ait − 1) + β2i (ΔPPEit/Ait − 1) + εit.


In order to verify that the presence by the Big Four auditing companies restricts earnings management,following Ghosh et al. (2010) and Jouber and Fakhfakh (2012), we included a categorical independentvariable to indicate the auditing presence by a Big Four company (AUDIT). As stated in Hypothesis 3, weexpected that the coefficient sign of the AUDIT variable would be negative, denoting the restrictive powerof auditing on earnings management.

To test Hypothesis 4, we included the categorical independent variable Corporate Governance (GC; inPortuguese Governança Corporativa), which we expected would be negative, based on the assumption thatentities with a higher level of GC show less earnings management than others.

To evaluate Hypothesis 5, we included the control variable REG, to control if the economy sectorregulation by a regulatory agency, besides CVM, could have restrictive effect on earnings management.

We included the following control variables: the firm's size (SIZE), according to which the coefficientsign cannot be foreseen, following Ghosh et al. (2010) and Jouber and Fakhfakh (2012) among otherauthors; the indebtedness level (LEV), which we expected to be positive and according to Rodríguez-Pérezand Van Hemmen (2010) and Nardi et al. (2009) increases in debts produces incentives for managers tomanipulate earnings.

Table 5Main general regression statistics of Eq. (4) for the Jones Model modified by Dechow et al. (1995).


(ΔREV − ΔREC)/A 0.03906 0.03916 0.06153 −0.02727 −0.02295 0.04554PPE/A −0.16293** −0.24955** −0.14148** −0.02167 −0.09227** −0.07837**Constant 0.04832 0.09040 0.02385 −0.02340 0.04011 0.00983R2 0.06657 0.10412 0.06482 0.00537 0.04321 0.05749Adjusted R2 0.05719 0.09555 0.05624 −0.00244 0.03387 0.05038F statistic 7.09632 1.21453 7.55499 0.68774 4.62859 8.08235Prob (F statistic) 0.00106** 0.00001** 0.00067** 0.50364 0.01081** 0.00039**Akaike info criterion −0.67337 −0.38773 −0.77294 −1.40823 −1.28298 −1.60739Schwarz criterion −0.62424 −0.34023 −0.72681 −1.36692 −1.23484 −1.56719Hannan–Quinn criterion −0.65349 −0.36853 −0.75432 −1.39162 −1.26351 −1.59124DW statistic 1.83729 2.15444 1.92169 1.81561 1.71737 1.95763

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + εit.

Table 6Main general regression statistics of Eq. (4) for the Polynomial Jones Model modified by Dechow et al. (1995).


(ΔREV − ΔREC)/A 0.09564 −0.02783 0.14634 -0.03430 −0.15499 0.03241(ΔREV − ΔREC)/A2 0.17813 0.40676 0.07166 0.17675 0.07639 −0.16847(ΔREV − ΔREC)/A3 −0.12001 0.02503 −0.42461 -0.11924 0.11329 0.26010(ΔREV − ΔREC)/A4 −0.03884 −0.13537 0.23546 -0.15252 −0.04932 −0.08598PPE/A −1.81690 −1.80972* −0.94734 -0.72724** -0.36370 0.03220PPE/A2 5.60360 3.52717 2.36717 1.65675** 0.71219 −0.38822PPE/A3 −6.53945* −2.77168 −2.54468 -1.38833* −0.64332 0.39948PPE/A4 2.43521* 0.74413 0.91716 0.34099* 0.20091 −0.10246Constant 0.14924 0.24121** 0.07978 0.04700 0.06819* 0.00762R2 0.12310 0.24218 0.09465 0.13729 0.07347 0.06860Adjusted R2 0.08675 0.21231 0.06049 0.10957 0.03622 0.03983F statistic 3.38663 8.10911 2.77049 4.95300 1.97245 2.38459Prob (F statistic) 0.00116** 0.00000** 0.00622** 0.00001** 0.05162 0.01704*Akaike info criterion −0.67644 −0.49848 −0.75106 -1.50401 −1.25742 −1.57447Schwarz criterion −0.52904 −0.35599 −0.61268 -1.38007 −1.11301 −1.45388Hannan–Quinn criterion −0.61680 −0.44089 −0.69518 -1.45418 −1.19903 −1.52603DW statistic 1.85214 2.10240 1.90032 1.75614 1.73076 1.95413

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + εit.


Additionally, we included the control variable HOLDING to identify Brazilian public companies that arerun by firms based in countries that adopted IFRS prior to 2010, the year when it was adopted in Brazil. Wedid this because of the claim by Bentwood and Lee (2012) that in the adoption period a big amount ofadjustments (intentional or unintentional) were made in the information generated by the financialstatements. Thus, to control if the oscillations in DA could have been influenced by unintentional errors,we included HOLDING. We expected that the coefficient sign of HOLDING would be negative, for thecompanies whose holding adopted IFRS before Brazil did, in accordance with its prior experience, wouldtend to err less and would not have the unintentional effect on earnings management.

Thus, to test if Hypothesis 5 had been applied, we used Eq. (12), in which besides the categoricalvariables it has used effects of interaction between the categorical variables IFRS and AUDIT and HIB andAUDIT:

Table 7Main ge

Indep

(ΔREConstR2

AdjusF statProbAkaikSchwHannDW s

*Signifi(TAit/Ait

DAitj j ¼ αþ β1iIFRSit þ β2iHIBit þ β3iAUDIT it þ β4i IFRSitxAUDIT itð Þ þ β5i HIBitxAUDIT itð Þþ β6iGCit þ β7iREGit þ β8iHOLDINGit þ β9iSIZEit þ β10iLEV it þ εit : ð12Þ

neral regression statistics of Eq. (5) for the Jones Model modified by Teoh et al. (1998).

endent variable 2006 2007 2008 2009 2010 2011

V − ΔREC)/A 0.07106 0.05506 0.06909 −0.02781 −0.03255 0.04545ant −0.02454* −0.01872 -0.03856** −0.03194** 0.00883 −0.01734*

0.00808 0.00500 0.00887 0.00258 0.00452 0.00601ted R2 0.00312 0.00026 0.00435 −0.00132 −0.00031 0.00227istic 1.62847 1.05574 1.96040 0.66130 0.93566 1.60820(F statistic) 0.20340 0.30537 0.16289 0.41686 0.33453 0.20585e info criterion −0.62249 −0.29223 -0.72389 −1.41319 −1.25296 −1.56167arz criterion −0.58974 −0.26056 -0.69313 −1.38564 −1.22086 −1.53487an–Quinn criterion −0.60924 −0.27943 -0.71147 −1.40211 −1.23998 −1.55090tatistic 1.91820 2.15713 1.93554 1.81255 1.73549 1.98088

cant at 5%. **Significant at 1%.− 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + εit.

Table 8Main general regression statistics of Eq. (5) for Polynomial Jones Model modified by Teoh et al. (1998).


(ΔREV − ΔREC)/A 0.14094 0.02777 0.14437 0.02353 −0.16812 0.01363(ΔREV − ΔREC)/A2 0.24250 0.48686 0.18796 0.22571 0.06023 −0.11123(ΔREV − ΔREC)/A3 −0.28267 −0.01590 −0.59518 −0.21115 0.12065 0.26189(ΔREV − ΔREC)/A4 0.03089 −0.15047 0.29350 −0.22170 −0.04932 −0.10489Constant −0.03312** −0.03548 −0.04856** −0.03804** 0.01820 −0.01464R2 0.03209 0.06768 0.01685 0.01796 0.02758 0.00863Adjusted R2 0.01244 0.04966 -0.00136 0.00244 0.00842 −0.00645F statistic 1.63305 3.75650 0.92527 1.15683 1.43938 0.57209Prob (F statistic) 0.16741 0.00567** 0.45007 0.33044 0.22227 0.68312Akaike info criterion −0.61730 −0.32899 −0.70482 −1.40547 −1.24755 −1.54191Schwarz criterion −0.53541 −0.24982 −0.62793 −1.33662 −1.16732 −1.47492Hannan–Quinn criterion −0.58417 −0.29699 −0.67377 −1.37779 −1.21511 −1.51501DW statistic 1.94124 2.09680 1.92169 1.80769 1.73593 1.99404

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + εit.


In this equation:

IFRSit a categorical variable; designated a 1 if firm i's financial statements were elaborated inconsonance with IFRS in year t, and designated a 0 in other cases;

HIBit a categorical variable; given a 1 if firm i's financial statements were elaborated according tostandards existent in the Hybrid Period in year t, and given a 0 in all other cases

AUDITit a categorical variable; attributed a 1 if entity i was audited by one of the Big Four firms in year t,and attributed a 0 in other cases

GCit a categorical variable, designated a 1 if entity i had a CVM classification of GC (NM, N1 or N2) inyear t, and designated a 0 in other cases

REGit a categorical variable; attributed a 1 if entity i's sector (sectors: public utility; telecom; oil, gasand biofuel) was regulated by an agency in year t, and attributed a 0 in other cases (e.g.information technology, cyclic and non-cyclic consumption, basic materials, industrial goods,construction and transport)

HOLDINGit indebtedness level; measured by total liabilities (current and non-current) divided by firm i'stotal assets in year t

SIZEit size of firm; measured by the natural logarithm of firm i's assets in year tLEVit leverage of firm i in year t, measured by the ratio of total liabilities (current and long-term)

divided by the total assets.

Table 9Main general regression statistics of Eq. (8) for the Jones Model modified by Teoh et al. (1998) without the accounts receivablevariation.


ΔREV/A 0.08331 0.13007 0.14703** 0.04532 0.07184* 0.10500**Constant −0.02752* −0.03003* −0.05364** −0.03385** −0.00513 −0.02418**R2 0.01559 0.04163 0.04892 0.00757 0.02373 0.04057Adjusted R2 0.01067 0.03706 0.04458 0.00369 0.01899 0.03696F statistic 3.16770 9.12126 1.12653 1.95216 5.00777 1.12479Prob (F statistic) 0.07663 0.00284** 0.00093** 0.16356 0.02630* 0.00091**Akaike info criterion −0.63010 −0.32973 −0.76514 −1.41820 −1.27244 −1.59705Schwarz criterion −0.59734 −0.29807 −0.73438 −1.39066 −1.24035 −1.57026Hannan–Quinn criterion −0.61684 −0.31693 −0.75272 −1.40713 −1.25947 −1.58629DW statistic 1.90003 2.17319 1.94129 1.83008 1.88728 1.97310

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + εit.

Table 10Main general regression statistics of Eq. (8) for the Polynomial Jones Model modified by Teoh et al. (1998) without the accountsreceivable variation.


ΔREV/A 0.17907 0.26368** 0.33970** 0.11784* 0.23283** 0.19397*ΔREV/A2 0.20607 0.16053 0.16120 0.38174** −0.10721* −0.18032ΔREV/A3 −0.32417 −0.06134** -0.80581* −0.26055 −0.13056** 0.02902ΔREV/A4 0.08113 -0.03499 0.39013 −0.28822* 0.05810** 0.02213Constant −0.03823** −0.05318** -0.07643** −0.04775** −0.01784 −0.02587**R2 0.04539 0.12055 0.08311 0.05911 0.05413 0.05092Adjusted R2 0.02600 0.10355 0.06613 0.04424 0.03550 0.03648F statistic 2.34152 7.09342 4.89489 3.97376 2.90448 3.52724Prob (F statistic) 0.05637 0.00002** 0.00085** 0.00381** 0.02287* 0.00798**Akaike info criterion −0.63113 −0.38737 -0.77460 −1.44828 −1.27523 −1.58551Schwarz criterion −0.54924 −0.30820 -0.69772 −1.37943 −1.19500 −1.51851Hannan–Quinn criterion −0.59800 −0.35537 -0.74355 −1.42059 −1.24279 −1.55860DW statistic 1.91983 2.00641 1.91616 1.83192 1.90979 1.92205

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + εit.


4. Result and analysis

4.1. Results of the measurement of earnings management

As previously seen, in order to measure earnings management through looking at DA in this study, weapplied Eqs. (3)–(11). To estimate DA, we used a transversal sectioning criterion since it is the mostappropriate for estimating discretionary appropriations, according to Chen et al. (2012). Additionally, inorder to amplify the specifications of the models, we used polynomials of degrees 2, 3 and 4, as suggestedby Gujarati and Porter (2008: 225–226).

We used the EViews Statistic Software for obtaining the results, deriving a coefficient estimation byusing the ordinary least squares method and White's heteroscedasticity correction, as suggested byGujarati and Porter (2008: 414). Regarding normality of residuals, we pointed out that according to thecentral limit theorem, sufficiently big samples have an approximately normal distribution function(Gujarati and Porter, 2008: 119).

Based on the previous results, we used the following criteria to select models to calculate the residualsas discretionary appropriations: The smaller, the better (Akaike); the smaller, the better (Schwarz); thesmaller, the better (Hannan–Quinn); the bigger, the better (adjusted R2); and the level of statisticalsignificance (global p-value) of the model.

Table 11Main general regression statistics of Eq. (9) Jones Model modified by Teoh et al. (1998) with ROA variable.


(ΔREV − ΔREC)/A 0.05646 0.04842 0.06099 −0.03436 −0.03335 0.04111ROA 0.06976** 0.18402 0.16913 0.03556 0.03091 −0.01501Constant −0.02267 −0.02509 -0.03879** −0.03342** 0.00739 −0.01686R2 0.05394 0.02124 0.06401 0.00443 0.00594 0.00940Adjusted R2 0.04443 0.01187 0.05543 −0.00338 −0.00376 0.00193F statistic 5.67302 2.26748 7.45453 0.56727 0.61201 1.25777Prob (F statistic) 0.00402** 0.10612 0.00074** 0.56779 0.54325 0.28598Akaike info criterion −0.65993 −0.29925 -0.77208 −1.40729 −1.24476 −1.55762Schwarz criterion −0.61080 −0.25175 -0.72595 −1.36598 −1.19662 −1.51743Hannan–Quinn criterion −0.64005 −0.28005 -0.75345 −1.39068 −1.22530 −1.54148DW statistic 1.86620 2.14877 1.89730 1.80675 1.73620 1.98930

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ROAit) + εit.

Table 12Main general regression statistics of Eq. (9) for Polynomial Jones Model modified by Teoh et al. (1998) with ROA variable.


(ΔREV − ΔREC)/A 0.07833 0.03702 0.11811 −0.01232 −0.22272 −0.12786(ΔREV − ΔREC)/A2 0.32907 0.41924 0.21037 −0.00031 0.08128 0.08214(ΔREV − ΔREC)/A3 −0.29564 −0.01697 −0.53026 0.00436 0.15546 0.25407(ΔREV − ΔREC)/A4 0.03376 −0.13115 0.24131 −0.00221 −0.05933 −0.13643ROA 0.29908 −0.40040 0.27408 0.10046 0.25225 0.33193**ROA2 −0.55117 −1.16432 −1.63670* −0.20687 −1.41428** −0.40912ROA3 −1.19489 6.62617 −2.11337* 0.04114 -1.38704 −0.30834ROA4 0.15243 5.19241 −0.55986 0.74593 1.12524* −0.03522Constant −0.0355** 0.00089 −0.02908 −0.03759** 0.02414 −0.01780*R2 0.11122 0.18787 0.18274 0.05805 0.07557 0.10732Adjusted R2 0.07438 0.15586 0.15190 0.02779 0.03841 0.07974F statistic 3.01889 5.86994 5.92548 1.91820 2.03344 3.89199Prob (F statistic) 0.00323** 0.00000** 0.00000** 0.05781 0.04428* 0.00024**Akaike info criterion −0.66298 −0.42927 −0.85343 −1.41615 −1.25969 −1.61692Schwarz criterion −0.51558 −0.28677 −0.71504 −1.29221 −1.11528 −1.49633Hannan–Quinn criterion −0.60334 −0.37167 −0.79755 −1.36631 -1.20130 −1.56849DW statistic 1.89945 2.08072 1.92681 1.82126 1.74802 1.96356

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ROAit) + εit.


As shown in Table 3, the main statistics of the regression we obtained from applying the Jones Modeldemonstrate that based on the F statistic and p-value (Prob F statistic), the Jones Model is statisticallysignificant to the usual confidence intervals, except in 2009. According to the results shown in theHannan–Quinn and Schwarz criterion, the Jones Model is the best model in 2011. In general, theindependent variables are significant within the usual confidence intervals

As indicated in Table 4, the application of the polynomial Jones Model derives an improvement of R2

and for the majority of the years also ensures better information for the Akaike, Schwarz and Hannan–Quinn criterion. The global p-value (Prob F statistic) demonstrates that the models are statisticallysignificant in all studied periods.

In Table 5, according to the F statistic and p-value, the Jones Model modified by Dechow and otherresearchers (Dechow et al., 1995) is generally statistically significant within the usual confidence intervalsin the period that we studied, except for 2009. It is possible to note that the model does not present thebest information criteria when it is compared to the others models. When the polynomial model is used,the results are similar, according to Table 6.

The polynomial Jones Model modified by Teoh et al. (1998) without variation referring to accountreceivable and including ROA (Table 14) is the unique variation of Jones Model modified by Teoh et al.

Table 13Main general regression statistics of Eq. (10) for Jones Model modified by Teoh et al. (1998) without accounts receivable variationplus ROA variable.


ΔREV/A 0.07258 0.12642 0.13710** 0.04463 0.07114* 0.10257**ROA 0.06908** 0.17666 0.16211 0.00395 0.02111 −0.01247Constant −0.02560 −0.03630 −0.05331** −0.03400** −0.00608 −0.02381**R2 0.06064 0.05662 0.09947 0.00759 0.02439 0.04293Adjusted R2 0.05120 0.04760 0.09120 −0.00019 0.01487 0.03570F statistic 6.42356 6.27230 1.20393 0.97523 2.56262 5.94301Prob (F statistic) 0.00198** 0.00226** 0.00001** 0.37851 0.07957 0.00299**Akaike info criterion −0.66704 −0.33607 −0.81069 −1.41047 −1.26350 −1.59205Schwarz criterion −0.61791 −0.28857 −0.76456 −1.36916 −1.21537 −1.55185Hannan–Quinn criterion −0.64716 −0.31687 −0.79207 −1.39386 −1.24404 −1.57591DW statistic 1.84965 2.16707 1.90947 1.82957 1.88850 1.98009

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + β2i (ROAit) + εit.

Table 14Main general regression statistics of Eq. (10) for the Polynomial Jones Model modified by Teoh et al. (1998) without the accountsreceivable variation. including the ROA variable.


ΔREV/A 0.14474 0.27663** 0.29746** 0.09421 0.23608* 0.08676ΔREV/A2 0.26985 0.11404 0.17196 0.21477 −0.12089* −0.04152ΔREV/A3 −0.38369 −0.06334** −0.70977 −0.11776 −0.13194* 0.02293ΔREV/A4 0.10888 −0.02586 0.32920 −0.13852 0.06318* 0.00154ROA 0.26864 −0.46993 0.22650 0.07066 0.10176 0.26172*ROA2 −0.58280 −1.13008 −1.52336 −0.14468 −1.40573* −0.25387ROA3 −1.20252 7.17444 −1.83280 0.09782 −1.88972* −0.17942ROA4 −0.15296 5.54772 −0.47000 0.60649 1.41448* −0.01991Constant −0.03919** −0.01473 −0.05420** −0.04709** −0.00971 −0.02818**R2 0.12413 0.24363 0.23639 0.08666 0.07905 0.10934Adjusted R2 0.08782 0.21383 0.20758 0.05732 0.04202 0.08183F statistic 3.41894 8.17352 8.20371 2.95319 2.13501 3.97438Prob (F statistic) 0.00106** 0.00000** 0.00000** 0.00358** 0.03419* 0.00019**Akaike info criterion −0.67761 −0.50040 −0.92133 −1.44699 −1.26346 −1.61919Schwarz criterion −0.53021 −0.35791 −0.78294 −1.32305 −1.11905 −1.49860Hannan–Quinn criterion 0.61797 −0.44281 −0.86545 −1.39715 −1.20507 −1.57075DW statistic 1.88772 2.02254 1.94240 1.84106 1.89557 1.90265

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + β2i (ROAit) + εit.


(1998) that is statistically significant within the usual confidence intervals, in all periods that werestudied. This variation of Jones Model modified by Teoh et al. (1998) shows the better information criteriaamong the Jones Model modified by Teoh et al. (1998). This variation model also shows the betterinformation when it is compared to Jones Model modified by Dechow et al. (1995). The results of allvariations of Jones Model modified by Teoh et al. (1998) are demonstrated in Tables 7, 8, 9, 10, 11, 12, 13and 14.

The Jones Models modified by Kothari et al. (2005), without the variation of accounts receivable anduse of polynomials, and the KS Model with and without polynomials present better R2 and informationcriteria. According to the F statistic and p-value, both models and their versions are statistically significant,within the usual confidence intervals, in all periods that we studied. The results are demonstrated in theTables 15, 16, 17, 18, 19 and 20.

The Jones Model modified by Kothari et al. (2005), which uses polynomials and doesn't feature avariation in accounts receivable, and the polynomial KS Model present better information criterion whenthey are compared to the other models in this study.

Table 15Main general regression statistics of Eq. (6) for Jones Model modified by Kothari et al. (2005).


(ΔREV − ΔREC)/A 0.02076 0.03437 0.05388 −0.03395 −0.02324 0.04158PPE/A −0.17494** −0.24256** −0.13752** −0.02197 −0.09158** −0.07795**ROA 0.07621** 0.14528 0.16432 0.03630 0.00858 −0.01370Constant 0.05574 0.08231* 0.02187 −0.02479 0.03948* 0.01013R2 0.12099 0.11416 0.11682 0.00730 0.04331 0.06032Adjusted R2 0.10767 0.10139 0.10461 −0.00443 0.02924 0.04964F statistic 9.08445 8.93548 9.56795 0.62217 3.07860 5.64865Prob (F statistic) 0.00001** 0.00001** 0.00001** 0.60125 0.02857* 0.00092**Akaike info criterion −0.72354 −0.38957 −0.82111 −1.40242 −1.27347 −1.60293Schwarz criterion −0.65803 −0.32624 −0.75960 −1.34734 −1.20929 −1.54933Hannan–Quinn criterion −0.69703 −0.36397 −0.79627 −1.38027 −1.24752 −1.58140DW statistic 1.76791 2.14333 1.87131 1.80994 1.71794 1.96733

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + β3i (ROAit) + εit.

Table 16Main general regression statistics of Eq. (6) for the Polynomial Jones Model modified by Kothari et al. (2005).


(ΔREV − ΔREC)/A 0.03622 −0.01174 0.10979 −0.08421 −0.19746 −0.11637(ΔREV − ΔREC)/A2 0.24347 0.36352 0.10045 −0.09422 0.08960 0.03682(ΔREV − ΔREC)/A3 −0.10871 0.02051 −0.34956 0.14499 0.14101 0.25661(ΔREV − ΔREC)/A4 −0.04769 −0.12341 0.17639 0.11366 −0.05633 −0.12430PPE/A −2.19060* −2.12350** −0.80631 −0.73239** −0.40252 0.14855PPE/A2 6.64613* 4.83746* 2.04983 1.60526** 0.95933 −0.54443PPE/A3 −7.64270* −4.31590* −2.30275 −1.31880* -1.06957 0.45967PPE/A4 2.82639* 1.25675* 0.85805 0.32280* 0.41124 −0.10888ROA 0.28858 −0.42222 0.30699* 0.13032 0.16431 0.29484**ROA2 −0.45951 −0.72920 −1.51211** −0.23411 −1.53164* −0.45710ROA3 −1.09188 7.48134 −2.01420* 0.03697 −1.31759 −0.19816ROA4 −0.14014 5.58043 −0.53742 0.86646 1.13652* −0.01818Constant 0.18401 0.27320** 0.08329 0.05201 0.07820* −0.00988R2 0.22550 0.34371 0.25317 0.19476 0.11741 0.15177Adjusted R2 0.17632 0.30413 0.21008 0.15532 0.06310 0.11185F statistic 4.58560 8.68492 5.87578 4.93812 2.16170 3.80212Prob (F statistic) 0.00000** 0.00000** 0.00000** 0.00000** 0.01497* 0.00003**Akaike info criterion −0.76100 −0.60459 −0.90734 −1.54195 −1.26755 −1.63815Schwarz criterion −0.54810 −0.39876 −0.70745 −1.36293 −1.05895 −1.46396Hannan–Quinn criterion −0.67486 −0.52140 −0.82663 −1.46996 −1.18320 −1.56819DW statistic 1.81539 2.12172 1.89894 1.76536 1.74421 1.92334

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit - ΔRECit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + β3i (ROAit) + εit.


Thus, we used the residuals, per company and per period, as dependent variables (DA) in Eq. (12). Weobtained the residuals from the following models that had better results: (1) the polynomial KS Model(MKSP); (2) the polynomial Kothari Model (MKSRP) that doesn't include accounts receivable; (3) thepolynomial Kothari Model (MKP); (4) the polynomial Jones Model (MJP); and (5) the Jones Modelmodified by Teoh et al. (1998), which uses polynomials without a variation in accounts receivable andwith ROA (MTSRRP).

4.2. Results of the measurement of the effect of IFRS on earnings management

In Eq. (12), we used panel data analysis in order to follow the variables over extended periods. Wecame up with the definition of estimation methods among estimators for pooled data, fixed effects andrandom effects, based on the observations of Gujarati and Porter (2008: 587–610).

Table 17Main general regression statistics of Eq. (11) for Jones Model modified by Kothari et al. (2005) without the accounts receivablevariation.


ΔREV/A 0.03783 0.10455 0.12631* 0.04437 0.07376* 0.10168**PPE/A −0.1702** −0.22917** −0.13099** −0.02200 −0.09605** −0.07736**ROA 0.07541** 0.14073 0.15799 0.00496 −0.00179 −0.01123Constant 0.05167 0.06666* 0.00532 −0.02536 0.02837 0.00309R2 0.12339 0.13869 0.14720 0.01047 0.06595 0.09306Adjusted R2 0.11011 0.12626 0.13541 −0.00122 0.05222 0.08276F statistic 9.29008 1.11639 1.24856 0.89542 4.80154 9.03000Prob (F statistic) 0.00001** 0.00000** 0.00000** 0.44408 0.00297** 0.00001**Akaike info criterion −0.72627 −0.41764 −0.85611 −1.40562 −1.29742 −1.63840Schwarz criterion −0.66076 −0.35431 −0.79460 −1.35054 −1.23324 −1.58480Hannan–Quinn criterion −0.69977 −0.39205 −0.83128 −1.38347 −1.27147 −1.61687DW statistic 1.75759 2.16737 1.88766 1.83175 1.87475 1.95673

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + β3i (ROAit) + εit.

Table 18Main general regression statistics of Eq. (11) for the Polynomial Jones Model modified by Kothari et al. (2005) without the accountsreceivable variation.


ΔREV/A 0.10670 0.17688* 0.24731* 0.04812 0.21719* 0.09916ΔREV/A2 0.18094 0.11178 0.11982 0.20820 −0.08997 −0.06260ΔREV/A3 −0.23748 −0.02978 −0.52739 −0.17060 −0.11823* −0.00847ΔREV/A4 0.05759 −0.02958 0.24840 −0.17890 0.05357* 0.02458PPE/A −2.14087* −1.93884** −0.79382 −0.73519** −0.40827 0.10438PPE/A2 6.53130* 4.40388* 2.08824 1.71825** 0.85756 −0.47919PPE/A3 −7.52799* −3.88542* −2.33747 −1.46599** −0.81672 0.42740PPE/A4 2.78582* 1.11964* 0.85569 0.36245** 0.26369 −0.10354ROA 0.26030 −0.47598 0.26589* 0.10440 0.00555 0.21561ROA2 −0.48981 −0.76656 −1.42035* −0.13455 −1.48795* −0.30622ROA3 −1.10082 7.80683 −1.77852 0.10158 −1.67282 −0.06264ROA4 −0.14087 5.81505 −0.46164 0.59300 1.33752 −0.00184Constant 0.17439 0.23676** 0.05339 0.03626 0.05153 −0.01427R2 0.23169 0.36410 0.29067 0.19392 0.12757 0.15563Adjusted R2 0.18290 0.32576 0.24975 0.15444 0.07388 0.11590F statistic 4.74943 9.49534 7.10283 4.91159 2.37612 3.91678Prob (F statistic) 0.00000** 0.00000** 0.00000** 0.00000** 0.00703** 0.00002**Akaike info criterion −0.76903 −0.63616 −0.95886 −1.54090 −1.27913 −1.64272Schwarz criterion −0.55612 −0.43033 −0.75897 −1.36188 −1.07053 −1.46853Hannan–Quinn criterion −0.68289 −0.55297 −0.87815 −1.46892 −1.19478 −1.57275DW statistic 1.81068 2.08316 1.91477 1.77583 1.89553 1.86842

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = αit (1/Ait − 1) + β1i [(ΔREVit)/Ait − 1] + β2i (ΔPPEit/Ait − 1) + β3i (ROAit) + εit.


To comprise data on a short panel (six periods), therefore, the fixed effects model would be theindicated model. However, Eq. (12) consists of variables that do not vary in the course of time (CG, REG,HOLDING). Although the fixed effects model controls variables that do not change over time, it cannotestimate them directly. Unlike the fixed effects model, the random-effects model can estimate coefficientsthat do not change over time.

According to Table 21, it is possible to conclude that by using the Likelihood Ratio Test, with a p-valueequal to zero, there will be no redundancy of fixed effects. This proves that the estimators for the fixedeffects model are more appropriate than they are for the random effects model. We also note that in thespecification of the effects observed in the outputs of the regressions (random cross-section), the randomeffects model apparently has a greater proximity to the pooled data model than to the fixed effects model.Additionally, by using the Hausman Test, we observed that, with the exception of Eq. (12), with the use ofresiduals from Polynomial Jones Model (MJP) the random effects model is more appropriate. Thus,

Table 19Main general regression statistics of Eq. (7) for KS Model.


δ1 (REV) 0.10133 0.25117* 0.33393** 0.00786 0.29584* 0.16064*δ2 (EXPi) 0.00186** 0.09561 0.03053 -0.00914 0.01239 0.03869δ3 (GPPE) −1.71323* −1.67309** −0.40998 -1.48264** −0.57061** −0.84130**Constant 0.04421 0.02718 −0.05400 0.02598 -0.01402 −0.00958R2 0.15960 0.13478 0.07992 0.13876 0.14579 0.09043Adjusted R2 0.14687 0.12230 0.06720 0.12859 0.13322 0.08009F statistic 1.25341 1.08005 6.28301 1.36410 1.16054 8.74868Prob (F statistic) 0.00000** 0.00000** 0.00042** 0.00000** 0.00000** 0.00002**Akaike info criterion −0.76846 −0.41312 −0.78017 -1.54448 −1.38677 −1.63549Schwarz criterion −0.70295 −0.34979 −0.71867 -1.48940 −1.32258 −1.58190Hannan–Quinn criterion −0.74195 −0.38752 −0.75534 -1.52233 −1.36082 −1.61397DW statistic 1.76717 2.13727 1.88382 1.83506 1.83371 1.98454

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = θ0 + θ1 [δ1 (REVit/Ait − 1)] + θ2 [δ2 (EXPit/Ait − 1)] + θ3 [δ3 (GPPEit/Ait − 1)] + εit.

Table 20Main general regression statistics of Eq. (7) for the Polynomial KS Model.


δ1 (REV) −0.36094 1.46657 0.50104 0.26669 −0.16813 0.31753δ1 (REV)2 6.21644 −3.33099 −0.17901 −1.21546 4.21196 0.06275δ1 (REV)3 −1.70317 −0.34317 −1.11684 1.27007 −8.64619 −0.64322δ1 (REV)4 1.13850 1.55455 0.84040 −0.21315 4.32289 0.33194δ2 (EXP) −0.05297 0.02346 0.08276 −0.06801 0.04178 0.04588δ2 (EXP)2 0.00302* 0.24131 −0.05579 −0.05895 −0.00351 −0.28735δ2 (EXP)3 0.00050 0.15078 −0.02168 0.08190 −0.00214 −0.74032*δ2 (EXP)4 0.00000 0.01188 −0.00133 0.01635 −0.00012 0.71712δ3 (GPPE) −1.16735* −1.35410** 0.83519 −7.38873** −0.86974 1.76135δ3 (GPPE)2 1.89515 2.13670** −4.60502 1.18363* −6.92658** −8.95065δ3 (GPPE)3 −1.33735 −1.34773** 2.64820 −8.55598 9.27310* 1.04565δ3 (GPPE)4 2.98336 2.33647* −2.89349 2.05794 −1.93862* −3.91631Constant 0.17498* 0.11509 −0.04052 0.09523** -0.00023 −0.03921R2 0.30847 0.28902 0.18777 0.25875 0.23210 0.16697Adjusted R2 0.26456 0.24614 0.14091 0.22244 0.18485 0.12777F statistic 7.02544 6.74113 4.00712 7.12684 4.91163 4.25928Prob (F statistic) 0.00000** 0.00000** 0.00001** 0.00000** 0.00000** 0.00000**Akaike info criterion −0.87431 −0.52454 −0.82340 −1.62475 −1.40675 −1.65623Schwarz criterion −0.66141 0.31872 −0.62351 −1.44572 −1.19816 −1.48205Hannan–Quinn criterion −0.78817 −0.44135 −0.74269 −1.55276 −1.32241 −1.58627DW statistic 1.83145 2.20419 1.89122 1.82189 1.81932 2.04359

*Significant at 5%. **Significant at 1%.(TAit/Ait − 1) = θ0 + θ1 [δ1 (REVit/Ait − 1)] + θ2 [δ2 (EXPit/Ait − 1)] + θ3 [δ3 (GPPEit/Ait − 1)] + εit.


considering the observations of Gujarati and Porter (2008) and the tests we performed, we decided toestimate the coefficients of Eq. (12) using the random effects model.

Table 22 shows the main statistics that we obtained by using Eq. (12). We found the absolute values ofresiduals by using selected models.

In Table 22, we see that, when the p-value is equal to zero, it can be verified that the independentvariables are capable of explaining the behavior of the dependent variable (earnings management throughdiscretionary appropriations).

The models have very similar values for the adjusted R2 category, with MKP having the greatest value,followed by MKSRP.

The categorical variable IFRS is statistically significant at 1% for all of the selected models except MKSP,for which it is significant at the 5% level. As expected, IFRS's coefficient is negative. This suggests that in the

Table 21Analysis of fixed or random effects in model of pooled data.

Model Fixed effects vs.pooling

Random effects vs. fixed effects Conclusion–estimationof Most appropriatemethod

p-ValueLikelihoodratio Test⁎

Cross-sectionrandom⁎⁎

Adjusted R2 p-Value Durbin-Watson p-Value(HausmannTest)⁎⁎⁎

MJP 0.00000 0.14200 0.04621 0.00000 1.85653 0.00910 FixedMKP 0.00000 0.16810 0.04089 0.00000 1.85860 0.59620 RandomMKSP 0.00000 0.14900 0.03735 0.00000 1.90717 0.22140 RandomMKSRP 0.00000 0.16550 0.04046 0.00000 1.86603 0.47590 RandomMTSRRP 0.00000 0.19360 0.03187 0.00000 1.81133 0.13920 Random

⁎ Likelihood ratio test (TLR); if amount is b0.05, then the fixed effects model is better.⁎⁎ The closer to 1, the random model approaches the fixed; the closer to 0, the closest estimation by pooled data.

⁎⁎⁎ Hausmann Test (TH); if N0, the random effects model is better, with residues uncorrelated with the explanatory variables.Otherwise, the fixed effect model is better, with correlated with the explanatory variable residues.

Table 22Main general regression statistics showing impact of IFRS on Earnings Management Model.

Independent variable Expected sign MJP MKP MKSP MKSRP MTSRRP

Coeff. Coeff. Coeff. Coeff. Coeff.

IFRS − −0.041** −0.051** -0.031* −0.045** −0.038**HIB − -0.017 −0.024* -0.011 −0.023* −0.019AUDIT − −0.030* −0.031** -0.016 −0.027* −0.015IFRSXAUDIT − 0.028 0.043** 0.010 0.032* 0.018HIBXAUDIT − 0.014 0.020 0.000 0.018 0.011GC − 0.018* 0.019* 0.015 0.018* 0.025**REG − −0.028** −0.028** −0.028** −0.029** −0.031**HOLDING − -0.005 −0.001 −0.011 −0.002 −0.005LEV + 0.002* −0.001 0.000 −0.001 −0.001SIZE − −0.008** −0.007** −0.006** −0.007** −0.006**C N/A 0.221** 0.214** 0.194** 0.211** 0.193**R2 0.053 0.048 0.044 0.047 0.039Adjusted R2 0.046 0.041 0.037 0.040 0.032F statistic 7.628 6.833 6.308 6.768 5.503Prob(F statistic) 0.000** 0.000** 0.000** 0.000** 0.000**Durbin–Watson statistic 1.857 1.859 1.907 1.866 1.811

*Significant at 5%. **Significant at 1%.|DAit| = α + β1iIFRSit + β2iHIBit + β3iAUDITit + β4i(IFRSitxAUDITit) + β5i(HIBit xAUDITit) + β6iGCit + β7iREGit + β8iHOLDINGit +β9iSIZEit + β10iLEVit + εit.


Full-IFRS Period there is less earnings management through discretionary appropriations than in otherperiods. This result is consistent with Hypothesis 1.

It can be observed in the statistics for the HIB variable, although the sign of the coefficient is negative, asexpected, not enough statistical significance is shown to say that this result is conclusive. Except whenusing the residues of MKP and MKSRP, significant at the level of 5%, the other models are not statisticallysignificant. Thus, nothing can be stated about Hypothesis 2, referring to reduction of earnings manage-ment through discretionary appropriations in Hybrid Period.

The categorical variable identifying the presence of a Big Four audit (AUDIT) is statistically significantwithin the confidence interval of 1% based on using discretionary appropriations estimated by using MKPand 5% based on MJP and MKSRP. The coefficients that we obtained from the residuals are negative, asexpected. The results suggest that the Big Four audit is a limiting factor to earnings management throughdiscretionary appropriations, as suggested in Hypothesis 3, but the significance levels are not high enoughfor this to be a conclusive result.

The interaction between categorical variables, IFRS with AUDIT and AUDIT with HIB, also shows noconclusive results. Therefore, we cannot assert that the Big Four audit has stricter efficiency to avoidearnings management through discretionary appropriations in the Full-IFRS and Hybrid periods than itdoes in the Pre-IFRS Period.

According to Table 22, the GC variable is statistically significant at the 5% level in all models except inMTSRRP, in which it is significant at 1%. Unlike what we expected to find, the coefficients are all positive forthis variable. This suggests that companies that the Brazilian Securities and Exchange Commission has given agood GC rating have greater earnings management through discretionary appropriations. However, thestatistical levels that were obtained are not a sufficient basis on which to reject the Hypothesis 4, whichsuggests that entities with higher levels of GC present lower levels of earnings management.

The coefficients of REG are negative for all of the models and statistically significant at the level of 1%.This result confirms Hypothesis 5: Entities of a regulated sector of the economy have a lower degree ofearnings management through discretionary appropriations when they are compared with others.

Additionally, we observe that the earnings management of a subsidiary in Brazil is not affected by theholding company having adopted IFRS earlier, as demonstrated by the categorical variable HOLDINGshowing no significant statistical results.

The SIZE variable, which indicates how large a company is, shows statistical significance at the 1% levelin all models and has negative coefficients, as expected. Thus, it is possible to conclude that largercompanies have less earnings management through discretionary appropriations.


Finally, the categorical variable LEV, which indicates the level of liabilities, is not statistically significant.The level of financial leverage cannot be considered relevant in terms of effect on earnings managementthrough discretionary appropriations.

We found that the convergence to IFRS had a restrictive effect on earnings management in Brazilafter the full implementation of IFRS (the Full-IFRS Period), as observed in previous studies (Aubert andGrudnitski, 2012; Barth et al., 2012; Leventis et al., 2011; Zéghal et al., 2011). Thus, as expected, when weconsider earnings management through discretionary appropriations, we found evidence of theimprovement of the quality of accounting information in Brazil in the IFRS Period. However, we do notfind enough evidence to affirm a decrease in earnings management in the Hybrid Period when comparedto the Pre-IFRS Period.

We do not find sufficient evidence that being audited by a Big Four firm presents a restriction tomanagement results. This confirms previous findings by Martinez (2009), Martinez and Reis (2010) andSilva and Bezerra (2010) and international research done by Tsipouridou and Spathis (2012). Theseresults, while not conclusive, suggest that there is better quality accounting information when the entity isaudited by a Big Four company, as already observed by Ghosh et al. (2010), Jouber and Fakhfakh (2012)and Almeida and Almeida (2009).

The analysis of the effects of corporate governance in earnings management produces sufficientevidence to affirm the influence of corporate governance on earnings management through discretionaryappropriations. However, we highlight the results suggesting that firms with higher levels of governancehave higher levels of earnings management, which is contrary to what was expected. This confirms theprevious findings of G.E. Iatridis (2012) and Price et al. (2011) but contrasts the findings of Martinez(2009) and Almeida dos Santos et al. (2011) and international studies, such as A. Chen et al. (2010).

According to a study of Brazilian companies done by G.E. Iatridis (2012), there is not enough presenceof corporate governance mechanisms to reduce agency cost, including the reduction of earningsmanagement in Brazil. In accordance with this work, we found that the entities that the Brazilian SecurityExchange Commission finds to have the best corporate governance practices are those with higher levelsof earnings management through discretionary appropriations.

Among the restrictive factors to earnings management that we studied—getting audited by a Big Fourfirm, corporate governance and regulatory environment—we found that the most effective is theregulatory environment. Entities that are regulated by an agency beyond Brazilian Security ExchangeCommission have lower levels of earnings management than the others.

Additionally, contrary to the previous findings (Rodríguez-Pérez and Van Hemmen, 2010; Nardi et al.,2009), we found no evidence of the influence of financial leverage to earnings management throughdiscretionary appropriations.

We also noted that the fact that the parent companies had already adopted IFRS earlier does notinfluence the level of earnings management through discretionary appropriations of their subsidiaries inBrazil.

We found evidence that the larger companies have low level of earnings management.Cavalier-Rosa and Tiras (2013) observed that some aspects of Brazil environment cause opportunities

for earnings management. We found that the IFRS adoption represented an advance in accounting quality,considering the reduction of earnings management. This is an advancement that improves the investortrust in Brazil and could cause more influx of foreign investments, especially in regulated sectors studied(public utility; telecom; oil, gas and biofuel).

5. Conclusion

This study has sought to evaluate whether changes in accounting practices brought a reduction inearnings management of Brazilian non-financial listed companies resulting in the improvement of qualityin accounting information.

We found that the convergence to IFRS had a restrictive effect on earnings management in Brazil afterthe full implementation of IFRS. Additionally, among the restrictive factors to earnings management thatwe studied—getting audited by the Big Four firms, corporate governance and regulatory environment—wefound that the most effective is the regulatory environment. These findings suggest that those legislativelaw countries, like Brazil, present less earnings management in the sectors with more enforcement


influence. In conclusion, to constrain earnings management in legislative law countries, it is necessary tohave effective enforcement.

For investors in public company, the CVM enforcement is deemed to be enough for reducing theearnings management after IFRS implementation. Although, it is important to highlight that theimprovement in accounting quality is more efficient when considering enforcement by another regulatoryagency besides CVM.

For investors in entities without CVM regulation or other regulatory agency, there remains the doubt ifIFRS will cause improvement in accounting quality with less earnings management. Therefore, we suggestmore researches for measuring if there is a reduction in earnings management after IFRS in legislativecountries using a comparison among entities that have some enforcement with those that do not haveenforcement for regulatory agency. In addition, we suggest more researches for comparing earningsmanagement among countries with the legislative law and customary law structures, considering thevariable of enforcement for verifying the enforcement power in this different kinds of law culture.

The process of convergence to IFRS in Brazil is recent, then, this work was limited to short coverageperiods of these standards. Considering this limitation, we suggest that future research examiningearnings management in Brazil looks at a bigger period in order to see whether there is continuingevidence of the reduction in earnings management through discretionary appropriations in the light ofIFRS in Brazil.

References

Ahn, S., Choi, W., 2009. The role of bank monitoring in corporate governance: evidence from borrowers' earnings managementbehavior. J. Bank. Financ. 33 (2), 425–434.

Almeida dos Santos, P.S., Verhagem, J.A., Bezerra, F.A., 2011. Gerenciamento de resultados por meio de decisões operacionais e agovernança corporativa: análise das indústrias siderúrgicas e metalúrgicas brasileiras. Rev. Contab. Organ. 5 (13), 55–74.

Almeida, J., Almeida, J., 2009. Auditoria e earnings management: estudo empírico nas empresas abertas auditadas pelas big four edemais firmas de auditoria. Rev. Contab. Finanç. 20 (50), 62–74 (USP, São Paulo).

Almeida, J.E.F.D., Costa, F.M.D., Faria, L.H.L., Brandrão, M.M., 2009. Earnings Management no Brasil: grupos estratégicos como novavariável explanatória. Rev. Contab. Organ. 3 (6), 81–95.

Alves Oliveira, V.A., Lemes, S., 2011. Nível de convergência dos princípios contábeis brasileiros e norte-americanos às normas doIASB: uma contribuição para a adoção das IFRS por empresas brasileiras. Rev. Contab. Finanç. 22 (56), 155–173.

Aubert, F.,Grudnitski, G., 2012. Analysts' estimates: what they could be telling us about the impact of IFRS on earnings manipulationin Europe. Rev. Account. Financ. 11 (1), 53–72.

Ayers, B.C.,Ramalingegowda, S.,Eric Yeung, P., 2011. Hometown advantage: the effects of monitoring institution location on financialreporting discretion. J. Account. Econ. 52 (1), 41–61.

Balachandran, B., Chalmers, K., Haman, J., 2008. On‐market share buybacks, exercisable share options and earnings management.Account. Financ. 48 (1), 25–49.

Ball, R., Shivakumar, L., 2008. Earnings quality at initial public offerings. J. Account. Econ. 45 (2), 324–349.Barbosa Neto, J.E.,de Oliveira Dias, W.,Pinheiro, L.E.T., 2009. Impacto da Convergência para as IFRS na Análise Financeira: um estudo

em empresas brasileiras de capital aberto. Contab. Vista Rev. 20 (4), 131–153.Barth, M.E., Landsman, W.R., Lang, M., Williams, C., 2012. Are IFRS-based and US GAAP-based accounting amounts comparable? J.

Account. Econ. 54 (1), 68–93.Baxter, P., Cotter, J., 2009. Audit committees and earnings quality. Account. Financ. 49 (2), 267–290.Bentwood, S., Lee, P., 2012. Benchmark management during Australia's transition to international accounting standards. Abacus 48

(1), 59–85.Bona‐Sánchez, C., Pérez‐Alemán, J., Santana‐Martín, D.J., 2011. Defence measures and earnings management in an owner dominant

context. J. Bus. Financ. Account. 38 (7/8), 765–793.Botsari, A.,Meeks, G., 2008. Do acquirers manage earnings prior to a share for share bid? J. Bus. Financ. Account. 35 (5/6), 633–670.Bozec, Y., 2008. Ownership concentration, separation of voting rights from cash flow rights, and earnings management: an empirical

study in Canada. Can. J. Adm. Sci. 25 (1), i–xv.Cahan, S., Emanuel, D., Hay, D., Wong, N., 2008. Non‐audit fees, long‐term auditor–client relationships and earnings management.

Account. Financ. 48 (2), 181–207.Caramanis, C., Lennox, C., 2008. Audit effort and earnings management. J. Account. Econ. 45 (1), 116–138.Cavalier-Rosa, G.,Tiras, S.L., 2013. Brazil's adoption of IFRS: fertile ground for examining earnings management. Braz. Bus. Rev. 10 (4),

133–146.Chan, K., Ikenberry, D.L., Lee, I.,Wang, Y., 2010. Share repurchases as a potential tool to mislead investors. J. Corp. Financ. 16 (2), 137–158.Chang, S.C.,Chung, T.Y.,Lin, W.C., 2010. Underwriter reputation, earnings management and the long‐run performance of initial public

offerings. Account. Financ. 50 (1), 53–78.Chen, A.,Kao, L.,Tsao, M., 2010. To improve investors' valuation of accounting earnings in emerging markets: evidence from Taiwan.

Can. J. Adm. Sci. 27 (4), 376–390.Chen, A.S., Cheng, L.Y., Cheng, K.F., Chih, S.W., 2010. Earnings management, market discounts and the performance of private equity

placements. J. Bank. Financ. 34 (8), 1922–1932.Chen, J.,Liao, W.M.,Lu, C., 2012. The effects of public venture capital investments on corporate governance: evidence from IPO firms

in emerging markets. Abacus 48 (1), 86–103.

http://refhub.elsevier.com/S1566-0141(14)00040-5/rf0005












































Cheng, P.,Aerts, W.,Jorissen, A., 2010. Earnings management, asset restructuring, and the threat of exchange delisting in an earnings‐based regulatory regime. Corp. Gov. Int. Rev. 18 (5), 438–456.

Chi, J.D.,Gupta, M., 2009. Overvaluation and earnings management. J. Bank. Financ. 33 (9), 1652–1663.Choi, T.H., Pae, J., 2011. Business ethics and financial reporting quality: evidence from Korea. J. Bus. Ethics 103 (3), 403–427.Christie, A.A., Zimmerman, J.L., 1994. Efficient and opportunistic choices of accounting procedures: corporate control contests.

Account. Rev. 69 (4), 539–566.Chung, H.,Sheu, H.J.,Wang, J.L., 2009. Do firms' earningsmanagement practices affect their equity liquidity? Financ. Res. Lett. 6 (3), 152–158.Coelho, A.C.D., Lopes, A.B., 2007. Avaliação da prática de gerenciamento de resultados na apuração de lucro por companhias abertas

brasileiras conforme seu grau de alavancagem financeira. Rev. Adm. Contemp. 11, 121–144.Cohen, D.A.,Zarowin, P., 2010. Accrual-based and real earnings management activities around seasoned equity offerings. J. Account.

Econ. 50 (1), 2–19.Cornett, M.M., Marcus, A.J., Tehranian, H., 2008. Corporate governance and pay-for-performance: the impact of earnings

management. J. Financ. Econ. 87 (2), 357–373.Cupertino, C.M.,Martinez, A.L., 2008. Qualidade da auditoria e earnings management: risk assessment através do nível dos Accruals

Discricionários. Contab. Vista Rev. 19 (3), 69–93.Daniel, N.D.,Denis, D.J.,Naveen, L., 2008. Do firms manage earnings to meet dividend thresholds? J. Account. Econ. 45 (1), 2–26.DeAngelo, L.E., 1986. Accounting numbers as market valuation substitutes: a study of management buyouts of public stockholders.

Account. Rev. 61 (3), 400–420.Dechow, P.M.,Dichev, I.D., 2002. The quality of accruals and earnings: the role of accrual estimation errors. Account. Rev. 77, 35–59.Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting earnings management. Account. Rev. 70 (2), 193–225.Degeorge, F., Patel, J., Zeckhauser, R., 1999. Earnings management to exceed thresholds*. J. Bus. 72 (1), 1–33.Ettredge, M.,Scholz, S., Smith, K.R., Sun, L., 2010. How do restatements begin? Evidence of earnings management preceding restated

financial reports. J. Bus. Financ. Account. 37 (3/4), 332–355.Fernandes, P.F.M., 2011. O impacto da entrada em vigor das IFRS na gestão de resultados: a experiência Ibérica(Dissertation of the

Master) Faculty of Economy of the University of Porto.Fischer, P.E.,Louis, H., 2008. Financial reporting and conflicting managerial incentives: the case of management buyouts. Manag. Sci.

54 (10), 1700–1714.Formigoni, H.,Paulo, E.,Antunes, M.T.P., 2009. Diferença entre o lucro contábil e lucro tributável: uma análise sobre o gerenciamento

de resultados contábeis e gerenciamento tributário nas companhias abertas brasileiras. Braz. Bus. Rev. 6 (1), 44–61.Gargouri, R.M., Shabou, R., Francoeur, C., 2010. The relationship between corporate social performance and earnings management.

Can. J. Adm. Sci. 27 (4), 320–334.Ghosh, A.,Marra, A.,Moon, D., 2010. Corporate boards, audit committees, and earnings management: pre‐and post‐SOX evidence. J.

Bus. Financ. Account. 37 (9/10), 1145–1176.Gong, G., Louis, H., Sun, A.X., 2008. Earnings management, lawsuits, and stock-for-stock acquirers' market performance. J. Account.

Econ. 46 (1), 62–77.Guenther, D.A., 1994. Earnings management in response to corporate tax rate changes: evidence from the 1986 Tax Reform Act.

Account. Rev. 69 (1), 230–243.Gujarati, D.N., Porter, D.C., 2008. Econometria Básica, 5th edition. McGraw-Hill, New York.Guthrie, K., Sokolowsky, J., 2010. Large shareholders and the pressure to manage earnings. J. Corp. Financ. 16 (3), 302–319.Hadani, M.,Goranova, M.,Khan, R., 2011. Institutional investors, shareholder activism, and earnings management. J. Bus. Res. 64 (12),

1352–1360.Hazarika, S.,Karpoff, J.M.,Nahata, R., 2012. Internal corporate governance, CEO turnover, and earnings management. J. Financ. Econ.

104 (1), 44–69.Healy, P.M., 1985. The effect of bonus schemes on accounting decisions. J. Account. Econ. 7 (1), 85–107.Herbohn, K.,Ragunathan, V., 2008. Auditor reporting and earnings management: some additional evidence. Account. Financ. 48 (4),

575–601.Hutton, A.P.,Marcus, A.J., Tehranian, H., 2009. Opaque financial reports, R 2, and crash risk. J. Financ. Econ. 94 (1), 67–86.Iatridis, G., 2012a. Hedging and earnings management in the light of IFRS implementation: evidence from the UK stock market. Br.

Account. Rev. 44 (1), 21–35.Iatridis, G.E., 2012b. Audit quality in common-law and code-law emerging markets: evidence on earnings conservatism, agency costs

and cost of equity. Emerg. Mark. Rev. 13 (2), 101–117.Ibrahim, S.S., 2009. The usefulness of measures of consistency of discretionary components of accruals in the detection of earnings

management. J. Bus. Financ. Account. 36 (9/10), 1087–1116.Jaggi, B.,Leung, S.,Gul, F., 2009. Family control, board independence and earnings management: evidence based on Hong Kong firms.

J. Account. Public Policy 28 (4), 281–300.Jeanjean, T., Stolowy, H., 2008. Do accounting standards matter? An exploratory analysis of earnings management before and after

IFRS adoption. J. Account. Public Policy 27 (6), 480–494.Jiang, J.,Petroni, K.R.,Wang, I.Y., 2010. CFOs and CEOs: who have the most influence on earnings management? J. Financ. Econ. 96 (3),

513–526.Jones, J.J., 1991. Earnings management during import relief investigations. J. Account. Res. 29 (2), 193–228.Jorion, P., Shi, C., Zhang, S., 2009. Tightening credit standards: the role of accounting quality. Rev. Acc. Stud. 14 (1), 123–160.Jouber, H., Fakhfakh, H., 2012. Earnings management and board oversight: an international comparison. Manag. Audit. J. 27 (1),

66–86.Kang, S.H.,Sivaramakrishnan, K., 1995. Issues in testing earnings management and an instrumental variable approach. J. Account. Res.

33 (2), 353–367.Kang, W.S., Kilgore, A.,Wright, S., 2011. The effectiveness of audit committees for low-and mid-cap firms. Manag. Audit. J. 26 (7),

623–650.Kim, J.B.,Yi, C.H., 2009. Does auditor designation by the regulatory authority improve audit quality? Evidence from Korea. J. Account.

Public Policy 28 (3), 207–230.Kothari, S.P., Leone, A.J.,Wasley, C.E., 2005. Performance matched discretionary accrual measures. J. Account. Econ. 39 (1), 163–197.








































































Labelle, R., Gargouri, R.M., Francoeur, C., 2010. Ethics, diversity management, and financial reporting quality. J. Bus. Ethics 93 (2),335–353.

Lee, G.,Masulis, R.W., 2011. Do more reputable financial institutions reduce earnings management by IPO issuers? J. Corp. Financ. 17(4), 982–1000.

Leventis, S., Dimitropoulos, P.E., Anandarajan, A., 2011. Loan loss provisions, earnings management and capital management underIFRS: the case of EU commercial banks. J. Financ. Serv. Res. 40 (1/2), 103–122.

Martinez, A.L., 2001. “Gerenciamento” dos resultados contábeis: estudo empírico das companhias abertas brasileiras(Doctoralthesis) Faculty of Economy, Business and Accounting of the University of Sao Paulo.

Martinez, A.L., 2008. Detectando earnings management no Brasil: estimando os accruals discricionários. Rev. Contab. Finanç. 7–17.Martinez, A.L., 2009. Novo mercado, auditoria e o gerenciamento de resultados por escolhas contábeis e por decisões operacionais no

Brasil. 9th Congress of Accounting and Controlling of University of São Paulo.Martinez, A.L., 2011. The role of analysts as gatekeepers: enhancing transparency and curbing earnings management in Brazil. Rev.

Adm. Contemp. 15 (4), 712–730.Martinez, A.L., Reis, G.M.R., 2010. Rodízio das firmas de auditoria e o gerenciamento de resultados no Brasil. Rev. Contab. Organ. 4

(10), 48–64.Martins, O.S.,Brasil, A.M.S., 2008. A Contabilidade Internacional e a Convergência às Normas Internacionais de Contabilidade do IASB.

Qual. Rev. Eletrônica 7 (2).Martins, O.S.,Paulo, E., 2010. Reflexo da adoção das IFRS na análise de desempenho das companhias de capital aberto no Brasil. Rev.

Contab. Organ. 4 (9), 30–54.Matsumoto, A.S., Parreira, E.M., 2007. Uma pesquisa sobre o Gerenciamento de Resultados Contábeis: causas e consequências.

Contab. Gest. Gov. 10 (1).McInnis, J., Collins, D.W., 2011. The effect of cash flow forecasts on accrual quality and benchmark beating. J. Account. Econ. 51 (3),

219–239.Mora, A., Sabater, A., 2008. Evidence of income-decreasing earnings management before labour negotiations within firms. Investig.

Econ. 32 (2), 201–230.Nardi, P.C.C.,Nakao, S.H., 2009. Gerenciamento de resultados e a relação com o custo da dívida das empresas brasileiras abertas. Rev.

Contab. Finanç. 20 (51), 77–100.Nardi, P.C.C., da Silva, R.L.M., Nakao, S.H., do Valle, M.R., 2009. A Relação entre gerenciamento de resultados contábeis e o custo de

capital das companhias abertas brasileiras. Rev. Universo Contábil 5 (4), 6–26.Nguyen, V.T., Xu, L., 2010. The impact of dual class structure on earnings management activities. J. Bus. Financ. Account. 37 (3/4),

456–485.Niskanen, J., Karjalainen, J., Niskanen, M., Karjalainen, J., 2011. Auditor gender and corporate earnings management behavior in

private Finnish firms. Manag. Audit. J. 26 (9), 778–793.Paulo, E.P., Leme, J.R., 2009. Gerenciamento de resultados contábeis eo anúncio dos resultados contábeis pelas companhias abertas

brasileiras. Rev. Universo Contábil 5 (4), 27–43.Paulo, E.,Martins, E., Corrar, L.J., 2007. Detecção do gerenciamento de resultados pela análise do diferimento tributário. Rev. Adm.

Empres. 47 (1), 46–59.Price, R.,Román, F.J.,Rountree, B., 2011. The impact of governance reform on performance and transparency. J. Financ. Econ. 99 (1), 76–96.Prior, D., Surroca, J., Tribó, J.A., 2008. Are socially responsible managers really ethical? Exploring the relationship between earnings

management and corporate social responsibility. Corp. Gov. Int. Rev. 16 (3), 160–177.Rayburn, J., 1986. The association of operating cash flow and accruals with security returns. J. Account. Res. 24, 112–133.Rezende, G.P.,Nakao, S.H., 2012. Gerenciamento de resultados ea relação com o lucro tributável das empresas brasileiras de capital

aberto. Rev. Universo Contábil 8 (1), 6–21.Rodríguez-Pérez, G., Van Hemmen, S., 2010. Debt, diversification and earnings management. J. Account. Public Policy 29 (2),

138–159.Santos, A.D.,Grateron, I.R.G., 2003. Contabilidade criativa e responsabilidade dos auditores. Rev. Contab. Finanç. 14 (32), 7–22.Sawicki, J., Shrestha, K., 2008. Insider trading and earnings management. J. Bus. Financ. Account. 35 (3/4), 331–346.Shelton, S.W., Owens-Jackson, L.A., Robinson, D.R., 2011. IFRS and US GAAP: assessing the impact of reporting incentives on firm

restatements in foreign and US markets. Adv. Account. 27 (1), 187–192.Silva, J.O., Bezerra, F.A., 2010. Análise do gerenciamento de resultados eo rodízio de firmas de auditoria nas empresas de capital

aberto. Rev. Bras. Gest. Neg. 12 (36), 304–321.Sun, J., Liu, G., Lan, G., 2011. Does female directorship on independent audit committees constrain earnings management? J. Bus.

Ethics 99 (3), 369–382.Teoh, S.H.,Welch, I.,Wong, T.J., 1998. Earnings management and the underperformance of seasoned equity offerings. J. Financ. Econ.

50 (1), 63–99.Tsipouridou, M.,Spathis, C., 2012. Earnings management and the role of auditors in an unusual IFRS context: the case of Greece. J. Int.

Account. Audit. Tax. 21 (1), 62–78.United Nations Conference on Trade and Development, 2013. Global Investment Trends MonitorAvailable at: http://unctad.org/en/

PublicationsLibrary/webdiaeia2013d6_en.pdf.Vasconcelos, C., Reis, D., Miyashiro, I., Paulo, E., 2008. A influência da cobertura das empresas de rating sobre o gerenciamento de

resultados das companhias abertas brasileiras. Rev. Contab. Organ. 2 (3), 37–52.Wang, Y.,Campbell, M., 2012. Corporate governance, earnings management, and IFRS: empirical evidence from Chinese domestically

listed companies. Adv. Account. 28 (1), 189–192.Wang, L., Yung, K., 2011. Do state enterprises manage earnings more than privately owned firms? The case of China. J. Bus. Financ.

Account. 38 (7/8), 794–812.Watts, R., Zimmerman, J., 1986. Positive Accounting Theory. Prentice Hall.Wilson, M., Wang, L.W., 2010. Earnings management following chief executive officer changes: the effect of contemporaneous

chairperson and chief financial officer appointments. Account. Financ. 50 (2), 447–480.Wilson, M.,Wu, Y.A., 2011. Do publicly signalled earnings management incentives affect analyst forecast accuracy? Abacus 47 (3),

315–342.


























































http://unctad.org/en/PublicationsLibrary/webdiaeia2013d6_en.pdf

http://unctad.org/en/PublicationsLibrary/webdiaeia2013d6_en.pdf













Yu, F.F., 2008. Analyst coverage and earnings management. J. Financ. Econ. 88 (2), 245–271.Zéghal, D.,Chtourou, S., Sellami, Y.M., 2011. An analysis of the effect of mandatory adoption of IAS/IFRS on earnings management. J.

Int. Account. Audit. Tax. 20 (2), 61–72.Zhao, Y.,Chen, K.H., 2008. The influence of takeover protection on earnings management. J. Bus. Financ. Account. 35 (3/4), 347–375.Zonatto, V.C.D.S.,Santos, A.C.D., Junior, M.M.R.,Bezerra, F.A., 2011. Fatores determinantes para a adoção de padrões internacionais de

contabilidade no Brasil: uma investigação em empresas públicas e privadas do setor de energia elétrica. Rev. Contab. Organ. 5(12), 26–47.








The effect of IFRS on earnings management in Brazilian non-financial public companies

Documents

Transcript of The effect of IFRS on earnings management in Brazilian non-financial public companies