A Note on Competitiveness and Structural Transformation in Pakistan

8/22/2019 A Note on Competitiveness and Structural Transformation in Pakistan

1/46

Economics and REsEaRch dEpaRtmEnt

a ne cevee

sruurl trfr

pk

Jesus Felipe

December 2007

RD WoRking PaPER SERiES no. 110


2/46


3/46

ERD Wrking Paper N. 110

A NoteoN CompetitiveNessANd struCturAltrANsformAtioNiN pAkistAN

Jesus felipe

deCember 2007

Jesus Felipe is Principal Economist at the Central and West Asia Department, Asian Development Bank. This paper

represents the authors views and does not represent those o the Asian Development Bank, its Executive Directors,or the countries they represent. Participants at seminars given at the Pakistan Institute o Development Economics,

Pakistans Planning Commission, Planning and Development Board o Punjabs government, and State Bank o

Pakistan made very useul comments and suggestions. The author also benefted rom conversations with ShujatAli, Ashaque H. Khan, Shaukat Hameed Khan, Aqdas Ali Kazmi, Sadar Parvez, and Norio Usui. Gemma Estrada and

Raquel Rago provided excellent research assistance. Any remaining errors are solely the authors.


4/46

Asian Development Bank6 ADB Avenue, Mandaluyong City

1550 Metro Manila, Philippineswww.adb.org/economics

2007 by Asian Development BankDecember 2007

ISSN 1655-5252

The views expressed in this paperare those o the author(s) and do notnecessarily reect the views or policies

o the Asian Development Bank.


5/46

FoREWoRD

The ERD Working Paper Series is a orum or ongoing and recently completed

research and policy studies undertaken in the Asian Development Bank or onits behal. The Series is a quick-disseminating, inormal publication meant tostimulate discussion and elicit eedback. Papers published under this Seriescould subsequently be revised or publication as articles in proessional journals

or chapters in books.


6/46


7/46

CoNtENts

Abstract vii

I. IntroductionI. Introduction 1

II. Structural Transormation in PakistanII. Structural Transormation in Pakistan 5

III. Exports and Structural Transormation 1III. Exports and Structural Transormation 19

I. The Product Space n Monkeys and orests 2I. The Product Space n Monkeys and orests 23

. Industrial Policy as a Central Part o Pakistans. Industrial Policy as a Central Part o PakistansDevelopment Strategy 24

I. Conclusions 2I. Conclusions 27

Selected Reerences 34


8/46


9/46

AbstRACt

This paper examines Pakistans economic prospects by analyzing the degreeo structural transormation during the last three decades and by comparing itwith that o other countries in Asia. This way, it intends to contribute to theliterature on Pakistans growth. The discussion is important in order to raise

awareness among policymakers so that sound policies or economic transormationare implemented. Pakistans degree o structural transormation has been lowerand slower than that o other countries in Asia. Accelerating the rate o structural

transormation should be a key objective o Pakistans policymakers. Sector-specifc reorms that oster competition should be viewed rom the point o viewo transorming the economy. This requires a dose o good policy, as the marketalone cannot accomplish this. The paper argues that a well-developed plan in the

orm o industrial policy, understood as strategic collaboration between publicand private sectors, is an important tool.


10/46


11/46

Progress is impossible without change,

and those who cannot change their mindscannot change anything.

George Bernard Shaw

I. INtRoDUCtIoN

The successul Asian economies (i.e., the newly industrializing economies [NIEs];1Malaysia;Thailand; or the Peoples Republic o China [PRC] and iet Nam more recently) present two maindierences between three decades ago and today. The frst one is about size they are signifcantlybigger. The second one relates to their structure they are signifcantly dierent.Indeed, what sets

the perormance o the successul Asian economies apart rom that o many other countries is notonly the ormers high growth rates but also their ast and intense structural transormation theiroutput and employment structures have changed dramatically; resources have been transerred to

higher value-added sectors (i.e., rom agriculture to industry and services); production has diversifed;frms have learned how to produce and export a more sophisticated and technologically advancedrange o products; and their levels o labor productivity have increased signifcantly.

Growth and structural transormation are related in a circular way, since countries do notgrow by simply reproducing themselves on a larger scale. Generally (unless all sectors o theeconomy grow at identical rates), countries become dierent as they grow, not only in terms o

whatthey produce, but also howthey do it, i.e., by using dierent inputs including new methods

o production. The successul economies have generated new activities characterized by higherproductivity and increasing returns to scale. The transition across dierent patterns o productionand specialization involves upgrading to higher value-added activities within each sector throughthe introduction o new products and processes. These changes entail ar-reaching transormations

in terms o, among other things, economic geography and skill content o output. It is the countriesthat can sustain multiple transitions across dierent stages o their structural transormation thatgrow successully.

This paper is limited in scope and does not delve into the many intricacies and causal relationsunderlying Pakistans development constraints and causal relations. Rather, it attempts to providea descriptive analysis o Pakistans economy and its prospects rom the viewpoint o its structural

transormation during the last three decades, and compares this with the experience o other countriesin Asia. It intends to contribute to the literature on Pakistans growth (see or example, Easterly2003 and Haque 2006). Haque (2006, 2) has argued that the lack o debate on this subject has

1 Hong Kong, China; Republic o Korea; Singapore; Taipei,China.


12/46

December 2007

A NoteoNcompetitiveNessAND structurAl trANsformAtioNiNpAkistAN

Jesusfelipe

kept Pakistans growth strategy uninormed o the new research on economic growth.2 Thereore,an update and discussion is important in order to raise awareness among policymakers so thatsound policies are implemented.

During the last our decades, Pakistan has grown at an average rate o slightly above 5% per

annum (although with requent boom and bust episodes) (igure 1).3 The average growth rateduring 20022007 has been 6.3%; the trend growth rate has picked up; and in three out o nineyears since the 1960s, growth was at least 7% (2004, 2005, and 2007). Moreover, historically, growth

was driven by private consumption, but in 20062007, investment was the largest contributor tooutput growth.

10.0

9.0

8.0

7.0

6.0

5.0

4.0

3.0

2.0

1.0

0.0

FIGURE 1

PAKISTANS LONG-RUN GROWTH RATE

1962 1967 1972 1977 1982 1987

Actual Trend

Note: The trend growth was estimated using the Hodrick-Prescott filter with a smoothing parameter of 50.

Source: Authors calculations.

1992 1997 2002 2007

Although these growth rates are substantially higher than those o many other countries in

the world, they all short o the rates achieved by the East and Southeast Asian economies in thelast 30 years; or those achieved by the PRC and iet Nam more recently. Moreover, Pakistan hashad ewer episodes o rapid growth since the mid-1960s (and growth during these periods hasbeen slower too) than PRC, Indonesia, Malaysia, or Thailand. Even Lao PDR and iet Nam have had

almost as many rapid growth episodes in about hal the time (Table 1).4

2 Haque (2006) argues that Pakistan lacks an academic community and serious debate, leading to a stunted development

o the economics proession. or long-run sustained growth, the country needs a new a new strategy based on latest

research (e.g., on the role o government as creator o growth-promoting institutions).3 Although as igure 1 shows, it has not had a single year o negative growth during the last 45 years.4 Rapid growth is defned as three consecutive average annual growth rates o at least 5%. Average annual growth rates

reer to rolling seven-year periods (see ADB 2007, Table 3.1.8).


13/46

sectioN i

iNtroDuctioN

erD WorkiNgpAper seriesNo. 110

tAble 1rApid GrowthANd ACCelerAtioNs

periodCovered

Numberof rApidGrowthepisodes

AverAGeGrowthduriNGrApid Growth

(%)

Numberof Growth

ACCelerAtioNs(yeArs)

Growthbefore(%)

GrowthAfter(%)

GrowthACCelerAtioN(perCeNtAGe

poiNts)

PRC 19652004 28 8.84 2 (1981, 1991) 6.36 10.87 4.51

8.50 10.73 2.23

India 19652004 15 5.57 1 (1982) 3.55 5.84 2.29

NIE

Hong Kong, China 19652004 21 7.52 1 (1975) 7.47 9.76 2.29

Korea, Rep. o 19652004 24 7.72 1 (1984) 6.44 8.75 2.31

Singapore 19652004 28 7.76 1 (1987) 6.08 9.17 3.09

Taipei,China 19702004 22 9.33 1 (1984) 8.16 12.01 3.85

ASEAN 4

Indonesia 19652004 24 7.00 1 (1988) 5.24 7.85 2.61

Malaysia 19652004 22 7.32 1 (1987) 4.50 8.95 4.45

Philippines 19652004 8 5.62 1 (1987) 0.15 3.11 2.96

Thailand 19652004 24 7.31 1 (1986) 5.30 9.68 4.38

oher suhea Aia

Lao PDR 19842004 9 6.13 1 (1991) 4.26 6.33 2.07

iet Nam 19842004 11 7.20 1 (1991) 4.63 8.03 3.40

suh Aia

Bangladesh 19652004 1 (1975) 0.09 3.80 3.89

Bhutan 19802004 13 6.59 0

Maldives 19952004 1 6.96

Nepal 19652004 1 (1983) 2.34 5.29 2.95

Pakistan 19652004 12 6.18 1 (1977) 3.53 6.66 3.13Sri Lanka 19652004 8 5.19 0

Source ADB (2007, Table 3.1.9)

However, since fscal year 2001 when the growth rate was just below 2%, growth has accelerated,and in fscal years 2005, 2006, and 2007, growth peaked at 9.0%, 6.6%, and 7%, respectively. A keyquestion in the minds o policymakers is whether Pakistan will be able to sustain this growth rate.5

ne possible way to address this question is by analyzing the extent and depth o structural changeo Pakistans economy. What sort o economic transormation does Pakistans economy require inorder to sustain high growth that is compatible with a reorm process? I there are lessons rom theexperience o developing Asias successul economies, where should Pakistan direct its eorts?

Recent work by Hausmann and Rodrik (2006) and Rodrik (2004 and 2006) suggests that there isnothing inevitable about either growth or structural change, and in act only a handul o countries

outside the industrialized world have been able to sustain both. In this light, new perspectives arenow stressing the importance o country-specifc actors, history, institutions, and coordinationdifculties, i.e., the act that the return to one investment depends on whether another investment

5 A lead article in one o Pakistans leading newspapers was entitled No choice but to match India growth ( The Nation2007).


14/46

4 December 2007


Jesusfelipe

is in place. This perspective also highlights the linkages between what a country produces, howit produces it, and the opportunities or sustaining growth. It also underlines the crucial role opolicy. inally, it emphasizes the importance o competitive advantage built on task ragmentation(i.e., frms can split production by being part o international production networks or global value

chains), and the importance o diversifcation, as opposed to traditional comparative advantage(Imbs and Waczciarg 2003).

Analysts today agree that structural transormation is a lengthy and difcult process that

most developing countries ind very diicult to traverse. What retards or even prevents thistransormation? There are two classes o problems (market ailures). The frst involves coordinationailures. or example, containerization o reight may not take hold i the right kind o acilities or

handling containers at ports, railway stations, and trucking depots do not already exist. Likewise,the development o a hotel industry is contingent on the existence o basic inrastructure, includingairports and roads. The second class o problem concerns inormation spillovers. I the frst moverinto a new activity fnds that it is proftable, then other frms will most likely ollow. But i the frst

mover ails, he or she will bear the loss (the private return to this investment is lower than thesocial return). Labor training is another example o this class o problem, as the frm that trains aworker may lose him or her to a frm oering a higher salary.

ne implication o the modern thinking is that structural transormation is about creatingnew activities that can use the capabilities that an economy already possesses. The importantpoint that the new literature emphasizes in this respect is that unless purposeul action is taken

to develop these new activities, countries will not overcome these market ailures that constrainstructural transormation. The market, by itsel, will not resolve these problems. rom this pointo view, policy has a vital role to play. Policy, in this context, is about creating the conditions orstructural transormation and about providing the inputs that are specifc to subsets o activities

(Hausmann and Rodrik 2006). These inputs are property rights, institutions, and inrastructure.These are complementary inputs to the market whose specifc orm will vary depending on thecircumstances. Successul policy making requires an experimental approach and a degree o creativity.Causes and outcomes o structural change are summarized in igure 2.

The remainder o the paper is structured as ollows. Section I provides an overview o Pakistansdegree o structural change by benchmarking it to other countries in Asia. Sections II, III, and Idiscuss three recent research areas on growth and structural transormation. These areas can be

viewed as tools that can help conceptualize Pakistans development challenges. Section II is onexport sophistication, Section III on product space, and Section I on industrial policy. Policymakersin Pakistan should beneft rom understanding the implications o this new literature. The last

section summarizes the main fndings.


15/46

sectioN ii

structurAl trANsformAtioNiNpAkistAN


FIGURE 2STRUCTURAL CHANGE: CAUSES AND OUTCOMES

Logistic pattern

Causes

Income elasticities

of demand: social

imitative; biological

Capital accumulation

Human capital

Physical capital

New industrial policy

Market failures

coordination failures

information spillovers

Factor relocation

Innovation

Structural Change

Outcomes

Change in output andemployment composition

Increase in laborproductivity

Upgrading to highervalue added activities

(production and exports);generation of new

activities; introduction ofnew methods of organization

II. stRUCtURAL tRANsFoRMAtIoN IN PAKIstAN

Decadal growth rates by sector as well as by output and employment shares are shown inigures 3 and 4. The service sector is by ar the most important contributor to Pakistans overall

output growth. Table 2 provides inormation about the distribution o Pakistans service sector. It isworth noting that since the 1970s, wholesale and retail trade was already the largest subcategory.In 20012005, transport, storage and communications plus wholesale and retail trade representedmore than 50% o the service sector, and about 29% o total GDP. These two categories tend to

be nontradables. This is mostly the result o the act that since the 1970s, services have been thelargest sector o the economy in terms o output, accounting now or over 50% o total GDP, and,simultaneously, having grown relatively ast. Agriculture and industry account or about 25% each.Agricultural growth is, however, very low. The result is that now the service sector contributes more

than 50% to overall output growth (igure 5).6

6 This is also the case o India and the Philippines. In Thailand, industry and services contribute about the same, about47% each. In PRC, Indonesia, Malaysia, and iet Nam, industry is the largest contributor to output growth.


16/46

December 2007


Jesusfelipe

8

7

6

5

4

3

2

1

01970s

Total

Source: Authors calculations based on data from World Development Indicators online database (World Bank,

downloaded 4 August 2006).

1980s 1990s 200004

Agriculture Industry Services

FIGURE 3PAKISTAN: ANNUAL GROWTH BY SECTOR (PERCENT)

Source: Authors calculations based on data from World Development Indicators online database (World Bank,downloaded 4 August 2006).

1970s 1980s 1990s 200004 1980s 1990s 200004

60

50

40

30

20

10

0

60

50

40

30

20

10

0

Agriculture Industry Services Agriculture Industry Services

FIGURE 4PAKISTAN: OUTPUT AND EMPLOYMENTSHARES (PERCENT)

Output Shares Employment Shares

tAble 2pAkistAN serviCes seCtor, output shAre

19717 19780 19818 19890 19919 19900 0010

Transport, Storage andCommunication

7.49 7.66 9.22 8.32 10.08 10.38 12.57

Wholesale and Retail Trade 14.89 15.62 15.48 16.62 16.43 15.32 16.77

inance and Insurance 2.29 2.57 2.98 2.99 3.18 3.22 3.51

wnership and Dwellings 4.16 3.73 4.88 4.57 4.35 4.50 3.56

Public Administration and Deense 7.36 7.83 8.38 9.30 7.87 7.89 6.80

Community Services/thers 7.47 8.30 7.31 7.37 7.78 8.79 9.14

Total 43.65 45.71 48.25 49.18 49.69 50.10 52.34

Source Authors calculations based on data rom CEIC Data Company Ltd. (downloaded 13 December 2006).


17/46

sectioN ii


erD WorkiNgpAper seriesNo. 110 7

Note: Data for Viet Nam only covers the period 19852004.Source: Authors calculations based on data from World Development Indicators online database (World Bank,

downloaded 4 August 2006).

Agriculture Industry Services

FIGURE 5GROWTH ACCOUNTING, 19702004 (PERCENT)

60

50

40

30

20

10

0PRC India Indonesia Malaysia Philippines Thailand Viet Nam Pakistan

It is important to note that Pakistan is an agricultural economy in terms o structure oemployment. Although the share o agricultural employment has declined over time, over 40% oPakistans labor orce is still employed in the primary sector (rom over 60% in the 1980s). This is

a sector that suers rom low productivity crop yields. This is due to, among others, lack o seedvarieties, drought and high temperature stress, soil nutrient mining, and salinity/water logging(PIDE 2006).7 The service sector employs slightly over 30% o the labor orce, while industryemploys less than a fth.8

igure 6 shows the contributions o intrasectoral labor productivity growth and intersectorallabor productivity growth (i.e., reallocation o labor) to overall labor productivity growth. ollowingChenery et al. (1986), the economywide growth rate o labor productivity can be decomposed into

two parts one, the sum o the growth rates o labor productivity within sectors (weighted by thesectors share in output); and second, the eect o labor relocation across sectors o dierentproductivity, calculated as the sum o the changes in the employment shares o the sectors receiving

employment moving out o agriculture (i.e., industry and services) multiplied by the dierential inlabor productivity with respect to agriculture.9

7 PIDE (2006, 23) argues that sustained growth in agriculture o 56% is needed to ensure rapid growth.8 A logistic cross-sectional regression indicates that Pakistans agricultural output share (in 2004) was about what it

should be (i.e., equal to the actual share), 22.5%, given the countrys income per capita. Actual employment share

in agriculture (in 2000), 48.5%, is slightly above what the regression predicts, 45.5%. In the case o industry, the

result is similar, with the output share (in 2004) being about what it should be (24.5%). Employment (in 2000) was2.5 percentage points higher than what it should be (18% versus 15.5%). In the case o services, the actual outputshare in 2000 (51.2%) is signifcantly above the predicted (43.4%) given the countrys income per capita; while the

employment share (in 2000) is about what it should be given the countrys income per capita, 33.5%.

9 The ormula used is ( )

( )( )

( ) q k qq q

q

q q

qi i I I

I AS S

S A= + + l l l l 0 0

, where q is the growth rate o overall labor

productivity; qi is the growth rate o each sectors labor productivity; kthe respective output shares; ( )/q q qI A is the dierence between the levels o labor productivity in industry and agriculture divided by the overall level


18/46

8 December 2007


Jesusfelipe

Share of Intra Share of InterHongKong,China

Singapore

Taipei,ChinaKorea

Thailand

Malaysia

Indonesia

Philippines

VietNamIndia

Pakistan

PRC

FIGURE 6

INTRASECTOR AND INTERSECTOR SHARES OF LABOR PRODUCTIVITY GROWTH

Kyrgyz

1009080

706050403020100

-10

Note: Time periods: PRC: 19872002; Hong Kong, China: 19782004; Singapore: 19702003; Taipei,China: 19652004; Korea:19702004; Thailand: 19712004; Malaysia: 19802004; Indonesia: 19762002; Philippines: 19712004; Viet Nam:19912004; India: 19832000; Pakistan: 19732002; Kyrgyz Republic: 19872002.

Sources: Authors calculations based on data from LABORSTA Labour Statistics Database (International Labour Organisation,

downloaded 9 August 2006); and World Development Indicators online database (World Bank, downloaded 4 August2006).

The frst part o the decomposition is the intrasector or pure productivity growth eect. Thiscomponent is unaected by changes in the employment share and thus isolates the contributiondue solely to productivity improvements within sectors. In general, the output share o agriculture

tends to decline over time; that o industry rises in the intermediate stage o growth; and that oservices increases in the fnal stage. As the growth rate o labor productivity tends to be highestin industry, the rise o the output share o this sector is part o the explanation o the growth

acceleration that takes place during this phase. Similarly, the rise o the output share o the

tertiary sector at the expense o industry in the later stages o growth is part o the explanationor the growth deceleration in that stage o growth. In most countries in igure 6 intrasectoralproductivity growth is more important. In Pakistan, it accounted or 83% o the total growth in

labor productivity during 19732002.10

Intrasectoral contributions to labor productivity growth (i.e., the subcomponents o intrasectoralproductivity growth) are shown in igure 7. During 19732002, Pakistans labor productivity grew at a

rate o 2.59% per annum. More than our fths o this increase was due to labor productivity growthwithin or intrasector. Productivity in agriculture grew at a rate o 2.21% per annum, accountingor 31% o the intrasectoral increase; industry 2.57% per annum (manuacturing labor productivity

grew at 2.91% per annum), accounting or 24% o the intrasectoral increase; and services at 1.97%per annum, representing 45% o the intrasectoral increase.

o productivity; ( ) /q q qS A is the dierence between the levels o labor productivity in services and agriculturedivided by the overall level o productivity; and l denotes the employment shares o each sector. The frst term o

the decomposition represents the component o overall growth that is due to the growth o labor productivity within

each sector (weighted by the output shares). The remaining two terms represent the eect o the relocation o laboracross sectors o unequal productivity (measured with respect to the productivity o the agriculture sector).

10 Given the high level o aggregation used (the three sectors), the decomposition hides shits within manuacturing andwithin services, or example.


19/46

sectioN ii




HongKong,China

Singapore

Taipei,China

Korea

Thailand

Malaysia

Indonesia

Philippines

VietNam

India

Pakistan

PRC

FIGURE 7SECTOR CONTRIBUTIONS TO LABOR PRODUCTIVITY GROWTH (PERCENT)

Kyrgyz

Agriculture ServicesIndustry

100

80

60

40

20

0

-20

The remainder o the overall growth in labor productivity (the second part o the decomposition)in Pakistan during this period (about 17%) was due to labor reallocation rom agriculture intoindustry and services (igure 8). I the shit o labor is rom a sector with low labor productivity to

a sector o higher labor productivity, this relocation makes a contribution to overall growth, overand above the growth o labor productivity within sectors. Since in general labor productivity islowest in agriculture and highest in industry, part o the acceleration o growth in the early stagesis due to the shit o labor out o the agriculture sector and into the other sectors. Likewise, part

o the deceleration in growth during the later stages is due to the shit o labor rom industry into

services. ne reason or the slower growth o labor productivity in agriculture and services lies inthe act that possibilities o mechanization o these two sectors are limited (especially services);

while the growth o labor productivity is much higher in industry because it oers much greaterpossibilities or mechanization. However, it is incorrect and misleading to assume that all servicesactivities have productivity growth rates lower than those in all individual manuacturing industries,or even below the average or all manuacturing industries.


20/46

10 December 2007


Jesusfelipe


HongKong,China

Singapore

Taipei,China

Korea

Thailand

Malaysia

Indonesia

Philippines

VietNam

India

Pakistan

PRC

FIGURE 8

BAUMOLS STRUCTURAL BONUS: INDUSTRY VERSUS SERVICES

Kyrgyz

Industry Services

120

80

40

0

-40

The eect o the transer o labor on the level o productivity is what Baumol et al. (1985)and Baumol et al. (1989) call the structural bonus. Backward economies with a large pool oemployment in low-productivity activities (normally agriculture) experience a bonus rom structural

change. This occurs because the transer o labor rom low to high productivity activities automaticallyincreases the productivity level o the economy (i.e., a composition eect). This happens even ithis transer o resources is mainly a shit rom agriculture to services (where productivity mightnot be signifcantly higher). However, as the logistic pattern o structural change drives resources

toward services, and given that productivity growth in this sector is usually slower than in industry,countries experience a structural burden. This burden means that the process o structuralchange has a negative impact on productivity growth. As the share o labor in services increases,the average rate o growth o the economy decreases. In the limit, as most o the labor orce has

moved into the services activities, economies experience asymptotic stagnancy as productivitygrowth is mostly determined by the services sector. This means that structural change increases thelevel o productivity o the economy but, at the same time, as the economy matures, productivity

growth slows down. The decomposition used here does not consider the structural burden.

igure 8 indicates that services have absorbed much more labor transerred out o agriculturethan industry. In Pakistan, reallocation o labor into industry accounted or only 17% o the

intersectoral increase in labor productivity growth, while the remaining 83% was due to reallocationo labor into services.

It is possible to analyze trends in per capita income by decomposing it into the product o

the employmentpopulation ratio and labor productivity.11 Regarding the frst part, Pakistan has

experienced alling employment absorption capacity, as the ratio o employment to working agepopulation has declined over time (igure 9). Between 1973 and 2002, it ell rom 55.8% to 47.8%.This is the result o a combination o actors. irst, the absorption capacity o industry has been at.

Second, the decline in the absorption capacity o agriculture has not been matched by an equivalentincrease in the absorption capacity o services (in the last ew years the absorption capacity o

11 Algebraically (Y/P)=(L/P) x (Y/L), where (Y/P) is income per capita, (L/P) is the employmentpopulation ratio, and(Y/L) is labor productivity.


21/46

sectioN ii



services has also been at). Third, the problem is aggravated by a relatively high population growthrate (2.69% per annum or 19811998). While this entails a demographic bonus it also poses aserious burden as long as the economy does not generate enough employment.

Sources: Authors calculations based on data from LABORSTA Labour Statistics Database (International Labour Organisation,downloaded 9 August 2006); and World Development Indicators online database (World Bank, downloaded13 December 2006).

FIGURE 9PAKISTAN: EMPLOYED PERSONS AS PERCENT OF WORKING-AGE POPULATION (PERCENT)

60

50

40

30

20

10

0

TotalIndustryManufacturingConstructionTrade/Hotels and RestaurantsFinance and Business Services

AgricultureMiningElectricity, Gas, and WaterServicesTransportation, Storage, CommunicationsPublic Administration and Others

1973 1976 1979 1982 1985 1988 1991 1994 1997 2000

Successul structural transormation must lead, in the long run, to increases in labor productivity,

the key determinant o living standards. Labor productivity in Pakistan is increasing, although itis still very low compared to that o the NIEs (and certainly to the average or countries o the

rganisation or Economic Cooperation and Development), indicating that there is plenty o roomor catch-up (igure 10). Pakistans level o labor productivity increased by a actor o 2.17 between

1973 and 2002 (which implies a growth rate o almost 2.6% per annum). This indicates that with aslowly increasing level o labor productivity and declining employmentpopulation ratio, Pakistansincome per capita cannot increase ast. Especially worrisome is the act that Pakistans employmentabsorption capacity is not contributing to the increase in living standards.

Historically, structural change in the industrialized world was associated with the expansiono the secondary sector (industry). The role attributed to manuacturing in particular in theprocess o take-o and subsequent catch-up is usually a key element o sector studies o growth.

Recently, Rodrik (2006) has argued that sustained growth requires a dynamic industrial base, anidea well established in the literature. It was probably economist Nicholas Kaldor (1966 and 1967)

who provided the most thorough explanation o why industry, in particular manuacturing, playsthe role o engine o growth (see elipe et al. 2007 or an analysis o Kaldors views applied todeveloping Asia)12 a country with a relatively high growth rate o manuacturing productivity willhave a relatively high growth rate o labor productivity in the economy as a whole.12 The importance o industrialization as the key driver o structural change in Pakistan is recognized by PIDE (2006),

which contains detailed sectoral analyses. This document talks about the need to establish an industrial strategy

whose overriding aim is to develop a ast growing, internationally competitive, and export-driven industrial sector thathelps in poverty reduction through the provision o adequate employment opportunities to the growing labor orce

(PIDE 2006, 23).


22/46

1 December 2007


Jesusfelipe

FIGURE 10TOTAL LABOR PRODUCTIVITY (US$2,000), LOGARITHMIC SCALE

100000

10000

1000

100

100000

10000

1000

100

100000

10000

1000

100

OECD versus PRC and India OECD versus NIEs

OECD versus ASEAN-4 OECD versus Other Asian Developing Countries

OECD

OECD

Indonesia

Malaysia

Philippines

OECD

Azerbaijan

Kyrgyz Rep.

Pakistan

Viet NamThailand

PRC India

197075 7679 8085 8689 9095 9699 20004 197075 7679 8085 8689 9095 9699 20004

19 70 7 5 76 7 9 8 0 85 86 89 90 95 96 99 2 00 0 419 70 7 5 7 6 79 80 85 86 8 9 90 9 5 96 9 9 20 00 4

OECD

Singapore

Hong Kong, China

Taipei,China

Rep. of Korea

Note: The 19801985, 19861989, 19901995, and 20002004 data for India refer only to 1983, 1988, 1994, and 2000 figures, respectively. Similarly,the 20002004 data for PRC, Indonesia, Kyrgyz Republic, and Pakistan refer only to 20002002; the 19861989 data for Indonesia only to 1989;the 19761979 data for the Philippines only to 1978; and the 19701975 Figure for Pakistan only to 19731975.

Sources: Authors calculations based on data from LABORSTA Labour Statistics Database (International Labour Organisation, downloaded 9 August2006); World Development Indicators online database (World Bank, downloaded 4 August 2006); and Directorate General of Budget,Accounting and Statistics (http://eng.stat.gov.tw/public/, downloaded 13 December 2006).


23/46

sectioN ii



ne can, thereore, speak o the logic o industrialization (Nixson 1990, 313) and understand whymany developing countries have adopted (or desire to adopt) strategies toward rapid industrialization,oten starting with industries that use relatively simple technologies and that have the potentialto be labor-intensive and thus absorb labor, such as textiles, clothing, and shoes. The experience

o the industrial economies shows that establishing a broad and robust domestic industrial baseholds the key to successul development, and the reason why industrialization matters lies in thepotential or strong productivity and income growth o the sector. This potential is associated

also with a strong investment drive in the sector, rapidly rising productivity, and a growing shareo the sector in total output and employment. The presence o scale economies associated withthe secondary sector, gains rom specialization and learning, as well as avorable global marketconditions, imply that the creation o leading industrial subsectors, along with related technological

and social capabilities, remains a key policy challenge. Today, there is wide variety across countriesin terms o resource endowments, pace o capital accumulation, and policy choices. This impliesthat there is ample room or diversity in industrial development.

igure 11 shows the relative stagnation o Pakistans manuacturing sector. While the share

o this sector in GDP increased signifcantly in countries like Indonesia, Malaysia, or Thailandduring the last 30 years, it has hardly changed in Pakistan, where it has remained at about 15%since the 1970s. Something similar has occurred i we look at the sector rom the point o view o

employment (igure 12). It is important to stress that, while not small (except when compared tothat o Malaysia), the share o the sector has not increased.13

13 A logistic cross-sectional regression indicates that Pakistans manuacturing output share in 2000 was about what it

should be, 14.5%, given the countrys income per capita, trade share in GDP, and population. The same regressionindicates that Indias manuacturing share, while similar to that o Pakistan, is about 4 percentage points below what

it should. n the other hand, PRC, Indonesia, Malaysia, and Thailand have actual shares signifcantly higher than the

predicted ones. It is also worth mentioning that similar regressions or 1975, 1985, and 1995 indicate Pakistans incomeelasticity o manuacturing output declined signifcantly rom 0.44 in 1975 to about 0.25 in the subsequent periods.

This is a clear sign o development. In poor developing countries, the income elasticity o manuacturing output tendsto be high; but as countries develop it becomes smaller, even negative in rich countries (elipe and Estrada 2007).


24/46

14 December 2007


Jesusfelipe

1970s 1980s 1990s 200004

40

35

30

25

20

15

10

5

0

FIGURE 11

MANUFACTURING OUTPUTSHARES

PRC India Indonesia Malaysia Philippines Thailand Viet Nam Pakistan

Note: The 1980s data for Viet Nam refer only to 19851989.


1980s 1990s 200004

25

20

15

10

5

0

FIGURE 12

MANUFACTURING EMPLOYMENTSHARES

PRC India Indonesia Malaysia Philippines Thailand Viet Nam Pakistan

Note: The 1990s data for Viet Nam refer only to 1996 to 1999.Sources: Authors calculations based on data from World Development Indicators online database

(World Bank, downloaded 4 August 2006); and Anant et al. (2006).

Looking deeper into the structure o manuacturing, one can see that this sector is heavily

dependent on ood and beverages and textiles (Table 3). These two categories represented about58% o the sectors value-added in the 1970s. Industrial chemicals represented about 11% o thetotal. The share o electrical and nonelectrical machinery plus transport equipment was a small8%. By the 1990s, ood and beverages plus textiles still represented 48% o total manuacturing

output; the share o industrial chemicals had increased to 15.5%; and that o electrical, nonelectricalmachinery, plus transport equipment represented 11% o the total. While this indeed denotes aair degree o transormation and upgrade in the manuacturing sector, the share o the last three


25/46

sectioN ii



branches is substantially smaller than, or example, in Malaysia, where the combined share o thesethree categories increased rom about 16% to about 40% o total manuacturing output during thesame period (see Appendix Table 1 or detailed data).14

tAble 3

pAkistAN mANufACturiNG brANChes (perCeNt)

1970 1980 1990

ood and beverages 30.45 30.94 22.89

Textiles 27.78 18.14 25.06

Apparel, leather, and ootwear 2.04 2.37 2.80

Wood and wood products 0.26 0.39 0.37

Paper and paper products 1.61 1.15 1.54

Printing and publishing 1.22 1.06 2.00

Industrial chemicals 11.20 14.29 15.50

Petroleum and coal products 5.27 6.01 3.26

Rubber and plastic products 1.80 1.80 1.42

Nonmetal mineral products 4.43 7.75 7.76Basic metals 3.06 6.20 5.13

Metal products 1.62 1.06 0.81

Nonelectrical machinery 1.84 2.14 2.09

Electrical machinery 3.31 3.26 5.43

Transport equipment 2.99 2.89 3.05

thers 1.11 0.55 0.88

Total 100.00 100.00 100.00

Source Authors calculations based on data rom Industrial Statistics International Standard Classifcation Revision 2 (United NationsIndustrial Development rganization 2005).

igure 13 graphs the degree o specialization in the manuacturing sector vis--vis income

per capita.15

As noted above, the new literature emphasizes the importance o diversifcation asopposed to traditional comparative advantage (i.e., the idea that as countries open up trade, theywill specialize in those activities that use intensively those actors that are in abundant supply).Imbs and Wacziarg (2003), however, present evidence that suggests that at low levels o income percapita, economies tend to diversiy and subsequently, as their income rises, they specialize whatever

drives economic development, it is not comparative advantage. Graphically, this corresponds to a U-shape relationship between specialization and income per capita, which igure 13 corroborates.

14 During a presentation at Pakistans Institute o Development Economics, aizullah Khilji noted that census datawould show that the shares o nonelectrical machinery, electrical machinery, and transport equipment have actuallydeclined.

15 The degree o specialization was constructed using UNID 3-digit manuacturing data. The degree o specialization

h is defned as hs s

hi ii= +

100 1

( ln )

max, where hmax ln= (no. sectors), si is the share o the i-th branch in

total manuacturing value-added. By construction 0 100 h . I the shares o all sectors are equal, the degree ospecialization is 0; and i only one sector exists, then the value o the indicator is 100.


26/46

16 December 2007


Jesusfelipe

90

70

50

30

FIGURE 13

SPECIALIZATION INDEX OF DEVELOPING ECONOMIES IN ASIA:

MANUFACTURING VALUE-ADDED

5 6 7 8 9 10

Note: The estimated regression line is:Specialization = 203.653 - 39.163 GDP per capita(log) + 2.554 (GDP per capita(log))2t-stat: (17.78) (-12.39) (12.13)R2: 0.30; No. of observations: 387

Source: Asian Development Bank (2007).

BangladeshMalaysia

Sri Lanka

Hong Kong, ChinaNepal

Taipei,China

IndiaPRC

Thailand

IndonesiaPakistan

Singapore

KoreaPhilippines

Degree

ofSpecialization

Log GDP per Capita

Figure 13 shows, though, that individual country experiences in developing Asia do not ft theU-shaped pattern o specialization or value-added or employment suggested by Imbs and Wacziarg(2003). This, on the other hand, is not surprising as data cover a short period o time. Figure13 shows increasing diversifcation as income per capita increases (at low levels) in Bangladesh,India, Indonesia, Nepal, Pakistan, and Thailand. There is no economy that becomes more specialized

within comparable low income ranges. Increasing specialization is only detected at higher incomelevels in Korea; Malaysia; Singapore; and Taipei,China. The fgure also shows that, at similar incomelevels, the PRC has a more diversifed pattern o manuacturing output. Kochhar et al. (2006) haveshown that India has a more skill-based and capital-intensive pattern o production than the PRC.Pakistans value-added specialization index has a trend toward diversifcation: in 1963 Pakistanhad a very high index, 74, denoting a very high concentration in the structure o manuacturing.By 1996, the index had a value o 63. The fgure also shows a trend toward diversifcation in thePRC and India. Thailand exhibits increasing specialization. While the trend toward specializationremains in Malaysia and Singapore, the index barely changes in Korea.

Pakistans manuacturing sector has a relatively low technology and scale index, compared tocountries like the PRC and India, or example, given the countrys per capita income (Figure 14).16

16 This index is calculated by dividing all manuacturing branches into our groups according to level o technology andscale. Then the share o each group is weighted by 1, 2, 3, or 4 (1 or the lowest technology and scale level, and 4or the highest) and summed up. Group 1 corresponds to the manuacturing branches with the lowest technology and

scale economies, e.g., ood and beverages, tobacco, wearing apparel, leather products. Group 2 consists o plastic andrubber products, paper, among others. Group 3 consists o iron and steel, nonmineral products, among others. Group 4consists o products with the highest technology and scale economies, such as electrical and nonelectrical machinery,industrial chemicals, proessional equipment, transport equipment, among others.


27/46

sectioN ii



Pakistans technology index increased rom 1.77 in 1963 to 2.26 in 1996 (the last years or whichdata are available), and at todays income per capita, its index is well below those o while thoseo the PRC and and India (about 3).

The share o medium- and high-technology and economies o scale manuacturing products

in Pakistans GDP has been stagnant since the 1970s (Group 4 in igure 15; see ootnote 16), wellbelow 10% o the total; while in Malaysia the share o this group increased rom less than 10% inthe late 1980s to about a fth o total GDP in 2005. The same applies to Korea. igure 16 shows

the shares or PRC; India; Indonesia; and Taipei,China. The share o group 4 is higher in these ourcountries. nly the share o Group 4 in the Philippines (shown in igure 15) is similar to that oPakistan.17

Log GDP per Capita

Scale

and

Technology

In

dex

FIGURE 14

TECHNOLOGY AND SCALE INDEX

Hong Kong, ChinaSingapore

Korea, Rep. ofTaipei,China

4.5 5.0

PRC and India

ASEAN-4

NIEs

Other South Asia

4.0

3.5

3.0

2.5

2.0

1.5

4.0

3.5

3.0

2.5

2.0

1.55.5 6.0 6.5 7.0 7 8 9 10

Scale

and

Technology

In

dex

Log GDP per Capita

5 6 7 8 9Log GDP per Capita

Scale

and

Technology

Index

3.5

3.0

2.5

2.0

1.5

1.0Scale

and

Technology

Index

3.5

3.0

2.5

2.0

1.5

1.0

Log GDP per Capita

IndonesiaPhilippines

MalaysiaThailand

BangladeshPakistan

NepalSri Lanka

PRC India

Source: Authors calculations based on data from Industrial Statistics International Standard Classification Revision 2 (United Nations Industrial

Development Organization 2005).

5.0 5.5 6.0 6.5 7.0

17 ne must not orget that Pakistan is capable o making nuclear weapons. This means that the country has developed

the capability to manuacture very sophisticated products. This, however, is dierent rom what is required to sustaingrowth in a country like Pakistan. More will be said about this in Section .


28/46

18 December 2007


Jesusfelipe

FIGURE 15SHARES OF MANUFACTURING GROUPS IN GDP BASED ON

TECHNOLOGY AND SCALE (PERCENT)

Pakistan

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

Malaysia

Philippines Korea

1970 74 78 82 86 90 94 98 20021981 84 87 90 93 96 99 2002 05

1987 91 95 991965 70 75 80 85 90 2003

Source: Authors calculations based on data from Industrial Statistics International Standard Classification Revision 2(United Nations Industrial Development Organization 2005).

121086420

15129

630

20171411

852

-1

211815129630

181512963

0

FIGURE 16SHARES OF MANUFACTURING GROUPS IN GDP BASED ON

TECHNOLOGY AND SCALE (PERCENT)

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

Group 1Group 2

Group 3Group 4

1970 74 78 82 86 90 94 98

9

6

3

0

181512963

0

India Taipei,China

PRC Indonesia

1980 83 86 89 95 1981 84 87 90 93 96 99 2002 05

1815129630

1981 84 87 90 93 96 99 2002 05

Sources: Authors calculations based on data from Industrial Statistics International Standard Classification Revision 2(United Nations Industrial Development Organization 2005); Directorate General of Budget, Accounting and Statistic(http://eng.stat.gov.tw/public/, downloaded 13 December 2006); and CEIC Data Company Ltd.(downloaded 13 December 2006).


29/46

sectioN iii

exportsAND structurAl trANsformAtioN


It is impossible to understand East Asias phenomenal growth between the mid-1960s andthe fnancial crisis o 19971998 without bringing into the picture productivity gains derived romthe assimilation o technologies rom the developed countries (Nelson and Pack 1999, elipe andMcCombie 2001). The gains that this process brought were not the result o hard-core research and

development, but o behind-the-rontier improvements. or example, Hobday (1995) described indetail how East Asian frms climbed the ladder by slowly learning

East Asian latecomers did not leaprog rom one vintage o technology to another. n the

contrary, the evidence shows that frms engaged in a painstaking and cumulative processo technological learning a hard slog rather than a leaprog. The route to advancedelectronics and inormation technology was through a long difcult learning process,

driven by the manuacture o goods or export Hobday (1995, 1188).18

Kim (1997) described Hyundais eorts to produce a car ater it had purchased the oreignequipment, hired expatriate consultants, and signed licensing agreements with oreign frms as

ollows

Despite the training and consulting services o experts, Hyundai engineers repeated trialsand errors or ourteen months beore creating the frst prototype. But the engine block

broke into pieces at its frst test. New prototype engines appeared almost every week,only to break in testing. No one on the team could fgure out why the prototypes keptbreaking down, casting serious doubts even among Hyundai management, on its capabilityto develop a competitive engine. The team had to scrap eleven more broken prototypes

beore one survived the test. There were 2,888 engine design changes. Ninety seventest engines were made beore Hyundai refned its natural aspiration and turbochargerengines. In addition, more than 200 transmissions and 150 test vehicles were created

beore Hyundai perected them in 1992 (Kim 1997, 129).

Pakistans Medium Term ramework 20052010 targets the expansion o the share o engineeringproducts, electronics and chemicals, and other high-technology-based and value-added manuacturing

industries between 20042005 and 20092010 by 3.5 percentage points (see Haque 2006, Box 1).Haque (2006, 13), however, shows some skepticism about how the Government is going to achievethis objective and claims that it is not clear what instruments will be used. However, i pastpractice is an indicator, it will be a mix o subsidy and tari protection.19

III. EXPoRts AND stRUCtURAL tRANsFoRMAtIoN

In conventional theory la Heckscher-hlin, countries export products that are more intensive inthe actors in which they are relatively abundant. As development takes place, countries accumulateactors o production at dierent rates, and the products that countries produce and export become

more intensive in these actors. In this theory, the pattern o specialization is uniquely determined

by actor endowments in the sense that, independently o initial conditions, the economy converges18 Nevertheless, case study analysis based on interviews by Hobday et al. (2004) covering 25 Korean frms in seven

sectors leads the authors to conclude that although some o the leading chaebols are reaching the innovation rontier

in some product areas, and are getting into new product design and research and development, there is reason to call

into question the strength o leading Korean frms in international leadership capabilities such as new product creationand capital goods technology.

19 PIDE (2006) presents a strategy o what Pakistan needs to do in order to embark on the same path as PRC, Japan,and Korea to develop its engineering industry.


30/46

0 December 2007


Jesusfelipe

to a pattern o specialization that can be ully explained by the economys actor endowments. Atthe other extreme o the theoretical spectrum, some new trade theory models treat productivitygrowth as the result o learning by doing and assume away actor endowments as a determinanto comparative advantage (e.g., Krugman 1987). Here, the pattern o specialization cannot be

determined independently o initial conditions and history. In these circumstances, real as wellas monetary shocks become important as they can inuence the pattern o trade specialization.Industrial policy (see Section ) also becomes important in acquiring new comparative advantage,

independently o actor endowments.

In recent work, Hausmann et al. (2005) have argued that specialization patterns are partlyindeterminate and may be shaped by idiosyncratic elements. They have shown that there is a very

strong relationship between the level o income o a country and a measure that they constructo the sophistication o exports (see below). ne implication o this relationship is that as theprocess o development takes place, countries change their export package, i.e., they must undergostructural transormation. Indeed, structural change appears reected in the improvements in

quality and diversifcation o a countrys exports. This implies, frst, that the range o goods thatan economy ends up producing and exporting changes in time. Second, and more important, whileor the standard theory o comparative costs the transition rom the production and export o onetype o goods into a dierent one is a unction o the changing actor endowments with no role or

active policy, the range o goods in Hausmann et al.s (2005) model is determined not just by actorendowments. It is also a complicated process that is determined by the number o entrepreneursthat can be stimulated to engage in cost discovery in the modern sectors o the economy.20 The

key and primary source o international gains is international learning leading to technical progressand increased labor productivity. However, upgrading and diversifcation will not occur automatically,as they require proactive measures. The cost o discovery is the uncertainty that an entrepreneuraces when producing a good or the frst time in a developing economy. Even in case the good is

produced with a standard technology, the institutional reality o many developing countries requiresa substantial degree o adaptation. Thereore, an entrepreneur will explore the underlying cost

structure o the economy. This is a process that entails considerable positive externalities or otherentrepreneurs since, i the investment is successul, other entrepreneurs will ollow and emulatethe incumbent. This way, the private returns to the frst entrepreneur are socialized. I, however,the frst entrepreneur ails, his cost remains private. Thereore, the knowledge externality impliesthat investment levels in cost discovery are suboptimal unless the industry or the government fnd

some way in which the externality can be internalized.

Thus, the mix o goods that a country produces and exports may have important implicationsor economic growth. In act, not all goods are alike in terms o their consequences or economic

perormance specializing in some products will bring higher growth than specializing in others. Inother words, exports can be seen both as a tool o development and as a test o a countrys success.In this ramework, the authors argue, government policy has a potentially important positive role

to play in shaping the production structure. The authors show that countries that latch on to a seto goods that are placed higher on the quality spectrum tend to perorm better.

20 Skarstein (2007) argues that underdeveloped countries aiming at development ollowing the precepts o the theoryo comparative costs (e.g., specialization in, or example, agriculture), will have a hard time, as this theory does not

explain reality, namely, that absolute advantage owing to disparities in technological levels and labor productivitiesplays an important role in international trade.


31/46

sectioN iii

exportsAND structurAl trANsformAtioN


Hausmann and Rodrik (2006) also note that countries capabilities are quite specifc to eachactivity, and that countries export some products but do not export others that either seem to be verysimilar or are manuactured and exported by countries that also seem to have similar endowments.or example, they note that Pakistan exports soccer balls but not hats, while Bangladesh does the

opposite. While this may be an anecdote, an in-depth understanding o Pakistans export structureis undamental in order to understand its prospects or structural transormation.

Analysis o Pakistans export sector also reveals important characteristics on the extent and

depth o the transormation o the economy. Table 4 reveals that although Pakistans exports havebecome less concentrated (the share o the top ten exports has declined rom 62.8% in 1986 to49.7% in 2004), nine o the top ten export categories are textiles and apparel.

Table 4 also shows that the weighted average o the per capita GDPs o the countriesexporting Pakistans top 10 exports, denoted by PRDY, has declined signifcantly (rom a value o5,014 in 1986 to 3,458 in 2004), indicating that the country is stuck in exports that are being

exported by ever poorer countries. And, the income level o Pakistans exports, denoted by EXPY, aproxy or its export complexity (one can think o the idea o export complexity as an indicator o

competitiveness), has not shown the increase expected rom a country that is undergoing the kindo structural transormation that leads to aster growth, i.e., as that seen in economies that promote

exports o more sophisticated goods (igure 17).21 Pakistans index in 1986 (4,664) is the same asin 2004 (4,628), which is about the same that Indonesia, Malaysia, Philippines, and Thailand hadin the late 1970s; and well below the levels o the PRC and India in the 1980s.22

21 To construct PRDY and EXPY, the methodology in Hausmann et al. (2005) and data rom the United Nations Commodity

Trade Statistics Database at the 5-digit level are used. irst, the proxy or the income level o a commodity PRDYis constructed as the weighted average o the per capita GDPs o the countries exporting a given product (i.e., each

product has an associated PRDY) asPRODY Y

x X

x Xk j

jk j

jk jj

j

=

( / )

( / ) , where Yj is the per capita GDP o country j; xjk

denotes exports o country jo product k;Xj is total exports o countryj, that is,X xj jk

j

= . Note that the weight

o income per capita Yj is the revealed comparative advantage o each country in each good it exports. The advantageo using this weighted average is as ollows a country may export less o a given product than another country in

absolute terms. This may, however, be more in relative terms (i.e., revealed comparative advantage). This way, the

frst countrys income per capita will be weighted more heavily in calculating the productivity level associated withthe given product. The second step consists in constructing the productivity level associated with a countrys export

basket, EXPY, as the weighted average o the PRDY associated with each product, that is,

EXPY PRODY x

Xi k

j

jk

= . Certainly, there is, by construction, a relationship between EXPY and income per capita.

However, this is not a mechanical outcome. Hausmann et al. (2005, 10) note that calculating country specifc PRDYs

by excluding own exports rom the calculation o these measures does not change the results much.22 There is perhaps something wrong with the data. Pakistans EXPY remained stagnant between 1981 and 1989 (at about

4,500). Then in 1990 it jumped to 6,000 and stayed there until 1993. There is no data or 1994. Then in 1995 the

index took on a value o 5,594 and continued declining thereater. In 2004 it stood at approximately the same levelas in the 1980s.


32/46

December 2007


Jesusfelipe

tAble 4pAkistANs top 10 exports

198

perCeNtof totAlexports

004

perCeNtof totAlexports

Rice, semimilled or milled (unbroken) 15.4 the r woven ab ric s, le ss 85% o co tton,bleached, etc., fnished

8.2

ther woven abrics, less 85% o cotton,

bleached, etc., fnished

9.9 Bed linen o cotton7.8

Carpets, carpeting and rugs, knottedo woolor fne animal hair

9.6 Cotton yarnmeasuring, per single yarn, rom14 to 40 km/kg, not or retail

7.5

Bed linen o cotton 6.2 Rice, semimilled, or milled (unbroken) 5.9

ther woven abrics with 85% or more grey

cotton, not mercerized

5.9 Toilet and kitchen linen o cotton4.4

Toilet and kitchen linen o cotton 4.1 Undergarments, knitted or crochetedo cotton, not elastic nor rubberizedmens and

boys, shirts

3.9

Goat and kid skin leather 3.4 ther woven abrics with 85% or more greycotton, not mercerized

3.6

abrics, woven, less 85% o continuous synthetic

textile materials

3.0 uterwear knitted or crocheted, not elastic

nor rubberizedother, clothing accessories,

nonelastic, knitted or crochetedo cotton

3.0

Womens, girls, inants outerwear, textile, notknitted or crocheted; other outer garments o

textile abrics, not knitted, crocheted

2.7 Mens and boys outerwear, textile abrics notknitted or crochetedtrousers, britches, and

the likeo cotton

2.8

Medical, surgical, and veterinary instrumentsand appliances

2.7 Knitted or crocheted textile articles, nes, notelastics, etc.

Percent share o top 10 exports 62.8 Percent share o top 10 exports 49.7

PRDY o the top 10 exports (weighted average

o the per capita GDPs o the countries exportinga given product)

5,014 PRDY o the top 10 exports (weighted average

o the per capita GDPs o the countries exportinga given product)

3,458

EXPY (weighted average o the PRDYs); income

level o a countrys export basket

4,664 EXPY (weighted average o the PRDYs). Income

level o a countrys export basket4,628

Source Authors calculations based on data rom the United Nations Commodity Trade Statistics Database at the 5-digit level (UnitedNations, various years).


33/46

sectioN iv

theproDuctspAce: oNmoNkeysAND forests


FIGURE 17

EXPORTCOMPLEXITY SCORE

1977

12000

10000

8000

6000

4000

2000

0

12000

10000

8000

6000

4000

2000

0

12000

10000

8000

6000

4000

20000

IndonesiaMalaysia

PhilippinesThailand

80 83 86 89 92 95 98 2001 04

1977 80 83 86 89 92 95 98 2001 04

PRC India

1977 80 83 86 89 92 95 98 2001 04

Bangladesh Pakistan Sri Lanka

Source: Authors calculations based on data from Commodity Trade Statistics Database (United Nations, various years).

IV. tHE PRoDUCt sPACE: oN MoNKEYs AND FoREsts

This section and Section discuss two recent research areas in the feld o structural change

that should prove useul to Pakistans policymakers identifcation o product space and industrialpolicy.

Developing new activities and products is not an easy task or most developing countries. Lack

o markets and coordination problems can create insurmountable difculties. or this reason, newactivities are rarely created in a vacuum, as their development has to take place in the contexto existing capabilities, which are useul to the extent that they are similar to those o the newproducts and activities. However, the degree o similarity o capabilities may vary between any pair

o activities. or example, one may think o a country that has developed capability, inrastructure,regulatory ramework, a trained labor orce, and knowledge in general in the production o garments.These capabilities may well be useul to develop a shoe industry, but are unlikely to be o much

use in developing a new steel industry. Capabilities are specifc to each activity.

or this reason, one can think o products as trees in the orest (the product space) beingat some distance rom each other in terms o capabilities (Hausmann and Klinger 2006). irms are

monkeys that live on trees. The distance between the trees reects the similarity o the required

capabilities. In general, it is easier or new activities to develop close to the areas where monkeysalready live, since they will require similar capabilities. Producing very dierent new products (i.e.,at larger distance) requires capabilities that the monkeys (frms) do not possess.

Hausmann and Klinger (2006) and Hidalgo et al. (2007) have proposed a measure o the distancebetween products, based on the act that i the capabilities (e.g., inrastructure, technology) neededto produce two dierent products are similar, this would be revealed in the act that countries that

are good at one are also good at producing the other one. This idea o proximity (i.e., the idea


34/46

4 December 2007


Jesusfelipe

that the ability o a country to produce a product depends on its ability to produce other products)allows the authors to map the space o products in the world.23 What are some characteristics othe orest or product space? There is, undamentally, a high degree o heterogeneity, with a core anda periphery; some parts o the orest have trees (products) with many frms (monkeys), while other

parts contain ew. This implies that the development o comparative advantage in new products isstrongly aected by the distance between the products being currently produced (i.e., a unction othe existing comparative advantage) and the potential new ones. In their words, monkeys tend to

jump short distances.The key to progressing in terms o sophistication is to position the countryin a well-connected part o the orest where it is easier to move to other products.

This research shows that the way countries develop comparative advantage is ar rom random.

Some developing countries produce in the periphery o the product space with ew opportunitiesor diversifcation; others have developed capabilities easily deployable in a wide range o productscreating a path to convergence. Indeed, progress is very slow in the area o the orest with ewtrees (ew frms). irst, the probability o jumping to an individual tree is strongly aected by

how ar the occupied trees are rom it. And second, the average distance to the trees eventuallyoccupied is much smaller than the distance to a randomly chosen tree. Moreover, the position o acountry in the orest at a point in time is a good predictor o how quickly the country upgrades.Evidence indicates that this is not an easy process and that using the existing capabilities to

develop new products is not automatic. In act, when monkeys jump to new treesor frms to newproductsthey land on the lower branches o the tree. Then, they move up move easily. This meansthat once a frm can start developing a new activity, it is easier to tackle coordination ailures and

gain the specifc knowledge required. However, when a frm moves toward new products it has tostart producing without the required capabilities, and this can be done only at lower qualities. orthis reason, moving within products is easier and more easible than between products. As Hidalgoet al. (2007) indicate their work has important policy implications

the incentives to promote structural transormation in the presence o proximateopportunities are quite dierent rom those required when a country hits a dead end. Itis quite difcult or production to shit to products ar away in the space, and thereore

policies to promote large jumps are more challenging. Yet it is precisely these long jumpsthat generate subsequent structural transormation, convergence, and growth (Hidalgoet al. 2007, 487).

V. INDUstRIAL PoLICY As A CENtRAL PARt

oF PAKIstANs DEVELoPMENt stRAtEGY

Pakistans mission statement, according to the Planning Commission, is to help createknowledge led, well governed enterprising and prosperous Pakistan through realistic and innovativepolicies and programmes delivered in the most cost eective ashion.24

Upgrading a countrys production and export structures does not happen without a well-thoughtout and crated push. Indeed, structural change in the successul Asian economies did not comenaturally or easily. Though it is common to describe the transormation o the East Asian economies

as a shit rom garments into electronics and automobiles, the reality o what happened is much

23 Strictly speaking, the proximity between two products is the minimum o the pairwise conditional probability o a

country exporting a good given that it exports another one.24 This mission statement is engraved at the entrance o Pakistans Planning Commission in Islamabad.


35/46

sectioNv

iNDustriAl policyAsA ceNtrAl pArtofpAkistANsDevelopmeNtstrAtegy


more complex. All this was a painstaking process that entailed careul identifcation o products thatthe country could produce proftably using existing capabilities; learning how to use new inputs andmethods o production; and exploring new activities that required new capabilities. Indeed, in aworld where agents have to explore, learn, and adapt to an unolding and unpredictable environment,

economic modernization is hardly preordained. The structural transormation o the successul Asianeconomies was the result o purposeul action in the orm o industrial policy(IP) (Hausmann andRodrik 2006, Rodrik 2004 and 2006). In ootnote 17, it was noted that Pakistan has a technological

level that allows it to manuacture nuclear weapons (like the old Soviet Union), while Singaporedoes not have that level. This, however, is no guarantee that the country can undergo successulstructural transormation. The capabilities and institutions needed to develop a broad-based andsophisticated manuacturing sector that employs millions o workers are very dierent rom those

required to be able to make nuclear weapons. While both require political will, the latter requiresthe mobilization and coordination o a much wider group o participants in society in the contexto a market economy, i.e., where globalization, ast technical change, competition and demand (i.e.,what do consumers want?) are the rules the game. Singaporean companies understand this.

What is IP? ar rom being interpreted as an exercise in picking the winners, in modernparlance, IP is an exercise in coordinating the dierent activities between public and privatesectors with a view to identiying market ailures that impede structural transormation. IP is also

a process o eliciting inormation on signifcant externalities, and about constraints that exist andthe opportunities available. This requires strategic collaboration between public and private sectorsas well as development o the appropriate institutions or change. It does not apply only to the

activities in the industrial sector but also those in agriculture and services.25

Why IP? The reason is that markets alone are likely to undersupply the incentives and demandor new activities necessary to transorm the economy. Today there is little economic determinism

in development (i.e., countries ollow a predetermined homogeneous trajectory) and a substantialrole or policy is lacking. or example, many projects require simultaneous investments in order tobe viable (e.g., it will be difcult to build a hotel i transportation inrastructure is not adequate).The coordination o investment decisions requires a signaling device to transmit inormation about

present plans and uture conditions. Moreover, uncertainty arising rom lack o communication,that is, rom one decision maker having no way o fnding out the concurrent decisions and plansmade by others, complicates things to the point that, i sufciently great, may inhibit investment

decisions and arrest growth.

In reality, many developing countries have adopted industrial policies in an attempt to acceleratethe rate o industrialization and economic growth. The results have been mixed i judged rom the

varieties o growth perormance under similar policies. This explains why the eectiveness o thesepolicies is controversial; and why dierent views coexist on whether they made a dierence and,i so, whether this was positive or negative (see, or example, Pack and Sagi 2006). This is thecase even though observers and policymakers alike have amply documented the role o industrial

25 Amsden (2000) and Amsden and Hikino (2000) argue that the new rules o the World Trade rganization allow countriesto promote their industries, including the manuacturing sector in particular, under the umbrella o advancing science

and technology (e.g., by setting up technology and parks). Subsidies in exchange or monitorable, results-oriented

perormance standards are acceptable. Countries can, or example, target national champions. The hurdles that developingcountries ace are o the ollowing nature (i) inormal political pressures by the developed countries in avor o market

opening; (ii) countries that make use o World Trade rganization rules to promote their industries are subject toreciprocal control mechanisms; and (iii) their lack o vision.


36/46

December 2007


Jesusfelipe

policy in ostering a ast rate o industrialization in East Asia (e.g., Amsden 1989, Wade 1990).An important reason or this state o aairs seems to be that consensus is lacking on preciselythe key issue o how and under what conditions industrial policy can signifcantly alter the rate ocapital accumulation and growth.

The successul policy interventions geared toward accelerating the transition between dierentpatterns o specialization is the basis or Rodriks (1994) interpretation o how Korea and Taipei,Chinagot rich.26 During these transitions, old comparative advantages were being eroded while new

ones emerged slowly. Market incentives alone would have been unable to move the economy to ahigh growth path.27 Indeed, Rodrik (1994) argued that, more than their export orientation, thedistinguishing eature o these growth experiences was the sharp and sustained increase in their

investment rates in the early 1960s. Through an array o government interventions, by subsidizingand coordinating investment decisions, government policy was successul in reallocating resourcestoward modern capital-intensive industries. With increasing returns in these activities, this reallocationraised the rate o return on capital and pushed the economy into a high growth path. utward

orientation ollowed because the higher investment rates increased the demand or imported capitalgoods. The relatively high level o skills o the labor orce in both countries was a condition orthe success o industrial policy.

ne important lesson that Pakistan can learn rom the successul Asian economies is how todevelop an appropriate IP program that addresses questions such as how can the country develop aculture o cooperation between the public and the private sector? Which private sector agents should

be consulted frst? Moreover, since structural transormation is also about ostering competition,what sector-specifc reorms should be considered? Since structural transormation entails substantiallabor reallocation and imbalances across sectors, how are these changes and their implicationsgoing to be managed? What complementary inputs to the market can the public sector provide?

How should public spending on inrastructure and training be prioritized? What oreign frms shouldbe targeted by oreign direct investment promotion agencies? And, given Pakistans employmentproblem, what role could IP have in order to improve the situation? 28

What does it take or IP to succeed? It is impossible to provide a recipe, but it appears thatwillingness o the government to monitor the results and change its policies as needed is a keyingredient. Many industrial policy successes are the outcome o programs that produced poor resultsinitially. Governments had to fne tune them in light o what they had learned. A government too

involved in the industry cannot eectively do this, since the current benefciaries o the ailedpolicies would tend to block the changes. But, likewise, a government that simply watches rom aarand keeps frms at arms length does not accumulate the evidence to be eective. Rodrik (2004)

argues that the basic elements o an institutional architecture o institutions or industrial policyare (i) placing political leadership at the top; (ii) setting up coordination and deliberation councils;

26 There is no doubt that these two countries used industrial policy in the sense o picking the winners and that mistakeswere made. However, as Rodrik notes The trick or the government is not to pick winners, but to know when it has

a loser (Rodrik 2004, 12).27 Determination o their policymakers to sustain rapid industrialization was a undamental actor.28 PIDE (2006) can be understood as an IP proposal. This document reers to the need to identiyspecifc sectors in which

Pakistan should invest to achieve high growth and development and to integrate them in the national developmentplans (PIDE 2006, 21; italics added). The identifcation process should not be an exercise in picking the winners.


37/46

sectioNvi

coNclusioNs


and (iii) setting up mechanisms o transparency and accountability.29, 30 In the fnal analysis, itis undamental that the main stockholders o this process be simultaneously involved. These aregovernment, policymakers, and entrepreneurs. Each plays a key role in the design and implementationo a successul industrial policy program. While the government must show political commitment

to structural change, policymakers will have to design the program, and entrepreneurs will haveto implement it. The key to a successul transormation o Pakistans economy lies in embeddingprivate initiative in a ramework o public action that encourages restructuring, diversifcation, and

technological dynamism beyond what market orces on their own are capable o generating.

VI. CoNCLUsIoNs

This paper has provided an analysis o Pakistans structural transormation during the lastthree decades rom the perspective o recent literature on the topic. The basic tenets o this newliterature are as ollows.

(i) Industrialization is the key to structural transormation and growth. This requires purposeulaction, i.e., policy.

(ii) Rapidly growing countries are those with large manuacturing sectors.

(iii) Economic development requires diversifcation, not specialization. Specialization patternstend to be indeterminate and be possibly shaped by idiosyncratic elements, not by actor

endowments.

(iv) Growth accelerations are associated with structural changes in the direction omanuacturing.

(v) Countries that promote exports o more sophisticated goods grow aster.

(vi) Developing new capabilities and activities is a difcult task. Some specialization patterns

are more conducive than others to industrial upgrading.

(vii) Industrial policy, viewed as an exercise in coordination between public and private sectors,can be an important development tool.

The empirical analysis provides valuable inormation about the possible areas where Pakistanspolicymakers have to look into, in order to understand the relationship between structural transormationand the sustainability o a high growth rate. The main fndings are as ollows

29 Rodrik proposes 10 design principles or the ormulation o industrial policy (Rodrik 2004) (i) incentives should be

provided only to new activities; (ii) there should be clear benchmarks or success and ailure; (iii) there must be abuilt-in sunset clause; (iv) public support must target activities, not sectors; (v) activities that are subsidized must

have the clear potential o providing spillovers and demonstration eects; (vi) the authority or carrying out industrial

policies must be vested in agencies with demonstrated competence; (vii) the implementing agencies must be monitoredclosely by a principal with a clear stake in the outcomes and who has political authority at the highest level; (viii)

the agencies carrying out promotion must maintain channels o communication with the private sector; (ix) optimally,mistakes that result in picking the losers will occur; and (x) promotion activities need to have the capacity to renew

themselves, so that the cycle o discovery becomes an ongoing one.30 An example o IP is the case o auto parts makers in the PRC. irst, the government required oreign companies to

enter into joint ventures with local enterprises, ensuring the transer o technological capacities. Second, domestic

content requirements compelled vehicle manuacturers to source more o their inputs domestically. inally, state-ownedenterprises proved to be breeding ground or private entrepreneurs.


38/46

8 December 2007


Jesusfelipe

(i) Industry and services output growth rates are signifcantly higher than agricultures, andgiven the service sectors high share in output, services are the major contributor to totaloutput growth rate (over 50%). The contributions o agriculture and industry are signifcantlysmaller. Growth in services labor productivity growth also accounts or the largest share o

total labor productivity growth.(ii) Pakistan is a service economy rom the point o view o its output structure, but an

agricultural economy rom the point o view o its employment structure.

(iii) Intrasectoral labor productivity growth has contributed substantially more than reallocationo labor rom agriculture into the other two sectors to overall labor productivity growth.

The latter actor in act plays a minor role accounting or overall labor productivity growth.Reallocation o labor rom agriculture into services has been much more important thanrom agriculture into industry.

(iv) Pakistan suers rom alling labor absorption as the absorption capacity o the servicesector is not enough to compensate the alling capacity o agriculture and the stagnationo industry.

(v) Although Pakistans manuacturing share is not low, given its income per capita, trade ratioin GDP, and population, the share o this sector in total output has been stagnant sincethe 1970s. This contrasts with what has occurred in Indonesia, Malaysia, and Thailand, orexample, which have seen their shares increase.

(vi) The manuacturing sector is heavily concentrated in ood and beverages and textiles. Together,they account or close to hal o the sectors value-added.

(vii) The level o technology o Pakistans manuacturing sector is relatively low compared to thato other countries in Asia. Moreover, the share o manuacturing value-added accounted orby high-technology products is low and has remained stagnant during the last 40 years.

(viii) Pakistans level o labor productivity has increased very slowly since the 1970s. There isstill plenty o room or catch-up with the developed world.

(ix) The share o the top 10 exports in Pakistans total exports has decreased signifcantly in the

last two decades, although it is still highly concentrated in textiles. The weighed averageo the per capita GDPs o the countries ex

A Note on Competitiveness and Structural Transformation in Pakistan

Documents

Transcript of A Note on Competitiveness and Structural Transformation in Pakistan