Industrial Policy in the Middle East and North Africa : Rethinking the Role of the State 9781936190232, 9789774160509

Citation preview

Industrial Policy in the Middle East and North Africa

Industrial Policy in the Middle East and North Africa Rethinking the Role of the State Edited by Ahmed Galal

An Egyptian Center for Economic Studies publicationg

The American University in Cairo Press Cairo New York

First published in 2008 by The American University in Cairo Press 113 Sharia Kasr el Aini, Cairo, Egypt 420 Fifth Avenue, New York, NY 10018 www.aucpress.com Copyright © 2008 by The Egyptian Center for Economic Studies Nile City Towers, North Tower, 8th floor Corniche El Nil, Cairo 11221, Egypt All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Dar el Kutub No. 15633/06 ISBN 978 977 416 050 9 1 2 3 4 5 6 7 8

12 11 10 09 08

Designed by Joanne Cunningham/AUC Press Design Center Printed in Egypt

Comparative Assessment of Industrial Policy

v

Contents

Acknowledgments

vii

Contributors

ix

1. Comparative Assessment of Industrial Policy in Selected MENA Countries: An Overview Ahmed Galal

1

2. Do Governments Pick Winners or Losers? An Assessment of Industrial Policy in Egypt Ahmed Galal and Nihal El-Megharbel

11

3. Incentives or Compensation? Government Support for Private Investments in Turkey Hasan Ersel and Alpay Filiztekin

35

4. An Empirical Analysis of Industrial Policy in Morocco Najib Harabi 5. The East Asian Industrial Policy Experience: Implications for the Middle East Marcus Noland and Howard Pack 6. The Political Economy of Industrial Policy in the Middle East and North Africa Mustapha K. Nabli, Jennifer Keller, Claudia Nassif, and Carlos Silva-Jáuregui Index

51

81

109

137

v

Comparative Assessment of Industrial Policy

vii

Acknowledgments

This volume is genuinely a group effort and would not have materialized without the significant contributions of several individuals. I would like to single out the chapter authors who provided the research that gives this volume its real value: Hasan Ersel, Alpay Filiztekin, Najib Harabi, Jennifer Keller, Nihal El-Megharbel, Mustapha K. Nabli, Claudia Nassif, Marcus Noland, Howard Pack, and Carlos Silva-Jáuregui. My deep appreciation also goes to those who made insightful comments on the papers themselves: Heba Handoussa, Bernard Hoekman, Hanaa Kheir-El-Din, Samiha Fawzy, Gouda Abdel-Khalek, and Mounir Abdel Nour. In addition, I am grateful to the participants in the ECES conference, and in particular those who ably moderated the sessions and contributed to the discussion: Galal El Zorba, Sultan Abu Ali, and Adel Beshai. My thanks also go to the ECES administrative and financial staff for a very well-organized conference. Last but not least, I would like to express my appreciation to Yasser Selim for carefully editing and reviewing the manuscript and final proofs, and to Neil Hewison and Abdalla Hassan at the American University in Cairo Press for their commendable efforts in the production of this volume.e

vii

Comparative Assessment of Industrial Policy

Contributors Hasan Ersel, TEPAV/ EPRI, Turkey Alpay Filiztekin, Sabanci University, Faculty of Arts and Social Sciences Ahmed Galal, Egyptian Center for Economic Studies Najib Harabi, World Bank Jennifer Keller, World Bank Nihal El-Megharbel, Egyptian Center for Economic Studies Mustapha K. Nabli, World Bank Claudia Nassif, World Bank Marcus Noland, Peterson Institute for International Economics Howard Pack, Wharton University, Pennsylvania Carlos Silva-Jáuregui, World Bank

ix

ix

Comparative Assessment of Industrial Policy

Replace Manualy

xi

xii

Ahmed Galal

Comparative Assessment of Industrial Policy

1

CH A P T ER 1

Comparative Assessment of Industrial Policy in Selected MENA Countries: An Overview Ahmed Galal

No debate in the development literature has persisted as long and with such intensity as that related to government intervention in economic activity. In the last half century alone, views and actual policies have changed considerably. In the 1950s and 1960s, it was believed that markets failed widely and government intervention was necessary to speed up the process of economic transformation and the rate of economic growth. Most developing countries, including those in the Middle East and North Africa, adopted import substitution strategies in conjunction with high levels of protection, central planning, public ownership, and nonuniform policies across sectors and activities. In the 1970s and 1980s it became increasingly evident that governments fail too, and in many cases they fail even more than markets. So, the pendulum swung in the opposite direction. Promarket reforms were adopted, especially in the 1980s, frequently with the support of the World Bank and the IMF. The Washington Consensus emphasized macroeconomic stability, trade and price liberalization, privatization, and competition as key ingredients for rapid economic growth. The experience of the last couple of decades has given grounds for rethinking the balance between governments and markets. On one hand, the colossal failure of the socialist system in Eastern Europe and the Soviet Union supported the argument that markets were the best mechanism for allocating resources and motivating economic agents. On the other hand, the failure of market reforms in Latin America in achieving high and shared economic growth demonstrated that government intervention could do 1

2

Ahmed Galal

some good (De Ferranti et al. 2002). The pendulum has thus swung to the middle. Extreme views aside, there is a broad consensus that markets and governments can play positive and complementary roles in achieving more rapid and shared economic progress. The evolution in thinking about the role of the state in economic development is very similar to the evolution in thinking about industrial policy, defined by Pack and Saggi (2003) as “any type of selective intervention or government policy that attempts to alter the sectoral structure of production toward sectors that are expected to offer better prospects for economic growth than would occur in the absence of such intervention, i.e. in the market equilibrium.” Currently, supporters of industrial policy argue, for various economic reasons that will be discussed below, that selective intervention is essential for economic diversification and long-term gains in productivity. They cite the success of industrial policy in East Asia as evidence in support of their view (World Bank 1993; Rodrik 2006). Opponents doubt the success of industrial policy anywhere and argue that the contribution of selective intervention in economic progress in East Asia is very modest (Krueger 1980; Noland and Pack 2003). The broad conclusion of the empirical evidence is mixed. In the MENA region, systematic empirical evidence regarding the consequences of industrial policy is rare if not nonexistent. Nevertheless, many countries are actively engaged in attempts to alter the structure of their economies in an effort to gain new competitive advantages. Surely, the level of intervention has certainly subsided in the last couple of decades, but the legacy of selective interventions lingers on. Differentiated protection from foreign competition continues, and preferred sectors continue to receive tax exemptions, implicit subsidies of inputs (such as energy), or explicit tax exemptions. Moreover, investment promotion agencies are busy advocating certain projects or zones over others. These policies may or may not be good for economic development, no one really knows. The purpose of this volume is to contribute to the debate on industrial policy in general and in the MENA region in particular. The following five chapters attempt to address these broad questions: 1. Has industrial policy worked in the MENA region? 2. What lessons can the region derive from the experience of East Asia? 3. Finally, what are the political economy factors that produced the current industrial policy in the region? The first question is addressed by carefully assessing the experiences of Egypt, Morocco, and Turkey in Chapters 2 through 4. The second ques-

Comparative Assessment of Industrial Policy

3

tion is addressed by reviewing the relevance of the East Asain experience to the MENA region in Chapter 5. Finally, the political economy question is addressed in Chapter 6. The remainder of this overview chapter does two things: First, it offers a brief account of the arguments for and against industrial policy; Second, it offers succinct answers to the three questions posed above—without pretending to be a substitute for the more careful and detailed analysis found in subsequent chapters. I. The Industrial Policy Debate The literature on industrial policy, both theoretical and empirical, is extensive and inconclusive. We have no intention of reviewing that literature here.1 Rather, we are interested in analyzing the following case studies within the context of the current debate. For this reason, we offer only a brief summary of the rationale for and arguments against industrial policy, followed by our analysis of the case studies of Egypt, Morocco, and Turkey. The Rationale for Industrial Policy Traditionally, the rationale for industrial policy was linked to the infant industry argument. This argument is based on the notion that new industries will not be able to compete against their rivals, especially foreign competitors, because they initially incur high production costs. Protection and other forms of direct and indirect subsidies (such as tariffs or cheap credit) enable these firms to grow, increase productivity, and reduce the cost of production over time. Without support, Baldwin (1969) argued, entrepreneurs would not have the motivation to invest in knowledge acquisition because of knowledge spillover, train their workers because of labor mobility, produce new products with static positive externalities because they could not internalize the benefits, or undertake new projects if the initial cost of assessing these projects is high. From the perspective of society, extending support to such activities is justified as long as the discounted stream of benefits generated from learning-by-doing outweigh the discounted stream of subsidies (Pack and Saggi 2006). Hausmann and Rodrik (2003) and Rodrik (2004) reformulated the arguments for industrial policy, emphasizing two types of market failures that weaken the motivation of entrepreneurs to diversify in low-income economies: information externalities and coordination externalities. With

4

Ahmed Galal

respect to information externalities, they point out that the diversification of the productive structure requires entrepreneurs to ‘discover’ the cost structure of new activities through random experimentation with new products and the adaptation of foreign technologies to local conditions. By providing support to this process, countries would be able to move beyond specializing in products in which they currently have a comparative advantage to products in which they could acquire one. Once the discovery is made by one entrepreneur, it is followed by imitative entry by others. In support of the randomness of the process, they point out several examples of countries that have very similar factor endowments, but end up specializing in different types of products. Among these examples, they cite Bangladesh and Pakistan, two countries that appear to have similar initial resources. Yet Bangladesh exports a substantial number of hats while Pakistan exports virtually none. At a higher level of income, they indicate that Korea is a major exporter of microwave ovens but exports few bicycles; this pattern is reversed in Taiwan. A similar point is made regarding the successful cases of garments in Bangladesh, cut flowers in Colombia, and the IT industry in India. Finally, they use the Chilean experiment in promoting the salmon industry to point out that a state entity can successfully act as entrepreneur. Beyond the examples cited above, perhaps the most compelling argument in favor of industrial policy is the observation that few developed or newly developed economies have made the transition without some kind of industrial policy that ignited a process of diversification. And it was because of diversification, spillover, and the sharing of knowledge that these economies were able to move to a higher and more sustainable level of economic growth and prosperity. With respect to coordination externalities, the point is simple but compelling. Many projects require simultaneous, large-scale investment in order to be profitable. Investment in one project is not profitable without investments in other related projects. If the fixed cost of these other projects is high and no one is playing the role of coordinator, none of the investments in that industry will take place. The coordination failure is particularly acute where new industries exhibit scale economies in the presence of nontradable inputs (Rodrik 1996). It is also a common characteristic of most lowincome countries. The need to coordinate investment and production decisions, especially in the early stages of development, is not new. The idea finds its origin in the ‘big push’ strategies of development, and more recently in the concept

Comparative Assessment of Industrial Policy

5

of clusters in particular sectors, such as tourism and pharmaceuticals. The practical problem is that most industries tend to operate as clusters, although many of them can operate without clusters as pointed out by RodriguezClare (2004). This observation led Rodrik (2004) to argue against extending support to specific sectors. Instead, he argues in favor of supporting the adoption of new technologies, the development of new products, and training that equips workers with the skills associated with new techniques. Supporting existing establishments or traditional products does not necessarily help the process of diversification and economic growth, and may result in lost resources to society. The Case against Industrial Policy Notwithstanding the strong appeal of arguments for industrial policy, the counterarguments seem equally powerful. To begin with, the success of industrial policy hinges on the assumption that the government is better informed than the private sector about potential winners, their geographical location, and the appropriate technologies. It is also based on the assumption that governments can identify instances of coordination failures and design support schemes that generate more benefits than costs. However, both assumptions may not necessarily hold in practice, and the private sector may in fact be better informed on these counts. Imperfect information on the part of the government is further exacerbated by the lack of penalties for bureaucrats who make bad decisions. Bureaucrats rarely pay for their mistakes and politicians are not typically penalized through the ballot box in less than democratic countries. Thus, societies may be better off if their governments refrain from adopting an active industrial policy, and focus instead on only what government can and should do. The latter includes protecting property rights, enforcing contracts, and providing sound policies. The second counterargument is that governments may not always do what is best for advancing the development process. Motivated by the desire to stay in power, governments are likely to use industrial policy to favor their political supporters at the expense of their opponents. In addition, because industrial policy favors some business ventures and not others, it could lead to corruption and rent-seeking behavior on the part of some bureaucrats and private entrepreneurs (Nogues 1990). The above arguments are not merely fear-based but are in fact supported by empirical evidence, which broadly suggests that industrial policy has

6

Ahmed Galal

either been ineffective or has been abused, with minimal returns to society (Krueger 1980; Pack 2000; Noland and Pack 2002). Without restating the findings of this literature, suffice it to note that it is widely believed that industrial policy in Latin America resulted in the inefficient allocation of resources, discrimination against exports, and even a deterioration of income distribution (Edwards 1994; Noland and Pack 2003). Even in East Asia, where industrial policy is believed to have worked, there is evidence that industries that received support did not experience higher productivity growth in comparison with those that received none (Pack 2000). There is also evidence that industrial policy promoted capital-intensive sectors at the expense of employment creation, or low export performers. More recently, the Asian crisis of the late 1990s has been partly blamed on earlier government direction of credit (Noland and Pack 2002). The Bottom Line It is clear from the above discussion that there are strong arguments for and against industrial policy. The empirical evidence is equally divided, offering support for the claims of both advocates and opponents. The dilemma is that industrial policy is needed to enable developing countries to escape the trap of specialization in a few traditional commodities. But there are no guarantees that this policy works. Even if the right intervention is made, politics, rent-seeking behavior, corruption, and weak institutions could stand in the way of the benefits of industrial policy. From the perspective of learning from experience—which is the subject matter of this volume—one way to resolve the above dilemma is to modify the question being posed. Instead of merely asking whether industrial policy is needed or not, it is probably more useful to ask whether or not it has worked in a specific context and to identify the conditions that have produced observed outcomes. These are essentially the questions addressed in the case studies of this volume. II. Summary of Findings No summary can do justice to the rich details and rigorous expositions presented in the chapters that follow. Nevertheless, it is useful to capture the essence of the answers they provide to the three questions raised at the outset of this introduction. These questions and answers are briefly elaborated below.

Comparative Assessment of Industrial Policy

7

Has Industrial Policy Worked in the MENA Region? The short answer for Egypt, Morocco, and Turkey is ‘no.’ Active industrial policy may have increased diversification in the early stage of import substitution, given the initial narrow industrial base. However, the evidence does not support the view that industrial policy in recent years is effective in achieving further diversification, let alone improving total factor productivity over time. In Chapter 2, Galal and El-Megharbel assess the merits of selective intervention in the Egyptian manufacturing sector, disaggregated into sixteen industries, over the period spanning 1980–2000. They measure the extent of diversification in the manufacturing sector during this period along with changes in total factor productivity (TFP). The results do not support the view that industrial policy in manufacturing in Egypt led to increased diversification in the 1980s and 1990s. Moreover, they find that industrial policy variables were insignificant explanatory variables of the variations in the changes in TFP of different industries during the same period. They attribute the poor results to the poor design of industrial policy. The story of Turkey is not much different. In Chapter 3 Ersel examines the effectiveness of government support to private investment in Turkey during the period 1980–2000. He does so by quantitatively assessing the effects of incentives on TFP, employment growth, and investment, and finds no correlation between industrial policy and outcomes. He then conducts a survey of the private sector to find out whether investment incentives played an important role in their decisions; he concludes that other factors were more important. He deduces that investment incentives in Turkey were not implemented to promote or guide investments per se, but as a form of compensation for deficiencies in the investment environment and possibly to gain the support of allies. Finally, in Chapter 4 Harabi conducts a comparable exercise in Morocco and reaches a similar conclusion. As in the previous two cases, he evaluates empirically whether industrial policies in Morocco have contributed to the growth of private firms or not. His analysis uses models of optimal firm size as a theoretical framework and relies on a field survey of 850 firms carried out under the auspices of the World Bank in 2004. The sample includes firms of different sizes and covers all major manufacturing industries. His main finding is that industrial policies played a very modest role in boosting firm growth, and possibly efficiency.

8

Ahmed Galal

What Lessons Can the Region Derive from the Experience of East Asia? In Chapter 5, Noland and Pack analyze the experiences of Japan, South Korea, and Taiwan, which are regarded as primary examples of countries that have derived great benefits by pursuing decidedly nonneutral policies with respect to the promotion of specific sectors and activities. They address a series of questions: Was industrial policy a major source of growth in these three economies? Can these outcomes be duplicated in the Middle East today, or do special circumstances or changes in the international policy environment prevent replication of the East Asian experience? Given the revealed costs and benefits, is replication advisable? And if not, are there other positive lessons that Middle Eastern countries can learn from the experiences of the East Asians? Their answers are not encouraging to policymakers in the MENA region currently engaged in selective intervention to promote specific sectors. To start with, the authors point out that their sample of countries adopted a coherent set of industrial policies, focusing on promoting research and development (R&D), through direct and indirect subsidies, encouraging the adoption of new technology through foreign direct investment (FDI), rewarding export performance, and in some cases according domestic industries some rents until they matured. The impact of these policies on performance was positive but modest relative to their remarkable overall growth. The rapid accumulation of human and physical capital played a much more central role. On the basis of the above conclusions, recent changes in the global rules of the game, and institutional capacity in the MENA region to design and implement effective industrial policy, Noland and Pack (2003) argue that the region will be better off focusing on other reforms. Their concluding sentence is quite revealing: “Improvements in the economic environment in the macroeconomic, microeconomic, and institutional dimensions are more likely to contribute to accelerated growth and enhanced welfare than sector-specific ‘picking winner’ strategies.” What are the Political Economy Factors that Produced the Current Industrial Policy in the Region? In Chapter 6, Nabli, Keller, Nassif, and Silva-Jáuregui have taken on the difficult issue of the political economy of industrial policy in the MENA region. They provide a framework to explain why the MENA region has lagged behind other regions in reducing selective intervention over time,

Comparative Assessment of Industrial Policy

9

review the history of industrial policy in the region in the context of a larger social contract between the state and citizens, and offer a way forward. They attribute the slowdown in the transition from excessive vertical intervention in the MENA region relative to Latin America, East Asia, Central Asia, and Europe to two main factors. The first is that the results of these policies have not been ‘bad enough.’ The region has not undergone the dramatic political changes of Eastern Europe or the economic crises of Latin America and East Asia. Moreover, oil has provided a cushion and a sense of economic health, even if growth has been sluggish on average, and unemployment high. The second factor relates to the weak position of those with an interest in lobbying for a shift toward horizontal industrial policy, including exporters. At the same time, the privileged networks that have the capacity to influence policies to preserve their rents are relatively strong. No wonder the region is trapped into an industrial policy that does not necessarily generate the potential benefits from such policies, even if well implemented. In conclusion, no wonder the region is trapped into an industrial policy that does not necessarily generate the potential benefits from such policies, even if well implemented. The above review and the following chapters reveal the need for a serious rethinking of the role of governments in the MENA region, especially in the industrial policy area. Good intentions and national aspirations are certainly necessary to make progress, but it is essential to bolster those efforts with the empirical lessons of economic development experiments elsewhere. The objective of this volume is to make a contribution to this rethinking of the role of the state.

Notes 1.

For a recent survey, see Pack and Saggi (2006) and Rodrik (2000).

References Baldwin, Robert. 1969. “The Case against Infant Industry Protection.” Journal of Political Economy 77 (3): 295–305. De Ferranti, David, Perry E. Guillermo, Daniel Lederman, and William F. Maloney. 2002. From Natural Resources to the Knowledge Economy. Washington, DC: World Bank. Edwards, Sebastian. 1994. “Trade and Industrial Policy Reform in Latin America.” NBER Working Paper 4772. Cambridge, MA: National Bureau of Economic Research.

10

Ahmed Galal

Hausmann, Ricardo, and Dani Rodrik. 2003. “Economic Development as SelfDiscovery.” Journal of Development Economics 72 (2): 603–33. Krueger, Anne. 1980. “Trade Policy as an Input to Development.” NBER Working Paper 466. Cambridge, MA: National Bureau of Economic Research. Nogues, Julio. 1990. “The Experience of Latin America with Export Subsidies.” Weltwirtschaftliches Archiv 126 (1): 97–115. Noland, Marcus and Howard Pack. 2002. “Industrial Policies and Growth: Lessons from International Experience.” Working Paper 169. Chile: Central Bank of Chile. _____. 2003. “The Asian Industrial Policy Experience: Implications for Latin America.” LAEBA Working Paper 13 (April). Tokyo: Latin America/ Caribbean and Asia/Pacific Economics and Business Association. Pack, Howard. 2000. “Industrial Policy: Growth Elixir or Poison?” World Bank Research Observer 15 (1): 47–67. Washington, DC: World Bank. Pack, Howard and Kamal Saggi. 2006. “The Case for Industrial Policy: A Critical Survey.” World Bank Policy Research Working Paper 3839. Washington, DC: World Bank. Rodriguez-Clare, Andres. 2004. “Clusters and Comparative Advantage: Implications for Industrial Policy.” Washington, DC: Inter-American Development Bank, mimeo (June). Rodrik, Dani. 1996. “Coordination Failures and Government Policy: A Model with Applications to East Asia and Eastern Europe.” Journal of International Economics 40 (1): 1–22. _____. 2000. “Trade Policy Reform as Institutional Reform.” Unpublished paper. Kennedy School of Government, Harvard University. _____. 2004. “Industrial Policy for the Twenty-First Century.” KSG Faculty Research Working Paper RWP04-047. Kennedy School of Government, Harvard University. Wade, Robert. 1990. Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization. Princeton, NJ: Princeton University Press. World Bank. 1993. The East Asian Miracle: Economic Growth and Public Policy. London: Oxford University Press.

CH A P T ER 2

Do Governments Pick Winners or Losers? An Assessment of Industrial Policy in Egypt Ahmed Galal and Nihal El-Megharbel

As in other developing countries, selective government intervention in Egyptian economic activities dates back to World War II. At that time, a shortage of products in the global markets and the drive for rapid diversification and industrialization led Egypt to erect high trade barriers to protect domestic industries and provide support to large national projects. This trend peaked in the late fifties and early sixties when the government nationalized most industries, adopted differentiated and high levels of protection, provided subsidies, and controlled prices of inputs and outputs as well as interest and exchange rates. For almost a decade and a half, markets played a modest role in resource allocations and planning was the name of the game. Notwithstanding some gains in diversification and productivity, the import substitution strategy eventually began to show its limits. And in the mid-1970s, Egypt opted for a partial liberalization of the economy under the name of infitah or ‘open door policy.’ Ironically, partial liberalization amounted to increased selective intervention as segments of the economy were liberalized while others were left unaffected. The next change came in the early 1990s, when Egypt embarked on a program that involved price and trade liberalization, privatization, reduction of subsidies and taxes, and deregulation. This process is still unfolding today, but selective intervention can still be seen, for example, in the form of implicit energy subsidies, tax exemptions, subsidized interest rates for SMEs, and underpriced land. With such a long history of experimentation with industrial policy in Egypt, it is important to ask: Has this policy worked? If yes, can this success 11

12

Ahmed Galal and Nihal El-Megharbel

be attributed to industrial policy variables or something else? Finally, what lessons can we draw from this experience to keep in mind when designing industrial policies in the future? These are the questions we address in this chapter. In answering these questions, we assess the effectiveness of industrial policy in two ways. First, we measure the performance of the entire manufacturing sector—disaggregated into sixteen industries—over the period spanning 1980–2000, using different measures of diversification and changes in total factor productivity (TFP) as yardsticks. Second, we explore the extent to which industrial policy has contributed to observed variations in performance by regressing TFP change on a host of industrial policy and sector-specific variables. These results provide information about the effectiveness of industrial policy, but offer no explanation as to why this policy may or may not have worked. Thus, we devote Section III to assessing the design of industrial policies in Egypt, and offer concluding remarks and policy recommendations in Section IV. Our bottom line is that the manufacturing sector in Egypt did not witness significant diversification or improved TFP during the last two decades. The variations in performance across different industries cannot be attributed to industrial policy variables. This result is due in large measure to the way industrial policy was designed, especially in comparison with recommended designs and the experience of East Asia. We thus argue that Egypt now needs to reorient its industrial policy to focus on new products, measurable performance targets, and broad rather than sector-specific support for prespecified durations. I. Performance of the Manufacturing Sector in Egypt, 1980–2000 If industrial policy had been effective, we should have observed a greater diversification of the Egyptian economy in general and the manufacturing sector in particular over time. We should also have observed TFP improvements as firms acquired or adapted new technologies and know-how, as information sharing and knowledge diffusion took place, and as coordination problems were resolved. The question we address in this section is whether these expectations were fulfilled or not. We answer this question by measuring diversification and TFP change for the entire manufacturing sector over the period 1980–2000. The analysis covers sixteen specific industries. Although we would have liked to assess the performance of all productive sectors in the Egyptian economy since 1960,

13

Do Governments Pick Winners or Losers?

we believe that the diversity within the sixteen industries and the time span covered offer sufficient variations to test the effectiveness of industrial policy. Diversification One measure of diversification is the Herfindahl-Hirschman Index (HHI), which is calculated as follows: 2

∑ ⎛⎜ X i X ⎞⎟ − 1 N N

HHI =

i =1

⎝

⎠ 1− 1 N

where Xi is the share of output of the ith industry in total output X, and N is the number of industries. The HHI takes on the value of 0 in the case of complete diversification and the value of 1 in the case of maximum concentration. The results of the calculation are shown in Figure 1. Figure 1. HHI Index of the Manufacturing Sector in Egypt, 1980/81–1998/99 HHI of Manufacturing Output

0.30 0.25 0.20 0.15 0.10

0.250.25 0.18

0.05 0.00 1980/81

1983/84

1986/87

1989/90

1992/93

1995/96

1998/99

Source: Authors’ calculations.s

The most striking observation in Figure 1 is that the manufacturing sector in Egypt has become more concentrated over time.1 That is not to say that there were no new products or technologies during the 1980s and 1990s. Casual observations suggest otherwise. For example, Egypt now produces new IT products and exports ceramic products. However, these and other new products were not produced on a large enough scale to make the index of the manufacturing sector more diverse over time.

14

Ahmed Galal and Nihal El-Megharbel

Another way of looking at the issue of diversification is by checking whether the list of major export items continues to be concentrated in a few products or not. On the basis of the information provided in Table 1, the trend is disappointing. In fact, the twelve most important export items accounted for 59 percent of total exports in 2003 compared with only 30 percent in 1983. Moreover, although there were a few new products on the list (such as inorganic chemicals, sanitary products, and coal), their export values were not very high. For most of this period, petroleum products, textiles and clothing, and iron and steel were the main export items. Table 1. Top Twelve Export Items in Egypt, Ranked by 1983 Exports (US$ m)8 1983

1986

1992

1995

2000

2003

Petroleum Products

469

348

153

496

1,555

2,110

Textile Yarn, Fabrics, Made-up Articles

253

432

395

571

413

279

Aluminum

96

165

188

198

128

92

Oils and Perfume Materials, Toilet and Cleansing Preparations

33

19

29

35

53

52

Manufactures of Metals

23

15

52

48

Clothing

18

35

163

253

314

233

Sugar, Sugar Preparations, and Honey

14

20

Printed Matter

13

19

Vegetables, Preserved or Prepared

12

8

17

25

Iron and Steel

10

16

138

160

133

376

Fertilizers

8

2

44

65

78

70

Medical, Pharmaceutical Products

8

10

29

35

50

51

Rice

57

57

104

150

Footwear

20

10 82

124

358

37

46

57

70

59

Inorganic Chemicals Sanitary, Plumbing, Lighting Fixtures, and Fittings Coal, Coke, and Briquettes Share in total exports (%)

30

37

37

Source: UN International Trade Statistics Yearbook, different issues.

42

Do Governments Pick Winners or Losers?

15

Finally, we compare the level of diversification in Egypt with that of other countries with a similar level of per capita income, using a diversification index compiled by UNCTAD/WTO. The data provided in Table 2 indicate that the manufacturing sector in Egypt is less diversified on average than the sample of listed countries.2 The concentration in the manufacturing sector in Egypt is particularly noticeable when it comes to sophisticated products like chemicals, nonelectric machinery, and electronic components. The gap is also more apparent when Egypt is compared with countries like Brazil, China, and Indonesia with respect to most products.

7 8 11 17 6 4 11

21 38 24 6 2 6

14 11

34 25

5 5 2 12 9 4 5 6 2 6

8 4 1

9 6

19 35 5 29 3 6 28 1 90 61 23 15 30 23 23 9 13 12 6 6 2 13 12 20 20

14 35

34

41 12 13 10 22

31 19

3 14

27 20

19 3

35

10

7 18 25

2 16

12 44 14 37 5 38 12 5 59 28 19 24 34 14 10 15 2 15 36 31 7 5 6

18 15 22 6 8 25 22 2 60 27 19 26 37 15 16 5 6 28 18 3 2 15

5 20

23 18

Source: UNCTAD/WTO, International Trade Center: http://www.intracen.org/countries. * Higher index value reflects higher product diversification.

Minerals

4 32 14 5

Misc. Manufacturing

9 12 50 38 31 3 28 13 43 20 26

Clothing

Electronic Components

4 4 9 7

Nonelectric Machinery

8 40 28 16 28 5 66 4 137 47 7 3 19 13 11 10 3 13 27 31 8 4 4 6 20 23

Basic Manufactures

24 43 21 33 36 25 17 3 72 91 12 24 6 13 7 20

Leather Products

7 12 9 10 4 12 7 6 29 25 8 6 15

Chemicals

17 5 14 18 2 9 7 3 39 17 9 11 10 7 14 9 5 6 23 8 3 1 10 4 6 10

Textiles

9 7 4 9 10 8 7 5 26 12 8 12 9 8 2 9 6 3 13 7 2 3 4 3 4 8

Wood

Processed Food

Egypt Indonesia Colombia Romania Peru Sri Lanka Brazil Bolivia China Thailand Tunisia Morocco Ukraine Syria El Salvador Jordan Jamaica Ecuador Bulgaria Philippines Paraguay Armenia Honduras Kazakhstan Guatemala Average

Fresh Food

Table 2. Product Diversification Index of Manufacturing Products in Egypt Compared to Lower Middle-Income Countries, 2003*m

3 4 3 2 6 3 5 3 7 6 2 5 8 1 2 2 1 1 2 2 2 3 2 2 3

16

Ahmed Galal and Nihal El-Megharbel

Collectively, the HHI index, the composition of major export items, and the UNCTAD/WTO diversification index suggest that industrial policy in Egypt has not led to a level of diversification consistent with Egypt’s level of per capita income and long history of active industrial policy. It could be argued that this outcome is due to the limited time horizon analyzed in this study. Had the analysis been carried out using data from 1960 onward, industrial policy would have been associated with increased diversification. Furthermore, it could be argued that the observed concentration in recent years is due to promarket reforms in the 1990s and the maturation of the Egyptian economy over time. We will deal with these two arguments in turn. The first argument seems to be valid, even without further analysis. After all, we know that one of the slogans of the 1960s was “We produce everything from the needle to the rocket,” and policies were put in place to make that slogan come true. However, the question is whether or not the diversification that must have taken place then was always justified. The example of the auto industry suggests otherwise. Support to this industry for almost half a century produced only a number of relatively small factories, mainly for assembling imported parts to sell in the domestic market, behind high levels of protection. All of these factories operate at a much lower scale of operation than the industrial minimum found elsewhere. The other issue is whether or not industrial policy was designed to minimize production costs and maximize benefits? We return to this point in Section IV. As for the second point, we are less convinced that the reforms of the 1990s are to blame for the increased concentration in the manufacturing sector over time, given that industrial policy influences the future pattern of the structure of the sector, not its past. So, if weakened selective intervention in the 1990s were to have an impact, that impact would probably be felt a decade later. As for the argument that the Egyptian economy has matured enough to move into a phase of focusing on a few sectors, as observed by Imbs and Wacziarg (2003), the fact is that Egypt is still too poor to have moved into that phase yet.3 Total Factor Productivity (TFP) If industrial policy is about providing support to industries that initially incur high costs on the premise that performance will improve over time, we would expect industries that received support to perform better than

Do Governments Pick Winners or Losers?

17

those that did not. To find out whether this assumption is true, we estimated TFP for the sixteen industries comprising the manufacturing sector in Egypt over the period spanning 1980–2000. The methodology we used is explained in the technical appendix at the end of the paper. Suffice it to note here that our TFP estimates are made using the Malmquist index, using a data-envelopment analysis (DEA) for cross-industry analysis of TFP growth. This approach requires fewer restrictions than other approaches. It is based on constructing a linear production frontier for each year. The frontier production function is constructed by the solution of a sequence of linear programming problems, one for each year. The degree of technical inefficiency is the distance between the observed data point and the frontier. The data requirement to compute TFP was extensive. Furthermore, information about output, intermediate inputs, capital, and labor was not readily available in a convenient format for immediate use. Even more demanding was the calculation of the price indices to deflate outputs and inputs. The data set we used, its sources, and manipulations are given in the technical appendix of this chapter. The TFP estimates are given in Table 3 below for each industry of the manufacturing sector. These estimates point out that: • TFP change averaged 0.75 percent a year over the period 1980– 2000. • The peak of productivity improvement was seen in the first half of the 1990s, a period during which active industrial policy is supposed to have diminished. • The standard deviation is quite high, especially in the 1980s. This variance is even more noticeable for individual industries, turning frequently from negative to positive TFP change.

Ahmed Galal and Nihal El-Megharbel

18

Table 3. TFP in Manufacturing Industries in Egypt, 1980/81–2000/01 TFP Growth Sector

Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Mean Standard Deviation

1980/81– 1994/85

1985/86– 1990/91

1991/92– 1995/96

1996/97– 2000/01

1980/81– 2000/01

-0.46 -0.04 0.67 1.61 -1.25 0.46 1.72 0.55 0.96

1.48 0.96 2.16 -0.27 0.62 -0.30 0.75 -0.30 5.39

1.42 1.72 1.89 -0.90 2.44 1.70 -0.42 1.11 -0.57

0.67 0.59 0.59 1.32 0.77 5.44 1.17 1.06 -0.24

0.75 0.81 1.33 0.44 0.65 1.83 0.81 0.61 1.39

1.36

2.40

2.78

-0.65

1.47

0.10 0.57 1.55 1.76 -0.06 1.29 0.67 0.84

2.33 0.30 -1.56 -1.29 1.92 0.86 0.97 1.64

3.01 0.88 -0.75 0.85 1.91 -0.48 1.04 0.26

-2.48 -0.14 -0.92 0.02 -1.38 -0.96 0.30 0.67

0.74 0.40 -0.42 0.34 0.60 0.18 0.75 0.53

Source: Authors’ calculations.

Overall, productivity improvements were modest and the results exhibit significant variations across sectors and over time. These variations provide the basis for explaining what may have caused them, which is what we do in the next subsection. II. The Contribution of Industrial Policy to Performance Notwithstanding the low average TFP change in the manufacturing sector in Egypt throughout the 1980s and 1990s, there were significant variations in performance across industries. In this section, we attempt to explore whether TFP improvements are associated with active industrial policy or not. To this end, we ran a two stage least square regression (2SLS) of the following equation:

TFP = β it

where

it

+γ

X

it

+ φ Z it + U it , i = 1,……….N, t = 1,………T

(4)

Do Governments Pick Winners or Losers?

TFP

X

it

it

19

is the TFP change of industry i in period t,

denotes a set of sector specific and exogenous variables,

Z

it

denotes a set of industrial policy variables, and

U

it

is the error term.

The 2SLS technique was used to overcome the possible presence of the endogeneity of different policy variables. The specific equation that we estimated and the results obtained are shown below: TFP = -1.7 - 0.03*ERP - 14.91*SUB+ 0.66*GDP + 0.06*KL + 0.03*FIRM (-1.9) (-6.3) R-square: 0.96

(-19.4)

(9.7)

(23.4)

(0.72)

DW: 1.99

The estimated regression is quite satisfactory, with R-square explaining 96 percent of the variations in TFP. However, the results do not support the hypothesis that industrial policy variables were associated with improved TFP change. More specifically, the estimated equation includes the three industrial policy instruments that are most frequently used in Egypt to support specific industries: namely, effective rates of protection (ERP), subsidies (SUB), and barriers to entry (Firm).4 ERP is estimated using Corden (1966), subsidies are defined as the ratio of explicit transfers to each industry divided by the total subsidy to all industries, and market structure is measured by the ratio of the number of firms in each industry relative to the total number of firms in the manufacturing sector. These variables were estimated for each industry for the entire period analyzed. Our expectations were that these variables would have a positive sign if industrial policy had been effective in improving performance over time.5 This hypothesis was not supported by the data. Industries that received greater protection and subsidies performed less well than industries that did not. Similarly, industries that operated in relatively less competitive markets performed less well than industries that faced greater competition. Rather than benefiting

20

Ahmed Galal and Nihal El-Megharbel

from support to overcome the initially high costs of production, supported industries seem to have relaxed and exerted less effort than was needed for industrial policy to be beneficial. In addition to the above industrial policy variables, we included two other variables: capital intensity (measured by KL ratio) and GDP growth rate. Capital intensity was included to neutralize the effect of variations in technology across industries, and GDP was included to capture the effect of demand on capacity utilization, thus TFP. Both variables were found to have a positive sign, as expected, and both were significant. Of course association is not the same as causation, and the results should be interpreted with caution. Nevertheless, the analysis can at least be taken to question the usefulness of industrial policy in Egypt with respect to TFP improvement. As for diversification, although the policy may have made a contribution at an earlier stage, protection of infant industries may have lead to a permanent state of infancy in some instances. Subsidies may have led to excessive expansion and little motivation to improve productivity. In the process, industrial policy may have encouraged some rent-seeking behavior and resistance to openness. The next question is: why was industrial policy in Egypt less successful than hoped for? III. Assessment of Industrial Policy Design in the Manufacturing Sector in Egypt Starting from the premise that the Egyptian economy needs an effective industrial policy to grow more rapidly and become more diversified and dynamic, we focus in this section on understanding why previous manufacturing-related industrial policies did not work. Our assessment is based on contrasting Egypt’s industrial policy with lessons gleaned from the literature and the experiences of East Asia and Latin America. In a very fundamental way, the design of industrial policy is about resolving tensions between different choices. The four most important areas of tension are the following:6 • The tension between supporting old versus new activities • The tension between providing support on the basis of convictions versus measurable outcomes • The tension between providing open-ended support versus timebound support • The tension between supporting activities versus supporting sectors

Do Governments Pick Winners or Losers?

21

The way the above tensions are resolved arguably makes the difference between effective and ineffective industrial policies. Indeed, it has been pointed out that the success of East Asia is due to a good resolution of these tensions, while the limited success or even failure of industrial policy in Latin America is due to the poor resolution of these tensions. In general, most analysts believe that industrial policy will be successful if it targets new rather than old activities, rewards entrepreneurs for measurable outcomes, extends support for a prespecified period of time, and supports activities rather than specific sectors (Rodrik 2004; World Bank 1993; Amsden 1989; and Wade 1990). The question is: how did Egypt resolve these tensions? Supporting Old versus New Activities With respect to the tension between supporting old or new activities, Egypt’s record is somewhat mixed and seems to have evolved over time. At the initial stage of industrial policy in the 1960s, the tension was resolved in favor of supporting new activities, perhaps because there were very few to begin with. This policy led to diversification and, in some instances, the successful creation of new areas of comparative advantage. With a weak or weakened private sector, the government took it upon itself to play the role of an entrepreneur, initiating new industries such as iron and steel, pharmaceuticals, and auto manufacturing. These initiatives were backed by high levels of protection, barriers to entry, and price control. In retrospect, public ownership was not always successful, but it left a legacy of local knowhow and physical infrastructure in a wide range of industries that allowed the private sector to flourish subsequently. From the mid-1970s to the end of the 1980s, the tension was resolved with a dual policy. First, existing activities continued to receive the support they had before. The trade regime remained highly protective, price control was pervasive, subsidies persisted, and restrictions on entry were common. Second, a new set of incentives was introduced, including tax exemptions for relatively large projects, sometimes because they were foreign and frequently because they were in certain sectors or geographical locations. Not much attention was paid to whether these projects expanded the capabilities of the economy to produce new products or new technology. More recently, especially starting in 1991, industrial policy began to decline. At the beginning of the decade, prices were liberalized, quantitative restrictions on imports removed, trade liberalization adopted in phases (the latest of which occurred in 2004), public ownership partially dismantled,

22

Ahmed Galal and Nihal El-Megharbel

and investment incentives somewhat streamlined. Nevertheless, there has been no conscious effort to provide support to new activities that have the potential to expand the capabilities of the Egyptian economy into new areas of comparative advantage. On the contrary, the highest effective rates of protection continue to be given to traditional industries like textiles and clothing, and leather products (see Table 4). Furthermore, one area where a new industrial policy is being pursued rigorously is with regard to small and medium enterprises (SMEs). While this support may be justified on other grounds, like employment creation, it is clearly derived from the notion of size rather than diversification into new products or the adoption/adaptation of new technology. Table 4. Nominal and Effective Protection in Manufacturing in Egypt, 2000 and 2004 (%) Nominal

Effective

Manufacturing sectors 2000

2004

2000

2004

Food

10.4

7.8

15.4

9.3

Textiles

24.0

9.2

27.6

10.3

Clothes and Footwear

38.3

26.7

43.4

31.6

Wood and Wood Products

12.9

7.3

12.4

6.9

Paper and Printing

15.6

10.2

15.0

9.7

Leather and Leather Products

30.0

29.5

34.4

36.1

Rubber

29.1

13.6

32.7

14.9

Chemicals

10.6

4.8

8.9

3.2

Nonmetallic

23.1

14.7

26.2

16.7

Basic Metal

12.5

5.9

11.0

3.7

Machinery & Equipment

14.3

8.7

14.1

8.8

Transport

33.6

18.1

38.3

20.4

Simple Average

21.2

13.0

23.3

14.3

9.8

8.0

11.9

10.5

Standard Deviation

Source: Galal and Refaat (2005).

Support on the Basis of Convictions versus Measurable Outcomes With respect to the tension between providing support on the basis of convictions or measurable outcomes, industrial policy in Egypt seems to have been based on the former. The only exception is a recent subsidy program

Do Governments Pick Winners or Losers?

23

that linked payment of subsidy to exports. Otherwise, all other support instruments were not linked explicitly to measures such as productivity, exports, or employment. The experience of East Asia was very different. Subsidized credit was conditioned upon meeting certain export targets. This performance-based reward must have put pressure on the recipients to improve productivity in order to compete internationally. No such system was adopted in Latin America, which may explain why industrial policy was not as effective in that region. The problems of failing to link support to measurable outcomes are obvious. It becomes difficult to judge the success or failure of a given policy intervention. As a result, both good and bad performers benefit at the expense of the rest of society. At the same time, substantial effort and resources tend to be expended on securing these advantages or lobbying against their removal. On their part, bureaucrats can claim their programs are successful in order to receive continued funding. Open-ended Support versus Time-bound Support The merit of announcing ex ante that support will be withdrawn at a certain date in the future is that beneficiaries will realize from the start that they will have to survive on their own at a given point in time. This prior knowledge would motivate them to do their best to succeed. An additional merit to including a sunset clause is that it saves scarce financial and human resources that could be put to other uses. Equally important, the time limit puts an end to activities that fail to generate a new comparative advantage once sufficient time for experimentation is allowed. It helps to cut losses rather than letting inefficient industries drain the rest of the economy. In the Egyptian context, industrial policy has had no sunset clauses. Perhaps the only exception can be found in the tax breaks given to investment, for durations specified by law. Otherwise, trade protection, subsidies, and entry restrictions have all been open-ended. To be sure, these policies changed from time to time, but the changes were brought about for reasons other than rationalizing industrial policy and phasing it out as industries matured. Supporting Activities versus Supporting Sectors The merit of focusing support on activities rather than sectors is that industrial policy can then be guided by the principle of correcting instances

24

Ahmed Galal and Nihal El-Megharbel

of market failures. Furthermore, the benefits of the policy will cut across different sectors, rather than benefiting some sectors and not others. By comparison, a focus on certain sectors is problematic because it is difficult to agree on which sectors to pick and which sectors to leave behind. It is true that industrial policy, even in East Asia, was not always indifferent with respect to sectors, but most of the support was either linked to performance targets or generic. Sufficient public resources were allocated to R&D, technical training and subsidized credit. Industrial policy in Egypt seems to favor specific sectors, with a particular focus on investment incentives. The preferred sectors range from tourism to land reclamation. It is true, however, that the government also allocates public resources to the development of science and technology, training, and subsidized credit. The problem with these programs concerns their effectiveness. Most expenditure on R&D is not oriented to meeting private sector needs. Training is not generally demand driven and subsidized credit now goes essentially to SMEs, which, as noted earlier, may be justified on social grounds, but is not necessarily compatible with moving the Egyptian economy into new areas of specialization. IV. Implications of the Findings for Future Industrial Policy in Egypt The above analysis, together with the previous findings related to the limited effectiveness of industrial policy, have strong implications for future industrial policies in Egypt, at least for the manufacturing sector. The most obvious implication is broad: there is a strong case for rethinking industrial policy as part of a process of reconsidering the role of the state in economic activity. The new version of industrial policy should aim at moving the economy into areas of new comparative advantage that go beyond the current patterns of production. Perhaps the most important principles of the new industrial policy are those that seem to have characterized the successful experience of East Asia. The main features of this policy are: • targeting new activities rather than existing ones; • rewarding entrepreneurs on the basis of measurable outcomes rather than on prior convictions; • providing support only for a prespecified period of time rather than making open-ended commitments; and • supporting activities with broad benefits rather than targeting specific sectors.

Do Governments Pick Winners or Losers?

25

These suggestions are general in nature and need further work in order to be translated into specific reform programs. Moreover, they abstract from a discussion of the best institutional arrangements to carry them out in such a way as to shield public officials from influence while engaging the private sector in a constructive dialogue about the best opportunities for a prosperous economy. Nevertheless, we hope that these suggestions can serve as a good starting point for a productive discussion about future industrial policy in Egypt, not only for the manufacturing sector but also for other productive sectors of the economy. V. Technical Appendix In this appendix, we briefly explain the methodology used to calculate total factor productivity (TFP). We also document the sources of our data and any adjustments we made to facilitate the analysis, as well as the data that might be utilized by other researchers. Total Factor Productivity Estimation We estimated TFP growth using the Malmquist index, calculated using the data-envelopment analysis (DEA). The DEA is a nonparametric mathematical programming approach to frontier estimation, which has an advantage over parametric techniques of assuming no specific functional form for the production function to estimate its parameters. The approach is based on constructing a linear production frontier for each year in the sample by the solution of a sequence of linear programming problems, one for each year. Technical inefficiency is determined by the distance between the observed data point and the frontier. The model starts by solving the following linear programming problems assuming constant returns to scale: max ö h ö, ë

n

s.t. ö h Yhq - ∑ ë i Yip

≤0

i =1

n

∑ë

i

K ip

≤ K ip

i =1 n

∑ ëL

ip

≤ L hq

i =1

ë 1 ,....., ë n

≥0

Ahmed Galal and Nihal El-Megharbel

26

The Malmquist TFP index, which was first introduced by Caves, Christensen, and Diewert (1982), measures the TFP change between two data points by calculating the ratio of the distances of each data point relative to a common technology (Krüger 2003). The estimation does not require information about input prices, nor does it require equating prices and marginal products. The index can be decomposed into two components: the first represents the change in productive efficiency and the second the rate of technological progress. The Malmquist index is calculated as the geometric mean of the ratio of two distance functions, which gives the maximum increase of output in one period to reach a boundary of the technology set in a previous period. Following Färe et al. (1994), the output-oriented Malmquist TFP change index between period t and period t+1 is given by: 1/2 ⎡ Dt ⎛ xt + 1 , y t + 1 ⎞ Dt + 1 ⎛ xt + 1 , y t + 1 ⎞ ⎤ ⎜ ⎟ ⎟ ⎜ ⎢ ⎥ h⎝ h h ⎠ h ⎝ h h ⎠ M t + 1 ⎛⎜ xt , y t , xt + 1 , y t + 1 ⎞⎟ = ⎢ ⎥ h ⎝ h h h h ⎠ t t t Dt + 1 ⎛⎜ xt , y t ⎞⎟ ⎥ ⎢ Dh ⎛⎜ xh , y h ⎞⎟ h h h ⎝ ⎠ ⎦ ⎝ ⎠ ⎣

(1)

which has an equivalent representation as:

(

)

M ht +1 x ht , y ht , x ht +1 , y ht +1 =

( ( E F

where

(

)⎡ D (x , y ) D (x , y ) ⎤ ) 1⎢⎣ D4 4(x4 4, y44 2)D4 4(x4 ,4y 4)4⎥⎦3

D ht +1 x ht +1 , y ht +1 Dt x t , y t 1 4 h4 2h 4 h43 t +1 h

t h t +1 h

t +1 h t +1 h

t +1 h t +1 h

T P

t h t +1 h

t h t h

t h t h

1/2

t +1 h

)

D ht +1 x ht , y ht denotes the distance from period t observation to the period t+1 technology.

E F E T P

t +1

is the change in productivity efficiency. h t +1 h

is the rate of technological change between the two periods; t and t+1.

If the index has a value greater than one, this indicates a positive TFP growth from period t to period t+1, while a value less than one indicates a decline in TFP. Data and Data Sources To compute TFP, we collected data on output and inputs for sixteen

Do Governments Pick Winners or Losers?

27

manufacturing industries in Egypt at the three-digit level of the ISIC, rev. 3 classification (see Table A.1 for a listing of these industries) over the period 1980/81–2000/01. The sources and data for each variable were obtained and processed as follows: • Output, material inputs, and labor: Data on output, intermediate inputs, and labor for the sixteen industries were compiled from the Annual Industrial Statistics Bulletin issued by CAPMAS. The data covered both public and private sector firms. The labor input was measured as the number of workers per industry. Material inputs data included local and imported inputs, packing materials, fuel, electricity, and spare parts. • Capital: As is customary, we employed the perpetual inventory method (PIM) to construct the capital stock series for different industries. Data on gross capital formation were obtained from the Annual Industrial Statistics Bulletin. The calculation of the capital stock involved estimating an initial capital stock for each industry. Starting from the initial capital stock, additions to the stock were added and depreciation was subtracted to obtain the capital stock for subsequent years (1981/82–2000/01). Gross capital formations used in the calculations were deflated using the GDP deflator. The initial capital stock was calculated for two categories of assets—land and buildings, and machinery and equipment, using the formula:

K = I (1 + g )/ (g + δ ), t

t

,where

K is the initial capital stock for period t, which was 1980/81 in our case. I is the gross capital formation for the base year taken as an average t

of investments over five years. g is the average growth rate of output over five subsequent years. δ is the depreciation rate (2.5 percent for buildings, and 8 percent for machinery and equipment). The decomposition of gross capital formation of each industry into these two categories was based on their shares in fixed assets as calculated from the CAPMAS publication Financial Statistics and Indicators. These shares are given below:

28

Ahmed Galal and Nihal El-Megharbel

Shares of Fixed Assets Components in Total Assets (%) ISIC Rev. 2

Industries

31 32 33 34 35 36 37 38 39

Food, Beverages, and Tobacco Textile, Apparel, and Leather Wood, Wood Products, and Furniture Paper and Paper Products, and Printing Chemicals, Petroleum, and Plastic Products Nonmetallic Mineral Products Basic Metals Fabricated Metal Products, Machinery, and Equipment Other Manufacturing Industries

Land & Buildings

Machinery & Equipment

28 28 40 20 23 23 24 27 14

72 72 60 80 77 77 76 73 86

Source: Authors’ calculations.

Price indices: Several price indices were calculated (Paasche, Laspeyres, Fisher, and Divisia), but the latter was the index used to deflate outputs and inputs of different industries in the process of estimating TFP. The price index of each industry was constructed on the basis of information about the values and quantities of more than two hundred commodities under each industry for each year of the twenty years under consideration. In addition to data to estimate TFP, the following variables were constructed to test the relevance of industrial policy variables to changes in TFP: • GDP growth rates: GDP growth rates for the period 1980/81– 2000/01 were calculated using data from the World Development Indicators. • Share of subsidies to total output: Data on direct subsidies obtained from the Annual Industrial Statistics Bulletin were used to compute the ratio of subsidies to output for the sixteen industries over the period 1980/81–2000/01. • Distribution of firms by industry: Data from the UNIDO industrial database on the number of firms by industry were used to calculate the share of the number of firms to total industrial firms. This index reflects the degree of concentration in different industries. • Effective rates of protection (ERP): ERPs for different industries were obtained from Refaat (1999). Finally, all data were filtered using the HP Filter (Hodrick and Prescott 1980) to smooth the data and to correct for real business cycles’ fluctuations. The actual data used are given Tables A.2 to A.9.

29

Do Governments Pick Winners or Losers?

Table A.1. List of ISIC Codes and Description of the Different Manufacturing Industries ISIC. Rev. 2

Manufacturing Sectors

311, 312 321 322 323 324 331 332 341, 342

Food Manufacturing Textiles Clothing and Apparel Leather and Leather Products Footwear Wood and Wood and Cork Products Furniture Paper and Paper Products, Printing, and Publishing Industrial Chemicals, Other Chemical Products, Petroleum Refineries, Miscellaneous Products of Petroleum, and Coal Rubber and Plastic Products Pottery and China Glass and Glass Products Other Nonmetallic Mineral Products Iron, Steel, and Nonferrous Metal Industries Fabricated Metal Products, Machinery, and Equipment Transport Equipment

351, 352, 353, 354 355, 356 361 362 369 371,372 381, 382, 383 384

Source: UNIDO 2003.

Table A.2. Output and Input Data: 1980/81–1984/857 Average Sector

Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average of All Industries Standard Deviation

Outputs (LE 000)

Inputs (LE 000)

Capital (LE 000)

5,488 3,553 258 75 120

4,440 2,410 170 61 65

2,215 4,793 380 178 213

145 311 9 4 7

104

72

126

7

97 1,118 4,825 657 85 102 806 2,428 2,960 962 1,477 1,810

65 744 2,798 502 37 75 353 1,374 2,055 909 1,008 1,291

47 1,112 8,420 353 50 103 2,852 3,675 2,140 1,380 1,752 2,300

5 33 96 19 4 12 47 65 79 32 55 79

Source: Authors’ calculations using CAPMAS data.

Workers (000)

30

Ahmed Galal and Nihal El-Megharbel

Table A.3. Output and Input Data: 1985/86–1990/91 Average Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average of All Industries Standard Deviation

Output (LE 000)

Inputs (LE 000)

Capital (LE 000)

8,303 5,898 484 104 186 181 146 1,392 7,222 1,044 241 243 1,858 3,724 3,417 1,606 2,253 2,705

6,347 3,610 371 74 118 117 93 960 4,158 714 147 136 750 2,451 2,590 732 1,460 1,861

2,779 4,275 235 235 271 104 44 1,033 6,068 368 66 135 5,680 1,702 1,354 4,303 1,791 2,136

Workers (000) 181 274 18 3 10 8 7 38 113 23 8 14 45 73 105 60 61 76

Source: Authors’ calculations using CAPMAS data.

Table A.4. Output and Input Data: 1991/92–1995/96 Average Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average of All Industries Standard Deviation

Outputs (LE 000)

Inputs (LE 000)

Capital (LE 000)

11,592 6,606 1,434 87 166 206 209 2,121 16,007 1,287 217 419 3,407 4,676 6,617 2,145 3,575 4,620

8,928 4,611 481 68 107 113 132 1,480 8,234 862 139 217 2,064 3,049 3,615 1,463 2,223 2,849

3,300 3,889 362 93 288 65 56 1,377 8,767 301 125 338 5,114 4,582 2,404 1,164 2,014 2,511

Source: Authors’ calculations using CAPMAS data.

Workers (000) 201 277 34 3 8 7 9 38 125 28 7 15 61 71 130 54 67 79

31

Do Governments Pick Winners or Losers?

Table A.5. Output and Input Data: 1996/97–2000/01 Average Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average of All Industries Standard Deviation

Outputs (LE 000)

Inputs (LE 000)

Capital (LE 000)

11,671 5,454 776 63 89 13 467 2,040 14,656 1,363 297 509 2,357 3,946 4,540 1,719 3,123 4,295

8,471 3,522 469 47 51 6 314 11,740 1,439 904 158 207 2,216 2,443 1,903 1,551 2,215 3,303

3,914 2,923 506 52 470 116 92 1,982 11,489 515 174 481 3,623 3,847 4,038 861 2,193 2,931

Workers (000) 188 236 64 3 7 1 19 38 118 25 5 14 60 61 84 26 59 69

Source: Authors’ calculations using CAPMAS data.

Table A.6. Output and Input Data: 1980/81–2000/01 Average Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average of All Industries Standard Deviation

Outputs (LE 000)

Inputs (LE 000)

Capital (LE 000)

9,111 5,331 710 82 140 130 225 1,663 10,315 1,081 204 307 2,055 3,631 4,289 1,589 2,554 3,230

6,933 3,501 360 62 86 80 147 3,600 4,100 741 117 154 1,299 2,252 2,517 1,144 1,693 1,990

3,046 4,029 366 146 304 103 59 1,382 8,549 385 101 256 4,293 3,363 2,427 2,015 1,926 2,328

Source: Authors’ calculations using CAPMAS data.

Workers (000) 177 274 30 3 8 6 10 37 113 24 6 14 52 68 98 43 60 75

32

Ahmed Galal and Nihal El-Megharbel

Table A.7. Share of Subsidies to Output: 1980/81–2000/01 Share of Subsidies to Output (%) Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramic Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation Average

1980/81– 1984/85

1985/86– 1990/91

1991/92– 1995/96

1996/97– 2000/01

2.11 3.68 0.14 3.46 0.07 0.06 0.00 0.06 2.74 0.00 0.00 0.03 0.78 0.00 0.02 0.00 0.82

1.71 0.30 0.00 2.70 0.01 0.00 0.00 0.12 0.67 0.00 0.00 0.01 0.17 0.06 0.04 0.00 0.36

0.67 0.06 0.01 0.02 0.00 0.00 0.00 0.04 0.05 0.02 0.00 0.09 0.08 0.24 0.49 0.00 0.11

0.10 0.03 0.21 0.00 0.02 0.00 0.02 0.13 0.04 0.01 0.08 0.00 0.09 0.83 0.82 0.00 0.15

Source: Authors’ calculations using CAPMAS data.

Table A.8. Effective Rates of Protection in Manufacturing Industries: 1986–97 Effective Rates of Protection Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramics Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation

Source: Refaat 1999.

1986

1994

1996

1997

17 788 348 35 160 40 296 36 75 563 214 54 1 120 39 628

7.5 68.2 87.3 79.6 94.1 6.8 128.8 17.6 9.2 50 90.8 39.4 29 26.4 22.5 65

6.3 49.8 61.8 52.7 53.6 6 95.2 18.3 9.1 45.6 60.9 23.8 18.4 19.4 15 57.8

6.4 47.6 55.9 47.6 50.8 6.1 83.8 17.8 9.2 43.1 56 23.2 18.5 18.1 14.5 55.6

Do Governments Pick Winners or Losers?

33

Table A.9. Number of Firms by Industry to Total Number of Firms: 1980/81–2000/01 Share of Subsidies to Output (%) Sector Food Processing Spinning and Weaving Readymade Garments Leather and Leather Products Footwear Wood and Wood Products Furniture Paper and Printing Chemicals Rubber, Plastic, and Related Products Porcelain, China, and Ceramic Glass Products Nonmetallic Products Steel, Iron, and Metal Products Machinery and Equipment Means of Transportation

1980/81– 1984/85

1985/86– 1990/91

1991/92– 1995/96

1996/97– 2000/01

0.46 0.16 0.02 0.01 0.02 0.02 0.01 0.03 0.03 0.02 0.00 0.01 0.12 0.01 0.09 0.01

0.49 0.12 0.04 0.01 0.01 0.01 0.01 0.03 0.03 0.03 0.00 0.01 0.07 0.02 0.09 0.01

0.49 0.11 0.05 0.01 0.01 0.01 0.01 0.03 0.03 0.03 0.00 0.01 0.07 0.02 0.09 0.02

0.58 0.09 0.05 0.01 0.01 0.00 0.02 0.03 0.03 0.03 0.01 0.01 0.06 0.01 0.07 0.01

Source: UNIDO, several issues.

Notes 1. 2. 3.

4.

5.

6. 7.

This finding is consistent with that of Abdel Khalek (2001). This conclusion is also reached by Kheir-El-Din (2001). Imbs and Wacziarg (2003) found that the patterns of sectoral concentration and diversification in a large cross section of countries indicate that as poor countries get richer, sectoral production and employment become less concentrated and more diversified. This process continues until relatively late in the process of development, when economies mature and per capita income increases significantly. Only then do the patterns become more concentrated. Although tax exemptions have been used extensively, it was not possible to estimate a consistent index for this variable for each industry during the period analyzed here. We also tried other industrial policy variables such as the share of FDI in total investment, and the share of public investment in total investment. But both variables were found to be insignificant. Rodrik (2004) lists ten key preconditions for success, including the four listed here. Data are filtered using HP Filter.

References Abdel-Khalek, Gouda. 2001. Stabilization and Adjustment in Egypt: Reform or De-industrialization? Northampton, MA: Edward Elgar Publishing Inc.

34

Ahmed Galal and Nihal El-Megharbel

Amsden, Alice H. 1989. Asia’s Next Giant: South Korea and Late Industrialization. New York and Oxford, UK: Oxford University Press. Caves, D.W., L.R. Christensen, and W.E. Diewert. 1982. “The Economic Theory of Index Numbers and the Measurement of Input, Output, and Productivity.” Econometrica 50 (6): 1393–1414. Central Agency for Public Mobilization and Statistics (CAPMAS). Annual Industrial Statistics Bulletin, several issues. Corden, W. Max. 1966. “The Structure of a Tariff System and the Effective Protective Rate.” Journal of Political Economy 74:221–37. Färe, Rolf, Shawna Grooskopf, Mary Norris, and Zhongyang Zhang. 1994. “Productivity Growth, Technical Progress, and Efficiency Change in Industrialized Countries.” American Economic Review 84 (1): 66–83. Galal, Ahmed, and Amal Refaat. 2005. “Has Trade Liberalization in Egypt Gone Far Enough or Too Far?” Policy Viewpoint 16 (June). Cairo: The Egyptian Center for Economic Studies. Hodrick, R.J., and E.C. Prescott. 1980. Postwar U.S. Business Cycles: An Empirical Investigation. Unpublished manuscript. Pittsburgh, PA: Carnegie Mellon University. Imbs, Jean, and Romain Wacziarg. 2003. “Stages of Diversification.” American Economic Review 93 (1): 63–86. Kheir-El-Din, Hanaa. 2001. “Economic Diversification: The Case of Egypt 1969/70–1999/2000,” in Economic Diversification in the Arab World. Beirut: United Nations Economic and Social Commission for Western Asia. Krüger, Jens J. 2003. “The Global Trends of Total Factor Productivity: Evidence from the Non-Parametric Malmquist Index Approach.” Oxford Economic Papers 55. Oxford, UK: Oxford University Press. Refaat, Amal. 1999. “New Trends in Egypt’s Trade Policy and Future Challenges.” Working Paper 16. Cairo: The Egyptian Center for Economic Studies. Rodrik, Dani. 2004. “Industrial Policy for the Twenty-First Century.” KSG Faculty Research Working Paper RWP04-047. Cambridge, MA: Kennedy School of Government, Harvard University. UNCTAD/WTO. International Trade Center: http://www.intracen.org/countries/ UNIDO. 2003. Industrial statistics database CD-ROM. United Nations (UN). International Trade Statistics, various issues. Wade, Robert. 1990. Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization. Princeton, NJ: Princeton University Press. World Bank. 1993. The East Asian Miracle: Economic Growth and Public Policy. London: Oxford University Press.

CH A P T ER 3

Incentives or Compensation? Government Support for Private Investments in Turkey Hasan Ersel and Alpay Filiztekin1

The history of government support to private investment in Turkey is not a new phenomenon. In fact, legislation concerning such government interventions can be traced back to the Ottoman era, when in 1913 the Provisory Law on Supporting Industry was enacted. The weakness of the private sector, coupled with the 1929 world crisis, led Turkey to adopt the etatist strategy, that is, government-led industrialization, in the 1930s. Although this strategy lost its initial momentum and consistency, particularly during the 1950s, it remained in effect until 1960. The adoption of the idea of development planning in the early 1960s systematized economic policymaking in Turkey. The planning approach was based on the writings of Tinbergen (1964, 1967) and emphasized consistencies at the macro, sectoral, and project levels. Since plans were only indicative for the private sector, the desired outcomes could only be achieved if private decision makers could be encouraged to take actions that supported the objectives of the plan. This motivation led Turkish policymakers to focus on incentives to promote economic activity in line with macro and sectoral targets. The outcome of this new thinking was a rather complex system of incentives to promote economic activity and private investments. Over time, the complexity of the system increased and its coverage expanded. As Arslan (2001), Duran (2002), and Togan (2003) have stressed, these features of the investment incentives made them nontransparent even for those who hoped to benefit from them. Despite the disillusionment of the public bodies—as reflected in Duran (2002), for example—and strong criticisms at the academic level, as in Togan (2003), concerning the negative effects of 35

36

Hasan Ersel and Alpay Filiztekin

Hasan Ersel & Alpay Filiztekin

such incentives on competition, government support to the private sector continued in the form of investment incentives. The purpose of this paper is to examine the reasons behind the continued practice of investment incentives, in spite of their alleged inefficiency. I. A Bird’s Eye View of Developments in the Turkish Manufacturing Industry The Turkish industrial sector was already well-diversified in the mid-1970s (the Herfindahl-Hirschman Index (HHI) is given in Figure A.1 in the appendix). However, the HHI indicates that the degree of diversification was almost stable thereafter; fluctuating only within the narrow band of 0.06 and 0.08. Turkey witnessed a very difficult period in 1970s. The political turmoil that brought the country to the edge of a civil war, coupled with a severe balance of payments crisis triggered by the oil shock, had a devastating effect on the performance of the manufacturing industry. In the early 1980s the situation improved as a result of a drastic shift in economic policies, from inward-looking industrialization to export-oriented growth. Turkey launched a comprehensive program to liberalize its foreign trade and financial system, and political stability coupled with policy credibility helped the economy to recover rather quickly. As can be seen from Figure A.2, export and import penetration ratios increased sharply beginning in 1980. Turkey’s reform strategy was based on the classical sequencing approach, and in 1990 it liberalized the capital account of its balance of payments after ‘completing’ reforms in the areas of finance and trade. This move was considered premature by many observers, who cited the lack of progress in reforming the public sector, which was already giving signals of falling into a debt trap. During the 1990s, the manufacturing industry’s performance was less than impressive. In the first half of the decade, the severe policy mistakes of the ruling government led the economy into a crisis in 1994, and in the second half, Turkey was deeply impacted by the contagion effect of the Russian crisis. As can be seen from Table A.1 (in the appendix), from 1980–2000 the manufacturing industry grew at an average rate of 8.9 percent.2 During these two decades, the rate of capital accumulation was 7 percent. Employment in manufacturing grew at 4.05 percent for the period as a whole. Although the rate of employment growth did not differ much between the first and second halves of the period, both the rate of value-added growth and the rate

Government Support for Private Investments in Turkey

37

of capital accumulation declined after 1991, that is, after Turkey opened up its capital account.2 In terms of productivity, the performance of the economy was far from satisfactory. As shown in Table A.1, labor productivity growth when measured in man hours (per employee) was 6.24 percent (7.20 percent) during 1981–1991, but declined to 3.14 percent (3.43 percent) during 1992– 2000. A similar decline is observed in capital productivity, from 3.52 to -0.09 percent; and in total factor productivity (TFP), from 4.07 to 0.53 percent—when man-hours are used to measure labor input—and from 4.32 to 0.6 percent when the total number of employees is used. During the period under consideration, investment incentives were used as a tool to influence/guide industry, with continuous revisions in terms of coverage, rates and types of incentives offered. One method was to abolish cash incentives (including preferential credits) and rely increasingly on tax exemptions and investment allowances. II. Did Investment Incentives Play Their Expected Role at the Sectoral Level? A typical legislation concerning investment incentives will include almost all the items from the following nonexhaustive list of objectives: 1. Increase investment/GDP ratio; 2. Increase employment; 3. Improve productivity; 4. Allocate investments to favored sectors; 5. Enhance regional development; 6. Encourage technological change; 7. Increase export capacity; and 8. Improve environmental protection. Although expressed in the form of a multi-objective decision-making problem, neither the legislation nor the administrative apparatus is designed to deal with the intricacies of such a problem. Moreover, even if this problem can be solved, it is still difficult for private decision makers to distinguish between these objectives and to calculate the effects of these incentives when they face a rather lengthy and complicated legislation.3 Nonetheless, in this section we investigate the effects of investment incentives on TFP growth, investment volume, and employment. Data on manufacturing industries are from annual manufacturing industry surveys collected by the State Institute of Statistics. In the analyses below, we only use data

38

Hasan Ersel and Alpay Filiztekin

Hasan Ersel & Alpay Filiztekin

for private manufacturing industries that employ ten or more people. While original data are on (ISIC Rev. 2) twenty-nine industries, we had to reduce the number to sixteen in order to achieve data consistency. TFP is in fact what is known as Solow residuals, that is, the residual value-added growth obtained after correcting for factor accumulation. Solow residuals are measured under standard assumptions of constant returns to scale, unit elasticity of substitution between capital and labor, and perfect competition.4 However, later in the regression analysis we allow a variable, namely price-cost margin (PCM), calculated as (value added - total wage bill) / (value added + value of inputs), to control for market imperfections. Capital stock data for each industry is obtained using data on sectoral investment and applying the perpetual inventory method. Employment is measured as total persons engaged. Our analysis focuses on the number of jobs created, rather than the number of hours worked. Moreover, while data on total hours worked are also available, in the early years of our sample, data were not collected for small establishments, and thus the missing values had to be extrapolated. Trade variables, export-output ratio and import-penetration rate are calculated for each industry separately. Value added generated by public establishments is for aggregate manufacturing, because the state was not competing with the private sector, but rather supporting it by supplying cheaper inputs. Effective protection rates (EPR) are obtained from Togan (1994, 1997). Data on incentives are from the State Planning Organization. The investment incentives variable is defined as the total volume of investment certificates to the actual investment volume. This variable stands as a proxy for the intensity of the government’s intent to supply investment incentives to the sector in question. Therefore, it is assumed that in making their investment decisions, firms can gather information by looking at this ratio concerning government intentions. Our goal here is to estimate the effects of incentives in the long run. Therefore, we took five-year averages of variables and formed a panel of sixteen industries, each with four observations. Each specification was then estimated using the fixed effects model. Major findings are as follows: Total Factor Productivity (TFP) Growth Table A.2 provides the regression results of TFP growth. First, we simply regressed TFP growth on the initial value of TFP. While the industries that are lagging behind the others might have an advantage in growing

Government Support for Private Investments in Turkey

39

faster, endogenous growth models predict just the opposite. The coefficient of initial TFP, as shown in the first column of the table, is negative and significantly different from zero, implying the advantages of backwardness. We then included the aggregate value of TFP growth. This variable could approximate two different effects: the existence of linkages across industries and/or stability of the economic environment that favors or disfavors productivity growth. The coefficient on this variable is significantly positive. In the third column, we included price-cost margin and trade-related variables. Market structure seems to be an important factor in productivity growth. Markets with less competition induce higher growth. None of the trade variables seem to have a significant impact on TFP growth. The fourth regression introduced investment incentives. The coefficient is negative but statistically insignificant. There seems to be no effect from investment incentives on TFP growth. Thus, one important reason for providing incentives—to enhance productivity growth—is not justified by the data. We also estimated the regression by including the R&D volume, which is only available for the 1990s, and the results are unchanged. Finally, to control for the possible endogeneity of some variables, we included lagged values. Once again, the results were robust: only the initial levels of TFP, aggregate TFP growth, and market structure had significant effects on TFP growth. Employment Growth A further concern of the investment incentive scheme is to increase employment. To test the effects of incentives on employment, we repeated the previous exercise, this time for employment growth. The results, presented in Table A.3, are very similar to those for TFP growth. The initial level of employment is significantly negative in all specifications. None of the other variables seem to have any significant effect on employment growth in our full specification [specification (4)]. The only exception is when we introduce R&D into the regression. Notice that this specification (5) restricts the sample to the 1990s due to a lack of R&D data for the 1980s. In this case, aggregate employment growth is positive and significant. Two other variables (PCM and IMPPEN) also become marginally significant (at a 90 percent confidence level), implying possible sources of employment growth. Market structure has a positive impact on employment and import penetration reduces the employment level. Using lagged trade variables and investment incentives instead of contemporaneous averages produces strange results. While aggregate employment

40

Hasan Ersel and Alpay Filiztekin

Hasan Ersel & Alpay Filiztekin

growth is still positive, effective protection rate and R&D (at a 10 percent significance level) has a negative impact on employment growth. More interestingly, investment incentives now have a negative and significant coefficient. These findings contradict our expectations and it is very difficult to explain them. Investment Growth Finding that investment incentives have no effect (or even adverse effects) on productivity and employment raises the question of whether the main motive for these incentives was to increase investments. To test this hypothesis, we repeated the exercise with investments as the dependent variable. Two different variables were used to represent the movements in real investments. The first was the change in the real investment ratio and the second was the change in investment-value added ratio. The results are qualitatively similar regardless of the variable used on the left-hand side of the equation; therefore only the equation that uses the latter as a dependent variable will be presented. As reported in Table A.4, the initial investment value-added ratio is negative and significant, as expected. Industries that have already had large investments in the previous period now invest less. Focusing on our favored specification (4), all trade variables are significant and have expected signs. Industries that are protected from foreign trade have a larger investment value-added ratio, as industries that face higher competition from abroad (larger import penetration ratios). On the other hand, industries that have a higher export-output ratio invest less. Turning to our main hypothesis, the coefficient of investment incentives is once again significantly negative. Using lagged values only reduces the significance level of trade variables and the incentive variable becomes insignificant (though it still has the wrong sign). The above findings point out the insignificant, if not adverse, effects of investment incentives in explaining sectoral investment, employment, and TFP growth. III. How do Businessmen View Investment Incentives? Survey Findings How does the business community evaluate industrial incentives? In order to get an answer to this question, a survey was conducted in the second half of October 2005.5 The main concern was to get the impressions of those

Government Support for Private Investments in Turkey

41

who actually have experience getting incentives for their investments. The questionnaire (consisting of nineteen questions) was sent to the local members of the Union of Chambers and Commodity Exchanges of Turkey6 and was completed only by those businesspeople that benefited from investment incentives within the last ten years. As was expected, most of the responses came from relatively small companies. Of the 2,510 businesspeople that responded, 252 had received investment incentives over the past decade.7 According to survey results, 61 percent of respondents claimed that domestic market concerns played a more important role than incentives in their investment decisions. Investment projects that were granted the largest portion of incentives were for enlargement (37.8 percent), followed by complete renewal (25.6 percent), and modernization investments (25.4 percent). A much smaller portion of the incentives was granted for R&D (7.6 percent) and even less for environmental protection (3.6 percent). According to the survey, 73.3 percent of respondents benefited from value-added tax support, 71.2 percent from investment allowances, 48.3 percent from exemptions of taxes, fees and duties, and 47 percent received customs duty exemptions. The share of those who received subsidized credit was smaller, at 21.6 percent, and the percentage of those who received other forms of incentives was even lower. Of those surveyed, 44.8 percent believed that incentives that reduce investment cost (such as customs duty exceptions) are most important, 34.1 percent placed more emphasis on incentives that bear results after the completion of the project (such as investment allowances), and 17.5 percent did not express an opinion. Respondents were asked to rank five different government motives for offering investment incentives. As shown in Table A.5, the figures8 indicate that businesspeople consider technological improvement” the most important reason for a government to offer investment credits, followed by promoting investment irrespective of its sectoral and regional impacts. Channeling investments to underdeveloped regions is ranked as the third motive. The last two motives, respectively, are sectoral allocation of investments, and “compensation for the negative effects on investments that stem from the difficulties in the supply of public services and/or their high cost.” One interesting aspect of this question is the inclusion of the last motive, which was not discussed at all in the public domain. Nevertheless, it was not discarded as irrelevant. In fact, it is considered a reasonable secondary cause for investment incentives. A major objective of the survey was to get a sense of the role incentives

42

Hasan Ersel and Alpay Filiztekin

Hasan Ersel & Alpay Filiztekin

play in business decision making. For this purpose, those surveyed were invited to respond to a question referring to a counterfactual: would you change your decision to invest (choice of technique, choice of location) if such an incentive was not offered? The distribution of the responses is given in Table A.6. Despite the difficulties inherent in interpreting the answers to a question that involves counterfactuals, the large difference between yes and no responses indicates that incentives seem to play a much more minor role in shaping investment decisions than the designers assumed.9 In line with the previous questions, the survey participants were also asked whether they would reconsider their investment locations in light of the very recent law that expanded the definition of “regions with special priority for development.” In contrast to the government’s high hopes, but consistent with their revealed behavior in the previous set of questions, 64 percent stated that they would not revise their choice of location. The findings of this survey are not sufficient to claim that investment incentives have no effect on business decisions. However, when coupled with the supporting statistical findings of the previous section, they may raise some questions as to the efficiency and desirability of incentives. IV. Conclusion: Why Do Governments Continue to Offer Investment Incentives? Both the econometric findings of the paper and the results obtained from the October 2005 survey indicate that investment incentives, as they are, can hardly be considered efficient tools to influence the level of investment and its sectoral allocation or to promote efficiency growth.10 Such a negative conclusion is not surprising given the general perception of investment incentives in Turkey. In fact, the incentive problem has always been a hot topic in the media and in bureaucratic circles, but amazingly less so at the academic level. Nevertheless, with the exception of Togan’s (2003) wellstructured critique of investment incentives, the focus seems to be on the implementation side of the issue and not on the incentive concept itself. What is puzzling is the behavior of political decision makers and the business community. Despite the headache incentives have created and their apparent inefficiency, no government has ever attempted to introduce a radical change in the system, and the business community has never expressed such a demand. In fact, the only visible trend in investment incentives is the shift toward using tax reductions and investment allowances from cash supports.

Government Support for Private Investments in Turkey

43

However, this can hardly be considered a deliberate choice on efficiency grounds. Instead, it is one of the reflections of the fiscal crisis of the state in the 1990s. It should also be noted that it is quite difficult to quantify such incentives. They are not accounted for as public expenditures but as taxes foregone, which are difficult to estimate even for their beneficiaries. In light of these issues, it may be more rewarding to look at the political economy side of the problem. For this purpose, the following conjecture is proposed: suppose the performance of the economy fails to keep it on a warranted growth path (say, in the sense of Harrod), due to constraints the existing economic environment is imposing on business decisions. The government would face the dilemma of either bearing the political and financial costs of launching a major reform program or accepting political responsibility for an economic failure. While the latter is never a choice for the incumbent, the former route may also be too risky. In such an environment, governments may opt to offer ‘incentives’ in order to compensate for at least part of the external costs the firms are facing. In other words, incentives in this framework are not for guiding businesses but for convincing them to implement their own plans; they are offered as side payments. If that is so, then the list of objectives attached to incentives and conditions for eligibility lose their importance, and as witnessed in Turkey, may be subject to frequent changes. If this is the case, then the inefficiency of incentive variables should not be a surprise. V. Appendix I. Definition of Subsectors A311+312+313

Food and Beverage

A314

Tobacco

A321+322

Textiles and Apparel

A323+324

Leather

A331+332

Wood and Furniture

A341+342

Paper and Printing

A351+352

Chemicals

A354

Misc. Petroleum Products

A355

Rubber

A361+362+369

Pottery, Glass, and Minerals

A371+372

Iron and Steel and Nonferrous Metals

A381

Fabricated Metal

A382

Machinery

A383

Electrical Machinery

A384

Motor Vehicles

A390+356+385

Plastics nec., Instruments and Others

Hasan Ersel and Alpay Filiztekin

44

Hasan Ersel & Alpay Filiztekin

Source: Authors’ calculations. Source: Author’s calculations.

Figure A.2.Export-Output Export-Output Ratio Import Penetration Figure A.2. Ratio andand Import Penetration Source: Authors’ calculations.

Source: Author’s calculations.

Table A.1.Industry Manufacturing Industry Table A.1. Manufacturing Developments, 1980–2000 (Rate of Change %) 1981–2000 Value Added Capital Labor (MH) Labor (PE) Labor Productivity (MH) Labor Productivity (PE) Capital Productivity Total Factor Productivity (MH) Total Factor Productivity (PE)

8.90 7.00 4.05 3.39 4.84 5.51 1.89 2.48 2.65

Source: Authors’ calculations. Notes: PE: Persons Employed; MH: Man-Hour.

1981–1991 10.24 6.72 3.99 3.03 6.24 7.20 3.52 4.07 4.32

1992–2000 7.26 7.35 4.12 3.83 3.14 3.43 -0.09 0.53 0.60

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

1970

1968

1966

1964

1962

1960

1958

1956

1954

1952

1950

Figure A.1. Figure A.1.Herfindahl-Hirschmann Herfindahl-HirschmanIndex Index

0.40795

-0.1325 (0.0233)***

0.47648

-0.1184 (0.0229)*** 0.6919 (0.2820)**

(2)

0.56144

-0.0000 (0.0000) 0.0020 (0.0023) 0.0008 (0.0018)

-0.1539 (0.0286)*** 0.7889 (0.2903)*** 0.0037 (0.0018)

(3)

0.57118

-0.0000 (0.0000) 0.0013 (0.0024) 0.0012 (0.0018) -0.0022 (0.0023)

-0.1573 (0.0288)*** 0.8501 (0.2974)*** 0.0037 (0.0018)**

(4)

0.85607

-0.1515 (0.0704) 1.2792 (0.9961) 0.0140 (0.0052)** -0.8595 (1.7647) 0.0001 (0.0001) 0.0017 (0.0082) 0.0070 (0.0052) -0.0043 (0.0039)

(5)

Dependent Variable: TFP Growth

Standard errors in parentheses. Fixed effect terms are not reported. * Significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

R-squared

Lag. Incentives

Lagged IMPPEN

Lagged EXPOUT

Incentives

IMPPEN

EXPOUT

EPR

R&D

PCM

Agg. TFP Growth

Initial TFP

(1)

Table A.2. TFP Growth

0.56164

0.0005 (0.0017) 0.0021 (0.0019)

-0.0000 (0.0000)

-0.1520 (0.0272)*** 0.6420 (0.2936)** 0.0027 (0.0019)

(6)

-0.0021 (0.0032) 0.0054 (0.0030)* -0.0032 (0.0030) 0.55260

-0.0000 (0.0000)

-0.1097 (0.0521)** 1.0843 (0.5294)* 0.0003 (0.0025)

(7)

-0.0027 (0.0064) 0.0006 (0.0074) 0.0046 (0.0062) 0.77933

-0.0807 (0.0747) -0.4112 (0.8193) 0.0175 (0.0063)** 2.7359 (3.0152) 0.0000 (0.0001)

(8)

Government Support for Private Investments in Turkey 45

0.42578

0.42755

0.48376

Standard errors in parentheses. Fixed effect terms are not reported. * Significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

R-squared

Lag. Incentives

Lagged IMPPEN

Lagged EXPOUT

(5)

0.91806

0.0155 (0.0119)

(0.0176)*

-0.0339

(0.0251)

0.0464

(0.0002)

0.47546

-0.0004

0.50575

0.93817

-0.0397 (0.0138)**

-0.0086

(0.0159)

0.0163

(0.0178)

0.0009

(0.0002)**

(0.0096)

0.0104 (0.0079)

0.0052 (0.0056)

-0.0044 (0.0104)

0.0066

(0.0000)

(0.0056)

(0.0000)

-0.0001

(0.0121)

0.0147

(1.0706)**

3.4745

(0.2484)***

(6.7893)

-0.0002

-0.0000

(0.0114)

0.0132

(1.1332)

1.7735

(0.2354)***

(8) -1.8247

-14.3899

(0.0075)

0.0059

(0.7326)

0.6235

(0.1618)***

(7) -1.0997

-2.1093

-0.0051

0.48962

(6) -0.7080

(5.3569)

(0.0137)*

0.0273

(1.6330)**

4.5612

(0.2531)***

-1.6419

(0.0074)

0.0047 (0.0058)

0.0040 (0.0056)

0.0047 (0.0076)

0.0066

(0.0000)

(0.0000) (0.0071)

-0.0000

-0.0000

0.0055 (0.0075)

0.0058

(0.6745)

-0.1055

(0.1799)***

-0.8105

(4)

Dependent Variable: Employment Growth

(0.0074)

(0.6558)

(0.5253)

(0.1760)*** -0.0101

(0.1540)***

(0.1036)***

-0.7890

(3)

-0.1984

-0.6542

-0.6115

(2)

Hasan Ersel and Alpay Filiztekin

Incentives

IMPPEN

EXPOUT

EPR

R&D

PCM

Agg. Empl. Gr.

Ln Init. Empl.

(1)

Table A.3. Employment Growth

46 Hasan Ersel & Alpay Filiztekin

0.46002

0.53737

(4)

0.60850

(5)

0.0005

0.85381

-0.0590 (0.0266)*

(0.0378)

-0.0229

(0.0442)

0.0105

(0.0004)*

0.0001

0.64597

0.0003

0.75567

0.88055

-0.0147 (0.0325)

0.0180

(0.0348)**

0.0937

(0.0288)

-0.0495

(0.0003)

(0.0124)

0.0261 (0.0104)**

0.0252 (0.0096)**

-0.0099 (0.0130)

-0.0151

(0.0000)

(0.0093)

(0.0000)*

0.00003

(0.0294)

-0.0417

(0.3293)

-0.0109

(0.3032)***

-14.4778

(0.0127)**

0.0305

(0.2231)***

0.6587

(0.1238)***

(8) -1.4962

(16.0467)

(0.0110)

0.0118

(0.2109)**

0.4690

(0.0893)***

(7) -0.8122

-3.4595

-0.0306

0.65747

(6) -0.6348

(10.5230)

(0.0308)

-0.0212

(0.6131)

0.3963

(0.4256)**

-1.3870

(0.0126)**

0.0211 (0.0098)**

0.0163 (0.0102)

-0.0249 (0.0131)*

-0.0141 (0.0131)

0.000016 (0.0000)*

0.000031 (0.0000)*

0.0125 (0.0101)

0.0117

(0.2245)

0.3248

(0.0879)***

-0.5894

(0.0107)

(0.2345)*

(0.1884)***

(0.0927)*** 0.4059

(0.0936)***

(0.0999)***

-0.6016

(3)

Dependent Variable: Change in Investment Value-Added Ratio

0.5226

-0.6195

-0.6320

(2)

Standard errors in parentheses. Fixed effect terms are not reported. * Significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

R–squared

Lag. Incentives

Lagged IMPPEN

Lagged EXPOUT

Incentives

IMPPEN

EXPOUT

EPR

R&D

PCM

Ch. in Agg. Vol

Init. Volume

(1)

Table A.4. Investment Growth

Government Support for Private Investments in Turkey 47

Hasan Ersel and Alpay Filiztekin

48

Hasan Ersel & Alpay Filiztekin

Table A.5. Ordering Different Motives of the Government for Offering Incentives to Investments (% distribution)nn Ranking

Investment Growth

Sectoral Development

Regional Development

Compensation for Inefficiencies in Public Services

Technological Improvement

1

7.2

5.1

6.2

4.8

8.4

2

3.0

4.2

4.5

4.4

4.5

3

3.0

4.3

4.5

2.8

3.5

4

2.4

2.6

2.8

3.9

2.3

5

4.5

3.9

2.6

3.3

1.2

Table A.6. Investments Even if the Incentives Were Not Offered? (% distribution)l “Would you stick to your investment decision even if the incentives were not offered?” Yes

No

No Answer

Decision to Invest

63.5

32.5

4.0

Choice of Technology

77.0

18.3

4.7

Choice of Location

77.0

18.3

4.7

Notes 1.

2.

The authors gratefully acknowledge the contribution of Subidey Togan (Bilkent University), who kindly allowed them to use his effective protection estimates for Turkey and his advice on various points. The authors also extend their gratitude to the Union of Chambers and Commodity Exchanges of Turkey and TEPAV/ EPRI for their generous help in conducting the survey and in obtaining its results. The authors also wish to offer their thanks to Heba Handoussa (Egypt’s Human Development Report), Kamil Yilmaz (Koç University), and to the participants of the workshop organized by the TEPAV/EPRI in Ankara on November 9, 2005 and the ECES Conference entitled Rethinking the Role of the State: An Assessment of Industrial Policy in MENA, held on November 12, 2005 in Cairo. Last but not least, the authors would like to thank Ahmed Galal for initiating the idea of this topic as well as his continuous encouragement and help. From this point on, the paper focuses on the post-1980 period that is, after liberalization reforms. Besides the logical difficulties of comparing two qualitatively different periods, in which economic policies as well as the behavior of economic agents differed considerably, the data for the 1970s seem much less reliable than those that are available for subsequent years. Although our findings are qualitatively in line with our expectations, the numerical values calculated for the 1970s were not comforting. Our guess is that due to the severe foreign exchange constraint that the country was facing, especially in the second half of the 1970s, the use of the official exchange rate

Government Support for Private Investments in Turkey

3.

4.

5.

6. 7. 8.

9.

49

in calculating the investment deflator and the amount of investment may have been distortionary. Turkey opened up its capital account in 1990. However, the effect on the economy was not instantaneous, but with a lag. For the purpose of this paper, the full effect of the liberalization of the capital account is assumed to have been realized in 1992. Arslan (2001) and Togan (2003) estimated the rate of subsidy by taking into account the existing legislative structure. Their work clearly demonstrates that it is practically impossible for an entrepreneur to fully integrate incentives into her/his investment-decision framework. In order to see the relevance of the first two assumptions, the following procedure was followed. First, a variable elasticity of substitution (VES) production function was estimated by using pooled data for the sixteen sectors. The coefficients that indicated variability in the elasticity of substitution were found to be statistically insignificant. In the second step, a CES function was estimated using the same data set. The findings indicated that the unit elasticity of substitution hypothesis could not be rejected. On the other hand, although the statistical findings indicated decreasing returns to scale, the numerical value of the scale parameter was very close to one. Therefore, TFP measures were calculated under the assumption of a CobbDouglas production function; labor shares, total payments to workers over value added, are Divisia Indices yielding average elasticities of 0.65 and 0.35, for capital and labor, respectively. See Deliveli and Ersel (2005) for the details of the survey. The Union of Chambers and Commodity Exchanges of Turkey (UCCET) has 364 local members, scattered all over Turkey, representing 1.2 million companies. The forms were distributed at the monthly meeting of the boards of the local chambers. Since the members of the boards of local chambers are democratically elected, the total sample can be considered as random. The low frequency of the responses can be attributed to the dominance of service sector related firms in the total. The figures can be formulated in various ways. A very simple way of doing it is by calculating the following ratio: EMBED Equation.3 The figures can be formulated in various ways. A very simple way of doing it is by calculating the following ratio: 3

η j = ∑ a ij i =1

5

∑a i =3

ij

for all j.

It can be seen that the value of the ratio is 2.34 for technological improvement, 1.55 for regional development, 1.33 for investment growth, 1.26 for sectoral development, and 1.19 for compensation. It can be seen that the value of the ratio is 2.34 for technological improvement, 1.55 for regional development, 1.33 for investment growth, 1.26 for sectoral development, and 1.19 for compensation. 10. The equality of the second and third rows is a pure coincidence. 11. Two of our commentators, independently, drew attention to the fact that almost one-third of the businesspeople indicated that their decisions, one way or another, were influenced by the availability of investment incentives. It is obvious that such a score can hardly be comforting for a policymaker, as two-thirds of the incentives were wasted. However, this criticism calls for further research at the micro level to understand the reasons behind the different responses.

50

Hasan Ersel and Alpay Filiztekin

Hasan Ersel & Alpay Filiztekin

References Arslan, I. 2001. “Investment Incentives in Turkey,” in S. Togan and N. Balasubramanyam, eds., Turkey and Eastern European Countries in Transition. Houndmills: Palgrave. Deliveli, E. and H. Ersel. 2005. Investment Incentives Survey, TEPAV/EPRI (Forthcoming). Duran, M. 2002. “Türkiye’de Yatırımlara Sa lanan Te vikler ve Etkinli i” (Investment Incentives in Turkey and their Efficiency). Republic of Turkey, Undersecreteriat of Treasury. http://www.hazine.gov.tr/arastirma_yayin.htm. Tinbergen, J. 1964. Central Planning. New Haven: Yale University Press. _____. 1967. “Methodological Background of the Plan,” in S. ilkin, and Ö. inanç, eds., Planning in Turkey. Ankara: METU–IIF Publications, 9:71–77. Togan, Subidey. 1994. Foreign Trade Regime and Trade Liberalization in Turkey During the 1980s. Aldershot: Avebury Ashgate Publishing Ltd. _____. 1997. “Opening up the Turkish Economy in the Context of the Customs Union with EU.” Journal of Economic Integration 12:157–79. _____. 2003. “Investment Incentives and Conditions of Competition in Turkey,” in S. Togan and H. Kheir-El-Din, eds. Competitiveness in the Middle Eastern and North African Countries, Cairo: Economic Research Forum for the Arab Countries, Iran, and Turkey Publication.

CH A P T ER 4

An Empirical Analysis of Industrial Policy in Morocco Najib Harabi1

Until recently, two extreme views often seem to have dominated the discussions of the role of the government in economic development. The first view has been that effective government is not only necessary due to market failure but possibly even sufficient to achieve economic development. At least implicit in this view is the argument that if a particular political regime could not be counted on to perform completely and honestly in this process, either the regime would be forced to do so as a result of building political pressures or else it would lose power, through elections if available or through other means if not. The second view, associated with the neoclassical counterrevolution or new orthodoxy school, which has its roots in the thought of Friedrich von Hayek, was developed in the ideas of James Buchanan and was applied to development policy by Anne Krueger, Deepak Lal, and others. In this view, participants in government, such as politicians and bureaucrats, were as selfish and self-interested as owners of firms and other assets but lacked the competitive climate of markets to restrain them. Even when the economy was locked in a poverty trap, government itself played a role in that bad equilibrium. While these points might enjoy a broad agreement under some circumstances, this approach led to the strong conclusion that, as a rule, at least beyond a minimum rule, governments could only make things worse.2 These two extreme views become clearer when one looks at the particular field of industrial policy in developing countries. From the second point of view, industrial policy elicits very strong reactions, while the first view sees it somehow as the magic bullet for resolving urgent problems of economic growth. Those who believe strongly in the efficient working 51

52

Najib Harabi

of markets (second view) see any argument in favor of industrial policy as fiction or, worse, an invitation for all types of rent-seeking activities. On the other side, people who believe market failures are pervasive in developing countries think that any path to economic development requires a liberal dose of industrial policy (first view). In order to shed more light, or even settle the debate in favor of one group, several authors have either reviewed the analytical literature (see for instance, Pack and Saggi 2003, or Rodrik 2004) or the empirical evidence (Noland and Pack 2003), or both. The results of these literature surveys, as noted in Chapter 1, are not conclusive: “While there certainly exist cases where government intervention coexists with success, in many instances industrial policy has failed to yield any gains. Above all, the real issue is that the relevant counterfactuals are not available. Consider the argument that Japan’s industrial policy was crucial for its success. Since we do not know how Japan would have fared under laisser-faire, it is difficult to attribute its success to its industrial policy. Maybe it would have done still better in the absence of industrial policy or maybe it would have done much worse” (Pack and Saggi 2003). Given this basic difficulty, we can only hope to obtain indirect clues regarding the efficacy of industrial policy. In this chapter, I add more empirical evidence to the debate on industrial policy in developing countries. By looking at the case study of Morocco, I hope to offer additional indirect clues regarding the efficacy of this policy. The study is organized as follows. In section one, I describe important aspects of industrial policy in Morocco since its independence. Then, in section two, I try to evaluate this policy. At the center of this section is the question whether industrial policy has contributed to the economic growth of the private sector in Morocco. In section three, I summarize the study and provide some concluding remarks. I. Industrial Policy in Morocco Since its independence in 1956, the Moroccan state has practiced selective interventions in favor of specific private entrepreneurs, firms, whole industries, and regions. For this purpose it has used all kinds of policy instruments, ranging from strong interventionist policies in the 1970s to more liberal ones in the 1980s and 1990s. Based on criteria related to the intensity of state intervention in the economy, I will divide the history of Moroccan industrial policy into two periods: 1960–1982 and 1983–2005.

An Empirical Analysis of Industrial Policy in Morocco

53

Activist Industrial Policy (1960–82) After its independence and the subsequent transition period between 1956 and 1960/62, the Moroccan state has developed different activist policies aimed at restructuring the economy through the process of picking winners and losers among firms, industries, and regions. These policies include: incentive programs for investments in the industrial sector, subsidized loans for investors in other selected sectors, public procurement policy in favor of certain firms in specific industries, and transfer of foreign ownership to Moroccans (known as the “Politique de Marocanisation”). The main purpose of the first set of policies was to promote an importsubstitution industrial strategy through means such as according investment privileges and customs protection for industrial goods and services. The investment privileges were codified in Investment Codes,3 whose main thrust was to change the relative prices in the economy in favor of nationally manufactured industrial goods. To the same end, the newly introduced Customs Code imposed heavy tariffs on certain imported industrial goods, and quantitative limitations or even import prohibition on others. In addition, the Moroccan state has pursued its policies of picking winners and losers through its credit policy. Since most financial institutions at that time were in state ownership and run by bureaucrats, subsidized credit and loans were attributed to investors in the national manufacturing sector and other selected sectors of the economy, such as tourism, hotels, housing, and agriculture. The benefits to private investors in those industries consisted of allowing relatively long credit repayment periods, financing between 60 and 70 percent of investment through governmental credit, and agreeing to a system of fixed interest rates in a period of high inflation rates. In other words, the real cost of capital had been held superficially lower than the market level. A further means of protecting specific firms in specific industries is the channel by which firms get access to public procurements. During the period under consideration, the Moroccan state favored national enterprises in selected industries such as construction and public works, and metallic and semimetallic furniture industries. Only those firms had access to contracts offered by the state; they were able to grow and prosper under its protection. A better known government policy of picking winners and losers is the policy of ‘Moroccanization’, codified in the law (‘Dahir’) of March 2, 1973. According to this law, two lists of economic activities were established

54

Najib Harabi

by the state. On the first list the following activities were included: trade activities (ranging from import to retail activities), all activities in construction and public works, the automobile industry, leasing and advertising activities, credit institutions, warehouses, facility management (especially the management of real estate), the food industry, and the fertilizer industry. The formal ownership of all businesses involved in these activities had to be transferred to Moroccan hands (“Moroccanized”) before the end of 1974. The second list encompassed activities that had to be “Moroccanized” before May 1975. It included businesses in the banking and insurance sectors, the flour-milling industry, foodstuffs (pastes), cork, assembling of vehicles, electric and electronic materials, etc. This vast operation of ownership transfer from foreigners to nationals is unique in the recent economic history of Morocco. It is a policy of capital substitution—substitution of foreign capital by national capital—that has created, overnight, a new segment of Moroccan firms in very important industries. These industries include not only the ones mentioned above, but also—via contagion and interlinkage effects—other sectors like agriculture4 and manufacturing industries (for more details, see Saadi 1989). More Liberal Industrial Policy (1983–2005) In 1983, Morocco started a program of structural adjustment policies, designed under the auspices of the IMF and World Bank. The purpose of those ambitious macroeconomic reform programs was to promote an open, market- and export-oriented system, in order to stabilize the economy—by achieving a more stable currency, lower inflation, lower budget deficits, lower balance of payment deficits etc.—and to obtain higher growth rates. The means for actualizing these goals were macroeconomic stabilization programs, liberalization of trade and selected domestic markets, and privatization of public companies. With respect to macroeconomic stabilization, Morocco has indeed achieved very low inflation rates (below 2 percent), a relatively low budget deficit (ranging from 11.6 percent in the 1980s to 3.8 percent in 2003), and a significant surplus in the balance of payments (due mainly to tourism and money transfers from Moroccans residing abroad). Trade reforms have also been launched. The application of rates resulting from tariffing in 1996 put an end to the imposition of quantitative import restrictions on the majority of products. The continued computerization of customs procedures, the development of customs clearance warehouses and areas, and the creation

An Empirical Analysis of Industrial Policy in Morocco

55

of domiciliation offices are on-site customs clearance procedures that have substantially reduced the time taken for customs clearance and enhanced transparency in this area. Customs duties have been lowered on certain nonagricultural products. In 2000, the fiscal import levy (PFI) was incorporated into the customs tariffs with the aim of simplifying imposition at the border. Morocco has bound all its tariffs lines solely at ad valorem rates ranging from zero to 380 percent; duties on nonagricultural products have been bound at 40 percent. In 2004, the simple arithmetic average of the bound rates should be 42 percent. Other duties and taxes have been bound at 7.5 or 15 percent. Morocco has also bound market access for certain agricultural products by introducing tariff quotas which, in practice, are not applied, all imports of the products concerned being subject to the out-of-quota rates. Subsidies have been abolished for the majority of products, with the exception of locally produced sunflowers and sugar not intended for industrial use. A number of fiscal, customs, and financial benefits are given to investors, especially exporting firms, through the 1995 Investment Charter, customs regimes, the free export zone regime, and the Hassan II Fund for Economic and Social Development created in 2002. Subsidies are also granted for the promotion of exports of certain agricultural products by air freight. However, levies are applied to exports of maize, plant fiber, and crude phosphates. In the automobile assembly industry, 60 to 70 percent of locally made components are required in exchange for certain advantages. Compulsory reserves of petroleum products and pharmaceuticals must be kept. Privatization of state-owned companies has also made progress. Started in 1993, the privatization program has until 2003 covered sixty-six (out of 113 planned) entities and thus generated revenues for the state of fifty-five billion Moroccan dirhams. Price controls and marketing monopolies have also been abolished for almost all goods and services, with the exception of certain transport industries, such as rail transport, port, and airport services, and crude phosphates. Legislation on government procurement and competition policy entered into force in 1999 and 2001 respectively. The Government Procurement Code enhances transparency and incorporates provisions to combat corruption; a price preference of up to 15 percent is given to Moroccan firms for work contracts and related design. In addition to these more general policies, the Moroccan state has pursued industrial policies that are targeted at specific sectors, ranging from agriculture and manufacturing to services. To illustrate this point, a few examples will be mentioned here. The first example is the tourism industry.

56

Najib Harabi

Being the second most important source of foreign currency after transfers from Moroccans residing abroad, this sector has been a major preoccupation of the Moroccan state in recent decades. A private-public partnership program called Plan AZUR has been set up, with the aim of attracting ten million tourists by the year 2010. In order to achieve this goal, an additional eighty thousand hotel rooms will have to be built, seventy-two thousand professionals trained, air transport upgraded, and new marketing and environmental organizations established. For financing these initiatives, both private and public monies have been mobilized. Another industry benefiting from government policies is the textile and clothing industry, which is the country’s largest export industry and its biggest employer. This industry that for many years had taken advantage of preferential treatment from Europe is now facing tremendous competitive pressures from China and other international suppliers. In order to alleviate the resulting problems, the state has designed, in collaboration with representatives of the profession, a set of industry-specific policies, ranging from import tariff reductions to export and other subsidies. In addition to sectoral policies, Morocco has introduced other instruments of industrial policy. Some, like the Regional Investment Centers (Centres Régionaux d’Investissement), are targeted at establishing new firms, others at promoting small and medium-sized enterprises (SMEs) through l’Agence Nationale de la Petite et Moyenne Entreprise and others at upgrading (mise à niveau) the managerial, technological, and organizational infrastructure of private firms in specific industries. These programs have been financed partly by the European Union and other international donors. To complete the whole spectrum of government policies of picking winners and losers, one has to mention the phenomenon of promoting ‘national champions.’ As in French industrial policy, Morocco tries to pick up flagship firms and promote them through all kinds of protective measures until they achieve an ‘optimal size’ for competing internationally. In this regard two examples can be mentioned. First, the semipublic company Société Nationale d’Investissement (SNI) has been privatized and taken over by the group ONA, in order to create a ‘national champion.’ The second operation consisted of merging two big banks—Banque Commerciale du Maroc (BCM) and Wafabank—in order to create the largest financial group in the country, another ‘national champion’ in the financial sector. After this short review of some of the main instruments of Moroccan industrial policies, I now ask how these policies can be evaluated.

An Empirical Analysis of Industrial Policy in Morocco

57

II. Evaluation of Industrial Policy in Morocco To assess Moroccan industrial policies, I will use a positive, rather than a normative, approach and ask whether industrial policies have affected, positively or negatively, the economic performance of private firms. Economic performance can be measured by different indicators. The most popular ones are: measures of total factor productivity in general and of labor productivity in particular; measures of profitability; and measures of economic growth. For data availability reasons, I will be using measures of firm growth as indicators for the economic performance of Moroccan firms. Aggregate economic growth is commonly broken down into two components: growth due to factor accumulation and growth due to an increase in total factor productivity. At the microeconomic level each of these components requires a further distinction. Aggregate factor accumulation can occur through the entry of new agents (such as firms, farms, banks, and households) or through the expansion of existing ones. Aggregate total factor productivity can rise because the most productive agents expand their activities at the expense of the less productive ones, or because some agents innovate and their innovations are adopted by others. From the perspective of firms there are thus four sources of growth: 1. organic growth (through investment) of existing firms; 2. successful formation of new firms operating in existing activities; 3. growth through concentration of firms’ activities (for instance through mergers and acquisitions); and 4. growth through innovation and diffusion of new products and processes. This study examines the growth experience in Morocco from the perspective of private firms. Concentrating primarily on the first microeconomic source of growth, it attempts to identify those factors positively or negatively influencing the growth process of private firms. Among these determinants of growth, selective government policies are especially highlighted. This should contribute to an empirical understanding of how governmental policies affect the growth performance of private firms in Morocco. This part of the chapter is organized in four subsections. The first reviews the theoretical and the empirical literature that examines the major factors influencing the growth process of private firms, including policy variables. The second develops an empirical framework for both systematically organizing our thoughts about the major factors influencing the growth process

58

Najib Harabi

and estimating the quantitative contribution of each. The third subsection summarizes the econometric results, based on data from Morocco. Theoretical Background The enormous literature on the theory of the growth of firms is summarized in standard textbooks such as Scherer and Ross (1990), as well as in extensive surveys such as You (1995), Trau (1996), Sutton (1997), Geroski (1999), and Hart (2000). There is also a large number of empirical studies concerning how firms grow.5 For several reasons, mainly related to data availability, I will concentrate on models of optimal firm size as the theoretical framework.6 Models of optimal firm size postulate that profit-maximizing firms can achieve an optimal size if they behave rationally. That size depends on the market structure in which the firm operates, that is, whether the setting is one of perfect competition or one of imperfect competition (monopoly, oligopoly, or monopolistic competition). In perfectly competitive markets, firms with a U-shaped average cost curve will grow until they reach the size corresponding to the lowest point on the curve; there is no incentive for them to grow beyond this size. Thus the sizes of perfectly competitive firms will be very narrowly dispersed, with any variation attributable to disequilibrium or managerial error, and this dispersion will diminish over time as firms converge toward the equilibrium size. One major conclusion of this theory is that small firms grow faster than larger ones until they reach what is called the minimum efficient scale (MES) of production. If firms have market power (that is, there is imperfect competition), their optimal size may differ from this optimal cost position. In this situation the limit on a firm’s growth is determined by the demand for its unique product rather than by cost considerations. The typical firm faces a downwardsloping demand curve for its products. In practice, this constraint does not limit the growth of a firm because it can always introduce another product line. Product diversification is therefore another determinant of firm growth. Relaxing the assumptions of this neoclassical theory of the firm permits many other explanations of growth. The two that this study considers are economies of scale and goals other than profit maximization. Economists distinguish among four different economies of scale: technical, pecuniary, external, and dynamic. All of these affect the growth process of firms and its determinants.

An Empirical Analysis of Industrial Policy in Morocco

59

The theory discussed so far assumes that all firms aim to maximize profits. Other assumptions about the goals of firms have different implications for growth. For example, Sargant (1943) suggested that many owner-managed companies adopt ‘satisficing’ rather than ‘maximizing’ policies;7 instead of maximizing profits or sales, these firms opt for a quiet life and hence tend to employ fewer people than they could. Satisficing theories were subsequently developed by Simon (1959) and Cyert and March (1963). Baumol (1959) postulated that firms maximize sales subject to the constraint that profits satisfy their shareholders and the company’s plowback policy. A firm’s goals might also change over its life cycle, in response to conflict between its principals and their agents (Mueller 1972). Young, dynamic firms have rapid growth and high profitability, and managers and shareholders are happy. But as a company matures and its investment opportunities decline, a conflict arises: managers may attempt to maximize growth at the expense of profitability. In summary, there exist several theoretical hypotheses about the determinants of optimal firm size and firm growth. Some of these hypotheses have been tested empirically, as shown in the next section. Empirical Framework Several economists have tried to translate the numerous theories of optimal firm size presented above into a simple, empirically testable model (Geroski 1999; Geroski and Gugler 2001). The model can be stated as follows:

ΔSi(t) = Si* + βSi(t - 1) +μi(t),

(1)

where Si(t) is the actual size of firm i at time t, Si* is the long-run steady-state size of firm i, β is the speed with which firm i converges toward Si* when Si ≠ Si*, and µi(t) is a normally distributed iid white noise error process. Before equation (1) can be used for empirical work, one has to specify S*. The most common approach is to write

Si*(t) = c + αX(t) + ηi(t),

(2)

where ηi(t) is a white noise error process and X(t) is a set of observable exogenous drivers of S*(t). Substituting equation (2) into equation (1),

60

Najib Harabi

ΔSi(t) = c + αX(t) + βSi(t - 1) + νi(t), where νi(t)

(3)

≡ μi(t) + ηi(t).

If α = 0, equation (2) says that S* is constant over time and the same for all firms (up to a stochastic term). If α = 0, S* also depends on a set of exogenous variables X(t). Based on our theoretical discussion and on other sources in the literature (cited below), these observable exogenous variables might include, in addition to size, the age of the firm, its legal form, its location, whether it engages in innovative activity, the diversification of its product line, its internal organization, the size of its market, the structure of its market, factors specific to its industry, state regulations and policies (our major emphasis), and others.8 The major problem with using equation (2) or equation (3) is omitted variables. Most studies, including this one, cannot accurately correct for all of the possible determinants of Si*, and, as a consequence, it is often difficult to avoid the suspicion that α is estimated with bias. Despite this limitation I discuss below some of the determinants of firm size just mentioned. Age: Recent empirical studies suggest a negative correlation between firm age and firm growth. Decreasing returns to learning over time are one major reason. The probability diminishes that an aging firm will achieve additional efficiency gains (Jovanovic 1982; Ericson and Pakes 1995; Das 1995; Farinas and Moreno 2000). This negative association has also been confirmed for German firms (Harhoff, Stahl and Woywode 1998; Steil and Wolf 1999). Legal form: Theoretically, firms are legally constituted so that the owners enjoying limited liability have a greater incentive to pursue risky projects, and therefore expect higher profits and growth rates, than other firms (Stiglitz and Weiss 1981). This hypothesis has been tested empirically, for instance in Germany by Harhoff, Stahl and Woywode (1998), and has not been rejected. Those authors argue that the legal liability of a firm, which is determined by the legal form chosen for it, influences its growth rate. They also show that firms with limited liability have above-average growth rates. Location: Several researchers suggest that agglomeration effects—in the form of both regional concentration of a specific industry and regional concentration of several unrelated economic activities—can produce net positive externalities up to a threshold. Once this threshold is reached, however, negative net externalities can be expected: high traffic, high land prices,

An Empirical Analysis of Industrial Policy in Morocco

61

environmental problems, and others. Geography matters, but its impact on firm growth cannot be determined ex ante.9 Innovative activity: Technical innovations can be divided into product and process innovations. The introduction of product innovations normally results in a new demand, and that of process innovations in a reduction of costs. Both elements affect the growth process of the innovating firm positively (for a survey of the literature see Cohen 1995). Diversification: As already mentioned, diversification also affects the growth process positively. It helps firms to cope with demand constraints on a specific product line and creates new opportunities for growth. Internal organization: In her classic study on firm growth, Penrose (1959) advanced the famous “managerial limits to growth” hypothesis. This argument starts with the premise that management is a team effort in which individuals deploy specialized, functional skills as well as highly team-specific skills that enable them to coordinate their many activities in a coherent manner. As a firm expands, it needs to recruit new managers and must divert at least some existing managers from their current operational responsibilities to help manage the expansion of the management team. This places a constraint on the firm’s growth process. Market size: Numerous empirical studies have confirmed the importance of market demand for a firm’s innovative activities and growth (Cohen 1995; Kleinknecht 1996). It is assumed here that there is a positive correlation between market size and firm growth. Market structure: As discussed above, market structure is a major force behind a firm’s growth. The growth process of firms in competitive markets is driven by different forces than those driving the process in firms under imperfect competition. Industry-specific environment: The variability of firm growth rates may also differ from industry to industry, depending upon the nature of the product, the character of competition, and so on. Dunne, Roberts and Samuelson (1989) show that firms’ growth rates vary significantly among the different industries in the manufacturing sector in the United States. Harhoff, Stahl and Woywode (1998) confirm sectoral differences in growth rates in Germany. Their study also shows that firms in the services sector in particular are characterized by above-average employment growth. Brüderl, Preisendörfer and Ziegler (1996) confirm significant sectoral differences in employment growth rates. Johnson, Baldwin and Hinchley (1997) find a close relation between growth dynamics within a sector and firms’ growth rates. They argue that growth rates of firms in growing sectors should be

62

Najib Harabi

higher than those of firms in stagnating or declining sectors. Young and growing markets are, as a rule, characterized by low barriers to entry, and thus by high rates of entry and exit. Individual firms therefore have different growth potentials as determined by their sector’s life cycle. State regulations and policies: As the framer of the legal environment within which firms operate, and as the largest single domestic customer for goods and services, government affects the ability of firms to grow in a sustainable manner. Empirical Specification This section uses the models of optimal firm size presented above to examine empirically the major forces behind the growth process of Moroccan firms, including government policies. The variables used in this analysis are summarized in Table 1 and described further below. The dependent variable, S(t) from equation (3), can be measured in different ways: as the average annual growth rate of a firm’s sales (this variable is here called SALESG), or as the average annual growth rate of employment (EMPLOYG). On the whole, I estimate two empirical models using each of the above specifications of the dependent variable. The following explanatory variables are drawn from the theoretical and empirical literature described above. Firm size: Firm size in the previous period, corresponding to the variable Si(t - 1) in equation (3), is designated here as FSIZE and measured as the logarithm of firm sales, defined as the average of firm sales in the years 2000–2002. Theoretically this variable could have a positive or a negative impact on firm growth, depending on the characteristics of the firm and the market in which it operates. It depends on the speed—that is, on parameter β in equation (3)—with which Moroccan firms converge toward their longrun steady-state size. The set of observable exogenous variables, X(t) in equation (3), are the following: Firm age: The age of a firm (AGE) is defined as the absolute number of years of existence since start-up. Theoretically it is assumed that younger firms grow faster. Firm location: On the basis of responses to the questionnaire, firms were grouped into six geographical categories: 1.Grand Casablanca (accounting for 60.5 percent of all firms interviewed); 2. Tangier-Tetouan (8.8 percent); 3. Rabat-Sale-Zemmour (5.8 percent); 4. Fès-Boulmane (12.2 percent); 5. Oriental (3); 6. Chaouia-Ouardigha (5.2 percent); This

An Empirical Analysis of Industrial Policy in Morocco

63

information was used to construct five dummy variables: FLOCATION1 takes the value of 1 for firms in the second geographical category and 0 otherwise; FLOCATION2 takes the value of 1 for firms in the third category and 0 otherwise; FLOCATION3 takes the value of 1 for firms in the fourth category and 0 otherwise; FLOCATION4 takes the value of 1 for firms in the fifth category and 0 otherwise, and FLOCATION5 takes the value of 1 for firms in the sixth category and 0 otherwise. This leaves firms in Grand Casablanca, the largest concentration of firms in Morocco, as the benchmark or omitted variable. From the earlier theoretical discussion, firms in large urban centers should grow faster than firms in smaller locations. Legal form: The questionnaire distinguishes among seven different legal forms: single proprietorships, partnerships, cooperatives, privately held corporations, limited liability corporations and public limited companies. Firms with the legal form ‘limited liability’ account for a large majority (80 percent) of all firms interviewed. From this information a dummy variable FSTATUS1 was constructed that takes the value of 1 if the legal form is that of a limited liability company and 0 otherwise. Innovative ability: Another major source of firm growth is the ability to innovate. The two dummy variables PROINNOV and PROCESIN control for this important capability. The first dummy takes the value of 1 if the firm reports engaging in product innovation and 0 if it does not; the second one also takes the value of 1 if the firm reports engaging in process innovation and 0 if it does not. Product and market diversification: A further source of a firm’s growth is the ability to diversify both its existing products and services and its markets. The qualitative variables DIVERS1 and DIVERS2 address this ability. The first variable indicates that the firm diversifies its existing products and services and is offering a certain number of different goods and services. The second takes the value of 1 if the firm is able to diversify its product market and is exporting to foreign markets, otherwise it takes the value of 0. Access to inputs: The ability of firms to obtain access to major inputs is also of paramount importance for their growth. Such assets would include managerial inputs, reflecting Penrose’s “managerial limits to growth” hypothesis (Penrose 1959). The following five variables were constructed to deal with these issues: LWORK measures a firm’s access to qualified workers, LFINANCE assesses its access to external financial resources, LINFRAST gauges its access to good infrastructure (for instance, telecommunications), LINFRAS2 evaluates its access to electricity and LLAND measures its access to industrial land. Each of these variables is measured on a 0–4

64

Najib Harabi

(Likert) scale, where 0 indicates that access to the input is not a major obstacle to growth, and 4 represents a major obstacle. Market structure: A major outcome of an industry’s market structure is whether firms can compete in product markets or not. A concrete expression of this market competition is the existence of a large number of firms competing in the same market. The variable DCOMPETE indicates the absolute number of domestic and foreign competitors. In addition, the variable MCOMPETE is measured on a scale from 0 to 4, where 0 means that the firm is not facing a severe competition and 4 indicates that it is facing severe competition, especially from the informal sector. Finally the variable PCOMPETE measures price elasticity of demand in the relevant market. This is an indicator of the nature of competition in product market. Theoretically, competitive markets are characterized by a perfectly elastic demand. Market demand: Expected demand in a firm’s product market enters the equation through the variable MDEMAND, measured as a dummy variable that takes the score of 1 if the firm reports that it has positive expectations of either domestic or foreign demand, otherwise 0. Theoretically, it is anticipated that greater expected market demand will enhance firm growth. State regulations and policies: In the survey, firms were asked whether each of the following types of regulations and government policies (or consequences of poor policies) was not a major obstacle for growth (value of 0) or, on the contrary, a major obstacle (value of 4): • Regulation of foreign trade: level of customs duties and management of the customs services • Tax regulation (relationship with tax administration) • Level of taxes • Regulation of the labor force • Interest rate policy • Inflation and volatility of exchange rates • Effectiveness of government policies in providing public goods (infrastructure, transportation, security, etc.) • Corruption Table 3 summarizes the average responses to each of these eight questions. On the whole, state regulations and policies are considered obstacles to doing business in Morocco. Their signs cannot be, however, predicted ex ante, since their impact on corporate growth depends on the specific situation of the firm and the industry it belongs to. Interindustry differences: Theoretical and empirical studies suggest substantial interindustry differences with respect to firm growth (see the

An Empirical Analysis of Industrial Policy in Morocco

65

discussion above). To control for these differences, industry dummies have been included in the regression analysis. According to the survey data, the garment industry was the most frequently cited branch of activity (42 percent). This industry is therefore used here as a benchmark. For the remaining industries—textile, leather and footwear, rubber and plastics, food processing, chemical industry, and the electrical industry—dummy variables were constructed, and assigned the value of 1 when the firm’s principal activity was in that industry, otherwise taking a value of 0. Data Ideally, the empirical model of firm growth should be tested on the basis of panel data, to more fully reveal the growth dynamics of Moroccan firms. Unfortunately, panel data for all the variables described above do not yet exist. What is available is a cross-sectional data set, based on a field survey of 850 firms carried out under the auspices of the World Bank in 2004. The survey sample covers firms of different sizes: large (more than 100 workers), medium (50 to 100 workers), and small (10 workers or more).10 It also addresses all of the major manufacturing industries in Morocco. The sample of firms under consideration is, for various reasons, not statistically representative of the universe of Moroccan firms. One reason is that the universe of firms is itself not really known but varies, according to the source, between 270,888 (from the 1995 patent registry) and 900,687 firms (from the official statistical yearbook for 1996). In addition, the sampling method and the number of units drawn are not statistically adequate. Despite these shortfalls, the sample allows an explorative analysis of firm behavior in Morocco. Econometric Problems A significant problem relates to the noise in the data. This is mostly due to the fact that almost all of the variables have the measurement properties of categorical data. To be useful in the econometric analysis, these responses have to be converted into dummy variables. A second problem is that there are missing values for firms in the data set that cannot be included in our estimate of equation (3). Since the remaining observations with no missing values were not selected randomly, this gives rise to sample selection bias in the data. In the presence of this specification error, the ordinary least squares procedure cannot be used to

66

Najib Harabi

estimate equation (3). An alternative procedure is the full information maximum likelihood (FIML) method developed by Heckman (1976).11 This method corrects for the specification error due to sample selection bias. Results As already mentioned above, two regression equations, using different specifications of the dependent variable (SALESG and EMPLOYG), were estimated (see Tables 4 and 5). Due to the better econometric quality of the EMPLOYG equation, I will present and discuss the results of this equation only (see Table 5): • Firm size (FSIZE) seems to have a positive impact on firm growth: the larger a firm was in 2000–02, the higher the probability of it being expected to grow. An acceleration of the convergence process toward a long-term steady-state size takes place. In other words, larger firms grow faster than smaller ones. This result, which is in this case statistically significant, is not consistent with some previous empirical findings in developing countries, as discussed above.12 • Firm age (AGE) has, in contrast, a negative impact on firm growth. Younger firms grow faster. Other research has shown that they are also the ones that are more likely to export than older firms (Fafchamps, El Hamine, and Zeufack 2002).13 • Firm location (variables FLOCATION1–FLOCATION5) also matters. Compared with firms located in the large urban centers (Grand Casablanca, Fès-Boulmane etc.), those in medium-size urban centers and especially those in smaller centers (for instance, Chaouia-Ourdigha) expect less growth. The regional dimension of firm growth is also important in Morocco, as one would expect. • The legal form of the enterprise (FSTATUS) also affects the growth process. Being a limited liability company is negatively correlated with the firm’s growth prospects. • There is some evidence indicating that the ability of a firm to innovate (as measured by the variables PROINNOV and PROCESIN) is not positively correlated with employment growth, but the two variables, especially the one related to process innovation, are not statistically significant. • A further positive source of growth is a firm’s ability to diversify its existing products and services: the variable DIVERS1 is positively

An Empirical Analysis of Industrial Policy in Morocco

•

•

•

•

•

correlated with growth, although not statistically significant. On the other hand, firms that try to diversify their product markets and export are even more successful: the sign on DIVERS2 is positive and numerically stronger. Access or no access to at least some major inputs also has an impact. Lack of access to external financial resources (LFINANCE) and to basic infrastructure, such as electricity (LINFRAS2) seems to be detrimental to the growth process of Moroccan firms. Less severe impediments are lack of access to qualified labor force (LWORK), industrial land (LLAND), and telecommunications (LINFRAST). The market structure and the competitive environment under which firms are operating affect their behavior regarding the quantity of products to be produced and pricing policy. In our case, this market structure measured by the number of competitors in a specific market (DCOMPETE) and by the qualitative perception of competitive pressure by firms interviewed (MCOMPETE), has a negative impact on a firm’s employment growth. Market demand seems to exert an important impact on firm growth: the MDEMAND variable shows a positive and statistically significant coefficient. Firms operating in industries such as leather and footwear, rubber and plastics, food processing, chemical industries, and electrical machinery have experienced a less favorable growth environment than those in the garment (benchmark) and textile industries. Industryspecific factors, as measured here by industry dummies, do matter for the growth processes of firms operating in those industries. State regulations and policies appear to have mixed effects. Effective government policies aiming at improving the quantity and quality of public goods, such as infrastructure, public transportation, security, education, and public health, seem to affect firm growth positively (the sign of the variable GOVERNM1 is positive and statistically significant at the 5 percent level). All other government policies included in our econometric analysis seem, however, to have a negative impact on firm growth. Tax policy toward firms and interest rate policy seem to be especially detrimental to the employment growth of Moroccan firms (the signs of the variables GOVERNM4 and GOVERNM6 are negative and statistically significant at the 1 percent level).

67

68

Najib Harabi

III. Summary and Conclusions The purpose of this study was firstly, to describe the major instruments of industrial policy in Morocco since its independence in 1956, and secondly, to evaluate them empirically. Regarding the second objective, the research question has been whether government policies have contributed to the economic growth of private firms. This question has been raised and tested in a broader model of firm growth. Using firm-level data for 2004, the principal factors positively affecting firm growth in Morocco were found to be the following: firm size, with larger firms growing faster than smaller firms; business strategies that focus on product diversification and market diversification, especially export markets; location (in large urban centers); strong and predictable demand for the firm’s products; and government policies that are aimed at improving the quantity and quality of public goods, such as infrastructure, public transportation, and security. The principal factors that affect firm growth negatively are the following: firm age (younger firms grow faster); legal status as a limited liability company; ability to innovate (technological innovation has not been a source of growth in the Moroccan context, therefore only a few firms—less than 5 percent—are technologically innovative); lack of access to external finance and basic infrastructure; market structure and competition; location in small population centers, and most government policies that have been included in the econometric analysis, such as tax policy toward firms, customs policy, and interest rate policy. If confirmed by further analysis, these results have significant policy implications for both business leaders and policymakers in Morocco. For business leaders, it is important to emphasize that an explicit and sound growth strategy matters. Essential aspects of such a strategy include choosing the right location and legal form, and markets with sufficiently strong and expanding demand. A promising growth strategy for firms in Morocco is to diversify both the products and services offered and their markets, via export. For policymakers, the analysis suggests several policy areas where improvements may be needed. First, the regulatory and administrative framework should be adjusted to become more responsive to the needs of firms that are able and willing to grow. In this respect, competition policy has an important role in ensuring fair play among competing firms. Second, policies regarding education and professional training must be targeted to the needs of firms. It is striking that in a country where thousands of college and university graduates are unemployed, lack of access to qualified workers and managers

An Empirical Analysis of Industrial Policy in Morocco

69

constitutes a major hindrance to firm growth. The mismatch between the skills supplied by the labor force and the skills demanded by employers has to be fixed. Third, regional disparities with regard to infrastructure (roads and utilities, among others), availability of manpower, life and work quality have to be addressed, because these imparities present major obstacles for growth seeking firms located in certain parts of the country such as Kenetra, Oujda, Marrakech, Khemisset, Larache, and Skhirat. However, excessive government intervention is not conducive for firm growth. A major lesson of this case study is that there are indirect clues indicating the inefficacy of industrial policies in Morocco, measured by their impact on firm growth.

Notes 1.

I would like to thank Mr. Najy Benhassine of the World Bank for authorizing me to use the 2004 World Bank data set of firms. 2. Krueger, A. (1990), Deepak Lal (1995), Hayek, von (1994). 3. The Moroccan state had introduced four codes of industrial investments between 1958 and 1982. 4. Around five hundred thousand hectares of agricultural land have been sold by foreign owners, mostly French, to Moroccans. 5. In reference to the United States, see Evans (1987a, 1987b), and Hall (1987); the United Kingdom, see Hart and Oulton (1995, 1996, 1998), Dunne and Hughes (1996), and Geroski (1998); Germany, see Wagner (1994), Brüderl, Preisendörfer and Ziegler (1998), Brixy and Kohaut (1999), Steil and Wolf (1999), and Almus (2000); and concerning Switzerland, see Harabi (2002). 6. There are, of course, other theoretical perspectives on firm growth. The most important are evolutionary models of firm growth (Neslon and Winter, 1982) and stochastic growth models; for a survey of these models, see Sutton (1997). 7. The word ‘satisficing’ was invented by Herbert Simon (1959) as a hybrid of the words ‘satisfy’ and ‘suffice.’ 8. For work on the effects of age, see Evans (1987), Dunne and Hughes (1994), and Das (1995); of R&D expenditures, see Hall (1987) and Liu, Tsou, and Hammit (1999); for mergers and acquisitions, see Ijiri and Simon (1974); for the internal organization of firms see Dunne, Roberts, and Samuelson (1989), and Variyan and Kraybill (1992). For recent overviews of the literature, see Sutton (1997) and Hart (2000). 9. Authors who have studied the relationship between location and firm growth include North and Smallbone (1994), Storey (1994), and Henderson (1994). 10. The size distribution in the World Bank sample is as follows: 40 percent small firms, 38 percent medium size and 22 percent large firms. 11. See also the exposition in Greene (2000, 693–96). 12. The result that firm size is negatively correlated with growth in Morocco has also been found in many other developing countries. It has been established both through cross-country analysis (Leidholm and Mead 1987; Banarji 1987), and through analysis across time within countries (Little, Mazumdar, and Page 1987; Steel 1993).

70

Najib Harabi

13. The same source finds that old firms are unlikely to switch to exporting, even in response to changes in macroeconomic incentives to export.

Figure 1. Sample Description ICA 2004

An Empirical Analysis of Industrial Policy in Morocco

71

Table 1. Variable Description Variable

Description

Dependent Variable SALESG

Logarithm of the average annual rate percentage of growth of sales from 2002 to the year of firm establishment.

EMPLOYG

Logarithm of the average annual rate percentage of growth from 2002 to the year of firm establishment.

Independent Variables Firm-specific FSIZE

Logarithm of firm sales as an average of firm sales in 2000, 2001, and 2002.

FLOCATION1

Dummy variable with value of 1 if firm is headquartered in Tangier-Tetouan, otherwise 0.

FLOCATION2

Dummy variable with value of 1 if firm is headquartered in Rabat-Sale-Zemmour, otherwise 0.

FLOCATION3

Dummy variable with value of 1 if firm is headquartered in Fès-Boulmane, otherwise 0.

FLOCATION4

Dummy variable with value of 1 if firm is headquartered in Oriental, otherwise 0.

FLOCATION5

Dummy variable with value of 1 if firm is headquartered in Chaouia-Ouardigha, otherwise 0.

AGE

Number of years of firm’s existence, between 2004 and the year of its establishment.

FSTATUS1

Dummy variable with value of 1 if firm is established as a limited-liability corporation, otherwise 0.

PROINNOV

Dummy variable with value of 1 if firm reports that it engages in product innovation, otherwise 0.

PROCESIN

Dummy variable with value of 1 if firm reports that it engages in process innovation, otherwise 0.

DIVERSE1

Number of the different products a firm is producing, in absolute terms.

DIVERSE2

Dummy variable with value of 1 if firm reports that it is exporting, otherwise 0.

Independent Variables Access to Inputs LWORK

Access of the firm to qualified labor force, measured on a 0–4 scale. A score of 0 indicates that access is not a major obstacle; and 4 represents a major obstacle.

72

Najib Harabi

LFINANCE

Access of the firm to outside bank financing, measured on a 0–4 scale. A score of 0 indicates that access is not a major obstacle; and 4 represents a major obstacle.

LINFRAST

Access of the firm to telecommunication infrastructure, measured on a 0–4 scale A score of 0 indicates that access is not a major obstacle; and 4 represents a major obstacle.

LLAND

Access of the firm to industrial land, measured on a 0–4 scale. A score of 0 indicates that access is not a major obstacle; and 4 represents a major obstacle.

LINFRAS2

Access of the firm to electricity, measured on a 0–4 scale. A score of 0 indicates that access is not a major obstacle; and 4 represents a major obstacle.

Independent Variables Market structure

DCOMPETE

Number of competitors in the market, in which a firm is operating, in absolute terms.

PCOMPETE

Price elasticity of domestic demand in which a firm is operating, measured on a 1–4 scale. A score of 1 indicates that the elasticity is almost 0 and 4 indicates it is important.

MCOMPETE

Dummy variable that indicates the severity of competition from the informal sector, measured on a 0–4 scale. A score of 0 indicates that this kind of competition is not a major obstacle; and 4 represents a major obstacle.

MDEMAND

Dummy variable that takes the score of 1 if the firm reports that it has positive expectations of either domestic or foreign demand, otherwise 0.

Independent Variables Industry Dummies

TEXTILE

Dummy variable with value of 1 if firm reports that its primary activity is textile.

LEATHER

Dummy variable with value of 1 if firm reports that its primary activity is leather and footwear.

RUBBER

Dummy variable with value of 1 if firm reports that its primary activity is rubber and plastics.

FOOD

Dummy variable with value of 1 if firm reports that its primary activity is food processing.

CHEMICAL

Dummy variable with value of 1 if firm reports that its primary activity is chemical industries.

ELECTRIC

Dummy variable with value of 1 if firm reports that its primary activity is electrical machinery

An Empirical Analysis of Industrial Policy in Morocco

73

Independent variables Policy Variables GOVERNM1

Effectiveness of government policies regarding the provision of public goods, such as infrastructure, public transportation, security, education, and health. A score of 1 indicates that the policy is very effective and 6 shows that it is very ineffective.

GOVERNM2

Importance of Customs policy for firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle; and 4 represents a major obstacle.

GOVERNM3

Importance of the relationship between firm management and customs authorities for firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle; and 4 represents a major obstacle.

GOVERNM4

Importance of tax rate policy for firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle and 4 indicates that it is a major obstacle.

GOVERNM5

Importance of labor code policy to firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle and 4 indicates that it is a major obstacle.

GOVERNM6

Importance of interest rate policy for firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle; and 4 represents a major obstacle.

GOVERNM7

Importance of inflation and exchange rate policy for firm growth, measured on a 0–4 scale. A score of 0 indicates that the policy is not a major obstacle; and 4 represents a major obstacle.

GOVERNM8

Corruption and firm growth, measured on a 0–4 scale. A score of 0 indicates that corruption is not a major obstacle; and 4 represents a major obstacle.

Source: Author’s model specifications.

Table 2. Reported Impact of State Policies on Firms in the Sample Type of Regulation or Policy Customs Relationships with tax administration Level of taxes Regulations on labor force Corruption Interest rate policy Effectiveness of government policies in providing public goods (infrastructure, transportation, security, etc.) Inflation and volatility of exchange rates

Average Response (4 = Severe Obstacle, 0 = No Obstacle)

74

Najib Harabi

Table 3. Descriptive Statistics Variable sales Lsales AGE worker Lworker Lbirth Sbirth SALESG EMPLOYG FSIZE FSIZE2 FLOCATION1 FLOCATION2 FLOCATION3 FLOCATION4 FLOCATION5 LAGE FSTATUS PROINNOV PROCESIN DIVERSE1 DIVERSE2 LWORK LFINANCE LINFRAST LINFRAS2 LLAND TEXTILE LEATHER RUBBER FOOD CHEMICAL ELECTRICA OTHERIND DCOMPETE MCOMPETE PCOMPETE GOVERNM1 GOVERNM2 GOVERNM3 GOVERNM4 GOVERNM5 GOVERNM6 GOVERNM7 GOVERNM8 MDEMAND

N

Mean

Std Dev

Minimum

Maximum

799 799 850 803 803 850 850 799 803 799 803 850 850 850 850 850 850 850 850 850 848 850 850 850 850 850 850 850 850 850 850 850 850 850 345 850 484 849 850 850 850 850 850 850 850 850

113103342 16.1677876 18.2176471 124.5421337 4.1118932 0 0 1.3426366 0.3811207 16.1677876 4.1118932 0.0494118 0.0247059 0.0505882 0.1070588 0.1211765 2.6171099 0.7658824 0.4482353 0.3447059 4.0931604 0.6423529 1.3694118 3.0694118 0.3494118 0.7082353 1.9694118 0.1882353 0.0941176 0.0905882 0.0847059 0.0705882 0.0341176 0.0447059 101.6608696 1.7388235 2.8636364 3.6548881 1.0800000 1.7894118 2.6000000 1.1776471 3.1411765 1.4011765 0.9764706 0.4282353

1561022324 1.6079559 13.8153196 202.5754507 1.1499249 0 0 0.9723374 0.3393893 1.6079559 1.1499249 0.2168538 0.1553186 0.2192844 0.3093701 0.3265244 0.7911986 0.4236954 0.4976060 0.4755521 19.1892224 0.4795897 1.2960049 1.2991366 0.6993390 1.0807461 1.4402937 0.3911301 0.2921642 0.2871916 0.2786075 0.2562866 0.1816382 0.2067790 164.7236880 1.6387601 0.7761987 1.0672451 1.2019615 1.3630254 1.2581887 1.2007925 1.1913799 1.3398833 1.3316521 0.4951144

64509.67 11.0745704 2.0000000 4.6666667 1.5404450 0 0 0.2256233 0.0443645 11.0745704 1.5404450 0 0 0 0 0 0.6931472 0 0 0 1.0000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1.0000000 1.0000000 0 0 0 0 0 0 0 0

40300862738 24.4196387 80.0000000 1970.67 7.5861272 0 0 6.9207902 2.3263818 24.4196387 7.5861272 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 4.3820266 1.0000000 1.0000000 1.0000000 500.0000000 1.0000000 4.0000000 4.0000000 4.0000000 4.0000000 4.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1050.00 4.0000000 4.0000000 6.0000000 4.0000000 4.0000000 4.0000000 4.0000000 4.0000000 4.0000000 4.0000000 1.0000000

An Empirical Analysis of Industrial Policy in Morocco

75

Table 4. The MODEL Procedure Nonlinear FIML Summary of Residual Errors DF

DF

Equation

Model

Error

SSE

MSE

Root MSE

adj R-Square

R-Sq

SALESG

37

147

58.2868

0.3965

0.6297

0.6272

0.5358

Nonlinear FIML Parameter Estimates Approx

Approx

Parameter

Estimate

Std Err

a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37

1.65711 -0.04319 -0.00653 0.010953 -0.19828 -0.22699 0.031682 -0.00255 0.167101 0.091866 0.019128 -0.00307 -0.05588 0.067304 0.362904 -0.03988 -0.00485 0.14694 0.489865 0.440694 -0.00043 -0.04306 0.022344 0.073642 -0.04156 -0.02448 -0.09105 -0.00214 0.078265 -0.04539 -0.00087 0.384035 -0.03009 0.74232 0.320544 -0.20917 0.190454

0.4150 0.00394 0.0557 0.1972 0.1079 0.1166 0.1316 0.00813 0.1060 0.0454 0.0641 0.0702 0.0536 0.0342 0.1629 0.1860 0.1807 0.2286 0.2093 0.3757 0.000392 0.0307 0.0648 0.0460 0.0588 0.0548 0.0480 0.0500 0.0742 0.0441 0.0428 0.2670 0.2890 0.3349 0.1632 0.3097 0.1016

t Value 3.99 -10.96 -0.12 0.06 -1.84 -1.95 0.24 -0.31 1.58 2.02 0.30 -0.04 -1.04 1.97 2.23 -0.21 -0.03 0.64 2.34 1.17 -1.10 -1.40 0.35 1.60 -0.71 -0.45 -1.90 -0.04 1.05 -1.03 -0.02 1.44 -0.10 2.22 1.96 -0.68 1.87

Pr > |t| 0.0001 |t| 0.5766