Finance and Economic Development

Published on July 2016 | Categories: Documents | Downloads: 63 | Comments: 0 | Views: 711
of 136
Download PDF   Embed   Report

Comments

Content

THREE ESSAYS ON FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH
DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Pilhyun Kim, M.A. ***** The Ohio State University 2006

Dissertation Committee:
Dr. Paul Evans, Adviser Dr. Masao Ogaki Dr. Pok-sang Lam

Approved by

Adviser Graduate Program in Economics

ABSTRACT

The primary part of my dissertation investigates the potential effects of financial sector development on economic growth. In order to reveal the nature of these effects, I focus on the potential channels of influence from the financial to the real sector. I investigate the link between the financial sector and economic growth focusing on the role of the financial sector in funding innovative activities. To this aim, I construct a model where the economy is driven by innovative activities that require both human capital and external funding. My analysis shows that when certain conditions are satisfied, there exists a unique equilibrium where the growth rate of the economy is jointly determined by the levels of human capital and financial development. An implication of this is that financial liberalization policies that do not adequately address the fundamentals of the economy can cause bank failures and possibly a financial crisis. Furthermore, the model suggests that, depending on the parameter values of the economy, there may be two forms of poverty traps, one with a small number of bankers and the other with a large number of bankers. Also, I examine empirically whether financial development has any effect on the rate of technological innovation using patent applications as a proxy for innovative output. For a sample of twenty eight countries from 1970 to 2000, my analysis shows that financial development is indeed significant in raising the growth rate of innovative output. ii

In addition, I investigate whether financial development enhances investment efficiency. The efficiency channel hypothesis states that financial development may increase the efficiency of investment by directing the funds to the most productive uses. I examine if there is any evidence of financial development positively affecting the efficiency of aggregate investment using developing countries as a sample. Compared to the volume channel, the efficiency channel has received relatively little attention until recently. I address the issue of the efficiency channel using two alternative measures of aggregate investment efficiency. I find that, for developing countries, financial development significantly and positively affects productivity of investment.

iii

To my parents and my family

iv

ACKNOWLEDGMENTS

I am indebted to Dr. Paul Evans, my advisor, for his valuable and expert guidance, insightful comments and encouragement during the course of this study. Without his patient guidance my dissertation would have been impossible. Special gratitude is extended to Dr. Masao Ogaki and Dr. Pok-sang Lam for their comments and valuable help to improve my study. Financial support from the PEGS Research Grant is greatly acknowledged as well as the Graduate Teaching Assistantship the Department of Economics have offered me throughout my residence at the Ohio State University. In addition, my special thanks go to my wife Jihee and my family who have been always supporting and praying for me. Most of all, I would like to express my deepest appreciation to my father and mother for their belief in my abilities and undying support throughout my life.

v

VITA

January 9, 1969 . . . . . . . . . . . . . . . . . . . . . . . . . . . . Born - Gwanju, Korea 1991 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.A. Economics, University of Wisconsin, Madison 1998 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M.A. Economics, The Ohio State University 2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ph.D Candidate, The Ohio State University 1998 - 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graduate Teaching Associate, The Ohio State University 2004 - 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lecturer, The Western Washington University

FIELDS OF STUDY
Major Field: Economics Studies in: Money Macroeconomics Applied Econometrics Economic Growth

vi

TABLE OF CONTENTS

Page Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vita . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapters: 1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Models of the finance-led growth theory . . . . . . . . . . . . . . . 2.2 Empirics of the finance-led growth theory . . . . . . . . . . . . . . 3. A Finance-led Growth Hypothesis: Revisited . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . 3.2 Model . . . . . . . . . 3.2.1 Environment . 3.2.2 A formal model 3.3 Discussion . . . . . . . 3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 5 6 9 12 12 16 16 18 27 30 ii iv v vi x xi

vii

4.

How Does Financial Development Promote Growth? . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4.2 Related Literature . . . . . . . . . . . . . . . . . . . . 4.2.1 Theories of the finance-led growth hypothesis . 4.2.2 Empirical studies . . . . . . . . . . . . . . . . . 4.3 Theoretical background . . . . . . . . . . . . . . . . . 4.3.1 Final goods sector . . . . . . . . . . . . . . . . 4.3.2 Intermediate goods sector . . . . . . . . . . . . 4.3.3 The research sector . . . . . . . . . . . . . . . . 4.3.4 The growth of the economy . . . . . . . . . . . 4.4 Empirical analysis . . . . . . . . . . . . . . . . . . . . 4.4.1 Patents as a proxy for technological innovation 4.4.2 Data . . . . . . . . . . . . . . . . . . . . . . . . 4.4.3 Methodology . . . . . . . . . . . . . . . . . . . 4.4.4 Estimation . . . . . . . . . . . . . . . . . . . . 4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35 35 37 37 39 41 42 42 44 45 47 47 50 54 61 63 71 71 74 77 79 81 84 84 87 90 92 93 94 95 96

5.

Investment Efficiency and Financial Development . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Investment, output growth, and financial development . . . . 5.2.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Estimation strategy . . . . . . . . . . . . . . . . . . . 5.2.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Investment efficiency . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Measures of investment efficiency . . . . . . . . . . . . 5.3.2 Investment efficiency estimates . . . . . . . . . . . . . 5.3.3 Determinants of investment efficiency . . . . . . . . . . 5.3.4 Estimation . . . . . . . . . . . . . . . . . . . . . . . . 5.3.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.6 The effects of political stability and legal environment 5.3.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

Appendices:

viii

A. B.

Data Sources for Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . 113 Countries Used in Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . 115

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

ix

LIST OF TABLES

Table 4.1 Sample Countries Used for Patent Regression . . . . . . . . . . . . . 4.2 Fixed Effects Estimation I . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Fixed Effects Estimation II . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Fixed Effects Estimation with Selected Independent Variables . . . .

Page 67 68 69 70

5.1 Investment Regression I . . . . . . . . . . . . . . . . . . . . . . . . . 100 5.2 Investment Regression II . . . . . . . . . . . . . . . . . . . . . . . . . 101 5.3 Investment Regression III . . . . . . . . . . . . . . . . . . . . . . . . 102 5.4 Investment Efficiency Estimates using GDP . . . . . . . . . . . . . . 103 5.5 Investment Efficiency Estimates Using IVA . . . . . . . . . . . . . . . 105 5.6 Investment Efficiency Regression I . . . . . . . . . . . . . . . . . . . . 107 5.7 Investment Efficiency Regression II . . . . . . . . . . . . . . . . . . . 108 5.8 Investment Efficiency Regression III . . . . . . . . . . . . . . . . . . . 109 5.9 Investment Efficiency Regression IV . . . . . . . . . . . . . . . . . . . 110

x

LIST OF FIGURES

Figure 3.1 Screening Costs of Competing Bankers . . . . . . . . . . . . . . . . . 3.2 Equilibrium When α + θ < 1 . . . . . . . . . . . . . . . . . . . . . . . 3.3 Effects of Deterioration in Education . . . . . . . . . . . . . . . . . . 4.1 Long Run GDP Growth vs. Long Run Patent Growth . . . . . . . . .

Page 32 33 34 65

4.2 Movements of Financial Development Indicators for Selected Countries 66 5.1 Output Growth and Investment in Cambodia . . . . . . . . . . . . . 5.2 The Share of Industry Value Added in GDP in Developing Countries 98 99

xi

CHAPTER 1

Introduction

The primary part of my dissertation investigates the potential effects of financial sector development on economic growth. In order to reveal the nature of these effects, I focus on the potential channels of influence from the financial to the real sector. The nature of the interaction between the real and the financial sectors has been hotly debated among researchers. Those who favor the finance-led growth hypothesis argue that the existence of a vibrant financial sector has growth-enhancing effects. In this literature, an economy can grow faster due to an efficient allocation of resources by the financial sector, mainly banks. A number of channels of influence have been proposed in the literature, which include increased savings, increased investment, increased efficiency thereof, increased human capital accumulation, and positive effects of the financial sector on innovation processes. Investigations of the validity of these channels as true agents of long-run growth, so far, have yielded mixed empirical results. In chapter 2, I review the vast literatue on this subject to examine how the literature has evolved over time. In chapter 3, I investigate the link between the financial sector and economic growth focusing on the role of the financial sector in funding innovative activities. My 1

motivation is based on the research by Easterly and Levine (2001) in that it is the residual that accounts for most of the income and growth differences across nations. Broadly speaking, innovative activities require both human and financial capital. They act as complements in the production function of ideas. However, the interaction between the two has been largely ignored in the literature. In order to improve our understanding of how the financial sector interacts with the real sector, the nature of interactions between two major components of innovative activities needs to be examined more closely. In this chapter, I pursue this goal by constructing a model where the economy is driven by innovative activities that require both human capital and external funding from the financial sector. Similar to King and Levine (1993), it is assumed that the role of the financial sector in this model is to screen the innovators for their probabilities of success. In formalizing the model, I depart from the conventional literature in four important ways. Firstly, I define innovation as a success not when it is realized but when it is commercially successful. This distinction is motivated by observations that not all innovations are implemented in the production processes. Secondly, I assume that the magnitude of technological change an innovator comes up with is a function of that innovator’s human capital. Thirdly, it is assumed in this model that external finance is needed not for R&D activities but for utilization of innovations. Finally, it is assumed that the financial sector pays no setup costs. My analysis shows that when certain conditions are satisfied, there exists a unique equilibrium where the growth rate of the economy is jointly determined by the levels of human capital and financial development. An interesting implication of this is that financial liberalization policies that do not adequately address the fundamentals of the economy can bring about bank failures and possibly a financial crisis. Furthermore, 2

in addition to showing that poverty traps can be explained without introducing setup costs, the model suggests that, depending on the parameter values of the economy, there may be two forms of poverty traps, one with a small number of bankers and the other with a large number of bankers. In chapter 4, I examine empirically whether financial development has any effect on the rate of technological innovation. A flood of empirical studies began to appear in the 1990s to test the validity of the finance-led growth hypothesis. A typical test strategy involves regressing some indicator of financial development on some aggregate growth measures such as investment growth, GDP growth or total factor productivity growth. By and large, the current empirical literature lacks one crucial element in that it does not consider the channels of influence suggested by theoretical models and fails to show how financial development affects economic growth. In order to examine the validity of the finance-led growth hypothesis, I depart from the conventional literature. Instead of estimating a relationship between aggregate growth measures and financial development indicators, I test the validity of the finance-led growth hypothesis by focusing on the innovation channel of influence, using patent applications as a proxy for innovative output. Under the framework of ideas-driven growth, the hypothesis I test is that financial development enhances innovation, which is the main engine of economic growth. If the finance-led growth hypothesis is right, as the financial sector develops over time in a certain country, the growth rate of innovation should be higher, which would, then, lead to faster economic growth due to a rising level of productivity. Using panel data on twenty eight countries from 1970 to 2000, my analysis shows that financial development is indeed significant in raising the growth rate of innovative output. 3

According to the finance-led growth hypothesis, financial development affects investment in two ways. Firstly, a better developed financial sector may raise the investment rate by pooling and risk sharing. This is the so-called volume channel. Secondly, the efficiency channel hypothesis states that financial development may increase the efficiency of investment by directing the funds to the most productive uses. In chapter 5, I examine if there is any evidence of financial development positively affecting the efficiency of aggregate investment using developing countries as a sample. Compared to the volume channel, the efficiency channel has received relatively little attention until recently. In this chapter, I address the issue of the efficiency channel using two alternative measures of aggregate investment efficiency. I find that, for developing countries, financial development significantly and positively affects productivity of investment. Further, I depart from the existing studies by focusing on the banking sector to measure the degree of financial development. In chapter 6, I conclude.

4

CHAPTER 2

Literature Review

The literature on the finance-led growth hypothesis is vast. Accordingly, a few survey papers have been written on this subject. See, for example, Levine (1997) and Tsuru (2000). The focus of this review is, therefore, not to present an exhaustive review of the literature as it would be rather redundant, but to assess critically how the literature has evolved over time and to identify the remaining issues that need to be resolved. Theoretical models of the finance-led growth hypothesis are, in general, modified versions of endogenous growth theories with risky investment opportunities. Due to uncertainty about the outcome of investment, the allocation of resources is suboptimal. The financial sector enters this world to reduce welfare loss resulting from this uncertainty by providing information, risk-pooling, and liquidity. As the financial sector develops, possibly as a result of feedback from economic growth, the provision of these services becomes more efficient so that a faster rate of economic growth is realized. The overriding research theme that came out of the theoretical models and that has occupied the attention of the empirical researchers for the past decade is a question of 5

whether growth rate of an economy is positively correlated with the level of financial development. A typical method employed to test this theory has been to regress some aggregate growth variables on financial development indicators that are based on some ratio of monetary aggregates to GDP. Simple as it may be in its approach, this line of research has produced an impressive amount of evidence for the finance-led growth theory. In what follows, I examine more closely how the literature has evolved over time both theoretically and empirically.

2.1

Models of the finance-led growth theory

An initial theoretical interest lay in the effects of financial development on the efficiency of capital accumulation. Greenwood and Jovanovich (1990) is among the frontiers of this approach. They assume an economy where growth is driven by capital accumulation, and the risky nature of investment prevents agents from allocating resources in an efficient manner, resulting in welfare loss in the absence of the financial sector. However, as the economy grows, it becomes able to pay the setup costs to establish the financial sector of which the role is to collect and process information and to pool risks across investors/savers. Once the financial sector is established, it allows a higher rate of return to be earned on capital, promoting economic growth. Bencivenga and Smith (1991) pursue a similar vein and present a model where random liqudity shocks raise the fraction of savings invested in liquid but unproductive assets. The financial sector enters this world exogenously, contrary to Greenwood and Jovanovich, to provide liquidity to economic agents and makes it possible for them to invest a larger portion of their savings in productive and illiquid assets. Although the specific types of risk assumed in these models are different, the nature of

6

the fundamental role the financial sector plays is essentially the same. By processing information and reducing risks, the financial sector reduces uncertainty associated with investment and directs funds to their most productive uses, and consequently, enhances the rate of economic growth. As doubts about the effectiveness of capital accumulation in promoting long-term growth arose, researchers shifted their focus to potential positive effects financial development may have on productivity growth. The main sources of productivity growth considered in this literature are purposeful innovative activities and market specialization that are risky. Uncertain nature of the outcomes of innovative activities (Fuente and Marin, 1996) and market specialization characterized by increasing number of firms (Galetovic, 1996; Greenwood and Smith, 1997) make monitoring necessary. Since monitoring is assumed to be costly, there is an incentive for the financial sector to endogenously emerge to economize on the monitoring costs as in Diamond (1983). The provision of monitoring services by the financial sector then leads to increased levels of innovative activities and market specialization, which result in enhanced economic growth. The main weakness of this type of model is that there seems to be no empirical evidence of the financial sector conducting active monitoring (Allen and Gale, 2001). Furthermore, the proposition that the financial sector actively monitors the outcome of investment may have a weaker ground in those countries where the banking sector provides a large share of external finance with debt contracts.. Since debt contracts require the investor/borrower to repay a fixed amount after a certain period independent of the outcome of investment, the banking sector does not have an incentive to actively monitor the activities of the investor/borrower except ascertaining the state of the outcome of such activities at 7

maturity. As such, the models above may not be applicable to the experiences of many, especially developing, countries. King and Levine (1993) and Harrison et al (1999) explore different possibilities. Instead of the monitoring role, King and Levine (1993) focuses on the screening role of the financial sector. Again, innovation that drives economic growth is assumed to be risky. Each innovator has a different ability to innovate and hence different probability of success. The financial sector in this case enhances growth by weeding out those innovators with low probability of success by screening and improving the aggregate probability of success. Harrison et al (1999) departs from the assumption of asymmetric information and considers the effects of agency costs. In their model, the economy grows by investment aimed at promoting productivity growth. Higher agency costs due to a poorly developed financial sector hinders growth by reducing the portion of savings that is channeled into investment. Financial development reduces the costs of intermediation and promotes growth by allowing productivity improvement. In turn, economic growth raises bank profits and induces more entry of banks. As more banks enter, more specialization of banks occurs that reduces costs of intermediation further, which, in turn, raises the investment rate feeding back to still higher economic growth. In general, the models described so far suffer from three drawbacks. Firstly, as mentioned above, the assumption of setup costs, needed to introduce poverty traps, seems empirically unjustified. Secondly, and related to the first point, is the fact that finance-growth relationship is likely to be nonlinear. That is to say that the effects of financial development on economic growth seems to get weaker as the level of financial

8

development rises, possibly suggesting diminishing marginal returns. This evidence directly contradicts the existing theories where economic growth is a (monotonically) increasing function of the level of financial development. Finally, the models where the financial sector actively monitors do not seem to accurately reflect the experiences of many countries.

2.2

Empirics of the finance-led growth theory

First systematic investigation of the relationship between finance and growth was conducted by King and Levine (1993). In their study, it was found that the level of financial development was positively correlated with growth variables such as GDP per capita growth, investment rate, and total factor productivity growth. Notwithstanding the rigorous statistical analysis they conducted to reveal the relationship, their study was more significant in that it brought the issue of simultaneity up to the center stage. It was pointed out that the use of lagged values of financial development indicators, as King and Levine did in their study, does not resolve the simultaneity bias if the agents behaved in forward-looking manner, which would make their results hard to interpret. By contrast, De Gregorio and Guidotti (1995) argued that the use of lagged levels of financial development indicators is justified in cross-sectional studies by noting that "the theories suggest that economic growth induces growth in the financial system but this has no implications regarding the size of the financial system with respect to GDP." Further, their Barro-type cross-sectional estimation found that the positive effects of financial development on growth vary over time periods, regions, and income levels, which suggested that the relationship might be nonlinear.

9

As the issues of simultaneity and nonlinearity persisted, it was suggested that a time-series study focusing on a small group of countries would prove to be beneficial. The results from the early attempts in this direction cast doubt on the validity of the finance-led growth hypothesis. For example, Demetriades and Hussein (1996), who were among the first researchers to employ time-series approach, found, using cointegration tests, that the relationship between finance and growth is bidirectional and that this relationship is country-specific. Similar results were obtained by Arestis and Demetriades (1997), Luintel and Khan (1999), and Shan et al (2001) using VAR estimation. However, more recent studies have reestablished finance as an important source of economic growth. To cite a few, Xu (2000) found evidence for the finance-led growth hypothesis using multivariate VAR, directly contradicting the findings of the previous time-series studies. Calderon and Lie (2003) agree with Xu, after conducting Geweke decomposition test on pooled data of 109 countries, and conclude that finance generally leads growth albeit some evidence of bidirectional Granger-causality. More recently, Christopoulos and Tsionas (2004) applied panel unit root and cointegration tests, threshold cointegration test, and panel VECM to find support for unidirectional causality from finance to growth. Another strand of research that has been pursued is the use of panel studies. Compared to the approaches mentioned above, panel study was generally regarded as more advantageous (Temple, 1999). Using dynamic GMM to control for simultaneity and unobserved country-specific effects, Beck et al (2000) found that financial development promotes growth by improving productivity. Rioja and Valev (2004) adapt a similar method to investigate whether financial development affects growth differently according to the income levels. While they concede that developed countries’ 10

growth is enhanced by finance-stimulated productivity growth as in Beck et al, they argue that the effects of financial development on growth of developing countries are via capital accumulation. Overall, the current trend in this area can be summarized as the following. First, the consensus on the bidirectional causality seems to be gaining an increasing support. This may be a natural course of work since the financial sector is, after all, a part of an economic system. Second, there is an increasing emphasis on the need to employ a time-series approach when considering the relationship between financial development and growth. Third, along with the second trend, researchers are increasingly putting an effort to incorporate infrastructural environments into the analysis. Fourth, the importance of incorporating microeconomic mechanism in modelling the behavior of the financial markets is gaining acceptance among researchers in this area. Given that there has been an increased emphasis on the microeconomic environment in macroeconomic topics, this seems rather late. At least, the research in this area seems to be going in the right direction. Finally, since 1996, there has been more attention on the channels of influence from financial development to economic growth

11

CHAPTER 3

A Finance-led Growth Hypothesis: Revisited 3.1 Introduction

Ever since Schumpeter highlighted a potentially growth-enhancing role of banks as efficient allocators of funds in 1911, the relationship between financial sector and real sector has been a subject of heated debates. In his argument, banks help an economy achieve first-best outcome by providing efficient markets for funds. In contrast to this argument, there are also a group of economists who view the financial sector as something that merely mirrors the real sector. Most notably, Robinson (1952) argued that "...finance does not lead growth. Growth leads finance..." The main argument of those who oppose finance-led growth theories is that, simply put, financial markets evolve in response to increased demands for better services from a growing economy. Therefore, the development of financial markets only mirrors that of real sectors. This led Lucas to state that economists badly over-stress the role of financial sectors. These two polar positions on the role of financial sectors have led economists to consider the issue of causality extensively. The approach I take to investigate the link between the financial sector and economic growth focuses on the role of the financial sector in financing innovative activities. Our motivation is based on the research by Easterly and Levine (2001) in that 12

it is the residual that accounts for most of the income and growth differences across nations. Further, the idea that innovative activities need financial assistance is quite intuitive. We can easily imagine what might have happened to IT companies if there had been no venture capital. It clearly illustrates the possibility of positive effects the financial sector may have on the real sector activities. Broadly speaking, innovative activities require both human and financial capital. They act as complements in the production function of ideas. However, the interaction between the two has been largely ignored in the literature. An exception is the study by Outreville (1999) where he shows empirically that there is a positive relationship between human capital and financial development, although no formal theory was offered to explain the finding. In order to improve our understanding of how the financial sector interacts with the real sector, the nature of interactions between two major components of innovative activities needs to be examined more closely. In this chapter, I pursue this goal by constructing a model where the economy is driven by innovative activities that require both human capital and external funding from the financial sector. Similar to King and Levine (1993), it is assumed that the role of the financial sector in this model is to screen the innovators for their probabilities of success. However, unlike King and Levine’s version of the financial sector that plays an active screening role of weeding out bad entrepreneurs, the financial sector I consider is a more passive one. Specifically, it is assumed that the role of the financial sector is to assess and convey the probability of success to the innovators so that the latter can make optimal borrowing decisions. It is passive in a sense that it does not refuse to lend to those innovators with low probability of success. Rather, it penalizes them by imposing higher repayment requirements. Therefore, contrary to King and 13

Levine, the financial sector does not improve the aggregate probability of success by screening. In formalizing the model, I depart from the conventional literature in four important ways. Firstly, I define innovation as a success not when it is realized but when it is commercially successful. This distinction is motivated by observations that not all innovations are implemented in the production processes. For example, we know from history that the steam engine was invented in Ancient Greece two-thousand years ago. However, it obviously did not ignite the industrial revolution at that time as it did in England millenniums later. A similar example can be found in the United States at the turn of the 20th century. Around the time when mass production of automobiles was about to be started, an automobile that could run on electricity was invented. It drove quietly, emitted less pollution and was technologically superior to gasoline-driven automobiles. However, we all know how the electric-automobile industry fared in the end. Although it was technologically superior to other competitors, it did not survive the forces of the market because it lacked commercial appeals. Gasoline-driven cars were simply much faster and more powerful than electric ones, and hence were more attractive to automobile buyers. In a sense that the aim of innovation is to improve the welfare, in terms of higher incomes or more convenience, of economic agents, a steam engine in Ancient Greece or electric cars in the early 20th century can hardly be regarded as a success. This is especially true if one regards the purpose of innovation to be improvements of production technology. To put it rather bluntly, a technology that is not used commercially is the same as a technology that is not invented as far as its effects on output production is concerned. Accordingly, in the model considered here, it is assumed that innovators invent new technologies 14

with certainty. What is uncertain is whether the new technology would be successful commercially. Then, the job of assessing the probability of commercial success is left to the financial sector. Secondly, I assume that the magnitude of technological change an innovator comes up with is a function of that innovator’s human capital. The idea is simply that the smarter the innovator is, the bigger the improvements over the existing technology will be. It should be noted in advance that although the magnitude of technological change is endogenized, the resulting implication in terms of economic growth remains similar to the existing literature. Thirdly, it is assumed in this model that external finance is needed not for R&D activities but for utilization of innovations. Venture capital markets justify this assumption. Finally, it is assumed that the financial sector pays no setup costs. The assumption of fixed setup costs has been used in the literature to explain poverty traps and bidirectional causality between finance and growth. However, historical experiences of developed countries do not support this assumption (Galetovic, 1996). My analysis shows that one does not need the assumption of setup costs to establish bi-causality or to explain poverty traps. My analysis shows that when certain conditions are satisfied, there exists a unique equilibrium where the growth rate of the economy is jointly determined by the levels of human capital and financial development. An interesting implication of this is that financial liberalization policies that do not adequately address the fundamentals of the economy can bring about bank failures and possibly a financial crisis. Furthermore, in addition to showing that poverty traps can be explained without introducing setup costs, the model suggests that, depending on the parameter values of the economy,

15

there may be two forms of poverty traps, one with a small number of bankers and the other with a large number of bankers. The rest of the chapter is organized as follows. In section 2, a formal model preceded by a sketch is presented. In section 3, the model’s implications are discussed. In section 4, I conclude.

3.2 3.2.1

Model Environment

Imagine a small island populated by groups of workers and bankers who are riskneutral. While bankers move freely to and from other islands, workers are assumed not to be mobile. People on this island live infinitely, but their planning horizon is on a daily basis. In particular, there are L number of identical workers with a zero population growth rate. The worker’s daily routine consists of three events in the morning, afternoon, and evening. At every morning, he is endowed with one unit of manna for the day. Notice that the manna depreciates completely in one day so that there is no saving for tomorrow. The worker can use the manna for two things. He can use some fraction of it to accumulate human capital which will enable him to become an innovator in the afternoon. With what remains, he produces the consumption good using the old technology. After obtaining education, the worker innovates over the existing technology. At this time, the worker/innovator does not know the likelihood of commercial success of his invention. Once he comes up with an idea as to how to improve the old technology, he goes to the banker to finance his production of consumption goods. Notice that the magnitude of improvement depends on the fraction of the manna he spent to receive education. That is to say,

16

the smarter the worker/innovator is, the bigger the improvement will be. When he meets with the banker, the probability of commercial success is revealed through the screening of the banker. Then the worker/innovator borrows from the banker to finance his production activity. In return he promises to pay a certain fraction of the consumption goods he expects to produce in the afternoon as a repayment to the banker. In the afternoon, the state of his innovation’s commercial success is revealed. If it is a success, the worker uses the new technology to produce consumption goods of which some fraction is paid back to the banker as a repayment. If it is a failure, the worker returns what was borrowed, instead of the agreed amount of repayment, to the banker. In the evening, the innovator/worker consumes what is left, and goes to sleep to start a new day tomorrow. There is an arbitrarily large number of identical bankers on this island. Since they can move freely across different islands unlike the workers, exactly how many bankers are on the island at any particular time is not relevant. What is relevant in this economy is how many bankers decide to finance the innovators. In the morning, the banker is endowed with the capital good. He is also endowed with a technology that can transform his endowment to the consumption good at the rate of r, and this process is assumed to take a day. With the endowment, the banker must figure out whether he should open a bank or use the endowed technology to transform it to consumption goods for his own use in the evening. Once he decides to open a bank, he does so without paying setup costs. Then, he waits for the innovators to come and ask for loans. When the innovator comes in for a loan to implement his innovation, the banker screens him to gauge the likelihood of the commercial success of the innovator’s idea. Based upon the probability of commercial success, which is 17

different across the innovators, the banker decides on the amount of repayment, in terms of the consumption goods that he must receive for lending a given amount of capital to the innovator. In the afternoon, he receives the proceeds from investing in the worker/innovator. In the evening, he consumes what he received in the afternoon. In the next morning, the whole cycle starts anew. With this description in mind, we move to formalize the model in the next section.

3.2.2
Workers

A formal model

Initially, the island is endowed with a certain level of technology denoted by A0 at time 0. In each day t, there exists L number of workers who live infinitely. It is assumed that there is no population growth and that L is some arbitrarily large number less than infinity. As mentioned above, the worker i’s daily routine consists of three events. In the morning, he is endowed with one unit of manna. He can use the manna for two activities: innovation and/or production with an old technology. His problem is to figure out how to allocate the manna between these two activities. If he decides to innovate, he spends a fraction vi of the manna to receive education that will enable him to become an innovator in the afternoon. The interpretation of vi is that it measures the amount of sacrifice the worker needs to make to make an innovation as it could have been used to produce the consumption goods with the old technology. Notice that the education process is assumed to be instantaneous for simplicity.

18

The worker i improves upon the old technology that is available for everyone based upon the following: Ait = γ it At−1 . (3.1)

where γ it = (δvit )θ . The parameter δ stands for the effectiveness of education process and is greater than zero. θ is an idea production parameter between 0 and 1. This specification of γ implies that the magnitude of the technological change is an increasing function of human capital acquired through education. In essence, it is similar to Aghion and Howitt’s model (1999) where the economy is driven by innovation and creative destruction. In their model, the magnitude of technological change is a function of some exogenous parameter (their γ) and the number of people working in the research sector. The difference is that I model γ to be solely a function of human capital and leave out the exogenous component. In terms of model implications, nothing is lost by the explicit modelling of γ in this fashion. An attractive feature, on the other hand, of specifying γ this way is that it allows for the possibility of diminishing marginal returns. R&D literature shows that as research efforts increase, captured implicitly by v here, the number of innovations that results grows at a decreasing rate. Aghion and Howitt’s specification does not coincide with this empirical finding. Admittedly, the specification of γ in my model is based on observations only and rather ad-hoc. However, I argue that specifying γ explicitly rather than taking it as exogenous is a step in the right direction toward understanding the impacts that innovations have on the economy. Recall that the fact the worker comes up with a new technology does not mean that the innovation is successful commercially. As discussed in the previous section,

19

whether the innovation is a success is judged by its commercial usefulness. After coming up with a new technology, the worker/innovator is randomly and uniformly put into a location around the circumference of the island. The circumference of the island is assumed to be equal to 1. In this model, I posit that the probability of commercial success, p, is a function of the distance between the worker/innovator and the banker who finances him, and that this distance is not known to the worker/innovator until he is screened by the banker. This idea is motivated by observations from the venture capital market. In the case of venture capital, it is accepted that the likelihood of success for a new idea, business plan, and/or product is critically dependent upon how close relationship the innovator has with those who provide funds. In this model, the nature of the relationship between the borrower and the lender is captured by the distance between the two. An exact specification of the probability of commercial success will be given in the next section. Afterwards, the worker/innovator goes to the banker to borrow capital to finance his production activity using the new technology, after which the probability of commercial success is revealed through the screening of the banker. Given the nowrevealed probability of the commercial success for his innovation, the worker/innovator decides the optimal amount of capital, xit , to borrow and promises to repay Iit to the banker if the innovation is commercially successful. Note that the capital is borrowed after the innovation occurred as described in the previous section. In the afternoon, the probability of the innovation’s commercial success is realized to everyone. If the innovation is commercially successful, the worker uses the new

20

technology, Ait , and the capital, xit , borrowed from the banker to produce consumption goods according to a production function:
H yit = Ait xα = (δvit )θ At−1 xα it it

(3.2)

where α is between 0 and 1. H stands for the production sector that employs a new technology. Note that after one day, the new technology becomes available to everyone. If the innovation is unsuccessful, then the worker/innovator returns the borrowed capital/consumption good, x, instead of I, to the banker. At the same time, he spends 1 − vit to produce consumption goods with the old technology according to the production function:
L yit = At−1 (1 − vit )

(3.3)

where L designates the production sector with the old technology. In the evening, the day is complete with the worker consuming what is left. Bankers In this model, the bankers are assumed to be symmetrical and to maximize profits. Similar to the workers, each banker’s daily routine consists of three events. In the morning, the representative banker is endowed with arbitrarily large units of capital/consumption goods less than infinity.1 It should be pointed out that the exact amounts of endowments do not need to be specified for the same reason that the number of the bankers present on the island need not be specified. It is implicitly assumed in this model that the bankers can potentially finance as many workers as possible as long as it is profitable. Considering that in the real world, the amount
1

This endowment will be designated as a capital good from now on for the sake of convenience.

21

of funds that can be potentially available to the innovator is significantly large especially when there is a relatively free flow of funds across countries, I believe that this assumption is reasonable. In addition, making this assumption allows us to abstain from having to deal with the effects of inter-island transfer of capital/consumption goods. In the morning, he has to decide what to do with the endowment. He can use it to invest in the worker/innovator or have it transformed at a rate of r. To use an analogy, his decision can be described as having to choose between risk-free bonds that pay an interest of r and risky bonds with a higher rate of return than r. Notice that opening up a bank does not cost the banker anything. Once he opens the bank, he meets and screens the worker/innovator to assess the probability of commercial success of the innovation which is defined as: p(zit ) = e−zit , 0 < zit < dt (3.4)

where zit is the distance between the banker and the worker/innovator i at time t.dt is the maximum distance the worker/innovator can be located from the banker. The screening is assumed to be costly. Intuitively, the further away the worker/innovator is from the banker, the more costly it would be for the banker to assess the probability of commercial success. Hence, the screening costs would be a function of z. Another way to think about it is to imagine that, in practice, it would cost more for the banker to find out the likelihood of success if the borrower is of a somewhat dubious nature, which z represents. To capture this, I define the screening cost to be:

SC(zit ) = βezit , 22

0 < zit < dt

(3.5)

where β is a exogenous policy variable that measures the stringency of the government policies regarding the screening processes. A higher level of β implies that more strict policies are in place, and hence the banker incurs higher costs for screening. Figure (3.1) illustrates the relationship between two neighboring banks in terms of the screening costs. It can be seen that for those innovators whose distances z are between −dt and dt , the banker A has a monopoly power over them since it has a cost advantage over all other banks and in particular over its adjacent bankers B and B’. It can be seen that since the circumference of the island is equal to 1, and the bankers will open their banks along the circumference, bankers open for business at time t. Once the screening is done, the banker informs the worker/innovator of the probability of commercial success so that the latter can make a optimal decision on how much to borrow. Then the banker meets the worker/innovator’s demand by supplying him with xit . In return, it is promised to the banker that he would receive I from the worker/innovator if the innovation is commercially successful, and xit if it is not. In the afternoon, the probability of success is realized and the banker receives either I or xit depending on the state of the worker/innovator. In the evening, he consumes what he has received from the worker/innovator along with the rest of the endowments. Equilibrium In order to solve the model, we start with the worker. Given the setup above, the problem the worker i faces at time t is to choose vit to maximize his consumption in
1 2dt

= Nt , where Nt is the number of

23

the evening. Formally, in the morning, he maximizes pit (Ait xα − Iit ) − (1 − pit )xit + At−1 (1 − vit ) it by choosing v. Since zit is not yet known at the time of decision, the problem above becomes in effect E{pit (Ait xα − Iit ) − (1 − pit )xit } + At−1 (1 − vit ). it (3.6)

Before solving the worker’s problem, it will be convenient to specify I at this point. When lending xit , the banker charges Iit to hedge against the potential loss in case the innovation is a failure so that (1 + r)xit = pit Iit + (1 − pit )xit , where r is the world interest rate and exogenous. Solving Eq. (3.7) gives Iit as: ¶ µ r xit . Iit = 1 + pit (3.8) (3.7)

One can see from Eq. (3.8) that the banker needs to be compensated for the risk he takes by financing the worker/innovator. As long as pit is less than one, the return he gets from investing in the worker/innovator is greater than what he would get from the endowed technology. If pit is equal to 1, then there is no uncertainty in this world, and therefore the banker does not need to be compensated for investing in the worker/innovator, which is as safe as the endowed transformation technology. Also, note that those innovators with low probabilities of success are imposed higher repayment requirements. In this world, the bankers do not weed out the bad innovators as in King and Levine (1993). Instead, they impose higher penalties and let the borrowers make the optimal decision. 24

Going back to the worker’s problem, When z is revealed in the afternoon, the probability of success is now known to the worker/innovator. Then, the worker’s problem after z is revealed is pit (Ait xα − Iit ) − (1 − pit )xit . it Differentiating Eq. (3.9) with respect to xit gives: xit = µ αAit pit 1+r ¶ . (3.10) (3.9)

Using Eq. (3.10) and Eq. (3.6), the worker’s problem before z is revealed can be described as: o n 1 max CE (Ait pit ) 1−α
v

(3.11)

where C ≡

around the circumference of the island, E(pit
1 1−α

¡

α Rα

1 ¢ 1−α ¡ 1−α ¢

α

. Since the innovators are randomly and uniformly located

1 )= dt

Z

dt

e 1−α dzit .

−zit

(3.12)

0

Noting that Ait = (δvit )θ At−1 , Eq. (3.11) becomes: max(δvit )
v
θ 1−α

At−1

1 1−α

µ

¶ −dt 1−α (1 − e 1−α )C, dt

(3.13)

where C is as defined above. Differentiating Eq. (3.13) with respect to v gives:
∗ vit = D

Ã

1 − e 1−α dt

−dt

α−1 ! α+θ−1

(3.14)

mal fraction of manna to be spent on receiving education depends on d. Note that by
∗ ∗ symmetry, vit = vjt for i 6= j.

ª 1 © where D ≡ δ −θ A−α (1 − α)α−1 α−1 R−α(α−1) α+θ−1 . Eq. (3.14) shows that the optit−1

25

Next, we examine the behavior of the banker. Given the setup from the previous section, the banker will decide to enter and finance the worker/innovator only if the expected average profit from financing is greater than expected average screening costs. Hence, the entry condition for the representative banker is E {pit Iit + (1 − pit )xit } = E(βez ). It follows then that the decision rule for the banker is given by: vit = where R = 1 + r. On this island, the equilibrium values of vit and Nt (=
1 ) 2dt

(3.15)

µ

β 1−α

¶ 1−α µ θ

Rα αδθ A

1 ¶θ (

edt − 1

1 − e 1−α

−dt

) 1−α θ

(3.16)

are jointly determined

by the decisions of the worker/innovator and the banker, represented by Eq. (3.14) and Eq. (3.16) respectively. Differentiating Eq. (3.14) with respect to d, we find that how v responds to changes in d depends on the values of two parameters, α and θ. Specifically, if α + θ 6 1, the optimal level of education, v, rises as more bankers enter since ∂v/∂d 6 0. In other words, the workers would devote a larger fraction of the manna to obtain education if there is a larger number of bankers in the economy. If α + θ > 1, the opposite situation occurs. By contrast, Eq. (3.16) shows that the banker’s optimal level of d rises as the workers increase their efforts to receive education. In the steady state, vit = v ∗ and dt = d∗ (or Nt = N ∗ ) by symmetry. These relations imply that pit = pi and xit = xi because zit = zi . Then, since the average output per worker that’s produced using a new technology is At E(pi xα ), the total i

26

output produced by using new technologies is given by:
H yt = L(δv∗ )θ At−1 E(pi xα ). i

(3.17)

Total output produced using old technologies is given by:
L yt = LAt−1 (1 − v ∗ ).

(3.18)

Then the total aggregate output in equilibrium at time t is given by £ ¤ yt = At−1 L (δv∗ )θ E(pi xα ) + (1 − v ∗ ) . i Finally, the growth rate of the economy at the steady state is: log µ yt+1 yt ¶ = log µ At At−1 ¶ (3.20) (3.19)

= θ log(δv)

Hence, the growth of the economy is driven by the level of efforts the workers exert to receive education (v), which is jointly determined by the bankers and the workers, the idea production parameter, θ, and effectiveness of the educational system, δ.

3.3

Discussion

As noted above, the behavior of the economy in equilibrium is dependent upon the summed values of two parameters, α and θ. Evidence from economic growth literature tells us that the value of α is typically estimated to be roughly between 0.3 and 0.4. As for the value of θ, there is no practical way to measure θ as its estimation essentially requires, among other things, separating out and measuring the technological changes that results from human capital accumulation. However, it is reasonable to argue that since there is no reason to believe that the value of θ,a 27

parameter that governs the degree of diminishing marginal returns to human capital, would be much different from that of α, the sum of α and θ is likely to be less than 1. Accordingly, the model has an implication that for a given value of v, there exists a unique value of d that is optimal, and vice versa for reasonable parameter values (Figure 3.2). The model presented in this chapter has some interesting policy implications. Firstly, the tougher the government toward the financial sector, the better it is for the economy. A rise in the stringency of policy rules, β, shifts graph B in Figure (3.2) to the left, inducing a higher level of human capital accumulation and a faster rate of economic growth. Another implication is that the economy of this island will not benefit from those financial policies that aim to raise the number of banks without paying attention to the fundamentals of the economy. When d is artificially lowered, say, by the government, the value of v that is optimal to the workers becomes greater than that of v optimal to the bankers. Hence, over time, the bankers will begin to suffer losses, and, as a consequence, some bankers have to exit in order for the economy to restore its equilibrium. This is in line with experiences of Latin American and Asian countries. Beginning around the mid-1980s, these countries have implemented financial policies that aimed to encourage competition by inducing more entry, but often failed to address the issue of whether and how the fundamentals of their economies would adapt to such policy changes. One of the results of these policies was, as we know, rather catastrophic. In some instances, the failure of the banking sector was so severe as to cause an economy-wide financial crisis. This model provides an explanation for how financial liberalization policies that ignore the fundamentals of the economy can lead to bank failures, and ultimately financial crisis. 28

Secondly, the model is able to explain the existence of poverty trap without introducing setup costs. Furthermore, it suggests that two kinds of poverty traps may exist. Firstly, deterioration in the effectiveness of the education system induces the workers to receive less education, pushing graph W to the left, while shifting graph B in the same direction. On net, the deterioration of the education system leads to a lower equilibrium level of d. On the other hand, its effect on v is uncertain. However, if the workers are more susceptible to deterioration of the education system than the bankers are, the economy would be stuck in the poverty trap despite the presence of a large number of banks (Figure 3.3). Therefore, it is possible that a large financial sector coexists with slow growth of the economy. Secondly, in contrast to the preceding case, there can be a poverty trap with only a small number of bankers in the economy as well. This happens when the financial regulations become lax. For instance, if the regulations regarding the screening processes become relaxed, perhaps as part of ill-designed liberalization policies, it causes a drop in the equilibrium values of the number of bankers present in the economy and human capital accumulated, leading to a slower growth. Another possibility that has been ignored so far is the case of α+θ > 1. If this were the case, then the bankers and the workers on this island live in a strange world. Since it is not possible to analytically examine this case because of the number of parameters involved, only a qualitative discussion is provided. First of all, depending on the parameter values, indeterminacy could arise. Secondly, if the parameter values are such that the bankers are more inelastic than the workers, a rise in β actually reduces the human capital accumulation and, thus, hinders economic growth. However, it should be reminded again that this scenario is an unlikely depiction of the real world 29

given the commonly accepted idea of what the values of these two parameters ,α and θ, should be.

3.4

Conclusion

The exact nature of the relationship between finance and growth has been the subject of heated debates for many years. Consequently, many models have been proposed to shed light on this issue. One strand of this literature investigates the role of the financial sector in promoting innovative activities. This chapter is part that literature. The approach I take to investigate the link between the financial sector and economic growth focuses on the role of the financial sector in financing innovative activities. It is motivated by increasing empirical evidence that technological advancement is a key to economic growth. To study this link, I start with the proposition that human and financial capital act as complements in the production function of ideas. Similar to King and Levine (1993), it is assumed that the role of the financial sector in this model is to screen the innovators for their probabilities of success. However, unlike King and Levine’s version of the financial sector that plays an active screening role of weeding out bad entrepreneurs, the financial sector I consider plays a passive role of assessing and conveying the probability of success to the innovators so that the latter can make optimal borrowing decisions. In formalizing the model, I depart from the conventional literature in four important ways. Firstly, I define innovation as a success not when it is realized but when it is commercially successful. This distinction is motivated by observations that not all innovations are implemented in the production processes. Secondly, I assume that 30

the magnitude of technological change an innovator comes up with is a function of that innovator’s human capital. Thirdly, it is assumed in this model that external finance is needed not for R&D activities but for utilization of innovations. Finally, it is assumed that the financial sector pays no setup costs. My analysis shows that when certain conditions are satisfied, there exists a unique equilibrium where the growth rate of the economy is jointly determined by the levels of human capital and financial development. An interesting implication of this is that financial liberalization policies that do not adequately address the fundamentals of the economy can bring about bank failures and possibly a financial crisis. Furthermore, in addition to showing that poverty traps can be explained without introducing setup costs, the model suggests that, depending on the parameter values of the economy, there may be two forms of poverty traps, one with a small number of bankers and the other with a large number of bankers.

31

Screening costs for Banker B’ A B’ B

Screening costs for Banker B A

β

β Screening costs for Banker A

β

-2dt Banker B’

-dt

0 Banker A

dt

2dt Banker B

Note) The vertical axis measures the costs of screening. The horizontal axis measures the distance between adjoining bankers. Figure 3.1: Screening costs of competing bankers

32

v (B’)

(B)

Rise in β v' v* (W)

d’

d*

d

Note) The figure above describes the equilibrium of the economy when α + θ < 1 . (W) represents the decision rule of the worker/innovator given by Eq. (3.14). (B) represents that of the banker given by Eq. (3.16). Figure 3.2: Equilibrium when α + θ < 1

33

v (B’)

(B)

v* v' (W)

(W’) d’ d* d

Note) The figure above describes how the economy responds to deterioration of the education system when α + θ < 1 . (W) represents the decision rule of the worker/innovator given by Eq. (3.14). (B) represents that of the banker given by Eq. (3.16). Figure 3.3: Effects of deterioration in education

34

CHAPTER 4

How Does Financial Development Promote Growth?

4.1

Introduction

What makes an economy grow? Much research has been done to identify the determinants of economic growth. Among suggested factors, researchers found that generally investment is one of the most important determinants of economic growth. This is not surprising. As a child needs a constant supply of nourishment to develop properly, so does an economy to realize sustained development and growth. Only that for the economy, nourishment would be in the form of investment in factors of production such as physical and human capitals and technologies. I start from this basic proposition that investment in factors of production is one of the fundamental forces that drive economic growth. Investment in factors of production generally necessitates the use of funds. The problem is that the amount of funds is often limited compared to the number of investment opportunities available and therefore should be allocated to more productive investment opportunities. Then, who is responsible for allocating these funds among various investment opportunities? Any undergraduate student who took a money-and-banking class will tell you with certainty that the answer to the question

35

is the financial sector.2 Indeed, that is what conventional textbooks teach us: one of the basic roles of the financial sector is to allocate funds efficiently. However, the existence of the financial sector by itself does not guarantee that the allocation will be efficient. Financial sectors exist in various forms and differ in their level of sophistication across countries. Then, in the presence of asymmetric information, those financial sectors that are relatively more developed will be more able to efficiently screen out bad investment opportunities. Then, to the extent that investment is an important determinant of economic growth, it follows that the degree of development of the financial sector should matter for economic growth. However, albeit much research, there is much controversy among professional economists regarding the role of the financial sector in promoting economic growth. And, those studies that do show a positive relationship between GDP per capita growth and a financial development indicator have been criticized for not accounting for a possible endogenous relationship in their estimations. Furthermore, it was argued that the existing empirical studies in general do not show how the financial sector does affect growth. The goal of this paper is to tackle these two main issues. The strategy I take is to focus on a potential channel of influence from the financial sector to the real sector. By doing so, I am able to ameliorate the problem of endogeneity as well as to shed some light on the largely ignored question of how the financial sector affects economic growth.
In this paper, terms such as financial intermediation, financial sector, and banking sector will be used interchangeably when there is no confusion.
2

36

4.2 4.2.1

Related Literature Theories of the finance-led growth hypothesis

The origin of the finance-led growth hypothesis can be traced back to Bagehot (1873). Early studies by Goldsmith (1969), McKinnon (1973), and Shaw (1973) constitute a first systematic investigation into the link between the financial sector and the real sector. Although they found a positive relationship between these two sectors, the finance-led growth hypothesis did not receive much attention at the time due to the lack of a formal theoretical foundation to back the empirical findings. Under the exogenous growth regime, the only way the financial sector could affect the growth rate of an economy was via technological innovation, which was not modeled adequately by then-existing growth theories. With the arrival of endogenous growth theories, the finance-led growth hypothesis received renewed attention. In a world governed by endogenous growth theories, the growth rate of an economy can be enhanced not only by an increase in productivity growth but also by either an increase in the efficiency of capital accumulation or an increase in the savings rate. Diamond and Dybvig (1983) and Bencievenga and Smith (1991) observe that the primary role the banking sector plays is the provision of liquidity and argue that, by providing liquidity, the banking sector enables more investments in illiquid/productive assets and thereby enhances the efficiency of capital accumulation and economic growth. Roubini and Sala-i-Martin (1995) study an alternative way the banking sector can enhance the efficiency of capital accumulation where a reduction of agency costs due to financial development allows a larger share of savings to be channeled into investments. However, whether or not financial development positively affects savings rate is not a clear-cut issue. For example, Devereux and 37

Smith (1994) show that a reduction in idiosyncratic risk and the rate of return risk may either reduce or increase savings rates depending on the degree of risk aversion of economic agents. Furthermore, Japelli and Pagano (1994) show that reducing liquidity constraints reduces savings since the younger generation in their model borrow much more in the absence of liquidity constraints. Saint-Paul (1992) builds a model where the financial sector allows more specialization in productive and risky technology by reducing idiosyncratic risks. More to the spirit of ideas-driven economic growth, but not quite, Galetovic (1996) studies the interaction between the financial sector and the research sector. In his model, the financial sector plays the indirect role of economizing the costs of monitoring research firms for investors. Contrary to these models where the financial sector promotes the process of learning by doing, King and Levine (1993) consider the financial intermediaries as actively involved in the production of ideas itself by screening and monitoring innovative projects. It is rather odd to note that the theory of the finance-led growth was revived by the birth of endogenous growth theories, and yet the researchers have paid only scant attention, with an exception of King and Levine (1993), on the effects of financial development on innovative activities in examining the validity of the finance-led growth hypothesis. In the context of both exogenous and endogenous growth theories, the ultimate engine of economic growth is a rise in the productivity level as an outcome of either learning by doing or intentional effort to come up with a better technology. Also, Easterly and Levine (2001) show that the ”residual” rather than factor accumulation accounts for most of the income and growth differences across countries. Further, the importance of purposive technological innovation in enhancing productivity is well illustrated by the amount of effort a large number of countries 38

put into establishing a robust R&D sector. Based on these observations and the fact that investment in R&D requires funds, at least partly, from the financial sector, I argue that if the finance-led growth hypothesis is correct, then we should be able to detect evidence of positive effects of financial development on innovation. Therefore, my aim in this paper is to examine this particular channel of influence. In doing so, I hope to shed further light on the validity of the finance-led growth hypothesis.

4.2.2

Empirical studies

A flood of empirical studies began to appear in the 1990s to test the validity of the finance-led growth hypothesis. Unfortunately, theoretical models of finance-led growth do not provide empirical researchers with structural guidelines on which they can base their estimation. As a result, one is forced to use reduced-form estimation and test the general conclusion of these models. Since the implication of all the theoretical models in this area is basically that a better-developed financial sector enhances economic growth, regardless of the channel of influence, a typical test strategy involves regressing some indicator of financial development on some aggregate growth measures such as investment growth, GDP growth or total factor productivity growth. A first attempt in this direction was made by employing cross-sectional estimation (King and Levine, 1993, and De Gregorio and Guidotti, 1995). Using a cross-sectional framework, they find a positive relationship between financial development and economic growth. However, a potentially endogenous relationship between financial development and economic growth made interpretation of these results difficult. Demetriades and Hussein (1996) and Odedokun (1996) take a Granger causality

39

approach to avoid these problems and present mixed results. They find that the effects of financial development on growth are country-specific. Based on their findings, they argue that a robust test of the finance-led growth hypothesis should incorporate the time dimension of the data under consideration. Benhabib and Spiegel (2000) employ a panel GMM method to reach a similar conclusion. By and large, the current empirical literature lacks one crucial element in that they do not consider the channels of influence suggested by theoretical models and fail to show how financial development affects economic growth. Furthermore, timeseries approach, while potentially resolving endogeneity issue, does not tell us exactly what the relationship between these two is. In addition, their results are as hard to interpret as cross-sectional estimation because Granger causality does not really provide an answer for the causal relationship between financial development and economic growth. Researchers who conduct causality tests in this area argue that, for some countries, economic growth causes financial development, when what they really need to say is that economic growth Granger-causes financial development. This does not really address the question of what causes what, especially when one considers that, statistically, Christmas card sales Granger-cause Christmas. In sum, the current empirical literature suffers from two problems. First, as long as one regresses GDP growth on a measure of financial development, the issue of endogeneity is not satisfactorily resolved. Second, the channel of influence has not been specified so far, thus, limiting our understanding of how the financial sector affects growth. In order to examine the validity of the finance-led growth hypothesis, I depart from the conventional literature. Instead of estimating a relationship between aggregate growth measures and financial development indicators as is commonly done 40

in the current literature, I test the validity of the finance-led growth hypothesis by focusing on the innovation channel of influence, using patent applications as a proxy for innovative output. Under the framework of ideas-driven growth, the hypothesis I test is that financial development enhances innovation, which is the main engine of economic growth. If the finance-led growth hypothesis is right, as the financial sector develops over time in a certain country, the growth rate of innovation should be higher, which would, then, lead to faster economic growth due to a rising level of productivity. Using panel data on twenty eight countries from 1970 to 2000, my analysis shows that financial development is indeed significant in raising the growth rate of innovative output. In section 2, I provide a theoretical motivation for the empirical analysis by extending a standard Romer-type growth model to include agency costs and draw a testable implication. In section 3, I discuss the estimation strategy employed and the issues that need to be addressed. In section 4, I conclude.

4.3

Theoretical background

One of the common assumptions made in finance-led growth theories is that the financial sector (mainly the banking sector) actively monitors borrowers of funds. Allen and Gale (2001) show that evidence is to the contrary. They show that in most cases, the banking sector does not serve as an active monitor. The rationale for this is that often times the banking sector makes a debt contract with the borrowers in which profits of a lending bank are not dependent upon the borrower’s degree of success. Rather, they simply depend on whether the borrower succeeds or not.3 Therefore,
Of course, this is not the case for equity contracts. However, in most cases, equity markets are relatively small in terms of intermediating funds and are ignored in this paper.
3

41

the welfare of the lending bank will depend more on how well it screens out the bad borrowers and less on its effectiveness as a monitor. Based on this observation, I extend a standard Romer-type endogenous growth model and include the financial sector as a provider of funds for researchers with agency costs to illustrate how the degree of financial development affects the rate of technological innovation. Note that agency costs in this model represent the costs of screening for the financial sector. The goal of the model is, thus, to show that high agency costs discourage innovation and growth of the economy.

4.3.1

Final goods sector

A perfectly competitive final goods sector produces a single homogenous consumption good by combining labor and intermediate goods. The production function for the final goods sector is given by Y = ZA
0

xα dj, j

(4.1)

where A is the number of intermediate goods used and x is the amount of intermediate good j used and is between 0 and 1. Given this production function, and normalizing the price of final goods to one, a firm in the final goods sector maximizes RA R its profit; xα dj − 0 pj xj dj, where pj is the price of an intermediate good j. Profit j pj = αxα−1 . j (4.2)

maximization gives the price of an intermediate good j as

4.3.2

Intermediate goods sector

The intermediate goods sector consists of monopolistic firms that buy designs from the research sector to be used in production of intermediate goods. These firms are 42

monopolistic since the designs they buy are protected by patents that exclude others from using the same designs. Therefore, each monopolist produces only one type of intermediate good. With the design in hand, the monopolist produces intermediate goods using a one-to-one production function. In other words, the monopolist requires one unit of capital to produce one unit of intermediate good. Formally, the monopolist maximizes the profit function given by: pj (xj )xj − rxj , (4.3)

where r is the interest rate for borrowing capital. The firm’s supply of xj derived from the profit maximization, together with the demand schedule in Eq.(4.2), determines the price of xj to be equal to r/α, which implies that xj = x and, consequently, that Y = Axα . Using Eq.(4.2) and pj = r/α, we get x = α2 Y /Ar. Then, the profit for each monopolist can be specified as; π I = (p − r)x Y = α(1 − α) . A (4.4)

intermediate goods sector, (1 − aK )K, x is equal to (1 − aK )K/A. Note that (1 − aK )

Further, since the total amount of the intermediate goods used in the final goods RA sector 0 xi di = Ax, should be equal to the total amount of capital spent in the is the portion of capital stock used in the intermediate goods sector, and K is the total stock of capital in the economy. Finally, the production function turns out to be Y = A1−α [(1 − aK )K]α . 43 (4.5)

4.3.3

The research sector

Recall that my primary goal in this section is to provide a theoretical sketch of how financial development affects innovation. I aim to show here that the share of capital spent in research sector is increasing in the degree of financial development as proxied by lower agency costs. In this model, each researcher faces a similar problem as the monopolist in the intermediate goods sector. In other words, each researcher borrows capital from the financial sector to finance her innovation. What is different from the monopolist’s case is that the researcher’s cost of borrowing capital is not r but r + c where c is the exogenous agency costs. The researcher pays an additional cost of c because the financial sector has to screen the researchers when they borrow funds. With this environment, the researcher tries to maximize her profit based on her production function. I make a standard assumption that when the researcher innovates, she takes the actions of other researchers and the knowledge stock as given so that she faces the arrival rate of δ defined as: δ = A1−β [aK K]β−1 , (4.6)

where δ stands for the arrival rate of new technology per unit of capital spent on innovation at the individual researcher’s level. When a new technology is developed, assuming that the design lasts forever, the researcher receives a price, pA which is equal to the monopolist’s profit discounted by r, (1 − α)x/α. Since x = α2 Y /Ar, pA = α(1 − α)Y . rA 44 (4.7)

Given the price pA and the arrival rate δ, the marginal product of capital spent in the research sector equals simply pA δ. Equating this to the marginal cost of capital, r + c, and noting that r = sector as: µ " µ ¶ # ¶1−α ¶ µ ¶β 1−α µ A 1 K 1−α (1 − aK ) = a1−β α2 +c . K α A K 1 − aK (4.8)
α2 Y , (1−aK )K

I get the share of capital used in the research

Although the Eq. (4.8) cannot be solved explicitly for aK , it can be seen that there exists a unique value of aK by noting that the LHS of the equation is decreasing in aK and that the RHS of the equation is increasing in aK for the entire range of aK .

4.3.4

The growth of the economy

Given the production function (4.5), I need to specify how the economy evolves over time. First, technological innovation occurs by the following law of motion:
1−β At = At [aK Kt ]β , ·

(4.9)

where β is between 0 and 1. This specification assumes that technological innovation is a function of capital devoted in the research sector. However, the production of new technologies faces diminishing returns in K. Furthermore, it assumes that the stock of knowledge in the economy also contributes to production of new technologies. Following Solow (1956), I assume that the capital stock evolves by K t = sYt ,
·

(4.10)

where s is a constant investment ratio. Also, assume, for simplicity, that there is no population growth so that n = 0. Substituting (4.5) into (4.10) and divide both sides 45

by Kt , the growth rate of capital is given by: K = s(1 − aK )α K
·

µ

At Kt

¶1−α

≡ gK .

(4.11)

Similarly, the growth rate of technology is given by: A = aβ K A
·

µ

Kt At

¶β

≡ gA .

(4.12)

Along the balanced growth path, the usual steady state condition applies so that gK = gA = g ≡ the growth rate of the economy. Combining Eq. (4.11) and Eq. (4.12), the steady state growth rate of the economy, dropping time index, is given by: g = [s(1 − aK )α a1−α ] 1−α+β , K and g ≥ 0 if aK ≤ 1 − α. Note that the steady state growth rate of the economy is increasing in aK as long as the share of the capital spent in the research sector is less than or equal to 1 — α. Further, it can be seen from Eq. (4.8) that the equilibrium amount of capital spent in the research sector is decreasing in the agency cost, c. It is shown in this model that the steady state growth rate of the economy is increasing in aK , which is itself a decreasing function of the agency cost. The testable implication is that a poorly developed financial sector impedes economic growth by reducing the amount of capital devoted to technological innovation. Also, note that that the share of income used for investment in capital accumulation, s, was assumed to be constant in this model. Hence, this model illustrates that for a given a level of investment, a smaller share of that investment will be devoted to R&D activities resulting in a lower rate of technological innovation if the financial sector is relatively less developed. 46
0 β

(4.13)

4.4

Empirical analysis

In the previous section, I presented a simple model in which high agency costs discourage technological innovation, which is a main engine of economic growth. In this section, I empirically examine whether financial development does have any significant effect on the rate of technological innovation as the model suggests.

4.4.1

Patents as a proxy for technological innovation

The first question that needs to be addressed in order to conduct the investigation is: how do you measure the underlying technological change in a given country? Variables such as R&D expenditure, the number of scientists and engineers, or the number of articles published in scientific journals have been proposed as measures of technological innovations. The problem with these variables is that they are not widely available except for a few developed countries. This limits the number of countries a researcher can include in her sample to a selective few. An alternative and less direct measure of technological innovation that has been popular among researchers is Total Factor Productivity. By assuming a cobb-douglas type aggregate production function, one can easily estimate this for a number of countries over a long period of time. However, this residual measure is only distantly related to technological innovation (Griliches, 1990). In other words, it simply is something that economists do not have full understanding of how changes in total factor productivity are brought about. This ambiguous nature of total factor productivity (TFP) as a measure of technological innovation, maybe harmless in some contexts, can prove to be not so harmless in other cases. When the relationship under consideration is the

47

one between development of the private financial sector and total factor productivity, we first need to have a well-defined idea about how the former affects the latter. The theoretical literature is not clear on this. What the literature suggests is that financial development encourages more efficient allocation of investments among innovation projects. As innovative activities are enhanced due to development in the financial sector, productivity, and ultimately the economy’s growth rate, rises. What is clearly defined in the theoretical literature is how financial development affects innovative activities. It does not tell us how financial development affects productivity as measured by total factor productivity. In this study, I use patent data as a proxy for technological innovation. A patent is generally defined as a document issued by an authorized governmental agency, granting the right to exclude anyone else from the production or the use of a specific new device, apparatus, or process for a stated number of years. Using patent as a proxy for technological innovation to examine the effects of financial development on innovative activities has a number of advantages. Firstly, patent data is the most direct measure of innovative output. Contrary to other proxies of technological innovation mentioned above, "a patent does represent a minimal quantum of invention that has passed both the scrutiny of the patent office as to its novelty and the test of the investment of effort and resources by the inventor and his organization into the development of this product or idea, indicating thereby the presence of non-negligible expectation as its ultimate utility and marketability." (Griliches, 90) Also, "patents are a direct outcome of the inventive process, and more specifically of those inventions, which are expected to have a commercial impact... a particularly appropriate indicator for capturing the propriety and competitive dimension of technological change." (Archibugi 48

and Pianta, 96) An inventor will apply for a patent right only if the benefits she expects to receive from it outweigh the costs of obtaining patent protection. By its nature, patent data alludes us to qualities of innovations that are produced. Stern et al (2000) argue that patents, by its very nature, reflect an important portion of the innovative output by a country and are the most concrete and comparable measure of innovative output across countries and time. Secondly, as mentioned above, patent data bears the closest resemblance to innovative output as described in the theoretical literature in this area. Furthermore, the relationship between financial development and innovative output is well-defined. It is also intuitively appealing to argue that patent production requires investments that are intermediated by the financial sector. Thirdly, and related to the second argument above, using patents allows a more concentrated focus on the effects the private financial sector has on technological innovation. Research shows that industries heavily involved in government contract work tend to patent fewer inventions of a given quality than those which pay for their own research (Comanor and Scherer, 1969). This indicates that, compared to other measures of technological innovation, patent data reflects to a lesser degree the influence of government activities that are not associated with financial development. Having argued that patent data is the best measure of technological innovation that serves the conceptual purpose of this study, there still remains a doubt if I may be barking at the wrong tree. An example is so-called Fox paradox. It states that in a given country, patent production may come from a small industry that does not affect the country’s economic growth while a large industry that is actually responsible for the country’s economic growth remains dormant in terms of patent

49

production.4 Therefore, despite its advantages mentioned above, using patent data will not serve the empirical purpose of this study if it is not related to economic activities. Research shows that indeed this is not the case. Comanor and Scherer (1969) argue that a simple count of the number of patents reflects not statistical noise but a meaningful message in the results of studies using patents by showing that the correlation between patents and the value of new product sales is significant. On an aggregate level, using patent data as a measure of technological change, it is shown that a higher intensity of technological activities has a generally positive impact on national growth (Archibugi and Pianta, 1996). Porter and Stern (2000) reach a similar conclusion in their study that there is a positive link between ideas production and realized productivity growth. Further, micro-based studies indicate that patents are actively utilized in production processes. They show that the share of patents actually used by firms range from 40% to 60% (Napolitano and Sirilli, 1990). EPO survey found that the majority of European firms utilized their patents most of the time. Also, it found that 84% of patenting firms cited patents in the case of products and 71% in the case of processes as their usual means of protecting new products and processes (Archibugi and Pianta, 1996). These studies further strengthens the validity of using patent data as a proxy for technological innovation.

4.4.2

Data

The patent data I use is based on the Technology Assessment and Forecast Report, compiled by the United States Patent and Trademark Office (USPTO) and reported to the World Intellectual Property Office on an annual basis. The motivation for using the U.S. patent data is based on the evidence that the U.S. has the lowest granting rate
4

Heterogeneity across countries such as this can be taken care of by using country dummies.

50

in the world (Griliches, 1990). It indicates that the U.S. has the highest standard for granting patent rights and gives us a hint about qualities of inventions for which patent protection is asked. Porter and et al (2000) argue that because USPTO approval requires that patents constitute novel, non-obvious inventions, patenting captures a sense of the degree to which a national economy is developing and commercializing new-to-the-world technologies and that by only including inventions that are granted patent protection in the U.S., we can be confident both that a relatively common standard has been applied and that the counted inventors are, in fact, near the global technological frontier. More importantly, using patent data from one source allows me to avoid the issue of heterogeneity across databases collected by various agencies. When one tries to combine the databases of several agencies, she needs to deal with different classification systems each agency has as well as quality differences in patents granted by different patent agencies. Many attempts have been made to handle these problems with only limited success. When a conceivable advantage of this kind of comprehensive data is basically a larger dataset, the benefit hardly outweighs the costs of having to deal with aforementioned problems. It is especially so if the number of countries covered in the U.S. dataset is reasonably large. USPTO has six categories for patent grants; Utility, Design, Plant, Reissue, DEF, and Statutory invention registration. Among them, what is relevant for this study is utility patents. According to the USPTO definition, these are patents that are issued for the invention of a new and useful process; machine, manufacture, or composition of matter, or a new and useful improvement thereof. Therefore, utility patents have a direct bearing on industrial production processes. Once granted, they provide twenty

51

years of protection. This data contains both developed and developing countries, and the country of origin is based on the residence of the first named inventor. In the model, the degree of financial development is represented by agency costs. When the financial sector is relatively less developed, it imposes higher agency costs to innovators for borrowing funds, leading to a lower rate of technological progress. Empirically, finding a single measure that captures every aspect of financial development is nearly impossible due to complexities involved with functions the financial serves in the economy. Therefore, I use three different indicators proposed in the literature to capture various aspects of financial development. They are • The ratio of private credit by deposit money banks to GDP (henceforth, PC) • The ratio of liquid liabilities to GDP (henceforth, liquidity) • The ratio of deposit money bank assets to GDP (henceforth, DMBA)5 PC measures the activity in channeling savings to investors and is equal to the ratio of claims on the domestic private sector by deposit money banks to GDP. The assumption is that as the financial sector develops, it will be able to channel more funds from savers to investors. De Gregorio and Guidotti (1995) also suggest that PC represents more accurately the role of financial intermediaries in channeling funds to private market participants. Since the main role of the financial sector I emphasize in this section is to serve as an effective intermediary of funds, PC is the variable that I will be primarily interested in. Liquidity is the most commonly used indicator of financial development and usually referred to as "financial depth". It measures the overall size of the financial sector
5

Data for these variables are from Beck et al (99)

52

without distinguishing between the financial sectors or between the uses of liabilities (Beck et al, 99). It is equal to the ratio of currency plus demand and interest bearing liabilities of banks and other financial intermediaries to GDP. However, the use of liquidity as a financial development indicator has come under attack recently. De Gregorio and Guidotti (1995) argue that it is conceivable that a high level of monetization (implied by a high level of liquidity) is a result of the lack of alternative assets that would serve as stores of value. Eastern Europe and the former Soviet Union provide evidence for this scenario. To overcome this problem, one can use a broader measure of monetary aggregates such as M3. However, to the extent that M1 is included in M3, it does not resolve the problem. Moreover, the fact that M3/GDP is also the inverse of the velocity of circulation of the broad money stock suggests that a positive association between the level of financial development, proxied by liquidity, and real GDP is tantamount to a downward trend in the velocity of circulation and may simply reflect an income elasticity of the demand for money with respect to GDP which is greater than unity (Demetriades and Hussein, 96). So, I use liquidity for the sake of completeness and with reservations. Finally, I also use DMBA as another measure of financial development. DMBA is a so-called absolute size measure and reflects the importance of the financial services performed by the banking sector6 .
All the measures used here are the ratios of a stock variable to a flow variable, which creates problems with correct timing and in terms of deflating correctly. Beck et al (99) address these problems and calculate each measure by ³ ´ F Dt−1 F Dt 1 2 CP Ie,t + CP Ie,t−1 , GDPt
CP Ia,t 6

where e: end of period, a: average for the period. The end year of year CPI is either the value for December or, if not available, the value for the last quarter. For additional information, see their paper.

53

4.4.3

Methodology

There are well-known concerns with respect to using patent data for economic analysis. First, there is an issue of heterogeneous quality of patents. Naturally, patents differ greatly in their technical and economic significance over time. Scherer (1966) suggests that the way to get around this issue is to invoke the law of large numbers. The idea is that by the law of large numbers, the economic significance of any sampled patent can be interpreted as a random variable with some probability distribution. Furthermore, the problem of differing qualities does not apply only to patent data. Other measures of technological innovation such as a simple count of scientists and engineers, or R&D expenditures are also prone to this problem. (Comanor and Scherer, 1969). Given this fact, using patent data as a proxy for technological innovation may have more merits than other measures for it allows one to examine a wide range of countries. Secondly, each country has a different propensity to patent arising from differences among countries in terms of their industrial composition. Thirdly, and related to the above, using USPTO patent data may exclude those innovations that are novel to a country but have been already discovered elsewhere, or those innovations that are not worthwhile patenting internationally. In order to take care of these two problems, I follow Eaton and Kortum (1996) and Porter and Stern (2000) and make an assumption similar to the one above. It states that the value of value of innovations is distributed according to a fixed distribution across economies and a constant fraction of innovative output turns out to be valuable enough to justify an international patent. To the extent that this fractional value varies across countries, it is overcome through the

54

use of fixed country specific effects in the regression. The preceding discussion tells me that panel estimation is a proper way to conduct the empirical analysis. Patrick (1966) argues that as the process of real growth occurs, the supply-leading impetus generally becomes less important and the demand-following financial response becomes dominant. Similarly, Fritz (1984), Jung (1986), and Dee (1986) suggest in their studies that developing countries have rather a supply-leading causality pattern of development than a demand-following pattern. What these studies say is that the developing countries should provide a fertile testing ground for finance-led growth hypothesis. If the hypothesis is not valid, a measure of financial development would not enter significantly in estimation. Furthermore, there is no reason to believe that choosing developing countries as a sample would present an upward bias in my estimation. In sum, if finance-led growth hypothesis is valid, the effects of financial development will be strong and significant for developing countries, and if it is invalid, they will be insignificant regardless of the countries chosen. Therefore, I select 28 developing countries from USPTO database to be included in the sample.7 Table (4.1) shows a list of countries that are included in the sample. Figure (4.1) shows the relationship between averages of real GDP growth rates and patent growth rates over 1970 - 2000. It shows that the latter is closely related to the former, in line with the previous discussion on the effects of patents produced on economic growth. In order to construct a panel data, I compute average growth rates of patent applications per million persons for six periods: 1970-74, 75-79, 80-84, 85-89, 90-94, 95-2000 as a proxy for the rate of technological innovation. To measure the degree of financial development, I use the initial levels of financial development indicators for
7

A criterion for selecting developing countries is from IMF.

55

each corresponding period to ameliorate endogeneity.8 There are a couple of handwaiving arguments against this as a solution for endogeneity. Firstly, if economic agents were forward-looking, the use of the initial levels of financial development indicators would not eliminate endogeneity (Rajan and Zingales, 1998). The idea is that, if the economy were expected to grow in the future, forward-looking economic agents would step up lending now hoping to take advantage of a economic boom in the future. Then, the level of financial intermediation today is going to be affected by future states of the economy. Hence, endogeneity still exists even if one uses the initial values of financial development indicators. This argument needs two assumptions satisfied to be valid. One is that, empirically, the dependent variable used in the regression is such that economic agents could observe its behavior, for example, its growth rate, easily enough so that the economy-wide change in lending activity could occur in response to the changes in its behavior. The other is that this change in lending activity is sufficiently big so that it affects the total volume of credit in a non-negligible manner. Aggregate growth measures typically used in this literature such as GDP per capita growth satisfy these assumptions. Thus, the argument by Rajan and Zingales (1998) is true if what the analysis focuses on is the relationship between the financial sector development and aggregate growth measures. However, the focus in this paper is on the relationship between the rate of technological change and the degree of financial development, and their argument does not apply well for two reasons. As for the first assumption mentioned above, the inherent nature of uncertainty associated with R&D activity makes it difficult for
8

For a discussion of simultaneity, see Tsuru (2000)

56

economic agents to observe the changes in the rate of technological innovation. In addition, there is no theoretical background to support that future economic conditions cause changes in the current rate of technological innovation. As a matter of fact, the consensus is that it is technological innovation that determines future economic conditions. Further, even if the first assumption is met, the resulting lending activity is not sufficiently big enough to affect the total volume of credit in a non-negligible manner since research is only a small part of what bank lending finances. Secondly, if financial development indicators are correlated across time, then using the initial values of financial development indicators would not remove endogeneity since they would simply be proxies for their contemporaneous levels (Demetriades and Hussein, 1996). Figure (4.2) shows movements of financial development indicators for sample countries across time. Significant variations in these indicators suggest that the use of initial values of financial development indicators is justified. In summary, the regression equation that I estimate is the following: (Patent growth rate)it = αi + β(Financial development indicator)it + γ(control variables)it + εit where i = 1...28, t = 1...6. If β comes out positive and significant, it will provide support for the finance-led growth hypothesis in a sense that the degree of financial development is positively associated with the rate of technological innovation, the driving engine of economic growth. Control variables (4.14)

57

R&D literature indicates that R&D effort is a significant determinant of technological innovation. The most commonly used measure of R&D effort for the developed countries is the number of researchers in the research sector. However, the data is not available for a long time span for developing countries. Therefore, I instead use the human capital measure compiled by Barro and Lee (2000). Among various measures they compiled, I use the percentage of the population 25 years of age or older who have attained "higher" education (HIATT25) as a proxy for the number of researchers in the research sector assuming that the number of researchers in the research sector is positively correlated with the extent of college education in the population. Although this is a somewhat indirect measure of R&D effort, I believe it to be reasonable to think that much of the innovation stems from college-educated innovators. For my analysis, I use the initial levels of HIATT25 for each corresponding period. This variable, according to the literature, is expected to have a positive effect on the patent growth rate. Therefore, the estimated coefficient should be positive. Intuitively, people alone cannot generally come up with technological breakthroughs if they lack infrastructure supporting their R&D activities. This infrastructure is generally funded by institutions such as government, private businesses and academic institutions. The volume of funds that supports R&D activities from these institutions is conventionally measured by R&D expenditure data. This data is generally widely available for developed countries but not for developing countries. Perforce, I use the real domestic private investment data to proxy for investment in R&D in my analysis. Then, I compute the average levels of domestic private investment in the five years immediately preceding each period in the sample (INVEST). The literature is not clear on what the sign of the estimated coefficient should be. Depending on the 58

nature of returns to scale, the estimated coefficient can be either positive or negative. So, I will let the data speak for itself. Knowledge production may also depend on the past knowledge stock. All the models of endogenous growth incorporate this intertemporal spillover effect in one way or another. I use the initial level of real GDP per capita for each corresponding period (Knowledge stock) to capture this potential effect. This variable also captures the ability of a country to translate its knowledge stock into a realized state of economic development and so yields an aggregate control for a country’s technological sophistication (Porter and Stern, 2000). Also, compared to other measures of the past knowledge stock such as the past patent stock, real GDP per capita provides a more comprehensive measure of the knowledge stock in the economy as the past patent stock may be industry-specific and does not convey information on the economy-wide knowledge stock. According to endogenous growth theories, the sign of the estimated coefficient should be positive. According to neoclassical growth theories, it should be negative. So, it would be interesting to see what the data says. In addition to intertemporal spillover effects, there can also be a cross-country spillover effect. This will be especially true if a country is an active importer of sophisticated technologies as Japan had been in the 50’s and 60’s. Hence, an important way a country can learn from a technologically more developed country and hasten its own knowledge production is by importing technology embedded in goods9 . Based on this observation, I use Openness measured by the average levels of the sum of exports and imports divided by GDP in the five years immediately preceding each period in the sample to capture the potential cross-country spillover of knowledge.
Foreign direct investment might also be a major channel of technology transfer has it not been that a majority of FDI is among developed countries.
9

59

Again, the sign of the estimated coefficient is uncertain due to the fact that what I am considering here is patent applications filed in the U.S., the foreign country. If a catch-up effect is dominant, Openness should have a positive effect on the country’s knowledge production. On the other hand, if a raising-the-bar effect is dominant, it should have a negative effect. Wurgler (2000) suggests that the effectiveness of the financial sector also depends on the magnitude of government intervention. He argues that with a bigger government, there is an increased incentive to overinvest in declining industries rather than improving the supply of finance to growing industries. The effects of government intervention is proxied by the average levels of the share of total government consumption in GDP for each corresponding period in the sample (Government size). I expect it to have a negative coefficient. There is increasing evidence in the literature that legal structure such as intellectual property protection plays a non-negligible role in knowledge production processes (La Porta et al, 1996). This may be especially true for patent production. I use the initial values of a legal index compiled by Gwartney et al (2002) to reflect the degree of IP protection within each country (Legal). This is a composite index of judicial prudence, impartial courts, protection of intellectual property, military interference in rule of law and the political process and integrity of the legal system. It provides a more comprehensive coverage of the legal system than typical indexes used in the literature, which represents only a particular aspect of the legal framework of a given country. The literature predicts that it should have positive effects on knowledge production.

60

Finally, I include the average levels of inflation using the GDP deflator (PI) for each corresponding sample period to reflect the economic conditions of a country at the time when a patent application is filed. Generally, an inventor will be less likely to utilize her invention during times of economic turmoil. Hence, I expect the estimated coefficient to carry a negative sign to the extent that inflation signals economic turmoil.10

4.4.4

Estimation

Table (4.2) shows the results of estimation. PC and DMBA come out positive and significant, suggesting that financial development does have a positive effect on the rate of technological change. On the contrary, liquidity comes out insignificant. It may be due to the possibility that liquidity is a poor measure of financial development as discussed in the previous section. Human capital (HIATT25) is generally positive and significant as expected. Note that it has a bigger effect on knowledge production than financial development. Openness comes out with a negative sign and is insignificant, which suggests that the raising-the-bar effect is dominant. This is in line with Porter and Stern (2000)’s finding that a cross-country spillover is weakly negative. Investment is negative and insignificant. This suggests that, although a higher investment may produce more technological innovations, it does not necessarily yield a faster rate of technological change. A possible interpretation, then, would be that there is decreasing returns to investment in terms of the rate of knowledge production. Knowledge stock is also negative and insignificant suggesting convergence in terms of knowledge production
Data for the variables are obtained from World Development Indicators, CD-ROM, 2000 unless noted otherwise.
10

61

among these countries. Government size is generally negative as expected. Inflation (PI) comes out significant and weakly positive, which is in line with the R&D literature. Table (4.3) shows the results of estimation when legal structure (Legal) is included in the regression. Note that due to data availability, I lose eleven observations here. Overall, the results are similar to what I have seen in the preceding regression. Financial development indicators generally come out positive and significant to confirm the conclusion we reached above. What is interesting is that Legal comes out negative and significant. One plausible interpretation would be that a weak legal system of a home country induces firms/inventors to seek patent protection from abroad (the U.S.). In other words, if a home country provides adequate legal protection for innovations, firms/inventors will be less motivated to apply for patent protection elsewhere. If this were true, then the frequency with which a foreign firm/individual applies for a patent protection in the U.S. may be negatively correlated with strength of the legal system in the home country. Hence, the negative sign I get here may simply be an artifact of using USPTO data instead of home country patent data. Table (4.4) shows the results of estimation using only those variables that turned out to be significant in the previous regressions. The fit is better although the overall results are similar to the previous regressions. Overall, my analysis provides support for the finance-led growth hypothesis11 . More to the point, it provides an answer as to how the financial sector promotes
I also ran the regression with distance dummies as the sample contains a large number of Latin American countries, which could introduce a geographic bias into the analysis. Including distance dummies did not change the results. Moreover, these dummies were insignificant. These results are available upon request.
11

62

economic growth. The answer I get from this analysis is that financial development enhances economic growth through, at least partially, promoting innovative activities.

4.5

Summary

This paper investigates the relationship between the financial sector and economic growth by focusing on a specific channel of influence, innovation. The basic idea of this paper starts from the proposition that investment is an important determinant of economic growth. Then, the degree of development of the financial sector that provides fuel to it should also matter for economic growth. This is the essence of the finance-led growth hypothesis. In order to examine this issue more carefully, I conduct an empirical investigation of how the financial sector development could affect economic growth focusing on the channel of influence the theoretical literature put forward. More specifically, I ask if the financial development is positively associated with the rate of technological innovation. If development of the financial sector does have growth-promoting effects as the finance-led growth literature contends, the answer should be yes. With this idea in mind, I first provide a simple theoretical model where the rate of economic growth is partly governed by the share of capital used in R&D sector, which in turn depends on the degree of financial development proxied by agency costs. The model indicates that the more developed the financial sector is, the faster the rate of technological innovation will be. Using patent growth rates as a proxy for the rate of technological innovation, I find that financial development positively affects the rate of technological innovation rendering support for the finance-led growth hypothesis.

63

My approach has two advantages over the current studies. Firstly, the channel of influence from the financial sector to the real sector is explicitly specified, which has been missing in the current literature. Further, by identifying the channel of influence from the financial sector to the real sector, this paper is able to produce an interesting policy implication. That is, when a country liberalizes its financial sector making it more market-oriented12 , the potential benefit could be in the form of a higher rate of technological change. However, this benefit may not be fully realized if the country is not equipped with a viable R&D sector, which is the case for most developing countries. Secondly, my approach resolves the issue of endogeneity, at least better than the existing literature, so that interpretation of the results is clearer. In summary, this paper sheds further light as to how development in the financial sector affects growth. My analysis indicates that financial development affects economic growth by promoting technological change, a fundamental driving engine of economic growth, rendering further support for the finance-led growth hypothesis.
Whether a country should allow free movements of financial capital across borders is a whole new different question. Capital account liberalization usually entails financial market liberalization. However, financial market liberalization does not necessarily imply capital account liberalization.
12

64

Real GDP per capita growth 2.5 2.0 1.5 1.0 0.5 0.0 Haiti 0.0 -0.5 Thailand Singapor Korea

-2.0

2.0

4.0

6.0

growth rate of patent per million

Figure 4.1: Long Run GDP Growth vs. Long Run Patent Growth

 

65

0.3 0.25 0.2 0.15 0.1 0.05 0 1901 1902 1903 1904 1905 1906 Private credit Brazil Liquidit dmba

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 1901 1902 1903 1904 1905 1906 Mexico

0.6 0.5 0.4 0.3 0.2 0.1 0 1901 1902 1903 1904 1905 1906 Philippines

1.2 1 0.8 0.6 0.4 0.2 0 1901 1902 1903 1904 1905 1906 Malaysia

Figure 4.2: Movements of Financial Development Indicators for Selected Countries.

 

66

Argentina Brazil Chile Colombia Costa Rica Dominican Republic Ecuador Egypt Guatemala Haiti

Indonesia India Iran Iceland Israel Jamaica Korea Mexico Malaysia Pakistan

Peru Philippines Singapore Thailand Trinidad & Tobago Uruguay Venezuela South Africa

Table 4.1: Sample Countries Used For Patent Regression

 

67

(1) PC DMBA Liquidity Human Capital Openness Inflation Investment Knowledge Stock Government Size N
R
2

(2)

(3)

1.610(3.062) 1.070(2.470) 0.706(1.826) 2.112(1.719) -0.287(-0.700) 0.023(2.178) -5.416(-1.398) -0.247(-0.842) -1.246(-1.096) 156 0.459 2.251 27.578 1.726(1.395) -0.272(-0.625) 0.018(1.734) -4.980(-1.335) -0.045(-0.261) -0.918(-0.817) 156 0.443 2.223 26.205 2.023(1.791) -0.316(-0.895) 0.024(1.936) -3.830(-1.054) -0.009(-0.060) 0.136(0.140) 156 0.444 2.249 26.289

DW F-Statistic

Note) The numbers in the parenthesis are t-statistics. The dependent variable is the average growth rate of patent applications per million. PC is the initial values of the ratio of private credit by deposit money banks to GDP for each corresponding sample period. Liquidity is the initial values of the ratio of liquid liabilities to GDP for each corresponding sample period. DMBA is the initial values of the ratio of deposit money bank assets to GDP for each corresponding period. Human capital (HIATT 25) is measured by the initial values of percentage of the population 25 years of age or older who have attained higher education for each corresponding sample period. Openness is measured by the average levels of the sum of exports and imports divided by GDP in five years immediately preceding each period in the sample. Inflation (PI) is the average levels for each corresponding sample period. Investment (INVEST) is the average levels of domestic private investment in the five years immediately preceding each period in the sample. Knowledge stock is the initial levels of real GDP per capita for each corresponding period. Government size is the average levels of the share of total government consumption in GDP for each corresponding sample period.

Table 4.2: Fixed Effects Estimation I

 

68

(1) PC DMBA Liquidity Human Capital Openness Inflation Investment Knowledge Stock Government Size Legal N
R
2

(2)

(3)

0.909(1.765) 0.499(1.241) 0.291(0.755) 2.154(1.932) -0.244(-0.583) 0.026(2.318) -5.917(-1.470) 0.045(0.248) -0.851(-0.751) -0.048(-1.750) 145 0.492 2.282 24.901 2.010(1.816) -0.184(-0.411) 0.023(2.220) -5.380(-1.367) 0.116(0.658) -0.534(-0.490) -0.051(-1.877) 145 0.491 2.265 24.855 2.208(2.016) -0.237(-0.606) 0.024(2.374) -4.505(-1.148) 0.128(0.741) 0.044(0.045) -0.035(-1.408) 145 0.509 2.274 26.369

DW F-Statistic

Note) The numbers in the parenthesis are t-statistics. The dependent variable is the average growth rate of patent applications per million. PC is the initial values of the ratio of private credit by deposit money banks to GDP for each corresponding sample period. Liquidity is the initial values of the ratio of liquid liabilities to GDP for each corresponding sample period. DMBA is the initial values of the ratio of deposit money bank assets to GDP for each corresponding period. Human capital (HIATT 25) is measured by the initial values of percentage of the population 25 years of age or older who have attained higher education for each corresponding sample period. Openness is measured by the average levels of the sum of exports and imports divided by GDP in five years immediately preceding each period in the sample. Inflation (PI) is the average levels for each corresponding sample period. Investment (INVEST) is the average levels of domestic private investment in the five years immediately preceding each period in the sample. Knowledge stock is the initial levels of real GDP per capita for each corresponding period. Government size is the average levels of the share of total government consumption in GDP for each corresponding sample period. Legal is the initial values of an legal index compiled by Gwartney et al (2002).

Table 4.3: Fixed Effects Estimation II

 

69

(1) PC DMBA Liquidity Human Capital Inflation Legal N
R
2

(2)

(3)

0.892(2.400) 0.696(2.206) 0.879(2.256) 2.285(2.463) 0.015(2.150) -0.034(-1.164) 145 0.529 2.273 65.824 2.424(2.456) 0.012(1.883) -0.037(-1.295) 145 0.533 2.246 66.734 2.097(2.262) 0.019(2.420) -0.021(-0.838) 145 0.523 2.270 64.424

DW F-Statistic

Note) The numbers in the parenthesis are t-statistics. The dependent variable is the average growth rate of patent applications per million. PC is the initial values of the ratio of private credit by deposit money banks to GDP for each corresponding sample period. Liquidity is the initial values of the ratio of liquid liabilities to GDP for each corresponding sample period. DMBA is the initial values of the ratio of deposit money bank assets to GDP for each corresponding period. Human capital (HIATT 25) is measured by the initial values of percentage of the population 25 years of age or older who have attained higher education for each corresponding sample period. Inflation (PI) is the average levels for each corresponding sample period. Legal is the initial values of an legal index compiled by Gwartney et al (2002).

Table 4.4: Fixed Effects Estimation with Selected Independent Variables

 

70

CHAPTER 5

Investment Efficiency and Financial Development

"The proximate causes of economic growth are the efforts to economize, the accumulation of knowledge, and the accumulation of capital." — Lewis (1955)

5.1

Introduction

According to the finance-led growth hypothesis, financial development affects investment in two ways. Firstly, a better developed financial sector may raise the investment rate by pooling and risk sharing. Specifically, it makes available more saving opportunities to savers so that there is a larger pool of available funds that can be used for investment. Although it seems intuitive, the theoretical basis for this view is ambiguous (Devereux and Smith, 1994; Japelli and Pagano, 1994). Alternatively, instead of raising the savings rate, financial development may increase the investment rate by allowing more efficient transformation of already existing funds (savings) into investment (Roubini and Sala-i-Martin, 1995). This is the so-called volume channel. Secondly, the efficiency channel hypothesis states that financial development may increase the efficiency of investment by directing the funds to the most productive uses. As it is the role of the financial sector to distribute funds among

71

various investment opportunities, a better developed financial sector can identify productive investment opportunities more efficiently, enabling more efficient allocation of capital (Greenwood and Jovanovich, 1990; Bencievenga and Smith, 1991).13 So, it is not necessarily that people invest more with a well-developed financial sector but that they invest more wisely. This is the focus of this chapter. I examine if there is any evidence of financial development positively affecting the efficiency of aggregate investment using developing countries as a sample. The hypothesis that financial development positively affects the investment rate has first been examined by King and Levine (1993). In response to the criticism that this study suffers from simultaneity bias, Levine (1998) uses legal environment as an instrumental variable to support his previous conclusion.14 Interestingly, a more recent study by Beck et al (2000), using the dynamic GMM technique to control for simultaneity and unobserved country-specific effects, concludes that the long-run link between financial development and physical capital growth is tenuous. Thus far, the overall impression from the literature is that the nature of the relationship between financial development and the volume of investment is, at best, controversial. Compared to the volume channel, the efficiency channel has received relatively little attention until recently. Wurgler (2000) looks at the effect of stock market development on the pattern of capital allocation across industries, and finds that financially less developed countries tend to overinvest in declining industries and
13

For more recent survey, see Tsuru (2000).

14 Benhabib and Spiegel (2000) reach a similar conclusion using panel data although the effects of financial development is sensitive to the inclusion of country fixed effects which suggest that the financial development indicators widely used in the literature are proxying for the broader country characteristics.

72

underinvest in growing ones. Fisman and Love (2003) find that countries have more highly correlated growth rates across sectors when both countries have well-developed financial markets. Although these studies address the interesting question of whether a better-developed financial market allows firms to take advantage of growth opportunities more easily, it does not directly relate to the issue of investment efficiency. As the goal of investment is the accumulation of a factor of production, a more suitable criterion for investment efficiency is whether a given unit of investment results in a higher output level. Accordingly, a more pertinent question is whether financial development allows a bigger bang for the buck in terms of the change in output. In other words, a more efficient financial sector enhances productivity of capital measured by the change in the output. Secondly, the stock markets these studies consider are not a major source of funds for firms in most countries. By focusing on the stock markets only, an important role the banking sector plays in these countries is largely ignored. In this chapter, I address the issue of the efficiency channel using two alternative measures of aggregate investment efficiency. I find that, for developing countries, financial development significantly and positively affects productivity of investment. Further, I depart from the existing studies by focusing on the banking sector to measure the degree of financial development. The rest of this chapter is organized as follows. In section 2, I examine whether investment is an important determinant of output growth when adjusted for its allocative quality. In section 3, based on the results from section 2, I construct two measures of investment efficiency and analyze their relationship with financial development. In section 4, I conclude.

73

5.2

Investment, output growth, and financial development

Developing countries around the world make significant investments annually into what economists regard as important for economic growth in the hope of achieving a better standard of living. According to the World Bank data, developing countries spent roughly 0.6% and 4.6% of GNP, on average, for research and development and education respectively. On the other hand, they devoted approximately 22% of their GDP for gross domestic fixed investment on average.15 While this number may exaggerate a bit the degree of emphasis these countries place on physical capital accumulation, I believe that it clearly shows that developing countries have chosen mainly the strategy of investment-driven economic growth. Not coincidentally, this strategy is perfectly in line with what international organizations such as the World Trade Organization (WTO) and the Asian Development Bank emphasize as a vital component of sustained economic growth. For example, Mike Moore, WTO’s director-general in 1999, stressed, in his speech to a group of trade ministers of least-developed countries, the importance of investment to development. Similarly, in its 2000 report of the Country Economic Review — Cambodia, the Asian Development Bank suggested that “boosting investment rates, currently low and overly dependent on foreign savings, is critical to achieving the government’s goal of sustained economic growth.” Figure (5.1) shows the relationship between investment and output growth for Cambodia in recent years. Although its investment rate as a share of GDP is lower than most countries, it consistently remains at between 4 and 6 percent of GDP. However, the real GDP per capita growth rate fluctuates
15

For countries whose data are available.

74

significantly between 14 percent and -2 percent. Looking at this picture does not convey a message consistent with the statement of the Asian Development Bank. It seems that something other than investment is driving output growth, or lack thereof, in this country. Unfortunately, the same story applies to most of the developing countries. The strategy of investment-driven economic growth does not seem to have worked successfully for them. In sum, investment did not lead to the higher standard of living for a majority of developing countries (Easterly, 2002). This lack of evidence that investment is an important element in economic growth strengthened the position of neoclassical growth theorists who have been arguing, since Solow’s work, that the focus was misplaced. They argue that, rather than focusing on investment, less-developed countries should devote more resources to devise a way to promote knowledge creation, which is deemed to be the only reliable source of long-run growth. Until endogenous growth theorists arrived on the scene, those who believed that investment matters for economic growth, particularly development economists, didn’t have a good answer with which to throw back at them. Now, they were given new tools, more rigorous and detailed than the Harrod-Domar type models to justify their claims that investment does have a role in economic growth. If one hoped to have a more definitive answer to the growth question by leaving the theoretical battlefield and looking at what the empirical evidence shows, she would be disappointed. Both camps, innovation-based growth and investment-based growth, gathered up enough empirical studies to justify and support their positions. De Long (1992) and Sala-I-Martin (1997) provide support for investment-driven growth. On the innovation-driven growth side, King and Levine (1994) and Easterly et al (2001) argue that physical capital accumulation is not the answer to the long-run 75

growth question. Recently, Madsen (2002) tests for causality between investment and economic growth and finds that growth is largely driven by investment in machinery and equipment. Amidst this clamor, a new consensus is emerging that some of the economically successful countries, of which the economies seem to be significantly driven by technological innovation nowadays, started out their paths onto prosperity initially by accumulating physical capital. For instance, Korea was able to grow much faster than a majority of other developing countries despite the apparent lack of sufficient technological sophistication (Gylfason, 2004). Hayami and Ogasawara (1999) find that the main force behind growth for Japan remains to be physical capital accumulation. Similar evidence is presented by Toh et al (2002) for Singapore’s growth experience. Acemoglu et al (2000) provide a theoretical support for this view and argue that the reason these countries pursued investment-based growth is because of relative technological inferiority which made innovation-driven growth more costly. Then the question that begs our attention is: why the accumulation of physical capital seems to work in some countries while it is failing in a large number of other developing countries? As Easterly (2002) suggests, the fact that investment does not seem to have its intended effect on output growth in developing countries may have to do with how capital is allocated. In other words, it is not necessarily the volume of investment but the efficiency with which investment is made that counts. This view is supported by Blomstrom (1996) who, while finding no evidence that the volume of fixed investment is the key to economic growth, takes a position that whether investment is efficient seems to be one of the chief foundations for economic growth.

76

As it turns out, this is one common ground between growth theorists and development economists. After all, technological advance, THE answer to prosperity, can be viewed as one form of increased efficiency. So, now the question can be rephrased. We should ask not how to increase output growth but how to increase efficiency. It is in this context that the financial sector can play a significant role as a distributor of funds to investment opportunities. In this section, I investigate if investment is a significant determinant of output growth in the presence of the financial sector by introducing an interaction term of the financial development indicator and investment. A similar approach was adopted by Burnside and Dollar (2000) to show that aid can be effective in the presence of good policy framework.

5.2.1

Data

There is evidence showing that, in identifying the determinants of output growth, the results can be sensitive to what is used to measure the output level (Temple, 1999). Accordingly, the researcher should choose the output measure with the goal of the analysis at hand in mind. For developing countries, it is reasonable to assume that a majority of investment would be channeled into industrial sectors that take up a relatively small fraction of an economy’s GDP. Figure (5.2) shows the share of industry value added in GDP for developing countries considered in this analysis. It varies greatly across countries ranging from 13% to about 62% of GDP with a mean of 27%. Consequently, it would underestimate, in the context of the current discussion, the effects of investment if one measured the output only by GDP. For these reasons, I use real GDP per capita and real industry value added (henceforth IVA) for ISIC 77

divisions 10 — 45 to measure the output. GDP data are obtained from Penn World Table 6.1, and IVA data are from World Development Indicator, 2000.16 I divide it by the population to get IVA per capita. I use the share of gross fixed capital formation from IMF’s IFS in CD-Rom in GDP to measure the investment rate.17 The optimal measure of financial development would be a variable that reflects the changes in the cost of financial intermediation that occur as a result of the financial sector becoming more efficient. The measure that is a close enough proxy for the cost of providing financial intermediation is arguably the real interest rate (Furstenberg and Fratianni, 1996). However, in practice, the use of the real interest rate data is not feasible because of the lack of available data for developing countries. Also, it carries some undesirable characteristics that are hard to reconcile (Benhabib and Spiegel, 2000). Therefore, I use the so-called quantity indicators that are widely used in this literature. Although these size indicators are not perfect measures of financial development, I argue that they reasonably accurately capture the degree of financial development for developing countries.18 A caveat is that it is necessary to make the assumption that the size of the financial sector is negatively correlated with the costs of providing financial services. For developing countries, I think that this is a reasonable assumption. For the current study, I choose two measures of financial development. One is the ratio of deposit money bank assets to GDP (henceforth DMBA), and the other is the
16 17

Refer to appendix for more detailed description of all the data used in this chapter.

I use the population instead of the number of workers, although the latter would be more appropriate, due to the lack of available data. Size indicators may not be a good proxy for financial development in developed countries as the financial sectors in these countries seem to have reached a mature stage in terms of their size as early as in the 1960s.
18

78

ratio of private credit by deposit money banks to GDP (henceforth PC). DMBA is classified as an absolute size measure that reflects the relative importance of deposit money banks in providing financial services in the economy. PC measures somewhat different aspect of financial development that relates to the activity of the financial institutions. These two variables will be used throughout this chapter to measure the degree of financial development. Although I use two different measures of financial development, I should note that the preferred measure is DMBA in this chapter. Gregorio and Guidotti (1995) argue that what type of financial development indicator is to be used in this type of analysis should be governed by the focus of the research at hand. Since our focus is on the efficiency aspect of financial intermediation, I believe that the size of the deposit money bank assets, which reflects the efficiency of banking management, captures the efficiency aspect of financial intermediation better than the volume of lending the banking sector is engaged in, which is what PC represents.

5.2.2

Estimation strategy

The list of countries that are considered in this analysis is provided in the appendix. The choice of countries to be included was based strictly on the availability of the necessary data. I start out with 53 developing countries from 1970 to 1998.19 Output is measured by real GDP per capita and real industry value added per capita as explained above. Note that when IVA is used to measure the output, I lose five countries in my sample. I use five-year averages to compute output growth, which gives me six observations across time for each country: 1970 — 1974, 1975 — 1979, 1980 — 1984, 1985 — 1989, 1990 — 1994, and 1995 — 1998. Note that the last period is made up of four years due to data availability.
19

Note that for some countries, the data cover shorter ranges. See the appendix for details.

79

To capture the allocative quality of investment, I interact the investment rate with the financial development indicators. Hence, the equation I estimate is the following: git = αi + β 1 Iit + β 2 F Dit + β 3 (I × F D)it + γ(controls)it + εit (5.1)

g is the growth rate of output per capita in real terms. α is a country dummy that capture the country-specific effects. Investment rate (I) is the log of the initial value of the share of gross fixed capital formation in GDP for each period. Similarly, to avoid simultaneity, the log of the initial value of the financial development indicator (F D) for each period is used to represent the degree of financial development at the beginning of each period. The equation is estimated by the FGLS. I start with a base regression that includes only the investment rate and human capital. To measure the level of human capital, I use the literacy rate. The widely used measure of human capital is the Barro and Lee data (2000) that lists the percentage of population, 15 and over, that received a certain degree of education. Although it is popular among growth researchers, I believe that literacy rate is a better measure of the level of human capital for developing countries. For example, in a country like Kenya, between 1980 and 1985, the percentage of secondary school enrollment in the population, 15 and over, fell from 14.5% to 9.5% while the illiteracy rate fell from 43.7% to 36.1%. This example illustrates that the level of human capital can be better reflected by the outcome of the education (which literacy rate captures) than by the number of people who receive education (which Barro and Lee data captures). So, although the two variables are generally positively correlated across the sample countries, I use the literacy rate to measure the degree of human capital. The values used in the analysis are the initial values for each period. Column (1) of table (5.1) 80

shows the results. The investment rate is negative and insignificant in line with the literature. On the other hand, human capital proxied by literacy rate turns out to be significantly positive. However, as can be seen in the column (2) of table (5.1), adding the initial level of income to the regression to capture the possible convergence effect renders the human capital variable positive but statistically insignificant. It seems that for the group of developing countries in the sample, the convergence effect seems quite large. The third and fourth columns of table (5.1) show the results of including the financial development indicators in the regression. In addition to these variables, I include three other variables to control for the policy aspect of the countries. The share of exports plus imports in GDP, obtained from Penn World Table, 6.1, is used to measure the degree of openness in the economy. Government size, measured by the government share in real GDP, is included in the regression to capture the public sector. The monetary policy aspect of a country is proxied by the rate of inflation, measured by the percentage change in GDP deflators. All variables are in logs.

5.2.3

Discussion

Overall, the results are generally consistent with the literature with the government size and inflation having negative effects on output growth. The coefficient of the financial development indicator is negative and insignificant. The effect of the investment rate on real GDP per capita growth seems to get quite big but is still statistically insignificant. Note that the interaction term, investment rate*financial development indicator, our variable of interest, is positive and significant for both measures of the financial development indicators, PC and DMBA, indicating that investment is a significant determinant of output growth when its allocative quality is

81

taken into account. However, its effect on output growth, compared to other variables that are significant, seems relatively small. When I use IVA per capita growth as the output measure, the overall results do not change. The results are shown in table (5.2). The interaction term is still significantly positive although its effect is smaller than in the preceding analysis. Interestingly, both the investment and the human capital are significant determinants of industrial output per capita growth. It may hint at the possibility that the accumulation of physical and human capital does what policy-makers believe it does after all, but that its effect is not evident in the typical growth analysis of developing countries using GDP per capita growth in which the share of the industrial output is small. The view that the legal aspect of the economy needs to be included in the analysis of the role of the financial sector is increasingly gaining acceptance among researchers. According to this view, the effective functioning of the financial sector is closely linked to the type of legal framework of the economy so that any investigation on the role of the financial sector in the economy should take into consideration the integrated nature of these two variables. On the other hand, one can argue, based on some evidence that the financial development indicators lose their significance when the country-fixed effects are included (Benhabib and Spiegel, 2000), that financial development is just a proxy for underlying fundamental economic infrastructure, including the legal framework governing the market activities. In order to account for this argument, I run another regression with two additional control variables. For this purpose, I first obtain data on legal system and property rights from Economic Freedom of the World: 2002 Annual Report (Gwartney et al, 2002) to capture the legal

82

environment of a country. The degree of the economy’s infrastructural development is difficult to measure since it involves many different aspects of the economy. In this study, I use the number of telephone lines per 1000 people to proxy for the infrastructural development. Although it is an imperfect measure of infrastructure, it has the advantage of being widely available for a long time span.20 Note that when the legal environment and infrastructure variables are added to the regression, the sample period is shortened from 1970 — 1998 to 1980 — 1998 due to the lack of available data on legal environment for most of the developing countries prior to 1980.21 Table (5.3) shows the results. Adding these additional variables does not change the overall picture although the model fit is slightly better. The effects of the interaction term on both real GDP per capita growth and IVA per capita growth remain about the same with most of the coefficient estimates of other control variables coming out with expected signs and generally being consistent with the preceding results. Overall, the results I obtain from this analysis appear to confirm that investment does work with a presence of the properly developed financial sector to distribute capital in an efficient manner. Although this is admittedly not conclusive evidence for the importance of investment in enhancing output growth because of the inherent sensitivity of this type of study to model specification and the sample being used, it seems to provide a reasonable support for the argument that the reason investment has not been effective
An alternative proxy for the infrastructure that is popular is infant mortality rate. When I ran the regression with the infant mortality rate to control for the infrastructure, the results did not change.
21 Additionally, some countries are dropped from the sample completely due to the lack of data on the legal environment. A list of those countries dropped is in the appendix. 20

83

in promoting economic growth for a large majority of the developing countries is the lack of mechanism to efficiently allocate capital.

5.3

Investment efficiency

The findings from the preceding section indicate that the financial sector positively affects output growth by enhancing the allocative quality of investment for developing countries. With this background, my goal in this section is to directly examine the effects of financial development on the efficiency of investment and test the validity of the efficiency channel of influence proposed by the finance-led growth literature (see Greenwood and Jovanovich, 1990; Bencievenga and Smith, 1991).

5.3.1

Measures of investment efficiency

The literature typically defines (aggregate) investment efficiency as a ratio of the output to the capital stock — or some minor variation of this ratio. The most commonly used measure is the so-called incremental capital output ratio, generally known as ICOR, which looks at the units of capital needed to increase the output level by one unit. It is essentially the inverse of the marginal product of capital. For instance, Jun (2003) uses this measure to analyze the investment-growth nexus between 1978 and 2000 for China. Odedokun (1996) adapts a slightly different version of this ratio and uses a change in output divided by a change in capital stock to examine the effects of development banking activities on resource allocation in less developed countries. This ratio can be modified to reflect the belief of the researcher about what the capital accumulation is thought to achieve. For example, Toh et al (2002) replace the aggregate output measure with estimated total factor productivity to examine whether the capital is efficiently utilized in Singapore compared to other Asian countries. 84

In order to investigate whether investment is an effective tool for promoting growth and whether its seeming failure to affect growth can be attributed to other macroeconomic variables, I start out by taking the comments of the Asian Development Bank (ABD henceforth) on Cambodia seriously. The readers will recall from the introduction of this chapter the following statement. “boosting investment rates, currently low and overly dependent on foreign savings, is critical to achieving the government’s goal of sustained economic growth. What ABD meant seems quite clear. In order to realize a favorable growth rate, investment rate has to be raised. It implies that there is a systematic relationship between investment rate and growth rate of an economy. A direct testable implication of this is that when investment rate rises by a certain percentage point, there will be a corresponding change in the growth rate of the economy. However, whether the aggregate investment is efficient in achieving this goal depends, to a large degree, on the macroeconomic environment. Thus, as this macroeconomic environment changes over time and across countries, a given unit of investment will produce dissimilar result in terms of output growth. Given that the efforts of developing countries to achieve a higher standard of living by focusing on investment have been, thus far, unsuccessful, and that this failure may be attributed to the inefficient manner in which the funds were allocated, identifying those elements of the macroeconomic environment that affect the efficiency of investment has important policy implications. To this end, I estimate the following: ¶ µ Iit ∆yit + εit = αi + β it ln ∆yit−1 Iit−1 (5.2)

where y is the output and I is the investment rate. αi is a dummy variable that captures country-specific unobserved effect. β i is an elasticity. It measures the extent 85

of the change in growth rates in response to a change in investment rate in country i. If investment is made efficiently over time, perhaps due to the presence of a better developed financial sector, a rise in investment rate of a given unit will result in a bigger change in the growth rate of the economy. In other words, a rise in investment rate in Cambodia will result in a higher rate of economic growth than otherwise if a good allocative mechanism is in place to make sure that additional investment is put into its best uses. Hence, a higher value of β implies that investment is utilized efficiently. A possible objection to this interpretation is that a lower value of β may represent not inefficient investment but the possibility that agents are looking at a longerhorizon than the next year. To incorporate this possibility, I estimate βs for 10-year horizon in this paper. Therefore, one can think of βs as a long-run trend relationship between investment rate and output growth rate. This is an advantage over the conventional measure of the investment efficiency that is measured on a yearly basis and, thus, does not reflect the possible long-run horizon investment decisions. It is possible that a rise in growth rates can cause the investment rate to grow. For example, a 2-percent rise in growth rate may induce the economy to raise its investment rate from 5% to 7% hoping to take advantage of (or accelerate further) the economy taking off if economic agents are forward-looking and expect the trend to continue. If this were the case, endogeneity makes interpretation of the results difficult. Although it is a possibility, however, it is highly unlikely for developing economies. Empirical evidence shows that the growth experiences of the developing countries are highly volatile, and there does not seem to be a noticeable upward

86

trend in their growth rates (Temple, 1999). In an environment where growth has been volatile, it is unlikely that the economic agents behave in a forward-looking manner and raise investment rate in response to what could be a temporary change in the growth rate. Before we proceed, a warning is in order. The βs in this study should not be interpreted strictly as measuring the causal effect of investment on output growth. Rather, given the evidence that investment that is adjusted for its allocative quality is a significant determinant of output growth, the emphasis is on providing an explanation as to why the output growth seems to respond to investment better in some countries than in others. In addition to this measure, I also use the ratio of the real GDP to the capital stock. This is essentially the approximation of the marginal product of capital assuming that the income shares of the factors of production are constant. The advantage of this measure over the marginal product of capital as a measure of investment efficiency is that it does not require any functional assumptions. In this study, this more conventional measure is used for comparability and completeness.

5.3.2

Investment efficiency estimates

In order to estimate the equation (5.2), I use two measures of output, the real GDP per capita and IVA per capita, as defined and motivated in the preceding analysis. Similarly, investment I is measured by the gross fixed capital formation. To avoid endogeneity, I use the lagged values of I. The data are gathered for 53 developing

87

countries from 1970 to 1998.22 Then, I divide the sample into three sub-periods: 1970 — 1979, 1980 — 1989, and 1990 — 1998 so that I get three estimates of β for each country. The idea is that I want to incorporate as much time dimension as possible into my analysis. Then, the equation is estimated by OLS with country dummies. Table (5.4) and (5.5) report the results. A larger value of β implies that the growth rate was more sensitive to changes in investment rate. Looking at the results, we can see that there are some incidences where an increase in investment rate not only failed to raise growth rate of the output but actually is associated with a slower rate of growth. However, for the entire sample, investment efficiency is positive indicating that a rise in the investment rate is generally positively associated with a rise in output growth rate. And, the impact of raising investment rate is much greater on the industrial sector output growth rate than on the national-level output growth rate. In fact, for the entire sample, the mean of estimated β with the real GDP as the output measure, henceforth β GDP , is 0.05 and 9.5 with IVA per capita as the output measure, henceforth β IV A . A much lower value of β GDP may indicate why other growth studies have failed to pick up investment rate as a determinant of economic growth. So, despite the negative experiences of some countries, raising investment rate does seem to, in general, promote a faster growth of the economy although its effect is more pronounced in the industrial sector. Furthermore, these countries seem to be getting better at utilizing investment. The number of the incidences when investment rate is negatively associated with output growth rate declines in the 1980s and 1990s. Also, the estimates of βs show
This is the same group of countries used in the preceding analysis. This is done intentionally to keep the sample as consistent as possible throughout the paper. For details about the sample, refer to the appendix.
22

88

that investment efficiency improves over time. When we compare the estimated investment efficiency between 1970s and 1990s, the average of the estimated β GDP rises from 0.008 in 1970 to 0.055 in 1990s while that of the estimated β IV A rises from 5.37 to 10.17. Thus, again, it seems that the more relevant task for growth economists should be not asking whether investment matters for economic growth but understanding why growth rate of an economy seems to respond well in some countries and, for these developing countries, better over time. On the other hand, the estimated investment efficiency displays wide variations across countries. For the whole sample, the standard deviations are 0.15 and 17.34 for β GDP and β IV A respectively. Over time, the magnitude of these differences across the countries rises as well. The standard deviations rise from 0.13 to 0.8 and from 13.7 to 19.7 for β GDP and β IV A , respectively, from the 1970s to the 1990s. Across time, very few countries seem to maintain a high level of investment efficiency. For the whole sample, the correlation between the estimated βs for two consecutive decades is generally low. Between the 1970s and 1980s, they are 0.61 when output is measured by real GDP per capita and 0.51 when it is measured by real IVA per capita. Between 1980s and 1990s, the correlation becomes weaker although it is not clear why this happened. They are 0.32 when output is measured by real GDP per capita and 0.14 when it is measured by real IVA per capita. This dynamic and volatile nature of the relationship between investment rate and output growth rate illustrated by the data further emphasizes the importance of introducing time dimension to the analysis of developing countries and question the suitability of cross-sectional studies.

89

By and large, the estimated βs paint a picture that is in line with the findings of the current growth literature, which allows me to be confident that they carry useful information about the constraints these countries face in terms of investment decisions.

5.3.3

Determinants of investment efficiency

Investment may promote output growth via two channels. Firstly, a rise in investment allows greater production of output by making more capital, a factor of production, available to be utilized. Clearly, this change in the volume of investment may not, in itself, be able to support a sustained rate of output growth as diminishing returns eventually kick in. However, investment also creates positive externalities. The accumulation of capital, embodied with knowledge, allows learning-by-doing of the workers, which results in the gain in the overall efficiency of production processes. For example, suppose you have been adding 5 machines to your existing stock of 100 machines every year. This year, you change your plans and start adding 10 machines instead of five. In the past, there were five more people who knew how to operate the machine. Now you have ten more people instead of five. Eventually, you will run out of new people to run newly-added machines (diminishing marginal returns). However, the pool of people who know how to operate the machine is greater with ten-percent investment rate than with five-percent. As each person is equipped with knowledge, and there are more of them, the probability for the occurence of positive externality is also greater. In other words, when investment rate is raised, the rate of learning-by-doing for the population also increases. This is a second channel through which investment can affect growth even in the long-run. Investment enhances the

90

efficiency of overall production processes. Hence, any changes in the economic environment that hinder the spillover effects of investment (the second channel above) or prevent investment to reach its best possible uses (the first channel above) would lead to a decline in the productivity of investment. A high rate of inflation distorts the returns to investment and, consequently, the incentives of investors. Specifically, persistently high rates of inflation may induce the economic agents to invest in liquid and yet unproductive assets which leads to inefficient allocation of resources. However, the literature is silent on exactly at what rate of inflation this would occur. Coupled with this, an unstable monetary policy that is prone to higher rates of inflation could hinder the effective functioning of the financial sector as an allocator of funds to various investment opportunities (Savvides, 1995). In addition, with more government involvement in the private market activities, resource allocation is more likely to be guided by political motives and less by valuemaximization (Wurgler, 2000). As a result, relatively more suboptimal investment decisions would occur when the government share of the economic activities is greater. Another important element of the economic environment that governs the productivity of investment is the level of human capital available in the economy. I posit that human capital may serve not necessarily to affect the growth rate of output directly but to make it possible for capital to be utilized more efficiently. Controlling for the macroeconomic environment with these variables, I test whether financial development indicators enter significantly in determining investment efficiency in the next section.

91

5.3.4

Estimation

In order to test the hypothesis that financial development affects investment efficiency, I estimate the following equation: Investment Efficiencyit = αi + γ(F D)it + z(controls)it + εit (5.3)

where α captures country-specific effects. As noted above, investment efficiency is measured by i) estimated βs from above and ii) average product of capital which is the ratio of output to capital stock (Y/K). To compute the average product of capital, I obtain the capital stock data from Penn World Table 5.6. Note that the use of this measure significantly reduces the sample size. Due to the lack of data availability, when Y/K is used as a measure of investment efficiency, the time horizon is from 1970 to 1989 instead of 1970 to 1998. Also, I lose 27 countries from my initial sample of 53 countries. The list of countries used in this case is reported in the appendix. To incorporate as much time dimension as the data allows and following the conventional way of dividing up the time horizon, I compute the five-year averages of Y/K for 1970 — 1974, 1975 — 1979, 1980 — 1984, and 1985 — 1989. The control variables are inflation, government size, and human capital as mentioned above. Inflation is measured by the growth rate of GDP deflator. Government size is measured by the share of government expenditure in GDP. Human capital is measured by literacy rate. The degree of financial development is measured by two variables, PC and DMBA as discussed above. All variables are the initial values for each period to avoid simultaneity. Equation (1.2) is estimated with White’s hetero-skedasticity consistent estimator.

92

5.3.5

Discussion

Table (5.6) reports the results when investment efficiency is measured by β. All variables come out significant and with expected signs except for inflation. The results show that a more government involvement reduces investment efficiency confirming what Wurgler (2000) found in his study. Human capital is also a significant determinant of investment efficiency as predicted. More importantly, both of the financial development indicators are significant in improving how efficiently investment is made. In addition, the effects of financial development are the biggest among the variables included in the analysis. On the contrary, inflation does not seem to be an important factor of investment efficiency. While it comes out with the expected sign with β GDP , its sign changes when the dependent variable is β IV A . Interestingly, the effects of these variables are more pronounced when the output is measured by IVA. Considering that the significance of these variables in raising output growth has come under fire despite the intuitive and strong argument for their roles in enhancing output growth, the current study may shed a new light on exactly how they would affect economic growth. It seems that one possible way that these factors, government size, human capital, and financial development, affect the output growth is by improving the efficiency of investment rather than directly affect the output growth itself. Furthermore, their effects on output may not be adequately captured in the conventional studies using aggregate income measures of which the industrial output takes, in some cases, only a small fraction. Table (5.7) reports the results of using the average product as a dependent variable. The model fit is much better than the preceding regression. The financial 93

development indicators are still significantly positive as is human capital. However, the government size is now insignificant although it comes out with a right sign. Contrarily, inflation, which was not a factor before, enters significantly. Overall, the results thus far paint a picture that is consistent with what the literature predicts. Especially, the role of the financial sector seems to be robustly related to investment efficiency.

5.3.6

The effects of political stability and legal environment

Recently, there has been increasing attention on the sociopolitical factors as determinants of macroeconomic performance. These factors range from cultural aspect of an economy to religious beliefs of a dominant majority of the population in an economy and geographical location of a country. Indeed, the possibilities they represent to the researchers are nearly endless. But, the common thread that connects all these variables is that they do not necessarily affect macroeconomic performances directly but do so indirectly by impacting the institutional arrangements that govern economic activities (Rodrik, 2004). It is particularly in this context that the researchers question the role of the financial sector in the economy. One cannot disregard the possibility that financial development is working not as a source but a mere medium through which the unobserved social and economic characteristics exert influence on investment efficiency. Then, ascertaining the validity of the finance-led growth hypothesis requires that these characteristics are accounted for in the analysis. The most pertinent characteristic in this regard seems to be legal framework of an economy. It not only governs how well the financial sector functions but also determines the social arrangement under which economic activities occur. Therefore, I add a 94

legal variable, as in the previous section, to account for this possibility. In addition to a legal environment, political stability can also affect how effective the investment can be (De Long, 1992). In a politically unstable economy, there is a higher degree of uncertainty that economic agents or investors must find a way to reconcile with. I use the average number of assassinations for each sub-period to proxy for the political stability of a country.

5.3.7

Discussion

Table (5.8) and (5.9) report the results of adding these two additional variables to the equation. The effects of legal environment in some instances are significantly positive, but overall, they don’t seem to matter much, at least for the group of countries considered here, for investment efficiency. However, it does not necessarily suggest that the legal environment does not matter for investment efficiency. It may as well be that the changes in legal environment for these countries have been only marginal and have not reached the crucial tipping point where they begin to influence investment efficiency in any meaningful way. On the other hand, political stability has highly significant negative effects on investment efficiency when the output is measured by IVA although, interestingly, for the economy as a whole, it does not seem significant. Most importantly, adding these two additional variables do not change the result that financial development is a significant determinant of investment efficiency. The size of the coefficient is somewhat smaller but is, nonetheless, significantly positive as predicted by the finance-led growth literature.

95

5.4

Conclusion

Theories of the finance-led growth state that the financial sector contributes to the output growth by distributing capital to the most productive uses. The hypothesis consists of two testable implications. Firstly, in order for the financial sector to work its magic, investment needs to be an important determinant of output growth, the proposition that’s been hotly debated. Secondly, investment efficiency across countries should be positively correlated with the level of financial development. In this paper, I test these implications using developing countries. The nature of the growth process of the developing countries as noted in the literature provides an ample testing ground for the hypothesis of financial-sector-assisted investment-driven growth. Using two measures of investment efficiency, I find evidence that investment adjusted for its allocative quality is an important determinant of economic growth and that investment efficiency is positively affected by financial sector development. The analysis shows that the positive effect of the financial sector on investment efficiency is more pronounced in the industrial sectors than in the economy as a whole. Given the already strong emphasis the developing countries place on investment, and the relatively low costs it involves (compared to technological innovation), promoting efficiency of investment looks to be a very promising avenue for the lessdeveloped countries to pursue. The current study shows that in order to achieve this goal, these countries can take a number of policy measures. Firstly, to make investment more productive or efficient, careful attention needs to be paid to raising the general educational level of population. Secondly, a larger share of the economy 96

needs to be liberalized so that investment is made not based on political motives of the government but on the value-maximization of private economic agents. Thirdly, in order to promote investment efficiency, efforts need to be made to provide stable political environment for investors so that the resulting investment decisions are optimal. Lastly, developing financial sector should be one of the first priorities for these countries.

97

16 14 12 10 8 6 4 2 0 -2 -4
The GDP per capita is in real terms (1996 dollars), and investment rate is the share of real investment in GDP. The data for the GDP per capita growth and investment rate are from Penn World Table 6.1. Figure 5.1: Output growth and investment in Cambodia

GDP per capita growth

Investment rate

1994

1995

1996

1997

1998

1999 Year

98

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0

Br a Ca zil m Co eroo n te d'I vo ir Ec e ua do r

A lg er ia

Figure 5.2: The share of industry value added in GDP in developing countries

99

Gu Fiji at em al a In di a Ja m ai Ko ca re a, Re M p. ala ys ia Ne pa l Pa ki sta n Pa ra gu ay Rw an da Se yc he l Sr les iL an k Th a ai la nd U ru gu ay

Dependent variable: Real GDP per capita growth (1) Financial development measured by: (2) (3) Private credit (4) Domestic money bank assets

Variable Investment rate -0.3099(-0.5712) 0.0703(0.1168) 0.3307(1.0213) 0.3855(1.2204) Human capital 0.6381(2.8295) 0.0153(0.0074) 0.0658(0.1856) 0.0854(0.2396) Initial Income -2.9996(-2.9743) -3.2506(-2.2796) -3.2307(-2.2567) Finance -0.1601(-1.1092) -0.1253(-0.2591) Finance*Investment 0.1348(1.9973) 0.1274(2.3471) Openness 0.4744(1.2668) 0.3567(1.0113) Government size -1.4784(-2.7529) -1.3925(-2.5760) Inflation -0.0513(-2.1944) -0.0508(-2.301) Nobs. 317 317 303 305 2 0.1895 0.2159 0.520258 0.5334 R DW 2.18 2.1083 2.5001 2.5109 Prob(F-statistic) 0.0000 0.0000 0.0000 0.0000

Note) The equation is estimated by FGLS. The numbers in the parenthesis are t-statistics. Table 5.1: Investment Regression I

100

Dependent variable: Real Industry value added per capita growth (5) (6) (7) Private credit (8) Domestic money bank assets

Financial development measured by:

Variable Investment rate -0.1375(-1.6104) 0.0909(1.7055) 0.1298(2.4752) 0.1417(2.6738) Human capital 0.1423(3.2804) 0.0901(1.9953) 0.1254(2.8735) 0.1383(3.0470) Initial Income -0.3823(-8.5597) -0.4046(-7.7167) -0.3984(-7.6494) Finance 0.0277(0.2389) 0.0317(0.8801) Finance*Investment 0.0754(1.7650) 0.0563(2.3530) Openness 0.0697(1.2005) 0.0570(0.9829) Government size -0.1615(-2.8158) -0.1586(-2.7841) Inflation -0.0271(-2.0182) -0.0292(-2.2253) Nobs. 278 278 270 270 2 0.1934 0.2436 0.4478 0.4494 R DW 2.3510 2.2320 2.2965 2.3020 Prob(F-statistic) 0.0000 0.0000 0.0000 0.0000

Note) The equation is estimated by FGLS. The numbers in the parenthesis are t-statistics. Table 5.2: Investment Regression II

101

Dependent variable Financial development measured by: Variable Investment rate Human capital Initial Income Finance Finance*Investment Openness Government size Inflation Infrastructure Legal environment Nobs.
R DW Prob(F-statistic)
2

Real GDP per capita growth Domestic money bank assets

Real Industry value added per capita growth Private credit Domestic money bank assets

Private credit

0.2692(0.8820) 0.1426(1.7112) -2.9229(-2.2798) -0.1757(-1.2353) 0.1230(1.7546) 1.2258(2.1695) -2.8246(-4.5539) -0.1337(-2.3136) 0.0976(2.1133) 0.4004(1.6279) 163 0.5776 2.3554 0.0000

0.2974(0.9954) 0.1461(1.9618) -2.6055(-1.8442) -0.1521(-1.2199) 0.1088(1.9662) 1.1365(1.9915) -2.6603(-4.1237) -0.1411(-2.3111) 0.07402(1.7424) 0.3390(1.4652) 163 0.5734 2.3661 0.0000

0.2130(2.8697) 0.3167(2.3955) -0.4874(-3.6727) 0.0042(0.1093) 0.0516(1.8940) 0.1806(0.8658) -0.1462(-1.8952) -0.0125(-1.2062) 0.0072(0.1181) 0.0432(0.7037) 142 0.4544 2.2032 0.0000

0.2103(2.7962) 0.3108(2.3398) -0.4799(-3.6111) 0.0184(0.4246) 0.0347(2.0081) 0.1747(0.7532) -0.1409(-1.7988) -0.0128(-1.2897) 0.0105(0.1739) 0.0459(0.7553) 142 0.4583 2.2017 0.0000

Note) The equation is estimated by FGLS. The numbers in the parenthesis are t-statistics. Table 5.3: Investment Regression III

102

Country Algeria Argentina Bolivia Botswana Brazil Burkina Faso Burundi Cameroon Chile Colombia Costa Rica Cote d`Ivoire Cyprus Dominican Republic Ecuador Egypt El Salvador Fiji Ghana Guatemala Haiti Honduras India Indonesia Iran Israel Jamaica Jordan Kenya Korea

GDP 1970s 0.0959(0.8067) -0.0281(0.0064) 0.0914(0.0078) 0.2462(0.0432) n.a. -0.0593(0.0436) 0.0489(0.1721) 0.0539(0.2293) -0.0441(0.0524) -0.0252(0.0203) -0.0005(0.0241) -0.0439(0.0585) 0.3592(0.0725) 0.0176(0.0554) 0.0079(0.0477) -0.1516(0.0592) -0.0177(0.0142) -0.1863(0.0570) -0.0364(0.1340) -0.0729(0.0107) -0.0028(0.0723) -0.1998(0.0904) 0.1147(0.1781) 0.1079(0.0691) 0.0026(0.3700) 0.0977(0.0884) 0.0174(0.1006) 0.2868(0.1936) 0.2811(0.5376) 0.0910(0.0249) 1980s 0.0788(0.0852) -0.0315(0.0243) -0.0066(0.0021) 0.0883(0.0523) 0.0068(0.0152) 0.1508(0.0998) -0.1220(0.2945) 0.3013(0.0315) 0.2068(0.0541) 0.0718(0.0161) -0.1696(0.1712) -0.0922(0.1746) 0.1375(0.0070) -0.0263(0.0152) 0.0564(0.0588) -0.0786(0.0320) 0.2126(0.0549) -0.1770(0.1687) 0.0148(0.0513) 0.0345(0.0125) 0.1915(0.0831) -0.0400(0.0284) 0.0685(0.0290) 0.1442(0.0220) -0.0092(0.2964) -0.0105(0.0078) 0.1386(0.0545) 0.0581(0.0926) 0.2122(0.0590) 0.3238(0.1289) 1990s 0.0329(0.0186) -0.0301(0.0256) -0.0516(0.0039) 0.3050(0.0239) 0.0012(0.0090) 0.0392(0.0435) -0.0380(0.0700) 0.1388(0.0368) 0.1542(0.0371) 0.0374(0.0045) 0.1953(0.0400) 0.0015(0.0441) 0.1779(0.1986) -0.1117(0.0962) 0.0490(0.0153) 0.0516(0.0276) 0.0643(0.0194) -0.0411(0.0222) -0.0104(0.0096) -0.0656(0.0022) 0.3374(0.0381) -0.0068(0.0822) 0.0718(0.0697) -0.6386(0.3362) 0.0542(0.0260) 0.1610(0.0051) -0.0388(0.0160) 0.0123(0.0763) 0.0494(0.0217) 0.3485(0.0497)

Table 5.4: Investment Efficiency Estimates Using GDP

103

Table 5.4: Investment Efficiency Estimates Using GDP (continued)
Lesotho Madagascar Malaysia Mauritius Morocco Nepal Nicaragua Niger Pakistan Panama Papua New Guinea Paraguay Peru Philippines Rwanda Senegal Singapore Sri Lanka Thailand Togo Trinidad &Tobago Uruguay Venezuela Nobs
R DW
2

0.0260(0.1585) -0.0244(0.0321) 0.1450(0.0426) -0.0225(0.0474) 0.0337(0.0073) 0.0578(0.0022) n.a. -0.1247(0.3071) -0.0510(0.0202) 0.0618(0.0137) -0.1041(0.0320) 0.0138(0.0288) -0.0864(0.1069) -0.0099(0.0064) -0.0656(0.0323) 0.3792(0.0949) 0.1851(0.0210) 0.0400(0.0045) 0.1460(0.0519) -0.4699(0.1593) -0.1529(0.1345) 0.0417(0.015) 0.1054(0.1152) 437 0.1944 2.1556 0.0017

-0.0265(0.0287) 0.0395(0.0181) 0.1720(0.0098) 0.0635(0.0901) 0.1658(0.1628) 0.0100(0.1057) -0.0350(0.0084) 0.1095(0.1007) 0.2127(0.0282) 0.1757(0.0392) -0.1148(0.0258) 0.0832(0.0471) -0.0718(0.0337) 0.1998(0.0225) -0.1024(0.2519) 0.4705(0.4218) 0.2457(0.0403) -0.0335(0.0224) 0.2082(0.0254) -0.1105(0.2084) 0.0679(0.0955) 0.2236(0.0624) 0.0661(0.0511) 477 0.2678 2.0156 0.0000

-0.2076(0.1008) -0.0177(0.0087) 0.1039(0.0015) -0.0868(0.0064) 0.2049(0.6229) 0.0569(0.0414) 0.0047(0.0023) 0.0385(0.0343) 0.1339(0.0205) 0.1056(0.0054) -0.0664(0.5518) 0.0966(0.0409) -0.0198(0.0106) 0.1625(0.0452) 0.5560(0.2035) 0.1509(0.0105) 0.4337(0.1907) 0.0372(0.0884) 0.3058(0.0047) -0.0172(0.1372) -0.1366(0.3050) 0.0035(0.0376) 0.0205(0.0536) 424 0.5565 2.1113 0.0000

Prob(F-stat)

Note) βs are estimated from equation (5.2) using OLS. The numbers in the parenthesis are standard deviations.

104

Country Algeria Argentina Botswana Brazil Burkina Faso Burundi Cameroon Chile Colombia Cote d`Ivoire Dominican Republic Ecuador Egypt El Salvador Fiji Ghana Guatemala Haiti Honduras India Indonesia Iran Jamaica Jordan Kenya Korea Lesotho

IVA 1970s -2.2125(0.1649) -1.2066(0.9041) 0.4697(2.2334) n.a. -9.9135(14.8534) 1.9567(7.7122) -1.4125(14.7897) -3.8672(4.9868) -3.5007(3.7782) 14.2528(24.7746) 15.3899(10.8853) -7.6419(8.7375) 9.7605(2.0121) 1.4479(2.9971) -33.2566(4.7908) 6.1689(7.6931) 14.0998(1.6034) 9.1138(6.8254) 12.5832(3.7347) 9.4180(7.7300) 1.2447(0.6117) 34.0102(25.0239) 15.0687(6.3684) n.a. 2.8326(68.2315) 36.2086(5.7064) 3.3637(18.4002) 1980s 3.9202(0.9031) -3.626(2.5835) -7.4819(7.7584) 0.7703(2.238) 8.4580(2.9813) 2.6930(0.4997) 5.6848(1.9119) 19.8963(5.1658) 5.7932(4.5881) 5.5477(10.2416) 25.1306(9.7111) -11.7707(12.3296) -2.6693(4.7574) 36.7105(7.6437) -15.5363(33.8456) 17.3982(17.4023) 19.6717(3.9835) 27.7791(7.1284) 8.0876(2.9952) -0.9967(6.5817) 12.933(12.6262) 32.1233(47.562) 19.3543(10.5561) 1.2563(7.3034) 8.379(1.1517) 64.0368(20.2605) 7.6509(1.2903) 1990s -3.4958(3.5071) -2.4746(2.203) 2.017(0.1054) -0.5822(1.2991) 3.6546(13.9438) 1.9071(17.8271) 1.8284(0.0225) 17.876(3.5581) 8.7870(0.9752) 16.5364(26.1071) -7.069(21.7371) 17.2677(2.1373) 15.8823(2.323) 8.2869(4.6504) -0.0512(3.2347) 0.2249(0.2847) 5.378(1.5506) 36.1285(17.6488) 23.5131(13.2606) 6.9128(1.5872) -84.4014(59.5649) 38.5135(5.6775) 1.7493(0.7418) 3.8924(1.4171) 2.6557(1.237) 41.0446(1.2586) 2.9995(1.5160)

Table 5.5: Investment Efficiency Estimates Using IVA

105

Table 5.5: Investment Efficiency Estimates Using IVA (continued)
Madagascar Malaysia Mauritius Morocco Nepal Niger Pakistan Panama Papua New Guinea Paraguay Peru Philippines Rwanda Senegal Singapore Sri Lanka Thailand Togo Trinidad &Tobago Uruguay Venezuela Nobs
2

-5.4658(8.2415) 9.6869(10.8412) 14.8768(5.8593) 11.3532(0.8569) -2.4395(0.8408) -15.7091(7.2691) -0.2184(7.4174) n.a. n.a. 5.4260(2.4496) -0.1970(3.1712) 3.3955(1.1552) -1.9850(0.2441) 31.3559(7.9633) 38.3329(6.2111) 21.8508(3.7982) 8.6302(5.9762) 6.9709(11.5213) -3.4754(3.8257) 8.6821(1.6441) -18.9878(12.4996) 367 0.1526 2.1752 0.0000

20.2000(26.441) 13.8038(4.6745) 26.5417(16.8879) 8.0180(2.3918) 0.2222(1.7956) 14.4015(21.1447) 6.6894(13.2983) 24.1498(1.3229) -0.1054(18.9467) 42.2938(16.9868) -5.5300(3.1704) 42.8161(2.8438) -37.5001(59.9429) 22.7896(22.644) 34.4393(7.1699) -0.5164(0.6829) 34.3772(7.3525) 22.4846(17.6556) 7.2363(6.4540) 24.9211(3.8076) 17.8963(4.2571) 432 0.2838 1.9342 0.0000

0.5844(0.6739) 10.2202(0.3020) 5.412(0.4324) 17.2071(11.5815) 8.3668(3.9818) 8.8712(3.6192) 13.6878(4.3302) 22.5887(1.9309) 17.7253(195.0098) 27.6978(8.9578) -4.1787(0.9699) 17.5358(6.0656) 59.6687(21.6652) 7.7239(14.1205) 25.4722(6.9614) 12.021(1.773) 38.1589(0.4704) 28.2358(3.0011) 8.3891(2.4667) -4.1574(3.186) 7.8113(8.104) 380 0.4862 1.7000 0.0000

R DW Prob(F-stat)

Note) βs are estimated from equation (5.1) using OLS. The numbers in the parenthesis are standard deviations.

106

Dependent variable: βGDP Variable Inflation Government size Human Capital Finance Nobs.
R DW Prob(F-statistic)
2

Private credit as a measure of financial development -0.1112(-1.0898) -0.2205(-2.2836) 0.1547(5.0508) 0.4660(1.9767) 155 0.4309 2.5519 0.0000 Dependent variable: βIVA Private credit as a measure of financial development 0.0544(1.0039) -0.5632(-1.7989) 0.3120(3.4340) 1.2073(2.6193) 139 0.3742 2.0201 0.0000

Domestic money bank assets as a measure of financial development -0.0987(-1.4745) -0.3720(-2.2187) 0.1793(4.9848) 0.3563(1.9333) 156 0.4240 2.6866 0.0000 Domestic money bank assets as a measure of financial development 0.0494(1.0010) -0.5954(-1.8243) 0.2242(3.2577) 1.0532(2.2750) 140 0.3703 2.1304 0.0000

Variable Inflation Government size Human Capital Finance Nobs.
R DW Prob(F-statistic)
2

Note: the numbers in the parenthesis are t-statistics. The equation is estimated by white heteroskedasticity-consistent estimator. Table 5.6: Investment Efficiency Regression I

107

Dependent variable: Average product of capital Variable Base regression Private credit as a measure of financial development -0.0449(-3.0064) -0.1995(-1.1103) 0.3729(4.2119) 0.2572(2.004) 103 0.6585 2.2987 0.0000 Domestic money bank assets as a measure of financial development -0.0455(-3.1122) -0.1936(-1.0857) 0.3534(4.0776) 0.2772(1.9767) 103 0.6690 2.3109 0.0000

-0.0359(-2.5863) Inflation Government size -0.1956(-1.0916) 0.4467(5.7401) Human Capital Finance 104 Nobs. 2 0.6355 R DW Prob(F-statistic) 2.2637 0.0000

Note: the numbers in the parenthesis are t-statistics. Average product of capital is the ratio of the average real GDP to the average capital stock for five-year horizon. The equation is estimated by white heteroskedasticity-consistent estimator. Table 5.7: Investment Efficiency Regression II

108

Dependent variable: β (GDP) Variable Inflation Government size Human Capital Finance Assassinations Legal environment Nobs.
R DW Prob(F-statistic)
2

Private credit as a measure of financial development -0.0827(-1.8206) -0.2628(-2.4071) 0.3336(4.0347) 0.3571(1.9877) -0.0174(-1.3651) 155 0.4312 2.5154 0.0000 -0.0531(-1.8395) -0.3083(-2.0627) 0.4316(1.9623) 0.3001(1.0393) -0.0110(-1.1631) 0.3370(0.8740) 119 0.4430 2.4914 0.0006

Domestic money bank assets as a measure of financial development -0.0355(-0.6663) -0.3525(-2.2158) 0.3857(3.5601) 0.2730(1.8211) -0.0126(-0.9256) 156 0.4394 2.5428 0.0000 -0.0256(-1.1360) -0.5227(-2.5030) 0.4184(1.7498) 0.2129(1.7792) -0.0129(-1.5372) 0.3870(1.1297) 120 0.4476 2.5039 0.0008

Dependent variable: β (IVA) Variable Inflation Government size Human Capital Finance Assassinations Legal environment Nobs.
R DW Prob(F-statistic)
2

Private credit as a measure of financial development -0.0372(-1.4044) -1.3864(-1.8963) 0.2549(3.1395) 1.1202(2.1215) -0.1569(-2.0842) 139 0.3809 2.0104 0.0000 -0.0265(-0.9653) -0.9877(-1.8532) 0.2389(3.4301) 1.0888(2.4447) -0.1661(-2.0889) 0.1004(1.1345) 107 0.3775 1.9975 0.0000

Domestic money bank assets as a measure of financial development -0.0261(-1.3968) -1.2503(-1.7640) 0.2221(3.4440) 1.3576(2.0013) -0.1308(-2.4793) 140 0.3782 2.0776 0.0000 -0.0276(-1.2755) -0.7591(-1.7773) 0.2006(3.3205) 1.2589(2.0001) -0.1200(-2.0982) 0.2156(0.9504) 108 0.3848 2.0460 0.0000

Note: the numbers in the parenthesis are t-statistics. The equation is estimated by white heteroskedasticity-consistent estimator. Table 5.8: Investment Efficiency Regression III

109

Dependent variable: Average product of capital Variable Inflation Government size Human Capital Finance Assassinations Legal environment Nobs.
R DW Prob(F-statistic)
2

Private credit as a measure of financial development -0.0217(-3.3231) -0.2416(-4.6977) 0.3357(3.7767) 0.2145(1.9454) -0.0417(-3.9008) 103 0.7152 2.1633 0.0000 -0.0678(-3.1653) -0.1495(-2.7244) 0.2198(2.8352) 0.2598(2.0902) -0.0507(-5.6346) 0.1654(3.1379) 84 0.7154 2.5294 0.0000

Domestic money bank assets as a measure of financial development -0.0224(-3.4635) -0.2347(-4.5103) 0.3300(3.0456) 0.2064(1.8106) -0.0400(-3.8843) 103 0.7597 2.3569 0.0000 -0.0407(-1.7668) -0.1941(-2.9712) 0.2537(2.0419) 0.1163(2.6014) -0.0605(-4.1289) 0.2730(4.9516) 84 0.7701 2.4257 0.0000

Note: the numbers in the parenthesis are t-statistics. Average product of capital is the ratio of the average real GDP to the average capital stock for five-year horizon. The equation is estimated by white heteroskedasticity-consistent estimator. Table 5.9: Investment Efficiency Regression IV

110

CHAPTER 6

Conclusion

The primary part of my dissertation investigates the potential effects of financial sector development on economic growth. In order to reveal the nature of these effects, I focus on the potential channel of influences from the financial to the real sector. The nature of interaction between the real and the financial sectors has been hotly debated among researchers. Those who favor finance-led growth hypothesis argue that the existence of a vibrant financial sector has a growth-enhancing effects. In this literature, an economy can grow faster due to an efficient allocation of resources by the financial sector, mainly banks. A number of channels of influence have been proposed in the literature, which include increased savings, increased investment rate, and efficiency thereof, increased human capital accumulation, and positive effects of the financial sector on innovation processes. Investigations of the validity of these channels as true agents of long-run growth, so far, have yielded mixed empirical results. I investigate the link between the financial sector and economic growth focusing on the role of the financial sector in funding innovative activities. I pursue this goal by constructing a model where the economy is driven by innovative activities that require 111

both human capital and external funding from the financial sector. My analysis shows that when certain conditions are satisfied, there exists a unique equilibrium where the growth rate of the economy is jointly determined by the levels of human capital and financial development. An interesting implication of this is that financial liberalization policies that do not adequately address the fundamentals of the economy can bring about bank failures and possibly a financial crisis. Furthermore, in addition to showing that poverty traps can be explained without introducing setup costs, the model suggests that, depending on the parameter values of the economy, there may be two forms of poverty traps, one with a small number of bankers and the other with a large number of bankers. In addition, I examine empirically whether financial development has any effect on the rate of technological innovation. I test the validity of the finance-led growth hypothesis by focusing on the innovation channel of influence, using patent applications as a proxy for innovative output. Using panel data on twenty eight countries from 1970 to 2000, my analysis shows that financial development is indeed significant in raising the growth rate of innovative output. Lastly, I examine if there is any evidence of financial development positively affecting the efficiency of aggregate investment using developing countries as a sample. I address the issue of efficiency channel using two alternative measures of aggregate investment efficiency. I find that, for developing countries, financial development significantly and positively affects productivity of investment.

112

Appendix A

Data Sources for Chapter 5

Real GDP per capita; Population; Openness as measured by the share of exports plus imports in GDP in constant 1996 prices; Government size as measured by the government share in real GDP in constant 1996 prices Source: Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP), October 2002. Industry value added: Industry corresponds to ISIC divisions 10-45 and includes manufacturing (ISIC divisions 15-37). It comprises value added in mining, manufacturing (also reported as a separate subgroup), construction, electricity, water, and gas. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. The origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 2. Data are in constant 1995 U.S. dollars. Source: World Development Indicator CD-Rom, 2000 Gross fixed capital formation: The total value of a producer’s acquisitions, less disposals, of fixed assets during the accounting period plus certain additions to the value of nonproduced assets realized by the productive activity of institutional units. Fixed assets are tangible or intangible assets produced as outputs from processes of production that are themselves used repeatedly or continuously in other processes of production for more than one year. Data in local currency. Source: International Financial Statistics CD-Rom, 2000 Deposit Money Bank Assets to GDP: Claims on domestic real nonfinancial sector by deposit money banks as a share of GDP, calculated using the following deflation method: (0.5)[Ft / Pet + Ft −1 / Pet −1 ] /[GDPt / Pat ] , where F is deposit money bank claims, P_e is endof period CPI, and P_a is average annual CPI. Raw data are from the electronic version of the IMF's International Financial Statistics (IFS line lines 22, a-d). Data on the deflators is from the electronic version of the IFS (line 64M..ZF or, if not available, line 64Q..ZF) 113

and annual CPI (line 64..ZF). Data on GDP in local currency (lines 99B..ZF or, if not available, line 99B.CZF) Source: http://www.worldbank.org/research/projects/finstructure/database.htm
Private credit by deposit money banks to GDP: Private credit by deposit money banks to GDP, calculated using the deflation method shown above. Raw data are from the electronic version of the IMF's International Financial Statistics (IFS lines 22d). Data on GDP in local currency (lines 99B..ZF or, if not available, line 99B.CZF), end-of period CPI (line 64M..ZF or , if not available, 64Q..ZF), and annual CPI (line 64..ZF) are from the electronic version of the IFS. Source: http://www.worldbank.org/research/projects/finstructure/database.htm Illiteracy rate: Adult illiteracy rate is the percentage of people aged 15 and above who cannot, with understanding, read and write a short, simple statement on their everyday life. Source: World Development Indicator CD-Rom, 2000 Inflation: Inflation as measured by the annual growth rate of the GDP implicit deflator. GDP implicit deflator measures the average annual rate of price change in the economy as a whole for the periods shown. Source: World Development Indicator CD-Rom, 2000 Legal: Legal system and property rights Source: Gwartney, James and Robert Lawson with Walter Park, Smita Wagh, Chris Edwards, and Veronique de Rugy. Economic Freedom of the World: 2002 Annual Report. Vancouver: The Fraser Institute, 2002. Data retrieved from www.freetheworld.com Telephone mainlines: Telephone mainlines are telephone lines connecting a customer's equipment to the public switched telephone network. Data are presented per 1,000 people for the entire country. Source: World Development Indicator CD-Rom, 2000 Assassinations: The number of any politically motivated murder or attempted murder of a high government official or politician Source: Arthur S. Banks Cross National Time-Series Data Archive downloaded from the web site: http://www.worldbank.org/research/growth/GDNdata.htm (2005)

114

Appendix B

Countries Used In The Sample

Countries used in the investment regression (with the real GDP per capita growth as a dependent variable) Algeria(1) Argentina Bolivia Botswana*(1) Brazil(2) Burkina Faso* Burundi Cameroon Chile Colombia Costa Rica Cote d`Ivoire* Cyprus* Dominican Republic Ecuador Egypt El Salvador Fiji* Ghana Guatemala Haiti Honduras India Indonesia(2) Iran Israel Jamaica Jordan(2) Kenya Korea Lesotho*(1) Madagascar* Malaysia Mauritius* Morocco Nepal* Nicaragua(2) Niger Pakistan Panama Papua New Guinea*(1) Paraguay Peru Philippines Rwanda* Senegal Singapore* Sri Lanka Thailand Togo Trinidad &Tobago Uruguay Venezuela

* Countries that are dropped when infrastructure and legal variables are added in the regression. (1) Countries for which the sample period is 1975 – 1998 (2) Countries for which the sample period is 1980 – 1998; For Indonesia, the sample period is 1970 – 1998 when DMBA is used as the financial development indicator. 115

Countries used in the investment regression (with real Industry Value Added per capita growth as a dependent variable) Algeria(1) Argentina Botswana*(1) Brazil(2) Burkina Faso* Burundi Cameroon Chile Colombia Cote d`Ivoire* Dominican Republic Ecuador Egypt El Salvador Fiji* Ghana Guatemala Haiti Honduras India Indonesia(2) Iran(3) Jamaica Jordan(2) Kenya Korea Lesotho*(1) Madagascar* Malaysia Mauritius* Morocco Nepal*(1) Niger Pakistan Panama(2) Paraguay Peru Philippines Rwanda* Senegal Singapore* Sri Lanka Thailand Togo Trinidad &Tobago Uruguay

Papua New Venezuela Guinea*(2) * Countries that are dropped when infrastructure and legal variables are added in the regression. (1) Countries for which the sample period is 1975 – 1998; (2) Countries for which the sample period is 1980 – 1998; For Indonesia, the sample period is 1970 – 1998 when DMBA is used as the financial development indicator. (3) For Iran, the sample period is 1975 – 1994

116

Sample countries used when investment efficiency is measured by Y/K Argentina Bolivia Botswana*(1) Chile Colombia Cote d’Ivoire* Dominican Republic Ecuador Guatemala Honduras India Iran Israel Jamaica Keyna Korea Madagascar* Mauritius* Morocco Nepal* Panama Paraguay Peru Philippines Thailand Venezuela

* Countries that are dropped when legal environment variable is added to the regression. (1) Countries for which the sample period is 1975 – 1989.

117

BIBLIOGRAPHY

Aghion, P. and H. Peter (1999). Endogeneous Growth Theory. London, England, The MIT Press. Aizenman, J. (2005). "Financial liberalizations in Latin America in the 1990s: A reassessment." NBER Working Paper Series(11145). Allen, F. and D. Gale (2001). Comparing Financial Systems. Cambridge, Massachusetts London, England, The MIT Press. Archibugi, D. and P. Mario (1996). "Measuring technological change through patents and innovation surveys." Technovation 16(9): 451-468. Arestis, P. and P. Demetriades (1997). "Financial Development and Economic Growth: Assessing the Evidence." The Economic Journal(107): 783-799. Arestis, P., P.Demetriades, et al. (2002). "The impact of financial liberalization policies on financial development: Evidence from developing economies." WP. Arestis, P. and D. Panicos (2002). "Financial Policies and the Aggregate Productivity of the Capital Stock: Evidence from Developed and Developing Economies." The Levy Economics Institute of Bard College, Working Paper No.362. Badawi, A. (2003). "Private Capital Formation and Public Investment in Sudan: Testing the Substitutability and Complementarity Hypotheses in a Growth Framework." Journal of International Development 15(6): 783-799. Barro, R. J. and L. Jong-Wha (2000). "International data on educational attainment: Updates and Implications." CID Working Paper(42). Beck, T., D.-K. Asli, et al. (1999). "A new database on financial development and structure." World Bank Policy Research Working Paper Series(2146). Beck, T., D.-K. Asli, et al. (2001). "Law, Politics, and Finance." World Bank Policy Research Working Paper Series(2585).

118

Beck, T., D.-K. Asli, et al. (2001). "Legal theories of financial development." Oxford Review of Economic Policy 17(4): 483-501. Beck, T., D.-K. Asli, et al. (2002). "Law, Endowments, and Finance." NBER Working Paper Series(9089). Beck, T. and R. Levine (2002). "Stock Markets, Banks, and Growth: Panel Evidence." NBER Working Paper Series No. 9082. Beck, T. and L. Ross (2000). "New Firm Formation and Industry Growth: Does Having a Market- or Bank-Based System Matter?" World Bank Policy Research Working Paper Series(2383). Beck, T., L. Ross, et al. (2000). "Finance and the Sources of Growth." Journal of Financial Economics 58(1-2): 261-300. Bencivenga, V. R. and D. S. Bruce (1991). "Financial Intermediation and Endogenous Growth." Review of Economic Studies 58: 195-209. Benhabib, J. and M. S. Mark (2000). "The role of financial development in growth and investment." Journal of Economic Growth 5: 341-360. Blomstrom, M., E. L. Robert, et al. (1996). "Is Fixed Investment the Key to Economic Growth?" The Quarterly Journal of Economics 111(1): 269-276. Bond, S., L. Asli, et al. (2004). "Capital Accumulation and Growth: A New Look at the Empirical Evidence." WP. Calderon, C. s. and L. Lin (2003). "The direction of causality between financial development and economic growth." Journal of Development Economics 72: 321-334. Chari, A. and H. Peter Blair (2002). "Capital account liberalization: Allocative efficiency or animal spirits?" NBER Working Paper Series(8908). Christopoulos, D. K. and G. T. Efthymios (2004). "Financial development and economic growth: evidence from panel unit root and cointegration tests." Journal of Development Economics 73: 55-74. Comanor, W. S. and F.M.Scherer (1969). "Patent statistics as a measure of technical change." Journal of Political Economy 77: 392-397. Crosby, M. (2000). "Patents, Innovation and Growth." Economic Record 76(234): 255-262.

119

De Long, J. B. (1992). "Productivity Growth and Machinery Investment: A Long-Run Look, 1870 - 1980." The Journal of Economic History 52(2): 307-324. Deidda, L. and F. Bassam (2002). "Non-linearity between finance and growth." Economics Letters 74: 339-345. Demetriades, P. O. and A. H. Khaled (1996). "Does financial development cause economic growth? Time-series evidence from 16 countries." Journal of Development Economics 51: 387-411. Diamond, D. W. (1984). "Financial intermediation and delegated monitoring." Review of Economic Studies 51: 393-414. Driffill, J. (2003). "Growth and Finance." The Manchester School 71(4, Special Issue): 363-380. Easterly, W. (2002). The Elusive Quest for Growth: Economists' Adventures and Misadventures in the Tropics. Cambridge, Massachusetts; London, England, The MIT Press. Easterly, W. and L. Ross (2001). "What Have We Learned from a Decade of Empirical Research on Growth? It's Not Factor Accumulation: Stylized Facts and Growth Models." World Bank Economic Review 15: 177-219. Easterly, W., L. Ross, et al. (2004). "Aid, Policies, and Growth: Comment." The American Economic Review 94(3): 774-780. Ergungor, O. E. (2004). "Market- vs. bank-based financial systems: Do rights and regulations really matter?" Journal of Banking and Finance 28: 2869-2887. Evans, A. D., C. J. Green, et al. (2002). "Human Capital and Financial Development in Economic Growth: New Evidence Using the Translog Production Function." International Journal of Finance and Economics 7: 123-140. Fatas, A. (2000). "Do business cycles cast long shadows? Short-run persistence and economic growth." Journal of Economic Growth 5: 147-162. Fisman, R. and L. Inessa (2003). "Financial development and the composition of industrial growth." NBER Working Paper Series(9583). Fisman, R. and L. Inessa (2004). "Financial development and growth in the short and long run." NBER Working Paper Series(10236). Fry, M. J. (1995). Money, Interest, And Banking In Economic Development. Baltimore and London, The Johns Hopkins University Press. 120

Fuente, A. d. l. and M. Jos'e Mari'a (1996). "Innovation, bank monitoring, and endogenous financial development." Journal of Monetary Economics 38(269): 301. Galetovic, A. (1996). "Specialization, intermediation, and growth." Journal of Monetary Economics 38: 549-559. Gapinski, J. H. (1996). "Heterogeneous Capital, Economic Growth, and Economic Development." Journal of Macroeconomics 18(4): 561-586. Goldsmith, R. W. (1969). Financial Structure and Development. New Haven, CT, Yale University Press. Greenwood, J. and J. Boyan (1990). "Financial Development, Growth, and the Distribution of Income." The Journal of Political Economy 98(5, Part 1): 1076-1107. Greenwood, J. and B. D. Smith (1997). "Financial markets in development, and the development of financial markets." Journal of Economic Dynamics and Control 21: 145-181. Gregorio, J. D. and E. G. Pable (1995). "Financial development and economic growth." World Development 23(3): 433-448. Griliches, Z. (1990). "Patent statistics as economic indicators: A survey." Journal of Economic Literature 28(4): 1661-1707. Gwartney, J. D., R. A. Lawson, et al. (1999). "Economic Freedom and the Environment for Economic Growth." Journal of Institutional and Theoretical Economics 155(4): 643-663. Gylfason, T. (2004). "To grow or not to grow: Why institutions must make a difference." CESifo Dice Report, Journal for Institutional Comparisons 2(2): 3-15. Hamilton, J. D. and J. Monteagudo (1998). "The Augmented Solow Model and the Productivity Slowdown." Journal of Monetary Economics 42(3): 495-509. Hansen, H. (2000). "Aid Effectiveness Disputed." Journal of International Development 12: 375-398. Harrison, P., O. Sussman, et al. (1999). "Finance and Growth: Theory and New Evidence." WP. Hayami, Y. and O. Junichi (1999). "Changes in the Sources of Modern Economic Growth: Japan Compared with the United States." Journal of the Japanese and International Economies 13(1): 1-21.

121

Hendriks, L. (2000). "Equipment investment and growth in developing countries." Journal of Development Economics 61: 335-364. Hirschey, M. and J. R. ernon (2004). "Are scientific indicators of patent quality useful to investors?" Journal of Empirical Finance 11: 91-107. Jacoby, H. G. and E. Skoufias (1997). "Risk, Financial Markets, and Human Capital in a Developing Country." The Review of Economic Studies 64(3): 311-335. Johnson, D. K. N. and D. Popp (2003). "Forced Out of the Closet: The Impact of the American Inventors Protection Act on the Timing of Patent Disclosure." Rand Journal of Economics 34(1): 96-112. Judson, R. (1998). "Economic Growth and Investment in Education: How Allocation Matters." Journal of Economic Growth 3: 337-359. Jun, Z. (2003). "Investment, Investment Efficiency, and Economic Growth in China." Journal of Asian Economics 14(5): 713-734. Khan, M. S. and S. S. Abdelhak (2000). "Financial development and economic growth: An overview." IMF Working Papers(00/209). King, R. G. and R. Levine (1993). "Finance, Entrepreneurship, and Growth: Theory and Evidence." Journal of Monetary Economics 32(3): 513-542. King, R. G. and R. Levine (1994). "Capital Fundamentalism, Economic Development, and Economic Growth." Carnegie-Rochester Conference Series on Public Policy 40(0): 259-292. King, R. G. and L. Ross (1993). "Finance and Growth: Schumpeter Might Be Right." The Quarterly Journal of Economics 108(3): 717-737. Lee, J.-W. (2005). "Human capital and productivity for Korea's sustained economic growth." Journal of Asian Economics 16: 663-687. Levine, R. (1997). "Financial development and economic growth: Views and agenda." Journal of Economic Literature 35: 688-726. Levine, R. (1998). "The Legal Environment, Banks, and Long-run Economic Growth." Journal of Money, Credit and Banking 30(3): 596-613. Levine, R. (1999). "Law, Finance, and Economic Growth." Journal of Financial Intermediation 8(1-2): 8-35.

122

Lopez, R. (1995). "Synergy and Investment Efficiency Effects of Trade and Labor Market Distortions." European Economic Review 39(7): 1321-1344. Luintel, K. B. and K. Mosahid (1999). "A quantitative reassessment of the finance-growth nexus: evidence from a multivariate VAR." Journal of Development Economics 60: 381-405. Madsen, J. B. (2002). "The Causality between investment and economic growth." Economics Letters 74: 157-163. Mankiw, N. G., D. Romer, et al. (1992). "A Contribution to the Empirics of Economic Growth." The Quarterly Journal of Economics 107(2): 407-437. McKinnon, R. I. (1973). Money and Capital in Economic Development. Washington, DC., Brookings Institute. Ndikumana, L. o. (2005). "Financial development, financial structure, and domestic investment: International evidence." Journal of International Money and Finance 24: 651-673. Odedokun, M. O. (1996). "International evidence on the effects of directed credit programmes on efficiency of resource allocation in developing countries: The case of development bank lendings." Journal of Development Economics 48: 449-460. Outreville, J. F. (1999). "Financial development, human capital and political instability." UNCTAD Discussion Papers(142). Popp, D., T. Juhl, et al. (2004). "Time In Purgatory: Examining the Grant Lag for U.S. Patent Applications." Topics in Economic Analysis & Policy 4(1). Porter, M. E. and S. Scott (2000). "Measuring the "ideas" production function: Evidence from international patent output." NBER Working Paper Series(7891). Rioja, F. and V. Neven (2004). "Finance and the sources of growth at various stages of economic development." Economic Inquiry 42(1): 127-140. Romer, P. M. (1990). "Endogenous Technological Change." The Journal of Political Economy 98(5): S71-S102. Ross, L. (2001). "International financial liberalization and economic growth." Review of International Economics 9(4): 688-702. Rousseau, P. L. and V. Dadanee (2005). "Finance, investment, and growth: Time series evidence from 10 Asian economies." Journal of Macroeconomics 27: 87-106.

123

Rousseau, P. L. and R. Sylla (1999). "Emerging Financial Markets and Early U.S. Growth." NBER Working Paper Series No. 7448. Saint-Paul, G. (1992). "Technological choice, financial markets, and economic development." European Economic Review 36: 763-781. Saint-Paul, G. (1993). "Productivity growth and the structure of the business cycle." European Economic Review 37: 861-883. Sala-I-Martin, X. (1997). "I Just Ran Two Million Regressions." The American Economic Review 87: 178-183. Savvides, A. (1995). "Economic Growth in Africa." World Development 23(3): 449-458. Shan, J. Z., G. M. Alan, et al. (2001). "Financial development and economic growth: An egg-and-chicken problem?" Review of International Economics 9(3): 443-454. Shaw, E. S. (1973). Financial Deepening in Economic Development. New York, Oxford University Press. Stern, S., E. P. Michael, et al. (2000). "The determinants of national innovative capacity." NBER Working Paper Series(7876). Tadesse, S. (2005). "Financial development and technology." William Davidson Institute Working Paper(749). Toh, M. H. and N. Wai Choong (2002). "Efficiency of Investments in Asian Economies: Has Singapore Over-Invested." Journal of Asian Economics 13(1): 52-71. Tsuru, K. (2000). "Finance and Growth: Some theoretical considerations, and a review of the empirical literature." OECD Economics Department Working Paper(228). Wachtel, P. (2001). "Growth and Finance: What do we know and how do we know it?" International Finance 4(3): 335-362. Wαlde, K. and W. Ulrich (2004). "R&D expenditure in G7 countries and the implications for endogenous fluctuations and growth." Economics Letters 82: 91-97. Wurgler, J. (2000). "Financial markets and the allocation of capital." Journal of Financial Economics 58: 187-214. Xu, Z. (2000). "Financial development, investment, and economic growth." Economic Inquiry 38(2): 331-344.

124

Yamano, N. and T. Ohkawara (2000). "The Regional Allocation of Public Investment: Efficiency or Equity"." Journal of Regional Science 40(2): 205-229.

125

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close