THE ECONOMIC IMPACT OF FINANCIAL DEVELOPMENT

THE ECONOMIC IMPACT OF FINANCIAL DEVELOPMENT IN DIFFERENT REGIONS OF KAZAKHSTAN A Thesis submitted to the Graduate School of Arts and Sciences at Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy By Dinara Berdigulova Washington, DC April 15, 2010

THE ECONOMIC IMPACT OF FINANCIAL DEVELOPMENT IN DIFFERENT REGIONS OF KAZAKHSTAN Dinara Berdigulova Thesis Advisor: Luis F. Brunstein, PhD ABSTRACT This thesis investigates the economic impact of the financial development in different regions in Kazakhstan. Data from the Agency of Statistics of the Republic of Kazakhstan was used to construct a sample of regional economic indicators for the period 2003-2008. The analysis estimates the effect of loans issued to the private sector by the banks on GDP per capita in 16 regions divided in groups by main types of economic activity. Results suggest that loans on average have a significant and positive effect on GDP per capita growth, especially in high-income oil and gas producing regions. Moreover, loans also have a significant and positive effect on the level of GDP per capita in agricultural regions and other regions with different types of activities. However, the effect on GDP per capita in oil and gas regions as well as regions with developed mining industries is insignificant. These findings suggest that the Government of Kazakhstan should pursue policies that increase the availability of bank loans in low-income areas. This will help to overcome the disparity in economic growth among regions with different types of economic activity. ii

I want to thank my advisor, Dr. Luis Brunstein, for his great guidance, wise advice and incredible patience, and my husband for his understanding and support. iii

TABLE OF CONTENTS Chapter 1. Introduction... 1 Literature Review... 4 Conceptual Framework and Hypothesis... 8 Data and Methods... 9 Data... 9 Analysis plan... 12 Chapter 2. Results... 16 Descriptive results... 16 Regression results... 19 Discussion... 27 Conclusion... 29 References... 30 iv

Chapter 1. Introduction After reading Paul Collier s book The Bottom Billion: Why the Poorest Countries Are Failing and What Can Be Done About It, I became concerned with the future of the poorest countries. Collier does not list these bottom billion countries but he states that they are located in Africa and Central Asia. In his opinion, globalization does not bring these countries to the well-trodden track of development. On the contrary, their sharp economic disparity with the rest of the world isolates them from development. Collier refers to four so-called traps, which are basic obstacles to growth: the conflict trap, the natural resources trap, bad governance, and being landlocked with other poor countries. He stresses that these traps keep countries in a ghetto of misery and discontent. However, each of the trapped countries has a potential to overcome these barriers through a variety of means but especially the development of effective financial institutions. The financial development of the country is an important factor in today s globalized world. An example is Kazakhstan, a landlocked Central Asian country with a natural resources trap. After independence in 1991, Kazakhstan was one of the poorest countries in the world. Today it is a rapidly developing country with the annual GDP growth around 9% for the period 2000-2007 according to the World Bank data 1. 1 World Bank Data and Statistics for Kazakhstan. http://go.worldbank.org/kzlz6olgt0 1

My research is an attempt to estimate the role of financial intermediation in the regional development of the country. Since Kazakhstan is a large country with diverse economic activities in its regions, I think it is a perfect example for analysis. Based on results of previous crosscountry studies of financial systems I believe that financial policy will have the most significant effect in the poorest regions of the country. If this is true, then financial policy can also play an important role in promoting the economic growth of the neediest states. Moreover, with intensification of the globalization processes, the effect of financial development may become even stronger. The purpose of this paper is to examine the empirical relationship between financial development and economic expansion in 16 different regions of Kazakhstan 2. Kazakhstan is the ninth largest country in the world with a vast territory divided by climatic conditions, availability of natural resources, density of population and types of economic activity. Its history as an independent state with a market oriented economy and financial liberalization started in 1992. This paper analyzes the impact of financial intermediation on different regions of the country. For the purposes of this analysis, the country s 16 oblasts (regions) are divided into 4 groups based on the main characteristics of their economic activity: agricultural regions, oil and gas regions, mining industry regions, and others. 2 Official territorial division of Kazakhstan according to the Agency of Statistics of the Republic of Kazakhstan. www.stat.kz 2

The analysis uses Barro s (1991) model of cross-country growth updated by De Gregorio and Guidotti (1995) with another explanatory variable measuring the ratio of bank credit to the private sector to GDP. The paper tries to explain the different growth effect of financial intermediation on regions of the country specializing in different industries. As such, the research is intended to fill a gap in earlier research, which does not establish the direction of the relationship between the financial development, investment, and output growth in various sectors of the economy. 3

Literature Review Numerous studies have addressed the issue of the empirical relationship between the economic growth and financial development. For example, research by Goldsmith (1969), McKinnon (1973) and Shaw (1973) indicates a strong positive correlation between the level of financial development and economic output. These studies focus on different sides of the relationship: Goldsmith on the relationship between financial development and efficiency of investment, and McKinnon and Shaw on the role of financial liberalization in increasing savings. However, some economists believe that finance is a relatively unimportant factor in economic development. Robinson (1952) concluded that financial development simply follows economic growth. And Lucas (1988) finds the relationship between financial and economic development "over-stressed." In the 1990s, a new group of studies on the relationship between financial development and growth has received attention. In an attempt to analyze formally the interactions between financial markets and long-run economic growth, these works have concentrated on the role of financial factors in models of endogenous growth. Greenwood and Jovanovic (1990) present a model in which both financial intermediation and growth are endogenous. Their research supports the idea of a positive two-way causal relationship between economic growth and financial development. Growth stimulates individuals and companies to participate in financial markets, which promotes the development and expansion of the financial system. 4

Conversely, financial development allows potential investors to introduce projects more efficiently, which stimulates growth. Bencivenga and Smith (1991) developed a model, which implies that the development of financial intermediation will increase real growth rates. They suggest that the presence of financial intermediation reduces the risk associated with the liquidity needs, which increases economic growth by directing savings into activities with high productivity returns. They argue that although individuals have uncertain liquidity needs, banks face a predictable demand for liquidity and can, therefore, allocate investment funds more efficiently. King and Levine (1993a, 1993b) use various financial indicators to show that there is a positive relationship between financial indicators and growth, and that financial development is correlated with subsequent rates of growth, capital accumulation, and economic efficiency. As the indicators of the level of financial development, they used the size of the formal financial intermediary sector relative to GDP, the importance of private banks relative to the central bank, the percentage of credit allocated to private firms, and the ratio of credit issued to private firms to GDP. Their results are uniform and unambiguous: financial indicators are strongly and robustly correlated with growth, the rate of physical capital accumulation, and improvements in the efficiency of capital allocation. Greenwood and Smith (1994) suggest that financial development has a dual effect on economic growth. They suggest that development of the financial markets of country may enhance the efficiency of capital accumulation. In addition, financial 5

intermediation can contribute to raising the savings rate and, thus, the investment rate. The former effect was first emphasized by Goldsmith (1969), who also found some positive correlation between financial development and the level of real per capita GNP. De Gregorio and Guidotti (1995) also find that financial development leads to improved economic performance; however, this effect differs across countries and over time. Moreover, they note that Latin American experience in the 1970s and 1980s show instances when unregulated financial liberalization and expectations of government bailouts can led to a negative relationship between the degree of financial development and growth. The model in the present study is based on Barro s (1991) model of crosscountry growth updated by De Gregorio and Guidotti (1995) with another explanatory variable measuring the ratio of bank credit to the private sector to GDP. In contrast to De Gregorio and Guidotti (1995) the McKinnon-Shaw hypothesis suggests that the level of financial intermediation should be closely related to the prevailing level of real interest rates. According to their view, a positive real interest rate stimulates financial savings increasing the supply of credit to the private sector. This, in turn, stimulates investment and growth. However, it seems likely that in Kazakhstan extremely high interest rates do not reflect improved efficiency of investment but rather various forms of country risk. Therefore, a better indicator of the importance of financial intermediaries may be the indicator introduced by De Gregorio and Guidotti (1995) that measures credit granted to the private sector. 6

A special role in the model is attributed to human capital. According to Romer (1990), human capital is the key input to the research sector, which generate the new products and ideas that underlie technological progress. So, regions with higher accumulations of human capital tend to grow faster. This paper uses the level of education coverage as the indicator of human capital. Other variables, such as the index of assassinations and the number of coups and revolutions, used by Barro (1991) as a measure of political stability are irrelevant to Kazakhstan because there were no coups or revolutions during the history of the independent country. Rather, model will use the level of criminality to measure favorability of conditions for business development. 7

Conceptual Framework and Hypothesis My hypothesis is that financial intermediation leads to increased growth. However, intermediation has different effects on the growth rates in different regions of the country. I would expect that financial development within the banking sector will have a larger effect in nonindustrial regions. For the purposes of this paper, Kazakhstan s 16 oblasts (regions) are divided into 4 groups based on the main characteristics of their economic activity: agricultural regions, oil and gas regions, mining industry regions, and regions, where GDP is comprised of other types of economic activity such as construction or services. Mining and oil and gas regions already experience the majority of the foreign investment and government spending for the development infrastructure. So, the effect of the financial development within the banking sector should be stronger in agricultural and other regions. Since, for Kazakhstan, agricultural and other regions except for two major cities Astana and Almaty are low-income, we may suppose that financial development of the banking sector has a stronger effect on them than on high-income regions. The effect of financial development in the non-banking sector is hard to measure. But referring to De Gregorio and Guidotti (1995) it is likely that the two forms of financial development, bank and nonbank, must be positively correlated. So, the presence of financial development outside the banking system will show up in the form of a smaller coefficient of the indicator on the banking system. 8

Data and Methods Data My analysis uses a data set prepared by the Agency of Statistics of the Republic of Kazakhstan and published in Regions of Kazakhstan in 2007 and 2008. This statistical collection contains data about social and economic development of the country for the period of 2003-2008. The data set covers the whole population for a majority of the indicators. For example, according to the Agency of Statistics of Kazakhstan population of the country s regions is measured through a national census conducted every 10 years 3. However, the number of people in each region is updated annually by adding the number of newborn and number of people, who moved to the region minus those who left the region or died 4. The Agency of Statistics conducts annual surveys of households to get updated information about the situation in the regions. These surveys use a random sample of 12,000 households to get information about level of income, consumption, employment, level of education, standard of living, and other information. Specialists in regional departments of statistics conduct interviews in households. Households in the sample cannot be replaced during the year, and, unfortunately, the data set does not provide information about the response rate or rate of attrition. 3 Agency of Statistics of the Republic of Kazakhstan. www.stat.kz 4 Methodology used by the Agency of Statistics. http://www.stat.kz/digital/naselsenie/pages/default.aspx 9

The Agency of Statistics estimates Gross Domestic Product (GDP) for regions based on the System of National Accounting 5. However, specialists at the agency note that it is difficult to construct such estimates. Even though regions of Kazakhstan represent more or less independent economic systems, it is impossible to consider each region as an independent country since ties between regions are extremely close. Specialists use the product approach as the basic method to calculate the amount of production of goods. However, they complement this method with the income approach to calculate services produced in the region. In the calculation of regional GDP, specialists do not use expenditure method because they do not have sufficient data at the regional level to make estimates using this method. In addition to conducting national and regional surveys, the Agency of Statistics collects information from other governmental agencies. For example, the amount of government spending and foreign investment is collected from the local government representatives. Local police departments provide information about the level of criminality in different regions to the Committee of the Criminal Statistics of the Prosecutor General s Office, which then shares it with the Agency of Statistics. The present study also uses data available on the website of the National Bank of Kazakhstan on the amount of loans issued by banks at all levels to the private sector for the period 2004-2008. 5 Methodology used by the specialists of the Agency of Statistics. http://www.stat.kz/metodologiya/standart_arks/pages/metodolog.aspx 10

The main limitation of my data is that a majority of indicators at a regional level are available only for the last five years. Since Barro s model measures long-run growth of per capita GDP, a five year data set may seem not enough for such analysis. However, for a country such as Kazakhstan, with a total of 17 years of independent history, analyzing a recent five-year period may provide substantial information on economic development. All the observations were changed to the real terms using a GDP deflator. 11

Analysis plan My analysis addresses the question of whether financial development leads to improved growth performance in the Kazakhstani regions and if this relationship has significant effect on the economy of its various regions. Therefore, I will focus on the whole population of the country divided by place of residence. My model is based on Barro s (1991) model of cross-country growth, updated by De Gregorio and Guidotti (1995), who included an explanatory variable measuring the ratio of bank credit to the private sector to GDP. The model is an ordinary least square (OLS) regression measuring the impact of a number of different variables on economic growth. Barro also used OLS regressions to estimate the determinants of per capita GDP growth. De Gregorio and Guidotti (1995) use the ratio of domestic credit to the private sector to GDP as a proxy for the degree of financial intermediation. They focus specifically on the amount of credit granted to the private sector by the central bank and commercial banks as a fraction of GDP, and they refer to this variable as Credit. De Gregorio and Guidotti prefer Credit to other monetary aggregates because it excludes credit to the public sector, and provides an accurate measure of the role of financial intermediaries in passing funds to the private sector. Therefore, Credit may be a definition of financial intermediation that is closely related to the level and efficiency of investment, and to economic growth. Consistent with the model estimated by De Gregorio and Guidotti (1995), my OLS model is as following: 12

GDP per capita growth = Credit + GDP per capita + Investment rate + Government Spending + Foreign investment + Level of Criminality + Education coverage + Constant, where GDP per capita growth is the numeric value of an increase in real domestic product per capita, Credit is the ratio of domestic credit to the private sector to GDP, provided by the National Bank, GDP per capita is a numeric value of the real Gross Domestic Product per capita calculated for each region by the Agency of Statistics, Investment rate is the amount of the nominal investment divided by nominal GDP, Government spending is the real amount of government funds allocated to different projects in the region to real GDP, Foreign Investment is the real amount of foreign funds invested into a region to real GDP, Level of criminality is measured by the ratio of the crimes to the number of people in the region, provided by the Agency of Statistics, and Education coverage is the percent of residents from age of 6 to 24 covered by education. GDP per capita growth is an acknowledged measurement of the rate of growth in each country. In my analysis GDP per capita growth, measured as a rate of increase of 13

GDP per capita, is used for each region as a dependent variable. GDP per capita is an approximation of the value of goods produced per person in each region, measured in local currency. Credit is my main explanatory variable, measuring the level of financial intermediation in each region. Education coverage is my measurement of the quality of the human capital in the region. Investment rate is a measurement of the rate of return on investments in each region, which is important for analysis of the investment climate of the region. Government spending measures the amount of government projects in the region, which in the majority of earlier research did not have a significant impact on growth. Still, my model includes this variable to see if the effect of government spending differs from region to region. Foreign investment represents the amount of foreign funds invested in a region. Though foreign investment is not a part of the original Barro model, it is included in the present model to observe effect of foreign investment on economic growth. Kazakhstan s economy, to large extent, depends on its oil sector. The majority of foreign investment is channeled to the country through the oil industry. Thus, measuring the effect of foreign investment on the overall economic growth produces interesting results. All my economic and financial indicators are measured in local currency in order to escape errors related to exchange rate conversion. Other variables such as the index of assassinations and the number of coups and revolutions as a measure of political stability used by Barro (1991) are irrelevant to Kazakhstan. However, my model uses 14

the level of criminality to measure the favorability of conditions for development and growth. And, finally, for the purposes of this paper, Kazakhstan s 16 oblasts (regions) are divided into 4 groups based on the main characteristics of their economic activity: agricultural regions, oil and gas regions, mining industry regions, and other regions. Two major cities, Almaty and Astana, considered as separate oblasts by the Agency of Statistics, are included in the group of other regions. The mining and oil and gas regions already receive the majority of government spending for the development of infrastructure and get all the foreign investment. So, the effect of financial development within the banking sector is likely to be stronger in agricultural and other regions. Also, the fact that in Kazakhstan agricultural and other regions are generally low-income, it is a second reason why we may predict that the financial development of the banking sector has a stronger effect for them rather than for high-income regions with developed oil and gas and mining industries. So, the model helps to estimate the effect of financial intermediation on the increase of economic output per capita growth in each regional group. Additionally, it is used to estimate the effect of Credit on economic output, using GDP per capita as an explanatory variable. The data are available only for five years, and this may not be a long enough time period to catch the effect of certain measures on the growth variable, but it is helpful to estimate the effect of the independent variables on the output per person. 15

Chapter 2. Results Descriptive results The data used in the analysis cover all 16 regions of the country for the period 2003-2008. Table 1 presents general summary statistics of the data. Table 1: Summary Statistics Variable Observations Mean St. deviation Minimum Maximum GDP per capita growth GDP per capita 80 0.273 0.276-0.449 1.632 96 586.35 491.75 86.48 2437.55 Credit 96 0.230 0.298 0.007 1.728 Investment rate Government spending rate Foreign investment rate 96 0.269 0.151 0.102 0.883 96 0.042 0.039 0.009 0.179 95 0.053 0.095 0 0.514 Criminality 96 90.06 29.08 42 164 Education coverage 96 80.17 15.79 60 137.9 16

Table 2 presents the correlation matrix of the variables in the model. Table 2: Correlation Matrix Variables GDP per capita Credit Investment rate Government spending Foreign Investment Crimin ality Educatio n coverage GDP per capita 1 Credit 0.510 1 Investment rate 0.434-0.163 1 Government spending rate 0.011-0.012 0.187 1 Foreign Investment rate 0.348-0.180 0.775-0.150 1 Criminality 0.054 0.363-0.201-0.213-0.153 1 Education coverage 0.474 0.525-0.039 0.016-0.063 0.277 1 17

Since my sample size is relatively small there is a possibility of multicollinearity, which can result in poor OLS estimators. The correlation matrix indicates a high correlation between investment rate and foreign investment (0.775), which makes sense because foreign investment is closely related to investment rate. Foreign investment is included in the model to see its effect on the economic growth. Another high positive correlation can be seen between the main explanatory variable, Credit, and the level of education coverage (0.525). The relationship between these indicators is not so straightforward, but it is quite possible that in regions with higher education coverage the levels of salaries are higher, which increases accessibility to loans. This would make the ratio of the amount of loans issued to GDP higher in these regions than in regions with lower education coverage. In an effort to reduce multicollinearity I could have dropped one of two highly correlated independent variables from the model. However, dropping a variable that belongs in the population model can lead to bias, so we have to look very cautiously at the change in the regression results. 18

Regression results Tables 3 and 4 present regression estimates of the effect of the independent variables in the model on economic growth in Kazakhstan and every regional group. Table 3 shows the results of three regressions for Kazakhstan as a whole. Table 3: Results of models (1), (2), (3) Estimates of the Effects of Key Variables on GDP per capita growth in Kazakhstan Regression (1) (2) (3) Credit 0.417* (0.246) 0.417* (0.232) 0.387* (0.226) GDP per capita -0.000017 (0.0001) -0.000017 (0.00008) -0.000028 (0.00008) Investment rate 0.349 (0.542) 0.349 (0.509) 0.041 (0.231) Government spending rate Foreign Investment rate Criminality Education coverage Intercept -0.662 (1.144) -0.573 (0.794) -0.00079 (0.0012) -0.0067 (0.0051) 0.743** (0.313) -0.662 (1.424) -0.573 (0.893) -0.00079 (0.0013) -0.0067* (0.0045) 0.743** (0.305) -0.186 (1.088) -0.0008 (0.0012) -0.0056 (0.0042) 0.702** (0.299) R2 0.06 0.06 0.05 Number of observations 79 79 80 *p<0.1, **p<0.05, ***p<0.01 19

Table 4 shows the results of four regressions for the four major regional groups in my analysis. Table 4: Results of models (5), (6), (7), (8) Estimates of the Effects of Key Variables on GDP per capita growth in Kazakhstan s Regions Regression 5 (oil&gas regions) 6 (mining regions) 7 (agricultural regions) 8 (other regions) Credit 2.178*** (0.761) -0.769 (1.862) 0.146 (0.346) 0.151 (0.511) GDP per capita 0.000094 (0.00015) 0.0013** (0.00021) 0.00078*** (0.00025) 0.00024 (0.00017) Investment rate -0.115 (0.625) 6.675 (3.295) -0.721 (1.835) 0.063 (1.770) Government spending -4.307 (5.008) -66.175* (25.825) 1.236 (2.780) -1.984 (3.827) Foreign Investment 0.731 (1.056) -7.213* (2.917) 7.591** (3.650) -0.721 (2.244) Criminality 0.006 (0.0047) 0.0015 (0.0014) -0.0018 (0.0013) 0.0096* (0.0059) Education coverage 0.056*** (0.015) -0.0032 (0.026) 0.0068 (0.011) -0.0188 (0.0133) Intercept -5.275*** (1.250) -0.0958 (2.031) -0.2897 (0.941) 0.9755* (0.622) R2 0.53 0.90 0.62 0.46 Number of observations 20 10 30 19 *p<0.1, **p<0.05, ***p<0.01 20

Table 3, regression (1) shows the results of a simple regression based on the observations for the whole country. The results show that the growth of GDP per capita is positively correlated with Credit and that this relationship is statistically significant at 90% confidence level. In other words one unit increase in Credit is predicted to increase per capita growth by 0.417 points, on average. The results support the findings of King and Levine (1993b) and De Gregorio and Guidotti (1995) that financial intermediation by the banking system has a positive effect on GDP per capita. The model predicts that the flow of credit increases growth in per capita output. It supports the hypothesis that a developed credit system stimulates productivity and output growth of businesses promoting economic growth. Other independent variables in the model are statistically insignificant. The signs of the coefficients of these variables are in line with those found by Barro (1991). However, the magnitudes are different, as was expected. Differences in definition and measurement of the observations employed may explain these divergent findings. Probably, the availability of data for the limited time period also affected the significance of the results. Barro (1990) found a negative relationship between government expenditure and growth. His argument was that government consumption had no direct effect on private productivity, but lowered saving and growth through the distorting effects of taxation or government-expenditure programs. In case of Kazakhstan his finding of the negative effect of government spending on the GDP per capita growth seems justified. 21

Since heteroskedasticity can be an issue across regions, Table 3 regression (2) presents standard errors for the coefficients of a second regression computed using robustness test. And in all the following regressions robust standard errors are presented in the tables. These standard errors differ from those obtained by simple ordinary least squares regression. However, they do not change the statistical significance of the previous results. Only the coefficient on education coverage becomes significant at 90% confidence level. This finding supports and simultaneously contradicts the neoclassical growth theory that human capital has a positive effect on the growth since its increase positively effects total productivity. According to Barro (1990) the level of school enrollment has a positive effect on the growth rate. In the present model, however, education coverage has significant effect on the economic growth, but education coverage is negatively correlated with the GDP per capita growth. This paradox can be explained by referring to the history of Kazakhstan. Being part of the Soviet Union, the country had a very high rate of literacy and education coverage not typical for a developing country. After gaining independence and switching to market economy, education coverage has decreased because access to free higher education became limited. However, measures undertaken by the government on economy liberization and introduction of private ownership along with development of the private sector had a strong effect on productivity growth. Most likely, when the economy comes to the normal condition of development after overcoming the shock of switching from a planned to a market economy, then education 22

coverage as the important indicator of the human capital development will become positively correlated with output growth. According to the correlation matrix, the investment rate and foreign investment variables are highly correlated. To prevent possible multicollinearity, therefore, in regression (3), the foreign investment variable has been excluded. However, the results show no change in the significance of the explanatory variables. Table 4 shows results for regressions conducted on four groups of regions divided by types of economic activity. The main dependent variable of interest is significant at 99% confidence level in regions with developed oil and gas industries. The sign of the coefficient on Credit is as expected supporting the idea that a developed credit system has positive effect on GDP per capita growth. However, the results do not support the main hypothesis that Credit has a significant effect in low-income regions. So, these results may suggest that financial development in high-income areas, which is true for the oil and gas regions, is at the stage of increasing returns. It implies that the oil and gas regions, in comparison with other regions, have reached that stage of infrastructure development when they can turn credit inflow to output growth. Other regions, most likely, lag behind in their output growth capacity. 23

The previous analysis shows that Credit has a positive and significant effect on GDP per capita growth in Kazakhstan as a whole, and, specifically, that it has a strongly positive and significant effect in high-income regions with developed oil and gas industries. The following regressions will use the GDP per capita rather than GDP per capita growth as a dependent variable. Tables 5 and 6 present the results of the regressions estimating the effect of Credit on GDP per capita on the whole country and on regional groups respectively. Table 5: Results of models (9), (10), (11), (12) Estimates of the Effects of Key Variables on GDP per capita in Kazakhstan Regression (9) (10) (11) (12) Credit Investment rate Government spending rate Foreign Investment rate Criminality Education coverage Intercept R2 Number of observations 1130.93*** (168.36) 1742.78*** (445.40) -1164.25 (1071.95) 183.19 (675.06) -1.388 (1.348) -2.647 (3.133) 1130.93*** (172.50) 1742.78*** (548.23) -1164.25 (1304.52) 183.19 (1116.8) -1.388 (1.315) -2.647 (2.869) 1111.2*** (170.86) 1861.03*** (336.19) -1261.26 (1110.98) -1.180 (1.265) -2.519 (2.798) 983.39*** (161.86) 821.60 (1208.74) 2327.90*** (873.51) -1.530 (1.457) 1.354 (2.656) 238.83 238.83 192.63 236.59 (245.14) (239.96) (233.15) (249.67) 0.55 0.55 0.55 0.47 95 95 96 95 *p<0.1, **p<0.05, ***p<0.01 24

Table 6: Results of models (13), (14), (15), (16) Estimates of the Effects of Key Variables on GDP per capita in Kazakhstan s Regions Regression (13) (oil&gas regions) (14) (mining regions) (15) (agricultural regions) (16) (other regions) Credit -325.595 (463.567) 1038.18 (1316.9) 638.0* (385.98) 1655.03*** (353.47) Investment rate 1684.77*** (548.195) -4118.37 (2502.34) -121.09 (657.52) 1380.86 (1177.54) Government spending -6927.78* (4080.17) 37176.54* (20541.89) 998.99 (921.29) 2909.76 (2132.55) Foreign Investment -495.31 (402.28) 2442.39 (3522.29) 717.66 (1550.35) -3160.65** (1474.13) Criminality -16.02*** (3.21) -4.465* (2.340) 0.108 (1.041) -4.704 (3.274) Education coverage -42.61*** (8.388) -10.66 (11.59) -4.045** (1.745) -9.571 (7.664) Intercept 5409.58*** (763.87) 1774.11 (1271.74) 439.24* (266.72) 751.59* (458.78) R 2 0.81 0.89 0.28 0.80 Number of observations 24 12 36 23 *p<0.1, **p<0.05, ***p<0.01 Regression (10) presents results with robust standard errors. As was expected, Credit has significant positive effect on overall economic output. In regressions (11) and (12), two highly correlated explanatory variables, Investment rate and Foreign investment rate are dropped in turn. The results show that only investment has a 25

significantly positive effect on GDP per capita at the 99% confidence level. Other variables including government spending do not have significant effects on output. The regional analysis of the effect of financial intermediation on GDP per capita in Table 6 indicates that this effect is statistically significant in agricultural and other regions, and does not have significant effect in regions with developed oil and gas and mining industries. This important finding supports my hypothesis that in high income oil and gas and mining regions, attracting the majority of foreign investment, credit does not have significant effect on the GDP per capita. However, low-income agricultural regions and other regions, as well as the two most important cities Astana and Almaty (Kazakhstan s capital and financial center respectively) Credit has significant effect on GDP per capita. The strong positive correlation between Credit and GDP per capita in these regions suggest the importance of financial intermediation in low income regions. The variable Criminality is statistically significant in two groups of regions: the oil and gas and mining regions. This variable was used to measure political instability instead of the two variables used by Barro (1990): revolutions and coups per year and an index of assassinations. Barro interpreted these two variables as adverse influences on property rights, and thereby as negative influences on investment and growth. Kazakhstan is a politically stable country and the number of revolutions or coups for its 19 years of independent history is zero. Therefore, I used the level of criminality as a measure of stability in the region. The finding that Criminality is significant in these two groups of regions makes sense. In these regions the disparity of income is higher. Thus, criminality has deeper roots. 26

Discussion The empirical analysis suggests that the financial intermediation measured as the ratio of the amount of loans issued to the private sector to GDP has a significant effect on GDP per capita growth. Moreover, the analysis shows that Credit has significant effect on growth in regions with developed oil and gas industries. Most likely, other regions, which have lower levels of income, do not have the capacity to transform credit flow into output growth. Analysis of the effect of Credit on GDP per capita indicates an important effect of financial development in the bank sector on the agricultural regions and other regions that do not have developed oil, gas and mining industries. Probably because these regions do not experience large financial inflows from government spending or foreign investments, loan funds are extremely important for their development. This finding implies that financial development plays an important role in the economic development of less developed regions. The results of this study suggest that Kazakhstan s Government should implement policies that expand the credit system in low income regions such as agricultural and other regions that do not get investments from foreign companies and do not experience high level of government expenditures. If the main explanatory variable is statistically significant in the less developed regions, in our case it is the agricultural regions, then financial development is also an important instrument for overcoming economic disparities among regions. 27

The most important limitation of the analysis is that data only available for six years. All the previous studies measuring the effect of financial development on economic growth focused on long run growth rather than short term effects. If there were data available for a longer period, we could be more confident that the analysis had escaped the effect of possible unusual short-term conditions. Another possible issue can be endogeneity. As suggested by Goldsmith (1969) the degree of financial intermediation may be endogenous, i.e. the independent variable is correlated with the error term in a regression model, therefore, it is necessary to be careful in interpreting the results. To avoid the endogeneity problem, I tried to focus more on correlation than causality while interpreting the acquired empirical results. The current analysis is in line with previous findings that financial development has significant and positive effects on output growth. Thus, government policies to encourage financial intermediation may have an important causal effect on long-run growth. According to De Gregorio and Guidotti (1995) financial intermediation affects growth through two channels: efficiency and volume of investment. This distinction assumes the difference in efficiency of private investment in regions with different types of economic activity. This issue goes beyond the scope of this paper but is an important topic for future analysis. In Kazakhstan, a country with less than 20 years of market economy history, financial policy may play a crucial role in economic development. 28

Conclusion The findings of this research indicate that bank loans to private companies and individuals have a measurably significant effect on economic growth; thus, supporting the hypothesis of this paper. Generally these results support earlier findings of King and Levine (1993b) and De Gregorio and Guidotti (1995) that increasing the amount of loans to the private sector contributes to GDP per capita growth. The growth effect in Kazakhstan is especially significant for those regions which have developed oil and gas industries. This finding contradicts the hypothesis of the paper that Credit should have a stronger effect in low income regions. There is a plausible reason for this result. Regions with lower income in comparison with regions that have oil and gas industries lag behind in their output growth capacity. The regressions on GDP per capita confirm this assumption. Credit has a significant effect on per capita output in low income agricultural regions and other regions that do not have any developed industries. This finding suggests that the Government of Kazakhstan should act to increase the availability of bank loans in low-income areas. This will help to overcome the disparity in economic growth among regions with different types of economic activity. 29

References Books and Articles Barro, Robert J. 1991. Economic growth in a cross section of countries. Quarterly Journal of Economics, Vol. 104, No. 2, pp. 407-433. Barro, Robert J., N. Gregory Mankiw and Xavier Sala i-martin. 1994. Capital mobility in neoclassical models of growth. American Economic Review. Bencivenga, Valerie and Smith, Bruce. 1991. Financial Intermediation and Endogenous Growth. The Review of Economic Studies. Vol. 58, No. 2, pp. 195-209. Dawson, P.J. 2003. Financial Development and Growth in Economies in Transition. Applied Economic Letters 10, No, 13, pp. 833-836. De Gregorio, J. 1992. Economic growth in Latin America. Journal of Development Economics, Vol. 39, No. 1, pp. 58-84. De Gregorio, José and Pablo E. Guidotti. 1995. Financial Development and Economic Growth. World Development, Vol. 23, No. 3, pp. 433-448. Goldsmith, Raymond W. Financial Structure and Development. 1969. New Haven, CN: Yale University Press. Greenwood, Jeremy and Jovanovic, Boyan. 1990. Financial Development, Growth, and the Distribution of Income. Journal of Political Economy. XCVIII, 1076-107. 30

Greenwood, Jeremy and Smith, Bruce. 1994. Financial Marlets in Development and the Development of Financial Markets. Journal of Economics, Dynamics and Control. King, Robert G. and Ross Levine. 1993a. Finance and Growth: Schumpeter might be right. Quarterly Journal of Economics, Vol. 108, No. 3, pp. 717-738. King, Robert G. and Ross Levine. 1993b. Finance, entrepreneurship, and growth. Journal of Monetary Economics, Vol. 32, No. 3, pp. 513-542. Levine, Ross. 1992. Financial structure and economic development. Working Paper WPS 849. Washington DC: The World Bank. Lucas, Robert E., Jr. On the Mechanics of Development Planning, Journal of Monetary Economics, XXII (1988), 3-42. McKinnon, Ronald I. Money and Capital in Economic Development. Washington, DC: Brookings Institution, 1973. Neusser, K. and M. Kugler. 1998. Manufacturing Growth and Financial Development: Evidence from OECD countries. Review of Economics and Statistics 80, No. 4, pp.638-646. Robinson, Joan. 1952. The Generalization of the General Theory. The Rate of Interest and Other Essays. London: MacMillan. 31

Romer, Paul. 1986. Increasing returns to scale and long-run growth. Journal of Political Economy, Vol. 94, No. 5, pp. 1002-1037. Shaw, Edward S. Financial Deepening in Economic Development. New York: Oxford University Press,1973. Solow, Robert. 1956. A contribution to the theory of economic growth. Quarterly Journal of Economics, Vol. 70, No. 1, pp. 65-94. Other sources Agency of Statistics of the Republic of Kazakhstan. 2007. Regions of Kazakhstan. http://www.stat.kz/publishing/documents/стат_сборники/регионы%20(с%20интер.с одержанием).pdf (Retrieved March1, 2010). [in Kazakh and Russian]. National Bank of the Republic of Kazakhstan. 2009. Credit System. http://www.nationalbank.kz/?docid=821&uid=4c5772b3-2219-b830-8a82c2048d264745 (Retrieved February 1, 2010). [in Kazakh and Russian] 32