**THE EFFECT OFFINANCIAL DEVELOPMENT ON ECONOMIC GROWTH: EVIDENCE FROM ASIAN COUNTRIES**

**Trần Thanh Giang (VNP18, 2012-2014)**

**ABSTRACT**

The purpose of this study is to examine the effect of financial development on economic growth in Asian countries in the period 2000 to 2011 by using panel data analysis and first difference GMM. The findings indicate that financial development affects the growth. However, its impacts depend significantly on proxies for financial development and estimation techniques. Overall, financial depth and credit to private sector have negative influences on growth. Moreover, a positive relationship between the ratio of commercial - central bank assets and the growth rate is shown by Fixed effects model whilst this indicator is not related to growth in dynamic panel model.

Key words: Financial development, economic growth, relationship, effect, endogeneity, Asian countries.

(*This is the thesis summary, The full is now at Library of Vietnam-Netherland Progamme: 1A Hoang Dieu, Phu Nhuan Dist, Ho Chi Minh city, Vietnam*)

**1. Introduction**

Contributing to the growth and the process of economic development in a country requires the combination of related sectors. It has been claimed that finance is one of the important channels affecting to economic situation. Especially, under the influences of liberalization and global economic integration, countries are more concerned with the role of finance. The impact of financial development on economic growth has been studied in many aspects by many economists since several decades.

Even though this is not a new topic, some arguments have been discussed surrounding this subject in theoretical and empirical studies. The main debates often focus on measures of financial development and channels of transmission from financial development to economic growth. In addition, there are four outstanding schools of thoughts in the direction of relationship between these two issues: the supply – leading hypothesis (Christopoulos & Tsionas, 2004; King & Levine, 1993a; Levine, Loayza, & Beck, 2000; McKinnon, 1973), demand – following hypothesis (Goldsmith, 1969; Jung, 1986; Shaw, 1973), bidirectional association (Greenwood & Jovanovic, 1989) or no linkage (Lucas Jr, 1988). The conflict issues may result from various conclusions. While some studies emphasize the positive contributions of financial development to growth, other researches provide opposite evidences. Enormous contributions of financial development to economic growth were explained by (McKinnon, 1973; Shaw, 1973). As given by (Levine, 1997), financial development is able to improve better investment opportunities, reduce transaction costs and mobilize savings, hasten technological innovation and diversify risks for investors. Studying 109 developing and industrial countries in 1965 - 1994, (Calderón & Liu, 2003) suggested that the contribution of financial development to growth is higher in developing countries than industrial countries. Moreover, this study also supports for view that financial development can accelerate economic growth through rapid capital accumulation and technological change. In particular, the causal relation from finance to TFP growth and capital accumulation is stronger in developing countries, but the direction from TFP growth and capital accumulation to finance is stronger in industrial countries. On the other hand, after combining both cross sectional and time series data, the result of (Christopoulos & Tsionas, 2004) exhibited that although there was no bidirectional causality between financial development and output growth in both long - run and short – run, the positive causality from financial development to growth was still manifested in 10 developing countries from 1970 – 2000.(Loayza & Ranciere, 2006), however, suggested that financial system is able to cause economic recession due to financial crisis. Specifically, financial development is not always positively related to economic growth. Differences in results are because its effects depend on the characteristic of financial system of each country, the institution, the study period or the measures of financial development (De Gregorio & Guidotti, 1995; Hassan, Sanchez, & Yu, 2011; Khalifa Al-Yousif, 2002; Wu, Hou, & Cheng, 2010). Consequently, finance and economic growth are always an interesting subject with economists and researchers. If the role of financial development is clarified, policymakers can outline the judicious development directions and propose appropriate policies to speed up the national growth.

Rising from the financial crisis in the United State during 2007 – 2009, it has spread to other countries to create the global crisis. The consequence is the collapse of financial system in a number of countries and leads to macroeconomic instabilities. According to the level of financial development and specific characteristics of each country, the effect of financial crisis on economic growth will be different in those countries. Asian countries, as a whole, have been benefited from financial liberalization. It yields a more efficient financial system, intensifies the efficiency and the adjustability of monetary policies. Thus, Asian countries succeeded in raising the role of financial system in promoting growth in the period before 1995. Following this, Asian economy underwent some large fluctuations because financial crisis and serious economic recession have consecutively happened since 2000. Especially, 1997 Asian financial crisis, later on, economic crisis in 2007 – 2008 caused adverse effects on both financial system and growth in most Asian countries. For these reasons, this study examines, under such circumstances, how financial development affects economic growth in Asian countries in the period 2000 – 2011. From the obtained research findings, this research will provide some suggestions to promote the development of financial sector.

**2. Literature Review**

Financial development is defined as an improvement ịn quality, quantity and efficiency of financial system. This process involves the combination of many activities and institutions. According to (Levine, 2005), financial development happens when financial instruments, markets and intermediaries are improved better, then they affect the information, enforcement and transaction costs. Therefore, financial development includes the improvement in allocating resource, monitoring investment, mobilizing saving, diversifying and managing risk. Besides, the enhancement in controlling corporate governance after providing finance and the facilitation in exchanging goods and services are also stimulants to financial development. Each of these functions is likely to influence saving and investment decisions, thereby they affects economic growth. The importance of financial intermediaries to economic growth appears to be supported by (King & Levine, 1993a; Levine et al., 2000; McKinnon, 1973; Shaw, 1973). They emphasize that the essential reason for different economic growth of each country derives from the differences in the quality and quantity of services provided by financial intermediations.

(Pagano, 1993) based on the endogenous growth model to study the impact of financial development on economic growth. From the equation g = Aθs – δ (Pagano, 1993) indicates the development of financial sector is likely to impact economic growth through three ways. The first one is an increase in θ that is the proportion of saving channeled to investments. If the banking sector can be operated in more competitive environment and the transition from saving to investment is implemented more efficiently, which cause θ can be increased, thereby growth rate is higher. Secondly, financial institutions, such as the bank, can increase the productivity of capital (A) by allocating funds to the projects having the highest margin product of capital and distributing savings more efficiently. Hence, higher growth results from higher productivity of capital. The third channel is private saving rates (s).

Similarly, in the model of (De Gregorio & Guidotti, 1995), the growth rate of output y_{t} = s_{t}θ_{t} is a function of saving rate s and marginal productivity of capital θ. This function also proves that financial development affects growth through two channels. The first, the development of domestic financial market may improve the efficiency of capital accumulation (increasing θ) (Goldsmith, 1969). The second, financial sector contributes to raising saving rate, thereby higher investment rate (increasing s_{t})(McKinnon, 1973; Shaw, 1973). Because of the increase in savings and the improvement in the efficient investments, financial development results in economic growth. (Bencivenga & Smith, 1991; Greenwood & Jovanovic, 1989). Moreover, the author also supports evidence that the effect of financial development on economic growth is exhibited essentially via the efficiency of investment rather than its volume.

In summary, from studies above demonstrate that the foundation of economic growth is the factors of production and the efficient use of those factors.

Due to the informational asymmetries between borrowers and lenders, the lender must spend more costs to verify the production capacity and the financial situation of firms. Therefore, the presence of financial intermediaries may perform financial contracts that have lower monitoring and enforcement costs can in turn induce higher return on investment and increase investment efficiency. As a result, financial arrangements which reduce asymmetric information and ameliorate corporate control will better the allocation of resources. This is the foundation to further faster capital accumulation and economic growth.

On the other hand, the appearance of asymmetric information and transaction cost will increase liquidity risk. Financial market with capacity for diversifying portfolio will move savers to the projects with higher expected yields. Thus, by eliminating risk of liquidity, financial intermediaries, such as bank, stock market and fund can also enhance savings and increase investment in high return projects and illiquid assets, so they may stimulate the growth.(Saint-Paul, 1992) developed a model to clarify how the financial market affects the technological choice. The first strategy is high flexibility and allows the productive diversification but has low productivity. The other strategy is more productive and more specialize. The financial market helps individuals who hold diversified portfolio against negative demand shock and choose high productive technology. Because the individual can be diversified their portfolio via the stock market, they can reduce risk and increase the proportion of resource allocated to firm and encourage the specialization of production, thereby economic growth is higher.

Hence, the well function of risk diversification not only improves the efficient resource allocation and higher saving, but it also affects capital accumulation and technological innovation. Following (Beck, Levine, & Loayza, 2000; Levine, 1997), better allocation of savings and rapid capital accumulation, which can be achieved by increasing domestic saving rates and attracting foreign capital, foster technological innovation. Before that, the importance of financial development on technological innovation was also mentioned in (King & Levine, 1993b). Following this, financial market can identify and evaluate the potential innovative projects, thus allocating resources efficiently. Therefore, the nation with a better function of financial system can conduct a successful innovation. On the contrary, a decrease in the rate of innovation produced by policies distorting financial system will decline growth rate. Consequently, as to (Beck et al., 2000; King & Levine, 1993b; Levine, 1997), the innovation is the main channel of transmission between finance and growth. In conclusion, as to endogenous growth literature, financial development heightens economic growth through capital accumulation and technological innovation.

In accordance with (Levine, 2005), the development of financial market and institutions can promote economic growth through some channels: increasing the liquidity, managing and diversifying risks, establishing the payment system, trading goods and services more easily, reducing information costs and transaction costs, providing information about enterprise to allocate capital efficiently. Especially, the function of savings mobilization from individual and investor in the economy contributes to raising capital accumulation, thereby economic development is faster.

**3. Research Methodology**

**3.1 Model Specification**

In order to construct a regression model examining the impact of financial development on economic growth, this study bases on previous researches to choose suitable model and variables. Normally, cross – country regressions and panel data techniques are common methods for investigating the finance – growth nexus. Cross section analyses, however, have some potential limitations because it is likely to suffer from omitted variable bias, or the regressors are endogenous. Moreover, the unobserved country specific effects are not able to control by cross section analyses. If cross section estimator is performed, it only controls for the endogeneity of financial development but cannot control for the endogeneity of all explanatory variables. The coefficients of financial development indicators can be biased and inconsistent. Consequently, to overcome these shortcomings, recent papers have been suggested using dynamic panel data approach instead of cross section. Dynamic panel data model not only allows to address the omitted variable bias and accounts for the unobserved country specific effects, but this model also deals with endogeneity of one or more regressors. As has been shown in reviewing the empirical studies, (Beck et al., 2000; Levine et al., 2000) also performed dynamic panel data model to estimate the relationship between finance and growth.

The general dynamic panel model which examines the effect of financial development on economic growth is described as follows:

y_{it} = a + βy_{i,t-1} + γx_{it} + n_{i} + λ_{t} + u_{it} (3.1)

where y_{it} is economic growth, y_{i,t-1} is the lagged values of dependent variable. X_{it} is explanatory variables, including financial development indicators and other determinants of economic growth. n_{i } is unobserved country specific effects. λ_{t} is a time – specific effect and u_{it} is the error term.

**3.2 Measures of financial development**

To evaluate the impact of financial system on economic growth, it has to consist of the combination of various measurement methods. Generally, three important criterions are used to manifest the development level of financial system and the extent and the efficiency of financial intermediation. Those are the size, activity and the efficiency of the financial sector (Beck et al., 2000).

In order to measure the size of financial sector, (Beck et al., 2000; Goldsmith, 1969; King & Levine, 1993a; McKinnon, 1973) concentrated on liquid liabilities ratio (DEPTH)calculated by the currency plus demand and interest bearing liabilities financial intermediaries and non -bank financial intermediaries divided by GDP. This indicator is often used for representing financial depth which means not only higher productivity of capital but also higher saving rates, thereby the volume of investment is higher. This indicator, however, may not exactly represent the effectiveness of financial sector in improving asymmetric information and reducing transaction costs. Moreover, this measure shows the extent of transaction services provided by financial system rather than evaluating the capacity of financial system in transferring funds from depositors to investment opportunities.

The second indicator of financial development is designed to indicate the role of commercial bank compared with central bank or the importance of bank in the allocation of society’s savings. Commercial – Central bank assets(BANK)equal the ratio of asset of deposit money banks divided by the sum of assets of deposit money banks and central bank assets. Because the intermediaries of commercial banks have functions such as identifying profitable investment projects, banks can increase the efficient use of funds. Also, commercial banks can monitor risks and mobilize savings, thus they are better than central bank. On the other hand, this indicator does not directly appraise the effectiveness of banks in examining firms, trying to control corporations, mobilizing savings, facilitating transactions and providing programs of risk management. As a result, it cannot directly measure the quality and quantity of financial services provided by financial intermediaries.

Another indicator measuring the activity of financial sector as well as the common measure of the efficiency of resource allocation in the financial sector is the ratio of credit to the private sector - credit by deposit money banks and other financial institutions to the private sector divided by GDP (CREDIT). Although this indicator also focuses on credits issued to the private sector, it excludes credits issued by the central bank (Levine et al., 2000). Hence ,it measures relatively the role of financial intermediaries in channeling funds to the private sector, which affects the investment as well as the efficiency of investment (De Gregorio & Guidotti, 1995). Furthermore, because credit to the private sector not only bring more productively than credit to public sector, but it also indicates aspects of both the quality and quantity of investments. In general, higher level of CREDIT manifests a higher level of financial services, thus it is synonymous with greater financial development.

Besides the credit market, the stock market, an important financial institution, contributes to the development of financial sectors and economic growth(Pagano, 1993). The appearance of stock market can help investors holding financial portfolio reduce transaction costs and diversify their risks. Because risk diversification encourages investors to hold more their individual assets in productive capital, it contributes to hastening growth. In addition, technological changes can be affected by the risk diversifications. For this reason, the last indicator, which is also used to measure financial development, is the ratio of stock market capitalization to GDP (STOCK)(Jun, 2012; Levine & Zervos, 1996; Ndlovu, 2013; Wu et al., 2010). It equals the total value of all listed shares in a stock market as a percentage of GDP.

**3.3 Data source**

This research applies four indicators to measure financial development, including liquid liabilities to GDP (DEPTH), commercial – central bank (BANK) and credit to private sector divided by GDP (CREDIT) and the ratio of stock market capitalization to GDP (STOCK).Besides indicatorsof financial development are the most important explanatory variables, other factors affecting economic growth are also involved in the research model. This study uses the secondary school enrolment ratio (EDU) to represent the human capital. Also, the annual growth rate of the GDP deflator(INF) and the ratio of general government final consumption expenditure to GDP (GOV) are used as proxies for macroeconomic stability. To capture the degree of trade openness of an economy, we utilize the sum of exports and exports as a percentage of GDP (TO). In addition, as given by the endogenous growth theory, the annual growth rate of real GDP per capita at constant price 2005 (GROWTH) is performed to estimate economic growth.

Most data of variables in this research are mainly gathered from list of available countries from World Development Indicators of World Bank’s databasein the period of2000to 2011.Unfortunately, all countries do not have sufficient data for all years. Many countries do not even have data ofthe finance and the secondary school enrollment ratio. Particularly, among financial development indicators, data of ratio of stock market capitalization to GDP has many missingvalues. Therefore, after eliminating countries not havingsufficient data, this studyonly focuses on investigating the impactof finance development on growth based on unbalance panel data of 32 countries. List of countries is described in detail in the table A.1 (Appendix A).

From the equation (3.1), this study establishes the general model researching the effect of financial development on economic growth as follows:

GROWTH_{it} = a + βGROWTH_{i,t-1} + χ1F_{it} + χ2INF_{it} + χ3GOV_{it} +χ4TO_{it} + χ5EDU_{it} + n_{i} + λ_{t} + u_{it }(3.2)

Where i denotes the country and t denotes the time. GROWTH_{it} is the growth rate of real GDP per capita. GROWTH_{i,t-1} is the one – year lag of growth rate of real GDP per capita. Other explanatory variables consist of financial development indicators (DEPTH, BANK, CREDIT, STOCK), the inflation rate (INF), the ratio of government expenditure to GDP (GOV), the trade openness(TO) and the human capital stock of an economy (EDU).

**3.4 Research Methodology**

Generally, The Pooled OLS, Fixed effects method (FEM) and Random effects modeL (REM) are three common methods estimating models for panel data. However, it need to be noticed that, because dynamic panel model contains the lagged dependent variable (GROWTH_{i,t-1}), which is considered as a regressor in the model, the result of OLS estimation, FEM and REM will be biased and inconsistent due to the correlation of regressor with the error term. For this reason, the lagged dependent variable (GROWTH_{i,t-1}) is excluded from the model when estimating regression model with OLS, FEM and REM. Furthermore, because there is a high correlation between financial development indications, this study willregress separately each indicator of financial development with other variables associated with growth. We have the following models:

GROWTH_{it} = a_{0} + a_{1}DEPTH_{it} + a_{2}INF_{it}+a_{3}GOV_{it} + a_{4}TO_{it} + a_{5}EDU_{it }+ u_{it }(Model 1)

GROWTH_{it} = a_{0} + a_{1}BANK_{it} + a_{2}INF_{it}+a_{3}GOV_{it} + a_{4}TO_{it}+a_{5}EDU_{it }+ u_{it } (Model 2)

GROWTH_{it} = a_{0} + a_{1}CREDIT_{it} + a_{2}INF_{it}+a_{3}GOV_{it} + a_{4}TO_{it}+a_{5}EDU_{it }+ u_{it } (Model 3)

GROWTH_{it} = a_{0} + a_{1}STOCK_{it} + a_{2}INF_{it}+a_{3}GOV_{it} + a_{4}TO_{it}+a_{5}EDU_{it }+ u_{it } (Model 4)

**3.4.1 The common constant method**

The general model ofthe common constant method is described as follows:

Y_{it}_{ =}a_{0} + a_{1}X_{it} + u_{it}_{, }i = 1,....,N; t = 1,....,T (3.3)

The common constant method, which is also called the pooled OLS method, can estimate a common constant a for all cross – sections. In other words, this estimation assumes that the differences between the estimated cross - sections do not exist. Moreover, the error term u_{it }is assumed to be independently and identically distributed (iid), u ~ iid(0,s^{2}). Specifically, the Pooled OLS has some important assumptions as follows:

- The parameters of a_{o}, a_{1} and the error term u_{it }are linear.

- The observations are independent across individuals but they are not across time.

- There is no correlation between the idiosyncratic error term u_{it,} and the explanatory variables of all past, current and future time periods of the same individual. It means we rule out the lagged dependent variables. In other words, the idiosyncratic error is uncorrelated with the individual specific effect. In this case, the variables are assumed to be strict exogeneity.

- The error variance is homoscedasticity and no serial correlation.

Under assumptions above, the estimated parameters a_{o} and a_{1 }of the Pooled OLS are unbiased in small samples. In the samples with a larger number of individual, the pooled OLS estimator is consistent and approximately normally distributed. However, this method is not efficient. On the other hand, the standard errors are also not correct, and the tests such as t-, F-, z-, wald test are no valid. In addition, if the true model is the fixed effects model, the Pooled OLS estimators of a_{o} and a_{1} are biased and inconsistent. The reason is due to the presence of obmitted variables and the potentional correlation between individual – specific effect and the other regressors.

In summary, The pooled OLS estimator is consistent if the Pooled OLS is appropriate and regressors are not correlated with the error term. However, in the case of the true model is the fixed effects model y_{it} = a + x_{it}β + (a_{i} - a- u_{it}), if the individual effect a_{I }is correlated with the regressors x_{it}, the pooled OLS estimation will be inconsistent. Although this is a simple estimation in panel data model, it still has some drawbacks.

Firstly, comparing with fixed effect model and random effect model, the constant intercept and the constant slope coefficents achieved by the pooled OLS model have more restrictions in panel data estimation. As to the basic assumptions of this technique, there are no serial correlation in observations, and the errors are homoscedasticity:

E(u_{it}) = 0, Var (u_{it} ) = s^{2}, Cov(u_{it},e_{js}) = 0 if i ¹ j or t ¹ s.

Secondly, assumptions of independently and identically distributed error term neglect the panel structure of data.

Thirdly, the intercept of each country is the same, and coefficients of independent variables of each country are also the same and unchanged by time.

These issues can distort the real relationship among the dependent and explanatory variables in the model. Therefore, the common constant method can be inconsistent in studying panel data.

**3.4.2 The random effects method (REM)**

The general model of random effects method is described as follows:

Y_{it} = a_{1} + a_{2}X_{it} + u_{it} where u_{it =}ɛ_{i} + v_{it} (3.4)

The error term of random effects method involves two parts. The first component ɛ_{i} called individual effect is individual time – invariant. The second component v_{it}that is individual time – varying differs unsystematically across individuals and over time. Moreover, the time – invariant individual effect ɛ_{i} is assumed to be uncorrelated with x_{it}. This can be wirtten: cov(ɛ_{i,}, x_{it}) = 0. Therefore, in the random effects model, the individual specific effect is a random variable that isuncorrelated with the independent variables of all past, current and future time periods of the same individual. Also, REM also assumes the individual specific effect has constant variance.

On the other hand, REM can include time – invariant variables into the model because this method can estimate the effects of time – invariant variables. However, REM cannot control for omitted variables, the estimated results can be biased. Besides, REM assumesthat different characteristics of each country will be presented through the error term.

Consequently, under the random effects model, the random effects estimator is absolutely efficient. Nonetheless, it is biased and inconsistent if the fixed effects model is the true model because the individual specific effect is obmitted and potentially correlates with the other explanatory variables.

**3.4.3 The fixed effects method (FEM)**

The general model of fixed effects method is described as follows:

Y_{it} = a_{i} + βX_{it} + u_{it} (3.5)

The fixed effects model indicates that slopes are constant and the intercept can differ across countries but it does not vary across time. The coefficient a is the unchanged coefficients among cross sections and time. Unlike REM, FEM assumes that the differences across sections are exhibited by the intercept. Moreover, FEM assumes that there is correlation between predictor variables and error term. On the other hand, it permits the time – invariant individual effect is correlated to x_{it, }cov(a_{i}, x_{it}) ¹ 0 while it continues assumingthat x_{it} is not correlated with the idiosyncratic error u_{it}.

If the model has omitted variables and these variables are correlated with variables in the model, fixed effect estimator can control for omitted variables bias and remove the characteristics of time – invariant from the independent variables. Therefore, unlike REM, FEM cannot estimate the effects of variables that its value does not change over time, such as gender, religion, geographical factors. culture, etc because these variables are perfectly collinear with the fixed effect.Ultimately, the estimated coefficient of FEM cannot be biased.

The FEM is also expanded by including a set of time dummies. This is known as the two way fixed effect model that shows effects vary over time but are common across the whole panel.

FEM is also called the least squares dummy variable (LSDV) because it can measure the characteristic of each country through dummy variables. In other words, LSDV allows different constants for each group. This estimation includes a set of N-1 dummy variables which identify the individuals. One of individual dummies is dropped because we include a constant. For instance, this study consists of 33 countries, there are 32 corresponding dummy variables. However, this is also disadvantage of this estimation because there are many dummy variables, which lead to reduce degree of freedom. Hence, if the number of parameters goes to infinity as N " ∞, LSDV estimator will be inconsistent.

**3.4.4 The choice of panel regression model**

To suggest the appropriate panel model, the research can rely on some tests. Firstly, F – test which bases on loss of goodness- of – fit is applied to choice an appropriate technique between fixed effects method and Pool OLS. The regression model of fix effect method is described by y_{it}= a + µ_{i} + βX_{it} + e_{it}, thus the null hypothesis is H_{0}: µ_{1 }= ….. = µ_{n-1} = 0. F test can be calculated as follows:

Where RSS is the restricted residual sum of squares of polled OLS model, and URSS is the unrestricted residual sum of squares of LSDV regression.

If the null hypothesis is rejected, it means that at least one group or time specific intercept µ_{i }is not zero. Hence, we can conclude that fixed effect model is better than the pooled OLS.

In order to decide to choose between random effect model and pooled OLS, this research can rely on the Breusch – Pagan LM test for random effects. The purpose of this test is to examine whether the individual or time specific variance components equal zero. The null hypothesis is H_{0}: var(µ) = 0. The LM test can be calculated by:

If the null hypothesis is rejected, the random effect model is favored over pooled OLS, otherwise pooled OLS is preferred.

As represented above, the difference between fixed effects model and random effects model is due to the difference in assumption of correlation between time – invariant individual effect a_{i} and explanatory variables x_{it}. If there is correlation between a_{i } and x_{it }, FEM is consistent and efficient, but REM is not consistent. In contrast, if a_{i} are uncorrelated to x_{it}_{,},REM will be efficient and consistent, whereas FEM is consistent and inefficient.

Therefore, on order to choose the appropriate model between FEM and REM, it is necessary to consider whether a_{i} and x_{it} are correlated or not.

The results of Hausman test can be relied on making decisions.

(β_{FE} – β_{RE})’[Var(β_{FE}) - Var(β_{RE})]^{-1} (β_{FE} – β_{RE}) ~ χ^{2} (K)

The null hypothesis Ho: the random effects are consistent and efficient

H_{1}: the random effects are inconsistent.

Ho means that a_{i} is not correlated x_{it}, H_{1}implies that a_{i } and x_{it} are correlated, which prefers to FEM.

In the case of the value of statistic is large, the estimates are significant differences, thus the null hypothesis can be rejected. It means that the fixed effects estimator is better. Inversely, if the statistic value of Hausman test is small, the random effects estimator should be employed.

In short, in order to examine the appropriate model for a panel data, the study has to perform several tests. There are three basis tests. First of all, the F test is implemented for fixed effects against pooled OLS. Secondly, LM test is executed to take the random effects against pooled OLS. Finally, applying the Hausman test makes a choice between fixed effects and random effects model.

**3.4.5 The generalized method of moments estimation (GMM)**

In order to examine the finance – growth nexus with panel data, (Levine et al., 2000) added the lagged value of dependent variable to the model, thus model becomes the dynamic panel model. In this case, if techniques of Fixed effects and Random effects are used to estimate model, results will be bias and inconsistent. Therefore, (Levine et al., 2000) suggested using the dynamic panel estimator – the GMMto deal with serial correlation, heteroskedasticity and endogeneity of explanatory variables in the model. Besides, this method can overcome the biases due to omitted variables. The general dynamic regression model is written as follows:

Y_{it} = a + βY_{i,t-1} + γX_{it }+ µ_{i} + λ_{t} + e_{it} (3.6)

Where Y is the dependent variable, Y_{i,t-1 }is the lagged of dependent variable and X is the explanatory variables. e is error term and µ_{i} is an unobserved country – specific effect, λ_{t} is a time – specific effect.

Two main econometric issues need to be noticed from equation (3.6). Firsly, the unobserved country – specific effect µ_{i} is a part of error term. Simultaneously, the lagged value of dependent variable Y_{i,t-1} is also included in the model. Thus, there exists a positive correlation between omitted fixed effects and lagged dependent variable Y_{i,t-1}. For this reason,the estimated coefficient will be biased and inconsistent in OLS estimation. Hence, to control for the presence of unobserved country specific effects, (Anderson & Hsiao, 1982) suggested eliminating fixed effect by taking first differences. The second issue comes from the potential endogeneity of explanatory variables. To deal with these problems, (Arellano & Bond, 1991) proposed the first differenced GMM estimator.

After we take first difference of equation (3.6) to exclude the country – specific effect, model for first difference GMM can be written as follows:

Y_{it} – Y_{i,t-1} = β(Y_{i,t-1} - Y_{i,t-2}) + γ(X_{it} – X_{i,t-1}) + (e_{it} - e_{i,t-1}) (3.7)

Although equation (3.7) removed the unobserved country – specific effect µ_{i}, a new construction shows the correlation between (Y_{i,t-1} - Y_{i,t-2}) and (e_{it} - e_{i,t-1}). To solve this correlation, it is necessary to find a valid instrument for (Y_{i,t-1} - Y_{i,t-2}). According to (Arellano & Bond, 1991), the lagged values of Y_{i,t} in two periods or more are correlated with (Y_{i,t-1} - Y_{i,t-2)}, but it is uncorrelated with (e_{it} - e_{i,t-1}), thus its values are valid instruments for equation (3.7). Similarly, there are some cases of explanatory variables X_{it }:

If X_{it} is a strictly exogenous, valid instruments for equation (3.7) will be the past, recent and future values of of X_{it}.

If X_{it} is endogenous, twice lagged level of X_{i} and more are instruments for equation (3.7).

GMM approach has two important assumptions. First, the error term e is not serially correlated. The second, the explanatory variables are weakly exogenous. Based on these two assumptions, the first differenced GMM estimator has following moment conditions:

E[Yi,t-s(eit- ei,t-1)] = 0 for s ³ 2; t = 3,...,T (3.8)

E[Xi,t-s (eit - ei,t-1)]= 0 for s >=2, t = 3....T (3.9)

The dynamic regression model for panel data investigating the effect of financial development on economic growth, which employs the first differenced GMM estimator, can be written as follows:

rGROWTH_{it} = β_{0} + β_{1}rGROWTH_{i,t-1 }+ β_{2}rDEPTH_{it} + β_{3}rINF_{it }+ β_{4}rGOV_{it }+ β_{5}rTO_{it }+ β_{6}rEDU_{it }+ u_{it } (Model 5)

rGROWTH_{it} = β_{0} +β_{1}rGROWTH_{i,t-1} + β_{2}rBANK_{it} + β_{3}rINF_{it} + β_{4}rGOV_{it }+ β_{5}rTO_{it }+ β_{6}rEDU_{it }+ u_{it } (Model 6)

rGROWTH_{it} = β_{0 }+ β_{1}rGROWTH_{i,t-1} + β_{2}rCREDIT_{it} + β_{3}rINF_{it} + β_{4}rGOV_{it }+ β_{5}rTO_{it }+ β_{6}rEDU_{it }+ u_{it } (Model 7)

rGROWTH_{it}= β_{0 }+ β_{1}rGROWTH_{i,t-1 }+ β_{2}rSTOCK_{it }+ β_{3}rINF_{it} + β_{4}rGOV_{it }+ β_{5}rTO_{it }+ β_{6}rEDU_{it }+ u_{it } (Model 8)

Where r denotes the difference in the estimated variable, and the expected sign of these variables are similar with FEM. The symbol of variables used in dynamic regression models and the definitions of these variables are described in detail in Table A.3 (Appendix A).

**4. Empirical Results**

Because we reject the null hypothesis of F – Test and LM Test, FEM and REM are better than Pooled OLS. Thus, Hausman test is performed to choose an appropriate model between them.

Since the p-value of Hausman Test for all measures of financial development equals nearly zero, the null hypothesis can be rejected at the significant level of 1%. Ultimately, the fixed effect model is considered as the most appropriate model for the regression model.

After being estimated by fixed effect model, the result is reported that the signs of DEPTH, CREDIT and STOCK are negative, but BANK is positive. However, most coefficients of financial development are statistically significant except the ratio of stock market capitatization to GDP. Thus, there are negative impacts of DEPTH and CREDIT on economic growth. Simultaneously, banking system has also important role and positive influence on economic growth in Asia region.

The results of first difference GMM in table 4.6 indicate some differences in the results of FEM and GMM. Firstly, if the estimated coefficient of BANK is positive and statistically significant at 5% level, but the finding from GMM is not. Secondly, similar to statistically insignificant coefficient of STOCK in FEM, the finding of GMM also suggests that rSTOCK is statistically insignificant. (Beck & Levine, 2004; Levine & Zervos, 1996) also put stock market in the model. Their findings argued that the stock market can boost growth quickly because this market allows individual to diversify portfolio, manage liquidity and risk. Hence, advantages of stock market encourage companies to invest more. In other words, if the functions of risk diversification and the reduction in transaction costs are not conducted efficiently, they will not indirectly contribute to promoting economic growth in this area.

In addition, the estimated coefficients of CREDIT is negatively related to GROWTH in FEM, applying GMM technique also reports an analogous result. As interpreted in previous section, the negative coefficient of rCREDIT implies that the larger volume of credit allocated to private sector does not contribute to higher growth rate. 1% change in CREDIT will cause -0.173% change in GROWTH. On the other hand, a negative coefficient of rDEPTH in GMM is similar to FEM. 1% change in DEPTH will lead to -0.138% change in GROWTH. It means that deepening of financial market may impede growth rates. This supports the view of negative effect of financial development on economic growth (Bencivenga, Smith, & Starr, 1995; McKinnon, 1973; Shaw, 1973). As mentioned in literature review, a negative relationship between ratio of liquid liabilities to GDP, ratio of credit to private sector and economic growth manifests some issues. Although more credits have been transferred to different sectors in economy, they do not effectively contribute to increasing the productivity of capital. It does not also expand higher saving rate and enhance quantity and quality of investment. Ultimately, economic growth is lower. In other words, the function of resource allocation of financial sector is not efficient. Besides, because the ratio of domestic credit to private sector represents the role of financial intermediaries in channeling funds to private sector, this ratio is a more closely linked to the level and the efficiency of investment. This measurement also concentrates on the development through banking system. Thus, this indicator has a weak relation with growth when financial development occurs with a large extent outside banking system. As a result, higher growth does not necessarily derive from higher volume of credit to private sector(De Gregorio & Guidotti, 1995). According to (King & Levine, 1993a), DEPTH relates with the magnitude and the efficiency of loans, This can be explained that if loans for investments are not monitored well, they may not bring about higher productivity of capital and lead to loan loss and financial crisis (Singh, 1997).

In addition, as to (Khalifa Al-Yousif, 2002), a negative linkage between finance and growth is due to the result of the business cycle. Concurrently, other reason is that financial intermediaries operate in a weak regulatory environment. As you know, financial intermediation mainly affects economic growth through mobilizing savings and allocating these funds to productive investment projects to create high returns. If financial intermediaries tend to expect that the government will bail out failing banks, they may neglect to manage risk and control the use of capital. The consequence is inefficient resource allocation and lower economic growth. On the other hand, the fragile financial system and the weak regulatory environment may result in financial crises such as the crisis in East Asia in 1997 – 1998. Furthermore, as to the McKinnon – Shaw hypothesis, financial repression, such as the control of interest rates and banking sector restrictions and direct credit program, intervened strongly in financial system. Because restrictions of government may affect the quality and quantity of investment, it causes an adverse impact on development of financial system. However, it does not mean that financial integration is always positive influence on the growth. Its effect depends on the development of financial market in each country, macroeconomic stability and the quality of institution as well as difference in policy implementation. In other words, different countries will have different effects of financial development on the growth because the differences in level of financial development are due to differences in policies and institutions. Therefore, while most studies have emphasized that economic growth can be positively influenced by the development of financial sectors, some other papers have had reverse conclusions.

Besides, inflation is also not an important determinant of economic growth in these countries when this estimated coefficient does not have statistical significance. This result is consistent with (Beck et al., 2000) because inflation affects on growth through the financial sector performance. Thus, when the research verifies the level of financial intermediary development, inflation has an insignificant effect. The relation between finance and growth has been discussed in some studies. Firstly, inflation decreases real return to savings and leads to more adverse selection problems in capital markets. Later on, it causes a high degree of credit ratio and negatively affects on financial development. Also, this study finds a negative effect of government spending and a positive impact of trade openness on economic growth. Consequently, the increase in trade openness combines with the reduction in government expenditure will help improve higher GDP. For this reason, the role of the trade and the government spending policies is quite important in the process of economic development.

From the results above, we may conclude that financial development is negative effect on economic growth through two indicators: DEPTH and CREDIT. As discussed above, the role of the financial depth and the credit to private sector in the growth of economy is important. Because an improvement in risk diversification combines reduction of transaction costs with enhancement in quality and quantity of investment, it is able to stimulate a rise in investment efficiency. By doing this, it yields higher efficient use of capital and greater economic growth. However, if one or more functions above are carried out efficiently, such as, high transaction costs and the inefficiency in the risk management and the quality of investment, they cause indirectly the decline in growth rates. Besides, Asian countries should concentrate on promoting trade and managing macroeconomic stability. In short, this research finds evidence which supports the negative impact of financial development on growth.

**5. Conclusion and policy implication**

The purpose of this study is to examine the effect of financial development on economic growth based on a dataset of 32 Asian countries over the period 2000 - 2011. By employing the pooled OLS, Fixed effects model and Random effects model, and especially, first difference GMM, several main results are found in this study. In general, despite using FEM or first difference GMM, this study still finds evidences for a negative effect of financial development on economic growth , which is consistent with (Ang & McKibbin, 2007; Bencivenga et al., 1995; De Gregorio & Guidotti, 1995; Khalifa Al-Yousif, 2002; McKinnon, 1973; Shaw, 1973).

This conclusion is based on a negative effect of the ratio of liquid liabilities and credit to private sector on growth rates. It can be also explained that the financial institutions do not well perform the function of saving mobilization and its allocation effectively, thus it does not lead to increase the efficiency of capital. Ultimately, it does not contribute to enhancing the growth rates in Asian countries. Concurrently, the volatility of financial system can derive from 1997 Asian financial crisis and global financial crisis as well as economic recession in the period 2000 - 2011. Moreover, the fragility of financial system, legal system, quality of institution and financial repressions can be factors that cause financial system negatively effects on economic growth. Besides, policies should also concentrate on stabilizing macroeconomic situation and encouraging trade openness.

In accordance with (Levine et al 2000), the level of financial intermediary development is different across countries because of the differences in the legal rights of creditors, the efficiency of contract enforcement and accounting standards. The degree of legal system that supports the creditor’s rights will directly affect the contract of finance and the activities of financial intermediaries. Countries with higher level of creditor’s rights not only reflect stronger creditor rights, but they also prove better financial intermediary development. Besides, the quality and the effectiveness of the legal system in enforcing contracts will affect the activities of financial sector through supporting activities of banking system. Moreover, investors are concerned with information about firms to determine the best investment opportunities. Hence, the role of accounting standards is also a foundation to create financial contracting when it yields interpretability and comparability of information across enterprises. Overall, creditor’s rights, contract enforcement and accounting standards are determinants of financial intermediary development. Therefore, those are important factors to boost the development of financial intermediaries, and so, growth is faster.

From the findings of this research, in order to better developed financial system, financial intermediaries need to improve information and transaction costs to generate an efficient allocation of sources, thereby stimulating economic growth. In other words, it is necessary for countries to reform their laws and regulations. They help ensure a high priority for creditors to build their trust more. Concurrently, legal system must be transparent and effective. Outside investors can easily collect comprehensive and comparable information about corporate financial statements to achieve the best investment chances.As discussed in previous chapters, policies should reduce financial repression and favour financial liberalization

Anderson, Theodore Wilbur, & Hsiao, Cheng. (1982). Formulation and estimation of dynamic models using panel data. Journal of econometrics, 18(1), 47-82.

Ang, James B, & McKibbin, Warwick J. (2007). Financial liberalization, financial sector development and growth: evidence from Malaysia. Journal of Development Economics, 84(1), 215-233.

Arellano, Manuel, & Bond, Stephen. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The review of economic studies, 58(2), 277-297.

Beck, Thorsten, & Levine, Ross. (2004). Stock markets, banks, and growth: Panel evidence. Journal of Banking & Finance, 28(3), 423-442.

Beck, Thorsten, Levine, Ross, & Loayza, Norman. (2000). Finance and the Sources of Growth. Journal of financial economics, 58(1), 261-300.

Bencivenga, Valerie R, & Smith, Bruce D. (1991). Financial intermediation and endogenous growth. The Review of Economic Studies, 58(2), 195-209.

Bencivenga, Valerie R, Smith, Bruce D, & Starr, Ross M. (1995). Transactions costs, technological choice, and endogenous growth. Journal of Economic Theory, 67(1), 153-177.

Calderón, César, & Liu, Lin. (2003). The direction of causality between financial development and economic growth. Journal of Development Economics, 72(1), 321-334.

Christopoulos, Dimitris K, & Tsionas, Efthymios G. (2004). Financial development and economic growth: evidence from panel unit root and cointegration tests. Journal of development Economics, 73(1), 55-74.

De Gregorio, Jose, & Guidotti, Pablo E. (1995). Financial development and economic growth. World development, 23(3), 433-448.

Goldsmith, Raymond William. (1969). Financial structure and development (Vol. 1): Yale university press New Haven.

Greenwood, Jeremy, & Jovanovic, Boyan. (1989). Financial development, growth, and the distribution of income: National Bureau of Economic Research.

Hassan, M Kabir, Sanchez, Benito, & Yu, Jung-Suk. (2011). Financial development and economic growth: New evidence from panel data. The Quarterly Review of economics and finance, 51(1), 88-104.

Jun, Sangjoon. (2012). Financial Development and Output Growth: A Panel Study for Asian Countries. Journal of East Asian Economic Integration, 16(1), 97-115.

Jung, Woo S. (1986). Financial development and economic growth: international evidence. Economic Development and cultural change, 333-346.

Khalifa Al-Yousif, Yousif. (2002). Financial development and economic growth: another look at the evidence from developing countries. Review of Financial Economics, 11(2), 131-150.

King, Robert G, & Levine, Ross. (1993a). Finance and growth: Schumpeter might be right. The quarterly journal of economics, 108(3), 717-737.

King, Robert G, & Levine, Ross. (1993b). Finance, entrepreneurship and growth. Journal of Monetary economics, 32(3), 513-542.

Levine, Ross. (1997). Financial development and economic growth: views and agenda. Journal of economic literature, 688-726.

Levine, Ross. (2005). Finance and growth: theory and evidence. Handbook of economic growth, 1, 865-934.

Levine, Ross, Loayza, Norman, & Beck, Thorsten. (2000). Financial intermediation and growth: Causality and causes. Journal of monetary Economics, 46(1), 31-77.

Levine, Ross, & Zervos, Sara. (1996). Stock markets, banks, and economic growth: The World Bank.

Loayza, Norman V, & Ranciere, Romain. (2006). Financial Development, Financial Fragility, and Growth. Journal of Money, Credit & Banking (Ohio State University Press), 38(4).

Lucas Jr, Robert E. (1988). On the mechanics of economic development. Journal of monetary economics, 22(1), 3-42.

McKinnon, Ronald I. (1973). Money and capital in economic development: Brookings Institution Press.

Ndlovu, Godfrey. (2013). Financial Sector Development and Economic Growth: Evidence from Zimbabwe. International Journal of Economics & Financial Issues (IJEFI), 3(2).

Pagano, Marco. (1993). Financial markets and growth: an overview. European economic review, 37(2), 613-622.

Saint-Paul, Gilles. (1992). Technological choice, financial markets and economic development. European Economic Review, 36(4), 763-781.

Shaw, Edward S. (1973). Financial deepening in economic development (Vol. 39): Oxford University Press New York.

Singh, Ajit. (1997). FINANCIAL LIBERALISATION, STOCKMARKETS AND ECONOMIC DEVELOPMENT*. The Economic Journal, 107(442), 771-782.

Wu, Jyh-Lin, Hou, Han, & Cheng, Su-Yin. (2010). The dynamic impacts of financial institutions on economic growth: Evidence from the European Union. Journal of Macroeconomics, 32(3), 879-891.