CHAPTER 4: ESTIMATION RESULT ANALYSIS AND DISCUSSION
4.2 Regression results and discussion
4.2.1 Impact of economic growth on financial development
To examine the impact of economic growth on financial development, we treat financial development (FD) as the dependent variable. The independent variables are the set of determinant variables of the economic growth (EG), such as Foreign direct investment (FDI); General government consumption (GOV); Trade openness (TRAO);
and one year lagged GDP per capita growth (GDPGRTHt-1). Determinants of the financial development are also included in the right hand side of the estimated equation (4).
In this estimation, we choose one year lag of the Ratio of domestic credit to the private sector to GDP (DOMCREDITGDPLAG1) to diminish the multicolinearlity problem as reported in the correlation matrix of Table 5.
We assume that the instrumental variables for financial development variable (FD) include the Ratio of Broad money to GDP (M2/GDP), Ratio of credit offered by bank to private sector to GDP (BANKCREDIT/GDP), Ratio of Gross domestic savings to
Page 56
GDP (GROSSDSA), and one year lag of Ratio of domestic credit to private sector to GDP (DOMCREDITGDPLAG1).
The result of regressing economic growth on financial development is summarized in Table 6 as bellow:
Table 6: GMM estimation of Economic Growth and Financial Development Independent variables: Dependent variable: FD
(IV-GMM)
Constant -7.619407 ***
(1.766992) (p-value = 0.000)
EG
1.213075 **
(0.4956303) (p-value = 0.14)
GROSSDSA
1.107113***
(0.0767307)
(p-value = 0.000)
M2GDP
0.986736***
(0.052102) (p-value = 0.000)
BANKCREDITGDP
1.243086 ***
(0.0497255) (p-value = 0.000)
DOMCREDITLAG1
0.7242191 ***
(0.0689256) (p-value = 0.000)
R-squared 0.9950
Page 57 J statistic (Hansen test)
Test of over-identifying restrictions
4.07701*
(p-value = 0.132)
Robustified DWH test testing endogeneity
7.24548 **
(p-value = 0.0074)
Observations 456
Note: * Significant at 10% ; ** Significant at 5% ; *** Significant at 1%
Table 6 reports the results from the estimation of the equation with financial development (FD) as the dependent variable. The coefficient of economic growth is significant at 5 percent and positive for financial development. The coefficient of economic growth (EG) is 1.213, implying that a 10 percent increase in the growth rate of GDP per capita would increase the average financial development (FD) by 12.13%
annually holding other things unchanged. This result is consistent with the findings in previous empirical literatures (Calderón and Liu., 2003; Hassan et al., 2011;
Habibullah and Eng.,2006; Levine et.al., 2000), which confirm the positive relationship between financial development and economic growth, and emphasize the important role of economic growth in financial development.
In addition, the coefficient of the Ratio of Gross domestic savings to GDP (GROSSDSA) is positive and significant at the 1% level for financial development.
The coefficient of GROSSDSA is 1.107, implying that the ratio of Gross domestic savings to GDP increases 10 percent, the average financial development would increase by 11.07% annually provided that holding other things unchanged. The ratio of Gross domestic savings to GDP indicates that a high percentage of savings used for investment leads to positive enhancement in economic growth. Because investment is considered as a one channel through which financial development fosters economic growth (Hassan et al.,2011).
Page 58
Furthermore, the ratio of Broad money to GDP (M2/GDP) is significant at the the 1%
level. As expected, the coefficient is positive for financial development. This M2/GDP indicator is implied that the financial system achieves the development if the ratio of broad money to GDP is high. Hence, financial sector becomes larger (Hassan et al., 2011; Calderón, C., & Liu, L., 2003).
In addition, the coefficients of the ratio of credit offered by bank to private sector to GDP (BANKCREDIT/GDP) and the ratio of domestic credit to the private sector to GDP (DOMCREDIT/GDP) are positive and significant at the 1% level for financial development. As argued by Hassan et. al., (2011), the ratio of domestic credit to the private sector to GDP (DOMCREDIT/GDP) measures how much financial intermediaries provide financial resources to the private sector for consumption, investment and working capital finance. Similarly to the ratio of credit offered by bank to private sector to GDP (BANKCREDIT/GDP), this indicator implies higher financial development due to banks tend to well perform the five financial functions as discussed in Levine (1997). In other words, a high ratio of domestic credit to GDP indicates a higher level of domestic investment will lead to a higher development of the financial system. Allocating more credit to private sector requires the financial system must well perform its basic five functions such as researching borrower firms, exerting corporate control, providing risk management control, facilitating transactions, and mobilizing savings. This requires a higher degree of financial development (Hassan et. al., 2011).
In the Table 6, the result from testing endogeneity in this case show that the DW-h test is robust and significant at the 5% level (i.e. p-value (0.0074) < 0.05). This result leads to strong rejection of the null hypothesis (H0) that the variables are exogenous. Hence, we conclude that the variable economic growth (EG) is endogenous.
We test the validity of the instrumental variables by using the indicator Hansen-Sargan test of over-identifying restrictions. The null hypothesis (H0) indicates that all instrumental variables are valid. The J statistic, shown in Table 6, is significant at the
Page 59
10% level (i.e. p-value (0.132) > 0.1). Therefore, we fail to reject the null hypothesis (H0). We come to the conclusion that the instrumental variables in the estimated equation are uncorrelated with the error terms. The instrumental variables are valid.
In addition, since we fail to reject the null hypothesis that all instrumenting regression variables are weak instrumental variables at 1 percent level of significance (the p-value is 0.0000. We report the results of testing for weak instrument in Appendix C), therefore, we recognize that the instrumenting regressions are not weak instruments.
The explanatory power of these instruments are high.