Time series data from the first quarter of 1980 to the second quarter of 2008 are used for the analysis. The definitions for variables and data sources are presented in Table 4. The data were sourced from the Korean National Accounts and ECOS of The Bank of Korea. The dependent variable used is real private investment and since the quarterly data for real private investment for 1980-2003 are not available, they were estimated using the yearly data of private fixed capital formation and the quarterly structure of real fixed capital formation. All variables were converted into real terms. The GDP deflator is used for interest rates, stock prices and capital prices, the CPI for exchange rates and the PPI for domestic credit. All series, except real interest rates and outward FDI, are logged for the estimation of the model.
Table 4
92
4.3 Empirical Results
To estimate the private investment function in Korea, an error correction model was used to show long-run cointegration relationship and fit of short-run dynamic relationships, with the inclusion as the regressors, the error terms derived from the long-run equations. As a preliminary to estimation, the unit-root hypothesis was carried out using the Augmented Dickey-Fuller (ADF) test and the Phillips- Perron (PP) test. The results are presented in Appendix Table A-10. The results show that none of the variables can reject the null hypothesis of a unit-root at 5% or less and they are integrated of order one. This means that this time series data has a non stationarity problem and an estimation without considering the problem would result in a spurious regression.
On this basis, cointegration tests are required. The unit-root tests on the error terms derived from the long-run equations in Tables 5 and 6 were conducted and the results are shown in Appendix Table A-11. The test results show that all the error terms can reject the null hypothesis of a unit-root at 5% or less. This means that there exist stable long-run_cointergrating_relations between investment, GDP, domestic credit, interest rates, exchange rates, stock prices, capital prices and outward FDI as presented in Tables 5 and 6, during the first quarter of 1980 to the second quarter of 2008.
The estimated long-run equation (8) shown in Table 6 suggests that a 1% increase in GDP, domestic credit and stock price stimulates investment spending in the long run by 0.36%, 0.48% and 0.05%, respectively. A 1% rise in the exchange rate and capital prices leads to a 0.25% and 1.01% decrease in investment, respectively. The negative sign of the exchange rate’s coefficient implies that the negative effect of a rise in exchange rate through increasing import prices is stronger than the positive effect stemming from the increase in exports. A 1% rise in interest rates reduces investment spending in the long run by 0.63%, which confirms the policy channel of investment fluctuations, that is, an increase in interest rates a decrease in investment. The negative coefficient (-0.20) of outward FDI suggests that overseas direct investment has a crowding- out effect on domestic investment. In addition, the negative coefficient of the dummy implies that there was a structural change in firms’ investment behaviour after the crisis which shows the stagnation of investment following the currency crisis
Table 5
Estimation Results of Private Investment Function
(Long-run equations)
Notes: 1) The t-values are in parentheses.
94
Table 6
Estimation Results of Private Investment Function
(Long-run equations)
Notes: 1) The t-values are in parentheses.
To capture the short-run dynamic adjustment process among the variables, based on the long-run cointegration equations fitted above, the error correction model was set up as in equation (2). The lag variables were selected according to the general to specific method.
(2)
where INV: real private investment, EC: the errors term of the long-run cointegration equations, GDP: real GDP, DC: real domestic credit to private sector, IR: real interest rates, ER: real exchange rates, SP: real stock prices, CP: real capital prices, OUTFDI: outward FDI and Dummy: crisis dummy. The estimation results in Tables 7 and 8 show that GDP and domestic credit have positive effects on investment demand while interest rates, exchange rates, capital prices and stock prices have negative effects. The estimates of the error correction terms are statistically significant with negative signs, implying that investment growth tends to return to equilibrium from the next period after deviating from the long-run equilibrium in response to external shocks and, at the same time, confirming the reliability of the long-run equations fitted above. A closer look at the short-run dynamic equation (8) in Table 8 suggests that a 1% rise in GDP increases investment growth by 1.74% while a 1% increase in domestic credit growth leads to 0.26%p increase in investment growth. Furthermore, interest rates, exchange rates, capital prices and outward FDI have negative signs as well as being statistically significant as expected, having coefficients of -0.33, -0.28, -0.31 and -0.07 respectively.
96
Table 7
Estimation Results of Private Investment Function
(Short-run dynamic equations)
Notes: 1) The t-values are in parentheses.
2) *, **, *** means that the estimates are significant at the 10%, 5% and 1% levels,
Table 8
Estimation Results of Private Investment Function
(Short-run dynamic equations)
Notes: 1) The t-values are in parentheses.
98