Macroeconomic Fundamentals and the Volatility of Foreign Investors’ Net Purchase in Korean Stock Market

Abstract

We employ the GARCH-MIDAS model to examine whether low-frequency macroeconomic variables help to explain the high-frequency volatility of the foreign investors’ net purchase in Korean stock market. The estimation results show that business cycle expansion along with high production growth, high inflation and low unemployment rate predicts high volatility in the near future. Higher interest rate and lower money growth are also likely to lead to higher volatility of foreign investors’ net purchase. We also find that domestic macroeconomic variables, relative to the US variables, have a stronger correlation with the future volatility of foreign investors’ net purchase.

KEYWORDS:

• GARCH-MIDAS
• volatility
• foreign investor
• macroeconomy

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The correlation coefficient between the changes in KOSPI and the foreign investors’ net purchase is 0.53 for the sample period from January 2005 to June 2022.

2 Choe et al. (1999) find no evidence that foreign investors’ trading had a destabilizing effect on the stock market in Korea around the 1997 currency crisis.

3 Appendix provides the estimation results for the GARCH-MIDAS model with other selected macroeconomic variables.

4 Appendix shows that the estimated ’ for retail sales and employment are also positive and statistically significant.

5 We examine the effects of the volatility of macro variables in Appendix. We use the squared residuals from AR(1) regressions as the measure of the volatility of monthly macro variables. The estimates on the volatility of real sector variables are negative and statistically significant but the coefficient estimates are very small except CPI. To the extent that macroeconomic volatility is relatively low during business cycle expansions, the negative estimates on the volatility are consistent with positive association between business cycle and net purchase volatility. we also consider uncertainty measures (Economic Policy Uncertainty index and VIX) and sentiment index (economic sentiment index and business survey index). We find negative coefficient estimates on the uncertainty measures. As long as uncertainty is low in business cycle expansion, this result is consistent with the finding of the positive association between business cycle and the future volatility. Since the sentiment index is pro-cyclical, the coefficient on the sentiment index is estimated to be positive.

6 Figure 2 shows that macro variables cannot predict the sharp increase in volatility during the crisis periods, implying that the high volatility in recessions (crisis periods) are mostly from short-term component, not associated with macro fundamentals. The empirical results, however, indicate a positive relation between business cycle and the long-term component of the volatility although business cycle is negatively associated with the (total) volatility which includes both the short-term and the long-term component.

7 The correlation coefficient between interest rates on 10-year Korean government bond and 10-year US Treasury bond is 0.8351 during the sample period.

8 If we include multiple macro variables simultaneously, the model often fails to converge. Given the correlations among macro variables, we need great caution when comparing the marginal effects.