다변량 GARCH 모형을 이용한 한국, 러시아, 브라질 주식시장에서의 충격전이효과 분석

Abstract

Listen

Translate article

This study analyzes inter-regional shock spillover effects among three countries Korea in Asia, Russia in Europe, and Brazil in Latin America using the multivariate GARCH-BEKK(generalized autoregressive conditional hetreroscedasticity-Baba, Engle, Kraft, and Kroner) model to find out the long-term portfolio diversification effects as well as volatility spillover effects in association of time-varying volatility correlations. The whole sample period of this analysis consisits of 4,072 days from January, 1997 to August, 2012 and we divide this total period into four sub-samples composed of three major crisis times and one normal period to take into detailed consideration on further differentiated effects of structural break times and normal times. The three major financial crisis times are composed of 649 days of Asia-Russian FX Crisis Times(January 1997∼June 1999), 522 days of Latin-American Currency Crisis Times(January 2002∼December 2004), and 1,204 days of Global Financial Crisis Times including recent European Sovereign Debt Crisis Times(January 2008∼August 2012). The one normal subsample period comprise 1,043 days from January 2004 to December 2007. Additionally, we use separate multivariate asymmetric GARCH model by GJR(Glosten, Jaganathan, and Runkle) to check further asymmetric volatility transmission effects between up shock and down shock. Major findings from the estimation results over the five entire sample groups composed of one total sample and four sub-samples for the impacts among the three stock market returns and their volatilities by using both symmetric and asymmetric GARCH-BEKK models are as follows. First, VIX, the US stock market volatility index, VIX as a global risk factor in the mean equation model over the whole sample period shows differentiated impacts by individual stock markets contrary to the prior expectation. Namely, VIX is shown to have significant negative impacts on both Korean and Russian stock market return conforming to the expectation from the volatility feedback hypothesis, but have a significant positive impact on Brazilian stock market return. On the while, changes in FX in the basis of dollar value as a country-specific risk factor clearly show significant negative effects on all the three sample stock market returns indicating that currency depreciation as a result of weakened fundamentals affects negatively stock market performance. Second, no clear shock spillover effects from the whole sample period are shown in the estimation of symmetric model of variance equations representing long-term portfolio diversification effects among these three countries. In the asymmetric model estimation over the same sample, shock from Russia is shown to affect positively the other two stock markets. Third, contrary to the total sample period estimation results, those from the 3 crisis times sub-sample periods, Asia-Russia FX Crisis, Latin American Currency Crisis, and Global Financial Crisis including European Sovereign Debt Crisis represent much differentiated phenomena. During the Asia-Russia Crisis period, in spite of the significant positive impact of shock spillover from Russia to Brazil and from Brazil to Korea in the symmetric model, no obvious downward shock spillovers are shown in the asymmetric model, contrary to the prior expectation that Asia-Russian Crisis can be surely to have a positive spillover effect to the global market.