Bear Periods Amplify Correlation: A GARCH BEKK Approach

Abstract

The aim of this paper is to see how correlation changes across time across different indices. We have used a sufficiently large benchmark period of 20 years to have a better understanding as to how correlations1have changed. We compared the correlation in the 20 year period with 3 sub periods namely the Dot Com crisis (1999-2002), the Bullish period (2004-mid 2007) and the Financial Crisis (mid 2007-mid 2009). The results suggest that time varying correlation increases in bearish spells whereas bullish periods do not have a big „statistical‟ impact on correlation. This will have implications for geographical equity diversification since the premise of diversification has been that it lowers risk but a high correlation would imply risk might not be reduced to a certain extent as expected. Therefore, fund managers should take this into account when coming up with equity allocations.