Is Correlation Constant After All? (A Study of Multivariate Risk Estimation For International Equities)

Abstract

Several studies published since the 1987 stock market crash have found evidence of structure and variability in the correlation of international equity market returns. This evidence has undermined the argument for international diversification and has spawned the development of complex multivariate models to explain variance/covariance. In spite of the well-known evidence for structure in volatility, many of these correlation studies fail to adjust for volatility clustering. After rectifying that problem, this study finds that correlation can be assumed constant with some qualifications. After standardising to allow for changing risk, correlations of daily MSCI returns from 1982-1996 are generally stable. In this context, the extreme market movements of 1987 are seen as an aberration. Implications for multivariate risk estimation, asset allocation decisions and international diversification strategies are examined.