CROSS-SECTIONAL CORRELATION: NEW EVIDENCE ON CHANGING CORRELATIONS AND CORRELATION BREAKDOWN IN EQUITY MARKETS

As an alternative to the traditional time-series based correlation measure Solnik and Roulet (2000) suggests a different correlation measure based on the cross-sectional dispersion of asset returns. This paper focuses on this cross-sectional correlation and finds that the level of correlation in the European equity market has increased between 1970 and 2000. In addition, in the light of the increased country linkages since the October crash 1987, it is found that diversification along industry lines is a feasible alternative to international diversification. Finally, using simulations, it is shown that the cross-sectional correlation measure, just as the traditional sample correlation, faces the problem of a spurious correlation breakdown bias when the absolute average return is large. The size of the bias depends on how volatility is estimated, however, and the bias disappears when the volatility is approximated with the constant unconditional volatility.