Working Papers 2003 – Abstracts

Diversification Meltdown or the Impact of Fat tails on Conditional Correlation?

Rachel Campbell, Catherine S. Forbes, Kees Koedijk and Paul Kofman

A perceived increase in correlation during turbulent market conditions implies a reduction in the benefits arising from portfolio diversification. Unfortunately, it is exactly then that these benefits are most needed. To determine whether diversification truly breaks down, we investigate the robustness of a popular conditional correlation estimator against alternative distributional assumptions. Analytical results show that the apparent meltdown in the benefits from diversification could be a consequence of assuming normally distributed returns. A more realistic assumption - the bivariate Student-t distribution - suggests that constant correlation may be sustained over the full support of the multivariate return distribution

Keywords: Conditional correlation, Truncated correlation, Bivariate Student-t correlation.

Next Abstract