Synchronize Your Data or Get Out of Step with Your Risks

Daily stock market data are used for many purposes, ranging from the valuation of mutual funds to risk models for trading desks and multifactor risk models offered by commercial software vendors. However, daily returns for global equity portfolios also display spuriously low correlations, as trading time across major financial centers imperfectly overlaps. Although this fact is well known (especially since Burns, Engle, Mezrich [1998] offered a practical solution), it has received little attention among practitioners. We show that underestimating the correlation leads to increasingly bad value at risk (VAR) estimates during periods of rising volatility. The credit crisis in 2008 and 2009 provides the backdrop for our case study.