Blog - Investing Notes

February 13, 2009 - The Downside Risk Factor

Is downside risk (beta as measured only during overall market declines) better than normal beta as an indicator of differences in future returns among individual stocks? In their February 2009 paper entitled "Sorting Out Downside Beta", Thierry Post, Pim Van Vliet and Simon Lansdorp explore empirically the power of downside beta to explain the variation in future returns across stocks. They measure beta and downside beta based on five years of historical stock and market returns. Using return data for a broad sample of U.S. common stocks spanning 1926-2007 (82 years) and various measures of downside beta, they conclude that:

• Sorting stocks by downside market beta better explains the differences in future returns of these stocks than sorting by regular market beta.
  • A value-weighted hedge portfolio, rebalanced annually, that is long (short) the tenth of stocks with the highest (lowest) beta generates an average annual return of 3.7% before trading frictions.
  • A value-weighted hedge portfolio, rebalanced annually, that is long (short) the tenth of stocks with the highest (lowest) downside beta generates an average annual return of 5.5% before trading frictions.
• Downside beta outperforms beta in explaining future returns of individual stocks in four subperiods, but the explanatory powers of both kinds of beta vary considerably. During 1970-1988, low beta/downside beta stocks tend to outperform high beta/downside beta stocks.
• The explanatory power of downside risk persists after controlling for firm size, book-to-market ratio and momentum.

The following chart, constructed from data in the paper, summarizes the variations in U.S. common stock future returns (in excess of the one-month Treasury bill yield) with beta and downside beta over the entire 1926-2007 sample period. The horizontal axis segments sample stocks into ten annually rebalanced portfolios based on level of beta/downside beta as derived from the previous five years of individual stock and overall stock market returns. Downside beta measurements come only from intervals during which the market is declining.

Results appear uncompelling in support of the headline conclusion that downside beta works better than beta in explaining differences in future returns among individual stocks. This conclusion seems critically dependent on the very high average return for stocks with the highest tenth of downside beta. Conclusions based on other deciles are often different.

In summary, evidence weakly suggests that downside risk (downside beta) may be a better indicator of differences in future returns among individual stocks than normal beta.

For related research, see Blog Synthesis: Volatility Effects.

To discuss this research, go to the Volatility Effects Forum.