Attempting to follow long stock market trends is a common investment approach, with much guru attention focused on calling long-term tops and bottoms. Is this approach meaningful for investors as an avenue to improve upon buy-and-hold performance? In the December 2006 version of his paper entitled “Analyzing Regime Switching in Stock Returns: An Investment Perspective”, Jun Tu investigates the potential importance to investors of exploiting differences between bull and bear markets within a Bayesian framework that accommodates considerable uncertainty. Using monthly value-weighted stock return and volatility data for July 1963 to February 2006 (512 observations), he finds that:
- The stock market switches between bull and bear regimes with mean annual returns of +12.9% and -11.0%, respectively. Annual overall stock market volatility is 11.4% during bull markets and 21.1% during bear markets.
- The stock market is in the bull regime a little over two-thirds of the time. The monthly probability of switching from a bull (bear) regime to a bear (bull regime) is less than 10% (20%). Switching appears to track the business cycle, with bear markets related to recessions.
- The size effect is roughly 6.1% in bull markets and -3.5% during bear markets.
- The value premium is 3.5% during bull markets and 9.6% during bear markets.
- Because of the substantial differences between the two regimes, portfolio allocations that statistically incorporate bull/bear regime switching are economically superior to those that ignore regime switching.
The obvious hurdle is determining when the regime has switched (or, even better, will switch) between bull and bear. The author’s method (applying Gibbs sampling to a Bayesian inference approach) for determining the regime state of the market is mathematically and computationally too complex for the typical investor. The following chart, taken from the paper, shows the probability of being in a bear market regime based on that method over the entire sample period. The chart also denotes economic recessions (based on National Bureau of Economic Research definitions) by shaded areas on the horizontal axis with vertical dashed and solid lines for recession beginning and end dates.
In summary, portfolio management based on statistically reliable characterization of the long-term trend of the stock market offers an economically significant advantage over approaches that ignore the long-term trend.
