Serial Correlations

For investors, serial correlation, also referred to as autocorrelation, measures predictability of returns from one period to the next. For example, if ABC returned 3% annualized over the past three years and then averaged close to 3% for the subsequent three years it would have a high serial correlation.

 

U.S. and foreign securities have close to a zero correlation—past returns are a poor indicator as to future performance, whether the period is a month, quarter, year, or decade. Serial correlation figures are typically based on multi-year or multi-decade returns. As noted by Clements (2000), “Past performance is a rotten guide to future results. But that hasn’t deterred investors.” Nevertheless, serial correlation for short-term debt instruments maturing within 1-2 years can be moderate-to-high.

 

One appeal of value investing is mean reversion, also known as the reversal effect. A value investor buys an asset at a discount to its expected value with the expectation of future appreciation. Conversely, the momentum effect is based on a belief good-per-forming investment will continue its appreciation short-term.

 

Asset allocation and risk models assume at least short-term stability return correlations, but “actual covariance and correlation relationships fluctuate dramatically. Correlations tend to increase in volatile periods, which reduces the power of diversification when it might most be desired…an increase in market volatility increases the relative importance of systematic risk compared with unsystematic component of returns…a large portion of the variation in correlation structures can be attributed to variation in market volatility…market volatility contains enough predictability to construct useful forecasts of return correlation…correlation and covariance structures vary dramatically over time—so much so, in fact, and with such high frequency that one must wonder whether asset allocation and risk management models can be of any use whatsoever.” According to Jacquier and Marcus (2001), more than 2/3 of U.S. sector return correlation is due to market volatility.