Industry equi-correlation: A powerful predictor of stock returns

Abstract

We show that the detrended equi-correlation of the returns of industry portfolios is a strong predictor of excess returns to the S&P 500 Index. Using a sample from 1927 to 2015, our monthly industry equi-correlation (IEC) index produces an out-of-sample R 2 of as high as 0.888%. For an investor with mean–variance utility, the IEC index can generate utility gains of 120.5 basis points over the benchmark model of the historical average. The return predictability of the IEC index is stronger than that of all of the popular predictor variables. Furthermore, we find that incorporating IEC in a univariate predictive regression with a popular predictor can significantly improve the out-of-sample forecasting performance of the individual models and their forecast combinations. These findings are confirmed by a large battery of robustness checks. • The equi-correlation of industry return (IEC) can predict aggregate market return. • The monthly IEC index produces an out-of-sample R 2 of 0.888%. • The IEC index generates utility gains of 120.5 bps over the benchmark model. • Adding an IEC index to a popular predictor model improves predictive ability.