Time-varying stock return correlation, news shocks, and business cycles

The cross-sectional average of pairwise correlations across stocks traded on the NYSE, AMEX, and Nasdaq is a powerful predictor of U.S. economic activity at a horizon of one to four years. Its predictive ability is on a par with the slope of the yield curve and significantly exceeds that of some other widely used financial indicators. The macroeconomic effects of an innovation to stock return correlation in a vector autoregression are nearly identical to those of a news shock about future productivity. Thus, market-wide changes in return correlation contain information about changes in future technological developments.