Limited information-processing capacity and asymmetric stock correlations

Through an orthogonalized impulse-response analysis, I studied the relationship between the variance risk premium, market variance and stock correlations in the French stock market from September 2002 through September 2006, using high-frequency data-based measures. Variance risk premium is estimated using realized variances and index options-implied variances and used as a state vector to proxy investors perceived uncertainty. I found that a shock to variance risk premium causes long-lasting increases in the market variance pointing to the limitedness of investors information-processing capacity. At the same time, the shock generates consecutive increases in realized correlations between individual stocks and the market portfolio. I propose this as a possible explanation for the asymmetric/counter-cyclic behaviour of stock correlations.