Intraday Variation in Cross-sectional Stock Comovement and Impact of Index-based Strategies

Abstract

We investigate how stock comovement changes during a day using a large high-frequency data set of S&P 500 stocks and estimators that are efficient in the high-frequency setting. We measure cross-sectional stock comovement by average pairwise correlation and cross-sectional dispersion in beta. Both measures exhibit substantial intraday variation. In the recent decade, the stock correlation increases across the trading session, while the cross-sectional beta dispersion decreases. These patterns suggest that stocks exhibit stronger comovement near the market close. We explain the time-varying stock comovement by the intraday variation in the composition of index-based and firm-based strategies, providing new evidence for the demand-based view of comovement. We empirically support this mechanism using intraday trading volume and change in comovement pattern on unusual trading days. Finally, we develop a cross-section market impact model that generates the observed intraday variation.

Keywords: intraday stock comovement, index-based investing, market microstructure, highfrequency estimation, big data

JEL Classification: G11, G12