Comovement revisited

Evidence of excessive comovement among stocks following index additions (Barberis, Shleifer, and Wurgler, 2005) and stock splits (Green and Hwang, 2009) challenges traditional finance theory. We show that the bivariate regressions in this literature provide little information about the economic magnitude of excess comovement, with coefficients that are sensitive to unrelated factors. Using robust univariate regressions and matched control samples, almost all evidence of excess comovement disappears. In both examples, the stocks exhibit strong returns prior to the event, akin to momentum winners. We document that winner stocks exhibit increases in betas, generating much of the apparent excess comovement.