Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Leveraging the systematic variations in investor clientele within a day, we validate an adapted version of the Hong and Stein (1999) model that addresses the consequences of slow information diffusion in China. The model predicts that overnight returns, rather than total returns, strongly forecast future returns, as informed overnight clientele underreact to value-relevant signals. Empirically, we establish a consistent overnight trend phenomenon: Firms with a strong overnight trend reliably outperform those with a weak overnight trend in the subsequent month. The phenomenon is more pronounced among stocks with higher levels of information asymmetry, valuation uncertainty, and relative mispricing. Furthermore, the overnight trend predicts positively firm fundamentals in the cross section.