Investigating the Sources of Day and Night Returns

Abstract

Recent research has shown that the premiums of many popular factors are earned at different times of the day. I examine how time of day affects the performance of the Fama and French (1996) and Fama and French (2015) factor models. I find that factor models composed of nighttime returns perform significantly better in explaining average portfolio returns but models using exclusively daytime returns better explain total portfolio variance. In investigating this further I find that daytime returns are sensitive to changes in investor sentiment whereas overnight returns are dependent on macroeconomic conditions largely consistent with Daniel, Hirshleifer, and Subrahmanyam (2001). Additionally, as predicted by the theoretical literature, I find that an increase in personal consumption substantially decreases returns occurring during the daytime when investors are most active.