Purpose
– The purpose of this paper is to examine the January effect, a well-documented capital markets pricing anomaly in which January return premiums are observed to be on average higher than in other months of the year. Extant literature focusses primarily on investor trading behaviors and incentives. This study is different in that it investigates the link between the unusually high returns characteristic of the January effect and accounting earnings, a popular measure that investors use to judge firm value.
Design/methodology/approach
– The empirical model used in this study is derived from the analytical framework of Ohlson (1995) and Feltham and Ohlson (1995), which explains returns as a function of current and future accounting earnings. Isolating firms that exhibit January effect return premiums from those that do not offers a deeper look at the characteristics of the anomaly. Regression analyses are carried out using a modified Fama-MacBeth (1973) methodology. Quarterly earnings and returns data are drawn from Compustat and CRSP.
Findings
– The main finding is that the association between January returns and first quarter earnings is unexpectedly and significantly negative, not positive as predicted by the model. Coefficient signs for the other three quarters behave as expected. Additional analyses highlight a difference in the returns-earnings association between firms affected by the anomaly and those that are not. Robustness checks indicate that the findings are not spurious.
Originality/value
– Rather than applying trading or multifactor economic models that rely on some level of market inefficiency or irrational investor behavior, this study uses an accounting valuation approach that relies on neither. The unexpected negative association between January effect returns and earnings suggests that other factor(s) besides earnings may play into valuation judgments for investors in such firms, and invites further research.
A model for a large investor trading at market indifference prices. II: Continuous-time case
