The Asymmetry of Market Efficiency for Positive and Negative Earnings Surprises

Abstract

I presume that due to conditional conservatism of investors, market efficiency is greater for bad news than for good news. I test the validity of this presumption by using earnings surprises as representatives for news. Positive earnings surprises represent good news and negative earnings surprises represent bad news. My research design is essentially an analysis of earnings response coefficients (ERC) and post earnings announcement drift (PEAD), i.e., I regress short term cumulative abnormal stock returns (CAR) around earnings announcements and long term CAR after earnings announcements on standardized unexpected earnings (SUE). However, unlike most prior ERC or PEAD studies, I use dummy variables to discriminate between positive and negative SUE. The regression results support my presumption because they show an asymmetry in the return reaction to earnings surprises. Positive earnings surprises have a positive effect on both short term CAR around earnings announcements and long term CAR after earnings announcements. Negative earnings surprises, on the other hand, have a negative effect only on short term CAR around earnings announcements. My finding that only positive earnings surprises have an effect on long term CAR after earnings announcements implies that the results of PEAD studies that do not discriminate between positive and negative earnings surprises can be misleading.