The Calm Before the Storm (3/08/13) argues that stocks with low abnormal volume convey unfavorable value-related information, indicating negative earnings surprises. Traditionally the post-event drifts for bad news are weaker than that for good news, and as shown in previous research, bad news is priced in more quickly than good news. Being able to relate low abnormal volume with bad news could be useful, especially for short sellers.
Data and methodology
US stock data for 1980-2011 is obtained from Compustat, CRSP, 13-F filings, I/B/E/S and OptionMetrics. Abnormal volume is measured by the ratio of the average share turnover of a stock in a 5-day event period to its turnover in a 50-day reference period prior to quarterly earnings announcement dates. A stock is classified as one with low (high) abnormal volume if this ratio is in the bottom (top) quintile. Earnings surprises are defined as the average of 3-day cumulative abnormal returns (CAR) around announcement, and also as the standardized unexpected earnings forecasts (SUEF). Earnings surprises are regressed on two dummy variables indicating low or high abnormal volume, together with control variables such as past earnings surprises, size, turnover, etc. The analysis is repeated with a constraint on short selling, where stocks with low institutional ownerships or stocks without an exchange-traded option are regarded as more difficult to short. The author repeats the analysis using another 5-day event period with a 5-day gap prior to the announcement date in order to check if the relation between earnings surprises and abnormal volume remains significant. If so, arguments such as changes in visibility (i.e. investors’ attention to the stock) and compensation for risk cannot be used to explain such relation, and it will reinforce the argument based on value-related information.
The author finds that stocks with low abnormal volume have significantly more negative earnings surprises, and such relation is stronger for stocks with more binding short sale constraints. Low abnormal volume continues to show predictive power even when measured with a 5-day gap prior to announcement, but high abnormal volume loses its predictive power possibly because of the increase in visibility. Due to this asymmetry, compensation for risk cannot explain the relation between earnings surprises (weather measured by CAR or SUEF) and abnormal volume.