Option Trading and the Relation between Price and Earnings: A Cross-Sectional Analysis

Prior research suggests that option trading affects the availability and timeliness of predisclosure information about firms and that the price-earnings relation is influenced by characteristics of the predisclosure information environment. Motivated by these research findings, this study (1) examines various firm-specific attributes that are likely to explain the different information environments of firms with and without exchange-traded options, and (2) investigates the price-earnings relation of such firms. The price-earnings relation is examined in both "event study" (short-window) and "association study" (long-window) contexts. The short-window analysis tests the hypothesis that the "surprise" in quarterly earnings reports (measured by the abnormal stock return variability around earnings announcements) is greater for nonoption firms than for option firms (hypoth. H1). The long-window analysis tests the hypothesis that security prices signal future earnings changes earlier for option firms than for nonoption firms (hypoth. H2A). The hypothesis also predicts that nonoption firms are more likely than option firms to exhibit post-FYE (fiscal year-end) drift (hypoth. H2B). The results indicate that option firms are associated with five firm-specific attributes: (1) larger firm size, (2) higher institutional concentration, (3) higher analyst coverage, (4) higher trading volume, and (5) more Wall Street Journal Index news releases. Based on 3,721 quarterly earnings announcements of 431 firms during the 1980-1983 period, the results support the first hypothesis; abnormal return variability surrounding quarterly earnings announcements is significantly greater for nonoption firms than for option firms. To examine how the five firm-specific attributes provide alternative explanations for the observed results, the empirical tests are repeated by using a unilateral matching approach. Five subsamples (each based on one of the five variables) are constructed. The results for each support the first hypothesis. In addition, a control portfolio is constructed that conservatively controls for the five proxy variables (i.e., the nonoption firms in this portfolio are associated with larger firm size, higher institutional concentration, higher analyst coverage, higher trading volume, and more Wall Street Journal Index news releases). With a relatively smaller sample size, the control portfolio results also support the first hypothesis. Finally, multiple regression results indicate that option trading possesses incremental explanatory power over the other variables in explaining the differential content of earnings releases for option and nonoption firms. The only other variable that is significant in the regression analysis is the number of Wall Street Journal Index news releases, which is negatively related to measured information content. This suggests that press coverage is also an important variable in explaining cross-sectional differences in information content, at least for the sample firms. With respect to the second hypothesis, the results for the entire sample, the matched subsamples, and the control portfolio indicate that the security prices of option firms anticipate accounting earnings earlier than do those of nonoption firms. For option firms, about 50 percent of the price change associated with economic events contributing to the current year's earnings change is realized in the previous fiscal year. For nonoption firms, however, a significant portion (about 70 percent) of the price change associated with the earnings change occurs in the current fiscal period and in the 12 months following the fiscal year-end. In addition, the magnitude of the post-FYE drift is greater for nonoption firms than for option firms. Certain important caveats apply to the results. First, given that the tests for the matched subsamples attempt to control only for one factor at a time, the results are subject to a missing-variable problem. Second, while the option-trading variable exhibits statistical significance in the regression analysis, the explanatory power of the full regression is low (although it is similar to that achieved in typical cross-sectional analyses of abnormal security returns). Finally, given the lack of a formal theory concerning option trading and information flows, it is difficult to infer that option trading causes the differences in the price-earnings relation across firms.