A review of Kothari, S. P., S. Shu, and P. D. Wysocki, 2009, “Do Managers Withhold Bad News,” Journal of Accounting Research 47 (1), 241-274.
I. Research question and its importance
Based on efficient market assumption, and continues the work of Hermalin and Weisbach (2007) this paper investigates whether managers delay the disclosure of bad news relative to good news. Namely, the authors examine whether bad news (i.e., dividend decrease and management forecast revision) lead to higher stock price reactions that the information content of the news is proxied by. This research question is fundamentally important because it fills a literature gap where extant works have focused on the accounting conservativism in recognition, while few studies (e.g., Miller (2002)) focus on conservativism in disclosure. Moreover, the disclosure literature before this paper concerned primarily with litigation risk as the incentive to accelerate disclosure of bad news. Several alternative explanations were, nevertheless, not empirically tested. Hence, this paper advances the study by not only offering a baseline analysis but also ruling out four major competing explanations documented in the literature before landing at its conclusion.
II. Method and findings
In their main model, the dependent variable is stock market reaction measured by cumulative abnormal returns (CAR), while the two independent variables are dividend change announcement and management earning forecast. OLS was employed for the analysis. The authors test their hypotheses on both full sample and a sub-sample that ownership data was available for. There are three major findings. First, the stock market reaction to good and bad news is asymmetric. Second, the release of bad news is, on average, delay. Finally, this asymmetry decreased after the implementation of Regulation Fair Disclosure.
III. Future research
The paper distinguishes itself by utilizing parsimonious design (i.e., descriptive statistics and OLS) in answering the research question. The authors’ elaboration of arguments was impressive and helps in justifying their choice of measurements. In the meantime, they manage in finding supports for their arguments simply with two kinds of tests: t- and F-statistic. Their choice of data and measurement follows convention, and hence the conclusion is well grounded. Yet, its fundamental assumption remains a matter of subjectivity. Conceptually, future research might examine the proposition in a different setting, such as one deviating from efficient market assumption which this paper was based on. As the recent sentiment literature points out, stock market is not as efficient as assumed. Investors might not have the ability to “rationally refer the remaining good news based on the disclosed portion” as the authors postulated on Page 267. Furthermore, under such different assumption about market efficiency, more sophisticated methodologies become essential. For instance, event study that exploits voluntary disclosure to assess change in CAR can be one way to measure stock market reaction. Ultimately, the data which can provide more direct evidence of good/bad news, such as conference call scripts, is becoming more available. Future work might leverage new data as well as methods to better address this research question under different market efficiency assumptions.