Short selling prior to going concern disclosures

Abstract

We provide insights into how the market processes going concern audit opinions based on the trading of some well-documented sophisticated investors – short sellers. We find that abnormal short selling increases significantly upon impending going concern disclosures. While prior literature attributed much of short selling around some corporate events to private information, we find evidence that pre-going-concern announcement short selling reflects both privately informed trading and processing of public information by short sellers. Further, a negative relation between pre-announcement short selling and post-announcement short-term stock returns exists for stocks with less short sale constraints. We also find moderate evidence associating short selling with subsequent bankruptcy to some extent. Overall, these results suggest that short sellers front run going concern announcements based on private information and fundamentals, although trading constraints prevent them fully impounding the severity of negative information in the short run, providing a partial explanation for the long-run price drift post-going concern.

Keywords:

• Going concern audit opinion
• Short sale
• Informed trading
• Short sale constraints

JEL Classification:

• G12
• G14
• M42

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 We would like to caution the audience early on to the issue that a large amount of information is disclosed simultaneously in corporate financial reports. While this is a common concern to the analysis of similar nature, we control for concurrent financial ratios, filing delays, auditor choices, as well as excess stock returns, and the effect we document for audit opinion releases is thus incremental to the above-controlled factors.

2 D’avolio (2002) shows that low-priced stocks belong to a constrained category and are difficult to short.

3 Securities and Exchange Commission (SEC) (1999). Short Sales (SEC Concept Research No. 34-42037. File S7-24-99. October.) Washington, D.C.: Government Printing Office.

4 As pointed out in previous studies (Boehmer and Wu 2013, Diether et al. 2009), we note that the SHO database of the daily shorting volume lacks information on shorting cost and when and how short sellers cover their position. Thus, we cannot determine whether short sellers’ trades are profitable.

5 In un-reported analysis, we measure the average short selling using an alternative benchmark period of (-150, -61) and (-60, -31). The results hold.

6 Note that given the sample size of going concern firms as well as the daily short-selling data, we choose to apply a simple matching procedure instead of calculating the propensity score based on variables in studies such as DeFond et al. (2002). We also tried to use all variables suggested by DeFond et al. (2002). However, the requirement of these additional variables (e.g., bankruptcy Score, beta, and future finance) causes a significant portion of our sample unmatched.

7 For the majority of firms, GCOs are released in the annual report which have to be filed with the SEC within 60–90 days after the company's fiscal year end (90/75/60 days for non-accelerated filer/accelerated filer/large accelerated filer).The exact release date is thus unknown to the public ex-ante but up to negotiation between the auditor and its client firm, not to say it is of uncertainty whether GCOs would be issued by the auditor in the first place, making it different from some other events such as earnings announcements and management earnings forecasts which can be expected within a predictable and repetitive timeframe. For example, Clinical Data Inc. released its annual report on 6/27, 6/29, and 6/19 for the fiscal years of 2005, 2006, and 2007, respectively. Thus, it is generally hard to pin down the exact release date on a repetitive basis.

8 A caveat is that short sellers can anticipate that release will happen more likely closer to the 90th day than earlier, therefore leading to possible path dependency and expectations in terms of timing of the release of information. We attempt to alleviate the issue by including a timing variable—Filing Delay which measures the gap between filing date and fiscal year-end date. Interestingly, we find this variable to be negatively related to the level of short selling, which partially alleviates our concern regarding short sellers are timing late filers.

9 We appreciate an anonymous referee raising this alternative possibility.

10 Other less commonly cited reasons include industry downturn, loss of major customers, difficulty in competitive environment, failure of mergers, lawsuits, etc.

11 We also tried the cutoff using the mean ($2.303) or the median price ($1.194), and the results of Table 7 remain qualitatively similar.