How Stock Option Backdating Can Be Detected

In our study completed in June 2009, we offer an analysis of the economic consequences of stock option backdating into areas that previous researchers have been unable to address.  We are the first study to explore the economic consequences associated with backdating that has remained undisclosed to the public (and therefore to the research community).  We find that undisclosed backdating companies suffer negative stock returns, lower accounting returns, and a rate of negative stock market delisting reasons that are all at least as bad, in some cases worse, than disclosed backdating companies.  This has significant theoretical implications, suggesting that market participants are able to identify and punish backdating companies even if regulators do not, and that vigorous disclosure and enforcement may actually mitigate losses associated with backdating.

 

Previous authors (e.g., Lie 2005) have asserted, and the community has apparently accepted, that public filings and stock price data alone cannot be used to establish a sample of companies that committed backdating.  Thus, previous studies have either directly aggregated grant-level data (e.g., Lie 2005, Heron and Lie 2009) or used samples of publicly-disclosed backdating companies (e.g., Bernile and Jarrell 2009, Armstrong and Larcker 2009).  Our study now makes clear that both approaches have (a different set of) significant shortcomings, and that neither can be used to investigate the economic consequences to undisclosed backdating. 

 

We have spent the last two years developing a statistical test that can determine if a company’s entire pattern of grants is abnormally favorable at a pre-determined cutoff level.  This allows us to derive the first sample of both disclosed and undisclosed backdating companies that is free from biases due to the disclosure process.  While the technique and the sample are not ends in themselves, they are necessary means to an end of determining the economic consequences to undisclosed backdating companies.

 

We also expand the analysis to include, for example, associating characteristics with undisclosed backdating companies and investigating aspects of the disclosure process itself.  We find that small market capitalization companies are strongly overrepresented in the sample of undisclosed backdating companies, indicating a bias in the discovery and disclosure process.  We have also produced the most rigorous measure to date of the ratio of disclosed to undisclosed backdaters.  This is a hotly debated question in the backdating literature, and our extensively supported results (we devote many pages to this and the related issues of error analysis, control testing, and extrapolation from the tested sample to the parent population) strongly challenge the conclusions of a recently published study (e.g., Heron and Lie 2009).

 

In our study, we also perform a series of control tests, reported in Section 4.7., that increased our confidence and allowed us to slightly relax the sample cutoff from 0.04% to 0.05%.  That expanded the high probability (of) backdating sample from 131 to 141 companies.  We have also shortened and improved the efficiency of the SAS code provided in the appendix. 

 

The two additional applications of the statistical technique are to test for backdating at the managerial level (e.g., CFO, CEO, inside and outside directors) and at the individual level, leading to a new finding that managers backdate much more frequently than outside directors.  This last result argues against the suggestion (Dierker and Hemmer 2008, Gao and Mahmudi 2008) that backdating is an efficient contracting mechanism knowingly entered into by managers and directors.  At the individual level, we offer empirical evidence that an individual-level analysis can detect stock option backdating, in some cases, when a company probability value is higher. 

 

In sum, we offer a direct test of whether companies or individuals appear to have obtained extremely favorable pricing of stock option awards.