Even though the basic fiduciary allegations in backdating suits are state law claims, they can be filed in federal court either by alleging diversity among the parties or by adding a federal cause of action, (128) such as one based on faulty proxy disclosures. (129) As Table (1) shows, shareholders filed a majority of backdating suits outside of the incorporation state regardless of the state of incorporation.
Figures 3.A and 3.B present trees of case outcomes. Each node in the trees show the number of cases that progressed to the procedural stage indicated. Of the 255 in-sample firms that appeared on a public list of backdaters, 161 firms were sued. Only thirty-three suits were dismissed outright, with an additional dismissal occurring after a special litigation committee (SLC) was formed. Interestingly, of the thirty-nine cases that made use of an SLC, only eight SLCs recommended that the case be dismissed. In the majority of cases surviving the motion to dismiss, settlement was the most common outcome.
While settlement amounts in securities class actions are readily available, (130) settlements in derivative backdating suits often involved forms of settlement consideration that are hard to value. In reviewing settlement agreements, we found that settlements often involved cancellation of some options, repricing of others, payments by certain defendants, payments by D&O insurers, and corporate governance changes. Moreover, companies often pursued some or all of these remedies outside of the process of settling the derivative claims, but nevertheless still during the pendency of the claims. Plaintiffs' attorneys sometimes executed a settlement agreement that mentioned no benefits to the company, but in their fee applications, sought to claim credit for other developments. At Semtech, for example, the plaintiffs' attorneys argued that their efforts "confer[red] substantial benefits upon Semtech and its shareholders in the form of the cancellation and/or repricing of options with a realizable value of over $9 million and the implementation of significant corporate governance reforms, internal controls measures and equity award procedures and practices." (131) The repricings and cancellations at Semtech, however, had occurred long before the settlement of the derivative claims. (132) The settlement agreement did not even mention repricings or cancellations. (133) When settlements did reprice or cancel options, they were seldom explicit about which grants were cancelled, making valuation impossible.
Thus, we rely on attorneys' fees as a proxy for settlement magnitude in derivative cases. We observe attorneys' fees in 86 cases. The mean attorneys' fee in the settlement of backdating claims was $3,006,000 and the median was $3,751,000. When claims were pending in multiple courts, companies would sometimes reach separate settlements with the attorneys in each jurisdiction, or sometimes the settlement would expressly allocate the fees among the various attorneys. (134) We measure fees as the total settlement attorneys' fees for each targeted company.
This Part presents the results of our analyses. We examine which companies were targeted by backdating lawsuits, the number of complaints filed, which claims were dismissed, settlement size for securities class actions and the related measure of attorneys' fees for derivative suits, and the use and outcomes of special litigation committees.
The number of firms implicated publicly in options backdating is smaller than our estimate of the number of firms that engaged in backdating with high probability. A majority, but not all, of publicly-implicated firms were sued derivatively, and only a subset of those firms were targeted by securities class actions. This makes the question of which firms were sued for backdating an interesting one, as some companies that backdated with high confidence were never subject to suit, or even publicly implicated in backdating. (135) Our empirical estimate of the number of firms that engaged in backdating provides a partial explanation of why some firms may have escaped suit: it is not possible to determine, based purely on reversal around grant dates, whether a firm engaged in backdating, because some firms will have high reversal by happenstance. (136) While we can estimate the total number of firms that likely backdated options, identifying specific firms is another matter. Nevertheless, the question of which firms were targeted for private suit, and how those firms compare to those that are, say, targeted by the SEC is an interesting one.
We begin by plotting, in Figure 4, relative densities of Firm-Level Backdating Probability for four groups of firms: firms investigated by the SEC, firms sued privately, firms implicated in options backdating by lists compiled by the Wall Street Journal and other publications described above, (137) and the full sample of firms.
Firms may fall into more than one of these categories; for example, a firm may have been investigated by the SEC and also sued privately. In this plot, firms in multiple groups are treated as members of both groups. Thus, the curve for private suits includes, for example, firms that were both sued and the subject of SEC investigations.
If option grants were assigned randomly, the density plot would be flat, and firms would be lucky and unlucky in equal numbers. The plots, however, are skewed with a greater density of firms on the higher end of the probability distribution, showing that firms implicated in backdating were indeed "luckier" in their grants than other firms. Figure 4 also reveals that sued firms and firms investigated by the SEC had even more striking grant reversals than firms publicly implicated. Such a relationship is elementary: implicated firms were very likely to have backdated, as were firms sued or investigated for backdating. What is more surprising is that the kernel density for private suits is above the curve for implicated firms. That suggests that plaintiffs' attorneys exercised some discretion. While there are many more private suits than SEC investigations, the private suits, in the aggregate, do not appear to be of much lower quality.
Figure 5 plots slightly different kernel density curves, this time using non-overlapping groups so that the comparison is between firms with SEC investigations and privately-sued firms with no SEC investigation. The likelihood of backdating for the privately-sued firms that were not investigated by the SEC appears to be lower than that of the SEC-investigated firms but higher than that of firms publicly implicated in backdating that were not sued, once again suggesting some selection on the merits. In this plot, the curve for sued firms is well above the curve for implicated-but-unsued firms.
Figure 6 compares backdating probability in securities class actions with backdating probability in derivative suits. The figure shows that class action complaints were more likely to be brought against companies with a significantly higher likelihood of backdating than derivative suits. The class action defendants appear to be the most egregious backdaters among the firms implicated.
An important question suggested by the density plots is the extent to which plaintiffs' attorneys were selective in targeting firms for backdating lawsuits based on the merits, controlling for other covariates that might affect selection. Tables 4.A and 4.B addresses this question using logit regressions, with the incidence of a private lawsuit as the dependent variable. The question is whether variables that measure merit matter once we include covariates that are not directly related to merit. The table presents two regressions for Firm-Level Backdating Probability and the logarithm of Total Abnormal Reversal, (138) with and without controls for SEC investigations, and also includes controls for logarithm of firm size. The sample for these regressions is the subset of firms publicly implicated in the options backdating scandal. (139) The regressions therefore implicitly control for public identification as a backdater.
Firm-Level Backdating Probability and Total Abnormal Reversal are strongly predictive of a derivative lawsuit in Figures 1 and 3. This suggests that plaintiffs' attorneys chose, from among the publicly implicated cases, a set of firms to target for suit that were more likely to have backdated, and to have done so more egregiously than the typical firm implicated in the scandal.
Models 2 and 4 include a control for an SEC investigation. In Models 2 and 4, the backdating activity measures continue to be significant, albeit weaker, when the control for SEC investigations is included, suggesting that plaintiffs' attorneys add additional selectivity on the merits. That is, plaintiffs' attorneys did not simply piggyback on SEC investigations, which would have nevertheless been a form of merit-selectivity, but independently identified cases of relatively high backdating activity.
The control for firm size is highly significant in the Firm-Level Backdating Probability models. The control is less significant, though still positive, in the Total Abnormal Reversal regressions. The reduced significance for firm size in Models 3 and 4 is likely due to a strong correlation between firm size and the logarithm of Total Abnormal Reversal, reflecting the higher value of option grants at larger firms with higher-paid executives. Nevertheless, there appears to be a correlation between firm size and the likelihood of being targeted in three of the four models, even controlling for backdating activity, potentially suggesting a deep-pockets effect.
Table 4.B presents parallel results for securities class action cases. In this set of regressions, we replace the Total Abnormal Reversal measure with the variable Backdating Revelation Abnormal Return. Backdating Revelation Abnormal Return is the cumulative abnormal return...