I'm delighted to have all the commentary on my paper. Following custom, I'll respond to the posts individually in the comments here.

Dear Prof. Rose,

Thank you very much for your comments. Your point about the completeness of my data is one that I worry about. As you suggest, there is a particular set of SLC decisions that don't make it into the disclosure documents: If an SEC filer has derivative litigation against it that it deems immaterial, it won’t disclose the litigation, and any SLC formed to control the litigation won’t be disclosed either. That type of litigation might be especially prone to an SLC decision to dismiss, and searching the SEC filings would miss them. As you indicate, to try to get a handle on how big this problem was I searched case reports for SLC decisions that could have been disclosed in EDGAR and found 31 SLC decisions mentioned in cases. I had identified 26 of them (84%) through EDGAR. That gave me confidence in what I had, but as you say what I’m missing might be uniformly decisions to dismiss. There’s no way to know, though, because settlements aren’t going to appear anywhere if they’re not disclosed in the SEC filings.

The content of the settlements is the other big issue, and that’s one thing I’m working on this summer. I have settlement data from the filings – filings include it, but it’s patchy. (And I cannot tell, based on the settlement values alone, whether it was a good-faith settlement.) I may use this data as a stand in for a while, but I am ultimately trying to collect data from court filings. I had high hopes for having it done by now, but (like everything on this project) it’s taking a lot longer than I planned.

Your idea to get reactions from practitioners is a promising one. A Delaware litigator is looking at it now, but expanding the number of practitioners who give the findings a gut check would definitely be valuable.

Thanks again,

Minor

Dear Prof. Henderson,

Thank you very much for your comments. Pulling apart the outcomes by governing law – Delaware, New York, or MBCA – is another of the projects I’m working on now. The limiting reagent here is good settlement data -- I can say something about SLC decisions by state law, but I’d like to back it up with settlement figures, so I’m holding off. Once I’ve got the settlement data, I should have something to say about the relationship between state law and SLC decisions and outcomes.

I’ve been editing the paper a lot recently and thinking about what does and does not belong in the paper. I used to have too much in the text, and now the pendulum may have swung too far in the other direction. I’ve been using the footnotes as a waystation along the road to total elimination of material, and what’s there now will probably either be cut or returned to the text. I agree that the paper as it is cannot stand, half text and half talking footnotes.

On my Bluebooking, please accept my apologies. I certainly don’t consider myself above good Bluebooking! I’ve just burned myself in editing before (putting in complete x-refs or Id.s initially and then moving the referenced material, creating general chaos) so I leave things blank as a reminder to myself to fix things once things are in place. But I do not want to burden reviewers any more than necessary, so I’ll be more mindful of proper citations.

Thanks again for your comments.

Minor

Dear Prof. Fairfax,

Thank you very much for your comments. The companies in the data with no publicly traded equity are a puzzle. There are only 9 of them, and I’m not sure that the population of private companies that file with the SEC is representative of private companies generally, so I don’t think there’s too much to be learned from those nine, but their presence in my data is by itself pretty weird. SLCs for those companies don’t make different decisions than SLCs generally, but I will dig a little deeper on those to explain exactly what’s going on there.

You are also right that there is tension between the rate at which SLCs decide to dismiss entirely (60%) and the rate at which claims are ultimately dismissed (20%). That may suggest that there is some strategic behavior going on there or maybe some bias, and this is one of the issues I want to grapple with as I push the project further.

I really like your point about the divergence of views on SLCs and independent directors between courts and others. Courts still endorse the Auerbach approach and the use of disinterested directors in other circumstances, so that suggests that some judges are of the view that SLCs are not all that problematic. I’m not sure how to account for their disagreement with most other people, but I will definitely dig into that point (probably by starting with the opinions of courts that have adopted NY/Auerbach – why do they think it’s okay?).

I hesitate to offer my own views yet on whether SLCs are a good idea because I want to wait until I’ve done everything I want to do with this project. In other words, I’m not sure myself. I can see the serious risks on both sides – of allowing directors to police themselves and of precluding SLC review wholesale. Once I’ve got the settlement data under control (sorry for using that as a shield so often), I’ll have better answers on this. Similarly, I’m not sure yet about whether altering the scrutiny for independent directors is a good idea. My preliminary hunch: is that Zapata and the MBCA may work, Auerbach may not.

Thanks again for your comments.

Minor

Minor, this paper makes a very good start on an important inquiry. I think I still lean toward conventional wisdom at least on the point that bracketing settlements, SLCs generally dismiss claims. As others have mentioned, I'm very interested in these settlements. Are they nonmonetary? Symbolic? If so, then they may be dismissals in settlement clothing. Nevertheless, I think your paper adds to the conversation in a very substantive way.

Dear Prof. Black,

Thank you very much for your comments. Demand is a huge obstacle to even getting the SLC apparatus up and running. Given the increasing independence of boards, a successful claim of futility or wrongful refusal may be quite tricky. The only circumstance that might alter things is when the independent directors are among the potential wrongdoers – thus, although they are independent for NYSE purposes, they are not disinterested for demand purposes. But that's a very tough row to hoe. I agree with you demand not the SLC is the big hurdle for a shareholder-plaintiff.

My paper might not prompt a wholesale rethinking of the use of independent directors in non-SLC contexts, but I do hope that it generates a reconsideration of the SLC context (and attempts to generalize from it). That said, I agree with you entirely about the possibility of collusive settlements. All of the settlements I’ve found might be just a deal between directors in the hotseat and plaintiff attorneys with bills to pay—where shareholders lose. On this view, though, I think litigation subject to SLCs just ends up looking like regular derivative litigation. The incentives created by derivative litigation might be the culprit, not special litigation committees. That’s definitely not encouraging, but it’s a different story about SLCs than the one in the literature now.

I hope that settlement data will resolve some of these issues. And I hope that it helps (along with some event studies around SLC decisions and case developments like settlement) differentiate bad settlements from good ones. That’s my best approach to clearing things up (and of course I welcome suggestions for improvement from anyone).

Thank you again for your comments.

Minor

My data on settlements is not ready for prime time, but I will post something about what I've got after I respond to Prof. Lawless's comments.

Dear Prof. Lawless,

Tough criticism indeed, and I'm grateful for it! As you point out, everything I have to say flows from how I code the data, so my coding decisions deserve attention. They are hugely important.

One important coding decision was to count as the same thing (1) explicit SLC decisions to settle and (2) instances where litigation settled prior to SLC decision without any indication of the SLC’s role. It was a mistake not to code for that. As I’ve mentioned in other responses, I’m trying to get better data from case dockets, and that will help for purposes of resolving what role the SLC played in settlements and for determining the content of the settlements. If I cannot resolve it that way, though, I will go back and recode for this, because you’re right that without knowing for sure I should not treat them as the same thing.

If though I find that the SLC is not involved at all in a settlement but it has not concluded it investigation, I think that fact still suggests something about SLC behavior (though you are right that I should not treat it as an SLC decision). If the conventional wisdom on SLCs is correct, a defendant has no reason to settle a claim prior to the SLCs decision (costs of defense are covered, D&O insurance available for settlement later if SLC loses in court, etc.). You are right that such settlements do not tell us what SLCs do (in other words, they should not be treated as SLC decisions), but they does suggest something about what kind of treatment defendants expect from SLCs.

The other big coding decisions are how to classify decisions as pursue, settle, or dismiss. I adopted the any-settlement-code-as-settle convention (and same for pursue) because of the unequivocal nature of the conventional wisdom on SLCs. I can see how that is rigging the results, but my view is that it's a fair test of the conventional wisdom. Plus, if the settlements are fig-leaves, I hope that the data on settlement (and the stock price reaction too) will bear that out. That I think will be the real test of the settlements, not the number of claims pursued against the number of defendants.

Thank you for your comments and pushing me on my assumptions.

Minor

Here’s a picture of the settlement data that I collected from filings.

First, I want to thank Profs. Black, Fairfax, Henderson, Lawless, and Rose again for their feedback, which has been extremely helpful and will improve my paper. Thank you also to Christine and the Conglomerate for arranging the workshop and lining up the commentators – participating in a forum like this is extremely rewarding to a new person like me.

The data I discuss below is not what I will rely on ultimately (I hope), and I have not done everything with it that I want to do. But my hunch is that it points in the direction of where the data will ultimately take me.

The filings disclosed all or some of the settlement content in 52 of the 64 settlements. I’ll first describe them in the aggregate and then separate them by SLC decision.

I. Settlements in the aggregate

Looking at the 52 settlements in the aggregate, things look a lot like what Prof. Romano found in her earlier study of shareholder suits. In 2005 dollars (throughout), the median monetary recovery here to the company is approximately $96,900. The settlement recovery amounts are heavily skewed by a small number of large settlements; the mean is $10.2 million (slightly higher than Prof. Romano’s mean of $9.9 million). The largest financial recovery was $117 million. Taking the settlement recoveries as a percentage of firm market value shows again that the recoveries are small but skewed. The median monetary recovery to the company was 0.02% of firm market value, and the mean recovery was 1.8% of firm value. Twenty-six (50%) of the settlements involved no financial recovery for the company. Here again, the data resemble the findings of Professor Romano (where forty-five percent of settlements (37 of 83) involved no financial recovery). Thirty-eight of the settlements—73%—involved a payment to the plaintiff attorneys. As in Professor Romano’s study, awards are paid to attorneys more frequently than to shareholders. Of the twenty-six settlements with no financial recovery to the company, fourteen of them involved some sort of structural relief. Some of these forms of relief are deeply important for shareholders; in one case, for example, a company’s dual-class stock structure was dismantled and the controlling owners relinquished their control. Some of these forms of relief were largely meaningless; for example, a railroad’s commitment to form a committee to help promote its “commitment to rail safety.” In nine instances, attorney fees were the only relief disclosed. This rate of attorney-fees-only settlements (15%) is nearly twice the rate found in Professor Romano’s study.

II. Settlements by category of SLC decision

The aggregate statistics may obscure what is actually going on in the cases. The pattern of recovery when broken down by SLC decision and further by court treatment shows substantial differences based on the procedural posture of the case. The claims SLCs pursue are far more likely than other avenues to result in a meaningful payment to the company upon resolution of the claims. Decisions to settle also generate non-zero recoveries in the median instance. When the SLCs moved to dismiss, however, the median recovery was zero no matter what the court does. (The mean recovery when a motion to dismiss was denied was high, but this is a result of skewed data, not regularly meaningful recoveries.) Here are the details:

The settlements in cases where the SLC determined to pursue are large. The median recovery was $4.3 million. The largest recovery was $47.5 million; the smallest was $0. The mean financial recovery is $13.9 million. (The settlement involving the dismantling of the dual-class structure was an SLC decision to pursue.) One of these claims is still outstanding.

When an SLC decided to settle, the median recover was $2.2 million. The mean recovery was $15 million, and the largest recovery was $117 million. Seven of the nineteen settlements involved no financial recovery. Twelve settlements indicated some kind of structural recovery.

The median recovery when an SLC decided to dismiss was $0, no matter what the court ruled on the m/t/d. There are three categories here: (1) after the decision to dismiss but before a court’s ruling on the m/t/d; (2) after a court denies the motion in whole or in part; or (3) after a court grants the motion entirely. In category 1, the median financial recovery was $0, and the mean was $3.71 million. The largest settlement was $25.4 million. In category 2, the median recovery was $0, and the mean was $11.5 million. (The results are skewed by two very large settlements, one for $85 million and the other for $54 million.) The smallest settlement recovery is zero, and six of the ten settlements involve no financial recovery. In category 3, none of these settlements involved financial recovery, structural relief, or attorneys fees.

All of this evidence is consistent with the explanation that SLCs generally pursue the strong claims, settle close ones, and dismiss the bad ones, but a big problem with this explanation is that I cannot tell what drives the association between SLC decision and monetary recovery. Maybe the SLCs decide to pursue the strongest and most valuable claims; but maybe the causation runs the other way: the strongest and most valuable claims are whichever ones the SLC decides to pursue. This turns on the relationship between settlement value and the underlying strength of the claims. If settlement values are related to the underlying merit of the claims, this suggests that the claims here—where the median recovery is about $96,000 and 0.02% of a firm’s market value—are generally of little merit and SLCs are not bad at screening them. If settlement values are unrelated to the underlying merit but are just a reflection of the claim’s procedural posture at settlement, then these data show that differences in the procedural posture generates large differences in the settlement value.

As I’ve said, I’m working on ways of improving the underlying data. I’m also trying to think of clever ways to try choose among the explanations (I have high hopes for some event studies), and I hope that some of the whopper outliers can be instructive as well. I would welcome any suggestions!