In a prior post, I discussed Fordham Law Professor Richard Squire's April 2012 article entitled “How Collective Settlements Camouflage the Costs of Shareholder Litigation” (here). After my post appeared, Professor Squire communicated to me his concerns about my comments regarding his paper. Because his comments and concerns about my post were quite substantial, it seemed that the best approach would be simply for Professor Squire to publish his response in full in a separate blog post. I have set out Professor Squire’s response below. I am very grateful to Professor Squire for taking the time to present his views in full here. I encourage readers to review Professor Squire’s paper as well. As Professor Squire indicates at the conclusion of his guest post, he and I will be participating in a session at Fordham Law School on May 8, 2012 to discuss his paper. Information about the session can be found here . Here is Professor Squire’s response:
I am most grateful to Kevin both for devoting a full blog entry to pay my current article on D&O insurance, titled “How Collective Settlements Camouflage the Costs of Shareholder Litigation,” and for giving me here the chance as a guest blogger to respond to his comments. This has been a valuable opportunity for me to learn more about the D&O field from Kevin, a recognized expert whose knowledge of the market for D&O insurance greatly exceeds my own.
In his generously extensive comments about my article, Kevin identifies several places in which he thinks I describe the dynamics of D&O insurance settlements accurately. At the same time, however, he expresses numerous concerns with my proposed method for reforming such settlements. In this blog entry, I wish to focus on his concerns, and in particular to say why I think several of them are not warranted or are addressed by other aspects of my article that Kevin’s comments do not mention.
To provide context, it will be useful to begin with a summary of my article, which has five interrelated points as follows:
1. In a shareholder lawsuit against a defendant with one or more D&O insurance policies, the current practice is to require any settlement of the suit to be a single, collective resolution that binds all defense-side parties, meaning the defendant and each of its insurers.
2. The current system of collectivized settlements encourages strategic liability-shifting among defense-side parties. This strategic behavior imposes a variety of costs on shareholders, only some of which have been previously recognized by judges and academic commentators.
3. The collective action problem we see in shareholder lawsuit settlements is not inevitable. Rather, it is “contrived” in the sense that it is a byproduct of how D&O insurance contracts are written and enforced.
4. At least in theory, settlements could be de-collectivized in a manner that greatly reduced or eliminated the costs to shareholders from strategic liability-shifting. My article describes such an approach, which I calls “segmented” settlements.
5. Even if de-collectivizing the settlement process would benefit shareholders, corporate managers would oppose it because the change would reduce the managers’ ability to use D&O insurance to shift settlement liability to insurers and thus to insulate corporate earnings reports from the impact of the managers’ conduct that gives rise to shareholder litigation.
Kevin’s comments suggest that he agrees with me on points 1, 2 and 5. Most of his criticisms of my paper are aimed at points 3 and 4—my arguments that settlements could in theory be de-collectivized and that such a change would benefit shareholders. His comments also stress how “policyholders” would resist my proposal, though in so doing he does not acknowledge the degree to which he and I are in agreement on this point as long as by “policyholders” we mean the corporate managers who actually make the decisions to buy D&O insurance on behalf of themselves and their corporations.
Since most of Kevin’s concerns are focused on the desirability of my alternative system of segmented settlements, I will illustrate my proposal here with a very simple example. Imagine a corporate manager who is covered by a single D&O policy with a limit of $1 million. This means that if the manager is sued in a shareholder lawsuit, the insurer is responsible for the first $1 million in liability, and the manager is responsible for the excess, if any. Under the current system of collective settlements, the insurer could not settle with the plaintiff unless the plaintiff also agreed to waive his rights to collect from the manager, including his right to collect damages in excess of the $1 million policy limit. Conversely, the manager and plaintiff could enter into a settlement for, say, $1.2 million, and then use the “duty to contribute” (a duty not previously identified in the academic literature, but whose existence and importance Kevin confirms) to force the insurer to “tender” (pay) its $1 million policy amount in support of the settlement. In this way, any settlement must be “collective”—i..e, jointly binding on both the insured (the manager) and the insurer.
One of the main problems with this system of collective settlements is that it can lead to plaintiff overcompensation. Since the manager and plaintiff can enter into a settlement whose costs are borne mostly by a third party—i.e., the insurer—they can jointly gain at the insurer’s expense by entering into a settlement that exceeds the expected damages at trial. In my article I call this the “cramdown” dynamic. The manager’s incentive to engage in this settlement is that it avoids the risk of a trial at which the total damages in case of a verdict for the plaintiff may exceed the $1M policy limit by a large amount, leaving the manager with greater personal liability than she incurs by settling.
Under a system of “segmented” settlements, by contrast, the manager and insurer could settle separately out of the case, and by so doing could neither shift liability onto the other nor bind the other in a settlement without the other’s consent. So, for example, the insurer could settle separately with the plaintiff, in which case the insurer would pay the plaintiff a settlement amount and the plaintiff in exchange would waive his right to collect the first $1 million in damages awarded at trial (if any). If the manager did not also settle, then a trial would occur, but only damages awarded in excess of $1 million would be collectible. Conversely, the manager could settle separately from the insurer, whereby the plaintiff would waive his right to collect any damages awarded that exceed $1 million. Under this system, “cramdown” settlements could not occur, and collective-action costs would be reduced for the reason that defense-side parties would no be able to shift liability onto each other.
Many corporate defendants have not just one D&O policy, but rather a primary policy plus one or more excess policies, forming an insurance “tower.” Adding these additional insurance layers increases the opportunity for strategic conduct but does not otherwise alter the basic conflict of interests created by a collectivized settlement approach nor the benefits of the alternative approach I describe.
With that overview for context, I’ll now turn to Kevin’s specific concerns and criticisms.
Holdouts: Kevin argues that strategic holding out by insurers would still occur under if settlements were de-collectivized. To illustrate, he gives an example of a case in which all defense-side parties have settled except for one mid-level insurer. I actually anticipate something close to this hypothetical on pages 29 and 30 of my article. Contrary to Kevin’s argument, under the system I describe a mid-level insurer in that position would not have an incentive to hold out, as by doing so the insurer could not externalize liability onto the policyholder or other insurers.
To make the example concrete, let's assume a policyholder with a tower of five D&O policies worth $1M each. We’ll assume further that all defense-side parties, including the policyholder, have settled with the plaintiff except for the excess insurer occupying the $2M to $3M slice of the tower. This means that if a trial occurs, the plaintiff could collect only those awarded damages (if any) that fall between $2M and $3M, and these would be collectible solely from the holdout insurer. Thus, regardless of whether the holdout insurer ultimately settles or goes to trial, no damages liability can be shifted to the policyholder.
Kevin expresses in connection with this example a concern that defense costs (i.e., attorneys’ fees and similar expenses) might reduce this "sole remaining layer" of coverage, leaving the policyholder exposed (though, to be sure, only to defense costs plus damages in the $2M to $3M range—i.e., the unsettled remaining slice). I anticipate this concern on pages 29-30, but I offer a simple solution: "We can predict that [if settlements were de-collectivized] liability policies would be written so that the non-settling insurers in such cases [in which the policyholder had separately settled] bore the defense-side trial expenses without regard to policy limits, as otherwise those insurers could externalize onto other defense-side parties some of the costs of their refusal to settle."
Kevin earlier in his post had expressed “trepidation” about the “world of academic analysis,” which seems to him “unbound by constraints that operate in the world to which I am accustomed.” But there is nothing otherworldly about my simple solution to the holdout problem with respect to defense costs. Under most other types of liability insurance (such as auto and homeowners insurance), defense costs do not count toward the policy limit. It is D&O insurance that is unusual in its use of "burning candle" policies. Thus, by suggesting that defense costs would no longer count toward policy limits in situations in which the policyholder has settled out of the case, I am incorporating a solution to the cost-shifting hazard that will already be familiar to people with real-world knowledge of liability insurance.
After presenting his holdout hypothetical, Kevin goes on to write that my proposal would "substitute a different cramdown dynamic for the existing one" and would "put the insurers in the position where they were jockeying to force loss costs elsewhere, including in particular onto their insured." It seems to me that his concerns here stem from this same misunderstanding about how holdouts would be handled under my proposal. For this reason, I don't think these concerns are warranted. Under my proposed system, defense-side parties would have significantly less ability to engage in strategic cost-shifting than they do now.
Distributional Impact: Kevin observes that the distributional impact of separate settlements would be to reduce liability for primary insurers and increase it for excess insurers. This observation is accurate, as my article acknowledges at several points. But it is not grounds for concern, as the excess insurers would adjust by charging higher premiums ex ante, and so they would not on net be worse off. However, because the current system’s cramdown dynamic would be eliminated, overall settlements would be lower, as there would be a reduction in plaintiff overcompensation (a phenomenon which Kevin acknowledges occurs under the current system). So overall insurance costs will be lower, a benefit to policyholders.
Expected Trial Liability: Kevin devotes several paragraphs to criticizing my reliance on the concept of "expected trial liability" (meaning expected damages if the case goes to trial). In particular, he criticizes my article for arguing that this figure is "objective" rather than "subjective" and can be reduced to a "single knowable measure." Here I think Kevin is attacking a straw man. Nowhere does my article claim that expected trial damages are knowable with certainty ex ante or that parties will form identical estimates of them. To the contrary, I acknowledge that estimates will differ, and on pages 35 and 36 I model what settlement negotiations look like when they do.
Similarly, Kevin criticizes me for not taking into account "myriad factors" that affect negotiations, including "whether some insurers believe they have unique coverage defenses." But on pages 37-40 I do model the impact of coverage exclusions on settlement negotiations.
To be sure, I don't model the impact of another factor Kevin identities—namely, the pendency of related lawsuits. But the fact that a model does not include every conceivably relevant factor does not mean that we can derive no insight from its results. Indeed, the article’s models are what revealed to me how the cramdown effect under the current system can lead to plaintiff overcompensation, a result whose accuracy Kevin confirms. More generally, while Kevin calls into question the usefulness of abstract models in the study of complex insurance negotiations, he does not identify any specific results produced by the article’s models that he thinks are inaccurate or unrealistic.
Delay: Kevin worries that separate settlements would introduce delay. But delay results from holdout problems, and de-collectivizing the settlement process would greatly reduce the advantages to holding out. Indeed, under my system, as each defense-side party settles out of the case, the incentive both for the plaintiff and for the remaining defense-side parties to settles increases. This is because each settlement reduces the plaintiff's potential recovery at trial but not the trial expenses that both sides would have to incur if trial occurred. Returning to the example above involving the mid-level insurer holdout, that insurer's incentives to settle are maximized when it is the only defense-side party left in the case, since at that point it will bear all of the defense-side trial costs. And the plaintiff's incentive to settle is maximized as well since the damages recoverable at trial have been pared down to a thin slice, but winning that slice would still require the plaintiff to incur the full costs of putting on his case. Since there is no advantage to anyone under my proposed system of delaying resolution of a lawsuit, we can expect less rather than more delay than we see now.
Achievability: Kevin criticizes me for writing that "segmented settlements could easily be achieved contractually," a statement he says "lacks a connection to the insurance marketplace" because insurance buyers would resist such a change. But all I meant was that nothing prevents segmented settlements as a matter of contract law. I acknowledge full well that D&O insurance buyers—i.e., corporate managers—benefit from the current system of collectivized settlements. See, for example, my abstract: "Yet corporate managers probably prefer the status quo." That's why on pages 42 and 43 I argue that, if reform were to occur, it probably would have to be initiated by courts.
Reinsurance: In my article I make the uncontroversial observation that excess insurance and reinsurance are substitutes: both help insurers diversify their risk exposure. A simple example will illustrate. Imagine that Company A buys a $10 million primary policy and a $10M excess policy. Meanwhile, Company B buys a $20M liability policy, and its insurer then purchases coverage for the top half of this policy from a reinsurer. In both cases we have a division of risk between two insurers.
I think it interesting that in the D&O market we see Company As rather than Company Bs—that is, we see towers rather than reinsurance. I raised the question in my article whether this might reflect a preference among insurance buyers for a system that encourages covered settlements through the cramdown dynamic. Kevin criticizes me on this point, arguing that there are "a finite number of reinsurers and they require a spread of risk every bit as much as the insurers do." But the fact that currently there is a relatively small number of reinsurers is not a criticism of my hypothesis; rather, it is a restatement of the question my hypothesis seeks to answer—i.e., the question why this particular insurance market has developed to rely on towers rather than reinsurance. And his claim that reinsurers also "require a risk spread" is misleading since reinsurance is, by definition, risk-spreading, as my example of Companies A and B illustrate.
Finally, Kevin claims that the presence of towers is explained solely by the preferences of insurers, but this claim is in tension with his earlier claim that D&O insurance is a buyers' market, and it fails to explain why the carriers should prefer the tower model over the reinsurance model.
As I said at the beginning of my comments, I am most grateful to Kevin for not only highlighting my article on his blog but also giving me the chance to respond as a guest as I’ve done here. As Kevin mentioned, he and I will be co-panelists at a conference at Fordham Law School (where I’m privileged to teach) next month, where I’ll have the opportunity to be able to discuss these matters with him further. I hope Kevin’s readers have found my exchange with him interesting. If any reader has any comments or questions on the subject matter discussed here, I’d welcome hearing from you at firstname.lastname@example.org.