We know that people respond to incentives, and that behavior will adjust in response to relative changes in price. But I think it’s commonly assumed that the only relevant price change attributable to disclosure regulations is the nominal change in direct costs of compliance. Sure, we all understand that if shareholder or regulatory pressure is brought to bear on corporate actors as a consequence of disclosure, that pressure can change behavior. But for some reason we’re unduly optimistic about this change; we just assume it will be for the better (you know, because “sunlight is a good disinfectant.” Brandeis said so).
But I want to point to another, I think overlooked, aspect of this dynamic: Behavioral changes taken by corporate actors to minimize disclosure-induced, unfavorable (by which I mean privately-costly) consequences. These adjustments can be costly. They may whittle away whatever benefits we think might arise from the disclosure obligation in the first place, and, ironically, they may actually staunch the flow of information.
What I’m interested in is how disclosure regulations might alter substantive behavior, particularly in unintended fashion. It was probably once the case that disclosure obligations were intended merely to inform and not to regulate. But I doubt that is any longer the case, although rhetoric (which is to say, compliance with statutory authority) is sticky. The idea that targeted disclosure regulation has substantive effect is not new—my dad mentioned it back in 1974 (and probably before); Steve Bainbridge more recently in The Creeping Federalization of Corporate Law. I’m sure there are many other examples. But we should also consider how disclosure has unintended, substantive effect as a function of actors’ efforts to avoid disclosure.
Proponents of disclosure regulation recognize that such regulations may “work� if the private benefit of conduct (say, taking an exorbitant pay package) decreases by dint of disclosure—either directly because of pressure to reduce compensation, or indirectly if the executive feels shame (or some other psychic cost) once his compensation is made more public. What proponents seem to forget is that rational actors will spend up to the expected amount of that decrease in order to avoid it.
So here’s the basic abstract idea, call it the hydraulic theory of disclosure regulation: If required disclosure of activity A increases the cost of activity A relative to that of activity B, actors will shift some of their conduct into activity B. Pretty straightforward, right? It’s a basic substitution effect. Behavior A plus disclosure is more costly to some actors than behavior B without disclosure, so there is a shift toward behavior B. Will behavior B always be preferable to evil behavior A? I don’t think we can know that ex ante, and sometimes the answer must be “no.”
And regardless of the relative merits of behaviors A and B in the abstract, if actors shift from regulated behavior A to unregulated behavior B, we have not increased information flow. In fact, quite to the contrary, we have induced a reduction in available information. Hardly a winning result for regulation aimed at disclosure of information. I’m pretty sure this sort of behavioral response to disclosure regulations is rarely considered, perhaps because disclosure is presumed not to be costly. But that’s patently false—not because of nominal compliance costs, but simply because others’ knowledge of one’s behavior can make that behavior less attractive.
So there are two costly aspects to this. First, as disclosure rules impose costs on the underlying behavior subject to disclosure, where behavior can be altered at lower cost than the cost of disclosure, disclosure rules will induce behavioral changes rather than increased information flow. This can be costly, to say nothing of contrary to regulatory intent (but maybe we have to count that in the benefit column . . . ). Second, as disclosure rules impose costs on the underlying behavior subject to disclosure, where behavior can be altered in a manner that lessens the cost of disclosure, disclosure rules will induce behavioral changes. These changes may not be desirable.
Here’s an example: SOX §406 requires companies to disclose waivers of their codes of ethics in an effort to bring the magical disinfectant of sunlight (and deterrence) to bear on particularly worrisome conduct. But there is an implicit assumption animating the presumption of deterrence that conduct will be relatively static—that codes of ethics won’t change in response. In fact, the more likely outcome is that codes of ethics will be (and have been) re-written in order to minimize the need for waivers, in the event actually stemming rather than improving the flow of information. In other words, disclosed waivers are (privately) costly, and it may be less (privately) costly to amend codes of ethics than to seek and publicize waivers. Underlying behavior may not change one whit (or it may get worse), but either way less of it will be disclosed. (On this aspect of SOX, see my colleague John Kroger’s article, Enron, Fraud and Securities Reform: An Enron Prosecutor’s Perspective.)
Or what about this? An executive is receiving a nice, pay-for-non-performance compensation package. His plan is complex enough that under current disclosure rules no one can really tell what it’s worth, nor precisely how much pay comes from which source. Enter the new disclosure proposal. Let’s assume it does what it’s intended to do (a dubious proposition, I know). Assume some analysts are now paying attention, the directors are getting squeamish, and a couple of big shareholders are squawking. Maintaining the “exorbitant” package might become much harder (and I’d wager, rhetoric to the contrary, for many this and not mere disclosure is the intent of the regulation). There are two options: Reduce/re-organize pay (which is to say, in the Bebchuk & Fried managerial power model, give back some of the ill-gotten rents to the shareholders) or find an alternative mode of compensation which is either a) not subject to disclosure or b) less susceptible of negative interpretation, even if disclosed. What might this look like? In the obvious (but maybe unrealistic) case, shirking comes to mind. Can’t maintain the outright $10 million package for the work you did before? How about the $10 million performance-based (riskier) package for less work? Or how about $8 million, but we go ahead and build that new building you’ve been pushing for, the one designed by Frank Gehry? Or we could upgrade the fleet of jets like you’ve been asking. Any of these forms of compensation would be undisclosed. I imagine the options are limitless, and I have to believe that they will be opted for. Oh, and if the options aren’t immediately apparent to the executive and/or the board, then be sure to add in the additional cost of determining what they are.
Dealing realistically with the possibility that mandatory disclosure has costs is not the norm. One recent exception is Mike Guttentag’s work. (Find his articles on disclosure here). Although I don’t think Mike considers the sorts of costs I suggest here, he at least takes seriously the possibility that disclosure is not costless. We would surely do well to have more work of this sort.
By the way, Larry comes close to making this point today when he notes that perks and other illegitimate compensation won’t likely make it within the ambit of the new executive compensation disclosure rules. He need only have added, “and thus the new rules will exacerbate the problem.� Chicken soup, indeed.
RSS - Posts
Geoff. All excellent comments. I can tell you how I try and address your concerns, but I think you have nicely summarized a set of plausible challenges to what I say. With respect to disincentives caused by disclosure – my vote is for “produce and disclose� rules, such as those that require financial statements disclosures. I try in the article to specify exactly which kinds of additional information should be required to be disclosed by looking to disclosure requirements in private transactions. The thought is that this type of rule is better than a conditional “disclose if produced� rules, which would inevitably mess up information gathering incentives. The second point I would make is that it would be silly to simply require the disclosure of all competitively disadvantaging information, and I think this is where Fox errs. The model attempts to show that there are particular categories of information for which non-disclosures is especially socially costly, because, for example, of the impact of non-disclosure on agency costs. Finally, it is not obvious if the world becomes less competitive or more competitive if this information is required to be disclosed.
Does the SEC do what I recommend? Certainly not in a systematic, thoughtful, or careful way, but they may be stumbling in the right direction, if I am to be believed.
I didn’t mean to exclude your justification for mandatory disclosure from my brief (2 example) mention of justifications in my comment — there I thought of the credible commitment argument because it is an argument rooted in firms’ (theoretical) view of what’s best for themselves rather than someone else’s, and B&F because I have been thinking about executive compensation. The interfirm externalities argument you propound is certainly a plausible one.
Now, am I convinced by it? Of course not (but that’s just my own intransigence talking). But really, I’m not convinced because (unless I misunderstand your article) the claim that interfirm externality correction justifies disclosure does not take account of the ways in which required disclosure might alter firm behavior in socially costly ways. Not that I have a model to capture the effect, but I can easily imagine that this might occur. If you require firms to disclose competitively-disadvantageous (but, in theory, socially advantageous) information, how will that affect their behavior? Will it not affect behavior? Could the offset eat up all or most of the benefit? I guess my sense is that, although permitting firms to keep secret information that could be valuable to competitors is a “cost” of non-disclosure, it is the price we pay for spurring the animal spirits and maximizing incentives to innovate and compete. Requiring every firm to operate like they were all part of one big firm, not competing in capital, labor and product markets, would be disastrous. Not that this is precisely what you are advocating, but, again, unless I misunderstand, it must be a move in that direction (ok — so they don’t act just like one big firm because they do still compete in these markets, but if we magnify the cost of failure and minimize the gains appropriable from success (by both managers and by firms at large), how much competitive incentive do we lose? I just don’t see how that factors in to your model. But I’m no expert on formal models — maybe I’m missing something.
Geoff. You were very kind to cite my article: An Argument for Imposing Disclosure Requirements on Public Companies, but I take it you were not impressed by the conclusion. In my article, as the title suggests, I argued that there is a substantial market failure in public company disclosure practices resulting from the inability of firm’s making disclosures to capture the benefits these disclosures provide to the firm’s competitors, so called interfirm externalities. This is an argument different than the credible commitment argument offered by Edward Rock (among others) and the reporting consistency argument alluded to above by William Goodwin (and discussed in greater detail by John Coffee in 1984 and Douglas Diamond in 1985).
The possible significance of interfirm externalities is not new to me. Merritt Fox made much of this as the possible cause of a significant market failure in firm disclosures, but most of the arguments he offered were effectively rebutted by Roberta Romano in her papers arguing for issuer choice in securities regulation. I take up the Romano and Fox dialogue in my article. One of the shortcomings that Romano pointed out in Fox’s argument was the lack of an adequate financial model linking interfirm externalities and gains from securities regulation. I believe I have rectified this shortcoming by developing such a model, which is posted on my SSRN cite, work that was completed subsequent to the publication of the article you cite.
Is there no market failure that justifies mandatory securities regulation? I think this is still an open issue.
Isn’t one obvious benefit of making disclosure mandatory that it harmonizes the reporting of all public firms and therefore reduces search costs for investors trying to compare potential investments? If, as many fundamental investors do, you view excess executive compensation (that is, compensation that is well above peer-group pay without any corresponding outperformance) as a sign that a company’s corporate governance is weak, and that therefore it may be headed for trouble, you want to be able to access the information about compensation as quickly and efficiently as possible, and you want to be able to aggregate the data across a wide range of companies and classes. Mandatory disclosure makes that significantly easier.
Another benefit, obviously, is that it makes securities pricing more efficient, by releasing more information to the marketplace. Given the fundamental importance to society of the public markets in allocating capital, there is an abiding social interest in ensuring that that allocation is as efficient and as accurate as possible. The marginal cost of the regulations needs to be weighed not just against the marginal benefit to investors, but also against the social benefit of having more efficient securities pricing.
Mike: Thanks for the comment. The important question, as always, is the marginal one (and the one for the marginal firm). I guess it’s possible that mandatory rules get us to the optimal point, but it’s extremely unlikely, given the inherent paucity of information possessed by regulators and the wide disparity among firms. Plus, even if so, the real question is, what do we get by making disclosure mandatory? Ed Rock has a nice, but to me not-fully-persuasive answer that at least identifies a reason firms may opt for a mandatory regime. Bebchuk and Fried would argue that we need it to correct an agency problem, but now we’re back to guessing whether the marginal gain of increased disclosure is worth the marginal cost (and, as you point out, whether that’s what’s going on in the private agreements). One big difference between the private/public contexts is that when compensation is voluntarily disclosed there is probably less incentive to subvert the disclosure than when it is imposed from outside. I’m sure there’s a name for this cognitive effect. Still, I think looking at private party agreements would be a good source of data here, but ultimately not all that compelling on the mandatory question.
Your point about politics is right on — in fact I stole my “hydraulic theory” moniker from Pam Karlan and Sam Issacharoff writing about campaign finance reform.
I think your post raises interesting issues, and, as I recall, my article that you cite does not deal well with the kind of dynamic substitution effects you discuss here.
However, I do think that looking at disclosure requirements in private party transactions might suggest how significant these substitution effects are. Wouldn’t we expect that venture capitalists and other investors would try and put together an efficient package of disclosure requirements, and Paul Mahoney’s work suggests that requiring the disclosure of compensation information is a common feature of disclosure requirements, even in the absence of mandatory regulations? This need not mean that these types of disclosure rules substantially reduce agency costs, perhaps for the reasons you suggest. It does, however, suggest that there are net gains from imposing these but not other disclosure rules. Of course, this logic probably does not address the SOX code of ethics example you provide, which looks like regulation without evidence to support it.
I would add that the concerns you are raising here may also be supported by recent events outside the corporate management setting. Maybe we see the concerns you raise in recent events in the political and medical settings. In both settings, we have implemented efforts to try and limit the ability of third parties to directly purchase certain assets, political influence or doctor’s recommendations. In both cases, what happened? Instead of being purchased for cash, political support was purchased by giving legislators boondoggles, campaign funds, and other indirect forms of compensation. Similarly, I read that a recent article in the Journal of the American Medical Association suggested that the same kind of substitution effects you are concerned about lead to expensive indirect forms of compensation provided by drug companies to insure doctors’ support of their products.