Academic

I just read ?Say on Pay?s Bundling Problems? by Andrew Lund (Pace University Law School). I didn?t know of Prof. Lund or his work before this, so in addition to learning a lot from his manuscript, I also am alerted to someone to whom I should pay attention. The article is comprehensive without being exhausting or repetitive, and it makes a nice point. I don?t believe ?say on pay,? either ex post (as is now the law) or ex ante (as Prof. Lund proposes), is sensible or needed, but that is a topic for another time. For now, I will try to (but fail here and there) limit my comments to Lund?s specific proposal about the timing of the vote.

Lund wants to move the shareholder vote on pay from after the year in question to before it, so as to ?decouple? the vote from the signal it will send to a particular executive. The fear he rightfully identifies in my view is that good performance will reduce the likelihood of a negative vote on pay, even when pay practices are ?bad.? (?Bad? here defined by the latest academic fad or guess at what those betting their own money should be doing with it.) Executives won?t like a negative vote when they do well, and such a vote may do more harm than good. This is the moral from the old yarn about the goose that laid the golden egg. Good CEOs are like golden-egg-laying geese, and it may be best to avoid making them go away or stop what they are doing, even if one thinks they might do it for less.

While this seems sensible enough, I have some quick reactions based on just a few minutes of thinking about this. My biggest issue has to do with the implications flowing from a key assumption that drives the argument. Lund acknowledges that an ex ante vote is likely to be superior only where the shareholders voting are ?effectively shield[ed] from the identity of the future CEO,? on whose pay they are voting. This raises several issues.

First, votes about pay for future executives will make the bargaining process unwieldy and unnecessarily rigid, even if only advisory. The shareholders? guess about what a future CEO might demand as part of their compensation contract is likely just a shot in the dark. And, insofar as the board is constrained on the margin by the vote?otherwise, what is the point??it will be a guess with potentially bad consequences. The process of hiring a CEO is highly idiosyncratic to particular firms, at particular times, and for particular individuals. These negotiations happen quickly, are quite fragile, and often turn on subtle and very hard to predict issues. Once a firm identifies its top candidate, it needs maximum flexibility in the amount and mechanisms of pay, since it does not know what the person will demand. Although one might think it makes sense to take some things that (some) academics don?t like off the table, since they seem inefficient, this may be a value-destroying move in a multi-issue negotiation. Maybe to attract the top candidate, the firm must give her the right to reload options, the right to backdate options, the right to guaranteed bonuses, or the like. While any of these might be objected to in an isolated case, the firm might rationally give one or more of them knowing it can get concessions on other issues, such as the amount of pay, the term of the contract, severance payments, or the like. The same issue is also true for amounts of compensation that may be offered. In short, constraining the board from offering X may be suboptimal, since X may be needed to get the person the board wants, and X may be efficient within the context of the negotiation. I can?t imagine telling the GE board they can?t hire Jack Welch way back when because shareholders expressed condemnation of a particular type of compensation that was essential to getting his services. It is one thing to give shareholders a non-binding way to express frustration with the firm and the CEO, which is what ex-post say on pay gives you, and quite another to ask shareholders to prescribe what might be needed in the future to attract someone they don?t know under facts they couldn?t possibly predict.

Second, the more general point is that the ex ante vote will either be too general, as in the case of option grants voted on by shareholders, or unnecessarily restrictive. This then points to a potential weakness in Lund?s analysis. He does not offer specifics about what an ex ante vote would cover. If it is just option amounts available on the shelf, that is obviously covered, since shareholders get to vote about dilution of their shares. If it is specific pay practices, like the type of options that will be given, this runs into the problem identified above. Shareholder votes are costly and time intensive, and are very unlikely to be able to respond nimbly when negotiations turn on provisions and issues no one expects. It may be argued that boards in that case could simply disregard even an overwhelming shareholder vote on a particular pay practice, but if this is the case, then, again, what is the point? If the board simply wants to know what general pay practices Profs. Bebchuk and Fried don?t like or RiskMetrics doesn?t like, it can read their books and reports. A generic ex ante vote on pay practices will give the board nothing more than this.

Third, it seems evident, as Prof. Lund argues, that ex ante votes are likely to alter the bargain struck, while ex post votes are likely to judge the bargain struck. He sees this as a bug of the current law, but it may be a feature. An ex ante vote may simply redirect compensation to other places or create opportunities for boards and executives to arbitrage what the shareholders have suggested. There will be ambiguities in what is prescribed and opportunities for boards to deliver pay in  ways not condemned that may be worse than those that are. If boards and CEOs are as bad and devilish as their critics suggest, having shareholders vote against X doesn?t mean pay packages will be more optimal; it just means we won?t necessarily see X. We may see X? instead. Unless the vote can be comprehensive, expect evasion and worse. In contrast, an ex post vote is a judgment on the package given, a judgment on the board, a judgment on firm performance (short of selling, which may be limited for large holders), and a judgment on the executive. This can be valuable, even apart from the details about pay, and it may, if nothing more, say that the pay is too high. This is harder to do ex ante: ?high? relative to what?

Fourth, even soft constraints on pay may be information about what or how a board is willing to pay in a particular case. Lund describes these votes as setting a floor, and I think that is right. But if the floor is meaningful, it may be set too high. If it is set low enough to give the board the necessary wiggle room, then it will be set too low for Prof. Lund. Such a floor may unnecessarily bind firms in bad times, and it may anchor them too low in good times. Outside CEO candidates might have many outside opportunities or few; they might have idiosyncratic preferences about pay or mainstream ones. The vote of shareholders in general is unlikely to be a good match to unknown future contingencies.

Fifth, if this is a sensible thing for shareholders to do, as a vote on options is, then we would see firms doing it already, voluntarily. There might be stories about why already-public companies may not amend their charter in this way, but nothing prevents a company doing an IPO from putting such an ex-ante-say-on-pay provision in place. If shareholders value such precommitment on pay on the margin, they should be expected to pay more for the shares of such a company. Not seeing these or any other contractual constraint on pay, other than the option grant requirement Lund describes, we should be skeptical of proposals to install them by government fiat. If we can?t identify a market failure, there is no reason to give people what they don?t want.

Finally, the application of this proposal to the top five executives makes little sense. Lund rightfully limits his proposal to new executives, but imagine a new CEO where the other four top executives remain the same. Is a shareholder vote for the new CEO going to bind him or her, while letting the other four keep their same contracts, perhaps influenced by say on pay votes many years earlier? As circumstances change, pay may need to change, for no reason other than to avoid unnecessary rivalries in the executive suite. But the ex ante proposal doesn?t work well in this (typical) case. Ex-post say on pay, for all its faults, doesn?t have this problem. It is shame that can apply to one and all at the same time.

There are many other issues one could raise. Here are just a few:

Lund cites just eight adverse say-on-pay votes in the UK in eight years for 1000 firms. That is a lot of process, cost, and distraction for ?discipline? 0.1 percent of the time. One wonders if the game is worth the candle.

Lund also seems to adopt the views of the critics of pay?like Bebchuk & Fried?without much addressing the opposing views of folks like Steve Kaplan and, well, me. I understand this is not crucial to his claims about whether the vote should be now or then, but nodding to the substantial literature showing this whole process is not economically (as opposed to doctrinally) sound, would make the paper stronger. In my Northwestern Law Review piece, for instance, I find that the Bebchuk & Fried agency cost story doesn?t hold much water, and in a forthcoming piece entitled ?Implicit Compensation,? I provide more empirical evidence of board bargaining.

There are some telling language choices too. The word ?constrain? appears about twenty times in the manuscript, but the words ?optimize? and ?efficient? appear hardly at all. Again, Lund isn?t debating the efficiency issue here, but any proposal discussing say on pay must acknowledge the debate more vigorously, in my view. For instance, does this difference between ex post and ex ante matter in a world in which, as I believe, current contracts are second-best efficient, meaning the best we can likely hope for?

A final point is worth making. In an era of increasing government review of pay, be it from Special Master Ken Feinberg or the SEC/DOJ clawing back, it is likely that boards will move away from rigorous algorithms, which lend themselves to ex post scrutiny or ex ante prescription, and lean more in the direction of discretion, like performance bonuses determined by the board. Such a move will help insulate the executive from this external oversight, which may be efficient for shareholders, who will, after all, have to pay a risk-averse CEO less in that case. If I?m right that this is the future, then ex ante say on pay would suffer against ex post say on pay, which can simply judge the board?s decision as it comes.

All in all, Prof. Lund gives us lots to think about. I?m not at all convinced any of this matters or amounts to anything other than window dressing and hand wringing, whether the vote comes before, during, or after the pay period in question. But this is a nice contribution to a sadly robust and timely debate.