Principal's direct welfare v. bias arbitrage

Bias arbitrage does not preclude an agent from taking actions that reduce the objective risk, so the agent may both attempt to objectively reduce a risk and to subjectively appear to reduce the risk. But the potential for bias arbitrage may affect an agent's choice of which risks to address when not all can be dealt with. To minimize objective risk, the agent should address the risk that is most-remediable (i.e., highest return per cost of addressing the risk). To minimize perceived risk, the agent should address the risk that is most over-estimated by the principals.

There is typically greater uncertainty as to the impact an action will have on an objective risk (i.e., uncertainty as to the success of an effort to reduce an objective risk) than as to whether a risk is over-estimated by the principal (i.e., uncertainty as to the success of an effort to reduce perceived risk).Footnote ⁴ Thus, when confronted with a choice between the most-remediable risk and the most over-estimated risk, agents are likely to address the latter.

Because the bias arbitrage incentive may direct the agent to act differently (i.e., address different risks) than the incentive to maximize the principal's direct welfare, the two incentives are separate and potentially conflicting.

Narrow self-interest v. bias arbitrage

An agent has an incentive to conduct bias arbitrage in order to receive credit for reducing the (over-estimated) risk. Why not, then, consider bias arbitrage as a subset of the agent's self-interested incentives?

The reason for distinguishing bias arbitrage from other self-interested incentives the agent faces (shirking, stealing, seizing an opportunity belonging to the principal, etc.) is that the latter reduce the principal's welfare, while the former typically increases it.

Individuals who over-estimate a risk are likely to act in suboptimal ways, such as:

1. (i) excessive avoidance of activities that expose them to the risk (e.g., excessive avoidance of corporate investments as a result of over-estimating the risk of corporate fraud);

2. (ii) excessive self-help (e.g., over-monitoring of the corporate executives by the shareholders); or

3. (iii) excessive pressure for institutional intervention (e.g., shareholders pressing government for unnecessary government regulation or pressing directors for cost-ineffective internal compliance programs).

Thus, mitigating the principal's misperception of a risk should increase social welfare.

Principals self-correct their misperceptions of risk over time (indeed, if they did not, agents would have no incentive to bias arbitrage). However, bias arbitrage speeds up the pace of this correction by giving principals a plausible reason to self-correct their perceptions and by focusing their attention on the misperceived risk.

Not every risk perception manipulation increases the subject's welfare. For example, if shareholders under-estimate the probability of fraud in the corporation (or the adequacy of controls on executive compensation), and the directors’ actions further increase their complacency, shareholders welfare will be harmed.Footnote ⁵ Such actions, however, are not bias arbitrage because they are not sustainable: under our assumption that individuals self-correct risk misperceptions over time, principals will gradually observe that their optimism was unfounded. Indeed, the more over-optimistic the principals are, the faster they would realize that their perception is wrong (since it would take fewer observations of negative events to cause them to question their current perception). Such realization would discredit the agent, and thus is not a sustainable risk manipulation strategy.

Conversely, bias arbitrage is sustainable when it attempts to reduce a perceived risk that is currently over-estimated, because the principal's self-correction (observing that the risk is lower) corroborates the agent's claims that the risk is lower, lending credibility to the agent (and to the agent's claim that it is her actions that resulted in the lower observed risk).

Thus, the agent's incentive to bias arbitrage differs from narrow self-interest in that it increases rather than decreases the principal's total welfare, and it is different from the agent's incentive to maximize the principal's direct welfare in that it leads the agent to address the risk most over-estimated by the principal rather than the risk most remediable by the agent.

Legal analysis of directors’ decisions under state corporate law

Below is an abstracted sketch of the manner in which Delaware corporate law (which is the most influential jurisdiction for American public companies) evaluates corporate directors’ decisions, such as their decision to approve executive compensation.

The analysis has three stages. First, the directors must have authority to make the decision – that is, the decision must comply with whatever explicit requirements the corporate statute demands.

This stage tends to present a low hurdle: directors are authorized to make nearly all corporate decisions (in most cases, regardless of shareholders’ wishes), and this authority is qualified in relatively few ways, procedural (e.g., rule requiring a quorum of directors for the meeting to be considered valid) or substantive (e.g., rule limiting when dividends may be distributed). Currently, directors face few constraints in approving executive compensation; indeed, they are even specifically authorized to determine their own salaries.

Second, even authorized actions must not breach directors’ fiduciary duties. Fiduciary duties represent the duty to use corporate powers only for a proper purpose, which under most American jurisdictions means the maximization of shareholders’ welfare. Whereas authority is an ex ante, legislation-determined, general constraint on directors’ discretion, fiduciary duties are ex-post, judge-determined and case specific constraints. The fiduciary duty analysis is itself split into two parts:

1. (i) Under a legal principle known as the Business Judgment Rule (‘BJR’), the court defers to the directors (i.e., presumes fiduciary duties were not breached) unless the board acted in bad faithFootnote ⁶ or directors did not sufficiently investigate and deliberate prior to reaching their decision.

2. (ii) If one of the exceptions to the BJR applies, suggesting an acute agency problem, the court may find that the decision breached fiduciary duties if directors were grossly negligent or if the decision is deemed unfair to the shareholders.

The third and final stage of analysis is ratification: fully informed shareholders may ratify (i.e., approve) directors’ decisions, validating them even though they failed one of the two stages above.

While Delaware courts condemn decisions to award executive compensation where directors have a conflict of interest or have merely ‘rubber stamped’ a compensation package determined by others, so far they have tolerated even seemingly over-generous executive compensation awards, if the directors awarding it thoroughly considered the compensation package, relied on appropriate expert advice, and did not appear to have a conflict of interest with the shareholders.

How will corporate law treat directors’ bias arbitrage?

Because corporate law currently portrays directors' conflict of interest as having only two conflicting incentives, it has not afforded separate consideration for directors’ actions that amount to bias arbitrage (i.e., maximizing shareholders’ indirect welfare). I conclude this comment, therefore, with preliminary thoughts as to how corporate law would likely address bias arbitrage. Bias arbitrage raises the following doctrinal questions regarding the legal analysis of directors’ decisions:

1. (i) Does a director's incentive to bias arbitrage amount to a conflict of interest?

2. (ii) Is an action that maximizes shareholders’ indirect welfare at the expense of shareholder's direct welfare inherently unfair to shareholders?

3. (iii) Bias arbitrage deludes the principal as to the reason for the reduction in the principal's perceived risk. Does this amount to fraud (and thus illegality)?

An evaluation of the permissibility of bias arbitrage by directors presents a tension between two conflicting themes of corporate law. The first theme, which underlies the BJR, is the deference afforded to informed decisions by directors who are not self-interested (in the narrow sense). Following this theme, if directors favor action A over action B and both actions are reasonably in shareholders’ interest, courts are highly unlikely to question the decision even if evidence suggests that action B is more favorable to shareholders. Bias arbitrage is almost always likely to increase shareholders’ indirect welfare (since, as discussed above, harmful bias arbitrage that addresses risks shareholders already under-estimate is unsustainable and thus unbeneficial to the director). Thus, a decision to bias arbitrage will increase shareholders’ welfare. At worst, it would not maximize shareholder welfare, if it precludes actions that increase shareholder welfare even more (such as actually mitigating the risk).Footnote ⁷

The second theme – at odds with bias arbitrage – is the importance of truthful disclosure of information to shareholders. Judicial intervention is seen as a last resort to curtail the agency problem in the corporation. Corporate law relies on structural mechanisms (e.g., shareholder voting for directors, performance-based compensation structures) and market mechanisms (e.g., the market for corporate control, directors’ need to raise additional capital in the future) as first lines of defense and deterrence against agent malfeasance. Such mechanisms can be undermined by directors’ manipulative disclosure.

Yet bias arbitrage, like medical placebos, requires the manipulation of the subject's perception to succeed. Shareholders are unlikely to correct their misperception if directors simply told them the risk was misperceived, rather than offering shareholders a reason for the mitigation of the risk. And directors are less likely to attempt to correct shareholders’ misperception if they gain nothing from it.

Thus, while bias arbitrage is welfare-enhancing and self-enforcing, corporate law's acceptance of it is uncertain.