Print
E-MailThis post comes from Alex Edmans at the University of Pennsylvania, Xavier Gabaix and Tomasz Sadzik, both of New York University, and Yuliy Sannikov of the University of California, Berkeley.
In our paper, Dynamic Incentive Accounts, which was recently updated after being presented at the Harvard Law School / Sloan Foundation Conference on Corporate Governance in March, we study how executive compensation might be reformed to address a number of issues that were important contributors to the recent financial crisis. We consider a setting in which the CEO can manipulate short-term earnings at the expense of long-run value (e.g. by writing sub-prime loans that become delinquent several years later, or scrapping investment projects) and may undo the contract by privately saving. In addition, shocks to firm value may reduce the incentive effect of securities that the CEO is given as part of his contract – for example, if the stock price declines, options may fall out of the money and have little incentive effect.
We solve for the optimal contract in such a setting. We find that it involves a “deferred reward principle”: since the agent is risk-averse, it is efficient to spread the reward for effort across all future periods rather than concentrating it in the current period. The relevant measure of incentives is the percentage change in CEO pay for a percentage change in firm value; in real variables, this is the fraction of CEO pay that comprises of stock. The required fraction of stock represents the contract’s sensitivity and is both scale- and time-independent: it does not depend on firm size or total pay, and is the same in each period. With a finite horizon, the required fraction of stock is now increasing over time – the “increasing incentives principle.” As the CEO approaches retirement, there are fewer periods in which to spread the reward for effort, and so the reward in the current period must increase.
The possibility of manipulation has two effects on the optimal contract, which must change to prevent such behavior. The CEO’s income is now sensitive to firm returns even after retirement, to deter him from inflating the stock price just before he leaves. In addition, the contract sensitivity now rises over time, even in an infinite-horizon model. The CEO benefits immediately from short-termism as it boosts current consumption, but the cost is only suffered in the future and thus has a discounted effect.
In practice, the optimal contract can be implemented in a simple manner. When appointed, the CEO is given a “Dynamic Incentive Account”: a portfolio of which a given fraction is invested in the firm’s stock and the remainder in cash. The Dynamic Incentive Account contains two key features. The first is rebalancing, to ensure that the CEO always exerts effort. As time evolves, and firm value changes, the portfolio of cash and stock is constantly rebalanced, to ensure the fraction of stock remains sufficient to induce effort at minimum risk to the CEO. For example, if the stock price falls and the fraction of stock drops, the CEO is “reloaded” by exchanging some of his cash for stock. Importantly, this additional equity is not given for free – it is fully paid for by a reduction in cash. This addresses a key concern with the current practice of “reloading” CEOs by repricing stock options that fall out of the money – the CEO is rewarded for failure.
The second feature is gradual vesting, to ensure that the CEO never wishes to manipulate. The Dynamic Incentive Account vests gradually: the CEO is only allowed to withdraw a fraction of his cash and shares in each period. The account continues to vest slowly after retirement, to dissuade the CEO from inflating the stock price and then cashing out upon departure. The paper thus provides a theoretical framework to guide a potential reform of executive compensation (by either shareholders or the government), to prevent the problems that manifested in the recent crisis.
The full paper is available for download here.
One Comment
I like the theory behind this paper, but what’s to stop the CEO from transferring his non-vested interest through some options contracts post-retirement?
I doubt there’s a form of contract that can keep CEOs from robbing Peter (where Peter is a prospective future shareholder) to pay Paul (where Paul is a current shareholder). Shareholders have to want CEOs to take a long-term view. Some do, some don’t.
Getting the average shareholder to take a longer view means making the time-horizons of various capital expenditures and expected IRR more transparent. Paradoxically, I believe this may require a return to something more like cash accounting. Accrual accounting gives CFOs (who report to the CEOs!) too much discretion, and also doesn’t scale with the complexity of the organization. In terms of information, the financial statements made available by a firm like Citi include far less information per unit of market cap than do the financial statements made available by a firm like Coke.