Tunnel-Proofing the Executive Suite

In my forthcoming Review of Financial Studies paper “Tunnel-Proofing the Executive Suite: Transparency, Temptation, and the Design of Executive Compensation”, I model the problem of providing incentives for the faithful performance of the CEO’s role as a custodian of a complex bundle of assets whose value is opaque to outsiders. In particular, I model a CEO who has an opportunity to divert corporate resources to personal consumption, an activity I term “tunneling.” Although I assume that the financial accounting system makes total output observable and contractible, I impose two conditions that make tunneling feasible. First, the firm’s investment capital is subject to diversion. Second, only the CEO knows the marginal return on invested capital. Outsiders only know the distribution of possible marginal rates of return. Because outsiders only know the distribution of project returns, not the return realization, they cannot distinguish between low returns caused by tunneling and low returns caused by bad luck.

If the CEO did not have private information regarding the firm’s production process, then my analysis would produce the standard result—high-powered compensation for risk-neutral CEOs. In this case, very high cash flows would be a very strong signal that the CEO did not divert funds. In which case, high-powered compensation would optimally link compensation with the desired CEO action—not diverting funds. However, when CEOs have intimate knowledge of the production process, very high cash flows indicate not only that (a) the CEO did not divert but also that (b) the marginal return of capital (observed privately by the CEO) is very large. However, when the marginal return on capital is relatively large, the “bribe” required to induce the CEO not to divert capital from investment is smaller as each dollar diverted “under the table” lowers output and thus the CEO’s “over the table” compensation more. Hence, when the marginal return on capital is opaque to outsiders, the optimality of high-powered compensation is no longer obvious.

In fact, when I derive shareholder-value maximizing compensation contracts for CEOs, I find that these contracts are very different from high-powered option-like compensation. Optimal designs have the following form: They fix a minimum output threshold. When the CEO is unable to prevent output from falling below this threshold, which I term the “cutoff rate,” it is optimal for the CEO to divert corporate resources to private consumption. Otherwise, the CEO does not divert. At the cutoff threshold, the CEO receives base rate compensation. Above the cutoff, CEO compensation increases with firm output but at a decreasing rate; in fact, the asymptotic rate of increase is zero. Thus, the optimal custodial design resembles the 80/120 bonus components of executive compensation documented in, for example, Murphy (1986). My conclusion that this sort of nonlinear bonus-like compensation is the optimal is opposed to the conclusion of Jensen (2003) that nonlinear contracting misaligns incentives and generates corporate malfeasance. The base rate of compensation above the cutoff is an increasing linear function of firm size. My work also shows that the better the firm’s ex ante prospects, the higher the cutoff return the firm sets for deterring diversion. This implies that an increase in ex ante firm prospects can actually increase the ex post likelihood of tunneling, and, consequently, make high market/book growth firms more susceptible to tunneling.

The full paper is available for download here.