CEO Horizon, Optimal Pay Duration, and the Escalation of Short‐Termism

• DOI: http://doi.org/10.1111/jofi.12770
• Date: 01 August 2019
• Author: IVAN MARINOVIC,FELIPE VARAS
• Published date: 01 August 2019

THE JOURNAL OF FINANCE •VOL. LXXIV, NO. 4 •AUGUST 2019

CEO Horizon, Optimal Pay Duration, and the

Escalation of Short-Termism

IVAN MARINOVIC and FELIPE VARAS∗

ABSTRACT

This paper studies optimal contracts when managers manipulate their performance

measure at the expense of firm value. Optimal contracts defer compensation. The

manager’s incentives vest over time at an increasing rate, and compensation becomes

very sensitive to short-term performance. This generates an endogenous horizon prob-

lem whereby managers intensify performance manipulation in their final years in of-

fice. Contracts are designed to encourage effort while minimizing the adverse effects

of manipulation. We characterize the optimal mix of short- and long-term compensa-

tion along the manager’s tenure, the optimal vesting period of incentive pay, and the

dynamics of short-termism over the CEO’s tenure.

SHORT-TERMISM IS PREVALENT AMONG MANAGERS. Graham, Harvey, and Rajgopal

(2005) find that 78% of U.S. CEOs are willing to sacrifice long-term value

to beat market expectations. For example, Dechow and Sloan (1991) argue

that, by the end of their tenure, CEOs tend to cut R&D investment, which,

though profitable, has negative implications for the firm’s reported earnings.

Managerial short-termism has been the suspect of concern for many years,

but it has assumed a particularly prominent role in recent years following the

Enron scandal and the financial crisis in 2008.

To understand this phenomenon, the theoretical literature has adopted two

approaches. One approach studies CEO behavior, taking managerial incentives

as given, and thus is silent about optimal incentives (see, e.g., Stein (1989)).

However, the complexity of CEO contracts in the real world (which include

accounting-based bonuses, stock options, restricted stock, deferred compensa-

tion, clawbacks, etc.) suggests that shareholders are aware of potential ma-

nipulation by CEOs and design compensation to mitigate the consequences

∗Marinovic is with the Graduate School of Business, Stanford University. Varas is with Fuqua

School of Business, Duke University. We thank Philip Bond (Editor), Associate Editor, and two

anonymous referees for constructive comments that have significantly improved the paper. We

also thank Jeremy Bertomeu, Tim Baldenius, Peter Cziraki, Alex Edmans (discussant), Simon

Gervais, Ilan Guttman, Zhiguo He, Francois Larmande (discussant), Gustavo Manso (discussant),

Ross Morrow, Adriano Rampini, Florin Sabac, Alex Storer, Vish Vishwanathan, and Jaime Zender

(discussant) for helpful comments; and seminar participants at Alberta (Canada), Duke University

(Fuqua), Bocconi, NYU, SFS Finance Cavalcade, Tilburg, FIRS Conference, TexasFinance Festival,

WFA, U. of Vienna,U. of Mannheim, and the 10th Accounting Research Workshop for their helpful

feedback. We have the Journal of Finance’s disclosure policy and have no conflicts of interest

to disclose.

DOI: 10.1111/jofi.12770

2011

2012 The Journal of Finance R

of such manipulation. An alternative approach studies optimal compensation

contracts that are designed to fully remove CEO manipulations. In this class of

models, manipulation is not observed on the equilibrium path (Edmans et al.

(2012)). This approach is particularly helpful in settings in which CEO ma-

nipulation is too costly to the firm or easy to rule out, but it cannot explain

why manipulation seems so frequent in practice or why real-world contracts

tolerate or even induce manipulation (see Bergstresser and Philippon (2006)).

In this paper, we study optimal compensation contracts when CEOs exert

hidden effort but can also manipulate the firm’s performance to increase their

compensation, sometimes at the expense of firm value. Building on Holmstrom

and Milgrom (1987), we consider a setting with a risk-averse CEO who can save

privately and consume continuously, and who exerts two costly actions: effort

and manipulation. Both actions increase the CEO’s performance in the short

run, but manipulation also has negative consequences for firm value. As in

Stein (1989), we assume that these consequences are not perfectly/immediately

captured by the performance measure but rather take time to be verified,

potentially creating an externality when the CEO tenure is shorter than the

firm’s life span.

Our paper makes two contributions to the literature. First, on the normative

side, we study the contract that maximizes firm value in the presence of ma-

nipulation. We characterize the optimal mix of long- and short-term incentives,

the duration of CEO pay over CEO tenure, and the ideal design of clawbacks

and postretirement compensation. Second, on the positive side, we make pre-

dictions about the evolution of CEO manipulations along CEO tenure, and we

establish the existence of an endogenous CEO horizon problem.

We study the timing of manipulation: how it evolves over CEO tenure and

whether optimal contracts generate a horizon effect whereby the CEO distorts

performance at the end of his tenure. Previous literature shows that in dynamic

settings one can implement positive effort and zero manipulation at the same

time (unlike in static settings) by appropriately balancing the mix of short-

and long-run incentives. However, in our setting, inducing zero manipulation

is not optimal; rather, tolerating some manipulation is desirable because doing

so allows the firm to elicit higher levels of effort than a manipulation-free

contract. Furthermore, to fully discourage manipulation, the firm would have

to provide the CEO with a large postretirement compensation package that

ties his wealth to the firm’s postretirement performance. Such postretirement

compensation is costly to the firm, as it imposes risk on the CEO during a period

when effort does not need to be incentivized and the CEO must be compensated

for bearing this extra risk (see, e.g., Dehaan, Hodge, and Shevlin (2013)).

In our model, performance pay at some date thas the benefits of providing

incentives at tand of deterring manipulation prior to t. On the other hand,

it has the cost of encouraging manipulation at time t. This trade-off shapes

the contract design and the evolution of performance pay along the CEO’s

tenure.

In the absence of manipulation, short-term incentives—measured as the

contract’s pay-performance sensitivity (PPS)—are constant over time, as in

CEO Horizon, Optimal Pay Duration, and the Escalation 2013

Holmstrom and Milgrom (1987). Unfortunately, the simplicity of this contract

vanishes under the possibility of manipulation. A constant PPS contract is no

longer optimal because it induces excessive manipulation, particularly in the

final years in office. Indeed, offering the CEO a stationary contract would lead

him to aggressively shift performance across periods, boosting current per-

formance at the expense of firm value. To mitigate this behavior, an optimal

contract implements lower levels of short-term compensation and higher lev-

els of long-term compensation, measured roughly as the present value of the

contract’s future slopes.

Also, in the absence of manipulation, CEO incentives vest deterministically,

whereas under the possibility of manipulation, vesting depends on firm perfor-

mance. This is empirically relevant. Bettis et al. (2010) assert that, even though

restricted stock awards with time-vesting provisions account for the major-

ity of performance-based pay in U.S. companies, shareholder advocacy groups

and proxy research services have expressed concern that these provisions do

not provide sufficiently strong incentives and have suggested that compen-

sation contracts include performance-based vesting conditions. In fact, since

the mid-1990s, U.S. firms have increasingly issued option and stock awards

with sophisticated performance-based vesting conditions. Our paper provides

a rationale for this phenomenon. Under the possibility of manipulation, the

optimal contract defers compensation and includes performance-based vesting

provisions. In the absence of manipulation, the vesting date of incentives is

known at the start of the CEO’s tenure and is independent of the firm’s perfor-

mance. When the CEO can manipulate performance, the optimal contract in-

cludes performance-based vesting. Thus, the duration of incentives is random:

vesting accelerates with positive shocks and is delayed with negative shocks.

Random vesting is helpful in the presence of manipulation because it allows the

principal to change the level of long-term incentives without having to simul-

taneously distort short-term incentives to avoid creating an imbalance, which

would trigger extra manipulation. Hence, performance-based vesting provides

the principal with an additional degree of freedom to reduce the CEO’s long-

term incentives without having to distort effort to contain manipulation.

The optimal contract also includes a postretirement package that ties the

manager’s wealth to the performance of the firm, observed for some time after

his retirement. This contracting tool is helpful but has limited power when the

CEO is risk-averse: even when the firm has the ability to tie the manager’s

wealth forever—and to any degree—to the firm’s postretirement performance,

the contract generally induces some manipulation. Although it would be possi-

ble to defer compensation long enough to deter manipulation altogether, firms

might not do so given the cost. A key insight of this paper is that firms find it

more beneficial to defer compensation, while the CEO is still on the job rather

than after he retires. This result implies that long-term incentives are larger

at the beginning of the CEO’s tenure and decay toward the end.

Under the possibility of manipulation, optimal CEO contracts are nonlinear,

unlike in Holmstrom and Milgrom (1987). Following Edmans et al. (2012),

we first characterize the optimal contract within the subclass of contracts