Lower Salaries and No Options: The Optimal Structure of Executive Pay

1
How Important Are Risk-Taking Incentives in
Executive Compensation?
Ingolf Dittmann Ko-Chia Yu Erasmus
University Shanghai University of Rotterdam Finan
ce and Economics
2
MotivationRelation between risk and incentives

• Informativeness principle (standard agency
  theory)
• More risk ? less incentive pay
• Mixed empirical evidence (Prendergast, 2002)

Firm Risk
CEO incentives
3
MotivationRelation between risk and incentives

• Solid evidence that CEOs respond to risk-taking
  incentives
• Hedging Tufano (1996) Knopf et al. (2002)
• Investments Rajgopal and Shevlin (2002)
• Leverage Coles et al. (2006), Tchistyi et al.
  (2007)
• Acquisitions May (1995), Smith and Swan (2007)
• Stock and bond holders anticipate CEO
  risk-taking DeFusco et al. (1990), Billett et
  al. (2006)

Firm Risk
CEO incentives
4
Research question

• Do shareholders provide risk-taking incentives on
  purpose?
• Or are risk-taking incentives just an unimportant
  side effect of effort incentives?
• Is it important to take into account risk-taking
  incentives when designing a CEO compensation
  package?

5
Approach

• We model the endogeneity between risk and
  incentives
• Principal-agent model
• Effort-averse agent chooses effort and firm
  strategy.
• Firm strategy affects firm value and volatility.
• Incorporate informativeness and risk-taking
  incentives
• Calibrate the model to individual CEO data.
• Model predicts
• Optimal compensation structure for each CEO
• Savings firms could realize by switching
• Compare model predictions with observed contracts
• Better than a model without risk-taking
  incentives?

6
Results (1)Consistence with compensation practice

• Savings from recontracting are small (mean
  10.4)
• Average distance between the observed contract
  and the predicted contract is small. (mean 8.0)
• Much better fit than models with effort aversion
  alone
• Dittmann Maug (2007) find up to 54 savings and
  28.8 difference in distances
• Conclusion Risk-taking incentives play an
  important role in executive compensation practice.

7
Results (1)Consistence with compensation practice
8
Results (2)Application In-the-money options are
optimal

• Replace stock ATM options by ITM options

• Small savings

• If U.S. taxes are taken into account
• Observed contract is optimal for 93 of the CEOs
• Results consistent with the universal use of
  at-the-money options

9
The modelStandard assumptions

• Time t 0 Contract is signed.
• Time T End-of-period stock price PT is realized
  and wage w(PT) is paid.
• Immediately after t 0, the agent chooses effort
  e.
• Firm value E(PT) is increasing and concave in e.
• Agent incurs costs of effort C(e) that are
  increasing and convex in e.
• Stock price is lognormally distributed.
• Agent is risk-averse (CRRA-parameter ?).

10
The modelAdditional assumptions

• In addition to effort, CEO chooses firm strategy
  s.
• Combination of many different actions (e.g.,
  project choice, MA, financial transactions)
• Affects firm value E(PT) and firm risk s.
• Risk-averse CEO with monotonic wage contract will
  choose a strategy (s) that maximizes E(PT) given
  s.
• Choice of s is equivalent to choice of s.
• Reduced form assume that CEO chooses s.
• First-best strategy is associated with risk
• E(PT) is increasing and concave in s if
• E(PT) is weakly decreasing in s if

11
Principal-agent models with effort and
risk-taking incentives

• Agent gathers information and makes project
  choice
• Lambert (1986), Core Qian (2002)
• Agents effort affects mean and variance of stock
  price
• Feltham Wu (2001), Lambert Larcker (2004)
• Continuous effort and volatility choice
• Hirshleifer Suh (1992), Flor, Frimor Munk
  (2006)
• Models in continuous time
• Hellwig (2008) assumes risk-neutral agent

12
The principals problem

• Assume that first-order approach holds, so that
  the incentive compatibility constraint can be
  written as

13
Calibration method

• The full model cannot be calibrated to data,
  because P0(e,s) and C(e) are unknown.
• Solve a simpler problem (first stage of Grossman
  and Hart, 1983) Search for a new contract with a
  given shape that
• provides the same utility to the agent,
• generates the same effort incentives,
• provides the same risk-taking incentives, and
• is as cheap as possible.
• If the model is correct, the new contract must be
  equal to the observed contract.

14
Dataset Construction

• Use CompuStat ExecuComp
• Require 5 years of continuous history
• Estimate wealth from previous years income
• Construct approximate option portfolios
• Aggregate into representative option with same
  value, same option delta and same option vega
• We are left with 727 CEOs (for the year 2006).
• Estimate volatility from daily CRSP returns

15
Dataset Descriptive Statistics
Variable Variable Mean Std. Dev. 10 Quantile Median 90 Quantile
Stock () nS 1.83 4.94 0.04 0.32 4.68
Options () nO 1.37 1.62 0.14 0.92 3.17
Fixed Salary (m) phi 1.64 4.47 0.51 1.04 2.43
Value of contract (m) pi 159.63 1,700.06 4.58 24.97 172.74
Non-firm Wealth (m) W0 62.8 667.0 2.5 12.0 72.2
Firm Value (m) P0 9,294 22,777 377 2,387 20,880
Moneyness () K/P0 70.1 21.7 41.2 72.0 100.0
Maturity T 4.6 1.4 2.8 4.4 6.4
Stock Volatility () sigma 30.0 13.4 16.4 28.3 45.5
Dividend Rate () d 1.24 2.25 0.00 0.63 3.30
Age 56.0 6.8 47 56 64
Return 2001-2005 () 11.8 15.6 -5.7 11.4 28.7

• Table 1

16
Optimal Contracts with Risk-Taking Incentives
Table 3
17
Optimal Contracts that Consist of Salary, Stock,
and Options

• Consider contracts that consist of base salary,
  stock and one option grant.
• Principal minimizes contracting costs over
• Base salary
• Number of shares
• Number of options
• Option strike price

18
Optimal Contracts that Consist of Salary, Stock,
and Options

• Table 6

19
In-the-money options and the U.S. tax system

• IRC 409A Executives must pay a 20 penalty tax
  on the intrinsic value of the option when it
  becomes exercisable.
• Neglect other rules like IRC 162(m).
• Repeat analysis with this tax penalty
• Observed contract is optimal for 76 to 93 of
  all CEOs (depending on assumptions)
• US tax system prohibits in-the-money options.

20
Robustness test Loss-Aversion Utility Function

• Dittmann, Maug, Spalt (2010) showed that if CEOs
  are loss-averse, the principal agent model is
  able to explain current compensation practices.

21
Optimal Contracts with Loss-Aversion
22
Loss Aversion with Risk-Taking Incentives
23
Conclusions

• Optimal compensation structure from a
  principal-agent model where the agent chooses
  effort and firm-volatility
• Model performs much better than a model w/o
  risk-taking incentives.
• Small savings (10.4 vs. 54 w/o risk-taking
  incentives)
• Small distance from observed contract (8 vs.
  28.8 w/o risk-taking incentives)
• Optimal contract is convex over some regions
• Risk-taking incentives are not a issue in LA
  models, but the RTI explanation is less
  susceptive to parameter choices.
• Risk-taking incentives are a major objective in
  executive compensation practice.