Chapter Summary • 431
CHAPTER SUMMARY
Agency problems arise when (1) a principal hires an agent to take actions that affect the payoff to the principal, (2) the agent’s interests differ from those of the principal, and (3) there is hidden action or hidden information.
Agency problems can be addressed by direct monitoring of the agent’s actions or information. Monitoring is typically imperfect and expensive, and, if the monitor is an agent as well, results in adding a layer of agency to the organization.
Performance-based incentives—where the agent is paid more when the payoff to the principal is high—work by aligning the interests of principal and agent.
If a performance measure is affected by random factors, linking pay more closely to performance places more risk on an employee. Since employees are risk averse, they dislike jobs that involve risky pay, and the firm must compensate the employee for bearing this risk. This means that there is a trade-off between risk and incentives.
Performance measures may fail to reflect activities that the firm wants the employee to pursue while rewarding activities that the firm does not want the employee to pursue. According to the multitask principle, stronger incentives will cause employees to focus more on activities that are measured at the expense of activities that are not measured.
Selecting from among performance measures often involves trading off these costs against each other.
For many jobs, firms can improve on explicit incentive contracts by using implicit incentive contracts. This is true when the available verifiable performance measures are noisy, reward activities that the firm does not want employees to pursue, or fail to reward activities that the firm does want employees to pursue.
Implicit incentive contracts allow firms to make use of performance measures that cannot be verified by external enforcement mechanisms such as judges or arbitrators. A firm that fails to follow through on promises made as part of an implicit contract will lose its reputation as a good employer, and employees will not respond to future incentives based on implicit contracts.
Often firms use supervisors’ subjective assessments of employees’ actions as performance measures in implicit contracts. If supervisors find it difficult to make sharp distinctions among employees, all employees might end up with similar evaluations. This weakens incentives to be a top performer.
Strong incentives can be provided through the use of promotion tournaments. The strength of incentives provided by tournaments depends on the size of the prize—that is, on the difference between the wages earned by the tournament’s winner and losers.
Firms can also provide incentives by threatening to fire underperforming employees. The strength of these incentives depends on the value of the job to employees. Firms that pay efficiency wages make their jobs more valuable to employees and thus increase incentives for effort.
Firms can motivate employees to work together by using team-based performance measures. Such measures can suffer from free-rider problems, however. Firms can combat free-rider problems by keeping teams small, allowing employees to work together repeatedly, and making sure employees working together can observe one another’s actions.
432 • Chapter 12 • Performance Measurement and Incentives
QUESTIONS
1.Using your own experience, if possible, identify three types of hidden information that could affect an agency relationship. Identify three forms of hidden action as well.
2.In the United States, lawyers in negligence cases are usually paid a contingency fee equal to roughly 30 percent of the total award. Lawyers in other types of cases are often paid on an hourly basis. Discuss the merits and drawbacks of each from the perspective of the client (i.e., the principal).
3.Suppose that you were granted a “risky job” of the type examined in this chapter. The job pays $40,000 with probability 1/2 and $160,000 with probability ½. What is your certainty equivalent for this risky payoff? To answer this question, compare this risky job to a safe job that pays $100,000 for sure. Then reduce the value of the safe job in $1,000 increments until you are indifferent between the safe job and the risky job. What is your certainty equivalent for a job paying $10,000 or $190,000, each with equal probability?
4.In the United States, lawyers in negligence cases are usually paid a contingency fee equal to roughly 30 percent of the total award. Lawyers in other types of cases are often paid on an hourly basis. Discuss the merits and drawbacks of each from the perspective of the client (i.e., the principal).
5.Suppose that a firm offers a divisional manager a linear pay-for-performance contract based on the revenues of the division the manager leads. The manager’s pay includes a fixed yearly salary F and a fraction of the division’s revenue that is paid to the manager. Suppose that the demand for this type of divisional manager increases, meaning that the firm has to increase this manager’s pay in order to retain her. Should the firm do this by increasing the salary F, the commission , or both? Explain.
6.Regulated firms, such as electric utilities, typically have limited discretion over the prices they charge. Regulators set prices to guarantee a fixed return to the firm’s owners after gathering information about operating costs. Studies of executive pay practices have consistently shown that the compensation of utility CEOs is significantly less sensitive to the firm’s performance than that of nonutility CEOs. Explain why, using the trade-off between risk and incentives.
7.Firms often use quotas as part of compensation contracts for salespeople. A quotabased contract may stipulate, for example, that the salesperson will receive a $10,000 bonus if yearly sales are $1 million or more, and no bonus otherwise. Identify actions a firm probably does not want pursued that the employee will be motivated to pursue under such a contract.
8.While in principle it is feasible for business schools to write explicit pay-for performance contracts with professors, this is rarely done. Identify the drawbacks of the following performance measures for this job:
•Number of research articles published
•Students’ ratings of professors’ courses
•Dollar value of research grants won
•Starting salaries of students after graduation
9.Suppose that Minot Farm Equipment Corporation employs two salespeople. Each covers an exclusive territory; one is assigned to North Dakota and the other
Questions • 433
to South Dakota. These two neighboring states have similar agricultural economies and are affected by the same weather patterns. Durham Tractor Company also employs two salespeople. One works in North Carolina, while the other is assigned to Oregon. Farm products and methods vary considerably across these two states. Each firm uses the dollar value of annual sales as a performance measure for salespeople. Which of the firms do you think would benefit most from basing pay on its salespeople’s relative performance? Why?
10. Consider a potential employee who values wages but also values the opportunity to pursue non–work-related activities. (You may think of these activities as relating to family obligations, such as child care.) Suppose that other jobs available to this employee pay $100 per day but require him to work at the company’s facility, effectively eliminating his ability to pursue nonwork activities. Ignore “effort” for the purposes of this problem and assume that the only agency problem pertains to how the employee allocates his time. Suppose that if the employee allocates all of his time in a day to “work,” he creates $150 worth of value (gross of wages) for the firm. The employee may also have access to two forms of “nonwork” activities:
(1) a high-value nonwork activity (think of unexpected child-care needs) and (2) a low-value nonwork activity (think of leisure—playing video games or watching TV). The employee values the ability to complete the high-value nonwork activity at $200 and the ability to complete the low-value nonwork activity at $50.
(a)Suppose first that the low-value nonwork activity does not exist and that the high-value nonwork activity exists with probability 0.10. (Interpretation: There is a 10 percent chance that the employee will need to perform an important child-care duty each day.) Suppose that your firm is considering offering this employee a telecommuting job. Assume here that if the employee telecommutes and the high-value activity arises, he spends all his time on this activity and creates no value for the firm that day. If the highvalue nonwork activity does not arise, he spends all his time working on behalf of the firm. What daily wage should you offer? What will your profits be? Is your firm better off than if it offered the employee a nontelecommuting job? Why?
(b)Suppose now that the low-value nonwork activity does exist. Unlike the highvalue activity (which only arises with some probability), the low-value activity is always present. Suppose also that the firm cannot pay this employee based on individual performance because the available performance measures are of insufficient quality. Suppose that your firm offers the telecommuting job you described in part (a). According to the multitask principle, how will the employee spend his time? Are your profits higher offering the telecommuting job or the nontelecommuting job?
(c)Next suppose that the firm does have access to a good measure of individual performance. It can make pay contingent on whether the employee works on the firm’s activity. Which job (telecommuting or nontelecommuting) and compensation arrangement (fixed or fixed plus some variable dependent on output) will maximize the firm’s profits? Comment on the types of jobs in which one might expect to see firms offering telecommuting.
11. Giganticorp, a large conglomerate, has just acquired Nimble, Inc., a small manufacturing concern. Putting yourself in the shoes of Nimble’s employees, what concerns do you have about the implicit incentive contracts that governed your relationship with Nimble before the merger? Now place yourself in the position
434 • Chapter 12 • Performance Measurement and Incentives
of Giganticorp’s merger integration team. How might concern about implicit contracts affect your dealings with Nimble’s employees?
12. Oil companies such as British Petroleum and Royal Dutch Shell sell gasoline through their own branded gas stations. In some cases, these companies own their gas stations; in others, the stations are owned by local franchises. How might the following factors affect the choice of corporate versus local ownership?
(a)The gas station also does a lot of automotive repairs.
(b)The gas station is located on an interstate highway.
(c)The gas station has a large convenience store.
ENDNOTES
1“O’Neal Out as Merrill Reels from Loss; Startled Board Ditches a Famously Aloof CEO,” The Wall Street Journal, October 29, 2007.
2Principal–agent relationships have long been a subject of study for economists. See, for example, Jensen, M., and W. Meckling, “Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure,” Journal of Financial Economics, 3, 1976, pp. 305–360, or Holmstrom, B., “Moral Hazard and Observability,” Bell Journal of Economics, 4, 1979, pp. 4–29.
3Berle, A., and G. Means, The Modern Corporation and Private Property, New York, Macmillan, 1932.
4I. Barankay, Q. Bandiera, and I. Rasul, “Social Connections and Incentives in the Workplace: Evidence from Personnel Data,” Econometrica, 77(4), 2009, pp. 1047–1094.
5To derive this, we differentiate the employee’s objective with respect to e and find the value of e at which the resulting quantity is equal to zero. This yields 10 5 e 2 40, or e 5 50.
6Here, we have calculated the employee’s commission payment ($14,000) less effort costs (1/2 (100)2 5 $5,000) as equal to $9,000. Hence, even if the employee is required to pay the firm $8,000 in order to take the job, the employee will be indifferent between this job and other jobs available in this labor market.
7This example is drawn from von Drehle, D., Triangle: The Fire that Changed America, New York, Grove Press, 2003.
8If you think you prefer the risky job, consider the following thought experiment. Suppose that you and a friend each take the safe job. You can easily convert these two safe jobs into two risky jobs by betting $60,000 on a coin flip at the end of the year. Would you be willing to do this? Note that it would be easy for people to inject randomness into their wealth in this way, but we rarely observe anyone doing so. This suggests that most people do indeed dislike large random fluctuations in their wealth.
9See Holmstrom, B., and P. Milgrom, “Aggregation and Linearity in the Provision of Intertemporal Incentives,” Econometrica, 55, 1987, pp. 308–328, and Holmstrom, B., and P. Milgrom, “Multitask Principal–Agent Analyses: Incentive Contracts, Asset Ownership and Job Design,” Journal of Law, Economics and Organization, 7, 1991, pp. 524–552.
10This specification of the employee’s certainty equivalent is useful for expositional purposes, but it does oversimplify preferences somewhat. In particular, the specification ignores the possibility that an individual may become less risk averse as he or she becomes wealthier.
11The marginal cost of effort is the first derivative (or slope) of the cost function with respect to e. The cost function is ½ (e 2 40)2, which has derivative (e 2 40). To find the value of e for which marginal benefit to marginal cost, we have 100 5 e 2 40, which yields e 5 40 1 100 .
12Note that if the variance of sales is 10,000, the standard deviation of sales is 100. 13Shi, L., “Productivity Effect of Piece Rate Contracts: Evidence from Two Small Field
Experiments,” Working Paper, University of Washington, 2007.
Endnotes • 435
14This example is discussed in Holmstrom and Milgrom, “Multitask Principal–Agent Analyses.”
15Note, however, that teaching to the test does happen in practice. According to Catalyst, an independent publication assessing public school reform in Chicago, some schools have “narrowed their curriculum in the pursuit of (standardized test) gains. At one South Side elementary school, the principal told her faculty to cut science, social studies, and writing from the curriculum and ‘just prepare for the test,’ an eighth grade teacher reports.” See “Accountability Impact Both Positive, Negative,” Catalyst: Voices of Chicago School Reform, October 2000.
16See “Incentive Pay Can Be Crippling,” Fortune, November 13, 1995.
17To illustrate, consider a setting in which two salespeople’s individual performance measures depend on their effort, their individual luck, and the condition of the local economy. Suppose that employee A’s individual performance is eA 1 A1 1 2, where A1 is a random variable representing individual luck and 2 is a random variable representing the condition of the local economy. Suppose similarly that employee B’s individual performance is eB 1 B1 1 2, and that A1 and B1 are independent. In this case, the total variation in employee A’s performance is A1 1 2. This is positively correlated with the variation in employee B’s performance because2 affects both. If the firm uses the difference between the employees’ individual output as a performance measure, employee A’s pay will depend on eA 2 eB 1 A1 2 B1. If the variance of2 is large relative to that of A1 and B1, this relative measure exposes employees to less risk than does the absolute measure.
18Zwiebel, J., “Corporate Conservatism and Relative Compensation,” Journal of Political Economy, 103, 1995, pp. 1–25.
19This discussion draws from a survey article written by Candice Prendergast. See “The Provision of Incentives in Firms,” Journal of Economic Literature, 37, 1999, pp. 7–63.
20Parsch, H., and B. Shearer, “Piece Rates, Fixed Wages, and Incentive Effects: Statistical Evidence from Payroll Records,” International Economic Review, 41, 2002, pp. 59–92.
21Gaynor, M., J. Rebitzer, and L. Taylor, “Physician Incentives in Health Maintenance Organizations,” Journal of Political Economy, 2004, pp. 915–932.
22Drago, R., and G. Garvey, “Incentives for Helping on the Job: Theory and Evidence,”
Journal of Labor Economics, 16, 1998, pp. 1–25.
23“Knowledge Management Sweeping Korea’s Corporate Landscape,” Korea Herald, June 22, 2002.
24“Rank and File Attrition Isn’t Working, So Best-to-Worst Grading is Gaining,” Time, June 18, 2001.
25“More Firms Cut Workers Ranked at Bottom to Make Way for Talent,” USA Today, May 30, 2001.
26Lazear, E., and S. Rosen, “Rank Order Tournaments as Optimal Labor Contracts,”
Journal of Political Economy, 89, 1981, pp. 841–864.
27Our discussion here omits a subtle aspect of tournament theory. Because it is the best loan officer who earns the promotion, the probability that loan officer 1 will win the tournament depends not just on his own effort but also on that of the second loan officer. Officer 1’s optimal effort choice therefore may depend on the effort choice made by officer 2. The simultaneous choice of efforts by competitors in a tournament is conceptually similar to the simultaneous choice of quantities by Cournot duopolists. Lazear and Rosen derived reaction functions for tournament competitors and solved for the Nash equilibrium of this game.
28Rosen, S., “Prizes and Incentives in Elimination Tournaments,” American Economic Review, 76, 1986, pp. 921–939.
29Baker, G., M. Gibbs, and B. Holmstrom, “The Wage Policy of a Firm,” Quarterly Journal of Economics, 109, 1994, pp. 921–956.
436 • Chapter 12 • Performance Measurement and Incentives
30Main, B., C. O’Reilly, and J. Wade, “Top Executive Pay: Tournament or Teamwork?
Journal of Labor Economics, 11, 1993, pp. 606–628.
31Eriksson, T., “Executive Compensation and Tournament Theory: Empirical Tests on Danish Data,” Journal of Labor Economics, 17, 1999, pp. 262–280.
32This example is based on Brown, J., Journal of Political Economy, forthcoming. We thank Professor Brown for writing the example.
33The wages w and w** in this model can be interpreted rather broadly. The wage w, for instance, can be viewed as the net present value of the employee’s future employment prospects conditional on retaining her current job today. The wage w** can be interpreted as the net present value of future employment prospects conditional on being fired from the current job. Many factors could cause w** to be less than w; firing may result in a long and costly period of unemployment, a black mark on a résumé, or a lower-paying next job.
34Shapiro, C., and J. Stiglitz, “Equilibrium Unemployment as a Discipline Device,”
American Economic Review, 74, 1984, pp. 433–444.
35Quoted in Allan Nevins, Ford: the Times, the Man, the Company, New York, Charles Scribner’s Sons, 1954. See also Raff, D., and L. Summers, “Did Henry Ford Pay Efficiency Wages?” Journal of Labor Economics, 5, 1987, pp. 57–86.
36Gaynor, M., and M. Pauly, “Compensation and Productive Efficiency in Partnerships: Evidence from Medical Group Practice,” Journal of Political Economy, 98, 1990, pp. 544–573.
37Leibowitz, A., and R. Tollison, “Free Riding, Shirking, and Team Production in Legal Partnerships,” Economic Inquiry, 18, 1980, pp. 380–394.
38Knez, M., and D. Simester, “Firm-wide Incentives and Mutual Monitoring at Continental Airlines,” Journal of Labor Economics, 19, 2001, pp. 743–772.
39This example is drawn from Gant, J., C. Ichniowski, and K. Shaw, “Working Smarter by Working Together: Connective Capital in the Workplace,” Working Paper, Stanford University, 2003.