Incentives

Chapter Study Outline

14.1 Incentives: An Overview

• The inherent conflict of interest between principals and agents necessitates incentives.
  • This conflict exists from the chief executive officer (CEO) at the top of the organization all the way down to the independent contractor at the bottom.
  • Principals: the owners or shareholders of the firm. They want to maximize the firm’s profits, or value.
  • Agents: supply labor services to the firm. They seek to maximize their own utilities, which depend upon their own consumption and leisure and not the firm’s value per se. Thus agents may not be motivated to work hard.
    • Incentive mechanisms are needed to align the utility-maximizing interests of agents with the principals’ value-maximizing goals. The mechanisms reward some types of behavior and punish others.
• The firm will use one type of technology for production and another type for the monitoring that will provide information about the performance of workers.
  • Better performance measures usually correspond with more effective incentive schemes.
  • The value of output produced by a given worker is denoted $V. In the simplest case, the worker exerts an effort level, e, to produce homogeneous output, y, that is sold at a price, $p, per unit. In this case, $V = p ∙ y.
  • Hard information refers to objective performance measures, denoted h. They are verifiable in court.
  • Soft information refers to subjective performance measures, denoted s. They are inherently nonverifiable.
• The incentive mechanism does not attempt to subvert agents’ utility-maximizing goals. Instead it is designed to harness and channel the agents’ rational self-interests in ways that are more congruent with those of the principal.
  • The incentive mechanisms available to the principal are bounded by the nature of the information she possesses.

14.2 Risk Sharing and Incentives

• Model 14.1: The Principal-Agent Model
  • (a) A risk-neutral firm (the principal) hires a risk-averse worker (the agent) for a single time period to produce a homogeneous good, y, by exerting effort, e ≥ 0. Each unit of output sells for the constant price of $p per unit.
    • e may be interpreted as intensity of work or attention to detail. If e = 0, the agent performs no more than his perfunctory duties at work.
  • (b) If the agent exerts the ex ante level of effort, e, then the ex post realized level of output, y, is governed by y = e + ϵ.
    • ϵ captures the effects of random shocks to output. E(ϵ) = 0, and its variance is σ².
      • The good shocks cancel out the bad shocks, and the variance captures the riskiness of the environment.
  • (c) The agent suffers the disutility from effort C(e), which is increasing and convex in e and satisfies C(0) = 0.
    • The agent’s reservation utility is u₀. It is the utility that the agent derives from the next best alternative use of her time.
    • The agents marginal disutility of effort, denoted MC(e), increases with the level of effort (e).
• Assumption 14.1: Information
  • (a) Ex ante, the impending value of ϵ is unknown to anyone.
  • (b) The level of output, y, and the price, p, are verifiable hard information ex post.
  • (c) Ex post, the principal can neither see the shock, ϵ, nor the agent’s effort, e.
    • The principal does not know the agent’s effort level, so he must design and implement an incentive-payment scheme, denoted w(y), which ties the agent’s earnings, w, to her verifiable performance level, y.
    • This scheme should encourage the agent to exert an appropriate amount of effort.
    • The scheme should also be sensitive to the agent’s risk aversion and the uncontrollable factors causing fluctuations in her output.
• Assumption 14.2: The Piece-Rate Payment Schedule
  • The agent’s earnings, w(y), are governed by the piece-rate contract, w(y) = w₀ + β ∙ y, where $w₀ is a base salary and 0 ≤ β ≤ p is the piece rate.>
    • The piece rate, β, determines the strength of the link between the agent’s performance, y, and his remuneration, $w.
    • Salaries: if β = 0, then w(y) =w₀, which means the agent is salaried; his earnings do not depend on his current performance.
    • Commissions: If β = β₀ ∙ p, then w = w₀ + β₀(p ∙ y). The agent is on commission, and he is paid $β₀ for each $1 of additional revenue he generates.
    • Leasing and Fixed Rent Contracts: If β₀=p and w₀<0, then w(y) = w₀ + py. The agent rents the firm’s productive capital but keeps all the revenues. This arrangement is common for taxicab companies.
  Once the principal has chosen the baseline salary, w₀, and the piece rate, β, the ex post value of the agent’s earnings is w(y) = w₀ + β ∙ y = w₀ + β ∙ (e + ϵ), where y = e + ϵ.
  • The optimal value of β balances the benefits of greater work incentives against the costs of exposing the agent to costly risk.
  • The optimal value of w₀ drives the agent’s utility down to her reservation utility, u₀.
  • In a sharecropping contract the tenant (agent) pays his landlord (principal) some agreed fraction of his crop yield and retains the rest for himself, as in the risk-sharing/incentives model.
    • Sharecropping arrangements often used fixed-rent agreements in which the tenant pays the rent, w₀ < 0, but accrues all of the returns from his efforts, β = p.
  • Measurement costs are also important in establishing the proper piece rate.

14.3 Extensions of the Principal-Agent Framework

• Firms may use salaries or explicit objective incentive payment (OIP) schemes such as piece rates.
  • OIP schemes have drawbacks.
    • They encourage agents to emphasize quantity at the expense of quality.
    • They require costly monitoring technology.
  • In 2002 only 6% of all private-sector U.S. employees had their earnings determined by an explicit OIP.
    • Still, the incidence of OIP schemes is high in certain occupations such as sales, footwear, and clothing.
• Lazear (1986) explained firms’ varying payment schemes based on heterogeneous worker abilities.
  • If all workers exert the same effort, e₀, and everyone is salaried, everyone receives a common salary equal to the average worker’s productivity.
    • Workers with high ability will pay the monitoring cost themselves in order to obtain a performance pay contract, which will pay them more than the salary payment.
    • As high-ability workers leave the salaried worker pool, the average salary decreases with average ability and hence productivity.
    • Equilibrium occurs when no additional salaried workers can benefit by switching to the OIP scheme. Both payment schemes coexist in equilibrium.
  • Payment schemes are designed to encourage higher levels of effort. Alternatively, suppose that effort levels vary, e ≥ e₀) and some firms offer the piece rate contract, w(y) = w₀ + β ∙ y.
    • Salaried workers pick the lowest level of effort, e₀.
    • Piece-rate workers will increase their effort with the piece rate, β, and their own innate abilities, aϵA.
      • Piece-rate workers now earn more because of both their greater effort and ability.
  • Using a rich data set, Lazear (2000) found that piece-rate payment schemes do increase worker output and earnings.
• Given the advantages of OIP schemes, it might seem odd that they are not often used. The ratchet effect offers one explanation.
  • Assumption 14.3: The Ratchet Effect
    • (a) The principal is less well informed than her agents about the inherent difficulty of the tasks they perform.
    • (b) The economic relationship extends over several periods of time.
    • (c) The principal is unable to precommit not to tinker with the incentive scheme as new information is revealed to her over time.
      • Part (c) results in the ratcheting effect, which creates an incentive for the manager to underperform in order to conceal the easiness of his tasks and avoid the ratcheting up of output targets, as in the former Soviet Union.
      • Though piecework was simple in theory, in practice, unless workers collectively restricted output managers would cut pay (Assumption 14.3c) and workers would find themselves working much harder for only slightly more pay.
  • Carmichael and MacLeod (2000) explain that firms will not commit to a fixed piece rate (Assumption 14.3c) because of competitive pressures.
    • If a firm paid its workers more for greater output arising from cost-saving innovations, other firms would take the innovations and gain a cost advantage.
• In reality individual workers function as part of a team; this can create difficulty in measuring individual performance.
  • Often principals reward the team as a whole.
    • Gainsharing plans set a piece rate for the entire team’s output.
    • Discretionary team bonuses are possible rewards.
    • Profit-sharing schemes reward individuals on the basis of the organization’s profits.
  • However, the “1/N problem” undermines team-based incentives by giving individuals an incentive to slack off due to the large number of people among which the reward is divided.
    • Peer pressure may counteract this shirking effect.
    • Often, heterogeneous-ability teams are the most productive, suggesting that there are spillover effects.
• Due to monitoring costs, there may be an efficiency wage that provides workers with a level of utility u* > u₀ while using terminations to punish workers caught shirking.
• The worker has an incentive to work hard because he values his higher-paying job.
• The principal recoups some of her costs from lowering monitoring costs.
• Model 14.2: Efficiency Wages
  • (a) A risk-neutral firm hires a group of risk-neutral workers for a single period.
  • (b) An agent can choose to work, e = 1, or shirk, e,=,0, on the job. If he works, he produces output valued $y; if he shirks, he is worthless.
  • (c) Each agent’s utility is denoted u(e) = c + (1 - e) ∙ G.
    • $c is consumption; $G is goofing-off utility from shirking.
  • (d) The principal apprehends shirkers with exogenous probability q. Those caught shirking are immediately dismissed and earn the reservation wage, $v₀. Those not caught shirking earn $w.
    • In this framework the principal has extremely limited information about agent performance levels.
    • The principal knows that the agent will not shirk only if doing so is counter to the agent’s own self-interest, which requires:
    • u(1) ≥ u(0)
    • $w* = v₀ + {(1 - q) ∙ (G/q)} > v₀, where $w* is an efficiency wage. It is the lowest wage offer that can induce the agent to exert effort.
      • The term in braces represents the worker’s utility surplus due to imperfect monitoring.

14.4 The “You Get What You Pay For” Principle

• Incentive schemes often have dysfunctional and unintended consequences.
  • Employees will respond to the incentives in a payment scheme whether or not those incentives are in the best interests of the organization: if agents are paid to do X, they will do X.
    • Sears mechanics convinced customers that unnecessary services were vital since the mechanics’ pay was tied to average revenue generated per customer.
    • In the public sector, the No Child Left Behind policies led teachers to teach to the test since their careers were tied to their students’ performance on standardized exams.
      • Jacob and Levitt (2003) also found that gearing teachers’ rewards to test scores created powerful incentives to cheat.
    • The 1982 Job Training Partnership Act (JTPA) encouraged employment agencies to engage in cream skimming: since the agencies’ rewards were tied to the success of the people they admitted, they accepted only the most able workers instead of the neediest.
  • Kerr (1975) explained the aptly-named “Kerr’s folly” in “On the Folly of Rewarding A, While Hoping of B.”
    • Payments are conditional on a verifiable performance measure.
    • Agents respond to incentives in a predictable way.
    • With the benefit of hindsight, their responses often run counter to the principal’s interests.
      • Rewarding only verifiable performance measures may encourage agents to neglect other tasks that can be costly to the principal.
  • Thus OIP schemes may only work in simple jobs with a narrowly defined set of tasks and easily measured performance.
    • Since most jobs in the United States are much more complicated, incentives are provided as promotions or bonuses, which depend on subjective evaluations.
      • Even when objective measures are available, subjective measures often provide a more holistic picture of performance.
    • Subjective measures may be hindered by agents’ mistrust of principles, who have an incentive to withhold deserved rewards in order to cut costs.

14.5 Tournaments

• In athletics, tournaments can provide powerful incentives for competitors to perform to the best of their abilities.
  • The tournament principle of “To the victor goes the spoils” also applies to promotions in business hierarchies.
  • Assumption 14.4: Tournaments
    • (a) A given group of contestants compete over a predetermined set of prizes.
    • (b) The contestants have some influence over the probabilities of winning each prize.
    • (c) The final ranking of contestants (and hence the assignment of people to prizes) is based on ordinal rather than cardinal performance measures.
      • Ordinal: relative
      • Cardinal: absolute
      • Both a tennis tournament such as Wimbledon and a firm’s promotion scheme fit this description of a tournament.
• Tournaments have a number of advantages:
  • It is often much cheaper to acquire an ordinal measure of performance than a cardinal measure.
  • Subjective performance rankings give a more holistic view of an agent’s performance.
  • Even though subjective measures are not verifiable in court, the preordained set of prizes fixes the principal’s costs in advance.
  • The tournament insulates risk-averse agents from common productivity shocks.
    • Ceteris paribus, the level of effort depends positively upon the spread between the prizes, rather than the levels of the prizes themselves.
    • For any given prize spread, the level of effort depends on the strength of the nexus between effort and the probability of winning.
• Extensions
  • As the number of contestants increases, the probability that chance events will affect the outcome increases.
    • The tournament model predicts that the optimal prize spread must increase with the number of competitors in order to maintain high-effort incentives.
  • Tournaments reduce agent risk by making all competitors subject to common shocks. They increase agent income risk by using a prize spread.
  • When contests have heterogeneous-ability contestants, handicaps or separate tournaments may be necessary to restore incentives.
    • If agents must choose among options with heterogeneous riskiness, low-ability agents will choose high-risk projects and high-ability agents will choose low-risk projects.
    • The low-ability agents expect to lose anyway, and they maximize their unlikely gains.
    • The high-ability agents expect to win, and they minimize their risk.
    • Either situation may be costly to the principal.
  • In a multi-period elimination tournament, each round is analogous to promotion within a company, until the undefeated winner attains the position as the company’s CEO. Rosen (1986) examined the optimal structure of payments:
    • The winner enjoys the value of the given wage increment Δw_n.
    • At each of the stages n < N, the winner also enjoys the option value of potentially winning the other prizes that remain in the tournament, since the losers are knocked out.
    • As a given worker advances up the career ladder, the future is less attractive since there are fewer stages left to win. The two final competitors recognize that one of them will win and the tournament will end in the last stage where n = N.
      • As a result of this declining option value, Rosen predicts that there will be a large jump in winnings from the second-to-last stage to the final stage.
• Problems
  • Tournaments may encourage too much risk taking.
  • They create an incentive for collusion among agents.
  • Agents may sabotage opponents’ efforts instead of outperforming them, which could be very costly for the principal.