Peer Evaluation: Incentives and Coworker Relations

• Published date: 01 March 2016
• Author: Joeri Sol
• DOI: http://doi.org/10.1111/jems.12134
• Date: 01 March 2016

Peer Evaluation: Incentives and Coworker Relations

JOERI SOL

Amsterdam Business School

University of Amsterdam

Amsterdam, Noord-Holland

j.sol@uva.nl

In many workplaces, coworkers have the best information about each other’s efforts. This paper

studies a principal who only observes the joint output by two limitedly liable agents, while agents

receive signals about each other’s effort levels. The principal attempts to exploit this information

through peer evaluation; agents are asked to report their signal and may receive a bonus for

being evaluated positively. Lying aversion ascertains that truthful evaluation is possible, while

interpersonal relations between colleagues give an incentive to misreport. This paper shows that

peer evaluation gives an incentive for effort, even when the evaluations are not truthful. The peer

evaluation bonus is constrained by more intensive coworker relations. Still, the optimal contract

always includes a peer evaluation bonus, sometimes complemented with a team bonus. Coworker

relations have nonmonotonic effects on profits in the optimal contract.

1. Introduction

In many employment relations, managers lack information about employees’ individual

effort or performance; for example, in highly interdependent work teams, with work on

location, or in case of experts. In an attempt to obtain a more complete picture of em-

ployee performance, many firms have turned to multisource feedback; in business better

known as 360◦evaluations. A 360◦evaluation can include performance assessments by

subordinates, peers, supervisors, customers, or other stakeholders. According to survey

data, about 90% of Fortune 1000 firms use some form of multisource feedback, often in-

cluding evaluation by peers (Edwards and Ewen, 1996). Moreover, in many companies

peer evaluation schemes partially determine personnel decisions regarding promotions

and performance pay (Bohl, 1996).1

Evaluation by peers has intuitive appeal when employees have valuable informa-

tion about one another’s performance. However, when seen from an economic theory

perspective, the widespread use of peer evaluation is less obvious. A major concern

is that employees may benefit from providing the employer with invalid performance

evaluations of their colleagues. For instance, employees can provide invalid ratings so

as to help their friends, hurt their competitors, or “game the system” through collusion

(Edwards and Ewen, 1996; Kozlowski et al., 1998). In line with this, Tsui and Barry

(1986) find that performance evaluations are more lenient when raters experience more

I gratefully acknowledge comments and suggestions by Daniel Arce, Josse Delfgaauw, Robert Dur, Amihai

Glazer, Vladimir Karamychev, Navin Kartik, Mirjam van Praag, Paul Steffens, and seminar audiences at

Erasmus University and at University of Amsterdam, and participants of the “Social Relations and Incentives

in the Workplace” workshop in Rotterdamand the IMEBE 2011 in Barcelona.

1. Bohl (1996) finds that one-fifth of 750 surveyed companies reports to use peer evaluations, of which 90%

includes the peer evaluation in personnel decisions. Similarly,in a survey among 280 firms, roughly one-fifth

of the firms have adopted some form of peer evaluation (Antonioni, 1996).

C2015 Wiley Periodicals, Inc.

Journal of Economics & Management Strategy, Volume25, Number 1, Spring 2016, 56–76

Peer Evaluation 57

personal affect toward the ratee.2Love (1981) even asserts that: “The most common

rationale given by organizational personnel for avoiding the use of peer assessment is

that it is simply a ‘popularity contest’.” (p. 451). This paper develops a formal agency

model so as to study when and how peer evaluation should be used under different

intensities of coworker relations.

I study peer evaluation in a model of team production by two homogeneous agents,

who are protected by limited liability. The production suffers from a pure moral hazard

problem; that is, the principal is unable to distinguish individual inputs, and can only

offer a team bonus for output. A team bonus results in inefficiently low effort levels,

since limited liability makes the budget breaking solution infeasible. Agents, on the

other hand, receive a signal about their colleague’s effort provision. The principal can

attempt to gather the information from these signals by employing peer evaluation, and

consequently attempt to offer more efficient incentives. A bonus is paid to those agents

that receive a positive evaluation report. As a starting point, I assume ideal circumstances

for peer evaluation: Agents have an aversion toward lying about their signal. A small

cost of lying is sufficient to ensure truthful evaluations, as the agents’ reports only

affect the payoff of their colleague.3Peer evaluation motivates agents to exert effort

so as to increase the likelihood that their coworker receives a positive signal and rates

performance accordingly. Under these ideal circumstances, peer evaluation outperforms

a team bonus; in fact, peer evaluation performs as well as individual incentives would.

Next, I allow for interpersonal relations between colleagues, either good or bad.

Social preferences can give agents an incentive to lie about their signal; that is, coworker

relations cause a (dis)likeability bias in the evaluations. In case, the internalized utility

from a bonus to a friend outweighs the cost of lying, agents in good relationships give

positive evaluations regardless of their signal. Likewise, agents in bad relationships

may begrudge their colleague a bonus and lie upon receiving a positive signal. The

anticipation of invalid peer evaluation still provides incentives for effort, as agents

also internalize of a colleague’s lying costs. For example, friends who give each other

positive evaluations irrespective of performance still provide effort so as to limit the

expected lying costs of the befriended colleague. However, there is only a limited range

of parameter values for which nontruthful peer evaluation is optimal, and exclusive to

workplaces with bad coworker relations.

The principal can ensure truthful evaluation by adjusting the bonus for peer eval-

uations downward, such that coworker relations do not affect the evaluation decision.

In other words, social preferences constrain the effectiveness of peer evaluation. Never-

theless, the optimal contract always includes peer evaluation. The principal will com-

plement incentives for effort by including a team bonus in case peer evaluation becomes

severely constrained. The combination of peer evaluation and a team bonus gives an

interesting comparative static result regarding the effect of coworker relations on prof-

its. The effectiveness of the team bonus increases with better coworker relations, as

coworker relations mitigate incentives to free ride on a colleague. Peer evaluations suf-

fer from more pronounced coworker relations, through the earlier discussed likability

bias. Taken together, coworker relations have nonmonotonic effects on profits.

2. Similar significant correlations between likeability and leniency in peer evaluation can be found in

Sonnentag (1998) and Love (1981). Antonioni and Park (2001) show that peer evaluations suffer more from

such a likeability bias than traditional performance evaluations.

3. This paper studies how peer evaluation may suffer under different intensities of coworker relations.

In doing so, it abstracts from a number of other motivations that may lead to invalid evaluations, such as

competition and collusion between workers. I come back to these motivations in the concluding remarks.