The role of performance appraisals in motivating employees

• Date: 01 June 2018
• DOI: http://doi.org/10.1111/jems.12241
• Author: Otto H. Swank,Jurjen J.A. Kamphorst
• Published date: 01 June 2018

Received: 3 November2015 Revised: 2 October 2017 Accepted: 3 January 2018

DOI: 10.1111/jems.12241

ORIGINAL ARTICLE

The role of performance appraisals in motivating employees

Jurjen J.A. Kamphorst Otto H. Swank

Erasmus School of Economics and Tin-

bergenInstitute, Rotterdam, The Nether-

lands (Email: kamphorst@ese.eur.nl;

swank@ese.eur.nl)

Abstract

Workers' rewards and career perspectives often depend on how their supervisors per-

ceive their performance. However, evaluating a worker's performance is often diffi-

cult. We develop a model in which a worker is uncertain about his own performance

and about his supervisor's ability to assess him. The supervisor gives the worker a

performance appraisal aiming to affect both the worker's self-perception and his own

credibility in assessing the performance. We examine how performance appraisals

affect the worker's future performance. Our model's predictions are consistent with

empirical findings. Supervisors give, on average, “too” positive appraisals, and both

positive and negative feedback can (de)motivate workers.

1INTRODUCTION

In 2003, a large financial service provider (FSP) in the Netherlands introduced a newincentive system for its workers (Bol, 2008).

An important feature of the new incentive system wasthat compensation of workers depended on various subjective performance

measures. Managers rated subordinates on a five-point scale on issues like cooperative behavior, business development skills,

and organization skills. Bol (2008) examined how managers at the FSP rated their workers.She found that these managers were

subject to two kinds of biases. First, they provided their subordinates higher ratings than was warranted by their performances.

This tendency is known as the leniency bias.1Second, managers tended to discriminate too little. This phenomenon is known

as the centrality bias.2Bol also examined the determinants of the biases. She found that managers who had less information

about their workers' performances were more subject to both biases. Finally, Bol examined the consequences of biased ratings

for workers' future performances.3She found that the leniency bias on average positively influenced future performances of

workers, whereas the centrality bias on average negatively affected future performances.

Measuring performance is an essential part of any compensation system. Objective indicators of all aspects of a complex job

are rarely available. For those jobs, firms often use subjective performance evaluation.Subjective performance evaluation, by its

very nature, requires that supervisors form perceptions. Supervisors have been found to vary in their skills to appraise workers

(see, apart from Bol, 2008, Napier & Latham, 1986; Tziner, Murphy, & Cleveland,2001). This finding has potentially important

implications when workers' rewards or career perspectivesdepend on how their super visors perceive their performance. Doubts

about a supervisor's ability to assess performance accurately weaken a worker's incentives to exert effort. As a result, when

providing performance appraisals, the supervisor's reputation for being capable of correctly assessing performance is at stake.4

In this paper, we develop a model in which supervisors differ in their ability to appraise workers' performances. We use this

model to better understand how supervisors appraise their workers, and how workers respond to these performance appraisals.

We thank Jasmijn Bol, Francis Bloch, Josse Delfgaauw, Silvia Dominguez-Martinez, Robert Dur, Kris de Jaegher, Botond Köszegi, Victor Maas, Menno

Middeldorp, John Morgan, Sander Onderstal, Ronald Peeters, Canice Prendergast,Ar no Riedl, StephanieRosenkranz, Dana Sisak, Wim van der Stede, Roland

Strausz, Roland van Weelder, Utz Weitzel, Bastian Westbrock, two anonymous referees and coeditor, as well as participants of the UMR-GAEL seminar

(February 2013), EALE 2012, EEA-ESEM 2012, the Workshop on Economic Theory and Game Theory (December 2011), and the ESE Brown Bag seminar

for their helpful comments. Also, we gratefully acknowledgethe financial support of this research by the NWO (Nederlandse Organisatie voor Wetenschappelijk

Onderzoek) through grant no. 400-07-122.

J Econ Manage Strat. 2018;27:251–269. © 2018 WileyPeriodicals, Inc. 251wileyonlinelibrary.com/journal/jems

252 JOURNAL OF ECONOMICS & MANAGEMENTSTRATEGY

The contribution of this paper is twofold. First, to our knowledge our paper offers the first model that explains both the

empirical results on supervisors' behavior on providing appraisals, and how workersrespond to performance appraisals in a single

setting. Second, this paper shows that cheap-talk messages to motivate workers can contain information when the supervisor is

concerned about her reputation for being able to assess performances correctly.5

Indirectly our paper contributes to the literature on compensation schemes, as measuring performance is an important aspect

of compensation schemes. We refer to an earlier version of our paper for an analysis of how imperfect evaluations of workers

should affect compensation schemes.6Important for the results of the present paper is that, if recognized by his supervisor,a

better performance benefits a worker.

The model we develop has four key characteristics. First, at the beginning of the game, both the supervisor and the worker

form a perception of the worker's ability.We model this formation of perceptions by assuming that the supervisor and the worker

receive private signals.7Second, we assume that supervisors differ in their abilities to assess the worker's performance correctly.

The motivation for this assumption is that supervisors have been found to vary in their beliefs about their skills to appraise their

subordinates. In our model, it is important that a supervisor who is better at observing the worker's ability is also better in judging

the worker's performance. Third, we assume an environment where the worker's utility depends on his supervisor's assessment

of his performance.8For instance, the worker may desire recognition,9or the assessment may affect the assignment of tasks,

the worker's bonus or his (internal) career opportunities. At the FSP studied by Bol (2008), a worker's bonus depends on his

supervisor's rating. Key is that a better performance benefits the worker only if it is recognized by the supervisor. Finally, the

worker's ability and his effort are complements. The implication of this last characteristic is that the more confident the worker

is about his ability, the more effort he exerts.

We derive several results. Our first set of results pertains to a situation where the worker knows his own ability. In this

extreme situation, performance appraisals only provide information about the supervisor's ability to assess the worker's ability

correctly. A supervisor who gives an incorrect assessment of a worker's performance loses credibility. A direct implication is

that a good supervisor, who observes a worker's performance, has no incentive to rate it incorrectly. This would only damage

his credibility. The workerwould doubt whether his future performance would be correctly assessed. For a bad supervisor, who

does not observe a worker's performance, three forces are at work. First, she has an incentive to give an appraisal that is most

likely to be consistent with the worker's perception. Second, as the worker's ability and his effort are complements, it is more

important for the supervisor that her evaluation is correct in case the worker is more able. This force leads to a positive bias in

performance appraisals. Finally, a bad supervisor wants to come across as good. This gives her an incentive to give an appraisal

that able supervisors give relatively frequently. We show that this third force tends to dampen the total effect of the first two

forces.

The second set of results is derived from the version of the model in which we have relaxed the assumption that the worker

knows his own ability. In this setting, apart from the incentives discussed above, a supervisor has an incentive to give positive

appraisals. The reason for this incentive is that the worker's effort is an increasing function of his belief about his ability. This

result explains the leniency bias often found in the empirical literature on performance rating. The idea that supervisors give

positive appraisals to boost workers' perceptions of their abilities to make them work harder is not novel. Bénabou and Tirole

(2003), for instance, show that giving a challenging task to a worker signals confidence and thereby motivates. New is that

simple cheap-talk messages may motivate workers.10,11

Our results on supervisors' incentives to give particular appraisals are important for understanding workers' responses to

appraisals. For example, we show that a positive appraisal motivates a worker more when he has a positive perception of his

own performance. A negative appraisal, by contrast, motivates a worker with a negative perception of his performance more

than a worker with a positive perception.

Apart from the business literature on performance appraisals, this paper is most closely related to the literature on subjective

performance appraisals (important early papers are Baker, Gibbons, & Murphy, 1994; Bull, 1987; and Gibbs et al., 2004; see

Prendergast, 1999; and Bol, 2008, for reviews of the literature, and Ederhof, Rajan & Reichelstein, 2011, for a synthesis of the

recent literature on discretionary rewards with an emphasis on the accounting literature). Key notion in this literature is that most

people do not work in jobs where all aspects of a worker's performance are verifiable. Contracting upon subjectiveper formance

appraisals is problematic as supervisors have incentives to save labor costs by underreporting performance. However, repeated

interaction may allow for an implicit contract in which rewards are based on unverifiable information.

Zábojník (2011) models the interaction between an objective and a subjective performance measure. Like us, he assumes

that the supervisor has superior information about the worker's ability. Unlike us, the supervisor is the residual claimant. As a

result, the supervisor has an incentive to underreport performance. Zábojník (2011) presents very interesting results about the

pros and cons of committing to specific distributions of evaluations. His model is less relevant for situations where supervisors

are not residual claimants. This also holds for Suvorov and vande Ven (2009) who examine how the size of a reward may signal