«IZA DP No. 3901 A Behavioral Account of the Labor Market: The Role of Fairness Concerns Ernst Fehr Lorenz Goette Christian Zehnder December 2008 ...»
Survey evidence on the prevalence of different compensation forms indicates that there seem to be strong limits to the usage of explicit incentive contracts. Explicit performance pay is still the exception rather than the rule. The vast majority of workers is compensated with hourly wages or salaries and not governed by formal incentive contracts (MacLeod & Parent 1999).
The fact the many firms seem to be constrained to pay flat wages obviously does not mean that they are unable to provide incentives to their workers. Even if the worker's compensation does not explicitly depend on performance, the employer may use the threat of dismissal in case of poor performance or the promise of promotion for extraordinary performance to motivate the worker. In order for these implicit incentives to be effective, it is essential that the worker judges remaining with his current employer to be more attractive than his next best alternative. In other words: the prospect of keeping his job (or getting promoted) must provide the worker with a non-competitive rent. The literature on conditional contract renewals in repeated interactions between employers and workers shows that such implicit incentives can endogenously enforce the provision of non-minimal effort under incomplete employment contracts (MacLeod & Malcomson 1989; MacLeod & Malcomson 1993; MacLeod & Malcomson 1998; Shapiro & Stiglitz 1984). The basic mechanism in all these models is that the employer provides incentives to perform by conditioning the renewal (or the terms) of the employment contract on the worker's past performance.
3.2 The Role of Fairness Concerns under Incomplete Relational Contracting In the ultimatum game, concerns for fairness and reciprocity cause rejections of unfair offers and induce proposers to make fair offers. In the context of contractually incomplete employment relations such concerns are also likely to affect workers' effort choices. The reason is that the preference for fairness or reciprocity induces workers to choose a higher effort to raise the firm's payoff in response to wage increases (Akerlof 1982; Akerlof & Yellen 1990; Benjamin 2005). However, as we discussed in section 2, there is considerable individual heterogeneity in the preference for fairness and reciprocity, and a significant share of subjects often behaves in a completely selfish manner. As a consequence, the one-shot effects of fairness may not be strong enough to make the provision of non-competitive wage premia profitable.
The prevalence of repeated interactions in relational labor contracts, however, greatly amplifies the possibilities for fairness concerns to play a decisive role. This holds even if employers and employees know that they will only interact for a limited number of periods, i.e., even in cases of very limited time horizons. In a finitely repeated context, the presence of even a small fraction of fair-minded workers may also motivate purely selfish workers to exert effort in response to receiving a wage rent (Brown et al 2004; 2008; Kreps et al 1982).
The intuition is that selfish workers have an incentive to exert effort in order to maintain a good reputation and make the employer believe that they are (at least potentially) fair-minded.
Such a reputation is valuable for selfish workers because finite repetition implies that the employer only pays non-competitive rents to workers who have not yet been identified as selfish. The firm anticipates that a selfish worker will always shirk in the final period of the interaction, which unravels all incentives to pay a rent in any period if a worker is known to be selfish. Fair-minded workers, in contrast, exert effort whenever they are paid a fair wage rent. Thus, if the belief about an individual worker's fair-mindedness is high enough, the employer will be willing to pay a high wage even in the last period. Note that the employer's willingness to pay a rent to fair workers disciplines the selfish worker because if the worker shirks he will be identified as a selfish type which, in turn, implies that he will not be paid a rent in the future.
It is important to stress that the existence of reputation equilibria in which all workers – selfish and fair-minded – put forward high effort despite the absence of explicit incentives does not require a large share of fair-minded workers. It is perfectly possible for the fraction of fair-minded individuals to be so small that firms would optimally refrain from paying noncompetitive wage premia in one-shot interactions, while finding it profitable to pay such premia in the reputation equilibria of finitely repeated interactions.
It is sometimes argued that the assumption of the existence of fair-minded individuals in infinitely repeated games is not needed to prove the existence of equilibria with noncompetitive wage rents.6 Even if all people are selfish, non-competitive rents can be equilibrium phenomena, provided that interactions are infinitely repeated. While, this argument is true, it is not important for our purposes because the existence of fair-minded subjects is an empirical fact. Physicists also don’t argue that we can explain sunset and sundown by assuming that the sun orbits around the earth although this incorrect assumption can provide a superficially plausible explanation for these phenomena. In addition, as the next section shows, it is an empirical fact that selfish subjects mimic fair-minded behaviors in repeated interactions (see e.g. Fehr, Brown, and Zehnder 2009). Moreover, because there are The papers by Shapiro and Stiglizt (1984) and Malcomson and MacLeod (1989, 1998) demonstrate the existence of such equilibria.
fair subjects, the range of parameter values (e.g. discount factors) for which equilibria with non-competitive wage premia exist is greatly expanded. Finally, fairness concerns may play a role as an equilibrium selection device, as argued in MacLeod and Malcomson (1998)7 and judgments regarding the fairness of performance evaluations are important whenever performance cannot be objectively assessed (MacLeod 2003).
4 Motivating Workers: Evidence on the Role of Fairness and Reputation In this section, we discuss the empirical role of fairness preferences for the motivation problem in contractually incomplete employment relationships. The cleanest evidence comes from laboratory experiments which implement the essential strategic features of the interaction between employer and worker. In interpreting the lab studies, there is a natural concern that the effects of fairness identified in the lab may not carry over to the labor markets outside the laboratory (Levitt & List 2007). However, we can allay these doubts by illustrating that the laboratory results are consistent with the findings in field experiments which implement explicit wage manipulations in real-life work environments and other field evidence from instances when firms changed the conditions in employment relationships.
In the first part of the section, we review evidence on the impact of fairness in one-shot encounters between employers and workers. Since reputational considerations cannot play a role in these situations, this evidence reveals the extent to which fairness motives alone help to overcome incentive problems in employment relationships. While there is clear-cut evidence that a non-negligible number of individuals are fair-minded and respond to higher wages by exerting more effort, the evidence also shows that many people are quite selfish and fairness concerns alone may not be sufficient to overcome the motivation problem in Another argument that is sometimes put forward is that because many interactions in real life are repeated, people apply rules of thumb or habits of fair behavior in one-shot interactions. However, while it is true that many interactions are repeated, people in modern societies also face many one-time encounters. Moreover, for our account of the labor market the origin of fairness preferences is not important. Whether many of us behave fairly because our parents imputed these values into our brains, or whether genes favor fair behaviors, is not important for our purposes. We have to take such behaviors seriously regardless of their evolutionary or educational origins.
incomplete employment contracts. In fact, if fairness concerns are the only force that drives effort above non-minimal levels, there are in general large unexploited efficiency gains.
In the second part, we discuss how labor market outcomes change if employer and worker have the option of interacting repeatedly with each other. We show that reputational incentives in endogenously formed long-term relationships greatly magnify the impact of fairness on performance. In fact, the data indicate that the interaction between fairness and reputation effects can be sufficiently strong to sustain high levels of efficiency, even when the
impact of fairness preferences alone is very weak.
4.1 Fairness in Spot Interactions A well suited game for studying the impact of wage rents on effort is the so-called giftexchange game, which was introduced by Fehr et al (1993). The gift-exchange game is a twoplayer game that captures the basic strategic features of an incomplete employment contract.
The structure of the game is as follows: the employer offers a wage to the worker and requests a certain level of effort. While the wage payment is binding, the requested effort level is not enforceable. The worker can either accept or reject the employer's offer. In case of acceptance, the worker can choose his actual effort level. The worker can choose any effort level and is not restricted by the employer's request. If the worker rejects the offer, he receives an unemployment benefit. The employer's profit is equal to the returns generated by the worker's effort minus the wage payment. The worker's payoff, in turn, is calculated as the wage minus the cost of effort. The parameters are usually chosen in such a way that the efficient outcome is achieved when the worker chooses the maximal effort level.
Since providing effort is costly, the self-interest model predicts that the worker always provides the smallest possible effort. However, if fairness considerations matter and workers are willing to reciprocate a generous wage with higher effort, it may be profitable for employers to offer wages that exceed the worker's outside-option.
A typical example for a one-shot gift exchange experiment is the baseline treatment in Brown, Falk and Fehr (2004). Each employer can at most hire one worker in each period and each worker can have in maximum one job. The matching of workers and firms occurs in a simple labor market with an excess supply of workers. A match occurs when a worker accepts a firm's offer. Although the market runs for several periods, reputation formation cannot play a role, because employers cannot distinguish workers from each other when they make wage offers. The results confirm that higher wage offers by firms, on average, induce workers to provide more effort. However, while effort is significantly higher than the minimal effort predicted by self-interest, it is also far below the efficient level8. The reason for the limited impact of fairness on performance is found in the huge inter-individual differences across subjects. Although there are the fair-minded workers, there is also a substantial share of workers who make mostly selfish choices. The relationship between wages and effort is steep enough to render wage offers above the worker's outside option profitable, but the presence of selfish agents restrains many principals from offering wage rents which would be high enough to induce efficient effort levels from fair-minded workers.
The finding that the impact of fairness alone on market performance is positive but small is very robust and has also been confirmed in a number of other laboratory studies using students as participants (Charness 2004; Charness et al 2004; Falk & Gaechter 2002; Fehr et al 1993; Hannan et al 2002). All report evidence from one-shot gift-exchange experiments and find results very similar to those reported above: wages and effort are always positively correlated, but the realized effort level is far from efficient. Fehr et al (1998) and List (2006) replicate the gift-exchange findings in laboratory experiments with non-student subject pools (soldiers and sportscard enthusiasts, respectively) and find that the realized effort levels are similar to those reported in the previous literature.
In order to assess the external validity of the laboratory findings on the impact of fairness on performance in employment relationships, several recent studies report evidence from field Workers can choose effort levels between 1 and 10 where 10 is the efficient effort level. The realized average effort level is about 3.
experiments on gift-exchange. These studies aim to test the proposition that higher wages are perceived as fairer and consequently elicit higher effort in real-life work environments.