عملکرد پرداخت، مرتب سازی و انگیزش اجتماعی

ariable pay links pay and performance but may also help firms to attract more productive employees. Our experiment investigates the impact of performance-pay on both incentives and sorting and analyzes the influence of repeated interactions between firms and employees on these effects. We show that (i) the opportunity to switch from a fixed wage to variable pay scheme increases the average effort level and its variance, and (ii) high skill employees concentrate under the variable pay scheme; (iii) however, in repeated interactions, efficiency wages reduce the attraction of performance-pay. Social motivation and reputation influence both the provision of incentives and their sorting effect

In the last two decades the prevalence of different forms of variable pay has increased substantially in many firms and countries. Variable pay schemes are typically motivated by the incentive effect they exert on the employees by linking pay to performance. It is, however, frequently forgotten that performance-related pay may also help firms to attract the most productive employees and weed out the less productive ones. In contrast, fixed pay schemes not only have less incentive power, but they are also less able to sort employees since they are usually implemented when it is difficult and costly to measure individual performance. If this self-sorting effect is not accounted for, the higher efficiency observed when comparing a piece-rate compensation scheme relative to an hourly wage scheme may be unduly attributed to the incentive effect of the variable wage; see Lazear (2005). The sorting effect arises for two reasons. Highly productive employees will prefer performance-pay to hourly or monthly pay in as much as they know that their productivity is higher than that of the other employees in the firm. Thus, performance-pay enables them to increase their income from work, whereas the less productive employees will tend to quit or avoid joining firms that use performance-pay schemes because such arrangements are not attractive to them. A first aim of this paper is to report the results of a laboratory experiment designed to investigate the incentive and the sorting effects of performance-pay by studying how the employees self-select their incentive scheme and choose their level of effort. This is motivated by the fact that due to dearth of natural data, the distinction between the selection and incentive effects of pay schemes cannot be easily documented empirically. As regards the incentive effect of performance-pay, one strand of the literature emphasizes the repetitive game nature of employment contracts (see Malcomson, 1999 for a survey), labor market conditions (MacLeod and Parent, 1999), or competition in the product market (Bertrand, 2004). Thus far, only a few studies have tried to identify a sorting effect of performance-pay distinct from its incentive effect.2 The best way to isolate the two effects is to use personnel files of firms that have changed their compensation schemes. This has been done in a widely cited econometric case-study of the Safelite Company by Lazear, 2000 and Lazear, 2005. Barro and Beaulieu (2003) study a large hospital company that switched from a salary to a profit-sharing plan compensation, and they also find increases in output as well as selection effects. However, following the strategy in Lazear (2000) is, in most cases, difficult as the switch from one compensation scheme to another can only rarely be considered as exogenous and finding good instruments for the choice of compensation scheme is extremely difficult. An additional problem is finding data on individual performance under both fixed wage and variable wage schemes.3 Experimental methods may help to circumvent some of these difficulties by guaranteeing the exogeneity of the introduction of performance-pay. The relationship between risk attitude and sorting has been recently documented by means of field experiments (Bellemare and Shearer, 2006) and laboratory experiments in which employees can choose between various payment schemes (Dohmen and Falk, 2006, Cadsby et al., 2007 and Eriksson et al., 2008). The laboratory experiment presented in this paper has been designed to test the key predictions derived from Lazear's (2000) model in a framework in which effort is observable but not contractible and in which the risk attitude of employees cannot play any role. The observability of effort would of course influence the offer of the payment scheme by the employers, but here we are interested in the selection made by the employees. The predictions are the following: (i) a switch from a fixed to a variable wage increases the average level of effort, (ii) introducing the possibility for the employees to move to a variable wage scheme increases the variance in effort, (iii) the gain in productivity is due to both an incentive effect and a sorting effect, and (iv) the possibility to choose between a fixed wage and a variable pay leads to a segmentation of the labor market with high skill employees concentrated in the performance-pay firms and low skill employees populating the fixed-pay firms. Our experimental methodology provides an opportunity to observe the incentive and sorting effects when both a fixed wage and a variable reward system are available to the employees. It both allows us to control the environment (skill levels, structure of pay) and provides unbiased measures of the agents’ productivity and mobility. This experiment also accounts for the observation that employment relationships usually are long-term relationships. Repeated interactions frequently involve social motivation that could influence both the incentive and the sorting effects of pay schemes. In repeated games where reputation building may occur, the interaction of reciprocally motivated subjects and selfish subjects may lead to the enforcement of labor contracts even in the absence of a formal incentive system (Falk et al., 1999 and Fehr et al., 1997). In such a context, the introduction of explicit (positive or negative) incentives may even crowd out voluntary cooperation (Frey and Oberholzer-Gee, 1997, Gneezy and Rustichini, 2000, Houser et al., 2008 and Dickinson and Villeval, 2008). Therefore, we test the additional hypothesis that (v) when interactions are repeated, the difference between the fixed and the variable pay schemes in terms of incentive provision and sorting is weakened. Indeed, a generous (non-equilibrium) fixed wage might also be able to attract skilled employees and reciprocity might motivate them to work hard. Our experiment involves firms and workers with two possible skill levels who have an opportunity to contract and work under different pay schemes. It consists of two treatments, a “fixed wage” treatment and a “menu” treatment. Two conditions have been implemented, a “market” condition and a “partner matching” condition, which are intended to represent a spot labor market and a long-term employment relationship respectively. Under the market condition, we implement a fixed wage treatment in the first part of the sessions. Each firm posts its fixed wage offer. Each worker then successively chooses an offer and decides on her actual level of effort. In the second part of the sessions, we implement the menu treatment in which the firm offers both a fixed pay and a variable pay scheme in which the wage increases in the ex post level of effort. Under the partner matching condition, the only difference is that pairs are fixed throughout the session. Our findings support the hypotheses mentioned above. Under the market condition, when a variable pay scheme becomes available, the average effort as well as its variance increase substantially compared to the situation where a fixed wage is imposed. The increase arises mainly because the high skill employees select the variable pay scheme as they can improve their payoff by exerting a higher effort. As a consequence, the labor market becomes segmented with the low skill employees remaining in fixed pay firms and most high skill employees working under a variable pay scheme and putting forth high effort. In the repeated interaction game, however, and relative to the market condition, most firms offer higher fixed wages, workers exert a higher average effort under the fixed pay scheme, and the high skill employees are less inclined to switch to a variable pay scheme. Thus, repeated interactions affect contractual choices and alter the sorting effect of performance-pay. This suggests that efficiency wages may be a reason why firms do not adopt variable pay schemes. The remainder of the paper is organized as follows. Section 2 outlines the experimental design and procedures. The empirical results are presented in Section 3. Section 4 summarizes and concludes.

Economic theory of performance-pay schemes predicts that the switch from a fixed to a variable pay scheme should increase the average output per worker because of incentive effects. Moreover, if workers differ with respect to ability, the high skill employees should be more attracted by the performance-pay than the low skill employees since it allows them to receive a higher wage by exerting more effort. Our experimental evidence confirms the coexistence of the sorting and the incentive effects of payment schemes as in Lazear (2000). It shows that a switch from a homogenous environment to an environment in which both a fixed and a variable pay schemes coexist entails an increase in both average effort and the variance of effort. Effort only increases for the high skill employees because the variable pay scheme induces them to work harder. The low skill employees are not attracted by the variable pay scheme when firms offer non-minimum fixed wages. This observation suggests that there is a limit on the adoption of performance-pay despite its incentive effect. This adoption increases team heterogeneity within firms, with a widening wage gap that may generate conflicts. Consequently, variable pay schemes are less likely implemented in firms promoting teamwork. The comparison between our market and partner conditions shows that the employees who choose the variable pay scheme are not necessarily the most ambitious ones. Provided the rent offered is sufficiently high, in a repeated interaction, a non-negligible fraction of the high skill employees also opt for the fixed pay. This indicates that if a firm pays efficiency wages to attract reciprocal employees and to allow for less unequal payoffs, then the variable pay scheme becomes less attractive to the skilled employees relative to the fixed pay scheme. Thus the social motivation of the subjects influences not only the provision of incentives, but may also affect their incentive and sorting effects. Although our design does not leave much room for reciprocity due to the limited space of strategies, we nevertheless observe a smaller ability-based segmentation of employees when firms make generous offers and a smaller productivity gap when employees accept these efficiency wages and try to build a reputation of reciprocity. Therefore, we would expect that a design allowing for a larger scope for reciprocity would weaken this ability-based segmentation even further. An avenue for further research would be to look for a sorting effect based on social preferences and to analyze how social preferences and ability interact in the sorting process. This could contribute to explaining why we observe less segmentation of labor markets according to ability than expected by theory. Our results already suggest that we would expect to observe less performance-pay in long-term employment relationships. This, in turn, could imply that differences across countries with respect to the employment protection legislation may be accompanied by differences in the prevalence of performance-pay schemes.