پرداخت عملکرد، نگرش های ریسک و رضایت شغلی

We present a model in which workers with greater ability and greater risk tolerance move into performance pay jobs to capture rents and contrast it with the classic agency model. Estimates from the German Socio-Economic Panel confirm testable implications drawn from our model. First, before controlling for earnings, workers in performance pay jobs have higher job satisfaction, a proxy for on-the-job utility. Second, after controlling for earnings, workers in jobs with performance pay have the same job satisfaction as those not in such jobs. Third, those workers in performance pay jobs who have greater risk tolerance routinely report greater job satisfaction. While these findings support models in which workers capture rent, they would not be suggested by the classic agency model.

Performance pay has been shown to increase worker effort, earnings and risk (Booth and Frank, 1999, Lazear, 2000, Lemieux et al., 2009, Oettinger, 2001, Paarsch and Shearer, 2000, Parent, 1999 and Shearer, 2004). Yet, distinguishing between competing models of performance pay depends on identifying the influence of performance pay on job satisfaction — something that has not been widely recognized in the existing literature. In the classic agency model of performance pay, the principal has full bargaining power. Facing a reservation utility constraint, she sets the earnings contract to just offset effort and risk costs (e.g., Holmstrom and Milgrom, 1987 and Gibbons, 1998). This exact offset implies that performance pay should not influence job satisfaction. In contrast, Lazear (1986) models performance pay as a self-sorting process by workers of heterogeneous abilities. Firms face a zero profit market constraint allowing workers to capture the rent associated with their ability by sorting into performance pay. The rent capture implies that performance pay should generate greater job satisfaction. Thus, a careful examination of job satisfaction may suggest the extent of rent capture and so the relative importance of the classic agency and sorting models of performance pay. We extend the sorting models of Lazear, 1986 and Booth and Frank, 1999 to account for heterogeneity in both ability and risk attitudes. While standard agency analyses are limited to risk averse workers, our model allows risk neutral and risk loving workers. Our sorting model predicts that the more able and more risk tolerant sort themselves into performance pay schemes in order to capture rents and that this increases their on-the-job utility. Moreover, the two critical sorting dimensions interact. Capturing the rent on ability requires sorting into the performance pay sector and among those sorting into performance pay, workers with the greatest risk tolerance will receive the greatest on-the-job utility.1 Among those remaining in the time rate sector, there should be no relationship between risk tolerance and utility. Finally, the model presents ambiguous predictions as to whether or not workers on performance pay will continue to receive greater utility once the positive influence of higher earnings is removed. However, the positive relationship between risk tolerance and utility for those on performance pay should remain independent of the influence of higher earnings. The empirical testing exploits a unique question on risk tolerance in the German Socio-Economic Panel (GSOEP). Using job satisfaction as an indicator for on-the-job utility, we confirm each of the predictions of our sorting model. Performance pay emerges as a positive determinant of job satisfaction. Among those receiving performance pay, greater risk tolerance is associated with greater satisfaction whereas risk tolerance plays no role in job satisfaction among those on time rates. Finally, controlling for earnings causes the coefficient on performance pay to move to statistical insignificance while it does not change the link between risk tolerance and job satisfaction. These results hold in robustness checks that control for an expanded range of ability correlates and that account for worker specific fixed effects generated from a panel earnings regression. They also persist in explorative instrumental variable estimates accounting for the possible endogeneity of performance pay. While these findings fit the predictions of our sorting model, they are not easily reconciled with the standard agency model. As the principal sets earnings just to offset effort and risk costs, the classical agency model predicts that performance pay should not influence job satisfaction and that after earnings are controlled for, job satisfaction should be lower in the performance pay sector than in the time rate sector. Moreover, the typical agency model predicts that there should be no relationship between risk attitude and job satisfaction in the performance pay sector. If a worker has a lower degree of risk aversion, the employer reduces the earnings premium that compensates for the disutility of bearing an income risk. The next section sets the context by briefly examining past research and isolating our area of interest and value added. The third section details our extension of the sorting model in which workers sort on both ability and risk preferences. It draws predictions and testable hypotheses. The fourth section presents our data and variables while the fifth section presents the estimation results. A sixth section discusses robustness checks and a final section draws conclusions.

This study uses job satisfaction as a measure of on-the-job utility in order to contrast a sorting model with rent capture from the classic agency model. In the latter, the workers retain no rents. The additional earnings they receive from performance pay exactly offsets the utility lost from being subject to earnings risk and from exerting effort. Thus, workers should receive the same utility in each sector and after controlling for earnings, those receiving performance pay should have lower utility. Instead, our empirical results suggest higher job satisfaction for those receiving performance pay both in the simple comparisons and the parsimonious regressions. Once earnings, and ultimately many other controls, are included, this advantage becomes insignificantly different from zero. In none of our estimations, can we find lower job satisfaction for those receiving performance pay. These results accord with our sorting model in which the more able and more risk tolerant capture rents. We also isolate the role of risk tolerance in the sorting model. The model predicts that it should matter only among those receiving performance pay and should do so with or without controlling for earnings. Indeed, we confirm this prediction using the unique risk tolerance variable. Greater risk tolerance is a strong positive determinant of job satisfaction among those receiving performance pay but plays no role among those not receiving performance pay. We recognize that contrasting the classic agency model with the sorting model leaves excluded alternative models that could predict a relationship between performance pay and job satisfaction. First, the standard agency model can be amended in various ways to suggest that performance pay workers retain a rent. Perhaps first among these amendments is the limited liability assumption. Interestingly, performance pay in face of a limited liability constraint has implications similar to those analyzed in the efficiency wage literature (Foster and Wan, 1984 and Laffont and Martimort, 2002: pp. 174–175; Jirjahn, 2006). Workers queue for jobs in which they can receive a rent while employers will be reluctant to invest in creating such jobs. On the other hand, other theories have suggested that performance pay should be associated with lower utility. Thus, workers may care not only about their own earnings but the implications of the greater earnings disparity associated with performance pay (Kennedy, 1995). This disparity can be sufficient to lower both morale and productivity. Alternatively, McCausland et al. (2005) suggest that workers may see performance pay as form of control and that the resulting loss of autonomy lowers utility. While lower moral and loss of autonomy may happen in some circumstances, our results find no support for the general contention that performance pay is associated with lower job satisfaction but instead that the higher earnings bring higher job satisfaction to those on performance pay. Again, even holding earnings constant, performance pay is associated with roughly similar job satisfaction as other forms of payment.