Academic journal article Economic Inquiry

Do Employers Pay for Consistent Performance?: Evidence from the NBA

Academic journal article Economic Inquiry

Do Employers Pay for Consistent Performance?: Evidence from the NBA

Article excerpt


There are many economic models of labor contracting that feature job performance monitoring by employers when there is asymmetric information about worker behavior. These models focus on a variety of agency issues and generate implications about contract terms such as initial wages, wage-tenure profiles, termination rates, and involuntary unemployment.(1) It is typically assumed in these models that a worker's Marginal Revenue Product (MRP) is known at the time of hiring, but that costly monitoring is necessary to enforce contractual terms. This paper deals with another issue: suppose there is no agency problem, but both worker and firm are equally uninformed about MRP when contracts are signed and work begins.

How are initial wages and other contractual terms influenced by symmetric uncertainty about MRP? Contracts are often formulated on the basis of various provisional predictors of MRP such as grades in college, letters of reference from teachers and former employers, work experience and performance test scores. Typically, employers expend resources to acquire such information at hiring and continue to engage in costly performance evaluation afterward. Under such conditions the purpose of monitoring is discovery, not contract enforcement. Information acquired from on-the-job monitoring is used to make decisions about adjustment of contractual terms, job assignment and retention. For example, tenure-track college professors are often evaluated yearly and must submit information about various aspects of on-the-job performance such as student evaluations of teaching, research papers completed, committee assignments, etc. On-the-job monitoring of workers is costly and is likely to affect the worker's pay. In fact, a worker may agree to pay for his monitoring if the information revealed by monitoring has alternative uses elsewhere in the labor market.

In this paper we develop and test a model of a firm facing costly on-the-job monitoring when there is symmetric information and we develop a number of novel implications for initial salaries. The central feature of our model is that monitoring can be costly when a worker's performance over time is variable. Notably, the model predicts that there will be an unambiguous inverse relationship between variance of MRP and initial wages. We obtain this and other predictions without resorting to assumptions about employers' and employees' utility functions that risk aversion models require. We regard our explanation of the relationship between variance and wages to be theoretically more robust than what the risk aversion literature would predict because that literature predicts an ambiguous relationship. For example, Stiglitz [1974, 232] uses a risk aversion model to demonstrate that wages and risk are inversely related only under very restricted conditions, specifically if the worker is more risk averse than the firm and if labor's elasticity of output does not exceed some threshold level. In contrast, our prediction is based on an information/transactions costs model with no restricted conditions.

We build upon earlier work by Bodvarsson [1987; 1989], who created a salary determination model in which a worker's output is variable over time. Bodvarsson characterized on-the-job monitoring as a sequential sampling experiment, where the employer periodically samples from the worker's MRP distribution and stops when she believes there is enough information to determine the true underlying MRP. A key implication of Bodvarsson's model is that expected monitoring costs are higher for workers with variable output and, therefore, if workers pay all or part of the cost of monitoring, salaries of such workers will be lower. In his earlier paper, this proposition was tested using a share contracting model and data on commercial fishing. In the later paper, a fixed wage model with educational signaling was developed, but never tested.

We extend Bodvarsson's fixed wage model and test it on an occupation that has many characteristics described by our model: professional basketball players in the National Basketball Association (NBA). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.