Performance Evaluations and Efficient Sorting
Much of the production in firms takes place over time. This paper seeks to understand the value of interim performance information on long projects. In particular, the model explores the sorting effects of performance evaluations. Conducting an interim performance evaluation increases efficiency by providing the option to end projects with low early returns. The main result: It is efficient to allocate more resources towards the end of a project. This result holds under a variety of scenarios: when the worker has unknown ability, when the outside options vary with output, and even under an agency framework with a risk-averse agent.
The production of goods and services takes time. It takes time to build cars, write software, develop drugs, formulate strategy, market products, audit clients, and panies. In fact, many firms anize their production in the form of projects that run for weeks, months, or even years at a time before the firm can sell the finished product in a market. For example, software development runs through design, implementation, and testing stages before final release. This paper examines the problem of performance evaluation before a project is finished. The main idea is that performance evaluations generate efficient sorting. They provide the firm with the option of ending projects with low early returns. A performance evaluation conducted at the interim stage sorts employees into two groups: stay or quit. It is efficient to quit if the early returns are weak, and to stay otherwise. If the agent’s early output is low, it is in the interest of both the firm and the agent to discontinue work and collect their respective outside options. In this sense, the performance evaluation assigns an agent to his most efficient use. The role of the evaluation, therefore, is to perform this sorting.
Interim performance evaluations change the efficient allocation of re-sources over time in surprising ways. The main result shows that it is efficient to assign more resources to the later stages of the project. The evaluation creates the possibility of termination after the early stage, which reduces the marginal return from first-stage effort, and so the agent shades his effort downward in the early stage. Once the agent advances, the possibility of termination vanishes and the marginal return to effort rises, so he works harder. In sum, it is efficient to stay if the early returns are high, and for those that stay, it is efficient to work harder.
The usual analysis of performance evaluations takes place in agency models that span an enormous literature in accounting, economics, and finance (see Baiman [1982, 1990], Indjejikian , Lambert , Prendergast for reviews). Agency theory has enjoyed wide theoretical attention over the last 20 years, and has dramatically advanced our understanding of perf