The contract selection effects of performance evaluation error and disclosure policy: An application in public accounting.

Abstract

When faced with various contract options, better workers self-select to those that pay according to performance (Salop and Salop 1976; Demski and Feltham 1978; Guasch and Weiss 1980; Chow 1983; Waller 1985; Waller and Chow 1985; Dillard and Fisher 1990). Similarly, this study suggests that public accounting firms, characterized by up-or-out contracts where workers are promoted or terminated based on the relative rank of their performance, design contracts that will attract better workers. This study hypothesizes that a low-skill worker's expected value of an up-or-out contract: (Hypothesis H1) is positively related to the error associated with employee performance measurement and (Hypothesis H2) is lower when the performance of terminated workers is disclosed versus when it is not. As a consequence of reducing the expected value, low-skill workers will tend to select other contracts. Hypothesis H1 is related to the analytical implications of tournament research by Nalebuff and Stiglitz 1980; Lazear and Rosen 1981; O'Keeffe, Viscusi, and Zeckhauser 1984; McLaughlin 1988). Two experiments were used to test the hypotheses. Both asked subjects to compare and state their preferences regarding two contracts; however, the first emphasized control while the second emphasized mundane realism. The data of both experiments strongly supported hypothesis H1. In the second experiment, this support was stronger yet for the responses of the subject's with the more extreme skill ratings. Results related to hypothesis 2 of the first experiment were weakly significant and of the second experiment were significant but opposite from the prediction.