Organizational sociologists have found that earnings and promotions in the later stages of careers are strongly correlated with those in the early stages, even when controlling for the effects of observable characteristics such as education. In Discussion Paper No. 637, Research Fellow Margaret Meyer considers the design of sequential promotion contests and shows that these observed correlations can result from an organization's optimal adaptation to the constraints on its available information on employee performance. Many organizations keep new employees in junior positions for several `periods' before making major job assignment decisions, which depend critically on their abilities to evaluate employees' performance on an interim basis and differentiate their subsequent treatment according to its results.

Meyer considers a sequence of two contests for a pair of employees and considers whether and how an organization may `bias' the second contest. They may artificially increase one employee's relative probability of success by assigning them tasks of unequal difficulty, placing them in different work environments, providing them with different amounts of training or supervision or different equipment.

One potential source of the documented correlations between early and later success is persistent differences in ability that are not captured by the control variables. Meyer cites her previous work on the design of contests to learn about such differences and thereby make efficient promotion decisions, which found that the average quality of promotion decisions is highest when the second contest is biased in favour of the winner of the first, which will reinforce the existing correlation between early and later success.

She then shows that an organization may wish to induce such a correlation even when it knows that employees' abilities are identical, in order to reduce the costs of monitoring how hard employees work. In this setting with `moral hazard', the organization designs the sequence of contests to minimize the monetary costs of incentives for its desired levels of effort; it will again bias the second contest in favour of the initial winner. Asymmetric treatment reduces employees' incentives for effort in the second contest, but this effect is outweighed, if they are risk averse, by their increased incentives in the first contest. Transitory differences in employees' early performance that reflect only transitory random factors and not their ability can therefore have persistent effects on their careers.

Biased Contests and Moral Hazard: Implications for Career Profiles
Margaret A Meyer

Discussion Paper No. 637, February 1992 (AM)