Portal de Periódicos da CAPES

Information, Screening and Human Capital

1976; American Economic Association; Volume: 66; Issue: 2 Linguagem: Inglês

1944-7981

John G. Riley,

Economic Policies and Impacts

Given the large and growing body of research into the nature and extent of human investment decisions, it is somewhat surprising that until the recent work of Michael Spence and Joseph Stiglitz there has been little discussion of the information transmission process. Certainly in all the theoretical modeling of human capital accumulation it has been implicitly assumed that throughout the life cycle employers are aware of each individual's marginal value product.' That is, traditional human capital theory has included the assumption that information costs are negligible. However, it is by no means clear that a firm can evaluate cheaply the productivity of an individual worker, especially when the nature of the job is nonspecific (e.g., the management trainee). Plausibly information about an individual's value often unfolds only slowly with time on the job. Plausibly also, the costs associated with placing an individual in a job for which he is ill-suited are far from negligible. If so, firms have a strong incentive to offer salary contracts in which earnings are contingent upon long-run performance. But such offers will only be completely successful in screening out the less productive if job seekers are either risk neutral, or have very tight prior probabilistic beliefs about their own lifetime productivity levels. Since it seems reasonable to reject both assumptions, contingent contracting of this type seems likely to be severely restricted. An extension of this argument, emphasized by Stiglitz, suggests that firms are unlikely to incur large expenditures for on-the-job evaluations of their employees' potential. First of all, risk aversion makes job seekers unwilling to bear these costs in the form of considerably lower initial salaries. Secondly, the possibility of raids by other firms makes each firm unwilling to itself bear the costs of identifying top talent. The question then arises as to whether firms might exploit information about an individual's general educational achievements at school and college in attempting to predict productivity on the job. Taking this to the extreme, might firms make initial job offers based entirely upon educational credentials which on average succeed in attracting workers of the desired productivity level? The key element in answering this question is the manner in which marginal costs of education vary across individuals. With everyone staying in school until the marginal increase in earnings resulting from additional education is just offset by marginal costs, those whose costs are lower will plan to accumulate higher credentials. Then if the marginal cost of an additional unit of education is highly negatively correlated with productivity on the job, it will be the more skilled individuals who accumulate more education. But it is surely reasonable to argue that the more productive workers are also, on average, the faster learners hence those with lower opportunity costs. Intuitively then, it seems qute possible that educational screening by firms might result in * University of California, Los Angeles. The helpful comments of Sherwin Rosen, Michael Darby, Dennis de Tray, Jack Marshall and Finis Welch are gratefully acknowledged. 1For a brief summary of recent theoretical and empirical advances, see F. Welch.