This paper analyzes an occupation in which there is substantial uncertainty about individual performance and past performance is correlated with future outcomes. In equilibrium, only the young enter the occupation and only the successful stay on. The young are numerous and earn incomes well below what their current alternative o ffers; success is rare and rewarded highly. The implied distribution of earnings must have positive skewness.

This paper develops a simple stochastic job matching model and uses it to derive a set of testable restrictions on the conditional probability with which a worker will be observed to change jobs over time. The restrictions describe the manner in which this probability varies with observable characteristics-i-current wage, labour market experience, tenure on current job, and past mobility. Econometric methods are also discussed, and some illustrative calculations provided. 1.

THIS PAPER PRESENTS a simple model of the manner in which person-specific information on productive capabilities is put to use in the firm. The analysis is then employed to examine the impact of improved information quality on equilibrium output, wage rates, and degree of specialization. Much effort has been devoted to studying the process through which personspecific information is accumulated. Burdett-Mortensen [1], MacDonald [8], Prescott-Visscher [10], Hartog [3], and Johnson [5] examine the process of learning about person-specific parameters through investment of resources in activities that yield information. Jovanovic [6] analyses the more specific problem of inferring the quality of a particular job-worker match. Little attention has been given to the question of just how the firm utilizes this kind of information. For information to play an interesting role in production, two things are necessary. One is that workers be heterogeneous in a meaningful sense. That is, in the space of productive characteristics, workers must not all be simply scalar multiples of one another. Second, the firm must have some choice about the kind of activities in which workers are engaged. If either of these conditions fails, the optimal assignment of workers is not a problem.2

between schooling and wages: Schooling raises wages. The standard empirical questions of when are wages raised and by how much have gone unaddressed. The purpose of this paper is to demonstrate that the conventional efficiency units model of human capital accumulation provides answers to these questions. The model deals with investment both in school and on the job. The existence of post-schooling investment implies a path of wages that rises over time. Proposition 1 is that the marginal impact of schooling on the log of wages at each point in time is a constant equal to the interest rate if and only if the human capital production function is locally unit elastic in accumulated stocks of capital. A constant marginal effect on log wages is what is usually assumed in empirical work. Further, extrapolating from a model with no post-schooling investment, the constant effect is expected to equal the interest rate. Proposition 1 indicates that this assumption severely restricts the underlying structure. Propositions 2 and 3 deal with intertemporal variation in the marginal effect of schooling on wages. For example, if the output elasticity of accumulated stocks in the human capital production function falls short of one, the marginal impact of schooling on log wages is shown to decline over time. Further, under a reasonable additional assumption, the marginal impact of schooling on the level of wages rises over time. The propositions arise from the fact that wealth maximization involves maximization of an appropriately discounted flow of rents. Wages at a point in time provide information on the current flow of rents. The relationship between wealth maximization and the implied optimal pattern of flow rents yields the results. The model is laid out in Section 2. Section 3 deals with optimal schooling choice. Propositions 1-3 are presented in Section 4. In Section 5 the results are employed to discuss several stylized facts in the empirical literature on the wage-schooling relation. Proofs of the propositions are straightforward and are therefore presented in an appendix.

The paper provides analysis of an occupation in which there is substantial uncertainty about individual performance, and past performance is correlated with future outcomes. In equilibrium, only the young enter the occupation and only the successful stay on. The young are numerous and earn incomes well below what their current alternative offers; success is rare and rewarded highly. The implied distribution of earnings must have positive skewness.

Heterogeneity on both sides of the labor market implies that the correct matching of individuals to firms is of importance. If there is uncertainty about individual productive characteristics, there are both private and social returns to activities that generate information facilitating the assortative matching process. A model of individual investment in information is presented and analyzed. A key assumption is that there are no individuals who have an absolute advantage in all jobs. Under this assumption, all individuals view more accurate information as beneficial and therefore invest in its production. Comparative statics are derived and the model is compared to human capital and signaling-screening models.

Heterogeneity on both sides of the labor market implies that the correct matching of individuals to firms is of importance. If there is uncertainty about individual productive characteristics, there are both private and social returns to activities that generate information facilitating the assortative matching process. A model of individual investment in information is presented and analyzed. A key assumption is that there are no individuals who have an absolute advantage in all jobs. Under this assumption, all individuals view more accurate information as beneficial and therefore invest in its production. Comparative statics are derived and the model is compared to human capital and signaling-screening models.

The familiar model of free-entry, competitive equilibrium has long played a central role in applied analysis of product markets. Among its stylizations is the restriction that firms produce a single output. For many situations, however, it is necessary to relax this assumption. While models permitting multiproduct firms exist, so far there is no framework that begins to rival the singleproduct analysis in terms of the ease with which it may be manipulated and extended to deal with specific applications. This paper develops such a model.' Specifically, in Section I a two-good model is set out that parallels the classic singleproduct analysis very closely. The sole difference is that alongside the single-product (specialized) technologies, which would usually be permitted, a multiproduct (diversified) technology is available. The model's competitive equilibrium is then characterized, and it is shown that this equilibrium may take one of three forms. Obviously, if diversification offers large cost advantages, no specialized firms can operate in equilibrium, and conversely if there are sizable disadvantages. The only other possibility is that diversified firms and exactly one type of specialized firm operate contiguously. Section II shows that the model is easy to manipulate, extends straightforwardly, and simple as it is, offers some new propositions. This demonstration involves examining the predictions the model offers in both its most general form and several extensions. To illustrate, basic features of equilibrium in the standard single-product environment are that price is determined by the cost of production with all operating firms producing the same level of output. Also, demand variation has no effect on either price or the actions of these firms. In contrast, in any equilibrium *Centre for Decision Sciences and Econometrics, Social Science Centre, University of Western Ontario, London, Ontario, N6A 5C2. Comments from David Donaldson, Ignatius Horstmann, Boyan Jovanovic, Peter Lloyd, Michael Parkin, Charles Plott, Edward Prescott, Sherwin Rosen, Christopher Robinson, Hugo Sonnenschein, and the referees are gratefully acknowledged. 'Research allowing multiproduct firms comes in a variety of forms. Early work by Roy G. D. Allen, 1938; John Hicks, 1939; Paul Samuelson, 1947; as well as more recent efforts by Keith Laitinen, 1980, analyzed in detail the isolated behavior of firms having access to an m-input/n-output production technology. Also, the Arrow-Debreu-McKenzie general-equilibrium model allows each producer a distinct production set, so that firms might choose to produce many goods. More recently, the literature (surveyed by Elizabeth Bailey and Ann Friedlander, 1982) analyzes a setting in which firms produce more than one good. Of all the contestability material, the work of William Baumol et al., 1982, Ch. 9, is the most closely related to the present analysis. Therein firms are permitted to choose a set of goods to produce, and a condition is provided that is necessary and sufficient for the (otherwise exogenously imposed) symmetric outcome to be supported in equilibrium. That this condition is indeed a relevant restriction is shown by means of a two-good numerical example in which the condition fails and the equilibrium is asymmetric. Finally, there is what might be termed the where there is sawdust there may be ' pressed logs' approach, dating back at least to Alfred Marshall, 1920, pp. 321-22, in which joint products are the result of unstructured technological complementarities.