Journal of Political Economy
1925 1942 1964 1973 1974 1978 1979 1980 1981 1983 1984 1985 1986 1987 1989 1991 1994 1995 1996 1997 1998 1999 2000 2001 2002 2004 2007
Опубликовано на портале: 24-01-2007Andrew Weiss Journal of Political Economy. 1983. Vol. 91. No. 3. P. 420-442.
This paper presents a sorting model of education in with individuals are tested in school. By assuming that higher-ability individuals are more likely to succeed on a given test one can construct a sorting model of education that does not hinge on the more able having lower nonpecuniary costs of schooling. Nash equilibria always exist in this model (even with a continuum of types of individuals); however, some are "unreasonable." To eliminate these unreasonable Nash equilibria, more restrictive definitions of equilibrium are proposed. I also show that when schooling affects productivity - and therefore a worker's probability of passing the test--a sorting equilibrium may be characterized by too little investment in education. This paper extends the important work of Spence (1974), Stiglitz (1975), and Riley (1979a, 1979b) on sorting theories of education by modeling the educational choices of individuals in game-theoretic terms and making two assumptions: (1) individuals are not perfectly informed about their own productivity, and (2) individuals are tested upon their completion of schooling. I also combine the sorting and human capital analyses by allowing education to increase productivity and show that if education increases the productivity of workers as well as enabling the more able workers to sort themselves, these sorting effects may lead to underinvestment in education. This result contradicts the main normative result of screening models of education: If skills are hierarchical, so that if Joe is more productive than Jim at any job he is more productive at all jobs, there is overinvestment in schooling as workers use education to signal their abilities. Although in a pure sorting model education always leads to overinvestment in education while human capital models lead to optimal investment, when both effects are modeled there may be too little investment in education. This result holds even if the more able learn faster in school.