The achievement of full employment or, as it is sometimes presented, the minimization of unemployment has been a major goal of public policy since the economic cataclysm of the 1930's. This goal reflects the implicit belief among policymakers that achieving full employment is the appropriate concern of a manpower policy responsive to the needs of individuals and society. This perception, along with the labor force concepts used to measure progress toward the full-employment objective, has its origin in the surroundings of the depression era and the Keynesian revolution. The events of this period, marked by mass unemployment and related economic hardship, and their conception in economic theory continue to shape contemporary economic policies and labor force concepts. What proved to be an adequate measure for one set of perceived problems may prove to be inadequate for another, necessitating a change in concepts or methods of measurement. In particular, the relevance of depression era policies and labor force concepts to the present is a question of major importance. Current surroundings have changed, with the growth of income transfer programs and multiple earner families weakening the link between unemployment and economic hardship. As the surroundings have changed, so has economic theory. Led by the resurgence of neoclassical theory and the development of neo-Marxist theories of segmentation, the perception of unemployment and its causes has changed over time. This paper traces the evolution of economic theory and events, and their impact upon labor force concepts. The relationship of current concepts of employment and unemployment to Keynesian theory and events of the depression era is described and the implications of post-Keynesian theories for these concepts explored. The argument is advanced that current labor force concepts lag behind contemporary economic theories and events. Some directions for change are suggested.

Introduction and summary, 304. — The Ashenfelter-Pencavel model, 305. — Testing the Ashenfelter-Pencavel model, 307.

T HE inferiority of southern black schools (especially rural schools) alleged by the Coleman Report (1966), coupled with the mass migration historically of southern blacks to northern cities, provides one potential explanation of the generally low returns to black education and of urban poverty in the nonSouth. Evidence from the 1960 Census suggests, however, that black migrants to the metropolitan North had higher incomes and less unemployment than blacks born there, even after controlling for differences in age, years of school completed, and a number of other variables (Masters, 1972). More recent evidence, confirming this pattern, from the 1967 Survey of Economic Opportunity (Weiss and Williamson, 1972) and the 1970 Census (Long and Heltman, 1974) also discounts the inferiority of southern black schools as an explanation of urban poverty in the non-South. In fact, the overall effect of a northern or even a large southern ghetto environment may be more harmful to black economic progress than a rural southern origin (Weiss and Williamson, 1972). In this paper we present results, using data from the National Longitudinal Surveys, that support the economic disadvantage of a nonsouthern ghetto environment for young black males. Controlling for differences in age, years of school completed, region and character of current residence, we find the mean earnings of young black males educated in the metropolitan non-South are substantially less than those of their peers educated in the rural South. We are unable to confirm this disadvantage for older black males, however. Several attitudinal and labor force characteristics of young blacks were examined to account for this pattern. The results suggest that a major problem in reducing black poverty lies in improving the environment of the nonsouthern ghetto. We also extend the analysis to whites in order to examine the rural-urban dimensions of environment and migration and their effect upon racial earnings differentials. The National Longitudinal Surveys, which provide the primary data for this paper, constitute a five-year longitudinal study of the labor market experiences of four subsets of the U.S. population: men 45 to 59 years of age, women 30 to 44 years of age, young men 14 to 24 years, and young women 14 to 24 years of age.1 For each of these cohorts a national probability sample of the noninstitutionalized civilian population was drawn by the Bureau of the Census. The present study is based upon data collected in the first round of interviews in 1966 with the two cohorts of men. Analysis is restricted to men whose current or last job reported in the survey week of 1966 was as a wage or salary earner. The self-employed are excluded to overcome the difficulty of separating income received as returns to physical capital from that received as returns to human capital. An additional universe restriction is included for the younger men's cohort to ensure they had been out of school for a minimum of 12 months. In section I we build upon earlier earningsfunctions studies by introducing variables which identify geographic origin of schooling. Section II explores the implications of our findReceived for publication January 28, 1974. Revision accepted for publication July 9, 1975. *This paper was prepared under a contract with the Manpower Administration, U.S. Department of Labor. We are indebted to Bennett Harrison, Ray Marshall, Herbert S. Parnes, and a helpful referee for comments on an earlier version of this manuscript. We especially thank Clarice Conger-Thompson, Gary Schoch, and Keith Stober for research and computational assistance. As is customary, the views expressed in this article are our own. I For a description of these surveys see Parnes (1972).