Much of the recent literature on unemployment in segmented labour markets is based on the Harris-Todaro theory, which assumes that the unemployed are better able than secondary-sector workers to search for primary-sector jobs and predicts that unemployment tends to the level at which the expected returns from secondary- sector employment and from unemployment are equalized. Such explanations of the bulk of unemployment seem implausible, however, for several reasons. Primary- and secondary-sector jobs generally exist in close proximity; the unemployed appear to make little use of their greater time to search; there is little correlation between search intensity and unemployment duration; and on-the-job search is more common than an intervening period of unemployment for workers changing jobs, and indeed more common in the secondary than the primary sector.

In Discussion Paper No. 523, Assar Lindbeck and Research Fellow Dennis Snower propose three alternative explanations of unemployment with an imperfectly competitive primary sector (combining salient characteristics of the efficiency wage, insider-outsider and bargaining theories) and a perfectly competitive secondary sector. They account for unemployment and labour market segmentation in terms of the heterogeneity of workers' preferences, productivities and endowments. In the first model these are exogenous, and unemployment and secondary-sector employment may coexist if the unemployed are those who find work in the secondary sector particularly onerous or whose secondary- sector productivity is low. In their second model, workers that have previously been employed in the primary sector have a sense of status that makes secondary-sector work more onerous relative to unemployment, so they are more likely to accept unemployment on losing primary-sector jobs. In the third model, workers' productivities in the secondary sector depend on the length of their job tenure and workers without previous secondary- sector employment have lower productivity and hence a stronger incentive to accept the alternative of unemployment.

In most efficiency wage, insider-outsider and union models, a rise in primary-sector employment must reduce unemployment, since there is no secondary sector; while in the Harris-Todaro model, a rise in primary-sector employment increases unemployment by raising the expected return from unemployment relative to work in the secondary sector. In Lindbeck and Snower's models, in contrast, this relationship may be positive, zero or negative, depending on the source of labour market segmentation.

Segmented Labor Markets and Unemployment
Assar Lindbeck and Dennis J Snower

Discussion Paper No. 523, April 1991 (AM)