First published in 1979. This thesis describes the theoretical impact on labour markets of a process of adverse selection similar to that described in outline by George Arthur Akerlof. It concerns the information conveyed to potential employers by the fact that any new worker, except for one just entering the labour force, has either left or is prepared to leave his latest Job. If an employer is able to identify his good workers more accurately than the market at large and is generally successful in retaining them, then the group of workers leaving him will contain a disproportionately small number of good ones. For similar reasons this pool should also contain an unusually large number of bad workers who have been either flied or induced to quit. Thus, workers who change jobs should on average be less able ones. Since the market failures that result have potentially significant consequences in the labour market, this study is devoted to examining their influence on the structure of wages and job tenure, and on the operation and efficiency of labour markets. This title will be of great interest to students of economics and business studies.

We show that adverse selection in the labor market may generate negative assortative matching of workers and firms. In a model in which employers asymmetrically learn about the ability of their workers, high-productivity firms poach mediocre workers, whereas low-productivity firms retain high-ability workers. We show that this flipping property is caused by information asymmetry alone. Our model has a number of positive and normative predictions: External promotions are not an indication of high talent, within-job wage growth is higher in industries with more revenue dispersion, and non-compete clauses are inefficient in industries with significant firm heterogeneity.

The main objective of this study is the application of an adverse selection model to verify the existence of discrimination in a competitive labor market caused by asymmetric information. The most important result obtained is when a group of workers with different productivities earn the same wage characterizing discrimination.

We show that in labor market models with adverse selection, otherwise observationally equivalent workers will experience less wage growth following a period in which they change jobs than following a period in which they do not. We find little or no evidence to support this prediction. In most specifications the coefficient has the opposite sign, sometimes statistically significantly so. When consistent with the prediction, the estimated effects are small and statistically insignificant. We consistently reject large effects in the predicted direction. We argue informally that our results are also problematic for a broader class of models of competitive labor markets.