Learning-by-doing and unemployment dynamics

This paper attempts to assess the impact of skill loss on the persistence of cyclical unemployment. The observations from the Current Population Survey and the Bureau of Labor Statistics suggest a countercyclical total unemployment that exhibits high persistence. The paper develops a model to identify the underlying market interactions that are critical in generating the observed behavior. A framework that features search frictions is developed in which the paper extends the learning-by-doing framework to consider endogenous skill loss by the unemployed, or a process of loss-of-learning-by-not-doing.

Endogenous skill acquisition and loss are introduced, in a framework of learning-by-doing. These interactions are captured in a dynamic stochastic general equilibrium model that features search frictions. In this framework, the unemployed choose their search intensity for occupations. Firms create vacancies, and a standard matching function matches workers with occupations. Workers accumulate skills through past work experience, or a process of learning-by-doing, similar to the one introduced by Chang et al. (2002). In their work, learning-by-doing is found to provide an important propagation mechanism in real business cycle models. Their framework, however, did not consider equilibrium unemployment. This paper extends their analysis to consider skill loss by the unemployed, or loss-of-learning-by-not-doing. Thus, this paper is considered the only one that integrates learning-by-doing in a framework of equilibrium unemployment. The extension allows the paper to succeed in capturing the observed unemployment persistence.

An adverse aggregate technological shock induces workers to reduce their search intensity and firms to reduce their creation of vacancies. As unemployment increases, workers lose their accumulated skills. The skill obsolescence causes a decline in the future marginal productivity of workers. The decline in productivity causes persistence in the cyclical downturn, and a delay in the recovery of the economy. This allows the model to capture the observed unemployment persistence. The model is calibrated and the impact of an aggregate shock is simulated. The impulse responses and the cross correlation coefficients show a pattern where the unemployment rate exhibits a level of persistence close to the one that is observed in the data.