This paper examines the effects of the distribution of unemployment by duration on the level of unemployment. It explores one central as- sumption that is observed empirically: when the share of long-term (short-term) unemployed increases, the unemployment rate increases (decreases). By embodying this assumption in a standard equilibrium unemployment model we characterize the joint behavior of unemploy- ment and the distribution of unemployment by duration. In the theoretical part of the paper an extension of the standard equilibrium unemployment model is provided where the average job finding prob- ability depends on the distribution of unemployment by duration. In the empirical section we estimate a panel data model for 16 OECD countries to test the implications of the theoretical model. The main results are: First, in a steady state equilibrium the flow rates are larger, the larger the share of short-term unemployed. Second, out of steady state the unemployment rate increases (decreases) with the share of long-term (short-term) unemployed. Third, the larger the duration dependent job finding probabilities the larger the share of short term unemployed. Fourth, panel estimates for 16 OECD countries provide robust and significant evidence that an increase in the share of short- term (long-term) unemployment decreases (increases) the unemploy- ment rate. Therefore, the empirical evidence supports our hypotheses whereby (a) the reemployment probability decreases with an increasing spell of unemployment and (b) average exit rates have international differences. Hence, the different experiences with unemployment on both sides of the Atlantic can be explained by observed aggregated duration dependence and differences in the average exit rate.
Distribution of Unemployment by Duration Duration Dependence