Unemployment Insurance and the Business Cycle

Abstract

We develop a numerically solved equilibrium model of the labor market to study the effect of unemployment insurance (UI) over the business cycle. This model combines sequential job search, optimal job offer, layoff, and recall decisions, an aggregate productivity cycle, and details of an actual (namely, the Canadian) UI system. Optimal worker (rm) behavior is characterized by a dynamic programming problem conditional upon beliefs about the behavior of rms (workers). In equilibrium beliefs are consistent with the optimal decisions of other agents. The equilibrium beliefs are found using a nested algorithm in which simulations of the economy are used to iterate on beliefs while resolving for optimal decisions. Some of the model's parameters are used to match simulated moments to data on labor market outcomes for young Canadian men. Simulations of recent changes to the UI system suggest that they will raise average unemployment rates and increase short-term lay-offs and recalls among young Canadians. Eliminating UI altogether would significantly lower the unemployment rate among young men as well as lower average observed wages. Under the previous UI rules each month of UI is associated with .86 more months of unemployment than without UI. Under the new rules the ratio is 1.46: each two people on UI can be thought of as generating a third unemployed person not receiving UI through the changes in rm and worker decisions generated by the UI policy. In general, UI policy is found to have complicated effects on the timing of cycles in wages and other variables relative to the productivity cycle.