This paper empirically analyses the impact of the unemployment insurance system upon the insured unemployment rate and the average duration of unemployment. It employs a simultaneous equation framework because of possible feedback effects between the insured unemployment rate and the average duration of unemployment. Based on a pooled crosssectional time-series model (covering all the 50 states in the USA for the years 1967-88) that corrects for heteroscedasticity and autocorrelation, and the results show some support for the hypothesis that the unemployment insurance system, by providing workers with a safety net, increases both the insured unemployment rate and the duration period.