Abstract: | Empirical evidence on the sign of the slope of the hazard rate from unemployment is obtained from a fixed effects model based on the gamma distribution for unemployment duration. The data used are pairs of unemployment spells for adult males in the control and experimental groups of the Denver Income Maintenance Experiment. The sample selection issue involved in selecting the pair of spells used in the empirical work is discussed. The empirical results suggest that for these samples the hazard function is monotone decreasing and support the assumptions that the first two spells of unemployment are identically distributed and can be modelled using the gamma distribution. |