On the Distributional Shape of Unemployment Duration
使用加速失效时间模型和扩展广义伽马分布，分析1984年1月当前人口调查中失业工人的数据，发现回归参数对分布假设敏感，且不同数据子集呈现不同的分布形态和持续时间依赖性。
Using an accelerated failure time model, the extended generalized gamma distribution is deployed to discriminate among various special cases of that distribution. Displaced worker data from the January 1984 Current Population Survey are analyzed within a regression framework that accommodates the stochastic nature of the point of censoring associated with incomplete spells of unemployment. Not only are the regression parameters sensitive to distributional assumption but also there is evidence of different distributional shapes and hence duration dependencies in subsets of the data.