We use a choice-based subsample of Social Security Disability Insurance applicants from the 1978 Social Security Survey ofDisability and Work to test the importance of policy variables on the timing of application for disability insurance benefits following the onset of a work limiting health condition. We correct for choice-based sampling by extending the Manski-Lerman (1977) correction to the likelihood function of our continuous time hazard model defined with semiparametric unmeasured

more »

... eneity and find that this correction significantly affects the results. We find that economic variables-the size of the disability benefit, expected wage earnings and accommodation-matter. JEL CLASSIFICATION: D6, C4 Abstract We use a choice-based subsample of Social Security Disability Insurance applicants from the 1978 Social Security Survey of Disability and Work to test the importance of policy variables on the timing ofapplication fordisability insurance benefits following the onset of a work limiting health condition. We correct for choice-based sampling by extending the Manski-Lerman (1977) correction to the likelihood function of our continuous time hazard model defined with semiparametric unmeasured heterogeneity and find that this correction significantly affects the results. We find that economic variables-the size ofthe disability benefit, expected wage earnings and accommodation-matter.