ESSAYS ON THE MICROECONOMETRICS OF LABOR MARKETS AND CRIMINAL BEHAVIOR

The subject of the present thesis is to apply microeconometrictechniques to labor economics and criminal behavior. There are twoindependent chapters. The aim of chapter one is to build andestimate an econometric model of recidivism for the United Statesusing data set of released prisoners from state prisons in 1983.The data set is unique in the sense that it contains informationon recidivism for both federal and state crimes in contrast ofpast studies restricted only to state samples with few or noinformation about federal crimes. The paper extends the literatureon the application of econometrics to criminal behavior bytreating the process of recidivism as a competing risks modelwhere dependence arises from unobserved heterogeneity. Theeffects of regressors are estimated in a model that considers thenon-monotonic hazards feature of criminal recidivism data sets. Anaccelerated time model is compared to the traditional Coxproportional model. The former appears to perform better, and isshown to capture some features of recidivism behavior reasonablywell. Chapter two evaluates the effect of a reemployment program forex-convicts on their recidivism behavior. The econometric modelhas to deal with the following institutional aspects of theprogram. First, upon release, each ex-inmate looks for a job andthis duration is represented by T_{s}. Second, at the samerelease-time the ex-criminal could commit a crime and the time ofrecidivism is represented by T_{c}. Third, the differencebetween controls and treatments samples only occurs conditional onfinding a job. Hence, the random search duration, T_{s},represents the timing of receiving the treatment. Finally, theavailable duration data is grouped. Our paper extends theliterature on econometric program evaluation by building andestimating a model that considers the conditional feature of thetreatment. We use a sample from the 1985 Employment Services forEx-Offenders Program and find that the program helps reducecriminal activity.