Essays on labor supply model evaluation and Internet job search

This dissertation consists of three essays in labor economics. The first two chapters evaluate the predictive ability of an important class of structural labor supply models. The last chapter studies the effect of Internet job search on employment outcomes.;Chapter 1 assesses the ability of a structural labor supply model to predict the impacts of a welfare policy change by studying two state welfare reform experiments conducted in Minnesota and Vermont during the mid-1990s. I estimate and evaluate a structural static labor supply model that assumes a discrete choice set of working hours and welfare participation and allows for heterogeneity in preferences. Static discrete-hours labor supply models have been commonly used for welfare and tax policy simulations without their forecast ability being examined. The availability of two experiments provides a rare opportunity for a thorough evaluation of the model. The utility parameters of the model are estimated using data from the Minnesota control group. First, the model's predictions on the effects of the policy change are compared with the experimental impact estimates in Minnesota. Next, I apply the parameter estimates to the Vermont control group sample and compare the predicted and observed impacts of the policy change in Vermont. The results show that the model fits the estimation sample very well, but is unable to replicate the observed treatment effects from the two experiments. Consequently, the increase in net government costs is underestimated by 31 to 93 percent in Minnesota. The prediction biases are even bigger for the Vermont sample.;As shown in Chapter 1, the within-sample fit could be a misleading evaluation tool of a labor supply model's predictive ability since even a labor supply model with a very good within-sample fit often produces large out-of-sample forecast errors. However, this result relies on the structural parameter estimates obtained from the internally inconsistent wage imputation procedure. Chapter 2 studies whether a proper and internally consistent approach to address the missing wage problem would lead to more sensible parameter estimates and better prediction results. Wages are incorporated by the joint estimation of the wage equation and the labor supply model. The joint estimation approach yields more plausible estimates and improved within-sample fit on labor supply but not on welfare participation. Similarly, the within-state evaluation results indicate that labor supply predictions are slightly improved but the downward bias in welfare participation predictions gets substantially worse. The cross-state evaluation results, on the contrary, show that the joint estimation approach substantially reduces the biases in predicted treatment effects on both labor supply and welfare participation. The results reiterate no correlation between within-sample fit and out-of-sample prediction performances found in Chapter 1.;Chapter 3 investigates whether using the Internet for job search helps unemployed workers find jobs. The self-selection of Internet job searchers is addressed by the instrumental variable (IV) estimation strategy. I isolate potentially exogenous variation in individuals' Internet job search status by exploiting the variation in adoption of the Internet across occupation groups. The fraction of unemployed workers using the Internet to search for jobs increased more rapidly in occupations with a higher computer use rate before the introduction of the Internet. The analysis sample consists of unemployed workers from the September 1992 Basic Monthly Current Population Survey (CPS) and the December 1998, August 2000, and September 2001 CPS Computer and Internet Use Supplements. The unemployed workers are longitudinally matched with their employment outcomes from subsequent CPS files. The panel structure of the CPS allows following the individuals for the subsequent 15 months. The IV results suggest that unemployed workers searching for jobs online are around 14.1 percentage points more likely to be employed during the 15 month follow-up period than unemployed workers who do not engage in Internet job search. This implies that using the Internet for job search raises the 15-month job finding rate by around 26 percent at the mean.