Three essays in labor economics

The first two chapters of my dissertation analyze the New Hope Project, a random-assignment experiment implemented in two inner-city areas in Milwaukee, Wisconsin. In the project, eligible applicants were randomly assigned to a treatment group or a control group. The treatment group members could freely use the project benefits and services, including earnings supplements, health insurance subsidies, child care subsidies and referrals to community service jobs. Starting in 1994, the project was operated for three years and participants were followed for five additional years after the project ended.;The first chapter evaluates the short- and long-term effects of the program treatment components on employment and earnings of those who actually took up the treatments. A multivariate probit model is used to estimate the treatment utilization patterns jointly for the four treatments and to describe self-selection into treatment uses. An extension of Heckman's estimator is applied to account for the potential selection bias problems in the estimation of the treatment effects on the treated. The estimator includes multi-dimensional selection factors that are imputed through a simulation process. The analysis suggests that some unobserved factors associated with selection into treatment uses are correlated with unobservables that influence the economic outcomes. The results from the multinomial selection correction models indicate that earnings supplements and community service jobs increased employment in the short-run, and childcare subsidies boosted earnings, but these positive effects faded when the subsidies ended. Another important finding was that health insurance subsidies led to reductions in employment and earnings, and some of the negative effects lasted beyond the experimental period.;The second chapter examines heterogeneity in effects of the New Hope Project by participants' pre-program attachment to the labor force and barriers to employment. It finds that the project resulted in significant increases in employment and earnings-related income of the quartile that was the least attached to the labor force before program participation. In this quartile, participants were not employed, had no pre-program earnings, and had no prior full-time work experience before program enrollment. Over the three experimental years and five follow-up years, there is no evidence that the project had significant impacts on other pre-program labor force attachment quartiles. Moreover, this study finds that benefits and services provided by the New Hope Project led to significant and persistent increases in employment and earnings-related incomes of participants who had only one barrier to employment: lacking a GED certificate or a high school diploma. The cost-effectiveness analysis undertaken in this study demonstrates that the program was more cost-effective in increasing earnings-related income for the quartile with the least pre-program attachment to the labor force and the subgroup with low education as the only barrier to employment than for the whole treatment group.;Using data from the National Longitudinal Survey of Youth 1979 (NLSY79), the third chapter examines impacts of maternal employment on the development of children in families led by working single mothers. A contemporaneous specification and a value-added specification are implemented for modeling the production functions for cognitive and non-cognitive development of children. In the estimation models based on the contemporaneous specification, instrumental variables and fixed-effects models are used to identify the effects of mother's recent work hours and work weeks. No significant effects of contemporaneous maternal employment are found. The value-added specification includes a lagged outcome that represents all past inputs to the production of outcomes of children. There is little evidence of significant effects of maternal employment in any of the models estimated.