The effect of welfare-program benefits on the earnings of public-housing residents
Tinsley, Karen Lee
MetadataShow full item record
The purpose of this dissertation is to investigate empirically the work disincentives of the cash-assistance program for public-housing residents. A two-equation system determining labor earnings and welfare benefits is estimated in a simultaneous-equations limited-dependent variable framework, where both earnings and welfare benefits are treated as endogenous, censored variables. With a sample of public-housing residents from the 1997 wave of the National Survey of American Families, a two-stage Tobit model of annual earnings is estimated. The predicted value from a Tobit regression determining welfare benefits is used as an explanatory variable in the earnings equation. Standard errors are corrected to account for use of this instrumental variable. The model is also estimated within a self-selection framework. Specifically, a bivariate probit selection mechanism determining work and welfare participation, which takes into account any correlation of the error terms due to unobserved characteristics that may affect both choices, is used to augment the annual earnings equation. Accounting for welfare benefits endogeneity, the empirical results indicate that an increase in labor earnings replaces only 35 to 75 percent of a reduction in welfare benefits; a smaller effect than was found assuming exogenous benefits. The selfselection model suggests that unobserved characteristics that negatively affect labor supply (earnings) also positively affect the magnitude of welfare benefits. That is, women who are more likely to have higher welfare benefits are also more likely to have lower earnings, after controlling for measured determinants of both earnings and benefits. The findings imply that HUD’s budget would absorb approximately 25 to 65 percent of any decrease in welfare benefits experienced by women living in public housing.