Search Results

Author: Kim, Sungteak
Resulting in 1 citation.
1. Kim, Sungteak
Duration Analysis of Welfare Spells: With Application to the NLSY and the NLS Young Women Data
Ph.D. Dissertation, Brown University, 1998. DAI-A 59/04, p. 1283, Oct 1998.
Cohort(s): NLSY79, Young Women
Publisher: UMI - University Microfilms, Bell and Howell Information and Learning
Keyword(s): Aid for Families with Dependent Children (AFDC); Family Structure; Heterogeneity; Modeling, Fixed Effects; Modeling, Hazard/Event History/Survival/Duration; Modeling, Mixed Effects; Parents, Single; Test Scores/Test theory/IRT; Welfare

mThis dissertation studies the determinants of welfare duration and the effects of the duration on the conditional probability of exit. It provides a theoretical background to understand the exit decision of welfare participants and the determinants of welfare duration, and explores the surroundings concerning welfare duration. The proportional hazard model and the accelerated failure time model are applied to the NLSY data (1979-1993), and the results are examined by two specification techniques-split sample comparison, and numerical test using a part of score test statistic. The fixed-effects partial likelihood model is applied to the multiple-spell NLSY data in an effort to solve the problems raised by the unobserved heterogeneity. Finally the welfare trends and the baseline hazard distributions over the 1970s and the 1980s are compared. This study reports that there is little evidence of negative duration dependence but strong presence of heterogeneity in the population, although some of the welfare recipients stay in the welfare program for a long period. The estimation results from the two hazard models are similar in the case of the whole sample. But the results from the split samples show very diverse estimates across the corresponding groups, and provide an evidence to the seriousness of the unobserved heterogeneity. Despite of the indication of decreasing hazard rate from the estimation results, the specification tests point the exponential distribution as the true underlying hazard rate except the 'self-income decrease' group. The multiple-spell application with the fixed-effects partial likelihood method presents more number of significant estimates and smaller standard errors than the single-spell application. Among the explanatory variables, the family income and the total welfare income consistently show significant effect on the hazard rate consistently, while the self-income and the maximum AFDC benefit reports insignificant effect. The policy imp lication from the results suggests to keep the family structure which can financially support the disadvantaged single mother, and to reduce the extent of connection among the welfare programs. The comparison of the welfare trends over the two decades reports that the young women in the 1980s start the welfare participation earlier, to return more often to welfare programs, and to work less hours than in the 1970s.
Bibliography Citation
Kim, Sungteak. Duration Analysis of Welfare Spells: With Application to the NLSY and the NLS Young Women Data. Ph.D. Dissertation, Brown University, 1998. DAI-A 59/04, p. 1283, Oct 1998..