Table 13-19 presents probit estimates of the probability of being employed for at least one quarter in the year after leaving welfare. Only those cases that left welfare are included. We expect that, controlling for all else, those with more work experience prior to leaving are more likely to be employed after leaving welfare. We also expect that those with shorter welfare receipt histories are more likely to be employed after leaving than those with longer welfare receipt histories.
Results presented in the first model are contrary to expectations in that both long-termers and cyclers are more likely to be employed after leaving welfare than short-termers. This controls for the age of the case head, the age and number of children, the SSI status of leavers, and other variables that also might be associated with employment. The results are, however, consistent with findings from Cancian et al. (1999).
In the next model, the long-term, short-term, and cycler distinctions were "unpacked"; that is, variables for asl and asl-squared along with the dummy variables for the number of spells were included. Results are similar to those in the first model in that longer spells of benefit receipt are positively associated with the probability of employment after leaving. However, the relationship is nonlinear as the coefficient on the variable for average spell length squared is negative and significant. As spell length increases, the marginal increase in the probability of employment gets smaller.
Instead of using the cycler distinction for measuring the frequency for which a case goes on and off AFDC, the second model includes a series of dummy variables for the number of spells of AFDC receipt, as explained earlier. In the first model, cyclers (three or more spells regardless of spell length) were significantly more likely to be employed within a year after leaving welfare than short-termers. In the second model, we see that relative to those with no prior AFDC spells, those with one to three spells of AFDC receipt are significantly less likely to be employed after leaving welfare. Those with more than three spells are no less likely to be employed than those with no prior AFDC spells. The results of the first two models do not conflict with each other because their reference groups are different. The reference group in the second model includes those with no prior AFDC receipt, which may include those who are slightly better off than short-termers because they have not had to rely on AFDC prior to July 1995 (the short-termer group includes some with no prior AFDC receipt, but it also includes some with some prior AFDC receipt). The result that those with at least four prior spells of receipt are no less likely to find employment after leaving than those with no prior AFDC receipt and the positive and significant sign on the cycler variable in Model 1 are still a bit perplexing. One hypothesis is that those who cycle on and off welfare also cycle between employment and unemployment. Because this group has some work experience, its members may have a relatively easier time finding jobs after leaving. This hypothesis is only supported to the extent that those with many spells are more likely to be employed after leaving than those with a few spells. However, those with a few AFDC spells (between one and three) are less likely to find employment after leaving than those with no prior AFDC spells.
|Independent Variable||Model 1||Model 2||Model 3|
|Cycler (reference group is short-termers)||+(a)||0.013|
|Average spell length||+(a)||0.002|
|Average spell length squared||-(a)||0|
|One spell of receipt||-(a)||-0.04|
|Two or Three Spells of Receipt||-(a)||-0.031|
|Four or more spells of receipt (reference group is no prior spells)||-||-0.023|
|No earnings prior to leaving & Short-term welfare recipient||-||-0.006|
|Some earnings prior to leaving & Long-term welfare recipient||+(a)||0.034|
|No earnings prior to leaving & Long-term welfare recipient||+||0.012|
|Some earnings prior to leaving & cycler||+(a)||0.014|
|No earnings prior to leaving & cycler (reference group for this series is short-termers with earnings prior to leaving welfare)||-||-0.055|
|# quarters withearnings before leaving||+(a)||0.01||+(a)||0.011||+(a)||0.01|
|Age of case head||-(a)||-0.007||-(a)||-0.007||-(a)||-0.007|
|Age of case head squared||-(a)||0||+||0||+||0|
|Hispanic (reference group is white)||+(a)||0.028||+(a)||0.028||+||0.029|
|No high school diploma||-||-0.003||-||-0.003||-||-0.003|
|At least some college (reference group is high school diploma)||+(a)||0.011||+||0.01||+(b)||0.011|
|Age of youngest child||+(a)||0.002||+(a)||0.002||+(a)||0.002|
|# of children under age 5||-||-0.003||-||-0.004||-||-0.003|
|# of children over age 5||+(a)||0.012||+(a)||0.011||+(a)||0.012|
|Other adult present in household||-||-0.006||-||-0.005||-||-0.006|
|Milwaukee County resident||+||0||-||-0.002||+||0|
|Resident of other urban county (reference group is rural county resident)||-||-0.005||-||-0.005||-||-0.005|
|Child receives SSI||-||-0.007||-||-0.008||-||-0.007|
|Mother receives SSI||-(a)||-0.219||-(a)||-0.218||-(a)||-0.219|
|Unemployment rate in county in 1996||-(a)||-0.01||-(a)||-0.01||-(a)||-0.01|
|Left AFDC 4th quarter 1995||+||0.009||+||0.008||+||0.009|
|Left AFDC 1st quarter 1996||+||-0.001||-||-0.002||-||-0.001|
|Left AFDC 2nd quarter 1996 (reference is left AFDC 3rd quarter 1995)||+||-0.004||-||-0.005||-||-0.004|
|Log Likelihood (restricted log likelihood is -6,471.46)||-5,605.20||-5,597.75||-5,604.41|
|Likelihood ratio index||0.134||0.135||0.134|
|Percent of observations predicted correctly||88.4||88.4||88.4|
|a Coefficient is statistically significant at the 5-percent level.
b Coefficient is statistically significant at the 10-percent level.
The third model attempts to flesh out the results in the first two models with respect to welfare receipt history. The model includes a series of dummy variables for the earnings and welfare receipt history of leavers. The third model combines the welfare receipt and work history variables. The sample is categorized into six groups: short-termers with no prior work experience, short-termers with at least one quarter of prior work experience, long-termers with no work experience, long-termers with some work experience, cyclers with no work experience, and cyclers with some work experience. The reference group consists of short-termers with some work experience.
Results from this model are useful in explaining the peculiar results in Models 1 and 2. Those with long-term welfare receipt histories and at least one quarter of work experience prior to leaving still have higher employment probabilities than short-termers with work experience. However, employment rates of long-termers with no prior work experience are not significantly different from the employment rates of short-term recipients with prior work experience. Likewise, cyclers with some prior work experience have higher probabilities of employment after leaving welfare than short-termers with some work experience. However, cyclers with no work experience do not have different employment rates than short-termers with some work experience. This group of leavers with long-term welfare receipt histories clearly have characteristics or faces economic or policy conditions that are associated with increased employment compared to those who have used welfare less. These results need further investigation.
For all three models of the probability of employment, those with more work experience are more likely to be employed after leaving welfare, as expected. The coefficient is positive and strongly significant. A one quarter increase in prior work experience increases the probability of employment after leaving by 1 percentage point. One other variable of interest is the dummy variable for whether or not a case was sanctioned from benefit receipt. In all three models, sanctioned cases were significantly less likely to be employed than nonsanctioned cases. The marginal effect of a sanctioned case decreases the probability of leaving welfare by more than 4 percentage points. This is as expected and is initial evidence that sanctioned cases may have a tough time finding employment.
"01.pdf" (pdf, 472.92Kb)
"02.pdf" (pdf, 395.41Kb)
"03.pdf" (pdf, 379.04Kb)
"04.pdf" (pdf, 381.73Kb)
"05.pdf" (pdf, 393.7Kb)
"06.pdf" (pdf, 415.3Kb)
"07.pdf" (pdf, 375.49Kb)
"08.pdf" (pdf, 475.21Kb)
"09.pdf" (pdf, 425.17Kb)
"10.pdf" (pdf, 424.33Kb)
"11.pdf" (pdf, 392.39Kb)
"12.pdf" (pdf, 386.39Kb)
"13.pdf" (pdf, 449.86Kb)