The model used to estimate the influence of employment liabilities can be used not only to examine the relative effect of each liability on employment status but also to simulate various strategies designed to increase work among TANF recipients. For example, the model can be used to examine how much the share of TANF recipients working 30 or more hours per week would increase if all child care problems were eliminated. The answer to this question depends on the proportion of recipients with a child care problem and the extent to which child care reduces the chance of working. The answer may also be influenced by the presence of other liabilities that may have independent effect on the likelihood of employment. In Table IV.6, we present the results of policy simulations that address one liability at a time. We then present simulations of strategies that address multiple liabilities, as in a policy or program that emphasizes development of human capital.
Under the current constellation of liabilities, about 28 percent of TANF case heads work 30 hours a week or more. Our simulation results suggest that no strategy that addresses just one specific liability would go far in increasing this rate. Eliminating the negative influence of limited work experience, possibly by placing recipients in transitional jobs or work experience programs, would increase the proportion of TANF recipients working 30 or more hours per week by 6 percentage points to 34 percent. Eliminating all child care problems would raise the proportion working by 4 percentage points to 32 percent.
Strategies that address multiple liabilities that may be related to one another would further increase the proportion of recipients working 30 or more hours per week. Still, these comprehensive strategies, even if successful, would raise the level of substantial employment to no more than two-fifths of the caseload. The comprehensive strategies that would go the furthest toward increasing work participation are those that address human capital liabilities and/or logistical challenges. A strategy that increases the level of work experience and eliminates the logistical challenges of child care and transportation would increase the proportion of the caseload working 30 or more hours per week by nearly 14 percentage points, to 41 percent. A strategy that eliminates all human capital liabilities (e.g., education and work experience) and addresses special learning needs (e.g., learning disabilities and language barriers) would increase the proportion to 36 percent. A strategy that eliminates all child care and transportation problems would raise the proportion of workers by about 6 percentage points. A strategy that focuses only on physical health and mental health problems and chemical dependence would raise the proportion of workers by only 3 percentage points. However, if these personal challenges contribute to the logistical and situational challenges faced by recipients, such a strategy could increase considerably the proportion of the caseload that is substantially employed.
|Change from Current Situation||Simulated Percent Employed 30+||Hours Per Week|
|Individual Liabilities Eliminated|
|No high school diploma or GED||29.0||+1.1|
|Limited recent work experience||34.3||+6.4|
|Performed fewer than four common job tasks||28.4||+0.5|
|Physical health problem||30.6||+2.7|
|Mental health problem||28.4||+0.5|
|Severe physical domestic violence in past year||26.9||- 1.0|
|Signs of a learning disability||27.5||- 0.4|
|Difficulty with English||28.2||+0.3|
|Child or other family member or friend with a health problem or special need||27.2||- 0.7|
|Child under age one in household||28.7||+0.8|
|Child care problem||32.2||+4.3|
|Multiple Liabilities Eliminated|
|All human capital and learning issues||35.9||+8.0|
|All personal challenges||31.6||+3.7|
|All logistical challenges||33.8||+5.9|
|Physical health, mental health and chemical dependency||31.2||+3.3|
|Work experience and logistical challenges||41.4||+13.5|
|Source: Simulations based on the results of a logit model predicting the probability of working 30+ hours per week using data from 2001-02 survey of Illinois TANF cases and Illinois administrative data.|
The simulation results should be interpreted with caution. First, if the influence of unobserved variables (such as personal motivation) is reflected in observed variables (such as work experience), then the simulations overestimate the effects of a particular policy change, such as increasing work experience among TANF recipients. Second, because the results are based on cases that were on TANF at a point in time (and consequently may oversample long-term cases, as noted earlier), they do not capture the effects of strategies that are already successful at moving cases both into substantial employment and off of TANF.
1. The liabilities analyzed in this chapter differ slightly from those presented in Chapter III. For purposes of the multivariate analysis, the measure of work experience omits the study quarter, the fourth quarter of 2001. The liability is, therefore, defined as employed in less than four of the seven quarters preceding selection into the study. Three liabilities--perceived discrimination, a criminal conviction, and perceived problems within one's neighborhood--are excluded from the analysis presented in this chapter. We did not include convictions because, based on our initial analysis, arrests are the more important determinant of employment. Perceived discrimination was not included because it was not asked of recipients who were not working, and perceived problems in one's neighborhood were not included because many respondents did not respond to all the components of the question that were necessary in developing the summary measure.
2. For purposes of counting liabilities we grouped child care problems, pregnancy, and having a child under one year old in the household together as one.
3. This predicted probability is different than that for the model that includes the number of liabilities. The difference in the model specification results in a different set of coefficients used to produce the predicted probabilities.