-
Chapter 8. Racing Up to the Bottom
- Brookings Institution Press
- Chapter
- Additional Information
This chapter evaluates the performance of the work support system at the state level. This is a significant concern because states have discretion over eligibility and benefit standards for many work support programs. In addition, their administrative practices influence program participation, even in programs that are governed by national standards. The data demonstrate significant variation in state-level performance, in terms both of how widespread and extensive work support program participation is and of how generous state-level benefits are. Data The General Accounting Office (GAO) has noted that the data sets available to study poverty and program participation at the state level suffer from two potential limitations: they either do not have a sufficient number of cases to draw reliable conclusions or do not cover of all fifty states.1 The GAO notes that although most Census Bureau data sets have a sufficient number of cases to make reliable generalizations at the national level, individual state samples are insufficient for state-level analyses. But surveys that provide richer samples for individual states, such as the National Survey of American Families, limit the number of states studied. To answer these concerns in estimating participation in the Children’s Health Insurance Program (CHIP), the size of the Current Population Survey (CPS) sample was expanded during 2000 and 2001, so as to “produce statis145 8 Racing Up to the Bottom 1. General Accounting Office (2001b). 08-8191-1 ch08 11/2/05 4:21 PM Page 145 tically reliable annual state data.”2 The general CPS sample was increased from 55,000 to 60,000“to meet the national and state reliability criteria,”and an additional 12,000 households from the District of Columbia and the thirty-one states with the smallest samples were added to the CPS March supplement . Analysis of labor force participation statistics included in the CPS indicates that these changes improved those estimates as well, especially in states to which the additional 12,000 cases were allocated. Consequently, we used the March supplement of the 2002 CPS to evaluate state-level work support program participation and supplement the CPS data with the estimates of work support benefit generosity presented in chapters 4 and 5. Consistent with the rationale presented in chapter 5, we adjust the state-level benefit estimates for differences in the cost of living (using median housing value as the basis for adjustment) to better indicate the real value of the benefits that states provide. State-Level Work Support Participation In this section we measure how widespread is participation (the proportion of households that receive work supports) and how extensive is participation (the proportion of households that receive three or more work supports) among the households in our sample.3 Significant differences exist in state-level participation rates. Table 8-1 illustrates these differences by reporting state-level summary statistics for the earned income tax credit (EITC), medical assistance (Medicaid and CHIP), school lunch, food stamps, child care grants, Temporary Assistance for Needy Families (TANF), and rental assistance. Consistent with the analysis at the national level, participation rates at the state level are greatest for the EITC, medical assistance, and school lunch programs . The highest EITC participation rate was in West Virginia, where 69.2 percent of sample households reported they received EITC benefits. This contrasts with a low of 51.4 percent in Nebraska. However, in spite of these differences, the coefficient of variation indicates that the federal EITC program exhibits the least interstate variation of any work support program. Medical assistance had the greatest range in state-level participation rates. The highest rate (56.0 percent, in Vermont) is nearly 40 percentage points higher than the lowest rate (16.3 percent, in Nevada). However, the coefficient of variation suggests that there is little interstate variation; Nevada is an outlier in the 146 Evaluating Work Support Performance 2. Census Bureau (2002a). 3. Sample selection is described in chapter 4. 08-8191-1 ch08 11/2/05 4:21 PM Page 146 [3.236.18.23] Project MUSE (2024-03-19 14:43 GMT) distribution. School lunch participation rates were highest in Mississippi (39.0 percent) and lowest inVermont (17.5 percent), and the coefficient of variation indicates a small amount of interstate variation. Thus, the results of the statelevel analysis are consistent with the national analysis in two important ways: the same three programs emerge as the most popular work supports and (with the exception of the occasional outlier) the states exhibit little variation in...