Skip to main content

Currently Skimming:

14 Experienced-Based Measures of Heterogeneity in the Welfare Caseload
Pages 473-500

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 473...
... This heterogeneity has assumed even greater importance in the welfare reform environment of the l990s. The new reforms are, generally speaking, aimed at raising employment levels and promoting work, particularly off the welfare rolls.
From page 474...
... Thus, for example, leaver outcomes before and after 1996 may differ because of the business cycle rather than because of welfare reform.3 Heterogeneity in the caseload can be characterized in many ways. A straightforward approach is simply to examine the distributions of characteristics thought to be related to labor market skill, income-generating potential, and general coping capabilities.
From page 475...
... The first group is composed of recipients with long spells of receipt and hence heavy dependence on welfare; the second group is composed of recipients who have short spells and are on welfare infrequently, leading to relatively mild dependence; and the third group consists of women who frequently move on and off the rolls and may, in the end, accumulate enough total time on welfare that they should be classified as welfare dependent even though their spells are fairly short on average.5 This view, again, suggests that the types of women who have high turnover and short spells are different than those who have low turnover and long spells, even though they might have the same total-time-on. The reason that one might expect differences among recipients with different turnover rates will be discussed in the text of this paper.
From page 476...
... We will be interested in this chapter not in these structural transition rates, but rather in the distributions of welfare participation outcomes that is, the types of patterns of participation that occur that result from them over a particular calendar interval. Nevertheless, that different women have different patterns over such an interval necessarily arises from differences in the underlying hazards, and those hazards are a function of the variables denoted.
From page 477...
... Hence the existence of earnings fluctuations around each individual mean will lead to more movements on and off the rolls for those with mean earnings close to the cutoff point for leaving or entering the rolls than for those with mean earnings farther away from that cutoff, assuming that the variance of the fluctuations is the same for all. This simple model would lead to the presumption that short-termers have the highest labor market skill, with mean earnings sufficiently high that only significant negative earnings declines result in participation; long-termers have the lowest labor market skill, with mean earnings sufficiently low that only significantly positive earnings increases lead to an exit 6We do not list H(i,t)
From page 478...
... If only total-time-on causes such deterioration, we should find that labor market skill even though it is partly a result, not a cause, of welfare participationshould be negatively related to an individual's amount of total-time-on but not to turnover or spell lengths, holding total-time-on fixed. A different model is one in which different individuals experience different degrees of fluctuation in earnings (i.e., different variances)
From page 479...
... A fourth and final model is one in which noneconomic considerations play a larger role in welfare turnover, unlike the models so far that tie welfare participation decisions closely to earnings levels. Noneconomic events like marriage, divorce, childbearing, and changes in personal situation, all can affect welfare turnover rates.
From page 480...
... To the extent that the labor market skill measures that will be the main focus of our analysis are more weakly correlated with these censored spell lengths than their uncensored counterparts, our correlations of spell length with skill will be weakened. However, as we shall describe, our calendar interval is 10 years long and hence there are few censored spells relative to the total number of spells.
From page 481...
... The vast majority of studies of welfare dynamics present estimates of the determinants of exit from welfare spells or entry onto welfare or, sometimes, of rates of reentry onto welfare after an exit. These econometric models are not set up to distinguish the determinants of turnover per se from the determinants of total-time-on because they impose a restrictive relationship between the effects of the independent variables on turnover rates, total-time-on, and spell lengths.
From page 482...
... In a study using administrative data from Wisconsin, Ver Ploeg (this volume: Table 13-5) examined employment rates of individuals on Temporary Assistance for Needy Families (TANF)
From page 483...
... Table 14-A3 shows how these average spell lengths differ by the number of spells in the period. Interestingly, those with larger numbers of spells tend to have larger numbers of medium spell lengths, but smaller numbers of both long and short spells.
From page 484...
... 484 o oo CO CM ~ CM o oo CO uao~ad
From page 485...
... 485 CO 11 (a CM tl I ~ Cal o o o o o ~ Us ~ CO CM ~ue0~0d En Q U)
From page 486...
... , annual earnings, weekly wages, and hourly wage rates, all measured only over periods off AFDC. The earnings and wages are measured only over periods off AFDC because they are intended to represent earning capability; including the AFDC periods would bias the measures in this sense, for earnings and wages are always lower on welfare than off.21 In traditional economic analyses, an individual's potential hourly wage and weekly wage are usually considered to be the best indicators of labor market skill.
From page 487...
... 487 o CO Cal o ~ o Cal ~ ~ On ~ o ·_.
From page 488...
... and L The variation in the characteristics with all three variables goes in the expected direction: Those with greater dependence, more spells, and longer average spell lengths tend to have lower levels of education, are more likely black (but not Hispanic)
From page 489...
... 489 ca V, ¢ Cal C)
From page 490...
... Thus the division is not quite equal across the three groups, but deviates from an equal division only through a somewhat greater-than-one-third long-termer group and smaller-than-one-third cycler group. The percent of the population classified as long-termers may seem high, even for the 1.5-year spell definition, where more than one-third of ever-on recipients are so classified.23 However, it should be noted that there have been no previous calculations of these distributions in the literature, and hence it is difficult to find comparisons in past work.24 Still, a smaller fraction of long-termers clearly could be obtained by requiring longer average spell lengths than 1.5 years, but at the cost of including as short-termers those with such long average spell lengths; and a 1.5-year spell does not seem to fit the notion of a short-termer.
From page 492...
... Although the educational and racial distributions are approximately the same, and while employment rates off AFDC for long-termers are indeed somewhat lower than those for cyclers, cyclers have lower annual earnings, weekly earnings, and hourly wages than long-termers. This result is quite surprising in light of the conventional wisdom in the literature based on the model discussed earlier in the paper, which assumes that cyclers are somewhat better off than long-termers by virtue of having sufficiently greater earning power to leave the welfare rolls periodically to enter the labor market.
From page 493...
... Moreover, even though the minimum spell length for long-termers used in Definitions 1, 2, and 3 is not large, the median and mean spell lengths are still in the range of 2- to4 years (see Table 14-A4~; so long-termers by these definitions typically indeed have very long spells. For all these reasons, it does not appear that any reasonable
From page 494...
... Each of the four key labor market characteristics the employment rate, annual earnings, weekly wages, and the hourly wage, all measured only over periods off AFDC is regressed, in the first case, on T and N or on T and L (all three are not included in the same equation because they are definitionally related to each other)
From page 495...
... Appendix Panel on Data and Methods for Measuring the Effects of Changes in Social Welfare Programs AGENDA Workshop on Data Collection for Low Income and Welfare Populations December 16-17, 1999 Georgetown-Holiday Inn (Mirage I) 2101 Wisconsin Avenue, NW Washington, D.C.
From page 496...
... The analysis shows that the single most consistent predictor of those characteristics is the total amount of time a woman has been on welfare. However, whether that time arises from a larger number of shorter spells, or a smaller number of longer spells, is less consistently important; that is, neither turnover per se nor the length of individual spells of welfare receipt is always related to labor market characteristics holding constant the total time the individual has been on welfare.
From page 497...
... Unpublished paper presented at the conference on Rural Dimensions of Welfare Reform, Joint Center on Poverty Research, Washington, DC, May 4-5. Ellwood, D
From page 498...
... TABLE 14-A2 Distribution of Number of Spells (N) (percent distribution)
From page 499...
... No. of Spells 1 2 3 1-6 25.5 32.1 21.2 12.2 7-12 19.7 19.7 13.9 28.6 13-18 11.9 7.2 11.5 16.3 19-24 9.5 9.2 5.5 20.4 25-30 6.8 2.4 11.5 14.3 31-36 7.0 2.4 10.9 5.6 37-48 7.4 5.6 13.3 5.3 49-60 5.8 4.0 12.1 0 61-84 1.8 3.6 0 0 85+ 4.7 9.6 0 0 Mean 23.8 27.2 24.1 16.8 Median 14.9 12.0 23.5 15.7 TABLE 14-A4 Distribution of Total-Time-On and Spell Lengths Among Long-termers, Short-termers, and Cyclers Definition 1 Definition 2 Definition 3 Long- Short- Long- Short- Long- Short termer termer termer termer termer termer Cycler Total-Time-On: 1-6 - 80.5 - 50.3 - 42.2 1.0 7-12 16.6 17.0 - 36.5 - 36.7 3.0 13-18 9.8 2.5 8.0 7.4 - 14.2 5.0 19-24 10.8 - 10.2 4.8 12.1 4.0 9.0 25-36 13.9 - 17.3 1.1 11.6 8.0 14.0 37-48 7.8 - 10.2 - 11.6 - 11.0 49-60 9.0 - 12.9 - 15.3 - 13.0 61-72 8.1 - 11.7 - 12.7 - 19.0 73-84 3.7 - 4.9 - 5.8 - 10.7 85-96 6.4 - 8.4 - 10.1 - 7.0 97+ 13.1 - 17.3 - 20.6 - 8.0 Mean 47.9 4.8 58.8 8.0 65.5 10.5 54.5 Median 34.2 4.0 52.0 6.0 60.0 8.0 56.5 Spell Length: Mean 34.9 3.5 42.7 6.1 47.9 7.6 14.9 Median 26.5 3.5 35.0 5.5 38.0 6.0 13.5


This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More information on Chapter Skim is available.