Final Synthesis Report of Findings from ASPE "Leavers" Grants. How Much Are Welfare Leavers Working?

11/27/2001

Fourteen of the ASPE-funded leaver studies included here use administrative data from their states Unemployment Insurance (UI) systems to examine the employment and earnings of TANF leavers in the months and years following exit. A fifteenth study, the District of Columbia's, accessed data from the National Directory of New Hires (NDNH) to obtain quarterly employment information for its TANF leavers. This data source cuts across state lines and includes federal workers; thus, given D.C.'s geography and employment patterns, NDNH has more complete and more useful information for D.C. than state UI records.

Post-exit employment rates calculated from administrative data appear in Figure III.1more detailed information appears in Table III.1 Employment rates are remarkably consistent across studies.

Figure III.1:
Employment Rates of Single-Parent Welfare Leavers

Figure III.1: Employment Rates of Single-Parent Welfare Leavers

Notes: The graph shows the minimum, maximum, and median employment rates as reported across the studies. The shaded box represents the range in which the middle 50% of reported employment rates fall. Not all studies provide data for all post-exit quarters. See table III.1 for more information.

Table III.1:
Employment of Single- Parent Welfare Leavers: Select Administrative Data Findings

State/Study

Exit Cohort Post Exit Quarter (%) Worked All Four Quarters (%) Ever Worked After Exit (%) Working in Pre-Exit Quarter (%) Working in Exit Quarter (%)
Q1 Q2 Q3 Q4

Arizona

1Q98 53 51 52 501 n.a 73 1 47 54

District of Columbia2

4Q97 62 66 57 60 39 79 n.a n.a

Florida

2Q97 50 51 53 54 31 71 n.a 39

Georgia

1Q99 64 60 59 57 n.a n.a n.a 61

Illinois

3Q97 - 4Q98 54 53 54 55 39 70 49 55

Iowa

2Q99 57 42 39 38 25 69 44 57

Massachusetts2

Dec 1998 - Mar 1999 60 61 51 n.a n.a 68 n.a 57

Missouri2

4Q96 58 58 59 58 n.a n.a n.a n.a

New York

1Q97 50 49 48 48 40 62 n.a 50

South Carolina2,3

Oct 1998 - Mar 1999 67 68 67 63 34 90 61 69

Washington

4Q97 57 57 58 57 n.a n.a 50 61

Wisconsin2

2Q98-4Q98 67 65 67 67 n.a 82 55 68

Cuyahoga Co.3

3Q98 68 64 67 64 47 82 n.a n.a

Los Angeles Co.3

3Q96 47 46 46 47 35 57 43 46

Bay Area

4Q98 55 55 55 n.a n.a n.a 51 58

1Data from report differ from revised data in public use data file. Revised fourth quarter employment is 51% and ever worked after exit is 70%. Arizona added 17 new cases to the data file one year after the report was published.
2Report employment data for all cases, not just for single-parent cases.
3Los Angeles Co., Cuyahoga Co. and South Carolina require a leaver to have at least $100 in earnings to be considered working while others require only $1.

In the first quarter after exit, employment rates range from a low of 47 percent in Los Angeles to a high of 68 percent in Cuyahoga County. The median first post-exit quarter employment rate across the studies is 57 percent, and many studies cluster tightly around the median. By the fourth post-exit quarter, most studies' employment rates remain tightly clustered around 57 percent. Wisconsin's leavers have the highest fourth quarter employment rates (67 percent) while Iowa's leavers have the lowest (38 percent).

In most studies, post-TANF employment rates as reported using administrative data remain fairly stable over the first post-exit year. The two exceptions to this are Georgia, which shows a modest decline in employment rates from 64 to 57 percent between the first and fourth post-exit quarters, and Iowa, which shows a substantial decline from 57 to 38 percent.

It is important to note that while overall employment rates for TANF leavers hover just below 60 percent, this does not imply that the same individuals who worked in the first quarter continue to work throughout the year. Indeed, there is a considerable amount of employment "churning" in the welfare leaver population. In the eleven studies that report information on leavers who ever worked over the first post-exit year, we see that the share of leavers who ever worked after exiting ranges from a low of 57 percent in Los Angeles County to a high of 90 percent in South Carolina; the median "ever worked" employment rate is 71 percent. Further, across the eight studies reporting the share of leavers who worked in all four post-exit quarters, the median study finds that only 37 percent of leavers worked in all four post-exit quarter. The 'all four-quarters' employment rates range from a low of 25 percent in Iowa to a high of 47 percent in Cuyahoga County.

Although there are some methodological differences between studiesfor example, the South Carolina, Los Angeles and Cuyahoga County studies require a leaver to have at least $100 in earnings to be considered working while others require only $1these differences do not account for much of the meager variation across studies. Indeed, Cuyahoga County consistently reports high employment rates despite having a higher threshold for employment. Thus overall, these findings from administrative data suggest that the majority of welfare leavers work or have worked since exiting, but about one out of four have never worked in the year following exit.

All these employment measures are fairly broadeven in the three studies using the $100 earnings threshold, a leaver would be considered employed if she earned minimum wage and worked for just one half of one week out of a 13 week quarter. The public use data files made available by Arizona, the District of Columbia, and Iowa allow an examination of employment rates using a stricter definition: an earnings requirement of at least $500 in a quarter to be considered employed. This higher threshold basically requires a leaver to have worked the equivalent of two full-time weeks and be paid about $6.25 an hour to be counted as having worked in a given quarter. Table III.2 shows that under this tighter rule, employment rates in the first post-exit quarter are 6 to 11 percentage points lower than under the "any earnings" criterion. In the fourth post-exit quarter, the employment rates based on the $500 rule are 6 to 10 percentage points lower than the rates based on the "any earnings" for the three studies. Finally, the share of leavers who ever earned more than $500 dollars in any of the first four post-exit quarters (the "ever worked" employment rate) (see Figure III.2) is also lower than when using the "any earnings" rule: in Arizona, only 62 percent of leavers "ever worked" during the first post-exit year if the "$500 rule" is applied, compared with 70 percent using the "any earnings" rule.

Figure III.2:
Percent of Single-Parent Welfare Leavers Who Ever Worked in Year After Exit Using Alternative Definitions of Work

Figure III.2: Percent of Single-Parent Welfare Leavers Who Ever Worked in Year After Exit Using Alternative Definitions of Work

Notes: See table III.2 for more information.

Table III.2:
Employment Rates of Single- Parent Welfare Leavers Using a $500 per Quarter Earnings Threshold:Administrative Data Findings

State/Study

Exit Cohort Quarter Relative to Exit (%) Worked All Four Quarters (%) Ever Worked After Exit (%)
Q-1 Q1 Q2 Q3 Q4

Arizona1

1Q98  

Any Earnings

  47 53 51 52 51 32 70

$500 in Earnings

  34 44 43 45 44 25 62

District of Columbia1,2

4Q97  

Any Earnings

  63 62 n.a. 57 60 39 79

$500 in Earnings

  56 56 n.a. 53 54 35 74

Florida

2Q97  

Any Earnings

    50 51 53 54 31 71

$500 in Earnings

    42 43 45 46 n.a 63

Illinois

   

Any Earnings

  49 54 53 54 55 39 70

$500 in Earnings

  38 47 46 47 48 n.a. 63

Iowa1

2Q99  

Any Earnings

  44 57 42 39 38 25 69

$500 in Earnings

  32 49 36 33 32 20 61

South Carolina2

   

Any Earnings

  61 67 68 67 63 34 90

$500 in Earnings

  44 56 58 54 53 n.a. n.a.

1Data calculated from public use data files.
2Report employment data for all cases, not just for single-parent cases.
Source: See Appendix B for a complete listing of the leavers studies referenced.

The gaps for DC, Florida, Illinois, and Iowa are similar in size. This suggests that over 10 percent of working leavers really do not work much, at least in jobs covered by unemployment insurance.

One explanation for this discrepancy may be that leavers who only work a little end up returning to the welfare rolls. Therefore, we may expect to see higher employment rates and more employment growth for "continuous leavers"families that remain off welfare for at least an entire year after they exit. Figure III.3 and Table III.3, however, show there is little difference in employment between continuous leavers and leavers in general in three out of four studies that either report this information or have made data available that allow us to compute it.

Figure III.3:
Percent of Single-Parent Welfare Leavers Who Ever Worked in Year After Exit--Continuous Leavers v. All Leavers

Figure III.3: Percent of Single-Parent Welfare Leavers Who Ever Worked in Year After Exit--Continuous Leavers v. All Leavers

Notes: See table III.3 for more information.

Table III.3:
Employment Rates of Single- Parent Welfare Leavers-- Continuous Leavers vs. All Leavers: Administrative Data Findings

State/ Study

Exit Cohort Post Exit Quarter (%) Worked All Four Quarters (%) Ever Worked After Exit (%)
Q1 Q2 Q3 Q4

Arizona

1Q98  

All Leavers

  53 51 52 51 32 70

Continuous Leavers

  53 53 53 52 36 68

District of Columbia1

4Q97  

All Leavers

  62 66 57 60 39 79

Continuous Leavers2

  63 68 62 64 44 80

Iowa

2Q99  

All Leavers

  57 42 39 38 25 69

Continuous Leavers

  55 42 40 39 28 67

Washington

Oct. 1997  

All Leavers

  57 57 58 57 n.a. n.a.

Continuous Leavers

  57 57 57 56 n.a. n.a.

1Report employment data for all cases, not just for single-parent cases.
2Published data are incorrect.
Note: All data calculated from public use data files.
Source: See Appendix B for a complete listing of the leavers studies referenced.

In Arizona, for example, the fourth post-exit quarter employment rate for continuous leavers is 52 percent, only 2 percentage points higher than the rate for all leavers. Similarly, in Iowa and Washington, the fourth post-exit quarter employment rates are 39 and 56 percent, respectively, for continuous leavers and 38 and 57 percent for all leavers. In DC, however, continuous leavers have slightly higher employment rates than leavers in general: 64 versus 60 percent for the fourth post-exit quarter. Interestingly, although continuous leavers are slightly more likely to have worked in all four post-exit quarters than leavers in general, they are just about as likely to have ever worked.

That the employment rates for continuous leavers are so similar to those for all leavers is somewhat surprising. Indeed, these findings may suggest that continuous leavers are more likely to have some form of support other than work and welfare than leavers in general. This support may come from a spouse/partner or from another public program like SSI. Alternatively, it is possible that continuous leavers have simply disappeared from administrative records, for example, by leaving the state. Since they do not show up back on the state's TANF rolls or in its UI records, they will appear to be unemployed continuous leavers.

Finally, it is important to note that administrative data likely under-represent the amount of work performed by TANF leavers. First, leavers who work across state lines or move to another state entirely will not appear in a states UI system. Second, not all jobs are reported to a states UI systemfor example, leavers who are self-employed or who work in certain jobs in agriculture or in the federal government are not included in UI data systems. And lastly, leavers who are domestic service workers (like nannies or house cleaners) may not appear in the UI system because their employers fail to report them. Thus, some of the leavers who appear to have never worked in administrative data may actually be bringing in earnings in some form.

Because eleven jurisdictions report survey findings on current employment status of TANF leavers, it is possible to assess the extent to which work is under-reported in administrative records. The responses of leavers generally refer to employment about 6 months to a year after exit (see Appendix Table B). Table III.4 compares these self-reported employment rates with fourth quarter post-exit employment rates computed from administrative data. The District of Columbia, which used administrative data from the NDNH, reports fairly similar employment rates from both survey and administrative data sources.15 Interestingly, South Carolina, Washington state, and the Bay Area Study also find similar employment rates using the two data sources. Arizona, Georgia, Illinois, Missouri, and Cuyahoga County find employment rates that are 6 to 10 percentage points higher in survey data than in their state UI records, indicating some coverage gaps in administrative data. In Iowa and Massachusetts, however, the employment gap between survey and administrative data is alarmingly large: surveys find employment rates that are 20 or more percentage points higher than those computed from administrative data. Iowa, because of its large agriculture sector, may have more under-reporting in its UI system than other jurisdictions, but it is unlikely that "true" under-representation would be this much larger than the under-representation in other Midwestern states with heavy agriculture employment, such as Illinois and Missouri.

Table III.4:
Employment of Single-Parent Welfare Leavers: Comparison of Administrative and Survey Findings

State/Study

Exit Cohort Timing of Survey Post Exit Employment Rate (%)
Survey Data Administrative Data1

Arizona

1Q98 12 - 18 months 57 50 3

District of Columbia2

4Q97 and 4Q98 4 ~ 12 months 60 60

Georgia

Jan 1999-June 20004 ~ 6 months 69 59 5

Illinois

Dec. 1998 6 - 8 months 62 54

Iowa

2Q99 8 - 12 months 60 6 37 6

Massachusetts2

Dec 1998 - Mar 1999 ~ 10 months 71 51 7

Missouri2

4Q98 26 - 34 months 65 58

South Carolina2

Oct 1998 - Mar 1999 12 months 59 63

Washington

Oct. 1998 6 - 8 months 59 57

Cuyahoga Co.

3Q98 14-21 months 70 64

Bay Area

4Q98 6-12 months 57 55 8

1Based on employment rate from the 4th post-exit quarter.
2Employment data reported for all cases, not just single-parent cases.
3Data from report differ from revised data in public use data file. Revised fourth quarter employment is 51%. Arizona added 17 new cases to the data file one year after the report was published.
4The exit cohorts in DC and Georgia are different for administrative and survey data. In DC, the survey cohort exit period is 4Q98 while the administrative data period is 4Q97. In Georgia, the survey cohort exit period is Jan99- June 2000, while the administrative data period is 1Q98.
5Administrative employment rate based on 2nd post-exit quarter.
6Only survey respondents included.
7Administrative employment rate based on 3rd post-exit quarter.
8Based on employment rate from 9th month.
Source: See Appendix B for a complete listing of the leavers studies referenced.