Skip to main content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.


The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

A Cross-State Examination of Families Leaving Welfare: Findings from the ASPE-Funded Leavers Studies

Publication Date

Paper presented by ASPE
at the National Association for Welfare Research and Statistics (NAWRS)
40th Annual Workshop in Scottsdale, AZ
August 1, 2000
Revised Nov. 6, 2000(*)

Julia B. Isaacs and Matthew R. Lyon
Division of Data and Technical Analysis
Office of the Assistant Secretary
for Planning and Evaluation (ASPE), HHS


*.В  The views expressed in this paper are those of the authors and should not be construed as representing the views of the Department of Health and Human Services, or any office therein. Small revisions were made in November, to reflect improved information on the Medicaid and health insurance status of Missouri leavers.



As large numbers of recipients leave the welfare rolls, there is widespread interest in their circumstances: How are they faring without cash assistance? Are they working? Are they moving out of poverty? To what extent to do they return to welfare? To what extent do they continue to need and to receive assistance and supportive services through other programs?

In September 1998, the Office of the Assistant Secretary for Planning and Evaluation (ASPE), Department of Health and Human Services, awarded approximately $2.9 million in grants to study the outcomes of welfare reform for individuals and families who leave the TANF program, or who apply for TANF but are not enrolled. The grants were awarded to ten states and three large counties or consortia of counties under a May 1998 competitive grant announcement. Somewhat comparable studies were funded in Wisconsin, through the Institute for Research on Poverty (IRP), and in South Carolina, resulting in a total of fifteen studies of former recipients funded by ASPE in FY 1998. In these studies, researchers are using a combination of linked administrative data sets and surveys of former recipients to monitor the economic status and general well-being of families leaving welfare. Most studies are reporting on outcomes for at least two cohorts of leavers, an early cohort that left welfare in 1996 or 1997, and a later cohort that left in 1998 or 1999, after the transition of welfare from the Aid to Families with Dependent Children (AFDC) program to the Temporary Assistance for Needy Families (TANF) program. ASPE awarded an additional $2.6 million in welfare outcomes grants to states and large counties in FY 1999.

This paper presents preliminary outcomes for families leaving welfare, based on reports released by the HHS-funded studies as of July 2000. As of this date, eleven of the fifteen studies initiated in 1998 had released interim reports with administrative data findings for the early leavers and four of these eleven studies had also released final reports, which generally contain both administrative data and survey data findings for the later cohort of leavers.(1) An April 2000 ASPE staff paper summarized administrative data findings from the interim reports.(2) This paper draws heavily on the earlier summary, and like it, focuses on employment and earnings, returns to cash assistance, and program participation in Medicaid and food stamps. Findings from administrative data are supplemented with more detailed findings from the recently completed surveys of former recipients. In addition, the paper presents preliminary survey findings related to the household income and experiences of material hardship of former recipients, an important set of topics not measured in administrative data.

The survey data presented here are from only four states - Arizona, Illinois, Missouri, and Washington - and so may not reflect the experiences of welfare leavers more generally across the country. Additional HHS-funded grantees will be releasing final reports with survey data over the next twelve months, adding to our knowledge about the circumstances of families who have left cash welfare programs.

Cross-State Comparisons

The full set of studies reviewed in this paper are from a diverse set of states and counties, including: Arizona, the District of Columbia, Georgia, Illinois, Missouri, New York, Washington, Wisconsin, and Cuyahoga County in Ohio, Los Angeles County in California, and San Mateo County in California. Before examining outcomes across the various sites, it is important to acknowledge how these differences can be driven by differences in study methodologies, state policies, and underlying economic and demographic conditions.

Rather than impose a uniform study design across the various sites, ASPE funded a set of locally designed studies of welfare outcomes. While most studies examined both "early" and "late" cohorts of leavers, the studies reviewed in this paper include cohorts that exited as early as July 1995 and as late as December 1998 (see Table 1). Most studies used both administrative and survey data, but the study design varied across grantees. For example, survey sample size ranged from 300 cases to over 4,000 cases (with most falling between 600 and 1200 cases), and time of interview varied from 6 months to 36 months after exit from cash assistance. Most grantees conducted mixed mode surveys, consisting of telephone interviews of between 20 and 40 minutes for the majority of households, supplemented by in-person follow-up for those who could not be reached by telephone. Three of the four survey reports reviewed here achieved response rates of 72-75 percent; one study, however, had a response rate of only 51 percent. Finally, all the studies examined outcomes across a broad array of areas, including employment and earnings, program participation, health insurance, child care arrangements, household income, experiences of material hardship, and general child and family well-being. Certain studies, however, emphasized some of these areas more than others, and the exact wording of items varied across surveys.

Although each ASPE-funded grantee followed its own proposed study methodology, certain steps were taken by ASPE to promote comparability among the studies. Chief among these was developing consensus around a common definition of the "leaver" study population as "all cases that leave cash assistance for at least two months." This definition excludes cases that re-open within one or two months, because such cases are more likely closed due to administrative churning" than to true exits from welfare. A second step toward comparability was to encourage researchers to report administrative data outcomes for "single-parent" leavers, in addition to reporting them for the grantee's entire study population  which for some grantees consisted of single-parent leavers only, but for others included small percentages of "two-parent" cases, that is, families where both parents are receiving cash assistance, as shown in Table 1.(3) Grantees also agreed to focus the main study population on cases with one or two adult recipients, excluding closures of "child-only" cases.

Cross-state differences in TANF policies are likely to cause differences in the leaver populations across states. A state with a policy of sanctioning non-compliant families off TANF, for example, is likely to have a different leaver population than a state that imposes partial sanctions that do not result in a case closure. Also, the level of earnings which leads to case closure differs from state to state, depending on maximum benefits and earnings disregard policies. Work requirements, time limits, and differences in Medicaid and other program policies also contribute to the varied experiences of leavers in different parts of the country.

Finally, caution must be made in interpreting cross-site differences because of differences in the underlying economic, social and demographic conditions of the states and counties under study. Leaving welfare may be a different experience in an urban as opposed to a rural region, in an area of high unemployment as compared to low unemployment, for a married versus a never-married recipient, and for a non-English speaking recipient compared with a fluent English speaking recipient. Although this paper focuses primarily on aggregate findings for the "average" welfare leaver in a state, it is important to recognize that outcomes vary significantly across various subgroups, and that the demographic characteristics of welfare recipients and leavers differ significantly across the various study areas.

Employment and Earnings

Interestingly, despite the many differences in studies, many of the findings from the eleven studies are quite consistent. This is particularly true in the area of employment, the first set of outcomes examined in this paper. Employment outcomes as measured by administrative data in ten or eleven reports are examined first, followed by a comparison to employment measures in the smaller set of studies reporting survey data.

Employment Rates

About 45 to 65 percent of former TANF recipients were working after leaving TANF, according to administrative data from ten reports (see Table 2A). Most studies reported employment rates between 50 and 60 percent one quarter after exit, although employment rates were below 50 percent in one site (Los Angeles, based on preliminary data), and above 60 percent in two sites (Wisconsin and Georgia). There was a slight dip in employment rates in the second quarter after exit in most sites. In half the sites, this trend was reversed, and employment rates returned to initial levels by the fourth quarter after exit. Over the twelve-month period, some former recipients lost their jobs, while others found new employment, resulting in cumulative employment rates of 62 to 75 percent, measured as those who ever had earnings in the quarterly earnings databases within the first four quarters after exit. Only about 35 to 40 percent of leavers were employed in all four quarters, according to the four studies reporting this statistic.

A review of four states providing employment rates for two or more cohorts of leavers shows no clear trends over time (see Table 2B). When compared to the early cohort of leavers, the later cohort of leavers had higher employment rates in Washington, lower employment rates in Arizona, and similar employment rates in Illinois and Wisconsin.

Employment rates based on administrative data were defined in a similar manner across the ASPE-funded studies, as the percentage of leavers with positive earnings in the quarterly earnings records maintained by the state's unemployment insurance (UI) program. Although most jobs are covered by the state UI systems, some jobs are omitted, such as self-employment, employment in the military or federal government, certain agricultural employment, informal employment, and jobs over state boundaries. These employment rates, therefore, are likely to be under-estimates of employment rates as compared to employment information gathered through surveys of former recipients.

In fact, former recipients reported employment rates of between 1 and 10 percentage points higher than the rates based on administrative data, according to survey data from four reports (see Table 2C).(4) Across the four studies, self-reported employment rates varied from 58 to 65 percent. This represents employment at the time of interview, a single point in time. Respondents also were asked to report whether they were employed at any point since exit, and the vast majority - 85 to 90 percent - told interviewers that they had worked at least once between exit and time of interview (a time period ranging from 6 months to over 2 years, depending on the state's study design).

Self-reported cumulative employment rates were higher than employment rates observed over four quarters of administrative data. In Illinois, for example, 85 percent of the survey sample reported at least some employment in the 6 to 8 months since exit, even though only 70 percent of the universe of leavers had any earnings reported over four quarters worth of administrative data. While most of this difference may be due to employment that is not captured in the earnings records of the unemployment insurance system, it also is possible that some respondents overstated their rate of employment since exit, not wanting to admit that they had not held any job over that period. It also is possible that the Illinois survey respondents, who represented only 51 percent of those in the original survey sample, did have slightly higher employment rates than the universe of leavers. A comparison of respondents and non-respondents indicated that respondents differed in some respects, including having slightly higher education levels, more income reported to the agency at the time of exiting welfare, and a higher likelihood that their case was closed because of earned income.

Although only one of the four survey reports, Illinois, reported on continuous employment, the survey findings - 37 percent continuously employed throughout the six-to-eight months since exit - seem consistent with the administrative data findings from three other states of 35 to 40 percent employed in all four quarters after exit. In Missouri, survey respondents reported working a mean of 20 months and a median of 25 months over the 26-34 months since exit, again indicating that many former recipients experienced spells of unemployment as well as periods of employment.

Three of the four surveys asked about employment of other household members, important information that is not usually available from administrative data. Illinois and Missouri reported that between 72 and 80 percent of leaver households had at least one employed adult. In these two states, 60 to 65 percent of adult leavers were employed, 10 to 15 percent were unemployed but living in households where others were employed, and 20 to 30 percent lived in households with no earned income.(5)

Evidence from the leavers studies suggests that about three-fifths of individuals leaving welfare were employed at any point in time (with slightly higher rates according to the survey data, and slightly lower rates when examining administrative data only). Employment rates rose to over 80 percent when measured over extended periods of time or across the entire household. However, former recipients experienced a fair amount of job turnover and spells of unemployment, which calls into question their ability to support themselves and their families on earned income alone. This concern is reinforced by information about the level of their earnings.


Mean quarterly earnings in the quarter immediately following exit from TANF ranged from about $2,200 to over $3,400, according to administrative data from the Unemployment Insurance system for leavers in nine study areas (see Table 3A). Median quarterly earnings, that is the midpoint in the earnings distribution of employed leavers, were slightly lower, ranging from approximately $2,000 to about $3,250. In every location, earnings steadily rose over the course of the year following exit. Earnings were highest (over $3,000 per quarter) in New York and the two California counties of San Mateo and Los Angeles.

These averages were calculated over those leavers who found employment (i.e., leavers with positive earnings in a quarter). Grantees generally reported earnings in current or nominal dollars, except for San Mateo County, which presented earnings in November 1998 dollars.

The UI system collects data on aggregate quarterly earnings, without providing underlying information about actual wages, hours worked, or months worked in the quarter. Much of this additional detail was collected in surveys of former recipients, as shown in Table 3B. Hourly wages averaged about $7.50 to $7.70 in Arizona and Washington, with median hourly wages somewhat lower. Former recipients with jobs worked an average of 35 to 39 hours per week across the four studies, and a median of 40 hours per week.

Survey respondents who were employed reported monthly earnings that averaged about $1,000-$1,100, ranging from $958 in Washington to $1,101 in Missouri (see Table 3B). Illinois reported a slightly higher estimate of $281 per week. If leavers worked three months continuously, these earnings levels would translate into $3,000-$3,600 per quarter, somewhat higher than the $2,500-$3,000 quarterly earnings observed in administrative data in these four studies. Some of this difference is probably explained by job loss and turnover; one would expect quarterly earnings to be less than three times monthly earnings for individuals who do not work continuously over a three-month period. In addition, the self-reported monthly earnings estimates include earnings from jobs that are not covered by the UI system and so are not reported in the administrative data. Taxes are a third factor complicating the comparison of survey and administrative data on earnings. Whereas quarterly earnings from t he UI system represent gross earnings, only two of the surveys asked about gross earnings, before taxes; the other two surveys (in Arizona and Illinois) asked about take-home pay, net of taxes.

The mean monthly and quarterly earnings observed in these studies suggest that annual earnings of employed welfare leavers were quite low; averaging no more than $10,000-$13,000, despite the fact that many former recipients had jobs involving full-time work. With these levels of earnings, many former recipients are in need of continued supportive services, such as food stamps, Medicaid and child care. The extent to which families leaving welfare actually receive supportive services was one of the questions explored by all the ASPE-funded leavers studies. In addition, the studies examined the extent to which job loss and low wages caused families to return to cash assistance, as shown in the section below.

Program Participation

Families leaving welfare in the period under study (1996 through 1998) were exiting from cash assistance (AFDC or TANF) before they hit Federal or state time limits on benefit receipt. Thus they had the option of returning to cash assistance, as needed. Furthermore, they often retained eligibility for Medicaid, food stamps, child care subsidies, child support enforcement services, child nutrition programs, and other federal benefits and programs that are not subject to the five-year time limits. The section below examines the extent to which former recipients return to cash assistance, as well as the extent to which they continue to participate in Medicaid and food stamps, two major Federal assistance programs for low-income families.

Returns to TANF

A fair number of families leaving welfare returned quite quickly, within one quarter, according to administrative data from eight reports. The percentage receiving TANF one quarter, or three months, after exit ranged from 5 to 20 percent of leavers, as shown in Table 4A. Note that leavers in these studies are defined as those who remain off welfare for at least two months; the recidivism rates would be even higher if the study population included those with exits of one month or less.

Over the next three months, an additional 4 to 7 percent of leavers returned in most states, bringing the total proportion of former recipients receiving AFDC/TANF to between 10 and 28 percent at two quarters after exit. There was a small increase over the next six months, resulting in rates of welfare receipt of between 12 to 29 percent one year after exit. Because some people come back to welfare for a few months and then leave again, the proportion that ever returned for at least one month over the first twelve months after exit was somewhat higher, ranging from 23 to 35 percent.

Some of the variation in receipt of cash assistance is due to measurement differences. Most importantly, some grantees measured program participation by month, while others measured it as receipt over any of three months in a quarter. Quarterly measures are likely to result in higher participation rates because of the longer time period for observing benefit receipt.

Recidivism rates appear somewhat lower in the 1998 cohorts as compared with the earlier cohorts, in each of the four studies reporting such data (see (Table 4B). The change was relatively small; in Wisconsin, for example, 25 percent of the 1998 cohort returned to TANF at least once in the first year after exit, compared to 29 percent of the 1995 cohort. In two states, Illinois and Washington, the 1998 cohorts had higher employment rates as well as lower recidivism rates. In Arizona and Wisconsin, however, the reductions in recidivism occurred despite flat or declining employment rates across the early and late cohorts of leavers.

The survey data on TANF receipt are fairly similar to the administrative data, with 14-19 percent of respondents reporting being back on cash assistance at the time of interview (see Table 4C). When asked if they had ever been back on welfare, however, survey respondents reported lower rates of receipt than appears true from the administrative data. This is particularly true in Missouri, where 31 percent of respondents reported ever receiving welfare over the two and a half years since exit, compared to a 44 percent rate based on an analysis of administrative data on the survey respondents. Some of this difference may be due to stigma about welfare receipt and respondents' reluctance to report that they were back on assistance.

Of those who returned to TANF, at least one half did so because of job-related reasons, such as job loss, or decreases in work hours or wages, according to survey respondents across three of the states (see Table 4C). Other common reasons for returning to TANF were divorce or separation from a partner, pregnancy or birth of a new child, re-compliance with program regulations, loss of other income, problems with child care, and problems with health or medical benefits. This latter issue touches on one of the research questions explored by leavers studies, that is, the extent to which families leaving cash assistance have health insurance coverage through Medicaid or through private health insurance.

Medicaid and Health Insurance

In contrast to the employment, earnings, and recidivism findings, which were fairly consistent across studies, rates of Medicaid enrollment for former AFDC/TANF recipients varied dramatically across grantees. Three months after exit, the Medicaid enrollment rate for the early cohorts of adult leavers across the nine studies that reported this measure ranged from 24 to 76 percent, with most studies reporting rates between 35 and 60 percent (see Table 5A.)

In the year after exit from AFDC/TANF, participation in Medicaid declined over time in some study areas, dropping by as much as 10 to 20 percentage points for adult leavers between the first and fourth quarters after exit. Rates remained relatively stable, however, in three of the studies. A substantially higher percentage of leavers were ever enrolled in Medicaid than were enrolled in each of the four quarters. In Arizona, for example, administrative data show that only about 50 percent of leavers were enrolled in Medicaid in any particular month, but 85 percent were enrolled at some point in the first year after exit.

Enrollment rates were even lower among "continuous leavers" that is, leavers who did not return to cash assistance (see Table 5B). In Illinois, for example, 29 percent of continuous leavers were enrolled in Medicaid four quarters after exit, as compared with 40 percent of all leavers. The main reason for lower enrollment among continuous leavers is that this subgroup excludes the leavers who return to TANF, most of whom are re-enrolled in Medicaid at the time of their return to cash assistance. In addition, it may be true that continuous leavers have better jobs and higher rates of private health insurance coverage than other leavers, but such information is not available through administrative data.

A more complete picture of health insurance coverage is provided through survey data. Data from four studies suggest that one-third to one-half of adult leavers (33 to 53 percent) were covered by Medicaid at the time of interview and one-fifth to one-third were covered by employer or other health insurance (20 to 34 percent), leaving one-third (26 to 41 percent) of former adult recipients without any health insurance (see Table 5C). Missouri leavers had the highest enrollment rates in employer and other health insurance (32 percent) and the lowest enrollment in Medicaid (33 percent). The other three states reported little difference in the percentage covered by employer or other non-Medicaid health insurance, and so the differences in uninsured rates were driven by the differences in Medicaid enrollment rates. There may be more variation in private insurance coverage patterns when data are available from more states.

As one would expect, the two states interviewing leavers six to eight months after exit found higher Medicaid enrollment rates (47 to 53 percent) than those interviewing leavers one to two years after exit (33 to 39 percent). Note that the states did not report private health insurance for the subgroup of continuous leavers, and so we do not know whether they were more likely to have private health insurance than other leavers.

Medicaid coverage for children of leavers was considerably higher than coverage for adults, according to survey data from four studies. Enrollment rates for children of leavers were from 6 to 35 percentage points higher than rates for adults in these four studies, with between one-half to two-thirds (51 to 68 percent) of leavers reporting Medicaid coverage for their children. (All four studies treated coverage under the State Children's Health Insurance Programs (SCHIP) as Medicaid coverage.) Administrative data on Medicaid enrollment rates for children of leavers did not show as clear a pattern, possibly because only three administrative studies reported rates for children and adults. One of these studies, Missouri, showed much higher coverage for children than for adults, consistent with survey data from the same state, but the other two sites (San Mateo and New York) showed surprisingly little difference between adults and children, at least according to the available administrative data shown in Table 5A.

About one-fifth (20 to 23 percent) of leavers reported that their children had health insurance coverage other than Medicaid. This was similar to the percentage of adults with such coverage, but the sponsorship of the coverage was different, at least according to data from two surveys. Children were somewhat less likely than adults to have such insurance in the form of employer-sponsored insurance, and more likely than adults to have it in the form of other insurance, including coverage through absent parents, the Indian Health Service, or other health plans. Finally, between 8 and 29 percent of leavers reported that their children had no health insurance. As the states' reports observed, this finding is troubling, given the fact that the vast majority of these children should be eligible for health insurance under expanded Medicaid coverage and/or the State Child Health Insurance Programs (SCHIP).

Although many of the surveys asked the leavers why they were not covered by Medicaid, it is very difficult to summarize these responses, because of differences in how the questions were phrased and the responses reported. For example, many surveys asked about the extent to which welfare agencies were informing families of their potential eligibility for Medicaid. In Arizona, slightly more than half (56 percent) of leavers said they received information from the agency that they might still be eligible for Medicaid (data not shown). In Illinois, only one third of those without any Medicaid coverage said they had received information from the welfare office about potential eligibility, but note that this percentage is just among those without Medicaid coverage, not among all leavers. The Washington survey did not ask about outreach specifically, but did report that more than three-fourths (78 percent) of leavers were aware of possible continued Medicaid after leaving welfare.

A comparison of survey and administrative data reveals some apparent discrepancies between respondent perceptions of their Medicaid coverage and the program records in the Medicaid computer systems. In Washington state, researchers analyzed administrative records of the survey respondents, and found that 61 percent were found in the Medicaid database, whereas only 53 percent had said they had Medicaid coverage. The Illinois report also found higher Medicaid enrollment rates from administrative records as compared with survey data. In Arizona, however, the rates were the same from the two data sources, as they were for children in Missouri. Differences in the visibility of state Medicaid programs, the interface in welfare and Medicaid computer systems, and the wording of the survey questions about Medicaid enrollment (6) might explain some of these cross-state differences.

Some of the differences across studies in reported Medicaid enrollment rates can be explained by methodological issues, including type of data (survey vs. administrative), unit of analysis (adult, child, or any member of household), time period (monthly vs. quarterly), and sub-population (all leavers vs. continuous leavers). In fact, New York reported four different Medicaid enrollment rates, ranging from 34 to 45 percent, to show the effects of differences in unit of analysis and time period, as shown in Table 5A. Such methodological issues, however, account for only some of the wide variation in Medicaid enrollment rates.

Many of the remaining differences are likely to be due to differences in TANF policies, Medicaid eligibility rules and application procedures, and the state and local administrative practices that come into play when families transition off of cash assistance. As these policies change over time, there may be changes in Medicaid enrollment rates. The data shown in Table 5D suggest some changes over time in Medicaid enrollment among leavers; the trend is different, however, in each of the three states reporting on early and late cohorts. Enrollment in Medicaid increased between early and late cohorts in Washington, but decreased in Arizona, with little apparent change in Illinois. Preliminary findings of low Medicaid enrollment rates among leavers have prompted Federal and state efforts to review Medicaid and TANF eligibility and enrollment systems and procedures to ensure that families are not incorrectly denied Medicaid benefits when they leave or are diverted from cash assistance . In fact, some states have mentioned in their leavers' reports that they have made changes in policies or administrative processes that should result in higher Medicaid enrollment rates among future cohorts of leavers.

Food Stamps

The Food Stamp Program, like Medicaid, provides benefits to most welfare families along with their cash benefits, but eligibility is not tied to welfare receipt and many families remain eligible after they leave cash assistance. Families remain eligible if their household income is below the Food Stamp eligibility limits (130 percent of the Federal poverty guidelines) and they meet the work and other requirements of the Food Stamp Program. The majority of the ASPE-funded leavers studies have found that between one-third and one-half of former recipients of cash assistance received food stamps immediately after exit, as shown in Table 6A. This same range (33 to 50 percent) was reported by two surveys asking about food stamp receipt six to eight months after exit. By one year after exit, food stamp participation rates generally fell to between 33 and 40 percent, according to the longitudinal administrative data. The percentage of leavers participating in the Food Stamp program was lower than the percentage enrolled in Medicaid across almost all the studies.

There were, however, a few exceptions to the general trends. The San Mateo study found only 10 to 15 percent of former TANF recipients in the food stamp administrative data base in any quarter. Also note that food stamp receipt in Missouri was reported as high as 47 percent two and a half years after exit, according to survey data.

As with Medicaid, participation was much lower among those who did not return to TANF. Only 20 to 30 percent of continuous leavers participated in the Food Stamp program, according to the three studies reporting such data (see Table 6B).

Participation rates in the Food Stamp program are more noticeably affected by the unit of measurement than other program participation rates, probably reflecting the short-term nature of food stamp participation for many households. Participation rates one year after exit, for example, were 21 to 35 percent when measured monthly, compared to rates of 26 to 40 percent among grantees measuring program participation across a three-month quarter. The exception is San Mateo, which measured low rates despite observing participation over a three-month quarter. Possible explanations for the low measure in San Mateo include: higher incomes among the leaver population (because eligibility limits and maximum benefits are higher than average in California), less active food stamp outreach, or technical measurement challenges with the administrative data.

It is difficult to interpret participation rates on food stamps without more information on household poverty status and food insecurity. Low rates of food stamp participation can be viewed positively, if they indicate that families leaving TANF are moving out of poverty and out of the population that is eligible for food stamps. Low rates are viewed with concern, however, when there are indications that families leaving cash assistance remain poor and eligible for food stamps, but are not receiving the nutritional benefits to which they are entitled. Concern is heightened if there is evidence that families are experiencing food shortages as a result of insufficient financial resources. The concluding section of this paper examines these critical issues of household income, poverty, and material hardship.

Poverty and Material Hardship

Household Income and Poverty Status

Total household income cannot be measured from administrative data, which is limited to information on the earnings and government benefits received by the adult leaver. Thus the household incomes reported here are limited to the four studies with survey data. Across these four studies, total monthly household cash income averaged from about $1,050 (Illinois) to over $1,400 (Missouri), as shown in Table 7A. Median household incomes were even lower, less than $900 in Illinois and less than $1200 in Missouri. These estimates are not much higher than the $1,000-$1,100 average earnings estimates reported above, but recall that the earnings estimates were averaged across the subset of households with earnings. In contrast, the household income statistics are averaged across all leavers, including those with no earnings, or even no income.

More information on the distribution of income is provided for leavers in Illinois and Missouri. Very few leavers (2 to 4 percent) reported $0 in household income. About half the leavers reported incomes of between $500 and $1,500 per month. Only 11 percent of Illinois leavers and 20 percent of Missouri leavers had household income of more than $2,000 per month ($24,000 in annual terms). This included 8 percent with incomes of more than $3,000 per month in Missouri, where income was measured two and a half years after exit from welfare.

These income distributions suggest that many former recipients remain poor after leaving welfare. The state reports from Washington and Missouri actually calculated a poverty rate for former recipients, by multiplying monthly incomes by twelve and comparing these annual incomes to the Federal poverty thresholds for different family sizes. The estimated poverty rate of former recipients was the same - 58 percent - in both states. This poverty rate was measured 6 to 8 months after exit in Washington and about two and a half years after exit in Missouri. In contrast, the poverty rate for families remaining on TANF was 83 percent, according to calculations in Washington state.

More detailed analyses of poverty status in Missouri showed that many leavers with income above poverty were still near-poor; a total of 89 percent of former recipients had incomes below 185 percent of the poverty threshold. These data on poverty status suggest that a large proportion of former welfare recipients remained eligible for child nutrition programs (eligibility at 185 percent of poverty), the Food Stamp program (eligibility at 130 percent of poverty), and the Medicaid program (eligibility varies by state and reason for exiting cash assistance).

All four studies provided some information about the sources, as well as the aggregate amounts, of household income. Earnings strongly drove household income. In both Illinois and Missouri, for example, median household income was about $400 per month in households without any earners, $1,000-$1,100 per month in households with one earner, and $2,000-$2,300 in households with two earners. In fact, earnings accounted for 70 to 85 percent of aggregate household cash income, according to data analyses from Washington, Arizona, and Missouri (see Table 7B). This included about 45-50 percent from the adult leaving TANF, and 20-40 percent from others in the household. Other major contributors to aggregate household income included TANF, child support, and Supplemental Security Income (SSI), each of which accounted for between 3 and 8 percent of total household income across the three states.

The percentage of total household income contributed by each source depends upon both the percentage of households with the income source and the amount of income from that source, as shown in Table 7C. The effects of these two factors can be seen by contrasting the examples of child support and SSI income. In the case of child support, close to one-fourth of leavers have some child support income, but the average amount of income for those with any child support income was only between roughly $150 and $300. In contrast, fewer households had SSI benefits (4 percent in Washington and 12 percent in Missouri and Illinois), but the average amount received by these households was larger, $400-$500. In the end, both child support and SSI contributed about the same percentage (3 to 7 percent) to total household income averaged across all leavers.

Half (54 percent) of leavers relied on one source of income, according to the report on Missouri leavers (data not shown). This one source of income was generally earnings. One third (34 percent) of former recipients in Missouri relied on two sources of income, with the most common package being a combination of earnings and child support. Finally 10 percent of former recipients received income from three or more sources, while 2 percent had income from no source, e.g, zero income in the past month.

Before concluding this discussion of household income and poverty status, it is important to note the difficulty of measuring household income accurately and differences across the questionnaires. The Illinois questionnaire placed relatively little emphasis on the collection of household income, simply asking respondents at the conclusion of the survey to estimate total household income in the prior month. (Earlier questions had asked about receipt of TANF, SSI, child support, etc., but without dollar amounts). The Arizona and Washington surveys went into more detail, asking about household income from a variety of sources, with separate questions about the presence and amount of income from own earnings, others' incomes, SSI, child support, etc. Finally, the Missouri survey focused the most attention on household income, asking a long set of questions about various types of income received by the adult respondent, and separately, income of others in the household. After the complete cycle of questions, the Missouri interviewer shared the calculated total income with the respondent, to confirm whether the total calculated household income seemed right to the respondent.

Past research on the collection of household income information suggests that more detailed and probing questions will reduce the problem of under-reporting of household income. It is possible, therefore, that some of the $400 differential between Missouri and Illinois may be due to differences in income reporting, as well as possible true differences in the underlying income distributions. In fact, earnings were a smaller percentage of aggregate income (69 percent) in Missouri than in other states (83-85 percent in Arizona and Washington), suggesting the possibility that the Missouri questionnaire was more successful in capturing other sources of income. It also is true, however, that average earnings in Missouri were lower than in other states, at least according to administrative data reported in Table 2B, and so the lower contribution of earnings to household income in Missouri reflects labor markets in Missouri as well as methodological differences in income reporting.

Partly because of the complexity of accurately measuring household income, most of the leavers studies also asked questions about the direct effects of financial hardship on material well-being. These questions deal with such issues as experiences of food shortages and housing problems, as discussed below.

Material Hardship

Approximately one-fourth of former welfare recipients reported that they did not have enough to eat, or they cut meal sizes or skipped meals, in the period since leaving welfare, as shown in Table 8. Each of the three studies reporting these data attempted to compare the experience of leavers with the experience of welfare recipients. The most reliable comparison is in Washington, where a sample of ongoing recipients was asked the identical set of question as the sample of leavers, regarding experiences of food shortages over the last six months. The result was that leavers reported somewhat more experiences with food shortages than ongoing recipients; for example, 23 percent of leavers reported skipping a meal some time or often, compared with 18 percent of ongoing recipients, a 5 percentage point difference that is statistically significant. In Illinois and Arizona, where leavers were asked to compare the time period since exit with experiences in the six months prior to exit, t he comparison is hampered by challenges with respondent recall and observation periods of varying lengths. Still, these data suggest that food shortages were less common after exit than before exit in Arizona, and of the same frequency before and after exit in Illinois.

The Washington survey included some questions that probed at more severe signs of hunger among leaver households, including instances where children as well as adults were affected by food shortages. When asked how often both adults and children in the household skipped meals, 4 percent of leavers (and a similar percentage - 5 percent - of ongoing recipients) reported that this occurred some time or often in the last six months. In addition, 14 percent of Washington leavers reported that adults went without food all day at least once in the past six months. The comparable percentages were 9 percent for ongoing recipients, 1 percent for children of leavers and 2 percent for children of ongoing recipients.

Former recipients also reported problems with housing arrangements, although these were not as common as the overall food shortages reported above. The most common problems were loss of utilities (12 to 14 percent of former recipients across the three studies) and being forced to move (13 to 17 percent of former recipients according to two studies). Less often, former recipients were evicted (7 percent in Washington) or went to a homeless shelter (1 to 4 percent across the three studies). An estimated 8 percent of former recipients in Illinois and Arizona reported that their children were forced to live elsewhere, and 3 percent of leavers in Washington reported that their children went into foster care.

Comparisons between former and current recipients in Washington suggest that while both groups experienced similar levels of loss of utilities, former recipients were more likely than ongoing recipients to experience the more extreme conditions of being evicted or having no place to live more than once over a six-month period. The comparisons in Illinois and Arizona suggest the opposite, that is, that leavers experienced fewer housing problems after their exit from welfare than they had experienced while on welfare. Again, it is not clear if this reflects a true difference across states (perhaps tied to the adequacy of welfare benefits) or if the Illinois and Arizona surveys were less accurate in measuring hardships while families were receiving welfare. None of the states reported that separations of children from the family occurred more or less frequently after exit than while the family lived on welfare.

Finally, close to one-third (31 percent) of leavers in Illinois, and one-fourth (24 percent) in Arizona reported that since leaving welfare, either they or someone in their household was unable to get needed medical attention because they could not afford it. This may reflect the fact that 25-40 percent of respondents in these two states were uninsured. Somewhat smaller percentages of respondents (26 percent in Illinois and 14 percent in Arizona) reported an inability to get needed medical attention during the last six months of welfare receipt, a time period when almost all respondents all had health coverage through the Medicaid program.


The eleven administrative data reports and four survey data reports reviewed in this paper provide important information about individuals and families leaving welfare. The studies consistently found that about three-fifths of former recipients were working; data from two states suggested that another fifth lived in a household where someone else is employed. Earnings levels were low, however, resulting in low household incomes and estimated poverty rates of close to 60 percent in two different states.

In most states, slightly over one-fourth (25 to 30 percent) of former recipients returned to cash assistance at least once over the first twelve months after exit. Usage of Medicaid was much less consistent across the states, with enrollment rates varying dramatically, from 25 to 75 percent of former recipients. Survey data indicate that significant percentages of leavers were uninsured and in some cases unable to afford needed medical attention. Leavers were somewhat less likely to participate in the Food Stamp program than in Medicaid, with between one third and two thirds of leavers receiving food stamps immediately after exit. Data from three states suggested that approximately one-fourth of former recipients experienced some degree of food shortages. Extreme food shortages, however, appeared quite rare (1 percent of leavers in one state, for example, reported that children in their household went a full day without food in the past six months), and less than 5 percent of recipients in three states reported turning to a homeless shelter. From the available data, it is not clear whether families who left welfare experienced greater or less material hardship than families who remained on welfare.

While the descriptive statistics summarized here provide an important overview of the well-being of individuals and families leaving welfare, they do not make full use of the rich administrative and survey data collected by states and counties under ASPE's welfare outcomes grants. Those interested in learning more about families leaving welfare can be directed to three additional sources of information. First, the individual state and county reports (see attached list of reports) cover a more comprehensive array of topics, and include more analyses of differences across subgroups, than can be covered in a synthesis paper of this length. Second, researchers at the Urban Institute have recently started writing a comprehensive synthesis of findings from ASPE-funded leavers studies; a short paper is expected this fall with the full report by next summer. Finally, ASPE is working collaboratively with the grantees and a technical assistance contractor (ORC Macro International), to make to make the grantees' welfare outcomes data files available to researchers for secondary analysis.(7) Through these various means, the rich and painstakingly collected data gathered with the support of the HHS Welfare Outcomes grants are being made available to researchers, policy-makers, program administrators and advocates to help inform future policy decisions and program implementation.

List of ASPE-Funded Welfare Outcomes Reports Included in This Review

Many of these reports can be accessed at:


Westra, K. & Routley, J. (January 2000). Arizona Cash Assistance Exit Study: First Quarter 1998 Final Report. Arizona Department of Economic Security.

Ibid., (July 1999). Arizona Cash Assistance Exit Study: Cases Exiting Fourth Quarter 1996. Arizona Department of Economic Security.

Cuyahoga County

Coulton, C. & Verma, N. (May 1999). Employment and Return to Public Assistance Among Single, Female Headed Families Leaving AFDC in Third Quarter, 1996, Cuyahoga County, Ohio. Prepared for Cuyahoga Work and Training.

District of Columbia

Loprest, P., & Acs, G. (February 2000). The Status of TANF Leavers in the District of Columbia: Interim Report. The Urban Institute.


Foster, E. M. (April 1999). "Amended Quarterly Progress Report: Outcomes for Single Parent Leavers by Cohort Quarter." Georgia State University.


Julnes, G., Halter, A., Anderson, S., Frost-Kumpf, L., Schuldt, R., Staskon, F., and Ferrara, B. (forthcoming). Illinois Study of Former TANF Clients, Final Report. Institute for Public Affairs, University of Illinois at Springfield and School of Social Work, University of Illinois at Urbana-Champaign.

Ibid., (March 2000). Illinois Study of Former TANF Clients, Interim Report: Analysis of Administrative Data. Institute for Public Affairs, University of Illinois at Springfield and School of Social Work, University of Illinois at Urbana-Champaign.

Los Angeles

Verma, N. & Goldman, B. (January 2000). "Los Angeles County Post-TANF Tracking Project: Quarterly Progress Report." Manpower Demonstration Research Corporation. Data are preliminary.


Midwest Research Institute (June 2000), "Chapter 1 - Employment and Earnings of Former AFDC Recipients in Missouri, Chapter 2 - Household Income and Poverty, Chapter 3 - The Continuing Use of Assistance by Former Missouri AFDC Recipients, and Chapter 4 - Insecurity Among Former AFDC Recipients," Interim Reports for the Missouri Department of Social Services.

Ryan, S., Theilbar, M., Choi, S., Qu, J., Deng, M., and Ellebracht, L. (April 1999). Preliminary Outcomes for 1996 Fourth Quarter AFDC Leavers: First Interim Report University of Missouri-Columbia. (See also Preliminary Outcomes for 1996 Fourth Quarter AFDC Leavers: Revised Interim Report (September 1999). Most data reported here are from the first interim report, which reported results separately for single parents.)

Ryan, S., (October 2000). Phase I. Part C. Demographic Sub-Group Analysis. Includes recalculated estimates of Medicaid enrollment among AFDC leavers.

New York

Rockefeller Institute, New York State Office of Temporary and Disability Assistance, and the New York State Department of Labor (December 1999). After Welfare: A Study of Work and Benefit Use After Case Closing. Revised Interim Report. See also Interim Report issued in July 1999.

San Mateo County

Moses, A. & Macuso, D. C. (May 1999). Examining Circumstances of Individuals and Families Who Leave TANF: Assessing the Validity of Administrative Data. SPHERE Institute.


Forgarty, D. and Kraley, S., (March 2000). A Study of Washington State TANF Leavers and TANF Recipients, Findings from the Administrative Data and the Telephone Survey: Summary Report. Washington Department of Social and Health Services.

Du, J., with Fogarty, D., Hopps, D., and Hu, J.(February 2000), A Study of Washington State TANF Leavers and TANF Recipients. Findings from the April-June 1999 Telephone Survey. Final Report. Washington Department of Social and Health Services.

Ahn, J., Fogarty, D., Kraley, S., Lai, F., and Deppman L. (February 2000). A Study of Washington State TANF Departures and Welfare Reform. Welfare Reform and Findings from Administrative Data. Final Report. Washington Department of Social and Health Services.


Cancian, M., Haveman, R., Meyer, D., and Wolfe, B. (May 2000). Before and After TANF: The Economic Well-being of Women Leaving Welfare. IRP Special Report #77. Institute for Research on Poverty, University of Wisconsin-Madison.

Cancian, M., Haveman, R., Kaplan, T., and Wolfe, B. (January 1999). Post-Exit Earnings and Benefit Receipt Among Those Who Left AFDC in Wisconsin. Institute for Research on Poverty, University of Wisconsin-Madison.
(NOTE: Employment and program participation rates in this summary are drawn from a supplemental table provided to ASPE in June 1999. Adjustments made to the rates in the original study made them less comparable to rates in other studies.)


1.  For convenience, the administrative data reports are referred to as "interim" reports and the survey data reports are referred to as "final" reports. In fact, one of the eleven administrative reports is actually a final report (the Wisconsin IRP study) and one of the four survey reports is not, in fact, a final report (it consists of four chapters of a forthcoming report on the Missouri study).

2.  The staff paper, " Summary of Research on Welfare Outcomes Funded by ASPE: Administrative Data Findings from Interim Reports," was revised several times, as more interim reports became available (HHS/ASPE, August 1999; HHS/ASPE, October 1999; HHS/ASPE, April 2000).

3.  In most cases, there was little difference in outcomes between "single-parent leavers" and "all leavers," because single parent cases comprised the vast majority of leaver cohort. Even so, all outcomes reported in this paper are for single-parent leavers, except where noted otherwise.

4.  Washington, the state with only a 1 percentage point differential between self-reported employment and administrative data on employment, used two administrative sources to calculate employment rates. In addition to UI wages, the calculation of employment rates included wages reported to the welfare/food stamp/Medicaid system. Employment rates based on administrative data reports would have been about 5 percentage points lower if the administrative data had been limited to UI wage data.

5.  Washington, the third study collecting data on household member employment, reported that 20 percent of leaver households had earnings from other household members. This information is not enough to compute a "household employment rate" comparable to the rates in the Illinois and Missouri reports, however, because the Washington report did not specify how many of these Washington households had earnings from both leavers and other adults.

6.  As reported in an earlier version of this paper, preliminary reports from Missouri suggested that Medicaid enrollment was much lower in administrative data (14 percent of leavers in eighth quarter) than in survey data (33 percent of respondents reporting they were enrolled in Medicaid and 38 percent reporting either they or another adult in the household was enrolled). This apparent discrepancy was largely resolved when the administrative data were re-analyzed to include Medicaid eligibility of families returning to AFDC/TANF, resulting in a revised estimate of 26 percent of leavers enrolled in Medicaid in the eighth quarter after exit.

7.  More information on how to secure access to these data files can be secured by contacting the author at or checking the ASPE-sponsored web page on Leavers and Diversion Studies after fall 2000.

Product Type
Location- & Geography-Based Data
State Data