The enormous growth in welfare caseloads between 1960 and 1976 was primarily due to administrative changes that increased participation among eligible families.  The 34 percent rise from 1989 to 1994 was caused by an economic downturn, an increase in out-of-wedlock births, Medicaid expansions, and immigration.  The 36 percent decline after 1994 is discussed below.
These are the broad elements of the new world of welfare reform.  Each is discussed below.  Although there was no new job training money in the 1996 welfare reform law, the Balanced Budget Act of 1997 provided substantial new funding for the employment of long-term welfare recipients, as discussed later.
Immediate work or participation in work-related activities is required in 25 states (34 percent of the national caseload).   In about 35 states (50 percent of the national caseload), noncompliance with work, self-improvement, or child support requirements can lead to a complete termination of assistance (either immediately or progressively).
A time limit of two years or less is imposed in 18 states (30 percent of the national caseload).  Many are "periodic," allowing a specified period of benefit receipt within a longer time frame.  For example, Florida allows 24 months of receipt within a 60-month period.  Regardless of the time limit established, 39 states (86 percent of the national caseload) have exemptions or extensions.
Most states have expanded the amount of earned income that recipients can keep.  A total of 22 states (over half the national caseload) have adopted earnings disregards of 50 percent or more.  Past AFDC policy effectively imposed a 100 percent marginal tax rate after four months of work.  As we will see, more generous earnings disregards, when combined with a work requirement and a generous EITC, have made welfare less attractive and work more attractive.
Most states have rules intended to foster responsible parental behavior.  There are three major types: "family caps," to discourage additional childbearing while on welfare by not raising the welfare grant (20 states); and financial penalties for not sending children to school (18 states) or for not having them immunized (19 states).
The Balanced Budget Act of 1997 authorized grants to states and local communities to create new job opportunities for the hardest to employ welfare recipients.  The Taxpayer Relief Act of 1997 established a new Welfare-to-Work Tax Credit for hiring long-term welfare recipients.  As we will see, the combination of subsidies to employers can approach 100 percent of wages for up to two years.
In 1996, the combination of work supplementation (i.e., diverting the welfare grant to employers who hire welfare recipients) and the Work Opportunity Tax Credit could result in a one-year employment subsidy equal to about 50 percent of earnings.  In 1998, the combination of work supplementation (now with food stamp funds allowed to be included) and the Welfare-to-Work Tax Credit could result in a two-year employment subsidy approaching 100 percent of earnings for hiring long-term welfare recipients.
Despite major efforts (and billions in extra spending) to increase collections, the average child support payment for one child fell slightly from 1978 to 1991.  Moreover, the proportion of mothers with child support awards and those receiving some or all of the amount awarded remained relatively stable over this time.  There are no data available after 1991/93, but experts in the field fear that little progress has been made.
In a continuing effort to increase collections, the 1996 welfare reform law requires state child support enforcement agencies to create centralized automated disbursement systems.  States are also required to collect information on all new hires to better track those who avoid wage withholding through frequent job changes.  Under the old law, states had the right to receive payment against arrears (to offset past welfare payments) even though that meant mothers who left welfare would often get no child support payments.
The provision that requires states to permit courts to order "pay-or-work" programs is likely to gain increased attention.  It is unclear whether or not states will enforce support awards against the parents of non-paying teenage fathers. 
Mothers on welfare tend to have more children than those who are not.  The concern driving the provisions listed above is that too many babies are born while the mother is still on welfare.  Bonuses go to the states that most lower the illegitimacy ratio without having a rise in abortions.  There is great uncertainty about how HHS will measure the illegitimacy ratio since data on out-of-wedlock births and abortions are unreliable.
Mothers who have their first baby as unwed teens are likely to go on welfare and form the core of long-term recipients.  A new element in welfare reform–evidenced first in the President's original bill as well as the later Republican bill–is the serious effort directed toward reducing teen births and teen dependency in general.  In addition, the law allows states to require non-custodial teen parents under age 18 to participate in work activities.
We earlier reviewed the reasons for the growth in the welfare caseload.  Now we turn to the reasons for its 34 percent decline since 1994.
There are four main forces associated with the decline in welfare caseloads.  Each is discussed below.
From 1960 through the early 1980s, administrative and policy changes played an important role in the growth of the AFDC and Food Stamp programs.  But even in the 1980s, unemployment and welfare caseloads did not entirely track.  (Neither did food stamp caseloads).  Only the recent upturn and decline in the caseload seem closely associated with the economy. 
Expansions in the Earned Income Tax Credit, child care, and Medicaid (not shown) have increased the financial returns to work.  Spending on the EITC is now nearly the same as AFDC at its peak.  This may help explain the sharp increase in the labor force participation of all low-income mothers, and especially never-married mothers with young children, up 28 percent since 1993, as the graph indicates. 
People can learn by observing the behavior of others and the consequences – desirable and undesirable–that result. This kind of social learning seems to have contributed to broad attitudinal and behavioral changes in drug use, crime, teen sex and pregnancy, and other social behaviors that influence welfare caseloads.
The 1988 welfare law encouraged states to emphasize education and training activities, and many did.  TANF encourages "work first" activities related to immediate employment such as job search, job readiness training, and work experience.  Data from 7 states show an increase in the participation in "work first" activities as a percent of recipients in each state's employment, training, and education program.  The intensity of "work first" activities has probably also increased.
Under the new welfare law, a family's net income (earnings, welfare, food stamps, and WIC less work-related expenses, such as child care and taxes) rises faster with additional earnings than under AFDC.  This is largely because most states have expanded earnings disregards and child care subsidies.  The bar on the left shows the pre-TANF value of welfare, including the amount of unreported earnings many welfare families had.  The gray bar shows the reduced value of welfare caused by the imposition of a work requirement, which reduces both the mother's leisure time and her opportunity for unreported work.
The actual imposition of sanctions varies considerably across states.  For example, about 11 percent of Delaware's caseload is receiving a reduced grant due to a sanction, compared to just 2 percent for Michigan.  Both states have sanctions that progress from a partial reduction in benefits to case closure.  Cumulatively, nearly 9 percent of Delaware's cases were terminated, compared to about 2 percent for Michigan (for only a 1/3 shorter period).
The 1996 welfare reform law mandates states to require a rising proportion of their caseloads to participate in work or another activity, from 25 percent in 1997 to 50 percent in 2002.  However, the law also gives states credit for caseload reductions–so that the actual requirement states will have to meet is now much smaller, and likely to remain so in the foreseeable future.
Past research on caseload dynamics suggests that many welfare families will exhaust a 5-year limit on benefits, although it might take 6 or 7 years for their cumulative time on welfare to reach 5 years.  If they wish, states can use federal funds to keep 20 percent of their caseloads on for longer than 5 years, and may use their own funds to keep the rest on.  The dramatic caseload decline suggests that far fewer families will hit the time limit.
Many fear that welfare recipients will not become independent because they face "barriers" to employment.  But data from Indiana suggest that self-reported "barriers" need not prevent mothers from leaving welfare.  Of course, it is possible that differences in the type and degree of problems facing recipients is an important factor in their ability to leave welfare.
Consistent with a large body of literature, welfare mothers without a high school degree or limited work experience are less likely to leave welfare.  This pattern is evident in a comparison of the education and work history characteristics of Indiana welfare mothers who left TANF to those who remained on assistance.
Many recipients are drug users and are often serious addicts.  The new law requires states to deny TANF benefits to drug felons, although they may opt out of this requirement (13 states have opted out).  States are permitted to require drug testing as a prerequisite for receiving welfare, but only 8 states do so (under varying circumstances).
Once again, four major factors seem to be contributing to the decline.
To review, four elements of welfare reform may be helping to reduce caseloads.  If time limits (the fifth item in the list) are having an effect, it is only because recipients fear that they may eventually be dropped from the rolls.
If the forces described above are having an impact, what's actually happening to the people affected?  The data are limited, but it appears that mothers are increasing their work effort and they are relying more on their parents, other family members, and friends (including boyfriends) for financial and emotional support.  It is too early to tell, but, as of now, there is no evidence of an increase in the formation and maintenance of two-parent families.
The welfare block grant guaranteed states a set amount of federal funding, whether caseloads grew or fell, thus creating an incentive for states to try to control caseloads.  The sharp caseload decline has resulted in a "windfall" to states, perhaps as much as $10 billion in the last two years.  Not all this money is going to welfare-related activities (the amount varies by state), but enough does so that per capita expenditures on recipients, although not grant amounts, have risen.  (The windfall has been used to fund an expansion of child care services as well as substantial state innovation.) 
What is the future likely to bring?  Both current law and political realities suggest that few states will impose harsh time limits on recipients who "play by the rules," that is, who comply with agency rules and regulations (even if they do not find a job).  On the other hand, although the block grant framework has encouraged programmatic expansions and some interesting innovation, there has been little radical change.  Thus, it seems reasonable to predict that states will continue a process of sequential innovation, with states moving forward one step at a time, learning as they go.  Call it "sequential implementation" or "incremental innovation."

  Return to Agenda | Endnotes | Top of Page