The Welfare Indicators Act of 1994 declared that its purpose was "to provide the public with generally accepted measures of welfare receipt so that it can track such receipt over time and determine whether progress is being made in reducing the rate at which and, to the extent feasible, the degree to which, families depend on income from welfare programs and the duration of welfare receipt." One of the tasks assigned to the Department of Health and Human Services in carrying out this requirement was to assess the data needed to report annually on welfare indicators and predictors and the ability of existing data collection efforts to provide such data. With the one exception noted below, whether existing data collection efforts would have been able to provide such data for the AFDC program was made a moot point by enactment of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996.
Chapter I presented the Advisory Board's proposal for a definition of dependence which is used throughout this report. The proposed definition would include work required to obtain benefits under the category of work activities, and would consider recipients who work to obtain their benefits through either subsidized or unsubsidized work as less "dependent" than recipients who do not work at all. Unfortunately, current data does not permit making this distinction. With the increased emphasis on work under the PRWORA, it will become increasingly important to be able to identify the extent to which welfare benefits are associated with public or private sector work.
More generally, in creating the Temporary Assistance for Needy Families (TANF) block grants, the PRWORA significantly changed the landscape against which the viability of existing data collection efforts is evaluated. The indicators of dependence and predictors or risk factors associated with welfare receipt that are presented in the previous chapters rely on national survey data collected primarily by government agencies and on administrative data collected and reported by state and local administering agencies (except for SSI program data, which is collected by the Social Security Administration). Although existing sources of data on the AFDC, Food Stamp and SSI programs are not without their limitations, the advent of the TANF block grant1 will make the existing data collection efforts that provide the data needed for the Department of Health and Human Services to report annually on the indicators and predictors more difficult and potentially less reliable.
1 Although titles II, IV and VIII of the Personal Responsibility and Work Opportunity Reconciliation Act contain a number of major revisions to the Food Stamp and Supplemental Security Income programs (several of which are highlighted in Appendix A), those changes did not include modification of the funding structure to a block grant or changes in the data collection and reporting requirements.
-
Administrative Data Issues
-
Although the abundance of program options presents a challenge to any data collection system, it is clear that the collection and reporting of state data on the nature and amount of assistance provided to eligible families under TANF will be much more valuable and descriptive of welfare receipt than national survey data alone2. The administrative data that is required to be reported on federal TANF assistance is substantial and will be extremely useful in providing information on families who receive TANF assistance. However, it will not be a complete representation of the multi-faceted TANF program that can be meaningfully compared to the AFDC program it replaced.
One of the biggest causes of uncertainty about the viability of data collection efforts to capture TANF program information is the change that took place in the nature of the program. While the AFDC program had always allowed states the flexibility to set eligibility rules, passage of the PRWORA dramatically increased states' ability to create assistance programs designed to meet each state's unique needs. Under TANF, federal block grants are provided to states for use "in any manner reasonably calculated to accomplish the purpose" of the law. Federal TANF funds cannot be used for assistance to families without minor children, to individuals who fail to cooperate in establishing paternity and collecting child support, to families who do not assign child support rights to the state, to teenage parents who do not attend high school or training and live in adult-supervised settings, or for assistance to adults for more than five years. Federal TANF funds can be used for virtually any other purpose that would accomplish the objectives of the block grant.
Under TANF, states are required to maintain a certain level of historic state expenditures, but they are not bound by the same requirements as federal TANF funds as long as the federal and state funds are not co-mingled. Some states may utilize the flexibility provided within TANF to separate the federal TANF dollars from the state maintenance-of-effort (MOE) funds, and operate separate programs with each funding stream. In order to accurately measure dependence, it is critical that future indicators include data that describes both federally-funded TANF assistance and assistance provided with state MOE funds.3
Also, the flexibility provided by TANF will almost certainly increase the variation in the type of assistance received. Assistance to needy families can be provided in cash or through in-kind services such as wage supplements to employers, child care, one-time heating or cooling assistance, individual development accounts, and assistance for qualified aliens. Moreover, as time goes on TANF assistance will likely become more and more associated with performance of some kind of work activity, from work required to obtain benefits to training to subsidized employment and may be provided by the welfare office or an employer. The Advisory Board's proposed definition of dependence discussed in Chapter 1 calls for future reports to distinguish between assistance provided in relation to work and assistance provided unrelated to work. As states more fully develop their TANF programs it will become increasingly important to understand what types of assistance are provided and by whom.
Finally, TANF program parameters and operating rules vary across states, and sometimes within states. These variations are likely to increase as states have an opportunity to further utilize the flexibility provided under TANF. The development of data sources that can illustrate these indicators at the state level will become even more important as the "national" data provided by national surveys further mask the differences across states and present an incomplete picture of welfare receipt in the United States.
It is generally desirable to capture a sufficient level of detailed data about the TANF program and its recipients and benefits to permit the development of AFDC/TANF time series data for tracking several of the indicators and risk factors associated with welfare receipt. In addition to the general desirability, however, the quality and level of detail of TANF data takes on even greater importance in the context of this report's proposed primary indicator of welfare dependence. For purposes of discussion, this report proposes the following definition of the primary indicator of dependence:
A family is dependent on welfare if more than 50 percent of its total income in a one-year period comes from AFDC/TANF, Food Stamps and/or SSI, and this welfare income is not associated with work activities. Welfare dependency is the proportion of all families who are dependent on welfare.
Public debate on any number of fine points of the merits of this definition of welfare dependence will be influenced by the availability of TANF data. Several questions about the proposed definition come immediately to mind:
- What should be considered "income?" Cash or near cash (like food stamps) benefits only? Should the value of services provided to employed recipients and in-kind benefits be counted? Should one-time emergency assistance provided in lieu of a determination of on-going TANF eligibility be counted?
- What work-related activities should be considered? Should working to obtain benefits count as work? Should job skills training be counted? What about education directly related to employment? Should state-funded payments to employers for the purpose of supplementing the wages of employed recipients be counted, or only federally-funded payments?
Regardless of the choices made for this report or whether this proposed definition is ultimately determined to be a useful indicator of welfare dependence, the quality of the public debate will be influenced by the quantity and quality of the available data. Questions about TANF receipt and the proposed dependence definition abound. For instance, who will be considered TANF "recipients?" Will they include families who receive cash benefits only, or will families who receive some cash and some services also be counted? What about families who receive only services? Will "recipients" include only those families who receive federal TANF-funded benefits? If the definition includes families who receive benefits under state-funded programs that count toward the state's TANF MOE requirement, will data on those families also be collected and reported by states? Will TANF data be available to extend the time series on the proposed definition? Will the TANF data distinguish between cash and in-kind benefits? Will the data identify whether work was required to obtain cash benefits? Will the data on participation in work-related activities differentiate between work and training?
These data reporting questions and others like them are being debated for many reasons unrelated to this report. Certainly, the federal government's role in requiring states to collect and report such data in the context of block grants is a major consideration. However, the Welfare Indicators Act was not repealed by the PRWORA, and while it is neither the purpose nor intent of the annual Indicators reports to evaluate the outcomes of welfare reform, these reports continue to be potentially useful tools in tracking the impacts of the PRWORA on AFDC/TANF, Food Stamps, and SSI recipients, and in tracking trends in predictors or risk factors associated with welfare receipt.
2 Most national surveys are not currently designed to provide reliable estimates of state-level data. Those that are representative at the state level can currently only provide reliable estimates for the largest states.
3 Some examples of assistance states are considering providing to eligible families with state funds that qualify as TANF MOE expenditures include: benefits to families after time limits expire, benefits to individuals who are enrolled in 2- or 4-year college education programs, and state-funded food stamp benefits to legal immigrants who are ineligible to participate in the Food Stamp Program as a result of PRWORA.
-
-
National Survey Data Issues
-
It is of critical importance to understand the policy and program context that may surround changes in welfare dependence over time. As noted throughout this report, between-state, within-state and across-time variations are already happening as a result of the PRWORA provisions and are anticipated to become more diverse. Changes are expected in eligibility requirements (both income-and non-income-related), benefit levels and benefit types, work requirements and sanction policies, time limits, family caps and other areas. Within national surveys, reliable indicators of dependence must capture the realities of individual experiences with welfare receipt. While survey data complement administrative data in several ways, surveys present two main drawbacks: (1) most survey data are not currently representative at the state level, and (2) survey data have a significant time lag between the collection of data and the availability of data for analysis. Nonetheless, national survey data are critical for capturing indicators of adult labor force participation, earnings, program participation, fertility and child well-being, as well as complementing caseload data for tracking changes in dependence.
The PRWORA makes it critical that national surveys accurately measure welfare receipt. Under TANF, as discussed above, welfare receipt can take on many forms of assistance, including child care, wage supplements, and vouchers for services. National surveys are neither currently designed to capture this broader range of cash and non-cash assistance nor to estimate the value of noncash services. In addition, the TANF assistance programs replacing AFDC are taking on a proliferation of names across the states and are increasingly being administered by non-government organizations both of which make the measurement of welfare receipt more difficult. Finally, measuring welfare receipt is further complicated by the potential existence of state-funded assistance programs, as discussed above, that are separate from federally-funded state TANF programs.
For purposes of this report, the Survey of Income and Program Participation (SIPP) has been used the most extensively and is considered the most useful survey. Some of its characteristics which make it most useful are its longitudinal design, system of monthly accounting, and detail concerning employment, income and participation in federal income-support and related programs. These features make the SIPP particularly effective for capturing the complexities of program dynamics and many of the indicators and predictors, or risk factors, associated with welfare receipt. Planning is underway for the seven-year extension of the 1992-1993 SIPP panels, or the Survey of Program Dynamics, provided for by the PRWORA.
The Panel Study of Income Dynamics (PSID) is also used in this report, as are the National Longitudinal Survey of Youth (NLSY) and the Current Population Survey (CPS). The CPS measures income and poverty over a single annual accounting period, and provides important information regarding childhood poverty. Both the PSID and NLSY are longer-run surveys that provide vital data for indicators of intra-and inter-generational dependence and deprivation. The PSID and NLSY collect annual income data, including transfer income, that yields inter-generational indicators. While the PSID and the NLSY are critical for obtaining measures of long-term welfare receipt and inter-generational receipt, both surveys have currently moved to biannual data collection which may reduce their usefulness.
As with administrative caseload data, the SIPP and other national surveys could enhance their value for developing indicators of dependence and deprivation in several ways. As discussed above, national survey data must be able to accurately measure both cash and non-cash assistance as well as assistance under both state-funded programs and federally-funded state TANF programs. National surveys are not currently designed to accurately capture this information. Also, the proposed definition of dependence discussed in this report requires data that would distinguish welfare benefits received in conjunction with work from benefits received without work. As current survey data do not include this information, this report was not able to fully illustrate the recommended definition. Surveys must collect such information if future reports are to utilize the proposed definition of dependence.
The value of national survey data would also be improved by gathering more complete and comprehensive retrospective information regarding previous welfare spells. Realistically, surveys cannot completely account for welfare dynamics --no matter how long the observation window or the accounting period, there will always be welfare spells that occurred before the survey began and continue after the survey ended. This makes it all the more important that surveys take precautions to reduce the risk of providing an incomplete picture of lifetime welfare receipt. Retrospective questions regarding prior welfare receipt will help address this concern.
In addition, the PRWORA makes it even more important that national surveys contain questions to determine the factors involved in the ending of a spell during the observation period. As noted above, it is expected that the PRWORA will result in more diversity in the causes of caseload terminations. Cases may be closed due to increased work effort or as a result of sanctions or time limits. Information regarding the precise event that began or ended a welfare spell can provide critical guidance to policy makers in their efforts to reduce dependence and deprivation. Discussions should continue around ways to ensure that information regarding events that begin or end welfare episodes is not lost if the event occurs outside of the observation period.
Despite the need to collect state-level data in order to fully capture information on the dependence status of recipients of means-tested assistance and the current limitations of national surveys to provide reliable state-level estimates, national survey data are of critical importance in efforts to measure and track changes in dependence. Even when the administrative data collection questions discussed earlier in this chapter are resolved, some state data systems have limited capacity for modification and may be unable to provide the necessary data. Unfortunately, current resources for the SIPP and other national surveys may not be sufficient to fill in the existing gaps in administrative data or to compensate for any lacking state data.
-
-
Potential Risk Factors for Which Data Do Not Exist
-
As noted in Chapter III, the predictors/risk factors included in that chapter do not represent an exhaustive list. Rather, they are a sampling of available data that address in some way a family's circumstances on the deprivation/well-being scale. The range of possible risk factors is extremely wide, and until they are measured and analyzed over time, their predictive value will not be known. As the PRWORA changes are implemented, some of the risk factors may turn out to be simply correlates of welfare receipt, some may have a causal relationship, some may be consequences, and some may, in fact, have predictive value.
While the Advisory Board recommended that this first annual report focus on a smaller set of dependence indicators, it also recommended that the report take an expansive view toward predictors and risk factors. Two domains, in particular, were identified as potential risk factors that should be included, but for which data do not exist.
Adult literacy is related to success in the labor market. A risk factor on literacy would illustrate the risk of welfare dependence. Barton and Jenkins (1995) report that a large proportion of the welfare population have weak literacy skills. Unfortunately, a comprehensive survey of adult literacy was conducted in 1992 but has not been repeated since. It would be desirable, although expensive, to measure literacy on a more routine basis.
The physical and mental effects of domestic violence put the victims at serious risk of dependence. The Department of Justice collects data on domestic violence in its Crime Victim Survey, but it is widely believed that this data severely underreports the incidence of domestic violence. Four recent research studies compiled by The Taylor Institute found large and consistently high percentages of women on AFDC currently abused by partners. Although these studies range in methodological rigor, taken together they can begin to assist in better understanding the role that domestic violence plays in poor women's ability to become self-sufficient. The Balanced Budget Act of 1997 requires the General Accounting Office to conduct a study of the effect of family violence on the use of public assistance programs, and in particular the extent to which family violence prolongs or increases the need for public assistance. It would be desirable to collect high quality data on domestic violence on a more routine basis.
-