• Repeals federal AFDC requirements and establishes a new TANF program, with no entitlement to benefits.
• Replaces federal matching funds with a block grant to the states.
• Requires that state plans include a provision that at least one adult in a family that has been receiving aid for more than two years participate in work activities, as defined by the state.
• Requires states to reduce grants for recipients who refuse to engage in work (as defined by the state).
• Penalizes states for not meeting specified rates of participation by TANF recipients in work-related activities.
• Establishes a maintenance-of-effort requirement on state expenditures for needy families.
• Limits to five years the amount of time a family can receive federal TANF funds.

In October 1996, the Governor submitted a preliminary state plan to the federal government in order to implement the TANF program. By doing so, the state will receive federal block grant funds for 1996-97, resulting in increased federal funds of over $300 million for the current year, compared to what the state would have received under the prior law. The plan, however, indicated that the state would continue to operate its AFDC program during 1996-97, as provided by current state law.

Governor's Welfare Reform Proposal

Welfare Reform: Competing Goals

• Ensure that individuals and families do not live in a condition of poverty--a goal that suggests the provision of sufficient aid to bring family income above the poverty line.
• Minimize welfare dependency and associated public costs--a goal that suggests the obligation to work and the use of work incentives such as low levels of aid and time limits on eligibility to encourage welfare recipients to work.

Generally, in developing welfare systems, policymakers strike some balance between these goals. It is important to recognize, however, that the emphasis given to one goal over another is often based on philosophy as well as on cost/benefit analysis.

Family income is the primary determinant of eligibility for grants under the AFDC program. Currently, eligibility is based mainly on an income "need standard," also known as the Minimum Basic Standard of Adequate Care. The need standard for a family of three, as an example, is $735 per month in the high-cost counties under California's regional grant system.

Generally, eligibility for an AFDC grant depends on whether the family income--after excluding a portion of this income pursuant to an "earnings disregard" (discussed below in more detail) and allowances for work-related expenses--exceeds the need standard. Because of the disregard and work-related allowances, families with incomes above the need standard can qualify for a grant. However, anyone with a gross income above 185 percent of the need standard is automatically ineligible for the program.

We note that the need standard is set above the maximum grant level. For example, the maximum grant for a family of three is $565 (in high-cost counties), or $170 below the need standard. This acts as a work incentive feature, by allowing working recipients to keep any earnings between the need standard and the maximum grant ($170 in this case) without having their grant reduced or losing eligibility for the program. We assume, in our approach, the need standard as provided under current law.

Grant Levels

Current state law also includes the Maximum Family Grant provision for the AFDC program. Under this provision, grants do not increase for additional children born while a recipient is on aid.

Finally, current law includes the Alternative Assistance Program, under which AFDC applicants or recipients with earned income are permitted to receive Medi-Cal benefits and child care payments if they choose to decline cash grants.

What is the Appropriate Grant Structure? A review of the research does not provide clear guidance as to what constitutes the "right" grant level. Increasing the maximum grant would help bring nonworking families out of poverty, but would reduce to some extent the financial incentive to work (and, of course, would increase costs). Conversely, reducing the grant would tend to have the opposite effects. After considering these factors--in the context of balancing the objectives of providing income support while maintaining a work incentive and controlling costs--we chose to incorporate the grant levels under existing law, with certain exceptions discussed below.

Our approach retains the fill-the-gap budgeting structure but limits the $30 and one-third disregard. Specifically, the disregard would be applied in full to a recipient's first year of employment, reduced by half for the second year, and then eliminated. We include these modifications in light of research recently conducted in California and Minnesota.

California contracted for an evaluation of certain work incentive provisions enacted in 1991-92--(1) the expansion of the $30 and one-third disregard, (2) the elimination of a provision that prohibited persons working more than 100 hours per month from eligibility for AFDC (U) benefits, and (3) the establishment (through maximum grant reductions) of a "fill-the-gap" grant structure in which there is a gap between the need standard and the maximum grant (as previously described). In a report submitted in January 1997, the evaluators found that the work incentives had a positive effect on employment among AFDC (U) recipients--possibly due to the elimination of the 100-hour rule--but did not show a positive impact among AFDC (FG) recipients, who comprise over 80 percent of the AFDC caseload.

If the $30 and one-third disregard and the fill-the-gap structure do not bring about increased employment among recipients, they will result in a cost to government. This is because they will result in higher grant payments to working recipients, and in some cases recipients will remain on aid for a longer period of time. The evaluation provides some evidence that this has been the case in California.

An interim evaluation of a welfare reform program in Minnesota, which includes financial work incentives similar to California's, had more mixed results: the program had the effect of increasing the number of persons on aid who were working, but kept working recipients on aid for a longer period of time. The program also had the effect of increasing average grant expenditures, but the long-term impact on cost-effectiveness is not known at this time. We note, moreover, that the evaluation does not isolate the impacts of the financial work incentives from the impacts of other program components.

Thus, the research suggests that the earnings disregard may not have the intended impacts. We believe that to the extent the disregard does affect a recipient's decision to obtain a job, phasing it out is unlikely to alter that initial decision or to cause the recipient to give up the job when the disregard is reduced and eliminated. This is primarily because recipients who lose the disregard would still retain a significant financial benefit from continuing to work (for example, from the Earned Income Tax Credit [EITC], which provides a tax reduction or refundable credit for low-income working persons).

Wage Replaces Grant for Community Service Job Participants. Under our approach community service employment would be required for adult recipients who are on aid after 24 months and not working at least 20 hours per week. These participants would be paid the minimum wage and would be eligible for the EITC. For example, a person who works 20 hours per week would earn $498 in an average month and be eligible for $183 per month through the EITC (assuming a single parent with two children). After adjusting for social security taxes, the family's income would total $643 per month. When combined with food stamps, this family would have a monthly income of $953, or 88 percent of the federal poverty level. (We discuss community service employment in more detail below.)

Services

Currently, these employment, health, and social services are provided through a variety of programs, generally administered by counties, school districts, and community colleges. Counties indicate, however, that the availability of these services is limited and sometimes inadequate, due primarily to funding constraints; and there is relatively little coordination of such services for welfare recipients. The GAIN program provides some degree of coordination of job search, education, and job training services through its case management, but the program currently is available only to AFDC recipients and has never been funded to serve more than about 25 percent of eligible persons.

An independent evaluation of the GAIN program found that as implemented by Riverside County--one of the six counties studied--the program produced larger earnings gains and grant reductions than found in any previous large-scale study of welfare-to-work programs. The evaluation, as well as a related study covering programs in various states, found that an approach that emphasizes getting recipients into jobs as soon as possible--rather than one that emphasizes education and training--is more effective.

Even in the Riverside program, however, 41 percent of the participants were still on aid after three years. Nevertheless, the Riverside program was found to be cost-effective from the government's perspective, compared to the AFDC program without GAIN--generating $2.84 in savings and revenues for every dollar spent.

What Services Should Be Provided? The Riverside GAIN approach can be characterized as including a strong employment message to recipients, relatively more (and earlier) use of job search and job development activities, broader participation, and greater use of sanctions. Our model includes the basic features of the Riverside GAIN program, with some modifications, as summarized below. We note that the time lines for providing specific services are intended as norms, and case managers would have the flexibility to vary from these guidelines where it is determined that it would be cost-effective to do so.

• Participation requirements would be broader. All adults, with certain exceptions, would be expected to participate in activities designed to assist them in obtaining employment. The exceptions would be parents with children under age one, caretakers of disabled persons, and relatives who are the caretakers of children on aid. Under current law, parents with children under the age of three are exempt from the GAIN program. We note, however, that nonwelfare parents with children one and two years old frequently have jobs. We also note that AFDC parents who have very young children when they go on aid tend to remain on assistance for a longer time, indicating the need to establish a connection with the labor force at an early stage of welfare receipt. Finally, under our model, deferrals would be more limited than under GAIN (which includes instances that are difficult to document, such as "family crises")--specifically, deferrals would be limited to medically-verified illnesses or disabilities.
• All recipients would receive case management, as in the GAIN program, with the amount depending on the needs of the clients. This would include a case plan designed to achieve self-sufficiency, based on the needs and abilities of the recipients.
• During the first six months on aid, GAIN orientation would be followed by job search/job club services--supervised and unsupervised activities focusing on making job contacts and interviewing for positions, as well as group job clubs which include a classroom instruction component . This time frame differs in some respects from GAIN currently, where recipients may be referred to basic education or job training during this period. As indicated above, evaluations have found that the "labor force attachment" approach (emphasizing up-front job search) is more effective than the "human capital development" approach (emphasizing education and training). In order to maximize efficiency, case managers would have flexibility to determine which recipients would need more intensive job search or group job club assistance, based on the recipients' prospects of obtaining jobs without this level of assistance. The six-month time frame is established in recognition of the fact that a large number of recipients (about 25 percent historically) go off of aid without any services. Thus, in order to control program costs, case managers generally would delay the provision of more expensive services (such as assessment, education, and job training) until after six months.
• Job developers would assist in finding jobs by establishing direct contacts with potential private and public sector employers. This is a component of the Riverside program, but not used in many other counties.
• For persons still on aid after six months, an employment assessment would be conducted. Pursuant to the assessment, other services would be provided to address barriers to employment, with periodic job search. These services could include basic education, English-as-a-Second Language courses, job training, and services--such as counseling, home visits, and drug or alcohol abuse treatment--designed to address certain problems or disabilities. Case managers would have discretion in determining what services are appropriate, taking into account the cost-effectiveness of these activities. This is similar to how the GAIN program operates, but we note that while case managers in GAIN currently may refer recipients to providers of health and social services, typically they confine their activities to basic education and job training services.
• Basic education would be provided only through programs that can demonstrate that they are effective, and only to the extent progress is made by the recipient. Typically, basic education courses offered by adult education programs and community colleges are provided independently of vocational training courses. This delivery mode, however, generally has not been successful. For example, an evaluation of basic education provided to GAIN participants--typically through public education institutions--did not find significant educational achievement impacts in two of the three counties studied or welfare and employment impacts in any of the counties. An integrated approach appears to yield much better results--for example, the job training/basic education courses offered by the Center for Employment Training, a private organization based in San Jose.
• Case managers and job developers would identify opportunities for volunteer work, and place recipients, on a voluntary basis, in such positions. While this is currently not part of the GAIN program, this activity is one of the components of a welfare-to-work program (Project Match) in Chicago. Program administrators indicate that there are a large number of such work slots available in local areas.
• Child care and transportation expenses would be provided as needed. Child care reimbursement, however, would cover costs up to the 75^th percentile of regional costs, as provided under current law for non-GAIN AFDC recipients. The higher limit for the GAIN program--approximately the 93^rd percentile--would be eliminated. (For a discussion of this issue, see our review of the GAIN program in the Analysis of the 1995-96 Budget Bill.)

With respect to other services provided outside the context of the GAIN program, the model includes the following:

• The Cal Learn Program would be retained, pending the results of an evaluation currently in progress. This program serves teen parents on AFDC by providing case management and fiscal bonuses and penalties based on school attendance and performance.
• A new service--one year of transitional case management for recipients going off of aid due to employment--would be provided. This is designed to facilitate stable employment, in light of the relatively high rate of recidivism in the AFDC program. (For example, an estimated 26 percent of AFDC recipients in October 1995 had been on aid more than once.) The existing transitional child care and Medi-Cal benefits would be continued as provided by current state law--for up to two years after an AFDC recipient goes off of aid due to employment or marriage. (The second year of transitional Medi-Cal currently is subject to approval of a waiver of federal regulations.)

Community Service Employment

Under current state law, adult recipients who have been on AFDC for two years from the date of their GAIN assessment must participate in a work preparation assignment (similar to community service jobs) if made available by the county, unless they are working at least 15 hours per week. This provision became effective during 1995-96, so recipients will begin to be affected by the two-year time frame in 1997-98. It is important to recognize that the provision of work slots is at the option of the counties.

In addition, counties typically provide such jobs as part of a requirement that GA recipients work. Contra Costa County, for example, requires employable recipients (who are not otherwise working) to work in county-provided jobs within a few months of application for aid. A wide variety of county jobs are provided, such as paper recycling, clerical work, and roadside litter removal, as well as jobs in nonprofit organizations such as certain hospitals in the county.

Some other states have implemented, or plan to implement, welfare reform proposals that include community service employment as one of the components (Wisconsin and Vermont, for example). In fact, the federal welfare reform act requires states to include such provisions in their state plans, unless the Governor specifically chooses not to do so.(The Governor's 1997-98 welfare reform proposal would authorize counties to include community service as a work-related activity for recipients.)

The LAO Approach. In our approach, as noted above, parents (or household heads in the case of two-parent families) who have been on aid for 24 months, and not working at least 20 hours per week, would be placed in a community service job unless they have a child under one year, are caretaker relatives, are caretakers for a disabled person, or have a medically-verified illness or disability that precludes employment.

By design, the community service jobs would be operated in the same manner as nonsubsidized jobs. Recipients would be paid a wage, not given a welfare grant. The wage would be set at the minimum wage level ($498 per month on average, for a 20-hour work week) and the employees would be eligible for the EITC ($183 per month for a single parent with two children). They would also be eligible for Medi-Cal and food stamps.

Participation in these jobs would be supplemented by job search and other activities designed to address obstacles to obtaining nonsubsidized employment, if prescribed in the case plan. Work would generally be for 11 months, with periodic job search, followed by one month of intensive job search prior to placement in another work slot.

During the first year of community service jobs, recipients would work 20 hours per week. The number of hours would increase to 25 in the second year, and to 30 in the third year, with essentially no increase in the wage. This is designed to increase the incentive for recipients to find a nonsubsidized job. (In order to retain the federal EITC, this would be accomplished by reimbursing the recipient with a nominal grant--$10 per month--for the additional hours worked.)

Under our approach, case managers could redirect a recipient from a community service job into a short-term intensive job training program, if the recipient could benefit from such training. For example, the Center for Employment Training, a private organization that operates job training programs in 19 sites in California, provides courses where the student attends full time for three to eight months, depending on the course. Such job training programs could also be operated directly by individual businesses.

The Pros and Cons of Community Service Employment. Community service jobs constitute just one of a variety of activities that could be adopted to facilitate the movement of recipients into private or public nonsubsidized employment. Proponents of community service employment generally cite two broad reasons for such an approach: (1) the monetary benefits of moving recipients off welfare and into employment, and (2) the nonmonetary benefits, to recipients and society alike, from engaging in work. The specific benefits are summarized below:

• It can help prepare participants for employment by teaching them good work habits.
• Similarly, it can increase participants' chances of obtaining nonsubsidized employment by giving them an opportunity to gain work experience and demonstrate a good work record.
• By requiring welfare recipients to participate in such work (and counting this time against time limits on aid), it can increase the incentive to seek nonsubsidized jobs, when compared to a system where no such participation mandate exists.
• When compared to not working, such jobs can contribute to participants' self-esteem and enhance their status as role models for their children. We note, in this respect, that participants in unpaid work experience programs in the 1980s reported that the work was meaningful, not "make work" jobs.
• Some public value can be attributed to the work itself.
• It is consistent with the notion, held by many, that able-bodied welfare recipients should participate in work-related activities in exchange for the aid they receive.

Community service jobs, on the other hand, would result in significant administrative costs and additional costs for child care and transportation. It also may be difficult to develop enough work slots if implemented on a large scale.

Research on a few small "workfare" programs in the 1980s (in which recipients worked for their grant) did not find consistent positive effects on employment and earnings. The researchers indicated, however, that programs offered on a larger scale and with broader participation mandates could prove to be effective. We also note that better outcomes might be achieved if such programs were operated in the context of a time-limited aid environment, where the incentive to get a nonsubsidized job would be greater. Moreover, the research indicated that by assuming some public value from the output of the work, the benefits exceeded the costs, from the perspective of the taxpayer.

Finally, we note that the State of Virginia has implemented a community service job requirement for AFDC recipients who have been on aid for 90 days. This is part of a welfare reform program that includes a time limit of 24 months of assistance within any 60-month period. (Some recipients would be eligible for the state's General Relief program after the time limit.) Program administrators indicate that in those counties where the provision has been implemented, caseload reductions have occurred and many recipients found nonsubsidized employment prior to the time the community service job requirement took effect. Consequently, the state has had to develop a much smaller number of community service work slots than had been anticipated. The program, however, has not been evaluated.

On Balance, It's Worth Trying. After considering the factors summarized above, we believe the potential benefits justify including this activity in our approach. This recognizes both the potential monetary benefits associated with moving recipients into nonsubsidized jobs, as well as the nonmonetary societal benefits from a program that requires able-bodied welfare recipients to work rather than remain at home.

Because studies indicate that from 40 percent to 50 percent of AFDC recipients leave aid within two years, the community service job requirement would not begin until a recipient has been on aid for two years. This will help to control the costs of this component while targeting the activity to those recipients who are most likely to benefit from it. Maintaining the requirement until the time limit, moreover, will ensure that work-ready recipients of aid are participating in a work activity.

As is the case for most community service employment programs, there would be a provision that these jobs not displace workers in existing jobs. We estimate that approximately 93,000 community service work slots would need to be developed in the first year that the provision is implemented. To put this in some perspective, there are currently about 13 million jobs in California. Counties and cities are the employer for over 500,000 of these jobs, and over 700,000 persons are employed by nonprofit organizations.

Sanctions

Time Limits

It is important to note that the federal act places no time limits on the use of state funds. As a result, the state does not necessarily have to impose any time limits on recipients' eligibility for aid. Those on aid for more than five years could be funded entirely with state funds, and the federal funds that would have otherwise supported these recipients would, in effect, be shifted to other recipients. Thus, whether to impose a time limit on state funds is a key decision facing the Legislature.

How Many Persons Might Be Subject to Time Limits? As background for the consideration of time-limited aid, we note that according to the 1995 survey of AFDC recipients, 35 percent of the cases were on aid for five years or more. This consisted of 36 percent of AFDC (FG) cases and 29 percent of AFDC (U) cases. Viewed from a different perspective, the Department of Social Services estimates that of those AFDC (FG) recipients who began aid in 1988, about 30 percent were on aid five years later. Some other studies estimate that over 40 percent of the persons receiving AFDC eventually will accumulate five years of time on aid.

The large number of long-term recipients is probably related to the fact that these recipients tend to have lower levels of education and work experience prior to going on aid. The recent GAIN evaluation shows that the program can increase the number of recipients who work--including long-term recipients--but a significant number still do not obtain employment. Even in the best-performing county, for example, 33 percent of the participants did not work at any time during the three years of the study.

Thus, it appears that support services by themselves--even with the potential to impose sanctions as provided under the GAIN program--will not be sufficient to move a high percentage of recipients into stable employment and off of welfare. Whether time limits would prove to be more successful is not known.

We note that estimates of the number of recipients who could be affected by a time limit, such as those cited above, assume a continuation of the AFDC program as it operated in past years. It is important to keep in mind that welfare reform interventions--such as the GAIN program, community service jobs, targeted tax credits, and time limits themselves--are designed with the intent of increasing the number of participants who obtain employment, thereby reducing the number of recipients who actually reach the time limit.

What Do We Know About the Effectiveness of Time Limits? Time-limited aid has the potential to act as a powerful work incentive--in terms of encouraging recipients to seek work and participate in work preparation activities--but carries the risk of depriving families of any source of income. While several states are beginning to implement various forms of time-limited aid, no evaluations have been completed on such provisions. An interim report on Florida's time-limited welfare program should be available within a few months, but the findings will be preliminary.

What Should the State Do? It is clear that time limits entail considerable risk, but there is also much that can be gained. The relatively large number of able-bodied welfare recipients who are not employed on a regular basis, even when GAIN services are provided, suggests that there is room for improvement. As indicated previously, time limits may prove to be effective when combined with other interventions designed to increase employment. The state savings resulting from time limits, moreover, could help to finance the costs of providing services to recipients. Based on these factors, we believe that some type of time-limited aid is worth trying.

How Should Time Limits Be Applied? In considering time limits, the state could adopt the premise that aid should not be provided past the time that a recipient could reasonably be expected to obtain employment. Of course, this would vary considerably among welfare recipients. One way to address this issue would be to give case managers discretion to decide when aid would be terminated. Such discretion, however, would inevitably lead to problems of equitable treatment of recipients and possibly increased costs for administration. In addition, the behavioral effect on recipients (as a work incentive) might not be as great if they do not have a clear idea, from the outset, of when the limit will occur.

Another approach is to adopt differential time limits according to specific characteristics of recipients--for example, a shorter time limit for those who are the most "work-ready" based on measurable criteria. This approach may also lead to problems of equitable treatment (between those who fall just below and just above the threshold) and increased administrative costs to collect and verify the data on the criteria.

Given these difficulties--and after considering the time on aid data for AFDC recipients--our approach incorporates the same limit that applies to the federal funds--five years of total time on aid. (Time would be counted as of the date of program implementation.) Certain cases would be exempt--those where the adult recipient is caring for a disabled person or is a caretaker relative, and child-only cases with disabled parents or caretaker relatives. As discussed below, the model provides for a "safety net" for those families that reach the time limit.

The model assumes that time on aid is not counted against the limit where (1) the adult recipient is working in a nonsubsidized job at least 20 hours per week or (2) the state does not provide services pursuant to the recipient's case plan, as determined by the case manager. (Note that time in community service jobs would count against the limit.) In addition, the time limit would be extended if (1) jobs are not available, according to specified local labor market measures, or (2) local administrators find that a recipient has a medically-verified health or psychological impairment that precludes employment, and the recipient has made a good faith effort to comply with case plan provisions to address these problems.

Even if time limits were to prove to be relatively successful, it is likely that some individuals and families will reach the limit without securing employment. This raises the question of how "hard" the limit should be. For example, should there be a "safety net" such as the one provided by the existing GA program? The dilemma confronting policymakers is that the more restrictive the time limit, the more likely it is to be successful in terms of the number of recipients who become employed (assuming the behavioral effects work as intended) but the greater the risk of adverse effects to the extent it is not successful.

Our approach does not include a strict time limit that would cut off cash assistance to families completely. Instead, after the five-year time limit is reached for the TANF program, families would be eligible for cash benefits but at a much lower level. The grants would be set at $300 per month for a family of two, $375 for a family of three, and $450 for a family of four or more (generally about two-thirds of the existing grant levels).

These families would remain eligible for Medi-Cal benefits and food stamps. In addition, they would be permitted to use job search services at their option.

By reducing, but not eliminating, grants for families after five years on aid, our model strikes a balance between the objectives of achieving the behavioral effects associated with time-limited aid and providing some income support for families with children. Figure 6 shows the change in monthly income when a family of three persons shifts from the community service job component to the five-year "safety net" program.

There has been some discussion, for example, of the possibility of adopting a state EITC. As noted above, the federal ETIC provides a tax reduction or refundable credit for low-income working persons. Enacted in 1975, the EITC was expanded significantly in recent years.

Another alternative to consider is the adoption of targeted tax policies to bring about more employment among low-income persons. In August 1996, the President signed the Work Opportunity Tax Credit, which provides tax credits to employers who hire persons from specified target groups, including AFDC recipients. The program replaces the federal Targeted Jobs Tax Credit, which expired in 1994. These programs have the effect of subsidizing the wages paid to individuals in the target groups. Generally, research on the Targeted Jobs Tax Credit program indicates that it was not effective in increasing the employment of disadvantaged workers.

In addition, the President has proposed a new program in which employers could (1) claim a 50 percent tax credit on the first $10,000 of wages paid to long-term welfare recipients and (2) treat employer-provided education and training, health care, and dependent care spending as wages for purposes of claiming the tax credit.

Targeted wage subsidies, in which the government would subsidize the wages of welfare recipients employed in the private sector, have also been proposed as a way to increase employment among recipients. A study of a targeted wage subsidy program for welfare recipients in Dayton, Ohio--and a separate study of a similar program in two cities in Wisconsin-- found that the subsidy actually reduced the prospects of recipients obtaining employment.

Finally, we note that state law includes tax policies designed to increase the employment of certain target groups and stimulate investment in depressed areas. This includes programs offering economic incentives--such as tax credits for hiring disadvantaged persons--for businesses in 29 designated "enterprise zones" throughout the state. A study by the Bureau of State Audits in 1995 indicated that data were not sufficient to determine whether the enterprise zone program was effective.

Administration

As Riverside County demonstrated in its implementation of the GAIN program, a successful welfare-to-work program requires local welfare departments to change the focus of their mission from one that emphasizes eligibility and grant determination to one that emphasizes employment. In this respect, it is important to note that while the state would contract for administration, we envision that local administrators would have considerable discretion in how services are provided. The local administrators would develop case plans and would make decisions on resource allocation. For example, they would determine the distribution of funds allocated for services such as basic education and job training, and would have the option of contracting with public or private providers for such services. This allows for local innovation in areas such as developing collaborative arrangements between service providers (community colleges and private industry councils, for example) and ways to integrate services such as education and job training.

Funding Structure

Under our approach, the state would continue to have programmatic responsibility for the TANF program. Thus, one could argue that the state should also assume the full costs of the program. The advantages to such an arrangement would be that it would clarify that the state is responsible for program outcomes and would insulate the counties (which have a more limited revenue base) from the fluctuations in costs due to the effects of economic cycles on program caseloads.

We also note, however, that under the model the counties would continue to be partners in the new program--not only as administrative agencies but also as providers of related services such as mental health and drug abuse treatment. In this situation, giving counties some share of program costs could act as an incentive for counties to take actions that would contribute toward positive program outcomes as well as efficient administration.

After considering these factors, our approach incorporates a funding structure as follows: For administration and services, the state would pay 85 percent of the nonfederal costs and the counties 15 percent. The county share of costs is set at a level that is designed to encourage efficiency but not so high as to give counties an incentive to "underspend" for fiscal rather than policy reasons.

For grants and community service job wages, the state would also pay for most of the costs but--in order to give counties an incentive to maximize their efforts to get recipients off of aid--the county share of costs would increase gradually as recipients' time on aid increases, up to a limit. Specifically, for recipients on aid for up to one year (beginning with the date of implementation of the program), the state would pay for all of the grant costs. For recipients on aid from 13 to 24 months, the county share would be 5 percent. The county share would increase in a similar manner by 5 percentage points as recipients' time on aid increases in one-year increments, reaching a maximum of 25 percent for recipients on aid for more than five years.

In addition, the model includes a provision for performance incentives, whereby the counties would have their share of costs reduced by up to 5 percentage points for achieving positive program outcomes. The outcome measures, for example, could be based on employment and recidivism rates, possibly accounting for demographic and socioeconomic factors related to time on aid.

Basing the state and county shares of cost on recipients' time on aid requires the means to track this time on a statewide basis. The state is currently in the process of implementing a statewide automation system. According to the Health and Welfare Agency Data Center, counties will be able to use an existing statewide data base (the MEDS file) in 1997-98 to track time on aid.

Program Implementation

A phase-in approach would reduce the initial costs needed to finance the services and make the community service job component more manageable from an administrative standpoint. On the other hand, it would delay the long-term savings that result from these components. A phase-in approach also would entail the operation of a dual system of TANF for an extended period of time.

We assume, in our estimates of the fiscal effects of the model (discussed below), that the program would be implemented on a statewide basis.

Program Evaluation

Fiscal Effects of the Model

We base our cost estimates on data from a variety of sources, including the state's GAIN evaluation, the CALDATA study on drug and alcohol treatment, and research on community service employment. We assumed annual increases in employment--based in part on the GAIN evaluation--from the combined effect of the program components. We discuss these impacts below in more detail.

Fiscal Impact on State and Local Governments. As Figure 7 shows, we estimate that our TANF approach would result in net total costs initially ($360 million in the first year) and net savings in later years ($120 million in the fourth year, increasing to roughly $650 million per year in the sixth and seventh years). The costs result primarily from the various services and the administrative and support costs of the community service jobs. The savings result from the effects of these interventions and the time limits (which result in increased employment and earnings).Considering the stream of total costs and savings to the state and counties, the "payback" period--after adjusting the future stream of costs and savings for inflation--would be 5.5 years. In other words, this is the point where the initial costs are offset by the savings generated in later years.

Uncertainty in Projecting Savings. It is important to recognize that from a fiscal standpoint, our approach entails an element of risk. Simply stated, the costs are more certain than the savings. We have provided the underlying rationale for assuming positive impacts from the welfare interventions in the model, and the research on the GAIN program helps guide us in making such assumptions. However, for some of the program components--for example, mandated community service employment when provided in the context of time-limited aid--the data are not adequate to estimate the impacts with a high degree of confidence. Thus, the actual impact on employment levels could be significantly less--or more--than we have projected; and as a result, the costs or savings could vary from our projections. This variation could be on the order of magnitude of several hundred million dollars in combined net state and county costs or savings over the seven-year period.

Summary of the Model for General Assistance

Figure 9 summarizes the GA component of the model, compared to current law.

Under our approach, grants for adults without children (GA recipients) would be uniform statewide, except there would be provision for variation according to regional cost differences, such as the existing adjustments for the AFDC program. The grants would be set at a level that approximates the projected average grants in the existing GA program--roughly $225 per month. The grant levels are designed to maintain an incentive to obtain employment and to control the costs of the program.

Services and Community Service Jobs

Time Limits

Under our approach, any such time limit would be imposed on a statewide basis. Rather than impose a limit of 3 out of every 12 months, however, we suggest basing the limit for adults without children on continuous time on aid, consistent with the type of limit applied to families with children. We believe that this approach would be more conducive to the establishment of a comprehensive case plan for these recipients (for example, a plan that includes a five-month job training program). The time limit, however, would differ from the limit on families with children in two respects.

First of all, we assume a substantially shorter time limit-- two years of time on aid for able-bodied recipients, with essentially the same exceptions and extensions noted for the families with children. Our approach reflects a shorter time frame primarily on the basis of data indicating that GA recipients generally are on aid for a much shorter period of time than are AFDC recipients. Statewide data are not available, but a study conducted by San Francisco County indicates that of those GA recipients who entered the program in 1991, about 21 percent were on aid two years later (in contrast to about 50 percent in a similar study of AFDC recipients).

Secondly, our approach assumes a complete cut-off of cash assistance after the two-year limit. As indicated previously, our model does not have a strict cut-off for families, in large part because of our concern for the welfare of those children who might be affected. With respect to adults who have no children and are able to work, we believe that the balance shifts in favor of a strict time limit, with the same qualifications for extensions as the TANF component.

Finally, we note that individuals who lose cash assistance due to the time limit would, under our model, continue to be eligible for job search services, as well as food stamps (subject to federal limitations) and indigent health services.

Funding Structure

Program Implementation

Fiscal Effects

Figure 11 summarizes the fiscal effects of the model, assuming implementation of the GA component (adults without children) three years after implementation of the TANF component (families with children).

This report was prepared by Chuck Lieberman and Todd Bland. The Legislative Analyst's Office (LAO) is a nonpartisan office which provides fiscal and policy information and advice to the Legislature. To request publications call (916) 445-2375. This report and others are available on the LAO's World Wide Web site at http://www.lao.ca.gov. The LAO is located at 925 L Street, Suite 1000, Sacramento, CA 95814.