February 2, 2000

Comments on the Proposed Rule for the Bonus to 
Reward States for High Performance

Summary of Recommendations

The Center on Budget and Policy Priorities and the Center for Law and Social Policy are submitting these joint comments in response to the Proposed Rule for the Bonus to Reward States for High Performance. We support the broad directions reflected in the proposed rule. We agree that state performance should be evaluated on both work-related measures and measures of state effectiveness in providing supports for low-income working families. We also agree that part of the bonus funds should be allocated to addressing the TANF goal of encouraging the formation and maintenance of two parent families. While we agree with these broad directions, we are proposing a number of specific modifications, for both substantive and technical reasons. We have six primary recommendations:

1. Retain and increase the bonus for state effectiveness in providing food stamps to low-income working families, awarding bonuses both for states that demonstrate the highest level of performance and states that demonstrate the greatest improvement in performance;

2. Retain and increase the bonus for state effectiveness in ensuring that adults and children leaving TANF assistance maintain health insurance coverage through Medicaid and/or CHIP, awarding bonuses both for the highest level and the greatest improvement in performance;

3. Add a measure of state effectiveness in providing child care subsidy assistance to low-income working families, awarding bonuses both for the highest level and the greatest improvement in performance;

4. In measuring the work-related performance of state TANF efforts, maintain a focus on job placement, employment retention, and earnings gains, but modify the measures to more effectively measure sustained employment and wage growth.

5. Add a measure of work-related performance that is not limited to families participating in or leaving TANF cash assistance, but that measures participation in employment among the entire population of low-income families with children;

6. Maintain a bonus for family formation and maintenance, but award funds based on a panel-based competition that rewards demonstration programs or innovative policies designed to encourage the formation and maintenance of two-parent families.


The 1996 welfare law established a set of goals for the TANF Program, and directed HHS to develop criteria for awarding to states a High Performance Bonus based on the measurement of state performance in achieving the goals of the law.(1) The proposed HHS approach to the bonus has three principal components: measurement of job placement, employment retention and earnings gains by TANF recipients entering employment; measurement of participation in food stamp program by working poor families and of participation in Medicaid/CHIP by adults and children leaving TANF assistance; and measurement of the share of children below 200% of poverty in married families.

In our specific recommendations for the high performance bonus, we emphasize a set of cross-cutting themes:

Our starting point is that the process of designing a high performance bonus in 2000 should be informed by the experience of TANF implementation since 1996. Broadly, that experience tells us that there has been a dramatic caseload decline, a significant increase in employment among low-income families, and a new set of issues now emerging for states and families as a result. Specifically:

As we look at these trends together, they lead to a set of conclusions that have shaped our comments on the high performance bonus regulations. First, it now seems clear that state TANF efforts, in connection with a strong economy and other federal and state make-work-pay efforts, have been successful in raising employment rates for families receiving TANF cash assistance. While efforts to promote workforce participation need to continue, it also seems clear that much of the initial employment is in low-wage jobs. Thus, the emerging challenge for states involves how to support employment retention and wage advancement and how to ensure that families in low wage jobs receive the work supports needed to sustain workforce participation and enhance the well-being of low-wage working families. The need to address supports such as food stamps, Medicaid/CHIP and child care for working families is not a diversion from the focus on employment; it is an essential part of the national strategy to encourage work and reduce the poverty of working families.

We understand that questions have been raised about whether the linkage to work supports should be a measure of state success in TANF implementation. We believe it is entirely appropriate, because it advances the first goal of TANF, i.e., providing assistance to needy families. Moreover, it also advances the second goal of TANF, because it is hoped that the provision of work supports will help families sustain and stabilize their workforce participation, so that they can progress over time to a point where they may no longer need government benefits.

In every state, there is a state agency that is primarily responsible for administration of the cash assistance component of the state's TANF efforts. As discussed in our detailed comments, there are an array of activities that this agency can do to improve linkages to food stamps, Medicaid, and child care. We recognize that other state agencies may have primary responsibility for one or more of the above work supports. However, that should not diminish the appropriateness of these measures for the high performance bonus. One can also observe that the performance of a number of other state agencies-the workforce agency, the employment service, the education agency, the vocational rehabilitation agency, the child support agency, etc. -may all have significant impacts in affecting the quality of a state's TANF implementation, but one does not conclude that it is inappropriate to measure employment outcomes simply because those outcomes are partially dependent on the actions of other state agencies.

Effective TANF implementation necessarily involves the coordination and participation of a large number of state and local agencies, businesses, nonprofit groups, and numerous other actors. Ultimately, the measure of TANF performance is not intended to be the measure of the performance of one agency, but rather the performance of the state. Hopefully, the incorporation of work supports in the high performance bonus-like the incorporation of measures of employment retention and earnings gains-will spur improved coordination and cooperation in state development and implementation of comprehensive strategies to enhance the well-being and support the progress of low-income working families.

Our second cross-cutting theme is the need to look at performance both for the families receiving TANF cash assistance and to also use broader measures for the entire population of low-income families with children. For example, we are recommending looking at workforce outcomes both for recipients of TANF cash assistance and for all low-income families with children. And, the food stamp measure and our proposed child care measures would be based on participation by the entire eligible population. We are not proposing a similar Medicaid measure, primarily because of data limitations, but over time, our view is that the best measures of performance would look both at families in and leaving TANF cash assistance and at the broader low-income population.

There are three principal reasons why we emphasize the need to measure performance based on broader population measures. First, because there are no uniform federal eligibility requirements for TANF, there are large variations between states in what it means to be a TANF cash assistance recipient. States vary in income and resource eligibility, family composition rules, work requirements, sanction policies, time limit policies, diversion policies, etc. Thus, looking at the families receiving cash assistance means very different things in different states, making it very difficult to compare state performance. And, those states with the largest caseload declines are sometimes suggesting that their remaining families have more severe barriers to employment, further exacerbating the problems of comparing employment performance among states.

Second, as the caseload has declined, a smaller and smaller share of low-income families are participating in state TANF cash assistance programs. The picture is clearest for those states with the largest caseload declines, but we have already reached the point where most poor families with children are not participating in TANF cash assistance programs. Families may not be participating because of eligibility rules, because they are working (and either are ineligible or choose not to participate), because restrictive policies have made it difficult to participate, or for other reasons. In any case, as the numbers receiving cash assistance become steadily smaller, the measure of state performance based on the outcomes for those in the cash assistance programs becomes a steadily less informative measure of state performance.

Finally, the structure of TANF itself and how states are actually spending monies suggests that the measure of state performance should not be limited to those families in the TANF cash assistance program. TANF funds can be spent for an array of purposes outside the TANF cash assistance program. Federal guidance has made clear that states can set a definition of needy families that is substantially above the TANF cash assistance benefit level and can use TANF funds for benefits and services to needy families outside TANF. TANF expenditures under the third and fourth purposes of TANF (preventing out of wedlock pregnancies, encouraging the formation and maintenance of two parent families) are not limited to needy families (if the expenditures are for purposes other than providing assistance). Under final regulations, states are free to use TANF funds for providing "nonassistance" in a number of circumstances, and families that are not receiving assistance are not counted as part of the TANF cash assistance caseload. And, states are free to use maintenance-of-effort funds outside of the TANF cash assistance framework, and the outcomes for families in separate state programs are not measured in a calculation of outcomes for recipients of TANF cash assistance.

Over time, a steadily smaller share of TANF funds are being directed to TANF cash assistance. The identification of high performers, then, should not be limited to measuring outcomes for families in the cash assistance program, but should look to outcomes affecting low-income families generally. For example, if a state transfers 30% of its TANF funds to child care, the state is presumably hoping that through expanding child care subsidies, fewer families will be in need of TANF cash assistance. Or, if a state commits TANF funds to a state earned income tax credit, the state is presumably hoping that it will be encouraging employment and reducing the need for TANF cash assistance among working families.

Because TANF is a block grant, it presents to states a set of choices about how to use its resources. The state may choose to use those resources in or outside of the state's TANF cash assistance program, or may choose not to spend those resources at all. The best way to measure the results of the choices made by states is by looking at the resulting outcomes for all low-income families.

This leads to our third cross-cutting theme: the need to structure bonuses that reward both absolute levels of performance and improvement. We understand that states began TANF at very different starting points, and that there is substantial variation between states on virtually every measure of performance. States have different economies, different demographic compositions, differing histories and capacities and differing TANF populations. There is no easy way to adjust for these differences when identifying high performers. We conclude that it is nearly impossible to adjust for all these differences fairly. So, throughout our recommendations, we have sought to propose awarding bonuses both to the highest performers and to the states demonstrating the greatest improvement from year to year. We think the high performance bonuses can send an important set of signals to states about goals to strive for. Hopefully, our knowledge about the practices associated with high performance can be enhanced by identifying those states with the highest levels of performance in accomplishing the goals of TANF. At the same time, all states should always have an incentive to improve their performance over time. We believe that rewarding both absolute levels of performance and improvement reflects the best way to strike this balance.

Attached is a table summarizing our suggested measures followed by detailed comments.

Center on Budget and CLASP Recommendations




# of Awards

Work Related Measures




Improvement in the Earnings Gap– improvement from the previous fiscal year in the reduction of the earnings gap as measured by the aggregate amount of earnings all low-income working families need to earn to get to the poverty line using Census data




Job Entry Rate– the rate at which unemployed or underemployed (less than half-time) recipients enter the workforce and remain employed for two consecutive quarters earning at least half-time minimum wages




Improvement in the Job Entry Rate– improvement from the previous fiscal year in the Job Entry Rate (see above)




Improvement in the Earnings Gain– improvement from the previous fiscal year in the earnings among unemployed or underemployed recipients who enter employment (as defined by the job entry measure) and who remain employed for one year earning at least half-time minimum wage earnings in the last quarter.




Work-Support Measures




Child Care Rate– percent of CCDF-eligible children receiving child care assistance–allocated equally by level and improvement from the previous fiscal year




Food Stamp Participation Rate– percent of eligible low-income working families (working at least half time at minimum wages) receiving food stamps–allocated equally by level and improvement from previous year




Medicaid/CHIP Participation Rate– percent of eligible TANF recipients and leavers receiving Medicaid/CHIP 6 months after leaving TANF–allocated equally by level and improvement from previous year




Non-Work Measure




Family Formation– innovative policy or program demonstrations that may lead to the desired outcomes as determined by a panel of experts



a small number

2 "TANF" means a measurement that includes the TANF recipients or families that recently left TANF. "Population" means a population-based measurement that includes low-income families regardless of whether they are receiving TANF.
Allocations in millions of dollars.
Numbers of awards are suggestions. In general, we believe the number of states receiving awards should be reduced, and that measurements broken out by level and percent improvement should have equal numbers of awards for each component.

Detailed Comments




We support inclusion of a measure of states' success in providing food stamps to eligible low-income working families. We believe that food stamps, along with health care coverage and child care assistance, are critical supports that states can provide for low-income working families. We recommend increasing the amount of the high performance bonus funds dedicated to the food stamp measure to $30 million.

Studies of people who leave TANF for employment show that most are making relatively modest wages, often below the poverty line. A recent Urban Institute report found that two-thirds of families that left TANF remained eligible for food stamps, but only about 40 percent of those who were eligible actually received food stamps.(2) Studies by the Food and Nutrition Service (FNS) have shown that historically fewer than half of working families who are eligible for food stamps and do not receive cash assistance receive food stamps.

Food stamps are, however, an important support for low-income working families — both those with recent experience receiving TANF assistance and those who never receive TANF either because they are diverted or because they never apply. A family of three with a parent who works 30 hours a week at the minimum wage (with $250 in rent and utility expenses each month) is eligible for almost $230 a month in food stamps. Receipt of food stamps would increase the family's pre-tax monthly income by 34 percent. The expansions of the Earned Income Tax Credit (EITC) in the early 1990s were designed so that full-time year-round minimum wage earnings plus the EITC and food stamps would bring a family of four to the poverty line.

Between 1996 and 1999 the number of food stamp participants fell almost 30 percent. Some decline was expected because of eligibility changes in 1996 and the continued strength of the economy. But a consensus has developed among researchers that these two factors do not explain the bulk of the decline. An analysis by FNS of participation decline between 1994 and 1998 concluded that only about a quarter of the decline can be attributed to lower participation among able-bodied childless adults and immigrants, the two groups whose eligibility was restricted in 1996. Participation of elderly and disabled citizens was almost unchanged. Thus, the remainder occurred primarily in families with children.(3) If these families have income that makes them ineligible for food stamps, then their non-participation should not be viewed as a problem, and the proposed food stamp measure would not include them. It appears, however, from state studies that have attempted to track people who leave TANF, that many of the families who no longer receive food stamps are still eligible.

We are increasingly convinced that the dramatic decline in the proportion of low-income families who receive food stamps is a serious problem that is contributing to the declining incomes of the poorest single-mother families. According to Census data, between 1995 and 1998 the disposable incomes of the poorest one-fifth of single-mother families with children declined by $349. On average, each family lost $488 of TANF income and $314 in food stamps. Their increased earnings and EITC did not make up for the lower benefits.

Given this evidence, we strongly applaud that states will be measured on their success providing food stamps to all low-income working families rather than only those who recently left TANF. Under the TANF law regarding the high performance bonus, states are to be judged based on their performance in meeting the multi-faceted goals of TANF. These goals address the well-being of all needy families with children, not only those that receive TANF assistance. Many states have been quite successful in moving people from cash assistance to employment and in lowering their TANF caseloads. TANF's five-year time-limit goes into effect in 2002. It is therefore especially critical in the period covered by the proposed rule that the high performance bonus examine state's performance on the well-being of all needy families.

Awarding a part of the high performance bonus based on a state's success in enrolling working poor families in food stamps makes good sense because it can encourage states to find innovative alternatives to practices that make receiving food stamps difficult for these families. For example, some states require working families to get their employers to complete a "wage verification form" every few months. This both forces the new worker, who is anxious to make a good early impression, to identify her or himself to the employer as a food stamp recipient and imposes extra work on the employer. These forms, however, are not required by federal law. Indeed, the regulations give states flexibility to accept any reasonable documentary evidence provided by the household. States that rely primarily on households' wage stubs, for example, or on electronic verification of households' earnings through links with other state databases likely would make receipt of food stamps a viable option for more working poor families. This also would reduce the burdens on employers who have been willing to support welfare-to-work efforts by hiring people seeking to leave cash assistance programs.

Similarly, requirements that working parents make frequent visits to food stamp offices is a major barrier to access to the food stamp program. Guidance USDA issued last summer allows states to limit face-to-face interviews to one per year, regardless of the length of the household's certification period or other options the state may have taken. That same guidance allows states to limit working households' reporting obligations to changes in status: going from part-time to full-time work, changes of employer, and changes in hour wage rates. It also allows states to adopt quarterly reporting: requiring working families to submit a complete report every three months, but not requiring the family to report routine changes between quarterly reports. Even where states choose not to adopt these reporting changes, they can allow households not to report changes of less than $100 per month in their earnings. The $100 threshold, and certainly status reporting and quarterly reporting, should free both households and state agencies from most of the more onerous burdens associated with continued receipt of food stamps. This is likely to cause more working families to conclude that receipt of food stamps is worthwhile. States electing these options to reduce burdens on food stamp participation are serving important purposes of welfare reform and can properly be rewarded with the high performance bonus.

President Clinton has recently recognized the need to improve access to low-income working families through the food stamp initiatives he launched last summer. (See FNS guidance, July 14, 1999.) In addition to the reporting changes mentioned above, the new policy allows states to extend eligibility to low-income working families without regard to the value of a motor vehicle by expanding categorical eligibility. States have been troubled that working families who need reliable cars to commute to work often lose food stamp eligibility when their earnings reach a level that disqualifies them for cash assistance under TANF. Similarly, under the prior policy, a family that did not want to receive TANF assistance but did want food stamps would not be eligible for food stamps if it had a car valued above the food stamp vehicle threshold, but would be eligible for food stamps if it applied for and received TANF cash assistance. Now, under the administration's new guidance on categorical eligibility, states have a concrete way to remedy this inequity. In fact, this particular policy change is closely related to TANF in that states can now extend food stamp eligibility by using TANF funds to assist families that have started work and ceased to receive cash assistance.

In addition to taking advantage of these new options for expanding eligibility and reducing procedural barriers, there are many other tools available to states to increase the participation of low-income working families in food stamps. They can make sure that potential recipients are informed about their eligibility and simplify application processes. In particular, they can improve links with other programs that provide benefits to working families not receiving TANF, such as Medicaid and child care. In their TANF work retention efforts they can highlight the availability of food stamps and ensure that eligible families maintain their benefits.


Specific Recommendations

Under the proposed rule, states would compete for high performance bonus funds based on the number of low-income working families (defined as families with children under age 18 who have an income less than 130 percent of poverty and earnings equal to at least half-time full-year minimum wage) who received food stamps as a percentage of the number of such low-income working families in the state. HHS would award a bonus to the ten states with the greatest percentage improvement in this measure between the two prior calendar years, based on Census Bureau decennial and annual demographic program data. Ten percent of the annual total amount available for the high performance bonus (or $20 million) would be allocated to the food stamp measure.

Increase the allocation for the food stamp measure. We strongly recommend that at least $30 million of the high performance bonus funds be allocated to the food stamp measure. As discussed above, given the importance of food stamps to the well-being of families with children, the tools that states have to influence participation, and the ability to measure states fairly and accurately, an amount at least this large should be in the final regulations.

Reward states based on their level of performance on the food stamp measure, as well as on improvement. The final rule should be changed to award five bonuses based on the percentage point improvement over the prior year and five based on the level of the measure. With the current structure for the award, a state that goes from having 30 percent of its low-income working families receiving food stamps to 33 percent (a 10 percent increase) would receive a bonus, but one that goes from 60 to 65 percent (an 8 percent increase) would not. The bonus should be used to promote continued improvement of all states, which can be accomplished by splitting the bonus into two and providing awards to both the highest performers and those with the greatest improvement.

Eliminate the qualifying conditions. The proposed rule would instruct the FNS to determine whether a state is meeting four conditions in order for the state to be eligible for the food stamp portion of the bonus. We recommend that these conditions be dropped. The conditions, which are all required under current law or regulation, include (1) informing individuals of the opportunity to apply for food stamps on their first contact with the state agency, (2) making application forms readily accessible and available upon request, (3) complying with application processing time frames and expedited service rules, and (4) taking steps to prevent inappropriate denials and terminations of eligible food stamp participants who have lost TANF eligibility. On the first two conditions FNS would base its determinations on the state's policy instructions or regulations. On the second two conditions FNS would conduct administrative reviews in addition to relying on state policy materials.

While we ardently support the premise that states should be in compliance with these four features of current food stamp policy, we oppose requiring FNS to go through the additional effort to certify such compliance for purposes of the High Performance Bonus. Our concern is that the agency would not have additional resources to undertake the new determinations and, as a result, would certify that states are in compliance based on incomplete information. If, at a later date, the agency did discover noncompliance with these policies the earlier certification could interfere with administrative or legal actions it might want to undertake.

As an alternative, we suggest that the regulations require states to submit certifications to HHS that they are in compliance, but not require agency verification. Under the latter recommendation, if the agency had evidence that the state was out of compliance with a particular requirement it could deny the state a bonus.

Consider whether to adjust the data to account for different sized immigrant populations in different states. The measure proposes to use the number of working families with children with income below 130 percent of poverty to represent those families who are likely to be eligible for food stamps. Federal law makes most legal immigrants ineligible for food stamps. Without an adjustment, states with a disproportionate number of immigrants would be at a disadvantage unless they had chosen to provide state-funded food stamps to all otherwise eligible legal immigrants. States that have provided these state-funded benefits could be rewarded for providing the benefits, but states with high immigrant populations should not be penalized if they do not fully offset the effects of the federal treatment of legal immigrants using their own resources. The data under consideration should provide enough information on citizenship status to allow for an appropriate adjustment.

Be more specific in the final rule about what data will be used. Under the proposed rule HHS would measure performance using "Census Bureau decennial and annual demographic program data." We understand that the Department will be using data from the American Community Survey when those data becomes available, as well as data from Census 2000 supplementary survey and other data to gauge state performance on the food stamp outcome measure and the family formation measure. We agree that these data, particularly the ACS data, are appropriate for this purpose. They will be the best data sources for measuring food stamp outcomes for the low-income population with children at the state level. At full implementation in 2003, the ACS will have a sample size of three million households, far larger than any other Census survey. It will be able to provide statistically valid demographic data at the state level on an annual basis, something that has only been available in the past only once every ten years from the Decennial Census. The data necessary for the food stamp outcome measure — families with children less than 18 who have an income below 130 percent of the poverty level and earnings at a given level — are only available from the 2000 Census or the ACS.

Because these are new surveys and it is not clear that the funding will be available for them, the final rule should specify what alternative data would be used if these surveys are either not available or deemed not statistically reliable at the state level. One alternative would be to use data from the annual Food Stamp Integrated Quality Control System for the numerator of the measure and data from the Census Bureau's projections of the number of children in each state for the denominator. If the measure used these data it would have to measure only improvement in performance because states have very different economic and demographic circumstances. We would recommend, however, that if there is a need to use improvement only, that HHS measure improvement from a base period, such as 1996, so that states that performed well in periods before the measure became effective would not be disadvantaged. In addition, if other data must be used, it may be necessary to include all food stamp households with earnings as opposed to just those with earnings above half-time full-time minimum wage in order to increase sample sizes.




In general, we support the health-care-related provisions included in the proposed regulations on the high performance bonus. In particular, we endorse the notion that the success of a state's welfare program should be evaluated in part on whether families still have Medicaid or CHIP after leaving welfare. Families that have left welfare often do not have access to affordable employer-sponsored coverage. For these families, Medicaid and CHIP coverage can play a crucial role in helping them to maintain their health and well being, as well as in supporting their efforts to stay in the workforce. States that succeed in assuring that their welfare policies do not have the consequence of causing working families to lose out on health care coverage should be recognized and rewarded for their efforts. Accordingly, our comments on this measure are relatively minor and technical with the exception that we recommend increasing the amount of the high performance bonus funds dedicated to the health care measure to $30 million.

We recommend retaining the health care measure and increasing the amount of money dedicated to it for the following key reasons:

The proposed measure focuses on measuring a positive outcome that is clearly consistent with one of the most significant goals of the 1996 federal welfare law—ending the dependence of needy parents on government benefits by promoting work. A growing body of academic research suggests that providing health care coverage to low-income families makes it easier for families to stay in the job market rather than enroll in welfare. For example, a study conducted by the Minnesota Department of Human Services found that the state's health insurance program for low-income families effectively reduced welfare caseloads by 9.6 percent by deterring families from ever applying for welfare and by making it easier for families to leave welfare once they were enrolled.

In the absence of Medicaid and/or CHIP coverage, most of the families that leave welfare will have little access to affordable health insurance coverage. Numerous exit studies from various states indicate that parents leaving welfare for work frequently end up employed in low-wage jobs that are unlikely to receive employer-sponsored health insurance coverage. According to one recent study, only 43 percent of workers who make $7 an hour or less are even offered health insurance coverage. Moreover, even in a case where a family has access to and uses employer-sponsored coverage, it still is important that the family retain its Medicaid coverage after leaving welfare. Although the employer's health plan will pay much of the family's medical bills, Medicaid plays a crucial role by helping such a family meet premium and cost-sharing obligations, as well as in providing coverage for services not included in the employer's health plan.

States have extensive opportunities to influence the extent to which families leaving welfare remain enrolled in health care coverage. For example, states can help to assure that families leaving welfare for work receive the Transitional Medicaid Assistance for which they are eligible by training caseworkers about the availability of TMA, educating families about TMA, and minimizing the verification requirements necessary to establish that someone has left welfare due to earnings. States also have a number of options for assuring that families who leave welfare for a reason other than earnings receive the health care coverage for which they are eligible. For example, they can help assure that families do not inappropriately lose their Medicaid coverage when sanctioned under TANF. These steps include training caseworkers about the Medicaid implications of a TANF sanction, providing sanctioned families with understandable information about the circumstances under which they continue to be eligible for regular Medicaid, and updating their computer systems to assure that they do not automatically terminate the Medicaid coverage of someone who loses TANF due to a sanction. (Note that children, unless they are minor heads of households, cannot automatically lose their eligibility for Medicaid when their families lose TANF due to a sanction and, in all but 13 states, other family members also remain eligible for Medicaid.)

Both national and state studies indicate that parents and children leaving welfare (as well as those who are diverted from welfare) are at high risk of joining the ranks of the uninsured even though they generally remain eligible for coverage under Medicaid or CHIP.

These studies strongly suggest that states have room to improve their performance in helping families maintain their health care coverage after leaving welfare.


Specific Recommendations

Under the proposed rule, states would compete for high performance bonus funds based on the number of individuals receiving TANF benefits who are also enrolled in Medicaid or CHIP, who leave TANF and are enrolled in Medicaid or CHIP in the sixth month after leaving TANF. HHS would award a bonus to the ten states with the greatest percentage improvement in this measure between the two prior calendar years. Ten percent of the annual total amount available for the high performance bonus (or $20 million) would be allocated to the Medicaid/CHIP measure. In general, we agree with the decision to allow states to compete on the basis of their success in providing Medicaid and/or CHIP coverage to the members of low-income families who have left welfare. This measure captures an important aspect of states' performance, and we believe that states have the capacity to provide reliable data to support a fair evaluation of their performance on this measure. Here are our specific technical comments on the portion of the proposed rules relating to Medicaid participation.

Increase the allocation for the Medicaid/CHIP measure. We recommend that at least $30 million of the high performance bonus funds be allocated to the Medicaid/CHIP measure. As discussed above, given the importance of health care coverage to the well-being of families with children, and the tools that states have to influence participation, an amount at least this large should be in the final regulations.

Reward states based on level of participation in Medicaid/CHIP, as well as on improvement in participation rates. Although it is somewhat ambiguous, the proposed health care measure apparently would distribute bonus funds to states with the largest percent increase from one year to the next in the Medicaid/CHIP participation rate among people who left TANF assistance six months earlier.(6) This measure would make it difficult for states with a history of success on the health care measure to win bonus funds, even if they improve upon their successful records. For example, consider a state in which only 30 percent of people who have left TANF are enrolled in Medicaid or CHIP six months later. If this low performing state increases its CHIP/Medicaid participation rate to 40 percent, it would be more likely to win a bonus than a high performing state that increases its health care participation rate from 90 percent to 95 percent. Accordingly, we recommend that the health care-related bonus funds be divided among the set of states with the highest Medicaid/CHIP participation rates and the states with the highest percentage point increase in this participation rate.(7) By rewarding states for their performance, as well as for improvement in their performance on this measure, HHS would make the health-care related measure(s) more equitable and also increase the incentive for states that already perform moderately well on the health care measure to compete for bonus funds.

Measure individuals rather than families. We particularly commend the decision to evaluate states on the percentage of individuals that are enrolled in Medicaid or CHIP, rather than the percentage of families with one or more members on Medicaid or CHIP. The studies conducted to date on the percentage of individuals with health insurance coverage after leaving welfare highlight that often only selected family members retain their health care coverage. In particular, parents appear to be at much greater risk than children of losing out on the health care coverage for which they are eligible. We also recommend that states be required to provide data separately on the percentage of parents and children with Medicaid and/or CHIP leaving welfare.

Evaluate annually the feasibility of a broader health care measure. As indicated previously, we support a measure of health care coverage for families leaving welfare. At the same time, the proposed measure does not provide any indication of state performance in providing health care coverage to the members of low-income families diverted from TANF assistance or who elect not to apply for TANF. Like families leaving TANF, low-income families who have never been enrolled in TANF often need Medicaid and CHIP coverage in order to maintain self-sufficiency and to improve child and family well being. It is as important that these families receive the health care coverage for which they are eligible as it is to assure that families leaving TANF do not inappropriately lose health care coverage. At this time, we are not recommending an alternative health care outcome measure that addresses this issue because of the lack of reliable state-specific survey data that can be used to evaluate a state's success in assuring that low-income children and parents have health care coverage. However, we recommend that HHS review annually whether such data has become available, and, if it does, to consider a measure that allows states to compete on their success in assuring that low-income children and parents have health insurance coverage, regardless of their recent receipt of TANF.

Eliminate or amend the qualifying conditions. As currently drafted, Section 270.4(d)(1) requires a state to meet various "qualifying conditions" before it can compete on the health care measure. The "qualifying conditions" require states to comply with various requirements of federal Medicaid law primarily related to the delinking of TANF and Medicaid eligibility. They include the requirements that states (1) provide families with the opportunity to apply for Medicaid at first contact with a TANF agency; (2) inform families via a written notice of their eligibility for Transitional Medicaid Assistance when they become ineligible for regular Medicaid due to earnings; (3) retain individuals' Medicaid coverage until they have been determined not to be eligible for Medicaid under any other category; and (4) meet CHIP and Medicaid data reporting requirements.

Although we believe it is essential that HCFA assure states are in compliance with Medicaid law, we do not believe that the high performance bonus provides an appropriate vehicle for HCFA to evaluate or verify whether a state is in compliance with the qualifying conditions. We recommend that the final regulation either drop the qualifying conditions or, at a minimum, clearly establish that an HHS decision to consider a state for the health care measure does not mean the federal government has "evaluated" or "verified" that the state is in compliance with federal law in these areas. The preamble language also should be corrected to remove any references to HCFA "verifying" or "evaluating" that states competing on the health measure are in compliance with federal Medicaid law.(8)

Eliminate the qualifying options. As currently drafted, Section 270.4(d)(2) requires a state to implement at least two of seven "qualifying options" as a condition of competing on the health care measure. The options include simplifying enrollment and re-enrollment procedures; accepting mail-in or phone-in application for Medicaid; adopting "less restrictive methodologies" when evaluating families with children for eligibility for Medicaid; and using a definition of "unemployed parent" when evaluating the eligibility of two-parent families with children for Medicaid that includes parents who are employed more than 100 hours per month.

We agree states should be encouraged to take advantage of the options available under federal law to promote participation in Medicaid and/or CHIP, but many of the qualifying options listed in Section 270.4(d)(2) are not particularly meaningful. For example, it appears that a state could meet the "qualifying options" requirement by limiting coverage of two-parent families to those in which the principal wage earner works fewer than 101 hours (rather than 100 hours) a month and by adopting a "less restrictive methodology" of disregarding $95 of a worker's monthly earnings (rather than $90). HHS should eliminate the qualifying options.

Consultation. As part of its consultation with interested parties, we believe that HHS should indicate in the final regulation that it will post the data submitted by states competing on the high performance bonus measures on the Internet.




We recommend that in addition to measuring state performance relating to participation in food stamps and Medicaid, HHS should also include a measure of state performance in providing child care subsidy assistance to families eligible for assistance under federal Child Care and Development Fund (CCDF) eligibility rules. Specifically, we recommend that states be ranked on the share of CCDF-eligible children receiving child care subsidy assistance, and that bonuses be awarded for the five states in which the highest percentages of eligible children receive assistance and the five states demonstrating the greatest improvement (measured in percentage point increases) from the prior year. We further recommend that a qualifying condition for the bonus be that the state's payment rates for center-based care and family day care be set at levels not lower than the 75th percentile of the local market rates in the state.

The rationale for including a measure of performance in relation to child care is similar to the rationale for including such a measure for food stamps and Medicaid. Child care assistance can be a crucial work support in making it possible for low-income families to accept and maintain employment. Moreover, child care assistance can help to reduce the out of pocket costs faced by low-income working families, and in doing so, can increase the resources available to meet other family needs. Thus, the provision of child care subsidy assistance advances both the TANF goal of promoting work and the goal of providing assistance to needy families. Indeed, the preamble to the proposed regulations expressly states that "The new law reflects widespread, bipartisan agreement on a number of key principles" including the principle that "Parents should receive the child care and the health care they need to protect their children as they move from welfare to work." 64 Fed. Reg. at 68203.

The preamble to the proposed rules does not expressly state why HHS opted to include bonuses in relation to food stamps and Medicaid but not child care. It is possible that this decision was based, in part, on the fact that there has been considerable attention and concern based on evidence of declining participation in food stamps and Medicaid after families leave TANF assistance. While that declining participation certainly presents reasons for immediate concern, we also believe that in the long run, the challenge for all states is to focus attention on efforts to link low-income working families with important work supports, and surely child care is a critical part of that effort. Recent research suggests that there is room for improvement both in providing child care assistance for families leaving TANF and for other low-income working families. A review of state leavers' studies concluded that in those studies, the majority of families working after leaving TANF assistance were not receiving child care subsidy assistance, and that in most of the states with data, only about 25-30% of working leavers were receiving subsidies.(9)  And, HHS analysis indicates that only about 10% of children eligible under federal CCDF income guidelines were receiving CCDF-funded child care assistance in 1998.(10)

In developing the proposed rules, HHS may also have distinguished child care from Medicaid and food stamps because the latter programs are federal entitlements for eligible families while child care is not a federal entitlement. However, the fact that a particular benefit is not a federal entitlement should not preclude HHS from measuring state performance and awarding bonuses for superior performance. For example, there is no "entitlement" to retention or advancement services, yet HHS proposes to award bonuses for state performance in addressing retention and advancement. And, although there is not an open-ended federal matching funding stream for child care, states can directly affect the share of eligible families receiving subsidy assistance through their decisions about how to spend (and whether to transfer) TANF funds and how to use maintenance-of-effort funds.


Specific Recommendations

We are recommending that child care performance be measured based on the share of CCDF-eligible children receiving child care subsidies. We understand that an estimate of the numbers of CCDF-eligible children in each state can be derived from Census data. The number of children in a state receiving CCDF-funded subsidies can be calculated from CCDF administrative data reported by states to the federal government. We recognize that a state may also be providing child care subsidies through direct expenditure of TANF funds or through expenditure of state funds not included in CCDF reporting. We suggest a state be permitted to supplement CCDF reporting with reporting of participation in other non-CCDF-funded child care programs. If a state did not elect to supplement CCDF reporting, the calculation of state rankings could be conducted by HHS without the need for reporting of any additional data by states.

As in our food stamp and Medicaid measures, we are recommending that the bonuses be allocated both for the states with the highest level of performance and for states with the greatest improvement from the prior year. Our reasoning is similar to that relating to food stamps and Medicaid: it is important to recognize those states with the highest level of performance, but also important to structure a bonus for which all states can compete by improving their performance from year to year.

In structuring our recommendations, we considered whether it would be preferable to base the bonus on the share of CCDF-eligible children receiving assistance or the share of TANF recipients and employed TANF leavers receiving subsidies. We ultimately opted for the CCDF measure for three reasons. First, we recognized that it might be more difficult to obtain consistent and reliable information on work status and child care program participation by TANF leavers. Second, we appreciate that some states have made a policy decision to structure a "unified" child care system in which there is no eligibility distinction made between TANF recipients and leavers and other low-income working families. It would seem inappropriate to award a bonus to a state that was doing a "better" job of assisting TANF leavers simply because fewer child care resources were being made available to low-income families outside TANF. Finally, we recognize that a number of states have used TANF funds to significantly increase the availability of child care assistance outside the TANF cash assistance system in an effort to reduce the need for TANF cash assistance. The efforts of those states would not be reflected if performance were solely measured based on receipt of subsidies by TANF families and TANF leavers.

Finally, we recommend that a qualifying condition for the bonus be that the state's child care payment rates for center-based care and family day care are set at rates no lower than the 75th percentile, based on a market rate survey conducted within the past two years. In promulgating CCDF regulations, HHS had indicated that payments established at least the 75th percentile of the market rate would be regarded as providing equal access for CCDF-funded children to the care available to families that are not eligible to receive subsidy assistance. 63 Fed. Reg. 39959 (July 24, 1998). We believe that a qualifying condition of this nature is important because otherwise, a state that was providing minimal subsidy assistance to a large number of families would be measured to be a "high-performing" state.




The proposed regulations would measure state work performance by calculating a job entry rate and a success in the workforce rate. The success in the workforce rate would be based on a combination of scores from a job retention rate and an earnings gain rate. All of these measures would be calculated from a base of families receiving TANF cash assistance.

We agree that job entry, employment retention, and earnings gains are appropriate outcomes to measure in a performance bonus. However, we believe that the measurements as described in the proposed regulations can be improved to more effectively measure these outcomes. As described in Section 5, we are recommending that one measure be an earnings gap rate measured for all low-income families in a state. For those families receiving TANF assistance, we are proposing two measures: a sustained employment measure and an earnings gain measure.

This section outlines our concerns about the proposed measures, suggests ways of strengthening those measures, and describes the rationale and a proposed design for a sustained employment measure and an earnings gain measures as potential alternatives.


Specific Recommendations

Measure Sustained Employment. Under the proposed regulations, bonus funds would be awarded for performance on a job entry rate calculated based on the unduplicated number of adult recipients who were unemployed at some point in the performance year and who then enter not fully subsidized employment at some point in the performance year. The proposed regulations do not define what it means to be "unemployed at some point" and do not set any minimum threshold for what is required to count as having entered not fully subsidized employment.

We have concerns about the job entry rate, based both on the methodology as outlined in the proposed rule and on the initial experience with a job entry rate in the High Performance Bonus data released to date. The fact that there is no definition of "unemployed at some point" means that states could interpret the language differently. For example, one state might define "unemployed at some point" as requiring a month of unemployment, while another might require a day or week of unemployment. At minimum, there should be a clear and consistent definition.

Even with a consistent definition of "unemployed," the rate is still of limited utility if there is no minimum threshold requirement for employment. Again, a state could count an individual who was employed for a day (or an hour) as having been employed. If the rate is to be maintained, we suggest that a uniform minimum threshold for being "employed" should be established.

We are concerned about the utility of the proposed job entry rate, in part, because some seemingly anomalous results were apparent in the initial state reporting using a similar definition in High Performance Bonus reporting for 1998. In the 1998 results, there were extreme variations between states in reported job entry rates, with three states reporting rates of 20% or less and one state reporting a rate of 88%. While it is not impossible that performance could have varied so much, the magnitude of the range suggests the need to examine the data more carefully. And, other anomalies were also apparent. For example, the calculation of the job entry rate is supposed to be based on the unduplicated number of adult recipients who were unemployed at some point in the year. Since this is an unduplicated count, one would expect the figure to grow at some reasonable pace throughout the year. However, one again sees large variations: at one extreme, in one state, the fourth quarter figure is only 11% larger than the first quarter figure, while at the other extreme the figure is 115% larger. And, there seems to be no consistent relationship between the fiscal year denominator and the state's unemployed adult caseload for the fiscal year.

We are unable to determine whether these anomalies occurred because of problems in the accuracy or consistency of data submitted by states, or because states were interpreting the instructions differently or if there were other undetermined problems in the nature of the job entry measure. At minimum, these results suggest the need to clarify instructions so that states are consistently interpreting the directions as to which cases should appear in their numerators and denominators.

We also have concerns about the calculation of the job retention rate under the proposed rule. The job retention rate would be calculated based on the unduplicated number of employed adult recipients who were also employed in the first and second subsequent quarters. We do consider this proposed rate an improvement over the one used until now, which has measured job retention based on being paid for work performed in two consecutive calendar quarters. Our concern under this prior definition was that since there was no minimum hours or earnings requirement so long as the individual worked in the quarter, an individual could be measured as having worked in two consecutive quarters if, for instance, he or she was employed for two days, i.e., on March 31 and April 1. While this problem has been alleviated somewhat by requiring work in an initial quarter and a first and second subsequent quarter, we believe there are still two limitations to the proposed measure. First, it imposes no minimum earnings requirement to constitute having worked in a quarter. Thus, one will know the number of individuals that have worked in three consecutive quarters, but will not be able to tell if such individuals worked for nine months or three days. Second, the proposed measure does not appear to calculate employment retention in a consistent manner from the point of job entry. Rather, the individuals employed in Quarter 1 and tracked for the next two quarters could be individuals that had already been continuously working for an extended period of time before Quarter 1. Prior research has found that the highest risk of job loss occurs in the initial months of employment. Thus, mixing individuals newly entering employment with individuals who have already been working for some time will mean that the data will not provide a clear picture of the share of new job entrants retaining employment.

Thus it would be an improvement to modify the job retention measure to calculate retention in the first two quarters after the initial quarter of employment entry. However, the problem of having no minimum threshold for counting as "employment" would remain.

Our alternative suggestion would be to develop a single sustained employment rate in lieu of the proposed job entry and job retention rates. Under our proposal, the base would be the entire caseload of adult TANF recipients, excluding only those adults who were already engaged in employment with earnings equal to or greater than the equivalent of 20 hours times the minimum wage for the thirteen weeks of the quarter. Then, an individual would count as having obtained sustained employment if, during the next year, he or she had at least two consecutive quarters with earnings equaling or exceeding the threshold (i.e., 20 hours times minimum wage times 13 weeks).

This sustained employment measure would offer several advantages over the approaches in the proposed rule. Rather than rewarding brief, sporadic, unstable employment, it would reward those states that were most successful in helping individuals enter or retain employment at a substantial level over a two-quarter period. And, because the base would exclude adults already working at or above the 20-hour-a-week equivalent, it would measure movement from below a threshold to at, or above, a clear, consistent work threshold. Moreover, no state would win a bonus for job entries if the state was not also able to demonstrate that individuals who entered employment were retaining that employment for at least two quarters.

As with our other measures, we are recommending that bonuses be awarded both for absolute level of performance and for improvement from prior performance, with $25 million allocated to each measure.

Measure Earnings Gain Rate. The proposed regulations would require the calculation of a measure of earnings gain over a two-quarter period. Though earnings gain would be separately calculated, states scoring highest in earnings gain might not qualify for a high performance bonus, because the measure of earnings gain and the measure of employment retention would be consolidated into a single "success in the workforce" measure, with greater weight given to employment retention.

We agree that it is important to look at earnings gains, but are recommending some modifications to the measure and how it is used. Specifically, we recommend:

First, we agree that earnings gain should be measured and reflected in the determination of the high performance bonuses. A consistent finding across state studies of families leaving welfare has been that the initial wages of employed welfare leavers are higher than the minimum wage, but not sufficient to reach the poverty line. The Urban Institute's National Survey of America's Families (NSAF) reported median wages for welfare leavers of $6.61 an hour, and reports of average wages are often at a similar level.

In structuring their TANF programs, many states articulated a "work first" strategy, encouraging families to take the first available job. The underlying premise of this strategy was that it was important for parents to rapidly connect with employment, and that once employed parents could move up into better jobs, either through normal labor force progression or through participation in education, training or other advancement activities. In light of this commonly-shared approach, and in light of the fact that parents are typically beginning employment with below-poverty wages, it would be extremely valuable for HHS to encourage consistent data collection to better understand the extent of upward movement over time and to reward those states that are most successful in increasing the earnings of families entering employment.

We appreciate that HHS has already recognized the need to measure earnings gains in the initial High Performance Bonus competitions and in the proposed regulations. We are concerned, however, that the initial measure may need to be adjusted to more effectively measure wage gains, and are therefore suggesting several revisions.

The proposed regulations are not entirely clear as to how the calculation of earnings gain would be made. The proposed regulations indicate that a comparison would be made between the initial and second subsequent quarters for adult recipients employed in both of those quarters. However, the regulations do not define initial quarter. We assume that it is intended to mean the first quarter in the fiscal year in which employment earnings appear, but were unsure whether the intent was to measure the first quarter after a quarter of unemployment, or the first quarter after any period of unemployment. At a minimum, this needs to be clarified.

In any case, we think there are two principal problems with comparing the initial quarter of employment to the second subsequent quarter. First, the initial quarter of employment will typically be a partial quarter because the individual may not enter employment on the first day of the quarter. Thus, a part of the difference between the initial and second subsequent quarter is simply a reflection of the difference between a partial and full quarter. Second, the gap between the initial and second subsequent quarters is too short to meaningfully measure earnings gains. For example, some states are seeking to encourage participation in education or training after a parent has stabilized in employment, and some projects seek to help individuals move into higher paying jobs after demonstrating a sustained work history. The effectiveness of initiatives such as these would not likely be reflected in earnings gains within six months. While there is no clearly correct standard, looking at progress over a one-year period would be a more meaningful measure. Since this measure would be based on unemployment insurance wage record matching, it would not call for sustained tracking and follow-up by states not otherwise wishing to do so.

One approach to dealing with the partial quarter problem could just be to say that the base for calculating earnings gain would be the first quarter after the initial quarter of employment. While we believe that would be an improvement over the proposed measure, we are also suggesting a further refinement: we propose that earnings gains be calculated based on the increase from Quarter 2 to Quarter 6 (i.e., one year) for those parents who a) are either unemployed or have earnings below 20 times the minimum wage times 13 weeks in Quarter 1; and b) have earnings of at least 20 times the minimum wages times 13 weeks in Quarters 2 and 6.

The reason for our proposed measure is to attempt to better measure real gains in wages as opposed to the increased earnings that result from increased hours or weeks of employment. Earlier research on the employment experience of AFDC recipients has concluded that most of the gains in earnings were attributable to increased hours or weeks rather than increased wage levels.(11) Advancement strategies are typically seeking to increase job quality, and will not be well reflected by a measure of aggregate wages that cannot distinguish between earnings increases due to higher wages and increases due to more hours of work. There is no perfect resolution of this difficulty when using UI wage records, but we think that by imposing a work threshold (20 hours times minimum wage), the result is more likely to better reflect the extent of increases in wages.

Finally, we are recommending that earnings gain be provided its own bonus rather than combined with the employment retention measure. The proposed regulations would combine the two, giving earnings gain less weight. The preamble explains "We believe that earnings gain is dependent on job retention, and therefore, should be given a lesser weight." 64 Fed. Reg. 68213. This statement is true in the narrow sense that an individual cannot have earnings gain unless he or she remains employed. However, research findings are suggesting that the factors affecting employment retention and the factors affecting advancement are not the same, and that the policies that support employment retention may, in themselves, do little to promote advancement.(12) And, at least in the initial results for the high performance bonus, there is little if any relation between the states that had the highest rankings on retention and those with the highest rankings in earnings gains. For example, Hawaii ranked 1st on retention and 43rd on earnings gain; Connecticut ranked 2nd on retention and 31st on earnings gain. Conversely, South Dakota ranked first on earnings gain but 34th on retention, and Kansas ranked 2nd on earnings gain and 29th on retention. While there are many questions about how to understand these results, they certainly do not support a measure which treats earnings gains and retention as measuring similar or closely related outcomes.

We should observe that while we support an earnings gain measure, we have one concern that merits further examination over time. A broadly shared goal is that families are able to stabilize in family-supporting employment. Some states may attempt to reach this goal by focusing on rapid employment placement and post-employment services; other states may attempt to reach the goal by stronger emphasis on employment preparation activities and more focus on the quality of initial job placement. Ideally, a federal measure of performance would not "prefer" one strategy to the other, but would focus on the ultimate outcomes, e.g., at some subsequent point, what are the earnings of employed families in the state. We do have some concern that a measure of earnings gain could unintentionally have the effect of treating those states that begin with the lowest base earnings as the states that demonstrate the most "earnings gains," at the expense of those states paying more attention to initial job quality. At this point, the available data is probably not sufficient to know whether this concern is justified, but we would urge HHS to closely examine the results of this measure, and try to develop a better understanding over time of the factors associated with measured high and low performance.

Reconsider Treatment of Separate State Programs. The preamble to the proposed regulations expressly recognizes that the measure of a state's TANF performance could be affected by the state's choices about who is and is not assisted in the TANF assistance program. The preamble observes that a state could, for example, shift families with the most significant barriers to employment into a separate state program, thus artificially elevating the measured performance of the TANF program. It is true that such a result could happen, but that is just one of a number of ways in which the states' TANF and maintenance-of-effort policies could limit the comparability of performance across states. For example, a state with more restrictive sanction policies or more rigorous up-front requirements could make it difficult or impossible for those families with severe barriers to employment to ever receive or continue to receive assistance. A state that allows exceptions to time limits for families in which a parent or ill, disabled or elderly may have a lower rate of employment entries among the TANF population than a state with a more restrictive time limits. A state that categorizes cases in which a grandparent is caring for grandchildren as child-only cases may demonstrate a higher job entry rate among adults receiving assistance than a state that provides assistance for the grandparents.

There is no easy way to adjust for the fact that measured state performance will be affected by state eligibility rules, caseload demographics, policy choices about whether and how to assist families and a multitude of other factors. That is one reason why we emphasize the need for population-based performance measures. However, for any bonus that is based on performance relating to the TANF population, we do think that it is inappropriate to single out the use of separate state programs among the array of potential factors that could affect measured performance. Either HHS should make a commitment to scrutinizing the entire set of policy choices that affect the composition of a state's TANF assistance program, including those policies that have the effect of restricting assistance for families with barriers to employment or HHS should simply base its measures of performance relating to TANF assistance on the families that receive TANF assistance. We think that the best balance is to have a set of measures relating to families receiving TANF assistance and a set of measures relating to the entire low income population, while at the same time committing federal attention to efforts to better understand and focus public attention on how states are addressing the circumstances of those families with the greatest barriers to employment.




Under the proposed regulations, all work-related measures are based solely on measuring outcomes among the population receiving TANF assistance. We recommend a more inclusive set of work measures that will measure how well states are serving the entire universe of working families, not just families on TANF. We suggest that the final regulation, in addition to measuring performance relating to the TANF population, also include at least one measure of performance based on work participation by the state's entire low-income working family population.

As discussed in the overview on pages 5-6, there are several reasons why the work measures should become at least partially population-based. Caseloads have declined significantly and more former and current welfare recipients are working than ever before. More former recipients are among the working poor. Some states have experienced very steep caseload declines and have relatively few families left on TANF. These remaining families have a disproportionate number of barriers to employment. A diminished caseload base may handicap a state against winning a performance bonus based solely on TANF population.

At the same time, states that are serving relatively few of their poor families with TANF may be providing other work-based supports to low-income families not receiving assistance. For example, a state may have used its TANF funds to substantially expand the availability of child care or to develop an earned income tax credit or to develop other supports for the low-income working population.

States are serving a varying percentage of their total number of poor families and children with TANF. Because of this variation, a performance bonus measure that looks only at current TANF population will inherently miss some of the work a state has done to support working families. A performance measure that focuses solely on the TANF population cannot evaluate how well states have done in helping low-income families avoid returning to the rolls. For these reasons, we recommend that HHS consider making more of the work-related measures based on the total population of low-income families.


Specific Recommendations

Specifically, we propose that $25 million of the work-related bonus be allocated to a population-based measure that calculates an earnings gap. Bonuses would be awarded to the eight states with the most significant improvement in earnings gap from the previous fiscal year. Specifically, we propose:

The advantage of an earnings gap rate is that it is population-based and tells us how much more all poor families need to be earning before they can get out of poverty. It is an indirect measurement of how all low-income working families are succeeding in the workforce, at job entry, retention and earnings gain. In this measure, a state that increases the overall employment rate among low-income families will succeed in reducing its earnings gap, as will a state in which low-income parents work more weeks of the year, as will a state in which low-earning families attain increased wages.

We understand that the Department will be using data from the American Community Survey when those data becomes available, as well as data from Census 2000 supplementary survey and other data to gauge state performance on the food stamp outcome measure and the family formation measure. At full implementation in 2003, the ACS will have a sample size of three million households, far larger than any other Census survey. It will be able to provide statistically valid demographic data at the state level on an annual basis, something that has only been available in the past only once every ten years from the Decennial Census. Because of the advantages of the American Community Survey data, we recommend that HHS use these data for the Improvement in the Earnings Gap measurement.

The use of Census data and a population-based measurement makes it possible to adjust the data for differences between years for state economic changes and population changes in the numbers of families with children. This adjustment is possible because the population is defined similarly across the states. TANF-based population measures would be considerably more difficult to adjust for the reasons cited in the overview on page 6. A change in the number of families with children from one year to the next will affect the earnings gap but we can easily adjust for that change.

Adjusting the earnings gap for state-wide changes in the unemployment rate between the previous year and the measured fiscal year is more complicated. We suggest using a common regression technique that could relate the change in the earnings gap per poor family to changes in the state's unemployment rate and perhaps other variables as well. This relationship would be estimated using the latest data available plus several historical years. The coefficients from this model could be used to adjust the earnings gap. This model would allow a more fair comparison between a state where the unemployment rate shrunk from 5.0 percent to 4.5 percent and a state where the unemployment rate grew from 5.0 to 5.5. percent.




We appreciate HHS' intent to "stimulate successful state initiatives" in pursuit of the TANF goals to "promote marriage" and "encourage the formation and maintenance of two-parent families." The proposed rule would rank and then reward the top ten states that demonstrate an increase in the percentage of children below 200 percent of poverty who reside in married couple families. However we believe there are serious methodological and substantive problems with using the proposed performance measure as a basis for rewarding states on this issue. We also believe that quantitative measures are inappropriate and premature in an area of policy that is so untried and untested.

As an alternative, we propose that HHS establish a monetary awards program that would reward states for innovative policy initiatives or program demonstrations that have demonstrated, or can be reasonably expected to lead to the desired outcomes, namely more children being raised in stable, healthy marriages and two-parent family households.

Data and Measurement Problems in the Proposed Regulations.

Since state-produced vital statistics on marriages and divorces are not universally available, uniform, or reliable, the rule proposes using Census household survey data for the proposed performance measure.

We understand that Census data will be available to construct this proposed measure for each state through population estimates based on data collected in the 2000 Census long-form (the survey administered to 1 in 6 households) and, in the subsequent years, data collected by the new annual American Community Survey (ACS) and its precursors, assuming the necessary fiscal appropriations are enacted to fund it.

These Census survey data have a number of shortcomings in relation to their proposed use for this measure and awarding the bonus. We first list those problems that apply mainly to restricting the measure to families below the 200 percent of poverty. This restriction could produce misleading results and create perverse incentives:

 If the measure were made universal, and state rankings were based on the percentage increase in all children living in married-couple households some of these problems would be avoided. However the following problems would remain:

In short, the adoption of quantitative performance measures of these goals is premature and will not stimulate state initiatives that are likely to be successful in the long run.

In pursuing the work goals, states have had more than a decade of lessons from demonstration work/welfare programs to draw upon, and reasonable quantitative measures of performance can be constructed. However, the public sector has no record of program or policy experiments related to encouraging marriage and strengthening two-parent families. Hence states have no guidance as to the best approaches to use to pursue these goals. This absence of program experience and evaluation is likely to result in two responses:


Specific Recommendations

We believe the best way to encourage states to develop successful innovations in this area would be for HHS to establish an Innovations in Strengthening Marriage and Two-Parent Families Award program. This program could be modeled on existing awards programs such as the Innovations in Government Award program administered by the John F. Kennedy School of Government at Harvard University, in collaboration with the Council for Excellence in Government and the Ford Foundation. (In this program $100,000 is awarded each year to the 10 applicants from local, state or the federal government for outstanding innovative examples of solving a problem of significant concern to a portion of the American people. $20,000 awards are given to an additional 15 programs).

In the proposed Innovation Awards program states (and perhaps local governments) would be invited to compete on a voluntary basis for substantial monetary awards given for demonstration programs, innovative policies or a package of initiatives designed to promote marriage and/or encourage the formation and maintenance of two-parent families. These efforts could be quite new, or instituted within the last five years. We recommend awarding a total of $20 million to a small number of states. The applications would be reviewed and rated by a panel of judges according to criteria such as:

In announcing the proposed competitive awards program, states would be encouraged to use unspent TANF funds to design and implement these new initiatives as well as seek matching funds from the private sector.

It would be important to select a panel of judges who were widely respected and independent of political partisanship. At a minimum the panel should include nationally known experts in couples and marital research, experts in evaluation of innovative programs and individuals with experience in public sector human service programs.

End notes:

1. The goals of TANF are to:  (1) provide assistance to needy families so that the children may be cared for in their homes or in the homes of relatives; (2) end the dependency of needy parents on government benefits by promoting job preparation, work, and marriage; (3) prevent and reduce the incidence of out-of-wedlock pregnancies and establish annual numerical goals for preventing and reducing the incidence of these pregnancies; and (4) encourage the formation and maintenance of two-parent families.

2. Sheila R. Zedlewski and Sarah Brauner, Are the Steep Declines in Food Stamp Participation Linked to Falling Welfare Caseloads?, November 1999.

3. U.S. Department of Agriculture, Food and Nutrition Service, Office of Analysis, Nutrition, and Evaluation, Who is Leaving the Food Stamp Program? An analysis of Caseload changes from 1994 to 1998, August 1999.

4. Prepared by researchers at the Urban Institute and released in the January/February issue of Health Affairs.

5. For a review of initial findings from state leavers studies concerning declines in Medicaid receipt after leaving welfare, see Greenberg, Participation in Welfare and Medicaid Enrollment (Kaiser Family Foundation, 1998).

6.The proposed regulations indicate that states will be rewarded based on the "percentage" increase in their health care participation rates. We have assumed that this means "percent" increase, rather than the percentage point increase in participation rates.

7. We specifically recommend evaluating states for the improvement measure based on which have the largest percentage point increase in participation rates, rather than percent increase. If HHS were to use the percent increase in the CHIP/Medicaid participation rate as the measure, it would make it relatively easy for states with a history of poor performance to secure bonus funds. For example, under an improvement measure based on the percent increase, a state that increases its participation rate from 20 percent to 24 percent (a 20 percent increase) would be more likely to win bonus funds than a state that increases it participation rate from 60 percent to 65 percent (an 8.3 percent increase).

8. In particular, the following language in the preamble should be corrected if necessary to clarify that a state's participation in the health care measure does not constitute an evaluation by HCFA that it has complied with the requirements referenced in the qualifying conditions: On page 68210, the regulation indicates "HCFA will verify States' compliance through State documentation and the agency's ongoing oversight of the Medicaid/CHIP programs;" on page 68211, the regulation states "HCFA will verify compliance through the agency's ongoing review of the Medicaid/CHIP program;" and on page 68221, it states that "The qualifying conditions will be evaluated by HCFA based on State documentation and HCFA oversight of the Medicaid/CHIP programs."

9. Schumacher and Greenberg, Child Care After Leaving Welfare (Center for Law and Social Policy, October 1999).

10. U.S. Department of Health and Human Services, Access to Child Care for Low-Income Working Families (October 19, 1999).

11. See Friedlander and Burtless, Five Years After (Russell Sage Foundation, New York, 1995); Cancian and Meyer, Work After Welfare: Women's Work Effort, Occupation, and Economic Well-Being, Social Work Research, Forthcoming, 2000.

12. See Rangarajan, Schoet, and Chu, Employment Experiences of Welfare Recipients Who Find Jobs: Is Targeting Possible? (Mathematica Policy Research, Inc., Princeton, N.J., 1998); Cancian and Meyer, Work After Welfare: Work Effort, Occupation, and Economic Well-Being, Draft paper prepared for the Annual Meeting of the Association For Public Policy Analysis and Management (1997).