Part 1 |
Save page Remove page | Previous | 1 of 2 | Next |
|
|
small (250x250 max)
medium (500x500 max)
Large
Extra Large
Full Size
Full Resolution
|
This page
All
|
*k COMPLETED ti 9t>3: C fs/f ") j£5&s. United States H&j^flft Department of i|i Agriculture Food and Consumer Service Office of Analysis and Evaluation Participation in the Child Support Enforcement Program Among Non-AFDC Food Stamp Households Final Report February 1995 9S-M3X5" United States Food and Department of Consumer Agriculture Service 3101 Park Center Drive Second Floor Alexandria. VA 22302 Participation in the Child Support Enforcement Program Among Non-AFDC Food Stamp Households Gregory B. Mills Elsie C. Pan Elaine Sorenson Sandra J. Clark Margaret Weant A product of: Abt Associates Inc. 55 Wheeler Street Cambridge, MA 02138 and The Urban Institute 2100 M Street N.W. Washington, DC 20037 February 1995 This study was conducted under contract number 53-3198-1-017 with the Food and Consumer Service. U.S. Department of Agriculture. Points of view or opinions stated In this report do not necessarily represent the official position of the Food and Consumer Service. ACKNOWLEDGMENTS The authors would like to thank the many people who made important contributions to this report. The following individuals deserve special recognition. Diana Perez, Carol Olander, and Steven Carlson of the Food and Consumer Service provided overall leadership, direction, and support throughout the study. Linda Mellgren, Pat Hagen, and Carl Montoya of the U.S. Department of Health and Human Services gave many useful suggestions on the research plan and the presentation of findings. Many staff at Abt Associates assisted with the data collection and analysis, including Diane Porcari, Susan Kannel, Jay Leatherman, Michael Walker, Alan Werner, Kymn Kochanek, Sara Craddock, Jane Belcore, Lynn MacKenzie, and Marva Lopez. Chris Hamilton and Kathleen Flanagan provided technical direction and management support. Wendy Campbell, editorial consultant, made numerous valuable comments as the report neared completion. Susan Byers and Eileen Fahey skillfully produced the draft and final versions of the report. The project staff at the Urban Institute played an essential role through careful analysis of Census data and use of the TRIM2 microsimulation model. Freya Sonenstein gave helpful early guidance on the evaluation design. Sharon Hilliard, Brenda Brown, and Sonja Drumgoole provided steadfast clerical support and production assistance. Finally, the study could not have been completed without the cooperation of clients and staff from the Food Stamp Program and the Child Support Enforcement Program in the participating states. « » TABLE OF CONTENTS EXECUTIVE SUMMARY i CHAPTER ONE INTRODUCTION 1 1.1 Research Questions 3 1.2 Data Sources 13 1.3 Organization of This Report 24 CHAPTER TWO NEEDS FOR CHILD SUPPORT ENFORCEMENT 25 2.1 Food Stamp-Only Custodial Households 28 2.2 CSE Nonparticipation and CSE Needs 36 CHAPTER THREE POTENTIAL FOR INCREASED CSE PARTICIPATION 45 3.1 Potential Response to a Mandate or Outreach 47 3.2 Factors Associated with CSE Participation 55 CHAPrER FOUR BENEFITS AND COSTS OF TWO POLICY OPTIONS 69 4.1 Method of Estimating Changes in Child Support Status 71 4.2 Effects on Household Incomes 87 4.3 Effects on Government Expenditures 93 CHAPTER FIVE POLICY IMPLICATIONS 101 5.1 Assessment of the Problem 101 5.2 Mandate Versus Outreach 103 Prepared by Abt Associates Inc. ' * * Table of Contents APPENDICES APPENDIX A ANALYSIS OF CURRENT POPULATION SURVEY APPENDIX B ANALYSIS OF SURVEY OF INCOME AND PROGRAM PARTICIPATION APPENDK C INTEGRATED QC REVIEW SCHEDULE APPENDK D RECIPIENT SURVEY INSTRUMENT APPENDK E RECIPIENT SURVEY RESPONSES APPENDK F CASE RECORD ABSTRACTION INSTRUMENT APPENDK G TRIM MICROSIMULATION TECHNICAL DETAIL Prepared by Abt Associates Inc. W LIST OF EXHIBITS CHAPTER ONE 1.1 Accounting for the Benefits and Costs of the Two Policy Options 11 1.2 Recipient Survey Summary: Cases Sampled, Screened, and Interviewed . 21 CHAPTER TWO 2.1 Identification of the Target Population 26 2.2 CPS- and SIPP-Based Estimates of Food Stamp-Only Custodial Families . 28 2.3 Distribution of Food Stamp Housenolds by AFDC/Medicaid Receipt, Custodial Status, and State, Fiscal Year 1991 29 2.4 Food Stamp-Only Custodial Households as a Percentage of All Food Stamp Households in the Five Participating States 34 2.5 Percentage of Food Stamp-Only Custodial Parents and Households Reporting No Previous CSE Contact or No Current CSE Case, by Data Source 37 2.6 Reported CSE Contact Versus Presence of CSE Case 39 2.7 Initial Child Support Status of Food Stamp-Only Custodial Parents Not Participating in the CSE Program 41 2.8 CPS and SIPP Estimates of the Target Population 43 CHAPTER THREE 3. i Survey Respondents' Reasons for Not Participating in the CSE Program . . 48 3.2 Survey Respondents' Predictions of Their Responses to a CSE Mandate in the Food Stamp Program 50 3.3 Identification of Custodial Parents Who Potentially Would Respond to Improved Outreach 54 3.4 Characteristics of Custodial Parents: All Survey Respondents 56 3.5 Characteristics of Custodial Parents: Never on AFDC and Ever on AFD2 62 3.6 Regression Estimates 64 CHAPTER FOUR 4.1 Distribution of Food Stamp-Only Custodial Parents 75 4.2 Upper-Bound Effect of a Mandate on the Distribution of Food Stamp-Only Custodial Parents by Child Support Status 76 4.3 Characteristics of Custodial Parents: Participating in CSE, Responding to Mandate, and Responding to Outreach 80 4.4 Lower-Bound Effect of Mandate on the Distribution of Food Stamp-Only Custodial Parents by Child Support Status . . 82 Prepared by Abt Associates Inc. . / List of Exhibits 4.5 Upper-Bound Effect of Outreach or. the Distribution of Food Stamp-Only Custodial Parents by Child Support Status 84 4.6 Lower-Bound Effect of Outreach on Distribution of Food Stamp-Only Custodial Parents by Child Support Status 85 4.7 Annual Benefits and Costs of Policy Options 90 4.8 Decomposition of Benefits and Costs 98 Prepared by Abt Associates Inc. VI EXECUTIVE SUMMARY The Food and Nutrition Service (FNS) of the d.S. Department of Agriculture has undertaken this exploratory study to assess the needs for child support among food stamp households and evaluate two policy options for meeting those needs: • a CSE mandate—requiring that custodial parents participate in the Child Support Enforcement (CSE) Program as a condition of their Food Stamp Program eligibility; or • improved CSE outreach—requiring that state food stamp agencies provide CSE information to custodial parents and facilitate their application to CSE at the time of initial food stamp certification. Both options would aim to raise household incomes through increased child support payments, offset partially by reduced food stamp allotments. This study focuses on food stamp-only custodial households—that is, households with children of noncustodial parents that receive food stamps but neither Aid to Families with Dependent Children (AFDC) nor Medicaid benefits. These latter programs already require CSE participation as a condition of program eligibility for custodial parents. We address the following three research questions (in Chapters Two, Tr -e, and Four, respectively): • What are the needs for child support enforcement among food stamp-only custodial households not participating in the CSE Program?1 • What is the potential for increased CSE participation among those with child support needs, through either a mandate or improved outreach? • What are the benefits and costs ofthe two policy options, from the perspective of both program clients and taxpayers? To answer the first question, we employed a variety cf existing data sources, including the March 1990 Current Population Survey (CPS) and the corresponding April 1990 Child Support Supplement, the 1990 Survey of Income and Program Participation (SIPP), food stamp 1. The needs of those already participating in CSE represent a separate policy issue that is not addressed here. Prepared by Abt Associates Inc. i„ •v/l Executive Summary quality control (QC) data for Fiscal Year .991, and state administrative data for the - 1992 food stamp and CSE caseloads. The state administrative data were compiled from five states that were selected to participate in this study. These states—Alabama, Florida, New Jersey, Oklahoma, and Texas—were selected in part because they had sufficiently large non-AFDC food stamp caseloads and were able to provide the necessary automated program data. Each state provided administrative data files enabling us to analyze statewide patterns of CSE eligibility and participation among food stamp households. We also collected and analyzed new data for this study, including a survey of 414 food stamp-only custodial parents, abstracts from CSE case records, and interviews with program staff and client advocates. In each of the five participating states two food stamp offices were chosen as sites for the local data collection. Site selection was based on criteria that included the size of the food stamp caseload (preferably at least 150 food stamp-only custodial households), the diversity of food stamp administrative practices (preferably encompassing a rant, 5 of current CSE outreach activities at the time of certifying applicants' eligibility for food stamps), the measured effectiveness of the CSE Program (preferably average or above-average), and the expected level of cooperation from local program staff (preferably high). The sites selected were: • Etowah County (Gadsden) and Montgomery County (Montgomery), Alabama; • Jacksonville (Southside Service Center) and Lakeland, Florida; • Camden County (Camden) and Hudson County (Jersey City), New Jersey; • Cleveland County (Norman) and Tulsa County (Tulsa), Oklahoma; and • Garland and Lubbock (Parkway office), Texas. We employed the data collected from the survey of food stamp recipients and the CSE case record abstracts to estimate the potential for increased CSE participation through a mandate or through outreach. The survey and abstract data, coupled with microsimulation findings from the Urban Institute's TRIM2 model, then provided the basis for projecting the benefits and costs of the two policy options. Prepared by Abt Associates Inc. ii > * v 11 Executive Summary Needs for Child Support Enforcement With respect to the first research question posed above, our findings are as follows. Of the total number offood stamp households nationwide, 2.8 percent are in the target population of this study: food stamp-only custodial households not participating in the CSE Program but with needs for child support enforcement. These households have at least one child of a noncustodial parent and either have no support order or receive less than the full amount awarded. Based on the fiscal year 1992 national monthly caseload of approximately 11 million households, the estimated 2.8 percent figure implies a target population nationally of 300,000 households. The national estimate of 2.8 percent is the product of two factors. The first is the percentage of all food stamp households nationally that were food stamp-only custodial households. We estimate this nationally at 7 percent. The second factor is the percentage of food stamp-only custodial households that were CSE nonparticipants with ^hild support needs, estimated at 40 percent. We estimate the aggregate needs for child support within the target population at $900 million annually, in 1992 dollars. This represents the additional amount of support payments that would be collected annually if all 300,000 households in the target population received the full amount of support ordered. This calculation assumes an average annual award per household of $3,000, consistent with the TRIM2 simulations. The $900 million total, which includes the shortfall in payments to those who already have support orders as well as the potential payments to those who currently lack support orders, provides clear evidence that some consideration of a policy change is warranted. The aggregate total also provides a benchmark against which to assess a proposed policy's estimated yield in additional support payments. Potential for Increased CSE Participation To address the second research question, we estimated the extent to which hous~nolds in the target population might enter the CSE Program in response to either a CSE mandate or improved outreach. Specifically, we divided the target population into the following three groups: • those unlikely to become CSE participants with either a mandate or outreach; Prepared by Abt Associates Inc. iii Executive Summary • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants with outreach (and also, therefore, with a mandate, since outreach would be much less compelling of participation than a mandate). We hypothesized that the larger the first group, the weaker the case for either a mandate or outreach. The larger the second group, the stronger the case for a mandate. The larger the third group, the stronger the case for outreach. Our basic findings, derived from the responses of custodial parents to questions on the survey of food stamp recipients and from information collected subsequently through CSE case record abstracts, are as follows: • An estimated 24 percent of custodial parents in the target population would respond to neither a mandate nor outreach. These custodial parents are subdivided as follows: 9 percent who would leave the Food Stamp Program altogether if faced with a mandate, 6 percent who would seek a good cause exemption from the mandate, S percent who would accept a smaller food stamp allotment as a sanction for noncooperation, and 4 percent who would "do something else" (possibly complying with a mandate but opting to withhold information about the noncustodial parent). • An estimated 39 to 60percent of custodial parents in the target population would respond to a mandate, but not to outreach. These are custodial parents who indicate that they would cooperate under a mandate and whose current nonparticipation appears to reflect a deliberate informed choice. They thus seem unlikely to respond to any outreach effort. • An estimated 16 to 37percent ofcustodial parents in the target population would respond to outreach (as well as to a mandate). Unlike the previous group, these custodial parents indicate little or no knowledge of the CSE Program, or they appear undecided about CSE participation. They thus might respond to information, referral, or application assistance. The custodial parents in the second group, those responding to a mandate but not outreach, thus represent at least a plurality—and perhaps a majority—of the target population. To examine further the likely patterns of CSE participation, we also estimated a series of regression equations. The variables included in these equations explain 41 percent of the variation in CSE participation among households that had never received AFDC (and thus never been subject to a CSE mandate) and 26 percent of the variation among households that had Prepared by Abt Associates Inc. iv Executive Summary previously received AFDC (and thus been subject to a mandate). When we controlled for client demographic characteristics, households that had never been on AFDC were significantly more likely to participate in the CSE Program in Florida—where food stamp caseworkers routinely provide CSE information and referral—than in the other four states studied. This evidence provides empirical support for the presumption that more active outreach efforts would indeed yield an increase in CSE participation. Benefits and Costs of the Two Policy Options With respect to the third research question, we estimated the following two sets of benefits and costs of a CSE mandate and improved outreach: • What is the net effect on annual household incomes, through increases in child support payments, associated reductions in food stamp allotments, and the food stamp benefits forgone by those who might accept a sanction or leave the program altogether rather than comply with a mandate? • What is the net effect on annual government expenditures, through reductions in food stamp allotments and changes in food stamp and CSE administrative costs? This analysis is based on our definition of the basic elements of either a mandate or outreach strategy, as detailed in Chapter One. The actual impact of either policy change would of course depend on the specific provisions of federal statute and regulation, as well as the manner of implementation carried out by state and local agencies. Our findings, based on microsimulations of national child support payments and food stamp allotments (using the Urban Institute's TRIM2 model) and an analysis of administrative costs in both the Food Stamp and CSE Programs, are: • A mandate strategy, when simulated under relatively optimistic ("upper-bound") assumptions about the effects of increased CSE participation on child support orders and payments is likely to result in a net increase in annual household incomes of $126 million and a net reduction in annual government expenditures of $60 million, expressed in 1992 dollars (relative to a baseline simulation of current policy). Under less optimistic ("lower-bound") assumptions, a mandate might raise household incomes by $9 million, while reducing government costs by $15 million. • An outreach strategy is likely to result in a net increase in annual household incomes of $15 million to $36 million, accompanied by a net increase in annual government expenditures of $9 million to $10 million. Prepared by Abt Associates Inc. v Executive Summary Under either the upper- or lower-bound assumptions, a mandate would thus make both program clients and taxpayers better off. However, the lower-bound effects of a mandate are quite small. Government expenditures arc estimated to drop somewhat, making taxpayers better off, but clients might experience little overall income gain because of the food stamp benefits forgone by those choosing not to comply with the mandate. In contrast, the outreach estimates are mixed. Household incomes would likely rise slightly, making clients better off, but with higher government expenditures. Policy Implications The benefit-cost estimates presented above also allow us to estimate the degree to which either policy option might succeed in closing the $900 million gap between current support payments received by the target population and the estimated potential for payments. A mandate is likely to eliminate between 7 percent and 24 percent of the gap in aggregate support payments, whereas outreach would likely close between 2 percent and 5 percent of the gap, based on the lower- and upper-bound estimates, respectively. The gains achievable through either a mandate or outreach are thus modest relative to the size of the underlying problem. Nevertheless, with policy issues as large and as intractable as this one, one should not dismiss progress of any magnitude. The question then becomes whether a change in policy produces enough gains to be worthwhile. A fundamental distinction between the mandate and outreach options is the combination of yield and risk that each option presents. A mandate offers the prospect of large gains to both clients and taxpayers. However, there is a substantial possibility that a mandate could leave clients only slightly better off.2 In particular, under the lower-bound mandate estimates, the forgone food stamp benefits (among those sanctioned for noncompliance and those opting to leave the program rather than comply) nearly offset the income gains among those who do comply and come to receive additional child support payments. Moreover, the federal savings from a mandate would come largely (if not entirely, in the lower-bound estimates) from forgone 2. Because of data limitations, the lower-bound estimates for the mandate strategy were not empirically derived, but were based on relatively arbitrary estimates of the potential effectiveness of CSE participation for the target population. The risk cited here may therefore be much less, or even more, than we have predicted. Prepared by Abt Associates Inc. vi Executive Summary food stamp benefits. In contrast, outreach poses gains to clients that are at best quite limited, and generates some additional costs for taxpayers. One advantage offered by the outreach strategy is the flexibility of implementing the approach on a selective geographic basis. One possible approach to targeted outreach is an FNS initiative in its Southeast and Southwest regions, which account for more than 60 percent of all food stamp-only custodial households. As a possible first step toward a national outreach policy, a regional initiative would enable FNS to focus its own staff and resources more effectively on the relevant segment of the national caseload. Finally, in evaluating either a mandate or outreach, there are societal concerns that go beyond the scope of measurable benefits and costs. As a matter of social policy, it is important to reinforce the responsibilities of parents to provide for the well-being of their children. For this reason in particular, a mandate or outreach may deserve more attention than would be warranted by the short-term fiscal impacts. Prepared by Abt Associates Inc. vii t'lii CHAPTER ONE INTRODUCTION The Food and Nutrition Service (FNS) of the U.S. Department of Agriculture has undertaken this exploratory study to assess the needs for child support among households receiving food stamps and evaluate two alternative policy options for meeting those needs: a mandate to participate in the Child Support Enforcement (CSE) Program as a condition of food stamp eligibility, and improved outreach to encourage greater voluntary CSE participation. The purpose of both alternatives would be to raise ths incomes of food stamp recipients through increased child support payments, and also thereby reujce the need for food stamps. The study focuses on food stamp households that (a) include at least one child of a noncustodial (absent) parent and (b) receive neither Aid to Families with Dependent Children (AFDC) nor Medicaid benefits. Both the AFDC and Medicaid Programs already require participation in the CSE Program, whereas the Food Stamp Program does not. More specifically, a custodial parent's eligibility for AFDC or Medicaid benefits is conditional upon cooperation with the CSE Program, unless the parent qualifies for a "good cause exemp-tion"— which is allowed, for instance, when one could reasonably expect cooperation to result in physical or emotional harm to the child. For cases not granted a good cause exemption, the sanction for noncooperation is removal of the custodial parent (but not the rest of the household) from the AFDC or Medicaid assistance group.1 There is widespread and increasing recognition that poverty in the United States is importantly associated with families of unwed mothers, parental separation and divorce, and the failure of noncustodial parents to provide child support. Nearly all of the rise in the child 1. To meet the AFDC requirement to participate in the CSE Program, a custodial parent must assign child support rights to the state and must cooperate with the state CSE agency in establishing paternity and obtaining support payments. The custodial parent receives the first $50 in monthly collections for current support; the remainder is distributed between the state and federal governments according to their funding shares for AFDC benefit payments. For families receiving Medicaid but not AFDC—"Medicaid-only recipients "—the custodial parent must assign rights to the state for medical support and must cooperate with the state CSE agency in establishing paternity and in obtaining medical support from the noncustodial parent. When such a case enters the CSE system, the CSE caseworker will typically process it for both cash support and medical support. The cash support payments collected for the case go entirely to the custodial parent, as with any non-AFDC CSE case. Prepared by Abt Associates Inc. Chapter One: Introduction poverty rate during the 1970s and 1980s can be attributed to the declining proportion of children in two-parent families and the corresponding increase in families headed by never-married or divorced mothers.2 As of spring 1990, of the 10 million custodial mothers in the U.S. (living with children under 21 whose fathers are outside the home), only one-half had a court order under which they were to receive child support in 1989. Of these 5 million women, only one-half actually received the full dollar amount, one-quarter received partial payment, and the remaining quarter received no payment.3 The aggregate "child support deficit"—the difference between the total amount of support payments due and the total amount actually received by custodial parents—was $5.1 billion in 1989. Even for those receiving full payment, award amounts are often inadequate—that is, the award amount may not properly reflect the custodial household's needs or the noncustodial parent's income. The CSE Program, enacted in 1975 as Title IV-D of the Social Security Act, assists in locating noncustodial parents, establishing paternity, obtaining support orders, and enforcing support obligations. Services are provided automatically at no charge to those receiving AFDC or Medicaid, and are available to others upon request (in so..ie states, at a small charge to the custodial parent). Food stamp households that receive AFDC or Medicaid thus presently fall under the existing CSE mandate. Those food stamp custodial households that receive neither AFDC nor Medicaid— termed "food stamp-only custodial households"—participate in CSE on a voluntary basis, if they participate at all. Indeed, previous research indicates that only about one-third of food stamp-only custodial households receive support payments through the CSE Program.4 Recognizing the extent to which food stamp-only custodial households do not seek services thro h the CSE 2. Robert I. Lerman, "Policy Watch: Child Support Policies," Journal of Economic Perspectives, Vol. 7, No. 1, Winter 1993, p. 171. 3. U.S. Department of Commerce, Bureau of the Census, "Child Support and Alimony: 1989," Current Population Reports, Consumer Income, Series P-60, Number 173, September 1991, p. 1. 4. Unpublished tabulations by the U.S. Department of Health and Human Services, based on data from the 1988 Current Population Survey, showed that only 33 percent of non-AFDC food sump custodial households received support payments in 1987 through CSE collections. Another 27 percent received support payments outside the CSE system (some on a voluntary informal basis from the noncustodial parent). The remaining 40 percent received no support payments. Prepared by Abt Associates Inc. Chapter One: Introduction Program, FNS has undertaken this study to examine their needs for services and the benefits and costs of alternative approaches to increasing their CSE participation. A key question underlying the present study is whether the unmet needs for child support among food stamp-only custodial households merit any policy action. Given the availability of services through the CSE Program, policy action might either require CSE participation as a condition of food stamp eligibility or encourage greater voluntary use of CSE services through improved outreach. These two policy options raise further questions: • • To what extent would a CSE mandate prompt food stamp households to seek good cause exemptions, to accept a lower food stamp allotment as a sanction for noncooperation, or to leave the Food Stamp Program altogether, to avoid participating in child support enforcement? Might there be very little response to improved outreach efforts, if potential CSE clients currently choose not to participate for reasons other than a lack of program information, such as the uncertain identity of the father, fear of physical harm, or the low prospects for collecting any payment from someone who has little or no income? Given the likely characteristics of the households that would newly enter the CSE Program, what are the prospects for collecting support payments for them? No previous studies have addressed these issues in the specific context of food stamp-only custodial households. This absence of previous research and the hypothetical ("what if4) nature of the questions above mean that this study must be regarded as exploratory. In particular, given the limited data on which projections of national benefits and costs must be based, one must interpret such projections cautiously. 1.1 RESEARCH QUESTIONS As an exploratory study, this research seeks to provide useful insights into the needs for child support enforcement services among food stamp households and the extent to which the CSE Program might meet the needs of those not currently participating. The fundamental research question addressed in this study is whether two policy alternatives being considered by FNS arc likely to be cost-effective in increasing child support payments to food stamp-only custodial households. To answer this question, we have investigated the following three more specific questions: Prepared by Abt Associates Inc. 3 Chapter One: Introduction • What are the needs for child support enforcement among food stamp-only custodial households not participating in the CSE Program? • What is the potentialfor increased participation among those with needs for CSE services, through either a mandate or improved outreach? • What are the likely benefits and costs of the two policy options, a CSE mandate or improved outreach, for program clients as well as public agencies? The remainder of this section discusses each of these three questions in turn. Needs for Child Support Enforcement The first step in this study is to determine how many food stamp-only custodial households both need child support enforcement services and are not currently participating in the CSE Program. These households constitute our target population—the universe of food stamp-only custodial households that might potentially benefit from either a CSE mandate or improved outreach. We define a household as "needing child support enforcement services'1 if there is no child support order, or less than full payment on an existing order, for one or more children of a noncustodial parent. This definition is necessarily arbitrary and has some potential for errors of both inclusion and exclusion. One might argue that some households defined here as needing child support enforce-ment do not pose any potential for increased child support collections, and thus should not be included in the target population. In some situations, for example, there is no immediate basis on which to establish the paternity of the child, or the noncustodial parent has no foreseeable capacity to make support payments. Any effort by the CSE agency in these situations seems destined to yield no short-term result. However, one cannot unambiguously identify such households a priori; the potential for establishing paternity, obtaining an order, and collecting support payments in the long term is an empirical issue. The same situation arises with those who currently have no support order but either have a pending support arrangement or receive support payments informally; there is some potential for obtaining a support order and collecting additional support payments. We therefore include all such households in our target population and then proceed to estimate the likelihood of collecting payments for them. Conversely, one might argue that some households not meeting our definition of "needing CSE services" do pose a potential for increased collections. For example, there are Prepared by Abt Associates Inc. Chapter One: Introduction households not participating in the CSE Program that receive full payment on their support orders, but where the award amounts appear currently inadequate or where the award amounts are not subject to periodic review (as now required in CSE every three years for AFDC cases, effective October 13, 1993). For several pragmatic reasons we have excluded such households from our target population. First, there is no consensus about the adequacy of award amounts set under existing child support guidelines.5 Thus, even with all the necessary information, any attempt to quantify needs would entail judgments about adequate award levels. Second, all guidelines require information about the noncustodial parent's income. This information is unavailable through any existing data sources; nor was it feasible to collect such information in this study except to ask the custodial parent to estimate the noncustodial parent's income. Third, one would expect uneven implementation among states with the requirement for periodic review and adjustment of awards. Fourth, the review of a non-AFDC CSE case currently requires a request from the custodial parent, many of whom decline the opportunity for a review (as one would also expect among food stamp-only custodial households that enter CSE voluntarily in response to more active outreach). Finally, when reviews do take place, they may result in no change to the award amount or even in a downward adjustment to the award amount. For these various reasons, we have assumed no potential for increased collections among those who already receive full payment on their orders. Potential for Increased CSE Participation Once we identify the target population, the next task is to estimate the percentage of such households that would likely respond to actions that FNS might take to require or promote their participation in the CSE Program, and to identify the factors that might cause a custodial parent to decline to participate. In the course of visits to each of the states and localities involved in this study, we conducted interviews with Food Stamp and CSE Program staff and 5. The most common approach to setting award levels is the "income shares" guideline, under which children receive the same share of combined parental income as they would in an intact family. However, states differ in the formula used to compute the noncustodial parent's child support payment. In some states the payment does not depend on combined parental income, instead equaling a fixed percentage of the noncustodial parent's income. In other states the prescribed amount declines at higher levels of income. Differing formulas have differing implications for living standards, labor supply, and incentives for child-bearing and separation and divorce. Prepared by Abt Associates Inc. Chapter One: Introduction client advocates. The reasons for CSE nonparticipation cited in these conversations tended to cluster into the following five categories:6 • Desire to maintain a relationship with the noncustodial parent. The custodial parent wants to continue a situation in which the noncustodial parent has some contact with the children, makes occasional cash payments, or provides in-kind support (diapers, clothing, gifts); feels that the noncustodial parent is "doing as much as he can"; does not want to "drive him away"; does not want to hassle the noncustodial parent (with administrative hearings, paternity tests, court appear-ances, court-imposed requirements for job search as part of a support order, or embarrassing publicity), does not want to give up intermittent informal support for an even less certain situation in which a support order might not be enforced (or in which formal support payments might cause the loss of food stamp benefits), or (at the fraudulent extreme) wants to conceal the fact that the "absent" parent actually resides with the family. • Desire to avoid involvement with the noncustodial parent. The custodial parent has concerns for her safety (or the children's) or does not want the noncustodial parent to obtain visitation privileges in conjunction with a support order, or the caretaker (such as a grandmother) does not want to "go after" the noncustodial parent (such as a son-in-law). • Desire to avoid involvement with program agencies. The custodial parent perceives any involvement with the CSE Program or the judicial system as intrusive, hostile, demeaning, intimidating, stigmatic, or time-consuming. • Low prospects ofobtaining any support payments. The custodial parent perceives little hope of receiving support payments, because the noncustodial parent's identity or whereabouts are uncertain or because the noncustodial parent is unemployed, is incarcerated, or now has another family to support. • Lack of knowledge about the potential benefits and minimal costs of CSE participation. The custodial parent knows little or nothing about the CSE Program; does not understand the potential benefits of establishing paternity and obtaining a child support order (such as health insurance coverage and survivor benefits under Social Security); does not recognize that assistance in obtaining child support would be expensive to obtain through private attorneys or collection agents; or is unaware of the methods available to the CSE agency for enforcing a support order (such as the interception of state or federal tax refunds or unemployment compensation, income withholding, and property liens). 6. We presume in these situations that the custodial parent either has no current support order or receives less than full payment on an existing order. Prepared by Abt Associates Inc. Chapter One: Introduction Recognizing these various reasons for CSE nonparticipation, we consider two alternative strategies that FNS might adopt. One is a CSE mandate, requiring participation in CSE as a condition of the custodial parent's eligibility for food stamps (as it is for AFDC and Medicaid) for households with at least one child of a noncustodial parent. Unless a household receives a good cause exemption, noncooperation would disqualify the custodial parent from food stamp eligibility, thereby decreasing the household's food stamp allotment. (Depending on household size, this would reduce the monthly allotment by about $80, on average.) Under such a mandate, as under current food stamp policy generally, child support payments represent countable income to the household and normally reduce the monthly allotment by 30 cents per dollar of support payments. (A state may disregard the first $50 of support payments in computing countable income if the state agrees to fund the corresponding increase in monthly allotment. However, no states have adopted this program option.) As mandatory participants, food stamp-only custodial households would not pay any CSE application fees or charges for CSE services. The other strategy being considered is to have state food stamp agencies undertake more active CSE outreach efforts. For the purposes of this study, we define improved outreach as food stamp caseworkers providing information on CSE services, referring households to the CSE agency, and assisting households with their CSE application—all during the course of the household's initial application to the Food Stamp Program. Under current CSE policy, the states could charge these households fees of up to $25 for their CSE application and services. It is especially in situations of the last type listed above—where the client lacks knowledge of CSE—that outreach is most likely to influence the custodial parent's decision to participate. Even in the other situations listed, however, there remains the potential that improved information, referral, or counseling would have some effect, depending on the intensity of the outreach efforts. Consider, for instance, the custodial mother who has little hope of collecting payments from a currently unemployed father. A caseworker might explain the value of obtaining a support order so that payments can be collected when the father regains employment. In assessing the relative merits of a mandate versus outreach, it is useful to categorize the target population into the following three categories: Prepared by Abt Associates Inc. ■■■ Chapter One: Introduction • those unlikely to become CSE participants even under a mandate (and thus presumably not with outreach either); • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants under outreach (and presumably with the more compelling mandate, as well). We hypothesize that the larger the first group, the weaker the case for any change to a mandate or an outreach policy. Members of the first group would include those likely to seek a good cause exemption to avoid CSE participation under a mandate, those likely to accept a reduced food stamp allotment as a sanction for noncompliance under a mandate, and those likely to leave the Food Stamp Program and thus forgo their household's entire monthly allotment rather than participate in CSE.7 The choice between a mandate or outreach revolves importantly around the relative sizes of the second and third groups identified above. The larger the second group, the more attractive is a mandate. The larger the third group, the more attractive is outreach. The important empirical issue in this portion of the study is thus the distribution of the target population among these three categories. Benefits and Costs of the Two Policy Options This study aims ultimately to provide FNS with information on the merits of either a CSE mandate requirement or improved CSE outreach. In evaluating these options, we had to define them more specifically to establish a framework for attributing benefits and costs. The actual benefits and costs of any policy change made will of course depend on how federal statutes and regulations are written and implemented. For the purposes of the benefit-cost analysis, we defined a CSE mandate as follows. At the initial food stamp certification, the caseworker would identify food stamp-only custodial households not currently having a child support order or not receiving full payment under an 7. Why might a custodial parent, facing a mandate to participate in CSE, choose to withdraw entirely from food stamps even though the case could remain on the program with a lowered benefit? The household's reduced monthly benefit might be so low that the custodial parent would consider it no longer "worth the hassle" of dealing with the welfare agency and using coupons to make food purchases. Studies of food stamp participation have shown that many eligible households do not enter the program because of the low benefit amount that they would receive. Prepared by Abt Associates Inc. Chapter One: Introduction existing order and not participating in CSE. The caseworker would explain that, to receive food stamps, the custodial parent must register with the CSE agency and cooperate with efforts to obtain child support payments, unless good cause exists for an exemption. As with AFDC or Medicaid, the CSE food stamp eligibility worker would determine good cause based on documentation provided by the custodial parent. Following the current AFDC and Medicaid policy, circumstances considered to warrant a good cause exemption might include the following: • if cooperation with CSE can be reasonably expected to result in physical or emo-tional harm to the child or caretaker relative; • if the child was conceived through incest or rape; or • if legal proceedings are already under way for a third party to adopt the child. Those who do not apply for an exemption or whose application for exemption is denied would need to register with the CSE agency (as currently for AFDC cases). If the custodial parent failed to do so, the food stamp caseworker (perhaps at the first recertification) would then sanction the case, lowering the monthly food stamp allotment by disqualifying the custodial parent or caretaker relative but not the rest of the household from food stamp eligibility. Other food stamp clients might decide to forgo all food stamp benefits, removing the entire household from the program to avoid participating in CSE. Those who complied with the mandate would then become CSE cases and could be treated by CSE in the same manner as AFDC or Medicaid cases. For example, agencies could not charge clients any fees for application or services. Agency efiorts would proceed as necessary (and as feasible based on the limited information provided by some custodial parents) to locate the noncustodial parent, establish paternity, obtain a child support order, and CO lect support payments. Cases would be subject to the same requirements for periodic review and modification as those now applied to AFDC and Medicaid cases.8 8. The priority assigned to CSE services for food stamp-only cases might depend on whether, for the purpose of computing incentive payments to states under Title IV-D, the collections from such cases are treated as AFDC collections or non-AFDC collections. This issue is not addressed in any proposed legislation. Prepared by Abt Associates Inc. Chapter One: Introduction If support payments are collected for the case, all payments would go to the custodial parent.9 Because those payments would count as food stamp income, the monthly food stamp allotment would decline by about 30 cents for each dollar of collections. (This is unlike AFDC, where only the first $50 per month of support payments go directly to the custodial parent.) There are some households for which the 30 percent offset would not apply, such as those that have not fully used their allowable deductions in computing countable food stamp income or those rendered ineligible for food stamps by the size of the increase in child support payments. We define improved CSE outreach in the following way for the benefit-cost analysis. At the initial certification of a food stamp-only custodial household, the food stamp caseworker would determine whether the household had any children with a noncustodial parent and no support order or only partial payment on an existing order. If so, the caseworker would explain the services available through the CSE Program and the potential benefits to the household of establishing a support order and obtaining collections. The caseworker would refer the household to the CSE Program and perhaps assist the client in completing an application for CSE services. The aim of this strategy would be to provide more complete information to food stamp clients who might need services, so that they could make a more informed judgment about whether to participate in CSE, and then to facilitate the clients' CSE applications. The benefits and costs of outreach would be similar in nature to those of a mandate, but outreach would likely entail both lower benefits and lower costs than a mandate. Because the custodial parents responding to outreach constitute a subset of those entering CSE under a mandate, the effects on support payrr-nts and CSE administrative costs are lessened. Moreover, by definition, under outreach no household would have to forgo food stamp benefits to avoid CSE participation, since participation would be voluntary. In the benefit-cost analysis in Chapter Four, we will first account for the net impact on incomes to food stamp-only households. As shown in Exhibit 1.1, there are two main components of this calculation: the estimated increase in child support payments and the associated decrease in food stamp allotments. We next account for the net impact on 9. We will follow here the conventional benefit-cost practice of counting child support payments as benefits to the custodial parent, while not counting such legally-obligated transfers as costs to the noncustodial parent. Note also that, for those noncustodial parents who are also food sump recipients, child support payments are treated as deductions from food stamp countable income. Higher support payments thus would raise food stamp benefits for these noncustodial parents. We do not take any account of this effect. Prepared by Abt Associates Inc. 10 Chapter One: Introduction government expenditures for the Food Stamp and CSE Programs. The exhibit indicates the three main components of this calculation: the estimated decrease in food stamp allotments, the estimated change in food stamp administrative costs, and the estimated increase in CSE administrative costs. Exhibit 1.1 ACCOUNTING FOR THE BENEFITS AND COSTS OF THE TWO POLICY OPTIONS Percentage Distribution 1 Effect (Expected Sign) Total Federal State Net change in household incomes Change in child support payments (+) A — — Change in food stamp allotments" (-) B — — Total (+) A + B — — Net change in government expenditures Change in food stamp allotments (-) B 100% 0% Change in food stamp administrative costs (+ or -) C 50% 50% Change in CSE administrative costs (+) D 66% 34% Total (+ or -) B + C + D — - 1 Approximately 30 percent of the increase in child support pay menu. Several aspects of this benefit-cost framework are noteworthy. First, the decrease in food stamp allotments enters on both the household side (unfavorably, as a reduction to the incomes of food stamp households) and the government side (favorably, as a reduction in government expenditures). Second, we do not attempt to incorporate a variety of potential nonmonetary benefits that might accrue to the custodial household from establishing paternity and obtaining a support order: • Paternity establishment enables children bom to an unmarried couple to gain most of the rights and privileges of children born to a married couple. These include, Prepared by Abt Associates Inc. 11 Chapter One: Introduction in the event of the death of a noncustodial parent, inheritance rights and rights to possible benefits from life insurance, Social Security (survivors' benefits), and veterans programs. • Health insurance coverage for the children and the custodial parent must now be included as part of a CSE-obtained support agreement, when such coverage is available to the noncustodial parent at a reasonable cost. • When paternity and a legal support order are established, children may have better prospects for a meaningful relationship with their father. Third, we also do not account for the out-of-pocket fees and charges that food stamp households may incur as voluntary CSE participants. For a mandate, the exclusion of out-of-pocket fees and charge: serves to understate both the net benefits to clients and net government expenditures, as the CSE Program could no longer collect fees and charges from those who might have voluntarily participated otherwise. For outreach, the exclusion of out-of-pocket fees and charges serves to overstate both the net benefits to clients and net government expenditures. Fourth, a mandate or outreach may not result in higher food stamp administrative costs. In principle, the increased certification costs to administer a mandate or conduct outreach might be offset entirely by savings associated with households made ineligible through increased support payments and households choosing to forgo food stamps altogether under a mandate. Fifth, the benefit-cost analysis will account for the distribution of government expenditures between the federal and state levels, reflecting the cost-sharing provisions for different categories of program expenditures: 100 percent federal funding of food stamp allotments, SO percent federal funding of food stamp administrative costs, and 66 percent federal funding of CSE administrative costs, as shown in Exhibit 1.1. The fiscal implications are thus quite different from the standpoint of states versus the total (federal and state) government sector. The net savings in food stamp allotments would accrue entirely to the federal government. States would bear SO percent of the change in food stamp administrative costs and 34 percent of the change in CSE administrative costs. To summarize, whether for the mandate approach or the outreach approach, the benefit-cost analysis will account for net additional income to food stamp-only custodial households and net government expenditures to the Food Stamp and CSE Programs. Either option would certainly seem desirable if it promised a net increase in household incomes and a net reduction Prepared by Abt Associates Inc. 12 Chapter One: Introduction in government expenditures. We may find, however, that the net increase in household incomes is attainable only with a net increase in government expenditures. Under that scenario, the question becomes whether the gains to food stamp recipients would outweigh the associated losses to taxpayers. The answer will depend on value judgments and considerations beyond any strict benefit-cost accounting. 1.2 DATA SOURCES This section describes the data sources that we have used for this study. We first describe the sources of existing data used to analyze the needs for child support enforcement among food stamp households: the Current Population Survey (CPS) and Child Support Supplement (CSS), the Survey of Income and Program Participation (SIPP), the food stamp quality control (QC) data, and state administrative data from the Food Stamp and CSE Programs. We then describe the data collection efforts undertaken for this study: recipient survey, case record abstracts, and program staff and client advocate interviews. We used the data from these latter sources to estimate the potential for increased CSE participation and increased child support collections under either a mandate or improved outreach. Current Population Survey One source of our national estimates on the needs for CSE services among food stamp recipients is the household survey data from the March 1990 CPS and corresponding data on child support arrangements from the CSS administered in conjunction with the April 1990 CPS (see the description in Appendix A). For those women interviewed in the March 1990 CPS who were demographically eligible to receive child support (by living with one or more "own children" under age 21 of a noncustodial father), about 70 percent were also administered the April 1990 CSS. For this representative national sample, the merged March/April 1990 CPS-CSS file allows one to link information on child support status with detailed information on demographic characteristics, employment, income, and participation in income support programs including food stamps, AFDC, and Medicaid. The information on income, including the receipt of child support payments, pertains to calendar year 1989. The CPS findings reported in this study are based on a sample of 4,064 custodial mothers. Among these, 209 represented food stamp-only families (receiving food stamps but neither AFDC nor Medicaid), another 209 were Prepared by Abt Associates Inc. 13 Chapter One: Introduction AFDC/MA only families (receiving AFDC and/or Medicaid, but not food stamps), 909 were AFDC/MA food stamp families (receiving AFDC and/or Medicaid, plus food stamps), and 2,737 were "no aid" families (receiving neither AFDC nor Medicaid nor food stamps). The primary advantage of the CPS is its large national sample. However, the CPS-CSS merged file has the following limitations: • Income reporting on an annual basis (versus monthly or quarterly), with acknowledged under-reporting of child support payments and income from benefit programs such as AFDC and food stamps; • Exclusion from the April CSS of custodial fathers and ever-married mothers who had children either outside of marriage with other previous partners or in marriages prior to the most recent divorce or separation; • Lack of information on the noncustodial parent; and • For some variables (including key child support characteristics such as the presence of a support order), a substantial percentage (30 percent or higher) of imputed values based on other respondents, because of the limited overlap between the March CPS and the April CSS. Survey of Income and Program Participation A second source of national estimates is the 1990 SIPP. As described in Appendix B, SIPP is a nationally representative longitudinal survey of adults that provides detailed individual and household information on income, wealth, and pro,"ram participation. Persons chosen for the survey enter a panel that is interviewed every four months for approximately two and one-half years (i.e., in eight rounds or "waves" of interviewing). The 1990 SIPP panel included about 20,000 households. Each wave of interviewing collects information from the initial sample of adults and all other adults residing with the initial sample members at the time of the interview. Information about income, labor force participation, and program participation is collected on the individual and the individual's household for the four months preceding the interview. SIPP interviews typically include two components: a core questionnaire and one or more "topical modules," sets of supplemental questions on special topics. Information on child support arrangements is included in both the core questionnaire (administered at Waves 1 through 8) and a child support topical module (normally administered at Waves 3 and 8). The Prepared by Abt Associates Inc. 14 Chapter One: Introduction core questionnaire captures information on the amount of child support payments received in each month. The child support topical module captures information from the custodial parent (mother or father) on child support agreements and awards, custody and visitation arrangements, the location of the noncustodial parent, and the use of the public child support enforcement agency. Custodial parents interviewed in the child support topical module from Wave 3 of the 1990 SIPP panel were the ba*is for the SIPP findings in this study. The sample analyzed here includes 3,069 custodial parents, classified as follows (based on benefit receipt in month 12 of the survey year): 113 were food stamp-only families, 152 were AFDC/MA only families, 669 were AFDC/MA fooc stamp lamihes, and 2,135 were "no aid" families. Our sample included individuals living with one or more own children under age 21 whose other parent lived elsewhere. Additional information on these individuals was obtained on previous program participation (from the Wave 2 topical module on recipient history and from the Census Bureau's longitudinally edited core file) and on family characteristics (from the cross-sectionally edited core file). Although the CPS-CSS file has a larger sample size, SIPP offers more extensive information on child support arrangements and use of child support services. SIPP also achieves more accurate reporting of income from food stamps and other assistance programs than CPS.10 Also important here is that SIPP provides retrospective information on prior receipt of AFDC and Medicaid. Any indication of prior AFDC receipt by a food stamp-only custodial parent would suggest earlier participation in (or at least some foreknowledge of) the CSE Program. Such individuals, if currently not participating in CSE, would presumably be less Ukely to respond to outreach efforts than those with no prior AFDC receipt. 10. In neither the CPS nor SIPP analysis is there any attempt to correct for income under-reporting. The SIPP findings thus reflect more reliable income data than the CPS findings. However, in the TRIM2 microsimula-tions discussed later, correction factors are applied to the CPS source data to align baseline estimates of recipients and payment expenditures with the corresponding benchmark totals available through published national program data. Prepared by Abt Associates Inc. 15 Chapter One: Introduction Food Stamp Quality Control Data Also included in this report are national and state-by-state estimates from the Fiscal Year 1991 food stamp QC data, as collected by states under the National Integrated Quality Control System. This cross-sectional data file contains household-level and person-level information on 64,311 active food stamp cases for which QC reviews were completed for the sample months October 1990 through September 1991. The sample sizes range among states from 300 to more than 2,400, varying in relation to state food stamp caseloads. Information from the case record and the QC review findings are recorded by the QC reviewer on a four-page Integrated Review Schedule, as showr in Appendix C. This data source allows for state-by-state estimates of the target population for this study—food stamp-only custodial households. Because the AFDC, Medicaid, and food stamp benefit information comes directly from the case record, such information is not subject to the underreporting that occurs either in CPS or (to a lesser degree) SIPP. Its usefulness is limited, however, by the lack of person-level information that would unambiguously identify children of a noncustodial parent (requiring the use of approximating criteria) and the absence of information on either the receipt of child support payments or participation in the CSE Program. For instance, the QC-based identification of food stamp-only custodial households could either overstate or understate this subset of the food stamp caseload. The uncertain direction of this bias will depend on the relative numbers of false-positive and false-negative misclassifications that result from the lack of person-level relationship codes. State Administrative Data We have conducted an analysis of child support arrangements among food stamp households using administrative data provided by each of the five states that participated in this study: Alabama, Florida, New Jersey, Oklahoma, and Texas. The selection of these five states, completed in April 1992, was based on the following considerations: • The state should have a sufficient capability to provide automated program data, for both food stamps and CSE, for our use in conducting statewide descriptive analysis and in selecting a survey sample in two local food stamp offices. Prepared by Abt Associates Inc. 16 Chapter One: Introduction • The state's monthly food stamp caseload should include at least 5,000 non-AFDC cases with children, to enable selection of a sufficient sample of CSE-eligible food stamp-only custodial households. • Preferably, the state should administer its CSE Program under the same human services or social services umbrella agency as food stamps, should be above-average in measures of CSE performance (such as child support collections per dollar of CSE administrative cost), and should offer a high expected level of cooperation from state and local staff. We also sought geographic variation, but expected that the states would come predominantly from the Southeast and Southwest regions because of the lower-than-average AFDC benefit levels and the resulting larger numbers of non-AFDC food stamp cases (that is, cases eligible for food stamps but with too much income to qualify for AFDC) in those regions. As we report later, QC data indicate that the thirteen states in FNS' Southeast and Southwest regions together account for more than 60 percent of national food stamp-only custodial households. The five states participating in this study alone account for about 35 percent of the national total. Each participating state provided automated data for all food stamp cases and child support enforcement cases active in the month of July 1992. These are extremely large data files, as the monthly food stamp caseloads at that time were approximately 900,000 in Texas, 300,000 in Florida,11 200,000 in both Alabama and New Jersey, and 140,000 in Oklahoma. From these files, we examined the distribution of food stamp households by the following characteristics: • Whether food stamp-only (versus AFDC/MA food stamps); • • Whether CSE-eligible (i.e., having in the food stamp household a child of an absent parent); If CSE-eligible, whether participating in CSE; • If participating in CSE, whether a prior AFDC recipient; • If participating in CSE, whether a support order exists; and 11. The number of households io the Florida analysis file is considerably less than indicated in other program statistics for the state. The data set initially provided by the state contained records for about 600,000 total household records, consistent with other reported caseload counts. However, we found a very large number of duplicate cases in the initial file. We feel that the analysis file represents an accurate unduplicated accounting of the food stamp recipients included in the initial data set. Prepared by Abt Associates Inc. 17 Chapter One: Introduction • If support order exists, amount of the award and whether receiving payments. For the first three items, we computed the distribution of food stamp allotments as well as households. We also examined the demographic characteristics of household heads in CSE-eligible food stamp-only custodial households. The structure and contents of these data files differ greatly from state to state. Some states maintain their data at the household level; other states maintain data at the individual recipient level (requiring aggregation by case number to construct information by household). The task of identifying CSE participants was made easier in Florida by a data element on the food stamp record that indicated a corresponding CSE case. Conversely, in Alabama and New Jersey there was a data element on the CSE file that indicated a corresponding food stamp case. In these two states, because such data elements are not always updated, we also classified a food stamp household as a CSE participant if the Social Security number for any member of the food stamp household matched with the Social Security number of a child or custodial parent in a CSE case. In Oklahoma and Texas, where no data element existed on either the food stamp or CSE files to indicate participation in the other program, we conducted a match between the two files using Social Security numbers (as indicated above for Alabama and New Jersey) as the sole basis for identifying CSE participants among food stamp households. Recipient Survey In each of the five participating states, two food stamp offices were chosen as sites for local data collection. The sites were as follows: • Etowah County (Gadsden) and Montgomery County (Montgomery), Alabama; • Jacksonville (Southside Service Center) and Lakeland, Florida; • Camden County (Camden) and Hudson County (Jersey City), New Jersey; • Cleveland County (Norman) and Tulsa County (Tulsa), Oklahoma; and • Garland and Lubbock (Parkway office), Texas. The selection of two local sites in each state, completed in July 1992, was based on the following criteria: Prepared by Abt Associates Inc. 18 Chapter One: Introduction • The monthly food stamp caseload should include at least 150 non-AFDC food stamp cases with children. • The CSE office serving the local area should be generally recognized as having average or above-average performance, as perhaps indicated by measures such as child support collections per dollar of administrative cost. • The locality should be one where we could expect cooperation from program directors, supervisors, and caseworkers (for both food stamps and CSE) and where orderly CSE record-keeping would facilitate abstraction of casefile information. • If possible, the local sites should employ different approaches to informing food stamp clients about the availability of CSE services, in the interest of encompassing a variety of administrative practices across the ten sites. All criteria were met, although the fourth proved difficult, as most local food stamp offices have no established method for providing CSE information and referral through the food stamp certification process. Among the five participating states, additional state-specific considerations served to limit the range of feasible choices for local sites. In Alabama, for instance, we excluded the six counties currently participating in the demonstration project entitled Avenues to Self-Sufficiency through Employment and Training Services (ASSETS). In Florida, we excluded any county with less than 80 percent conversion of its CSE cases to the state's new automated system. Within each state, we sought geographic separation in the sites, in the interest of obtaining greater variation in client socioeconomic characteristics and agency administrative practices. We also encouraged states to avoid local areas with exceptional characteristics that might limit the generalizability of findings, such as a high proportion of cases residing on Indian reservations or military bases. We also indicated that states should nominate food stamp offices whose caseload is served by a single CSE office, to facilitate CSE staff interviews and case record abstraction. Finally, where several offices were otherwise comparable, we indicated a preference for the site with the larger non-AFDC food stamp caseload or the larger population center. In drawing the survey sample in each site, we started with an automated file of households receiving food stamps in July 1992. We then removed from this file any food stamp households of the following types: Prepared by Abt Associates Inc. 19 Chnpter One: Introduction • Single-person households; • Multi-person households with no members under age 18; or • Multi-person households with members under age 18, any of whom receive AFDC. In all ten sites, the cases that remained in the sample frame were thus multi-person food stamp households with at least one member under age 18, where none of those under age 18 received AFDC. In Alabama, Florida, and Oklahoma, the sample frame included those non-AFDC food stamp households where one or more of those under age 18 received Medicaid on a medical assistance only (MAO) basis.12 We then divided the sample frame into three strata, as follows: • CSE participants—Non-AFDC food stamp households with at least one member also appearing in a child support enforcement case within the state;13 • CSE nonparticipants of type A—Non-AFDC food stamp households with no associated child support enforcement case and with either only one adult member (i.e., only one member 18 years of age or older) or adult members of only one sex; and • CSE nonparticipants of type B—Non-AFDC food stamp households with no associated child support enforcement case and with adult members of the opposite sex. We then randomly sorted the cases in each site into replicates of CSE participants (ten cases per replicate) and CSE nonparticipants (eight cases of type A and two cases of type B per replicate). We stratified the nonparticipants into types A and B to minimize the effort expended in screening out cases of type B, which were less likely (than those of type A) to have children of 12. In New Jersey and Texas, the person-level indicator of AFDC receipt was a joint indicator of AFDC or Medicaid receipt. In these two states we thus effectively excluded from the sample the non-AFDC food stamp households where one or more of those under age 18 received Medicaid on a "medical assistance only" basis. 13. We described earlier in this chapter the approach taken in each state for identifying CSE participants. For the sites in Alabama, New Jersey, and Oklahoma, the automated match of Social Security numbers used statewide child support enforcement files. In Texas, because of the size of statewide CSE files, the match was with cases in (a) the CSE office(s) serving the local food stamp site, and (b) all CSE offices that serve adjoining geographic areas. Prepared by Abt Associates Inc. 20 Chapter One: Introduction noncustodial parents and thus to qualify for the survey. In computing all survey findings, we have appropriately weighted the survey observations to account for this stratification. Replicates were released to the survey staff as necessary to complete interviews with 20 CSE participants and 20 CSE nonparticipants per site. Once a replicate was released, work was completed on all cases in the replicate. This was to ensure that the completed interviews would constitute a random sample, while minimizing the chances of overshooting the target number of completed interviews per site. We provided each site with a pre-screening list of the cases in the initial sample replicates, for the purpose of verifying the client's address and telephone number and the household's continued receipt of food stamps as of November 1992. Exhibit 1.2 RECIPIENT SURVEY SUMMARY: CASES SAMPLED, SCREENED, AND INTERVIEWED Number of Cases Cases sampled Cases prescreened Found ineligible Outside office jurisdiction Not food-stamp-only Released for interview Cases not screened (not located) Cases screened Found ineligible Found eligible Interviews not completed Interviews completed 1,655 136 433 1,086 216 442 14 414 - Source: Abt Auociates, 1993 survey of food tump-only custodial parents. As shown in Exhibit 1.2, a total of 1,655 cases entered the survey sample. During pre-screening, office staff found 136 cases to reside outside the office jurisdiction, and 433 were found to no longer receive food stamps or to have become an AFDC food stamp case. The remaining 1,086 cases were released for interview. Of these, 216 could not be located, 442 Prepared by Abt Associates Inc. 21 Chapter One: Introduction were screened out as ineligible,14 and 14 were found eligible but did not complete the interview. Interviews were completed on 414 cases. The completion rate was 78 percent (414/534), if one assumes that the extent of eligibility among the 216 unlocated cases was the same as for those screened (i.e., 428/870 or 49 percent).15 Of the 414 completed interviews, interviewers completed nearly one-half by telephone and the remainder in person. Interviewers completed 40 (or more) interviews in nine of the ten sites—all but the Jacksonville site, where they exhausted the sampling frame after completing 32 interviews. Of the 414 completed interviews, 218 were sampled as CSE participants, 179 were sampled as CSE nonparticipants of type A, and 17 were sampled as CSE nonparticipants of type B.16 The survey was conducted over a ten-week period from mid-January to late March 1993. See Appendix D for a copy of the survey instrument. The responses to each survey item are tabulated in Appendix CSE Case Record Abstracts After completing the interviews, we abstracted information from CSE case records for those 239 respondents participating in the CSE Program in the interview month, based on whether a CSE case actually existed for the respondent in the local CSE office. To identify these respondents as participants, in each site we provided a list of the survey respondents (and their food stamp case numbers) to the CSE office and asked the staff to check them against the CSE caseload. This allowed us to identify CSE participants by their "true" CSE status, versus their self-reported status (from the relevant survey items) or the status by which they were sampled (that is, their sampling stiatum). 14. Either DO longer receiving food stamps, receiving (or applying for) AFDC, without children under 18 of an absent parent, with foster children only, or with a deceased case bead. 15. The estimated number of eligible cases released for interview, 534, is computed as follows: 414 + 14 + (216) [(414 + 14) / (414 + 14 + 442)]. 16. The 414 completed interviews included 147 interviews for which Medicaid covers either the custodial parent or a child of an absent parent included in the food stamp grant. Based on the CSE case record abstracts, 90 of those respondents participated in CSE, in keeping with the eligibility requirements for Medicaid. We have included the remaining 57 in our estimates of potential response to a mandate or outreach, even though they were already subject to a CSE participation requirement. Prepared by Abt Associates Inc. 22 Chapter One: Introduction Once the participating cases were identified, field staff conducted the case record abstracts during March and April 1993. (In the two New Jersey sites, state CSE staff completed the abstracts.) The abstraction proceeded on the basis of "case record units," which normally corresponded to a sibling group (children from the same custodial-noncustodial parent pair). A respondent could have more than one case record unit—most typically, if a woman had children from several noncustodial fathers. There were multiple case record units for about one-fourth of the 239 respondents participating in CSE. The total number of case record units among the 239 respondents was 307. We did not abstract case information (and we did not regard the custodial parent as a CSE participant) in situations in which the CSE activity was confined to collection of AFDC arrears payments only17 or in which all children in the case record unit were 18 years old or older. See Appendix F for a copy of the case record abstraction instrument. Program Staff and Client Advocate Interviews To understand better the program environment in each of the study sites, we conducted interviews with state and local staff in both the Food Stamp and CSE Programs, as well as representatives of client advocacy groups. Two-person teams conducted these interviews during site visits in the summer and fall of 1992. For each state, the interviews took place at the state office buildings in the capital city and in the food stamp and CSE offices for each study site. These interviews aided in formulating hypotheses about the factors affecting CSE participation, selecting appropriate variables for the multivariate analysis of CSE participation, and interpreting the site differences in CSE participation rates. 17. The HHS Office of Child Support Enforcement defines an "AFDC arrears only" case as "one in which the children to be supported are former recipients of IV-A [AFDC] payments and in which the absent parent is now delinquent in his or her reimbursement of these payments to the government." We excluded these cases from the analysis because there is no effort under way to collect current support for the children. Any amounts collected on these cases go entirely to the federal and state governments (according to their respective cost-sharing of AFDC benefits), not to the custodial household. CSE services to these cases and any resulting collections would thus have no impact on either the income of the food stamp household or costs to the Food Stamp Program. Prepared by Abt Associates Inc. 23 Chapter One: Introduction 1.3 ORGANIZATION OF THIS REPORT The next three chapters address in sequence the three research questions posed at the outset of this chapter. Chapter Two presents our findings with respect to the needs for child support enforcement among food stamp-only custodial households not currently participating in the CSE Program. Chapter Three discusses the potential for increased CSE participation through either a mandate or improved outreach. Chapter Four examines the potential benefits and costs associated with either a mandate or outreach, as defined in Section 1.1 of this chapter. Chapter Five considers the policy implications of the findings reported. Prepared by Abt Associates Inc. 24 CHAPTER TWO NEEDS FOR CHILD SUPPORT ENFORCEMENT This chapter examines the extent to which food stamp-only cases need child support enforcement services. The findings define the size of the target population of this study: food stamp-only custodial households not participating in the CSE Program that need a support order or collection on an existing order. The target population is thus restricted to those households that might potentially benefit from either a CSE mandate or improved CSE outreach. Specifically, we estimate the percentage of all food stamp households nationally that meet a'l of the following criteria: • food stamp-only—that is, receiving food stamps but neither AFDC nor Medicaid, • custodial—that is, including at least one child of a noncustodial parent, • not participating in the CSE Program, and • lacking a support order or not receiving full payment on an existing order. Our basic finding is that nationally, the target population constitutes 2.8 percent of allfood stamp households. In other words, 2.8 percent of the national caseload constitutes food stamp-only custodial households that are not participating in the CSE Program but have a need for child support enforcement. Based on the current national monthly caseload of approximately 11 million households, the estimated 2.8 percent implies a target population nationally of 300,000 households. As illustrated in Exhibit 2.1, one can regard the national estimate of 2.8 percent as the product of the following two factors. The first is the percentage of all food stamp households that are food stamp-only custodial households. We estimate this nationally at 7 percent. The specific national estimates are 6.7 percent from CPS, 6.4 percent from SIPP, and 9.2 percent from the 1991 food stamp quality control (QC) data. (The QC data indicate substantial interstate variation in this percentage, from below 2 percent to above 20 percent of households.) The 7 percent national figure represents a defensible mid-range assumption for subsequent calculations. The second factor is the percentage of food stamp-only custodial households that are CSE nonparticipants and that have needs for CSE services, estimated at 40 percent. This second Prepared by Abt Associates Inc. 25 Chapter Two: Needs for Child Support Enforcement Exhibit 2.1 IDENTIFICATION OF THE TARGET POPULATION FOOD STAMP HOUSEHOLDS (100%) -47% AFDC/MA FOOD STAMP HOUSEHOLDS (47%) -53% FOOD STAMP-ONLY HOUSEHOLDS (53%) -13% -87% WITH AT LEAST ONE CHILD OF OF A NONCUSTODIAL PARENT (7%) WITHOUT A CHILD OF NONCUSTODIAL PARENT (46%) -46% -54% NOT PARTICIPATING IN CSE (3.2%) -88% PARTICIPATING IN CSE (3.8%) -12% NEEDING SUPPORT ORDER OR ENFORCEMENT (2.8%) WITH iW ORDER AND FULL PAYMENT (0 .4%) Target population of this study Source: Abt Associates, FY 1991 national food tump quality control sample, 1993 survey of food stamp-only custodial parents, and accompanying CSE case record abstracts (sec Exhibits 2.3, 2.5 and 2.7). Notes: Percentages in parentheses arc computed as proportions of the total national food sump caseload. Percentages preceded by - are conditional estimates, computed as proportions of the caseload segment represented by the preceding box. Prepared by Abt Associates Inc. 26 Chapter Two: Needs for Child Support Enforcement factor is itself derived from a 46 percent CSE nonparticipation rate and an 88 percent rate of CSE needs among nonparticipants, as described below. The estimated 46 percent CSE nonparticipation rate reflects the estimate from the ten-site pooled survey/abstract data (46.3 percent), and is in the range of estimates obtained from statewide administrative data in three of the five participating States (45.5 percent in Alabama, 39.8 percent in New Jersey, and 54.3 percent in Texas). As discussed in Section 2.2, we regard as biased the higher estimates obtained from CPS and SIPP and the lower estimates obtained from the other two participating States (Florida and Oklahoma). The estimated 88 percent rate of CSE needs reflects the estimate from the survey/ abstract data (87.7 percent), which we consider more accurate than the estimates obtained from either the CPS (77.1 percent) or SIPP (81.1 percent). These estimates all indicate the proportion of custodial parents either needing a support order or not receiving full payment on an existing order, among food stamp-only CSE nonparticipants. The following sections provide detail on these estimates of the target population. First we discuss our national and state-by-state estimates of the proportion of the food stamp caseload comprised of food stamp-only custodial households, as obtained from the CPS, SIPP, and food stamp QC data. Then we present our estimates of CSE nonparticipation and CSE needs, as obtained from the CPS, SIPP, statewide administrative data from the five participating states, and the pooled survey/abstract data from the ten study sites. We use multiple data sources in this analysis because no single data set provides sufficiently detailed information for a nationally representative sample. The need for national estimates of the target population arises from our need in Chapter Four to evaluate the national benefits and costs of a CSE mandate or improved outreach. In Chapter Four, we will project onto the national food stamp caseload the effects of these policy options as estimated from the survey/abstract data in the ten study sites. To do this, it is essential to know the extent of the national caseload that corresponds to the surveyed population—food stamp-only custodial households. As we describe in detail throughout this chapter, the data sources used here have differing limitations. To proceed with our later analysis of policy impacts, we must consider the reliability of the varying estimates and then choose specific empirical values as the basis for subsequent benefit-cost calculations. The choices discussed in this chapter ultimately reflect Prepared by Abt Associates Inc. 27 Chapter Two: Needs for Child Support Enforcement subjective judgments, weighing a series of factors that mitigate against any precise, definitive answers. We consider this an appropriate research strategy, given the exploratory nature of the study. Recognizing the inherent imprecision, we later explore (in Chapter Four) the sensitivity of the national benefit-cost estimates to the particular empirical values estimated here. 2.1 FOOD STAMP-ONLY CUSTODIAL HOUSEHOLDS We derived three independent national estimates of the percentage of all food stamp households that are food stamp-only custodial households: 6.7 percent from the CPS, 6.4 percent from SIPP, and 9.2 percent from QC data. We discuss below the possible sources of bias in these estimates. One should note that the difference between the CPS and SIPP estimates, shown in Exhibit 2.2, is not statistically significant. However, both the CPS and SIPP estimates are significantly lower than the QC estimate, shown in Exhibit 2.3.' Exhibit 2.2 CPS- AND SIPP-BASED ESTIMATES OF FOOD STAMP-ONLY CUSTODIAL FAMILIES CPS (1990) SIPP (1990) Number Percent Number Percent All families on food stamps Families with a custodial parent on food stamps* Food stamp-only 7,821,141 100.0 2,628,201 33.6 6,918,264 100.0 3,024,814 43.7 524,197 6.7 440,565 6.4 !_ Source: Urban institute, 1990 Current Population Survey and the 1990 Survey of Income and Program Participation. * CPS estimate include! families with custodial mothers; SIPP estimate includes families with custodial mothers or custodial fathers. In our later benefit-cost calculations, we need a single point estimate and not a range of values for the percentage of total food stamp households represented by food stamp-only custodial households. This is necessary to convert the findings of the survey/abstract data into national impacts. Ideally, one would either choose a best estimate from among the three 1. The sample sizes used to derive these estimates were as follows: 1,118 for the CPS data, 782 for the SIPP data, and 64,311 for the QC data. Prepared by Abt Associates Inc. 28 Chapter Two: Needs for Child Support Enforcement Exhibit 2.3 DISTRIBUTION OF FOOD STAMP HOUSEHOLDS BY AFDC/MEDICAID RECEIPT, CUSTODIAL STATUS, AND STATE, FISCAL YEAR 1991 State United States Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware District of Columbia Florida Georgia Guam Hawaii luaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada Number of Households 8,862,066 186,152 10,134 137,498 93,938 807,135 94,672 69,953 15,209 32,012 401,704 243,569 3,294 34,447 23,420 460,226 131,115 71,307 60,061 183,322 261,434 50,533 129,899 172,325 408,046 119,476 186,862 189,540 22,805 39,171 27,909 AFDC/MA Food Stamp Households Food Stamp-Only Households Not Custodial Custodial Total Percentage ofAll Food Stamp Households 46.6 44.2 9.2 100.0 24.7 55.2 37.5 22.3 79.6 39.6 64.7 46.4 58.4 33.2 37.3 33.5 41.0 38.7 50.5 45.9 48.0 54.4 44.3 33.8 44.7 53.5 54.2 55.7 52.9 31.4 42.7 33.7 44.7 27.2 51.7 42.2 49.8 58.5 18.7 45.4 33.6 37.4 36.6 48.7 46.3 48.5 53.3 47.7 45.2 41.0 44.8 40.9 48.2 45.9 50.2 39.1 42.0 41.7 45.2 53.0 47.3 57.3 44.8 53.6 23.6 100.0 2.6 100.0 12.8 100.0 19.2 100.0 1.7 100.0 15.0 100.0 1.7 100.0 16.2 100.0 5.0 100.0 18.1 100.0 16.4 100.0 18.0 100.0 5.7 100.0 13.6 100.0 4.3 100.0 13.1 100.0 7.2 100.0 4.7 100.0 7.5 100.0 20.3 100.0 5.1 100.0 7.4 100.0 3.9 100.0 2.6 100.0 1.9 100.0 15.6 100.0 10.0 100.0 9.0 100.0 10.5 100.0 19.2 100.0 Prepared by Abt Associates Inc. 29 Chapter Two: Needs for Child Support Enforcement Exhibit 2.3 (continued) DISTRIBUTION OF FOOD STAMP HOUSEHOLDS BY AFDC/MEDICAID RECEIPT, CUSTODIAL STATUS, AND STATE, FISCAL YEAR 1991 0—' State New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virgin Islands Virginia Washington West Virginia Wisconsin Wyoming Number of Households AFDC/MA Food Stamp Households Food Stamp-Only Households Not Custodial Custodial Total Percentage of All Food Stamp Households 20,121 176,268 58,414 775,738 200,968 15,574 498,553 116,355 102,761 456,210 41.2 60.3 42.6 47.3 36.9 51.7 49.8 31.0 39.6 44.6 33,723 57.1 116,325 37.2 18,042 44.5 238,807 43.1 757,112 31.8 39,667 51.5 20,607 48.6 4,301 12.6 171,070 31.1 157,816 57.9 104,546 38.6 100,619 74.6 11.301 52.0 54.7 4.1 100.0 34.9 4.8 100.0 44.4 13.0 100.0 49.1 3.6 100.0 51.0 12.1 100.0 44.9 3.4 100.0 44.4 5.8 100.0 54.1 14.9 100.0 52.9 7.5 100.0 51.6 3.8 100.0 39.7 3.2 100.0 44.8 18.0 100.0 43.5 12.0 100.0 46.9 10.0 100.0 49.4 18.8 100.0 42.1 6.4 100.0 50.2 1.2 100.0 46.8 40.6 100.0 53.6 15.3 100.0 38.1 4.0 100.0 52.6 8.8 100.0 23.3 2.1 100.0 40.0 8.0 100.0 Source: Abt Associates, FY 1991 national food stamp quality control sample. Prepared by Abt Associates Inc. 30 j Chapter Two: Needs for Child Support Enforcement available values after weighing their relative strengths and weaknesses, or derive an adjusted estimate that reconciles their differences. However, in the analysis that follows we find each of the three estimates to have some potential bias, with no clear basis for reconciliation. We thus have sought to identify a single value that represents a reasonable assumption for further calculations. For reasons explained below, we have chosen this value as 7 percent. Assessing the Accuracy of the National Estimates We start here by noting the features of the available estimates suggesting that the true value may lie within their range—that is, above 6 percent but below 9 percent. To reiterate, the measure of interest is the ratio of monthly food stamp-only custodial households to total monthly food stamp households. The 6.7 percent CPS estimate may represent an underestimate of this ratio, on two grounds. First, the numerator excludes custodial fathers and some custodial mothers.2 Second, in classifying respondents as food stamp recipients, the March 1990 CPS uses a reference period for food stamp receipt that is calendar year 1989, an annual interval rather than a monthly interval. The longer the reference period, the greater is the likelihood of counting in the denominator short-term food stamp recipients who tend to be intact families, not custodial families.3 Another reason to believe that the true value may lie within the range of 6 to 9 percent pertains to a source of potential upward bias in the 9.2 percent QC estimate. In particular, the QC numerator represents an overestimate to the extent of false-positive identificati jn of custodial 2. Custodial fathers represent about 6 to 7 percent of the custodial parents in food stamp-only custodial households, based on the SIPP data and uV survey/abstract data. As to the exclusion of some custodial mothers, the skip pattern of questions asked . the CPS child support supplement results in the exclusion of ever-married custodial mothers with children fathered by someone other than the current or most recent spouse. For instance, the survey would miss a woman now married to her second husband also raising a child from her first marriage. The survey/abstract data do not provide sufficient marital histories to estimate reliably the incidence of such situations. 3. The fact that the QC period is more recent historically (than CPS or SIPP) would seemingly cause little difference in the estimates, for the following reason. During the period 1989-91, despite rapid caseload growth, there was relative stability in the percentage of non-AFDC households (about 58 percent) and in the percentage of single-adult female-headed households with children (about 39 percent) in the total food stamp caseload. See U.S. Department of Agriculture, Food and Nutrition Service, "Characteristics of Food Stamp Households: Summer 1991," January 1993, Appendix Tables A-31 and A-53, and corresponding tables from previous issues of the same report. Prepared by Abt Associates Inc. 31 Chapter Two: Needs for Child Support Enforcement households. Such misidentification could arise from the limited detail in the codes that identify familial relationships among food stamp household members. In using these codes, we ran the risk of identifying the following types of households as custodial: a single-parent household in which the other parent is deceased, a household in which both parents jce present but not married, and a three-generation household headed by a child's grandparent but also including both of the child's parents.4 Next, there are other aspects of the estimates suggesting that the true value may lie below the estimated range. First, both the CPS and SIPP estimates are based on the Census-defined family or subfamily—a group of two or more persons who not only reside together but also are related by birth, marriage, or adoption. The ideal unit of observation here is the food stamp-defined household—a group of persons (or a single individual) who occupy the same residence and who purchase and prepare food together. The family unit of observation would tend to bias upward the CPS and SIPP estimates, by excluding from the denominator food stamp recipients who reside alone and who by definition could not represent custodial households. Second, both the CPS and SIPP estimates classify a family as custodial according to whether it includes a child underage 21 of a noncustodial parent, consistent with Census definitions. This tends to overstate the intended numerator, which applies an age threshold of 18 years consistent with the administrative definitions in the CSE mandates for AFDC and Medicaid. Third, CPS and (to a lesser extent) SIPP each tends to under-report means-tested benefits such as AFDC, food stamps, and Medicaid. This by itself might bias upward the estimated percentage, if the undercount of recipients is more pronounced in the denominator (among multiple-benefit 4. The person-level identifying codes in the QC data relate each member of the household to the household head but not to other household members. We classified a household as a custodial case if it met the following conditions: • there were two or more persons in the food stamp household; and • either (a) the case included one or more stepchildren of the household head or (b) the case head had no spouse present and had one or more children (or grandchildren). Prepared by Abt Associates Inc. 32 Chapter Two: Needs for Child Support Enforcement recipients such as AFDC food stamp families) than in the numerator (among single-benefit recipients such as food stamp-only families).5 Finally, but probably of lesser importance, there is one aspect of the QC estimate suggesting that the true value may even lie above the estimated range. Recall that the false-positive identification of custodial households may cause an overestimate in the QC numerator. To the extent that false-negative situations could occur even more frequently, however, the QC numerator may represent an underestimate. For instance, our use of the relationship codes in the QC data would have caused us to falsely identify a household as not custodial in one particular situation: if the female head resides with her current husband and a child fathered by an earlier husband or other partner. In this situation, the child would have been coded as the son or daughter of the household head; we would have incorrectly regarded the woman's current spouse as the child's father.6 The issue of whether the QC estimate represents on overestimate thus rests on the question of whether the false positives outnumber the false negatives. To address this question, it is instructive to compare the QC estimates obtained for the five participating states with the corresponding estimates derived from the state administrative data. As a means of testing the validity of the QC estimate, the state administrative files offer several advantages over the QC data. First, because the state data files include the entire statewide food stamp caseload, there is no sampling variability in the estimated percentages. Second, the state files allow us to unambiguously identify as custodial households those food stamp cases with corresponding CSE cases. For four of the five states (all but Florida), the administrative data yield an estimate that lies outside the 95 percent confidence interval of the QC-based estimate (see Exhibit 2.4). For several reasons, one would expect some discrepancies in these estimates for any given state. 5. Consider the following example using AFDC and food stamps. Assume that the total food stamp caseload consists of X non-AFDC food stamp cases and Y AFDC food stamp cases. The true portion of non-AFDC cases is thus X/(X+Y). Assume that there is complete reporting of both food stamps and AFDC in the QC data. For the Census data (CPS or SIPP), assume that underreporting reduces the observed number of non- AFDC food stamp cases from X to aX and reduces the number of AFDC food stamp cases from Y to abY, where a and b (both less than 1) are the reporting rates for food stamps and AFDC, respectively. The observed percentage of non-AFDC cases in the food stamp caseload will equal aX/(aX+abY) in the Census data, which equals X/(X+bY), and thus is greater than the value X/(X +Y) derived from the QC data. 6. If instead the male spouse had been designated as the household head, with the child thus coded as a stepchild, we would have correctly classified the case as a custodial household. Prepared by Abt Associates he. 33 Chapter Two: Needs for Child Support Enforcement Exhibit 2.4 FOOD STAMP-ONLY CUSTODIAL HOUSEHOLDS AS A PERCENTAGE OF ALL FOOD STAMP HOUSEHOLDS IN THE FTVE PARTICIPATING STATES Food Stamp-Only Custodial Households QC Data Administrative Data State (FY 1991) (July 1992) Percentage ofAll Food Stamp Households 1 Alabama 23.6' 28.5 1 Florida 18.1 16.4 | New Jersey 4.8' 8.9 | Oklahoma 14.9s 17.3 1 Texas 18.8' 10.2 Source: Abt Associate!, FY 1991 national food stamp quality control sample and 1992 state administrative data for food stamps and CSE. 1 Significantly different at the .05 level from the estimate derived using the adminis-trative data. First, as described earlier, the limited demographic information in both the QC files and the administrative data required the use of approximating criteria for identifying custodial households. To recall, in the QC data a food stamp-only case was identified as a custodial household if two conditions were met: • Two or more persons in the food stamp household; and • Either (a) one or more stepchildren under the age of 18, or (b) the case head has no spouse present with one or more children (or grandchildren) under the age of 18. In the state administrative data, a food stamp-only case was classified as a custodial household if it met either of the following criteria: • CSE participant; or • A multiperson household with at least one member under age 18 and no opposite-sex members over age 18. One would expect these differing criteria to lead to differing estimates of the food stamp-only custodial population. On the one hand, the state administrative data would appear more reliable, Prepared by Abt Associates Inc. 34 Chapter Two: Needs for Child Support Enforcement as we can unambiguously identify a food stamp-only custodial household in those instances where a CSE case exists. On the other hand, among food stamp-only cases where no CSE case exists, the more detailed .elationship codes for each household member in the QC data enable a more accurate identification of custodial households than is possible with the state administra-tive files. Note that Texas, by far the largest of the five states, is the one for which the QC estimate significantly exceeds the corresponding administrative estimate for the percentage of food stamp-only custodial households. For the five states combined, the Texas overestimate (along with the much smaller and insignificant Florida overestimate) more than offsets the underestimates in the other three states. The five-state QC estimate (17.4 percent) thus exceeds the corresponding administrative estimate (14.2 percent). If we proportionally adjust the 9.2 percent national QC estimate to account for the extent of net upward bias apparent from the comparison with the administrative data, we obtain an adjusted national QC estimate of 7.5 percent. In summary, we have raised here a variety of issues pertaining to the available national estimates of the percentage of food stamp households that consist of food stamp-only custodial households. The weight of evidence implies a true value that lies below the QC estimate (9.2 percent) and may even lie below the values derived from CPS (6.7 percent) and SIPP (6.4 percent). We interpret the issues raised here as indicating that none of the three available estimates is a compelling choice. For the CPS estimate, there are potential sources of both upward and downward bias with uncertain net direction. In contrast, the SIPP estimate appears to overstate the true value for the same reasons as CPS, with no apparent offsetting sources of downward bias. This would seemingly call into question the even higher QC estimate. We have used the state administrative data as a means of validating the QC estimate and have derived an adjusted QC value of 7.5 percent. Choosing a Single Point Estimate In the end, we have adopted 7 percent as the national point estimate for subsequent calculations. We simply consider it more likely that the correct value lies toward the lower end of the range of the three estimates than either toward the upper end of the range, above the range, or below the range. If the true value lies outside the range, however, we consider it Prepared by Abt Associates Inc. 35 Chapter Two: Needs for Child Suppon Enforcement more likely to lie below the range than above the range. In Chapter Four we therefore check the sensitivity of the benefit-cost estimates to the 7 percent assumption by using an alternative 6 percent assumption. As a final observation here, note that there is substantial variation among states in the percentage of active food stamp cases that are food stamp-only custodial households, as estimated from the ^C. data for all states and from the administrative data for the five states participating in this study. The variation found in the QC data is especially noteworthy, since (compared to the state administrative estimates) the QC estimates reflect a greater degree of consistency in measurement definitions. Based on the QC data by state, as shown in Exhibit 2.3, the portion of total food stamp cases that are food stamp-only custodial households ranges from less than 2 percent in California, Connecticut, Minnesota, and Vermont to more than 20 percent in Alabama and Louisiana. (The Virgin Islands estimate, an extreme outlier at 40 percent, is based on a sample of less than 200 cases.) As one expects, these percentages are inversely related to AFDC benefit levels. The higher the AFDC benefit for a family of given size, the higher the income level at which the household no longer qualifies for AFDC and consequently the narrower the range of income in which the non-AFDC family would still remain income-eligible for food stamps. 2.2 CSE NONPARTICIPATION AND CSE NEEDS The next task is to estimate the percentage of food stamp-only custodial households that do not participate in the CSE Program but might benefit from CSE participation (i.e., which either lack support orders or receive less than full payment on existing orders). We have derived estimates from CPS, SIPP, the survey/abstract data, and state administrative data (Exhibit 2.5). The QC data indicate nothing about participation in (or contact with) the CSE Program or about the status of support orders or support payments. (Child support income is lumped together with "other unearned income.")7 7. It would be difficult to use "other unearned income" as a proxy measure for child support paymenu, for two reasons. First, this category of reported income also includes (to an unknown degree) the following sources: alimony, foster care payments, rental income, private pension benefits, union benefits, dividends, and interest payments. Second, the QC file reflects some pre-editing of this data element to reconcile household gross income with the summation of person-level income amounts. Prepared by Abt Associates Inc. 36 Chapter Two: Needs for Child Support Enforcement Exhibit 2.5 PERCENTAGE OF FOOD STAMP-ONLY CUSTODIAL PARENTS AND HOUSEHOLDS REPORTING NO PREVIOUS CSE CONTACT OR NO CURRENT CSE CASE, BY DATA SOURCE No Previous CSE Contact No Current CSE Case Percentage of Food Stamp-Only Custodial Parents I CPS (1990) SIPP (1990) 68.8 - 74.9 - Survey/abstract data (1993) 59.4 46.3 Percentage of Food Stamp-Only Custodial Households 1 State administrative data (1992) Alabama - 45.5 Florida - 13.2 New Jersey - 39.8 Oklahoma 15.9 Texas 54.3 Source: Urban Institute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation; Abt Associates, 1993 survey of food stamp-only custodial parents, accompanying CSE case record abstracts, and 1992 state administrative data for food stamps and CSE. CSE Nonparticipation As explained below, our estimate of CSE nonparticipation among food stamp-only custodial households is based on the survey/abstract data and is generally consistent with the state administrative data. For CPS and SIPP, the survey items on the custodial parent's involvement with the CSE Program appear to have limited usefulness for the purposes of this study. The items explicitly address prior client-initiated contact with the program, versus current participation. In particular, the April 1990 CPS Child Support Supplement asked the following question: "Have you ever contacted any government agency for aid in obtaining child support (for child(ren) of last divorce or separation)?" Correspondingly, the child support topical module administered in Wave 3 to the 1990 SIPP included the following item: Prepared by Abt Associates Inc. 37 Chapter Two: Needs for Child Support Enforcement "For any of [the noncustodial parent]'s children, has [the custodial parent] ever asked a public agency (such as the child support enforcement office or welfare agency) for help in obtaining child support?" For food stamp-only custodial families, the percentage indicating no prior CSE contact was similar in the two surveys: 68.8 percent in CPS and 74.9 percent in SIPP. For several reasons, these responses appear to overstate the level of nonparticipation in the CSE Program (i.e., understate CSE participation). First, because the question implies voluntary contact by the respondent, its accuracy in identifying all contact between the respondent and the CSE agency is questionable. The percentage of the custodial parents in AFDC/MA food stamp households who report CSE contact is implausibly low—only 38.6 percent in CPS and 47.1 percent in SIPP—despite the requirement for CSE participation among such households. Second, the survey/abstract data indicate that many custodial parents who report no previous voluntary contact with the CSE agency in fact do currently participate in CSE. The survey asked the following question, deliberately worded to resemble the CPS and SIPP items: "Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtaining child support for any of your children . . . ?" As shown in Exhibit 2.6, among those who answered negatively to this question, almost 40 percent have a case in the local CSE office. This is understandable because many food stamp-only cases are former AFDC recipients whose cases would have been continued automatically unless they requested that their case be closed. All contact between CSE and the recipient could have been agency-initiated rather than recipient-initiated. (Also understandable is the fact that, among those who answered "yes" to this question, 26 percent have no current CSE case. Such clients may have had contact with CSE but did not apply for services or may have had a CSE case that was subsequently closed.) We have used the survey/abstract findings to adjust the CPS and SIPP findings with respect to reported no previous CSE contact (i.e., the 68.8 percent CPS estimate and the 74.9 SIPP estimate cited earlier). If one assumes the same pattern found among survey respondents between reported CSE noncontact and actual CSE nonparticipation, the Prepared by Abt Associates Inc. 38 Chapter Two: Needs for Child Support Enforcement implied CSE nonparticipation rate is 49.6 percent for CPS and 51.7 percent for SIPP.8 Both of these adjusted estimates correspond generally with the 46.3 percent CSE nonparticipation rate determined from the survey/abstract data. Exhibit 2.6 REPORTED CSE CONTACT VERSUS PRESENCE OF CSE CASE Does a CSE Case Presently Exist for This Food Stamp-Only Custodial Parent? Question Response Yes No Total Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtain-ing child support...? Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtaining child support...? Yes No Total Yes No Total Number of Respondents 135 40 175 104 239 135 175 239 414 Weighted Percentage* 74.4 2^ 5 100.0 39.5 53.7 60.5 46." 100.0 100.0 Source: Abt Associates, 1993 survey of food s;imp-9only custodial parents and accompoanying CSE case record abstracts. * Row percentages sum to 100. Based on the food stamp and CSE administrative data obtained for this study, as shown in Exhibit 2.5, three of the five states have CSE nonparticipation rates that are in the same range as the 46 percent survey/abstract estimate: 45.5 percent for Alabama, 39.8 percent for New Jersey, and 54.3 percent for Texas. Nonparticipation estimates were much lower for the other 8. For CPS and SIPP, we compute the adjusted CSE nonparticipation rate as the sum of: (a) the percentage reporting no previous CSE contact multiplied by the (survey/abstract-based) conditional CSE nonparticipation rate for those reporting no prior CSE contact (.605), and (b) the percentage reporting previous CSE contact multiplied by the conditional CSE nonparticipation rate for those reporting previous CSE contact (.256). The CPS calculation is (68.8 x .605) + (31.2 x .256) - 49.6. The SIPP calculation is (74.9 x .605) + (25.1 x .256) - 51.7. Prepared by Abt Associates Inc. 39 Chapter Two: Seeds for Child Support Enforcement two states: 13.2 percent for Florida and 15.9 percent for Oklahoma. The latter estimates appear to reflect the fact that the computer systems in these states automatically generate a CSE case number when there is a child of a noncustodial parent in a food stamp household. However, these cases do not truly exist in the CSE Program if the client has filed no application and the CSE agency has provided no services. We encountered this situation for some survey respondents who were sampled as CSE participants (based on the automated files) but were found (at the time of abstraction) not to have any CSE case on file in the local CSE office. CSE Needs The next issue to address is the extent of need for child support enforcement services among those food stamp-only custodial parents who do not participate in CSE. At the extreme, these CSE nonparticipants would pose little concern if they currently receive full payment on privately-arranged support orders. To the contrary, however, more than three-fourths of the food stamp-only custodial parents not participating in CSE either have no support order or receive less than full payment on their order—and thus, by our definition, have needs for CSE services. The evidence from CPS, SIPP, and the survey/abstract data is generally consistent on this point. As shown in Exhibit 2.7, we find the percentage with no order or with less than full payment on an order is 77.1 percent from CPS, 81.1 percent from SIPP, and 87.7 percent from the survey/abstract data.9 Predominantly, such cases lack a support order, versus having an order on which they receive no payment or partial payment. The higher extent of CSE needs indicated in the survey/abstract data (versus either CPS or SIPP) reflects in part a difference in classifying a custodial parent who receives full payment on all support orders obtained, but who has one or more children not covered by an order. Because both CPS and SIPP identify such a parent as having an order and receiving full payment, the case is tabulated here as having no needs for CSE services. However, in the 9. We have based the CPS and SIPP estimates on food stamp-only custodial parents reporting no contact with the CSE agency. From the discussion above, we recognize that the self-reported information from these respondents may not accurately indicate their current CSE participation or nonparticipation. We implicitly assume here that the needs for child support enforcement services among these custodial parents reasonably approximates the profile of needs among those who truly do not participate in CSE. We have tested this assumption with the survey data by comparing the profile of needs among those who report no CSE contact with the profile among those who (based on the case record abstraction) are not CSE participants. The two profiles are very similar. Prepared by Abt Associates Inc. 40 Chapter Two: Needs for Child Support Enforcement Exhibit 2.7 INITIAL CHILD SUPPORT STATUS OF FOOD STAMP-ONLY CUSTODIAL PARENTS NOT PARTICIPATING IN THE CSE PROGRAM CPS" (1990) SIPI* (1990) Recipient Survey' (1993) Percentage of Custodial Parents n Have no order 57.9 65.1 65.1 Have an order, receive no payment 12.7 9.8 13.4 Have an order, receive partial payment £5 fij 2J Subtotal 77.1 81.1 87.7 I Have an order, receive full payment 219 18.9 m 1 Total 100.0 100.0 100.0 Source: Urban Institute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation; Abt Associates, 1993 survey of food stamp-only custodial parents and accompany-ing CSE case record abstracts. . * For custodial mothers reporting no previous CSE contact. b For custodial mothers and fathers reporting no previous CSE contact. c For custodial mothers and fathers with no current CSE case. Those respondents with no support orders for all children of a noncustodial parent were classified above as having no order. Those respondents with a support order were asked, for each order, "In the laat 12 months, how regularly have you received payments under this (agreement/order)?" Those answering "never" for all orders were classified above as receiving no payments. Those answering "seldom" or "occasionally" on at least one order, or answering "regularly but late" or "regularly and on time* for all orders but with at least one child of a noncustodial parent not covered by an order, were classified as receiving partial payments. Those with orders for all children of a noncustodial parent and answering "regularly but late" or "regularly and on time" were classified as receiving full payments. survey/abstract data we regard this parent as still having CSE needs; we therefore tabulate the case along with those having an order but receiving partial payment. As shown in Exhibit 2.7, the latter group with partial needs represents 9.3 percent of custodial parents in the survey/ abstract data, higher than the 6.5 and 6.2 percent estimated from CPS and SIPP, respect-ively. 10 10. We have not attempted to reconcile further the estimates from the survey/abstract data with the estimates from either CPS or SIPP, as the survey/abstract data pertain to six sites that were not chosen to be nationally representative. Prepared by Abt Associates Inc. 41 Chapter Two: Needs for Child Support Enforcement The findings reported in this section can be summarized as follows. Among custodial parents in food stamp-only households, an estimated 46 percent do not participate in the CSE Program. Among these CSE nonparticipants, approximately 88 percent need child support enforcement services—that is, they either lack a support order or receive less than full payment on an existing order. Combining these percentages, we conclude that about 40 percent of food stamp-only custodial households could potentially benefit from measures to require or encourage CSE participation. Because food stamp-only custodial households represent an estimated 7 percent of all food stamp households, these estimates imply that the target population for either a CSE mandate or improved outreach is approximately 2.8 percent of the total food stamp caseload. The specific national estimates obtained from CPS and SIPP are 3.5 percent and 3.9 percent, respectively. For completeness, we show these estimates in Exhibit 2.8, even though we regard them as over-estimating the extent of CSE nonparticipation." Among all states, the size of the target population appears to vary substantially as a percentage of total food stamp cases, in the range of 1 to 8 percent. 11. As described earlier, the extent of "no previous CSE contact" reported either in CPS or SIPP (68.8 percent and 74.9 percent, respectively) implies a substantially lower CSE nonparticipation rate (49.6 percent and SI .7 percent, respectively) when one accounts for the observed pattern of responses in the recipient survey and case record abstracts. Prepared by Abt Associates Inc. 42 Chapter Two: Needs for Child Support Enforcement Exhibit 2.8 CPS AND SIPP ESTIMATES OF THE TARGET POPULATION r CPS (1990) SIPP (1990) Number" Percentage1* Number" Percentage All food stamp families 7,821 Food stamp-only custodial families 524 that have contacted CSE 164 that have not contacted CSE 360 have an order, receive full pay-ment 83 have no order 209 have an order, receive no payment 46 have an order, receive partial payment 23 Subtotal 278 100.0 6.7 2.1 4.6 1.1 6,918 441 111 330 62 100.0 6.4 1.6 4.8 0.9 Target Population 2.7 215 3.1 0.5 32 0.5 0.3 21 0.3 3.5 268 3.9 Source: Urban Inititute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation. * All numbers are in thouiands of families. b All percentages are of all food stamp families nationally. Prepared by Abt Associates Inc. 43 Prepared by Abt Associates Inc. 44 CHAPTER THREE POTENTIAL FOR INCREASED CSE PARTICIPATION This chapter examines the extent to which households in the target population—food stamp iy custodial households not participating in CSE but with needs for CSE services-might enter the CSE Program in response to either a mandate or improved outreach. As indicated in Chapter One, the empirical task here is to decompose the target population of custodial parents into the following three groups: • those unlikely to become CSE participants with either a mandate or outreach; • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants with outreach (and also, therefore, with a mandate, since outreach would be much less compelling of participation than a mandate). The larger the first group, the weaker the case for either a mandate or outreach. The larger the second group, the stronger the case for a mandate. The larger the third group, the stronger the case for outreach, assuming that a mandate imposes higher costs (to agencies and clients) than outreach. As indicated below, the division between the second and third group is somewhat arbitrary, requiring interpretation of the data in identifying those clients whose participation decision might conceivably be affected by caseworker efforts to provide CSE information or make CSE referrals. Our basic findings, derived from the responses of custodial parents to questions on the recipient survey and from information collected subsequently through case record abstracts, are as follows: • An estimated 24 percent ofthe target population is unlikely to respond to either a mandate or outreach. These custodial parents are subdivided as follows—9 percent who say that they would leave the Food Stamp Program altogether, 6 percent who say that they would seek a good cause exemption, 5 percent who say that they would accept a smaller food stamp allotment as a sanction for noncooper-ation, and 4 percent who would "do something else." Prepared by Abt Associates Inc. 45 Chapter Three: Potentialfor Increased CSE Participation • An estimated 39 to 60 percent of the target population is likely to respond to a mandate, but not to outreach. These are custodial parents who indicate that they would comply with a mandate. Their current nonparticipation appears to reflect a deliberate informed choice, and they thus seem unlikely to respond to any outreach effort. • An estimated 16 to 37 percent of the target population is likely to respond to outreach, and thus, presumably, also to a mandate. Unlike the previous group, these custodial parents indicate little or no knowledge of the CSE Program, or they appear undecided about CSE participation and thus might respond to information, referral, or application assistance. The custodial parents in the second group, those responding to a mandate but not outreach, thus represent at least a plurality—and perhaps a majority—of the target population. To examine multivariate effects on CSE participation, we have also estimated a series of regression equations. The included variables in these equations explain 41 percent of the variation in CSE participation among cases never receiving AFDC previously and 26 percent of the variation among cases previously on AFDC. As one might expect, the pattern of effects is quite different between the never-on-AFDC and ever-on-AFDC groups. For the never-on-AFDC cases, CSE participation implies client initiative to enter the program. In contrast, for the ever-on- AFDC group, CSE participation is typically a default outcome; the client initially cooperated with CSE to receive AFDC benefits and subsequently (after losing AFDC benefits) took no action to leave the CSE Program. Importantly, the multivariate analysis of never-on-AFDC casesfound significantly higher CSEparticipation in the two Florida sites, where caseworkers routinely provide CSE information and referral at the time of initialfood stamp certification. These latter findings, which take account of differences among sites in case demographic characteristics, suggest that CSE participation may be influenced by administrative practices that link more closely the Food Stamp and CSE Programs. Other possible explanations—e.g., that differences in CSE effectiveness tend to attract or discourage CSE participation—are difficult to test, given the small number of participating sites. The first section of this chapter provides details on the key findings mentioned above with respect to the possible responses of custodial parents in the target population to either a Prepared by Abt Associates Inc. 46 Chapter Three: Potentialfor Increased CSE Participation mandate or outreach.1 The second section of the chapter then examines the descriptive characteristics of the target population pooled across the ten study sites, and presents our multivariate analysis of CSE participation among food stamp-only custodial parents. 3.1 POTENTIAL RESPONSE TO A MANDATE OR OUTREACH We examine here the extent to which the target population of custodial parents (food stamp-only, not participating in CSE, but with needs for CSE services) consists of those who (a) would respond to neither a mandate nor outreach, (b) would respond to a mandate but not outreach, or (c) might respond to outreach and thus also presumably to a mandate. As explained below, the easier issue is to identify the first subgroup; the more difficult question, requiring some interpretation of the survey evidence, is to distinguish between the second and third groups. Reasons for Nonparticipation In the recipient survey, we asked custodial parents about their contact with the CSE agency. For those having heard about the CSE agency but not currently participating in CSE, we asked them to indicate their main reason for not using the CSE Program. Here we review the pattern of responses for custodial parents in our target population—food stamp-only custodial parents not participating in CSE (based on our case record abstracts) but with child support needs (without support orders or receiving less than full payment on their orders). Exhib
Object Description
Title | Participation in the Child Support Enforcement Program among non-AFDC food stamp households |
Date | 1995 |
Contributors (individual) | Mills, Gregory B., 1951- |
Contributors (group) | United States Dept. of Agriculture Food and Consumer Service.;Abt Associates.;Urban Institute. |
Subject headings | Food stamps--United States;Child support--United States |
Type | Text |
Format | Pamphlets |
Physical description | 1 v. (various pagings) :ill. ;28 cm. |
Publisher | Alexandria, Va. : U.S. Dept. of Agriculture, Food and Consumer Service, |
Language | en |
Contributing institution | Martha Blakeney Hodges Special Collections and University Archives, UNCG University Libraries |
Source collection | Government Documents Collection (UNCG University Libraries) |
Rights statement | http://rightsstatements.org/vocab/NoC-US/1.0/ |
Additional rights information | NO COPYRIGHT - UNITED STATES. This item has been determined to be free of copyright restrictions in the United States. The user is responsible for determining actual copyright status for any reuse of the material. |
SUDOC number | A 98.2:C 43/6 |
Digital publisher | The University of North Carolina at Greensboro, University Libraries, PO Box 26170, Greensboro NC 27402-6170, 336.334.5304 |
OCLC number | 903978299 |
Page/Item Description
Title | Part 1 |
Full-text | *k COMPLETED ti 9t>3: C fs/f ") j£5&s. United States H&j^flft Department of i|i Agriculture Food and Consumer Service Office of Analysis and Evaluation Participation in the Child Support Enforcement Program Among Non-AFDC Food Stamp Households Final Report February 1995 9S-M3X5" United States Food and Department of Consumer Agriculture Service 3101 Park Center Drive Second Floor Alexandria. VA 22302 Participation in the Child Support Enforcement Program Among Non-AFDC Food Stamp Households Gregory B. Mills Elsie C. Pan Elaine Sorenson Sandra J. Clark Margaret Weant A product of: Abt Associates Inc. 55 Wheeler Street Cambridge, MA 02138 and The Urban Institute 2100 M Street N.W. Washington, DC 20037 February 1995 This study was conducted under contract number 53-3198-1-017 with the Food and Consumer Service. U.S. Department of Agriculture. Points of view or opinions stated In this report do not necessarily represent the official position of the Food and Consumer Service. ACKNOWLEDGMENTS The authors would like to thank the many people who made important contributions to this report. The following individuals deserve special recognition. Diana Perez, Carol Olander, and Steven Carlson of the Food and Consumer Service provided overall leadership, direction, and support throughout the study. Linda Mellgren, Pat Hagen, and Carl Montoya of the U.S. Department of Health and Human Services gave many useful suggestions on the research plan and the presentation of findings. Many staff at Abt Associates assisted with the data collection and analysis, including Diane Porcari, Susan Kannel, Jay Leatherman, Michael Walker, Alan Werner, Kymn Kochanek, Sara Craddock, Jane Belcore, Lynn MacKenzie, and Marva Lopez. Chris Hamilton and Kathleen Flanagan provided technical direction and management support. Wendy Campbell, editorial consultant, made numerous valuable comments as the report neared completion. Susan Byers and Eileen Fahey skillfully produced the draft and final versions of the report. The project staff at the Urban Institute played an essential role through careful analysis of Census data and use of the TRIM2 microsimulation model. Freya Sonenstein gave helpful early guidance on the evaluation design. Sharon Hilliard, Brenda Brown, and Sonja Drumgoole provided steadfast clerical support and production assistance. Finally, the study could not have been completed without the cooperation of clients and staff from the Food Stamp Program and the Child Support Enforcement Program in the participating states. « » TABLE OF CONTENTS EXECUTIVE SUMMARY i CHAPTER ONE INTRODUCTION 1 1.1 Research Questions 3 1.2 Data Sources 13 1.3 Organization of This Report 24 CHAPTER TWO NEEDS FOR CHILD SUPPORT ENFORCEMENT 25 2.1 Food Stamp-Only Custodial Households 28 2.2 CSE Nonparticipation and CSE Needs 36 CHAPTER THREE POTENTIAL FOR INCREASED CSE PARTICIPATION 45 3.1 Potential Response to a Mandate or Outreach 47 3.2 Factors Associated with CSE Participation 55 CHAPrER FOUR BENEFITS AND COSTS OF TWO POLICY OPTIONS 69 4.1 Method of Estimating Changes in Child Support Status 71 4.2 Effects on Household Incomes 87 4.3 Effects on Government Expenditures 93 CHAPTER FIVE POLICY IMPLICATIONS 101 5.1 Assessment of the Problem 101 5.2 Mandate Versus Outreach 103 Prepared by Abt Associates Inc. ' * * Table of Contents APPENDICES APPENDIX A ANALYSIS OF CURRENT POPULATION SURVEY APPENDIX B ANALYSIS OF SURVEY OF INCOME AND PROGRAM PARTICIPATION APPENDK C INTEGRATED QC REVIEW SCHEDULE APPENDK D RECIPIENT SURVEY INSTRUMENT APPENDK E RECIPIENT SURVEY RESPONSES APPENDK F CASE RECORD ABSTRACTION INSTRUMENT APPENDK G TRIM MICROSIMULATION TECHNICAL DETAIL Prepared by Abt Associates Inc. W LIST OF EXHIBITS CHAPTER ONE 1.1 Accounting for the Benefits and Costs of the Two Policy Options 11 1.2 Recipient Survey Summary: Cases Sampled, Screened, and Interviewed . 21 CHAPTER TWO 2.1 Identification of the Target Population 26 2.2 CPS- and SIPP-Based Estimates of Food Stamp-Only Custodial Families . 28 2.3 Distribution of Food Stamp Housenolds by AFDC/Medicaid Receipt, Custodial Status, and State, Fiscal Year 1991 29 2.4 Food Stamp-Only Custodial Households as a Percentage of All Food Stamp Households in the Five Participating States 34 2.5 Percentage of Food Stamp-Only Custodial Parents and Households Reporting No Previous CSE Contact or No Current CSE Case, by Data Source 37 2.6 Reported CSE Contact Versus Presence of CSE Case 39 2.7 Initial Child Support Status of Food Stamp-Only Custodial Parents Not Participating in the CSE Program 41 2.8 CPS and SIPP Estimates of the Target Population 43 CHAPTER THREE 3. i Survey Respondents' Reasons for Not Participating in the CSE Program . . 48 3.2 Survey Respondents' Predictions of Their Responses to a CSE Mandate in the Food Stamp Program 50 3.3 Identification of Custodial Parents Who Potentially Would Respond to Improved Outreach 54 3.4 Characteristics of Custodial Parents: All Survey Respondents 56 3.5 Characteristics of Custodial Parents: Never on AFDC and Ever on AFD2 62 3.6 Regression Estimates 64 CHAPTER FOUR 4.1 Distribution of Food Stamp-Only Custodial Parents 75 4.2 Upper-Bound Effect of a Mandate on the Distribution of Food Stamp-Only Custodial Parents by Child Support Status 76 4.3 Characteristics of Custodial Parents: Participating in CSE, Responding to Mandate, and Responding to Outreach 80 4.4 Lower-Bound Effect of Mandate on the Distribution of Food Stamp-Only Custodial Parents by Child Support Status . . 82 Prepared by Abt Associates Inc. . / List of Exhibits 4.5 Upper-Bound Effect of Outreach or. the Distribution of Food Stamp-Only Custodial Parents by Child Support Status 84 4.6 Lower-Bound Effect of Outreach on Distribution of Food Stamp-Only Custodial Parents by Child Support Status 85 4.7 Annual Benefits and Costs of Policy Options 90 4.8 Decomposition of Benefits and Costs 98 Prepared by Abt Associates Inc. VI EXECUTIVE SUMMARY The Food and Nutrition Service (FNS) of the d.S. Department of Agriculture has undertaken this exploratory study to assess the needs for child support among food stamp households and evaluate two policy options for meeting those needs: • a CSE mandate—requiring that custodial parents participate in the Child Support Enforcement (CSE) Program as a condition of their Food Stamp Program eligibility; or • improved CSE outreach—requiring that state food stamp agencies provide CSE information to custodial parents and facilitate their application to CSE at the time of initial food stamp certification. Both options would aim to raise household incomes through increased child support payments, offset partially by reduced food stamp allotments. This study focuses on food stamp-only custodial households—that is, households with children of noncustodial parents that receive food stamps but neither Aid to Families with Dependent Children (AFDC) nor Medicaid benefits. These latter programs already require CSE participation as a condition of program eligibility for custodial parents. We address the following three research questions (in Chapters Two, Tr -e, and Four, respectively): • What are the needs for child support enforcement among food stamp-only custodial households not participating in the CSE Program?1 • What is the potential for increased CSE participation among those with child support needs, through either a mandate or improved outreach? • What are the benefits and costs ofthe two policy options, from the perspective of both program clients and taxpayers? To answer the first question, we employed a variety cf existing data sources, including the March 1990 Current Population Survey (CPS) and the corresponding April 1990 Child Support Supplement, the 1990 Survey of Income and Program Participation (SIPP), food stamp 1. The needs of those already participating in CSE represent a separate policy issue that is not addressed here. Prepared by Abt Associates Inc. i„ •v/l Executive Summary quality control (QC) data for Fiscal Year .991, and state administrative data for the - 1992 food stamp and CSE caseloads. The state administrative data were compiled from five states that were selected to participate in this study. These states—Alabama, Florida, New Jersey, Oklahoma, and Texas—were selected in part because they had sufficiently large non-AFDC food stamp caseloads and were able to provide the necessary automated program data. Each state provided administrative data files enabling us to analyze statewide patterns of CSE eligibility and participation among food stamp households. We also collected and analyzed new data for this study, including a survey of 414 food stamp-only custodial parents, abstracts from CSE case records, and interviews with program staff and client advocates. In each of the five participating states two food stamp offices were chosen as sites for the local data collection. Site selection was based on criteria that included the size of the food stamp caseload (preferably at least 150 food stamp-only custodial households), the diversity of food stamp administrative practices (preferably encompassing a rant, 5 of current CSE outreach activities at the time of certifying applicants' eligibility for food stamps), the measured effectiveness of the CSE Program (preferably average or above-average), and the expected level of cooperation from local program staff (preferably high). The sites selected were: • Etowah County (Gadsden) and Montgomery County (Montgomery), Alabama; • Jacksonville (Southside Service Center) and Lakeland, Florida; • Camden County (Camden) and Hudson County (Jersey City), New Jersey; • Cleveland County (Norman) and Tulsa County (Tulsa), Oklahoma; and • Garland and Lubbock (Parkway office), Texas. We employed the data collected from the survey of food stamp recipients and the CSE case record abstracts to estimate the potential for increased CSE participation through a mandate or through outreach. The survey and abstract data, coupled with microsimulation findings from the Urban Institute's TRIM2 model, then provided the basis for projecting the benefits and costs of the two policy options. Prepared by Abt Associates Inc. ii > * v 11 Executive Summary Needs for Child Support Enforcement With respect to the first research question posed above, our findings are as follows. Of the total number offood stamp households nationwide, 2.8 percent are in the target population of this study: food stamp-only custodial households not participating in the CSE Program but with needs for child support enforcement. These households have at least one child of a noncustodial parent and either have no support order or receive less than the full amount awarded. Based on the fiscal year 1992 national monthly caseload of approximately 11 million households, the estimated 2.8 percent figure implies a target population nationally of 300,000 households. The national estimate of 2.8 percent is the product of two factors. The first is the percentage of all food stamp households nationally that were food stamp-only custodial households. We estimate this nationally at 7 percent. The second factor is the percentage of food stamp-only custodial households that were CSE nonparticipants with ^hild support needs, estimated at 40 percent. We estimate the aggregate needs for child support within the target population at $900 million annually, in 1992 dollars. This represents the additional amount of support payments that would be collected annually if all 300,000 households in the target population received the full amount of support ordered. This calculation assumes an average annual award per household of $3,000, consistent with the TRIM2 simulations. The $900 million total, which includes the shortfall in payments to those who already have support orders as well as the potential payments to those who currently lack support orders, provides clear evidence that some consideration of a policy change is warranted. The aggregate total also provides a benchmark against which to assess a proposed policy's estimated yield in additional support payments. Potential for Increased CSE Participation To address the second research question, we estimated the extent to which hous~nolds in the target population might enter the CSE Program in response to either a CSE mandate or improved outreach. Specifically, we divided the target population into the following three groups: • those unlikely to become CSE participants with either a mandate or outreach; Prepared by Abt Associates Inc. iii Executive Summary • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants with outreach (and also, therefore, with a mandate, since outreach would be much less compelling of participation than a mandate). We hypothesized that the larger the first group, the weaker the case for either a mandate or outreach. The larger the second group, the stronger the case for a mandate. The larger the third group, the stronger the case for outreach. Our basic findings, derived from the responses of custodial parents to questions on the survey of food stamp recipients and from information collected subsequently through CSE case record abstracts, are as follows: • An estimated 24 percent of custodial parents in the target population would respond to neither a mandate nor outreach. These custodial parents are subdivided as follows: 9 percent who would leave the Food Stamp Program altogether if faced with a mandate, 6 percent who would seek a good cause exemption from the mandate, S percent who would accept a smaller food stamp allotment as a sanction for noncooperation, and 4 percent who would "do something else" (possibly complying with a mandate but opting to withhold information about the noncustodial parent). • An estimated 39 to 60percent of custodial parents in the target population would respond to a mandate, but not to outreach. These are custodial parents who indicate that they would cooperate under a mandate and whose current nonparticipation appears to reflect a deliberate informed choice. They thus seem unlikely to respond to any outreach effort. • An estimated 16 to 37percent ofcustodial parents in the target population would respond to outreach (as well as to a mandate). Unlike the previous group, these custodial parents indicate little or no knowledge of the CSE Program, or they appear undecided about CSE participation. They thus might respond to information, referral, or application assistance. The custodial parents in the second group, those responding to a mandate but not outreach, thus represent at least a plurality—and perhaps a majority—of the target population. To examine further the likely patterns of CSE participation, we also estimated a series of regression equations. The variables included in these equations explain 41 percent of the variation in CSE participation among households that had never received AFDC (and thus never been subject to a CSE mandate) and 26 percent of the variation among households that had Prepared by Abt Associates Inc. iv Executive Summary previously received AFDC (and thus been subject to a mandate). When we controlled for client demographic characteristics, households that had never been on AFDC were significantly more likely to participate in the CSE Program in Florida—where food stamp caseworkers routinely provide CSE information and referral—than in the other four states studied. This evidence provides empirical support for the presumption that more active outreach efforts would indeed yield an increase in CSE participation. Benefits and Costs of the Two Policy Options With respect to the third research question, we estimated the following two sets of benefits and costs of a CSE mandate and improved outreach: • What is the net effect on annual household incomes, through increases in child support payments, associated reductions in food stamp allotments, and the food stamp benefits forgone by those who might accept a sanction or leave the program altogether rather than comply with a mandate? • What is the net effect on annual government expenditures, through reductions in food stamp allotments and changes in food stamp and CSE administrative costs? This analysis is based on our definition of the basic elements of either a mandate or outreach strategy, as detailed in Chapter One. The actual impact of either policy change would of course depend on the specific provisions of federal statute and regulation, as well as the manner of implementation carried out by state and local agencies. Our findings, based on microsimulations of national child support payments and food stamp allotments (using the Urban Institute's TRIM2 model) and an analysis of administrative costs in both the Food Stamp and CSE Programs, are: • A mandate strategy, when simulated under relatively optimistic ("upper-bound") assumptions about the effects of increased CSE participation on child support orders and payments is likely to result in a net increase in annual household incomes of $126 million and a net reduction in annual government expenditures of $60 million, expressed in 1992 dollars (relative to a baseline simulation of current policy). Under less optimistic ("lower-bound") assumptions, a mandate might raise household incomes by $9 million, while reducing government costs by $15 million. • An outreach strategy is likely to result in a net increase in annual household incomes of $15 million to $36 million, accompanied by a net increase in annual government expenditures of $9 million to $10 million. Prepared by Abt Associates Inc. v Executive Summary Under either the upper- or lower-bound assumptions, a mandate would thus make both program clients and taxpayers better off. However, the lower-bound effects of a mandate are quite small. Government expenditures arc estimated to drop somewhat, making taxpayers better off, but clients might experience little overall income gain because of the food stamp benefits forgone by those choosing not to comply with the mandate. In contrast, the outreach estimates are mixed. Household incomes would likely rise slightly, making clients better off, but with higher government expenditures. Policy Implications The benefit-cost estimates presented above also allow us to estimate the degree to which either policy option might succeed in closing the $900 million gap between current support payments received by the target population and the estimated potential for payments. A mandate is likely to eliminate between 7 percent and 24 percent of the gap in aggregate support payments, whereas outreach would likely close between 2 percent and 5 percent of the gap, based on the lower- and upper-bound estimates, respectively. The gains achievable through either a mandate or outreach are thus modest relative to the size of the underlying problem. Nevertheless, with policy issues as large and as intractable as this one, one should not dismiss progress of any magnitude. The question then becomes whether a change in policy produces enough gains to be worthwhile. A fundamental distinction between the mandate and outreach options is the combination of yield and risk that each option presents. A mandate offers the prospect of large gains to both clients and taxpayers. However, there is a substantial possibility that a mandate could leave clients only slightly better off.2 In particular, under the lower-bound mandate estimates, the forgone food stamp benefits (among those sanctioned for noncompliance and those opting to leave the program rather than comply) nearly offset the income gains among those who do comply and come to receive additional child support payments. Moreover, the federal savings from a mandate would come largely (if not entirely, in the lower-bound estimates) from forgone 2. Because of data limitations, the lower-bound estimates for the mandate strategy were not empirically derived, but were based on relatively arbitrary estimates of the potential effectiveness of CSE participation for the target population. The risk cited here may therefore be much less, or even more, than we have predicted. Prepared by Abt Associates Inc. vi Executive Summary food stamp benefits. In contrast, outreach poses gains to clients that are at best quite limited, and generates some additional costs for taxpayers. One advantage offered by the outreach strategy is the flexibility of implementing the approach on a selective geographic basis. One possible approach to targeted outreach is an FNS initiative in its Southeast and Southwest regions, which account for more than 60 percent of all food stamp-only custodial households. As a possible first step toward a national outreach policy, a regional initiative would enable FNS to focus its own staff and resources more effectively on the relevant segment of the national caseload. Finally, in evaluating either a mandate or outreach, there are societal concerns that go beyond the scope of measurable benefits and costs. As a matter of social policy, it is important to reinforce the responsibilities of parents to provide for the well-being of their children. For this reason in particular, a mandate or outreach may deserve more attention than would be warranted by the short-term fiscal impacts. Prepared by Abt Associates Inc. vii t'lii CHAPTER ONE INTRODUCTION The Food and Nutrition Service (FNS) of the U.S. Department of Agriculture has undertaken this exploratory study to assess the needs for child support among households receiving food stamps and evaluate two alternative policy options for meeting those needs: a mandate to participate in the Child Support Enforcement (CSE) Program as a condition of food stamp eligibility, and improved outreach to encourage greater voluntary CSE participation. The purpose of both alternatives would be to raise ths incomes of food stamp recipients through increased child support payments, and also thereby reujce the need for food stamps. The study focuses on food stamp households that (a) include at least one child of a noncustodial (absent) parent and (b) receive neither Aid to Families with Dependent Children (AFDC) nor Medicaid benefits. Both the AFDC and Medicaid Programs already require participation in the CSE Program, whereas the Food Stamp Program does not. More specifically, a custodial parent's eligibility for AFDC or Medicaid benefits is conditional upon cooperation with the CSE Program, unless the parent qualifies for a "good cause exemp-tion"— which is allowed, for instance, when one could reasonably expect cooperation to result in physical or emotional harm to the child. For cases not granted a good cause exemption, the sanction for noncooperation is removal of the custodial parent (but not the rest of the household) from the AFDC or Medicaid assistance group.1 There is widespread and increasing recognition that poverty in the United States is importantly associated with families of unwed mothers, parental separation and divorce, and the failure of noncustodial parents to provide child support. Nearly all of the rise in the child 1. To meet the AFDC requirement to participate in the CSE Program, a custodial parent must assign child support rights to the state and must cooperate with the state CSE agency in establishing paternity and obtaining support payments. The custodial parent receives the first $50 in monthly collections for current support; the remainder is distributed between the state and federal governments according to their funding shares for AFDC benefit payments. For families receiving Medicaid but not AFDC—"Medicaid-only recipients "—the custodial parent must assign rights to the state for medical support and must cooperate with the state CSE agency in establishing paternity and in obtaining medical support from the noncustodial parent. When such a case enters the CSE system, the CSE caseworker will typically process it for both cash support and medical support. The cash support payments collected for the case go entirely to the custodial parent, as with any non-AFDC CSE case. Prepared by Abt Associates Inc. Chapter One: Introduction poverty rate during the 1970s and 1980s can be attributed to the declining proportion of children in two-parent families and the corresponding increase in families headed by never-married or divorced mothers.2 As of spring 1990, of the 10 million custodial mothers in the U.S. (living with children under 21 whose fathers are outside the home), only one-half had a court order under which they were to receive child support in 1989. Of these 5 million women, only one-half actually received the full dollar amount, one-quarter received partial payment, and the remaining quarter received no payment.3 The aggregate "child support deficit"—the difference between the total amount of support payments due and the total amount actually received by custodial parents—was $5.1 billion in 1989. Even for those receiving full payment, award amounts are often inadequate—that is, the award amount may not properly reflect the custodial household's needs or the noncustodial parent's income. The CSE Program, enacted in 1975 as Title IV-D of the Social Security Act, assists in locating noncustodial parents, establishing paternity, obtaining support orders, and enforcing support obligations. Services are provided automatically at no charge to those receiving AFDC or Medicaid, and are available to others upon request (in so..ie states, at a small charge to the custodial parent). Food stamp households that receive AFDC or Medicaid thus presently fall under the existing CSE mandate. Those food stamp custodial households that receive neither AFDC nor Medicaid— termed "food stamp-only custodial households"—participate in CSE on a voluntary basis, if they participate at all. Indeed, previous research indicates that only about one-third of food stamp-only custodial households receive support payments through the CSE Program.4 Recognizing the extent to which food stamp-only custodial households do not seek services thro h the CSE 2. Robert I. Lerman, "Policy Watch: Child Support Policies" Journal of Economic Perspectives, Vol. 7, No. 1, Winter 1993, p. 171. 3. U.S. Department of Commerce, Bureau of the Census, "Child Support and Alimony: 1989" Current Population Reports, Consumer Income, Series P-60, Number 173, September 1991, p. 1. 4. Unpublished tabulations by the U.S. Department of Health and Human Services, based on data from the 1988 Current Population Survey, showed that only 33 percent of non-AFDC food sump custodial households received support payments in 1987 through CSE collections. Another 27 percent received support payments outside the CSE system (some on a voluntary informal basis from the noncustodial parent). The remaining 40 percent received no support payments. Prepared by Abt Associates Inc. Chapter One: Introduction Program, FNS has undertaken this study to examine their needs for services and the benefits and costs of alternative approaches to increasing their CSE participation. A key question underlying the present study is whether the unmet needs for child support among food stamp-only custodial households merit any policy action. Given the availability of services through the CSE Program, policy action might either require CSE participation as a condition of food stamp eligibility or encourage greater voluntary use of CSE services through improved outreach. These two policy options raise further questions: • • To what extent would a CSE mandate prompt food stamp households to seek good cause exemptions, to accept a lower food stamp allotment as a sanction for noncooperation, or to leave the Food Stamp Program altogether, to avoid participating in child support enforcement? Might there be very little response to improved outreach efforts, if potential CSE clients currently choose not to participate for reasons other than a lack of program information, such as the uncertain identity of the father, fear of physical harm, or the low prospects for collecting any payment from someone who has little or no income? Given the likely characteristics of the households that would newly enter the CSE Program, what are the prospects for collecting support payments for them? No previous studies have addressed these issues in the specific context of food stamp-only custodial households. This absence of previous research and the hypothetical ("what if4) nature of the questions above mean that this study must be regarded as exploratory. In particular, given the limited data on which projections of national benefits and costs must be based, one must interpret such projections cautiously. 1.1 RESEARCH QUESTIONS As an exploratory study, this research seeks to provide useful insights into the needs for child support enforcement services among food stamp households and the extent to which the CSE Program might meet the needs of those not currently participating. The fundamental research question addressed in this study is whether two policy alternatives being considered by FNS arc likely to be cost-effective in increasing child support payments to food stamp-only custodial households. To answer this question, we have investigated the following three more specific questions: Prepared by Abt Associates Inc. 3 Chapter One: Introduction • What are the needs for child support enforcement among food stamp-only custodial households not participating in the CSE Program? • What is the potentialfor increased participation among those with needs for CSE services, through either a mandate or improved outreach? • What are the likely benefits and costs of the two policy options, a CSE mandate or improved outreach, for program clients as well as public agencies? The remainder of this section discusses each of these three questions in turn. Needs for Child Support Enforcement The first step in this study is to determine how many food stamp-only custodial households both need child support enforcement services and are not currently participating in the CSE Program. These households constitute our target population—the universe of food stamp-only custodial households that might potentially benefit from either a CSE mandate or improved outreach. We define a household as "needing child support enforcement services'1 if there is no child support order, or less than full payment on an existing order, for one or more children of a noncustodial parent. This definition is necessarily arbitrary and has some potential for errors of both inclusion and exclusion. One might argue that some households defined here as needing child support enforce-ment do not pose any potential for increased child support collections, and thus should not be included in the target population. In some situations, for example, there is no immediate basis on which to establish the paternity of the child, or the noncustodial parent has no foreseeable capacity to make support payments. Any effort by the CSE agency in these situations seems destined to yield no short-term result. However, one cannot unambiguously identify such households a priori; the potential for establishing paternity, obtaining an order, and collecting support payments in the long term is an empirical issue. The same situation arises with those who currently have no support order but either have a pending support arrangement or receive support payments informally; there is some potential for obtaining a support order and collecting additional support payments. We therefore include all such households in our target population and then proceed to estimate the likelihood of collecting payments for them. Conversely, one might argue that some households not meeting our definition of "needing CSE services" do pose a potential for increased collections. For example, there are Prepared by Abt Associates Inc. Chapter One: Introduction households not participating in the CSE Program that receive full payment on their support orders, but where the award amounts appear currently inadequate or where the award amounts are not subject to periodic review (as now required in CSE every three years for AFDC cases, effective October 13, 1993). For several pragmatic reasons we have excluded such households from our target population. First, there is no consensus about the adequacy of award amounts set under existing child support guidelines.5 Thus, even with all the necessary information, any attempt to quantify needs would entail judgments about adequate award levels. Second, all guidelines require information about the noncustodial parent's income. This information is unavailable through any existing data sources; nor was it feasible to collect such information in this study except to ask the custodial parent to estimate the noncustodial parent's income. Third, one would expect uneven implementation among states with the requirement for periodic review and adjustment of awards. Fourth, the review of a non-AFDC CSE case currently requires a request from the custodial parent, many of whom decline the opportunity for a review (as one would also expect among food stamp-only custodial households that enter CSE voluntarily in response to more active outreach). Finally, when reviews do take place, they may result in no change to the award amount or even in a downward adjustment to the award amount. For these various reasons, we have assumed no potential for increased collections among those who already receive full payment on their orders. Potential for Increased CSE Participation Once we identify the target population, the next task is to estimate the percentage of such households that would likely respond to actions that FNS might take to require or promote their participation in the CSE Program, and to identify the factors that might cause a custodial parent to decline to participate. In the course of visits to each of the states and localities involved in this study, we conducted interviews with Food Stamp and CSE Program staff and 5. The most common approach to setting award levels is the "income shares" guideline, under which children receive the same share of combined parental income as they would in an intact family. However, states differ in the formula used to compute the noncustodial parent's child support payment. In some states the payment does not depend on combined parental income, instead equaling a fixed percentage of the noncustodial parent's income. In other states the prescribed amount declines at higher levels of income. Differing formulas have differing implications for living standards, labor supply, and incentives for child-bearing and separation and divorce. Prepared by Abt Associates Inc. Chapter One: Introduction client advocates. The reasons for CSE nonparticipation cited in these conversations tended to cluster into the following five categories:6 • Desire to maintain a relationship with the noncustodial parent. The custodial parent wants to continue a situation in which the noncustodial parent has some contact with the children, makes occasional cash payments, or provides in-kind support (diapers, clothing, gifts); feels that the noncustodial parent is "doing as much as he can"; does not want to "drive him away"; does not want to hassle the noncustodial parent (with administrative hearings, paternity tests, court appear-ances, court-imposed requirements for job search as part of a support order, or embarrassing publicity), does not want to give up intermittent informal support for an even less certain situation in which a support order might not be enforced (or in which formal support payments might cause the loss of food stamp benefits), or (at the fraudulent extreme) wants to conceal the fact that the "absent" parent actually resides with the family. • Desire to avoid involvement with the noncustodial parent. The custodial parent has concerns for her safety (or the children's) or does not want the noncustodial parent to obtain visitation privileges in conjunction with a support order, or the caretaker (such as a grandmother) does not want to "go after" the noncustodial parent (such as a son-in-law). • Desire to avoid involvement with program agencies. The custodial parent perceives any involvement with the CSE Program or the judicial system as intrusive, hostile, demeaning, intimidating, stigmatic, or time-consuming. • Low prospects ofobtaining any support payments. The custodial parent perceives little hope of receiving support payments, because the noncustodial parent's identity or whereabouts are uncertain or because the noncustodial parent is unemployed, is incarcerated, or now has another family to support. • Lack of knowledge about the potential benefits and minimal costs of CSE participation. The custodial parent knows little or nothing about the CSE Program; does not understand the potential benefits of establishing paternity and obtaining a child support order (such as health insurance coverage and survivor benefits under Social Security); does not recognize that assistance in obtaining child support would be expensive to obtain through private attorneys or collection agents; or is unaware of the methods available to the CSE agency for enforcing a support order (such as the interception of state or federal tax refunds or unemployment compensation, income withholding, and property liens). 6. We presume in these situations that the custodial parent either has no current support order or receives less than full payment on an existing order. Prepared by Abt Associates Inc. Chapter One: Introduction Recognizing these various reasons for CSE nonparticipation, we consider two alternative strategies that FNS might adopt. One is a CSE mandate, requiring participation in CSE as a condition of the custodial parent's eligibility for food stamps (as it is for AFDC and Medicaid) for households with at least one child of a noncustodial parent. Unless a household receives a good cause exemption, noncooperation would disqualify the custodial parent from food stamp eligibility, thereby decreasing the household's food stamp allotment. (Depending on household size, this would reduce the monthly allotment by about $80, on average.) Under such a mandate, as under current food stamp policy generally, child support payments represent countable income to the household and normally reduce the monthly allotment by 30 cents per dollar of support payments. (A state may disregard the first $50 of support payments in computing countable income if the state agrees to fund the corresponding increase in monthly allotment. However, no states have adopted this program option.) As mandatory participants, food stamp-only custodial households would not pay any CSE application fees or charges for CSE services. The other strategy being considered is to have state food stamp agencies undertake more active CSE outreach efforts. For the purposes of this study, we define improved outreach as food stamp caseworkers providing information on CSE services, referring households to the CSE agency, and assisting households with their CSE application—all during the course of the household's initial application to the Food Stamp Program. Under current CSE policy, the states could charge these households fees of up to $25 for their CSE application and services. It is especially in situations of the last type listed above—where the client lacks knowledge of CSE—that outreach is most likely to influence the custodial parent's decision to participate. Even in the other situations listed, however, there remains the potential that improved information, referral, or counseling would have some effect, depending on the intensity of the outreach efforts. Consider, for instance, the custodial mother who has little hope of collecting payments from a currently unemployed father. A caseworker might explain the value of obtaining a support order so that payments can be collected when the father regains employment. In assessing the relative merits of a mandate versus outreach, it is useful to categorize the target population into the following three categories: Prepared by Abt Associates Inc. ■■■ Chapter One: Introduction • those unlikely to become CSE participants even under a mandate (and thus presumably not with outreach either); • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants under outreach (and presumably with the more compelling mandate, as well). We hypothesize that the larger the first group, the weaker the case for any change to a mandate or an outreach policy. Members of the first group would include those likely to seek a good cause exemption to avoid CSE participation under a mandate, those likely to accept a reduced food stamp allotment as a sanction for noncompliance under a mandate, and those likely to leave the Food Stamp Program and thus forgo their household's entire monthly allotment rather than participate in CSE.7 The choice between a mandate or outreach revolves importantly around the relative sizes of the second and third groups identified above. The larger the second group, the more attractive is a mandate. The larger the third group, the more attractive is outreach. The important empirical issue in this portion of the study is thus the distribution of the target population among these three categories. Benefits and Costs of the Two Policy Options This study aims ultimately to provide FNS with information on the merits of either a CSE mandate requirement or improved CSE outreach. In evaluating these options, we had to define them more specifically to establish a framework for attributing benefits and costs. The actual benefits and costs of any policy change made will of course depend on how federal statutes and regulations are written and implemented. For the purposes of the benefit-cost analysis, we defined a CSE mandate as follows. At the initial food stamp certification, the caseworker would identify food stamp-only custodial households not currently having a child support order or not receiving full payment under an 7. Why might a custodial parent, facing a mandate to participate in CSE, choose to withdraw entirely from food stamps even though the case could remain on the program with a lowered benefit? The household's reduced monthly benefit might be so low that the custodial parent would consider it no longer "worth the hassle" of dealing with the welfare agency and using coupons to make food purchases. Studies of food stamp participation have shown that many eligible households do not enter the program because of the low benefit amount that they would receive. Prepared by Abt Associates Inc. Chapter One: Introduction existing order and not participating in CSE. The caseworker would explain that, to receive food stamps, the custodial parent must register with the CSE agency and cooperate with efforts to obtain child support payments, unless good cause exists for an exemption. As with AFDC or Medicaid, the CSE food stamp eligibility worker would determine good cause based on documentation provided by the custodial parent. Following the current AFDC and Medicaid policy, circumstances considered to warrant a good cause exemption might include the following: • if cooperation with CSE can be reasonably expected to result in physical or emo-tional harm to the child or caretaker relative; • if the child was conceived through incest or rape; or • if legal proceedings are already under way for a third party to adopt the child. Those who do not apply for an exemption or whose application for exemption is denied would need to register with the CSE agency (as currently for AFDC cases). If the custodial parent failed to do so, the food stamp caseworker (perhaps at the first recertification) would then sanction the case, lowering the monthly food stamp allotment by disqualifying the custodial parent or caretaker relative but not the rest of the household from food stamp eligibility. Other food stamp clients might decide to forgo all food stamp benefits, removing the entire household from the program to avoid participating in CSE. Those who complied with the mandate would then become CSE cases and could be treated by CSE in the same manner as AFDC or Medicaid cases. For example, agencies could not charge clients any fees for application or services. Agency efiorts would proceed as necessary (and as feasible based on the limited information provided by some custodial parents) to locate the noncustodial parent, establish paternity, obtain a child support order, and CO lect support payments. Cases would be subject to the same requirements for periodic review and modification as those now applied to AFDC and Medicaid cases.8 8. The priority assigned to CSE services for food stamp-only cases might depend on whether, for the purpose of computing incentive payments to states under Title IV-D, the collections from such cases are treated as AFDC collections or non-AFDC collections. This issue is not addressed in any proposed legislation. Prepared by Abt Associates Inc. Chapter One: Introduction If support payments are collected for the case, all payments would go to the custodial parent.9 Because those payments would count as food stamp income, the monthly food stamp allotment would decline by about 30 cents for each dollar of collections. (This is unlike AFDC, where only the first $50 per month of support payments go directly to the custodial parent.) There are some households for which the 30 percent offset would not apply, such as those that have not fully used their allowable deductions in computing countable food stamp income or those rendered ineligible for food stamps by the size of the increase in child support payments. We define improved CSE outreach in the following way for the benefit-cost analysis. At the initial certification of a food stamp-only custodial household, the food stamp caseworker would determine whether the household had any children with a noncustodial parent and no support order or only partial payment on an existing order. If so, the caseworker would explain the services available through the CSE Program and the potential benefits to the household of establishing a support order and obtaining collections. The caseworker would refer the household to the CSE Program and perhaps assist the client in completing an application for CSE services. The aim of this strategy would be to provide more complete information to food stamp clients who might need services, so that they could make a more informed judgment about whether to participate in CSE, and then to facilitate the clients' CSE applications. The benefits and costs of outreach would be similar in nature to those of a mandate, but outreach would likely entail both lower benefits and lower costs than a mandate. Because the custodial parents responding to outreach constitute a subset of those entering CSE under a mandate, the effects on support payrr-nts and CSE administrative costs are lessened. Moreover, by definition, under outreach no household would have to forgo food stamp benefits to avoid CSE participation, since participation would be voluntary. In the benefit-cost analysis in Chapter Four, we will first account for the net impact on incomes to food stamp-only households. As shown in Exhibit 1.1, there are two main components of this calculation: the estimated increase in child support payments and the associated decrease in food stamp allotments. We next account for the net impact on 9. We will follow here the conventional benefit-cost practice of counting child support payments as benefits to the custodial parent, while not counting such legally-obligated transfers as costs to the noncustodial parent. Note also that, for those noncustodial parents who are also food sump recipients, child support payments are treated as deductions from food stamp countable income. Higher support payments thus would raise food stamp benefits for these noncustodial parents. We do not take any account of this effect. Prepared by Abt Associates Inc. 10 Chapter One: Introduction government expenditures for the Food Stamp and CSE Programs. The exhibit indicates the three main components of this calculation: the estimated decrease in food stamp allotments, the estimated change in food stamp administrative costs, and the estimated increase in CSE administrative costs. Exhibit 1.1 ACCOUNTING FOR THE BENEFITS AND COSTS OF THE TWO POLICY OPTIONS Percentage Distribution 1 Effect (Expected Sign) Total Federal State Net change in household incomes Change in child support payments (+) A — — Change in food stamp allotments" (-) B — — Total (+) A + B — — Net change in government expenditures Change in food stamp allotments (-) B 100% 0% Change in food stamp administrative costs (+ or -) C 50% 50% Change in CSE administrative costs (+) D 66% 34% Total (+ or -) B + C + D — - 1 Approximately 30 percent of the increase in child support pay menu. Several aspects of this benefit-cost framework are noteworthy. First, the decrease in food stamp allotments enters on both the household side (unfavorably, as a reduction to the incomes of food stamp households) and the government side (favorably, as a reduction in government expenditures). Second, we do not attempt to incorporate a variety of potential nonmonetary benefits that might accrue to the custodial household from establishing paternity and obtaining a support order: • Paternity establishment enables children bom to an unmarried couple to gain most of the rights and privileges of children born to a married couple. These include, Prepared by Abt Associates Inc. 11 Chapter One: Introduction in the event of the death of a noncustodial parent, inheritance rights and rights to possible benefits from life insurance, Social Security (survivors' benefits), and veterans programs. • Health insurance coverage for the children and the custodial parent must now be included as part of a CSE-obtained support agreement, when such coverage is available to the noncustodial parent at a reasonable cost. • When paternity and a legal support order are established, children may have better prospects for a meaningful relationship with their father. Third, we also do not account for the out-of-pocket fees and charges that food stamp households may incur as voluntary CSE participants. For a mandate, the exclusion of out-of-pocket fees and charge: serves to understate both the net benefits to clients and net government expenditures, as the CSE Program could no longer collect fees and charges from those who might have voluntarily participated otherwise. For outreach, the exclusion of out-of-pocket fees and charges serves to overstate both the net benefits to clients and net government expenditures. Fourth, a mandate or outreach may not result in higher food stamp administrative costs. In principle, the increased certification costs to administer a mandate or conduct outreach might be offset entirely by savings associated with households made ineligible through increased support payments and households choosing to forgo food stamps altogether under a mandate. Fifth, the benefit-cost analysis will account for the distribution of government expenditures between the federal and state levels, reflecting the cost-sharing provisions for different categories of program expenditures: 100 percent federal funding of food stamp allotments, SO percent federal funding of food stamp administrative costs, and 66 percent federal funding of CSE administrative costs, as shown in Exhibit 1.1. The fiscal implications are thus quite different from the standpoint of states versus the total (federal and state) government sector. The net savings in food stamp allotments would accrue entirely to the federal government. States would bear SO percent of the change in food stamp administrative costs and 34 percent of the change in CSE administrative costs. To summarize, whether for the mandate approach or the outreach approach, the benefit-cost analysis will account for net additional income to food stamp-only custodial households and net government expenditures to the Food Stamp and CSE Programs. Either option would certainly seem desirable if it promised a net increase in household incomes and a net reduction Prepared by Abt Associates Inc. 12 Chapter One: Introduction in government expenditures. We may find, however, that the net increase in household incomes is attainable only with a net increase in government expenditures. Under that scenario, the question becomes whether the gains to food stamp recipients would outweigh the associated losses to taxpayers. The answer will depend on value judgments and considerations beyond any strict benefit-cost accounting. 1.2 DATA SOURCES This section describes the data sources that we have used for this study. We first describe the sources of existing data used to analyze the needs for child support enforcement among food stamp households: the Current Population Survey (CPS) and Child Support Supplement (CSS), the Survey of Income and Program Participation (SIPP), the food stamp quality control (QC) data, and state administrative data from the Food Stamp and CSE Programs. We then describe the data collection efforts undertaken for this study: recipient survey, case record abstracts, and program staff and client advocate interviews. We used the data from these latter sources to estimate the potential for increased CSE participation and increased child support collections under either a mandate or improved outreach. Current Population Survey One source of our national estimates on the needs for CSE services among food stamp recipients is the household survey data from the March 1990 CPS and corresponding data on child support arrangements from the CSS administered in conjunction with the April 1990 CPS (see the description in Appendix A). For those women interviewed in the March 1990 CPS who were demographically eligible to receive child support (by living with one or more "own children" under age 21 of a noncustodial father), about 70 percent were also administered the April 1990 CSS. For this representative national sample, the merged March/April 1990 CPS-CSS file allows one to link information on child support status with detailed information on demographic characteristics, employment, income, and participation in income support programs including food stamps, AFDC, and Medicaid. The information on income, including the receipt of child support payments, pertains to calendar year 1989. The CPS findings reported in this study are based on a sample of 4,064 custodial mothers. Among these, 209 represented food stamp-only families (receiving food stamps but neither AFDC nor Medicaid), another 209 were Prepared by Abt Associates Inc. 13 Chapter One: Introduction AFDC/MA only families (receiving AFDC and/or Medicaid, but not food stamps), 909 were AFDC/MA food stamp families (receiving AFDC and/or Medicaid, plus food stamps), and 2,737 were "no aid" families (receiving neither AFDC nor Medicaid nor food stamps). The primary advantage of the CPS is its large national sample. However, the CPS-CSS merged file has the following limitations: • Income reporting on an annual basis (versus monthly or quarterly), with acknowledged under-reporting of child support payments and income from benefit programs such as AFDC and food stamps; • Exclusion from the April CSS of custodial fathers and ever-married mothers who had children either outside of marriage with other previous partners or in marriages prior to the most recent divorce or separation; • Lack of information on the noncustodial parent; and • For some variables (including key child support characteristics such as the presence of a support order), a substantial percentage (30 percent or higher) of imputed values based on other respondents, because of the limited overlap between the March CPS and the April CSS. Survey of Income and Program Participation A second source of national estimates is the 1990 SIPP. As described in Appendix B, SIPP is a nationally representative longitudinal survey of adults that provides detailed individual and household information on income, wealth, and pro"ram participation. Persons chosen for the survey enter a panel that is interviewed every four months for approximately two and one-half years (i.e., in eight rounds or "waves" of interviewing). The 1990 SIPP panel included about 20,000 households. Each wave of interviewing collects information from the initial sample of adults and all other adults residing with the initial sample members at the time of the interview. Information about income, labor force participation, and program participation is collected on the individual and the individual's household for the four months preceding the interview. SIPP interviews typically include two components: a core questionnaire and one or more "topical modules" sets of supplemental questions on special topics. Information on child support arrangements is included in both the core questionnaire (administered at Waves 1 through 8) and a child support topical module (normally administered at Waves 3 and 8). The Prepared by Abt Associates Inc. 14 Chapter One: Introduction core questionnaire captures information on the amount of child support payments received in each month. The child support topical module captures information from the custodial parent (mother or father) on child support agreements and awards, custody and visitation arrangements, the location of the noncustodial parent, and the use of the public child support enforcement agency. Custodial parents interviewed in the child support topical module from Wave 3 of the 1990 SIPP panel were the ba*is for the SIPP findings in this study. The sample analyzed here includes 3,069 custodial parents, classified as follows (based on benefit receipt in month 12 of the survey year): 113 were food stamp-only families, 152 were AFDC/MA only families, 669 were AFDC/MA fooc stamp lamihes, and 2,135 were "no aid" families. Our sample included individuals living with one or more own children under age 21 whose other parent lived elsewhere. Additional information on these individuals was obtained on previous program participation (from the Wave 2 topical module on recipient history and from the Census Bureau's longitudinally edited core file) and on family characteristics (from the cross-sectionally edited core file). Although the CPS-CSS file has a larger sample size, SIPP offers more extensive information on child support arrangements and use of child support services. SIPP also achieves more accurate reporting of income from food stamps and other assistance programs than CPS.10 Also important here is that SIPP provides retrospective information on prior receipt of AFDC and Medicaid. Any indication of prior AFDC receipt by a food stamp-only custodial parent would suggest earlier participation in (or at least some foreknowledge of) the CSE Program. Such individuals, if currently not participating in CSE, would presumably be less Ukely to respond to outreach efforts than those with no prior AFDC receipt. 10. In neither the CPS nor SIPP analysis is there any attempt to correct for income under-reporting. The SIPP findings thus reflect more reliable income data than the CPS findings. However, in the TRIM2 microsimula-tions discussed later, correction factors are applied to the CPS source data to align baseline estimates of recipients and payment expenditures with the corresponding benchmark totals available through published national program data. Prepared by Abt Associates Inc. 15 Chapter One: Introduction Food Stamp Quality Control Data Also included in this report are national and state-by-state estimates from the Fiscal Year 1991 food stamp QC data, as collected by states under the National Integrated Quality Control System. This cross-sectional data file contains household-level and person-level information on 64,311 active food stamp cases for which QC reviews were completed for the sample months October 1990 through September 1991. The sample sizes range among states from 300 to more than 2,400, varying in relation to state food stamp caseloads. Information from the case record and the QC review findings are recorded by the QC reviewer on a four-page Integrated Review Schedule, as showr in Appendix C. This data source allows for state-by-state estimates of the target population for this study—food stamp-only custodial households. Because the AFDC, Medicaid, and food stamp benefit information comes directly from the case record, such information is not subject to the underreporting that occurs either in CPS or (to a lesser degree) SIPP. Its usefulness is limited, however, by the lack of person-level information that would unambiguously identify children of a noncustodial parent (requiring the use of approximating criteria) and the absence of information on either the receipt of child support payments or participation in the CSE Program. For instance, the QC-based identification of food stamp-only custodial households could either overstate or understate this subset of the food stamp caseload. The uncertain direction of this bias will depend on the relative numbers of false-positive and false-negative misclassifications that result from the lack of person-level relationship codes. State Administrative Data We have conducted an analysis of child support arrangements among food stamp households using administrative data provided by each of the five states that participated in this study: Alabama, Florida, New Jersey, Oklahoma, and Texas. The selection of these five states, completed in April 1992, was based on the following considerations: • The state should have a sufficient capability to provide automated program data, for both food stamps and CSE, for our use in conducting statewide descriptive analysis and in selecting a survey sample in two local food stamp offices. Prepared by Abt Associates Inc. 16 Chapter One: Introduction • The state's monthly food stamp caseload should include at least 5,000 non-AFDC cases with children, to enable selection of a sufficient sample of CSE-eligible food stamp-only custodial households. • Preferably, the state should administer its CSE Program under the same human services or social services umbrella agency as food stamps, should be above-average in measures of CSE performance (such as child support collections per dollar of CSE administrative cost), and should offer a high expected level of cooperation from state and local staff. We also sought geographic variation, but expected that the states would come predominantly from the Southeast and Southwest regions because of the lower-than-average AFDC benefit levels and the resulting larger numbers of non-AFDC food stamp cases (that is, cases eligible for food stamps but with too much income to qualify for AFDC) in those regions. As we report later, QC data indicate that the thirteen states in FNS' Southeast and Southwest regions together account for more than 60 percent of national food stamp-only custodial households. The five states participating in this study alone account for about 35 percent of the national total. Each participating state provided automated data for all food stamp cases and child support enforcement cases active in the month of July 1992. These are extremely large data files, as the monthly food stamp caseloads at that time were approximately 900,000 in Texas, 300,000 in Florida,11 200,000 in both Alabama and New Jersey, and 140,000 in Oklahoma. From these files, we examined the distribution of food stamp households by the following characteristics: • Whether food stamp-only (versus AFDC/MA food stamps); • • Whether CSE-eligible (i.e., having in the food stamp household a child of an absent parent); If CSE-eligible, whether participating in CSE; • If participating in CSE, whether a prior AFDC recipient; • If participating in CSE, whether a support order exists; and 11. The number of households io the Florida analysis file is considerably less than indicated in other program statistics for the state. The data set initially provided by the state contained records for about 600,000 total household records, consistent with other reported caseload counts. However, we found a very large number of duplicate cases in the initial file. We feel that the analysis file represents an accurate unduplicated accounting of the food stamp recipients included in the initial data set. Prepared by Abt Associates Inc. 17 Chapter One: Introduction • If support order exists, amount of the award and whether receiving payments. For the first three items, we computed the distribution of food stamp allotments as well as households. We also examined the demographic characteristics of household heads in CSE-eligible food stamp-only custodial households. The structure and contents of these data files differ greatly from state to state. Some states maintain their data at the household level; other states maintain data at the individual recipient level (requiring aggregation by case number to construct information by household). The task of identifying CSE participants was made easier in Florida by a data element on the food stamp record that indicated a corresponding CSE case. Conversely, in Alabama and New Jersey there was a data element on the CSE file that indicated a corresponding food stamp case. In these two states, because such data elements are not always updated, we also classified a food stamp household as a CSE participant if the Social Security number for any member of the food stamp household matched with the Social Security number of a child or custodial parent in a CSE case. In Oklahoma and Texas, where no data element existed on either the food stamp or CSE files to indicate participation in the other program, we conducted a match between the two files using Social Security numbers (as indicated above for Alabama and New Jersey) as the sole basis for identifying CSE participants among food stamp households. Recipient Survey In each of the five participating states, two food stamp offices were chosen as sites for local data collection. The sites were as follows: • Etowah County (Gadsden) and Montgomery County (Montgomery), Alabama; • Jacksonville (Southside Service Center) and Lakeland, Florida; • Camden County (Camden) and Hudson County (Jersey City), New Jersey; • Cleveland County (Norman) and Tulsa County (Tulsa), Oklahoma; and • Garland and Lubbock (Parkway office), Texas. The selection of two local sites in each state, completed in July 1992, was based on the following criteria: Prepared by Abt Associates Inc. 18 Chapter One: Introduction • The monthly food stamp caseload should include at least 150 non-AFDC food stamp cases with children. • The CSE office serving the local area should be generally recognized as having average or above-average performance, as perhaps indicated by measures such as child support collections per dollar of administrative cost. • The locality should be one where we could expect cooperation from program directors, supervisors, and caseworkers (for both food stamps and CSE) and where orderly CSE record-keeping would facilitate abstraction of casefile information. • If possible, the local sites should employ different approaches to informing food stamp clients about the availability of CSE services, in the interest of encompassing a variety of administrative practices across the ten sites. All criteria were met, although the fourth proved difficult, as most local food stamp offices have no established method for providing CSE information and referral through the food stamp certification process. Among the five participating states, additional state-specific considerations served to limit the range of feasible choices for local sites. In Alabama, for instance, we excluded the six counties currently participating in the demonstration project entitled Avenues to Self-Sufficiency through Employment and Training Services (ASSETS). In Florida, we excluded any county with less than 80 percent conversion of its CSE cases to the state's new automated system. Within each state, we sought geographic separation in the sites, in the interest of obtaining greater variation in client socioeconomic characteristics and agency administrative practices. We also encouraged states to avoid local areas with exceptional characteristics that might limit the generalizability of findings, such as a high proportion of cases residing on Indian reservations or military bases. We also indicated that states should nominate food stamp offices whose caseload is served by a single CSE office, to facilitate CSE staff interviews and case record abstraction. Finally, where several offices were otherwise comparable, we indicated a preference for the site with the larger non-AFDC food stamp caseload or the larger population center. In drawing the survey sample in each site, we started with an automated file of households receiving food stamps in July 1992. We then removed from this file any food stamp households of the following types: Prepared by Abt Associates Inc. 19 Chnpter One: Introduction • Single-person households; • Multi-person households with no members under age 18; or • Multi-person households with members under age 18, any of whom receive AFDC. In all ten sites, the cases that remained in the sample frame were thus multi-person food stamp households with at least one member under age 18, where none of those under age 18 received AFDC. In Alabama, Florida, and Oklahoma, the sample frame included those non-AFDC food stamp households where one or more of those under age 18 received Medicaid on a medical assistance only (MAO) basis.12 We then divided the sample frame into three strata, as follows: • CSE participants—Non-AFDC food stamp households with at least one member also appearing in a child support enforcement case within the state;13 • CSE nonparticipants of type A—Non-AFDC food stamp households with no associated child support enforcement case and with either only one adult member (i.e., only one member 18 years of age or older) or adult members of only one sex; and • CSE nonparticipants of type B—Non-AFDC food stamp households with no associated child support enforcement case and with adult members of the opposite sex. We then randomly sorted the cases in each site into replicates of CSE participants (ten cases per replicate) and CSE nonparticipants (eight cases of type A and two cases of type B per replicate). We stratified the nonparticipants into types A and B to minimize the effort expended in screening out cases of type B, which were less likely (than those of type A) to have children of 12. In New Jersey and Texas, the person-level indicator of AFDC receipt was a joint indicator of AFDC or Medicaid receipt. In these two states we thus effectively excluded from the sample the non-AFDC food stamp households where one or more of those under age 18 received Medicaid on a "medical assistance only" basis. 13. We described earlier in this chapter the approach taken in each state for identifying CSE participants. For the sites in Alabama, New Jersey, and Oklahoma, the automated match of Social Security numbers used statewide child support enforcement files. In Texas, because of the size of statewide CSE files, the match was with cases in (a) the CSE office(s) serving the local food stamp site, and (b) all CSE offices that serve adjoining geographic areas. Prepared by Abt Associates Inc. 20 Chapter One: Introduction noncustodial parents and thus to qualify for the survey. In computing all survey findings, we have appropriately weighted the survey observations to account for this stratification. Replicates were released to the survey staff as necessary to complete interviews with 20 CSE participants and 20 CSE nonparticipants per site. Once a replicate was released, work was completed on all cases in the replicate. This was to ensure that the completed interviews would constitute a random sample, while minimizing the chances of overshooting the target number of completed interviews per site. We provided each site with a pre-screening list of the cases in the initial sample replicates, for the purpose of verifying the client's address and telephone number and the household's continued receipt of food stamps as of November 1992. Exhibit 1.2 RECIPIENT SURVEY SUMMARY: CASES SAMPLED, SCREENED, AND INTERVIEWED Number of Cases Cases sampled Cases prescreened Found ineligible Outside office jurisdiction Not food-stamp-only Released for interview Cases not screened (not located) Cases screened Found ineligible Found eligible Interviews not completed Interviews completed 1,655 136 433 1,086 216 442 14 414 - Source: Abt Auociates, 1993 survey of food tump-only custodial parents. As shown in Exhibit 1.2, a total of 1,655 cases entered the survey sample. During pre-screening, office staff found 136 cases to reside outside the office jurisdiction, and 433 were found to no longer receive food stamps or to have become an AFDC food stamp case. The remaining 1,086 cases were released for interview. Of these, 216 could not be located, 442 Prepared by Abt Associates Inc. 21 Chapter One: Introduction were screened out as ineligible,14 and 14 were found eligible but did not complete the interview. Interviews were completed on 414 cases. The completion rate was 78 percent (414/534), if one assumes that the extent of eligibility among the 216 unlocated cases was the same as for those screened (i.e., 428/870 or 49 percent).15 Of the 414 completed interviews, interviewers completed nearly one-half by telephone and the remainder in person. Interviewers completed 40 (or more) interviews in nine of the ten sites—all but the Jacksonville site, where they exhausted the sampling frame after completing 32 interviews. Of the 414 completed interviews, 218 were sampled as CSE participants, 179 were sampled as CSE nonparticipants of type A, and 17 were sampled as CSE nonparticipants of type B.16 The survey was conducted over a ten-week period from mid-January to late March 1993. See Appendix D for a copy of the survey instrument. The responses to each survey item are tabulated in Appendix CSE Case Record Abstracts After completing the interviews, we abstracted information from CSE case records for those 239 respondents participating in the CSE Program in the interview month, based on whether a CSE case actually existed for the respondent in the local CSE office. To identify these respondents as participants, in each site we provided a list of the survey respondents (and their food stamp case numbers) to the CSE office and asked the staff to check them against the CSE caseload. This allowed us to identify CSE participants by their "true" CSE status, versus their self-reported status (from the relevant survey items) or the status by which they were sampled (that is, their sampling stiatum). 14. Either DO longer receiving food stamps, receiving (or applying for) AFDC, without children under 18 of an absent parent, with foster children only, or with a deceased case bead. 15. The estimated number of eligible cases released for interview, 534, is computed as follows: 414 + 14 + (216) [(414 + 14) / (414 + 14 + 442)]. 16. The 414 completed interviews included 147 interviews for which Medicaid covers either the custodial parent or a child of an absent parent included in the food stamp grant. Based on the CSE case record abstracts, 90 of those respondents participated in CSE, in keeping with the eligibility requirements for Medicaid. We have included the remaining 57 in our estimates of potential response to a mandate or outreach, even though they were already subject to a CSE participation requirement. Prepared by Abt Associates Inc. 22 Chapter One: Introduction Once the participating cases were identified, field staff conducted the case record abstracts during March and April 1993. (In the two New Jersey sites, state CSE staff completed the abstracts.) The abstraction proceeded on the basis of "case record units" which normally corresponded to a sibling group (children from the same custodial-noncustodial parent pair). A respondent could have more than one case record unit—most typically, if a woman had children from several noncustodial fathers. There were multiple case record units for about one-fourth of the 239 respondents participating in CSE. The total number of case record units among the 239 respondents was 307. We did not abstract case information (and we did not regard the custodial parent as a CSE participant) in situations in which the CSE activity was confined to collection of AFDC arrears payments only17 or in which all children in the case record unit were 18 years old or older. See Appendix F for a copy of the case record abstraction instrument. Program Staff and Client Advocate Interviews To understand better the program environment in each of the study sites, we conducted interviews with state and local staff in both the Food Stamp and CSE Programs, as well as representatives of client advocacy groups. Two-person teams conducted these interviews during site visits in the summer and fall of 1992. For each state, the interviews took place at the state office buildings in the capital city and in the food stamp and CSE offices for each study site. These interviews aided in formulating hypotheses about the factors affecting CSE participation, selecting appropriate variables for the multivariate analysis of CSE participation, and interpreting the site differences in CSE participation rates. 17. The HHS Office of Child Support Enforcement defines an "AFDC arrears only" case as "one in which the children to be supported are former recipients of IV-A [AFDC] payments and in which the absent parent is now delinquent in his or her reimbursement of these payments to the government." We excluded these cases from the analysis because there is no effort under way to collect current support for the children. Any amounts collected on these cases go entirely to the federal and state governments (according to their respective cost-sharing of AFDC benefits), not to the custodial household. CSE services to these cases and any resulting collections would thus have no impact on either the income of the food stamp household or costs to the Food Stamp Program. Prepared by Abt Associates Inc. 23 Chapter One: Introduction 1.3 ORGANIZATION OF THIS REPORT The next three chapters address in sequence the three research questions posed at the outset of this chapter. Chapter Two presents our findings with respect to the needs for child support enforcement among food stamp-only custodial households not currently participating in the CSE Program. Chapter Three discusses the potential for increased CSE participation through either a mandate or improved outreach. Chapter Four examines the potential benefits and costs associated with either a mandate or outreach, as defined in Section 1.1 of this chapter. Chapter Five considers the policy implications of the findings reported. Prepared by Abt Associates Inc. 24 CHAPTER TWO NEEDS FOR CHILD SUPPORT ENFORCEMENT This chapter examines the extent to which food stamp-only cases need child support enforcement services. The findings define the size of the target population of this study: food stamp-only custodial households not participating in the CSE Program that need a support order or collection on an existing order. The target population is thus restricted to those households that might potentially benefit from either a CSE mandate or improved CSE outreach. Specifically, we estimate the percentage of all food stamp households nationally that meet a'l of the following criteria: • food stamp-only—that is, receiving food stamps but neither AFDC nor Medicaid, • custodial—that is, including at least one child of a noncustodial parent, • not participating in the CSE Program, and • lacking a support order or not receiving full payment on an existing order. Our basic finding is that nationally, the target population constitutes 2.8 percent of allfood stamp households. In other words, 2.8 percent of the national caseload constitutes food stamp-only custodial households that are not participating in the CSE Program but have a need for child support enforcement. Based on the current national monthly caseload of approximately 11 million households, the estimated 2.8 percent implies a target population nationally of 300,000 households. As illustrated in Exhibit 2.1, one can regard the national estimate of 2.8 percent as the product of the following two factors. The first is the percentage of all food stamp households that are food stamp-only custodial households. We estimate this nationally at 7 percent. The specific national estimates are 6.7 percent from CPS, 6.4 percent from SIPP, and 9.2 percent from the 1991 food stamp quality control (QC) data. (The QC data indicate substantial interstate variation in this percentage, from below 2 percent to above 20 percent of households.) The 7 percent national figure represents a defensible mid-range assumption for subsequent calculations. The second factor is the percentage of food stamp-only custodial households that are CSE nonparticipants and that have needs for CSE services, estimated at 40 percent. This second Prepared by Abt Associates Inc. 25 Chapter Two: Needs for Child Support Enforcement Exhibit 2.1 IDENTIFICATION OF THE TARGET POPULATION FOOD STAMP HOUSEHOLDS (100%) -47% AFDC/MA FOOD STAMP HOUSEHOLDS (47%) -53% FOOD STAMP-ONLY HOUSEHOLDS (53%) -13% -87% WITH AT LEAST ONE CHILD OF OF A NONCUSTODIAL PARENT (7%) WITHOUT A CHILD OF NONCUSTODIAL PARENT (46%) -46% -54% NOT PARTICIPATING IN CSE (3.2%) -88% PARTICIPATING IN CSE (3.8%) -12% NEEDING SUPPORT ORDER OR ENFORCEMENT (2.8%) WITH iW ORDER AND FULL PAYMENT (0 .4%) Target population of this study Source: Abt Associates, FY 1991 national food tump quality control sample, 1993 survey of food stamp-only custodial parents, and accompanying CSE case record abstracts (sec Exhibits 2.3, 2.5 and 2.7). Notes: Percentages in parentheses arc computed as proportions of the total national food sump caseload. Percentages preceded by - are conditional estimates, computed as proportions of the caseload segment represented by the preceding box. Prepared by Abt Associates Inc. 26 Chapter Two: Needs for Child Support Enforcement factor is itself derived from a 46 percent CSE nonparticipation rate and an 88 percent rate of CSE needs among nonparticipants, as described below. The estimated 46 percent CSE nonparticipation rate reflects the estimate from the ten-site pooled survey/abstract data (46.3 percent), and is in the range of estimates obtained from statewide administrative data in three of the five participating States (45.5 percent in Alabama, 39.8 percent in New Jersey, and 54.3 percent in Texas). As discussed in Section 2.2, we regard as biased the higher estimates obtained from CPS and SIPP and the lower estimates obtained from the other two participating States (Florida and Oklahoma). The estimated 88 percent rate of CSE needs reflects the estimate from the survey/ abstract data (87.7 percent), which we consider more accurate than the estimates obtained from either the CPS (77.1 percent) or SIPP (81.1 percent). These estimates all indicate the proportion of custodial parents either needing a support order or not receiving full payment on an existing order, among food stamp-only CSE nonparticipants. The following sections provide detail on these estimates of the target population. First we discuss our national and state-by-state estimates of the proportion of the food stamp caseload comprised of food stamp-only custodial households, as obtained from the CPS, SIPP, and food stamp QC data. Then we present our estimates of CSE nonparticipation and CSE needs, as obtained from the CPS, SIPP, statewide administrative data from the five participating states, and the pooled survey/abstract data from the ten study sites. We use multiple data sources in this analysis because no single data set provides sufficiently detailed information for a nationally representative sample. The need for national estimates of the target population arises from our need in Chapter Four to evaluate the national benefits and costs of a CSE mandate or improved outreach. In Chapter Four, we will project onto the national food stamp caseload the effects of these policy options as estimated from the survey/abstract data in the ten study sites. To do this, it is essential to know the extent of the national caseload that corresponds to the surveyed population—food stamp-only custodial households. As we describe in detail throughout this chapter, the data sources used here have differing limitations. To proceed with our later analysis of policy impacts, we must consider the reliability of the varying estimates and then choose specific empirical values as the basis for subsequent benefit-cost calculations. The choices discussed in this chapter ultimately reflect Prepared by Abt Associates Inc. 27 Chapter Two: Needs for Child Support Enforcement subjective judgments, weighing a series of factors that mitigate against any precise, definitive answers. We consider this an appropriate research strategy, given the exploratory nature of the study. Recognizing the inherent imprecision, we later explore (in Chapter Four) the sensitivity of the national benefit-cost estimates to the particular empirical values estimated here. 2.1 FOOD STAMP-ONLY CUSTODIAL HOUSEHOLDS We derived three independent national estimates of the percentage of all food stamp households that are food stamp-only custodial households: 6.7 percent from the CPS, 6.4 percent from SIPP, and 9.2 percent from QC data. We discuss below the possible sources of bias in these estimates. One should note that the difference between the CPS and SIPP estimates, shown in Exhibit 2.2, is not statistically significant. However, both the CPS and SIPP estimates are significantly lower than the QC estimate, shown in Exhibit 2.3.' Exhibit 2.2 CPS- AND SIPP-BASED ESTIMATES OF FOOD STAMP-ONLY CUSTODIAL FAMILIES CPS (1990) SIPP (1990) Number Percent Number Percent All families on food stamps Families with a custodial parent on food stamps* Food stamp-only 7,821,141 100.0 2,628,201 33.6 6,918,264 100.0 3,024,814 43.7 524,197 6.7 440,565 6.4 !_ Source: Urban institute, 1990 Current Population Survey and the 1990 Survey of Income and Program Participation. * CPS estimate include! families with custodial mothers; SIPP estimate includes families with custodial mothers or custodial fathers. In our later benefit-cost calculations, we need a single point estimate and not a range of values for the percentage of total food stamp households represented by food stamp-only custodial households. This is necessary to convert the findings of the survey/abstract data into national impacts. Ideally, one would either choose a best estimate from among the three 1. The sample sizes used to derive these estimates were as follows: 1,118 for the CPS data, 782 for the SIPP data, and 64,311 for the QC data. Prepared by Abt Associates Inc. 28 Chapter Two: Needs for Child Support Enforcement Exhibit 2.3 DISTRIBUTION OF FOOD STAMP HOUSEHOLDS BY AFDC/MEDICAID RECEIPT, CUSTODIAL STATUS, AND STATE, FISCAL YEAR 1991 State United States Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware District of Columbia Florida Georgia Guam Hawaii luaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada Number of Households 8,862,066 186,152 10,134 137,498 93,938 807,135 94,672 69,953 15,209 32,012 401,704 243,569 3,294 34,447 23,420 460,226 131,115 71,307 60,061 183,322 261,434 50,533 129,899 172,325 408,046 119,476 186,862 189,540 22,805 39,171 27,909 AFDC/MA Food Stamp Households Food Stamp-Only Households Not Custodial Custodial Total Percentage ofAll Food Stamp Households 46.6 44.2 9.2 100.0 24.7 55.2 37.5 22.3 79.6 39.6 64.7 46.4 58.4 33.2 37.3 33.5 41.0 38.7 50.5 45.9 48.0 54.4 44.3 33.8 44.7 53.5 54.2 55.7 52.9 31.4 42.7 33.7 44.7 27.2 51.7 42.2 49.8 58.5 18.7 45.4 33.6 37.4 36.6 48.7 46.3 48.5 53.3 47.7 45.2 41.0 44.8 40.9 48.2 45.9 50.2 39.1 42.0 41.7 45.2 53.0 47.3 57.3 44.8 53.6 23.6 100.0 2.6 100.0 12.8 100.0 19.2 100.0 1.7 100.0 15.0 100.0 1.7 100.0 16.2 100.0 5.0 100.0 18.1 100.0 16.4 100.0 18.0 100.0 5.7 100.0 13.6 100.0 4.3 100.0 13.1 100.0 7.2 100.0 4.7 100.0 7.5 100.0 20.3 100.0 5.1 100.0 7.4 100.0 3.9 100.0 2.6 100.0 1.9 100.0 15.6 100.0 10.0 100.0 9.0 100.0 10.5 100.0 19.2 100.0 Prepared by Abt Associates Inc. 29 Chapter Two: Needs for Child Support Enforcement Exhibit 2.3 (continued) DISTRIBUTION OF FOOD STAMP HOUSEHOLDS BY AFDC/MEDICAID RECEIPT, CUSTODIAL STATUS, AND STATE, FISCAL YEAR 1991 0—' State New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virgin Islands Virginia Washington West Virginia Wisconsin Wyoming Number of Households AFDC/MA Food Stamp Households Food Stamp-Only Households Not Custodial Custodial Total Percentage of All Food Stamp Households 20,121 176,268 58,414 775,738 200,968 15,574 498,553 116,355 102,761 456,210 41.2 60.3 42.6 47.3 36.9 51.7 49.8 31.0 39.6 44.6 33,723 57.1 116,325 37.2 18,042 44.5 238,807 43.1 757,112 31.8 39,667 51.5 20,607 48.6 4,301 12.6 171,070 31.1 157,816 57.9 104,546 38.6 100,619 74.6 11.301 52.0 54.7 4.1 100.0 34.9 4.8 100.0 44.4 13.0 100.0 49.1 3.6 100.0 51.0 12.1 100.0 44.9 3.4 100.0 44.4 5.8 100.0 54.1 14.9 100.0 52.9 7.5 100.0 51.6 3.8 100.0 39.7 3.2 100.0 44.8 18.0 100.0 43.5 12.0 100.0 46.9 10.0 100.0 49.4 18.8 100.0 42.1 6.4 100.0 50.2 1.2 100.0 46.8 40.6 100.0 53.6 15.3 100.0 38.1 4.0 100.0 52.6 8.8 100.0 23.3 2.1 100.0 40.0 8.0 100.0 Source: Abt Associates, FY 1991 national food stamp quality control sample. Prepared by Abt Associates Inc. 30 j Chapter Two: Needs for Child Support Enforcement available values after weighing their relative strengths and weaknesses, or derive an adjusted estimate that reconciles their differences. However, in the analysis that follows we find each of the three estimates to have some potential bias, with no clear basis for reconciliation. We thus have sought to identify a single value that represents a reasonable assumption for further calculations. For reasons explained below, we have chosen this value as 7 percent. Assessing the Accuracy of the National Estimates We start here by noting the features of the available estimates suggesting that the true value may lie within their range—that is, above 6 percent but below 9 percent. To reiterate, the measure of interest is the ratio of monthly food stamp-only custodial households to total monthly food stamp households. The 6.7 percent CPS estimate may represent an underestimate of this ratio, on two grounds. First, the numerator excludes custodial fathers and some custodial mothers.2 Second, in classifying respondents as food stamp recipients, the March 1990 CPS uses a reference period for food stamp receipt that is calendar year 1989, an annual interval rather than a monthly interval. The longer the reference period, the greater is the likelihood of counting in the denominator short-term food stamp recipients who tend to be intact families, not custodial families.3 Another reason to believe that the true value may lie within the range of 6 to 9 percent pertains to a source of potential upward bias in the 9.2 percent QC estimate. In particular, the QC numerator represents an overestimate to the extent of false-positive identificati jn of custodial 2. Custodial fathers represent about 6 to 7 percent of the custodial parents in food stamp-only custodial households, based on the SIPP data and uV survey/abstract data. As to the exclusion of some custodial mothers, the skip pattern of questions asked . the CPS child support supplement results in the exclusion of ever-married custodial mothers with children fathered by someone other than the current or most recent spouse. For instance, the survey would miss a woman now married to her second husband also raising a child from her first marriage. The survey/abstract data do not provide sufficient marital histories to estimate reliably the incidence of such situations. 3. The fact that the QC period is more recent historically (than CPS or SIPP) would seemingly cause little difference in the estimates, for the following reason. During the period 1989-91, despite rapid caseload growth, there was relative stability in the percentage of non-AFDC households (about 58 percent) and in the percentage of single-adult female-headed households with children (about 39 percent) in the total food stamp caseload. See U.S. Department of Agriculture, Food and Nutrition Service, "Characteristics of Food Stamp Households: Summer 1991" January 1993, Appendix Tables A-31 and A-53, and corresponding tables from previous issues of the same report. Prepared by Abt Associates Inc. 31 Chapter Two: Needs for Child Support Enforcement households. Such misidentification could arise from the limited detail in the codes that identify familial relationships among food stamp household members. In using these codes, we ran the risk of identifying the following types of households as custodial: a single-parent household in which the other parent is deceased, a household in which both parents jce present but not married, and a three-generation household headed by a child's grandparent but also including both of the child's parents.4 Next, there are other aspects of the estimates suggesting that the true value may lie below the estimated range. First, both the CPS and SIPP estimates are based on the Census-defined family or subfamily—a group of two or more persons who not only reside together but also are related by birth, marriage, or adoption. The ideal unit of observation here is the food stamp-defined household—a group of persons (or a single individual) who occupy the same residence and who purchase and prepare food together. The family unit of observation would tend to bias upward the CPS and SIPP estimates, by excluding from the denominator food stamp recipients who reside alone and who by definition could not represent custodial households. Second, both the CPS and SIPP estimates classify a family as custodial according to whether it includes a child underage 21 of a noncustodial parent, consistent with Census definitions. This tends to overstate the intended numerator, which applies an age threshold of 18 years consistent with the administrative definitions in the CSE mandates for AFDC and Medicaid. Third, CPS and (to a lesser extent) SIPP each tends to under-report means-tested benefits such as AFDC, food stamps, and Medicaid. This by itself might bias upward the estimated percentage, if the undercount of recipients is more pronounced in the denominator (among multiple-benefit 4. The person-level identifying codes in the QC data relate each member of the household to the household head but not to other household members. We classified a household as a custodial case if it met the following conditions: • there were two or more persons in the food stamp household; and • either (a) the case included one or more stepchildren of the household head or (b) the case head had no spouse present and had one or more children (or grandchildren). Prepared by Abt Associates Inc. 32 Chapter Two: Needs for Child Support Enforcement recipients such as AFDC food stamp families) than in the numerator (among single-benefit recipients such as food stamp-only families).5 Finally, but probably of lesser importance, there is one aspect of the QC estimate suggesting that the true value may even lie above the estimated range. Recall that the false-positive identification of custodial households may cause an overestimate in the QC numerator. To the extent that false-negative situations could occur even more frequently, however, the QC numerator may represent an underestimate. For instance, our use of the relationship codes in the QC data would have caused us to falsely identify a household as not custodial in one particular situation: if the female head resides with her current husband and a child fathered by an earlier husband or other partner. In this situation, the child would have been coded as the son or daughter of the household head; we would have incorrectly regarded the woman's current spouse as the child's father.6 The issue of whether the QC estimate represents on overestimate thus rests on the question of whether the false positives outnumber the false negatives. To address this question, it is instructive to compare the QC estimates obtained for the five participating states with the corresponding estimates derived from the state administrative data. As a means of testing the validity of the QC estimate, the state administrative files offer several advantages over the QC data. First, because the state data files include the entire statewide food stamp caseload, there is no sampling variability in the estimated percentages. Second, the state files allow us to unambiguously identify as custodial households those food stamp cases with corresponding CSE cases. For four of the five states (all but Florida), the administrative data yield an estimate that lies outside the 95 percent confidence interval of the QC-based estimate (see Exhibit 2.4). For several reasons, one would expect some discrepancies in these estimates for any given state. 5. Consider the following example using AFDC and food stamps. Assume that the total food stamp caseload consists of X non-AFDC food stamp cases and Y AFDC food stamp cases. The true portion of non-AFDC cases is thus X/(X+Y). Assume that there is complete reporting of both food stamps and AFDC in the QC data. For the Census data (CPS or SIPP), assume that underreporting reduces the observed number of non- AFDC food stamp cases from X to aX and reduces the number of AFDC food stamp cases from Y to abY, where a and b (both less than 1) are the reporting rates for food stamps and AFDC, respectively. The observed percentage of non-AFDC cases in the food stamp caseload will equal aX/(aX+abY) in the Census data, which equals X/(X+bY), and thus is greater than the value X/(X +Y) derived from the QC data. 6. If instead the male spouse had been designated as the household head, with the child thus coded as a stepchild, we would have correctly classified the case as a custodial household. Prepared by Abt Associates he. 33 Chapter Two: Needs for Child Support Enforcement Exhibit 2.4 FOOD STAMP-ONLY CUSTODIAL HOUSEHOLDS AS A PERCENTAGE OF ALL FOOD STAMP HOUSEHOLDS IN THE FTVE PARTICIPATING STATES Food Stamp-Only Custodial Households QC Data Administrative Data State (FY 1991) (July 1992) Percentage ofAll Food Stamp Households 1 Alabama 23.6' 28.5 1 Florida 18.1 16.4 | New Jersey 4.8' 8.9 | Oklahoma 14.9s 17.3 1 Texas 18.8' 10.2 Source: Abt Associate!, FY 1991 national food stamp quality control sample and 1992 state administrative data for food stamps and CSE. 1 Significantly different at the .05 level from the estimate derived using the adminis-trative data. First, as described earlier, the limited demographic information in both the QC files and the administrative data required the use of approximating criteria for identifying custodial households. To recall, in the QC data a food stamp-only case was identified as a custodial household if two conditions were met: • Two or more persons in the food stamp household; and • Either (a) one or more stepchildren under the age of 18, or (b) the case head has no spouse present with one or more children (or grandchildren) under the age of 18. In the state administrative data, a food stamp-only case was classified as a custodial household if it met either of the following criteria: • CSE participant; or • A multiperson household with at least one member under age 18 and no opposite-sex members over age 18. One would expect these differing criteria to lead to differing estimates of the food stamp-only custodial population. On the one hand, the state administrative data would appear more reliable, Prepared by Abt Associates Inc. 34 Chapter Two: Needs for Child Support Enforcement as we can unambiguously identify a food stamp-only custodial household in those instances where a CSE case exists. On the other hand, among food stamp-only cases where no CSE case exists, the more detailed .elationship codes for each household member in the QC data enable a more accurate identification of custodial households than is possible with the state administra-tive files. Note that Texas, by far the largest of the five states, is the one for which the QC estimate significantly exceeds the corresponding administrative estimate for the percentage of food stamp-only custodial households. For the five states combined, the Texas overestimate (along with the much smaller and insignificant Florida overestimate) more than offsets the underestimates in the other three states. The five-state QC estimate (17.4 percent) thus exceeds the corresponding administrative estimate (14.2 percent). If we proportionally adjust the 9.2 percent national QC estimate to account for the extent of net upward bias apparent from the comparison with the administrative data, we obtain an adjusted national QC estimate of 7.5 percent. In summary, we have raised here a variety of issues pertaining to the available national estimates of the percentage of food stamp households that consist of food stamp-only custodial households. The weight of evidence implies a true value that lies below the QC estimate (9.2 percent) and may even lie below the values derived from CPS (6.7 percent) and SIPP (6.4 percent). We interpret the issues raised here as indicating that none of the three available estimates is a compelling choice. For the CPS estimate, there are potential sources of both upward and downward bias with uncertain net direction. In contrast, the SIPP estimate appears to overstate the true value for the same reasons as CPS, with no apparent offsetting sources of downward bias. This would seemingly call into question the even higher QC estimate. We have used the state administrative data as a means of validating the QC estimate and have derived an adjusted QC value of 7.5 percent. Choosing a Single Point Estimate In the end, we have adopted 7 percent as the national point estimate for subsequent calculations. We simply consider it more likely that the correct value lies toward the lower end of the range of the three estimates than either toward the upper end of the range, above the range, or below the range. If the true value lies outside the range, however, we consider it Prepared by Abt Associates Inc. 35 Chapter Two: Needs for Child Suppon Enforcement more likely to lie below the range than above the range. In Chapter Four we therefore check the sensitivity of the benefit-cost estimates to the 7 percent assumption by using an alternative 6 percent assumption. As a final observation here, note that there is substantial variation among states in the percentage of active food stamp cases that are food stamp-only custodial households, as estimated from the ^C. data for all states and from the administrative data for the five states participating in this study. The variation found in the QC data is especially noteworthy, since (compared to the state administrative estimates) the QC estimates reflect a greater degree of consistency in measurement definitions. Based on the QC data by state, as shown in Exhibit 2.3, the portion of total food stamp cases that are food stamp-only custodial households ranges from less than 2 percent in California, Connecticut, Minnesota, and Vermont to more than 20 percent in Alabama and Louisiana. (The Virgin Islands estimate, an extreme outlier at 40 percent, is based on a sample of less than 200 cases.) As one expects, these percentages are inversely related to AFDC benefit levels. The higher the AFDC benefit for a family of given size, the higher the income level at which the household no longer qualifies for AFDC and consequently the narrower the range of income in which the non-AFDC family would still remain income-eligible for food stamps. 2.2 CSE NONPARTICIPATION AND CSE NEEDS The next task is to estimate the percentage of food stamp-only custodial households that do not participate in the CSE Program but might benefit from CSE participation (i.e., which either lack support orders or receive less than full payment on existing orders). We have derived estimates from CPS, SIPP, the survey/abstract data, and state administrative data (Exhibit 2.5). The QC data indicate nothing about participation in (or contact with) the CSE Program or about the status of support orders or support payments. (Child support income is lumped together with "other unearned income.")7 7. It would be difficult to use "other unearned income" as a proxy measure for child support paymenu, for two reasons. First, this category of reported income also includes (to an unknown degree) the following sources: alimony, foster care payments, rental income, private pension benefits, union benefits, dividends, and interest payments. Second, the QC file reflects some pre-editing of this data element to reconcile household gross income with the summation of person-level income amounts. Prepared by Abt Associates Inc. 36 Chapter Two: Needs for Child Support Enforcement Exhibit 2.5 PERCENTAGE OF FOOD STAMP-ONLY CUSTODIAL PARENTS AND HOUSEHOLDS REPORTING NO PREVIOUS CSE CONTACT OR NO CURRENT CSE CASE, BY DATA SOURCE No Previous CSE Contact No Current CSE Case Percentage of Food Stamp-Only Custodial Parents I CPS (1990) SIPP (1990) 68.8 - 74.9 - Survey/abstract data (1993) 59.4 46.3 Percentage of Food Stamp-Only Custodial Households 1 State administrative data (1992) Alabama - 45.5 Florida - 13.2 New Jersey - 39.8 Oklahoma 15.9 Texas 54.3 Source: Urban Institute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation; Abt Associates, 1993 survey of food stamp-only custodial parents, accompanying CSE case record abstracts, and 1992 state administrative data for food stamps and CSE. CSE Nonparticipation As explained below, our estimate of CSE nonparticipation among food stamp-only custodial households is based on the survey/abstract data and is generally consistent with the state administrative data. For CPS and SIPP, the survey items on the custodial parent's involvement with the CSE Program appear to have limited usefulness for the purposes of this study. The items explicitly address prior client-initiated contact with the program, versus current participation. In particular, the April 1990 CPS Child Support Supplement asked the following question: "Have you ever contacted any government agency for aid in obtaining child support (for child(ren) of last divorce or separation)?" Correspondingly, the child support topical module administered in Wave 3 to the 1990 SIPP included the following item: Prepared by Abt Associates Inc. 37 Chapter Two: Needs for Child Support Enforcement "For any of [the noncustodial parent]'s children, has [the custodial parent] ever asked a public agency (such as the child support enforcement office or welfare agency) for help in obtaining child support?" For food stamp-only custodial families, the percentage indicating no prior CSE contact was similar in the two surveys: 68.8 percent in CPS and 74.9 percent in SIPP. For several reasons, these responses appear to overstate the level of nonparticipation in the CSE Program (i.e., understate CSE participation). First, because the question implies voluntary contact by the respondent, its accuracy in identifying all contact between the respondent and the CSE agency is questionable. The percentage of the custodial parents in AFDC/MA food stamp households who report CSE contact is implausibly low—only 38.6 percent in CPS and 47.1 percent in SIPP—despite the requirement for CSE participation among such households. Second, the survey/abstract data indicate that many custodial parents who report no previous voluntary contact with the CSE agency in fact do currently participate in CSE. The survey asked the following question, deliberately worded to resemble the CPS and SIPP items: "Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtaining child support for any of your children . . . ?" As shown in Exhibit 2.6, among those who answered negatively to this question, almost 40 percent have a case in the local CSE office. This is understandable because many food stamp-only cases are former AFDC recipients whose cases would have been continued automatically unless they requested that their case be closed. All contact between CSE and the recipient could have been agency-initiated rather than recipient-initiated. (Also understandable is the fact that, among those who answered "yes" to this question, 26 percent have no current CSE case. Such clients may have had contact with CSE but did not apply for services or may have had a CSE case that was subsequently closed.) We have used the survey/abstract findings to adjust the CPS and SIPP findings with respect to reported no previous CSE contact (i.e., the 68.8 percent CPS estimate and the 74.9 SIPP estimate cited earlier). If one assumes the same pattern found among survey respondents between reported CSE noncontact and actual CSE nonparticipation, the Prepared by Abt Associates Inc. 38 Chapter Two: Needs for Child Support Enforcement implied CSE nonparticipation rate is 49.6 percent for CPS and 51.7 percent for SIPP.8 Both of these adjusted estimates correspond generally with the 46.3 percent CSE nonparticipation rate determined from the survey/abstract data. Exhibit 2.6 REPORTED CSE CONTACT VERSUS PRESENCE OF CSE CASE Does a CSE Case Presently Exist for This Food Stamp-Only Custodial Parent? Question Response Yes No Total Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtain-ing child support...? Have you ever been in contact with a government agency responsible for the child support enforcement program for help in obtaining child support...? Yes No Total Yes No Total Number of Respondents 135 40 175 104 239 135 175 239 414 Weighted Percentage* 74.4 2^ 5 100.0 39.5 53.7 60.5 46." 100.0 100.0 Source: Abt Associates, 1993 survey of food s;imp-9only custodial parents and accompoanying CSE case record abstracts. * Row percentages sum to 100. Based on the food stamp and CSE administrative data obtained for this study, as shown in Exhibit 2.5, three of the five states have CSE nonparticipation rates that are in the same range as the 46 percent survey/abstract estimate: 45.5 percent for Alabama, 39.8 percent for New Jersey, and 54.3 percent for Texas. Nonparticipation estimates were much lower for the other 8. For CPS and SIPP, we compute the adjusted CSE nonparticipation rate as the sum of: (a) the percentage reporting no previous CSE contact multiplied by the (survey/abstract-based) conditional CSE nonparticipation rate for those reporting no prior CSE contact (.605), and (b) the percentage reporting previous CSE contact multiplied by the conditional CSE nonparticipation rate for those reporting previous CSE contact (.256). The CPS calculation is (68.8 x .605) + (31.2 x .256) - 49.6. The SIPP calculation is (74.9 x .605) + (25.1 x .256) - 51.7. Prepared by Abt Associates Inc. 39 Chapter Two: Seeds for Child Support Enforcement two states: 13.2 percent for Florida and 15.9 percent for Oklahoma. The latter estimates appear to reflect the fact that the computer systems in these states automatically generate a CSE case number when there is a child of a noncustodial parent in a food stamp household. However, these cases do not truly exist in the CSE Program if the client has filed no application and the CSE agency has provided no services. We encountered this situation for some survey respondents who were sampled as CSE participants (based on the automated files) but were found (at the time of abstraction) not to have any CSE case on file in the local CSE office. CSE Needs The next issue to address is the extent of need for child support enforcement services among those food stamp-only custodial parents who do not participate in CSE. At the extreme, these CSE nonparticipants would pose little concern if they currently receive full payment on privately-arranged support orders. To the contrary, however, more than three-fourths of the food stamp-only custodial parents not participating in CSE either have no support order or receive less than full payment on their order—and thus, by our definition, have needs for CSE services. The evidence from CPS, SIPP, and the survey/abstract data is generally consistent on this point. As shown in Exhibit 2.7, we find the percentage with no order or with less than full payment on an order is 77.1 percent from CPS, 81.1 percent from SIPP, and 87.7 percent from the survey/abstract data.9 Predominantly, such cases lack a support order, versus having an order on which they receive no payment or partial payment. The higher extent of CSE needs indicated in the survey/abstract data (versus either CPS or SIPP) reflects in part a difference in classifying a custodial parent who receives full payment on all support orders obtained, but who has one or more children not covered by an order. Because both CPS and SIPP identify such a parent as having an order and receiving full payment, the case is tabulated here as having no needs for CSE services. However, in the 9. We have based the CPS and SIPP estimates on food stamp-only custodial parents reporting no contact with the CSE agency. From the discussion above, we recognize that the self-reported information from these respondents may not accurately indicate their current CSE participation or nonparticipation. We implicitly assume here that the needs for child support enforcement services among these custodial parents reasonably approximates the profile of needs among those who truly do not participate in CSE. We have tested this assumption with the survey data by comparing the profile of needs among those who report no CSE contact with the profile among those who (based on the case record abstraction) are not CSE participants. The two profiles are very similar. Prepared by Abt Associates Inc. 40 Chapter Two: Needs for Child Support Enforcement Exhibit 2.7 INITIAL CHILD SUPPORT STATUS OF FOOD STAMP-ONLY CUSTODIAL PARENTS NOT PARTICIPATING IN THE CSE PROGRAM CPS" (1990) SIPI* (1990) Recipient Survey' (1993) Percentage of Custodial Parents n Have no order 57.9 65.1 65.1 Have an order, receive no payment 12.7 9.8 13.4 Have an order, receive partial payment £5 fij 2J Subtotal 77.1 81.1 87.7 I Have an order, receive full payment 219 18.9 m 1 Total 100.0 100.0 100.0 Source: Urban Institute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation; Abt Associates, 1993 survey of food stamp-only custodial parents and accompany-ing CSE case record abstracts. . * For custodial mothers reporting no previous CSE contact. b For custodial mothers and fathers reporting no previous CSE contact. c For custodial mothers and fathers with no current CSE case. Those respondents with no support orders for all children of a noncustodial parent were classified above as having no order. Those respondents with a support order were asked, for each order, "In the laat 12 months, how regularly have you received payments under this (agreement/order)?" Those answering "never" for all orders were classified above as receiving no payments. Those answering "seldom" or "occasionally" on at least one order, or answering "regularly but late" or "regularly and on time* for all orders but with at least one child of a noncustodial parent not covered by an order, were classified as receiving partial payments. Those with orders for all children of a noncustodial parent and answering "regularly but late" or "regularly and on time" were classified as receiving full payments. survey/abstract data we regard this parent as still having CSE needs; we therefore tabulate the case along with those having an order but receiving partial payment. As shown in Exhibit 2.7, the latter group with partial needs represents 9.3 percent of custodial parents in the survey/ abstract data, higher than the 6.5 and 6.2 percent estimated from CPS and SIPP, respect-ively. 10 10. We have not attempted to reconcile further the estimates from the survey/abstract data with the estimates from either CPS or SIPP, as the survey/abstract data pertain to six sites that were not chosen to be nationally representative. Prepared by Abt Associates Inc. 41 Chapter Two: Needs for Child Support Enforcement The findings reported in this section can be summarized as follows. Among custodial parents in food stamp-only households, an estimated 46 percent do not participate in the CSE Program. Among these CSE nonparticipants, approximately 88 percent need child support enforcement services—that is, they either lack a support order or receive less than full payment on an existing order. Combining these percentages, we conclude that about 40 percent of food stamp-only custodial households could potentially benefit from measures to require or encourage CSE participation. Because food stamp-only custodial households represent an estimated 7 percent of all food stamp households, these estimates imply that the target population for either a CSE mandate or improved outreach is approximately 2.8 percent of the total food stamp caseload. The specific national estimates obtained from CPS and SIPP are 3.5 percent and 3.9 percent, respectively. For completeness, we show these estimates in Exhibit 2.8, even though we regard them as over-estimating the extent of CSE nonparticipation." Among all states, the size of the target population appears to vary substantially as a percentage of total food stamp cases, in the range of 1 to 8 percent. 11. As described earlier, the extent of "no previous CSE contact" reported either in CPS or SIPP (68.8 percent and 74.9 percent, respectively) implies a substantially lower CSE nonparticipation rate (49.6 percent and SI .7 percent, respectively) when one accounts for the observed pattern of responses in the recipient survey and case record abstracts. Prepared by Abt Associates Inc. 42 Chapter Two: Needs for Child Support Enforcement Exhibit 2.8 CPS AND SIPP ESTIMATES OF THE TARGET POPULATION r CPS (1990) SIPP (1990) Number" Percentage1* Number" Percentage All food stamp families 7,821 Food stamp-only custodial families 524 that have contacted CSE 164 that have not contacted CSE 360 have an order, receive full pay-ment 83 have no order 209 have an order, receive no payment 46 have an order, receive partial payment 23 Subtotal 278 100.0 6.7 2.1 4.6 1.1 6,918 441 111 330 62 100.0 6.4 1.6 4.8 0.9 Target Population 2.7 215 3.1 0.5 32 0.5 0.3 21 0.3 3.5 268 3.9 Source: Urban Inititute, 1990 Current Population Survey and 1990 Survey of Income and Program Participation. * All numbers are in thouiands of families. b All percentages are of all food stamp families nationally. Prepared by Abt Associates Inc. 43 Prepared by Abt Associates Inc. 44 CHAPTER THREE POTENTIAL FOR INCREASED CSE PARTICIPATION This chapter examines the extent to which households in the target population—food stamp iy custodial households not participating in CSE but with needs for CSE services-might enter the CSE Program in response to either a mandate or improved outreach. As indicated in Chapter One, the empirical task here is to decompose the target population of custodial parents into the following three groups: • those unlikely to become CSE participants with either a mandate or outreach; • those likely to become CSE participants with a mandate but not with outreach; and • those likely to become CSE participants with outreach (and also, therefore, with a mandate, since outreach would be much less compelling of participation than a mandate). The larger the first group, the weaker the case for either a mandate or outreach. The larger the second group, the stronger the case for a mandate. The larger the third group, the stronger the case for outreach, assuming that a mandate imposes higher costs (to agencies and clients) than outreach. As indicated below, the division between the second and third group is somewhat arbitrary, requiring interpretation of the data in identifying those clients whose participation decision might conceivably be affected by caseworker efforts to provide CSE information or make CSE referrals. Our basic findings, derived from the responses of custodial parents to questions on the recipient survey and from information collected subsequently through case record abstracts, are as follows: • An estimated 24 percent ofthe target population is unlikely to respond to either a mandate or outreach. These custodial parents are subdivided as follows—9 percent who say that they would leave the Food Stamp Program altogether, 6 percent who say that they would seek a good cause exemption, 5 percent who say that they would accept a smaller food stamp allotment as a sanction for noncooper-ation, and 4 percent who would "do something else." Prepared by Abt Associates Inc. 45 Chapter Three: Potentialfor Increased CSE Participation • An estimated 39 to 60 percent of the target population is likely to respond to a mandate, but not to outreach. These are custodial parents who indicate that they would comply with a mandate. Their current nonparticipation appears to reflect a deliberate informed choice, and they thus seem unlikely to respond to any outreach effort. • An estimated 16 to 37 percent of the target population is likely to respond to outreach, and thus, presumably, also to a mandate. Unlike the previous group, these custodial parents indicate little or no knowledge of the CSE Program, or they appear undecided about CSE participation and thus might respond to information, referral, or application assistance. The custodial parents in the second group, those responding to a mandate but not outreach, thus represent at least a plurality—and perhaps a majority—of the target population. To examine multivariate effects on CSE participation, we have also estimated a series of regression equations. The included variables in these equations explain 41 percent of the variation in CSE participation among cases never receiving AFDC previously and 26 percent of the variation among cases previously on AFDC. As one might expect, the pattern of effects is quite different between the never-on-AFDC and ever-on-AFDC groups. For the never-on-AFDC cases, CSE participation implies client initiative to enter the program. In contrast, for the ever-on- AFDC group, CSE participation is typically a default outcome; the client initially cooperated with CSE to receive AFDC benefits and subsequently (after losing AFDC benefits) took no action to leave the CSE Program. Importantly, the multivariate analysis of never-on-AFDC casesfound significantly higher CSEparticipation in the two Florida sites, where caseworkers routinely provide CSE information and referral at the time of initialfood stamp certification. These latter findings, which take account of differences among sites in case demographic characteristics, suggest that CSE participation may be influenced by administrative practices that link more closely the Food Stamp and CSE Programs. Other possible explanations—e.g., that differences in CSE effectiveness tend to attract or discourage CSE participation—are difficult to test, given the small number of participating sites. The first section of this chapter provides details on the key findings mentioned above with respect to the possible responses of custodial parents in the target population to either a Prepared by Abt Associates Inc. 46 Chapter Three: Potentialfor Increased CSE Participation mandate or outreach.1 The second section of the chapter then examines the descriptive characteristics of the target population pooled across the ten study sites, and presents our multivariate analysis of CSE participation among food stamp-only custodial parents. 3.1 POTENTIAL RESPONSE TO A MANDATE OR OUTREACH We examine here the extent to which the target population of custodial parents (food stamp-only, not participating in CSE, but with needs for CSE services) consists of those who (a) would respond to neither a mandate nor outreach, (b) would respond to a mandate but not outreach, or (c) might respond to outreach and thus also presumably to a mandate. As explained below, the easier issue is to identify the first subgroup; the more difficult question, requiring some interpretation of the survey evidence, is to distinguish between the second and third groups. Reasons for Nonparticipation In the recipient survey, we asked custodial parents about their contact with the CSE agency. For those having heard about the CSE agency but not currently participating in CSE, we asked them to indicate their main reason for not using the CSE Program. Here we review the pattern of responses for custodial parents in our target population—food stamp-only custodial parents not participating in CSE (based on our case record abstracts) but with child support needs (without support orders or receiving less than full payment on their orders). Exhib |