School of Public and Environmental Affairs
Maureen A. Pirog
School of Public and Environmental Affairs
Indiana University, and
Daniel J. Evans School of Public Affairs
University of Washington
School of Social Work
University of Michigan
In this review, we focus on current research on the major welfare program in the United States, food security programs, Social Security, Social Security Disability, Unemployment Insurance, child support, and tax provisions such as the Earned Income Tax Credit (EITC) that provide substantial financial support for low-income households and other potentially vulnerable populations such as the elderly and the unemployed. Since many of these are programs specifically targeted at poor and low-income individuals, we also describe how poverty is defined in the United States, update readers on the ongoing debate over poverty measurement, and provide some comparison to how it is measured outside the United States. Looking across the various social policies addressed in this review and the associated recent research, one clear theme emerges: the United States is very concerned about work disincentives potentially embedded within these programs.
Social policy is a broad term that encompasses a wide range of policy interests including means-tested and social insurance programs that offer cash and/or in-kind assistance to individuals and families for the purposes of providing a basic income, maintaining and improving physical and mental health, food and housing security, and mitigating the adverse consequences of domestic violence, other crimes or disasters. Social policy can also include the panoply of education programs from preschool to college tuition assistance to adult literacy, vocational training and other learning programs. In this review, we focus on current research on the major welfare program in the United States, food security programs, social security, social security disability, unemployment insurance, child support, and tax provisions such as the Earned Income Tax Credit (EITC) that provide substantial financial support for low-income households and other potentially vulnerable populations such as the elderly and the unemployed. We recognize that this does not encompass all possible programs, but other review articles in the Policy Studies Journal are tackling areas we are omitting (e.g., health policy and education).
Many books have been written on various aspects of social policy. Even with our narrowed focus on the above-listed programs, social policy research can encompass public opinions about social programs and trends (Winter, 2008); the politics of race, gender, class and sexuality as they pertain to social programs (Reingold and McKee, 2012; Fox, 2012) ; policy feedbacks (Campbell, 2012); social activism (Ernst, 2010; Strolovitch, 2007) as well as policy analytic research on what is or is not working in social programs. Our emphasis is on what we have learned over the past several years about the effectiveness of the social programs which we are reviewing. This further narrowing of the scope of the review is intended to (1) limit the length of the review and (2) provide cohesion in our review given that the bulk of the material published on social programs in the past few years has had this emphasis.
Because many, certainly not all, social programs are targeted to low-income households and are means-tested, we begin by describing how poverty is defined in the United States, the most commonly used benchmark to determine eligibility. We update readers on the ongoing debate over poverty measurement and provide some comparison to how it is measured outside the United States. We then describe the latest research on means-tested programs, followed by the three major social insurance programs: social security (SS), social security disability insurance (SSDI), and unemployment insurance (UI). We include a description of the research on the child support program next, in part because it is a major income maintenance program for children with divorced, separated or never married parents irrespective of whether they are poor, rich or in-between. We conclude with recent research on provisions of the tax code that have important social policy goals. In this section, we discuss the EITC as well as recent proposals related to a negative income tax and a unified child credit.
Understanding whom is poor, and thus in need of assistance through various social policies, is a task that is challenging on both methodological and ideological grounds. In general, countries employ one of two techniques to measure poverty among their populace: a relative measure or an absolute measure, both of which are on income. European nations have tended to use relative measures, which define poverty relative to the standard of living within a particular country (for example, some fraction of median household income). The United States uses an absolute measure that sets a poverty line at a fixed amount of income. In both scenarios, individuals with income below either the fraction of median income or the fixed line are considered poor (Couch and Pirog, 2010).
However, deciding where to set the poverty line and what to count as income is challenging. In the U.S., the method for measuring poverty was developed in the 1960s. In short, the measure is based upon the amount of money needed to afford a low cost diet in 1955, multiplied three (since at that time, families spent approximately one-third of their income on food), and then adjusted based upon the number of related individuals in the household and the ages of those individuals (e.g., the elderly are assumed to need less food than younger adults and children). Each year, the threshold is adjusted to take into account increases in the cost of living. Individuals and families whose pre-tax cash income falls below the poverty threshold are considered "poor."
This method has come under criticism, both for its construction but also for what is counted as income for purposes of calculating who is poor. Families no longer spend one-third of their income on food (they spend less), and certain types of federal assistance, including benefits received through the various nutrition programs and the Earned Income Tax Credit (see below), among others, are not counted, even though they increase poor families' purchasing power and may in fact reduce poverty. Additionally, the current measurement scheme does not take into consideration variations in the cost of living across and within states or out-of-pocket expenses families may incur (Couch and Pirog, 2010; Hutto et al., 2011). In recent years, the U.S. Census Bureau, which tracks the poverty rate, has implemented an "alternative poverty measure" that attempts to correct for some of these inadequacies.
A recent article attempts to put into practice recommendations made by a National Academy of Sciences panel (Citro and Michael, 1995) and others (e.g., Blank and Greenberg, 2008) to address some of these shortcomings in poverty measurement. Hutto et al. (2011) find that using a number of alternate methods increases poverty in the U.S. in 2007 by 3.6 percentage points for all individuals and by 3.2 percentage points for children. However, after adjusting for the receipt of near-cash benefits (but not for out-of-pocket expenses), child poverty and poverty among African Americans goes down, suggesting that various in-kind anti-poverty programs do lead to reductions in poverty. Two other papers (Levitan et al., 2010; Zedlewski, Gianarelli, and Wheaton, 2010) attempt to implement the NAS recommendations at a state and local levels in high cost-of-living areas: Connecticut and New York City. Both analyses find that although poverty thresholds (i.e., the amount of income below which a family would be considered poor) would be higher under alternative constructs, the inclusion of income from tax refunds and in-kind benefits, at least for certain groups, offsets increases in poverty that might otherwise be seen.
Beyond income based measures of poverty, other types of measures have been considered, including those that: a) take assets into account; b) measure consumption patterns; and c) examine a broader range of well-being, such as experiences of material hardship and social inclusion. Brandolini and colleagues (2010) examine how the inclusion of wealth changes the level and composition of poverty in the U.S. and several European nations. They find that for all countries, including measures of net worth considerably reduces poverty. Yet, the authors caution that these findings are sensitive to the inclusion of housing (which may not easily be liquidated during periods of low income) as well as to data limitations. Recent scholarship (Marlier and Atkinson, 2010; Nolan and Whelan, 2010) has also drawn attention to social exclusion, which Marlier and Atkinson (2010) define as "the involuntary exclusion of individuals and groups from political, economic, and social processes, preventing their full participation in the society in which they live" (p. 285). Although the concept is not often used in discussion of poverty in the United States, it is gaining traction in Europe.
In describing the major programs that comprise U.S. social policy, we make the distinction between means-tested programs and those that are considered social insurance. Means-tested programs are those that require recipients to meet certain income or other resource-related requirements. Although there is variation across states, the two programs we consider here- Temporary Assistance to Needy Families (TANF) and the Supplemental Nutrition Assistance Program (SNAP) - require that recipients have very low income and little in the way of assets. Social insurance programs also have certain requirements that must be met to receive benefits, but the primary purpose of these programs is not to assist low-income individuals but rather to provide protection against the ups and downs of the labor market and to individuals who are unable to participate fully in the labor market (i.e., the elderly and the disabled).
Temporary Assistance to Needy Families (TANF). TANF is a program providing cash benefits to very poor families with children as well as services needed to help adult TANF recipients become and stay employed. Passed as part of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA), TANF replaced Aid to Families with Dependent Children (AFDC). Whereas AFDC was an entitlement program, meaning that those who met eligibility requirements received benefits, TANF is funded via a block grant to states. The level of funding is fixed, regardless of changes in demand or need for the program. The block grant funding mechanism also gives states more discretion in determining how to structure eligibility and fund various aspects of TANF.
Along with funding changes, PRWORA also contains a number of provisions that make TANF very different AFDC. First, adult TANF recipients may not receive federal benefits for more than 60 months in their lifetime, less at state option. Recipients must also be working or participating in work related activities. AFDC recipients could continue to receive benefits as long as they had a minor aged child and met other eligibility, and work requirements in the AFDC program were imposed on a very small fraction of the caseload (Weaver, 2000). These provisions were predicated upon a belief that recipients needed discipline imposed upon them, lest they linger on welfare rolls instead of seeking employment (e.g., Mead, 1986). Whether due to PRWORA's changes, other policies changes that were implemented at the same time (e.g., expansions to the EITC program, discussed below), the strong economy in which PRWORA was implemented, or other factors, the number of families receiving TANF benefits plummeted and has not increased much despite the recent economic downturn (Danziger, 2010).
Two recent works take stock of what has been learned since TANF's inception. Danziger (2010) performs a comprehensive review of social science research conducted since 1996, finding that the TANF program plays a very small role in poor families' lives. Historically low proportions of eligible families receive TANF benefits, and those who do receive it do so for very short periods of time. While many former TANF recipients left the program for employment, the jobs they hold tend to be very low-paying and unstable. Finally, a large body of work finds that a significant minority of current and former TANF recipients have serious health and mental health problems and/or experience serious domestic violence. These challenges may both impede women's ability to hold jobs as well as to interact with a TANF system that has complex rules.
Both Hands Tied (2010) by Collins and Mayer provides an in-depth exploration of these and other issues faced by former TANF recipients in Wisconsin. The authors argue that TANF's implementation must be considered alongside changes in the global labor market (i.e., the proliferation of service sector jobs) as well as an increased belief that the market should be free of regulation, which has served to dismantle many of the protections previously available to workers. The jobs recipients obtained, as Danziger's review highlights, were low-paying service sector positions without benefits and without the flexibility needed to care for their children. Yet, when jobs were lost, women were faced with a rather punitive welfare system whose main goal was to funnel them back into the low-wage labor market, rather than addressing the underlying problems that brought them back to TANF.
The organization of work and responsibilities within local TANF offices, and not just federal policy changes or recipient challenges, plays an important role in poor families' ability to access benefits, since these offices serve as a "gateway" to the TANF program (Brodkin and Majmundar, 2010). Several papers highlight various aspects of local TANF administration and its effects on potential recipients. Brodkin and Majmundar (2010) focus on the informal practices that offices may put in place to manage TANF caseloads, largely by making access to benefits difficult:
Caseworkers exercise what we term procedural discretion when they demand face-to-face meetings beyond those required by regulation, set appointment times without regard to claimant circumstances (such as pickup schedules for school children), or schedule multiple claimants simultaneously, producing long waiting times at welfare offices….when caseworkers misunderstand and/or misapply rules in disadvantageous way, lose documents and require resubmission, or do not return telephone calls. (p. 831).
While such practices occurred under AFDC, the increased discretion afforded by PRWORA, as well as more stringent rules mandated by federal law, may have served to increase such "procedural discretion."
Fording and colleagues (2011) examine how the frontline use of specific TANF policy tool- sanctions- intersects with race. PRWORA specified that states must develop and implement procedures to deal with recipients found to be in violation with TANF rules yet left it up to states to determine how they would "sanction" recipients. In most states, a sanction results in a gradual or immediate loss of some or the entire TANF cash grant. Previous research (e.g., Soss et al., 2001) found that states with higher proportions of African Americans have stricter sanction policies. The current research seeks to examine how race may affect the use of sanctions. The authors find that race plays a role in the decision to sanction, but local context matters; in places where TANF is administered in a more centralized fashion (e.g., less frontline discretion), race is unrelated to sanctioning. These findings are part of a larger argument the authors make, stating that sanctions and other punitive policies are part of a larger paternalistic effort to regulate the behavior of the poor (Soss, Fording, and Schram, 2011).
Food Assistance. As TANF caseloads declined, the federal Supplemental Nutrition Assistance Program (SNAP), formerly called food stamps, has become the nation's largest income support program (Klerman and Danielson, 2011). SNAP provides benefits via a debit card to qualifying individuals and families that can be used to purchase food. Since 1996, the program has undergone a number of changes, and Klerman and Danielson (2011) examine the effect of those changes, as well as changes in the economy, on use of the program and the composition of caseloads. They note that the program has shifted from one primarily serving families with children who were receiving AFDC/TANF to one that serves those without ties to cash assistance. They find that welfare policies, such as those described above, served to push down the proportion of the caseload that combined food stamps with cash benefits. Changes in the economy have time-lagged effects, driving up caseloads even after a period of economic recovery has begun.
As was the case with the old AFDC program, there is some concern that participation in SNAP could have unwanted behavioral effects, such as reducing employment; but unlike TANF, SNAP does not have large-scale work requirements. Yet, as Hoynes and Schanzenbach (2011) note, testing this premise is difficult, since the program is fairly uniform across states (so designs that would exploit policy variation are not possible). The authors instead employ a quasi-experimental design that uses variation in county adoption of the Food Stamp program during the 1960s and 1970s and find that while participation in the program has no labor market effects overall, single mothers were significantly likely to reduce their work effort.
While SNAP is by far the largest of all food assistance programs in the U.S. (in 2011 more than 44 million people received benefits), other programs provide food and related assistance to low-income families. The Women, Infant, and Children (WIC) program offers food and nutritional counseling to low-income pregnant, post-partum, and breastfeeding women and to low-income children under the age of five. Using a design similar to that described immediately above (WIC was gradually implemented during the early 1970s), Hoynes and colleagues (2011) find that the implementation of WIC was associated with increased birth weights.
The School Breakfast and Lunch programs provide subsidies to participating schools so that they can offer eligible low-income children free and reduced price meals. The original intent of the program was to ensure that children consumed a nutritious meal that they may not have had otherwise. Existing studies find both negative and positive short-term effects of participating in the program. The School Lunch program has existed in various forms since 1946, so researchers are now able to examine longer-run outcomes of participation. Hinrichs (2010) finds no long term effects on health among those participating in the lunch programs in the mid-twentieth century, but participation is associated with large gains in educational attainment. While this analysis cannot determine the mechanisms through which the effects emerged, the author speculates that lunches provided an incentive to attend school (increasing education) but simply displaced food that would have been consumed elsewhere (thus no health effects); or, health effects might have been positive in the short run, thus contributing to higher educational attainment.
The School Breakfast Program is newer, and participation is lower than in the Lunch Program, both because fewer schools participate and because fewer eligible families take up the benefit (Bartfield and Kim, 2010). On this second issue, Bartfield and Kim (2010) find that family characteristics, school/community characteristics, and program logistics are all associated with participation. Poorer families, families with more children, and families with two working parents (who may be pressed for time) are more likely to participate, as are children attending schools with higher proportions of lower-income students and children attending schools that make it logistically easy for children to eat breakfast. Increasing participation may be an important goal for policy, since another study (Bartfield and Ryu, 2011) finds that participation is associated with a reduced likelihood of skipping breakfast.
Social insurance programs are not means tested and are intended to smooth income over the life cycle, for example when individuals are disabled or separated from employment through no fault of their own. These are large and costly programs for the governments, and their financial sustainability continues to be closely scrutinized. This section of our review covers recent research on the Social Security, Social Security Disability Insurance and Unemployment Insurance programs.
Social Security. Social Security is the United States' largest social insurance program, accounting for approximately 29 percent of overall federal government revenues and 20 percent of expenses (Aaron, 2011). Created in 1935 under President Roosevelt, the program provides cash benefits to retirees, disabled persons, and survivors after the death of a qualifying family member. Social security is mainly funded through an earmarked payroll tax.1 The large size of the program and the fact that the program benefits some of the most vulnerable groups of individuals in the United States has attracted a great deal of interest from researchers and government officials. The program enjoys popular support but has also heavily criticized. Articulating the program's shortfalls and suggesting improvements have been a major thrust of many recently published articles on the topic.
One of the major concerns about Social Security is potential insolvency (Aaron, 2011; Kotlikoff, 2011, Benitez-Silva and Heiland, 2007; Sabelhaus and Topoleski, 2007), which may lead to as large as 16 trillion dollars in future budget shortfalls (Kotlikoff, 2011). The other problem is the disincentive for work among recipients (Goda, Shoven, and Slavov, 2011b; Weathers and Hemmeter, 2011 and Wachter, Song, and Manchester, 2011). Researchers also mention issues such as inappropriate adjustment of benefits for inflation (Goda, Shoven, and Slavov, 2011a), the complexity arising from 2,728 rules created for the program (making rational saving and labor supply decisions almost impossible), and the unfairness associated with the shift of resources from singles to married couples, from individuals with lower life expectancies to those with higher longevity (Kotlikoff, 2011), and from future generations to current as earlier cohorts (born before 1935) paid little in payroll taxes, but enjoyed large benefits at the expense of future generations (Kotlikoff, 2011 and Aaron, 2011). The problems the Social Security system faces and the possibility for reform is the focus of a number of papers in a recent issue of the National Tax Journal. While the authors of the various papers (Aaron, 2011; Kotlikoff, 2011; Goda, Shoven, and Slavov, 2011a) agree that reforms are necessary, they diverge regarding the essence and extent of transformations.
The problem of potential insolvency of the Social Security program is addressed by Aaron (2011) and Kotlikoff (2011). While Aaron (2011) argues that only minor changes are required to correct financial shortfalls and that large scale changes will be disruptive, Kotlikoff (2011) insists on the necessity of in-depth reforms. Aaron (2011) argues that the changes needed to correct the potential insolvency problem should be aimed at either immediate fifteen percent cut in benefits or two percentage point increase in payroll taxes, or a combination of two. Kotlikoff (2011), on the other hand, believes that the problems with Social Security are too great to be resolved with small modifications. This author proposes to freeze benefit accumulation under the current system and replace Social Security with an alternative "Personal Security System" (PSS) program. The PSS would feature a mandatory contribution from an individual's income to a personal savings account (or both an individual and spousal account for couples). The federal government would match contributions to those who are poor, disabled, and unemployed. PSS contributions will be invested in a global-market portfolio and sold to the PSS participant when they reach a retirement age.
Another problem facing Social Security is the issue of appropriate adjustment of benefits for inflation (Goda, Shoven, and Slavov, 2011a; Aaron, 2011). Currently, benefits are indexed using the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W). However, both Aaron (2011) and Goda, Shoven, and Slavov (2011a) argue that this index does not appropriately adjust for the cost of living for Social Security recipients. Using data from the Health and Retirement Study for the late 1990's early 2000's, Goda, Shoven, and Slavov (2011a) test whether retirees have higher out of pocket medical expenses than workers. Such expenses might be higher, first, because Medicare premiums are automatically deducted from Social Security retirement benefits, and these premiums have grown much faster than Social Security benefits, and, second, because retirees have high out of pocket medical expenses that leave them with even smaller real income left for non-medical purchases. Goda et al (2011a) find that indeed non-medical spending of retirees is going down. The authors also compare CPI-W to CPI-E, an experimental index based on the spending pattern of the elderly, produced by the Bureau of Labor Statistics. They conclude that while CPI-E better adjust social security benefits for inflation it will be more costly to the federal government.
Work disincentives facing Social Security recipients (Goda, Shoven, and Slavov, 2011 b) are another frequently cited problem in the literature. Social Security imposes an implicit tax on work beginning from the year 14 of employment for an average-earning worker. The authors define implicit tax as the situation when the increase in the present value of the Social Security taxes exceeds the discounted value of future social security benefits. (Goda et al, 2011, b). The implicit tax rate increases in several jumps for all Social Security retirement beneficiaries depending on earnings and career length. And, for many recipients, regardless of the salaries' range, who choose to work over 35 years, a further earnings' accumulation towards retirement is not happening at all. Hence, a full amount of social security payroll tax beginning from year 36 and beyond becomes an expense that provides no additional Social Security benefits.
Social Security Disability Insurance. Social Security Disability Insurance (SSDI) is also funded via a payroll tax and pays out cash benefits to qualifying workers who become disabled. (A separate program, Supplemental Security Income- SSI- is a means tested disability program paid for out of general funds).2 As is the case for Social Security, SSDI also has work disincentives embedded within its structure; SSDI recipients lose disability benefits if their earnings exceed a specific level, commonly termed a "cash cliff", or, in the other words, if they perform so-called substantial gainful activity (SGA). To mitigate work disincentives among SSDI recipients, four states (Connecticut, Utah, Vermont, and Wisconsin) experimented with an alternative "benefit offset" policy, in which recipients incurred a $1 reduction in benefits for every $2 increase in earnings. The "benefit offset" policy was tested on a self-selected sample using a random assignment design. Compared to the "cash cliff" policy, "benefit offset" recipients were more likely to have earnings above the SGA level. However, "the benefit offset" policy did not affect overall labor force participation (because under the "cash cliff" policy individuals can still work and receive benefits as long as their work efforts are below the SGA level) or earnings level of SSDI recipients (because many participants of the experiment with earnings above SGA under the "cash cliff" policy reduced their earnings under the "benefit offset" policy) (Weathers and Hemmeter, 2011).
Wachter, Song, and Manchester (2011) in their study of employment and earnings' trends of SSDI applicants and beneficiaries before and after the year of application similarly find an adverse effect of the program on employment. Their analyses find that many male beneficiaries are likely to work in the absence of SSDI. Older (45-64 years old), rejected male applicants have low labor force attachment and low earnings, while younger, rejected male applicants tend to remain attached to the labor force after application, despite substantial losses in earnings. The authors also observe an increase in the number of individuals applying for SSDI benefits during economic downturns.
Unemployment Insurance. Unemployment insurance (UI) is a federal and state cooperative program to smooth income for workers who become unemployed through no fault of their own. Each state sets the benefit levels and eligibility requirements. In general, workers must have earned between $1,000 and $3,000 over the past four quarters (termed the base period), the exact amounts and calculation of the base period varying by state. Some states exclude earnings from the current quarter, a policy that is thought to exclude some low-income workers from the program, since workers with short or intermittent work histories may not qualify for UI. Because of this, some states have adjusted the base period to include earnings from the current quarter. To be eligible for UI, there are also several nonmonetary requirements related largely to the conditions of the job separation. To qualify for benefits, workers must not have quit or been terminated for good cause. UI beneficiaries typically lose their jobs due to layoffs, the closing of plants or other involuntary reasons although states may choose to pay UI benefits if employees quit for good reason (e.g., to care for an ill relative, escape domestic violence, or accompany a relocating spouse). The duration of UI benefit receipt is generally extended during serious economic downturns.
Krueger and Mueller (2010) examine the job search behavior of UI recipients using time use data. As a requirement of receiving UI, recipients are expected to search for work; the authors find UI recipients spend about 41 minutes per day seeking employment, and job search is less intense when state benefits are more generous or workers expect to be recalled by their employers. Job search intensity increases just prior to UI benefit exhaustion. Interestingly, UI recipients appear to spend more time searching for work than their similarly unemployed European counterparts. This is remarkable, in part, because a number of European countries are experimenting with their UI programs, paying partial benefits if UI recipients take a part-time job, providing they continue to search for full-time employment, a policy that appears more beneficial to men than women (Kyrra, 2010).
Shaefer and Wu (2011) examine the participation of low-educated, single working mothers both before and after welfare reform. They find that the UI has become a more common form of cash assistance than welfare for this group of women, even though UI benefit receipt levels did not change vis-à-vis low-educated but childless women. The fact that UI participation did not increase over the time frame of this study-even though UI eligibility did increase- suggests that low-educated, working mothers my not have sufficient knowledge of the UI program or may lack understanding of the bureaucratic processes.
The Child Support Enforcement (CSE) Program, also known as Title IV-D of the Social Security Act (1975), promotes family self-sufficiency and the well-being of children. The IV-D program is operated by state and local governments in partnership with the federal government. Major activities include establishing paternities, medical and financial support orders, and enforcing those orders. The initial emphasis of the program was the recovery of welfare costs and early in the program's history, the caseload was primarily AFDC or former AFDC participants. Over time, the program has handled an increasing proportion of all child support activities, including the collection of support initially established by the courts for couples using private attorneys. In FY2010, the program served 15.9 million cases with 6.8 million cases that had formerly received TANF and 6.9 million cases that had never received welfare. A total of $26.6 billion in child support was distributed in FY2010. Only 22 percent of distributed child support went to cases that also received Medicaid (18 percent) or TANF (4 percent) (Office of Child Support Enforcement, US Department of Health and Human Services, 2011).
Recent research on child support has focused on the amount of support ordered, including the forgiveness of debt (Cancian and Meyer, 2011; Ha, Cancian and Meyer, 2010; Heinrich, Burkhardt and Shager, 2011). For example, Cancian and Meyer (2011) use the Wisconsin CSE caseload to examine the amount of child support owed in slightly more than 30 percent of the cases which involve multiple partner fertility, that is, cases in which a father has biological children with more than one mother (and thus may owe child support across different households). In complex families, should the amount of the child support order be based on birth order, or should older children be held harmless by the birth of new siblings or half-siblings? Because of the administrative burdens of revising old child support orders whenever a father has another child, the authors conclude that child support orders should be set at a fixed amount of the nonresident parents' income, even though this approach ignores the economies of scale in raising multiple children and can result in high burdens for child support obligors.
Ha, Cancian and Meyer (2010) use longitudinal data from Wisconsin to document the sensitivity of the amount of child support ordered to changes in the obligors' incomes. Over a five year period, the authors document than between an third and one-half of fathers experienced changes in income of 50 percent or more. Despite the fact that the incomes of many obligors changed substantially, only about one-third of the court orders for child support actually changed, documenting the inflexibility of court orders to even large upwards and downwards changes in incomes. Based on Wisconsin law, in year 3 of their study, 60 percent of the child support cases were eligible for a change in the court order, although only 8 percent were actually modified. The insensitivity of court orders to large changes in income can lead to obligors with diminished income owing unrealistically high amounts or, in the case of improvements in income, fixed orders leave children short changed.
Similarly, Heinrich, Burkhardt and Shager (2011) examine the effects of changing the amount of child support owed for obligors with substantial debt. With $107 billion in child support debt in FY2009, the federal and state governments are looking at debt forgiveness strategies in cases where both parents agree or in exchange for desired behaviors such as the on-time payment of current support. These authors report the findings of the Family Forward pilot debt forgiveness program in which obligors received 50 cents reduction in debt for every dollar of current support paid. Take-up of the program was low partly because of mistrust of obligors with large debts of the CSE program. Those who participated were older and had larger debts. The study findings were mixed. The experimental results found no significant differences between treatments and controls whereas their nonexperimental analysis using propensity score matching found lower debt balances of more than $2,500 by the end of the study.
Nepomnyaschy and Garfinkel (2010) look at total amount of child support received by low-income children whose parents never married. Using the Fragile Families data, the authors look at both formal, court-ordered support and informal cash and in-kind support. By the time court-ordered child support is received, it is offsetting informal support which tends to decline rapidly in the 15 months that parents live apart. The authors view formal and informal sources of child support as substitutes for one another and do not find significant impacts of strong child support enforcement on the total amount of support received after the first five years after a nonmarital birth in fragile families.
Other researchers have examined some of the behavioral and economic effects of effective child support enforcement. For example, Crowley, Jagannathan and Falchettore (2012) found that effective child support enforcement has unintended effects on the abortion rate per 1,000 women of childbearing age although no impacts on the ratio of abortions relative to the number of live births. The authors conclude that there results are sensitive to how one measures abortion (rate versus ratio) but that it appears that stronger child support enforcement likely increases the likely that women have more children as they expect that they may receive some financial assistance. Cheadle, King and Amato (2010) report that patterns of higher and lower father involvement among men with different levels of child support compliance. Similarly, Garasky, Stewart, Gundersen, and Lohman (2010) examine the impacts of impacts of in-kind and monetary support and find a strong relationship between in-kind support and father involvement. Ha, Cancian and Meyer (2011) report that although child support receipt can be intermittent, it generally improves the regularity of mothers' total family income, particularly for mothers in low-income families.
Because child support is a large program, scholars have examined a wide variety of issues related to how child support orders should be set, whether or not debt forgiveness can encourage great compliance with the program as well as behavioral outcomes such as abortion choices and father involvement. For a thorough historical overview of the program as well as a comprehensive review of this literature prior to 2006, readers are referred to Pirog and Ziol-Guest (2006).
Since passage of the Tax Reform Act of 1986, the federal income tax system has arguably became the primary means of providing cash assistance to low-income families with kids (Eissa and Hoynes (2011); Maag et al., 2010). Such assistance includes several tax credits that can be claimed only by the families with dependent children, such as the Child Tax Credit (CTC) and Additional Child Tax Credit (ACTC), rules within the tax code, such as the head of household tax filing status (that extends the tax brackets for single parents, lowers marginal federal income tax rates, and allows individuals to claim many of the mentioned credits for kids, not available for married couples filing separate tax returns); credits that benefit lower-income families with dependent children more, such as the Earned Income Tax Credit (EITC) and the Child and Dependent Care Credit (CDCC). Dependent exemptions, the head of household filing status, CTC, and CDCC simply reduce the amount of taxable income, while the EITC and ACTC are refundable. The EITC, ACTC, and CDCC are only available for lower income households with earned income. The EITC has been the most generous federal tax credit of all, which, not surprisingly, has received a lot of attention by researchers.
Recent research on the EITC falls into three broad categories: research on the characteristics of EITC recipients (Athreya, Reilly, and Simpson, 2010; Lim, Livermore, and Davis, 2010); whether and how the presence of the credit alters people's behavior (Fitzpatrick and Thompson, 2010; Lopoo and Raissian, 2012; Herbst, 2011); and possible outcomes of EITC structure changes (Eissa and Hoynes, 2011; Wheaton and Sorensen, 2010), including replacing the EITC with a different type of credit or other policy change (Rothstein, 2010, Maag, 2010). Each is discussed in turn.
Athreya, Reilly, and Simpson (2010) observe that while the earnings of non-recipients tend to be higher when their age is in the 30's and 40's ("hump-shaped"), the earnings of the EITC recipients are relatively stable (more linear) over their lifetime. In addition, the authors find that the EITC increases labor force participation, but not necessarily the number of hours worked. Finally, EITC households have much smaller wealth accumulations than non-EITC families. Lim, Livermore, and Davis (2010) find that the majority of EITC families experience some sort of "material hardship" (defined by the authors as "the inadequate consumption of… food, housing, clothing, and medical care", p.267). This likelihood of experiencing hardship increases for EITC households without a bank account.
Several recent papers examine behavioral changes associated with EITC receipt (Fitzpatrick and Thompson (2010); Lopoo and Raissian (2012); Herbst (2011). Fitzpatrick and Thompson (2010), for example, examine how geographic differences affect labor market responses by potential EITC recipients (in particular single mothers). The authors argue that variations in the cost-of-living, make the EITC more or less valuable depending upon one's residence. Since wages of low-skilled workers in high-cost areas tend to be higher than in low-cost areas, workers in higher cost areas may receive lower benefits or lose eligibility all together (the EITC increases as earnings rise, then plateaus and gradually phases out as earnings increase even more). Using a 1993 expansion of the EITC as a natural experiment, the authors find that the EITC increases labor participation among single mothers, but only in the lowest cost areas and has no effect in the highest cost areas of the country.
Lopoo and Raissian (2012) study the impact of EITC on fertility (EITC benefits are larger for families with two children or three or more children, compared to families with one child). The authors conduct a literature review that does not confirm the effect of the credit on birthrates. Crump, Goda, and Mumford, 2011, for example, do not find a combined effect of EITC, CTC, and personal exemptions on fertility), but find a drop in abortions associated with EITC eligibility (Herbst, 2011).
The effect of potential EITC structural changes is addressed by Eissa and Hoynes (2011) and Wheaton and Sorensen (2010). Eissa and Hoynes (2011) test the effects of alternative scenarios that involve a more or less generous credit and expanded eligibility. In general, the authors find expansion of the phase-out region to be more welfare improving than increasing the EITC rates. Wheaton and Sorensen (2010) examine a possible EITC extension to noncustodial parents who work and pay child support; this policy was enacted in the state of New York and Washington DC and in 2007 was proposed on a federal level by Senator Bayh and then-Senator Obama (S.1626). Should the extension be enacted on a federal level, the authors predict that a credit in the amount of $600 to $1,800 may benefit as many as 645,000 noncustodial parents. The number of noncustodial parents is much higher than that, but the proposed eligibility criteria would exclude many parents.
Finally, the possibility of replacing the EITC with other changes to the federal tax code is considered by Rothstein (2010) and Maag (2010). Rothstein (2010) studies the labor market responses and net transfers of the EITC and a hypothetical Negative Income Tax (NIT) for single and married women with and without children. Under the paper's scenario, the EITC can be claimed by low-income working families, while NIT eligibility depends only upon income and not employment status. Thus, the EITC has a potential to increases labor force participation among the low income population, which, in turn, could drive down wages. The hypothetical NIT, on the other hand, discourages work, which may induce transfers from employers to low-skilled workers. From the results of a series of simulations, Rothstein echoes findings of other researchers, namely that the EITC is a cost-effective mechanism for raising incomes of low-skilled women under the assumption of fixed wages. Under other scenarios, though, the NIT is more cost effective.
Maag (2010) discusses the possibility of introducing a Unified Child Credit, a single credit that would replace several tax credits and policies discussed above. The main idea behind the proposal was to separate the federal tax preferences by three functions, such as a subsidy for working, having children, and spending on child care. The eligibility for a work subsidy should not depend on whether the family has kids and tax payers have to be able to claim child credits even if they did not earn income. Some of the specific proposals of the Unified Child Credit include consolidation of benefits for having kids, such as the dependent exemptions, CTC, and the EITC (see also Crump, Goda, and Mumford, 2011), combining the CTC, ACTC, dependent exemption, and the benefit from the head of household filing status, as well as consolidating college subsidies, such as the EITC, dependent exemptions for college students, the American Opportunity Tax Credit, and the tuition and fees deductions. Maag discusses the literature available on the proposals.
Great Britain, for example, adopted a Child Tax Credit as part of a comprehensive set of reforms designed to eliminate child poverty. Waldfogel (2010) demonstrates that the combination of the Child Tax Credit, a working family's tax credit (similar to the EITC), early childhood education efforts, and other programs moved upwards of 1.6 million British children out of poverty.
In this paper we have reviewed current trends in social policy research. We targeted recent empirical publications (mainly published in 2010 and 2011) related to public means tested programs, social insurance programs, child support as well as social programs delivered via the tax system. Looking across the various social policies addressed in this review and the associated recent research, several themes emerge. One of the major themes is a problem of the work disincentives potentially embedded within social programs. This is true for TANF, SNAP, and even for Social Security, a benefit designed to be used upon retirement. Surprisingly, the most recent research on Unemployment Insurance has not directly addressed work disincentives, but as UI receipt remains quite high in the wake of the Great Recession, it is plausible that more work will be done in this area.
Besides work disincentives, researchers have described behavioral responses to social policies including the positive effects of EITC on labor force participation among single mothers in the lowest cost areas (Fitzpatrick and Thomson, 2010). Lower abortion rates have been associated with the EITC (Herbst, 2011) as well as strong child support enforcement (Crowley, 2012). Similarly, Hoynes et al. (2011) find higher birth rates associated with WIC. , but no effects on fertility (Lopoo and Raissian, 2012). Crowley et al. (2012) report lower abortion rates in the states with stronger child support enforcement Hoynes et al. (2011) find higher birth rates associated with WIC program and in their review of the literature on natalist policies in the US, Lopoo and Raissian (2012) conclude that many public policies have affected the fertility choices of families.
Structural changes in social policy programs have been the other point of interest of recent policy research. Researchers have studied alternative poverty measures to create a better baseline for the means tested program (Couch and Pirog, 2010; Hutto et al., 2011; Brandolini et al., 2010; and others), discussed alternative adjustments of Social Security benefits for inflation (Goda et al., 2011a, Aaron, 2011), proposed changes to deal with potential insolvency of the Social Security program (Aaron, 2011 and Kotlikoff, 2011), and to reduce work disincentives created by SSDI (Weathers and Hemmeter, 2011). On the tax frontier Eissa and Hoynes (2011) and Wheaton and Sorensen (2010) studied the alternative structures of the EITC, while Rothstein (2010) and Maag (2010) simulated the possibility of replacing the EITC with alternative tax policy instruments.
In conclusion, there has been a great deal of research on the efficacy, structure and outcomes associated with many social programs including those embedded in our tax codes. Our review is not exhaustive. We exclude some programmatic areas because they are covered by others in this very journal and still others in our attempt to focus on those areas that have received the most attention of scholar and policy makers over the past few years. Nonetheless, despite our somewhat narrow focus, there is a rich and growing body of information on the impacts of our country's collage of social programs.
1Aaron (2011) notes that Social Security revenues also “come from income taxes collected from the inclusion of some benefits in taxable personal income” (p.388)
2Policy makers have been concerned about work disincentives in SSI, particularly among noncitizens. PRWORA barred the majority of immigrants from receiving SSI. A recent study finds that while this policy is associated with increased employment of male noncitizens, a similar association does not exist for female noncitizens, who are in fact more likely to see declines in income (Kaushal, 2010).