Printer Friendly

Comments and Discussion.

COMMENT BY

JANET CURRIE This paper by Hilary Hoynes and Diane Schanzenbach should be required reading for policymakers facing decisions about social safety net programs in the United States. The bottom line is clear, and clearly put:
We are spending too little on children and their families.... Any cuts
to current programs that will reduce resources going to children would
have direct, negative effects on children in both the short and long
terms. It is also crucial to recognize that the modal recipient family
is combining safety net use with employment; the view that all spending
is welfare and going to out-of-work families is not the case.


The findings regarding spending, in particular, are the most authoritative to date in that they are based on administrative data, which captures much more spending on families at the bottom of the income distribution than traditional, survey-based measures.

Hoynes and Schanzenbach argue that we spend too little, because society has not recognized the returns to investing in safety net programs in terms of reductions in taxpayer expenditures (or increases in tax revenues) down the road. Children who grow up to be better educated, more likely to be employed, and more likely to lead healthy and productive lives will pay higher taxes themselves and be less likely to rely on taxpayer-funded social programs as they age. Although there is no doubt that safety net programs have an investment component, it is less clear that policymakers have failed to take future payoffs into account when arriving at current levels of social spending. Language about "investing in children" goes back at least to the Clinton administration. President Clinton's 2000 State of the Union Address repeatedly calls for investing in children and working families--stating, for example, "We must also make investments that reward work and support families. Nothing does that better than the Earned Income Tax Credit."

It is possible, in fact, that policymakers have taken the investment paradigm too literally when it comes to children. As Hoynes and Schanzenbach point out, the United States spends 2.1 percent of GDP on children, compared with 9 percent of GDP on the elderly. Yet we do not expect spending on the elderly to yield a return. We spend on the elderly because they are viewed as deserving of our support. The disparity in treatment of children and the elderly is embedded in the fact that programs for the elderly (especially Social Security and Medicare) are entitlements, with strong protection from the vagaries of federal budgeting, whereas programs for children are highly vulnerable to cuts in spending from their already-low levels.

Although the comparison between spending on children and spending on the elderly is instructive, focusing on intergenerational conflict may distract from the major driver of spending, which is health care costs. Hoynes and Schanzenbach's figure 6 shows that Medicaid is the largest single program for children in terms of costs, and this figure understates costs because it omits spending on pregnant women and the state Child Health Insurance Program (CHIP). Moreover, in programs such as public housing and the Supplemental Nutritional Assistance Program (SNAP), some share of the benefit goes to adults in the household, whereas Medicaid is targeted to the covered children.

However, the bulk of public health care spending goes to the elderly and disabled. Until recently, Medicaid itself was really two programs, one covering low-income children and their parents, and the other covering the elderly and disabled. And though children make up half of Medicaid beneficiaries, they account for only 19 percent of expenditures (Truffer, Wolfe, and Rennie 2016). In 2015, per-person personal health care spending for the elderly, disabled, and children was, respectively, $14,323, $19,478, and $3,389.' Now that substantial numbers of adults are also covered under the Affordable Care Act's Medicaid expansions, the share of Medicaid spending accounted for by low-income children can be expected to fall still further.

Given the gap in health care costs between children and the elderly, rising health care costs can be expected to widen gaps in spending between the two groups and will create budgetary pressures that will threaten all other forms of spending on children. According to the Centers for Medicare and Medicaid Services, health care spending has been growing faster than the rate of inflation for decades, and is projected to continue to do so. This rise in health care costs seems to be driven largely by higher prices in the United States relative to other countries.

Preserving the safety net is likely to require the reining in of health care costs. Because prices are so important to driving costs, measures that might help include imposing regulations to mandate price transparency and reforming Medicare to allow it to negotiate prices (including drug prices) with providers. Other measures that might help include using big data to identify providers that are outliers in the care they provide, and more systematically identifying best practices (Currie, MacLeod, and Van Parys 2016; Currie and MacLeod 2017).

Just as rising health care costs will continue to drive a wedge between spending on the elderly and spending on children, they will also continue to increase the share of spending on children in families just above the poverty line relative to children in poor families. This shift in relative spending is quite intentional. Even if we kept everything the same for poor families, expanding public health insurance for families above the poverty line (along with expansions of the Earned Income Tax Credit and other programs to families in this income category) would have had the effect of increasing the share of safety net spending going to families above the poverty threshold. Hence, I urge readers not to skip the authors' online appendix, and to focus on the figures that show amounts spent per child in the various income categories (for example, the top panels of the authors' figures 11 and 12), rather than to focus on the figures emphasizing expenditure shares for different economic groups. These figures also suggest flat periods in spending on families below poverty, along with declines in spending for families with no income between 1995 and 2005, but are less dramatic than the shifts in shares emphasized in the main text. On the whole, they indicate that the trend is for new safety net monies to be allocated to nearly poor families rather than for money to have been taken from poor families.

Hoynes and Schanzenbach focus on federal spending, but it is important to understand that this is only one strand of the safety net. State and local spending on the safety net is tremendously important, and also highly variable across states and over time. Medicaid is one of the most important components of state budgets (along with K-12 education, higher education, and prisons), so rising health care costs threaten to eat up a larger and larger share of state spending. This budgetary pressure will likely have negative consequences for public education, state Earned Income Tax Credit programs and child tax credits, and child protective services (which are chronically underfunded in many localities, even though child abuse and neglect is a leading cause of child injury and death). Other important state programs to protect working families include workers' compensation and especially unemployment insurance.

It is surprising that in a period when the safety net is increasingly geared toward parents who work, unemployment insurance systems in many states have become less and less generous, to the point where they offer very little insurance to working families in the event of a job loss. Data from the National Employment Law Project indicate that the fraction of the unemployed who receive any assistance ranges from a low of 11 percent in Florida to a high of 66 percent in North Dakota. The median state, Oklahoma, assisted only 28 percent of the unemployed (McKenna and McHugh 2016). Several states adopted significant benefit cuts after 2011, and currently nine states are offering fewer than 26 weeks of benefits. For example, in Florida in February 2016, a newly unemployed worker qualified for only 12 weeks of benefits. Given work by Jonathan Gruber (1997) and Raj Chetty (2008) showing that unemployment insurance smooths consumption and reduces liquidity constraints on households, this is a disturbing trend.

One reason that state and local safety net programs are generally neglected by researchers, despite the rich variation in these programs, is that good data are hard to come by. Even for federal programs, administrative data are patchy and incomplete. For example, Hoynes and Schanzenbach point out that there are no federal data that can be used to apportion Medicaid, Supplemental Security Income, and public housing expenditures by poverty group or parental employment status. Hence, even studying the components of the federal safety net in a consistent fashion requires making assumptions about how resources are being allocated.

The researcher wishing to study state and local programs must first assemble and harmonize data from many different jurisdictions, all with different data access policies and stances toward the use of their administrative data for research purposes. Thus, just building a data set becomes simply a monumental task that tends to shut down research before it can even get started. The creation of cross-state data depositories for state and local administrative data would likely have a tremendous impact on research in this area. One model is the Healthcare Cost and Utilization Project, which is managed under the auspices of the federal Agency for Healthcare Research and Quality. Participating states provide their hospital discharge data (that is, records of each hospitalization that occurs in the state, which are collected for regulatory purposes) to the project's central depository, which makes the data available in an anonymous and standardized format to health care researchers.

One reason to hope that more data could become available at the state and local levels is that these jurisdictions will be increasingly responsible for experimenting with the traditional safety net programs. Executive Order 13828, dated April 10, 2018, encourages states to implement stricter work requirements on programs, including SNAP and Medicaid, reduce the size of program bureaucracies, target programs more strictly to the neediest people, and eliminate programs they find to be duplicative or ineffective. It also promises to grant states flexibility to achieve these goals. These policies seem likely to reduce access to the safety net for many, and it will be important to assess their effects on children and families.

In summary, Hoynes and Schanzenbach offer a wonderful introduction and overview to federal safety net programs, as well as innovative analyses of administrative data to support their arguments. In this brief comment, I have tried to place the programs and trends they identify in a larger context, in which spending on the elderly is protected in entitlement programs while spending on children is not; spending on all nonhealth programs is increasingly threatened by rising health care costs; and variation in the generosity of the safety net depends on state and local policies, in addition to the federal programs and policies that garner the lion's share of research attention. Adding these dimensions to the analysis would not change their key conclusion--that we spend too little on children--but it would make clear how difficult it may be to spend more on programs that have been shown to make a difference.

REFERENCES FOR THE CURRIE COMMENT

Chetty, Raj. 2008. "Moral Hazard versus Liquidity and Optimal Unemployment Insurance." Journal of Political Economy 116, no. 2: 173-234.

Currie, Janet, and W. Bentley MacLeod. 2017. "Diagnosing Expertise: Human Capital, Decision Making and Performance among Physicians." Journal of Labor Economics 35, no. 1: 1-43.

Currie, Janet, W. Bentley MacLeod, and Jessica Van Parys. 2016. "Provider Practice Style and Patient Health Outcomes: The Case of Heart Attacks." Journal of Health Economics 47: 64-80.

Gruber. Jonathan. 1997. "The Consumption Smoothing Benefits of Unemployment Insurance." American Economic Review 87, no. 1: 192-205.

McKenna, Claire, and Rick McHugh. 2016. "Share of Unemployed Receiving Jobless Aid Remained at Record Low in 2015." Blog post, February 9, National Employment Law Project, New York. Truffer, Christopher J., Christian J. Wolfe, and Kathryn E. Rennie. 2016. 2016 Actuarial Report on the Financial Outlook for Medicaid. Baltimore: Centers for Medicare and Medicaid Services.

(1.) This is according to the Centers for Medicare and Medicaid Services' National Health Expenditures fact sheet for 2015.

COMMENT BY

GORDON B. DAHL (1) Hilary Hoynes and Diane Schanzenbach's paper serves as a valuable resource for both researchers and policymakers. It makes two contributions. First, it synthesizes the recent literature on the effects of early investments in children, with a particular focus on safety net spending directed toward children. Not so many years ago, there was scant evidence on long-term outcomes, and arguments for government transfer spending on children relied more on humanitarian and social insurance grounds. But as Hoynes and Schanzenbach document, there is now substantial evidence that spending on children has benefits for a variety of later-in-life outcomes. Some of these gains accrue privately, but others have positive spillovers to society due to increased tax revenue and lower government transfers in the future.

The second contribution is an analysis of how spending on children via the safety net has changed over time. The findings are both striking and relevant for policymaking. Total spending has remained fairly flat over time, but its composition has changed. Relative to 20 years ago, more spending reaches families near or above the poverty line, while less is spent on the poorest of the poor. There has also been a large movement away from unconditional transfers and toward benefits linked to work. Other studies have looked at how the child safety net has evolved, but this is the first based primarily on administrative data. This is an important contribution, given that survey data suffer from several issues--including sizable under-counting, a problem that is becoming more severe over time.

Although the long-term benefits of safety net spending on children documented by Hoynes and Schanzenbach are compelling and broad-based, I found it refreshing that the authors remained true to what the data can and cannot say in terms of policy recommendations. The authors rightly conclude that the fiscal benefits are unlikely to make increased expenditures on child safety net programs self-funding. Instead, the investment rationale still needs to be combined with humanitarian and social insurance motivations. Moreover, the authors recognize that the literature is not yet developed enough to estimate rates of return or provide guidance on how to optimally allocate funding across programs. This type of humility is admirable, but it should not detract from the authors' main policy conclusion that there is "a substantial investment component [to safety net spending], and because there have been positive returns from expansions in spending, the evidence suggests that we may be spending too little on the safety net for the young." At a more granular level, there is a solid case that returns to increased spending on children are especially large for the most disadvantaged, and that reallocating spending from later in life to earlier in life is likely to enhance efficiency.

Hoynes and Schanzenbach are experts on this topic. Their summary of the literature is comprehensive and up-to-date, and their analysis of spending trends is well executed. This is a great paper, with little to quibble over, so I instead focus my comments on three broadly related issues: program interactions, work requirements, and intergenerational issues.

PROGRAM INTERACTIONS The authors' analysis focuses on the tax and transfer benefits for seven of the largest programs affecting children. In the authors' figure 9, they summarize changes in universally available cash and near-cash programs between 1992 and 2015. The figure plots benefits for a single adult with two children in Colorado, and serves to highlight the shift over time toward programs tied to work.

An augmented version of the authors' figure 9 can also be used to illustrate program interactions, and the unintended incentives that can arise. In my figure 1, I have added three universally available noncash programs to the 2015 panel: Medicaid, the Children's Health Insurance Program (CHIP), and the Premium Tax Credit (PTC), which subsidizes health insurance under the Affordable Care Act (ACA). These three programs provide a patchwork of health insurance coverage for low-income families.

As background, all but two states cover children's health insurance up to at least 200 percent of the federal poverty level (FPL) via Medicaid coverage and CHIP. In addition, most states cover pregnant women past the federal minimum of 138 percent of the FPL via Medicaid and CHIP. In contrast, health insurance coverage for other parents varies widely across states. Thirty-two states currently cover parents up to 138 percent of the FPL, because these states have adopted the ACA Medicaid expansions. But 19 states have not expanded Medicaid, and among these nonexpansion states, the median eligibility limit is only 44 percent of the FPL. Premium assistance credits kick in after 138 percent of the FPL has been reached for all parents, and after CHIP eligibility ends for all children (Garfield and Damico 2017).

In my figure 1, I graph the case for a single adult with two children in North Carolina (as opposed to Colorado, in the authors' figure 9). (2) North Carolina was chosen because it illustrates the potential for perverse work incentives when the three health insurance programs are not well coordinated. North Carolina chose not to adopt the Medicaid expansions. Between 0 and 44 percent of the FPL, a parent in North Carolina qualifies for Medicaid; between 44 and 138 percent, a parent receives no coverage or subsidy; and between 138 and roughly 350 percent, a parent is eligible for marketplace subsidies through the PTC. This creates a gap in coverage for the parent, as shown in my figure 1.

To illustrate the type of work disincentives created by the canyon-shaped gap in coverage, consider a single parent in North Carolina with two children who earns the minimum wage of $7.25 per hour. If this parent works between 0 and 25 hours per week ($0 and $8,985 in yearly earnings), they would be covered by Medicaid. But they would have no coverage if they worked between 25 and 78 hours per week, as marketplace subsidies do not start until $28, 180 per year. This example makes clear the disincentive for full-time employment, as it entails a loss of Medicaid. Even for a single parent making twice the minimum wage ($14.50 per hour), there would be no assistance between 12 and 39 hours per week.

Does the ACA mandate that employers offer full-time workers health insurance coverage help fill in the gap? The answer is: only imperfectly. One challenge is that such a mandate creates an employer-based disincentive for hiring full-time workers. Moreover, 42 percent of working adult Medicaid enrollees work in a firm with fewer than 50 employees, and these firms are exempt from the mandate (Garfield, Rudowitz, and Damico 2018).

As shown in my figure 1, health insurance assistance for children does not have a similar gap. Even so, a parent's coverage can have spillovers to their children. The first reason is that when a parent does not have access to health care, they are more likely to become sick and less able to effectively care for their children. An additional spillover is that roughly 160,000 uninsured children have a parent in the coverage gap. This is potentially a problem, because parental coverage in public programs is associated with higher enrollment of eligible children (Sommers 2006).

Similar notches in the Temporary Assistance to Needy Families (TANF) program and Section 8 housing vouchers make the work disincentive problem even worse for some families. Other programs--such as the Special Supplemental Nutrition Program for Women, Infants, and Children and the National School Lunch Program--are also tied to the FPL, and therefore they affect a family's budget constraint. One caveat in the analysis of noncash programs is that individuals may not value them at the cost of provision. (1) If individuals value in-kind transfers such as health insurance or housing vouchers at less than their cost, this would make the canyon-shaped gaps in the budget constraint less pronounced. But the basic point remains that program interactions can have unintended incentive effects, especially when they create nonlinearities and dominate segments in the budget constraint.

As a side note, from an evaluation perspective, program interactions make it more difficult to estimate the effect of safety net programs. Programs can have offsetting incentive effects on an individual's budget constraint. For example, the phase-out portion of the Earned Income Tax Credit (EITC) coincides with the introduction of health insurance subsidies in my figure 1. Program interactions also pose a challenge for certain estimation approaches. Suppose a researcher was interested in utilizing the kinks in the EITC schedule to estimate labor supply elasticities. One approach would be to use a bunching estimator, looking for excess mass to the left of the first kink in the EITC schedule, for example. But my figure 1 makes clear that in this setting a bunching estimator will have issues, as the notch in Medicaid will limit the number of individuals with earnings in a neighborhood near the first EITC kink.

WORK REQUIREMENTS One of Hoynes and Schanzenbach's central findings is that there has been a shift toward requiring work for benefit eligibility, largely as a result of more reliance on programs like the EITC and less on cash transfers like the now-defunct Aid to Families with Dependent Children program. The authors recognize the importance of assistance programs that supplement low earnings during normal economic times, especially given wage stagnation in the lower end of the wage distribution. They argue that "it is crucial to preserve these programs' work incentives, which are currently quite strong."

Preserving work incentives is important, but the shift toward work requirements can have the wrong incentives if implementation is not well thought out. Consider recent proposals to link Medicaid to employment. Starting in January 2018, states were allowed to seek a waiver and impose work requirements for Medicaid eligibility. Kentucky was the first state to get approval, and other states are following (Goldstein 2018). For Medicaid nonexpansion states seeking waivers, like Kansas and Mississippi, meeting Medicaid work requirements through 20 hours of work at the minimum wage would actually lead to a loss of Medicaid eligibility, as income would be too high. One solution is to expand Medicaid coverage at the same time as imposing a work requirement, a proposal that was recently put forward as a political compromise in North Carolina. (4)

Moreover, it is important to recognize that not all social assistance programs are designed with a positive work incentive. Consider one of the largest social insurance programs in most countries, disability insurance (DI). In the United States, DI is administered through two programs, Supplemental Security Income and Social Security Disability Insurance. To qualify for DI in the United States, the primary requirement is that the individual is deemed not able to work, with individuals being disqualified if they earn more than a minimal amount. (5) DI is often considered a social insurance program, but it also has incentive effects and is a key part of the safety net. DI participation has been shown to generally rise during periods of high unemployment, even though it is unlikely that the latent amount of disability in the population has increased (Autor and Duggan 2003).

In the United States, an individual is either on or off DI, whereas in many European countries partial disability is allowed. For example, in the Netherlands roughly 40 percent of individuals are currently on partial disability benefits. One possible reform to the U.S. system would be to allow for partial disability, so that individuals with some ability to work could be gainfully employed. Research finds that many DI participants have substantial work capacity, both in the United States and Europe (French and Song 2014; Maestas, Mullen, and Strand 2013; Kostol and Mogstad 2014). The possibility of partial DI has the potential for cost savings that can be redirected elsewhere.

A detailed discussion of policy reforms to encourage part-time work for disabled individuals is beyond the scope of this comment. But other researchers have thoughtfully considered what types of reforms might work. Some of the more innovative proposals promote work through a mixture of firm incentives and individual accommodations to allow those with partial work limitations to remain employed or return to work (Autor and Duggan 2010; Burkhauser and Daly 2012).

How do DI programs interact with the rest of the social safety net provided to families? The first thing to note is that health insurance coverage is automatic if an individual is on DI in the United States. Combined with a replacement rate of 40 to 50 percent, this makes DI one of the more generous social assistance programs in the United States.

Recent research has also documented substantial social support substitution across programs. Lex Borghans, Anne Gielen, and Erzo Luttmer (2014) examine a reform in the Netherlands that tightened DI eligibility for existing claimants. Using a regression discontinuity design, they find that about 4 percent of DI participants exited DI due to the more stringent rules and that annual benefits fell by about [euro]1,000, or roughly 10 percent. Treated individuals exposed to the reform replaced over 60 percent of lost DI benefits with increased earnings in the labor market. Equally relevant, the drop in DI income was partly offset as individuals shifted to other government programs. The authors find that for each [euro] 1 of lost DI benefits, treated individuals collected [euro]0.30 from other social assistance programs in the short run (primarily unemployment insurance). This echoes the point made above that considering program interactions is crucial when evaluating the social safety net.

INTERGENERATlONAt ISSUES Hoynes and Schanzenbach's review of the recent literature documents compelling evidence for the positive effects of social safety net spending on children's outcomes. There are both immediate and medium-term benefits, as well as long-term improvements in a variety of health, human capital, and economic outcomes. When thinking about long-term effects, one additional consideration is whether a parent's participation in a program has an effect on their child's participation.

Parental participation in a social assistance program--such as TANF, SNAP, or DI--could influence a child's participation through a variety of channels. Parents could serve as role models, provide information about how to apply, demonstrate what it is like to be on a program, or even invest differentially in child development due to changing resource constraints. All these channels suggest a causal effect, where a parent's participation influences a child's outcomes in the long run. Conversely, the use of public assistance could primarily be due to environmental factors. Poverty, bad health, and reduced opportunities could persist across generations, in which case intergenerational links could simply reflect unobserved heterogeneity and not a behavioral response.

Until recently, it has been difficult to differentiate between correlation and causation. But a series of recent quasi-experimental papers suggests that children do learn from their parents. For example, using an instrumental variables approach, Robert Hartley, Carlos Lamarche, and James Ziliak (2017) find that a mother's use of welfare increases the chances that her daughter will participate as well. Using a random judge design, Dahl, Andreas Kostol, and Magne Mogstad (2014) find that children whose parents enter DI on appeal are more likely to themselves participate as young adults. And using a regression discontinuity design, Dahl and Gielen (2018) find that children whose parents are kicked off DI or have their benefits reduced are less likely to themselves participate 21 years later. Monique de Hann and Ragnhild Schreiner (2017) bound average treatment effects and find substantially smaller estimates compared with the local average treatment effects identified in the other papers, suggesting caution about extrapolating the large responses found to the entire population.

Taken together, these recent studies suggest that children do learn from and copy their parents. But the spillovers extend beyond program participation. Dahl and Gielen (2018) show that children whose parents are pushed out of DI or have their benefits reduced not only reduce their own participation in DI but also earn more in the labor market as adults. The increased taxes due to increased earnings by children exceed the cost savings from their reduced DI usage. Consistent with an anticipated future with less reliance on DI, the children of affected parents on average complete an extra 0.12 year of schooling. Although several interpretations of these intergenerational effects are possible, a consistent explanation is that children learn from their parents about the relative costs, benefits, and stigma associated with work versus government assistance. From a fiscal perspective, these intergenerational links matter. Ignoring parent-to-child spillovers understates the long-run cost savings of the Dutch reform by between 21 and 40 percent in present discounted value terms.

FINAL THOUGHTS Hoynes and Schanzenbach provide an excellent summary of the existing literature and a careful analysis of safety net investments in children. Their paper is a useful reference for academic researchers and policymakers alike. Though my comment has disproportionately focused on various aspects of incentives related to work, this should not be interpreted as an endorsement of policies to reduce or eliminate unconditional cash transfers. As the authors point out, "building a safety net around work leaves families with little protection during times of high unemployment." Creating effective incentives for work is important, but it is crucial to recognize that the social safety net also needs to take care of children with nonworking parents. Children whose parents are out of work are among the poorest of the poor, and the United States currently does not have a comprehensive safety net to cover them. Investments in these disadvantaged children have high returns, but policy recommendations about how to best structure programs to help children in these nonworking families are beyond the scope of this comment.

REFERENCES FOR THE DAHL COMMENT

Autor, David H., and Mark G. Duggan. 2003. "The Rise in the Disability Rolls and the Decline in Unemployment." Quarterly Journal of Economics 118, no. 1: 157-206.

--. 2010. "Supporting Work: A Proposal for Modernizing the U.S. Disability Insurance System." Washington: Center for American Progress and Hamilton Project.

Borghans, Lex, Anne C. Gielen, and Erzo F. P. Luttmer. 2014. "Social Support Substitution and the Earnings Rebound: Evidence from a Regression Discontinuity in Disability Insurance Reform." American Economic Journal: Economic Policy 6, no. 4: 34-70.

Burkhauser, Richard V., and Mary C. Daly. 2012. "Social Security Disability Insurance: Time for Fundamental Change." Journal of Policy Analysis and Management 31, no. 2: 454-61.

Dahl, Gordon B., and Anne C. Gielen. 2018. "Intergenerational Spillovers in Disability Insurance." Working Paper no. 24296. Cambridge, Mass.: National Bureau of Economic Research.

Dahl, Gordon B., Andreas Ravndal Kostol, and Magne Mogstad. 2014. "Family Welfare Cultures." Quarterly Journal of Economics 129, no. 4: 1711-52.

de Hann, Monique, and Ragnhild Camilla Schreiner. 2017. "The Intergenerational Transmission of Welfare Dependency." Working paper. https://sites.google.com/site/ragnhildschreiner/home/papers

Finkelstein, Amy, Nathaniel Hendren, and Erzo F. P. Luttmer. 2015. "The Value of Medicaid: Interpreting Results from the Oregon Health Insurance Experiment." Working Paper no. 21308. Cambridge, Mass.: National Bureau of Economic Research.

French, Eric, and Jae Song. 2014. "The Effect of Disability Insurance Receipt on Labor Supply." American Economic Journal: Economic Policy 6, no. 2: 291-337.

Garfield, Rachel, and Anthony Damico. 2017. "The Coverage Gap: Uninsured Poor Adults in States That Do Not Expand Medicaid." Issue Brief. Menlo Park, Calif.: Kaiser Family Foundation.

Garfield, Rachel, Robbin Rudowitz, and Anthony Damico. 2018. "Understanding the Intersection of Medicaid and Work." Issue Brief. Menlo Park, Calif.: Kaiser Family Foundation.

Goldstein, Amy. 2018. "Kentucky Becomes the First State Allowed to Impose Medicaid Work Requirement." Washington Post, January 12.

Hartley, Robert Paul, Carlos Lamarche, and James P. Ziliak. 2017. "Welfare Reform and the Intergenerational Transmission of Dependence." Discussion Paper no. 10942. Bonn: Institute of Labor Economics (IZA).

Kostol, Andreas Ravndal, and Magne Mogstad. 2014. "How Financial Incentives Induce Disability Insurance Recipients to Return to Work." American Economic Review 104, no. 2: 624-55.

Maestas, Nicole, Kathleen J. Mullen, and Alexander Strand. 2013. "Does Disability Insurance Receipt Discourage Work? Using Examiner Assignment to

Estimate Causal Effects of SSDI Receipt." American Economic Review 103, no. 5: 1797-829.

Sommers, Benjamin D. 2006. "Insuring Children or Insuring Families: Do Parental and Sibling Coverage Lead to Improved Retention of Children in Medicaid and CHIP?" Journal of Health Economics 25, no. 6: 1154-69.

(1.) I am grateful to my colleagues Jeff Clemens, Julie Cullen, and Roger Gordon for helpful discussions and suggestions.

(2.) Thanks to Hoynes and Schanzenbach for sharing their figure 9 with me. Program parameters for the Medicaid, CHIP, and PTC programs come from HealthCare.gov and the Kaiser Family Foundation.

(3.) For example, Finkelstein, Hendren, and Luttmer (2015) find individuals value Medicaid benefits between $0.20 and $0.40 per $1 of government spending, perhaps in part because the counterfactual is often not a complete lack of medical care but care from other sources, such as emergency rooms.

(4.) Although work requirements are generally waived for caregivers of young children, a work requirement would still affect a couple's work incentives.

(5.) There are some existing incentives for participants to exit DI and return to work. For example, participants can earn more money during a "trial work period" for Social Security Disability Insurance, but not Supplemental Security Income. Moreover, programs like the Social Security Ticket to Work program provide resources such as vocational training.

GENERAL DISCUSSION Robert Moffitt complimented the authors for bringing to bear new data on expenditures on children. He had two comments. First, he noted that the paper has two distinct parts: The first documents new evidence on the effects of transfers on children, and the second explores how the distribution of transfers has changed over time. He asked what the second part implies about the first--that is, given that transfers have benefited children, what does the change in the distribution of transfers imply about which programs should be expanded? For example, should we try to redesign programs to focus on the lowest-income families instead of those with slightly higher incomes? Second, he referred to work by Janet Currie showing that cash transfers do not have the same impact as transfers targeted specifically at children. (1) He wondered if it would be best to focus on programs like preschool education and the School Breakfast Program, which are more specific to children than cash transfers to families.

Katharine Abraham noted that certain programs not mentioned in the authors' literature review also have been shown to have an impact on outcomes for children. In particular, a recent paper by Fredrik Andersson and colleagues examines the long-term effects of growing up in public housing or receiving a housing voucher. (2) Abraham also drew attention to the present paper's findings on divergent trends in spending on children and the elderly, noting that, although there are strong political economy reasons to have universal assistance programs for the elderly, it would be interesting to know more about the incomes of elderly households receiving assistance.

Jeffrey Campbell asked about the complementarity of parental ability and public assistance. If more effective parents are able to put public resources to better use, then there may be some justification for moving the resources up the income distribution.

N. Gregory Mankiw noted that because the number of people in each section of the income distribution is changing, changes in the shares of benefits going to different segments of the income distribution are difficult to interpret. Mankiw also mentioned that he would be interested in hearing the authors' views on a universal basic income (UBI). Though freely admitting that a UBI was in no way politically feasible in 2018, he wondered how the kind of UBI conceptualized by Milton Friedman--or, more recently, by Chris Hughes--would compare with programs that already exist. (3)

Alice Rivlin mentioned that a common perception among those in the general public who oppose the social safety net is that they are "the hardworking folks who are supporting these lazy people." The present paper, she said, offers two messages about this perception. The first one, which should be reassuring to those who oppose public assistance programs, is that assistance has shifted toward working families. The second one, however, is that as income increases, it is very difficult to know what the work incentives are. In his comment, Gordon Dahl documented the seemingly impenetrable structure of work incentives in North Carolina. Rivlin asked what the paper's authors and other experts would do to make work incentives more sensible--suggesting, as Dahl did, that one option is to combine programs. The downside, she said, is that doing so would likely result in less money being allocated to the programs.

Isabel Sawhill praised the paper as "a great synthesis of the research and wonderful data," but she expressed concern that some of its findings on programs such as the Supplemental Nutrition Assistance Program are based on data that started being collected in the 1960s. An effect that occurred 40 or 50 years ago may not hold true today because of changes in contextual factors. For instance, malnutrition was more widespread and education was less ubiquitous in the 1960s than in 2018. Her preference is to use data on more recent cohorts of children from randomized controlled trials when available, or otherwise from quasi-experimental studies.

Picking up Rivlin's point about the shift in benefits toward working families, Sawhill remarked that she was not sure of the authors' normative position on this trend. She added that many of the families receiving assistance are probably female-headed, and that there had been a major change in female labor force participation over the last several decades--due in part to welfare reform, but mostly to changes in norms and opportunities for women. Finally, she was glad that the authors had focused on an "investment framework"--that is, on assistance programs as investments--but she cautioned that many of these programs may not be able to compete with other kinds of investment programs. This could be a reason to focus on motivations rooted in humanitarianism and fairness, she concluded.

David Autor drew the conversation back to work incentives, saying that one distinction between programs targeted at the young and those targeted at the elderly is that for the elderly, there is no danger of substitution away from work as a result of transfers. He noted that the transfer programs discussed in the paper were also, in a sense, labor market programs. Labor market shocks feed more strongly into social safety net programs than into intended labor market programs. For example, a trade shock in a local labor market will exacerbate a larger uptake of disability insurance, Medicare, Medicaid, and other transfer programs than of unemployment insurance or trade adjustment assistance. Just as social safety nets are ultimately forced to respond to changes in labor market conditions, labor market incentives are affected by public transfers, he concluded.

Kent Smetters pointed out that the median voter model could predict that transfers would increasingly go to the elderly, because the median voter will eventually become elderly but "is not going to be young someday." And yet, he said, $1 spent on Social Security is not $1 taken away from youth. Rather, Social Security is a "pay-as-you-go game."

Michael Klein asked whether it is possible to compare trends in the distribution of transfers to children and the elderly across countries. If so, he suggested that it may be worth looking into the political reasons for differences between the United States and other countries.

Richard Cooper agreed with Klein, stating that the paper "cries out for international comparison." He mentioned Canada and Sweden as potential comparisons. He noted that the Copenhagen Consensus asked panelists to choose from a long list of international public goods that they would like to fund, assuming they had $75 billion to spend over the next five years. The first choice was reducing child malnutrition, and the second was reducing childhood diseases. These priorities seemed consistent with those laid out in the present paper. Cooper also reinforced the point made by both Dahl and Currie that the "full picture" must also pay attention to state and local spending (though Currie noted that obtaining data on state and local programs would be a "daunting challenge"). For instance, though special education does not feature much in federal spending, Cooper noted that in Cambridge, Massachusetts, special education is roughly a quarter of the school system's budget.

Finally, Cooper objected to the authors' lumping together of Social Security and Medicare with the other safety net programs. Social Security and Medicare are part of a "social contract," whereby workers pay into the system during their lifetimes and receive the returns to the investment down the road. He thinks the paper fails to acknowledge the difference between public expenditures financed by dedicated taxes and those financed by general revenues.

Schanzenbach first addressed the point made by some macroeconomists earlier in the writing process (and by Mankiw earlier in this discussion) that the authors should express spending in terms of shares of the population living within a certain range of the poverty level. Doing so would require relying primarily on data from the Current Population Survey (CPS), which is well known to be plagued by measurement errors. The authors went through the tedious process of obtaining administrative data primarily to avoid using the CPS, though CPS-based calculations are included in their online appendix. Additionally, not much research exists on many of the policy questions in which the authors are most interested--for example, whether $1 is better spent on the group living at between 0 and 50 percent of the poverty threshold or on those living between 100 and 150 percent, or on which program, or on people whose parents have high cognitive abilities, as Campbell had suggested.

On the question of work disincentives, Schanzenbach pointed out that the paper discusses well-identified studies that have found the work disincentive effects for programs like the Supplemental Nutrition Assistance Program to be very small. Although this does not mean that doubling the safety net might not produce a larger effect, current research suggests that the programs discussed in the paper carry minimal work disincentives. On Mankiw's question about a UBI, she noted that between the Supplemental Nutrition Assistance Program and the Earned Income Tax Credit, the current social safety net is similar to Friedman's negative income tax.

(1.) Janet Currie, "Welfare and the Well-Being of Children: The Relative Effectiveness of Cash and In-Kind Transfers," Tax Policy and the Economy 8 (1994): 1-44; Janet M. Currie, The Invisible Safety Net: Protecting the Nation's Poor Children and Families (Princeton University Press, 2008).

(2.) Fredrik Andersson, John C. Haltiwanger, Mark J. Kutzbach, Giordano E. Palloni, Henry O. Pollakowski, and Daniel H. Weinberg, "Childhood Housing and Adult Earnings: A Between-Siblings Analysis of Housing Vouchers and Public Housing." Working Paper no. 22721 (Cambridge, Mass.: National Bureau of Economic Research, 2016).

(3.) Milton Friedman advocated his notion of a "negative income tax," which is conceptually similar to a universal basic income, in his book Capitalism and Freedom (University of Chicago Press, 1962). See also Chris Hughes, Fair Shot: Rethinking Inequality and How We Earn (New York: St. Martin's Press, 2018).
COPYRIGHT 2018 Brookings Institution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Brookings Papers on Economic Activity
Date:Mar 22, 2018
Words:7206
Previous Article:Safety Net Investments in Children.
Next Article:Jobs for the Heartland: Place-Based Policies in 21st-century America.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters