A key issue in economic policy is determining which goods or services the government should provide —either by producing them directly or by funding others to do so. Traditional economic theory suggests that, if markets are functioning properly, competitive market forces will generate efficient provision of goods without intervention from the government. But the conditions required for this efficiency result are quite strong, and there are a great many goods and services to which they do not apply. When they do not, there is no a priori reason to think that welfare will be higher when the good is left to the private market than when it is publicly provided.
We argue for a larger public role in the provision of two categories of goods that long experience has shown are drastically under-provided by private markets, for which theory and evidence each clearly indicate that public provision would improve welfare. Government should do more to support families in the raising and educating of children. It should also play a larger role in insuring against certain types of risks that individuals and families face.
We first discuss childhood investments, including expenses relating to child care and early education as well as post-secondary education. A key characteristic of these investments is that the costs are lumpy and are, to a large extent, incurred by parents, though most of the benefits accrue to the children. These services are also very expensive, are tightly concentrated in time, and in many cases occur decades before the benefits are realized. In addition, as we discuss below, research has shown that there are many public externalities associated with these investments. Efficient private provision, even assuming agency problems and externalities could be addressed, requires borrowing large amounts against the child’s future earnings. Such loans are not available on the private market, and even when the government can create a market for them (as for student loans), issues of asymmetric information create moral hazard problems in the absence of stringent government regulation. Together, these three issues ensure that without a larger role for government in providing and/or funding these services we will see dramatic underinvestment in child development.
The second category we discuss is insurance. There are many types of insurance that are efficiently provided by the (often carefully regulated) private market, such as car or homeowners’ insurance. However, other types of insurance that individuals would highly value either are not available at all on the private market or are available only at extremely high prices to a small share of the group that could benefit. These include insurance against job loss in recessions, against illness, against outliving one’s savings, and against long-term care expenses. We focus on these because these risks detract importantly from individual welfare, and, in the absence of insurance, lead to quite costly responses such as over-saving for retirement and going without needed health care. It is not likely that private insurance markets can be made to function well in these areas, even with aggressive regulation. Instead, these are natural candidates for public provision.
We want to emphasize that we are not the first to propose public provision of the types of services considered here. Indeed, many childhood investments are already publicly provided. Most obviously, we provide free public K-12 education, and most college students attend public institutions, albeit often with substantial and growing tuition bills. Similarly, we also provide public insurance against unemployment, longevity (via Social Security), and unforeseen medical costs. Our purpose is merely to articulate a common intellectual justification for these programs, and to advocate for expanding them to cover important needs that are not currently covered.
Child care and education
It is increasingly recognized that high-quality child care and early childhood education, prior to entering kindergarten, is a key investment with important implications for children’s long-run outcomes (e.g., Elango et al. 2016; Deming 2009). This investment has important impacts on others – a more educated child may benefit the rest of society through reduced reliance on public support, productivity spillovers, and reduced criminal activity – and one can make a strong case for public intervention on the basis of these externalities alone.
But even if we set aside potential impacts beyond the immediate family, there is little reason to expect private decisions to be optimal. The decision to purchase child care and early childhood education, along with the burden of paying for them, rests on parents, while it is the children’s futures that are at stake. This type of principal-agent problem could generate inefficient investment – parents may invest less in children’s education than the children themselves would, could their future selves be allowed to control the decision and bear the costs (Brown et al., 2012). Society has addressed this issue for children between the ages of 6 and 18 through the funding and provision of public education. However, public investments are heavily tilted toward older children: President Obama’s Council of Economic Advisers estimated that, in 2015, combined annual local, state, and federal expenditure per child was 63% higher for those between 6 and 11 than for those between 3 and 5 (Council of Economic Advisers 2016b).
Importantly, this early childhood period also coincides with a period when parents’ own earnings are lower and they are thus least able to afford the substantial expenses that young children bring. Because the benefits of investments in child development are not easily collateralized (as they appear through better longer-run outcomes far in the future, when the children are grown, and as children’s own future earnings cannot be encumbered), it may be difficult for parents to borrow to finance this investment, even if they wanted to (Caucutt and Lochner 2012).
These factors point to a need for government intervention. A range of existing programs help parents when children are young—the Head Start pre-school program is one example (Deming 2009) —but these programs are relatively small and tightly targeted to the very poor. The high expenses of early childhood are a burden not just for the poor but also for middle class families, who are at similar risk of under-investment. Broader based support, beginning with funding for high quality childcare and pre-K, would help ease this burden and ensure that children receive the appropriate investments when they are small. This funding should be accompanied by careful and thorough quality regulation, as research has shown the important benefits of high quality childcare and pre-school; lower quality programs are less effective. (Council of Economic Advisers 2016a, Chapter 4)
Higher education presents its own set of challenges. Children are much more involved in their own postsecondary education decisions, which at least partially aligns the benefits with the decision-making. However, financing this investment remains difficult—again, as the benefits are realized far in the future in terms of higher earnings for the students, private markets have trouble funding the loan without extensive government involvement. This, combined with evidence of positive externalities generated by education (Moretti 2004), suggests a need for government intervention.
We currently support a credit market for higher education by guaranteeing student loans and enforcing repayment (for example, by making the loans non-dischargeable in bankruptcy). But this creates its own problems. Institutions vary widely in quality in ways that are hard for students to perceive, and many institutions take loans on students’ behalf without providing education of commensurate quality (Deming et al. 2012). The inability of the government to effectively regulate quality has led to what some have termed a student loan crisis—student debt ballooned during and after the Great Recession, in large part due to increasing attendance at low-quality institutions that are unlikely to yield a positive return.
With debt financing, students bear the risk of regulatory failures, as they must repay their loans even if the institution they attended turns out not to have helped them succeed. There is little indication that this risk leads to better decisions, given the lack of information available to students. Welfare could be increased if the government simply insured against poor outcomes by reducing the costs of a college education. This could take any of a number of forms, ranging from increased public provision through growth of the public higher education sector (with restrained or eliminated tuition made up through additional investment of tax revenue) to larger public grant programs built on the existing Pell Grant. Given the evidence that much of the student debt problem is concentrated among individuals who never completed their degrees (CEA 2016c), one attractive solution is to make the first two years of college free. In addition to removing enormous leverage and risk from individual portfolios, this would also resolve an agency problem, creating appropriate incentives for government to regulate quality aggressively.
Risk is a fundamental part of life and can have major welfare consequences. Without insurance, even prosperous families would lack economic security, as they could risk being thrown into poverty at any time. Insurance can protect them from this risk and thereby promote security.
Many private insurance markets – for example, auto insurance, life insurance, and homeowners insurance –work fairly well, in part due to extensive regulation. But people face many risks for which there is no private insurance market. Economists have long studied the information and other failures that prevent well-functioning insurance markets. There are many types of market failures, but three are common:
- Moral hazard. People with insurance are cushioned against the negative consequences of bad outcomes, and may not act prudently to minimize their risk if someone else will bear the cost. Someone with car insurance may be more willing to park a car in a dangerous neighborhood, raising the risk that it could be stolen. This raises the likelihood of damage, making it very expensive to insure truly unavoidable risks.
- Heterogeneity in risk and adverse selection. Risk often varies across people in predictable ways – some may have a genetic propensity toward a particular disease, for example. When the potential purchaser of insurance has more information about his or her type than the insurer, the insurer must price insurance under the assumption that the consumer is of the high-risk type, setting the premium very high. This makes it difficult or impossible for those with low risk to buy insurance at any reasonable price, leaving many people uninsured. Even when risk type is verifiable, so that the insurer is able to sell insurance to both types, those with above-average risks can face very high prices, which may price them out of the market and reduce overall welfare. When insurance markets do exist, insurers have strong incentives to design policies narrowly, excluding preexisting conditions or denying coverage after the fact for costs that the individual had hoped to have covered, leaving the insured with less protection against risk than he or she would like.
- Common/correlated risks. The insurance business relies on risks being random but total cost being predictable: While we can’t know whose house will burn down in the coming year, we can estimate how many fires there will be, and the fire insurer can thus plan on predictable expenses. Some risks, however, are not idiosyncratic, but shared: Although a major earthquake may occur only once every 50 years, when it does a large share of policyholders will file claims. If the insurer does not maintain a large reserve fund, it will not have enough funds to pay claims when they arrive. Much insurance regulation centers around managing this problem, often unsuccessfully (Sjostrom 2009; Coffee Jr. 2011). The problem is much worse for risks associated with the business cycle, such as unemployment, where payout events tend to occur when capital is hardest to obtain, and this type of insurance is very expensive to provide.
These pathologies can cause insurance markets to break down, leaving many people uninsured. But it is important to remember that their presence, and the resulting absence of a private insurance market, is not a signal that the welfare value of insurance is low; people would prefer to insure themselves against consequential risk, but they find themselves unable to do so at a fair price.
Where adverse selection or correlated risks cause insurance markets to fail, there is a role for the government. By mandating the purchase of insurance at a price that is fair on average, or by providing the insurance out of general revenues, the government can generate risk coverage despite information asymmetries that would otherwise cause insurance markets to fail due to adverse selection. Similarly, the federal treasury has unique access to credit markets during recessions, when it typically faces very low – often zero – borrowing rates. This enables it to insure business cycle risks that private insurers cannot cover. (The government generally has no special ability to manage moral hazard, so where that is the primary source of failure there may not be a public role.)
These arguments are not new – there has long been a recognized role for the public sector in the provision of social insurance (Hacker, 2008). An example is disability insurance: those who become disabled often lose their livelihoods, but information asymmetries make it very difficult to purchase disability insurance on private markets. Where insurance is available, underwriters often require extensive medical exams, and refuse to write insurance for those with preexisting conditions. Through the Social Security Disability Insurance and Supplemental Security Insurance programs, the government provides insurance to all, providing monthly income to the disabled through (in the case of SSDI) premiums levied on the healthy. This necessarily entails some transfer from the healthy to the sick, which may be socially desirable in its own right. But it also enables everyone to obtain protection from a very serious risk to their livelihood, protection that would not otherwise be available.
But the scope of existing social insurance programs is tightly circumscribed. There are many important risks that reduce welfare that are not covered, even though the insurance market failures are similar to those that are. An expanded scope would be welfare improving (Hacker 2008). We review several leading candidates for new or expanded social insurance coverage below.
First, however, we address a common counterargument. It is often suggested that moral hazard makes it unwise to expand social insurance: With coverage, people will take on risks that they would not otherwise. There are two responses to this. First, in several of the cases that we discuss below, the degree of risk is well outside the individual’s control, so there is no scope for moral hazard. We are not concerned that longevity insurance will lead people to live longer than is efficient, for example, or that business cycle insurance for workers will lead them to court recessions. Second, and most importantly: There is nothing specific to social insurance that creates or magnifies moral hazard, which is equally likely to afflict holders of private insurance policies. This may limit the extent of insurance that it is prudent to provide, but absent some reason to expect moral hazard to be a particular problem in a case under consideration, it does not constitute an argument against the provision of insurance.
We next discuss several specific risks that are presently difficult to insure, where a new or expanded government role would increase welfare.
Unemployment insurance. A worker who loses her job will need to finance consumption, often for an extended time and generally with very limited ability to borrow against future income, until she finds a new job. In the absence of insurance, workers must maintain large savings against this possibility. There is little efficiency benefit of forcing people to bear the portion of this risk that reflects changes in their employers’ prospects outside of their control. Accordingly, a joint federal-state program has since 1935 provided unemployment insurance to workers, financed by payroll taxes. To limit coverage to the true risks and avoid moral hazard, payments are limited to those laid off from their jobs; those who quit or are fired for cause are generally not eligible. Another measure taken to limit moral hazard is time limits on benefits, typically 26 weeks, which are meant to encourage aggressive job search before the benefits run out. The idea here is to balance moral hazard against the need for insurance (Bialy 1978; Chetty 2008; Schmieder and von Wachter 2016).
This insurance is quite valuable to workers. But the uniform program described above does not address systematic differences in job availability. The chance that someone will be able to find a job by the end of 26 weeks of benefits, even with diligent search, varies enormously over the business cycle, with much higher rates of benefit exhaustion in recessions. The federal government often extends benefits in recessions, but these extensions are ad hoc and often come too late to be helpful. Expanding the existing system to automatically extend benefits when the economy weakens could significantly improve worker welfare (Landais et al., 2018a,b).
Such a policy offers two other benefits, beyond the basic risk protection. First, unemployment benefits act as Keynesian demand stabilizers, boosting consumption among those with high propensities to spend. Automatic extensions would ensure that that boost arrives when the economy needs the additional demand, not afterward. Second, the moral hazard argument for encouraging active job search by limiting the duration of benefits is much attenuated in recessions, when there are more job searchers than jobs and there would be little efficiency cost to reduced search effort among the unemployed.
Old age insurance. Someone who reaches old age without adequate savings has no good options. At that point, it is too late to go back to work, so the only choices are to sharply reduce consumption or to rely on transfers from family members.
If lifespans and investment returns were predictable, the private solution would be straightforward: People would set aside money during their working lives to consume during retirement. However, the unpredictability of lifespans and of investment returns creates substantial risk, and the only way people can protect themselves is by saving much more than they will likely need. This precautionary savings reduces welfare (Lusardi 1988; Abel 1995; Hubbard 1987; Kotlikoff et al. 1986; Mitchell et al. 1999).
There is little efficiency benefit of making people bear either of these risks. People would be much better off with insurance that guaranteed them a stable income as long as they lived, and there is no real moral hazard problem here. Annuities are financial products that provide insurance against these risks. However, annuity markets generally function poorly, for reasons that are not entirely understood but include adverse selection and the substantial complexity of the products on offer (Brown et al. 2008; Finkelstein and Poterba 2004; Brown et al. 2008). Annuities are generally priced well above their actuarial value, and few people buy them (Lockwood 2012; Mitchell et al. 1999).
The resulting market failure creates a clear public need that has long been recognized. Since 1935, Social Security Retirement Insurance has provided a mandatory retirement annuity to all American workers, with some modest redistribution from high- to low-earners and from younger to older cohorts. But Social Security was designed to be just one part of what was intended as a three-legged stool, with private pensions and individual savings providing the other legs. In the last decades, one of these legs has nearly disappeared – only one-fifth of full-time private-sector workers are covered by a defined-benefit pension. And the last leg never functioned well – less than 60% of those approaching retirement have any retirement savings, and over half of those have less than $100,000 saved – not enough to last through a long retirement. Those who do save still face uncovered risks from financial market volatility as well as the risk of outliving their savings.
Given these failures, there is a strong case for expanding the Social Security annuity, growing it to cover a much larger share of expected retirement consumption. Again, this would not distort private decision-making, except for those few who would prefer to dramatically reduce their consumption in retirement. To recognize this heterogeneity, one might combine an expansion of required Social Security with an optional public annuity, structured as an option to top up one’s Social Security benefits through voluntary additional contributions. This would be a much less risky way to save for retirement than via 401(k)s, and with even modest take-up would be welfare improving.
Health insurance. Medical care absorbs an ever-growing share of national expenditures, due in large part to incredible progress in medical science that makes it possible to treat a wide range of ailments that were previously untreatable, though often at a high cost. This makes health insurance essential to economic security. However, health insurance markets rarely work well, for two reasons: First, information asymmetries and adverse selection are particularly salient in health insurance, and insurance companies often deny insurance to those at high risk (for example, those with preexisting conditions). Second, the correct care is not easily observable and often depends on doctor discretion, creating a principal-agent problem that leads to extensive controls on care usage, frequent denial of coverage for legitimate claims, and enormous hassles for the policyholder. The end result is that, absent very strict regulation, many go without insurance altogether, and those who have insurance still face large risks of being bankrupted by charges that the insurance company denies (Swartz 2003; Himmelstein et al. 2000, 2009; Warren et al., 2000). In addition to the missing markets for those with the highest risk, there is another important case for government insurance. Health insurance and health care generates positive externalities to society, in that healthier citizens are more productive and those who do not have health insurance may ultimately depend on safety net programs when they get sick (Finkelstein et al. 2018; Brown et al. 2015).
We have already created a successful public health insurance program for the elderly, the hardest group to cover due to the high and predictable costs of aging. For those who do not yet qualify for Medicare, we rely primarily on private insurance markets, which must be extensively regulated. Regulators often fail to keep the markets functioning well, and even with careful regulation many remain uncovered. Public provision of health insurance would dramatically reduce bureaucracy and transaction costs in health care while ensuring universal coverage.
Long-term care insurance Related to health insurance is long-term care insurance. Many people need labor-intensive personal care at the end of their lives, and following major illnesses. This is a major expense that arrives when people are least able to accommodate it. In principle, healthy people could save against possible future long term care needs, but the range of possible outcomes is so enormous, and the variability so high, that few do this. Instead, people fall back on public welfare programs – stories abound of people intentionally exhausting their assets to qualify for Medicaid – or on their families, who are often called upon to provide enormous amounts of uncompensated care for which they are not well qualified.
Insurance against this major risk of old age would be extremely valuable and would not create serious moral hazard problems (Cohen et al. 2018). Private insurance markets exist, but are complex, hard to understand, and badly undersubscribed. Adverse selection seems to combine with individual optimization failures – people prefer to wait until they are sick to purchase long term care insurance, but at that point it is unaffordable (Brown and Finkelstein 2007, 2009; Brown et al. 2008). The existence of Medicaid as a fallback option also makes it more feasible to go without this type of insurance, though the low quality of Medicaid care leaves a large uncovered risk (Brown and Finkelstein 2008, 2011). Finally, this type of insurance creates serious time consistency problems, similar to but more severe than those in traditional health insurance: Insurers have strong incentives to collect premiums from the healthy, then deny coverage and/or change the terms of coverage as people approach the need to make a claim.
The net effect is that people face large uncovered risks, while the existence of the public program for the poor, via Medicaid, means that the government still winds up bearing much of the cost. Providing higher quality, universal public coverage, without requiring recipients to exhaust all assets before using it, would be expensive, but would much improve welfare (Cohen et al. 2018).
 This discussion focuses on efficiency rather than distribution. Concerns about inequality may provide their own motivation for government intervention, even in the absence of market failures.
 A version of this was recently proposed by Congressman Bobby Scott of Virginia with his “Aim Higher” proposal that would include two years of free community college.
 In addition to creating an opportunity to obtain insurance where none otherwise exists, government-provided insurance can also avoid many of the pathologies of private insurance in the presence of adverse selection, such as aggressive underwriting, preexisting condition exclusions, and ex post denial of apparently valid claims.
 Another possible form of insurance against job loss would be through wage insurance. Evidence suggests that individuals who lose their jobs often suffer wage losses in their next jobs (Davis and von Wachter, 2012). Wage insurance would offset some of these losses. Wage insurance was included in President Obama’s 2017 Budget Proposal.
 There was an attempt to do this via the Affordable Care Act, but the program was under-resourced and was eventually abandoned. The problem has not gone away, and welfare would be improved if we did this right.