Macroblog

June 29, 2016

Pay As You Go: Yes or No?

The Atlanta Fed's 2015 Annual Report focused on the graying of the U.S. economy. Part of the report and a follow-up webcast discussed how aging is driving the depletion of the U.S. Social Security and Medicare trust funds.

Based on current projections from the Congressional Budget Office, the Social Security trust fund is forecast to run dry around 2030 (see the chart); the Medicare trust fund in 2025. Barring a change in legislation, once the trust funds are depleted, benefits will be cut so that outlays match revenues. In the case of Social Security, this reduction will mean a 31 percent decline in benefits. To balance the Medicare budget, certain Medicare benefits will also face significant reduction.

As my coauthors and I explain in a recent Oxford University Press blog post, our research has found that pay-as-you-go programs for retirees such as Social Security and Medicare, on average, make people worse off, whereas means-tested social insurance programs for retirees, such as Medicaid and Supplemental Security Income (SSI), improve welfare.

These findings are based on comparing the welfare of individuals born into economies with different types of social insurance programs available. We find that, given the hypothetical choice between having or not having Social Security, the average individual would choose to be born into an economy without Social Security. However, when we ask if an average individual would prefer to be born into an economy with or without means-tested retiree programs, we find that he or she would strongly prefer the economy with these programs.

The preference for an economy without universal pay-as-you-go programs like Social Security is consistent with findings in the literature more generally. These programs are large (Social Security was 4.9 percent of U.S. gross domestic product [GDP] in 2013) and have distortionary effects. In standard economic models, the distortions lead to such large reductions in savings and labor supply that they tend to outweigh the programs' insurance benefits.

In contrast, means-tested social insurance programs for retirees, such as Medicaid and SSI, are much smaller. Together, outlays from these programs for the elderly were only 1 percent of GDP in 2013. These programs provide transfers only to individuals with limited income and assets or with impoverishing medical expenses. However, it is in these states of world, when one is poor and/or sick, that such transfers are most valuable, which is why we find that these programs improve welfare.

Researchers have found that means-tested transfer programs for working-age individuals are highly distortionary because they implicitly tax income and assets. However, we find that such distortions are less severe for means-tested transfer programs for retirees, since individuals cannot use these programs to finance working-age consumption and medical care.

Our findings suggest that one potential solution to the sustainability problems plaguing Social Security and Medicare may be to make these programs means-tested as well. Under such a system, the government would still provide protection against the risks of ending up old, sick, alone or poor, but with programs that are significantly less costly.

Of course, saying that individuals would prefer to be born into a hypothetical economy A instead of economy B is not the same thing as saying that current U.S. citizens want to make such a transition. Moving from the current system to one in which Social Security and Medicare benefits are means-tested would not be attractive to wealthier individuals who are already retired or on the verge of it. A compensation scheme would likely have to be devised and financed through taxes or government debt.

Once the cost of compensation is taken into account, we may find that such a transition is too costly to undertake. And as the population ages and the ratio of retirees to working-age individuals increases, the fraction of individuals in the economy who need to be compensated will increase further. This reality adds impetus to dealing with the Social Security sustainability issue sooner rather than later.

By Karen Kopecky, research economist and associate adviser in the Atlanta Fed's research department

• June 29, 2016 in
  • Social Security
• Permalink
• Comments

November 13, 2014

A Closer Look at Employment and Social Insurance

The Atlanta Fed's Center for Human Capital Studies hosted its annual employment conference on October 2–3, 2014, organized once again by Richard Rogerson of Princeton University, Robert Shimer of the University of Chicago, and the Atlanta Fed's Melinda Pitts. This macroblog post summarizes some of the discussions.

Social insurance programs in the United States and other developed countries represent a large and growing share of expenditures relative to gross domestic product (GDP). Assessing the costs and benefits of the diverse programs that make up the U.S. social insurance system is a key input into the design and implementation of effective programs. This conference featured seven papers that dealt with various aspects of this assessment. Although each program is designed to address specific issues and hence needs to be studied in the context of those issues, many of the same basic economic questions arise in each context. For example, what is the rationale for social insurance programs? Do they address inefficiencies, or are they mainly designed to redistribute from one group to another? Who benefits from specific programs? How do programs designed to achieve specific objectives distort economic outcomes? These are the questions that featured prominently in the conference.

A classic question in economics concerns the extent to which markets cannot achieve efficient outcomes without government intervention. It is well known that the so-called "invisible hand" can achieve efficient outcomes in a wide range of standard settings, but do these results extend to situations in which information asymmetries exist? In 1976, Michael Rothschild and Joseph Stiglitz's article "Equilibrium in Competitive Insurance Markets" suggested that in the presence of certain kinds of private information, insurance markets could not achieve efficient allocations. In fact, they argued that competitive equilibrium might not even exist in these settings. In "Adverse Selection Is Not a Justification for Social Insurance," Ed Prescott challenges this result and shows that competitive equilibrium exists and achieves efficient allocations in settings that include information problems such as Rothschild and Stiglitz's adverse selection problem. Key to this result is the presence of mutual insurance companies, and how this presence influences the contracts offered by insurance companies in equilibrium. In the Rothschild and Stiglitz environment, insurance companies were effectively agents with deep pockets that were outside the model.

Providing insurance to individuals in situations in which they face bad outcomes may distort individual behavior and lead to negative outcomes that outweigh the benefits of the insurance. This basic issue was addressed by three of the papers at the conference in three separate contexts. Jason Abaluck, Jonathan Gruber, and Ashley Swanson examined how prescription drug coverage through Medicare influences prescription drug usage; Hamish Low and Luigi Pistaferri studied the disability insurance (DI) system; and Bradley Heim, Ithai Lurie, and Kosali Simon examined whether the extension of health benefits to young adults as mandated by the Affordable Care Act (ACA) influenced the behavior of young adults.

In "Prescription Drug Use Under Medicare Part D: A Linear Model of Non-linear Budget Sets," Jason Abaluck, Jonathan Gruber, and Ashley Swanson study how prescription drug use responds to price changes associated with social insurance through Medicare. At the conference, Gruber discussed one key objective of their analysis: uncovering the elasticity of prescription drug use with respect to price. A large elasticity implies that providing insurance in the form of lower prices will distort behavior and lead to much higher drug use, and some recent papers have argued that this elasticity may be quite large. Their basic strategy is to study how changes in the details of Medicare coverage over time influenced individual choices. A novel feature of the estimation strategy is to take advantage of the fact that the marginal price people face depends on their overall annual expenditure on prescriptions, so that individuals can be sorted into groups based on histories of usage, interacted with changes in the details of coverage. A first key finding of this paper is that the elasticity is relatively small. A second key set of findings concerns the extent to which individual choices (in terms of plan selection and yearly expenditure conditional on plan choice) reflect departures from rationality, such as myopia or salience. The paper finds an important role for both of these effects.

Disability insurance (DI) represents a clear and classic example of the tension between insurance provision and insurance. While one would like to provide insurance to individuals who are unable to work, it can be difficult to assess the true ability of an individual to work, thereby creating the opportunity for people who are not disabled to also collect. Luigi Pistaferri addressed this issue in the paper he coauthored with Hamish Low, "Disability Insurance and the Dynamics of the Incentive-Insurance Tradeoff." This paper builds and estimates a structural model that incorporates labor supply, health shocks, earnings shocks, and the key details of the DI application process. The authors conduct various counterfactuals and assess the tension between insurance and incentives in the context of the U.S. DI program. Several results emerge. First, making the review process less strict would enhance welfare despite worsening incentives for people to misreport their health status. This is because the current system denies too many truly disabled individuals from collecting. But decreasing generosity would also increase overall welfare by decreasing the incentives for false collection.

One of the first measures of the Affordable Care Act (ACA) to be enacted was the provision that allowed dependent individuals to remain covered by their parents' healthcare plans until the age of 26. The paper by Bradley Heim, Ithai Lurie, and Kosali Simon, "The Impact of the Affordable Care Act Young Adult Mandate: Evidence from Tax Data," aims to assess the extent to which this provision has affected outcomes for young adults in terms of employment, wages, schooling, and marriage. As Simon described it at the conference, the novel aspect of this analysis is that it tracks outcomes using administrative IRS data, which affords a large sample size. The main empirical strategy is to compare the change in outcomes from before and after the provision was enacted for individuals below the age threshold with the change in outcomes for individuals just above the age threshold. The paper also reports estimates based on triple differencing that uses information on parental health insurance status. The main message from the analysis is that one cannot find robust, statistically significant effects of this ACA provision on outcomes for young individuals. One important qualification is that despite the large sample size, standard errors are still quite large, so that the analysis cannot rule out the possibility of economically significant effects.

Naoki Aizawa and Hanming Fang also considered the effects of the ACA in their paper "Equilibrium Labor Market Search and Health Insurance Reform." However, in contrast to the above papers that focus on how a particular program feature might influence individual choices, this paper focuses on how the creation of health insurance exchanges and the individual insurance mandate would affect the overall equilibrium in the labor market, taking into account the firms' decisions on whether to offer insurance and the wages that they offer to workers. In his presentation, Fang discussed building a structural equilibrium model of the labor market and estimating it using a variety of data sets. The authors find that the ACA will reduce the uninsured rate from about 20 percent to about 7 percent. But interestingly, the paper finds that the uninsured rate would drop even further if the employer mandate were dropped from the ACA. General equilibrium responses are key to understanding this result, illustrating the importance of studying these effects.

One of the rapidly growing social insurance programs is Medicaid. Mariacristina De Nardi, Eric French, and John Bailey Jones assess the benefits of this program in their paper "Medicaid Insurance in Old Age." As French described at the conference, this paper uses a structural approach to assess the extent to which households with different income and health status benefit from Medicaid. The analysis focuses on individuals from age 70 and forward using data from the Health and Retirement Study, emphasizing the risks that individuals face as a result of health shocks. Medicaid offers partial insurance against these shocks, particularly the large expenditures associated with nursing home care, and the paper assesses the value of this insurance for individuals in different positions in the wealth distribution at age 70. The paper has two main findings. First, the insurance value of Medicaid is substantial, and decreasing the size of the program would entail large welfare costs in excess of one dollar for every dollar of reduced spending. Second, expanding the size of the program would offer significant insurance value only to wealthy households. The authors conclude that in terms of managing the risks of the elderly, the current scope of Medicaid seems appropriate.

As the above discussion emphasizes, a critical input into the design and assessment of social insurance programs are data that allow us to reliably document the outcomes and groups that the insurance program wishes to help, as well as measure the efficacy of existing programs in achieving desirable outcomes. In the paper "Welfare Programs and Survey Misreporting: Implications for Income, Poverty and Disconnectedness," Bruce Meyer and Nikolas Mittag documented the serious shortcomings of several standard publicly available data sets when it comes to measuring the resources available to the poorer segments of the population. Meyer presented the paper at the conference, and it uses administrative data from New York State that allow them to link income and transfer data, both cash and in-kind, and compare the measures obtained using these administrative data with the measures obtained using data from the Current Population Survey (CPS), which is a standard source for publicly available data on the income distribution. The results are striking. Relative to analysis based on data from the CPS, analysis using administrative data shows better outcomes in terms of inequality and disconnectedness and yield larger effects from existing programs in terms of their ability to affect these outcomes.

Full papers or presentations for most of these papers are available on the Atlanta Fed's website.

By Melinda Pitts, director of the Atlanta Fed's Center for Human Capital Studies, Richard Rogerson of Princeton University, and Robert Shimer of the University of Chicago

• November 13, 2014 in
  • Employment
  • Labor Markets
  • Social Security
• Permalink
• Comments

October 9, 2006

If You Care About The Deficit, You Care About Social Security

Dean Baker has a bone to pick with Ben Bernanke:

The projected increase in Social Security spending is relatively modest over the next 45 years and in fact no larger than it was over the last 45 years. In addition, he also knows that workers have already largely paid for this projected increase in spending, paying a designated Social Security tax that exceeds current needs. The Congressional Budget Office projects that future tax revenue, plus the accumulated surplus over the last quarter century, will be sufficient to pay all scheduled Social Security benefits through the year 2046, with no changes whatsoever.

So, Mr. Bernanke was not being honest when he claims there is a problem with Social Security...

At the risk of being labeled one of those "unified budget types" that keep Angry Bear's pgl angry, I object.  The trick there is the stipulation that "future tax revenue, plus the accumulated surplus will be sufficient to pay all scheduled Social Security benefits."  It is fair enough to say that the Social Security "trust fund" is a promise to workers that the government ought not breach.  It is incorrect to say that it will finance "all scheduled Social Security benefits" in any economically meaningful sense. 

The relevant piece of information is this, from the 2006 report of the Social Security and Medicare Board of Trustees:  "Projected OASDI tax income will begin to fall short of outlays in 2017..."  In other words, the Social Security ceases to be self-financing out of payroll taxes in about 10 years.  Absent an increase in overall tax revenues or a reduction in government spending, the payment of scheduled social security benefits adds to the deficit. If you think that deficits are a problem, then logic compels you to treat the payment of accrued Social Security promises as a problem, and one that will arrive in fairly short order.

Note that the same sort of problem does not apply to a large chunk of the Medicare program.  Again, from the Board of Trustees:

Part B of the SMI Trust Fund, which pays doctors' bills and other outpatient expenses, and the recent Part D, which pays for access to prescription drug coverage, are both projected to remain adequately financed into the indefinite future by operation of current law that automatically sets financing each year to meet next year's expected costs.

Part A of the program, which covers hospitalization costs, remains a problem, of course, and it is big -- about 1/2 of all Medicare outlays.  And you might reasonably argue that the increasing share of medical expenditures in both government expenditures and GDP is worrisome. Though I think this subject to some dispute, I'm not inclined to object too vehemently. But I just don't buy the argument that this is reason for ignoring the very real imbalance that exists in the Social Security system.

Several bloggers I admire have consistently argued that, given the benefit promises they imply, it would be a very good thing to not commingle Social Security taxes with other sources of federal revenues.  pgl is in that group.  So is Calculated Risk and Andrew Samwick.  In the name of transparency, you can put me on that list as well.  But unless you harbor pretty firm Ricardian views -- in which case you believe that any discussion of the deficit per se is fundamentally off-topic -- the relevant economic measure is indeed the unified budget.  And for that, the trust funds don't mean a thing.

• October 9, 2006 in
  • Social Security
• Permalink
• Comments

February 8, 2006

This Week In Entitlement Reform

In this the federal budget week, blogland brings a couple of interesting discussions on social security and health policy reform.  First up (in reverse chronological order) is the latest Econoblog installment featuring Mark Thoma and Andrew Samwick.  This, from Andrew, neatly summarizes my thinking about the foundation on which social insurance reform must be built:

As global trade increases, the U.S. loses its ability to be the least-cost producer if it stipulates that employment contracts must include taxes for all manner of redistributive programs. If we are to purse both social insurance and economic growth, we need to consider alternatives to the employment relationship as a way to deal with our health and retirement needs.

Mark agrees, but is skeptical about solutions that rely primarily on the private sector:

There are substantial problems -- market failures -- in the private-sector provision of health and retirement insurance that are not easily overcome with market-based regulatory schemes.

For example, adverse selection issues, where high medical-cost individuals are excluded from coverage or are forced to pay extremely high premiums, plague health-insurance markets. High administrative costs of private health insurance are another problem, and there are problems in the private provision of retirement insurance as well. When markets fail, the insured often pay for the uninsured, and for these and other reasons I believe it's best to share the burden more generally through government programs that require individuals to contribute insurance premiums.

I confess that I don't quite buy that one.  As Andrew points out, there is a distinction between government regulation of an industry and government production of the service that the industry supplies:

The first mistake is to make insurance voluntary when we don't subsequently exclude those who need care from getting it at the public's expense. We should make health insurance mandatory, but we should do so by putting the mandate on the individual, not the employer...

The second mistake is to allow the tax code to distort the type of insurance offered. Premiums are fully excludable from taxation, but out-of-pocket expenses are only imperfectly tax deductible. This generates extremely generous, first-dollar coverage and little incentive for individuals to economize on the care they receive. Rather than the Bush administration's proposal to make out-of-pocket expenses deductible via expanded medical savings accounts, I favor removing the excludability of health-insurance premiums from taxable income.

The third mistake is to force young workers to subsidize older workers in group health-insurance markets. Insurance is supposed to transfer resources from those who have unpredictably low expenses to those who have unpredictably high expenses.

I agree with Mark that the government should mandate coverage, but that doesn't mean the government should centralize the provision of services or dictate their terms. I would prefer to fix some of the obvious mistakes before making such radical changes to the system.

Andrew's diagnosis gets a second from Dr. Becker in this week's installment of the Becker-Posner Blog:

...many of the problems in the health system are correctable with the right policies. I believe the three most important defects are the over 40 million Americans who are not insured, the weak incentives to economize on unnecessary medical spending by most people covered by some form of health insurance, and the tying together of health insurance with employment as a result of special tax privileges provided to employers.

Arguably the best parts of President Bush's State of the Union address are his suggested reforms in the health care system. They do not fully attack all the problems, but they do offer significant improvements. I will concentrate particularly on his proposals to extend Health Savings Accounts (HSAs), and to improve the portability of health insurance when workers change jobs.

Professor Becker goes on to an extensive discussion of the HSA proposal, the benefits of such a plan, and a very wise observation about at least one of the costs:

President Bush has proposed changes in the health care system that initially will reduce tax collections and increase federal spending at a time when the US government is already spending too much and running a sizeable budget deficit. However, by making the health delivery system more efficient, this important set of proposals in the State of the Union address might end up raising tax collections, and certainly would improve the efficiency of the American economy.

There are plenty of other interesting things in these two items -- the discussion, for example, of the Liebman-MacGuineas-Samwick social security reform proposal which I have endorsed (and which was met with a resistance I find as baffling as members of Congress giving themselves a standing ovation for doing absolutely nothing about fixing a system that clearly needs to be fixed).

Really, I beg you.  Read the whole things.

UPDATE: William Polley agrees the key question is "How much social insurance should be provided by the government and how much should be provided by markets." 

There is much relevant discussion at Angry Bear: here, here, and here.  It's fair to say we've taken different sides on this one.

Kevin at Truck and Barter points to research by Alice Zawacki and Amy Taylor on the relationship between employer characteristics and the provision of health insurance benefits. 

You'll find some thoughts about the Posner half of the Becker-Posner conversation at winterspeak. 

Max remains a consistent reform-skeptic (at least as it relates to social security).

• February 8, 2006 in
  • Health Care
  • Social Security
• Permalink
• Comments