The current discussion of ways to reform the U.S. Social Security retirement system is becoming increasingly polarized over the issue of "privatization." This divide unfortunately obscures the fact that the views of most Democrats and Republicans on the subject are not that far apart: a bipartisan solution should be achievable. More important, by lumping together all reforms that involve personal retirement accounts (PRAs), this polarization obscures critical differences between PRA-based plans that raise future standards of living and preserve the social-insurance features of the current system, and those that have few or no economic benefits to offset their high administrative costs, added risk, and reduced redistribution from rich to poor. Given that a PRA-based plan may well be enacted in the next few years, it is critical that the public understand that all such plans are not created equal.
Aging Americans, Anemic Savings
Retirement policy in the United States faces two challenges. The first is the long-term imbalance of the Social Security system. Although the system is currently running surpluses, the aging of the population implies that in about a dozen years expenditures will outstrip revenues. By 2050 -- around the time this year's Harvard seniors reach retirement -- scheduled benefits are projected to exceed revenues by 35 percent and the gap will continue to grow thereafter.
We should keep the magnitude of this shortfall in perspective. In 2050, the system will bring in enough revenue to pay 73 percent of scheduled benefits -- and would support benefits that are larger in inflation-adjusted terms than the benefits received by today's retirees. So the perception that Social Security will not be there at all for today's younger workers is a myth.
Moreover, the projected shortfall in 2050 is only 1.7 percent of projected gross domestic product (GDP). This is dwarfed by Medicare's long-run financial gap, and by the current shortfall of 5 percent of GDP in the federal budget excluding Social Security and Medicare. Indeed, simply returning the tax code to what it was when President Clinton left office would produce more than enough extra revenue to cover the entire Social Security shortfall. The irony is that the non-Social Security fiscal imbalances have become so large that reforming Social Security -- previously the untouchable third rail of American politics -- has become attractive in comparison.
The second retirement-policy challenge is that too few Americans have significant savings to supplement their income from Social Security. Social Security is designed to provide a solid foundation for financial well-being in retirement, not to cover a retiree's entire income needs. Private pensions and savings are meant to provide the rest. Unfortunately, for too many older Americans, Social Security is essentially all there is: one-third receive 90 percent or more of their income from Social Security and almost two-thirds receive more than half their income from the program. Thus, an important goal for retirement policy should be to encourage all workers to accumulate sufficient individual savings so the combination of Social Security and private nest egg allows them to maintain their standard of living during retirement. Proposals to introduce personal retirement accounts as part of Social Security can be seen as an attempt to deal with this second challenge.
Here is where the potential for a bipartisan solution comes in. Almost every reform plan, Democratic and Republican, includes some combination of benefit cuts and additional revenue to bring the traditional system into balance. And most Democrats and most Republicans support proposals to encourage savings on top of the traditional Social Security benefit. The substantive differences between the two parties' proposals involve the balance between tax increases and benefit cuts used to bring the traditional Social Security system into balance, and the extent to which the savings incentives are described as part of Social Security or as a separate program. Democratic reform plans tend to maintain higher benefit levels and therefore to rely more heavily on tax increases to restore the traditional system. And Democratic plans generally try to maintain a distance between savings incentives and Social Security. Republican plans rely more heavily on benefit cuts or on the hope that stock market gains will make tax increases and benefits cuts unnecessary, and these plans generally divert payroll tax revenue directly into PRAs. But these are not fundamentally different approaches; they are the kinds of differences that can be bridged in a legislative compromise.
The Political Debate
Why then is the debate over personal retirement accounts so contentious? Political grandstanding, of course, plays a role. But there is a substantive reason as well. People who place a high value on the social-insurance features of the current system worry that once the Pandora's box of personal retirement accounts is opened, it will be hard to control what will come out. Thus, even if it is possible in theory to design a PRA-based plan that preserves the best features of the current Social Security system while simultaneously increasing private savings, there is no guarantee that this is what the political process will yield.
The reason it is hard to predict what will happen is that three different groups support personal retirement accounts with three different goals in mind.
The first group sees PRAs as a politically feasible and economically efficient way to set aside additional resources to meet future retirement needs. This group believes that replacement rates (the ratio of retirement income to pre-retirement income) should be maintained or increased and that current generations should sacrifice some consumption to raise the standard of living of future generations. People in this group tend to support using new tax revenues on top of the existing 12.4 percent Old-Age, Survivors, and Disability Insurance (OASDI) payroll tax to fund personal retirement accounts.
The second group sees PRAs as a politically feasible way to cut Social Security benefits so as to limit the share of the nation's resources consumed by the elderly. This group believes that, given the aging of the population, replacement rates should fall -- that the economic costs of maintaining a system as generous as the current one are too high. Thus this group believes we should try to come as close as possible to living within the current 12.4 percent payroll tax. The typical plan from people in this group funds personal retirement accounts by diverting a portion of the payroll tax to PRAs.
The third group believes that Social Security was basically a mistake in the first place -- that it is not the government's job to provide for retirement, beyond perhaps a minimal welfare program. For this group, the mixed plans under consideration are the first steps toward total privatization.
The Bush administration appears to be dominated by people in the last two groups. Moreover, in promoting the virtues of PRAs, the Bush administration has not been forthcoming about the benefit cuts and revenue demands of its approach. Instead, it has emphasized features like "choice," voluntary participation, "ownership," and inheritability -- features that make Social Security reform appear painless and that are directly at odds with the social-insurance aspects of the current program. In this charged environment, many strong supporters of social insurance have become unwilling to advance their own PRA-based proposals for Social Security reform; instead, they have adopted the tactic of opposing all plans that include personal retirement accounts.
The Economic Effects of Personal Retirement Accounts
In addition to the risks of opening Pandora's box, there is also considerable uncertainty about the economic effects of any given PRA-based reform plan. PRAs have the potential to provide two main economic benefits. First, they can increase national savings, thereby raising future standards of living. Second, they can increase the perceived link between people's OASDI tax payments and their ultimate retirement income, thereby encouraging labor supply by motivating individuals to work longer, or more. But it is far from certain that a PRA-based plan will have these effects.
If PRAs are designed so as to reduce current consumption -- by requiring people to make contributions above the current 12.4 percent payroll tax to fund the accounts -- the direct effect of the plan will be to raise national savings. But if the PRAs are funded by diverting some share of existing payroll tax revenue and increasing the deficit -- as President Bush is apparently about to propose -- then the reform will have, at best, no direct effect on national savings, and may even reduce savings
Reform plans can have large indirect effects on savings as well. If Congress raises other taxes or cuts other spending in response to the increase in the deficit created by Social Security reform, even a plan like President Bush's could raise national savings. Recent news accounts suggest, however, that the administration is trying to come up with new budgetary rules that would hide the increase in the deficit created by Social Security reform. If true, then the Bush-style plan is unlikely to raise national saving through this indirect channel, either.
It is also possible that Social Security reform will affect the saving decisions of individual workers, but it is very hard to predict whether the net result will be to increase or decrease saving. Individuals might react to legislated reductions in Social Security's replacement rates by increasing their savings, hoping to ensure that they still have enough resources for retirement. Or they might believe that Social Security reform has increased the chance that they will receive significant benefits in retirement and therefore decide that they no longer need to save as much on their own. Individuals might learn the benefits of saving from watching their PRA balances grow and start making additional contributions to 401ks and IRAs. Or they might decide that because they have PRAs they no longer need to contribute to IRAs and 401ks, and net personal savings could decline.
The magnitude of the economic gains from increasing the perceived link between OASDI tax payments and ultimate retirement income is similarly uncertain. While PRA plans have the potential to make people feel that their OASDI contributions are going directly into their own accounts and are therefore not a tax, many such plans -- including the main plan of President Bush's 2001 Social Security commission -- are so complicated that they could conceivably reduce the extent to which people perceive a connection between what they pay in and what they get out of the system.
Moreover, the potential economic benefits from PRAs can, in theory, be accomplished without such accounts. To increase national savings, we could simply raise taxes or cut spending. To increase the perceived link between Social Security taxes and benefits we could adopt a national accounts system like those in Sweden or Italy, in which the basic pay-as-you-go structure is preserved, but the relationship between a person's contributions and retirement benefits is transparent. In other words, the main value of PRAs is as a mechanism to achieve economically desirable policies that would otherwise be politically infeasible to enact or sustain.
Privatizing with Protections Preserved
Given the uncertainties about what form a PRA-based Social Security reform plan will take and the economic impacts it would have, what should a strong supporter of social insurance do? One option is to reject personal retirement accounts and try to solve Social Security's problems with a combination of tax increases (for example, by repealing the cap on the level of earnings subject to the Social Security tax) and benefit cuts. After all, this has been the traditional approach to Social Security reform.
In my judgment there are two reasons to abandon the traditional approach and, instead, to embrace personal retirement accounts. The political reason is that in the current anti-tax environment, it is unlikely that significant tax increases are going to be enacted as part of Social Security reform. Therefore, the only feasible way to devote the extra resources to Social Security that will allow us to maintain replacement rates in the future is to do so via personal retirement accounts.
The economic reason is that we need to save the tax increases for Medicare and Medicaid. Over the next century, the share of national income that we will want to devote to healthcare will likely rise substantially and, because much of healthcare in the United States is paid for via the public sector, we will need to increase tax rates substantially to cover those costs. Although there is considerable debate about the magnitude of the economic costs of taxation, they are certainly not zero. We should therefore try to solve the Social Security problem with as little economic distortion as possible, relying on forced saving via PRAs and leaving the explicit tax increases to pay for future healthcare costs.
But is it possible to design a mixed system -- combining scaled-back traditional benefits with personal retirement accounts -- that preserves the best features of the current Social Security system? The answer is yes, provided the mixed system has the following six features.
Maintain replacement rates. The main function of Social Security is to prevent a large drop in living standards in retirement for people who, through bad luck or bad planning, do not reach retirement with sufficient savings. As wage levels rise over time, Social Security benefits therefore need to rise with wages. Otherwise, retirement standards of living would fall further and further behind those of the pre-retirement period.
The Bush administration is apparently going to propose moving from wage-indexing to price-indexing as a way to implement large benefit cuts in the traditional Social Security program. To pave the way for this proposal, the administration is trying to turn attention away from the concept of replacement rates and instead have people focus on real benefit levels. They are arguing that it is crazy for the financially insolvent Social Security system to provide real benefits in 2050 that are 40 percent higher than today's. But that argument misses the point of Social Security. Would we want today's seniors to have a 1950s standard of living?
There are, however, two serious arguments for letting replacement rates fall a bit over time -- though not nearly as far as the administration seems likely to propose. As the population ages, the tax rates necessary to pay Social Security benefits will have to rise, raising the amount of economic distortion caused by financing the program. We may therefore want to have a Social Security system that provides a bit less protection, rather than incur all of the incremental economic costs of higher taxes. In addition, because income levels rise over time due to productivity growth, we may reasonably be a bit less worried about people suffering a drop in their standard of living at retirement -- given that the amount of hardship caused by, say, a 40 percent drop in consumption is likely to be lower if the initial level of consumption is higher.
But these arguments assume that replacement rates are set optimally to begin with. As I mentioned earlier, we need to find a way to raise retirement-income levels for retirees who lack significant savings beyond Social Security. For the two-thirds or so of Social Security beneficiaries in this category, our goal for Social Security reform should be to have the sum of the traditional benefit and personal retirement accounts increase their replacement rates, not reduce them.
Maintain a substantial guaranteed benefit. In calculating the replacement rate provided by a reformed Social Security system, one needs to count both the income provided by the scaled-back traditional Social Security benefit and the income from the personal retirement accounts. These two components are not equivalent, however, because one provides a certain income stream free from market risk, while the amount of income from the other depends on how one's investments perform. Regardless of the overall replacement rate provided by the reformed system, it is important that a significant portion of the benefit remain free from market risk. Today, the replacement rate for a typical worker retiring at age 62 (the most common retirement age) is about 33 percent (a bit higher in after-tax terms). In a reformed system that combined a traditional benefit with PRAs, it would seem prudent to have the traditional benefit continue to provide an income-replacement rate of at least 20 percent with the PRA making up (or more than making up) the remainder. This is a very modest requirement: it would mean that for a worker who, through Social Security and private saving, manages to replace 80 percent of pre-retirement income -- a goal often suggested by financial planners -- the traditional Social Security benefit would be accounting for only one-fourth of the total.
Avoid the slippery slope. One of the main dangers of mixed Social Security systems that combine traditional benefits with PRAs is that they might become the first step toward total privatization. The scenario is easy to envision: this year, we divert 3 percent of the 12.4 percent payroll tax to personal retirement accounts, as proponents argue that this modest change preserves a strong traditional benefit. Then, assuming the stock market does well for the next couple of years, the 2008 Republican platform calls for diverting another 3 percent. Pretty soon, we have no traditional benefit left.
The way to avoid this risk is to establish a clear principle that none of the 12.4 percent payroll tax should ever be diverted to personal retirement accounts. Instead, the accounts should be funded with new resources above the 12.4 percent tax. In this scenario, there will still need to be significant cuts in traditional benefits -- because in the long run a payroll tax rate of 12.4 percent can support only about 60 percent of currently scheduled benefits. So we end up with a scaled-back traditional system funded with the 12.4 percent payroll tax and personal retirement accounts funded with, for example, an additional 3 percent of payroll contribution. Such a system, with a total cost of 15.4 percent of payroll, would allow us to maintain total replacement rates above current levels, even if the stock market performs somewhat less well in the future than it did during the past century.
Another benefit of insisting that the PRAs be funded with new money is that it maximizes the chance that the reform raises national savings.
Protect investors from high costs and poor investment decisions during accumulation. Individuals can mismanage their retirement savings in two ways. The first is to invest in mutual funds with high expense ratios. Investing in a mutual fund with an expense ratio of 1.25 percent reduces one's retirement resources by about 20 percent compared with investing in a fund with an expense ratio of 0.25 percent. In theory, competition should produce low fees. But in practice mutual funds manage to compete on non-fee dimensions (glossy brochures, sales incentives for distributors), and the average equity mutual fund charges fees of 1.25 percent. Normally, it is not the government's role to prevent consumers from making bad choices: for example, we don't stop people from purchasing Advil at one and one-half times the price of generic ibuprofen. But in the case of Social Security, the government has an interest in preventing people from making stupid decisions because people who end up destitute in old age will become a public responsibility.
The second way in which people can mismanage their retirement saving is to invest in an inappropriate portfolio. Judging the appropriate portfolio for each individual is difficult because it is hard to observe an individual's tolerance for risk. Nonetheless, there is some evidence that the main danger is that people will be too cautious -- say by investing only in bonds for their entire lives.
The solution to these problems is to allow people to invest only in a limited number of large broad-based index funds with regulations on fees and the minimum and maximum fraction of the portfolio that can be invested in stocks. Note that even this mixed system would produce income-replacement rates of only about 40 percent, so people would still need to continue to do additional retirement saving in IRAs and 401ks. People who desire a different overall asset allocation than that of the Social Security PRAs could do so by altering their other portfolios.
Finally, the accounts themselves need to be structured to minimize administrative costs, which are essentially a fixed amount per account. The cost does not depend on the account balances because nearly all of the expense comes from mailing account information to consumers, processing requests for changes in addresses or investment allocations, and answering customer questions via the telephone -- all factors independent of account balances. In addition, the typical PRA will have account balances that are smaller than those of the average mutual-fund investor today. So small personal-retirement-account plans are unlikely to be cost effective -- a plan probably needs to have contributions that are at least 3 percent of payroll to be worthwhile. Accordingly, administrative costs need to be assessed as a fraction of portfolio balances, not on a per-account basis, lest the fees consume a disproportionate share of the return for low earners. And there will likely need to be limits on the services provided: statements mailed annually rather than quarterly, restrictions on how frequently investors can alter their investment allocation, and perhaps even a charge per telephone inquiry.
Protect retirees during decumulation. Even though much of the debate over personal retirement accounts has focused on methods for structuring investments during the phase of life when workers are accumulating assets in their accounts, it is the design of the "payout" that largely determines whether a reformed system delivers a level of retirement-income security similar to that in the current Social Security system. In the current system, retirement benefits are paid in the form of real annuities that are indexed for inflation, last as long as a beneficiary lives, and are free from financial-market risk. In contrast, the most prominent legislative proposals for PRAs pay little attention to the payout phase, and even plans that specify a payout design tend to follow the Federal Thrift Savings Plan in not mandating annuitization and not guaranteeing benefits for surviving spouses. To fully replicate the protective features of the current system, full annuitization into fixed real annuities must be required of all workers, married couples must use joint-and-survivor annuities, and the government probably needs to be the annuity provider in order to keep administrative costs low.
Maintain redistribution. Finally, the Social Security benefit formula is progressive, providing a higher level of retirement income relative to lifetime earnings for low earners than for high earners. Although low earners tend to have shorter life expectancies than high earners and therefore receive benefits for fewer years, they still receive higher benefits relative to contributions than higher earners do. In contrast, a PRA plan that made deposits equal to 3 percent of payroll into each worker's account would not redistribute from high earners to low earners. Many different mechanisms could be used to maintain the existing level of redistribution in a Social Security system that includes PRAs. One would be to make higher deposits (relative to earnings) into the accounts of low earners. Another would be to make the traditional benefit formula more redistributive to offset the lack of redistribution in the PRA portion of the system. (But fans of the traditional system tend to be wary of this second approach because they feel that making Social Security too redistributive undermines its broad-based support.)
Including Protections from the Start
Some people argue that we do not need to worry about all of the details of a PRA system at the time it is introduced. After all, they maintain, there will be plenty of time to figure out the details as account balances accumulate, because it will be decades before people have to depend on PRAs for a substantial share of their retirement income. Indeed, many of the most important features of the current Social Security system -- survivor benefits, disability benefits, automatic cost-of-living adjustments -- were introduced after the original 1935 Social Security Act.
I believe that this argument is wrong. Although the protections in the current Social Security system were developed gradually over time, introducing those protections almost always involved politically popular benefit expansions. In contrast, in a defined-contribution PRA system, introducing protections over time will typically require reductions in expected retirement-income levels, and so will be perceived as the government taking money away from account holders. For example, requiring joint-and-survivor annuities to protect widows will reduce monthly payments compared to requiring single-life annuities. Similarly, introducing a requirement that a greater fraction of account balances be converted into an annuity upon retirement could be perceived by some as the government taking money away from them. Therefore, it is essential that these provisions be included in the PRA system from the start to ensure that a reformed system continues to provide the level of protection that Americans count on from Social Security.
Jeffrey B. Liebman, Ph.D. '96, is associate professor of public policy at the Kennedy School of Government. From 1998 to 1999 he served as special assistant to President Clinton for economic policy. For other readings on the subjects discussed in this essay, see http://ksgfaculty.harvard.edu/jeffrey_liebman.