Statement of Sylvester J. Schieber, Vice President, Watson Wyatt Worldwide

Testimony Before the Full Committee
of the House Committee on Ways and Means

May 12, 2005

Mr. Chairman and members of the Subcommittee on Social Security of the Ways and Means Committee of the U.S. House of Representatives, the following is a discussion about issues I believe you should consider during 2005 in your deliberations about the future operations of our Old-Age, Survivors and Disability Insurance system—what most people in our society call Social Security.  This discussion does not include a specific proposal for reform of our existing system.  If you want a specific proposal, I can offer you one but there are already many proposals available including a number of them that I have helped develop over the years.  Before offering you a new proposal, I would need to know what you wish to achieve with our Social Security program in the future including a set of principles that would serve as its foundation for future generations.

I have a set of principles that I offer as a starting point for discussion.  The remainder of my testimony here supports these principles.  In brief:

• The early survivor and disability insurance programs are term insurance and should be preserved and modified as appropriate.

• The important “safety net” or progressivity of the existing Social Security system is insurance for workers against bad labor market outcomes and should be preserved and enhanced.

• “Retirement savings” under the auspices of Social Security should be real savings and not loans to be redeemed out of our children’s consumption budgets.

• We should improve equity in the structure of benefits, especially between one and two-earner couples.

• We should continue to provide a floor of protection against longevity risk by providing basic benefits in the form of annuities.

• We should improve economic efficiency in the system, especially the linkage between contributions and benefits beyond foundation levels.

• We should assure long-term solvency, not simply postpone insolvency.

• We should assure that risks borne by individual participants are diversified and at tolerable levels—including skewing financial market risks toward those who are more able to bear it.

• Administrative costs should be kept at tolerable levels.

• Finally, fixing the system soon is extremely important.

Among other things in my career, I have studied the history of our Social Security program to a somewhat greater extent than most people who will come before you.  I wrote a book on Social Security in 1982 entitled, Social Security: Perspectives on Preserving the System published by the Employee Benefit Research Institute.  In 1998, I wrote a second book on the same subject with Professor John B. Shoven of Stanford University, The Real Deal: The History and Future of Social Security published by Yale University Press.  For the sake of full disclosure here, I advocated in both of these books that our Social Security pension system should include an element of personal accounts in its structure.  I did not come to this conclusion in either of these books because of ideological reasons.  I reached the conclusion because I believe that it is ultimately the only way that one of Franklin D. Roosevelt’s original and deeply held goals for the system can ever be realized.  I continue to believe that today and I continue to advocate that personal accounts should be part of our Social Security system because I agree with FDR’s strong belief in funding pension obligations as they are earned.

In the following discussion, I touch on a number of issues that I believe are important to your deliberations.  I start with FDR’s statement about what our Social Security pension system was intended to achieve.  I start here because I believe that FDR chose his language about this system carefully and that he deliberately meant what he said.  I move on to discuss a problem that arose in the implementation of FDR’s goals, a problem that many proposals today are attempting to correct.  Next, I revisit the “insurance principles” that FDR espoused in his vision of the system because I believe we would benefit to a great degree by returning to them.  I then take up a discussion about transition costs associated with reform of our Social Security system because I believe there is a great deal of confusion about how costs should be assigned to the rebalancing of the current system’s financing versus the costs associated with individual accounts.  In the final section of the discussion, I explore the differences in “carve-out,” “add-on” and hybrid financing of personal accounts.

Background

In June 1934, President Roosevelt established the Committee on Economic Security (CES) to explore the way in which our society could provide “security against the hazards and vicissitudes of life,” especially those associated with “unemployment and old age.”[1]  FDR indicated that he thought a program of “social insurance” was the way to address these problems.  The CES report, although dated January 15, 1935, was not formally submitted to President Roosevelt until two days later, January 17. He transmitted it to the Congress on the latter date along with recommendations on legislation.  The reason that there is a discrepancy in the dates on the CES report and its submission to the president is important in understanding FDR’s intentions about the operation of our Social Security program. 

President Roosevelt’s submission of the Social Security proposals to Congress was not the first time that he had been involved in developing public policy to provide income security to the elderly population.  While he had served as governor, New York had implemented a state assistance program for the elderly.  FDR considered the patchwork of state assistance programs as only a partial solution to the problems of income insecurity among the elderly.  In November 1934, he addressed an advisory committee to the CES and laid out certain tenets of the evolving legislation.  He said that when signing the Old-Age Pension Act while governor of New York he had expressed the “opinion that the full solution” to the old-age-income security problem could be achieved only on the basis of “insurance principles.  It takes so very much money to provide even a moderate pension for everybody, that when the funds are raised from taxation [that] a ‘means test’ must necessarily be made a condition of the grant of pensions.”[2]  By referring to “insurance principles” he was saying that he believed the new Social Security benefit would have to be funded in order to be viable on any grounds other than means testing.

On the afternoon of January 16, 1935, President Roosevelt was reviewing the final package that had been prepared by the CES for submission to Congress when he discovered a table in the report showing that the old-age insurance program would be running a significant deficit after 1965 that would require a government contribution over and above the payroll tax sometime later, around 1980.  He immediately suspected an error in the report and summoned Secretary of Labor, Frances Perkins, and the executive director of the CES, Edwin Witte, to help sort out the matter.  Upon being informed that the deficit was an element of the package as designed, FDR insisted that it had to be changed.  In regard to the prospect that the old-age insurance program he was proposing would require government subsidies in the future, Frances Perkins quotes FDR as saying: “This is the same old dole under another name.  It is almost dishonest to build up an accumulated deficit for the Congress of the United States to meet in 1980.  We can’t do that.  We can’t sell the United States short in 1980 any more than in 1935.”[3]

FDR’s statement ties back directly and consistently with his feelings at the time he had signed the old-age assistance law in New York while serving as governor when he said the full solution to the old-age problem could only be achieved through a program based on “insurance principles.”  It also follows from his statement to the Advisory Council the prior November when he said the old-age system had to be based on such principles.  FDR clearly envisaged and intended to develop a plan that was contributory and self-supporting with an accumulation of a trust fund roughly commensurate with accruing benefit obligations.

After the meeting between FDR, Secretary Perkins and Witte at the White House on the afternoon of January 16, the report was withdrawn from the President.  Secretary Perkins set about polling the members of the CES and all agreed that the President’s wishes on the funding matter were to be addressed. At the President’s insistence, the offending table was taken out of the report and the package was modified to indicate that the schedules of tax rates and benefits included were merely one approach to providing old-age benefits that Congress might consider.  The report was filed with the President the morning of January 17.[4]

The final provisions in the Social Security Act adopted in 1935 called for a schedule of payroll taxes to begin at a rate of 1 percent each on workers and their employers on the first $3,000 of annual earnings.  The initial payroll tax rate paid by workers and their employers was to increase in half-percentage point increments every three years until it reached 3 percent of covered wages in 1949.  The contributory funding was projected to be adequate so that no added government contribution would be required to finance the old-age insurance benefits.  By 1980, the trust fund was projected to grow to $47 billion.[5]

Under the proposal that had been put forward by the Roosevelt Administration, the trust fund was to invest only in government bonds.  For many people the thought of this accumulation, especially in the form of government bonds, was too fantastic to comprehend.  At the time it was being considered, the total outstanding federal debt was only $27 billion and no one thought of the government running future deficits that could accommodate such accumulations. After all, the government had accumulated only a total of $27 billion in debt in its first 159 years of operations, and no one expected it to accumulate another $20 billion in the succeeding 45 years.  Further, contemporary policymakers thought of paying down the debt after getting out of the Depression rather than seeing it grow in the future.  There were a number of potential problems in the projected accumulation of the Social Security fund. 

From one end of the political spectrum, the critique of the Social Security Act focused on the relative levels of benefits that would be provided through the federal Old-Age Benefits program in its early years of operations and the state administered old-age assistance programs.  The funding provisions, which President Roosevelt had insisted on when the Act was under development, meant that the old-age insurance program was not going to pay significant benefits until many years into the future.  From the other end of the political spectrum, the critique of the original legislation focused on the notion that a trust fund invested in government bonds is, in reality, a scheme to borrow from future generations at the expense of fiscal discipline today.  This argument was summarized by Senator Arthur Vandenberg:

The Treasury collects [a] billion in pay-roll taxes…  The Treasury gets a billion in cash.  It goes into the general fund…  Congress then takes it out of the Treasury by appropriating a billion to the reserve… So the Social Security Board hands the billion in cash back to the Secretary of the Treasury and takes from him a special… IOU… The Secretary of the Treasury has the billion of money… He can use the billion either to retire regular Government-debt obligations in the general market or… he can apply it on his current operating deficit.  As things are now going, we shall have deficits…

What has happened, in plain language, is that the pay-roll taxes for this branch of social security have been used to ease the contemporary burden of the general public debt or to render painless another billion of current Government spending, while the old-age pension fund gets a promise-to-pay which another generation of our grandsons and granddaughters can wrestle with, decades hence.

It is one of the slickest arrangements ever invented.  It fits particularly well into the scheme of things when the Federal Government is on a perpetual spending spree.  It provides a new source of current revenue, which while involving a bookkeeping debit, providentially eases the immediate burden of meeting current debts and deficits.[6]

The funding principles espoused by Franklin D. Roosevelt began to unravel as early as 1939. Because of the concerns about the implications of funding the system, President Roosevelt agreed to convene an Advisory Council to study the matter. Based on its recommendations, Congress adopted several amendments to the original 1935 legislation. Payments would begin in 1940 rather than 1942. The system would pay out benefits to spouses and other dependents of retirees or workers who died before retirement. Under the 1939 Amendments, the trust fund was projected to hold a balance of $6.9 billion in 1955 compared with $22.1 billion projected under the original legislation.[7]

During World War II, the system shifted even further away from advance funding.  Although President Roosevelt had gone along with the 1939 Amendments’ three-year delay in increasing the payroll tax, he opposed the subsequent delays.  When Congress was considering the delay in the tax increase scheduled for January 1, 1943, FDR wrote the chairmen of the Senate Finance and House Ways and Means Committees.  He argued that “a failure to allow the scheduled increase in rates to take place under present favorable circumstances would cause a real and justifiable fear that adequate funds will not be accumulated to meet the heavy obligations of the future and that the claims for benefits accruing under the present law may be jeopardized.”[8]  President Roosevelt vetoed the Revenue Act of 1943 because it included a delay in the payroll tax increase, but the veto was overturned.  At the end of 1944, in signing H.R. 5565 which delayed the increase in the payroll tax from January 1, 1945 to January 1, 1946, the President’s accompanying statement noted, “I have felt in the past and I still feel that the scheduled rate increase, which has been repeatedly postponed by Congress, should be permitted to go into effect.  The long-run financial requirements of the Social Security System justified adherence to the scheduled increases.”[9]

On April 12, 1945, President Roosevelt died leaving behind the Social Security program as the central foundation of the welfare state in America.  By the time he died, Social Security was well on its way to operating on the pay-as-you-go financing basis. By the mid-1950s, the concept was completely abandoned.[10]  After that, the program ran largely on a pure pay-as-you-go basis until the mid-1980s. 

At the beginning of the 1980s, the system was facing the prospect of coming up short on regularly scheduled benefit payments.  Early in 1983, Congress intervened just in time to avoid a partial default on current benefits, adopting a number of provisions to secure the program. At that time, there was absolutely no consideration of individual accounts as part of the solution. Over subsequent years, the trust fund has grown to approximately $1.7 trillion dollars. The trust funds are projected to peak at $3.6 trillion or so in 2022 in 2005 dollars ($5.7 trillion nominal that year), after which they will begin to decline.[11] 

The implications and import of the accruing trust fund assets continue to be controversial.  The general consensus seems to be that they do not add to national savings according to a number of empirical analyses.  Several researchers have concluded that surplus revenues generated in national retirement income systems held in government bonds result in larger deficit spending in other elements of those governments’ general fund accounts.[12]  That conclusion is not universally embraced,[13] although the folks that dispute it have not presented comparable empirical evidence to bolster their conclusion.

Interestingly, in the political arena, this modern day debate is the same one that the Arthurs Altmeyer and Vandenberg carried on back in the 1930s.  In almost the identical setting where Altmeyer and Vandenberg conducted the original debate, some 60 years later Senator Bob Kerrey (D-NE) and Ken Apfel, the Social Security Commissioner who served during the later years of the Clinton Administration, engaged in a parallel discussion in a Senate Finance Committee hearing. In this more recent version of the debate, Senator Kerrey summarized the conclusion that many observers have drawn over the last couple of decades:

We are not prefunding… Are we holding the money in reserve someplace?  We are not prefunding!  The idea in 1983 was that we would prefund the baby boomers.  We began to use it immediately for the expenditures of general government.  We didn’t prefund anything.  What we are doing is asking people who get paid by the hour to shoulder a disproportionate share of deficit reduction.  That’s what we’re doing!  And the beneficiaries on the other hand, they suffer under the illusion inflicted by us very often, that they have a little savings account back here.  They are just getting back what they paid in.  They don’t understand that it’s just a transfer from people that are being taxed at 12.4 percent.[14]

Revisiting the Insurance Principles That FDR Embraced

In 1935, when President Roosevelt insisted that what he called “my Social Security program”[15] be based on insurance principles he was thinking about the program in the context of providing retirement benefits.  The original law did not provide many of the sorts of protection that are included in today’s system.  Indeed, in signing that original law, Roosevelt spoke of the system it created as being “a cornerstone in a structure which is being built but is by no means complete.”[16]  Despite the fact that we have built on that cornerstone over the years, it may be worthwhile to review what FDR had in mind when he insisted that “insurance principles” be followed in the construction of Social Security.

Insurance is a mechanism whereby a group of individuals can join together to spread the risk that they each individually face in regard to some contingency that creates an economic loss for those who incur that particular “vicissitude” of life.  Consider, for example, a society comprised of 1,000 households where each family lives in their own home.  Assume, for simplicity, that every family’s home is worth $100,000 and that each year fire strikes one family’s home completely destroying it.  If every family attempted to cover this risk by itself, then each year one family would be faced with a devastating $100,000 loss.  On the other hand, if all of the families pooled together and each contributed $100 to a home-owners’ fire insurance fund, each family would invest $100 per year to assure that no family incurred such a devastating loss.

In order to understand what should be done in reforming our Social Security system, it is important to understand what it currently does and to rationally design reforms that preserve those elements we wish to preserve and to modify those that need to be changed to secure its ongoing operation.  The current Social Security provides insurance for four hazards that workers face.  It provides insurance for workers:

1. Who die and leave juvenile dependents;
2. Who become disabled and can no longer earn a living;
3. Who experience bad labor market outcomes; and
4. Who suffer from the myopia that workers have about making adequate protection for their own retirement needs. 

In addition, for retirees Social Security provides:

1. Longevity insurance because the benefits are paid in the form of an annuity;
2. Income protection against inflation in retirement because the annuity is indexed to account for increasing prices; and
3. Survivor benefits. 

In an insurance context, the nature of risks that are insured under Social Security vary considerably from one aspect of the program to the next.

Early-survivor insurance

The early survivor program provides insurance protection against the vicissitude of workers dying and leaving juvenile children with insufficient resources to meet their economic needs.  There are two factors that define the risk that workers incur in this case.  One is the probability that they will die and the second is the probability that they have children.  Over most of the working-age years, the probability of dying for most workers is quite low but rises gradually as workers age.  In the younger years of the working career, the probability of having children under the age of 18 is relatively high but drops off significantly as people approach retirement age.  To show what the exposure is here, I used the 1950 birth cohort life table from the Office of the Actuary and the incidence of individuals by age with dependent children under age 18 to derive Figure 1.  The left panel shows the variation on a scale of 0.00 to 0.25 percent. It is intended to show that there is some variation in the exposure here. In no year does the line get as high as 0.25 percent meaning that in no year were there more than 2.5 people per 1,000 dying in this birth cohort and leaving juvenile children over most of their exposure period. The right panel in Figure 1 shows the same distribution on a scale of 0.00 to 100 percent.   It is intended to show that the risk exposure to this particular contingency is extremely small in the overall scope of things.  That is not to say that when the contingency actually strikes a family that it has a devastating effect.  Indeed, this is a case a lot like our opening hypothetical example of people having house fires. 

The probability of workers dying and leaving juvenile children with the need for economic support is a contingency that can be covered without significant expense to active workers.  Indeed, there has been virtually no discussion of significantly modifying this element of the current system in any of the discussion about reforming it.  As we look at reform options, we need to make sure that modifications made to the existing system do not result in unintended consequences in this area.  There are certain public good features to the existing benefits and there are likely relative efficiencies that are realized by running them through government on a nationalized basis with mandatory participation for virtually all workers. 

 Figure 1: Probability of Death from One Year of Age to the Next for the 1950 Birth Cohort Times the Probability of Individuals Having Dependent Children under Age 18 in 2003

Sources: Calculated by the author from data published by the Office of the Actuary, Social Security Administration and U.S. Department of Commerce, U.S. Census Bureau, Current Population Survey.

Disability insurance

The case of disability insurance provided through our existing Social Security program is similar to early-survivor benefits.  The incidence of disability under the Disability Insurance (DI) program by age in 1968 is reflected in Figure 2. Once again, the DI program has not been widely discussed in the same context that reform of the retirement system has been although a number of proposals would have implicit implications on benefit levels in the program.  Part of the reason that DI has not been part of the discussion is that the incidence of disability is relatively low across much of the age spectrum and the overall cost of benefits is significantly less than in the case of the old-age retirement aspects of the system.  As with early-survivor benefits, there are almost certainly public good features to the existing benefits and there are likely relative efficiencies that are realized by running them through government on a nationalized basis with mandatory participation for virtually all workers. 

Just because the DI system has escaped the same scrutiny as the retirement program in recent discussion about Social Security reform does not mean that the current disability program should not be included in these discussions.  This element of the system is underfunded and contributes to the total underfunding in the combined systems.  In addition, the determination of eligibility in the current system is tied to a concept of being unable to work that may have made sense in the mid-twentieth century when it was postulated but makes much less sense in the “knowledge economy” of the twenty-first century. Finally, there are a variety of administrative issues that also plague the existing Disability Insurance system.  The potential reform of the disability programs is an issue that should be considered outside the realm of reform to the retirement plan or basic benefit structure of Disability Insurance or any other facet of Social Security.

 Figure 2: Incidence of Disability under the Social Security DI Program in 1998 by Age

Source: Office of the Actuary, Social Security Administration, “Social Security

Disability Insurance Program Worker Experience,” Actuarial Study No. 114, June 1999).

Insurance against bad labor market outcomes

While we have not characterized it that way, the redistributive structure of our Social Security benefit formula is the primary way we provide insurance against bad labor market outcomes.  At the outset of our careers, none of us knows for sure that we will succeed.  It is not hard to find many examples of people born into the most modest circumstances who go on to be dramatically successful in their careers.  It is not hard to find many other examples of people who seem to set off on a career marked for success who fail miserably along the way.  The element of our Social Security system that pays a relatively higher monthly benefit to people who have not been as successful in the labor market as those who have is our way of helping the less fortunate have a reasonable standard of living in their latter years. 

Table 1 shows estimated internal real rates of return that will be realized by a set of prototypical Social Security program participants reaching age 65 in 2008 according to estimates developed by the Social Security actuaries.  These workers are classified according to their marital and earning status and earnings levels over their working careers.  If you focus on any particular column, you will see that the rate of return on lifetime contributions declines the higher up the earnings distribution that a worker ends up.  This sort of “social insurance” provided by Social Security is not something that we can ever expect private insurance markets to provide.  To the extent that there is a concern that people who are unsuccessful in their working careers not be forced to live out a retirement at a socially unacceptable level of living, this sort of mechanism almost certainly will have to be part of our retirement structure.  Many reform proposals would maintain or strengthen this element of the current system.  Part of the reason for that general support is the result of the broad dependence on Social Security for income security among the portion of the workforce at the lower end of the earnings distribution. 

 Table 1: Internal Real Rates of Return for Various Earnings Level Scaled Workers Who Will Turn Age 65 in 2008 

Source: Office of the Actuary, Social Security Administration, “Internal Real Rates of Return under the OASDI Program for Hypothetical Workers,” Actuarial Note, Number 2004.5, March 2005, p. 6.

^a Career average earnings level wage indexed to 2003.

The extent to which selected workers benefit from the insurance against bad labor market outcomes can be seen from Table 2 which shows a distribution of Social Security Primary Insurance Amounts (PIAs) for actual workers who retired in 2003 in comparison to the PIAs of the prototypical workers considered in Table 1.  It is clear from this table that female workers tend to be skewed toward the lower end of the earnings distribution so they get a somewhat disproportionate share of this form of insurance provided by Social Security. While it is not reflected in the table, we know from other sources that older women in particular are at risk of living out their final years in poverty.  Reform options that move the Social Security system more toward operating purely as a retirement savings system should include elements to maintain or enhance the income security protections built into the existing system for some particularly vulnerable members of our society, namely those who have not had a particularly successful working career.

 Table 2: Distribution of PIAs of Actual Workers Who Retired in 2003 Relative to Prototypical Scaled Workers Developed by SSA Actuaries 

Source: Office of the Actuary, Social Security Administration, “Internal Real Rates of Return under the OASDI Program for Hypothetical Workers,” Actuarial Note, Number 2004.5, March 2005, p. 3.

^a Career average earnings level wage indexed to 2003.

There is another feature of Table 1 that policymakers ought to consider in any deliberations to modify Social Security.  Earlier we looked at the columns in the table to consider the insurance feature in the system intended to protect low earners.  It is also important to consider the lines in the table.  One thing that is apparent when looking at the table in this fashion is the disproportionately high returns that single-earner couple participants in the system receive.  This may have been an intended consequence back in the 1930s and even as recently as the 1960s and 1970s when most female spouses spent much of their prime working years as homemakers.  In a modern era when the vast majority of women work outside the home during their prime working years, it is no longer clear that this characteristic is equitable especially taking into consideration that many non-employed spouses live in households where total income is relatively high.  In this regard, spousal benefits may be considerably dampening the intended insurance feature of the system intended to skew benefits toward lower earners.

Another aspect of modern times that is remarkably different than when Social Security’s insurance features were configured back in the 1930s, 1940s and 1950s is the prevalence of other income protection features in our retirement system.  The dependence on Social Security for retirement security is not randomly distributed.  That means that some types of reform have the potential to disproportionately disadvantage certain groups. This point can best be understood by looking at people on the cusp of retirement as James Moore and Olivia Mitchell have done.[17]  Their analysis uses Health and Retirement Study (HRS) data.  The HRS is collecting longitudinal information on a representative sample of the U.S. population between the ages of 51 and 61 in 1992.  Sample members are being interviewed every two years.

Moore and Mitchell used the 1994 wave of the HRS interviews to estimate the participating households’ wealth levels just as most of them were approaching retirement.  They included four classes of wealth in their calculations: 1) net financial wealth, including savings accounts, investments, business assets, and non-residential real estate less outstanding debt not related to housing; 2) net housing wealth; 3) pension wealth, or the present value of employer-sponsored retirement benefits; and, 4) the present value of Social Security benefits under current law. 

Table 3 has been derived from Moore and Mitchell’s analysis. The wealth measure used here does not include net housing wealth because most homeowners do not sell their homes at retirement, or if they do, they tend to buy another one.  This definition of wealth includes business assets and non-residential properties.  We are interested in looking at the assets of these households that can be expected to generate a stream of income that can be used to finance consumption during retirement.

Table 3: Distribution of Wealth among the Near Elderly

Source: James F. Moore and Mitchell, Olivia S., “Projected Retirement Wealth and Savings Adequacy,” in Olivia S. Mitchell, P. Brett Hammond, and Anna M. Rappaport, eds., Forecasting Retirement Needs and Retirement Wealth.  Philadelphia: University of Pennsylvania Press, 2000, p. 72.

Table 3 shows that the people at the bottom tenth percentile of the wealth distribution hold almost all of their wealth in the form of Social Security retirement benefits.   Social Security benefits still account for almost two-thirds of total wealth for those households one-third of the way up the wealth distribution.   Those two-thirds of the way up have a rough parity in their wealth holdings between their social security annuity, employer-sponsored pensions and other financial wealth.  Those at the top of the wealth distribution have very limited dependence on Social Security. The point of the analysis here is to show that for many workers reaching retirement age in our society, a disproportionate portion of their wealth has been accumulated under the auspices of Social Security.  For many workers, however, Social Security is a relatively small share of their retirement security portfolio.  We should be mindful that rebalancing Social Security by means of across the board reductions in benefits will have a highly skewed effect on future retirees.  A 20 percent across the board reduction in Social Security benefits would reduce the total retirement wealth of those at the bottom 10^th of the wealth distribution in Table 12 by nearly 19 percent.  For those at the top 10^th of the wealth distribution, it would reduce the total retirement wealth by about 2 percent.  To the extent that we might shift toward individual accounts as a portion of the national base of our retirement security system, we should be mindful of how such a change might alter the insurance protection provided to those with low lifetime earnings. 

Insuring that workers make adequate provision for retirement income needs

The fourth sort of worker insurance provided by Social Security is distinctly different than the first three.  For the overwhelming majority of workers, the prospect of reaching an advanced age is a near certainty and retirement patterns developed during the twentieth century suggest most people will end up with a period at the end of their lives when they no longer earn a direct wage.  To the extent there is a public interest that elderly people maintain some minimum standard of living, it is reasonable to force people to "save" some portion of their earnings to provide for their needs when they no longer work.  If we do not require that workers save, they may fail to do so on their own and become wards of the state.  That was one of the most fundamental motivations for Social Security from Franklin Roosevelt’s perspective. 

One of the amazing achievements of the U.S. Social Security pension system is that it has succeeded so well in providing a broad base of protection for our elderly citizens.  Figure 3 shows the percentage of people ages 60 to 80 who reported receiving Social Security benefits in 2003.  When one takes into consideration that some 4 or 5 percent might still be in public pension plans outside of Social Security and that some simply failed to report correctly about their sources of income, it is clear the system is providing close to universal protection for those it is intended to cover.

 Figure 3: Percentage of People by Age Who Reported Receiving a Social Security Benefit in 2003

Source: Author’s tabulations of Current Population Survey

Given the relative certainty that most workers will get old and most will quit working before death this element of the program deserves separate and careful consideration.  If one harkens back to the example that we described at the outset the implications of the phenomenon in Figure 3 become apparent.  There, we had a case where one family in a thousand had their house burn down each year causing them a catastrophic loss.  In that case, the risk of such a loss could be pooled across a large number of people and everyone could be protected by small annual contributions.  Now consider trying to provide this same sort of protection where houses burned with certainty at a given age and where as many as one third to one half of them burned each year.  The cost of providing protection explodes with the virtual certainty of the contingency occurring for everyone and the approach for securing against this sort of loss would be significantly different than where the incidence of the problem is small. 

Given FDR’s fiscally conservative nature and the strong position he had taken on funding of the social insurance elements of the Social Security Act in 1935, he saw this legislation “as protection to future Administrations against the necessity of going deeply into debt to furnish relief to the needy.”[18]  We know that FDR felt strongly about the funding of “retirement” benefits when he first submitted Social Security proposals to the Congress.  His statements at the time he submitted the legislation and when he signed the bill indicate a clear concern about the long-run fiscal implications of running a pay-as-you-go system.  We know that he repeatedly resisted the shift toward pay-as-you-go financing of the system once it was up and in operation.  When he vetoed the Revenue Act of 1943 because of its provisions that shifted away from Social Security funding toward pay-as-you-go operations, it was only the second bill that he had vetoed in his 10-year tenure as president at the time.

To show how alternative pension structures operate from an economic perspective, consider a theoretical worker who begins working at age 25 earning $35,000 per year and attempts to save a bit of her annual earnings to provide for income needs during retirement. Assume this individual has perfect foresight and knows that her pay will increase 4 percent per year until she reaches age 65, when she will retire and receive a pension that is 35 percent of her gross earnings in here last year of employment, a pension indexed for inflation which we assume to be 3 percent per year.  To simplify the process of determining how much the worker should save, we assume she knows that she will live to be 81.5 years of age. We also assume the worker anticipates receiving an annual rate of return on his assets of 5 percent per year.  At retirement, roughly 60 percent of accumulated assets are attributable to interest earned on the lifetime contributions the worker has made.

If everything goes according to plan, this worker will earn roughly $161,600 in her last year of employment. After her retirement savings are put aside, her disposable income will be approximately $135,700 that year. As it turns out, this worker will need to save 10.3 percent of her annual earnings each year in order to fulfill her work and retirement plans. If she does that, she should be able to receive an annuity of $56,550 per year, a benefit that will grow from year-to-year during retirement at the rate of price inflation. The initial benefit will be about 39 percent of her disposable income in her final year of work where disposable income is her total wage minus what she has to contribute to a pension in order to finance her retirement income.

This pattern of asset accumulation and net balances are reflected in Figure 4. Over the working period, the worker’s steady saving plus interest accruing on accumulated assets gradually accelerate the growth in total assets. From a macroeconomic perspective, while the worker or the employer is contributing to the plan, these contributions are reflected as savings accruing in the economy.  After retirement, the assets are steadily depleted over the worker’s remaining lifetime and run out when he dies.  Net savings over the worker’s lifetime, in this example, are zero. Had she wished to leave a bequest to heirs, the worker would have had to save more during her working life or spend less during retirement.

Figure 4: Accumulated Savings of a Hypothetical Worker Participating in a Funded Pension Plan

Source: Calculated by the author.

If the same worker described above is covered by a pay-go retirement plan, the dynamics of her accumulating retirement wealth are considerably different than those in a funded pension plan. First, her annual contributions to the retirement system are paid out to current retirees. Second, rather than becoming part of an accumulation of capital that can be invested in the economy, in most cases her contributions merely purchase an entitlement to a retirement benefit. In other words, it results in an unfunded obligation —what Paul Samuelson has characterized as a “consumer loan” [19]— that future participants in the system are obligated to pay when the current worker retires. The pattern of this transaction is reflected in Figure 5, which turns out to be a mirror image of Figure 4. In this case, the worker’s “accumulated savings” from the worker’s perspective is the sum of the obligations she is owed. It grows on a gradually accelerating basis until the worker reaches age 65, and then is paid off over the remainder of her lifetime as annual retirement benefits.  From a macroeconomic perspective, however, deducting payroll taxes from a worker’s compensation may reduce his or her consumption at that time, but the benefit paid to a retiree is usually used largely for consumption purposes.  Thus, it has no positive effect on net savings in the economy.

Figure 2: Accumulated Savings of a Hypothetical Worker Participating in a Pay-As-You-Go Pension Plan

Source: Calculated by the author.

From the worker’s perspective, the accumulation of pension rights through a pay-go social security system is no different from accumulating wealth through personal savings or a funded pension. In the life-cycle context, the primary motivation for workers to save is to provide for their consumption after they retire.  

In both the funded and pay-go pensions, the worker is deferring consumption from the working period to the retirement period. In an economic context, however, there is an important distinction between the two approaches.  In the funded plan, the deferred consumption is used to purchase assets that will finance post-retirement consumption. In the pay-go plan, the deferred consumption establishes a claim on the productivity of the next generation of workers. If a significant share of their retirement consumption needs will be met by a mechanism that does not require savings, and indeed actually creates substantial liabilities, it has the potential to lower national savings rates. A funded pension system generates real savings.

In recent years, there has been a considerable debate among economists about whether our accumulating Social Security trust funds represent real savings that will help to ameliorate the burden that the baby boom generations pose on the retirement system. Looking at this discussion in the context of the comparison of funded versus pay-as-you-go financing helps to clarify the issue being debated. 

Looking back to Figure 4, it is clear that a retirement plan’s aggregate contribution to savings is the extent to which assets accumulate to cover its net obligations.  In an aggregate context, it is not the net of the annual contributions into a trust fund minus the payout of current benefits and administrative expenses. It is the extent to which accruing obligations in the plan are covered by the assets in the plan. In the case of private pensions, actuaries are required to estimate the accrued benefit obligations at each valuation, and plan sponsors are required to report the results to the federal government.  These periodic tallies of assets and obligations in plans can be used to track the contributions of the system to national savings.  Along similar lines, the Social Security actuaries have calculated something they have labeled the “maximum transition cost” for that system in recent years.  The actuaries report that this measure “represents the transition cost for continuing the Social Security program in a different form, with all payroll taxes for work after the valuation date credited to the new benefit form. The maximum transition cost is equivalent to the unfunded accrued obligation of plan designed to be fully advance funded at the time of plan termination.”[20] Once again, the tally of assets in the system and the accruing obligations allows us to assess the net effect of Social Security on national saving.

The results of the Social Security liability calculations and funding levels are presented in Table 4.  The results that are shown there are as of the beginning of each year listed in the table.  The asset values actually reflect those reported by the Social Security Actuaries as of the end of the prior year but one day’s income would be relatively trivial in the context of the discussion here.  The table shows that trust fund assets in the Social Security system grew by nearly $1.2 trillion between the beginning of 1996 and 2005, while total obligations increased by $6.8 trillion over that same period with unfunded obligations climbing by $5.6 trillion.  Some people look at the trust fund growth and conclude that, between 1996 and 2005, Social Security contributed $1.2 trillion to U.S. saving.  A number of studies cited earlier suggest that this accumulation of trust funds has actually been used to hide deficit expenditures elsewhere in the federal fiscal operations.[21]  Even if those dollars were accumulated to claim that they have added to national savings completely ignores the added $6.8 trillion of obligations created over the last decade for future generations of workers to finance.  I strongly believe we need to find a savings mechanism to secure future benefit accruals for this sort of insurance.  We need to return to the “insurance principles” that Franklin Roosevelt was advocating when he adamantly demanded that his Social Security program be funded.  My own personal conclusion is that the only way we can do that is to create a system of personal accounts that are part of our Social Security program that will allow us to segregate the assets and keep them from being used to finance other government operations.

Table 4: Social Security Unfunded Accrued Obligations, Trust Fund Assets and Under Funding

Sources: Author’s calculations of total plan obligations as sum of trust fund assets from the 2005 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds plus the unfunded accrued obligations from unpublished data from the Office of the Actuary, Social Security Administration.

Insurance protections provided in the retirement period

It is also important to consider the post retirement benefits in the current system and how they might be addressed in Social Security reform proposals.  The current system provides at least three sorts of insurance protection to the retiree population.  The first is longevity insurance—protection against outliving one’s resources—by providing its benefits in the form of an annuity.  The second form of insurance in this aspect of the system is protection against erosion against the standard of living achieved while working by providing a benefit indexed for inflation during retirement.  The third form of insurance is spouse and survivor protection provided to people in annuity status.

One characteristic of individual accounts is the strong sentiment of ownership that develops around the accounts.  This sense of ownership has the potential to be in conflict with the public interest in using the accumulated funds to provide retirement security on a broad basis.  For example, if policymakers wish to assure that people achieve a level of income at least equal to the official poverty line, they may require that some portion of accumulated account balances be annuitized when a worker reaches retirement age.  If policymakers do not impose an annuity requirement, some retirees will likely spend their resources prior to dying and then potentially present themselves for additional support at a cost to the public fisc – the classic moral hazard problem that often arises in situations like this. Similar concerns arise in regard to joint and survivor benefits.  These issues will have to be explicitly addressed if individual accounts are part of Social Security reform.

Transition Costs Associated with Social Security Reform

In the discussion about Social Security reform and the prospect that individual accounts might be part of it, there has been a great deal of misinformation spread about the costs associated with transition that we will incur in implementing such reforms.  In understanding the dynamics of transition costs associated with reform, it is important to segment the costs associated with various aspects of any reform.

For the sake of discussion, consider the potential for reforming the current system without including any element of personal accounts as part of that reform.  Each year the Social Security actuaries calculate the “open group unfunded obligation.”[22]  This measure is an estimate of the funding shortfall under current law to deliver benefits now scheduled over the projection period.  Traditionally, the actuaries have estimated this amount over the 75-year projection period covered in their annual valuation of the program.  At the beginning of 2005, they estimated the present value of this unfunded obligation to be $4.0 trillion.  The interpretation of this value is that, if the trust funds held an added $4.0 trillion on January 1, 2005, then the scheduled collection of taxes over the next 75 years in combination with the trust fund balance and expected returns would cover expected expenditures over the period. 

So, we face a $4.0 trillion transition cost under current law no matter what we do with the program. Under law, the program does not have deficit spending authority, so benefits over the projection period must be fully financed through program revenues and assets. In other words, to comply with the law over the next 75 years, we must come up with an additional $4.0 trillion in new revenues in 2004 dollars, cut scheduled benefits by that amount or some combination of the two.

In the 2005 annual trustees’ report on the Social Security system, the open group unfunded liability was also calculated for an “infinite” time frame. The estimate in this case was $11.1 trillion.  This estimate has come under considerable criticism in some circles, although it was included in the annual report at the Trustees’ insistence.  The problem is that our demographics today are far more favorable than they will be in the future.  That being the case, calculating adequate financing for the 75-year valuation period was different last year than it is this year.  The basic valuation released in early 2004 covered the period 2004 through 2078, and the one released in April 2005 covers the period 2005 through 2079.  Last year’s valuation included 2004 and this year’s valuation did not.  This year’s valuation includes 2079 and last year’s did not.  In terms of the actual calculations of the funding status of the program, 2004 was a good year because revenues exceeded expenditures, but in 2079, anticipated expenditures will significantly exceed revenues.  If policymakers devise a reform that balances the program’s finances over the current 75-year valuation period, it will be out of balance again next year because of this limited time period focus.

In 1983, policymakers adopted policies they believed would fully finance the program over the 75-year valuation period.  But lo and behold! We are once again confronting a program that is underfinanced, and a substantial share of the shortfall is due to the passage of time and the difference in valuation periods.  This is why many policy analysts have advocated that we consider policy options that will provide financing stability well beyond the fixed valuation period.  One way of measuring whether a particular adjustment to the program will deal with the long-term underfunding is to look at the infinite period.  An alternative way is to focus on whether a particular adjustment results in sufficient financing to get through the traditional 75-year valuation period in a way that there is a substantial trust fund balance at the end of the projection period and that projected net income to the system at that time will be in relatively close balance to projected expenditures.

Financing the transition costs embedded in current law

The most straightforward way to finance our way out of a $4.0 trillion pension hole reflected by the unfunded obligations from an “ongoing” perspective might appear to be either increasing contributions to the plan by that amount over some reasonable period, reducing benefits by that amount or some combination of the two.  Proposals to accomplish any one of these options in the context of the current plan structure should be relatively transparent in terms of revealing how the transition costs will affect various segments of the population. 

We could simply raise the payroll tax rate starting virtually immediately by something around 1.9 percent of payroll.  This would bring in approximately $4.0 trillion over the next 75 years in current present value terms, theoretically solving the financing problem for the formal 75-year valuation period.  But it would not resolve the structural weaknesses in the system in the out years, and 15 or 20 years down the road, the program would likely be facing as big a problem as it is today.  It would also lead to an even larger accumulation of trust fund assets over the next two or three decades than is now projected under current law.  Given that we have never been able to find an effective way to actually “save” these trust fund accumulations, I do not believe this approach will actually help us solve the current financing problems.

An alternative to raising the payroll tax rate would be to eliminate the cap on covered earnings against which payroll taxes are assessed.  This would represent a significant deviation from the underlying philosophy that has been the foundation of the program since its conception and initial passage in the 1930s.  Specifically, the system has always based benefits on the range of earnings covered under the payroll tax with the understanding that it made little sense to provide the foundation benefit intended under Social Security all the way to the top end of the earnings distribution.  If we are simply going to take the cap off of earnings covered under the system without providing a commensurate increase in benefits to high earners, we will be converting the program into a welfare transfer program.  To quote Franklin Roosevelt, the program would simply be the “dole under another name.” He never intended it to be that and it will likely lose further support if that is what it is to become.  If we intend to move in that direction, then one must ask why we would want to finance it simply by taxing high wage earners and not include general tax revenues from all people with high incomes.  While there may be resistance to completely eliminating the cap on earnings covered under the payroll tax, some proposals would significantly increase the tax cap or apply a partial tax on up the earnings distribution.[23]

Another option for covering the costs of retaining the existing program is to reduce benefits.  President Bush and most other advocates of reform have established principles that would largely concentrate any benefit reductions on future retirees.  The one potential exception to this generally accepted guiding principle is the occasional suggestion that the consumer price index (CPI) be modestly adjusted to correct for what many economists believe is a tendency to overstate the rate of price inflation.  The more likely mechanisms for reducing benefits would be to adjust the current benefit formula in some way or to raise the age(s) for benefit eligibility under the program.  Once again, without an effective way to actually “save” the resulting trust fund accumulations, I do not believe this approach will actually help us solve the current financing problems.

Finding a way to partially fund Social Security obligations

An Italian proverb says: “If a man deceives me once, shame on him.  If he deceives me twice, shame on me.”  In 1982, after my earlier reading of Social Security’s history and the difficulties of funding pension obligations as they accrued in the fashion that Franklin Roosevelt wanted, I proposed transferring trust fund accumulations projected for the baby boomers’ working careers into individual accounts.  Further, I proposed that these accounts remain locked until workers reached retirement age, at which point the benefits would offset a portion of the benefits from the traditional Social Security pension.[24]  We did not institute such accounts when we amended Social Security in 1983 nor did we seriously consider them.  In the intervening period, we have accumulated a trust fund that is estimated to be at $1.7 trillion today but we have not changed the fundamental pay-as-you-go nature of the foundation to our national retirement system.  If we insist on ignoring history, we will once again be condemned to repeating it.

Unquestionably, we could craft legislation tomorrow that would mathematically rebalance Social Security within the program’s existing framework.  But balancing the system in its current configuration would build up a much larger trust fund without doing anything to ensure that accumulations would be saved rather than squandered this time around.  My objection to these approaches is that history has proven that we cannot actually save these trust fund accumulations to pay retirement costs down the road.  What’s the point of pursuing approaches that will do nothing to resolve the basic dilemma?

The Road to Accounts: Carve Out’s, Add-On’s and Hybrid Approaches

There has been a great deal of discussion about the costs associated with creating personal accounts in the context of Social Security reform.  Once again, I believe that the discussion has added little illumination to the policy matters that need to be addressed in reforming the system.  Part of the problem is the use of the terms “carve out” and “add on” do not precisely describe what is often being accomplished under various proposals.

President Bush’s general framework for financing individual accounts has generally been described as a “carve out” from the existing system.  Indeed, his critics suggest that financing the benefits in the way he proposes to do so would cost trillions of dollars over the next decade or two.  I believe that this assertion is confusing the transition costs associated with rebalancing the current system that the President proposes with the costs associated with creating the individual accounts themselves. 

In the earlier discussion about dealing with the transition costs associated with rebalancing the current system, we looked at those transition costs without considering the implications of individual accounts.  Now to understand the implications of establishing individual accounts, it is important to look at them in isolation.  To the extent we are concerned about interaction effects, we can come back and consider them later.

Assume for the sake of discussion that we have a worker at age 55 direct $1,000 of his payroll taxes into the sort of individual account that President Bush has suggested.   Table 5 sorts out how this $1,000 will be treated under two alternative scenarios.  In both scenarios, I assume that the worker retires at age 65.  For the sake of developing this example, I have assumed there is no inflation.  Adding it would change the numbers but not the substance of the outcome.  Under the president’s proposal, at retirement, this worker would have his Social Security benefit determined under whatever benefit formula applies at that particular point in time.  The lifetime value of his Social Security annuity would be reduced by $1,343.92 based on the accumulated value of the $1,000 he had withdrawn from Social Security at age 55—that is, $1,000 compounded at 3 percent per annum over 10 years. 

Table 5: Benefit Dynamics Associated with Personal Accounts in President Bush’s Social Security Reform Recommendations

Source: Derived by the author.

In Case 1, we assume that the worker has received annual returns of 5 percent per year on his account.  Under this assumption, the account would accumulate to $1,628.89 by the time he reaches age 65.  Under the president’s proposal, the worker would be required to annuitize $1,343.92 of that to replace the withdrawal he or she had made at age 55.  The extra $284.97 that is left after the required annuitization would be left for the worker to dispose of as he or she saw fit.  As I look at this example, I do not see that there is any cost associated with this transaction.  Instead of characterizing this approach as a “carve out,” it would be more appropriate to characterize it as a diversion of the payroll tax.  This individual has simply carried out an asset swap in his retirement portfolio, moving from a share of his retirement assets held in the form of Social Security accumulations to that share being held in an alternative form of financial asset.  In this particular case, the extra $284.97 would be the return for undertaking this swap.  Some analysts contend that we cannot consider this $284.97 a benefit from modifying the current system because the worker has taken on added risk in investing in assets in the financial markets.[25] 

In Case 2, I assumed that the individual received only a 1 percent annual return per year on his investment in assets in his personal account.  At retirement, his $1,000 has accumulated to only $1,104.62 and he would be required to annuitize the whole amount.  Since this is less than the annuity reduction to his Social Security benefit of $1,343.92, this worker would end up with $293.30 in reduced lifetime benefits under the modified system relative to staying completely in the central defined benefit system.  In this case, the worker incurs a benefit reduction because he has decided to put a portion of his payroll taxes in the financial markets.  President Bush’s proposal would seek to minimize this sort of loss by requiring that workers invest in broad index funds and that they move toward fixed income investments as they approach retirement age. 

In the aggregate, I believe the added benefits associated with this sort of system would outstrip the losses but there is a concern about the distribution of gains and losses that policymakers should consider in constructing a complete reform package.  No matter which of these two outcomes were to play out, to attribute the diversion of $1,000 from the Social Security fund to the personal account as a $1,000 transition cost associated with the reform of the system is wrong.  Substantial numbers of workers would definitely benefit by participating in this sort of system.  Even for those who realize a loss, that loss would be relatively minor to the overall size of the diversion of assets from the Social Security fund to their personal account.

Another issue that has been somewhat controversial in considering the diversion of payroll taxes to finance personal accounts is the prospect that  that it will exacerbate the expected cash-flow shortfall in Social Security financing in the transition period.  When commentators suggest that introducing individual accounts as part of Social Security reform will incur massive amounts of new debt, they generally do not consider the net ramifications of reform on the system.  The principles that President George W. Bush has stipulated for reform have frequently led to this criticism. 

President Bush has said that he wants individual accounts for younger workers but that he opposes benefit reductions for current retirees or those close to retirement age.  He has also said that he opposes new taxes.  Some prognosticators look at this combination of principles and conclude that individual account financing has to come out of the current revenue stream supporting the system.  They contend that a system that is already under funded cannot sustain an even further drain on revenues to finance the individual accounts.  While that may prove to be true, the principles also imply that modifications on the benefit side of the current system will reduce revenue requirements over the long term.  In an economic sense, using a government bond to temporarily finance a shift in the structure of financing Social Security so as to reduce “statutory obligations” by an amount at least equivalent to the bond amount does not create a cost.  Once again, it is simply a swap of one sort of obligation for another. 

Using government bonds to help finance the transformation of Social Security may create larger federal budget deficits in the near term than would exist under current policy, even if the transformation eliminates the long-term financing shortfall.  This is because we do not account for Social Security obligations on an accrual basis, and issuing bonds would formally recognize obligations that are not recognized in the budgetary process today.  It is not clear how financial markets might react.  In one highly publicized private case a couple of years ago, a company issued billions of dollars of corporate bonds to raise the funds to cover unfunded pension obligations, and the financial markets seemed to recognize that the borrower was simply swapping one sort of obligation for another without any real financial implications. It is not clear that the financial markets or the public would react any differently if the federal government did exactly the same thing in restructuring Social Security.

Some policymakers and analysts argue that instead of having the sort of transaction that President Bush has proposed, we should have “add-on” financing of the personal accounts.  This implies that new monies would be found to finance the accounts.  President Clinton actually proposed the establishment of USA accounts outside the scope of Social Security to give low-wage earners a mechanism to accumulate personal account wealth.  He proposed that these accounts be financed out of general revenues.  A number of Social Security reform proposals would use “add-on” funds to create personal accounts within the scope of a reformed system.  In most cases, these proposals would use money from general revenues to help finance the accounts.  Personally, I am skeptical about such proposals because of the paucity of spare general revenues for as far as my eyes will allow me to see. 

I personally have been associated with two reform proposals that would require new contributions on covered earnings as a part of the transition to a system that includes personal accounts.  Under these proposals, a portion of the current payroll tax would also be diverted to the personal accounts.  In that regard, such proposals might be characterized as a hybrid to the proposals that might depend on financing personal accounts through one mechanism or the other.  The reason that I favor some new money to help finance the individual accounts is because I believe our retirement system generally is underfunded.  The creation of personal accounts alone under the auspices of Social Security will not sufficiently ameliorate our savings shortfall.  I also believe that the added contributions should be mandatory because there are individuals all across the earnings spectrum who are saving inadequately to meet their future retirement income needs.  Possibly my biggest problem with the suggestion that we can tap general revenues to finance individual accounts is that I was born and raised in the Show Me State.  Someone is going to have to show me the source of the significant general revenues that will be required to solve this problem. 

A number of interesting opportunities to address a myriad of concerns present themselves when new money is introduced into the system.  First among these is that the saving shortfall for workers who are not adequately saving for their retirement today can be ameliorated.  Second, it gives policymakers greater opportunities to create meaningful personal accounts while maintaining the desirable insurance features in the current system.  Third, it provides an opportunity to solve the current system’s financing problems without having to intrude on any other revenue sources to get through the necessary transitions from the current system to the new one.  Even to the extent there is some transition borrowing that might be required in this sort of reform, that borrowing could be financed completely with a temporary requirement that a portion of workers’ personal account balances be invested in temporary transition bonds that would gradually be paid off over a 30 or 40 year period.