I appreciate the willingness of the Social Security Commission to consider a wide range of views on the problems facing the program. This represents a very positive development. If its final report and recommendations are to be viewed seriously, it is essential that it take these views into consideration.
I will make three main points in my comments:
1) The commission has overstepped its bounds in questioning the value of the assets held by the trust fund. When Congress passed the legislation that led to the accumulation of bonds by the trust fund, it was assigning these assets to Social Security (there is no other plausible interpretation of this build-up). The budget issues associated with paying the interest and principle on these bonds were not the charge of this commission.
2) If the system actually faces a shortfall after the trust fund is depleted in 2038, it is easy to design measures that preserve the existing system while increasing its fairness. For example, if the cap on taxable wages was raised enough to cover the same percentage of wages as the Greenspan Commission considered appropriate in 1983, it would eliminate nearly 30 percent of the projected seventy-five-year shortfall in the program. If the cap was removed completely, the increase in revenue would be almost enough to make the program fully solvent for its seventy five year planning period.
3) The commission appears to be operating under the assumption that the stock market will provide substantially higher returns than the government bonds currently held by the trust fund. There are two reasons for believing that this is not true:
a) the Social Security trustees project that real corporate profits will grow at approximately half their historic rate over the program's seventy-five year planning period, and
b) the current price to earnings ratio of the stock market is nearly twice its historic average, which has depressed dividend yields to close to half their historic average.
The only stock projections which have actually been derived from the trustees' profit projections show that real stock returns will average less than 4.0 percent over the seventy five year planning period (see Dean Baker's 2nd Letter to Martin Feldstein on Stock Market Projections and Social Security). The difference between this projected return and the return projected for government bonds will not even enough to cover the administrative cost of individual accounts. If the commission expects a proposal for investing funds in the stock market to be treated seriously, it must derive its projections for the components of stock returns (capital gains and dividend yields) in the same rigorous manner as all the projections that appear in the trustees report.
Each of these issues will be addressed in turn.
The Trust Fund is an Asset for Social Security
The Commission was asked to evaluate the future of the SS program, it was not asked to design budgets for the federal government for the next thirty or forty years. The commission has gone to great lengths to emphasize that the bonds held by the Social Security trust fund, like all government bonds, are a liability of the federal government. This claim is true, but it has no obvious relevance to the Social Security program. Under current law, these bonds are assets of the Social Security program.
The 1983 Social Security commission, which was chaired by Alan Greenspan, proposed a schedule of taxes and benefits which was projected to lead to the accumulation of a large surplus in the Social Security trust fund. This plan was approved by an overwhelming majority in both houses of Congress, which presumably intended this surplus to be an asset for the program in future years (it is difficult to imagine what else it could have intended). As a result, the Social Security trust fund has accumulated over $1 trillion in government bonds in the last 18 years. It is currently adding more than $170 billion a year to the trust fund, as its current revenue far exceeds benefits.
If there were any doubts about the intention of Congress to keep the finances of the Social Security program separate from the overall budget, these should have been eliminated when Congress voted to remove the Social Security program from the overall budget in 1990.1 As a result of this separation, the current surplus of the Social Security program will not add to the surplus of the federal government, nor will the deficit projected in future years add to the government's deficit. Under current law, when the Social Security trust fund begins to rely on the interest and principle from its bonds to pay benefits, it has exactly the same effect on the federal budget as when any other holder of federal bonds chooses to sell its bonds. At that point, the federal government can either borrow this money from another source, leaving its total debt unchanged, or it can choose to run a surplus on its budget, and pay off the debt held by the trust fund as it comes due.
The treatment of the bonds held by the trust fund as real assets is more than just a semantic issue. These bonds were accumulated as a result of deliberately setting the regressive payroll tax at a level that was higher than necessary to pay current benefits. This could be justified, if this money would eventually be used to support Social Security's progressive benefit structure. However, this implies repaying the bonds held by the trust fund with money raised through general revenue. The vast majority of general revenue is obtained through progressive individual or corporate income taxes.
If the government defaults on the bonds held by the trust fund, an option that the commission apparently viewed as acceptable in its interim report, it would lead to an enormous transfer from the bottom 80 percent of the population (the families who pay most of the Social Security tax) to the top 5 percent of the population (the families who pay most of the income tax). In a recent study, I calculated that such a default would transfer over $370 billion to the richest 5 percent of families, if it took place immediately, with $270 billion going to the richest 1 percent.2 If the default did not take place until 2016, since the fund will continue to accumulate assets in the meantime, the transfer to the richest 5 percent will be over $1 trillion, in 2001 dollars. Given the size of the redistribution that would result from a default on the trust fund's bonds, the public would be right to view this as more than a semantic question.
Trustees Projections and Alternative Scenarios
As is widely known, the Social Security trustees currently project that the trust fund will be depleted in 2038, at which point the program will only be able to pay approximately 72 percent of scheduled benefits. However, it is less widely recognized that this projection rests on assumptions that are considerably more pessimistic than those used by other government agencies, or than recent experience indicates is plausible.
In the former category, it is worth noting the trustees assumption that annual productivity growth will average 1.5 percent annually is more than a full percentage point below the 2.7 percent annual rate of productivity growth assumed by the Congressional Budget Office (CBO) in its most recent set of projections (2001).3 The growth rate assumed by the trustees implies that productivity growth will effectively remain at the low rates seen over the years 1973-1995 for the indefinite future. While not all economists accept that the productivity upturn of the late nineties will be sustained, there is little reason to expect that over the next seventy five years productivity growth will be as low as the country has ever seen.
On this topic, it also worth making a comparison to projections concerning other nations. The 1994 World Bank volume, Averting the Old-Age Crisis, (which was produced under the direction of commission member Estelle James), included a table of projections which assumed that other industrialized nations would sustain a 2.0 percent average annual rate of real wage growth over the next forty years (World Bank 1994, p 159). Keeping with the trustees projections, this table assumed that real wage growth over this period in the United States would average just 1.0 percent. This difference in growth paths implies that, in 2040, real wages in France, Germany, and other western European nations will be close to 50 percent higher than in the United States. This view seems implausible on its face. Furthermore, if it were true, it suggests that the nation faces far greater concerns than the possibility of shortfall in its Social Security program in thirty eight years.
The more plausible scenario is that a reasonable projection of real wage growth for the United States would be close to the path the World Bank projected for other industrialized nations, and to the historic average for the United States. If the projection for the average rate of real wage growth was raised by 0.5 percentage points, it would eliminate approximately half of the shortfall projected over the next fifty years, and push the date of the projected depletion of the trust fund past 2045.
The trustees assumptions on immigration also appear to be very much out of line with the evidence. The trustees assume that immigration will average just 900,000 annually. This is a projection which is supposed to apply to a period when the United States will be experiencing a serious labor shortage due to the retirement of the baby boomer cohort. By contrast, according to data from the 2000 census, over the last ten years net immigration averaged 1,330,000 per year.4 It is reasonable to expect that we will see at least as much immigration during the decades when the trustees project that the country will be experiencing a severe labor shortage, as we saw in the last decade. Large inflows of immigrants, depending on their exact timing, can significantly reduce the shortfall currently projected for the program.
However, even if the trustees projections are assumed to be accurate, the picture is not nearly as dire as polls indicate the public believes. While many workers, and especially young workers, believe that Social Security will not be there for them, the projections from the trustees show a very different picture. The program could pay all scheduled benefits for the next thirty seven years, with no changes whatsoever. Even after the trust fund is projected to be depleted, it would still pay retirees a larger real benefit than workers receive today. In fact, the projections show that a worker who is near twenty today and earns the average wage, and expects to retire around 2045, can expect a benefit of $1,320 per month (in today's dollars), or 25.5 percent more than what an average retiree would receive currently.5 This would be the case, if nothing is ever done to change the program, and the trustees projections prove entirely accurate.
It is also worth pointing out that the real source of the projected problems in the Social Security program is simply the fact that workers are expected to live longer in the future than they did in the past. Longer life spans are the expected result of growing affluence and improving medical technology. If future generations of workers choose to take a portion of their increased longevity in the form of longer retirements, then it will cost more money. There is no way around this arithmetic, but it does not present any obvious crisis. We have dealt with this problem in the past -- thankfully life-spans have been increasing since the program was created. Past generations have chosen -- through their representatives in Congress -- to meet the costs of a longer retirement primarily through increasing the contribution or tax rate for Social Security during their working years. President Bush apparently does not consider this an option at present, but given the program's financial strength, there is no reason to be considering Social Security tax increases for the foreseeable future.
As far as the more long-term future, no one knows how future generations will view their options. There is certainly no evidence to suggest that future generations of workers will be more hostile to increasing the tax rate for Social Security than were previous generations. It is also important to recognize that an increase in the payroll tax rate is not the only or the best way to increase revenue for the program. The tax is currently capped, so that it does not apply to wage income above $80,400. If this cap were removed, it would raise an amount equal to a payroll tax increase of 1.50 percentage points, nearly eliminating the seventy five year projected short-fall in the program.6 Removing the cap, combined with an increase in the payroll tax of 0.3 percentage points on both employee and employer, phased in over the years 2011-2020 (if still needed), would be sufficient to keep the program solvent over its seventy five year planning horizon.
It is important to recognize that the increase in wage inequality over the last two decades has contributed to Social Security's financial problems, since a larger portion of wage income is now going to high end wage earners, who are over the cap. The Greenspan Commission set the cap at a level which covered 90 percent of wage income. Presently, just over 85 percent of wage income is subject to the cap. If the cap were raised enough to again cover 90 percent of wage income -- the level that the Greenspan Commission considered appropriate -- it would eliminate 0.54 percentage points of the projected shortfall, or nearly 30 percent. In other words, much of the work of this commission could be accomplished simply by restoring the share of wages subject to Social Security taxes to the level set by the Greenspan Commission in 1983.
There is one final point worth noting on the questions of generational equity that have often arisen in the context of the Social Security debate. All current projections show that future generations of workers will both live longer and enjoy higher living standards than the current generation of workers. The only question raised in this debate is how much better the living standards of future generations will be, on average.
The main determinant of the well-being of future generations will be the rate of growth of the real wage. The Social Security trustees have projected that real wage growth will average 1.0 percent a year. This means that the average real wage for workers in the year 2040 will be nearly 50 percent higher than it is at present. However, many people, most notably Senator Moynihan, the commission's co-chair, have argued that the consumer price index (CPI) overstates the true rate of inflation by approximately 1.0 percentage point a year. There are two important implications to this claim, regardless of whether or not it is true.
First, an overstated CPI implies that past and future rates of real wage growth are much higher than is shown by the official CPI. If Senator Moynihan's assessment of the CPI is correct, and the Social Security trustees projections prove exactly right, then the average real wage of workers in 2040 will be nearly 120 percent higher than it is for workers today. This would have an enormous impact on any assessment of generational equity, since clearly out children and grand children will be enjoying far higher standards of living than we do at present, regardless of what is done with Social Security.
The second important implication of Senator Moynihan's view is that the real rates of return on Social Security would be approximately 1.0 percentage point higher than the numbers conventionally used. The CPI is the yardstick by which payroll taxes and benefits are converted into a common measure. If Senator Moynihan is correct in his assessment of the CPI, then all calculations of rates of return must be adjusted to take into account the overstatement, including those done by this commission in its interim report.
I have argued extensively that there is little evidence to support the view that the CPI substantially overstates the true rate of inflation.7 However, if the commission draws the opposite conclusion on this issue, then it must be consistent and accept all the implications of its position. To do otherwise would invite public ridicule.
Stock Return Calculations
The final point that I would urge the commission to take seriously, is the need to derive projections of stock returns, based on the components (dividends and capital gains), from the other projections in the Social Security trustees report, if stock ownership is part of its final plan. The trustees produce very detailed projections for all the economic and demographic variables that are relevant to the financial health of Social Security. For example, table V.A1 includes decade by decade projections of life expectancy and birth rates. Table V.B1 includes decade by decade projections of wage growth, inflation, labor force growth, interest rates, and other relevant variables. The trustees report also includes careful explanations of how these numbers are derived and the sensitivity of the calculations to varying each assumption.
It would be possible to construct similar projections for capital gains and dividend yields as I have already done (see Dean Baker's 2nd Letter to Martin Feldstein on Stock Market Projections and Social Security). M.I.T. economist Peter Diamond has also written on this topic, explaining in great detail how projections of stock returns can be derived from the profit growth projections which are implicit in the Social Security trustees projections.8
This issue is important both as a question of consistency, and because the existing projections do not support the assumptions on stock returns put forward by advocates of investing Social Security money in the stock market. In terms of consistency, it makes no sense to have detailed projections for all the relevant demographic and economic variables, without making similar projections for stock returns. If all the other projections prove completely accurate, but the assumptions on stock returns turn out to be too optimistic, then the program will find itself with severe financial problems. If it is important that the projections for other economic or demographic variables be subject to rigorous scrutiny, then it must be important that the projections for stock returns are subject to the same sort of scrutiny.
The second reason why it is essential that the commission derive projections of stock returns from the other variables is that the existing projections contradict claims by proponents of stock returns as to what yields can be anticipated. My own work showed that real returns will average close to 3.6 percent annually, far below the 7.0 percent conventionally assumed by advocates of investing Social Security money in the stock market. Professor Diamond's work showed that 7.0 percent returns would be possible, but only after the market declined by close to 50 percent from its current levels.
The reasons why it will be impossible to get historic rates of returns given current price to earnings ratios are simple. First, the price to earnings ratios are presently close to double their historic average. Even with recent declines, the average price to earnings ratio for the market as a whole remains over 25 to 1. This compares to a historic average of approximately 14.5 to 1. As the price to earnings ratio has risen, the dividend yield (including share buybacks) has fallen proportionately. Historically the dividend yield has been close to 4.0 percent of the share price. It presently hovers near 2.0 percent (including share buybacks). This is an inevitable result of the higher than normal price to earnings ratio. Firms cannot pay out more money in dividends just because their share price is high. They generally pay out between 50 to 60 percent of after-tax profits as dividends or share buybacks. With the current price to earnings ratio, this implies a dividend yield of just over 2.0 percent.
The other reason why projections of stock returns derived from the Social Security trustees projections are lower than past returns is that the trustees project that profit growth will be considerably lower in the future than in the past. Over the last fifty years, real profit growth has averaged more than 3.0 percent annually, measured against the consumer price index. Real profit growth is projected to average just over 1.5 percent annually (also measured against the consumer price index) over the next seventy five years. Assuming that the price to earnings ratio stays approximately the same in the future (no one considers it plausible that it would rise indefinitely) the average real growth rate of stock prices must be approximately equal to the 1.5 percent annual projected growth rate of corporate profits.
Adding the 2.0 percent dividend yield to the 1.5 percent real growth in share prices gives a 3.5 percent real return. This is approximately half of the 7.0 percent real return assumed by proponents of investing Social Security money in the stock market. The difference between the 3.5 percent projected return on stocks and the 3.0 percent real return projected for government bonds is not likely to be large enough to even cover the administrative costs of individual accounts.
If there is anything missing from this analysis, it should be quite simple for the economists working with the commission to explain the flaw. It would take a competent economist no more than an hour to derive projections of stock returns from the Social Security trustees profit projections. Before, hundreds of billions of workers' Social Security money is placed at risk in the stock market, it seems reasonable to expect that the commission would produce an analysis that would at least show how it is possible to have the returns from the stock market which it is assuming.
In conclusion, I would urge the commission to produce a final document that is internally consistent and meets the same standards that the Social Security Administration applies to its work. This means first and foremost, that the report's assumptions about stock returns -- and other issues -- must be consistent with other assumptions that appear in the trustees report, and are accepted by the commission. A report that is not internally consistent cannot be taken seriously by Congress or the public.
3 The Economic and Budget Outlook: Fiscal Years 2002-2011. The importance of this difference for the Social Security projections is reduced somewhat by the fact that CBO assumes a considerably larger gap between the rate of inflation as measured by the GDP deflator and the rate of inflation as measured by the consumer price index. However, even after adjusting for this effect, the CBO projection would still add more than 0.5 percentage points to the rate of real wage growth projected by the trustees.
5 This number is derived from projected benefits that appear in the updated version of table III.B.5 in the Social Security trustees report. The projected benefit is multiplied by the percentage of benefits that are projected to be covered by tax income.
6 This figure is derived from the estimate in the Report of the 1994-1996 Advisory Council on Social Security (p 238), that raising the cap enough to cover 90 percent of wage income would reduce the projected shortfall by 0.5 percentage points of payroll. Since this rise in the tax would raise the share of wage income covered from approximately 85 percent to 90 percent, it is assumed that raising the share to 100 percent would reduce the projected shortfall by three times as much.
Dean Baker is the co-director of the Center for Economic and Policy Research (CEPR). He is the author of Plunder and Blunder: The Rise and Fall of the Bubble Economy. He also has a blog on the American Prospect, "Beat the Press," where he discusses the media's coverage of economic issues.