Shameful Moments in Economics I:
The Social Security Debate
by Dean Baker
June 12, 1999
When I first began studying economics as an undergraduate I occasionally came across references to it as "the dismal science." At the time, I thought this was a sarcastic commentary on the quality of the work in a discipline that had pretenses of being a science. It was only years later that I realized that the reference was to Malthus' gloomy prediction, that the bulk of the human race would always be at the edge of starvation, as population growth inevitably ran up against the limits of the food supply. However, the longer I work in economics, the more strongly I come to believe that my original understanding was closer to the mark.
For the last five years I have been working in Washington on issues directly or indirectly associated with Social Security. As a result I have often found myself near the center of public debate on this issue. What I have seen has not been pretty. I have witnessed a major national debate driven almost entirely by misconceptions. Scare tactics and exaggerations have figured more prominently than evidence and arithmetic. Basic principles of logic have been jettisoned.
The economics profession does not deserve the entire blame for this state of affairs. Politicians have been happy to exploit fears about Social Security for political advantage, thereby increasing public confusion on the issue. Too often they have succeeded in this strategy, because reporters covering the issue lack sufficient expertise in economics to independently evaluate the truth of competing claims. While economists cannot be held responsible for the character of politicians or the quality of economic reporting, they should be accountable for failing to stand behind basic truths of economics and arithmetic, in the same way that biologists would be expected to stand behind the theory of evolution in the face of an onslaught of creationism.
I will briefly examine four issues relevant to the Social Security debate in which the voice of most economists has been too muted or lacking altogether. The issues are:
1) the overall size of the problem,
I have selected these issues because they all raise points where reasonable people cannot disagree. In each case, there are simple truths, which would not be disputed among economists, but which have not yet permeated into the public debate. I apologize for the fact that much of the ensuing discussion is autobiographical, but I would be telling the story from where I sit in any case, so there seems little point in attempting to feign a more distant stance.
Adjectives in Economics: Is the Problem Big or Small?
One of the nice things about the Social Security debate is that we have a widely accepted estimate of the magnitude of the problem. We can all pick up the Social Security trustees report and see that there is seventy five year shortfall that is projected at 2.07 percent of taxable payroll. For purposes of the debate, virtually everyone has accepted this as a starting point. (I will qualify this point later in the discussion of the 4th issue.) Some people, like myself, have pointed to the fact that the projections of future GDP growth are quite pessimistic relative to past trends. Others have argued that the projections somewhat overstate the health of the program, because they underestimate increases in life expectancy. Nonetheless, the Trustees numbers form the framework for the debate. This means that there is wide agreement about the actual magnitude of the problem.
If there is agreement about the magnitude of the problem, then it should be possible to agree on how it is presented to the public. There will always be a considerable element of subjectivity in the choice of adjectives. One person's "crisis" may be another person's "problem." But there should be little issue about questions of ordering. In other words, a "crisis" should always be bigger than a "problem." And the degree of urgency involved in addressing the Social Security shortfall should bear some relationship to its potential impact on living standards.
Last year I did a very simple study in which I compared the projected impact of changes in demographics, the continuation of recent trends in inequality, and the projected increase in health care costs, on the median family's standard of living (Baker 1998). The comparisons focused on projections for the year 2030, the peak of the baby boomer retirement. My numbers showed that the projected increase in health care costs (for non-demographic reasons) would reduce the income that the typical family had left over, after paying taxes and health care costs, by more than twice as much as the costs associated with the aging of the population. The numbers showed that the continuation of recent trends in inequality would reduce the typical family's after tax income by more than three times as much as the costs associated with the aging of the population (see Figure 1).
Comparative Impact on Median Family After-Tax After-Health Income, Year 2030
(Source: Baker 1998)
It is important to realize that the costs of aging in this comparison included not just higher Social Security expenditures, but also the higher taxes that would be needed to pay for Medicare and Medicaid, as a result of an aging population. If the Social Security shortfall is a "crisis," then what should the public make of these other problems that will potentially have much larger impacts on their standard of living?
While others may produce somewhat different projections than I did in my study, the orders of magnitude cannot really be debated. The Health Care Financing Administration (HCFA) projected that per capita health care costs would rise by more than $160 annually on average over the next seven years, even after adjusting for the overall inflation rate (HCFA, 1998). For a family of three, this would imply, on average, an annual increase in health care costs of more than $480 in real terms. By contrast, if the full 2.07 percentage point tax increase needed to balance the Social Security program were put in place tomorrow, it would mean a one-time increase of approximately $600 in taxes for the average worker.
Similarly, there have been several years during the last two decades in which the real median wage has dropped by more than one percent, in spite of increases in average wages and productivity (Mishel, Bernstein, and Schmitt, 1999). If the counter-factual is that workers' wages keep pace with productivity growth, then their real wages should have risen by approximately 1.0 percent annually over the last two decades. This means, in effect that the median worker has often lost more than 2.0 percent of their wages in a single year due to a rise in inequality. By this standard, there have been many single years in which growing inequality has taken a larger bite out of the wages of a typical worker, than the full tax increase that would be needed to keep the program solvent for seventy five years.
In short, if the projected shortfall in Social Security poses a crisis, then the projected increase in health care costs and the prospect of the continued growth of inequality present something far worse than a crisis. It seems hard to justify focusing public debate on the relatively minor problem of the Social Security shortfall, when the problems of rising health care costs and inequality threaten to have a much larger impact on the typical family's living standards.
Some may object that the focus on Social Security is justified, since this is a problem that can be solved. The experience of other OECD countries suggest the opposite. The United States is alone in experiencing such an enormous run-up in health care costs. Every other industrialized nation has managed to achieve comparable, or better, health care statistics, while spending a much smaller share of their GDP on health care. Similarly, among OECD nations, only Britain and Canada have experienced the same sort of rise in inequality. The fact that other nations have largely or completely avoided the problems of rising health costs and growing inequality suggests that these are problems that can be addressed through public policy.
Whatever adjectives one chooses to apply to the projected Social Security shortfall, it is clear that other problems pose much greater threats to the living standards of the typical working family. By this criterion, economists should have been yelling much louder about these other problems than about Social Security. In fairness, there have been some prominent economists who have attempted to put the projected Social Security shortfall in perspective. Henry Aaron, Robert Reichsauer, and the late Bob Eisner, stand out in this respect. But there have been all too many who have been eager to jump on the "crisis" bandwagon to further other agendas. This has allowed the public debate to remain far removed from reality.
The Missing Stock Market Projections
Most of the plans currently on the table for restructuring Social Security provide for investing funds in the stock market in some form. President Clinton's proposal, which has been supported by many of the liberal economists in the debate, calls for investing in the stock market directly through the trust fund. On the other side are a wide range of partial privatization proposals that call for investing money in the stock market indirectly through private accounts. In both cases, the reason for looking to the stock market is the expectation that the returns on stocks will be far higher than the return on government bonds. The assumption of a higher return allows for the trust fund to be balanced with fewer benefit cuts or tax increases, or alternatively, allows private accounts to provide a far higher return than the existing system.
I have argued in detail that it will not be possible to for stocks to provide the same rate of return in the future as in the past, given current stock valuations and the Trustees projections of future profit growth (Baker, 1997b). At present, the price to earnings ratio in the stock market averages more than 30 to 1. This is more than twice its historic average, which has been less than 15 to 1. The growth rate of profits projected by the Trustees is less than 1.5 percent a year, this is also less than half its average over the last fifty years, which has been more than 3.0 percent annually. Based on current stock valuations and projected profit growth, I have projected that the price to earnings ratio would have to rise above 1000 to 1, by the end of the Social Security planning period, in order to produce the 7.0 returns assumed by advocates of investing in the stock market (Figure 2).
Based on the current record high price to earnings ratio (and therefore a record low dividend to price ratio), and the slow projected growth rate of profits, I have projected that real stock returns will average just 3.5 percent over the next seventy five years, not the 6.75-7.0 percent assumed by most proponents of investing in the stock market. With a 3.5 percent return, there would be little point in placing money in the stock market, since there is not much difference between this yield and the 3.0 percent return projected for government bonds. In fact, the difference in returns is probably not even large enough to cover the administrative costs in the case of private accounts.
I first put forward this argument nearly three years ago. I have discussed it in a variety of public forums including the Washington Post and Atlantic Monthly. The argument was also the topic of an essay by New York Times columnist Peter Passell ("Economic Scene." 3-6-97; D2). I have also tried to make a considerable pest of myself to all the proponents of investing in the stock market, challenging them repeatedly in private and public letters to produce projections that will support their assumptions.
I believe that I have put forward a solid logical argument that it will be impossible for returns in the stock market to average anything close to 7.0 percent, or even 6.0 percent, given the Trustees projections for profit growth, unless the stock market crashes in the near future. (No proponents of privatization are on record as assuming a market crash in the near future.) While my argument may be wrong, this could be easily demonstrated by producing an alternative set of projections of the components of stock returns, dividends and capital gains, which are consistent with the Trustees projections of profit growth. In fact, there are no projections of the components of stock returns that would support the returns routinely assumed by advocates of investing in the stock market.
Apparently, the economists involved in this debate do not believe it is necessary to make such projections. While the actuaries at the Social Security Administration examine in considerable detail their assumptions about wage growth, birth rates, life expectancies and other relevant variables, the consensus among the economists involved in this debate is that we can just plug in an assumption on stock returns that is simply a naive extrapolation from the past.
I have never been terribly fond of seventy five year projections, as I have frequently stated. But, if we make these projections for every other factor that is relevant to the solvency of the Social Security program, it is difficult to see any rationale for not making similar projections for dividends and capital gains, if the stock market is going to play a role in the financing of the program. No one honestly concerned about the future solvency of the program can argue differently.
As things stand now, very few people, including members of Congress and reporters at major newspapers, are even aware that the Social Security Administration has no projections for stock returns. In fact, it is often asserted that the Social Security Administration projects that stock returns will average 7.0 percent in the future or scored a plan based on its own projections of stock returns (see e.g. "House G.O.P. Chiefs Forgoing Revision of Social Security" by David E. Rosenbaum, New York Times, April 23, 1999, page A1). The economists involved in this debate deserve most of the blame for the fact that such a blatant inaccuracy has been allowed to become so pervasive in the public's perception.
The Consumer Price Index: The Yardstick of Generational Affluence
In July of 1995, the Senate Finance Committee appointed a commission to examine the accuracy of the consumer price index (CPI). The reason for their interest in the methodology used to construct the CPI, was that tax brackets and many government benefits, most importantly Social Security, are indexed to the CPI. If the CPI could be adjusted to show a lower rate of inflation, then tax collections would increase and government expenditures would decrease, without any legislative action whatsoever. In terms of eliminating the deficit that existed at the time, a lower CPI was every politicians' dream.
In December of 1996, in the midst of enormous media fanfare, the Boskin Commission (named after Michael Boskin, the commission's chairman) issued its long awaited report. It concluded that the CPI overstated the annual inflation rate by an average of 1.1 percentage points. Based on this assessment, it recommended that the indexation formulas be altered so as to more accurately reflect the true increase in the cost-of-living.
Had their recommendation been implemented, it would have had an enormous impact on Social Security beneficiaries. They would have seen a 1.1 percent reduction in their benefits each year, compared with a baseline where their benefits rose in step with the CPI in place at the time. After ten years, the cumulative reduction would have been 10.5 percent. After twenty years, the reduction would have been 19.8 percent. There were relatively few Social Security beneficiaries in their mid-eighties, who would not have experienced considerable hardship, if their Social Security benefit had been 19.8 percent lower.
There were two issues raised by the Boskin Commission's report. The first issue was its accuracy. Even a superficial reading of the report would undermine confidence in its conclusion. Many of its findings were based on nothing more than introspection or crude extrapolations from dated research. It is reasonable to ask for more substantive evidence before discarding a core government statistic. There were prominent economists who made this argument at the time, such as Barry Bosworth at Brookings. Katherine Abraham, as Commissioner of the Bureau of Labor Statistics, also conducted herself with great integrity, insisting that any changes to the index itself (Congress always had the option to change indexation formulas, regardless of the accuracy of the CPI) be guided by research findings rather than political convenience.
While the Boskin Commission's assessment of the accuracy of the CPI is a debatable point, the second issue, the implications of a significant CPI overstatement of inflation, is not. If the CPI overstates the true rate of inflation by 1.1 percentage points annually, then measures of real wage growth, which use the CPI, understate the annual rate of real wage growth by 1.1 percentage points. This is an inescapable logical truth.
Yet, virtually no economists were willing to discuss this point and its implications. If the true annual rate of real wage growth is 1.1 percentage points more rapid than is indicated by the current CPI, then workers will be far richer in the future than had previously been projected. In fact, they will be far richer in the future than we could have even hoped possible when we were looking at the world with our overstated CPI. Looking out from 1995, if the true annual rate of inflation was 1.1 percentage points less than the CPI in existence at the time, then the real wage projected for 2030, would be more than 45 percent higher than was being projected with the overstated measure. By comparison, the Congressional Budget Office projected that the impact of a large increase in saving, sustained over 35 years, would increase per capita income in 2030 by just 1.7 percent (see Figure 3) (1996, p 90).
The flip side of the more rapid projection of wage growth in the future is that wage growth must have been more rapid in the recent path. This logically implies that people had been much poorer in the recent past than is indicated by data deflated with the CPI. I did some extrapolations backward of income levels in the past assuming an overstated CPI and found that the median family income may have been below the current poverty line as recently as 1960 (Baker 1997a). This means that if the Boskin Commission is right about the CPI, most of the people currently receiving Social Security grew up in poverty.
In short, if the CPI substantially overstates inflation, then it logically follows that concerns about generational equity must be turned on their head. People will be far wealthier in the near future than our current projections show and much poorer in the recent past than current data indicate. If the Boskin Commission is correct in its assessment of the CPI, then there would be much weaker policy grounds for cutting Social Security benefits than if they were wrong.
Remarkably, virtually no economists have been willing to state this simple logical truth in the context of the Social Security debate. As a result, the claim that the CPI substantially overstates inflation often appears in an argument about protecting the well-being of future generations.
I once appeared on a radio show with former Wyoming Senator Alan Simpson. Based on his concern for future generations, Senator Simpson argued strongly that the cost of living adjustment should be reduced because the CPI overstates inflation. Not content with the Boskin Commission's 1.1 percentage point estimate, Senator Simpson argued that it could be as much 1.5 percentage points, or even 2.0 percentage points. As the estimate of the overstatement went higher, his projection of future living standards went lower. By the Senator's account, his 2.0 percentage point estimate would mean our children and grandchildren will be living in chicken coops.
The problems in constructing an accurate cost-of-living index are enormous. There are many areas of this debate where economists can have honest disagreements. But there can be no honest disagreement about the implications of a significant overstatement. There should have been no shortage of economists willing to tell Senator Simpson that he had gotten it completely backwards. If his 2.0 percentage point estimate is correct, our children and grandchildren will be living in penthouses, not chicken coops.
Falling Wage Growth Projections: Do We Only Get Bad News?
Earlier I stated that all parties to the debate accept the Trustees projections as defining the problem. While this is true, that does not mean that the projections should not be subject to serious scrutiny. Those who follow the debate closely know that panels of experts are periodically convened to evaluate the Trustees assumptions concerning the variables relevant to the program's solvency. The most recent panel was convened in connection with the 1994-1996 Advisory Council. The 1996 Trustees Report was the first one issued with the benefit of the recommendations from this expert panel.
In this report, the projection of long-run real wage growth was 1.0 percent annually. Since this report was issued three years ago, the projection for real wage growth has been revised downward by 0.5 percentage points (see Figure 4). This downward revision appears difficult to justify on its face, since real wage growth has been extremely rapid during the last three years, averaging 3.2 percent. While it may be unreasonable to revise seventy five year wage growth projections upward based on the experience of just three years, there was no new information that would provide any obvious justification for the large downward revision between 1996 and 1999.
In fact, virtually no one involved in the Social Security debate is even aware of this downward revision. The major reason for this ignorance is that most of the revision was concealed by changes in the CPI. Between 1996 and 1999, the BLS implemented changes in the CPI that reduced its measure of the rate of inflation by at least 0.4 percentage points annually. Some analysts have produced estimates of the impact of these changes which are considerable higher. For example, the Council of Economic Advisors estimated the impact at 0.68 percentage points (1999 Economic Report of the President, p 94). In 1996, the Trustees Report estimated that the rate of long-run real wage growth would be 1.0 percent, measured against the CPI in place at the time. Since the current CPI reports a rate of inflation that is at least 0.4 percentage points lower relative to the true rate of inflation than 1996 CPI, the real wage growth projection in the 1996 Trustees Report would be 1.4 percent measured against the current CPI.
In this case, as with the previous ones, the issue is one of logic and accounting, not a dispute about values or competing claims on evidence. The Trustees have lowered their projections of wage growth by 0.5 percentage points in the wake of the strongest real wage growth we have seen in three decades. This should at least have been a subject of some public discussion, with the Trustees being forced to explain why they have a so much more pessimistic view of the future in 1999 than they did in 1996.
If the 1996 projection had not been revised downward, the projected shortfall would be reduced by 0.52 percentage points, eliminating fully one quarter of the problem. The date at which the fund is projected to be depleted would have been pushed out to 2040, nearly forty years into the future. The whole shortfall in Social Security's finances would appear considerably smaller, if we were still looking at the same real wage growth projections in 1999 as we were in 1996. As it is, this downward revision in the wage growth projection has gone largely unnoted because of the failure of the economists working on the topic to call it to public attention.
Professional Standards: Does Economics Have Room?
The economics profession has served the public poorly in the Social Security debate. At the most basic level, it has failed to present the public with a realistic assessment of the magnitude of the problems faced by the program. As a result, politicians can freely invent any story they choose about the plight of the program without risk of contradiction by the experts in the field. Similarly, reporters operate largely in a sea of confusion, with little understanding of the issues they are trying to explain to the public.
Economists, like everyone else, bring their views to political debates. But in the discussion of economic issues, it should be possible to set out to the public a clear set of parameters on which we do agree. That has not happened in the Social Security debate. In fact, in some ways the economics profession has actually added to confusion, for example by implying that someone has made projections of stock returns for the future which show the same yields as in the past.
It should be expected that politicians will care little for accuracy in discussing policy issues. Similarly, the arguments put forward by various interest groups often bear only a casual relationship to the truth. Reporters generally view their job as writing down what their selected sources say, not attempting to verify whether what they are saying is true.
For the most part, economists have maintained no greater commitment to the truth than these other actors in the debate. As a result, any changes in Social Security that come out of the current debate will likely bear little relationship to what is actually needed or to what the public really wants. In the mean time, issues that will have far more impact on people's lives will have been pushed off the political agenda, as grossly exaggerated depictions of Social Security's financial problems crowd out more substantive discussions.
Figure 1 shows the percentage loss, due to each of the four factors shown, in the income that the median family would otherwise have left over in the year 2030 after paying their taxes and health expenditures. The counterfactual assumes that the demographic composition of the country is the same in 2030 as at present, that health care costs only rise in step with overall GDP growth and the aging of the population, and that there is no increase in inequality.
The first bar shows the percentage decline, measured against this baseline, in after tax after health care expenditure family income, due to the demographic changes that are projected to take place between 1995 and 2030. The second bar shows the loss in after tax after health care expenditure family income, due to the projected increase in health care expenditures in excess of the part attributable to aging and GDP growth. The third bar shows the loss in income attributable to the continuation of recent trends in inequality. The fourth bar shows the combined effect of the projected increase in health care costs and the continuation of recent trends in inequality. The construction of this chart is explained in more detail in Baker 1998.
Figure 2 shows projections of the price to earnings ratio in the stock market, assuming that it provides 7.0 percent real returns each year, and that profits grow at a 1.5 percent annual rate, approximately the rate projected by the Social Security trustees. It is assumed that 60 percent of profits are paid out as dividends each year and that the rest of the return is derived from capital gains.
Figure 3 shows real per capita income in 1960, 1995, and 2030 assuming that the Boskin Commission's estimate, that the consumer price index overstates the annual rate of inflation by 1.1 percentage point, is correct. This means that the annual growth in real per capita income is 1.1 percentage point larger than the numbers indicated with data deflated with the current CPI. For the 1960 per capita GDP calculation, the 1960 per capita GDP figure (1999 Economic Report of the President, table B-31) was converted into 1992 dollars using a CPI that shows 1.1 percentage point less inflation annually. A similar calculation was used to adjust the CBO's projections for capita income in the year 2030 (1996, p 90). The "CBO Good Saver" bar shows the impact of increasing the annual saving rate by an amount equal to approximately 2.5 percent of GDP. It reflects the increase in per capita GDP that CBO projected in a scenario where the budget was balanced every year, compared with the scenario where the debt to GDP was held constant over the thirty five year period.
Figure 4 shows the long-term rate of real wage growth that appears in both the 1996 and 1999 Social Security Trustees report. The third bar shows what the rate of real wage growth that appears in 1999 would be if it were measured against the same CPI as was used for the 1996 projection. The difference between the second and third bar reflects the methodological changes that have been implemented in the construction of the CPI over the last three years. These changes have lowered its measure of the annual rate of inflation by at least 0.4 percentage points.
Baker, D. 1998. Defusing the Baby Boomer Time Bomb: Projections of Income In the 21st Century. Washington, D.C.: Economic Policy Institute.
Baker, D. 1997(a). Getting Prices Right: The Battle Over the Consumer Price Index. Armonk, New Jersey: M.E. Sharpe Press.
Baker, D. 1997 (b). Saving Social Security With Stocks: The Promises Don't Add Up. New York: The Twentieth Century Fund
Congressional Budget Office, The Economic and Budget Outlook: Fiscal Years 1997-2006. Washington, DC: U.S. Government Printing Office, 1996.
Economic Report of the President, 1999. Washington, D.C.: U.S. Government Printing Office.
Health Care Financing Administration, 1998. National Health Expenditure Amounts.
Mishel, L, J. Bernstein, and J. Schmitt. 1999. The State of Working America 1998-99. Armonk, N.J.: M.E. Sharpe.
Center for Economic and Policy
Phone: (202) 293-5380 Fax: (202) 588-1356 Home: www.cepr.net