Published October 1, 1997
National Committee to Preserve Social Security and Medicare
The Economics of Pay-as-you-go Social Security,
and the Economic Cost of Ending It
a report by
Senior Vice President & Chief Economist
Lehrman Bell Mueller Cannon, Inc.
for the National Committee to Preserve Social Security and Medicare
1. The Inadequacy of “Neoclassical” Economic Theory
Economists are notorious for making assumptions that aren’t necessarily true — as in the old story about a group of economists stranded on a desert island with a case of wine, and no way to open it. “I know!” one economist chirps. “Assume we have a corkscrew.”
Rather than assuming things into existence, the debate over the economic theory of pay-as-you-go Social Security is mostly about assuming things away that really do exist. Those who seek to end pay-as-you-go[i] Social Security base their argument on an economic theory devised about 100 years ago, for quite a different purpose. They ignore the fact that this “neoclassical” theory was challenged nearly 40 years ago as inadequate to explain economic growth, and disproven by careful research more than 20 years ago. The trouble with neoclassical theory lies not in its logic but in its assumptions, which do not match the world in which we live. The theory entirely ignores the existence and economic value of investments in people.
The neoclassical economic theory and its policy prescriptions could accurately be described as follows:
“It seems that economic output and our standard of living can be explained by three factors: labor, capital, and technical progress. Frankly, we haven’t the foggiest idea of what causes people to raise more or fewer future workers; nor do we understand where technical progress comes from. We will therefore suppose that the size of the labor force grows at some constant rate, and that technical knowledge advances at some constant rate.
“It follows that the only remaining way to increase output or raise our standard of living is to increase net investment in business plant and equipment. If saving responds to the after-tax rate of return (and this appears to be the case), then the most efficient way to increase investment in plant and equipment is to reduce or abolish taxes on the income produced by such investment, while increasing taxes on workers’ incomes. With higher taxes, workers may decide to work fewer hours than before; but we don’t believe that this will be large enough to offset the positive effect of increased investment in plant and equipment.
“The approach may seem unfair, but it is really in the workers’ own best interest. It is only by reducing workers’ consumption for awhile — and this should not take more than two decades — that we can give them more tools to work with. And by our assumptions, increasing the ratio of machines to workers is the only sure way to increase workers’ wages and their future consumption.”
In a now famous address in 1960, economist Theodore W. Schultz challenged this “neoclassical” theory as an explanation of economic growth, pointing out that it was contradicted by the facts:
“The income of the United States has been increasing at a much higher rate than the combined amount of land, man-hours worked, and the stock of reproducible capital used to produce the income. . . . To call this discrepancy a measure of ‘resource productivity’ gives a name to our ignorance but does not dispel it. . . . Unless this discrepancy can be resolved, received theory of production applied to inputs and outputs as currently measured is a toy and not a tool for studying economic growth.”[ii]
2. Schultz’s Thesis: Total Capital, Human and Nonhuman
Schultz went on to present his own hypothesis. It is probable, Schultz said, that the productivity of labor and capital does not change much over time, if at all. Rather, the apparent rise in their productivity is due to the way in which these inputs are measured: “Investment in human capital is probably the major explanation of this difference. Much of what we call consumption constitutes investment in human capital. Direct expenditures on education, health, and internal migration to take advantage of better job opportunities are clear examples.”[iii]
Schultz argued that the main thrust of the neoclassical theory was therefore wrong: “Laborers have become capitalists not from a diffusion of corporate stocks as folklore would have it, but from the acquisition of knowledge and skill that have economic value.”[iv]
Schultz’s address greatly stimulated interest and activity in efforts to explain economic growth. Some economists, like Gary Becker, undertook to refine the theory of what Schultz had called”human capital.” Just as all property income is the return on investment in nonhuman capital, Becker argued, all labor compensation is the return on investment in human capital.[v]
This means that neither growth of the labor force nor rising productivity due to “technical progress” can be taken for granted. There can be little technical progress without investment in research and development. There can be no growth of the labor force without a prior investment in child-rearing. There can be no increases in income resulting from education, training, health, safety or mobility, unless there has been prior investment in education, training, health safety or mobility. And though public spending on education, safety. etc., is far from negligible, more than three-quarters of all investment in human capital is due to direct investments by families.[vi]
In the meantime, John W. Kendrick, already a pioneer in the field, decided to test Schultz’s hypothesis by careful measurement. In 1976, Kendrick published The Formation and Stocks of Total Capital, a body of research which he has periodically updated. After four decades of research, Kendrick concluded that Schultz’s hypothesis was proven: “the total capital approach. . . provides an effective explanation of most of the rate of growth of real (adjusted) GDP.”[vii] Kendrick showed that, while part of the economic growth left unexplained by neoclassical economic theory was due to investment in research and development, most was due — exactly as Schultz had argued — to investment in “human capital.”
Because this point is so central to the debate over Social Security, it is worth spending some time examining the facts in more detail.
3. Where Does Economic Growth Come From?
In the first place, Schultz’s theory and Kendrick’s research indicate that total (human and nonhuman) saving and investment is much larger than conventional (and official) measures, which confine saving and investment to the accumulation of plant and equipment.
In fact, total investment is nearly three times as large a share of the economy as the official measures suggest. Gross investment (before deducting depreciation of capital) ranges between 40% and 50% of gross domestic product, rather than 15% to 20% (see Graph 1). Net investment (after subtracting allowances for depreciation of capital) is in the range of 20% of net national product — also nearly three times as high as the conventional measures.[viii]
Moreover, while the share of income devoted to investment in plant and equipment has had a slightly declining trend over time, total investment has increased substantially as a share of the economy, though the absolute peak occurred around 1973.[ix] (See Graph 2.)
The stocks of human and nonhuman capital are the cumulative result of many years of investment. And the total stock of human capital is larger than the total stock of nonhuman capital (see Graphs 3 and 4).
Kendrick also showed that the relation between growth of total (human and nonhuman) capital and growth of total output is essentially one-for-one. From 1929 to 1990, total capital grew at an average rate of 2.9% a year in real terms, and total real output grew at a 3.0% rate. (See Graph 5.)
In the U.S. business economy (excluding government and households) — for which the data are usually more reliable, since relatively little has to be imputed — the relationship between growth of total capital and growth of total output is apparent, not just on average, but also during relatively short time periods, such as a business cycle. Faster or slower economic growth has coincided with faster or slower growth of total human and nonhuman capital (Graph 6).
Moreover, the growth rates of human and nonhuman capital speed up and slow down at the same time (Graph 7). This is because human and nonhuman capital are both necessary for any increase in output, in roughly constant proportion.
However, human and nonhuman capital do not contribute equally to output. Human capital produces about two-thirds, and nonhuman capital about one-third, of output and real income. In the domestic business economy (excluding government and households), human capital has contributed about 71% and nonhuman capital 29% of economic growth (24% due to investment in plant and equipment, and 5% to investment in research and development). (See Graph 8.) In the whole U.S. economy (including government and households), human capital has contributed about 63% and nonhuman capital 37% of economic growth (32% due to buildings and machines, and 5% due to research and development).[x] (See Graph 9.)
These relative contributions of human and nonhuman capital to total output explain why the returns to human and nonhuman capital — labor compensation and property compensation, respectively — are divided as shares of total national income in about the same proportion (Graphs 10 and 11).
However, Schultz and Kendrick pointed out that the national income acounts and the tax code both treat labor compensation and property compensation disparately. Net property income earned from investment in nonhuman capital is measured only after subtracting the costs of maintenance and depreciation. But forlabor compensation earned from investment in human capital, both costs are treated (and taxed) as if they were net income.[xi] Graphs 12 and 13 show how national income would be classified if labor compensation and property compensation were treated alike: at least half of labor compensation would be deducted to avoid double-counting of the same income.
Finally, Kendrick worked out average real pretax rates of return on investment for both human and nonhuman capital. In business cycle peak years, the real rates of return on net capital in the business economy averaged about 12.5% for human capital and 10.4% for nonhuman capital (Graph 14); in the broader total domestic economy, the real rates of return averaged 11.3% for nonhuman capital and 6.9% for nonhuman capital (Graph 15).[xii]
Comparing these real rates of return with growth of the capital stock indicates that investment in both human and nonhuman capital responds to changes in the real rate of return. Both seem to be about equally sensitive to variations in the rate of return.
4. Theories of Pay-as-you-go Social Security
The main difficulty in the debate over Social Security is that most antagonists on both sides still use the outdated neoclassical theory and its assumptions.
The first serious attempt to explain the economics of pay-as-you-go Social Security was by Paul Samuelson.[xiii] In 1958, Samuelson developed a simple theory showing that, in the long run, the rate of return on pay-as-you-go Social Security would be about the same as the rate of economic growth. And this rate of return would be higher than the average interest rate on lending and borrowing. The argument implied that pay-as-you-go Social Security increases economic welfare. However, in this simple model, all loans were assumed to be for the purpose of consumption.
In 1966 Henry Aaron, using assumptions more closely approximating those of neoclassical theory, reached a similar conclusion.[xiv] Aaron showed that pay-as-you-go Social Security increases total economic welfare, as long as the rate of economic growth exceeds the risk-free interest rate on other investments (which has been the case on average in the past).
In the 1970s, however, Martin Feldstein sought to overturn the conclusion that pay-as-you-go Social Security increases economic welfare.[xv] Feldstein ridiculed Samuelson’s theory as one-sided in assuming that there is no investment in plant and equipment. In its place, he offered an equally one-sided theory, which assumed that allinvestment consists solely of plant and equipment.[xvi]
During its start-up phase, Feldstein said, the rate of return on Social Security is much higher than the return on other investments. This causes people to save less and consume more, he argued. Feldstein attempted to measure the decline in saving which he believed Social Security must have caused.[xvii]
Samuelson was right about one thing, Feldstein said: the average future rate of return on Social Security must equal the rate of economic growth. But, said Feldstein, the appropriate rate of return for comparison with Social Security is not a risk-free interest rate, as Aaron had argued — but rather the rate of return on plant and equipment before all taxes, which is higher. Individual investors do not receive this full return on their financial investments, because of corporate and personal income taxes. But the value to society, Feldstein said, should be measured by the total pre-tax rate of return on investment in plant and equipment.
A pay-as-you-go Social Security system must therefore reduce economic welfare, according to Feldstein: “In reducing private saving, social security causes the substitution of a low-yielding implicit intergenerational contract [Social Security] for real capital investment with a higher social yield” [plant and equipment].[xviii]
The economic loss, Feldstein said, must equal the difference between these two rates of return, multiplied by the (alleged) reduction in saving caused by Social Security. Therefore, he concluded, there must be an economic gain from ending pay-as-you-go Social Security, equal to the present value of this amount, less the original economic gain received by the first generation of retirees.
Each of the theories just outlined ignores the existence of “human capital.” Before examining why this changes everything, we must note one more source of confusion.
5. Disagreement Among the “Privatizers”
The debate over the merits of pay-as-you-go Social Security is further confused by the fact that many who later joined Feldstein in proposing to end pay-as-you-go Social Security apparently do not understand his reasoning.
This group includes those associated with the Washington, D.C., based Cato Institute, who propose to end Social Security and replace it with individual retirement accounts, while leaving the tax code essentially unchanged.
Using what they apparently consider a variation on Feldstein’s argument, the Cato group argues that simply replacing pay-as-you-go Social Security with private individual savings accounts would bring a large economic benefit. The reason, they argue, is that the return on investments like common stocks has been higher than the growth rate of the economy.
In a separate paper, we show that this conclusion is untenable when risk is taken into account.[xix] In the past, the average rate of return on financial investments, adjusted for risk, has always been lower than the average rate of economic growth — which, as we have seen, approximates the average long-run return on pay-as-you-go Social Security.
This fact alone is sufficient to explain why replacing pay-as-you-go Social Security with mandatory financial accounts is a bad idea. But the Cato group does not seem to understand that, according to Feldstein also, merely replacing Social Security with financial investments would bring about an economic loss, not an economic benefit — even if risk could be ignored.
The reason, as Feldstein explains, is that “capital income taxes reduce the net return that individuals receive on private savings to approximately the implicit rate of return that they earn on social security tax contributions.”[xx]
This means that, even if the risks on Social Security and financial saving were exactly the same, “privatizing” Social Security would reduce economic welfare. The after-tax return on saving, the size of the capital stock and the size of the economy could be no larger than before. Thus there could be no way to pay for the large transition cost of ending pay-as-you-go Social Security. At least one generation would have to pay payroll taxes without receiving Social Security benefits paid by the following generation — and also save for its own retirement at a rate of return no higher than could be realized from pay-as-you-go Social Security.
6. Essential Feature of Feldstein’s Plan: A Tax Hike on Workers
This is why the plan advocated by Feldstein (and a few other economists) involves two parts: not only replacing pay-as-you-go Social Security with mandatory financial accounts, but also, at the same time, “a shift from the current income tax to a consumption tax or a labor income tax.”[xxi]
The details of Feldstein’s particular plan are ratherimpractical.[xxii] However, the main question is not whether Feldstein’s plan makes sense, but whether any plan along such lines could make sense. This is a serious and intelligent question; and the serious and intelligent answer is “no.”
As Feldstein explains, “The essential feature of the transition to a funded program of retirement benefits is a period of reduced consumption by employees during the early years of the transition so that a dedicated capital stock can be accumulated. This dedicated capital is then used to finance retirement benefits, thereby permitting lower taxes and more consumption by employees in later years.”[xxiii]
According to Feldstein, the reduction in consumption would be accomplished by a tax increase on workers’ incomes. Abolishing at the same time all Federal, state and local taxes on the private investment in plant and equipment funded by retirement saving would raise the return on retirement saving. The higher rate of return would bring forth additional saving and increase the size of the economy. In Feldstein’s view, the additional economic growth would pay for the “transition cost” of ending pay-as-you-go Social Security within a couple of decades.
The problem, as we have seen, is that the economic theory of the “privatizers” ignores the existence of human capital. The theory defines all income not spent to purchase machines or buildings as “consumption,” not investment. But as Kendrick’s research shows, the majority of such income is not devoted to “consumption,” but rather to investing in and maintaining human capital.
Feldstein’s plan therefore cannot work in any form, because it would substitute more investment in nonhuman capital for less investment in human capital — which yields a higher pretax rate of return. The result would necessarily be a smaller, not a larger economy.
7. Did Social Security Cause the Baby Boom?
To understand the effect of ending pay-as-you-go Social Security, perhaps the most helpful starting point is to consider what happened when pay-as-you-go Social Security began. And the first question we must ask is this: Did pay-as-you-go Social Security cause the Baby Boom?
This may seem like an odd question, and one difficult to answer. Social Security was enacted in 1935, workers began paying payroll taxes in 1936, the first benefits were paid in 1941, and the Baby Boom began in the 1940s. But the startup of Social Security also overlapped with many other major economic events: the end of the Great Depression, the Second World War, and the beginning of the Cold War, to name only a few.
However, the question will not go away. The point on which everyone agrees is that the first generations of workers covered under Social Security received extraordinarily high real rates of return on their contributions.
Also, demographers like Richard Easterlin have shown that the chief economic influence on the decision by any generation to have more or fewer children, compared with the preceding generation, is the real income of that generation, relative to the preceding generation.[xxiv]
A simple way to measure the rate at which Social Security affects the wealth of successive generations is to compare the Social Security benefits paid to retirees in a given year (as a share of taxable payroll) with benefits as a share of payroll about 25 years earlier — approximately the length of a generation. When we compare this “rate of intergenerational transfer” with the total fertility rate, we find an uncanny resemblance; moreover, the intergenerational transfer precedes the fertility rate by a few years. (See Graph 16.)
Whatever other factors may have been important, we at least cannot reject the theory that the “wealth effect” of pay-as-you-go Social Security — which the “privatizers” argue must have had a large effect on the behavior of American families — must have played an important role in the Baby Boom.
This should not surprise us. When working families can expect a higher (risk-adjusted) average return from Social Security than from nonhuman capital, they have more wealth to put into an investment they prefer to any other — their children.
As Kendrick showed, the economic rate of return on expenses likechild-rearing and education, in terms of increased future labor earnings, has been about 4.4 percentage points higher than the average return on all nonhuman capital (Graph 15). This has also been about 5 percentage points higher than the average real return on common stocks — yet with only a quarter of the risk caused by variations in return.
But many families, especially poorer families, do not have the means to finance all the investments in their children that would yield this return. And we can expect that this will always be the case (barring a return to the barbarities of human slavery or indentured servitude). It will always be harder to borrow for an investment that is embodied in a human being instead of a piece of property.[xxv]
Of its nature, therefore, pay-as-you-go Social Security is a method of financing more investment in human capital than could otherwise occur.[xxvi]
This expansion of investment in human capital, in turn, makes the economy larger than it would otherwise be. To see how this happens, let’s consider the effect of the Baby Boom on investment and economic growth, using Kendrick’s figures.
The rise of the birth rate during the Baby Boom coincided closely with an acceleration in investment of all kinds from 1948 to 1973 — which caused an acceleration in growth of real GDP (Graph 17). The strongest acceleration in investment occurred in human capital (Graph 18). But, according to the figures we examined earlier, investment in nonhuman capital also increased — both as a share of the economy and in its rate of growth (Graphs 1 and 2). Relatively more investment went into human capital. But both human and nonhuman capital are necessary for increases in output. As a result, there was an acceleration of investment even in nonhuman capital.
Thus, Feldstein is correct in arguing that pay-as-you-go Social Security profoundly affected the savings habits of American households. He is also correct to believe that it caused a substitution away from investment in nonhuman capital. But he is wrong to assume that the result was an increase in the share of the economy devoted to “consumption.” Consumption increased in absolute terms as the economy expanded. But the fact that total investment (as well as investment in nonhuman capital alone) increased as a share of the economy means that the share of national income devoted to consumption declined sharply after the introduction of Social Security, above all during the Baby Boom.
The evidence is unambiguous: pay-as-you-go Social Security financed more investment and more economic growth than could otherwise have occurred.
8. The Economic Cost of Ending Pay-as-you-go Social Security
Any effort to end pay-as-you-go Social Security must throw this process into reverse — resulting in a smaller population, lower total investment as a share of the economy, and a smaller economy.
Raising taxes on labor compensation, as Feldstein and other neoclassical theorists propose, cannot increase investment by reducing the consumption of workers — any more than raising taxes on property income would reduce the consumption of physical capital. What is taxed in each case is the return on investment in human or nonhuman capital; what is reduced in each case is investment in human or nonhuman capital.
Therefore, the notion that it is possible to abolish taxes on “investment” and tax only “consumption” is a pipe-dream. Every penny of income in the economy is the return on some kind of investment: there is nothing to tax but the return on investment. Removing all taxes from investment would mean abolishing taxes altogether. This means that the best fiscal policy is to make government spending as efficient as possible, and to levy any necessary taxes on the returns to investments in human and nonhuman capital as equally as possible.
Just as starting up pay-as-you-go Social Security caused a substitution toward more investment in human capital relative to nonhuman capital, ending pay-as-you-go Social Security would cause a substitution toward more investment in nonhuman capital relative to human capital.
But in the first case, the result was a larger economy; in the second case, the result would be a smaller economy. The reason is that human investment is responsible for about three times as much economic growth as investment in buildings and machines; and because the pretax rate of return on human capital is higher than the pretax rate of return on nonhuman capital.
To grasp this point, let’s consider the arithmetic of ending pay-as-you-go Social Security as Feldstein sees it.
Feldstein adopts the projections of the Social Security actuaries, who assume that, largely because of reduced future population growth, the average rate of economic growth over the next 75 years will be less than half the rate observed in the past — about 1.4% instead of about 3%.
If this happens, the next 75 years will amount to an Economic Ice Age. Yet, judging by the past relation between the fertility rate and economic growth, this projection about economic growth is not unreasonable — as long as the birth rate behaves as the actuaries assume. On the other hand, the birth rate has defied official projections for most of the past 60 years. And according to Easterlin, the same economic factors that contributed to the rise and fall of the birth rate during and after the Baby Boom should cause the birth rate to rise again in the future.[xxvii]
However, we don’t need to question the actuaries’ projections to show why Feldstein is wrong in believing that pay-as-you-go Social Security incurs a net cost rather than a net benefit.
If real GDP growth in the business economy averages 1.4% a year, then we can expect that 1.0% percentage point (71%) of the growth will be accounted for by investment in human capital; about 0.3% percentage point (24%) by investment in equipment and buildings; and about 0.1% percentage point (5%) from investment in research and development. (See Graph 19.)
Feldstein estimates that, as the result of raising taxes on workers and abolishing taxes on new investment in plant and equipment, the stock of tangible business capital would grow over the next 75 years by about 34% more than otherwise. Total real output and national income would be about 8% higher (a 34% increase in plant and equipment, times the 24% contribution of plant and equipment to economic growth). This represents an average increase in real GDP growth of 0.1% per year.
However, these calculations assume that the size of the stock of human capital would be unaffected.[xxviii] The conclusion reverses once we recognize that a tax increase on workers would reduce investment in child-rearing, education and other kinds of human capital.[xxix]
Let us conservatively suppose that the tax increase on labor compensation proposed by Feldstein would reduce investment in child-rearing, education, etc., by 17% — only half as much as Feldstein claims the stock of nonhuman capital would increase. Because human capital contributes about three times as much to economic growth as plant and equipment, the tendency for less investment in human capital must outweigh any tendency for more investment in nonhuman capital. Since both types of capital are necessary for production in roughly constant proportion, the value of both human and nonhuman capital would have to decline. After 75 years, the economy would be 4% smaller, not 8% larger, as a consequence of Feldstein’s plan.[xxx]
The future economic loss from ending pay-as-you-go Social Security can also be measured in another way. Feldstein claims that ending pay-as-you-go Social Security would reap an economic benefit equal to at least $3.2 trillion in today’s dollars.[xxxi] This calculation assumes that the average future return on investment in plant and equipment will be 8 or 9 percentage points higher than the average future return from Social Security.
The trouble with this calculation is that Feldstein is comparing the pre-tax return on nonhuman capital with the after-tax return on human capital. Just as the return on financial assets to individual investors is reduced by taxes already levied on the underlying investment in nonhuman capial, so also the return on Social Security received by workers is reduced by the taxes already paid on the underlying investment in human capital.[xxxii] To be consistent, therefore, Feldstein must use either the pretax or the after-tax rates of return for both human and nonhuman capital, but not mix them.
A tax increase on labor income, combined with a tax cut on property compensation, would reduce investment in human capital, in an effort to increase investment in nonhuman capital. But the average pre-tax real rate of return on human capital is significantly higher than the corresponding rate of return on nonhuman capital — as we have seen, about 4.4 percentage points higher in the total domestic economy.[xxxiii] Moreover, this difference in rates must also be adjusted for the fact that the average volatility risk of investing in nonhuman capital is more than twice the risk of investing in human capital.
Correcting this inconsistency reverses Feldstein’s calculation:Far from producing an economic gain, ending pay-as-you-go Social Security would cause an economic loss of about $3 trillion in today’s dollars.[xxxiv]
9. Conclusion: Keep Social Security Pay-as-you-go
The conclusions of our analysis are straightforward:
1. Neoclassical economic theory is totally inadequate for guiding policymakers on issues like pay-as-you-go Social Security — particularly since a better theory, confirmed by decades of careful research, has long been available.
2. The argument for abolishing pay-as-you-go Social Security rests on assumptions which are contradicted by the facts — particularly the neoclassical theory’s assumption that there is no such thing as investment in “human capital.”
3. Ending pay-as-you-go Social Security must incur a large economic cost, not an economic benefit.
Aaron, Henry J. (1966), “The Social Insurance Paradox,” Canadian Journal of Economics and Political Science vol. 32, August, 371-77.
Becker, Gary S. (1991), A Treatise on the Family, Harvard University Press, Cambridge, Mass.
Becker, Gary S. (1994), Human Capital, 3d ed., University of Chicago Press, Chicago; 1st ed. (1964), Columbia University Press for the National Bureau of Economic Research, New York.
Boskin, Michael J. (1986), “Theoretical and Empirical Issues in the Measurement, Evaluation, and Interpretation of Postwar U.S. Saving,” in Savings and Capital Formtion: The Policy Options, F. Gerard Adams and Susan M. Wachter, eds., Lexington Books, Lexington, Mass.
Cartwright, William S. (1984), “Saving, social security and private pensions,” International Social Security Review Vol. 37 no. 2, 123-138.
Chand, Sheetal K. and Albert Jaeger (1996), “Aging Populations and Public Pension Schemes,” Occasional Paper 147, International Monetary Fund, Washington, D.C., December.
Duggan, James E., Robert Gillingham and John S. Greenlees (1993), “Returns Paid to Early Social Security Cohorts,” Contemporary Policy Issues vol. XI, October, 1-13.
Easterlin, Richard A. (1980), “American Population Since 1940,” in The American Economy in Transition, Martin Feldstein, ed., University of Chicago Press, Chicago, 275-321.
Feldstein, Martin (1974), “Social Security, Induced Retirement, and Aggregate Capital Accumulation,” Journal of Political Economy, September/October, 905-926.
Feldstein, Martin (1977), “Does the United States Save Too Little?” AEA Papers and Proceedings, Vol. 67 no. 1, February, 116-121.
Feldstein, Martin (1982), “Social Security and Private Saving: Reply,” Journal of Political Economy Vol. 90 no. 3, 630-642.
Feldstein, Martin (1994), “Fiscal Policies, Capital Formation and Capitalism,” NBER Working Paper 4885, National Bureau of Economic Research, Cambridge, Mass., October.
Feldstein, Martin (1995), “Social Security and Saving: New Time Series Evidence,”NBER Working Paper 5054, National Bureau of Economic Research, Cambridge, Mass., March.
Feldstein, Martin (1996), “The Missing Piece in Policy Analysis: Social Security Reform,” AEA Papers and Proceedings, May, 1-14.
Feldstein, Martin (1997), “Privatizing Social Security: The $10 Trillion Opportunity,” Cato Institute SSP No. 7, Washington, D.C., January.
Feldstein, Martin and Andrew Samwick (1997), “The Economics of Prefunding Social Security and Medicare Benefits,” NBER Working Paper 6055, National Bureau of Economic Research, Cambridge, Mass., June.
Hubbard, R. Glenn (1986), “Pension Wealth and Individual Saving: Some New Evidence,” Journal of Money, Credit and Banking vol. 18 no. 2, May, 167-178.
Kendrick, John W. (1976), The Formation and Stocks of Total Capital, Columbia University Press for the National Bureau of Economic Research, New York.
Kendrick, John W. (1994), “Total Capital and Economic Growth,” Atlantic Economic Journal Vol. 22 no. 1, March, 1-18.
Leimer, Dean R. and Selig D. Lesnoy (1982), “Social Security and Private Savings: New Time Series Evidence,” Journal of Political Economy Vol. 90 no. 3, 606-629.
Kotlikoff, Laurence J. (1979), “Testing the Theory of Social Security and Life Cycle Accumulation,” American Economoic Review vol. 69, 396-410.
Kotlikoff, Laurence J. and Laurence H. Summers (1981), “The Role of Intergenerational Transfers in Aggregate Capital Formation,” Journal of Political Economy, vol. 89 no. 4, August, 706-732.
Mackenzie, G.A., Philip Gerson, and Alfredo Cuevas (1997), “Pension Regimes and Saving,” Occasional Paper 153, International Monetary Fund, Washington, D.C., August.
Samuelson, Paul A. (1958), “An Exact Consumption Loan Model of Interest with or without the Social Contrivance of Money,” Journal of Political Economy vol. LXVI no. 6, December, 467-482.
Schultz, Theodore W. (1959), “Human Wealth and Economic Growth,” The Humanist 1959 no. 2, 71-81.
Schultz, Theodore W. (1961), “Investment in Human Capital,” American Economic Review Vol. LI no. 1, March, 1-17.
Schultz, Theodore W. (1962), “Reflections on Investment in Man,” Journal of Political Economy Vol. LXX no. 5 part 2, October supplement, 1-8.
The Economics of Pay-as-you-go Social Security,
and the Economic Cost of Ending It
a report by
Senior Vice President & Chief Economist
Lehrman Bell Mueller Cannon, Inc.
for the National Committee to Preserve Social Security and Medicare
The “neoclassical” theory of economic growth, relied upon by those who seek to “privatize” Social Security, was challenged by Theodore W. Schultz nearly 40 years ago, and disproven by the research of John W. Kendrick more than 20 years ago.
The current paper recounts the neoclassical theory’s shortcomings as a guide to policy, analyzes the recent debate about Social Security, and shows how pay-as-you-go Social Security financed more investment, especially in “human capital,” than could otherwise have occurred. Ending pay-as-you-go Social Security would throw this process into reverse.
The paper concludes by calculating the economic cost of ending pay-as-you-go Social Security. After 75 years, the U.S. economy would be about 4% smaller. The present value of the economic loss is about $3 trillion.
[i] “Pay-as-you-go” means that each generation pays the retirement benefits of its parents’ generation.
[ii] Schultz (1961), 6.
[iii] Schultz (1961), 1.
[iv] Schultz (1961), 3.
[v] “Particularly in developed economies but perhaps in most, there is sufficient investment in education, training, informal learning, health and just plain child rearing that the earnings unrelated to investment in human capital are a small part of the total. Indeed, in the developmental approaches to child rearing, all the earnings of a person are ultimately attributed to different kinds of investment made in him.” Becker (1994), 111.
[vi] Becker (1994), 209.
[vii] Kendrick (1994), 16.
[viii] Since most economists are familiar with the idea of “human capital,” it seems that ignorance of the research done by Kendrick and others is the only explanation as to why so many prominent economists continue to define saving as plant and equipment, excluding investment in human capital (and in R&D).
Michael Boskin seems representative of this attitude: “such investment [in human capital] is likely to be important. However, those who argue for inclusion of human capital expenditures in savings figures ignore the depreciation of the stock of human capital. . . . While further refinements of measures of human investment expenditures and depreciation of the stock of human capital are desirable, they are unlikely to raise the net saving rate more than a small amount (perhaps 1% or so) and are unlikely to cause us to alter the conclusion that currently the net national saving rate in the U.S. is low, falling, and well below its optimum.” Boskin (1986), 28.
Yet Kendrick’s work, published 10 years earlier, carefully accounted for depreciation of human capital, and showed that including investment in human capital raises the net saving rate by about 10 percentage points — roughly doubling Boskin’s estimate. In fact, the actual net saving rate measured by Kendrick equals or exceeds Boskin’s calculation of the “optimal net saving rate” for the U.S. economy.
[ix] “In contrast to the declining secular trend shown by the conventional series, all of our measures of total capital formation show a significant rise in the proportion of income and product saved and invested between 1929 and 1969.” Kendrick (1976), 127. This conclusion remains true in the light of more recent data; Kendrick (1994).
[x] The reason for the lower contribution of human capital to output in the total economy, strange as it may seem, is that households and governments employ relatively more nonhuman capital — houses, cars, natural resources, defense equipment — than businesses.
[xi] Schultz (1961), 13. Kendrick (1976), 30, 122.
[xii] Average real rates of return on gross capital (before subtracting depreciation) were slightly higher. Kendrick (1976), 118-125; Kendrick (1994), 7, 13-15.
[xiii] Samuelson (1958).
[xiv] Aaron (1966).
[xv] Feldstein (1977).
[xvi] Feldstein called Samuelson’s argument about pay-as-you-go Social Security “brilliant but mischievous.” “I say mischievous because the economy that Samuelson considered has no capital stock and indeed no nonperishable goods. A social security system, or a system of money, was therefore the only way in which individuals could provide for their old age. In an actual economy, as I have emphasized, the unfunded social security program displaces private saving and real capital accumulation.” Feldstein (1994), 21.
The point is well taken: nonhuman capital contributes about a third of GDP. But of the two simplifications, Samuelson’s is the less unrealistic, because it ignores investment in the factor responsible for only one-quarter of economic growth (plant and equiment), while Feldstein ignores investment in the factor that provides three-quarters of economic growth (human capital). By saying that pay-as-you-go Social Security “displaces private saving and real capital accumulation,” Feldstein assumes that investment in “human capital” is not saving or “real” capital accumulation.
[xvii] Feldstein (1974).
[xviii] Feldstein (1977), 119. We note that, in Feldstein’s theory, investment in human capital is not “real capital investment.”
[xix] “Can Financial Assets Beat Social Security? Not in the Real World,” companion paper.
[xx] Feldstein (1994), 15-16. By “capital income taxes” Feldstein means taxes paid by corporations before interest and dividends are paid to individual investors: chiefly taxes on corporate profits.
But Feldstein also includes state and local property taxes, which violates standard accounting practice. Since “indirect” taxes are paid before any “factor income” — labor compensation or property compensation — there is no reason to believe that they fall more heavily on property income than on labor income. To be consistent, indirect taxes should either be omitted or else attributed proportionately to labor and property income. At the very least, if property taxes are attributed entirely to property income, all other (sales-type) indirect taxes should be attributed to pretax labor compensation.
[xxi] Feldstein (1994), 14. In economic terms, a “consumption” tax and a tax on labor income are essentially the same.
[xxii] Feldstein proposes to rebate all Federal, state and local taxes on investment in nonhuman capital financed by retirement saving. This, he argues, would provide a real rate of return of about 9% on retirement saving. But Feldstein has yet to grapple with the practical difficulties that this would involve. In its current form, Feldstein’s plan, taken literally, amounts to the Federal government guaranteeing each and every individual investor a 9% real return on any financial investment, without restriction. If so, the Federal governnment would suddenly find itself borrowing at a Treasury bill rate 9 percentage points above the inflation rate, while simultaneously guaranteeing a 9 percent real return on any private financial investment, regardless of risk — and, incidentally, freezing the relative prices of all private capital in place. The possibilities for making a financial killing at public expense, by legal or fraudulent means, would make the 1980s savings and loan fiasco look like a Sunday-school picnic. It seems highly likely that any future evolution of Feldstein’s plan must be in the direction of ever more complex and minute regulation of permissible investments and permissible rates of return on private investments.
[xxiii] Feldstein and Samwick (1997), 32.
[xxiv] Easterlin (1980). Gary Becker points out that this is not necessarily always the case. A rise in wealth will tend to cause young couples to have more children. But a rise in the cost of raising each child works in the opposite direction. Becker (1991), 135-178. Therefore, young couples will raise more children than their parents only if the “wealth effect is stronger than the “price effect.” As we will see, however, the “wealth” effect appears to have been considerably stronger, at least since pay-as-you-go Social Security began.
[xxv] “[U]nderinvestment in knowledge and skill, relative to the amounts invested in nonhuman capital, would appear to be the rule and not the exception for a number of reasons. . . . Where the capital market does serve human investments, it is subject to more imperfections than in financing physical capital. I have already stressed the fact that our tax laws discriminate in favor of nonhuman capital.” Schultz (1961), 14, 15.
[xxvi] “By combining publicly subsidized schooling with a social security system, countries may have found a very crude and indirect, but perhaps reasonably effective, way to provide loans to children that get repaid when the parents are old and collect retirement benefits.” Becker (1994), 22. Of course, public schooling is only one of many forms of investment in human capital, financed in this way.
[xxvii] Easterlin (1980), 295.
[xxviii] Feldstein sometimes assumes that higher tax rates on labor compensation would cause workers to work fewer hours, but he does not account for the fact that there will be fewer workers, less educated workers, etc. Feldstein and Samwick (1997).
[xxix] “Even a modest tax on births can have a large negative effect on the number of children.” Becker (1994), 23.
[xxx] The weighted effect on output of nonhuman capital would be an increase of 8% (.24 times .34) — if human capital were unchanged; but the weighted effect of human capital would be a decline of 12% (.71 times .17) — if nonhuman capital were unchanged.
[xxxi] “Consider an unfunded [pay-as-you-go] social security program that begins with an initial tax of T dollars that is then transferred to the then current retirees. If the marginal product of capital is r and the real growth rate of aggregate wages is g (the sum of the growth rates of the labor force and productivity per worker), the annual welfare loss in year t is (r-g)Tegt and the present value of this stream of welfare losses is I (r-g)Tegt = (r-g)T/(d-g) where d is the appropriate discount rate.” Feldstein (1994), 22n.
Feldstein uses values ranging from .09 to .12 for r, from .02 to .04 for d, from .01 to .03 for g, and from 5% to 6% of GDP for T. In Feldstein (1994), using r = .12, g = .03, d = .04 and T = $400 billion, he calculates the welfare loss due to pay-as-you-go Social Security to be $3.2 trillion [=(.12-.03)($400 billion)/(.04-.03)=$3.6 trillion, plus losing the original $400 billion gain]. With other assumptions, he calculates the welfare loss at up to $20 trillion: Feldstein (1997).
[xxxii] Feldstein is correct to observe that the return on financial investments to individual investors is reduced by corporate profits taxes. He ignores the fact that the return on Social Security is also reduced by the taxes paid on the underlying investment in human capital that produced such labor income.
Labor compensation is not subject to a separate corporate profits tax, but it is subject to at least two layers of tax not paid by owners of nonhuman capital: the costs of capital maintenance and depreciation are tax-exempt for nonhuman capital, but fully taxable for workers. As a result, the burden of personal income taxes alone, ignoring the payroll tax, falls more heavily upon human capital than the combined burden of personal income and corporate profits taxes falls upon nonhuman capital.
This can be proven by laborious calculation (for example, the author’s testimony to the National Commission on Economic Growth and Tax Reform, June 21, 1995). But the simplest way to see this is to recall Feldstein’s statement that the return on private saving for individual investors is about the same as the return on Social Security — and compare this with Kendrick’s calculations showing that the pretax rate of return on human capital is higher than on nonhuman capital.
[xxxiii] Kendrick (1976), 124, 240. Most economists would accept — even insist [Boskin (1986), 28] — that depreciation is as real a cost for workers as for machines. But some would argue against Schultz’s and Kendrick’s deduction of human “maintenance” costs (essentially, the cost of keeping body and soul together) from net labor compensation, on the ground that staying alive affords satisfaction, and so should be considered “consumption” (Eisner , 16).
Such an assumption would add about 10 percentage points to the real rate of return on human capital, meaning that its real rate of return is up to three times as high as the return on nonhuman capital. “The fact that the adjusted rate is much closer to the rate of return on nonhuman capital helps support the theoretical arguments for adjusting human as well as property returns to exclude maintenance expenses” (Kendrick , 122).
[xxxiv] We recall that Feldstein’s formula for calculating the gain or loss due to pay-as-you-go Social Security is (r-g)T/(d-g), where r is the pretax return on nonhuman capital, g (= real economic growth rate) is the return on Social Security, d is the discount rate, and T is the initial gain from starting up Social Security.
To be consistent, the first “g” in the equation should be replaced with “w,” signifying the pretax rate of return on the human capital displaced by the tax increase on labor income that Feldstein advocates. The pretax rate of return on net human capital, according to Kendrick, has been about 4.4 percentage points higher than the return on nonhuman capital in the total domestic economy. Moreover, the average volatility of the return to nonhuman capital (approximated by a balanced portfolio of stocks and bonds) is about three times the corresponding risk on nonhuman capital (approximated by the volatility of GDP). This further widens the risk-adjusted difference, by about 2 percentage points.
Since only the difference in rates of return on human and nonhuman capital matters, we will use the difference in Kendrick’s rates of return (adjusted for the difference in risk) to ensure comparability. We will leave the other numbers as in Feldstein (1994). For the total domestic economy r – w = -.063. In this case, ending pay-as-you-go Social Security incurs an economic loss of $2.92 trillion [=[-.063)($400 billion)/(.04-.03)=$2.52 trillion, plus returning the original $400 billion gain].