<- Retour

William Vickrey, 1996 Nobel Award in Economics

October 5, 1996

Government debt is thought of as a burden handed on from one generation to its children and grandchildren.

Reality: Quite the contrary, in generational terms (as distinct from time slices), the debt is the means whereby the present working cohorts are enabled to earn more by fuller employment and invest in the increased supply of assets, of which the debt is a part, so as to provide for their own old age. In this way the children and grandchildren are relieved of the burden of providing for the retirement of the preceding generations, whether on a personal basis or through government programs.
This fallacy is another example of zero-sum thinking that ignores the possibility of increased employment and expanded output. While it is still true that the goods consumed by retirees will have to be produced by the contemporary working population, the increased government debt will enable more of these goods to be to be exchanged for assets rather than transferred through the tax-benefit mechanism.

In some ways the result of such deficit financing is analogous to the extension of a social security retirement scheme to provide added benefits to middle and upper incomes beyond the existing caps to the wages and earnings subject to social security contributions and the corresponding benefits. There are important differences, however. The Social Security System is indeed often criticized as being in effect a kind of Ponzi scheme in which benefits to earlier cohorts are financed by taxes on later cohorts. The scheme is kept from collapsing by virtue of its being compulsory so that there will always be succeeding cohorts to foot the bill, though possibly by higher or lower tax rates, unlike private schemes, which tend to collapse when it is discovered that the emperor has no clothes and new contributors shy away.

This Ponzi element was, however, necessary to get the program off the ground during the depression. Retirees were given pension payments far beyond what would have been financed by their contributions and only a relatively small reserve fund was accumulated to allow for adventitious differences between receipts and outlays. Even so, the relatively brief lag between the onset of social security contributions out of payrolls and the beginning of substantial payments to retirees constituted a withdrawal from purchasing power, aggravated by the exclusion of the revenue in computing the formal deficit, adding to pressure to reduce government's net addition to purchasing power, and to overall pessimism stemming from the preception of deficits as symptoms of economic ill-health. These impacts substantially aggravated the drop in industrial production in the fall of 1937, by far the sharpest ever recorded.

Currently the amount by which the present value of expected future payments to current participants exceeds that of expected future contributions by them is a real liability of the government that is probably at least as inescapable as that represented by the formal debt. While the schedules of payments are subject to alteration by act of Congress, whether by changing the age of retirement, subjecting more of the payments to income tax, or otherwise, political pressures are likely to require at least some degree of indexation for inflation, so that on balance the real burden is likely to prove as unavoidable a real "entitlement" obligation as that of the formal debt, which is to a much greater extent subject to possible erosion through accelerated inflation. The amounts are not small; one estimate has put the capital value of governments entitlements, including military and civil service pensions, at over 3 years of GDP, though such estimates are necessarily subject to a wide range of uncertainty.

The situation could be formally regularized by a bookkeeping entry that would add to the assets of the Social Security System and to the explicit liabilities of the government. However, this would be a purely formal move that should in principle be of negligible practical significance, though a Congress obsessed with reducing the formal deficit might seize upon this recognition of a liability as an excuse for further inappropriate budgetary stringency. In any case the macroeconomic impact is measured not by the magnitude of the government liability, however calculated, but by the value placed on these entitlements by the potential beneficiaries in making decisions as to saving and consumption.
Many have even complained that the investment of the small actual social security reserves in special government securities amounts to the diversion of social security contributions to government expenditure. But the situation would be no different if the Social Security Administration were to invest in private securities instead, with the private insurance industry switching its reserve funds from private to government securities. The only real impact of moving the social security system "off budget" would lie in the reaction of Congress to the enlargement of the nominal deficit by the disregarding of the growth in the social security reserve. Should the Congress react to offset this increase by budget tightening, the result would be an increase in unemployment produced as a result of a notional rescuing of the social security reserve from being "squandered" in government expenditure.

Setting aside, as irremediable bygones, the subsidizing of the earlier cohorts, for those currently paying payroll taxes the relevant reality (as distinct from arbitrary accounting conventions) is that the relation between the taxes paid by or on behalf of any individual and the present expected value of future benefits is extremely loose. Overall, if one were to apply the rules currently on the books to a steady demographic state of a constant population with a constant expectation of life, with the relatively small social security reserve fund kept at a constant level, present value of benefits payable to a given cohort would fall short of the net present value of the taxes paid during its working life by the difference between the interest that would have been earned by a full actuarial reserve and the smaller amount of interest paid on the recorded reserve. From this viewpoint, looking only at the future, there would thus be a net contribution from the social security system to the general-purpose fisc, much larger, actually, than the amount involved in the charge that the addition to the small nominal reserve is being improperly appropriated to current government expenditures.

In terms of actual demographic changes, a growing population and a lengthening expectation of life both mean that if the reserve fund were held constant, current cohorts still gain at the expense of later cohorts. In practice this is somewhat modified by differentials between total current tax revenues and total current benefit payments, reflected in fluctuations in the reserve fund.

Within each cohort, the often arbitrary and even capricious operation of the complex formulas by which benefits are determined mean that the relation between taxes paid at any given time by a given individual and the consequent increase in expected eventual benefits varies widely and often capriciously. At one extreme, many of those who accumulate less than 40 quarters of covered employment over their working life will not become eligible for any benefits; their contributions are effectively a tax on their wages, whether nominally paid by themselves or their employer. Examples are women who start work at 18 but marry and leave the labor force at 25, or "empty nesters" who enter the labor force for the first time at age 54 or later; for such persons squeezing in a fortieth quarter of coverage could be extremely lucrative.
Even for most of those who do become eligible, there is an arbitrary exclusion from the formula of the 5 years of lowest indexed annual covered earnings, so that for these years the contributions are again a pure tax. This is particularly unfortunate in that these lowest years are in most cases the earliest years of employment, at ages for which unemployment rates are highest, and the effects of the tax most unfortunate.

Benefits are not paid on the basis of taxes paid but on the basis of covered wages, which means that those employed during years in which tax rates were low obtain benefits as though they had paid taxes at the later higher rates. On the other hand, in computing benefits wages are indexed, not by a price index or by a compound interest factor, but by a nationwide average wage, which has tended to grow at a rate significantly below an appropriate rate of interest. The result is that over a period of constant tax rates, taxes on earlier wages purchase fewer benefits in terms of present value than those on later wages.

Benefits are determined on a fairly steeply progressive basis, being roughly 90% of the first $5,000 of the individual's average indexed annual wages, 32% of wages between $5,000 and $30,000, 15% between $30,000 and $60,000, and 0% above $60,000. The result is a fairly substantial transfer from high-wage earners to low-wage earners. Low-wage earners may actually receive, as a group, benefits exceeding in present value that of the payroll taxes paid on their earnings, while a relatively large part of the payroll taxes paid on higher wages would be effectively a tax rather than a premium.
Because of this low return in terms of benefits on taxes on wages in the $30,000-$60,000 bracket, the fact that no payroll taxes are levied on wages above this $60,000 cap produces a highly anomalous dip in the combined marginal effective tax rate on earnings as earnings rise above this cap. Not only is this inversion of progression inefficient in terms of incentives, it even opens the door to an arrangement whereby an employer would agree with his employee to pay $20,000 and $100,000 in alternate years, instead of a constant $60,000. This would reduce the payroll taxes payable while producing only a relatively minor reduction in expected benefits. This might be partially offset by consequent increases in the individual's income tax unless some countervailing shifting of other income can be devised.

The impact of the Social Security System on the balance between the demand for and supply of assets and on employment is thus fairly complex. However, it does not depend so much on the intricate realities of the system as on the way it is perceived, both by its participants and by Congress. Many in Congress seem bemused by wildly irrelevant rhetoric concerning the supposed "diversion" of surplus social security revenues to government expenditure, and contentions over whether the system should be considered "off budget" or on. Most payroll taxpayers are only dimly aware of the relation of their "contributions" to eventual benefits. Most younger wage earners probably pay little attention to the prospect of benefits several decades in the future, and tend to treat their contribution as entirely a tax, though perhaps persisting under the delusion that the "employer's" share of the tax is actually borne by the employer.

Older low-wage workers are perhaps more likely to take future benefits into consideration in determining their attitude towards payroll taxes, expectations of benefits, and decisions on the level of expenditure. High-wage earners, on the other hand, may be more likely to regard payroll contributions as a tax, encouraged, in many cases, by propaganda showing how their contributions, if invested instead on an individual basis in private pensions or annuities, could yield substantially greater benefits, so that Social Security appears to be a bad bargain for them.

Another way of looking at it is to inquire what the equivalent is, in terms of individual wealth, of the interest of clients in the system. On the one hand the level of future benefits is not guaranteed, but is subject to modification by Congress, such as by subjecting benefits to individual income tax, increasing the normal age of retirement in terms of which benefits are calculated, increasing the cap on taxable wages, or even changing the benefit formulas themselves. While there is no guaranteed minimum below which benefits cannot be reduced, the political reality seems to be that taxpayers can rely on a fairly substantial wealth-equivalence. There is even a fairly well-established practice of indexing benefits by the consumer price index, so that social security wealth is likely to be less impaired by inflation than investment in long-term government securities.

Also, social security wealth is much less heavily concentrated among middle and upper classes than wealth in general, and thus tends to have a greater favorable influence on the level of consumption expenditure.