Peter Peterson
Peter Peterson; drawing by David Levine

The population of the US is getting older on the average. Why? Because families are having fewer children, people are living longer (and retiring earlier), and because the roughly twenty-year-long bulge of births that we call the baby boom has reached the fifth of Shakespeare’s seven stages and will soon arrive at the lean and slipper’d pantaloon. Much the same is true of other rich countries; the aging of the population is happening sooner and more suddenly in Japan and more gradually in Europe.

This demographic certainty will have all sorts of consequences for these societies, affecting the need for schools, the demand for housing, the amount of crime, the volume of health care, and, most broadly of all, the general standard of living their economies can support. Peter Peterson writes of “the wrenching economic and social costs that will accompany this demographic transformation—costs that threaten to bankrupt even the greatest of powers, the United States included, unless they take action in time.”

It is better to start a little less feverishly, so I begin with an abstract exercise.


Think about a completely artificial population. A million people are born every decade. For the first twenty years they are at home, in school, or otherwise dependent. In their third, fourth, fifth, sixth, and seventh decades of life they are in the labor force, working for a living. Everyone retires at age seventy, lives for ten more years, and dies promptly at eighty. The population thus amounts to eight million souls, and it is neither growing nor declining; a million people are born in every decade and a million people die.

There are five million workers (aged twenty to seventy) and three million dependents. The five million workers have to produce enough for all eight million people to live on. Of course they do not produce with their bare hands. The more machinery, computers, and other capital equipment they have to work with, and the more technologically advanced that capital equipment is, the more productive they will be, and the higher the average standard of living they will be able to provide for the whole society.

Now suppose that medical advances allow everyone to live to age ninety. After a while the population will grow to nine million, of whom the same five million work. But now they will have four million dependents to support, instead of three million. Unless something happens, the average income of the population will be roughly 11 percent lower. Try a different experiment: Suppose the age of retirement falls to sixty. This is a more drastic development, because it both adds to the dependent population and subtracts from the labor force. There would be four million dependents and four million workers; the average standard of living would fall by a whole 20 percent. Put both cases together and 4 million workers would have to support themselves and five million young and old dependents. (Keep in mind that a reduction in the work week from forty to thirty hours would not affect numbers of people at all, but would be equivalent to an increase in the “dependency ratio” all the same.)

Take something a little more complicated, a “baby boom.” Births jump to two million per decade for two decades and then revert to one million. The initial effect is an increase in the dependency ratio, the extra dependents consisting entirely of children. The boomers then graduate to working age, and for several decades the dependency ratio will be lower than it was before the baby boom. As the boomers retire, however, the dependency ratio rises again, but now the extra dependents are old people who need health care, nursing homes, and slippers, instead of schools and expensive athletic shoes. Eventually the baby-boom generation disappears; this restores the situation that ruled before they appeared. Some projections for the US suggest that the disappearance of the baby-boom generation will not in actual fact be accompanied by a fall in the dependency ratio, because increasing longevity and earlier retirement may offset the potential decline in the number of retired people.

A little pencil-and-paper work can easily trace out other scenarios. For instance, one might allow for the fact that each cohort of a million births consists of half a million women and an equal number of men. A rise in the proportion of adult women who work will, of course, reduce the dependency ratio. (But in fact there is not much more scope for this effect to take place in the US, because women already work in such large numbers.) With only a bit more patience, it is easy to work out the consequences of a continued rise in the number of births and eventual deaths. Except that they are very much more sophisticated, that is what demographers do.



I have deliberately begun with this spare arithmetic exercise to sharpen the contrast with the strident Chicken Little tone of Mr. Peterson’s book. Maybe it is all a matter of personal style. He was—with ex-Senators Warren Rudman of New Hampshire and the late Paul Tsongas of Massachusetts—one of the co-founders of the Concord Coalition, a lobbying group devoted to the merits of balanced federal budgets. During much of that time, I too thought the case for temperate but determined reduction of the persistent federal deficit (properly measured) was very strong. But I could never bring myself to join or support the Concord Coalition, for two reasons. First, its literature sounded as if the deficit was a catastrophe waiting to happen, instead of a nagging, slow drain on the US economy’s potential productivity. Second, the group concentrated on the budget deficit as if that were the only broad problem of economic policy worth thinking about, instead of one among several, and not at every instant the one with highest priority.

This book seems to me to have the very same temperamental faults. It is full of alarmist images—icebergs (think Titanic), tidal waves, explosions—that both stretch the truth and undermine any impulse to careful, sober analysis. And the book goes on as if the threat from an aging population were so overwhelming that there are no other considerations, no possible trade-offs, of comparable significance. It is sometimes said that the only way to rouse the political process of the US to timely action is to shout “crisis.” Maybe so: but it cannot be good to let Chicken Little drown out cool thought.

To show that this is not merely a matter of taste, I will cite a few instances where Peterson’s judgment is led astray by tunnel vision and excess of zeal. Here is a minor case: he resists the obvious notion that a slowing of population growth, although it increases the proportion of old dependents, also decreases the proportion of young dependents. He offers the judgment that children and old people are not the same. Of course no one has proposed that they are, only that both are dependent. The only possible reason for this lapse is the fear that mere mention of an offset to aging may weaken the alarmist message he wants to propagate.

But then Peterson gets trapped by his own rhetoric. One of his “new strategies” is to urge the raising of more children. This is foolish for several reasons. First, as the arithmetic shows, the initial effect would be to increase the dependency ratio. Second, if the birthrate eventually reverts to the maintenance level, the end result will have been a second baby boom; the new boomers will, for a while, lower the dependency rate when they are all working; they will then worsen it again when they retire, and eventually disappear. Third, if Peterson would really like birthrates to stay high enough to keep the population growing instead of stabilizing, he has found a “solution” that is probably worse than the problem it is supposed to solve. Faster population growth brings with it increased pollution, loss of open land, and pressure on food and water supplies. It seems an odd way to solve a different problem.

Another of Peterson’s “new strategies” is, as he says, to put more emphasis on filial obligation; that must mean to create incentives or requirements that would induce adult children to assume more of the costs of supporting their aged parents. Obviously this would do exactly nothing to reduce the aggregate burden of dependency; the relative size of the working and dependent populations would not change. It would, however, redistribute the burden: the parents of entrepreneurial children would have a more graceful old age than the parents of underpaid schoolteachers.

I would have thought that one of the successes of indexed Social Security is that it has allowed older Americans to exercise their evident preference to live independently of their adult children. One imagines that this preference is shared by the children. Only a kind of tunnel vision could have led Peterson to think that this is any of his business.

Many of Peterson’s recommendations for policy are entirely reasonable. They include providing better early education (to raise productivity), making work more attractive compared with dependency, tailoring retirement benefits more closely to need, and promoting voluntary saving. I want to comment on only one of these, probably the most sensible and popular.

It is to extend working lives by gradually raising the minimum age of eligibility for Social Security benefits. A provision to do so is already written into law, and many advocate going further in this direction. As people live longer, and stay healthy longer, it seems foolish that the law should encourage retirement at sixty-five, or even earlier. On the whole, I agree. But I think it is worth keeping in mind that nearly all argumentation to this effect comes from the sort of people who enjoy their work, and might feel a little lost without it. Many, perhaps most, wage and salary earners find their work unpleasant and boring; the news that they will have to keep it up for three or four years longer will not elicit sighs of relief. Prolonging working lives may be better than the next-best thing, but it is not without cost.



The beginning of wisdom in thinking about our aging population and the reform of Social Security is to realize that they are distinct problems. All by itself, the demography implies that the burden of dependency is going to increase: a proportionately smaller working population will have to support a proportionately larger total population. That is true no matter what kind of Social Security system we have, or if we have no public Social Security at all. Social Security is about something else: how much income gets transferred to retired people—and from whom—in a society in which about 70 percent of all income is income from work.

I have exaggerated a little. There are connections between the two generally separable questions. The most important is that the details of the Social Security system affect decisions about when to retire, and how much to save for one’s own old age.

This is important enough to be worth one more piece of arithmetic. A society’s (average) standard of living is its total income divided by the size of its population. (That is a narrow definition, but it is what we are talking about.) The arithmetic states that

(total income/population) = (total income/labor force) å´ (labor force/population).

We have broken down the standard of living into the product of two numbers: the first is income per worker, a measure of productivity, and the second is the fraction of the population that works, primarily a result of demography (plus some sociological trends and some consequences of the economics of retirement).

The problem with an aging population is that it makes that second factor—called the participation rate—smaller. If nothing else happens, the standard of living must fall in exactly the same proportion as the participation rate. The rules governing Social Security do not affect this outcome, except as they can induce a larger fraction of the population to work (by extending the normal working life, for instance, or by attracting more men and women into the labor market).

As a matter of economic policy, there is another possibility. Anything that the society can do to jack up productivity offsets the reduction in living standards enforced by a lower participation rate. If productivity can be made to rise by the same percentage as the participation rate falls, there need be no change in the standard of living at all.

Making productivity rise is not an easy matter. If it were, why would we not have done it already? The natural first thought is to save and invest more, to outfit each worker with more and newer capital equipment than he or she now has. That would certainly increase future productivity. One would expect a leader of the Concord Coalition to run with that ball somewhat further than Mr. Peterson does. But proposing fiscal measures that would increase future productivity is exactly what Mr. Clinton has done.

Just as a dollar of federal deficit dissipates a dollar of private savings, a dollar of surplus is a public addition to private savings. It works through the capital market in the following way: when the Treasury has a billion dollars of surplus in any year, it must allow a billion dollars’ worth of federal debt to expire on maturity, or else it must buy a billion dollars’ worth of unexpired bonds from private owners. (The only alternative is to pile up cash, in which case the Federal Reserve will have to buy up bonds to maintain the money supply.) Either way, individual savers or institutions or pension funds will find themselves holding a billion dollars less of Treasury bonds and a billion dollars more of cash than they held before.

They will generally want to invest that cash in assets that earn income. After all, if they had wanted to hold cash or to consume their wealth they could have done that yesterday, before the federal surplus appeared. With the Treasury not in the market, the billion dollars will find a home in privately issued stocks and bonds; the cost of capital will fall until corporations are willing to finance the additional investment that would ultimately increase productivity. Nothing goes quite that smoothly, of course. Glitches are possible, including recessions, but in principle that is how federal saving finances private investment and leads in the end to higher productivity.

The centerpiece of the Clinton administration’s program is to convert trillions of dollars of budget surplus expected over the next decade or two into private capital formation. Private owners now hold Treasury bonds worth a bit over 40 percent of GDP. In the years ahead, if the surplus forecasts come true, that will be reduced to about 7 percent of (a higher) GDP. Private portfolios will contain many fewer Treasury bonds and many more bonds and equities issued by productive businesses. The Clinton program also involves a certain amount of creative bookkeeping whose purpose is to make it very difficult for Congress to dissipate much of the ongoing surpluses in across-the-board tax reductions. (The idea is to donate to the Social Security trust fund the Treasury bonds bought from private holders. This transfer from one of the government’s pockets to another has no real significance, except that it makes a visible commitment that the Treasury will provide funds for Social Security when they are needed.) Income-tax cuts would almost certainly find their way into current consumption, and not into investment designed to cut costs and increase productivity. (In the last quarter of 1998, American households saved one tenth of one percent of their after-tax incomes.) In this context, the tax-cutter is the spendthrift, the grasshopper, not the ant.

As enormous as the Clinton proposal is, it is not enough to solve the problem. Suppose that in twelve or fifteen years the program succeeds in adding an extra $2.5 trillion to the stock of productive capital. The rules of thumb that economists use to make such calculations say that the productive capacity of the US would be about 3 percent higher than it would have been if the $2.5 trillion had been consumed rather than invested. I think this might be an underestimate, because it does not factor in fully the technology-enhancing effect of extra, newer, investment. Even so, 5 percent would be an outside estimate. So much for the productivity side of the equation.

By how much will the proportion of people who work—the participation rate—fall? Most contributors to the Social Security debate look at the ratio of the population of working age (defined as between ages sixteen and sixty-four) to the total population sixteen or older. This strikes me as off the mark in two respects: one should look at the whole population, including dependent children, although it might be reasonable to count a child’s portion of GDP as only a fraction of an adult’s; and “working age” in the US should probably begin nearer to eighteen than sixteen.

Even after such adjustments are made, however, the power of the demographic shift is so great that the participation rate is likely to fall by more than the 3 to 5 percent that productivity might rise. Maintaining the overall standard of living will take something slightly more heroic. Some of Mr. Peterson’s suggestions, like his advocacy of a greater flow of immigrants, strike me as unrealistic rather than heroic. Keep in mind that a lower (future) standard of living does not mean lower than now, but lower than would otherwise be the case then, a less painful prospect.


The more productive our economy is when the baby boomers start to retire, the easier it will be to support them in their retirement. Most of the debate about the future of Social Security will be about how to do that: What benefits should be offered, and who will pay for them? That is too long and complex a story to deal with here; but I should relate these debates to the prior and more fundamental issue of productivity.

Our Social Security system was designed to run on a pay-as-you-go basis, and still runs that way. Payroll taxes on today’s covered workers are paid into a trust fund, and then paid out again to today’s beneficiaries. The trust fund is just a buffer. When, as now, revenues exceed outlays, the trust fund grows. Later, when there are more beneficiaries and fewer workers, the trust fund will shrink.

In the meanwhile, the trust fund hands its surpluses over to the Treasury and gets interest-bearing Treasury bonds in return. When benefits begin to exceed revenues, the Treasury will have to redeem those bonds and accrued interest and turn over the money to the trust fund so that benefits can be paid. Where will the Treasury get the funds to meet its obligations? This will depend on what the President and the Congress decide: the funds can be obtained either by taxation or by selling other bonds to the private sector. So the trust fund owns no “real” productive assets; what today’s beneficiaries have is a claim on today’s workers.

When the trust fund “runs out,” which would happen at about 2030 under current law, Social Security revenues would be enough to cover only about 75 percent of currently scheduled benefits on a pay-as-you-go basis. There will surely be changes before then; even if there were not, nothing would stop the Congress from allocating general funds to meet Social Security’s obligations. Indeed, part of the Clinton administration’s proposal for creative bookkeeping already described is just a way of committing enough general revenues to cover Social Security obligations for a couple of decades beyond 2030. A permanent fix of the current pay-as-you-go system, presuming that the demography will not turn favorable, would have to involve increases in taxes and/or reductions in benefits, though probably not very onerous ones for most people.

The most heated discussion currently seems to center around two issues. Should part of the trust fund be invested in corporate equities (or eventually in corporate bonds)? And if the answer is yes, should those investments be managed by the trust fund itself or by each covered worker individually?

Use of a fraction of the trust fund (a quarter in the Clinton proposal) to buy equities has no direct effect on aggregate national savings and investment, and therefore no effect on productivity. The simplest way to see that is to notice that it amounts to a swap of Treasury bonds for an equal amount of equities in the private sector’s portfolio. This is so because the Treasury has to pay its bills; if it cannot sell bonds to the trust fund, it must raise funds by selling them to the public. There is government savings to add to private savings only when the government runs a cash surplus. It is that anticipated surplus that protects the standard of living. Simply to shift assets butters no parsnips.

So the proposal to invest part of the trust fund in equities has to be judged on other grounds. History and theory both tell us that doing so will increase the return earned by the trust fund on the average, and that is good for future beneficiaries. But equities will also add risk or volatility to the portfolio, and that is bad for beneficiaries. People rich enough to have private portfolios can always restore a preferred balance between risk and return by selling stocks and buying bonds on their own account. A much larger number of workers have no real opportunity to make that adjustment; Social Security and a house is all most of them have. For them, the swap of bonds for equity in the trust fund is probably a good thing; at least one imagines that a private financial adviser would think it unwise for such a person to own no stocks at all. There are further complications, but on balance this proposal sounds right.

Well, then, should the trust fund make these equity investments on behalf of future beneficiaries, or should there be an individual account for each beneficiary, who could then invest his or her “Social Security wealth” according to knowledge and temperament? Mr. Peterson is strongly (what else is new?) in favor of individual accounts, along with a certain amount of necessary regulation. Again, this decision does not bear on the national ability to save and invest.

The main arguments for individual accounts appear to be that they are in tune with American individualism, will instill in citizens a proper respect for and attachment to the capital markets, and will avoid the specter of political intrusion into corporate governance via the government’s voting of proxies and threats not to buy the shares of companies that do not toe a political line. The first two of these do not lend themselves to discussion; the third is usually countered by pointing out that the Clinton proposal would have the trust fund eventually owning about the same fraction of the equity market as Fidelity does now—some 4 percent—and by the possibility of having the portfolio managed by a nonpartisan body, restricted to certain index funds, and open to public scrutiny.

The main arguments against individual accounts are that the sheer clerical costs of running more than 100 million small portfolios would inevitably be very high, that many of the owners have no investment experience and would be vulnerable to high-pressure salesmanship or worse, and that any unwise or unlucky investor would have a poverty-stricken old age. The last argument is usually countered by allowing that the new system could include a relatively small safety net, more or less like the current system, and by observing that taking the “social” out of Social Security is exactly what one side of the debate is after.

These are complicated questions, and there are still other important aspects of Social Security that I have not even mentioned. But for anyone who wants calm, balanced exposition, Gray Dawn would not be high on my list. Here are some excellent alternatives. Evaluating Issues inPrivatizing Social Security, the report of a panel appointed by the National Academy of Social Insurance,1 lays out the different approaches to reforming Social Security with great acuity and clarity. Older and Wiser: The Economics of Public Pensions, by Lawrence Thompson,2 goes over some of the same ground and includes a useful look at research results, international comparisons, and related facts about social insurance. Is It Time to Reform Social Security? is by Edward M. Gramlich,3 the chairman of the Quadrennial Advisory Council on Social Security that produced a divided report in 1997. Unlike the others, this is not just an exposition; it argues for the author’s own preferences, which are not always mine, but it does so lucidly and temperately.

This Issue

May 6, 1999