FACING FACTS

The Truth about Entitlements and the Budget
A Fax Alert from The Concord Coalition
Volume V, Number 7 -- July 6, 1999

DEAD WRONG ABOUT MORTALITY?

Few developments have done more to shape the modern world than declining mortality-or, to put it another way, rising longevity. If it weren't for the reduction in mortality rates since the beginning of the century, one-quarter of all Americans alive today would already be dead; another quarter would never have been born (since their parents wouldn't have been around).

One of the most far-reaching consequences of this demographic revolution-the explosion in the number of elderly-still lies largely in the future. Declining mortality is not the only cause of the age wave.  But assumptions about future reductions in mortality rates have a bigger quantitative impact early in the next century than any other variable.   And most experts believe that the Social Security Administration's (SSA's) assumptions are too conservative.

In its "intermediate" scenario, SSA projects that mortality will decline at just half of its long-term historical rate. It thus assumes, at a time of stunning medical advances, that U.S. life expectancy a half-century from now will be no higher than Japanese life expectancy already is today. If SSA's pessimism proves unfounded, the Social Security cost problem will be bigger than officially projected-perhaps much bigger.

Over the past few years, SSA's assumptions have been questioned by demographers, criticized by the technical panel of the Advisory Council on Social Security, and ignored by the Census Bureau, whose projections assume faster reductions in mortality. It's time political leaders took note: The official projections they're relying on may be wrong-dead
wrong.

A Flawed Methodology

There are two big problems with the way SSA projects mortality, the first being its cause-of-death methodology. Let's leave aside the fact that cause-specific death data are notoriously unreliable and difficult to interpret, especially at older ages. Most demographers believe the methodology is flawed. In cause-specific projections, the total rate of mortality decline must eventually decelerate as the most slowly declining cause of death becomes the most widespread. According to this
methodology, the rising incidence of a cause of death will never persuade people to do anything more to reduce it. Consider a society in which the typical person dies of the plague by age 35. Very likely, its members will focus on surviving or curing the plague-while paying no attention to cardiovascular disease at age 65. SSA assumes that this society's priorities won't change, even centuries after the plague is eradicated and everybody is dying of heart attacks.

The second problem lies in SSA's ad hoc assumptions about "ultimate" rates of mortality decline. SSA assumes that reductions in cause-specific mortality will steadily slow until, twenty years from now, a set of "ultimate" rates of reduction is reached. SSA is vague about how these rates were calculated. What we do know is that they are much lower than historical rates, and that the total rate of reduction in mortality projected by SSA is thus much lower too. From 1900 to 1995, the U.S. age-sex adjusted mortality rate declined by 1.13 percent per year. Over the next seventy-five years, SSA projects that it will decline by just 0.56 percent per year.

Back in 1992, this startling divergence caught the attention of demographers Ronald Lee and Lawrence Carter. In their now-classic critique of the SSA projections, they cited historical data demonstrating that the total rate of morality decline in the United States has been remarkably consistent over time. They then showed that mortality projections based on historical trends point to bigger gains in longevity and higher Social Security costs than SSA's projections do.

Since then, the basic Lee-Carter approach has gained widespread acceptance. At an October 30, 1997, conference of the Society of Actuaries, a gathering of prominent demographers, economists, and actuaries was surveyed on just this issue. Of 59 respondents, only thirteen (twelve of them actuaries) thought that cause-specific forecasts were more accurate than forecasts based on aggregate data.  Two-thirds agreed that historical trends should be the "primary guide."

Alternative Scenarios

Recently, we made our own historical-trend projections, not just to confirm Lee-Carter's conclusions using more recent data, but to focus specifically on mortality among the elderly, where assuming an unchanged rate of progress seems most reasonable. We ran our projections using the Employee Benefit Research Institute's SSASIM2 model, which was originally developed for the Advisory Council on Social Security.

Specifically, we assumed that future mortality rates decline for everyone under age 65 as SSA projects, but that they continue to decline at their 1970 to 1995 average for everyone aged 65 and over. The results are dramatic. By 2070, life expectancy at birth rises to 85, three years further than in SSA's projection. Meanwhile, the cost of Social Security grows to 21.3 percent of payroll, compared with 19.6 percent. (Using rates of mortality decline over longer historical periods, all the way back to 1900, gives similar results.)

How does SSA respond to these kinds of historical-trend projections?  Upon occasion, SSA has pointed out that the outlook for Social Security won't be so bleak if mortality continues to decline at its historical rate for all age brackets.  The reason, as SSA explained to the technical panel of the Advisory Council, is that mortality has historically declined faster for the young than for the old. The extra cost of more retirees will thus be offset by the extra payroll contributions of more workers.

This rebuttal isn't very convincing. Mortality among the young is already very low, and is largely due to causes, such as accidents and homicides, that are as likely to go up as down. Mortality among the old, however, remains high, and is largely due to causes-from arteriosclerosis to hypertension to cancer-where enormous progress is now underway. In the Society of Actuaries survey, the great majority of respondents thought that the age-differential in mortality rates will narrow over time. Several thought that rates of morality decline for the old will eventually surpass those for the young.

If anything, this suggests that our projection may not go far enough. A number of demographers, notably Kenneth Manton, are now looking beyond the historical trend in mortality rates and are making projections that take into account changes in the underlying risk factors (from diet and lifestyle to affluence and educational attainment) that determine a population's general health. According to their models, a life expectancy of 100 is entirely possible-even without major breakthroughs in biogenetics that slow the aging process itself.

To allow for this brighter future, we ran a second projection in which rates of mortality for the elderly decline at twice their 1970 to 1995 average. What we found is that life expectancy rises to 89 by 2070-less than Manton predicts, but enough to push the cost of Social Security all the way to 24.5 percent of payroll.

And this is just the tip of a cost iceberg. Remember: Rising life expectancy will add not just to Social Security spending, but to
Medicare and Medicaid spending as well. In fact, the cost growth in these programs may be proportionally greater, since per capita health-benefit spending rises sharply with age and since declining elderly mortality translates directly into a higher average age for the elderly-a phenomenon demographers call the "aging of the aged." SSA projects that there will be 16 million Americans aged 85 and over by 2070, up from 4 million today. This is dramatic enough. But if elderly mortality continues to decline at its historical rate, the number will be 23 million; if it declines at twice that rate, the number will be 36 million.

Questioning Conventional Wisdom

To the extent there is any justification for SSA's pessimism, it lies in the seemingly reasonable conviction that longevity cannot keep rising indefinitely. At some point, we feel instinctively, medical progress must push us up against the natural limit to the human life span.

The conventional wisdom among biologists and health experts is that this is indeed the case. It is well known that many measures of human organ efficiency decline linearly after about age thirty, regardless of an individual's general health. It also appears that human cells are preprogrammed to reproduce themselves with accumulating imperfections, perhaps because there is little evolutionary advantage to living much past middle age. It thus seems reasonable to assume that everyone's reserve physiological strength must eventually fall to a level where even a minor trauma or illness will become life-threatening. If fatal,
we call this "dying of old age."

But in recent years, many experts have begun questioning conventional wisdom. One school of thought accepts the existence of a fixed genetic limit to the human life span, but stresses that there's a difference between knowing a theoretical limit exists and knowing the precise age at which it becomes of practical significance. The fact that the elderly keep living longer, in other words, may simply mean that the maximum human life span is considerably higher than we once assumed.

Another school of thought rejects the so-called fixity thesis altogether, since it implies a number of consequences that are not borne out by observation. If there's a limit to the human life span, mortality improvements for the oldest elderly age brackets should be slowing relative to those for younger elderly age brackets. At the same time, variations in life expectancy should be narrowing as more people bunch up against the limit.

None of this appears to be happening. Over the last couple of decades, the data-for different states, for the U.S. population as a whole, and for other developed countries-show no tendency for mortality improvements to slow at advanced ages, which is what gives the aging of the aged such an edge. Indeed, centenarians are the fastest growing of all age groups.

Nor are variations in life expectancy diminishing, whether we look at the data by country, by region, by income, or by education. For many years, Sweden and Japan have had a sizeable longevity advantage over other developed countries-and some states like Minnesota and Utah have had a huge advantage (less than half the mortality rate) over states like Mississippi and Louisiana. Such differences appear to be persistent. Everywhere, people are living much longer. Yet everywhere, some groups are living much longer than others.

Demographer James Vaupel sums up the evidence this way: "The Social Security Administration would prefer it if the low-cost future were true. But...there does not appear to be any genetic barrier to a substantially increased life expectancy."

Cognitive Dissonance

Is it certain that life expectancy will rise more than SSA projects? Of course not. It's easy to imagine a catastrophe that slows or even reverses the historical rate of mortality decline-though the most obvious candidate, a major war, would fall more heavily on the young than the old. But we should also keep in mind another possibility that breakthroughs in biogenetics, as futurist Marvin Cetron predicts in his bestseller Cheating Death, may unlock the secret of aging at the cellular level. We don't base our projections on such a scenario, but it is one America would be foolish to dismiss.

A decade ago, speculation about "post-mortal" societies was the preserve of science fiction writers. Today, the President of the United States dazzles audiences with talk about how technology may soon "raise the life expectancy of people...through 100 years or more." Apparently, no one cares that the administration's own SSA projections, on which it bases its Social Security reform plan, are massively inconsistent with Clinton's forecast. They show that U.S. life expectancy will never approach 100, even after centuries.

This kind of cognitive dissonance characterizes much of the debate over senior entitlements. We must get beyond outmoded notions about the three-box life cycle of education, work, and retirement. We must revise the expectation that future increases in longevity, no matter how large, will be automatically subsidized by taxpayers. At the very least, we should consider indexing Social Security's full benefit retirement age to changes in longevity. This idea has been incorporated into a number of reform plans now under consideration-and it was recently enacted into law in Sweden.

America is getting older. It's time to take an unflinching look at where we're heading, and face up to the cost of the new reality.

ALTERNATIVE LONGEVITY SCENARIOS

Notes:  Preliminary results from EBRI SSASIM2 model. Historical mortality decline scenario assumes SSA-projected rates of decline for
all ages under 65 and 1970-1995 average rates of decline for ages 65 and over; accelerated mortality decline scenario is identical, except that
elderly mortality rates decline twice as fast.

Facing Facts authors: Neil Howe and Richard Jackson
Concord Coalition Policy Director: Robert Bixby