A quarterly magazine of the IMF

September 2006, Volume 43, Number 3

Pension Challenges in an Aging World
Adair Turner

Except where fertility rates are very low, needed pension system adjustments look manageable

Pensions are high on the policy agenda in many developed countries and, increasingly, in developing countries also. This reflects the challenges that demographic changes are creating for pension systems, whether pay-as-you-go (PAYG) or funded. The broad directions of that demographic change are common, but the precise nature of the pension policy challenge in each country reflects two uncommon factors—the severity of demographic change and the pension systems with which countries start. In some countries, relatively modest adjustments to existing pension systems can ensure their sustainability; in others, more radical change is required.

Two fundamental demographic trends are at work in the world. In countries that are achieving reasonable economic success and are free from the extreme effects of AIDS, life expectancy is increasing, possibly without limit. And in all economically successful economies, irrespective of apparent cultural differences, fertility rates have fallen or are falling to replacement level and, in many cases, well below. The United Nations medium projections suggest that, within 15 years, the total fertility rate will be about 2.0 or fewer children per woman in countries as diverse as Brazil, Iran, and Turkey. (A birth rate of a little more than 2.0 is required to ensure that each generation is at least as large as that of its parents, with the required increment above 2.0 driven by the proportion of children who die before reaching child-bearing age.)

A similar trend is seen in the economically successful states of southern India and in China, whose fertility rate, at 1.74, is well below replacement level. That trend, of course, partly reflects the one-child policy, but low fertility rates in Hong Kong SAR, Singapore, and Taiwan suggest that the Chinese policy has simply brought forward an otherwise naturally occurring phenomenon. Indeed, the East Asian pattern suggests that, as China becomes more prosperous, its fertility rate could fall further.

Longer lives and lower fertility rates are dramatically increasing the proportion of the population above the ages we typically associate with retirement (see table). In the United Kingdom, the ratio of people older than 65 to those between the ages of 20 and 65 will almost double between now and 2050. In Korea, that ratio could increase more than four times.

The implication for countries with PAYG state pension systems (where contributions from current workers fund payments to current beneficiaries) is that some combination of three adjustments will be unavoidable: higher levels of tax or of compulsory contributions to pay for state pensions, lower pensions relative to society-wide average earnings, or higher pensionable ages. But it is important to note that moving to a funded private pension system (where workers save a part of their wages and draw on the accumulated funds after retirement) does not provide a complete and certain escape from this demographic challenge.

All pension systems, PAYG or funded, involve the transfer of resources to pensioners (who consume but do not produce) from workers (who produce more than they consume). As a result, in a funded system a change in the ratio of pensioners to workers must affect rates of return and asset prices. If people attempt to finance a longer retirement by saving at a higher rate, that will tend to generate a higher capital output ratio and therefore come at the expense of lower returns on capital. If a high-saving generation is followed by a generation smaller in number, asset prices must tend to fall (relative to a higher fertility scenario) as the first generation attempts to sell its accumulated assets. The theoretical magnitude of these demographic effects on returns and asset prices is extensively debated: the empirical analysis of correlations is hampered by the limited availability of data and multiple other factors. And, ultimately, in a global capital market, it is global demography that matters to the economics of funded systems, whereas for PAYG systems national demography dominates. But it remains unwise to see funded approaches as providing a full and certain answer to demographic challenges.

Faced with these demographic challenges, many countries have announced significant changes to pension policy or appointed policy commissions to propose solutions. The U.K. Pensions Commission, which I chaired from 2003 to 2006 and whose recommendations the government largely accepted in its May 2006 White Paper, is one example among many. Some countries have focused on reforming their PAYG systems; others have introduced a greater element of funding in the overall system. The appropriate way forward needs to reflect each country's starting point. But one common policy, a gradual rise in retirement ages, makes sense in all countries.

Fertility and immigration

An "aging world" creates pension system challenges in both PAYG and funded systems. But it is important to understand that aging itself—that is, people living longer—is not the key problem.

Indeed, if the only demographic change at work were increasing longevity, there would be no fundamental problem of pension policy design. Instead, there would be one straightforward, simple, though sometimes politically difficult, solution. If longevity increases, but fertility stays stable, a proportional rise in pensionable ages (that is, a rise that keeps stable the proportions of adult life spent working and in retirement) will balance a PAYG system, with no need for either higher contribution rates or lower pensions relative to average earnings.

The U.K. government, following the Pensions Commission's recommendations, has now committed itself to the principle of a proportionally rising state pension age. This principle should be a central feature of PAYG pension reform in all countries. In a funded system, it is also easy to illustrate that an analogous condition applies. If increasing longevity were the only factor at work, and if intertemporal discount rates and preferences as to the proportion of life spent in retirement were stable between generations, there would be no reason to anticipate that demographic factors would affect rates of return or asset prices.

The far more fundamental problem is falling fertility. All 20th century PAYG pension systems were in a sense "ponzi schemes," or pyramid selling schemes: for the mathematics of the relationship between contributions paid in and pensions paid out to work, there must be more people in the next link of the chain, the next generation. And the population pyramids are coming to an end. That is good news for world environmental sustainability but creates a major challenge for pension systems. And the severity of the demographic challenge to pension systems is a direct function of the fertility rate or, more precisely, of the combined fertility plus permanent immigration rate (immigrants being, to the pension system designer, functionally equivalent to new citizens who are born, on average, in their mid-20s).

If the combined fertility plus immigration rate is equivalent to a fertility rate close to or greater than 2.0, the affordability problems of PAYG pension systems are easily manageable. This is clearly the case in the United Kingdom (see box), which is likely to have a slowly growing population, and would be the case even if the outlook were for population stability or a very gradual decline. Pension system sustainability does not therefore require that the world reject the environmental benefits of population stability. But for countries with very low fertility rates, the challenges will be severe. Unless fertility rates rise or large-scale immigration is allowed, Italy (fertility rate 1.4), Korea (1.2), and Japan (1.4) will have to accept some mix of major increases in the percentage of GDP devoted to state pension expenditure, increases in the pensionable age that rise more than proportionately with rising life expectancy, or significant falls in pension income relative to average earnings.

A question of generosity

Fertility rates and permanent immigration levels thus determine the severity of the pension challenge. But the starting point of the pension system, state and private combined, in different countries determines the nature of the challenge. Across the rich developed world, there are major differences in the generosity of the state PAYG promise and in whether the system seeks solely to keep people out of poverty in retirement (through a flat-rate earnings-independent pension) or aims to provide earnings-related pensions in return for earnings-related contributions (see chart).

Most countries in Western Europe start with similarly generous earnings-related PAYG pension systems and therefore face a common challenge—how to reform those promises to make them affordable and sustainable in the face of demographic change. The severity of that challenge varies with the fertility rate (it is more manageable in Sweden than in Italy), but the nature of pension reform debates is the same. But in other countries, such as the United Kingdom, the problem is not one of fiscally unaffordable state pension promises. Rather, the U.K.'s problem is that the state pension system would, under current policies, do no more than keep people out of severe poverty; that people are failing to save enough in private pensions to achieve what they would consider adequate pensions; and that the means-tested nature of the state pension provision (with cash targeted at the poorest pensioners) itself impedes private saving. The U.K.'s future potential problem is not a fiscally strained state but poor pensioners. The Pensions Commission's proposed solution combines a slight increase in the generosity of the state pension system with strong encouragement to people to save in addition. The solution relies on automatic enrollment to overcome the barriers of inertia and myopia that make an entirely free market approach to pension savings untenable. In developing countries like China and India, starting with even more limited state systems, that balance of policies may well be required.

But the proposed U.K. state pension system, though more generous than currently planned, will still play a far smaller role in total pension provision than, for instance, Sweden's PAYG system. The United Kingdom will have only a flat-rate antipoverty state pension, whereas Sweden will have a reformed but still fairly generous earnings-replacement system. Both solutions, however, have been chosen after extensive analysis and cross-party debate, driven by independent policy commissions. Should we then conclude that one commission must have got the answer right, the other wrong? The answer is no, because pension policy solutions need to reflect the political context and culture. The appropriate balance between funded and PAYG solutions is not precisely given by economic theory.

PAYG versus funded systems

Both types of systems, as argued earlier, are ultimately challenged by changing demography, and many of the supposed advantages of funded systems can be achieved within a PAYG environment. If an increase in the national saving rate is appropriate, adjustments to PAYG contribution rates can achieve that as easily as compulsory savings, and under both systems the increase is attainable only if someone sacrifices current consumption. And PAYG systems, as Sweden's reforms have demonstrated, can be made robust in the face of uncertainty over both future longevity and future fertility. The risk of unexpected future changes in longevity can be shifted from the government to individuals by moving to a notional defined-contribution approach, where the state credits each individual's compulsory contributions to an account whose value increases with a defined rate of return and delivers a capital sum at retirement age that it converts into an annuity. And the risk of unanticipated future changes in fertility can be managed through devices such as Sweden's "automatic balance mechanism," which can adjust contribution rates and pensions paid in response to a slowdown in overall economic growth. The only inherent advantage of funded approaches is that they allow for the expression of diverse individual utility preferences between different combinations of risk and return; conversely, PAYG systems have the inherent advantage of very low administrative cost.

As a result, the choice of a balance between PAYG and funded elements of the system, and thus the appropriate generosity of the PAYG system, needs to reflect a country's political culture and, in particular, the acceptability of different levels of taxation and compulsory contribution. The solution proposed by the U.K. Pensions Commission reflected a judgment that British political culture would not accept the taxation or contribution levels required to support a continental-style earnings-related PAYG system. It therefore makes sense to focus the state's limited fiscal firepower on doing flat-rate provision well rather than (as for the past 30 years) combine an inadequate and means-tested flat-rate pension with an inadequate and overly complex state earnings-related scheme.

Sweden's solution has been to reform the PAYG earnings-related system to make it sustainably affordable and fair between generations (the system also includes a small compulsory funded element). This implies devoting a considerably higher level of tax and contributions as a percent of GDP to pensions than in the United Kingdom, but that level is acceptable in the Swedish political culture. Both solutions are sensible within their specific context. In other countries, the appropriate balance should also reflect specific starting points and political cultures. But in all countries, the principle of proportional rises in pensionable ages will be appropriate to cope with increasing longevity. And in all, to different degrees, systems will need to make the more difficult adjustments necessitated by the shift to replacement or below-replacement fertility rates.