Defusing the Pension Timebomb: What are the Policy Options?

Public pension programmes in many OECD countries are in difficulties. With an ageing population existing pension arrangements may be untenable. How should we prepare for the looming crisis – the so-called pensions timebomb? At a CEPR/Royal Economic Society discussion meeting sponsored by Morgan Stanley Dean Witter and held in London on 3 February 2000, a panel of researchers (Tito Boeri, Axel Börsch-Supan, Richard Disney, Kevin Gardiner and David Miles) explored the policy options.

Tito Boeri began the meeting by stating that European governments could not afford to underestimate the challenge of the demographic transition currently taking place. Public pension systems can no longer bear the full burden of providing pensions and, where necessary, efforts must be made to achieve a more balanced pension programme with a sufficiently large funded component.

Boeri highlighted three major factors that are essential in achieving this goal. First, social partnerships should be promoted by involving workers, government and industry in the provision of income for retirement. Second, individual responsibility for retirement saving should be encouraged. And third, the role of public pension systems should be confined mainly to redistributing resources from the lifetime rich to the lifetime poor, and the responsibility for full old-age insurance provision for workers should be shifted to the other social partners.

In addition, pension programmes should no longer encourage early retirement nor should they hamper labour mobility. Boeri stressed that global investment is the key to achieving flexible pension programmes vis-à-vis demographic and political crises. But a global diversification of resources can only be achieved if there are no barriers to capital movements, in terms of arbitrary constraints on the international diversification of the portfolio of pension funds.

Finally, Boeri outlined three areas that required immediate action from the European Commission. First, the development of a European household panel survey, possibly linked to administrative data. This would provide timely information on trends in contributors to the pension system and on the determinants of retirement decisions. Second, the harmonization of methodologies used in the various countries to report pension outlays and forecast future pension liabilities. And third, a definition of common standards as to the frequency of expenditure forecasts and the length of the forecast horizons. Boeri concluded that all generations can benefit from a more balanced (i.e. more funded) pension programme if the costs and the benefits of pension reforms are shared. In particular, compensation mechanisms can and should be designed to avoid the middle generations bearing any extra burden caused by changes in the system. This, he felt, would contribute to winning public support for reform.

Richard Disney began his presentation by comparing the current costs of public pension programmes with their projected costs for the year 2030. The following figures state public pension payments in 1995 as a percentage of GDP, the figures in parentheses are the OECD's projected costs for 2030: Japan 4.1% (13.4%), Germany 11.1% (16.5%), France 10.6% (13.5%), Italy 13.3% (20.3%), UK 4.5% (5.5%). So what has caused these adverse trends? According to Disney, the demographic transition to an aged society is only one factor among many. Forecasts of the consequences of demographics and of labour supply have often been much too optimistic. Improvements in longevity have been far faster than official actuaries projected. And the reduction in labour supply of older men has been much more rapid than was predicted in many countries.

Disney explored three possible options for reforming the pensions programme: parametric reforms, actuarially fair unfunded reforms and privatisation. For parametric reforms, the standard approach to financing an unfunded pension scheme uses the following identity: c=(B/L)(p/w), where c= contribution rate, B= number of beneficiaries, L= number of workers, p= average pension and w= average wage rate. Therefore, in order to reduce c you must reduce B and/or p, and/or increase L and/or w. In practical terms this can be achieved by raising the retirement age, cutting benefits, or increasing labour force participation. This reduces the pain of change but is susceptible to political interference and reversal, and relies on behavioural responses that may not happen. For example, if the state pension age is raised from 65 to 70, will people really work for an extra five years? This, said Disney, is the short-run solution, and will be adopted in many countries.

A second broad strategy for reform of an unfunded scheme is to link entitlements to amounts paid in contributions. This is the reform route adopted in Italy, Sweden and Poland. Although it may avoid the excessive generosity of previous schemes, it cannot guarantee fiscal sustainability as the government can still interfere with the rules governing contributions and benefits. Furthermore, the return on an unfunded scheme is intrinsically related to the growth of the labour force and its productivity, whereas a funded scheme can generate much higher returns on the capital market.

This leaves the third route of introducing a funded component to pensions, usually by privatising some of the scheme. The attractions are that assets are created to match liabilities, the scope for government interference is limited and investors can benefit from high returns on the world capital market. The drawback is simple: the transition costs fall heavily on certain generations as they will have to honour most of the existing liabilities and pay for their own future pensions. But, one way or another, somebody will have to pay to eliminate these growing liabilities.

The German public pension programme was not only the first but also the most successful pension scheme of the last 100 years. But times have changed. In his presentation, Axel Börsch-Supan stated that although the system may be able to limp through the coming decades in its present form, it will cease to provide generous retirement incomes at reasonable tax rates. He argued that the current policy proposals are insufficient but that a few decisive changes to its design and some degree of prefunding could rescue the present system's many positive aspects.

In particular, Börsch-Supan outlined several policy recommendations. First, the present system should minimize the tax portion of contributions by making the system as actuarially fair as possible. Cross-subsidies towards early retirement should be abolished; by increasing the tax burden, they increase the negative labour supply incentives and therefore only shift unemployment from one sector (i.e. elderly) to another (i.e. low income and self-employed).

Second, owing to the limited magnitude of labour supply and the even more limited labour market adsorption, the German pension system has no choice but to reduce benefits and to prefund the resulting pension gap. It is important to make this policy explicit as even a small degree of prefunding requires time. Workers need clear and steady policy signals to plan ahead and save, not modifications and reforms that create additional uncertainty.

Third, a decisive step towards prefunding could exploit the large differences in rates of return between 'pay-as-you-go' (PAYG) and a fully funded system. Since the German system is much less redistributive than other systems, a relatively large share of the PAYG system is actual insurance and can thus be privatised. Of course, there are reasons to be conservative in the degree of prefunding. PAYG systems have a built-in insurance against inflation and secular capital market failures. Since Germany has experienced the disastrous effects of hyperinflation and stock market crashes in a rather dramatic way, Germans are probably willing to pay a high price for this insurance.

The transition costs to a degree of prefunding that is palatable to the German public, say 50%, are relatively modest, even if the burden lands on a single generation. Germany is in a situation that makes such a transition particularly attractive. The extent of population ageing makes the difference in returns between PAYG and funded schemes very large, thus reducing relative transition costs.

The first three speakers had advocated a move to a funded system, partly for the reason that funded schemes are able to take advantage of the high rates of returns in equity markets. However, the main theme of the penultimate speaker, David Miles, was that although funded schemes give higher returns on average, they are accompanied by significantly higher risk. Using estimates of real stock returns in Europe, Miles presented his results of measuring the risk for investors with a 30-year investment horizon. He had simulated the returns for 1 million non-overlapping 30-year equity portfolios. Although the mean return was relatively high at 6.25% per year, the probability of negative returns could be as high 3%, while as many as 0.6% of the simulations generated returns of -2%. Repeating this procedure so as to measure the distribution of returns on a hypothetical fund that earns a return of real GDP growth (effectively the returns from a PAYG system) results in a mean return of 2.5% but a far lower dispersion. Hence, the results suggest that a funded system will generate higher pensions on average than a PAYG system, but at the cost of taking on extra risk.

A government could choose to ensure a minimum annual return by issuing a put option written on the average return of the underlying assets in the fund. But how costly would this be? Miles estimated that the cost of a put option that would guarantee a long-run average real return of at least 2% a year might be as high as 20% of the value of the fund. Of course this does not mean that the alternative unfunded schemes are riskless. There is uncertainty about the desire and ability of future governments to deliver on the implicit promises of earlier governments. The risks with unfunded pensions may be no lower than with funded systems, but they are certainly different. Hence, Miles concluded, one prudent solution may be to mix funded and unfunded systems, as currently occurs in the Netherlands. 

In addition, Miles highlighted the problems associated with the burden of cost for making the transition from unfunded to funded schemes. Some have argued that a switch to a funded system might generate a sufficient surplus that could be used to buy out the costs of transition. Miles argued that this view is false, presenting simulations to show both the burden of transition and how it might be distributed across generations. He suggested that, unless governments are allowed to incur current budget deficits and accumulate sufficient debt to tax future generations, current voters would be unlikely to vote for funded pension schemes.

The premise for the entire meeting so far had been that changes in demography had resulted in the unsustainability of the unfunded system. Yet this idea was challenged by Kevin Gardiner in the final presentation. Gardiner noted that the ultimate determinant of average living standards and of real pensions is the rate at which the economy can grow. This in turn depends on the amount and utilization of labour and capital resources and the pace of technical progress. Gardiner argued that there is no necessary shortage of labour facing the UK or continental Europe. Plausible changes in participation and unemployment rates can deliver a rising supply of labour, even in Euroland, where the population is projected to decline. This is before taking account of possible increases in retirement ages and working hours, let alone extra capital input or technical progress.

Specifically, US-style levels of labour utilization would permit existing levels of European pensions to be financed at lower average tax rates than at present. In the UK, economic dependency (carefully defined) has been higher on at least three occasions in the not-so-distant past (1981, 1986 and 1991) than would be the case in 2030 on unchanged participation and unemployment rates. Such episodes were brief, but their existence suggests that the territory ahead, even on the pessimistic assumption of 'no change' in labour utilization, is far from uncharted. But, Gardiner continued, more important than the supply of both labour and capital is the prospect of continuing growth in total factor productivity (TFP), or technical progress. Historically, TFP growth seems to have accounted for most output growth in the UK. If past trends continue and are supplemented by extra labour and capital input, per capita GDP growth could accelerate over the next three decades. The likelihood of an aggregate supply constraint biting on GDP and average living standards in the UK and Euroland is slim.

Gardiner concluded that policy-makers should place less emphasis on measures designed to raise savings: a higher aggregate savings ratio is not necessary to fund future pensions, and could even prove counterproductive if unaccompanied by measures encouraging higher investment. Instead, policy should focus on improving labour market flexibility and fostering economic growth. Rather than focusing on the possibility of a future shortage of labour, European politicians should focus on making bigger inroads into today's excess supply.

Tito Boeri's (Università Bocconi and CEPR) presentation drew on a report on pensions commissioned by the European Round Table of Industrialists.

Richard Disney (University of Nottingham and The Institute for Fiscal Studies): 'Crises in Public Pension Programmes in OECD: What are the Reform Options?', published in the February 2000 issue of the Economic Journal.

Axel Börsch-Supan (University of Mannheim): 'A Model Under Siege: A Case Study of the German Retirement Insurance System', published in the February 2000 issue of the Economic Journal.

David Miles (Imperial College, London, and CEPR): 'Risk-Sharing and Transition Costs in the Reform of Pensions Systems in Europe', written with Allan Timmermann (London School of Economics and CEPR) and published in the October 1999 issue of Economic Policy.

Kevin Gardiner (Morgan Stanley Dean Witter): 'Defusing the Demographic Timebomb', a report written for Morgan Stanley Dean Witter and published in October 1999.