Ageing and Policies: Pension Systems Under Pressure

Abstract

A considerable amount of research has already been devoted to examining the impact of ageing at the national level, for instance, on the pension system, economic productivity, and labour market supply. However, it is more difficult to make comparisons between countries, due to differences in models, assumptions, and definitions. In this chapter, I use a demography-based model, along the lines suggested by Calot (Le vieillissement démographique dans l’Union Européenne à l’horizon 2050. Étude d’impact du viellissement démographique. Directorate-General Employment and Social Affairs, European Commission, 1995), in order to assess the implications of ageing on the 27 EU member states and on four non-European developed countries (Canada, Japan, South Korea, and the USA) in the period 2008–2050. I measure the relative impact of a series of alternative policies that could be implemented to counter ageing: increasing the number of (autochthonous or immigrant) people in employment, delaying age at exit from the labour market, increasing the proportion of GDP spent on pensions, or modifying the so-called transfer ratio. Results show that immigration alone cannot fully counter ageing in developed countries. However, it can play a useful complementary role if combined with an increase in labour participation and delayed retirement. The best policy mix will likely differ in each country, depending on the initial situation, demographic trends, and policy preferences.

Appendices

Annex 1: How the Model Works

1.1 Formal Description of the Model

At the macroeconomic level, the funding of pensions is a question of gross domestic product (GDP) redistribution from those who are participating in economic life to those who are already retired. A pension system can be described as in equilibrium when the volume of contributions levied on workers equals the amount of pensions paid to retired people.

$$ {\hbox{Contributions}} = {\hbox{Pension benefits}} $$

(2.1)

or

$$ GDP \cdot c = R \cdot p $$

(2.2)

where:

• GDP = the wealth produced (gross domestic product).

• c = the ‘contribution rate’ or the share of gross domestic product necessary to finance pensions, comprising all forms of contribution to the system, i.e. employer and employee contributions, taxes, and other contributions. This broad definition of ‘c’ results from the underlying assumption that people in employment are the only producers and contributors within the system, and hence, there is no difference between systems financed more through taxation and systems financed more through social contributions.

• R = the number of retired people.

• p = the average pension.

(2.2) is equivalent to

$$ E\,\cdot \,GDP/E\,\cdot \,c = R\,\cdot \,t\,\cdot \,GDP/E $$

(2.3)

where:

• E = the number of people in employment (in full-time equivalents), i.e. the people producing the GDP from which the pensions are financed

• GDP/E = the average gross domestic product per employed person (GDP divided by the number of people working)

• t = the ‘transfer ratio’, defined here as the ratio of average pension to average gross domestic product per employed person $\left( {t = \frac{p}{{{{{GDP}} \left/ {E} \right.}}}} \right)$

(2.3) can be simplified further and then is equivalent to

$$ E\,\cdot \,c = R\,\cdot \,t $$

(2.4)

As the number of the retired (R), the number of the employed (E), and the share of contributions to the GDP (c) are known for any particular year, t can be estimated for 2008 (initial year of the projection model) by assuming equilibrium between expenditures and receipts:

$$ t = E/R\,\cdot \,c $$

(2.5)

In this model, pension sustainability is achieved when total resources equal total expenditures at the macroeconomic level during the considered period (in this case, 2008–2050). Within this framework, it is not relevant whether pension schemes are based on a pay-as-you-go (PAYG) or funding since the current GDP is shared every year between those who receive income directly from their participation in economic life and those who do not. If there are more retired people, the share of pensions in GDP is likely to be greater, whatever the funding base of the pensions: contributions, taxes, or financial yields.

1.2 Assumptions for the Variables and Parameters Used in the Model

The model works on the basis of the following assumptions:

• Pension schemes are assumed to be in financial equilibrium in the starting year (2008). Therefore, all findings should be related to this year as the reference point.

• The only external shock unbalancing the system is assumed to be the change in the number of retired people (defined as the number of persons aged over the average pension age in each country) due to population ageing.

• Data on population by age for the period 2008–2050 are taken from Eurostat demographic projections (convergence no-migration and convergence with-migration scenarios) for the EU member states and from UN World Population Prospects 2008 Revision (zero-migration and medium scenarios) for Canada, Japan, South Korea, and the United States. Two different sets of scenarios have been developed by Eurostat and UNPD: with migration and without migration from 2009 onwards (therefore, in zero-migration scenarios, population growth and structure is only determined by births and deaths). The latter has been preferably used here, as the number of (employed) immigrants (and their descendants) compensating ageing is one of the main outputs of the model. However, the more realistic scenario with migration has also been used to compare results and to assess the compensatory effect of migration levels forecasted by Eurostat and UNPD.

• E, the number of people employed and contributing to the system, is calculated by using the full-time equivalents (FTE). They have been calculated for each country by multiplying the number of people working part-time times the ratio of the average number of usual weekly hours of work of those working part-time to the average number of weekly working hours of those who are employed full-time. This ratio is around 0.5, but significant variations exist among countries. The resulting figures, plus the number of people currently working full-time, give the total number of full-time equivalents employed. Taking the FTE rate provides a better and more comparable insight, between countries and over time, of the ability of employed people to create wealth and to contribute to pension system funding, as part-time workers contribute less than full-time ones regardless of the type of pension system funding: directly through social contributions (since those are mostly more or less proportional to earnings) or indirectly through general taxation. Furthermore, using FTE allows for a more reliable comparison between countries, given the wide dispersion in the prevalence and average duration of part-time employment.

• c, the share of GDP levied to finance pensions, is estimated for the EU countries from the total expenditure on old age and survivors’ benefits, as given by Eurostat ESSPROS (European System of Social Protection Statistics) database. For the other developed countries, I used OECD data.

• R, the retired population, is assumed to be the part of the population above the average effective retirement age (ERA, as defined by Eurostat: average exit age from the labour market, weighted by the probability of withdrawal from the labour market). This assumption is acceptable as most elderly people actually have direct or derivative rights to pensions. The average effective retirement age has been estimated for each EU country by Eurostat from EU Labour Force Survey data on age-specific activity rates. For the four other developed countries, OECD statistics on average effective retirement age have been used (OECD 2009b).

• t, the transfer ratio, is equal to $E/R\,\cdot \,c$ when the system is in equilibrium (see Eq. 2.5) and provides a proxy measure to assess the evolution of the relative pension level. It is similar to the ‘net replacement rate’ used in other models, which is the ratio of the average pension to the average wage. However, t puts the average pension in relation to the average GDP per employed person, which is a direct measure of productivity—indeed, ‘t’ can decrease because the average pension diminishes or because productivity increases. Therefore, maintaining the relative level of average pension implies that the absolute level of the average pension should evolve at the same pace as productivity.

• Finally, as the model assumes that GDP is entirely produced by employed people and pensions are distributed among the retired population (as defined in R), it disregards people who are neither employed nor retired. It is hence assumed, for the sake of simplicity, that those people are neither contributing nor costing anything to GDP, although some of them may benefit from different forms of government transfers. A more complex model could incorporate the impact of other forms of government transfers and social policies—e.g. health-care contributions and expenditures, which are distributed across ages in a similar way as pensions or unemployment benefits.

Annex 2: Results for the EU and the 29 Countries Analysed in This Chapter

Figures 1–31. Retired people (R) and working-age population (wap) projected in the analysed country between 2008 and 2050 and number of employed persons (E ′) maintaining the model in equilibrium

Bulgaria	Belgium
Canada	Czech Republic
Denmark	Germany
Estonia	Greece
Spain	France
Ireland	Italy
Japan	Cyprus
Latvia	Lithuania
Luxembourg	Hungary
Malta	Netherlands
Austria	Poland
Portugal	Romania
Slovenia	Slovakia
South Korea	Finland
Sweden	United Kingdom
United States