Pension Systems in the EU — Some Policy Issues

To what extent a pension system can be judged fair given structural differences in life expectancy across socioeconomic groups is not straightforward and will depend on the exact design of the system and other social policies in place, and how these policies interact with the pension system.

system, ensuring an adequate pension is available to all and setting the age at which one can draw a pension. The contributions to this Forum illustrate these elements well. Franco and Tommasino (2020) trace the history of Italian pension reforms, starting from the overhaul in 1992, which addressed severe fi nancial unsustainability and a fragmented pension system, and through subsequent adjustments. They then look at the outcomes in terms of labour market participation of older workers, fi nancial sustainability, pension adequacy and supplementary private pension savings. Their overall assessment of the Italian pension system is (mildly) positive, though spending on pensions as a percentage of GDP will outpace that of other eurozone countries. As they discuss, the system is not without challenges, some of which are common to other countries' pension systems. In particular, they take up the issue of pension coverage of non-standard workers, a category of employees predicted to increase in the future (OECD, 2019a).
The French pension reform proposal, currently going through legislation and social partner deliberation, is covered in great detail in this issue by Boulhol (2020). As with the Italian case, fi nancial sustainability and simplifi cation of a fragmented pension system are important objectives. Employees in France are covered by a multitude of different pension schemes with varying rules on pension entitlements. The latter limits both transparency and job mobility that requires a change in the pension scheme. It also creates inequalities concerning pension entitlements for otherwise identical career trajectories and life time earnings profi les. These issues are, to some extent, addressed in the new proposal, and Boulhol discusses some ideas for further desirable improvement.
Valkonen, in his succinct description of the Finnish pension system, attributes its relative success in terms of fi scal sustainability and coverage to a "capacity to make extensive reforms when required" (2020, 92). Besides, the Finnish system has an automatic sustainability stabiliser linking the pension age to life expectancy -a feature now available in a number of EU countries including in the Italian system as well as the current reform proposal for the French pension system. Valkonen then goes on to discuss the implications of potential challenges to the Finnish pension system and pension systems in general. Like Franco and Tommasino, he touches on life expectancies among socio-economic groups and changes in the struc-Mikkel Barslund, Centre for European Policy Studies, Brussels, Belgium. DOI: 10.1007/s10272-020-0873-5 Pension reform has been a staple ingredient in economic and social policy-making in most European countries over the past three decades. Some countries, such as Italy, enacted sweeping reforms as far back as the 1990s, whereas others are still in the process of comprehensive pension reform (France). Population ageing -leaving relatively fewer people of working age to support pensioners and occurring with varying speed in different countries -has been the most important catalyst for reforms. Pension reforms have often been politically controversial and subject to intense political debate and public dissatisfaction. 1 Changes to pension rules often affect a large part of the population and due to the demographic situation most often involve a reduction in entitlements; hence large groups of the electorate can be mobilised. France is the proverbial example of this, with massive strikes over proposed pension reforms occurring once or twice a decade, most recently in 2019. Reforms in Italy following the fi nancial crisis also led to widespread social dissatisfaction.

Pension Systems in the EU -Some Policy Issues
There are two main dimensions to pension policy: a public fi nance dimension and a broader social policy dimension. In the face of population ageing, the critical public fi nance issue is that of sustainability. In Europe, most pension systems have a (substantial) pay-as-you-go (PAYG) element -current pensions are in part fi nanced by taxes levied on current workers. Securing public fi nance sustainability as the population ages comes down to a combination of increasing the effective retirement age and lowering the average pension amount paid out per person (Valkonen and Barslund, 2019). The social policy dimension is primarily about redistribution within the pension Forum approximation is to consider groups of people based on their educational attainment. Mosquera et al. (2019) did a structured literature survey of studies analysing differences in life expectancy at age 50 for EU countries. They found that differences in remaining life expectancy at age 50 between individuals with low (ISCED 0-2) and high (ISCED 5-8) educational attainment range from 1.6 years in Finland for women to 11.3 years in Estonia for men. Though these differences in life expectancy and health do not seem to be increasing, there is also no evidence that demonstrates that attempts to reduce socio-economic inequalities over the last two decades have been particularly successful (Mackenbach et al., 2016;Barslund and Ludolph, forthcoming).
To what extent a pension system can be judged fair given structural differences in life expectancy across socioeconomic groups is not straightforward and will depend on the exact design of the system and other social policies in place, and how these policies interact with the pension system. However, as pointed out by Franco and Tommasino (2020), if there is a tight link between labour market performance and the accumulation of pension rights (as, for example, in the Italian points-based sys-ture of labour markets. In Finland, a recent reduction in fertility challenges the long-term fi nancial sustainability of the pension system and Valkonen outlines potential options for adjustments. Werding (2020), writing about ageing and the pension system in Germany, assesses that the current system is fi nancially viable until around 2025, after which the effective retirement age will have to increase or pension entitlements must be reduced. A government commission is currently mulling over further reforms. One issue is that private -so-called third pillar -pension savings are not delivering the desired results. This is mainly due to low uptake and regulation that limits investment in equity. Werding argues that an expansion of occupational -second pillar -plans may be a promising path forward.

Financial sustainability of public pensions in a European perspective
All four contributions illustrate that when it comes to ensuring the fi nancial sustainability of pension systems, EU countries have come a long way over the past two decades. Ageing populations will lead to only a moderate overall increase in pension expenditures as a percentage of GDP towards 2040, as seen by comparing the lines from the 2015 and 2018 Ageing Reports (Figure 1). Contrast this with the fi rst European Commission report in 2001.
The main drivers of this development have been crosscountry decreases in the average pension paid out relative to average country wages (the pension benefi t ratio) and a substantial (expected) increase in the future effective retirement age. The increase in effective retirement is a function of a reduction in early retirement options and other pathways to early retirement, as well as an increase in the statutory pension age in most countries (Nordheim, 2016). This article discusses in brief two themes -touched upon in one or more of the articles in this Forum -of importance to current and future pension policy, namely differences in life expectancies and the issue of means testing of pension benefi ts and associated high implicit marginal tax rates.

The challenge of socio-economic differences in life expectancy
It is well established that there are (substantial) differences in average life expectancy among groups with different socio-economic status in most EU countries. While there is no one set defi nition of socio-economic status, one Forum very imprecise instrument to target differences in life expectancy and will probably introduce other inequalities as a result. Individual targeting and assessments of working capacity are alternative instruments to address health and associated mortality inequalities.

Means testing of pension benefi ts and high implicit marginal tax rates
Means testing of pension benefi ts occurs when eligibility for a given pension benefi t is subject to a test of other income or asset resources (means) available to the individual or the household. Means testing is often applied when a minimum or basic pension benefi t is in place to avoid old-age poverty, but has also been suggested as a means to reduce the costs of universal pension benefi ts (Cumbo, 2017). As a social policy instrument, means testing is cost effective in reaching targeted groups. However, it can have adverse behavioural effects and in the context of (private) pension savings can introduce very high implicit marginal tax rates (Sefton et al., 2008) when the means tested benefi t is tapered off as own resources increase. As an example, consider an individual close to retirement age with a labour market career of low wage employment and with many interruptions. This person may not, through the pension system in place, reach the minimum basic pension as he retires. He will therefore qualify for the means tested minimum pension once retiring. But, this also implies that any additional pension savings done before retirement (and which are included in the resources that the basic pension is means tested against) faces tem), converting these rights into pension annuities based on a population-wide estimate of life expectancy will short-change groups of individuals with lower than average life expectancy. 2 Heterogeneity in mortality rates also has a bearing on the calculation of pension adjustments in pension systems that allow for some fl exibility in retirement relative to a 'target' pension age, including where fl exible retirement is allowed only after the target age. 3 Postponing retirement beyond the target age increases future pension payouts in an actuarially fair manner, on average. Average life expectancy is the basis of this calculation. If a group of people have a shorter life expectancy than the population average, the pension adjustment for this group of people will be less than their shorter life expectancy warranted on strictly actuarial terms. Similar arguments can be put forward regarding a statutory pension age differentiated by differences in group life expectancy.
While it is clear that there are arguments in favour of taking group-specifi c mortality rates and life expectancies into account in the pension system design, it is much less clear whether such an approach is practically feasible. There is also the question of why this approach should be limited to the pension system, and not also, say, to health care usage or unemployment benefi ts, where there may also be population group differences in average usage.
An important aspect often missing in the discussion of group differences in mortality and health is that while there can be substantial differences in group means, the distributions of outcomes may have considerable overlaps. To illustrate this, consider the distribution of the number of chronic diseases for individuals aged 60-64 in EU countries by educational attainment (Figure 2). 4 The difference in means is visible; however, it is also shown that a signifi cant share of people with low educational attainment have better health outcomes than many people with high educational attainment.
The implication is that using educational attainment (or other broad socio-economic measures) is likely to be a an additional implicit tax equal to the means tested basic pension benefi t foregone as a result of the extra pension savings done prior to retirement.
In general, means testing can distort behaviour when it comes to decisions on how much to save (accumulate resources), and which assets to save in (which resources to accumulate) if not all assets are part of the means testing. A further complication is that means testing rules can be complex and not straightforward. This has the advantage of limiting distortive behaviour (as individuals cannot react to the adverse incentives), but may on the other hand lead to poor savings management by disadvantaged groups.
Means testing is prevalent in most EU countries (OECD, 2019b). It primarily affects asset-poor people, but in some instances can affect a larger share of the population. Andersen (2015) reports that for some types of means tested pension benefi ts in Sweden and Denmark, marginal effective tax rates on pension savings can be close to 100% at some pension wealth levels, once the tapering off of pension benefi ts is included. Means testing in those two countries also affects both low and middle income groups.
The trade-off with means testing is between targeting (and hence lower public expenditures) and behavioural distortions. The more precisely a benefi t is targeted, the steeper the tapering off of the benefi t, raising the implicit marginal tax rate on the accumulation of own resources. Managing this trade-off can only be done on a case-bycase basis. However, with many EU countries actively encouraging the development of private pension savings (either via occupational savings or private savings products), and with tight public fi nances, the importance of paying attention to means testing and distortionary effects will increase.

Concluding remarks
Pension systems differ across the EU due to historical legacies. Countries are also following different demographic trajectories and may, therefore, have different options available in adjusting to population ageing. Furthermore, pension systems interact with the tax code and many other aspects of social policy, including those related to long-term care and health services. Hence, the purpose of this Forum is to shed light on some key issues and features of pension systems relevant to most, if not all, EU countries by looking at the history of and policy discussion around the pension systems in the four countries of Finland, France, Germany and Italy.