Table of Contents
2. Pension Systems
2.1. Traditional PAYG Pension Systems
2.2. Notional Accounts
2.2.1. General Description
2.2.2. Risks in Notional Accounts Pension Systems
2.2.3. Example Italy
2.2.4. Transition Problems
2.3. Funded Pension System
2.3.1. General Description
2.3.2. Risks in Funded Pension Systems
2.3.3. Example Chile
2.3.4. Transition Problems
3. One-Pillar- versus Multi-Pillar Pension Systems
3.1. General Description
3.2. Example Switzerland
Numerous debates and ideas, concerning the pensions system and its reforms, have arisen during the last ten to 20 years all over the world. Moreover, in 1994, the World Bank published a book “Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth”. The diagnosis of this book, summarized by Michael Bruno, was that “systems providing financial security for the old are under increasing strain throughout the world”1.
“As long as the number of retirees does not grow faster than total earnings of employees [in the current system], the tax receipts can continue to pay the benefits of retirees”2. But because of important changes in the demographic situation and mis-constructions in present systems, this condition does not hold true anymore: first, the birth rate in most countries of the Organization for Economic Co-operation and Development (OECD) decreases continuously. Thus the number of young employees decreases, too. Second, the people, born in the time of the baby boom (after World War II, the so-called “baby boomers”), will retire during the next decade. Additionally, due to the progress in medicine and high living standards, longevity increases. According to a prediction of the World Bank (1998), the portion of people over 60 in OECD-countries will rise from 19.9% in 2000 to 31.2% in 2050. Almost a tripling from 7.3% to 20.7% is expected in Asia within this period3. Moreover, in many countries the pension system provides incentives for early retirement. For many “Western countries” it holds true that the net present value of pension benefits is significantly higher for a person retiring earlier than the legal retirement age. Therefore we can conclude that there seems to be an implicit penalty on delaying retirement. Furthermore Gruber and Wise stress that “in many countries, disability and unemployment insurance programs effectively provide early retirement before the explicit social security early retirement age”4. Also generous pension benefits have been developing in many countries during the last decades.
All of the reasons above make the dependency ratio, defined as number of workers per pensioner, deteriorate: the ratio declined from 3.5 workers per pensioner in 1950 to 2.5 in 1990 and it is expected to decrease further. This ratio is the most important measure for the sustainability of the pay-as-you-go (PAYG) pension system. Thus it is getting harder for governments to maintain financial stability for current pension systems. The status quo is that future benefits are larger and growing stronger than future contributions, meaning that the public pension debt can not be amortized by future contributions without any pension reform. The current system is financially not sustainable. One more drawback within current PAYGsystems: internal rates of return for contributors are low or even negative.
Another aspect to consider is international pressure through increased globalization. As some countries (e.g. Sweden, UK etc.) have already reformed their pension systems successfully, taxes for companies and consumers in those countries are lower than in countries with “old” pension systems. Thus a difference in efficiency of the pension system can lead to differences in the competitiveness of countries, which should be a serious issue for any government.
The problems mentioned above are most evident in OECD countries with PAYG pension systems. Other countries operating PAYG-systems with currently less severe demographic problems will experience similar problems, but at a later point in time. Due to the maturity and intensity of the problems in OECD countries (i.e. Germany or France), solving those problems can obviously not be done by devolving them to the following generations and have to be tackled in our days.
At this point we would like to stress that current PAYG-systems have not always been that inefficient. Current pension systems have been established during the first half of the last century, with the demographic and economic situation being different.
One possible solution, which has often been used to reduce financing problems of PAYGsystems, is parametric reform1. In this framework, the government can reduce benefits by increasing the normal and/or early retirement age, by reducing indexation of benefits or earnings, by changing the calculation-formula (e.g. changing the period used to calculate average lifetime earnings) for benefits, by increasing the tax on benefits or by increasing common social security tax receipts (contribution rates)2.
But as seen in the past in several countries, this kind of reform is only efficient in the shortterm. After a certain period of time a low dependency ratio evolved again and faces the systems with the same problems as before. Parametric reforms tackle the symptoms of the problems but not the cause. From our point of view an efficient approach to reform pension systems does not solely try to increase the dependency ratio, but tries to make the system more resistant against low dependency ratios. This can be done by a structural (fundamental) reform of the system. An assumption of the authors is that a structural pension reform of current PAYG-systems is superior. Hence the topic of this paper is to present pension systems suitable to tackle current problems of countries with mature PAYG-systems effectively. This is done by taking a closer look at notional account systems and funded systems. On the basis of this detailed analysis we will introduce a certain reform proposal, a mixed system consisting of three pillars. We also fortify our assumptions and statements by current examples of (successful) pension reforms, which have recently been introduced.
By combining a notional account pension system with a funded pension system a more sustainable and effective pension system will be created3. The reasons for this are better abilities to tackle demographic and incentive problems combined with a superior internal rate of return.
The second chapter provides a general idea of the three basic pension schemes: a brief overview regarding the PAYG-system and a more detailed consideration of notional accounts and funded systems. The third chapter talks about combining the systems into a multi-pillar approach. The forth chapter concludes this paper by merging the goals for a better system and the reform possibilities (mentioned in the second section) into a certain reform proposal.
2. Pension Systems
According to Willmore, formal pension systems have two dimensions1: The first dimension describes the way systems are financed, funded or unfunded. In funded systems individuals pay contributions into a pension fund, out of which their benefits will be paid. Within the unfunded system, individual’s contributions finance the benefits of retirees at present2. The second dimension refers to the way benefits are calculated, via defined benefit (DB) or via defined contribution (DC). According to DB, the level of pensions benefit is defined and guaranteed in advance. In contrast, benefits in DC-systems depend on the amount of contributions paid during working life. In this chapter, pension systems including different characteristics of the two dimensions will be introduced.
2.1. Traditional PAYG Pension Systems
The traditional pay-as-you-go (PAYG) system is an unfunded pension scheme with defined benefits. This scheme is managed publicly and “by far the most common formal pension arrangement in Europe”3. As the level of benefit is defined in advance by a certain formula, which is mainly based on earnings, there is no direct correlation between contributions paid and benefits received.
Contributions are paid via a social security tax, which is usually a certain percentage of the monthly wage/salary of the employee. Usually employees as well as employers contribute an equal part. Hence this system comes with higher labour costs, resulting in competitive disadvantages for companies in countries with a PAYG-system. Corporations try to pass on these costs to the employees by paying lower salaries/wages. If this should not be possible (due to downward wage rigidities) employers will eventually decrease their demand of labour. Alternatively some governments back up public PAYG systems by general budget revenues4 to assist employers.
Benefits are defined in advance and are calculated with a certain formula at retirement. They are mainly depending on the income of the individual earned during working life. To be more distinct, benefits are largely depending on a certain base income5. Meaning, there is an indirect correlation between earnings and benefits, but this correlation is limited, for instance by an upper boundary of earnings relevant for the calculation of benefits1. With this feature of the PAYG-system, which is incorporated into the calculation formula for benefits, redistribution comes into play, and redistribution is an important attribute of this system. Benefits paid are indexed to a certain rate, mostly at the wage growth rate or at the CPI growth rate.
Talking about the risks in PAYG-systems2, we would like to go with the risks compiled by Steven Ney3. Investment Risks, Company Insolvency Risks, Inflation Risk4, and Risks of Volatile Investment Returns do not exist, as no money is invested. While Longevity Risks, Political Risks and Demographic Risks have to be beared by the entire society, Individual Risks and Disability Risks are shared between the individual and the society.
To show how PAYG-systems do work in practice, we briefly consider the pension system in France, which is highly fragmented. It consists of 120 basic and additionally 400 supplementary schemes, all of which are PAYG-systems. Between those schemes, contribution rates, benefit payments, retirement ages (in most cases the retirement age is between 60 and 65), and other parameters do vary considerably. Private sector and public sector employees are treated fundamentally different: while private schemes comprise both levels (basic and supplementary), public pension schemes only consist of one level. In 1993 France reformed the public system via several parametric reforms. But as it turned out, main problems have only been delayed, not solved. In 2003 a further reform of public pension system was introduced, modifying parameters to calculate the level of benefits. Yet, the French pension system does not appear to be financially sustainable, which fortifies the authors’ assumption that parametric reforms are not suitable to reform pension systems efficiently.
2.2. Notional Accounts
2.2.1. General Description
The first scheme resembling notional defined contribution (NDC) was proposed by Buchanan in 19685.
The basic idea of a notional accounts system is to combine an unfunded PAYG-system with features of a DC system. Current benefits for retirees are financed by current contributions of the working generations (demographic funding). But the amount of benefits for each individual is closely tied to the contributions the individual has been paying during its professional career. Thus, a notional accounts system can also be considered as an insurance system.
Contributions are deducted from the monthly income of the individual at a fixed rate via a social security tax. In contrast to PAYG-systems, every individual has an own (notional) account/fund, to which the (notional) value of the contributions is accredited one to one. Alternatively points can be accredited to the account, with the number of points being dependent on the contribution and on other redistributional factors. Furthermore, there is a virtual (notional) annual interest rate, paid on the notional fund of accumulated virtual credits. In most cases this interest rate will be defined by the growth rate of the economy (e.g. in Italy) or by the growth rate of wages in the economy (e.g. in Sweden). The interest rate represents the internal rate of return in a NDC-system.
After having accumulated credits for contributions and notional interests, the notional fund is usually transferred into an annuity at retirement. Benefit payments are indexed to the wage or CPI growth rate. A further important aspect is that annuities are calculated on the basis of current life expectancies and are paid until the individual dies. This means that benefits for individuals getting older than the average life expectancy are financed with “unused” annuities of individuals dying early. But also other benefit-payout-models are imaginable. After retirement the total value of benefits is determined. Theoretically the individual should be able to customize benefit payments according to personal necessities, as long as the sum of benefits is not exceeding the final value of the notional account.
NDC-systems can also comprise redistributional components. For example, workers can be accredited points to their notional accounts during periods of disability, sickness, childbearing or unemployment. This can be “financed” by transferring those costs to current workers by reducing the credits in their accounts. Alternatively disability- and similar costs can be moved out of the system. In this case such costs would be financed out of the general governmental budget.
Talking about pros of notional accounts, an important feature is to closely link benefits and contributions. Benefits are even tied to the growth of the contribution base1. Moreover, the disincentive problem involving early retirement can be tackled with notional accounts (please refer to Appendix 1 for more details on the retirement decision). As individual benefits are defined by individual contributions, the net present value of total benefits is decreasing in case of early retirement (in contrast to an increase in PAYG-systems). In addition, the ageing problem is considered, because of the annuities being tied to current life expectancies. All of those three aspects increase the financial sustainability of notional accounts significantly, compared to traditional PAYG-systems.
Nevertheless there are still shortcomings with NDC-systems. Although the longevity problem seems to be tackled by making annuities dependent on life expectancy, we want to mention that it is hard to implement this feature: after assessing life expectancy and calculating the annuity, life expectancy may change during the long period of, on average, 20 years of retirement. Furthermore, notional accounts still have inherited several drawbacks of PAYGsystems, namely low internal rates of return and falling dependency ratios1, due to its unfunded nature.
2.2.2. Risks in Notional Accounts Pension Systems
In the following, the risks inherent in pension systems will be discussed with respect to the notional accounts scheme:
Investment Risks, Company Insolvency Risks: As NDC-systems are not funded, these risks do not exist.
Disability Risks: Whether these risks have to be borne by individuals or the society depends on the design of the system. But most implementations in practice transfer them to the society (e.g. via “imaginary contributions”).
Longevity Risks: There are two kinds of risks: the individual can die much later or earlier than expected. In the first case longevity risk is borne by the society, as annuity payments will be maintained until passing. In the second case the total value of benefits received by the individual will be smaller than the total value of the notional account at retirement. But the individual can reduce these risks by adapting the date of retirement into personal estimations about the own life expectancy.
Political Risk: A NDC-system comes with more emphasis on the individual (via the notional accounts) and a higher financial sustainability. Hence, incentives for politicians to alter future benefits are lower than in PAYG-system.
Inflation Risk: As current benefits are financed by current contributions, there is nearly no inflation risk2. Since credits in the notional accounts (as well as benefits) can possibly be indexed to the CPI growth rate, the notional accounts still provide a meaningful measure for the total value of benefits.
Volatile (Investment) Returns: Returns in this system do not come from investments in assets, but from the notional interest rate. Since interest rates (such as GDP- or wage growth) are not as volatile as equity and bond returns, this risk is rather low.
Demographic Risks: As in the PAYG-system, current benefits have to be financed via current contributions. Hence demographic risks are borne by the society.
Individual Risks: Individual uncertainty concerning someone’s career path is shared between individuals and the society, as some of those risks (i.e. childbearing and redundancy) will be financed by the society.
2.2.3. Example Italy
Popularity and application of the NDC-system increased in the last two decades. Notional accounts were introduced in countries like Latvia, Poland, Sweden and Italy. In this section, we consider the implementation of notional accounts in Italy.
The new pension system was installed in August 1995 during the “Dini-Reform”, named after the Prime Minister Dini, who was decisively involved in putting this reform into action. The previous Italian pension system was fragmented and very generous, thus being one of the most expensive old age security system in OECD-countries (with contribution rates of over 30% the system was still running a deficit).
The reform consists of several parts. First, a shift from an earning-based to a contributionbased system was implemented: contributions are credited into a notional account and revaluated (indexed) annually in line with a moving average of GDP-growth. At retirement pension benefits are determined, according to gathered credits in the personal account and consideration of the retirement age, residual life expectancy and expected productivity growth. Next, retirement age turned to be flexible, namely between fifty-seven to sixty-five years1. Hence, benefit payments were linked to the respective residual life expectancy at retirement. Benefits are indexed to either wage growth or CPI growth. Third, the contribution period was raised by five years. Although there is no ceiling on contributions, a ceiling on retirees’ income was set; thus entailing redistribution into the new pension system. Additionally a voluntary, private pillar was introduced, which was backed with financial incentives to stimulate private saving.
By improving the dramatic situation of the Italian pension system at least partially, the reform can be considered as a success. Nevertheless the entire pension system was still running significant deficits in the following years. This was primarily due to the long transition period allotted for the reform: new entrants participate in the new system, individuals who have been contributing for more than eighteen years, stay in the old system. The remaining group will be part of each system, relative to the duration of contribution to the old system2. The so-called “Prodi-Amendments” of 1997 tried to further decrease deficits, by limiting total pension spending to a certain portion of GDP.
2.2.4. Transition Problems
As the NDC-system is not funded, there will be only moderate financial problems regarding a transition from a PAYG-system. Current benefits can still be financed with current contributions. The most striking problem regarding a switch from a PAYG- to a NDC-system is that internal rates of return are usually positive in the notional account system (as described above). With several mature PAYG-system, for example those in France and Germany, yielding negative rates of return (benefits are lower than contributions for the individual). The question evolves: How can the strongly increased level of benefits be financed after a switch to notional accounts? One possibility to deal with this problem is to tie the notional interest rate to the annual wage growth of the total contribution base1, not to per capita rates. Actually this solution might result in low or even negative returns.
Next we want to take a look at other and less important aspects of a transition, e.g. transparency and comprehensibility: At first sight the notional account system seems to be transparent, as an intuitive link between contributions and benefits exists. But by taking a closer look it becomes evident that the calculation of benefits is rather complicated: numerous and sophisticated parameters have to be considered to derive benefits from the final value of the notional fund. As benefits are not previously defined it has to be made sure that individuals have the possibility to obtain estimations regarding future benefits. In Sweden the administrative system provides such a “service” to contributors. This means that there are higher costs of managing and administrating the system.
After considering notional account pension systems we come to the conclusion that NDCsystem are superior to traditional PAYG-systems. While the central issue of the PAYG system, redistribution, can still be implemented in the framework of notional accounts, NDC systems seem to be fairer and have better abilities in tackling some demographic and disincentive problems (early retirement and longevity). Hence NDC-systems are financially more sustainable.
2.3. Funded Pension System
2.3.1. General Description
As mentioned above, funded systems do not finance benefits of the retired generation via contributions paid by the “working generation”. Contributions of the individual are invested into a fund, out of which the pension of this individual will be paid from retirement on. This feature represents the most fundamental difference between the two pension systems. Therefore, in a funded system every individual is responsible for his/her own pension. A funded system can be organized either way, as a defined benefit system, or alternatively, as a defined contribution system. In this part of the paper we want to focus on the latter, due to two reasons. Firstly, we have already been talking about defined benefits (in the context of PAYG). Secondly, and more important, defined contribution systems are much more popular and well-established than defined benefit systems.
1 See Fox/Palmer “New Approaches to Multipillar Pension Systems: What in the World is Going On?”, page 90
2 See Gruber and Wise, “Different Approaches to Pension Reform from an Economic Point of View, page 50
3 See Fox and Palmer, table 3.2
4 See page 53 of their paper. For example, unemployed people can “retire” earlier than employees. 3
1 I.e., retaining the system but changing certain parameters.
2 See Gruber and Wise, section 2.2.2
3 This proposal will be recommended as “best structural reform” for countries with a mature PAYG-system. Moreover we emphasize, that this will not be the best solution for any country, as this depends on local circumstances.
1 See Ney, “Thinking about pension reform: Discourses, politics and public participation”, page 5
2 The unfunded system corresponds to “Umlageverfahren“ in Germany, the funded system to “Kapitaldeckungsverfahren”.
3 See Ney, page 6
4 See Ney, page 6
5 This base income can be calculated in various ways, i.e. the last five years of earnings just before retirement (in Italy before 1993) or the highest 35 years of earnings (in the US)
1 No additional benefits will be earned by the individual if this boundary of earnings is exceeded, but contributions are still paid as a percentage of earnings without any ceiling.
2 The following is just a very brief consideration of those risks, as PAYG-systems are not the main part of this paper. Risks will be discussed in more detail for NDC- and funded systems.
3 Steven Ney has listed these risks, consisting of risks defined by the World Bank (1994), Willmore (1999) and the ILO (1999) in his paper “Thinking about pension reform: Discourses, politics, and public participation”, page 4 -5.
4 In case of hyperinflation the period between the contribution and the distribution of the contributed funds via pension benefits incorporates a risk
5 Buchanan’s proposal included in general a replacement of social security payroll taxes by mandatory individual purchases of social security bonds, which would be credited with a rate of interest compounded over the working life. Please refer to Disney, “Notional accounts as a pension reform strategy: An evaluation”, for more details.
1 As benefits are also depending on the internal rate of return of the notional accounts, with those returns being defined by wage growth, there is a reasonable connection of future benefits and current contributions, which is not necessarily the case for PAYG-systems
1 Despite its ability to lower problems arising from early retirement and increased life expectancy, NDC-systems cannot deal with decreasing birth rates.
2 Although in case of hyperinflation the period between the contribution and the distribution of the contributed funds via pension benefits incorporates a risk
1 Yet, retirement age remains among the lowest in OECD-countries.
2 See Disney page 10, “Notional accounts as pension reform strategy: An evaluation” 9
1 The wage earned by the total contribution base is the wage of all contributors earned in one year. 10