Digest of Our Key Views on Pension Systems

5 November 2018

With the population ageing in most countries, having in place an efficient pension system becomes more and more important.

Three main objectives can be assigned to the pension system, and the proposals to reform existing ones must be judged according to these three criteria.

  • First, the “no regret objective”. Retired senior people must be broadly happy with the revenues they get. They must not regret in insight not having saved more during their active life. This primary objective is hard to attain since people tend to prioritize their short-term well-being (a sort of bias that behavioral economists call “hyperbolic discounting”). Left on their own, they may under-save despite knowing perfectly well that sooner or later they will regret it…
  • Second, the “minimal cost objective”. Reaching the “no regret objective” requires a very significant effort during the active life, call is “saving” or paying “social contributions” depending on the system. Obviously, these large costs should be minimized and this key “minimal cost” objective will lead us to compare, for a given level of future pensions, the cost-efficiency of pay-as-you-go systems mainly based on social contributions with the performances of the various sorts of fully funded systems based on prior saving. Along the way, some surprises may await us…
  • Third, and finally, the “minimal risk objective”. Depending on how the system is organized, future pensioners may face various sorts of risks. In “defined contributions” funded systems, they may face some large financial risks as their pensions may depend a lot on the financial markets’ performances. In all systems, they face a “human capital” risk as, due to health problems or some professional skills’ obsolescence, they may find difficult to stay active long enough to build enough pension rights. These risks can have a very large negative impact on workers’ and pensioners’ well-being and should be considered when one compares the efficiency of various systems.

Obviously, these three key objectives are not independent. The “no regret objective” and “minimal risk objective” are indeed very much complementary since a system characterized by a large amount of risks will sometimes produce bad outcomes in which many people will have a lot of regrets. Unfortunately, the “minimal cost objective” does not seem so harmoniously coherent with the two others. Protection is in general costly and thus there is apparently a fundamental trade-off between the desire to limit the risks born by future pensioners and the parallel desire to limit the pension system’s costs for the active workers. In some way, an efficient pension system is precisely a system that optimizes that trade-off between safety and costs, and we’ll see that unfortunately most existing systems probably fail to do that![1]

For a start, no pension system has been built from scratch based on a rational analysis of these three objectives. Many decades ago, pragmatic initiatives, sometimes public and sometimes private, have been taken in most counties to try to alleviate the disturbing obvious poverty of many old people. Like many other public interventions (see our discussion of the a-theoretical approach to financial regulation), most pensions systems result from the accumulation overt time of reforms by pragmatic governments trying to fix the most evident drawbacks of what they have inherited from past private or public initiatives. As a result of this chaotic process, a striking feature of existing pension systems is how diverse, and often uselessly complex, they are.

Ambitious governments which want to improve radically the efficiency of the system they have inherited from the past are confronted to the realities of vested interests and the economic and political costs of a transition to a better and less complex system.  It has always been the case in the past, and it is still the case today.

Yet, it makes sense to discuss what, with the benefit of insight, would be the main characteristics of an efficient system built from scratch. The interest of this exercise is not only theoretical: In France, despite the costs, M. Macron seems to be ready to implement the most ambitious reform ever made within the French pension system, and it is the economists’ responsibility to nurture the reflections of courageous politicians!

Five key points need to be considered in this discussion about an optimal pension system. The first three points will not look very original as they will insist on some basic aspects of how a pension system should work. The last two points are in some way less trivial and more interesting as they discuss some often ignored differences between funded and pay-as-you-go pension systems, and the best way to profit from financial markets’ opportunities to minimize the cost of providing pensions.

1/ Let’s start with the basics: the “no regret objective” probably means that workers should be forced one way or the other (saving, paying social contributions…) to prepare for their future retirement. If too much freedom is left to workers, it seems rather obvious that a large proportion would prioritize their short-term consumption. Thus, it is the responsibility of the state to devise a system which is indeed constraining and costly in the short-term but will provide retired people with a “no regret” sort of income. Moreover, a compulsory system may provide some positive economies of scale and help minimize the cost of providing pensions (more on that later when we’ll compare the efficiency of pay-as-you-go systems to that of various fully funded systems).

2/ Yet, in order to maximize people’s well-being, this “paternalistic” approach justified by the “hyperbolic discounting” fear should not be pushed too far. Towards the end of their professional life, people should become free to express their preferences between “leisure” and “money”. Some may choose to stop working relatively early, maybe because they have accumulated some personal saving on top of their pension rights, and others, maybe because they want to help their children or because they enjoy so much their work, may continue to work until an advanced age. The system should allow these different choices in a way which respects what is called “actuarial neutrality”: the sooner you ask your pension, the lower the pension is, and this “malus” should reflect precisely the cost for the pension system of your early retirement choice. Obviously, with supposedly myopic workers unable to assess the long-term consequences of their decisions, the ”no regret” objective may justify fixing a lower age limit (a floor of 60? 62? 65?) to this ability to draw on pension rights and retire with a discounted pension. But it seems clear that the “hyperbolic discounting” risk is much lower with people aged more than 60 thinking about their future financial situation in the not too distant future, let’s say the coming five or ten years, than with people aged 30 trying to plan rationally their consumption and saving 35 or 40 years in advance. Thus, a liberal approach respecting individual choices makes more and more sense as people advance in age towards the end of their professional career: a rigid “legal age of retirement” has little place in an efficient pension system. Yet, this is not well understood and most discussions addressing the reforms needed by pension systems turn around this issue of the “legal age of retirement”. In France, during the 2017 presidential election won by Mr. Macron, all the candidates positioned themselves mainly on this topic (with proposals going from a return to a legal age of 60 to a progressive increase to 67[2]). There were little discussions of “actuarial neutrality” and the benefits of putting in place or improving a consistent scale linking pension rights to the various possible ages of retirement. The benefits of respecting people’s preferences are both direct (i.e. welfare gains) and indirect. On top of avoiding some costly constraints, the flexible “bonus/malus” approach to pensions calculations has indeed another fundamental advantage: it may inform on people preferences and hopefully help the government to better control the pension system. Indeed, it is very hard for a paternalistic state to decide how longer human lives should be shared between an active professional period and an inactive period of retirement. Everyone agree that longer lives probably need to make the pension system less generous and to encourage a longer professional career, but how to calibrate precisely the needed reforms? Should we nevertheless accept in the process a large increase in the system’s cost (saving or social contributions) or should we prioritize an answer based mainly on longer working lives? A flexible scale of pension benefits would help to answer: dominance of early retirements despite sizeable malus would send some sort of signal that most workers are ready to bear some significant costs in order to enjoy a long period of retirement. A majority of delayed retirements in order to take full advantages of the proposed “bonus” would send the symmetric signal about preferences. Indeed, high average bonified pensions would make a progressive upward shift of the pension scale (i.e. lower pensions for a given age of retirement) politically much easier.

3/ The most difficult questions related to pension systems are, as often in economics, those linked to the optimal degree of insurance and risk-sharing. By the sheer impact of ageing, people are particularly exposed to health risks towards the end of their active working life. Moreover, professional risks are also rising overtime. Using the economists’ jargon, one can say that, as workers gain experience, skills are in general more and more specific to the job they have. This accumulation of “specific human capital” may either, for some lucky workers, justify high wages at the end of the career as they excel in their job or lead to a very difficult professional situation if these specific skills based on years of “learning-by-doing” lose their value (thanks to structural shifts in how companies or sectors operate). We should acknowledge that these risks may pollute significantly the signals that policy makers can get from the “malus/bonus” policy we advocated in the previous paragraph. Some older workers may retire early not because they have a strong preference for leisure, but because they simply don’t have access to satisfactory jobs. Obviously, the pension system may provide in many ways some sort of insurance against these professional risks. For example, under certain conditions, early retirement may be possible without penalties. Yet, one can argue that the pension system should not be overloaded with these risk-sharing issues and that other “specialized” public tools are better suited. Protecting older workers is more naturally part of the missions of the unemployment insurance funds. Obviously, the prevention issue is also very important: in sectors affected by big structural changes, is it possible to train workers, especially the older ones, to acquire the new required skills? There are also many key questions related to labor laws: should companies be required to provide some special protections/insurance to senior workers until a certain age? Indeed, in some countries, the controversial “legal retirement age” has a dual function that is not always well understood. On the one hand, it marks when it becomes possible to ask the payment of a pension. This is its more visible function. But, on the other hand, it may also fix the minimum age at which people with stable labor contracts may at last be made redundant with little justifications required form the employers. In this second function, the higher is the “legal retirement age”, the better workers with stable contracts are protected! These complex “insurance/risk sharing” questions are obviously very important, but we’ll not discuss here the comparative advantages and drawbacks of the various approaches to older workers’ protection[3].

Let’s now discuss the other key issues related to how financial markets may help (or not) minimizing the costs of providing pensions.

4/ First, a key observation: When interest rates on government bonds are lower than the long-term growth potential of an economy, the pay-as-you-go system is more cost-effective than a funded system. This is currently the current situation in most advanced countries and the difference between “g” and “r” is sometimes quite large. A stabilized and sustainable pay-as-you-go system provides workers with a return on their social contributions that is mechanically equal to the long-term growth rate of the economy. In a pure pay-as-you-go system, the social contributions paid today give the right to receive as pensions the larger contributions paid tomorrow by the next generation of workers. It is not difficult to show that, with stabilized social contribution rates, the internal rate of return that workers get on their contributions is thus just equal to the long-term average growth rate of the economy. In some ways, in a pay-as-you-go system, the government can be seen as issuing a special form of debt reserved specifically to workers/pensioners with a coupon equal to the long-term growth rate of the economy. In a fully funded system, workers or companies saving on their behalf have no access to this special form of debt and have to do with the yield available on the traded public debt. In the past, the funded system was sometimes providing quite a good deal as market interest rates were high (as in the 80s and part of the 90s). Yet, the current situation is in some countries no short of a catastrophe for the workers saving to prepare their retirement. In the UK, 10-year gilts indexed on inflation provide at time of writing (June 2018) a return of minus 1,6% in real terms. For 30-year gilts, the real return is minus -1.5%! With such low returns, the cost of a funded pension system explodes. One euro invested with a real return of 1.5% per annum will be transformed in 1.64 euros 35 years later, while it will shrink to 0.59 euros with a real (British) return of -1,5%… In France, the situation is less extreme. Yet, French 30-year government bonds provide a real yield slightly negative, while the pay-as-you-go system is probably able to provide a real annualized return close to 1.5%. It may sound a bit provocative considering all the criticisms pay-as-you-go systems have endured in the past, but in the current financial environment, having a well-managed pay-as-you-go system is quite a big advantage for the future pensioners. Yet, the situation may change in the future and countries like France or Italy must recognize that the new-found relative attractiveness of their pay-as-you-go pension system for workers depends a lot on real interest rates staying low (and for the record the aforementioned concept of actuarial neutrality being at last fully understood and implemented!). It is indeed a supplementary reason to protect European countries against the risk of higher interest rates and to avoid financing the public debts with low maturities bonds (see here our discussion of Eurozone reforms).

Maybe some readers may find the previous discussion a bit biased, or even dishonest, since funded pension systems don’t invest only in government bonds, and thus may benefit from the risk premia provided by other asset classes. And a few hundred basis points of risk premia will make a huge difference for pensions saving invested over decades. This fundamental remark would be right! Yet, risk premia are not provided as free lunches. They are here to compensate for the financial risks accepted by investors. And this reward against financial risk is obviously perfectly available for workers benefiting from a pay-as-you-go system and having thus access to the special growth-indexed bonds implicitly issued by their government. Benefiting from a safe pay-as-you-go pension, they may allocate their personal saving to rather risky investments and collect the available risk premia as well. Thus, it is 100% true that financial risk premia may be very helpful to cut the cost of pensions. For someone ready to accept some rather reasonable financial risk, the return provided by funded pension systems will be significantly above the return provided by risk-free government bonds. But for workers benefiting from a pay-as-you-go system, the same sort of bonus is available if they are ready to accept the same level of financial risk. In other words, it is important to acknowledge that in a well-devised system, financial risk premia may make it less expensive to prepare for retirement. Yet, there is absolutely no reason to believe that the benefits are necessarily much stronger in a funded system.

Indeed, some key fully funded systems are just doing a horrible job with respect to risk premia harvesting. We need here to say a few words of “Defined Benefits” (DB) plans sponsored by private companies. In the past, it was probably the dominant way to provide pensions in many countries where historically pragmatic companies rather than the government took the first initiatives in order to tackle old-age poverty. DB plans seem to offer the best of worlds. On the one hand, workers are supposed to be free of financial risks: expected pensions depend on workers’ careers, not on financial market performances. Companies guarantee the payment of these future pensions by holding in their pension funds a large stock of financial securities. As they are making some long-term investments, they can prioritize the holding of equities over bonds, and benefit from large risk premia. As a result, thanks to the high return provided by their investments, they hope to contribute little to these funds. Until the early 2000s, this was apparently a very successful strategy and, indeed, most pensions experts in the US or the UK were looking at the high cost of social contributions in pay-as-you-go countries like France with a lot of commiseration…. Yet, this success was based on a complete illusion! If companies guarantee the payments of pensions (with watertight “defined benefits”), they support all the investment risks present in their pension funds. They become some sort of highly leveraged financial institutions using the money implicitly lent by their workers to invest in risky assets. Obviously, the return provided by this leveraged strategy has no reason to benefit the workers but should mainly go to the shareholders who support the extra risks. At that time, this basic principle was not well understood and a 20-year equity bull market blinded shareholders who were happy to take some huge financial risks without being compensated for it. As we wrote in the spring of 2001, near the peak of the internet bubble and the DB mania, “a bear stock market could send a wake-up call to financial analysts and shareholders investing in companies with large defined-benefit pension funds” [4].  Indeed, the last 15 years have been a sad period for most defined benefits pension plans sponsored by private companies which have been either closed to new members or managed in a much more cautious way, i.e. invested in bonds providing a very low return.

5/ Harvesting risk premia is indeed a complex business, and several steps could be done to improve how funded or pay-as-you-go systems help workers to do so. As a starting point, it should be clear that, as illustrated by the DB private pension funds’ descent into hell, there is (almost) no free lunch: in order to benefit from risk premia, one must accept to bear financial risks. Yet, there are many ways to organize how people are rightly exposed to financial risks and some approaches may be much more efficient than others.

Let’s start by two extreme and opposite approaches for risk premia harvesting: the “pure” individualistic model and the “Sovereign Fund” approach. In the “pure” individualistic model, people choose the financial assets they invest in. Obviously, they may, or rather they should, be helped by investment professionals as advisors or managers of the mutual funds they buy (more on that later). But these investment professionals bear little risks: at the end of the day, workers and pensioners bear in a transparent manner most if not all the financial risks and benefit from the risk premia[5]. The pure individualistic model of risk premia harvesting can be fully integrated into a Defined Contribution funded pension system (see for example the 401k saving accounts in the US) or run parallel to a pay-as-you-go system (with often some special tax incentives when facultative savings are blocked until retirement). In the “Sovereign Fund” (SF) approach, the public sector is making large financial investments on behalf of the citizens who at some stage will share the benefits (or sometimes the costs!) of these investments. Obviously, there are many sorts of SF. They may be fully integrated into the pension system, which benefits from planned contributions from the SF. Or they may be separated from the pension system and make contribution to the overall state budget (as it is the case in Norway even though the very large fund is named The Government Pension Fund Global).

The SF approach has many advantages relative to the individualistic model. First, a large fund is probably able to benefit from economies of scales and reduce the various costs (including the administrative fees) linked to the management of individual accounts. Second, building SFs is a way to recognize that most people are unable to harvest efficiently the risk premia provided by various assets classes. In particular, the volatility of financial markets can be quite stressful. Even when the long-term risks are limited (for example on long-term bonds issued by solvent government), the large day-to-day changes in prices can be very anxiety-generating and lead people to take irrational decisions. Last but not least, for more fundamental reasons than the potential economies of scale already mentioned, the SF approach may be able to do something quite extraordinary: it may improve strongly the Sharpe ratio perceived by citizens, i.e. improve the excess return they get as a compensation for the risks taken. The main reason is that SFs may be able to spread risks over several generations: a bad financial outcome can be passed on pensions and/or contributions over several decades. This risk sharing capacity may improve for each generation the ratio between the return they get and the remaining risks they must accept[6].

Yet, this ability to share risks between generations raises the question of the rules used to fix year after year the contributions of the fund (whether to the pension system or to the state budget). This is a difficult problem of governance which is not specific to SFs, as it is important for other collective funds benefiting a community made of several generations (see the rather similar problem faced by universities’ endowment funds). Having large funds collectively owned presents some huge advantages (economies of scale, risk sharing….) but also introduces the risk of the misappropriation of these funds by special interests or a specific generation. We’ll not discuss this key point here, but obviously an efficient SF rests on a strong and transparent system of governance (with well-thought checks and balances). Thus, there is no way to issue a general rule on what is the best system. The optimal balance between the collective and individualistic components may well depend on the robustness of a country’s political and social institutions.

Moreover, it is important to note that the “pure” individualistic model and the SF approach we have just discussed are some sort of extreme “corner” solutions. Harvesting risk premia can involve some intermediary approaches. Some life insurance contracts, for example, can provide a small bit of intergenerational risk sharing in a broadly individualistic context. Equally, the collective approach does not need to be at the national level with a full coverage of all citizens. Many collective funds are organized to cover workers and pensioners in a specific economic and/or geographical sector, yet with a large degree of intergenerational risk-sharing[7].

Another interesting point is that whatever the system (individualistic or collective), it is not absolutely necessary to save massively in order to benefit from risk premia. We have already stressed that even pay-as-you-go systems with little funding may allow some significant risk premia harvesting. Indeed, with leveraged positions, using various sorts of derivatives, it is possible to bear financials risks, and collect the risk premia, with a limited amount of prior saving. Yet, it should be noted that it is much easier to introduce such an element of leverage in a collective framework. Indeed, many countries have some sort of sovereign funds making large risky investments, and hopefully harvesting risk premia, while at the same time maintaining a significant public debt.

Finally, independently of the main characteristics of the pension system, we cannot insist enough on how important the fund management industry is to an efficient process of risk premia harvesting. The key role of the industry is obvious when the system is individualistic as few individuals have the required skills to build alone an optimal risk premia strategy over the long-run. As explained at length in this website, efficient risk premia harvesting needs quite a lot of economic analysis of returns and risks across many asset classes. Thus, individual savers need to rely on fund management companies which should invest enough resources to devise an optimal risk premia strategy. And in this area, the current funds’ offering seems very perfectible! As discussed elsewhere in this website, progresses should be made in two key dimensions:

  1. The investment process across many asset classes, which should be based on a dynamic analysis of risk premia with a long-term horizon in mind.
  2. The reporting process which should help individual investors to evaluate the risk they take and the return they can expect over a long-term horizon. Thanks to a high-quality reporting process, the volatility of financial markets should become less painful for individual investors. When people are suffering from a decline in the value of their “risk premia” mutual funds, it is important to explain why, and how the fund manager sees the resulting outlook. If the fall is mainly attributable to an increase in risk aversion, it will have no lasting consequences and may even benefit flexible investors as a well-managed risk premia funds will be able to profit from more attractive valuations. Yet, the fall in value may result from real fundamental factors, for example a less favorable outlook for corporate profits, and be as a result a really bad news for investors. This sort of real-time analysis of returns, which is currently seldom done, is necessary, not only in order to correctly manage the funds but also to inform investors as they should be.

It may be tempting to believe that the efficient harvesting of risk premia is less dependent on the fund management industry when it relies on the SF approach. In general, large collective funds prefer to control themselves the key asset allocation decisions and delegate externally only some relatively minor decisions (stock-picking or short-term tactical decisions with a limited risk-budget). Thus, they are not especially dependent on the offering (or the lack of it!) of well-managed diversified funds investing across various asset classes. Yet, it is far from sure that the current balance between what is managed internally and what is externally delegated is optimal. A better offering of diversified funds may also help SF, but this is a rather complex issue that we’ll not discuss in this “digest of our key views…”.

November 1st, 2018

[1] On top of this optimal balance between safety and costs, “fairness” is also a secondary objective often assigned to the pension system. There should be a strong link between what people contributed during their working life and what they receive once retired (except maybe for some clear redistributive policies in favor of people having had some special difficulties during their professional career). Indeed, a key objective of M. Macron in France is to improve the fairness of a system which treats differently various categories of workers in the public and private sectors.

[2] The “legal age of retirement” is currently fixed at 62 in France.

[3] Obviously, “moral hazard” is the key issue. “Moral hazard” limits the possibility to insure risks which are partly under the insured control. Too much public insurance against professional risks may lead to inaccurate behaviors by workers and companies, and to job losses for senior workers which could have been avoided. In this respect, forcing companies to protect workers may limit “moral hazard” risks, as companies may have the right information and incentives to manage their workers’ skills. This key question is discussed (unfortunately in French) in O. Davanne and T. Pujol « Assurance et échanges de risque sur le marché du travail», Economie et statistique, N°291-292, February 1996.

[4] We attended in the spring of 2001 an international conference on pension systems sponsored by the German Bundesbank. In our contribution, while stressing the reforms needed in France, we also insisted on the widespread mismanagement of funded pension systems. See here this contribution published a bit later in the book “Ageing, Financial Markets and Monetary Policy”, edited by Alan J. Auerbach and Heinz Herrmann, Springer, 2002.

[5] Some investment professionals may be remunerated with performance fee and thus bear a small part of the financial risks involved.

[6] In general, as we stressed for DB private pension funds, there are little free lunches in finance, so it is important to understand why a collective instrument may radically improve on the outcome of an individualistic model. The reason is that the individualistic model is victim to a fundamental well-known “market failure”. In theory, to maximize the benefits of risk premia, people should be exposed to financial risks and collect the related risk premia over a very long period of time. This time-diversification (i.e. being exposed to financial risks at very different periods of times) improves considerably the risk/return profile of investments as does the traditional diversification across asset classes. Yet, by definition, in the individualistic model, people cannot be exposed to the financial risks taking place before they are financially active (mid-twenties?) and they will not benefit from the related risk premia. Indeed, in the individualistic model, different generations have often mechanically an opposite exposure to financial markets’ risks. A stock market crash will penalize the most heavily invested generations (people in their 40s, 50s and 60s) and provide opportunities for the younger generation able to profit from more attractive prices. This seems impossible to avoid and yet this is just the opposite of the optimal risk sharing efficient markets are supposed to offer! SFs may radically improve how financial risks are shared and as a result improve the Sharpe ratio perceived by all generations.

[7] While France is mainly a pure pay-as-you-go country, there are a few collective funds allowing a collective harvesting of financial risk premia for specific sectors. It has long been the case for independent professionals (doctors, architects…) and more recently (2005) such an approach has been introduced for public sector workers (“Retraite Aditionnelle de la Fonction Publique”).