The Research Training Network

Financing Retirement in Europe: Public Sector Reform and Financial Market Development

The RTN is a collaborative network funded by the European Commission to support young researchers wishing to undertake research in pensions, at a number of European institutions. It complements the work of the UBS Pensions Research Programme.

Participating institutions: Centre for Economic Policy Research [CEPR]; IDEI, Université de Toulouse; University of Amsterdam; Universitat Pompeu Fabra; University of Salerno; CORE, Université Catholique de Louvain; FMG, London School of Economics and Political Science.

The aims of this network, led by Ronald W. Anderson, Helmuth Cremer, Frank De Jong, Xavier Freixas, Marco Pagano, Pierre Pestieau and David Webb, are outlined in brief terms below.

Project objectives

1. Pensions, growth and risk bearing
2. Horizontal equity, integration and political economy
3. Pension reform and the development, efficiency and regulation of investment-based pensions

The research undertaken by the network will be highly original in that it combines the latest techniques of financial economics and public economics to address the concrete questions of institutional design posed by reforming the system of retirement funding in Europe. It will carry the study of retirement finance beyond the insights afforded by traditional tools (classic overlapping generations models and demographic simulation) in order to obtain policy conclusions regarding specific institutional features of pension finance. 

Specifically, this will involve applying recent developments in dynamic portfolio analysis in the presence of market frictions, political economy, and principal/ agent analysis in complicated stochastic environments. While some of these developments have begun to be applied to the study of U.S. pension reform (Campbell and M. Feldstein (2001)) many important methodological problems remain to be solved. Furthermore, increasing realism also implies that the modelling should incorporate the important features of the relevant institutional environment. Thus the European focus of our network will lead us to certain modelling choices that will oblige us to break methodological ground on several fronts. While our network does not follow the rigid form of a tightly planned project, there is considerable congruence in the scientific contributions and agendas of the members. These can be grouped according to three broad topics.

1. Pensions, growth and risk bearing

A young person attempting to prepare for old age is faced with risks in several dimensions that result in a set of life-time savings, portfolio and labour market choices that is fundamentally complex in nature:

1. They are exposed to large and persistent shocks to their labour market income.
   
   
2. The returns to financial assets are uncertain and subject to wide fluctuations for periods that span a typical period of economically inactive old age.
   
   
3. The length of life-time is stochastic.

The institutions of public sector pensions and market finance have very different ways of dealing with these risks. Changing the balance between PAYG defined benefits pensions and funded, defined contributions potentially will have a very large impact on the risks borne by individuals depending upon their age. Furthermore, the realisations of the random variables in their environment (e.g., a down-turn in the stock market or a substantial shock to labour affecting its productivity and earnings) can result in significant intergenerational inequality. To compensate for these altered risks brought on by pension reform, individuals may react with substantial changes in their savings and labour market choices.

Important, but partial, insights have been gained in some recent studies of retirement finance. Some of these have focused on uncertain rates of return on investments, but have ignored fluctuations in labour market income (Feldstein and Ranguelova). Alternatively, studies with stochastic labour incomes have typically ignored stochastic investment returns (Imrohoroglu et al (1995) ) A limited number of papers have dealt with joint uncertainty about labour earnings and investment returns combined in the same analysis of pension finance (Campbell et al (2001) and Miles and Cerny (2001)). The difficulty in this area of modelling is that most calibrated models have needed to make assumptions that certain parameters are fixed which would likely change if there were a major shift toward funded/ DC pensions. For example, the issue of the implications of correlation of labour incomes and investment returns is still an open and hotly debated issue (See, Jaganathan and Kocherlakota (1996). Bergantino (1998), Poterba (1998), and Brooks (2000)). Furthermore, the opportunities and constraints imposed by real world financial contracts have been reflected crudely, if at all, in the models (for an important step in the right direction see Storesletten et.al. (1998)). Thus much work remains in improving our understanding of the ramifications of shrinking PAYG/DB pensions and reinforcing funded, defined contributions pensions.

The choice of PAYG versus funded transfer systems has been studied in a recent paper by Bhattacharya, Fulghieri and Rovelli which, unlike most papers in this literature, focuses on transitional (out of steady-state) economies. This choice is important in an advanced relatively slow-growing economy, but it is particularly interesting economic environments of rapid capital accumulation, such as those in Eastern Europe. This examined the key tradeoffs and inter-generational conflicts arising in a dynamic environment with long-lived capital stock (which can be the basis for funded schemes), and agents having uncertain lifetimes (or time of justified retirement). However, weighing the trade-offs among possible solutions of these conflicts is hampered by the limitations of the modelling. There is a need to extend this analysis of the transition to environments with risky returns on capital and aggregate risk regarding demographic parameters. This is a necessary step toward building operational models that allow for the calibration of the impact of alternative pension schemes on the levels of savings, portfolio choice (risk-taking), market interest rates and risk-premia, and investors' welfare.

One of the conclusions of previous analyses of reducing PAYG, defined benefits pensions, is that risk averse agents may respond by increasing savings rates very significantly which can have a large differential impact on welfare of different generations. (See Campbell et al (2001) and Miles and Cerny (2001)). These analyses typically assume very underdeveloped financial sectors. Much further work needs to done to assess the sensitivity of the models to alternative configurations of the financial sector. A first step in this process will be to examine the consequences of the existence of complete markets for real annuities (Brown et al (1999)). This work needs to be extended to general equilibrium under more realistic specifications of the stochastic environment.

An important difference between a defined benefit plan and a defined contribution plan is that the first provides insurance against uncertain lifetimes while the second does not provide such insurance necessarily (See Fuster (1999a), (1999b) and (2000)). One issue in pension system design that bears on this is whether there should be constraints placed on the choice of the form of benefits. In particular, should all agents be required to take their benefits in the form of annuities or should this be left to the choice of the beneficiaries? In its Personal Security Accounts (PSA) proposal of the U.S. Advisory Council on Social Security, for example, places no restriction on the form of benefits receipt. In contrast, in the Chilean system part of benefits must be annuities and there is a limit to the size of monthly withdrawals of any part of the retirement amount in other form. Further work is need to better understand the consequences of these design differences in light of the possible adverse selection problems and the possibility of bequests.

2. Horizontal equity, integration and political economy

There is wide-spread recognition that the system of mandatory PAYG pensions has important distributive implications (See, Aaron and Shoven (1999) and Atkinson (1996)). Achieving equitable outcomes may justify the tolerance of distortions of incentives. Changing the balance between PAYG, defined benefits pensions and investment-based retirements raises the question whether the distributional objectives of the system are being compromised. In particular, many expect the extent of redistribution (both intra- and intergenerational) in national pension systems to decrease if funded, defined contributions pensions take greater prominence.

A significant part of the debate on retirement systems has been concentrating on the issue of retirement age. There exists a theoretical literature on retirement decision which focus on different aspects: disability contingent retirement rules, long-term labor contracts encompassing retirement rules and the implicit inducement to retirement of existing public and also private pension plans. (See, Cremer and Pestieau (1996) and (1997) and references therein.) This literature is positive; it tries to explain retirement behavior and to explain the observed evolutions in retirement practice. There is a need to expand these studies in several directions.

• To examine whether or not an optimal-tax-cum retirement policy of the government calls for distortions (bias) in the individual retirement decisions. In other words, does there exist a welfare economics justification for distorting retirement decisions? If yes, in which direction should the bias go? What policy towards social security benefits, payroll taxation and retirement age ought a utilitarian social planner to pursue with heterogeneous individuals differing in two unobservable characteristics: their level of productivity and their health status? How is such an optimal policy modified when life expectancy increases and when there are legal constraints such as a mandatory uniform retirement age?
  
  
• To study the role of retirement age when the political process is accounted for. This requires studying the political economy of PAYG pension systems when retirement decisions are taken into account. In particular, one needs to consider a vote over payroll taxes, when retirement decisions are endogenous. This should be analysed under two kinds of pensions systems. The first ones are called neutral systems, in the sense that they do not distort retirement decisions with respect to the case where no pension system exists. In the opposite case, the pension system is said to be biased. Of particular interest are the systems that implicitly tax labor earnings and therefore provide incentives to retire earlier. We shall study the voting equilibrium (if any) for a given bias. Then we shall investigate the impact of the introduction of a bias on equilibrium level of retirement benefits and on welfare.
  
  
• Given results on these questions further issues immediately arise. The first one consists in determining the issue of the voting procedure when the decision variable is not longer the payroll tax rate but the retirement age. A second direction is the political determination of PAYG systems in the context of a demographic transition. This issue is crucial with respect to the current situation faced by most developed countries.
  
  
• A different but complementary set of questions concerns the political economy of retirement insurance when individuals differ in income and in risk. The issue here is to study the political economy of retirement programs in an environment where individuals differ in risk and in income.
  
  
• Distributive effects of programs affect political support for them (See Casammata et.al. (1999)). For example, contributions may or may not vary with individuals income, while benefits may be linked to contributions (positively or negatively, in which case one often speaks of means-testing) or be the same for everybody. There is a need to study how all intra- and inter-generational redistribution within pension systems affects the political support for retirement programmes.

3. Pension reform and the development, efficiency and regulation of investment based pensions

The analysis surveyed in section 2.1 shows the sensitivity of behaviour and welfare to the financial markets environment in which a funded system operates. Finance theory shows some financial markets may not exist as the result of agency problems giving rise to adverse selection or moral hazard. (See Freixas and Rochet, Microeconomics of Banking). Furthermore, missing markets may the historical remnants from a past when financial markets were severely constrained by regulations. Despite important efforts to modernise, European financial markets remain underdeveloped in many respects. Consequently, there are open questions as to the appropriate pace of transition toward reliance on funded pensions.

There are both theoretical and historical reasons for thinking that markets would develop to satisfy at least some of the needs created by the growth of funded pensions. (See Mason, et al (1995), Tufano (1989), Bodie (1990)). Our network will explore these issues within the European context. Here a number of factors converge to make the situation particularly difficult. First there is likely to be strong demand for annuities with the growth of funded pensions. Constructing these annuities will require a large and liquid fixed income market. However, the growth of public sector debt is being reigned in by the constraints of monetary union. Furthermore, the corporate bond market in Europe, especially on the continent, is still quite small (See, R. Lee (2001)). Finally, the heterogeneous nature of the corporate bond market means that illiquidity is a considerable problem (see Ericsson and Renault (2001) for a theoretical analysis and Anderson and Sundaresan (2000) for empirical evidence).

Even if markets develop to be relatively complete ones, the problems of moral hazard or adverse selection may be obstacles to their efficient operation. The network will examine these issues within the specific context of funded pensions. Pension funds make investment decisions that have potentially enormous effects on their beneficiaries. Are the implicit and explicit incentives of fund managers aligned with the preferences of fund beneficiaries? If not, in what way can the behaviour of the managers hurt the beneficiaries? Can institutional changes correct these problems? Empirical work on mutual funds (e.g. Chevalier and Ellison (1997)) has identified systematic deviations from optimal portfolio management due to their incentive structure. Theoretical work, starting with Bhattacharya and Pfleiderer (1985), has looked at the agency problem between the investor and the manager. Prat and Palomino (2000) have studied the incentives for managers to undertake investments that are excessively prudent or excessively risky. Members of the network will build on this research to look at delegated portfolio management in the specific context of pension schemes, taking into account such important features as

1. extremely long-term horizons,
   
   
2. the need for minimal contractual payouts, and
   
   
3. the existence of multiple layers of control that impede effective monitoring.

In particular, it will address the incentives for risk taking and churning (portfolio turnover that is not justified by information on fundamentals) that would result from alternative contractual and market criteria for evaluating and rewarding fund managers.

Theoretical analysis of risk taking incentives by fund managers highlights the importance of defining clear, risk-adjusted performance standards. However, despite the development in investment management techniques, designing such standards is still not a settled issue. (See Blake and Timmerman (2000)). The researchers of the network will pursue this issue of benchmark definition and performance measurement to study the evidence of excessive trading or of variations of style in pension fund management.

One plausible scenario for pension reform is that Continental Europe will witness a gradual expansion in the funded component of social security. This suggests that it is useful to study the experience of the UK, where the funded component has already been operating for some time. The UK experience has been scrutinised in the Myners Report (2001) which raises many policy concerns about their fund operations. The activity of pension funds is of public policy interest for several reasons:

1. The management of the savings of millions depends on the internal governance of pension funds and on their investment policies.
   
   
2. Pension funds' investment policies also affect the performance of the companies in which they hold stakes through their monitoring of management (Smith, 1996). Views on the advisability of pension fund activism vary considerably (Hutton, 1995).
   
   
3. The increasing weight of pension funds will change the nature of trading activity. The outcome will depend upon whether pension funds are net suppliers or consumers of liquidity. (Sciubba 2001).

A large body of literature analyses the performance of US pension funds and, more recently, that of UK pension funds. The major result emerging from this literature is that in both countries "pension fund managers typically under perform external benchmarks that represent feasible passive investment vehicles" (Blake, Lehmann and Timmermann, 1998). This is a reason for concern. The Myners Report suggests that benchmarking is widely used in the UK, and that this induces pension fund managers to herd, and "makes meaningful active management near impossible."

Network members will pursue these issues with a work programme directed at understanding the investment and trading policies of pension funds, their mutual interactions in the marketplace ("herding"), and their effects on market liquidity and volatility.

In light of the difficulties of obtaining correct incentives for risk sharing through contracts, the questions naturally arises whether public regulations can help. Pension funds are typically regulated as to the portfolios they can select. However, the regulation vary widely across different national systems and there is little agreement among academics as to which regulatory regime is best. (See, Davis (2001)). This work should be advanced by the more explicit consideration of portfolio decision-making under alternative regulatory regimes. Similarly, in the face of possibly inappropriate incentives for risk taking by providers of annuity pensions, the issue of solvency regulation arises. (See Bodie et al. (1995)) These issues need to be examined within the context of the evolving capital standards for financial intermediaries.

Click here for References

^