How should pension liabilities be valued? Risk aversion and demographic uncertainty

09:00-09:05 Welcome by the Royal Society and lead organiser

09:05-09:45 The UK State Pension age review and demographic uncertainty

Colin Wilson, Government Actuary's Department, UK

Abstract

The 2017 review of State Pension age included an independent report (the "Cridland report") and projections of life expectancy by the Government Actuary. A particular feature of the latter was an indication of alternative scenarios reflecting some of the uncertainty in future longevity. Since these reports an apparent slowdown in longevity improvements seems to have crystallised, highlighting the importance of acknowledging the degree of uncertainty inherent in longevity projections. This session explains how the review allowed for uncertainty and some possible implications.

09:45-10:30 Mortality Estimation and Forecasting: Models, Methods and Issues

Professor Jon Forster, University of Southampton, UK

Abstract

The Royal Society has a long association with mortality estimation dating back to the original development of the subject by John Graunt, presented to the Society in the seventeenth century. Mortality forecasting, an essential component in the valuation of pension liabilities, has a much more recent heritage. Modern forecasting methods are typically based on stochastic models and allow realistic quantification of uncertainty and, where appropriate, the incorporation of expert opinion. This talk will present a selective review of model-based mortality forecasting approaches. Key issues in mortality forecasting will be considered, including cohort variation, smoothing, sparse data and the potential for borrowing strength by jointly modelling (sub)populations. In light of recent UK data exhibiting relatively weak mortality improvements, the question of how reactive forecasts should be to apparent shifts in observed mortality experience will also be discussed.

10:30-11:00 Coffee

11:00-11:45 Quantitative easing impacts - what may be the endgame?

Baroness Ros Altmann CBE, House of Lords, UK

Abstract

For the past ten years, long-term bond yields have remained exceptionally low, which has significantly impacted pensions. Pension liabilities have been artificially inflated by unconventional central bank ‘quantitative easing' (QE) policies, leading to rising deficits and increasing annuity costs. QE has focussed on driving long rates lower, by purchasing sovereign and other bonds with newly created money. This unprecedented monetary policy experiment has had a number of side-effects and the ultimate impacts are not yet known. Having bought significant proportions of fixed income issuance in order to provide artificial economic stimulus, what next? How might Quantitative Tightening play out in markets and what does this mean for pensions?

11:45-12:30 Intergenerational fairness: statistical analysis, assumptions & actuarial models

Professor Jane Hutton, University of Warwick, UK

Abstract

Intergenerational fairness is an important and ancient concept in society and government. With respect to pensions, intergenerational fairness as well as fairness to women of a certain age have been debated when considering changes in state pension age.

Funded defined benefit schemes which are estimated to have deficits are required to impose deficit recovery payments or change benefits. Current members pay not only for their own future pensions, but also for their predecessors' (and their own) accrued entitlements. It is often assumed that this means younger people paying for older people. However, older generations' pensions contributions have provided productive capital investment and infrastructure used by all ages. Strict intergenerational 'fairness' within a scheme might neglect wider social balance.

Actuarial models require assumptions in order to estimate assets, liabilities, life expectancy and other demographic factors. Multiple assumptions biased away from statistically valid estimates can substantially increase an estimated deficit. Consequences of these assumptions affect not only the particular scheme's stakeholders, but wider society. Money used for deficit recovery payments is diverted away from business investment and dividends. A large estimated deficit can bankrupt a company, and put many people out of work. If pension contributions are tax-exempt, the government's income is reduced.

Assumptions underlying actuarial models are not merely economic or statistical.

13:30-14:15 Inflation guarantees in pensions

Mr Andrew D Smith, University College Dublin, Ireland

Abstract

Most UK defined benefit schemes provide retirement benefits linked to inflation, with the link subject to maximum or minimum levels of increase. It may be required to report fair valuations of these liabilities allowing for the embedded inflation options. Investments (in conventional and index-linked bonds) may be informed by a sensitivity analysis, re-expressing liability cash flows into equivalent real and nominal components. The results of these calculations depend on many things: inflation assumptions, volatility assumptions, the valuation model used, sensitivity definitions especially near fixing dates, demographic assumptions, use of model points and other simplifying demographic assumptions. In this session, Andrew shows alternative modelling approaches, the differences in reported values and sensitivities, with explanations for these.

14:15-15:00 The macro-economic context for pensions: monetary policy and financial stability

Professor David Miles CBE, Imperial College London, UK

Abstract

Low yields on government bonds – particularly on indexed bonds – mean that estimates of pension deficits that depend on bond prices have been driven up. Are such low yields likely to persist and does it make sense to evaluate the sustainability and the risks that DB pension schemes run by reference to them? This talk will consider these issues and reflect on how concerns on financial and fiscal stability mean that the legitimate concerns of regulators will be particularly relevant.

15:00-15:30 Tea

15:30-16:15 Estimation of mortality in South Africa: challenges and methods

Emeritus Professor Rob Dorrington, University of Cape Town, South Africa

Abstract

Researchers or consultants new to the estimation of mortality rates and life tables tend not to appreciate that there is a need for life tables to be specific to the task at hand, and depending on this task, there are different data requirements and problems with the estimation of mortality. Researchers or consultants familiar with the use of, or even with some knowledge of, the derivation of life tables/mortality rates in a developed country setting are often unaware of the paucity and poor quality of data that are available, or of the methods required to produce best estimates, should they require these for work in developing (particularly African) countries. 

This presentation covers these issues by presenting, within the context of both national and insured population life tables in African countries in general, the availability of and quality of data; the methods for producing population (national and sub-national) mortality estimates when data are deficient; and what industry standards and national life tables have been produced in South Africa.

16:15-17:00 Panel discussion

09:00-09:45 The Affordability of the UK State Pension

Paul Sweeting, Hassana Investment Company, Saudi Arabia

Abstract

For some years the affordability of the UK State Pension has been an issue. And for some years, the most visible answer to this issue has been to raise the State Pension Age. However, increasing the State Pension Age has a much greater impact on the least well-off - and they are the ones who are most dependent on this benefit. Just as importantly, increasing the State Pension Age is unlikely to control the cost of State Pensions in the long run. In this session, Paul Sweeting investigates the concept of means-testing instead of using the State Pension Age alone to control costs. Means testing for the State Pension would not be unprecedented. In fact, when the Old Age Pension was introduced in 1909, it was brought in as a means-tested benefit. Furthermore, it is already used in Australia for state pension benefits there. Paul Sweeting demonstrates that using means-testing, whether alone or in conjunction with increases to the State Pension Age, could help to control the cost of the UK State Pension. He also argues that it could do so equitably, ensuring that those who most need this benefit are more likely to receive it.

09:45-10:30 Optimal design: From insurance policy to economic policy

Professor Mogens Steffensen, University of Copenhagen, Denmark

Abstract

We discuss how utility optimization can help to shape improved product design that appropriately unveils and takes into account the demand for life insurance among young individuals and the demand for annuities among old individuals. Classes of objectives lead to classes of optimal life-cycle profiles of consumption, investment, insurance and annuitization decisions that can help pension funds to better target the needs of an individual, both in the savings and the payout phase. To some extent these profiles can also help politicians make an improved design of the pension system as such. We discuss some of the homogeneous issues and dilemmas that turn up in pension policy making in different countries with heterogeneous pension systems. Can utility optimization really help, not only in insurance policy making but also in economic policy making?

10:30-11:00 Coffee

11:00-11:45 Monetary risk and prudence in pension fund valuation

Professor Saul Jacka, University of Warwick, UK

Abstract

Prudence is the first of the cardinal virtues of Plato (and then of Aquinas) and much misunderstood in modern times. It has come to mean something akin to 'caution' or even 'pusillanimity' whereas its original meaning is closer to 'wisdom'. We will argue that the original meaning is closer to what is required in the context of pension valuation. In particular, that valuation must have regard to facts and experience. This talk will attempt to rehearse the sources of the 'modern' prudential approach to pension fund valuation and explain why they are inadequate and often overly-pessimistic. In 1999, Artzner, Delbaen, Eber and Heath introduced the concept of a Coherent Risk Measure (CRM), in part to address problems with the popular measure of monetary risk - Value at Risk. CRMs are characterised by the three properties: cash-invariance, subadditivity, positive homogeneity/scaling. These are properties that any (prudent) valuation method has. Any CRM can be represented by probabilistic scenario analysis i.e. as the worst calculated mean under a range of probabilistic scenarios. The fundamental flaw with such measures is that they usually fail to be temporally consistent. This talk will give some discussions in terms of coherent (monetary) measures of risk and explain in principle how to address and how not to address the failure of time-consistency in these measures.

11:45-12:30 Defined benefit or defined contribution schemes: good or bad

Dr Con Keating, Brighton Rock Group

Abstract

This presentation will consider, from first principles, the appropriate valuation method for pension liabilities and the implications of this for scheme management concepts such as prudence and structural design elements such as “buffers”. In this light it will suggest that risk management using stochastic modelling and actions based upon this technique, such as “self-sufficiency” and funding to buy-out levels, are largely redundant. The relative efficiencies of the three major forms of scheme design will be discussed.

13:30-14:15 Actuarial valuation and risk management under P and Q

Professor Alexander J. McNeil, University of York, UK

Abstract

The value of complex insurance and pension liabilities, and of the assets that back them, depends on a number of risk factors, which may be financial, demographic and behavioural. Modern approaches to point-in-time valuation and to projection of surpluses and deficits typically blend market-consistent techniques (under the risk-neutral measure Q) with real-world statistical considerations (under the physical measure P). The subtle interplay of these measures make valuation and projection conceptually and practically challenging tasks.

This talk will present an overarching framework for these issues and consider the role and state-of-the-art of computational solutions, such as proxy models, least-squares Monte Carlo simulations and recent ideas from artificial intelligence.

14:15-15:00 How can people share their longevity risks?

Dr Catherine Donnelly, Heriot-Watt University, UK

Abstract

Modern tontines offer retirees the opportunity for a life-long, reasonable income by sharing their longevity risks. They can make cost-efficient use of pension savings.  While this can also be done in a life annuity, the annuitant is generally forced to buy the guarantees implicit in a life annuity contract. In contracts, tontines can be structured with and without guarantees.

We show one method for structuring tontines. The different methods have implications for the flexibility and choice for the members of the tontine.

This talk is based on joint work with Thomas Bernhardt, Montserrat Guillen, Jens Perch Nielsen and John Young.

15:00-15:30 Tea

15:30-16:15 Pension fund governance: fiduciary duties of trustees

Dr Anna Tilba, Durham University, UK

Abstract

UK Pension fund trustees are legally bound by common law fiduciary duty, which requires trustees to proceed with prudence and to act in the best interests of the beneficiaries when they make strategic investment decisions. The term ‘fiduciary duty’ has been receiving increasing attention by policy makers. More recently, fiduciary duty has been used as a tool of governance and investor stewardship. John Kay has encouraged institutional investors such as pension funds to observe ‘fiduciary standards in their relationships with their clients and customers’. However, our understanding of how this central concept shapes pension fund trustees’ approaches to investment is still lacking in clarity both in academic research and practice. As a concept ‘fiduciary duty’ is highly flexible, loose and uncertain and it seems to be understood and enacted by practitioners in a variety of ways. No studies have yet empirically investigated how pension fund trustees interpret and apply the concept of fiduciary duties in their investment practices. In this talk, I will share over a decade of my knowledge on this topic. I will discuss the plurality of trustee interpretations of the concept of ‘fiduciary duty’ and how these interpretations shape trustees’ understanding of the purpose of the pension fund and their subsequent investment strategies. I will also share my insights into behavioural biases that may be present in trustee decision-making and how that may be affecting the quality of their investment decision-making.

16:15-17:00 Panel discussion