I prefer true but imperfect knowledge, even if it leaves much indetermined and unpredictable, to a pretence of exact knowledge that is likely to be false. The credit which the apparent conformity with recognized scientific standards can gain for seemingly simple but false theories may, as the present instance shows, have grave consequences.Friedrich von Hayek, Prize lecture in Stockholm on receiving the 1974 Nobel Prize in Economics.
1974 was also the year I thought about pensions for the first time, although, like most people in their twenties, not for long. The long-standing provision of defined contribution pensions for university staff was replaced by a defined benefit scheme, USS. This was an era in which most large employers and many small ones were making similar moves. After a long political debate, Barbara Castle presided over the bipartisan introduction of an earnings-related tier to the state pension scheme, with provision for good occupational schemes to “contract out”.
Following the establishment of the ‘cradle to grave’ welfare state in the decades immediately after the Second World War, this extension of defined benefit occupational pension schemes was perhaps the greatest success of British social welfare policy. The system was not without problems. I recall in the 1980s helping Sir David Walker organise a seminar at the Bank of England to discuss issues in the governance of occupational pensions and their relationship to corporate balance sheets. But as schemes matured, the beneficial results of the growth of occupational pensions became apparent. In 1995, more than 60% of pensioners had below median incomes. By 2010 the distribution of pensioner incomes was in line with that of the working population, as the architects of defined benefit schemes had envisaged. Old age was no longer a major cause of poverty.
But then the foundations of defined benefit provision began to crumble. Famously, the scale of Robert Maxwell’s theft from Mirror Group Newspapers pension funds was revealed when he was found dead in the sea near Madeira in 1991. It became evident, as we had recognised at that Bank of England seminar, that more needed to be done to promote the security of pension promises and this led to the Goode Report in 1993. Less dramatically, in 1997 Exley, Mehta and Smith delivered a seminal address which introduced a generation of actuaries to modern financial economics. I had once been startled to find that actuaries knew little of finance theory – soon I would come to think they knew too much or at least took it too seriously. Accounting practice was shifting from historic cost to ‘fair value’ assessment of balance sheet values, with possible implications for the treatment of pension liabilities, which became a reality with the adoption of new accounting standards – in particular, FRS 17 in 2000 and IAS 19 in 2001. The outcome of these events was a stream – then a torrent, then a flood – of fresh complexity and regulatory intervention.
There were several components to the economic contribution to actuarial analysis. It had long been recognised that money in the future is worth less than money today. This idea was systematised in the calculation of discounted cash flow (DCF) valuations, which enabled a present value to be attached to a stream of prospective cash flows. Subjective expected utility (SEU) valued uncertain returns by reference to their variance and the ‘risk aversion’ of an agent, with such ‘risk aversion’ deduced from the second derivative of the function relating utility to income or wealth. Modern portfolio theory (MPT) and the capital asset pricing model (CAPM) emphasise that the uncertainty attaching to an investment portfolio depends not just on the uncertainties (variances) associated with returns on individual assets but on the covariances between them. Finally, the efficient market hypothesis (EMH) expressed the idea that market prices reflected all publicly available information. (Importantly, the empirical observation that historical equity returns substantially exceeded levels readily explicable by these models, described in 1985 by Rajnish Mehra and Edward Prescott as ‘the equity premium puzzle’, still lacks good explanation within this theoretical framework.)
This analytic framework was extensively developed in the 1950s and 1960s and contributed to the transformation of the world of finance from one in which well-bred gentlemen exchanged inside information with each other over long alcoholic lunches, to the professional activities of today that sweep many of the most able students of mathematics into the City of London and Wall Street (or Greenwich, Connecticut). And Nobel Prizes were awarded to Harry Markowitz (MPT), William Sharpe (CAPM) and Eugene Fama (EMH) for their contributions; LJ Savage (SEU) had died in 1971 aged only 53.
These models have been indispensable tools in developing our understanding of financial markets. But Hayek’s 1974 lecture gave prescient warning of the dangers of their misapplication.
He observed “I regard it in fact as the great advantage of the mathematical technique that it allows us to describe, by means of algebraic equations, the general character of a pattern even where we are ignorant of the numerical values which will determine its particular manifestation. We could scarcely have achieved that comprehensive picture of the mutual interdependencies of the different events in a market without this algebraic technique. It has led to the illusion, however, that we can use this technique for the determination and prediction of the numerical values of those magnitudes; and this has led to a vain search for quantitative or numerical constants…”
He continued: “…compared with the precise predictions we have learnt to expect in the physical sciences, this sort of mere pattern predictions is a second best with which one does not like to have to be content. Yet the danger of which I want to warn is precisely the belief that in order to have a claim to be accepted as scientific it is necessary to achieve more. This way lies charlatanism and worse. To act on the belief that we possess the knowledge and the power which enable us to shape the processes of society entirely to our liking, knowledge which in fact we do not possess, is likely to make us do much harm.”
‘Much harm’ of this kind was done in the events that led to the Global Financial Crisis of 2008. But this paper is concerned with the harm done by ‘the pretence of knowledge’ in the evolution of pensions in Britain. There has rarely been a better illustration of the adage (ironically, attributed without evidence to Mark Twain) ‘It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so’.
The only good answer to a question such as ‘what is the cost of a pension promise maturing fifty years hence’ is ‘I do not know’. That is not to say that nothing is known and, for example, it is possible to describe some of the variables – inflation rates, mortality experience, etc. – which might be relevant to the answer. But the uncertainties that surround estimates of these things are sufficiently great that no numerical answer likely to be accurate even to one significant digit is available or conceivable.
If anyone doubted that limitation, then the fact that massively different answers to the valuation of the liabilities of the same pension funds were given in 2020 and 2023 should dispel such illusions. The liabilities had not changed much – what had changed was the estimated net present value of these liabilities.
Let me spell the point out further, if that is necessary. In the past two years the yield on the 2071 gilt has been as low as 0.5% and as high as 5.0%. Discounting a liability that falls due in 2071 at 0.5% reduces its cost by 20%; at 5% by more than 90%. There are people working today who will not even have reached retirement age in 2071 and who can realistically expect to live into the twenty-second century.
This does not mean that commitments to pay future pensions are virtually costless if interest rates are 5% and prohibitively expensive if they are 0.5%. Whatever pension arrangements are in place – pay-as-you-go (PAYG), defined benefit (DB), defined contribution (DC) or some hybrid of these – the bread that a retiree eats in 2071 is baked by someone working in 2071 who very likely has not yet been born. And who will need to be compensated for it in the currency of the time.
Pension provision is necessarily an inter-generational contract and the fairness of that implicit contract is a fundamental issue. Once put in terms of inter-generational equity, I find it difficult to avoid the conclusion that my generation has done very well for itself at the expense of its grandchildren.
There is a second issue of distributional fairness within generations. This conditions the arrangements that are put in place in 2023 and 2043 and 2063, in anticipation of the realisation in 2071 of commitments made at these various earlier dates. The manner in which these preemptive arrangements are made and the manner in which the related commitments are ultimately honoured is a matter of contemporary, and no doubt evolving, conceptions of fairness. These are questions of equity and social and economic organisation and cannot, as Hayek emphasised, be answered by any pseudo-scientific calculation.
I need to know
Sir Denys Henderson, a lawyer by training, was chairman of ICI from 1987 to 1995. He accepted the honorific role of President of the Society of Business Economists and arranged a meeting so he could express his frustration at the inability of economists to make sufficient contribution to business. He began with a critique of the failure of economists to predict successfully the turbulent economic events of the 1970s and 1980s, a critique which had considerable justification.
The Society had lined up two of its members to respond. Alan Budd, who served in numerous roles in public and private sectors, including Chief Economic Adviser at the Treasury, explained that economic systems are complex and non-linear and that while patterns may be identified, reliable prediction was impossible. I had recently taken a chair at London Business School and followed Alan with a talk suggesting that microeconomic analysis of firms and markets might be more useful to practical business people than macroeconomic forecasting. By the end of the evening, Henderson was almost tearing out (what was left of) his hair in frustration. “I need to know” he fumed, reiterating his demand for reliable prediction.
It was an event I would remember often. Not least as I observed subsequent developments at ICI, then Britain’s leading industrial company but now defunct. (A story I have told elsewhere.) But the recollection came more often when someone approached me, as I practised economic consultancy, to ask a question like ‘what will the dollar sterling exchange rate be ten years from now?’
My response – and I strongly believe the only appropriate response – was on the lines of ‘if you tell me why you are asking that, I will try to help you formulate a more sensible question which it may be possible to answer.’ But this reply was not often well received. Frequently the problem was that the individual concerned had been asked the question by their boss, who – like Sir Denys – had to be placated. Or they were building a spreadsheet and this number was required to fill a vacant cell. And the caller might observe that if I could not give them an answer, there was someone else – probably at an investment bank – who would, and ring off.
Frank Knight is often described as the founder of the Chicago School of Economics. The maxim of the physicist Lord Kelvin, that “where you cannot measure your knowledge is meagre and unsatisfactory,” is engraved on the Social Sciences Building of that university. Anticipating Hayek, Knight, who must have observed Kelvin’s assertion almost every day, wrote that “as applied in mental and social science, (it) is misleading and pernicious. This is another way of saying that these sciences are not science in the sense of physical science and cannot attempt to be such without forfeiting their proper nature and function. Insistence on a concretely quantitative economics means the use of statistics of physical magnitudes, whose economic meaning and significance is uncertain and dubious. (Even wheat is approximately homogeneous only if measured in economic terms.) And a similar statement would even apply more to other social sciences. In this field, the Kelvin dictum very largely means in practice, “if you cannot measure, measure anyhow!”
Although we must abandon ‘the pretence of knowledge’, individuals and institutions need to act in the face of uncertainties. The proper response is not to insist on receiving answers to unanswerable questions – ‘I need to know’ – but to reformulate the problem in ways that allow the preparation of information that relates directly to the issues faced by decision-makers. So ‘what is the net present value of all my existing pension commitments?’ is not a useful question. Not only is the answer not knowable – even to within an order of magnitude – but the answer has no obvious implications for sensible action.
A more appropriate approach is to ask ‘how can I make my pension arrangements resilient to an unpredictable future?’ This approach is equally relevant for individuals and pension providers. And it provides a basis for the beginnings of informed and thoughtful decision making.
Thinking Clearly about Uncertainty
In applying economic models to issues such as pension provision, we need to end the imprecise and interchangeable usage of the terms ‘risk’ and ‘uncertainty’ (along with other words such as ‘likely’ and ‘probable’ which are often thrown around in the same context). This linguistic imprecision is accompanied by a tendency to treat volatility of anticipated outcomes (typically measured as the variance of a distribution) as a measure of all of these things. Incredibly, this lack of rigour is if anything more common among professionals engaged in managing risk and uncertainty than it is in everyday usage.
In our recent book, Mervyn King and I attempted to clarify these issues. We began with history. A century ago, John Maynard Keynes and Frank Knight, in very distinct contributions, each emphasised the distinction between risk and uncertainty. For both these authors, ‘risk’ could be described probabilistically, but ‘uncertainty’ could not.
For Knight, “A measurable uncertainty, or ‘risk’ proper, as we shall use the term, is so far different from an unmeasurable one that it is not in effect an uncertainty at all.” Keynes wrote similarly ‘By “uncertain” knowledge, let me explain, I do not mean merely to distinguish what is known for certain from what is only probable. The game of roulette is not subject, in this sense, to uncertainty; nor is the prospect of a Victory bond being drawn. Or, again, the expectation of life is only slightly uncertain. Even the weather is only moderately uncertain. The sense in which I am using the term is that in which the prospect of a European war is uncertain, or the price of copper and the rate of interest twenty years hence, or the obsolescence of a new invention, or the position of private wealth-owners in the social system in 1970. About these matters there is no scientific basis on which to form any calculable probability whatever. We simply do not know.”
But these arguments were overtaken by the axiomatic approach to choice under uncertainty pioneered by John von Neumann and Oskar Morgenstern and extended by Milton Friedman and LJ Savage. This became the foundation of the financial economics developed from the 1950s. It adopted a subjective Bayesianism which claimed – in the words of Friedman – ‘we may treat people as if they attached probabilities to every conceivable event’.
But rational people do not attach probabilities to every conceivable event – they recognise that there are many relevant things they do not know, about which others are likely to know more than they know, and about which – like the matters listed above by Keynes – no one may know very much at all. They follow the advice attributed by Damon Runyon to Sky Masterson’s father in Guys and Dolls:
‘On the day when I left home to make my way in the world, my daddy took me to one side. “Son,” my daddy says to me, “I am sorry I am not able to bankroll you to a very large start, but not having the necessary lettuce to get you rolling, instead I’m going to stake you to some very valuable advice. One of these days in your travels, a guy is going to show you a brand new deck of cards on which the seal is not yet broken. Then this guy is going to offer to bet you that he can make the jack of spades jump out of this brand new deck of cards and squirt cider in your ear. But, son, do not accept this bet, because as sure as you stand there, you’re going to wind up with an ear full of cider.”’
An alternative approach
King and I believe it important to restore the distinction between risk and uncertainty, but in a somewhat different sense from that employed by Keynes and Knight. Uncertainty is the result of imperfect information: about the past, present and future. Uncertainties may be resolvablein the light of further information or where there exists some reasonably objective basis for constructing a probability distribution of outcomes. Other uncertainties which are not resolvable in these ways we describe as radical.
Uncertain outcomes may be better or worse than expectations, but in ordinary language, risks are always negative. No one has ever said ‘There is a risk that I might win the National Lottery.’ Or even ‘There is a risk that I might not win the National Lottery’ because rational individuals do not expect to win the National Lottery. We define risk as an event or combination of events which derails a realistic reference narrative.
In the context of pensions, for example, investment returns and life expectancy are uncertain. Life expectancy is in our language resolvable, or close to being so; capable of being described by a probability distribution derived from observed frequencies. Investment returns are radically uncertain – although there are well-documented historical series, there is little reason to believe they are generated by a specifiable underlying process – although some financial economists make that claim, as in the Fama-French three factor model. And investment returns are sensitive to geopolitical events which are necessarily radically uncertain – not only do we not know what will happen, but we also have little capacity even to describe the range of things that might happen.
A review of the experience of the last thirty years forces the conclusion that the most important source of radical uncertainty in pension planning is the likelihood of changes in the tax and regulatory regime. These changes have been large, predictable only in the very short term, and are incapable of probabilistic description. As is also true of changes in the assumptions used in actuarial valuations.
The meaning of risk is idiosyncratic
In the context of pensions, an appropriate reference narrative for the individual is that provision is sufficient to maintain one’s standard of living in retirement, and the associated riskis that planned provision for retirement will prove inadequate for that purpose. Thus in pension policy, risk is concerned with the adverse extremes of outcomes, while uncertainty arises from the variability and covariability of many of the components that taken together determine outcomes.
Let me take a very pertinent example. It is conventional wisdom that at my age my portfolio should contain a significant proportion of cash and bonds. This is what almost any financial advisor will recommend. Indeed because it is conventional wisdom, an advisor who recommended otherwise would encounter problems with indemnity insurers and the FCA. If I were content to allow a fund manager to determine my asset allocation, a ‘target date’ or ‘lifestyle’ fund would implement that strategy on my behalf. While the Pension Wise website stops just short of recommending this (or any) course of action, you will certainly come away with the impression that this is what you should do.
The reasoning is that as one gets older, one should shift away from ‘risky’ assets. The meaning of risk here appears to be historical short-term volatility of nominal asset prices – as it is for the highly misleading Key Information Documents required of retailers of packaged investment products. But such volatility is not the relevant meaning of risk for me, and I do not hold bonds and only modest amounts of cash. Leave aside for the moment a possible bequest motive – the amount and importance of that motive is clearly highly specific to the individual concerned. And for practical purposes we might treat my wife and myself as one, since the cost of maintaining our standard of living is not much affected by whether we are one or two. The risk that concerns me is that my resources are insufficient to maintain that joint standard of living for the rest of our joint lives.
Both life expectancy and investment returns are uncertain. But over my lifetime they are likely to be covariant. Over time, the Sharpe ratio – variance relative to expected return – diminishes. Put simply, if I only live five more years – which is less than my expectation – it really doesn’t matter what I invest in. If I live for twenty-five, which is at the upper end of realistic expectation, then expected return is much more important than short-term volatility. And each time I visit my doctor, I will reassess the relevant covariances in the light of his – medical – advice.
It is not hard to develop a simple illustrative numerical model to demonstrate these points. And that is typical of how economic models are useful in practice. In developing insight and understanding, in confirming or refuting intuitions, rather than in delivering predictions or offering spurious precision when the only correct answer is ‘I do not know’.
Harry Markowitz, recipient of the Nobel Prize for his pioneering work on Modern Portfolio Theory, was asked by a journalist how he had invested his own retirement funds. He responded:
“I should have computed the historical co-variances of the asset classes and drawn an efficient frontier. Instead, I visualized my grief if the stock market went way up and I wasn’t in it – or if it went way down and I was completely in it. My intention was to minimize my future regret. So I split my contributions 50/50 between bonds and equities.”
Behavioural finance has been based on observing the degree to which otherwise ‘rational’ individuals fail to implement the findings of finance theory. But we should not jump, as many do, to the conclusion that the fault lies with the individuals rather than the theory. Markowitz, better equipped to implement the findings of modern financial economics than any other person alive (sadly, he died in June 2023), failed to find the necessary middle ground between populating a spreadsheet with imagined numbers and completely ignoring the ideas behind the analysis.
Yet with his emphasis on regret, Markowitz had identified a real human emotion – and one which is often useful. People in retirement may regret the choices they made – they saved too little or too much, they invested foolishly or made bad career choices. Today, they may wish to avoid such future chagrin, and that thought influences their current actions. And individuals making decisions for others or within organisations are sensibly mindful about decisions that might lead to bad outcomes for which they might, reasonably or unreasonably, be blamed. In large bureaucracies, of course, this often leads to decision paralysis in which all initiatives are rejected for fear that some person involved in the procedure might be held responsible for their failure. But all decision processes must be framed within a social context and sensitive to it.
Managing risk and uncertainty
The key to managing risk and uncertainty is not to attempt to eliminate uncertainty. Uncertainty is unavoidable and in a sense welcome since it is the existence of uncertainty which gives rise to profit opportunities in investment and business and adds interest to our personal lives. We look with pity, not envy, at the person whose daily routine is repetitive, who always eats the same food and always holidays in the same place. Our objective should be to embrace uncertainties while managing risk through strategies that are both robust to events or combinations of events that might derail our reference narrative, and are also resilient by allowing opportunity to regroup and recover when risks materialise – as sometimes they will. Robustness is mainly a matter of portfolio construction, and resilience much influenced by relevant time scales.
For thirty years I was responsible for investment strategy in my Oxford College. St John’s is almost five hundred years old and relevant time scales are long. There are many lessons but I want to focus on one. As horizons extend, covariances tend to decline. If I consider – say – an apartment block in Berlin, an office in San Francisco, and a technology spinoff from university research, short-term asset values are correlated, possibly strongly. Valuations change daily, but similarly, as market participants speculate – what will the Fed do, has inflation peaked, will there be a recession? Over twenty years, no one cares about these things and over a century the anticipated covariance between the investment performances of these assets is likely to be close to zero.
This points to the overwhelming importance of diversification as a means of minimising risk – where risk means threats to a reference narrative which allows the College to spend gradually increasing amounts on its educational activities for the indefinite future. Minimising risk does not mean eschewing ‘risky’ assets (meaning assets with high variance of expected return, or even just assets that might decline in value) – I remember a former Bursar remarking that legal restrictions on trustee investments had cost widows and orphans far more than they had ever lost from corrupt or foolish trustees. Indeed an ideal College investment was one offering little or no income but which might yield a jackpot twenty-five years from now, because few other institutions would be interested in this sort of proposition. There are not many such opportunities but we found some and are currently benefiting from the realisation of one.
(Put more formally, what my former colleague meant was that a portfolio of investments each of which had low variance and low return but which were strongly correlated with each other, was likely to exhibit both low return and high risk. In 2008, it was realised that had been true of collections of mortgage-backed securities. And it is not difficult to think of many other examples. Including from the world of pensions.)
The overarching lesson is that decisions under uncertainty are by their nature unique. They require an understanding of the overall environment in which they are made, and cannot be entrusted to algorithms – or made by reference to models whose parameters are necessarily speculative. Judgment is indispensable. And the views of regulators or consultants who offer generic advice without regard to the specific objectives of the individual or institution concerned should be treated with great scepticism.
Pension Policy in Practice
What is meant by ‘risk’ is specific to an individual or organisation and to the circumstances and aspirations of that individual or organisation. Thus there is no objective meaning or measure of risk. We must always ask ‘what risk?’ and ‘for whom?’. When different people have different measures and conceptions of risk they can profitably trade with each other, as they have done in insurance markets for centuries. An Oxford College may acquire an asset with a long time horizon and speculative return on advantageous terms because mainstream investors are with good reason not interested in such propositions.
The risks arising from uncertainties can be diminished by pooling – as through insurance – and diversification – as in portfolio management. The fundamental fallacy of pension policy in the last three decades has been the fruitless attempt to eliminate uncertainty altogether rather than to manage uncertainties with the objective of mitigating the consequential risks through pooling and diversification.
So we created an insurance scheme in the Pension Protection Fund (PPF) – but then actually introduced a regulatory objective of minimising recourse to it, as if stationing a fire engine in front of every house were a good alternative to fire insurance. If we make the close proximity of fire engines a condition of eligibility for insurance, the result will not be that there are no fires; the outcome will be no fire engines and no fire insurance, and householders will have to deal with the risks of fire unaided. And this is roughly what has happened in the British pension system, as the risk of inadequate pension provision has been systematically transferred from employers and financial institutions to individuals, with long-term consequences we have yet to recognise.
If we further make immediate access to fire engines a condition of the validity of existing insurance policies, we might even find an army of consultants advising on whether the services were sufficiently close and extensive to satisfy the changing and increasingly demanding requirements of inspectors. Does any of this sound familiar to those in the world of pensions?
We actively discouraged diversification by advocating attempts to match specific assets to specific liabilities. The outcome was not just to cripple company pension schemes, mostly beyond repair, but to render the cost of buyouts daunting and the purchase of annuities unattractive.
And not only did we raise the cost and diminish the security of pensions through these policies, but we were also careless as to the implications for the wider economy, as pension funds substantially redirected their portfolios away from investment in business and in real assets and towards bonds. In this, we neglected the hundred-year-old insights of Knight and Keynes, who both recognised that embracing uncertainty was key to the dynamism of the market economy.
‘If the animal spirits are dimmed and the spontaneous optimism falters, leaving us to depend on nothing but the mathematical expectation, enterprise will fade and die.’
The collapse of private sector defined benefit pensions may constitute the most serious avoidable policy disaster in Britain this century. We have damaged the retirement prospects of a generation and the growth potential of the British economy. We need to start again with a different philosophical approach.
Beware of his false knowledge: it is more dangerous than ignorance.G. B. Shaw, Man and Superman, 1904