The credit contraction has focused the attention of the powers that be on the levels of risk being run by the UK’s pension schemes. The Pension Protection Fund (PPF) in particular has expressed disappointment at the modest take-up of liability-driven investment strategies, and appears prepared to use its levy to kick-start de-risking.
“There are more and more speed cameras, road humps and penalties being imposed,” warns Dawid Konotey-Ahulu, partner & co-principal at Redington Partners, which advises on management of scheme-wide risks. “You’ll have to change the way you drive, because if you’re caught doing 40mph in a 30mph limit now it won’t be just three points and a £30 fine – they’re going to take your car away and crush it.”
As schemes do begin to transfer out more of their interest rate, inflation and equity market exposures, longevity makes up a greater proportion of overall risk. Redington Partners calculates that a scheme with 55 per cent of its assets in equities and no interest rate or inflation rate hedges sees about 0.7 per cent of its value-at-risk (VaR) coming from longevity risk; put a 100 per cent hedge on and weight equities at zero, and that shoots up to nearly 4 per cent.
Massive exposure
At that point, the sheer scale of the issue should be clear. Pension funds have the market’s biggest exposure to longevity – bigger even than insurance companies’.
The best estimates currently put longevity-related liabilities of UK pension funds at about £2000bn (€2,500bn), and according to the Pensions Regulator, each extra year of life expectancy at age 65 adds 3-4 per cent overall liabilities.
“Even if your fund managers generate good performance, if life expectancy increases more than you expect you get a pension deficit you hadn’t planned for,” says Professor David Blake of the Cass Business School.
Prof Blake’s recent research has been into the stochastic nature of mortality. It can make depressing reading – and if anything the stochastics are more depressing than the mortality. If we all shared the curse of the Struldbruggs in Gulliver’s Travels we may well not have a pensions system – but at least we would know where we stood. Instead, we are faced with “massive uncertainty about how long people are going to live in the future”.
Otto von Bismarck designed the first pension system in 1889, to pay annuities to railway workers when they reached 70. This was a great deal for the Reich, with the average German citizen expecting to die aged 45.
“Life-expectancy has improved linearly, by 15 minutes per hour, between 1840 and the present day,” says Prof Blake. “At that rate, your grandchildren could live beyond 150.”
The optimists assume that medical breakthroughs will continue, and ask why we should not all live like the Struldbruggs (official statistics have always predicted a levelling-off, and have always been wrong). But the pessimists point to things like the obesity epidemic and global warming to support the common-sense view that there has to be a limit somewhere.
Furthermore, within these broad population trends, it is well-known that gender, social class, year of birth and location are all correlated with life expectancy.
All this complexity is reflected in the tables used to project longevity, created by the Continuous Mortality Investigation bureau (CMI) of the Faculty and Institute of Actuaries. There are now three different sets of tables (and if a pension scheme has a statistically-significant membership it can generate a proprietary table), which are themselves subject to variation thanks to the application of two “reduction factors” – the “cohort effect” and the “underpin”.
The cohort effect derives from the fact that the “cohort” of people born around 1930 enjoyed a much greater improvement in mortality than the preceding or following generations. As they grow old this cohort moves like a wave through the pensions system – particularly hurting schemes with a high proportion of members born in the 1930s. Scheme actuaries have to decide how long it will be before the 1930s cohort dies out, and the CMI came up with three options - the short, medium and long cohorts – which assume this will happen by 2010, 2020 and 2040, respectively.
The underpin was introduced as a floor on mortality rates, accommodating the optimists’ scenario that we could all live like the Struldbruggs: a 1 per cent underpin represents a built-in assumption that mortality rates will keep reducing at 1 per cent annually.
Saps
In February, the Pensions Regulator warned that schemes submitting Recovery Plans using no underpin or the short or medium cohort would face “further scrutiny”. Several consultants despaired at the billions these assumptions would add to liabilities. But there was good news, too, in the shape of the new set of base tables – “Saps 03” – which the CMI published around the same time.
Saps stands for Self-Administered Pension Schemes: for the first time, the mortality data was taken from pension schemes rather than the insurance industry, as it was for the “92” and “00” series. Moreover, whereas the 92 series offered just two tables (one per gender) and the 00 series 14, Saps 03 has 20 tables split by gender, health status, wealth and marital status.
“People who have insurance policies tend to live longer, so using Saps 03 could actually lower your longevity projections - even after you have adopted the long cohort and a 1 per cent underpin,” observes Mr Gardner (see graph).
Once you have got to grips with your risk, what can you do to transfer them out of your scheme? As pressure builds to adopt the most stringent liability projections, buyout solutions look increasingly attractive. But longevity risk may be the stumbling block to getting a good value deal in this market.
“The more actives and deferreds you have, the less likely the buyout providers are to give a quote – and they will try to pass the tail-risk back to you,” warns Mr Gardner.
Fan charts of longevity risk demonstrate that uncertainty of survivorship increases markedly after age 75, reaching a maximum in the early 90s before falling off. In other words, the proportion who will survive to their late 80s and early 90s is highly uncertain, what Prof Blake calls a “toxic tail” for annuity providers.
Sure enough, the £800m buyout of the P&O scheme last December saw Paternoster happily buying-out 11,000 pensioners while leaving actives and deferreds with P&O.
Another solution lies in the potential development of a secondary market in longevity, facilitating synthetic longevity solutions that mirror the interest rate and inflation swap markets. For this, we need longevity indices, and the three leading ones are the Credit Suisse Longevity Index, launched in 2005; the Lifemetrics Index launched by JPMorgan in partnership with the Pensions Institute and Watson Wyatt in 2007; and, most recently, the Goldman Sachs QxX - Life Settlements Index.
“We need standardisation to generate liquidity, and it was to meet those needs that we started to publish LifeMetrics about a year ago,” says Guy Coughlan, global head of LifeMetrics and ALM advisory at JPMorgan. “With an index like this you can start thinking about a framework for transferring the longevity risk out of pension plans, with a hedge provider like an investment bank or insurer acting as a broker for your longevity and a credit intermediary. The cost of taking out a hedge like that would be around 1.5 per cent of the accounted liability – although costs will depend crucially on the profile of the membership and the nature of the benefits that are guaranteed.”
Messy factors
One might question how good a match of scheme-specific risk can be achieved with index-based derivatives. What about the effects of all those messy factors – gender, location, social class, cohort? An insurance-style solution, though costly and inflexible, offers a 100 per cent risk transfer. An index-based solution, more of a risk management paradigm, does not. But hedge effectiveness is not simply about how realised mortality rates among scheme members move relative to the national population, Mr Coughlan reminds us – a hedge does not need to transfer 100 per cent of your risk to be worthwhile.
“What matters is how much the hedge reduces the uncertainty around the risk of monetary loss,” he says. “We’ve found that that uncertainty can be reduced by 85-90 per cent using a standardised indexed approach, calibrated for scheme-specific risks.”
JPMorgan facilitates this calibration by maintaining a number of indices covering males and females in different age ranges across different countries (a growing list currently including England and Wales, the Netherlands, Germany and the US). Derivatives on these indices (called q-Forwards) work so that the pension scheme receives a fixed mortality rate in return for a realised mortality rate, and they can be combined to approximate scheme-specific risk.
“With the same set of building blocks you can hedge a pension plan of blue-collar or white-collar workers, workers in the North or the South, and so on,” says Mr Coughlan. “You establish the cashflows and sensitivities with your actuary. How will those cashflows change if mortality rates fall faster than expected? How will they change if it’s just male mortality that falls faster than expected? Or just female 65 year-olds’ rates that fall faster? Then you construct a q-Forwards portfolio to mirror those sensitivities.”
LifeMetrics and q-Forwards got an important vote of confidence in February, when buyout specialist Lucida entered into a derivatives deal with JPMorgan – the first of its kind involving the insurance industry.
The q-Forwards only have 10-year tenors, so they do not offer much in terms of hedging actives and deferreds. And although existing investors in insurance-linked securities like hurricane and mortality-catastrophe bonds will be glad of a way to fill the longevity-shaped gap in their portfolios, it may take a while for the broader investment community to come in and provide the demand for all that potential volume.
“It’s not clear that private-sector investors would be willing to take on that toxic tail risk,” adds Prof Blake, who has been lobbying for a government-led solution for this part of the market.
So the synthetic market in its current forms does not address all the issues – but we are still at an early stage, and important progress is clearly being made in a sphere of the pension industry which can only increase in importance.
