For David Blackwood, group treasurer of ICI, the moment of realisation came in 2001 when the deficit of the UK chemicals company's pension and health-care arrangements reached £500 million under the FRS17 accounting standard. At the time the deficit was seen as little more than an accounting quirk by ICI's board and shareholders, and Blackwood recalls that a trustee joked: "Is ICI good for that?" Joking aside, the trustee was speaking the language of a creditor.
For Jelles van As, chief investment officer for the EUR5 billion pension scheme of Dutch steel company Hoogovens, now part of the Corus group, the moment came after the market turmoil of 1998. "We learned that our solvency was not as stable as we expected," he recalls. "We had to increase contribution rates and we had to reduce indexation to Netherlands consumer inflation. We were really surprised by the volatility in equity markets and how heavily the surplus depended on equity markets."
For Graham Aslet, former chief actuary of Friends Provident, the UK life insurer, the moment came in early 2003 when his finance department showed him and the trustees of the £600 million group pension scheme a derivatives sales pitch from Merrill Lynch. The pitch suggested that the scheme deficit, then in single figures, was dangerously vulnerable to correlated moves in UK interest rates and inflation.
"From the trustees' point of view," says Aslet, "they believed firmly in the advantages of final salary pension schemes and wanted the company to continue to support that scheme. This scenario could increase the deficit by hundreds of millions, which would have made people seriously scared. That might prompt the employer to minimise the risk of things getting worse by closing the scheme to new members."
The moments of realisation may have been very different in their timing and nature. But the end result was the same. The managers of these three pension schemes, and a select few others in the UK and Netherlands, would end up wholeheartedly embracing liability-driven investment (LDI). Either directly or indirectly, they would become significant users of derivatives, and arguably would reach an enviable position of risk reduction.
Reaching this position was not easy, intellectually as much as physically, and it helped to have a head start wherever possible. As early as 1997, following research into asset-liability modelling, Van As persuaded Hoogovens to adopt a policy based on actuarial solvency, or the ratio of market value of assets to liabilities discounted at a fixed 4% rate. "We based the level of contributions, indexation and investment risk on this solvency ratio," he says.
Although not a fully market-consistent measure, Van As believes it gave him a head start in moving to fully market-based methods later on. "It helped that we had already started using the policy table based on solvency. Other pension plans were somewhat slower in recognising the volatility of the actuarial solvency situation."
While the 40,000 member Hoogovens scheme is split roughly equally between its active, deferred and retired members, the main ICI scheme - with nearly £7 billion of liabilities - epitomises a mature, closed scheme, with only 1,000 active members out of a total of 75,000. It represents nearly 85% of the group's FRS17 liabilities. Saddled with liabilities that could only increase for the foreseeable future, Blackwood had a stroke of fortune: the main scheme had largely sold its equities before the start of the bear market. "The UK scheme is blessed with good trustees and it made a proposal to dump a lot of equities around 2000, which I and others supported, and we moved into bonds," he says. "The objective was to get down to 20% equities, which was based on actuarial advice at the time. So things were improving in the way people were looking at these issues."
A now rare example of an open UK defined benefit scheme, Friends Provident is at the other end of the spectrum from ICI. Its head start came from the fact that UK life offices had already faced daunting risk management challenges. This made the company receptive to derivative solutions, as Graham Aslet recalls. "Realistic balance sheets were coming in and we had been doing various things in the life fund to reduce the economic capital requirements. It seemed to us as an employer that the kind of things we did for the life company wouldn't be out of place in the pension fund as well," he says.
Regardless of any risk management insights or advantages already in place, those who have taken the route to LDI agree that the process from realisation to execution is arduous. Having been beguiled for years by actuarial and accounting frameworks that unwittingly concealed risk from stakeholders, boards of companies often resist changes, which amount to accepting bad news.
Getting to grips with risk management means not just getting the right kind of information. It also means getting it more often that the pensions industry is accustomed, at least on the liability side. According to Van As: "Until 1998 we had only a yearly evaluation of our surplus position, at which point we changed it to quarterly. In 2000 it was monthly and by 2002 it was almost weekly. So the frequency of looking at our solvency position had changed dramatically, and so had our awareness of the volatility of that surplus."
The UK, where statutory triennial liability valuations somehow became accepted as best practice, serves as a cautionary tale. Both trustees and sponsors agree that triennial valuations are unnecessarily detailed and costly for risk management purposes. According to Martin Taylor, the chairman of WHSmith's pension fund: "We did a formal triennial valuation where the actuaries looked at every single account and sent us a hell of a bill. You're required to do that by statute, but it clearly isn't enough if you're running a scheme."
The more risk-aware UK schemes have been discreetly building their own 'desktop' liability valuation models, calibrated to their triennial valuations, which can be updated regularly. Taylor says: "Doing a desktop check of your liabilities is not very onerous. The further you get from the three-year point, the less accurate it becomes, but it doesn't get that inaccurate."
A further obstacle to risk management in the UK comes from accounting rules. The FRS17 standard has been applauded because it enshrines a market-consistent valuation of liabilities, using AA-rated bond yields as a discount rate. Less well known is the fact that generally the value of contributions required to fund the scheme is set according to the expected return on scheme assets, irrespective of risk. The consequence of this actuarial approach is that reducing risk has an adverse impact on cashflow, as ICI's Blackwood explains.
"It increases the rate of funding, because the actuary will say, 'if you've reduced the equity content in favour of a matching bond portfolio, the deficit goes up, because there are less return-seeking assets to close the gap'. Someone who thinks of cashflow as an important risk measurement outcome will say, 'that's a bad thing to do because I know that will increase my annual funding contribution'."
Some actuaries and investment consultants such as Watson Wyatt have attempted to patch up the market inconsistencies in the FRS17 contribution calculation by raising the confidence level for asset returns. However Blackwood believes that the conflict between reducing risk and reducing contribution cashflows remains an "intractable conundrum" for many pension schemes and company management.
For deficit-ridden closed schemes, the constraint of contribution rates has led many sponsors to negotiate longer time periods with trustees, often providing credit guarantees or letters of credit to assuage concerns over sponsor insolvency risk. ICI negotiated such a deal in 2003, providing a £250 million guarantee to the pension scheme backed by assets placed in a special-purpose vehicle. For Blackwood, it provided the vital breathing space he needed to think about risk. "We have a much lower level of cashflow than we would have done without the guarantee," he says.
But how does one think about risk? In the Netherlands, where pension schemes are incorporated, fully funded institutions, Hoogovens could afford to approach things systematically. According to Van As: "We started a finance committee, looking at the financing of our pension plan. We examined all the routes, including the use of derivatives. We looked at diversification of risks, understanding the liabilities, and whether to use matching for only interest rate risk or also inflation risk of the cashflows of only the pensioners, active members or both. In 2003 we decided to base our strategy on all the interest rate risk and inflation risk of all the cashflows because our goal is providing the real value of pensions, not the nominal value, to all the scheme members."
At Friends Provident, the work was done by the finance team, which already had extensive experience of derivatives markets. According to the head of the team, Paul Cooper: "We were looking at things from a banking perspective, where they're interested in cashflows and their sensitivities. We started to analyse the long-term interest rate and inflation sensitivity. For every one basis point shift at that long point in the curve, what was the potential impact on the scheme?"
At first glance, risk management might be understood as simply removing all risks from a scheme, but this is usually unrealistic. A full de-risking involves finding a life insurance company to buy out the pension scheme liabilities and replacing them with indexed annuities. Although this removes all inflation, interest rate and longevity risks, it is usually a prohibitively expensive option. For example, ICI's deficit on a buyout basis is £2.5 billion.
If a buyout is unfeasible, simply switching from equities into bonds meets many people's expectation of risk management. Indeed, many schemes with an advanced view of risk have already been heavy buyers of bonds, in particular inflation-linked government bonds. But bonds on their own are not enough, for several reasons. Firstly, bonds have insufficient duration to match liabilities, according to Van As. "When you take inflation into account you get a longer horizon because you get inflation on inflation in the projected liability cashflows," he says. "The cashflows from indexation are further away than the nominal part, and taking them into account is in our case almost one-third of the total cashflows. So the most efficient matching portfolio is built with bonds and swaps because there are all maturities available going out to 50 years and all interest rate and inflation exposures can be achieved independent of the surplus position or the capital invested."
Secondly, because the vast majority of inflation-linked bonds are issued by AAA-rated governments, the yields are very low. With AA-rated banks as counterparties, cash-collateralised inflation swaps pay a 30bp premium to governments. With such long durations at stake, that premium versus index-linked gilts is as valuable as gold dust, says Blackwood. "Are we getting that 30bp for nothing? The answer is yes, more or less. You might not quite get the Libor return, but you should keep most of the 30bp for essentially the same risk as gilts," he says.
The third reason is particularly relevant for mature schemes with deficits. By definition, such schemes have insufficient assets that they can sell in order to buy the liability-matching bonds they need. But swaps, which typically exchange floating-rate cashflows for nominal or inflation-linked payments, avoid the need for an immediate liquidation of assets. Instead, the swaps simply need to be funded over time, which allows for some targeted risk-taking across asset classes.
For this reason, ICI's Blackwood sees derivatives as the primary tool for liability-driven investment. "Once you realise pension cashflows aren't very complicated, you start to grasp the misleading nature of the discount rate. For given mortality assumptions, if you want a low-risk portfolio with the maximum immunisation of liabilities, you can get most of what you need in terms of maturity and better matching using the inflation and interest rate swap market." He adds: "In economic terms this will of course redefine the deficit."
Entering into a fully fledged liability-driven investment programme involves a final conceptual and practical step, as Van As explains. "We split our portfolio. In terms of cashflow-producing fixed income, we put that in our matching portfolio. All our credit risk, equity and real-estate risk we put aside in a return portfolio. We look at our matching portfolio and ask, which cashflows are we missing relative to our liabilities? Then we just add to it using swaps," he says.
This asset split suits the way that the swaps are funded. Unlike traditional interest rate swaps which pay a regular fixed-versus-floating rate coupon, inflation swaps are normally zero coupon in nature, with opposing Libor and inflation-linked payments compounded and cash-settled at maturity. The return-seeking portfolio can then be benchmarked against a Libor-based total return. At Hoogovens, this portfolio is allocated using a risk budget with a value-at-risk of EUR300 million.
Rather than execute their LDI strategy with an investment bank, both Hoogovens and ICI chose to use Barclays Global Investors as an intermediary, investing EUR2 billion and £1.75 billion respectively in funds structured to match their cashflows. Van As says that Hoogovens took this route in 2004 for reasons of price information and documentation convenience.
For ICI, using a fund structure was a matter of authorisation. "There's a problem a lot of corporates have," explains Blackwood, "which is that a derivative is probably not legally an investment, and may not therefore be covered by the trust deed - so the banks won't contract directly. Our scheme currently has this problem, so it has used inflation and interest rate swaps via a BGI fund." He adds that this framework is expected to change in the near future.
For Friends Provident, a completely different route was chosen, more suited to its concerns about potential future exposure. Cooper explains: "Instead of taking every single point on the curve and trading to negate that with inflation swaps, which involves a lot of transaction costs because there are a lot of swaps, our solution was to take a bucketed approach and go as long as we could, where there is a sufficiently liquid market, and put in place a swap to that date." Already comfortable with derivatives transactions, Friends Provident executed directly with Merrill Lynch late in 2003.
This approach of duration matching has its critics on the cashflow-matching side. In particular, Van As complains that duration matching exposes pension schemes to changes in yield curve shape which force regular and expensive rebalancing of inflation swap positions. Moreover, such a hedge contains nonlinear or convexity exposure to interest rates and is certainly not cost-efficient.
Cooper explains that Merrill Lynch neutralised convexity exposure by embedding a long-dated interest rate floor in addition to the main 30-year inflation swap hedge. However, the disagreement over duration versus cashflow matching says more about the level of flexibility that scheme managers expect after hedging.
As Van As and Blackwood put it, a cashflow match is effectively a 'one-time event' that particularly suits mature schemes unlikely to experience merger activity or new entrants that could skew the risk profile. Friends Provident on the other hand is an open scheme and Cooper says he wanted the flexibility to unwind the swap in response to a market or corporate event.
The week on Risk.net, October 6-12, 2017Receive this by email