Risk control in a fair value regime
Pension funds will be, as of 2006, confronted with a new regulatory framework, in which the pension liabilities are valued on market value. The change in valuation from a fixed discount percentage to a nominal valuation (fair value) is certainly in line with a more global development towards enhanced risk management. In some European countries, like Denmark, regulators have already adopted fair value and others, for example, Sweden, will follow shortly. Also in accounting, the new standards (IAS/IFRS) show a shift to fair value all along the line.
The fair value approach, in combination with the new regulatory framework, causes pension funds to rethink and upgrade their risk management abilities. This will provide them with a more profound understanding of the economical challenges they are facing: investing billions of euros, while taking into account their liabilities.
Understanding how to invest this money in such a way it not necessarily outperforms an asset only benchmark, but enlarges the possibility that the assets meet the (indexed) liabilities, is key. Due to the fair value approach the actual interest rate risk in the liabilities has become visible and shows the high sensitivity of most Dutch pension funds to interest rate fluctuation. For these funds a decrease in interest rates of 1% in most cases means a drop in funding level of over 10%, due to the large duration gap between assets and liabilities.
Since the average funding level of Dutch pension funds at the start of 2005 is between 110% and 115% and the new regulatory framework mandates a funding level of 105% (excluding solvency requirement), the interest rate risks the Dutch pension funds bear are quite obvious and huge.
Since pension funds have to take risk in order to meet their indexation ambition and keeping the contribution affordable, the main question is if this large interest rate risk is an efficient use of the risk budget of a pension fund. Studies for numerous Dutch pension funds show that this certainly is not the case. Hedging interest rate risk via duration extension of bonds or by a swap overlay, proves to be very efficient. For reasons of flexibility, we will assume a swap overlay.
The reason for taking on risk is to get returns higher than the risk-free rate. The common belief is that investing in (private) equity, real estate, and so on, involves taking risk, but also is rewarded in the long run. The expected extra return above the risk-free rate is called the risk premium. For interest rate risk this is not quite that simple. It is most likely that in the long run the 30-years swap rate will be on average above the six-month Euribor, since normally the yield curve will be upwards sloping. This implies that in the long run, expectation in lending money short and investing this money long will generate a return or so called yield pick-up.
Most banks take on this interest rate position. Pension funds have, at least until recently, the inverse position. They invest money in relative short bonds up to 10 years, while funding very long pension liabilities of up to 80 years. This inverse position will in the long run generate a negative return, while still generating risk. Strategically holding on to this position will not be a sensible allocation of the risk budget in the long run. Of course, some nuances are in place. First of all, the indexation ambition of the pension fund and secondly, the broad expectation that the current long interest rate is historically low, implying an expected increase of interest rates. We will discuss these two matters briefly.
Most Dutch pension funds have a conditional indexation policy in which, depending on the funding level, indexation is granted, normally linked to Dutch price or wage inflation. For a growing number of funds this not only applies for the inactive participants but also for the active participants (via the average career system). In case of higher funding levels, full indexation is assigned and the liabilities become more and more sensitive to the real interest rate and less for the nominal interest rate. For most Dutch pension funds however, funding levels are low and indexation policies are granting no or partial indexation, implying the interest rate sensitivity in downward risk, certainly for horizons up to 10 years, has a nominal nature.
The table showsthe results for a typical Dutch pension fund with a constant contribution rate, conditional indexation and an average asset allocation. This table shows clearly hedging can substantially lower the risks of the pension fund.
For an average Dutch pension fund a nominal hedge, possibly with a small part inflation-linked, reduces the risks considerable. For pension funds with higher initial funding levels a nominal hedge still reduces the risks considerably, but, due to the larger probability of indexation in the future, a more inflation-linked hedge can even improve these risk reductions. The results are based on inflation swaps on the European inflation, since this is the most suitable inflation that is readily traded, however, due to a lower correlation with Dutch inflation, less effective. This inflation market is growing exponentially but is nevertheless not yet as liquid as the market for interest rate swaps.
Although in the long run interest rate swaps will lead to, besides a substantial risk reduction, a positive return as substantiated above, an expected interest rate rise can change that. A widely accepted view on the interest rate market is that current interest rate levels are low in relation to their long-term mean reversion levels. An expected rise in interest rate exceeding the current forward interest rates implies an expected negative return on long interest rate positions like the suggested interest rate swap. In other words, not hedging the interest rate risk in the current situation where interest rates are considered low, will be rewarded with a risk premium.
An important question a pension fund should ask itself is whether this risk premium compensates enough for the enormous interest rate risk and thus whether it is an efficient use of the risk budget. For most pension funds the expected return of other asset classes, like eq-uity, are much higher at similar risk levels, making swaps, even at ‘low interest rates’, attractive. The results above, in which the expected interest rate rise was already incorporated, indicate this.
If the expected interest rate exceeds the forward rates, a swaption can be a good alternative. As long as the strike level of the swaption is not too far out of the money, the risk reductions will be largely preserved. However, due to the optionality, one still profits from the expected upward potential of rising interest rates and so the expected negative return on the short horizon, which was expected on swaps, is mostly averted. Independent of the subjective magnitude of the expected interest rate rise, a swaption strategy will be a very good ‘no-regret’ strategy in the short term for most pension funds, especially when it is changed to a swap strategy when interest rates have risen to levels closer to expected long-term levels.
It is clear that a multitude of factors influence the optimal strategy: the desired risk/return profile, the current interest rate level, the current funding level, the indexation policy, etc. Also, the exact asset allocation, with possible interest and/or inflation-correlated classes such as equity, real estate and commodities, is of relevance to the optimal strategy. Sound asset and liability modelling will have to determine the right combination of bonds, swaps and swaptions, and the right combination between nominal and inflation linked. Considering both short- and long-term effects is not an easy task, but an absolutely necessary stage that many Dutch pension funds go through in order to get a better feeling with the risk return profile under fair value. Fortunately, this will not only provide a better fit with new solvency regulations, but an optimal solution will also provide a much better risk return profile from an economic point of view.
Huub van Capelleveen is director of research at Cardano Risk Management, in Rotterdam www.cardanoriskmanagement.com