The Frijns and Goudswaard committees, created by the Dutch Ministry of Social Affairs to review the current pension fund system and advise on pension policy, have recommended a greater focus on risk management practices and better systems for pension fund management to create a ‘future-proof system’.

In this interview contributed by PIMCO, the firm’s product manager for liability-driven investing and investment solutions, Jeroen van Bezooijen (pictured), discusses what this means for pension funds in the Netherlands and examines how they can successfully meet the committees’ recommendations.

Q: The government has already commissioned two separate reviews of the Dutch pension fund industry in the direct aftermath of the financial crisis. Is the industry in turmoil?
 
JvB: There is indeed a fair amount of soul-searching going on. A new pensions law was introduced on January 1, 2007 after a multi-year consultation period, providing a new framework for regulating Dutch pension funds. In part, the framework was designed to ensure pension funds would have sufficient buffers to protect them against falling into deficits. Unfortunately, this element of the law – the Financieel Toetsingskader (FTK) – proved insufficient. Within two years of its introduction, it was already falling short of the mark.
 
According to the Dutch pensions supervisor, De Nederlandsche Bank (DNB), during the thick of the 2008 financial crisis, aggregate Dutch pension funding positions deteriorated from 136 per cent at the end of the second quarter to 95 per cent at the end of the year. In other words, the industry burnt through the buffers that were supposed to protect it against a one-in-40 year event within six months. It was this robustness – or lack thereof – of Dutch pension fund investment processes and policies that formed the focus of the Frijns Committee.
 
Q: What were the committees’ key findings, and how do you think they will impact on pension funds’ investment strategies?
 
JvB: In the first instance, the Frijns Committee criticises pension funds for focusing too much attention on returns and not enough on risk management. The committee advocates clearer discussions between trustees and members about what pension funds are trying to achieve and the risks involved. In our opinion, this suggests that liability-driven investing or overlay strategies will become increasingly important. Many pension schemes already have LDI strategies in place, but they will probably need to be reviewed.
 
Related to this is the issue of assessing liabilities, another important point raised by the committee. Currently, the FTK’s assessment of liabilities is based on a nominal framework, which includes discounting future pension payments and excluding any future indexation against the swap curve. This is at odds with what most pension funds are trying to achieve, which is to provide and protect the purchasing power of their members’ pensions.
 
In response, the committee argues that this should be reflected in the liability calculation to avoid the current disconnection between what is reported to the regulator and what pension funds strive to do. This would mean LDI strategies addressing both interest rate and inflation risks, rather than just interest rate risk, as is currently the case.
 
Finally, the Frijns Committee suggests the focus on risk management should be broadened to include not just strategic risks such as equity and interest rate risks, but also more operational risks related to strategy implementation or what the committee refers to as an “implementation shortfall.” Pension funds’ implementation shortfall is described as one of the drivers of the solvency level decline in 2008.
 
Q: A focus on risk management seems to make sense. Do you feel the two committees are addressing risk management issues appropriately?
 
JvB: I would say, yes, they are generally tackling risk management issues appropriately. However, the one area that remains unchallenged is the accuracy of asset-liability models (ALM) and other prevalent statistical risk models. The Frijns Committee advocates pension schemes form objectives, model them up in an ALM model and analyse which investment strategy delivers the desired outcome. This is a compelling structured approach but only works if we can assume that the models are accurate in their description of future events. If they are not, you end up misrepresenting risks and potential outcomes and set yourself up for the next big systemic failure.
 
History in general, and the 2007-09 financial crisis in particular, has illustrated how risk models do not accurately capture risk, due to their emphasis on normal distributions. Historical data overwhelmingly show that financial market variables are not normally distributed.
 
We can illustrate this with the assumptions underlying the FTK standard test for pension solvency, which is based on a number of shocks to market value that pension funds should be able to withstand without falling into deficit. These assumptions are designed to provide sufficient buffers against a one-in-40 years event.
 
At the start of 2008, the main parameters were a 25 per cent fall in equities, a 0.5 per cent widening in credit spreads and a 0.9 per cent fall in interest rates. In reality, the market saw a 40 per cent fall in equities, a 3.1 per cent widening of credit spreads and a 1.3 per cent fall in long-dated interest rates during 2008.
 
Assuming a normal distribution, the probability of this happening would be about 0.01 per cent, meaning that it should only occur once every 10,000 years or so. So was 2008 really a one-in-10,000 years event? Not exactly – the market moves of the early 1930s, around the time of the end of the First World War, and during the deep recession of the late 19^th century were all of a similar magnitude.
 
So while the Frijns Committee has spent a considerable amount of time on the important topic of implementation shortfall, I would argue that is only part of the picture. The other half is that risk models are not up to the task they are set out to address.
 
To illustrate, if we assume a Dutch pension fund was 40 per cent invested in the MSCI World index, 60 per cent in euro fixed income and hedged 35 per cent of its liabilities (in line with our estimate of the aggregate level of interest rate hedging), its funding position would have declined from 136 per cent in June 2008 to about 100 per cent at the end of 2008 (compared with 95 per cent for the aggregate pension fund industry). Therefore, market moves explain the majority of the decline in funding positions. Implementation shortfall is important but in most cases a secondary rather than a primary driver of risk.
 
In other words, most risk models ignore tail events like those that occurred between 2007 and 2009, which happen more frequently and are more severe than current risk models predict.
 
Q: What is PIMCO’s approach to risk management?
 
JvB: In general, we believe you need to be modest about what can be achieved with risk modelling. Quantitative analysis is important and useful, but it should not be relied upon completely. When investing, you should have long-term fundamental views about the global economy and its impact on asset classes.
 
We believe risk management starts with identifying key risk factors as drivers of uncertainty. Examples of risk factors include equity exposure, duration, inflation, credit spreads and liquidity. According to this approach, asset classes are carriers of risk factors. We think the key to better risk management lies with identifying and understanding these risk factors rather than relying on quantitative analysis of asset classes’ historical returns.
 
We also believe it makes sense for pension funds to incorporate explicit tail risk hedging strategies into their portfolios. Extreme events occur more frequently than you expect and their timing is hard to predict.
 
Q: So should Dutch pension funds adopt tail risk hedging strategies?
 
JvB: Yes, we believe so. Looking ahead, we expect to see a ‘new normal’ economic environment characterised by lower growth, lower returns and higher volatility. In this environment, developing an investment process based on robust medium- and long-term macroeconomic views combined with tail risk hedging strategies, should better meet pension members’ expectations.
 
We believe it makes sense to set a small budget aside each year to hedge against these tail events on an ongoing basis. Although pension funds give up returns in the short term because there’s a cost to implementing hedges, thinking in such terms over a longer investment horizon is short-sighted. These hedges could end up covering their cost by many multiples during a market crisis. So while you may be giving up returns in the short term, over the long term you have a strategy that may potentially result in excess positive returns. In PIMCO’s view, adopting tail risk strategies is a potential alpha generator for a long-term investor.
 
More specifically, Dutch pension funds are becoming more mature, which means their investment horizons have generally shortened. The Frijns Committee says that about 60 per cent of liabilities are due in the next 10 years, which puts a lot of strain on the system. If pension funds do not meet their return targets, they have to ultimately choose between paying out pensioners the full amount and leaving significantly less for the rest or cutting back on payouts altogether. In this context, the Frijns Committee likens pension funds to ‘ageing giants’.
 
There was less strain on the system 20 years ago when these funds were less mature. Now however, pension funds do not have several decades to recover and therefore need to protect themselves from shocks.
 
Q: How does PIMCO hedge tail risks?
 
JvB:Derivatives play the most important role in hedging tail risk. This includes options on broad equity market indexes such as the S&P 500, FTSE 100 and Euro Stoxx 50, interest rate swaps or swaptions, foreign currency options and credit default derivatives such as iTraxx and CDX. In addition, options or futures on commodity indices are also used to hedge tail risk.
 
The important thing to remember about tail risks is that while they may vary widely in their origins, they tend to pose similar macro risk and can have comparable impact on investment portfolios. We find that nearly all tail risks are systemic risks driven by investors’ desire for liquidity – the ability to trade easily – at a time when nobody is willing to provide it. This results in the simultaneous and substantial decline of equity markets, credit spread widening, increased market volatility and disorderly moves in the currency markets.
 
This increase in correlation allows risk to be broadly hedged at the portfolio level rather than at the security level, which is cost prohibitive. This concept is important when considering how to hedge tail risks. Think about it this way: When you purchase home and content insurance, you don’t insure each item of jewellery and clothing and every appliance individually. You insure the entire house and all its contents. At the very least you need to be ready to purchase insurance if it is attractively priced, even if your neighbour is telling you that you will never need it.
 
Q: What is the impact of implementation shortfall on active management strategies?
 
JvB: The Frijns Committee prescribes a rather rigid approach to devising and implementing investment policy – specifying objectives, devising an investment strategy using ALM models and implementing the strategy versus market benchmarks. This approach almost automatically results in a preference for passive strategies that replicate the benchmark and indices used in the ALM models, and any deviation from these benchmarks could result in an implementation shortfall.
 
We believe that taking fundamental views of where the world is heading – the backbone of active investment strategies – is imperative to sound risk management practices. You can’t just rely on statistical models, especially when many market benchmarks are likely to be based on arbitrary choices.
 
We agree that pension funds and their boards should understand what they invest in. At PIMCO we spend a lot of time explaining our portfolios and strategies to investors. However, fully relying on an ALM models and passively implementing its conclusions is as risky as venturing into asset categories and strategies that are not fully understood.
 
A good example of the shortcomings of passive investing can be gleaned from the current situation with government bonds. If you employ a passive strategy in government bonds at the moment, you will follow a market benchmark with a bias toward countries with the highest levels of issuance. Do you really want to be overweight developed markets that probably will generate the lowest returns and the most problems in terms of budget deficits? Meanwhile, you are underweighting many developing countries whose financial situations are considerably healthier and have better growth potential.
 
Q: How do you envisage the inclusion of future indexation in pension liabilities working in practice, and how will this impact Dutch pension funds?
 
JvB: That is an interesting question and needs to be looked at in detail. In principle, the approach would be either to increase all projected future pension payments in line with expected inflation between now and the time of payment, or to discount liabilities against a real yield curve.
 
The second approach – discounting against a real yield curve – seems more straightforward. The current nominal swap discount curve can simply be replaced with a real curve. The question is which curve to use? In theory, it appears straightforward to use a yield curve of inflation-linked bonds.
 
In reality, it is more challenging. Firstly, there is no established market of inflation-linked bonds linked to Dutch inflation, but government bonds linked to eurozone inflation issued by France, Germany and Italy. And even if you aim to devise a methodology based on these bonds, the size of the overall eurozone inflation-linked bond market is not large enough to supply Dutch pension funds. According to Barclays Capital, the overall size of the eurozone inflation-linked bond market is about EUR280bn, while the size of the Dutch pension fund market is about EUR650bn, according to DNB.
 
Therefore, if you go down this route you risk driving up the market as Dutch pension funds clamour to buy these inflation-linked bonds in an attempt to stay as close to liabilities as possible. This in turn will drive real yields lower and force liabilities up, creating a vicious cycle where everyone starts buying inflation-linked bonds that no-one can really get their hands on while driving liabilities up further. So while it is a good idea in theory, it is very challenging to implement.
 
An alternative is to prescribe the real discount rate, such as discounting liabilities against some fixed real yield, for example 1.5 or 2 per cent. However, the issue with this approach is that you effectively revert back to the system that existed before 2007 of discounting liabilities at fixed discount rates regardless of what happens in the wider market. The only difference is that the discount rate is substantially lower at 1.5 or 2 per cent than the 4 per cent used before 2007.
 
Finally, you can increase liability cash flows by assuming some level of inflation and discounting them against the swap curve. This is comparable to the current approach but with cash flows ‘inflated’ to reflect future indexation rather than based solely on current entitlements, as is presently the case. The main discussion point will be the assumed inflation level. This could be based on a number of variables, such as the European Central Bank’s long-term inflation target (2 per cent), most recent inflation observations and/or ‘break-even’ inflation levels priced into inflation-linked bond markets.
 
In summary, there are many questions that still need to be answered and details to be worked out. Given the complexity of the situation, working with an LDI manager with global expertise makes sense. PIMCO is a large player in global inflation markets and has had many years of experience running LDI strategies for pension schemes in the UK, where inflation is already integrated into the strategy. So we believe we are well positioned to assist our clients navigate their way through these complexities.