Investment: The limits of expectations
How are Dutch funds doing against their long-term targets?
- Top Dutch pension funds have lowered return assumptions because of poor 2018 results
- Long-term investment horizons mean that investment policies have not necessarily changed
- Benefits and premiums may be affected by factors other than return expectations
Losses – or at best, minimal investment returns – for the largest Dutch pension funds in 2018 have inevitably lowered expectations for future returns.
ABP, the Dutch civil service scheme and the Netherlands’ largest pension fund, reported a 2.3% loss for the year, compared with an expected average return of 2.8% over the 2018 – 2020 period.
But of course, pension funds invest over much longer time horizons, so they are cautioning against over-emphasis on short-term underperformance.
ABP usually focuses on a 15-year horizon, though it specifies yearly figures for one to 15 years ahead. It works with a central path for average expected interest rates and returns, to work out expected pension results and their distribution based on the usual volatility in markets and the economy.
“Last year’s expected return of 2.8% for 2018 – 2020 had an expected volatility, or standard deviation, of 7.5%,” says Paul Duijsens, head of strategy and asset liability management at ABP.
“We do not expect to hit the expected annual return each year, but we would be surprised if we were more than two standard deviations away. In this light, last year’s return is unremarkable.”
Likewise, PFZW usually takes a horizon of 15 years for ‘base case’ scenarios, using a stochastic approach to model uncertainty. For ‘stress’ scenarios, the horizon is reduced. The stress occurs in the first year and the impact is analysed on a five-year horizon.
“We do not forecast the expected return for just one year ahead, because we simply cannot,” says Jan Willem van Oostveen, investment director, PFZW. “Financial markets are just too volatile to say something about expected return. Moreover, since we do not believe in tactical asset allocation, we do not need one-year forecasts for defining our investment policy.”
Van Oostveen continues: “We can compare our long-term expectations with the 2018 results of course, but this does not provide us with useful insights. Almost all assets experienced negative returns, especially oil and emerging markets equity, where we expect positive results in the long term. This means the difference between expectation and result was relatively large.”
There are three scenarios that would prompt PFZW to modify its expectations:
● Changes in current valuations – for example, lower spreads mean lower expectations.
● Changes in the base case – for example, lower expected GDP growth causes lower returns.
● Changed insights – PFZW periodically carries out in-depth research on each asset class, which can lead to different estimated fair values.
At present, PFZW’s expected return on a 15-year horizon for the total portfolio, including hedges, is 4.2%.
PFZW’s stochastic modelling incorporates the uncertainty around return assumptions. The fund sets policies that not only take into account the positive influence returns can have on ambition, but also the risk of falling coverage ratios.
Van Oostveen says: “The assumptions are by themselves not set to be prudent, but can be seen as best estimates. We do not claim to predict returns with a high degree of accuracy in the short term.”
And he adds there is no direct linkage between expectations and investment policy. Updated expectations are used when discussing policy, but changes are the result of considerations at board level.
Likewise, ABP revisits its expected returns once a year, discussing whether any aspect of the set needs to be updated. This can be anything from the entire approach to formulating expected returns, to adjustments in parameters.
At the same time, it also takes account of the changed starting point of current market prices.
“This year, for example, we lowered the expected equilibrium interest rate in the UK in light of the reduced growth prospects after Brexit,” says Duijsens. “We also work out alternative expected returns based on different principles – market implied returns, deterministic scenarios – and discuss whether these tell us something about what markets are discounting.”
Then, throughout the year, ABP observes market movements; when these hit pre-defined triggers, it automatically re-assesses its assumptions and makes any necessary adjustments.
“In practice we rarely see the need to change the assumptions that underlie our long-term investment strategy during the year,” says Duijsens. “The tactical asset allocation team at APG, however, frequently changes its approach to rebalancing the portfolio after significant market moves.”
But are Dutch pension funds overall using a sufficiently prudent approach towards return expectations?
Harry Horlings, director, institutional business development Netherlands at NN Investment Partners, says: “We are seeing expected returns currently used in ALM studies which are below the expectations of a few years back and also below returns in past years. It comes down to a prudent approach to expected future returns – for example, using rather modest expectations for interest rate developments in line with the forward curve.”
He says NN’s annual update of expected returns also shows lower levels of returns compared with a couple of years ago.
“The excess return versus pension liabilities remains at reasonable levels for the longer term, but for the coming years we expect increased uncertainty in regard to these returns,” he says.
According to Horlings, pension funds with a sufficiently high coverage ratio are expected to keep their indexation potential.
“For low-coverage-ratio pension funds, however, increased return is urgently needed to regain indexation potential. But, the regulatory framework unfortunately limits these pension funds in increasing their risk profile.”
More pension funds are using lower returns in their ALM studies, according to Aon. The consultancy expects that an average fund will achieve a 0.4% lower return on its entire investment portfolio in the coming 10 years. Aon’s forecast is based on interviews with over 200 experts. So, for example a fund which made its calculations in 2015 with an expected return of 4.5% would now use 4.1%.
But Aon says pessimistic expectations are not leading to major changes in the investment portfolio or contributions.
Corine Reedijk, senior ALM consultant, Aon, says: “Many pension funds need to stay within defined risk margins because they are underfunded or have a reserve deficit. They cannot simply increase their risk profile because of an ALM study. These funds are more or less stuck. But pension funds with a better financial position are discussing reducing risk in order to secure future indexations.”
She continues: “If pension funds are changing their investment policy, they are looking for better diversification within the limits of the risk margins, for instance by adding another investment category, such as mortgages. However, adding an investment category requires considerable paperwork for the pension fund and therefore not all of them are willing to make the changes.”
And she says the lower yield expectations used in ALM studies have a limited impact on premium policy.
“Funds which have determined their premium on expected returns use the 2015 parameters from the Parameters Commission,” she says. The Parameters Commission is an advisory body to the Dutch government on return assumptions. “Many funds cannot do much with the outcomes, because they have made a fixed premium agreement with social partners, and there is often limited room for premium increases.”
Duijsens agrees: “Pension premiums are based on long-term return assumptions,” he says. “Changes in benefits and indexation are not based on return assumptions, but on the fund’s financial position, where liabilities are computed using the risk-free swap rate.”
Neither are return assumptions used to calculate funding levels, he says. These are based on the market value of assets and the swap curve, modified with an ultimate forward rate, the UFR.
“There is a role for expectations in signalling that future financial returns may be different from past performance, which helps to calibrate expectations,” Duijsens says. “By formulating expectations on both the average and the distribution, our policies do not just reflect the average, but also take account of the 5% worst outcomes. This creates a prudent baseline.”