The financial crisis revealed flaws in pension fund benchmarks. But until these tools evolve more, trustees need to consider how to best use them and avoid being led astray, finds Nick Fitzpatrick.
It has long been known that pension scheme benchmarks – those tools that funds typically measure their investment returns against and sometimes their liabilities – can lead fund managers astray.
The best-known example is how active managers, in trying to outperform an equity index such as the FTSE 100, might almost replicate it rather than beat it to try and ensure that they at least do not underperform their target.
Now the financial crisis is revealing a new set of problems with standard benchmarks that is sparking yet another re-think about how best to construct and employ them.
Geoff Reader, head of pensions and treasury management at the Bedford Borough Council Pension Fund, says: “Setting the benchmark is absolutely a huge decision, rarely do you give it the appropriate level of analysis or stress testing.” (See Pensions Head-to-Head, p20).
The Bedford scheme discovered during the financial crisis that benchmarks for real estate were not adequate because they encouraged an investment provider to use leverage – a situation not wanted in a down market.
Roger Gray, chief investment officer at the Universities Superannuation Scheme, says that a number of pension funds found that their supposedly low-risk non-government bond allocation had a surprisingly high exposure to subordinated financial sector debt as a consequence of drift in the underlying index composition.
“A benchmark should reflect what exposure you want in the scheme’s asset allocation,” he adds.
Concerns with indices, which are commonly used as benchmarks, are surfacing in current research by the Edhec Risk Institute. Felix Goltz of Edhec, says: “A survey of 100 institutional investors in Europe, which we are currently analysing, found investors are critical of existing indices. The problem is that pension funds typically use broad market indices that do not necessarily reflect their objectives.”
The key word there is “objectives”.
After all, a closed defined benefit pension scheme that has predictable liabilities might not need to beat the FTSE 100 or use leverage to meet its objective of paying those pensions. Therefore, it doesn’t necessarily need to set common benchmarks like the FTSE 100 for its asset managers to perform against.
Paul Jeffries, investment manager at Railpen Investments, says there is a distinction to be made between a benchmark and an index and this is sometimes forgotten.
“A pension fund will create a benchmark at the top of its decision tree. The scheme’s solvency ratio is the most widely used benchmark.
“Then, under this, there will be various other benchmarks put in place in order to meet the solvency requirement. These could be benchmarks for asset allocation and for liabilities, for example.”
Running further down the decision tree, investment managers are put in place for different asset classes and given indices to meet or beat. It is, of course, intended that the fund managers’ performance in aggregate will allow the scheme to
meet its ultimate benchmark, such as the solvency ratio.
“Because this distinction hasn’t always been recognised, there is often a focus on performance of the assets versus indices, and this comes at the expense of looking at the broader issues, such as asset allocation, or at the direction a fund is headed in. Whether a manager has beat the FTSE All Share is less of an issue in that context,” Jeffries adds.
David Calfo, head of group DC [defined contribution] pensions strategy for BNY Mellon Asset Management, says: “In the context of DC and, to some extent DB, benchmarks need to account ultimately for what your end goal is, and that end goal is to achieve an appropriate income stream.”
Although liability-driven investment (LDI) is changing perspectives, many trustees have failed to see that a benchmark is less about asset manager performance and more of a risk decision based on the maturity and cash flows determined by a scheme’s liabilities.
“The industry needs to think about benchmarks that account for more than just the asset side of the balance sheet. Traditionally, both sides of the balance sheet – assets and liabilities – are treated independently,” says Calfo.
Ashish Kapur, European head of institutional solutions at SEI Investments, says: “Looking at whether you have outperformed the FTSE is relevant, but it is not the first thing you should do. The more important thing is to look at how well your assets are doing in relation to your liabilities. That means thinking about benchmarks in a different way.”
For defined benefit plans, looking at assets and liabilities together and benchmarking against the scheme’s funding level makes volatility more apparent, he says.
“The funding level moves not just because of equities markets moving around, but also because of interest rate movements.”
He says that a typical UK scheme has a 20-year duration implying a 20% volatility of its liabilities. This means the liability can go up or down 20% for every 1% change in interest rate. “If assets perform well and return 20% and at same time liabilities move up by 20% then the funding level hasn’t changed despite the high return.”
Calfo, at BNY Mellon, says DC is a good illustration of the divide between asset and liabilities. “For example, the entire focus for years has been to have a big pot to retire on. But how big should that pot be? There is a greater need to consider longevity or purchase of an annuity, which entails volatility.”
In other words, more emphasis needs to be placed on decumulation as well as the accumulation phase.
“LDI is a good concept, but so far in DC it is virtually non-existent,” Calfo says.
S&P Indices is attempting to confront the DC benchmark challenge. Srikant Dash, head of channel management and solutions, says the firm is strategically investing in the issue globally. It has worked extensively on the DC market in the United States.
Dash, who is based in New York City, says: “In the US, about 50 cents in every dollar that is in DC goes into target-date products, but there have never been adequate benchmarks that either the plan sponsor or the participants could use.
“Traditionally, they have used a blend of indices and have based their asset allocation on this. But with research suggesting strongly that 80% of returns come from asset allocation, this is inadequate.”
Fund companies in the US target-date market have adopted a “glide path” model, which alters the asset mix through time. Like lifestyle funds, they will invest more in risky assets when the pension saver is young, and de-risk closer to retirement.
Standard asset mixes tend to be 60/40 equities/bonds, benchmarked against indices such as the S&P 500.
“The problem is that you are fixing the benchmark to represent your most important investment decision: your asset allocation. You’re not measuring the impact of asset allocation, which is really going to drive the investment outcome. It’s deeply flawed.”
Dash says S&P Indices has the industry’s first benchmark for this market. To create it, the firm collects data on all available glide paths and composes an index of the average asset allocation.
This means one provider can be compared with another and performance attribution, such as the effect of emerging markets, can be measured.
Dash says the liability side of the balance sheet is catered for because the glide paths factor in how much replacement income is needed in retirement, albeit this creates the issue of the assumptions used by actuaries.
Finding an ultimately appropriate benchmark presents problems beyond the technical factors of what assumptions are appropriate. There will also be the issue of getting a consensus on them. Clearly, index providers such as S&P Indices, and also MSCI, which has developed “minimum variance indices” to help trustees confront volatility, will be hard at work.
Investment managers will be watching with interest for any effect on them, too, particularly in the alternatives world where investors want more scrutiny about how much of an absolute return benchmark is risk free – in other words, delivered by factors other than the manager’s skill.
As Jeffries at Railpen points out: “A lot of hedge funds that take a performance fee on absolute returns do not provide a risk-free hurdle in their fee structure.”
©2011 funds europe