On the Record: Using and abandoning benchmarks
Geoff Reader, head of pensions and treasury management at the Bedfordshire Pension Fund; Marinos Gialeli, general manager at the Hotel Employees Provident Fund in Cyprus; and Gregor Hirt, head of investments at Switzerland’s Vorsorgestiftung Schroder & Co Bank pension fund share their views on benchmarks.
Bedfordshire Pension Fund, UK
• Invested assets: Just under £1.3bn (€1.6bn)
• Members: 49,000
• Defined benefit
• Date established: 1974
Bedfordshire Pension Fund tends to use benchmarks for all of the specific mandates and tries to fine-tune the benchmarks it uses to the particular mandates.
The fund views benchmarks as points of reference, which are also used in terms of formal performance targets.
Its portfolios hold a wide variety of benchmark, ranging from standard ones such as the UK Equities FTSE All-Share Index to ones where the mandate used offers the opportunity to match the fund’s long-term ambition on the inflation side to tie in with its liabilities.
The fund has been using some of the benchmarks for a long time, while others have changed as time evolved. One thing that the fund has discovered is that the industry evolves and advice on what benchmark to use changes.
When the fund first started to invest in indirect property, for example, only one index was available. But that index probably was not the most appropriate, and it was only with time that the fund obtained the knowledge to revise it and make sure the manager had a more appropriate benchmark.
The pension fund’s most interesting use of benchmarks is in the unconstrained global equities portfolio where it applies a long-term and short-term benchmark.
The long-term benchmark embodies where the fund wants to get to, while the short term one can explain any short-term movements.
Non-traditional benchmarks are interesting ideas but the fund has not yet invested in a complex portfolio where those benchmarks would be of value.
However, for diversification purposes it has started to look at fundamental indexing although with regard to those benchmarks, it is still at the information gathering stage.
In some areas, such as sovereign debt, there can be issues with benchmarks, which is why the fund has only one small benchmark for UK Gilts, the FTSE All-Share Gilt Index. The other sovereign debt exposure is via the multi asset absolute return mandates without any debt benchmarks. The fund introduced those mandates with an absolute return target of inflation plus only earlier this year so it is too early to tell how well they will perform. The other absolute return benchmarks the pension fund uses are for a fixed interest unconstrained portfolio and a global tactical asset allocation fund.
Hotel Employees Provident Fund, Cyprus
• Invested assets: €273m
• Total members: 12,860
• Defined contribution
• Date established: 1968
Our current asset allocation is split between bonds, equities, real estate, loans to members and cash.
For some of the asset classes - namely bonds and equities - we use traditional benchmarks. But, fundamentally, we are looking for absolute rather than relative returns.
For the global equities portfolio we are using the MSCI World, while for our global bond portfolio we are using the Barclays Capital Aggregate Bond index, previously the Lehman Aggregate Bond index.
Our global equity mandates are for a value and a growth fund, which have benchmarks of MSCI World Global index plus 2% and CPI plus 6% respectively. But if one generates 5% and the other -1%, any benchmarks are irrelevant.
The idea behind using the benchmarks is, of course, to outperform them. But initially the debate was about whether to invest passively or actively. We invest actively, which is why the benchmarks we use are for internal use and guidance only. In other words, we do not tell our asset managers to invest according to a benchmark. Neither do we regularly tell them whether they outperform or underperform one of our in-house benchmarks.
We are looking for unconstrained managers and want them to use their best ideas to perform well. We do not want them to follow an index or benchmark because, in our experience, 50% outperform the index.
However, when we realise there is a problem with their performance in comparison with one of the global indices, we send them a letter telling them they are underperforming relative to our internal benchmarks, and that unless they raise their game we will have to let them go.
In that sense, we are a dynamic investor. If a manager does not perform according to our expectations for a few quarters, we will replace him - we will not just watch them underperform.
We have yet to make use of non-traditional benchmarks such as style indices, and because we are a small fund looking at investments from a global unconstrained perspective, tailor-made benchmarks are of no use to us either.
We plan to invest approximately 2.25% of our overall portfolio in hedge funds and infrastructure respectively over the next couple of months.
To measure the success of those investments, we will use the three-month Euribor plus 4% as a benchmark for hedge funds and two different benchmark indices, the MSCI World Global index and inflation-plus 5%, for infrastructure.
Vorsorgestiftung Schroder & Co Bank, Switzerland
• Invested assets: CHF89m (€73.5m) including the BVG Stiftung
• Total members: 200
• Defined contribution
• Date established: 1985
Most Swiss pension funds are still benchmark-oriented - however, we no longer make use of traditional benchmarks for Schroders’ own pension fund in Switzerland.
There is simply too great a difference in the volatility between a fund that tries to beat a market-cap weighted benchmark and the liabilities of a pension fund, and the pension fund does not get compensated for this risk.
Instead, we use a target return of LIBOR plus 2% and aim to generate an asymmetric return profile, which is better suited to our liabilities and risk profile. Our active management allows us to put in place downside protection strategies when appropriate, and to minimise any large benchmark-driven market falls. In the absence of a benchmark, we manage the risk by fixing a maximum ex-ante value-at-risk for the portfolio.
Nevertheless, while we refrain from using benchmarks, there is always the risk that if the board of trustees learns about a mandate that has performed better they may no longer be satisfied with our performance, even though our volatility and drawdowns would have been much lower than those of the most commonly used benchmark indices, the Pictet LPP 25, the LPP 40 or the LPP 60.
Until 2006, we used Pictet’s LPP 40 index as our traditional benchmark although, even then, we did not make use of any sub-indices. But due to the large tracking error between the volatile benchmark and the stable underlying liabilities, the results in the wake of the correction of the tech bubble were disappointing. A decision was consequently made to reduce the downside risk and volatility of the fund.
In 2006, we decided to move away from the traditional benchmark approach to a more dynamic and risk oriented strategy, targeting absolute returns.
The Pictet 40 index remained as an internal yardstick for comparison until 2010 but we have since abandoned individual indices for bonds and equities.
Particularly on the bond side, it makes no sense to invest in capitalisation-weighted indices that allocate the largest amount of money to the most indebted nations or companies. With equities, we feel the 15% weighting of Swiss equities in the Pictet 40, for example, is too high and means that the index is too concentrated. The Swiss Performance index - the index for Swiss equities - is highly dependent on the fortunes of just six companies, which cannot be described as diversification when you have a potential global universe of 10,000 stocks.