Pension fund investors are demanding more of their benchmarks in the 21st century. The traditional qualities required of a benchmark index are: that it fairly represents an asset class; and is a measure which fund managers are happy to use in running and monitoring portfolios.
However, this is no longer enough and has mainly been driven by the more fragile finances of pension schemes. In the UK, it has been reinforced by the Myners principle of ‘appropriate benchmarks’ which requires trustees to “explicitly consider… whether the index benchmarks they have selected are appropriate; in particular, whether the construction of the index creates incentives to follow sub-optimal investment strategies”.
The Myners principles also encourage trustees to consider the bigger picture. This has prompted trustees to rethink the role of the benchmark in managing the overall risk in the pension fund’s investments – the risk that liabilities are not met. In the current process, using asset-liability modelling and other tools the trustees set the benchmark and the asset allocation. The fund manager then selects the investments to take risk and gain rewards against that benchmark.
This way of splitting the risk between trustees and fund managers may seem rather unbalanced. The trustees take responsibility for the lion’s share of the risk relative to the liabilities. The investment professionals focus on the smaller part of the risk and develop a portfolio of products all designed around these handy index and peer group benchmarks. This isn’t really what trustees had in mind. Trustees don’t want to take investment decisions – that’s why they employ a fund manager.
Most institutional investors would prefer a process in which the fund manager gets closer to understanding their liability-related risks, illustrated below. This involves adopting a benchmark which has been designed to better suit the liabilities.
To make this dream a reality, benchmarks would be as close as possible to a scheme-specific liability driven investment approach which has the following features:
o A duration driven by the scheme’s cash flows and an appropriate mix of fixed and inflation-linked assets to avoid unnecessary risks; and
o A ‘pick and stick’ (buy and hold) approach to avoid unnecessary transaction costs.
The fund manager is then asked to manage a multi-asset portfolio using this portfolio as the benchmark. Whether invested in bonds, property or equities, mandates would have the objective of outperforming a tailored bond portfolio with some risk tolerance. The objective for the fund manager is then more closely aligned with what the client wants. It is something along the lines of ‘beat the liabilities by x% with y risk’. The mandate focuses on credit risk guidelines and detailed implementation issues, rather than risk relative to some arbitrary index benchmark. To implement this strategy requires innovative solutions, including using cash flow swaps and the full range of bonds around the world.
As this is a long way away from current practice, it takes a brave trustee and a radical fund manager to make this work. While it doesn’t stop some early adopters making this change, a step in the right direction is to assess and set bond or equity benchmarks by looking at how ‘sub-optimal’ they would be when matched relative to liabilities.
New bond benchmarks
To examine what makes a good bond benchmarks in this liability-driven world, it’s worth looking closer at the characteristics of the bonds which best suit the liabilities. There are two reasons why choosing a bond index may be sub-optimal (to use Myners’ favourite benchmark word again): risk and return.
If the benchmark is to be as close as possible to the liabilities, a bond index simply doesn’t fit the bill. The cash flows, while varying from scheme to scheme, are usually much longer-term compared to the bond index. This means that the bond index and the need for reinvestment hold considerable risk for the trustees relative to their liabilities. If the duration of the liabilities averages 20 years and the bond index 13 years, measuring risk relative to the index would be missing the point. Moving bonds longer-term relative to the benchmark, maybe out to 17 years would actually reduce risk relative to the liabilities. Similarly, many managers regard the relative weightings in index-linked and fixed as tactical or macroeconomic positions. Instead, the long term mix between these two types is important in the context of the particular scheme’s benefits and liabilities.
From the return perspective, the sub-optimality stems from unnecessary transaction costs. If the benchmark is a bond index then, whether active or passive, most typical bond mandates are not hitting the mark. Passive managers buy bonds just because they are issued and sell bonds when they fall out of the index. This has no bearing on the liabilities yet still incurs costs. Active managers have even higher turnover (300% to 400% pa) and costs (0.2% to 0.4% pa).
For those who can afford to take more risk, there’s no need to stop at bonds. Other asset classes can also be used, but keeping one eye on how liabilities are assessed may lead to different benchmarks.
Most long-term investors would have welcomed a smoother ride through the funding roller-coaster at the end of the 20th century shaped by equity market volatility. If a lower volatility approach is possible without impinging too heavily on returns, this is an attractive proposition – particularly now that equity market volatility hits company balance sheet disclosures through FRS 17 or other market-related accounting standards.
For example, we carried out some analysis based on global equity indices drawn from the FTSE All World universe where:
o Economic groups were equally weighted (10% each);
o Stocks were chosen so that each group had the same number of stocks and each stock was equally weighted.
The rebalancing rules were based on intuitive ideas of reducing sector volatility through sector diversification. To reduce the likely impact of transaction costs, constituents and weights were reviewed only on an annual basis and in one index we relaxed the strictly equal-weighted approach, allowing the drift in weights away from their neutral position to be higher for individual stocks than for economic groups.
The chart illustrates the results and shows how £100 (E157) will have grown if invested in the various indices (with no reinvestment of dividends). The two new indices (“Equal weight broad ranges” in heavy pink and “Fully equal weight” in heavy green) both had lower volatility and higher returns for 1994 to the end of 2001 than the FTSE All Share and the FTSE All World indices.
As always, with the benefit of hindsight, it is always possible to come up with better ways of building portfolios. It’s worth noting that we came up with the detailed rules for constituents and rebalancing regimes – and the period for the analysis – before carrying out the calculations. They were based solely on intuition and not by any back-testing. Far better solutions could be found by those willing to look.
Benchmarks driven by corporate strategy
The greatest benefit of using a liability-driven approach to choosing benchmarks is that it helps the trustees to focus on the risks that they really want to manage. These are the risks that the finance director has to focus on when thinking about the pension scheme in the context of the company’s accounts. They are the risks that the HR also worries about because they threaten the continuation of the pension scheme and the stability of this particular aspect of the remuneration package.
The discussions with fund managers about risk management can be pulled up to a higher level by focusing on liability-driven benchmarks. Is the pension scheme right for the company’s financial and remuneration strategy or are the risks and costs unacceptable? All pension schemes, both defined benefit and defined contribution, are expensive and absorb a large amount of time and energy by corporate management. They need to be worth it. Those thinking longer term are coming up with clearer aims for the pension scheme. These filter down into the investment strategy and setting benchmarks for managing the risks and costs that really matter.
Sally Bridgeland is with the quantitative research team at Hewitt Bacon & Woodrow in London