The index gives the investment manager an essential reference point. It is a prime way in which the pension scheme instructs the fund manager what risks to take, whether the prime worry is losing money in nominal terms next month, or protecting the scheme against inflation, or maximising the long run real return, or something else. Consequently index choice is a crucial issue for a fund.
There has been strong trend towards consolidation among index providers. For example S&P has just taken over the Citi indices business and earlier the IFC emerging markets index, while FTSE purchased the Barings emerging markets indices. Consequently most pension funds are likely to choose indices provided by FTSE, MSCI, S&P, Dow Jones or Stoxx. These indices are also becoming more alike in their treatment of such issues as the adjustment for how much of a stock can be bought – the ‘free float’.
Plan sponsors are also likely to choose one provider for nearly all asset classes. This trend towards ‘families’ of indices ensures consistency of approach, and helps to avoid complications where one asset class ends and another begins – the border between large and small cap for example.
There are cases for and against each of the major suppliers, particularly over such issues as transparency – how understandable and consistent are the rules for each index – and how predictable are rebalances; lack of clarity can cost money. Independence is another issue, with some questioning MSCI’s ownership by a broker, Morgan Stanley. The differences may not be strong enough to justify a variation from the customary index choice in each country, so that most UK funds will choose FTSE, while many European plans will opt for MSCI. Nonetheless, it would be worthwhile for funds, especially larger plans, to discuss the appropriate provider with their investment managers and consultants.
Much more important, however, is the third trend, towards custom indices, and the related issue of understanding what indices are for. The discussion of the benchmark index with their advisers and investment managers is one of the most important the trustees will have. It should never be an afterthought, nor should it be a simple question of following the herd.
The benchmark or index – they are, or should be, the same thing – is a crucial part of determining appropriate investment risks. A few years ago Vodafone had over a 12% weight in the FTSE All-share Index. Was a 5% holding in Vodafone then low or high; very low because it was so far below index weight, or very high because it was so much greater than one divided by the number of stocks in the index? The fund manager must know in order to implement any view he might have on the stock.
The index tells the fund manager what to do if he has no view. And he should not be forced to take views; my own biggest losses as an investment manager have been made when I took a strong position on a weak view. But if it is not clear even whether a particular position is a big bet or not, the situation is even worse – the fund manager will regularly be taking large risks on no view at all.
At the TRW/Lucas pension scheme in the UK we developed a capped index for UK equities. The trustees wanted to benefit from the general performance of the UK and world economy, wanted the long run equity risk premium, and wanted the long term inflation hedge of equities. That meant there should be a benchmark based on UK equities, and that broadly speaking it should be weighted in line with shares in the economy, which is roughly similar to saying the index should be weighted by market capitalisation. But they were also concerned about taking too much single company risk.
The heavy Vodafone weighting in 2000 is not a new problem. There have been high index concentrations before, for example in 1900, 1920, and with oil stocks in 1980. But there is a much stronger index orientation now, so the problem of high exposure to individual companies is greater now. Barings shows the risk that such large weights with one firm can have; management error meant that overnight an apparently sound firm went bust. Low risk relative to a risky index does not equal low risk for a fund!
At TRW/Lucas we chose to have a 4% cap on the maximum exposure to any one company. This limit was set on the six monthly rebalance dates; in between exposure could rise above 4% with market movements. In 2000-2002 the 4% capped index outperformed the uncapped FTSE Allshare Index by over 5% (as Vodafone under-performed the index). But we would have liked this index even if it had under-performed; it was developed to reduce single company risk.
Some have proposed a multinational index. As a liquid means of gaining international exposure to large firms there is a case for such an index. But it is not a solution to the concentration problem, as it merely swaps one risk for another, replacing a BP problem with BT problem, and a Vodafone exposure with exposure to US high tech and Japanese banks. It fails to reduce exposure risk.
Certainly, asset weights should be reviewed. At TRW/Lucas we used roughly equal weights for the three main regions of the world. There is a case for GDP weighting of countries, and for equal weighting companies (if the fund is not too large). One large Dutch plan has custom groups of indices dividing Europe into four blocks: large and small cap for the UK and Ireland and for the rest of Europe.
One questionable area of custom development is ‘socially responsible investment’ indices. One leading SRI index uses highly dubious criteria, including a blanket exclusion of companies involved in tobacco, weapons and nuclear power. Such simplistic and often political judgements are unlikely to be in the interests of pension fund members.
Index problems have been especially great in the fixed interest area. Index provision has been much more fragmented and less developed than for equities, with indexes created by a wide variety of suppliers for local requirements, with few international standards. It is an area of increasing work by FTSE and others.
Bond indices also illustrate a problem with market capitalisation weightings. For bonds this means the highest exposure to the most indebted entities, not the most obvious requirement. There is no easy answer to this problem. One common solution has been to use a AAA government bond index – gilts for example – but to permit much greater latitude to buy assets not in the index than is commonly permitted with equity investment. This is often labelled ‘opportunistic bond management’.
In emerging markets, there have been indices weighted by GDP or in some other way that dampens the exposure to highly indebted countries.
Another possible concern is hedging. If a fund expects to make use of futures and options, it might – all else being equal – want to choose a benchmark index that is used in liquid traded derivatives markets. This might imply a more narrowly based index than might otherwise have been chosen.
The selection and use of indices is a crucial decision for pension trustees, where proper care will deliver dividends in both in optimal risk and return and in understanding of investment issues.
William McDougall is an independent investment and pensions consultant. Until last year he was the chief investment officer of TRW’s UK Pension Scheme (formerly Lucas)