In these days of specialist investment management, choosing an index or, more importantly, the right index is becoming more critical. The impact of indices on overall portfolio returns on a year-to-year basis can be significant.
During 1999, the annual return on the FTSE Japan Index was 78%, while the return on the MSCI Japan Index was 67%, a huge difference. There were several specific factors behind this differential, but less extreme examples have been seen for indices for other asset classes.
Clearly an investor’s choice of index will impact on the return of their portfolio, certainly on a year-to-year or month-to-month basis. Depending on how fundamental the differences are, performance differentials may even out over longer periods, but one of the first rules for institutional investors should be to be clear about their objectives.
The choice of indices is becoming more confusing by the month, with existing providers broadening the range of indices they offer to cover asset classes that are gathering favour with institutional investors. Micropal covers in the region of 6,000 separate indices. Asset classes that have seen a flurry of new index launches recently include socially responsible investing and high- yield bonds.
The choice of indices is also becoming more important as plan sponsors’ focus on risk control increases. Investors are typically specifying mandates more tightly than in the past as they expect to know what they are buying. This is often expressed as a set of limits of portfolio holdings or performance variations (actual or potential) from the index. This means that portfolio returns are becoming more index -driven. Most institutional investment managers view their job as being to beat the index they are given (and they know that they will risk being removed from the mandate if they underperform). There have been accusations in recent years that most institutional investment managers are ‘closet indexers’, a charge that is not entirely true but which does reflect the increased focus on selecting benchmarks. The industry has certainly moved a long way from the days when absolute return was the key objective.
Ideally, the choice of indices should be an integral part of investment decision-making and should be considered from the asset allocation stage. Comprehensive studies conducted on the investment returns actually achieved by pension funds show the overwhelming importance of asset allocation. Brinson Hood and Beebower found in their seminal work on performance attribution in 1986 that asset allocation accounted for 93.6% of the total variation in returns for a typical large pension plan. Complex models are often used to simulate the behaviour of investment markets, and therefore determine the most optimal asset allocation for the particular fund. This hard work can be undone if the fund doesn’t then invest against indices that provide a fair representation of the market.
At this stage a variety of points need to be considered. For example, if the fund is to invest in bonds, is this sovereign debt only, or will corporate and asset backed debt be used. If non-sovereign debt will be bought, what is the level of credit rating you are comfortable with? For a global equity portfolio there is another set of questions to be considered. Are you happy for your manager to invest in emerging markets and smaller companies?
New questions are being raised by indices that slice across traditional splits between asset classes. The FTSE multinationals index family treats global companies as one asset class, and this is an asset class that is not traditionally a part of asset models.

Choice of indices is also being made more complex by the increasing concentration of equity markets on a few large stocks. Many investors are concerned by the concentration in a few stocks in their portfolio that this could result in and so are considering limiting the weighting in these stocks to below index weight. Indeed, some investment vehicles have regulatory and legal requirements, which do not allow full weighting in some stocks as the market stands today.
This is a dilemma for investors. Investment managers will claim that they cannot be measured against an index which they can’t invest in, but if investors move away from the market weights, they may find themselves embroiled in estimating the impact on indices and having a less clear-cut analysis. Indices with limited weights in some stocks could be an answer to this.
Although the choice of index should ideally be an integral part of a fund’s strategic decision, in practice the choice of an index can sometimes come down to being the one that the manager in question uses most frequently. This can be a valid approach when the differences between indices are minimal. If a manager’s risk and reporting systems can most easily cope with a particular index, this could make life easier for all concerned. However, it is the responsibility of the plan sponsor to formulate long-term objectives and strategic investment policy.
Unless guided by clearly defined investment objectives and policies, the investment manager is tempted to manage all funds in virtually the same way in a ‘one size fits all’ approach. Then the portfolio will not meet the specific needs and objectives of each particular client.
However the large international managers are increasingly prepared to apply their process to any index the client requires. This links back in with their focus on adding value to a base index rather than structuring a portfolio from scratch.
Risk control against indices is becoming ever more complex. Periods where not holding a few of the right stocks, as we saw during the heady performance of technology media and telecom stocks recently, have certainly focused investors’ attention. Many investment management organisations are reviewing models and processes for monitoring the active positions they hold against the index and the potential impact of these. However, it is difficult to build this into a modelling process for periods when markets are unusually volatile.
The focus on benchmarks leads to a segmentation of the portfolio risk of a pension fund into two components:
q the risk inherent in the benchmark, which is the responsibility of the plan sponsor, and
q the active risk (tracking error) added to it by the portfolio manager, whose objective is to create a portfolio that is sufficiently different from the index to allow her to outperform it.

In terms of total portfolio risk the benchmark risk is by far the most substantial. Not surprisingly, plan sponsors increasingly pay more attention to choosing benchmarks using factors such as the timeliness of return data and costs of receiving this (for both the investment manager and the consultant/performance measurement company that will monitor performance), the transparency of construction methodology and the degree of use in the industry. The drive towards customisation requires the possibility to look beyond standard indices to the index constituents in order to reassemble the data to meet specific benchmarking or investing needs.
Recently, issues such as the treatment of corporate cross-holdings and the free float in benchmark construction have been more looked at more closely. In the case of most indices, stocks are given a full weighting even if a large proportion of the shares is still owned by the government, in the case of privatisations, or by a parent company. The use of these indices as performance benchmarks leads to the effect that investment managers sometimes are obliged to by the shares, whether they think the purchase is sensible or not.
So it is certain that the choice of index is critical. It should be at the top of your list of decisions to make in the investment process, whether it is used as a performance benchmark, as an asset allocation tool, or as a securities database for stock selection. Besides these applications is benchmarking, an essential prerequisite for a systematic and structured approach to risk management.
Frans Dooren is head of investment consulting at Aon Consulting in the Netherlands