Building an index of indices
The chief obstacle to the institutionalisation of hedge funds is the absence of proper performance evaluation. This is conclusion of a recent report from Edhec Risk and Asset Management Research Centre, part of the Edhec Business School, based in Lille and Nice.
The problem, the report says, is the lack of an acceptable and representative benchmark. Because there is so little information about hedge fund constituents, the normal rules of representativeness through market capitalisation do not apply. As a result finding a benchmark that is representative of a particular management universe is no easy matter.There are plenty of hedge fund indices to choose from, and their number continues to grow (see list) However, the Edhec report suggests there are serious drawbacks to their use.
The most serious is a lack of transparency. Since a hedge fund’s participation in a database is voluntary, the reliability of the data published is often questionable. Hedge fund index providers also have different selection criteria for inclusion in their databases. HFR for example excludes managed futures from its database, while the TASS and MAR databases include them.
As a result, the report suggests, data provided will not be representative of the same management universe.
Nor are indexes really representative of their own database. For example, CSFB uses 431 funds in its indices drawn from a TASS and Tremont database of 3,300 funds (see list). Since the different indices are constructed from different data the report concludes that investors cannot rely on competing hedge fund indexes to obtain a ‘true and fair” view of hedge fund performance.
In an attempt to improve this situation, a team from the Edhec Risk and Asset Management Centre floated the idea of an ‘index of indices’. The Edhec team says that, since it is impossible to arrive at an objective judgement of which is the best existing hedge fund index, it might be better to use a combination of all the competing indexes.
This idea has developed into the Edhec Alternative Indexes, the first series of indices of indices in the alternative universe in March last year. The aim of the new index is to provide a benchmark that is more representative and stable than the hedge fund indexes currently available on the market.
One way to build an index of indexes is to simply compute an equally-weighted portfolio of all competing hedge fund indexes available in the market. Such a portfolio would inevitably be more comprehensive than any of the competing indexes themselves.
The Edhec team have extended the logic of this by using factor analysis techniques. Factor analysis is a mathematical tool which can be used to examine a wide range of data sets. Edhec has used it to extract the best possible one dimensional summary of a set of competing hedge fund indexes to design what it calls ‘pure style indexes”
The Edhec approach is to move beyond the idea of an equally weighted portfolio. The aim is to look for the portfolio weights that make the combination of hedge fund indexes capture the largest possible fraction of the information contained in these indexes. In other words, the objective was to reduce as far as possible the loss of information that would result from combining indexes.
The designers of the Edhec Alternatives Indexes say that they have achieved this aim. The Edhec Event Driven and Merger Arbitrage indexes, for example, capture more than 80% of the information originally available in a set of eight and four competing analyses respectively. The Edhec Fund of Funds index also has an information loss of only 9%.
The design of the Edhec Alternative Indexes was done in stages. The first stage was to establish a list of manageable indexes. The Edhec team eliminated alternative strategies for which fewer than four competing indexes were available They then eliminated strategies with a narrow focus - for example the various sectors such as healthcare. This filtering left a list of 13 investment styles with between four to nine index providers for each style.
The second stage was to pick the index for each strategy to be included in the Edhec Alternative Index itself. To be included, a strategy had to have transparent methodology and management principles. So, in effect, it was a condition of inclusion that the selected indexes were publicly available.
Finally, some ‘backfilling’ of data was necessary, Backfilling enables index providers to acquire data from historical trends that they could not acquire in real-time. The historical data period for the Edhec Alternative Indexes began officially in January 2003. However, to be able to use a longer range of data, it needed to backfill from the historical data of the indexes.
Since not all the competing indexes have historical data that goes back far enough, the Edhec team selected only those indexes that have published monthly performances since January 1994. These are Altvest, CSFB/Tremont, HF Net, HFR and MAR. This enables Edhec to use monthly performances from January 1997 onwards, taking into account the three years required for factor analysis.
From January 2001 onwards, the Edhec Alternative Indexes are constructed using the same method, but incorporating four additional indexes in their composition - EACM, Hennessee, Van Hedge and Zurich.
So how representative is the Edhec index of hedge fund indexes? To test the representative qualities of the Edhec Alternative Indexes, Edhec constructed an equally-weighted portfolio for each of the strategies from a proprietary database made up of 7,422 funds. The portfolios therefore contained more than 600 funds on average, and were considered to be relatively representative of their management universe.
They then calculated the correlation coefficient of those portfolios with the Edhec Alternative Indexes over the period from January 1998 through December 2000. The higher the coefficient, the more representative the index is.
As a last step, they compared this correlation coefficient with the average correlation coefficient obtained by the indexes of the Edhec Alternative Indexes. The results, say Edhec, demonstrate that Edhec’s index of indexes was systematically more representative than the average of the competing hedge fund indexes.