Indexes 2

Another way that I think about an index is just as another investment manager. Remember that the index is not the average, it's the point that the average would like to beat. The index follows a simple investment strategy that most other managers just can't beat. The realised results of the average investment manager are probably the best that can be expected when managing many billions within an institutional framework. With restrictions on the degree of activity and with the pressures to preform it may be nothing but expected. But why is the "index manager" so good and why does the average active manager have so much trouble keeping up? (note that something like 80% of funds under perform the benchmark). Clearly consistently beating benchmark is incredibly difficult but there may be some reasons which start to answer why this is.

I get the feeling that most investment managers are more interested in building their business rather than focusing on their profession (of achieving investor returns). As with any business, if you get the process right then over the long run everything else will fall into place in time. There will be less chance of mistakes and the business will find life much easier. Remember that a process is just a way of defining your average result. Instead, as I understand it, most managers are focused too heavily on their benchmark tracking error which deviates thinking away from the long term average and focuses it on the short term. To paraphrase Keynes, many investors focus on predicting the market's psychology (speculation) rather than predicting long term return on investments (enterprise).

Why is it that the index manager, who has a simple criterion such as market capitalisation, price or a simple set of fundamental metrics, so powerful? Looking at the behaviour of these managers might lead to some clues. Indexes generally have very low turnover (which translates into low costs) and they have no pressure to outperform. In other words they have a simple buy and hold strategy. Yet despite their lack of detailed analysis and skill they still outperform.

The low turnover point is important. The returns from the stock market do not follow a normal distribution- the returns have fatter tails and many more smaller fluctuations. In other words we are looking at a leptokurtic distribution (i.e. excess kurtosis). I was reading the other day how the returns from the S&P from 1978 to 2005 were 9.6% (annualised) . If we knocked out the worst 50 days our returns would be 18.4%. If we knocked out the best 50 days our returns would have been just 2.2%. So I decided to run a similar test using the daily returns of the FTSE All Share from 10-Jan-1977 to 4-Jan-2008. The results I got were pretty similar. I had an actual annualised gain of 10.2%. If I knocked out the worst 50 days I got a return of 17.6%. If I knocked out the best 50 days I got only 4.3%.

This makes it pretty clear that the tail events have a lot of impact on returns and it highlights how important a longer term holding period is in order to catch all of the tail events. I'm not sure exactly how this will impact many of the standard pricing models but I would imagine that as they assume a normal distribution they could be problematic because they only work "most of the time". It seems that the tail events have a lot more to answer for than what I am told.

BTW, if you're interested in the spread sheet then just ask and I'll send it over.