QAIB 2008

Quantitative Analysis of Investor Behaviour (2008)

This post is longer than I would like, and I will break it up at a later stage. I basically just wrote thoughts down as I read through the DALBAR study. The study looks at the 20 year period ended December 31 2007, and the equity section focuses on investments in the US through S&P Index.

The intention is that the QAIB be used to establish benchmarks. It further suggest that you can contact DALBAR for further information on using the QAIB benchmark and making comparable calculations.

The benchmark attempts to simulate the behaviour of the 'average investor' by tracking net inflows/outflows into Mutual Funds. This allows them to establish a pattern in which a $10,000 investment was made relative to that of a systematic investment of $10,000 over time.

Does the Dollar Cost Averaging/Systematic approach assume level contribution in nominal or in real terms?

Most mutual fund companies use the return of a lump sum invested at the beginning of an investment period. This only really works for someone who starts at the beginning of those periods and makes one contribution. Perhaps there should be three measures:

1) Lump sum from start of period
2) Annual Contribution from start of period
3) Monthly Contribution from start of period

The issue then becomes, what are the regular contributions? Do you use $1 per period, or do you allow for inflation. For Domestic Funds, you could allow for inflation. For Global Funds it becomes a little more difficult. You could benchmark it against the same currency benchmark as the fund's benchmark and use weighted average inflation measures.

The results show the average investor over the last 20 years having gained 4.48% vs. 11.81% performance of the underlying S&P index. This is often used to suggest a passive investment strategy and regular contributions. I think those are separate considerations.

1) Do you remove your decision making from when to invest?
2) Do you believe your chosen asset manager can beat a passive index?

The big problem is that both questions are not easy to answer and overconfidence would lead people to answer Yes to both. My feeling is that it is possible for a manager to beat an index, but finding those managers is tough. Partly because the best ones seem to make a lot less noise than the average ones. I think the first question is a more difficult one to claim that you can do.

The Guess Right Ratio

The next measure that is a useful suggestion is one of how often an investor makes the write call. So, if there are inflows and the next month has positive returns, they score 1. They also score if there are outflows and the market goes down. The average score over the period under investigation was 61%. Naturally, the period under investigation was an incredibly long bull market. It will be interesting to see how next years results to the end of 2008 look! If you invest every month, you would expect to have a ratio greater than 50% in a rising market.

Perhaps a better measure would be to divide the `Guess Right Ratio' by the % up months? If the market went down, so you were wrong more often, the base would also go down, so your `Guess Better than Randomly Ratio' would stay close to 100%. If you destroy value, it would fall below and if you are adding value it would go above 100%. Both the GRR and the GBTRR aren't weighted by money, so you wouldn't be able to use them to check consistency with actual value added or destroyed.

Holding Period

`The holding period reflects the length of time the average investor holds a fund if the current redemption rate persists'

This measure is useful if you assume that it is more likely the investor will share the funds returns if they stick around and don't do much. That seems intuitive.

The paper gives a methodology for calculating the investor return which can be applied to markets other than the S&P. I would like to see if others have taken DALBAR's work and applied it to other markets, or have any critical comment on the way they have done it. I need to let what they have done wallow in my head for a while to think of my take. First take is that it provides a very useful start to the process of providing measures of client behaviour.