To what extent is past performance a guide to future performance?
If you go to a top restaurant, you might get a dodgy scallop one visit in 100 but the odds are you will get the fine dining experience that you expected. But what about investment? Billboards, newspapers and financial magazines are full of statistics on what has done well in the past but should we read much into this?
Last May, Vanguard published a paper called, Shopping for Alpha: You Get What You Don’t Pay For, which explores the relationship between past performance and future performance and then looks at other features of successful funds.
They first looked at a Morningstar database of actively managed funds in the US for a 20-year period from 1990 to 2010. They then calculated alphas (the portfolio’s risk-adjusted excess return versus its effective benchmark) relative to the stockmarket’s four common risk factors, namely:
- Market risk
- Small cap risk
- Value risk
- Momentum risk
For each rolling 36-month period, they grouped the funds in quartiles from lowest alpha to highest alpha and calculated the probability that the same fund would stay in the top 25 per cent over the next one, three, five and 10-year timeframes.
The result was that the probability of the highestalpha funds remaining in the top 25 per cent was no more than the result that chance would give.
In fact, the results were noticeably worse when survivorship bias was taken into account. Fund management groups are well known for closing their poor-performing funds or merging them with better funds to improve their published performance.
When these funds are taken from the overall performance figures, it makes active managers look significantly better than the true picture.
Vanguard also examined whether other factors might be a better guide to the future. Looking at Morningstar ratings, they looked at one-star (the “worst”) up to five-star (the “best”) funds using data from June 1992 to August 2009 to see if these helped identify future top-performing funds over the next 36 months. The result was that these star ratings were no guide to the future either.
They then looked at how other variables might impact on performance. Variables researched were:
- Fund expense ratios
- Portfolio turnover
- Fund asset size
- Fund age
The conclusion here was that, on average, for every 1 per cent increase in expenses, alpha declined by 0.78 per cent a year.
In other words, there was a significantly negative correlation between expenses and alpha. This comes as no surprise. If a fund manager spends an extra 1 per cent in costs, an additional 1 per cent return has to be delivered consistently just to break even.
Portfolio turnover rates produced a similar, albeit less pronounced, result with every 1 per cent increase in portfolio turnover giving a 0.22 per cent drop in alpha.
In other words, trading within a portfolio was also destroying alpha. The size of a fund and the age of the fund did not produce statis-tically significant results.
So our message would be that past performance is no guide to the future but that fund costs are.
A sensible approach would be to put together an asset allocation strategy that is right for you and then within each asset class or geographical region look for the lowest-cost, best diversified fund that you can find.
Investment should be boring and might only bear results over the long term – if you want excitement and a short-term kick, go to Vegas.
Jason Witcombe is a director of Evolve Financial Planning