PsyFi Search

Thursday 16 April 2009

Survivorship Bias in Magical Mutual Funds

The Magical Inflation of Mutual Fund Returns

Despite years and years of data showing that, on average, actively managed funds underperform their passive brethren by around the amount charged in fees – a point William Sharpe demonstrated using the most basic math back in 1991 (The Arithmetic of Active Management) – the darned things won’t lie down and be slain. Worse still, they use their excess fees to dream up smart new ways of improving their apparent returns.

Perhaps the cleverest and simplest method is to magically make the worst performers disappear from the record, a clever trick I wish I could perform with my portfolio. However, just like the investors in those disappeared funds, I’m stuck with my returns while the funds benefit through survivorship bias.

Survivorship Bias

Survivorship bias is simply the inflation in apparent returns due to the fact that some investments fail utterly and disappear from the record. For individual stockpicking investors this obviously damages their actual returns but when we look at the records of indices and funds these failures simply don’t appear. This causes their apparent returns to be biased upwards.

Survivorship bias applies at all levels in investment. At the most extreme level the returns quoted on stockmarkets across the 20th century only really cover those first world nations lucky enough to have traversed the century without a total breakdown in their economies. Investors in pre-Leninist Russia or pre-Peronist Argentina wouldn’t recognise the figures we idly bandy about here.

Similarly investment returns quoted for specific stockmarkets generally ignore the disappearance of individual stocks. Investing in index trackers neatly gets around this problem, since survivorship bias is built into the returns (the index is the index, after all). Individual stock pickers, however, need their survivors to outperform average market returns in order to even equal the indexes.

Mutual funds, however, are portfolios of stocks. They don’t go bust or get taken over by mad populist politicians or frenzied anti-capitalist philosophers with a point to prove. How in the world do such things get to exhibit survivorship bias?

Vanishing Funds

Well mutual funds don’t exist to serve their investors, they’re creations of the fund managers and the point of them is that they generate fees and attract funds from new investors. Failing funds do neither – they manage less and less money, generating less and less fees while a rating at the bottom of their sector doesn’t do much for garnering new money. Even most mutual fund investors aren’t so foolish as to back an active fund manager who’s “obviously” not much good.

So the fund managers perform their magic trick and merge the underperforming fund with a more successful fund and roll over the investors. For some reason that completely escapes me they don’t then adjust the historical record of the merged fund to a blended performance but simply keep the better fund’s numbers. The laggard fund’s numbers simply ... vanish.

The effect of this sleight of hand is to magnify the performance of mutual funds in comparison with index trackers, which rarely have any reason to disappear. There’ve now been lots of findings to this effect – see, for example, the Savant Capital / Zero Alpha study.

This study suggests that there are virtually no sectors of the market over any reasonable time period that don’t suffer from survivorship bias where the performance of the surviving funds is exaggerated by the silent slaying of the underperformers – in some cases by up to 2% a year, in excess of the normal underperformance. This may not sound like much but when you start to compound it up over many years it can end up making a startling amount of difference – over 100% over a decade in one case quoted.

The Lost Funds

The numbers of funds that simply go missing is astonishing. In A Random Walk Down Wall Street Burton Malkiel found that 15% of US funds disappeared between 1988 and 1991. In the UK a study by Luke Schneider showed that of 1008 funds available to the private investor in 1992 only 491 were still trading in 2003. This is not a minor problem.

Incidentally Schneider also showed that the average mutual fund investor even managed to significantly underperform this debased benchmark by around 2% per year through overtrading. Investing in active funds carries risks other than the obvious one of picking dubious funds, it also brings with it a mindset that it’s possible for an individual to outperform the index. Normally it isn’t.

This survivorship effect, added to the impact of active manager funds, drives most low cost index trackers (do watch out for those indexers with TER’s over 0.5%) up into the top performance echelons of their sectors. Even the long quoted outperformance of small cap active funds over index trackers turns out to be a figment of the data.

Instant History and Other Biases

The manipulation of survivorship bias to enhance relative and apparent performance isn’t the only trick pulled by active funds. After all, they have lots of money to spend on smart people to think up clever ideas once they’ve figured out that doing something as apparently simple as actually outperforming a benchmark is fundamentally hard. My favourite is something called “instant history bias”.

Instant history bias is what you get when you launch a fund that you’ve preselected to have an excellent track record. To do this you launch a series of funds, without marketing them to the public. At the end of some period of time – a couple of years, maybe – you pick the fund that’s done the best and ditch the rest. You then launch the new fund with a ready-made top notch prior performance record, spend a load of money advertising it and watch the funds come flooding in. Easy.

There are other types of bias that creep in as well. Watch out for easy data bias, which arises when specific low points in the market or a fund’s history are used as the starting point for performance figures. Then there’s backfill bias when a new manager is added to the lists of mutual funds reporting. Oddly enough this only happens when the new manager has a good record. Guess which way that changes the statistics?

Redrawing the Map

Way back when magic was first used, in the days of dungeons, dragons and damsels in distress, heroes were usually equipped with maps. These were often usefully annotated with signs in large, friendly letters which said helpful things like “Watch Out, Ogres About”, “Beware Of The Basilisk”, “Here Be Dragons” and "Warning, Wizard's Wand Testing Ground".

In our more modern, financial version of the map maybe it’s time to add a new type of warning for a new type of magician: "Be Fearful, For Here Be Fund Managers".

Related Posts: Fundamental Indexing Can't Save You From Aliens, You Can't Trust the Experts with Your Investments, Don't Give Index Trackers The Bird

Back to: The Psy-Fi Blog

1 comment:

  1. Great summary. While advertising is relatively cheap compared to returns, mutual funds aren't going to go away.

    I think another trick they'll use in the long-term will be to turn into life-cycle/wealth managers, that hide the poor performance of various elements of their investing by telling investors they're smoothing returns etc. (Similar to With Profits funds here in the UK.)