Share index and survivor bias

BY MARY HOLM

Q: A straight comparison of a share index change over a long period, as discussed in last week's column, may be less significant than it appears, because of "survivor bias".

For example, suppose the NZSE40 doubles in a 10-year period. With some simplification, this means that the shares that made up the index at the end were worth double those in the index at the start.

But some companies (many at some periods of history) would have gone broke and dropped out of the index during those 10 years, to be replaced with other survivors. Thus it is a measure of the performance of the survivors, not the whole market.

It is true that if you had been clever enough to invest only in those shares which remained successful for 10 years you would have doubled your money.

But it doesn't tell you as much about what a real investment, which would have included the good, the bad and the ugly if you were safely diversified, would have done.

For term deposits (which hardly ever fail), and for bonds (which sometimes fail), the quoted 10-year return is more likely to be what the average punter would have received.


A: Interesting point. But I'm afraid you're wrong.

Survivor bias can happen if you take a group of current investments and see how well they've done in the past.

For instance, if we looked at all the currently available New Zealand share funds that have been around for at least five years, we could come up with an average performance over that time.

The trouble is that any share funds that closed down during the five years would not be included.

We'd be looking only at the good ones, the ones that survived. That's survivor bias.

The same thing would happen if we took the shares in the current NZSE40 and looked back at how they have performed over the past five years, 20 years or whatever.

Some shares would have grown tremendously over that time, after starting out tiny. And there would be none of the "bad and the ugly", as they wouldn't be in the current top 40. Performance would be great.

But we don't do that. As you rightly say, the bad performers drop out of the index. Its composition is changed monthly to reflect changes in weightings, and quarterly to reflect changes in the biggest 40 companies on the Stock Exchange.

What you're ignoring is that as a company falls from, say, sixth biggest company to 41st - and hence drops out of the index - its share price will be falling and pulling the index down.

If lots of the shares fall fast at once, the index, too, falls fast.

You've only got to look at what happened to the NZSE40 after the October 1987 share crash to see exactly that happening.

You might protest that the index covers only the bigger end of the market. So it gets rid of the dogs when they're 41st biggest, before they become totally worthless.

But, by the same token, it doesn't include the tiny companies that grow to be huge until they've done a fair bit of their growing, and become 40th biggest.

There is an index, called just the NZSE Index, or All Index, that covers the whole market, and so includes both the dying dogs and the upstarts.

We probably should quote that index more often. The drawback is that it doesn't go nearly as far back as the NZSE40 and its predecessors.

One final point: regular readers will know that I favour index funds, which hold the shares in a market index.

They have several advantages, such as lower costs, fees and taxes.

If what you are claiming is true, they would also benefit from survivor bias. That would make them not merely the best way for most people to invest in shares, but by far and away the best way.

Unfortunately for index fund investors, there's no bias there.

* Mary Holm is a freelance journalist and author of Investing Made Simple. Send questions for her to Money Matters, Business Herald, PO Box 32, Auckland; or email: maryh@pl.net Letters should not exceed 200 words.