Correlation or causation?

A common mistake in investing, and in the broader economy, is to confuse correlation and causation. Correlation means that two phenomena tend to go hand in hand. Say, a recession and higher unemployment. Or high ice cream sales and drowning deaths (thanks for the example Wikipedia).

Causation, on the other hand, means we go so far as to say that a change in one variable caused the other. Sometimes that's right. But often we assume we understand causation when we've merely identified a correlation.

If I say that high ice cream sales caused an increase in drowning deaths, the mistake is obvious. But perhaps you'll believe me when I say that a recession caused the unemployment, when the reality might have been the other way around. Or they might have independently moved in the same direction. Or perhaps a little slip in one started a trend where they're both feeding off each other.

As a registered member of the Australian house price bears' club, I'm always coming across what I believe is a confusion of correlation and causation on the topic. Take this recent article on BusinessDay websites across the country:

Real estate slump will leave banks in pain, too

I don't want to single out Ian Verrender, he generally provides high quality commentary, including most of this article. But there's an all-too typical misunderstanding in there, one you can find most days in the newspaper. When talking about the potential for a bursting house price bubble in Australia, Verrender says:

'Given we are approaching full employment and with our terms of trade at record levels, the chance of that happening are slim, in the near term at least.'

In my opinion, and in the opinions shared in some of the comments below the article, this statement mistakes correlation for causation. It suggests that a strong economy caused the high house prices (causation). Ergo, house prices can't fall substantially unless or until the economy comes off the boil.

Few would argue that there's a correlation between high house prices and a strong economy, although I haven't crunched the numbers to see how strong the correlation is. But I agree it's unlikely that we can have a significant house price correction without a weaker economy.

But it's not a one-way street. Rising house prices—fuelled by a borrowing binge—can strengthen an economy (at least apparently), boosting construction and retail spending and creating a positive feedback loop, until something (anything) upsets it.

Similarly, overpriced houses—or any other overpriced asset for that matter—can run out of steam by themselves. They have a tendency to eventually self-correct. The process of self-correction, on its own, can lead directly to job losses in construction and the real estate industry, and potentially second order job losses in the retail sector as consumers shut their wallets as house prices fall (negative wealth effect), and further weakening as a real economic slowdown and further house price falls contribute to another positive feedback loop, this one on the downside (until it too stops).

Of course, it might not work this way in 2011. House prices mightn't fall at all. Or they might fall because of an exogenous shock—the sort that most pundits believe is the only risk to house prices. There is a lot to worry about in this area, and most concerns point to China.

But I can also envisage a house price correction coming first, followed by a slowing domestic economy. In fact, as the first commenter below Verrender's article points out, that's what seems to be currently occurring in Perth and Brisbane. The US housing and economic slowdown also seemed to work in that order—house prices started falling before the economic slowdown kicked in.

House prices and economic conditions are a little like the chicken and the egg—they generally go together, but it's hard to know which came first. Sometimes it's one, sometimes the other, and sometimes we can't know. That implies correlation only. Those who confuse correlation and causation are missing the bigger picture. And they'll only recognise the bursting of the bubble once its shows up in the unemployment numbers.