Monday, December 20, 2010

Heart attacks and housing bubbles

I was rereading Malcom Gladwell's "Blink" this week, and it occurred to me that his theory of "thin slicing" is an interesting way of looking at Australia's housing bubble. For those that haven't read the book, Gladwell's main argument is that in a wide range of fields -- from psychology to police work -- we can often make better judgments by training our minds to instantly focus on the most relevant facts—and that less input (as long as it's the right input) is often better than more.

One of Gladwell's examples is a cardiologist named Lee Goldman, who has developed a decision tree that, using only four factors, evaluates the likelihood of heart attacks much better than most trained cardiologists in emergency rooms. When a patient walks into an emergency room complaining of heart trouble, the typical cardiologist, says Gladwell, will logically and systematically find out as much about the patient as he can in order to make a quick diagnosis. Is the patient under stress? Does he smoke? Does he exercise? Is he overweight? Does he have a history of heart trouble? But interestingly, Goldman's research suggests that most of this information is next to useless.
All you need is the evidence of the ECG, blood pressure, fluid in the lungs, and unstable angina....that extra information is more than useless. It’s harmful. It confuses the issues. What screws up doctors when they are trying to predict heart attacks is that they take too much information into account.

There are many more interesting examples in the book about how experts can often go badly wrong by trying to take too much information into account, thereby losing the forest for the trees. Now, before you wonder where the hell I am going with this, the IMF's recent report on the Australian housing market strikes me as a classic example of the exact phenomenon Gladwell is taking about. Confronted with overwhelming evidence that Australia is in the midst of a massive speculative housing bubble -- and an army of property lobbyists and government officials in denial about this patently obvious reality -- after taking all the evidence into account, the IMF has concluded that the massive rise in Australian house prices is mostly explained by fundamentals.

Like the clueless emergency room doctor, the IMF has ignored the patient's vital signs -- in this case, simple metrics like the amount of mortgage debt to GDP or the price of housing relative to median incomes -- and instead devoted all its time to examining a number of extraneous factors such as interest rates, the terms of trade, and population growth. Meanwhile, the patient is having a coronary.

So, in the spirit of Gladwell, let's forget about Australia's population growth, supply constraints, the mining boom, low interest rates, platypuses, and Kevin Rudd's hair, and focus on what's REALLY important in determining whether the current level of house prices is justified. The key determinant of affordability in housing is the relationship between prices, rents, and incomes. In fact, you could go so far as to argue that nothing else really matters all that much.

Why? Because houses are financed primarily by borrowing, and these mortgage loans have to be serviced out of people's income. If house prices grow faster than incomes for an extended period of time, then eventually, the amount of debt required to finance the purchase of a home becomes unsupportable. Simple as that. The result of this is that prices inevitably crash. Similarly, if house prices are growing faster than rents for an extended period, as they have been in Australia, buying will become increasingly unattractive relative to renting over time, until we reach a tipping point, and, again, the market crashes.

Now, as Keynes once said, the market can stay irrational longer than you can remain solvent. But it is clear that over the long-term, the relationship between incomes, house prices and rents inevitably reverts towards an equilibrium. Like the price-to-earnings ratio in the stock market, the median house price-to-income ratio, and the price-to-rent ratio will not tell you when a bubble is going to burst. But in the long term, they act as a pretty reliable anchor for prices.

See the chart below from this article by the IMF's Prakash Loungani.

So how far out of whack are prices in Australia? In the chart below, again from the IMF's Prakash Loungani, you'll notice that at the end of 2009 Australian house prices were more than 70% overvalued according to the long-term price-to-rent ratio, and almost 50% too expensive according to the price-to-income ratio.

Interestingly, according to the price/income metric, US housing is now "fairly valued" -- which is consistent with anecdotal evidence that prices in many US cities appear to be finally bottoming.

Source: Calculated Risk

Now, in response to such a simple argument, the property spruikers will no doubt try to bamboozle us with reams of statistics about supply shortages, population growth, and so on and so on. But again, none of those "fundamentals" really matter if households are already tapped out with too much debt to force prices any higher.

If you really don't think price-to-income ratios are important, and that there is still room for growth in Australian house prices, lets perform a quick reality check with the chart below.

This chart forecasts the future path of the price-to-income ratio in Sydney (currently the world silver medalist at 9.1) for various scenarios of capital appreciation over the coming two decades. You can see that if house prices were to continue growing at 8%, by 2030 the median house would cost almost 20 times the median income. Just for perspective, this is equivalent to having to pay $2 million for a house if you make $100,000 today. Clearly a financial impossibility.

If we assume annual wage growth of 4%, then 4% annual price appreciation over the coming two decades -- which many investors would probably regard as a poor result -- would merely preserve the current state of hyper unaffordability. In fact, to get the price-income ratio back down below a reasonable level of 3 would require average annual price declines of 2% for the next two decades. How many negatively geared investors are prepared for this type of scenario?

Yes, I think the patient is having a heart attack.



  1. FF,

    I would be interested to know if your OECD Price To Rent ratio for Australia is based on ABS figures for rents, which I believe to be the case. If so, I think you will find they are simply wrong.

    ABS measure a different set of "dwellings" for house prices and for rents. Weird, but true.

    It is simpler to use the gross rental return expressed as a percentage of price. On this basis the price to rents ratio is about at its 30 year average.

    Happy to substantiate if required, I have spoken to ABS, RBA and others about this. None disagree.

  2. Nic - Thanks for reading. I'm not sure where the OECD gets its figures on rents from. I assume it's the ABS as you say. Where are you getting the gross rental return figures from?



  3. FF,

    REI quarterly (though I think they have stopped now), corporate proprietry data and personal experience.

    I noticed recently that Home Price Guide (now Domain Data) are publishing gross yields for houses and units though I don't think they go very far back.

    The advantage of gross yields is they track prices, which is what rents in fact do. The disadvantage of the ABS conclusions from their own data is that they do not use the same definition of "Dwelling" for their price and rents series, they are two different data sets.

    This, to me, is a fundamental error and it has found its way in to the Australian House Price debate. This is not the first time that incorrect ABS figures impacted national policy, for several years in the late 1990s (I think) they understated immigration by about 40%. Oops!

  4. " let's forget about Australia's population growth, supply constraints, the mining boom, low interest rates, platypuses, and Kevin Rudd's hair"

    But what about the hairy nosed wombat?

  5. Nic > I'll add this to my list of things to look into. I'd be pretty surprised if the correct measure was at its 30-year average given that every other measure you can think of (i could have looked at mortgage debt to income or any number of other metrics) points to significant overvaluation. But I'm certainly open to the possibility that the rent figures are wrong and welcome the feedback...

    Heyworth > Indeed, perhaps the IMF needs to redo its analysis!