What Goes Up . . . Usually Stays Up

Back in May 2003 The Economist said that many countries were in the midst of a housing bubble:

A SURVEY in The Economist in May predicted that house prices would fall by 10% in America over the next four years, and by 20-30% in Australia, Britain, Ireland, the Netherlands and Spain. Prices have since continued to rise, so have we changed our mind?…

I’ve already discussed the US; in this post I’d like to examine some foreign markets.  Nick Rowe has an excellent post on bubbles, and he argues that Canada did not experience a housing bubble.  Before considering Nick’s assertion, take a look at this graph of average Canadian housing prices (you must click on the graph link on the right.)

Nick argued that if there had been a bubble then you’d expect that once prices stopped rising and began falling, then people would worry that the bubble had burst and prices would fall sharply.  Prices did fall a bit when the worldwide recession hit Canada in late 2008, but then began recovering in the second half of 2009.  So he concluded there was no bubble.  My hunch is that he is right, although as they like to say on Wall Street “past performance is no guarantee of future success.”

My theory all along has been that the US housing bubble was widely misunderstood, and that if NGDP had kept growing at a 5% rate after mid-2008 then the US housing bust would have been far milder.  More specifically, in heartland markets like Texas there would have been no housing bust at all.  Most of the damage would have occurred in 4 key sub-prime states, and even in those states the price declines would have been far smaller.

One implication of my hypothesis is that in most other countries the housing crash should have occurred later than in the US, and should have been far milder.  The Economist, which ironically is the publication that I most strongly disagree with on this issue, has published a graph that strongly supports my hypothesis.  If you click on the link you will see an interactive graph that shows global housing trends since 1990.  I found it easier to read by moving the starting point to 2000:1, and go up to 2009:3, the last observation.  And I used real housing prices, which should make it easier to spot bubbles in countries that have inflation.

Many countries experienced no unusual increases (Germany, Switzerland, Japan, Hong Kong, China, Netherlands, etc.)  Other countries like France experienced a sharp increase, but only a very small decline.  Australia and (to a lesser extent) Canada also fit this pattern.  Furthermore the timing of the decline is clearly associated with the 2008-09 recession, which reduced NGDP in the US, Europe and Japan.  Also note that where housing downturns eventually did occur, they were generally more severe in countries that had severe recessions (UK and Ireland) and relatively mild in countries where the recession was much milder (Australia and Canada.)  So it looks like we can say there was (at best) a housing bubble in the US, and to a lesser extent a few other countries like Spain and Ireland.  But no global housing bubble.

I believe that portions of the US market, and perhaps to a lesser extent the Irish and Spanish market, became overvalued in early 2006.  The other English-speaking countries also saw large increases in real housing prices, just as in the US.  But because they experienced much milder declines, and because those declines were closely associated with the severe global recession, I don’t see how you can call them bubbles.  A severe recession is a fundamental factor that would be expected to reduce housing prices even if the EMH explained 100% of house price changes.

While my falling NGDP story explains most foreign downturns, The Economist’s bubble story seems inconsistent with these inconvenient facts:

1.  Prices in most other “bubble” markets did not decline when the US market turned down in 2006.

2.  The declines in other markets were generally much milder, even when the preceding price increases had been just as rapid as in the US.

3.  Housing prices turned up recently in places like Australia and Canada, despite their being “overvalued” according to The Economist.

So you shouldn’t believe people who tell you that we are in a bubble, and who defend that hypothesis by merely pointing to fast rising prices.  Ex post, people like Krugman and Shiller were right, but only about the US.  In many other countries large price increases did not lead to a major collapse, and prices are still quite high in early 2010.

Of course if you wait long enough prices will eventually fall in any country, that’s how markets work.  But it turns out that it is much harder than most people imagine to predict which prices are “obviously overvalued,” and hence bound to fall sharply.  We all know about confirmation bias.  Because most elite economists live in the US, and pay little attention to countries like Australia, they tend to assume that recent events have provided decisive support to their bubble hypotheses.  Yet from a global perspective bubble theories haven’t done well at all.  Indeed they performed quite poorly over the past 10 years.  Outside of the US and Ireland, you would have been better off if you had ignored The Economist’s bubble warning of 2003, and instead had gone out and bought a house.  But what about those who base their forecasts on fundamentals, and not merely the rate of price appreciation?

If you look at the specific countries cited in The Economist’s famous prediction, the results are far worse than even for the US.  Recall that in May 2003 The Economist predicted 20% to 30% prices declines in 5 foreign countries.  If you set the starting point at 2003:2 and use nominal prices, then you will see that prices rose by more than 10% in the Netherlands, and by anywhere from 30% to 55% in other 4 countries.  Not exactly the 20% to 30% price drops they expected.  Some of my commenters claimed that The Economist was right in the long run, because prices did eventually fall.  I don’t think that argument is right—surely you must be held to the terms of your prediction, otherwise there is no way to ascertain how good you are at forecasting.  But even with those very generous assumptions, they were still far off base.  Even today the Netherlands continues to show a greater than 10% gain.  Spanish, British and Australian housing prices are still up by 30% to 50% over 2003 levels.  Only Ireland is even remotely close, with its price appreciation having fallen to about 10%.  But even that prediction is far from the 20% to 30% price drop forecast by The Economist.  The most generous assumption of all would be to use real housing prices.  But even in that case none of the 5 showed price declines over the following 4 years, and only Ireland experienced a tiny decline between 2003 and late 2009.

So The Economist’s predictions of 10% to 30% price declines over 4 years were spectacularly off base.  Even over a 6 1/2 year time frame, only the US (the country they thought was the least overvalued) experienced a significant decline in house prices.  And then only in real terms.  So in all 6 countries their predictions were wildly inaccurate for the 4 year time window they specified.  And even under the most generous assumptions, using real housing prices and a 6 1/2 year time frame, The Economist’s predictions were still highly inaccurate in 5 of 6 countries.  What an awful record of forecasting housing prices through the use of ”fundamentals.”  This shows just how difficult it is to identify bubbles in real time.