How to spot a housing bubble before it bursts

BIS private debt and property price databases indicate which countries are at risk of a housing bubble, including the potential ‘big daddy’ bubble of them all.

After a bubble has burst, no one denies that it existed. But before it does, the popular refrain is that though bubbles existed elsewhere in the world, “there’s no bubble here”. So housing bubbles are admitted to have existed in Japan, the USA, Spain and Ireland – because they’ve already burst.

But the rest of the world – and especially Australia – is different. House prices in Australia, the UK, and everywhere in between where they are still rising, are justified by … (fill in your favourite fundamental reasons here) and are not in any way manifestations of bubbles.

My unpopular alternate refrain is that house price (and share price) bubbles are always and everywhere a phenomenon of rising bank leverage, and anywhere that has high and rising levels of household debt (relative to GDP) is a candidate for being a bubble economy. This is a non-mainstream argument in economics, because for decades the so-called 'efficient markets hypothesis' and 'capital assets pricing model' have dominated academic economics, and they both argue that leverage has no effect on the price of assets.

I always thought these theories were delusional in the extreme, and anyone who believed them had to have swallowed some mighty potent Kool-Aid. But since Kool-Aid drinkers dominated academic economics, no-one even looked at the data. Often, it wasn’t even collected by statistical agencies.

The Bank of International Settlements, which has been a bastion of sensible economic analysis ever since Bill White was its Director of Research, has recently come to the rescue not once but twice. It has a database on private debt levels that finally makes it easy to compare the private debt levels of different countries and a database of house price indices. They’re not perfect databases: the debt database misses some countries (such as New Zealand), and I wish it included specific data on mortgage debt as well as data on household and business debt; and the house price data ranges from incredibly long term (as for Norway) to minuscule (as for Greece and France). The Australian data the BIS uses appears to differ somewhat from the Australian Bureau of Statistics series. But both databases are a vast improvement on not having any cross-country-comparable data at all.

So let’s take a look at the data: firstly on house prices across the world, adjusted for consumer price inflation rates. This raises an important preliminary issue: what should the relationship be between house prices and consumer prices?

If my memory serves me rightly, a point of contention I used to have with Chris Joye was about whether there was or should be a trend. He argued there was and should be one; I argued that there wasn’t, and shouldn’t be one. My empirical evidence at the time was the Herengracht Canal Index – a record of house prices along Amsterdam’s most prestigious canal dating right back to 1628 (nine years before the most famous bubble of all, the Tulip Mania in Holland). Prices rose and fell over enormous stretches of time, but over the 350 years of the data the average index was 200, and there is no trend.

To that I can now add the US data after its bubble burst. By 2012, the US index had fallen back to its 1890 value, having peaked 75 per cent higher back when Alan Greenspan was dismissing talk of a bubble. I think that Figure 1 makes a pretty good case that over the long term, there is no trend.

Figure 1: Really long term trend on real house prices

Graph for How to spot a housing bubble before it bursts

As an aside, to my amazement and delight I now find that Chris Joye is today warning of the dangers of a house price bubble in Australia. This is a very different tune to some time ago, which may well reflect Keynes’s famous aphorism that “When the facts change, I change my mind. What do you do, sir?”. I wonder whether he’s changed his mind on the long-term trend too.

Back to the BIS. The next four charts show the BIS house price data, adjusted for inflation, across 15 countries for which the BIS has both house price and household debt data. The same scale is used on all four graphs to make comparisons easier.

Graph for How to spot a housing bubble before it bursts

Graph for How to spot a housing bubble before it bursts

Graph for How to spot a housing bubble before it bursts

Graph for How to spot a housing bubble before it bursts

The only clear conclusion from here is one that surprised me. On this data, Canada does not have and has not had a housing bubble. Its price index is exactly where it was thirty years ago. I’ve just spent some weeks in Toronto, and its horizon-to-horizon condominiums and cranes seemed to reek of a bubble – but apparently not.

Austria and Germany can also be ruled out – though based on a much shorter time series – and perhaps France and Greece (though more likely in the latter case a bubble had burst in 2007). Bubbles clearly have burst in Denmark, Ireland, Japan, Spain and the US (though prices are rising in the US again – the BIS data ends in 2012).

That leaves Australia, Belgium, The Netherlands, Norway and the UK as possible bubbles as I define them: where it isn’t just a rising price level that matters, but whether that rise has been fuelled by accelerating debt.

Of all these possible bubbles, the Big Daddy appears to be the UK – a fourfold increase in real house prices since the late 1960s, with most of that occurring from the Thatcher days when the so-called 'Big-Bang' deregulated British finance markets and led to it becoming the casino it is today.

But the fairest comparison is in the next chart, which dates all indices from 1985 – since a difference in the starting point for an index can have a dramatic effect. Witness the US and Herengracht data in Figure 1: it isn’t that the Dutch had a much bigger bubble, it’s just that the starting point of 1628 was depressed compared to the long-term average, whereas the US starting date of 1890 appears to have been pretty much spot on the long-term average.

That brings us to Figure 6, which collates the data for countries where the BIS records go back to 1985 (with some additional data tacked on from Australia and the US). Now it’s clear that the UK indeed does have the daddy of all bubbles – and now the UK government is trying to reflate it with its “Help to Buy” scheme (which I think is more aptly called “Help to Sell”). Japan clearly had the fastest rising bubble of all and has since suffered the longest slide.

Graph for How to spot a housing bubble before it bursts

And Australia? We were up there with the best until 2004 (on the BIS data – the ABS data however implies we peaked in June 2010 in real terms), and we’re still a member of the 'More than Double' club, with our house prices 2.4 times their 1985 level as at the end of 2012 (when the BIS data ends).

So on the bubble front, thus far Australia’s house prices walk like a duck. Next week I’ll see whether they quack like a duck: can the price rises be attributed to leverage, or are they really based on "the fundamentals”, as my ex-colleagues in academic economics (Kool Aid fans) used to say?

Read today's piece from Stephen Koukoulas on housing market affordability: First home buyers are not locked out.

Steve Keen is author of Debunking Economics and the blog Debtwatch and is developer of the Minsky software program.