I’m working on a neat series looking at the historical percentage of GDP derived from residential construction in different provinces and connecting that to the change in the percentage of the workforce employed in construction over time. It might give us some hints on how a potential housing correction might weigh on economic and job growth. Stay tuned for that.
In the meantime, I’m going to throw out two charts I’ve created in the hopes that some of our astute readers might make some connections for me. Tell me what these graphs say to you, and what are the potential implications…if any? These are all derived from Stats Canada data.
Exhibit 1: Inflation adjusted investment in residential structures in Canada.
Exhibit 2: Cumulative change in real house prices and CPI deflated mortgage debt
I’ll take a break doing the analysis for tonight and instead turn it over to my readers. Have at ‘er.
Also note that the website on the graphs is…
This is my new website. It is running and functional, but there are a few bugs to work out. All content should be ported over within a couple weeks. Check it out! It is pretty snazzy if I do say so myself.
Cheers,
Ben
Financial Insights 
Thanks, Ben… the coming analysis sounds very interesting.
regarding Exhibit 2:
It initially seems too large, but does the difference between the two plots represent housing backed debt spent into the economy that has NOT resulted in housing price rise (in other words, RE_ATM spending)?
As I say, seems like too large an effect… frightening if it is so.
Hi VREAA
The numbers are right. It’s not as shocking once you look at the rise in mortgage debt relative to house prices in the US and Australia:
http://www.debtdeflation.com/blogs/wp-content/uploads/2011/03/Keen2011AustralianHousePrices.ppt
I just realized that I left the years off the x axis. I’ll update that shortly.
Investment in RE should be increasing due to pop growth and real wage growth but I’m unsure how to normalize it to your first graph. I expect your upcoming series looking at re as % of GDP will be revealing, as will changes in construction employment vs this %age.
The mortgage debt and price time series is interesting. Notice the timing of the start of the deviation. Now look for what changed starting then.
I generally agree. Population growth has been relatively stable over decades, so why the sudden jump in the early 2000s? Is it a reflection of under-investment through the 90s and then a subsequent ‘catching up’? Also, by the early 2000s, the boomers, the largest cohort in our population, were aged 35-55, well outside the normal first time home buyer age. I’m a bit perplexed at the fairly sharp increase. I’d be equally inclined to look at the associated loosening of mortgage credit standards over the same period.
Remember what you said about credit availability and the formation of a bubble. Glass-Steagall was repealed in 1999 and this had a huge impact on not just the availability of credit, but the ability to deliver it right into neighborhoods as well as tremendous competition to do so.
I was looking at the timing of that divergence too. The divergence in Australia occurred at the exactly the same time. What changed in 1990 that both countries have in common?
The only thing I can come up with is 1990 was the peak of interest rates (for both countries) with mortgage rates running 17%.
Interest rates came down so quickly after 1990, and to such low levels relative to the previous twenty years, that consumers may have become much more comfortable with debt.
I don’t know much about Australia but 1990 was about the time the Canadian federal government, due in part to higher rates, started to swing a structural deficit into a surplus. Interesting how when government deficits were decreasing mortgage debt was increasing. The two are now both increasing.
The next slope change is around 2002 when rates were low. As rates were increasing in the latter half of the decade mortgage rules were relaxed most notably with the introduction of the 5/40 government-insured mortgage. Tax code changes may have also played a role.
I would like to be able to explain the coincidence. Canada only started reducing its deficit in 1993 under Jean Chrétien.
Taxation changes may play a role. GST was enacted in 1991, but I doubt it can take all the blame.
Taxation changes have gone beyond GST. Cuts to income taxes have also occurred and the GST was lowered by 2% a few years’ back. Governments are now subtly starting to claw back the other direction (well out in BC anyways).
We can add both private and public debt together to get a sense of the nation’s total debt load. (I don’t know how to include corporate debt into this.) While the government was getting its accounts in order, Canadians were levering up, so, without knowing more about it, I would suppose some of the burdens once carried by government have been partially shifted to households in the past 20 years.
Do these series extend further back in time? If so they should be extended as far back as possible.
That’s as far back as the Stats Canada series goes.
I wonder what these charts would look like using a log chart? Regardless, these charts show that since around 2000, mortgage debt has been more prevailent (which makes sense with the CMHC standards), which has been atributed to a rise in the real house price. And of course with any investment, the longer it is stable (ie. rising prices), the more risks people are willing to take, and we have seen people become more and more comfortable with real estate as an investment, so investment $$ has gone into real estate as it was rising, people panicked in 2007-2008, then continue to invest as real estate values stablized and people regained confidence in the market. The longer a market is stable, the more risk people are willing to take, and we can see by the constant flow of money as investment dollars into the market, people are very comfortable holding real estate as an investment. Also the sign of a market that is overextended and due for collapse.
Ben, where did this data set come from? Is this NHPI data? I ask because, as I wrote a couple of days ago, over the past 11 years the TeraNet index has gone up 100%, but your series only goes up 73% since 2000.
It is interesting to note that, as I wrote about earlier, in the 30 years covered by your data set, house prices are only up a little less than 100%. Inflation over the period has averaged 3%. House prices should be up 150%.
Compare this with Australia where house prices are up over 1,000% during the period. During roughly the same period Australia’s mortgage debt is up something more than 2,500% compared with our 400%. Canada’s economy is 28% larger than Australia’s. It would be very interesting to see how large these numbers are relative to their economies.
Which data set in particular? The first data set is CANSIM table 380-0010. The second has been pieced together using CPI data, CREA resales, and CANSIM table 176-0069 which highlights outstanding mortgage credit by large private lenders.
Double checked and the numbers seem to include all outstanding mortgages…..the graph is legit.
With regards to your facts on Australian house prices and debt, I’ve had a look at what Steve Keen has produced and it doesn’t seem to reflect what you are saying. Keep in mind that this graph indicates real prices and real increase in mortgage debt, not nominal. Are you sure you weren’t referencing the nominal charts when comparing Canada to Australia. Their debt bubble is certainly larger than ours, but I haven’t seen anything that suggests it is as large (relatively) as you suggest.
I may be. I’ll double check.