The tennis ball theory of the Canadian housing market

From Nick's absolutely wonderful post:

Real bubbles are unstable; they burst when you prick them. They don't spontaneously revert to their original size. Soap bubbles aren't like tennis balls. If the bubble metaphor means anything, it has to mean that. If asset price bubbles aren't unstable, and don't burst when you prick them, or re-inflate immediately, then the bubble metaphor is useless.

Below the fold, I'm going to provide some context to support the claim that the Canadian housing market looks more like a dented tennis ball than a burst bubble.

In a recent speech, the Bank of Canada's David Wolf offered these remarks:

Canada has not had the over-investment in housing that has created the need for painful adjustments in the United States and some other countries. The steady growth in housing investment (that is, the construction of new housing units) in Canada over much of the past decade was due largely to strong employment and sustained income growth supported by rising commodity prices. There was also a degree of catch-up from under-investment during the 1990s.

Emphasis added. For the most part, discussions about whether or not the Canadian housing prices were generated by a bubble during the 2000's haven't paid much attention to what happened before. This is probably a mistake. Here is a graph generated from The Economist's very useful gadget:

Although the 2000's was a good decade for the housing market, it followed an extended period of soft prices and weak demand following the housing bubble of the late 1980's.

Of course, the notion of a national housing market is a bit misleading. So let's take a look at three cities that play major roles in narratives about the housing market over the past decade: Vancouver, Calgary and Toronto. For each city, I've tracked down data for 'newly-completed and unoccupied housing' (Cansim Table 027-0010), which appears to be the best measure of inventories that I'm likely to find. These numbers are available for houses (detached and semi-detached) and multiple units, and I've graphed them against the Teranet house price index.

First up is Vancouver:

Throughout the 1990's, there was a significant excess supply of housing, and this kept a lid on prices for the better part of a decade. Once that inventory had been worked off, prices started to increase. Inventories rose as the recession hit, but to an extent that was nowhere near as important as during the mid-1990's. The reduction in prices seems to have been enough to absorb that extra supply, and now prices are on their way back up.

On to Calgary:

The sharp increase in prices in 2005-06 and the subsequent fall looks more like a standard story of supply and demand than a bubble. Tens of thousands of people moved to Calgary, exhausting the available inventory and driving up prices. Construction companies responded to the increase in prices by building more housing, bringing prices back down.

And now Toronto:

The data start two years after the peak in 1990, and we see that if there ever was a story of an overhang depressing a housing market for a long period of time, it is Toronto during the 1990's. The Teranet index isn't available for much of this period, but we can assume that prices had to be low enough to absorb all that excess inventory. Compared to the earlier part of the sample, current levels of inventories can hardly be considered to be consistent with what we'd expect after a bubble had burst.

And so it goes. Across the country, housing prices during the 2000's resembled a tennis ball that had been flattened during the 1990's and was returning to its usual form. So it shouldn't be surprising that after a sharp, brief whack - one much less severe than that of the early 1990's - housing prices are rebounding again.

Comments

[asp]

I think you forgot to put incomes on those graphs. And while you are at it, include rents.

[name]

And so it goes. Across the country, housing prices during the 2000's resembled a tennis ball that had been flattened during the 1990's and was returning to its usual form. So it shouldn't be surprising that after a sharp, brief whack - one much less severe than that of the early 1990's - housing prices are rebounding again.

Is that a prediction?

[Stephen Gordon]

No, it's an observation on the past few months.

asp, a couple of mouse clicks at that page on The Economist that I linked to will give you a graph of house prices against incomes.

[Mark]

I'm writing a briefing note on the housing market myself and this is very helpful. I've never come across the unoccupied dwelling series you are using.

An even better LHS metric would be unoccupied dwellings per total households. Getting that for the CMA level, non-annual would involve a lot of calcualation/extrapolation. If not, some other base to normalize it for changes in the size of the market (stock/population/labour force). I think then you'd see an even more startling bust/post-bust comparison.

[Rogue]

If indeed it was never in bubble territory before, then it can’t be in a bubble now.

And if we take the implication of Nick’s point (that lowered rates are a central bank’s policy to offset lower incomes, and an offset for lowered nominal prices), then the BOC’s current accommodative rates have only kept people (who are good credit buyers) from becoming too frightened to go to market.

[Just visiting from macleans]

Interesting analysis. It raises a few questions for me to ponder when I look at the city data. Your Teranet house price index indicates a national weighting of Toronto (42%) Vancouver (23.6%) and Calgary (10%). So, I presume the heavier weighting on Toronto numbers (more gradual increase and not as steep as Vanc/Calgary - the unequal scaling of the graphs is a bit visually misleading) is also reflected in the flatter Economist index for Canada as a whole.

So, I would ask, if the real estate markets in all three cities are indeed not bubbles but rather balls, are they all tennis, or rather squash, tennis and lacrosse?

So, to sort the balls, wouldn't one need to look a little closer at their regional economies, and see if there are any anomalies in the past decade that have driven their projectories higher than might be normally explained (such as a recovery from depressed 1990's prices)?

Here's my non-economist take.

Toronto - it seems to me a fairly diversified economy, and destination of choice for the majority of immigrants to this country due to its diversity. Gradual increase in prices (tennis- fairly resilient)

Calgary - dependent on the health of the oil and gas industry, with housing prices generally following the commodity price index Stephen posted a few blogs ago (squash? - hurts like hell if you ever get hit by one, and doesn't bounce much once it hits the wall)

Vancouver - a building boom fueled in large part by the Olympics, and resources to a lesser extent /gateway to the Pacific (lacrosse - somewhere between a tennis and a squash ball)

My two cents.

[Declan]

It was bad enough that the Economist used StatsCan's CANSIM 327-0005 as a house price index, but at least they have the excuse of being a large international organization that neither knows nor cares much about the Canadian housing market.

As people who live in Canada, let's not compound their error by taking them seriously. I could write a long explanation of why that series is a measure of cost pressures in the suburban construction industry, not a housing index, or I could point out how nobody who seriously looks at Canadian housing prices (Bank economics departments, the folks at the Sauder school, the Central Bank, etc.) uses that data for the purpose of tracking housing prices, but maybe it's simpler if I just point out that, according to CANSIM 327-0005 (the data used by the Economist), if you bought a house in Vancouver in 1994 it is currently worth less, in *nominal* terms, than what you paid for it (i.e. after inflation you've lost about half the value of your house). Try running that one by anyone who lives in Vancouver!

That Statcan series is just not a housing index as we normally understand the term, and anyone who takes it as one is irresponsibly confusing folks like 'Just Visiting' who don't (and shouldn't really be expected to) realize that comparing that series to a true housing index (such as Case-Shiller or Teranet) is like comparing Apples and Hockey Pucks.

[Declan]

"In a recent speech, the Bank of Canada's David Wolf offered these remarks:

Canada has not had the over-investment in housing that has created the need for painful adjustments in the United States and some other countries. The steady growth in housing investment (that is, the construction of new housing units) in Canada over much of the past decade was due largely to strong employment and sustained income growth supported by rising commodity prices. There was also a degree of catch-up from under-investment during the 1990s."

---

You mean this David Wolf?

"The Canadian housing market is headed for a meltdown because households finances are in even worse shape than in the U.S. or the U.K., said David Wolf, chief strategist Merrill Lynch & Co. Canada.

Wolf said in a report published today that Canadian families in 2007 carried an average net debt load that was 6.3 percent of their disposable income, more than in the U.K. and 'not far off' from the 2005 peak in the U.S.

'We fear that it may simply be a matter of time' before home prices start plunging, Wolf said. 'We're just now starting to see house prices fall in Canada, and sharp rises in unsold home inventories increasingly imply that this will not be a transitory phenomenon.'

---
What a difference a change of employment makes!

[Stephen Gordon]

Fair enough, but The Economists was doing the best it could with the available data; the Teranet Composite-6 only goes back to 1999.

[Just visiting from macleans]

irresponsibly confusing folks like 'Just Visiting' who don't (and shouldn't really be expected to) realize that comparing that series to a true housing index (such as Case-Shiller or Teranet) is like comparing Apples and Hockey Pucks.

Very well, but if one was to use the national Teranet series (as opposed to the Economist/STATSCAN), and then break it down by cities, are you suggesting my general comments still are irrelevant?

[Declan]

The Economist should have used the Royal Lepage data, used the Teranet data as far back as they could or just left Canada out altogether. They could have included a chart of The Maple Leafs annual point totals for their 'Canada' line instead of the StatCan data and it would have been an improvement, because at least then they wouldn't be misleading people into thinking they were looking at a Canadian Housing Price Index.

'Just visiting' - I was just commenting on your attempt to reconcile the two charts. Your comments on cities are OK, but you're not (directly) factoring in the changes in migration and income growth - both of which have been much higher recently in Calgary than Vancouver, and also, in my opinion, you're underestimating the extent to which the housing market follows its own logic - i.e. the buildng boom in Vancouver was largely a consequence of the bubble itself, not any external factor.

Stephen - if there *was* a housing bubble, what would expect to see in your charts? Wouldn't prices go way up and inventory levels go way down due to demand from speculators (e.g. people camping out to buy pre-sales)? And isn't that what the charts for Vancouver and Calgary show? How would you expect the charts to look different if there was a bubble in, say, Vancouver?

Also, just a point of clarification, although CMHC uses the word 'Unoccupied' for that data series, they really mean 'Unpurchased' right? Otherwise that data doesn't look right - 200 unoccupied condos in all of Vancouver during 2006/2007? I could see more than that from my window!

[Just visiting from macleans]

Q: How were you able to find out that the Economist was using CANSIM 327-0005? I couldn't find it in their notes. Maybe I was looking in the wrong place. Same caveat for the US series?

I'm not sure I was being as simplistic as you suggest. The Vancouver "buildng boom" I was speaking of was the one time billions of infrastructure spending pouring into Vancouver as a result of the Olympics (sea to ski highway, light rail to Vancouver airport, Olympic venues etc etc). This to me is like pouring billions into a few oil sands facilities in Fort Mac - though Vancouver is much more diverse, larger pop. and an attractive place to live. It also has physical barriers to where new housing development can occur (ocean/mountains/prime farmland in the Fraser Valley)which tends , I would think, to keep prices higher.

Calgary has no natural barriers - and as a result it has a bad case of sprawl. Yes, ultimately supply and demand for housing prices (and stiff competition for labour with other industries) - but there has to be an underlying economic base to sustain the high prices. Think I've seen an article a couple of months ago suggesting the reverse is happening - net migration out to Sask and back east due to the oil sands /gas boom being tempered.

Anyway, I was just lopping off a few pieces of the carcass - I'll leave the filleting to others.

[Declan]

They list Statscan as the source (click the Source button on the gadget), which made 327-0005 the most likely as it's the only housing index-y time series that Statscan produces. With a likely suspect it was easy enough to check the series and see that it matched the chart from the Economist. Oddly, Statscan doesn't specify whether the 327-0005 series is real or nominal (as far as I can tell), but the Economist treated it as nominal so I'm assuming they at least got that part right.

I wasn't saying you were being simplistic, and I certainly agree that the money for Olympic construction played a role in the Vancouver run-up, I was just pointing out some additional factors to consider.

"there has to be an underlying economic base to sustain the high prices"

True - that's why I think Vancouver is in an unsustainable bubble! Price increases can be self-sustaining as more and more money gets pulled in by psychological factors (greed + speculation), and in the case of a housing bubble in particular, you also get ever increasing borrowing which effectively means the printing of large amounts of money locally, but unless there are increases in rents or incomes eventually the coyote looks down and realizes he's run off the cliff.

Maybe there really are good reasons why Vancouver prices should be as high as they are (underground economy, importing wealth from other regions, rents are about to rise sharply, populaton growth is about to go up sharply, ?) but I'm not convinced.

[Just visiting from macleans]

Stampede from the oldtimers (early to mid 50's) from Calgary's oil patch cashing out. There's only so much to do in Kelowna.

[Stephen Gordon]

Stephen - if there *was* a housing bubble, what would expect to see in your charts? Wouldn't prices go way up and inventory levels go way down due to demand from speculators (e.g. people camping out to buy pre-sales)? And isn't that what the charts for Vancouver and Calgary show? How would you expect the charts to look different if there was a bubble in, say, Vancouver?

I'd have expected to see a big buildup of inventories, the sort that takes years to absorb. That's what happened in the late 1980's in Canada, and more recently in the US.

[Stephen Gordon]

What a difference a change of employment makes!

1) Deputy Governor Timothy Lane was originally scheduled to give the speech, but he had to cancel at the last minute.

2) There's nothing wrong with changing your views after new data arrive.

[Travis]

Just got back from Vancouver and a small house with peeling stucco and a pancaked garage is listed at 680, 000 (22nd). So figure a min of 100,000 in renos. That is what they are calling a bargain fixer upper. Sometimes one needs to get out of their car and take a look around.

[Just visiting from macleans]

Any bubblegum keeping the stucco attached?

[asp]

"Vancouver ... has physical barriers to where new housing development can occur..."

If that were a factor, why has the number of housing units grown faster then the population?

"The 2006 Census reported 870,992 total private dwelling units for Greater Vancouver, an increase of 84,715 units from 2001, and a 10.8% rate of increase. This is slightly lower than the 12.7% rate of increase from the previous census period, but is a greater rate of increase than population growth."

Metro Vancouver Population and Dwelling Counts

[Stephen Gordon]

And yet inventories have fallen since 2001.

This really is the point: the really bad thing about previous housing bubbles is that they generated inventory overhangs that knocked out construction employment for a very long time until it was worked off.

The data show the overhang for the previous cases where we're pretty sure there was a bubble, but they don't show it now. Are they missing something this time?

[Just visiting from macleans]

If that were a factor, why has the number of housing units grown faster then the population?

I think all that this indicates is that more housing units being built are small, pricey condos (one or two person) in high rise buildings as opposed to single family residences which probably is more representative of the older existing stock. Should be reflected in changes in population density, particularly in Vancouver proper (False Creek, West End, Coal Harbour), and along the SkyTrain routes. When there is limited land available, lots get smaller, and you build up, or knock down and build up rather than build out.

[Declan]

"the really bad thing about previous housing bubbles is that they generated inventory overhangs that knocked out construction employment for a very long time until it was worked off"

I would have figured that you only get the inventory overhang after the bubble bursts, not before.

1. Run out of new buyers
2. Lose price momentum
3. Speculators dump units on market
4. Downward price momentum
5. Local economy shrinks, reducing household formation
etc.

[Stephen Gordon]

If we want to use the US experience as a point of reference - and apparently we do - then inventories should have been building up before the peak.

[Just visiting from macleans]

And that therefor raises the question - can you have a real estate bubble without excess inventory leading up to the burst?

And in a market such as Greater Vancouver, can you have local bubbles in one part (say City of Vancouver), and non bubbles (say Surrey, Delta, Richmond)?

I'm always cautious about using high level data (I presume the Teranet data is for Greater Vancouver) which may drown out the sub divisional differences. But, sometimes your are forced to use what's available. I may say something more about this later after walking the dog etc. (btw, I've lived in all three cities at various stages, and fit more on the skool end of the spectrum rather than the dept end on the Nick scale)

[Just visiting from macleans]

Here's an interesting site to get a flavour of the type of development I was referring to earlier in Vancouver proper. I was wondering how many units would be released onto the market once the Olympics are over (athlete's village) as I knew there was some controversy over cost overruns ($1 billion development owned by the City of Vancouver), and in googling, found this site:
http://www.vancouver-real-estate-direct.com/blog/2009/10/vancouver-2010-olympic-athletes-village.html

If you click on the blog's name, lots of other postings of similar developments - many presold which may partly explain lack of inventory levels. The financial risks of undertaking these types of developments without having presold a good portion is significantly higher than single family residences in the burbs. I wonder in the runup of prices if there is some anticipation that the Olympics will give the city an international boost and draw foreign investors, boosting asking/resale prices?

Vancouver is a unique city in terms of development issues, and I'm not sure rules developed elsewhere apply here. In the early 90's when Gordon Campbell (current BC premier) was mayor, I would watch with fascination some Vancouver city council meetings. Many had to do with development issues - including the wholesale development of the Expo '86 lands (and later Coal Harbour, adjacent to Stanley Park). Issues like maintaining line of sight corridors to see the mountains impacted building heights, and the nature of redevelopment. Also, the city planners deliberately limited upgrades of the thoroughfares from the suburbs to create bottlenecks - forcing commerce to locate to places like Burnaby, Surrey etc. and force more people onto public tansport.

Interesting case study.

[JP Koning]

Stephen, in the US chart you provide, what is the name of the data series and the agency that maintains that series?

[JP Koning]

Stephen: "I'd have expected to see a big buildup of inventories, the sort that takes years to absorb. That's what happened in the late 1980's in Canada, and more recently in the US." AND "If we want to use the US experience as a point of reference - and apparently we do - then inventories should have been building up before the peak."

That may be true, but in your post and comments you're comparing total national U.S. inventories to those of three individual Canadian cities. What about the Canadian aggregate to that of the US?

If you take Cansim Table 027-0010 and plot out the data for all Canadian metropolitan and urban areas, you get the exact characteristics that you say define a bubble.

National house inventories (V732850) rose from 2002-2008, touching highs last seen in 1994, though they have collapsed over the last 12 months (they still hover above the lows set in 2001). Multiple-unit inventories (V732914) have also risen since 2002, and stand near the highs set in 1992 just after the data series was begun. So I'd have to disagree with your point that inventories have fallen since 2001; national data shows the exact opposite.

Does the national data demonstrate the sort of inventory overhang and bubble characteristics that you warn of in your last comment?

[Stephen Gordon]

The US graph is from Calculated Risk - the numbers are from the US Census Bureau.

The point about the national numbers is well-taken. But if the buildup in inventories (at least in housing) was something that could be absorbed in a matter of months, then it's not clear how much a problem it is.

Of course, the next question is: where is all that inventory? From a quick look at the entire table, it looks as though much of it is in Montreal, where prices never really stopped increasing.

Has anyone heard any bubble stories coming out of Montreal?

[JP Koning]

Stephen, thanks for the reference for the US chart.

I created a chart of national house and multi-unit inventories without Montreal.

See http://www.financialgraphart.com/public.html, and scroll to the bottom.

Even without Montreal, there's been significant growth in multi-unit inventories. And in early 2009, before falling dramatically, housing inventories (less Montreal) hit record highs.

So I still don't agree with your appeal to the US inventory chart as disproof of a US style housing market in Canada. Not necessarily similiar in scope, but the inventory buildup, particularly in condos and such, would seem to have been there for a number of years.

[Stephen Gordon]

Interesting, thanks. But where *is* is all that inventory located? And can a bubble story be told there? The point about looking at city data is that bubbles are local. Even in the US, not all cities had bubbles.

I wonder what the multi-unit graph would look like in the US; the CR graph is for homes (or does that include condos?).

[JP Koning]

Well, since January 2007, a total of 3,513 new multi-units have been added to Canadian inventory.

The largest contributors are:

Large urban*, 901
Vancouver, 818
Edmonton, 618
Ottawa, 618
Calgary, 401
Victoria, 330
Winnipeg, 286
Kelowna, 251
Sherbrooke, 212

*Note, "large urban" includes all sorts of smaller places like Belleville, Red Deer, Granby, and Kamloops.

Largest multi-unit drawdowns since 2007 are from:

Kitchener, -144
Hamilton, -158
Toronto, -645
Montreal, -727

So anyways, the new inventory coming online in multi-unit housing since 2007 is pretty broad-based. Large-urban, Edmonton, and Ottawa are surprisingly large contributors, compared to what one might expect from Toronto or Calgary. Too tired to do house data.

[Stephen Gordon]

Oh my goodness - you shouldn't have gone through all that for a comment on someone else's blog!

I'm going to revisit these numbers shortly. As I said, I'm trying to figure out how we can put those numbers in the context of a story in which a bubble knocks out the construction sector for several years in a row.

[JP Koning]

"Oh my goodness - you shouldn't have gone through all that for a comment on someone else's blog!"

Tell me about it!

Actually, I can put the information gleaned from blog-inspired research to duty in other lines of work. Like shooting two birds with one stone. This is a very interesting blog subject you've picked.

[Declan]

I'm a little late coming back to this, but just for the record, I agree that a fall in price will be preceded by a rise in inventory, I just believe that the end of the bubble will precede the rise in inventory, not the other way around. In your U.S. chart, inventory only broke out of its historical range in 05 and peaked in 06, the burst of the bubble was well underway by then.

[Just visiting from macleans]

Bubble, bubble, toil and trouble:

New report says urban land policies and rules making homes unaffordable.
Mon Jan 25, 3:44 PM
The Canadian Press

By The Canadian Press

A report says urban land use policies are making homes almost unaffordable in markets around the world including Vancouver, Toronto and Montreal.

The Demographia International report looked at 272 metropolitan markets in Canada, the U.S., the U.K., Australia, New Zealand and Ireland.

The report says Vancouver is the most unaffordable market in the world when median housing sale values are compared to median household incomes.

It says Toronto is in the severely unaffordable category and Montreal is classified as being seriously unaffordable because of contraints on land use.

Wendell Cox, one of the authors, blames urban consolidation policies that restrict the development of suburban residential neighbourhoods for high housing prices.

Cox says governments should allow more housing to be built on the fringes of urban areas to help keep costs down.

[Just visiting from macleans]

Vancouver to lead economic growth, Conference Board of Canada says

City's economy projected to top rankings this year, spurred by Olympics and the housing market; Toronto, Kitchener also shine

http://www.theglobeandmail.com/report-on-business/economy/vancouver-to-lead-economic-growth-conference-board-of-canada-says/article1445929/