Gross National Income and house prices and in Canada and in the US

I've been having difficulty finding ways in which we can directly compare the US housing market débâcle with what's going on in Canada, but I've recently been made aware of two new data sources:

• The Teranet-National Bank Housing Price Index, which appears to be a Canadian equivalent of the celebrated Case-Shiller index: the methodology makes use of how repeat sale prices (that is, sale prices for the same house) have evolved over time.
• Statistics Canada's Cansim Table 380-0062: Gross National Income (GNI) at market prices. This series takes into account the considerable increase in buying power generated by the improvement in Canada's terms of trade.

We'll start comparing apples to apples and oranges to oranges below the fold.

Here's a graph of the Teranet-National Bank composite index along with the Case-Shiller 10 and Case-Shiller 20 composite indices; all three are scaled so that January 2000 = 100:

When the US housing bubble finally popped at the end of 2005, prices had doubled in the space of six years, but they had increased by only about 50% in Canada.

Of course, rising house prices isn't a problem per se: if increasing housing prices reflected households' increasing ability to pay higher mortgage payments, then there wouldn't be much to worry about. But of course, that wasn't the case in the US: employment and real incomes have more or less drifted sideways since 2000. Things have gone much better up here.

So here's a graph of the housing price indices divided by GNI per capita; these ratios are all scaled so that 2000Q1 = 100:

Between 2000 and 2006, house prices had doubled in the US, but nominal incomes (a proxy for ability to pay) had increased by only 50% - hence the current mess. But in Canada, the increase in house prices only slightly outpaced the rate of growth of GNI.

I'm not sure what to think about the latter part of that graph. As a faithful reader of Calculated Risk (and this is the first moment I've been able to create to express my inexpressible sadness at the passing of Tanta), I'm aware of the fact that there's a considerable amount of excess housing inventory that has yet to be absorbed, so the US price-income ratio probably has a lot more to go before the housing market starts to rebound. But I'm not aware of a comparable series for Canada.

What I take from this is that although housing is extremely unlikely to be a source of growth for the Canadian economy for 2009, it's hard to see how it's going to be a drag of a size comparable to what's going on in the US. As I mentioned earlier, any prediction of a severe recession in Canada has to have a home-grown component: I don't see how the housing sector is going to provide it.

Comments

[Nick Rowe]

Seriously important post there Stephen. Demonstrates once again the point you've been making: Canada is not the US. Tempted to revise my predictions upwards! Shows as well the importance of good house-price data.

Minor point: shouldn't the relevant measure of affordability be house price divided by GNI *per person*? (I don't think this will affect the comparison much, since I think Canada and the US had about the same population growth rates.

Typo just above figure 2? Should be "scaled so that 2000Q1=100"?

[Stephen Gordon]

Whoops - typo fixed.

And you're right: I should scale by population. But that shouldn't qualitatively affect what those graphs say. I'll do it tomorrow.

Update: GNI is now scaled by population.

[anon]

Somewhere I've seen a 100 year graph (yes) of real housing prices in the US, where the trend is basically flat (i.e. no increase) for 90 years, and then almost doubles before the crash. Anything comparable available for Canada? Real prices have certainly gone up here a lot as well.

The trend would also suggest US prices have further to fall. Perhaps the shape is the same here, with lower amplitude, and with a lag.

[Nick Rowe]

I wouldn't trust any measure of real house prices over as long a period as 100 years. The size and quality of houses has changed so much, and the cumulative bias in the CPI also would make it very suspect. Plus, with (I would guess) technical change in houses being less than other goods, it wouldn't surprise me if real house prices (quality and size adjusted) were higher now than 100 years ago.

On the other hand, 100-year data on nominal house prices as a ratio to nominal income per capita might still be meaningful. But the Case-Shiller/Terranet-National Bank data would not be good for this (even if it were available), because it adjusts (as best it can) for changes in the average size and quality of houses. If we are looking at a 100 year ratio of house prices to income, what we are looking for is to see whether the average household can afford to buy the average house. (And even here, would we suppose that the long-run income elasticity of demand for housing is one? Maybe, but it's not obvious that it should be one.)

[Richard H. Serlin]

You might want to also look at graphs starting in 1997. Shiller, in his outstanding book Irrational Exuberance, 2nd Ed., 2005, puts that year as the start of the U.S. housing bubble (for example, page 13). I took a quick look at his free data and found a 16.5% increase in real prices from 1997 to 2000. The long term trend in his data is only a 0.4% real increase (1890-2004, from his book page 20). At 0.4%, real prices would only have increased by 1.2% between 1997 and 2000, not 16.5%.

It's interesting, though, to look at the average rate of increase in U.S. housing prices from the Case-Shiller data, ex-recent bubble, that is not counting 1997 on. In 1890, the inflation adjusted home price index starts at 100. 107 years later what do you think it's risen to? 109.6387! less than a 10% increase in 107 years! The annual return is 0.086%, less than 1/10th of 1%! At that trend rate, real homes would have only appreciated by 0.3% between 1997 and 2000, versus their actual real increase of 16.5%.

[anon]

That's the source for the graph I saw, and that's my point.

[Stephen Gordon]

Sadly, the Canadian series only goes back to 1999 or so, and one of the C-S indices (forget which) starts in January 2000. I would very much have liked to go back further.

[Patrick]

Seems to me that with so much of our production being sold in the US, it doesn't take much for their plummeting aggregate demand to affect us. But if we're looking for homegrown problems: how capex spending drying up. With risk appetite at zero and credit nowhere to be found, project cancellations are getting to be a problem here in Alberta, and I imagine it's much the same in the rest of the country. It'll take some time for the effect to be felt as existing projects wind down, but many tradespeople currently earning really, really good money and consuming most of it, could find themselves unemployed in the next 6 months to a 1 year.

Could be a good argument for infrastructure spending in Canada.

Alberta tracks project spending (I think other provinces do too):

http://www.albertacanada.com/statpub/1112.html

[Stephen Gordon]

That's an interesting resource; thanks.

But there's that whole timing thing. As you say, it'll take some time for the existing projects to unwind. There's not a lot of slack in the construction industry right now, so infrastructure projects would have to be put on hold until workers are laid off from the jobs they have now.

Infrastructure might be a good idea later on, if the recession is longer-lived than usual.

[Phillip Huggan]

They might inflate to balance budget and this would lower their dollar and suck more manufacturing jobs away from Southern Ontario. Resources should rebound long term, helping Western Canada. In the USA consumers are allowed to walk away from out-of-the-money mortgages while in Canada consumers can be sued for such a contract breach, so housing will always be more volatile there.

[Toronto condos]

Very nice illustration. This just shows what every good professional has known. Canada and the USA have different real estate markets. Our boom began hardly two years ago and the prices didn't jump so high at all. Mortgage structure is different, our urban structure is different...Even the decline is different. Prices are going down by few percents only, returning somewhere to year 2006. No fall, no collapse.
Take care
Elli

[David]

Very interesting and useful.

If the long term trend in real house price appreciation is .4%, how far have we deviated from trend in the last 8 years? Using the back of the envelope and assuming Q12000 = 100 and annual inflation since 2000 of 3%(?), the Teranet-National Bank Housing Price Index should be about 130. At 178 it is currently about 25% over valued.

So too is the house price / GNI per capita. From 1980 to 2000, US home values remained at about 3x income before spiking to 4.5x in 2006 (http://www.bergenjerseyforeclosures.com/blog/info/entry/where_should_house_prices_really.) Assuming that Canadian homes were also worth about 3x incomes in 2000(?), the index should still be about 100 in the above chart not 130.

And then there is overshoot.

The US is heading for a 40-50% drop in housing prices peak to trough. A 25%-35% drop in Canada looks about right to me.

[Popping Bubbles]

Just because we didn't drink as much as some of the other countries, it doesn't mean that Canada isn't in for a huge housing hang over, with price drops of over 30% from the peak the likely outcome. Why? Because of...

a) the fact housing affordability is at its worst level since the last housing bubble burst [RBC. Housing Affordability. Mar-2009, p.1]

b) the fact that real housing prices have increased substantially more than during the last three housing cycles dating back 40-years (all of which ended badly) [Scotiabank. Real Estate Trends, 26-Feb-2008, p.2]

c) the fact that real housing prices in Canada have risen more from trough to peak than in the U.S., where prices and the general economy are now tanking [Scotiabank. Real Estate Trends, 26-Feb-2008, p.2]

d) the fact that Canada's housing prices-to-rent ratio is higher than in any other OECD country save Spain and 90% higher than the long-run trend [OECD Economic Outlook No. 82, December 2007. Data table can be found in the housing price ratio tab of http://www.oecd.org/dataoecd/6/5/2483894.xls]

e) the fact that Canada's housing prices-to-income ratio is 32% above historic trends and substantially above the ratio which prevailed when the last housing boom bubble popped in the late 80’s / early 90’s [same source as (d)]. Indeed, statistics on the Toronto housing market show that prices are currently 4.18x the average household income. This is 45% higher than the ** PEAK ** of the last major housing bubble in 1989, when only 2.87x the prevailing average income was required to purchase a home in Toronto ( http://multimedia.thestar.com/acrobat/50/79/d004eda0453f8443bb506994c2ad.pdf )

f) the fact that the unprecedented run-up in prices have been fueled by a proliferation of risky lending practices such as (i) a decrease in the required down payment from 10% to 0%, (ii) an increase in the allowed amortization from 25-years to 40-years, (iiI) the proliferation of 7% cash back mortgages and other lending gimmicks (teaser rates, step mortgages, skip a payment, builder rate buy downs, etc.), (iv) the proliferation of home equity lines of credit, and (v) lenders not being on the hook for the vast majority of risky loans they write (CMHC guarantees low-down payment and/or extended amortizations)

g) the fact that studies show typical consumers do not fully understand the implications and risk of low down payment, long amortization and gimmicky (e.g. 7% cash back) mortgages. How many consumers do you think have run a scenario analysis which asks, “what would happen if interest rates went to 8%, 10% or even 12%? What would happen if my partner or I lost our job? What would happen if real estate prices dropped by 10%, 20% or 30%? What impact will extending myself for this house have on my retirement plans?”

h) the fact that housing bubbles around the world are beginning to deflate. Canada is not some magical island. By way of example, the UK (admittedly a worse market then ours) mortgage lending in the first quarter is down 40% to the lowest level in 33-years and things are only beginning to get rolling there. In New Zealand housing sales are down 53% year-over-year. Prices in Ireland, Spain and Australia are all getting pounded. And we all can see what is going on in the U.S.

i) the fact that housing construction is far in excess of household formation. CMHC data shows housing starts averaging 226,000 units per year from 2003 through 2007, 33% per year above the roughly 170,000 net new households formed each year [as estimated by TD Economics and others]. Based on housing permits and starts, this trend is expected to continue well into the future.

j) the fact that Canadian MLS housing inventory is at record highs while at the same time the number of sales is dropping dramatically [Canadian Real Estate Association (CREA)]

k) the fact consumer indebtedness is at record highs relative to disposable income [Vanier Institute. The Current State of Canadian Family Finances. 11-Feb-2008. p.28]

l) the fact that savings rates are close to nil even though the baby boomers should be saving for retirement [Vanier Institute. The Current State of Canadian Family Finances. 11-Feb-2008. p.9]

m) the fact that Canadian incomes have stagnated. Statistics Canada recently reported "that adjusted for inflation the earned income of the ‘average’ Canadian -- the so-called median income - was the same in 2004 as in 1982”

n) the fact that the economy is in recession. The U.S., which absorbs some 70% of our exports, is unlikely to recover in a meaningful way for some time. The rose is off the commodity boom (with further downside likely) and construction is next

[Nick Rowe]

Popping bubbles:

Your points a,b,c,d, and e are all based on house price data that are significantly worse than the new data Stephen is using. Those studies are now obsolete.

The Vanier Institute suffers from "micro-economists' fallacy". The net debt of a closed economy is zero. The net debt of Canada as a whole is slightly negative (foreigners owe more to us than we do to them). If we divide Canada into three sectors: households; business; and government, then since businesses and governments have positive net debt, the Canadian household sector must have negative net debt, despite what the Vanier Institute says.

[Declan]

Aren't PB's points a, b and e just historical comparisons so as long as the historical series is consistent, its level vs. another series shouldn't really matter? And point d) shouldn't really be affected much either. So adjust point c) so that the rise in Canada is almost as high as the one in the U.S. instead of slightly higher, and does that make you feel much better?

As for k) what difference does net debt make to anything? - the problem is the gross debt. If everyone in the country except one person was on the verge of declaring bankruptcy because they couldn't pay their debt, would it make things any better to know that all that debt was owed to somebody?

I'm with PB on this one.

[Nick Rowe]

Declan: but if the composition of houses that are sold varies over time, then the historical series won't be consistent (unless by sheer chance US and Canadian compositions change in the same ways and introduce the same time-varying biases). By comparing Canadian and US house prices using the same Case-Shiller method, we eliminate those biases.

on the net debt: it could be that one person's liabilities offset another person's assets, or it could be that one person has offsetting assets and liabilities. The Vanier Institute 2007 report gives average debt at $80,000 per family (presumably gross debt), and net worth at $395,500 (which is presumably real assets minus net debt). A distribution of net debt over people would be useful, but that's not what we get. And I would want to be sure that the cumulative distribution of net debt adds up to a number which is consistent with macro data, which means it should be negative.

By the way, I was wrong above when I said that Canada has negative net foreign debt. It was slightly negative, but is now slightly positive in 2008 (but very small compared to government or business debt).

[JKH]

Nick,

A couple of comments on your debt discussion above, which I found quite interesting.

There are intra-macro as well as macro-micro reconciliation challenges, I think.

At the macro level, for example, one of the components is a country’s net international investment position. I’m more familiar with the US position than the Canadian one. The US position of course is a “net liability”. But the corresponding gross liability includes a variety of financial claims, some of which are portfolio equity and FDI.

Yet people commonly refer to the US as “owing” the rest of the world so many dollars, essentially due to the negative US NIIP position, which incorporates the result of the cumulative US current account position, marked to market in some of its components, as well as the cumulative effect of trade in various types of financial assets.

Of course, a good chunk of government and corporate debt is also included in the net liability position.

My point is that in the case of this sector, people may well refer to a “net debt” position that in fact includes a variety of types of financial claims, including equity claims.

You can’t simply pull out the debt claims component of US NIIP and attribute the commonly referred NIIP net “debt” position to this debt claims component. There’s no basis for isolating debt claims within the larger sphere of financial claims in this way.

So if the term “net debt” is used in this case, it really can only be used correctly in the sense of “outside” wealth rather than “inside money”. “Inside money” would be defined as financial claims of all types, where a financial claim held as an asset is always offset by the same financial claim viewed as having been issued as a “liability”, including equity claims.

Moreover, the more proper term in my view for this net international position would be “net equity” (in the sense of outside wealth) rather than “net debt”, because the net position in the case of a current account deficit country is really a subtraction from the country’s equity position in term of outside money – i.e. its cumulative position in terms of net assets of all types – other things equal. In fact, if one were to draw up a complete gross balance sheet for the US, with all sorts of offsetting entries on both sides, the NIIP contribution in its complete accounting would include outside equity as an asset, which is negative equity.

My larger point is that I think it’s difficult to reconcile the aspect of gross debt claims within an economy, taken at any level, with full balance sheet integration of those claims in the context of the entire array of real and financial assets. Or at least it would take a lot of work to do this thoroughly (which I’ve certainly never attempted).

I’ve actually not looked closely at some of the gross debt ratios that are the cause of concern. I think Declan above makes a pretty fundamental point about it though. I interpret it this way: the only reason that gross debt measures and ratios are useful is that higher gross debt measures at the macro level tend to correlate with greater mal-distribution of negative and positive gross positions at the micro level, almost by definition I suppose. It’s the distribution that causes the credit problem, although distribution correlates with size.

A more thorough analysis would examine gross debt positions as well in the context of complete sectoral balance sheets. For example, it’s very interesting to watch the declining US household net worth position as measured in the quarterly Fed flow of funds report. There, the gross household debt positions (mostly mortgages and consumer debt) start to show up in the context of plunging asset values for both residential real estate and various types of financial claims (bonds, equities, mutual funds, pension reserves, etc.) held by households. There you can see the household leverage risk play out over of the course of the credit crisis and recession in a more complete macro interaction. The plunge as measured already is quite breathtaking, and will be even worse when 4th quarter results are released in March.

[Nick Rowe]

JKH: I agree with a lot of what you say, except:

"....higher gross debt measures at the macro level tend to correlate with greater mal-distribution of negative and positive gross positions at the micro level, almost by definition I suppose." That wouldn't be true by definition, since many people will have (say): mortgage debt and pension plan assets; or a transactions balance liability on their credit card, and long-term GIC's; a margin account with leveraged purchases of assets, at the same time. So it's not obvious that a rise in gross debt over time will signal an increased variance of net debt across the population, though it may be true.

Best avoid the words "inside money" and "outside money", because they have a very particular meaning in the Pecek & Saving vs. Gurley and Shaw debate. But I get your meaning. "Inside" vs. "outside" financial assets maybe?

Average household leverage is also zero (in a closed economy).

My point was the simpler one. We hear the words "the average Canadian household has debts of $80,000", mentally multiply by the number of households, and assume that Canadians have debts of about $1 trillion, which they owe to...errrr.....the Chinese? It gives a horribly distorted picture of reality, and we hear journalists echo those distortions every year. It's some sort of distortion of the "representative agent" concept. You are lead to believe they are talking about a representative agent, but they aren't.

I feel a post beginning to well up inside me..........

[JKH]

Nick,

I didn’t mean larger gross debt over time; rather larger gross debt relative to some norm (i.e. as in a ratio). But I guess the implication is pretty obvious in that sense.

Yeah, I’ve always been confused about inside/outside, among a billion other things in economics.

Household leverage – I don’t get your point. E.g. define aggregate household leverage in the US as ratio of debt to net worth – then it’s roughly 1:4 currently. If that’s the aggregate, how can the average be zero?

Keep on postin’ !

[Nick Rowe]

Household leverage: take a closed economy. The households in aggregate own all corporate liabilities (stocks and bonds), plus all government bonds, plus houses, land, and real assets. In an economy of 1 million households, the representative household owns 1 millionth of all those assets, and cannot be leveraged, because there is nobody to borrow from to leverage. The 1:4 figure must be wrong (unless it's the Chinese???)

The conceptual difficulty is in trying to talk about the debt (or leverage) of a representative household. It just won't work. It leads to conceptual confusion. But at the same time we (and I don't mean just economists by "we") somehow need to think in terms of a representative or "average" household (or individual). Still trying to get my head clear on this. It's frustrating me.

Peter owes Paul, and Paul owes Peter, but in aggregate, or in net debt on average, it all washes out to zero.

[JKH]

I see your point once more – I’ve been arguing both sides against the middle at different times I guess.

FYI, here is the link to the latest Fed flow of funds report:

http://www.federalreserve.gov/releases/z1/Current/z1.pdf

The numbers I’m talking about are on page 102, table B.100. In $ trillions:

Total Assets 71
Tangible 26
Financial 45

Total Liabilities 14
Mortgages 10
Other 4

Net Worth 56

On this basis, gross leverage is 14:56 or 1:4.

This interpretation is consistent with how leverage is viewed for corporate balance sheets.

To the extent that 14 in liabilities are offset by asset debt, then net household leverage could be viewed as 0, which I think makes your point. Financial assets include 7 in deposits and 4 in credit market instruments, which is reasonably close.

Or perhaps you are making a larger point and including all types of financial assets as the potential offset to gross household debt. The total includes equities, mutual funds, pension reserves, etc. How does one identify, interpret, or think about the ultimate debt or equity financial claims within this entire financial asset category for this sort of discussion? Mutual funds and pension reserves hold both debt and equity, for example. And if the argument is in this larger sense that gross debt by logic must be offset by some financial asset, debt or otherwise, why doesn’t it work the other way around to arrive at net 0 financial assets as part of household worth? (Here I’m getting quite confused.)

I suppose the bottom line is that you can always interpret the mere existence of positive household net worth as an indicator that there is no net “debt” or net “leverage” for households. That makes the point in the crudest way that households aren’t indebted on a net aggregate basis in the sense of being insolvent on an aggregate basis – “they owe it to themselves” sort of thing.

Other aspects:

The financial sector provides maturity and other risk transformation functions. One result is that the duration of aggregate household debt liabilities is longer than the duration of many household debt assets (e.g. mortgages versus bank deposits), among other things. This could be viewed as a sort of “duration leverage”.

Also, in terms of thinking about “average leverage” for households even in the gross sense, there is a zero bound (oh no!). One doesn’t normally think of negative leverage on a balance sheet – i.e. for those with no gross debt liabilities. This may be one of your points as well.

Of course, the US economy is open. Last time I checked the NIIP was roughly $ 2 or 3 trillion. That complicates things at the margin, but given the size of the rest of the numbers, certainly not enough to affect the thrust of the 1:4 leverage ratio to which I’ve referred.

[JKH]

Weird – I posted this and then it disappeared.

I see your point once more – I’ve been arguing both sides against the middle at different times I guess.

FYI, here is the link to the latest Fed flow of funds report:

http://www.federalreserve.gov/releases/z1/Current/z1.pdf

The numbers I’m talking about are on page 102, table B.100. In $ trillions:

Total Assets 71
Tangible 26
Financial 45

Total Liabilities 14
Mortgages 10
Other 4

Net Worth 56

On this basis, gross leverage is 14:56 or 1:4.

This interpretation is consistent with how leverage is viewed for corporate balance sheets.

To the extent that 14 in liabilities are offset by asset debt, then net household leverage could be viewed as 0, which I think makes your point. Financial assets include 7 in deposits and 4 in credit market instruments, which is reasonably close.

Or perhaps you are making a larger point and including all types of financial assets as the potential offset to gross household debt. The total includes equities, mutual funds, pension reserves, etc. How does one identify, interpret, or think about the ultimate debt or equity financial claims within this entire financial asset category for this sort of discussion? Mutual funds and pension reserves hold both debt and equity, for example. And if the argument is in this larger sense that gross debt by logic must be offset by some financial asset, debt or otherwise, why doesn’t it work the other way around to arrive at net 0 financial assets as part of household worth? (Here I’m getting quite confused.)

I suppose the bottom line is that you can always interpret the mere existence of positive household net worth as an indicator that there is no net “debt” or net “leverage” for households. That makes the point in the crudest way that households aren’t indebted on a net aggregate basis in the sense of being insolvent on an aggregate basis – “they owe it to themselves” sort of thing.

Other aspects:

The financial sector provides maturity and other risk transformation functions. One result is that the duration of aggregate household debt liabilities is longer than the duration of many household debt assets (e.g. mortgages versus bank deposits), among other things. This could be viewed as a sort of “duration leverage”.

Also, in terms of thinking about “average leverage” for households even in the gross sense, there is a zero bound (oh no!). One doesn’t normally think of negative leverage on a balance sheet – i.e. for those with no gross debt liabilities. This may be one of your points as well.

Of course, the US economy is open. Last time I checked the NIIP was roughly $ 2 or 3 trillion. That complicates things at the margin, but given the size of the rest of the numbers, certainly not enough to affect the thrust of the 1:4 leverage ratio to which I’ve referred.

[Nick Rowe]

Hi JKH: I'm switching this interesting discussion of debt to my new post.

[Lawrence]

The graphs would be better presented using a logarithmic scale for prices, so that doubling amounts would be equally spaced. This is used on stock market scales all the time. For example, doubling from 100 to 200 would equal the distance from 500 to 1000, whereas on a linear graph like the ones used, 500 to 1000 is five times the distance of 100 to 200.