Carleton University's Nick Rowe left a comment in a previous post that deserves a wider readership. With his permission, I'm recycling it as a guest post:
I am less optimistic than you, though I don't think it will be as bad here. But I wish I had better evidence for my belief/hope.
House prices in many Canadian cities have increased a lot in the last couple of years, and I see no reason why they can't decrease by the same amount in as short a time. If they did, many households would have negative equity, and many of those mortgages would default.
We need to look at the variance of house price declines, not just the average decline across the country. Let me explain why:
Assume that the percentage of mortgages which default is proportional to the drop in prices. Assume also that the percentage loss on a defaulting mortgage (when the bank sells the foreclosed house) is also proportional to the drop in prices. Then the percentage loss on all mortgages is proportional to the square of the drop in prices. So a 20% drop in prices would cause four times the losses of a 10% drop in prices. So a 20% drop in prices in half the country, with no drop in the other half, would cause double the losses of a 10% drop in prices across the whole country.
That's what worries me; even if average house prices across Canada are only slightly too high, the possibility of very big price declines in some parts of Canada could still cause big losses on mortgages.
As far as I can tell, all Canadian mortgages with original Loan To Value ratios greater than 80% must have mortgage insurance. As far as I can tell, there are three insurers of Canadian mortgages: CMHC, AIG (familiar name?) and Genworth (which is still in business, despite a very big fall in share price, and is trying to sell off its mortgage insurance business). I can't tell what shares of the total market those three have, but I suspect that CMHC has by far the biggest share.
According to its 2007 Annual Report (if I read it correctly) CMHC had $334 billion of mortgages insured, and about $7 billion capital. I couldn't find any information on the distribution of LTV ratios across those mortgages. On the one hand, since CMHC insurance is only required for LTVs greater than 80%, we might expect to see a lot of high LTV ratios represented. On the other hand, many of those insured mortgages might be old ones, that have been paid down over the years while house prices have risen, so will have low LTV ratios.
Here's a back of the envelope guesstimate of likely CMHC losses if house prices drop 20% in half of Canada:
Assume a uniform distribution of LTV ratios, so if prices drop 20%, 20% of houses will be in negative equity. Assume half of those mortgages default, so 10% default. The average defaulting mortgage would be 10% underwater, but the losses could be double that with transactions costs of foreclosure, so 20% losses on a defaulting mortgage. 10% defaults x 20% losses per default = 2%losses on the mortgages insured, and if the 20% price drop happens in only half of Canada, that's 1% of the $334 billion, which is $16 billion losses for the taxpayer (who owns CMHC). Or $9 billion if we keep the $7 billion CMHC capital off the books. Not too bad, and the banks come through fine, because the taxpayer pays all the losses. And it might be much worse, if my assumptions are wrong, or if house prices fall more.
As for the Canadian banks: I have heard a number of people say the banks are in good shape, but all of them either work for the banks, or are in government, so "they would say that, wouldn't they?" I have no reason to disbelieve them, and actually think they are probably right, but I would like to see some informed independent analysis.
My biggest fear for Canada though, is not house prices, falling commodity prices, falling export demand, etc., but sheer financial contagion. It's NOT just a US financial crisis (we read too much US news). There are very similar stories being told in the UK, Spain, Ireland, Australia, New Zealand, Eastern Europe, China too I believe, of falling house prices and financial institutions in trouble. The TSX seems to have a very high short term correlation with the S&P500 recently. Why, if we are safe here?
Quite apart from any fundamental channels for contagion, financial markets, and financial institutions, nearly always seem to have that multiple equilibrium (like Diamond/Dybvig bank runs) character. If the rest of the world sees a sunspot, and runs to the bank, I can't see Canadians failing to join the herd. We have already seen that run on asset backed commercial paper.
One reason for definite optimism is the superior fiscal position of the Federal Government. If we really need to, the government could borrow about one trillion, pushing the debt/GDP ratio up to 100%. We've been there before, and survived, and this time we would be borrowing to buy assets, so the net debt/GDP ratio would be less.
House prices in many Canadian cities have increased a lot in the last couple of years, and I see no reason why they can't decrease by the same amount in as short a time. If they did, many households would have negative equity, and many of those mortgages would default.
We need to look at the variance of house price declines, not just the average decline across the country. Let me explain why:
Assume that the percentage of mortgages which default is proportional to the drop in prices. Assume also that the percentage loss on a defaulting mortgage (when the bank sells the foreclosed house) is also proportional to the drop in prices. Then the percentage loss on all mortgages is proportional to the square of the drop in prices. So a 20% drop in prices would cause four times the losses of a 10% drop in prices. So a 20% drop in prices in half the country, with no drop in the other half, would cause double the losses of a 10% drop in prices across the whole country.
That's what worries me; even if average house prices across Canada are only slightly too high, the possibility of very big price declines in some parts of Canada could still cause big losses on mortgages.
As far as I can tell, all Canadian mortgages with original Loan To Value ratios greater than 80% must have mortgage insurance. As far as I can tell, there are three insurers of Canadian mortgages: CMHC, AIG (familiar name?) and Genworth (which is still in business, despite a very big fall in share price, and is trying to sell off its mortgage insurance business). I can't tell what shares of the total market those three have, but I suspect that CMHC has by far the biggest share.
According to its 2007 Annual Report (if I read it correctly) CMHC had $334 billion of mortgages insured, and about $7 billion capital. I couldn't find any information on the distribution of LTV ratios across those mortgages. On the one hand, since CMHC insurance is only required for LTVs greater than 80%, we might expect to see a lot of high LTV ratios represented. On the other hand, many of those insured mortgages might be old ones, that have been paid down over the years while house prices have risen, so will have low LTV ratios.
Here's a back of the envelope guesstimate of likely CMHC losses if house prices drop 20% in half of Canada:
Assume a uniform distribution of LTV ratios, so if prices drop 20%, 20% of houses will be in negative equity. Assume half of those mortgages default, so 10% default. The average defaulting mortgage would be 10% underwater, but the losses could be double that with transactions costs of foreclosure, so 20% losses on a defaulting mortgage. 10% defaults x 20% losses per default = 2%losses on the mortgages insured, and if the 20% price drop happens in only half of Canada, that's 1% of the $334 billion, which is $16 billion losses for the taxpayer (who owns CMHC). Or $9 billion if we keep the $7 billion CMHC capital off the books. Not too bad, and the banks come through fine, because the taxpayer pays all the losses. And it might be much worse, if my assumptions are wrong, or if house prices fall more.
As for the Canadian banks: I have heard a number of people say the banks are in good shape, but all of them either work for the banks, or are in government, so "they would say that, wouldn't they?" I have no reason to disbelieve them, and actually think they are probably right, but I would like to see some informed independent analysis.
My biggest fear for Canada though, is not house prices, falling commodity prices, falling export demand, etc., but sheer financial contagion. It's NOT just a US financial crisis (we read too much US news). There are very similar stories being told in the UK, Spain, Ireland, Australia, New Zealand, Eastern Europe, China too I believe, of falling house prices and financial institutions in trouble. The TSX seems to have a very high short term correlation with the S&P500 recently. Why, if we are safe here?
Quite apart from any fundamental channels for contagion, financial markets, and financial institutions, nearly always seem to have that multiple equilibrium (like Diamond/Dybvig bank runs) character. If the rest of the world sees a sunspot, and runs to the bank, I can't see Canadians failing to join the herd. We have already seen that run on asset backed commercial paper.
One reason for definite optimism is the superior fiscal position of the Federal Government. If we really need to, the government could borrow about one trillion, pushing the debt/GDP ratio up to 100%. We've been there before, and survived, and this time we would be borrowing to buy assets, so the net debt/GDP ratio would be less.
Worthwhile Canadian Initiative 
Could the federal government raise a trillion dollars? Wouldn’t that cause some pretty interesting fluctuations in exchange rates on the dollar?
CMHC also has $165 billion of MBS guarantees in force, so total at risk is $500 billion against capital of $7 billion.
2007 provisions for claims totals $500 million (0.1% of the book).
Some comments on the back-of-the-envelope calculation:
A 10% default (presumably you mean foreclosure) rate seems very high. The US is running about a 9% (6.41% delinquency + 2.75% foreclosure) rate in Q2.08.
Loss severity of 20% seems high as mortgages in Canada are recourse through deficiency judgments.
If house prices decline 20%, the average mortgage would only be 7.5% under water as CMHC only insures to 95% LTV. But as pointed out that overstates the case as some of the mortgages and guarantees are seasoned; i.e. the principal has been amortized and house prices have risen since origination.
Assuming a uniform rate of LTVs is not instructive. A 20% drop in house prices puts most (not 20%) CMHC mortgages underwater, according to the model.
The bottom line arithmetic is incorrect. A 1% – 2% loss on $334 billion is $3.34 to $6.68 billion.
Why would negative equity lead to default? Isn’t it the case that the default rate in the US is being caused by unqualified lending, which is leading to the decline in prices? While it is true that businesses might decide to writedown equities worth less than the debt held against them, people have to live somewhere, and people with the cashflow to pay the mortgage generally do without any regard to the value of the asset.
That’s sort what I’m thinking. Mortgage interest rates are still pretty low – lower than they were when the subprime crisis hit – so people aren’t facing significant jumps in their mortgage payments of the sort faced by US borrowers who signed onto teaser rates and negative amortisation schemes.
Replying to comments:
Andrew: I don’t see why the Federal government couldn’t borrow one trillion dollars. That would put the debt/GDP ratio up to 100%, and many countries, including Canada, have had debt/GDP ratios that high or higher in the past. Plus, if we used that trillion dollars to buy assets, the net debt would not rise by a trillion. But yes, I’m not sure what it would do to the exchange rate.
marmico: I get the definite sense that you know more about CMHC than I do! I would like to see your back-of-the-envelope calculation. I think that most US states have non-recourse mortgages (which means the bank can take back the house, but can’t come after the borrower for any losses if the sale of the house doesn’t cover the mortgage). So your noting that Canadian mortgages are recourse will make an important difference between Canada and the US, tending to reduce Canadian defaults as well as reducing the losses per default. I would really like to see some sort of distribution of LTVs in Canada. A uniform distribution is supposed to roughly work with UK houses (at least, those with a mortgage). As you say, it may not work well for CMHC-insured mortgages. Yep, arithmetic is not my strong suit; but that was an editing error on my part I think!
Kenneth and Stephen: even if he can “afford” the mortgage payments, if a homeowner has negative equity, if he walks away from the house and the mortgage his net worth increases, which is the incentive to default (mail the bank the keys, or “jingle mail” as it’s called). In a non-recourse jurisdiction the only disincentive is the damage to his credit rating, but that would be temporary. It’s cheaper to rent than to keep paying the mortgage, is another way of looking at it. In a recourse jurisdiction, like Canada, the bank (or CMHC) could still come after him for the remaining debt, but how likely are they to ever collect? No doubt some people would continue paying the mortgage, even though it would cost them more than renting, because of not wanting to move house for example. But if the price of oil drops, and the Alberta labour market worsens, so people need to move anyway, I don’t see that stopping them. That’s what happened last time, I think.
In addition to the valid points that Marmico made, a few more:
1) AIG has no real market presence in Canada.
2) The government guarantees 90% of all privately insured mortgages as well, so if Genworth defaults, the government is on the hook there too. Genworth is pretty big in terms of market share, not as big as CMHC I don’t think, but not that far off. Interesting to see how the government might react to a potential sale of Genworth’s portfolio to a less well-backed institution.
3) Your 10% additional cost estimate is probably low, 15% is probably a better estimate, especially during a housing downturn.
4) The argument that Canada didn’t have the same subprime lending as the U.S. is undermined by the fact that the homeownership rate peaked in the U.S. at 69% and was at 68% in Canada (the highest on record, having runup sharply in the last few years just like the U.S. rate) as of the 2006 census.
5) Don’t forget the insurance premium, which adds a few percent to the LTV of all insured mortgages.
6) A high percentage of the recent mortgages in the bubbly markets (e.g. Vancouver) are 40 year amortization periods, which take a long time for the principal to be paid down materially, so you might not have much paydown on the underwater mortgages.
7) Many banks insure mortgages with lower LTV’s through CMHC so that they can securitize them, so this will lower the average LTV of CMHC’s portfolio.
8) Not only will the LTV’s not be uniform, the price drops won’t be either – this will likely make the losses worse, although that depends on the circumstances.
I get the definite sense that you know more about CMHC than I do!
I don’t think so, Nick. I just quickly perused the 2007 CMHC Annual Report. FWIW, I don’t think that CMHC is at much risk for pre-July, 2006 vintage mortgages. The CMHC bumped up insurance from 75% to 80% LTV at that time.
The worst case scenario is 3% foreclosure rate X 30% loss severity or 1% of book or $5 billion in losses spread over several years.
For the 2 years post July 2006, the NHA MBS added about $90 billion to the book and insured mortgages added about $80 billion to the book, for a total of $170 billion. An armegaddon 5% foreclosure rate X 50% loss severity on the post July 2006 book is still only $4.3 billion.
You should review Section 9. Insurance of the Financial Statements (page 76 of the PDF linked upthread),a section of which, is italicized below:
To provide a further measure of sensitivity of the change in Provision for Claims, it is estimated that for every 5% change in the estimate of future claim severity or every 5% change in the estimate of future claim frequency, the effect on Income Before Income Taxes would be an increase/decrease of approximately $20 million by the end of the Corporate Plan horizon.
These sensitivities are hypothetical and should be viewed in that light. The relationship of a change in assumption to the change in value may not be linear. Changes in one factor may result in changes in another which might magnify or counteract the sensitivities.
FWIW, I believe national house prices will decline 10%, 5-10% in Ontario/Quebec, 20-30% in Alta, BC. And if you are interested you can check out the UBC/Sauder School historical house price data here.
Just a quick note to further complicate matters, using Vancouver as an example as it is very bubbly.
In Vancouver there has been a significant investment into homes and especially condos by people overseas. The amount of investment properties is hard to gauge on its own, the place can be a hard investment and sit empty or be rented or it may be a soft investment where it is owner occupied. Including condo assignments the number of hard investment condos is in the tens of thousands according to leaked BC Hydro data. The amount of foreign investment ownership is also anecdotal but can be confirmed through condo stratas. The main foreign buyers come from Australia and China which incidentally began their housing booms a little earlier.
So the new variable is owners who default, but obtained their mortgage from a foreign bank. The incentive to go with a foreign bank (not restricted to foreigners mind you) is familiarity, easier terms, lower mortgage rates, and lower transaction costs (limiting FOREX transactions)
One more variable is that among locals, many of the young buyers were financed by family, usually with wealthy parents covering a down payment to avoid private mortgage insurance. Worse, some homes are financed through a chain of home equity lines of credit.
Advertising for, and the use of home equity lines of credit is also a significant factor in Vancouver.
Finally, a 40% decline in Vancouver home prices is a very credible possibility. The law of real home price changes is months of inventory to sales (take MLS listings, divide by sales). Examining rent to buy, and income to house price ratios will give a base level to which prices will return once the excess inventory is absorbed.
Considering the leverage/margins, the stunning increase in prices, there have been 3 other real estate bubble crashes since 1988, and the correlation between economy and housing (realtors, construction, painters, furnishings, etc) really mean that the lack of subprime and alt-a are irrelevant. Subprime availability may enable a house price bubble, but it does not exaggerate declines (although it may accelerate it).
In parts of Vancouver, the decline will be over 50% (eg a $365,000 450 sq ft studio goes to $180,000)
If the blog readership is interested, I could compile some of the Vancouver data to give perspective on what will be the single biggest destruction of notional value in Canadian real estate. I believe I saw somewhere that from 2003-2007 an increase of $1 trillion real esate in provincial real estate (commercial included, likely excluding crown land). If you’re not too interested in where things are going, but where they have been nationally RBC has a good annual housing report
By all means!
I’d like to thank Nick and everyone who has posted; this discussion has taught me a lot. I hope it continues.