Does CMHC have big enough reserves?

This post is more of a bleg; because I'm sure someone reading this knows more about the answer than I do.

An insurance company doesn't need reserves if it is insuring a large number of short term uncorrelated risks. The Law of Large Numbers ensures that its annual claims will equal its annual premiums, if it prices its insurance correctly. Every year exactly 5% of all cars insured will crash, so if you price the insurance at 5% of the claim, plus admin costs, you will break even every year.

But suppose there is the same 5% annual risk of a car crashing, but the risks are perfectly correlated. Once every 20 years, on average, all the cars crash. Then you need reserves. If you price the insurance at 5% of the claim, plus admin costs, you would need to keep at least 20 years' worth of premiums, net of admin costs, in reserve. (That's a minimum, because you might get 2 or more crash years in a row.)

In the 19 fat years, when none of the cars crash, you can't just spend your "profits" from premium income. You can't spend anything, on average, because your expected premiums are exactly equal to your expected costs.

(It must be difficult to estimate accurately the expected losses, when there is a small probability of big losses; fat tails, black swans, blah, blah...)

CMHC insures high loan to value ratio mortgages against default risk. It seems to me that those default risks are likely to be highly correlated. A nationwide fall in house prices, or rise in unemployment, or both together like in 1982, would cause mortgage defaults to increase nationwide. CMHC is closer to the second example than the first. So CMHC needs reserves.

Furthermore, these are not short-term risks. You renew your car insurance every year, and pay a new premium. But when you insure a mortgage, you pay the premium when you first take out the mortgage, and it covers the life of the mortgage. It would be like paying for car insurance when you first buy the car, and it covers you for the life of the car.

Even if car insurance claims were uncorrelated, if you paid the premium up front, and cars lasted 30 years, you would need 15 years' worth of premiums in reserve in steady state, since the average insured car would be 15 years old and would stay covered for another 15 years.

CMHC does indeed have reserves. If I am reading its 2008 balance sheet (pdf) correctly (I'm probably not) it has reserves of $8.17 billion. And annual income from premiums and fees of $1.52 billion. Ignoring admin costs, that's a ratio of reserves to annual premiums of 5.32 years. Is that enough?

Insurance premiums range from 0.50% to 7.4% of the value of the mortgage. I do not know if that's an actuarily fair set of premiums, but it's doesn't sound obviously too low to me. A 5% premium would be about right, ignoring admin costs, if 10% of insured mortgages went bad, and the average loss were 50% of the original value of the mortgage. Not obviously priced too low.

If we buy my assumption that CMHC mortgage insurance premiums net of admin costs are actuarily fair (and they ought to be priced to be roughly fair), here's my question:

Has CMHC been spending any of its annual "profits" from insurance premiums (on "social housing" or whatever)? Or has it been saving those "profits" to add to its reserves?

(If it were giving its annual "profits" back to the taxpayer, I wouldn't worry too much, since the taxpayer is ultimately on the hook anyway for any losses. Though it would mean the national debt is a bit bigger than it is reported to be.)

 

Comments

[Neil]

I think your estimate of losing 50% of the original mortgage value is too high, and is therefore throwing off your estimate. The difference between home insurance and mortgage default insurance is that the home insurance must pay the replacement cost of the house in a loss event. Mortgage default insurance must cover the difference between the liquidation value of the home and the amount that remains owing on the mortgage.

Furthermore, CMHC mortgages are full-recourse, even in provinces where an uninsured mortgage would not normally be. So even if someone has defaulted on their home, unless they go through bankruptcy, CMHC can continue to come after them for the remainder, further reducing their payout requirements.

[Nick Rowe]

Neil: what you say sounds plausible. But that would mean (if you buy my other assumption about 10% of mortgages going bad), that a 5% premium would be too high (higher than actuarily fair).

[Nick Rowe]

I didn't know that CMHC mortgages were full recourse (even in Alberta). That's an interesting fact.

[Neil]

I actually have no real though about the 10% number, might be right, might be wrong. In any event, the 5% premium is an extreme case, I'm pretty sure that's not very close to the average. It would indicate someone who is amongst the highest risk groups that CMHC will cover (self employed, 10% or less down payment), so the likelihood of someone with that risk profile defaulting is likely higher than 10%, and they pay premiums accordingly.

As I understand it, the full recourse thing happens because, being a program of the federal government, they fall outside provincial jurisdiction. A private insurer would not be able to operate in the same way. I'm in Alberta and the full recourse thing was made quite clear by my lawyer - told me to never allow a buyer to assume a CMHC mortgage because if they default I can still be targeted by CMHC.

[Phillip Huggan]

Yep, non-recourse is hugely stabilizing to whole economy the more I think of it. Makes home owners act like home owners and not speculators.
One innovation is staggering redemption schedules since having them all or partially fall at different times is prefereable to all 5 yrs after a low interest rate spike triggers demand. You start off with a line or gentle curve of the interest rates you charge for mortgages, and then as more mortgages are purchased that reset or redeem in a given short duration, the same month or year, you charge a little surtax and the line or curve of future mortgage redemption interest rates rises.
If all mortgages reset rates smoothly over next 20 years is starting point, and banks decide to reset mortgage rates after 5 years, increasing failure risk, you give the first of these a 5% interest rate. The next 1000 5.2%, the next 1000 5.5%....this way either the banks or home buyers are incented to go with a 6 year schedule. This way the pain of simultaneous defaults is spread out over lingering depressing years. Can't go wrong with a novel finance hedging strategy.

[casanova]

I dont know how CHMC insuring morgages no canadian bank want to insure is a sound corporation!! Right now canadian banks are making a risk free profit by giving out morgages to people who cant afford a house and selling the morgage to CHMC.
All canadian banks are concerned is in making morgages that can be insured by CHMC.
They dont care about you nor do they want to know if you are able to pay, if you qualify for ChMC insurance then theye give you a morgage.
Does this sound similar to what happened south of the border? At least fanny and freddy were limited at morgages of value less than 477k USD wheras CHMC will insure all morgages.
Have a look here for some facts, this guy is into smth
http://americacanada.blogspot.com/2009/07/cmhc-and-our-government.html
I beleive we taxpayers are silently through CMHC subsidising the biggest real estate bubble in Canada. To be convinced of this, one has only to look at vancouver prices.
CHMC books look good because prices are still at their peak here, however if prices come down by 10 to 30% which is very probable, the CHMC will be the biggest scandal in canadian history as it will be bailed out by those working people who could not afford a house because of CHMC. What a paradox!

[casanova]

now, have a look at this,
http://www.realtor.ca/propertyDetails.aspx?propertyId=8688066
take a morgage, insure it by CHMC and hope that all goes well. The premium will be peanuts in the books of CHMC once fundamentals reassert themselfs as they have done everywhere inthe world.
CHMC will suffer huge losses, that is the canadian taxpayer.
I like the analasys of this guy too:

http://marketdepth.typepad.com/marketdepth/2009/09/welcome-to-kanada.html

[Patrick]

casanova:

I think your concerns are overblown.

Mortgages in Canada are recourse loans. You can't walk away. If you do, the bank will come after you. Including garnishing future wages. If your house is underwater and the bank forecloses, they will sell the house and come after you for whatever balance on the loan remains unpaid. The only way to get out is to declare bankruptcy. Bankruptcy is no joke; it's liquidation. They clean you out.

Also, as Stephen has pointed out numerous times, the real estate bubble in Canada was nowhere near as big as it was in the US and some other places (Ireland, Spain).

Given that the crazy lending never really became widespread in Canada, that mortgage interest is not tax deductible, and the rather severe consequences of default, I think it's very unlikely that CMHC will blow-up due to defaults.

[Nick Rowe]

There are really 3 possibilities:

1. CMHC is pricing its insurance high enough (given the risks of the people it is willing to insure) but doesn't keep enough reserves because it is spending its "profits" from premiums.

2. CMHC is underpricing its insurance (or willing to insure riskier people than its premiums would warrant), so even if it saves all its premiums its reserves will always be too small.

3. None of the above

[GA]

I can't answer your factual question (is that enough reserves?), but I can try to provide a bit more balance sheet/accounting interpretation. Your statement that there are $8.17 bln in reserves is not correct (although reserves is one of those accounting terms that means different things depending on circumstances). At minimum, the "unearned premiums and fees" has to be included (and really constitutes the 'reserves' in the classic insurance sense, whereas the Equity of Canada - shareholders' equity - is a buffer on top of that). So at minimum their ability to cover losses here is an extra $6.2 billion over what you have stated.

The key thing to understand here is that since CMHC charges for insurance up-front over the term of coverage and it accounts on an accrual basis, the money it gets when it signs a mortgage insurance contract is unearned premium. And, CMHC will expect to pay for its costs (losses on insurance contracts) out of this amount. (Put another way, this is an accounting entry designed to say "all those assets on the asset side equal to the amount of the unearned premiums will need to be used to pay our claims, and don't count as capital or profits yet").

Is this enough reserves? I'm not qualified to say. But we can dig a bit more.

More later (I'll write a bit more separately) but one more point of clarification: in both banking and insurance, 'reserves' and 'capital' are related but not the same thing, and are sometimes used interchangeably. The key extra point is to remember that there is 'regulatory capital', or, put a different way, 'ability to absorb losses before it hits the senior debt-holders, depositors and the like.' Regulatory capital usually includes regular capital (shareholders' equity) PLUS most types of reserves (plus some other odds and ends depending on context and regulations). So when talking about capital, particularly when it comes to banks/insurance activities, need to keep them straight.

[GA]

Ah, you also ask whether CMHC is spending its profit on government programs/social housing. Nope, government pays for those through direct appropriations (payments to CMHC) which CMHC pays out - on both income and expense side. Note that in 2008 the appropriation (grant) for "Housing Programs" to CMHC was $2.288 billion, the expenses for "Housing Programs" $2.288 bln - the government even reimburses CMHC directly for related expenses (CMHC probably has to track these operating expenses/overhead for Housing Programs separately from all other). CMHC also has a portfolio of what one could call 'policy loans', but these are guaranteed by the Federal or provincial governments - so in principle there is no hidden subsidy for interest rates in here for these programs (if anything CMHC keeps a small spread minus overhead - probably net close to nil effect or small profit).

All profits are consolidated in government's deficit/surplus (government doesn't really keep a balance sheet per se). Any profits government does not want to leave in CMHC it takes back; for 2007-2008, it left the profits in rather than dividending to itself, building up loss absorption capacity, what you could call capital reserves). This is _in addition_ to actuarial reserves (I neglected to note in my comment above there is a line item for $0.7 bln in provision for claims, this is another reserve line).

The other key point to note when looking at CMHC's balance sheet/accounts that CMHC _consolidates_ its securitizations (which, like Ginnie Mae, it guarantees). So this is not the off-balance sheet SPV structure global banks have been fiddling with - CMHC retains effectively all the risk and rewards of the underlying mortgages, so this should not raise the fears that exist in other markets.

The other big differences from US mortgage markets are:
-vast majority of CMHC's assets/liabilities are shorter-term (less than five years), and particularly in respect of interest rate risk (all Canadian mortgages are basically "ARMs" to use the US parlance). Long-tail risk is less prevalent when your balance sheet has shorter duration.
-Canadian mortgage insurance market very different than US: full fee is paid up-front. US mortgage insurers receive premiums annually, and generally stop receiving them when mortgage becomes "compliant" (generally when mortgage is repaid or refinanced - or falls below the loan-to-value trigger, usually 80%). So there is an adverse selection issue - lower risk customers prepay, leaving a steadily worsening book (of a given cohort). Given this, a simplistic assumption for analysis might be that any mortgage insurance still on CMHC's book after five years is nil - enough principal has been repaid (or other risk reduction elements such as increase in value of security) that they won't lose money on it even in a bad scenario. Of their $400 billion in nominal outstanding insurance how much does this account for? Half?
-They have a note that discloses how much they expect to lose for every 1% increase in unemployment, 'lower house price inflation', and (gain) for lower mortgage rates. Short form is that sustained high unemployment is by far their biggest risk - consistent with full recourse leading to bankruptcy and low recoveries. Deflation would be bad in many ways for CMHC (not just house price deflation).
-Another big difference from (pre-conservatorship) Fannie and Freddie is that CMHC basically has no liquidity risk and low repricing risk - their debt is backed by full faith and credit of the federal government, so no-one likely to refuse to buy their bonds (other info in the notes suggests quite the opposite, their is excess demand for their bonds).
-They also have no so-called hybrid capital (preference shares and subordinated debt), and hence no higher-coupon debt. (Insert rant here about Fannie/Freddie shareholders receiving dividends/pref share dividends for effectively benefitting from govt guarantees, even if not recognised as such - the most obvious definition of socialized capitalism I've ever seen, where govt keeps all the risks and none of the rewards).
-I could go into more detailed analysis and risk scenarios, and there are some parts of their financial statements I'm not sure I understand completely (see my question below). Suffice to say that both using their numbers (actuarial risk in note 8) and using fairly drastic scenarios, I can generate estimates of capital/reserve charges of $3-8 billion over five years in a quite deep recession/housing correction with unemployment 5% higher than at end-2008. Their capital plus reserves should be enough to cover that with quite a bit of room (allowing for some earnings from ongoing business, which is not really in doubt given full faith and credit). At the low end of a bad scenario, CMHC would have a couple of bad years that would not seriously impact capital.
-As with any long-term insurance company with potentially high correlation between losses (as you note), it would be possible to generate estimates of losses much higher than that, but I don't think they would be that credible. I'd estimate non-performing (defaulting) loans above 15% with loss given defaults above 30% before it really gets anywhere close to insolvency (or mathematically equivalent variants thereon).
-This would be, from the minimal historical data I have, far beyond what seems plausible in Canada (short great depression and worse scenarios). Given lower house price run-up in Canada, some regional variance (like large housing markets more tied to commodities), stricter lending standards, very low prevalence in Canada of neutron bomb loan types (interest only and option ARMs in particular), and mortgages mostly remaining on bank balance sheets, I don't buy it. Plus, of course, Canadian government/Bank of Canada have monetary and fiscal policy tools remaining (unlike, say, Spain and Ireland). Unless somehow intense deflation or stagflation comes to town again (high rates and high unemployment), hard to see major risks for CMHC. If Bank of Canada had to pull a Volcker for an extended period of time, CMHC would probably face its biggest challenge. No sign of that for time being.

Nope, however much one might want to hate CMHC and predict disaster, it's position is not that bad.

Now, my question: there is a very interesting disclosure in Note 10 - 'The carrying amount of CMB at 31 December 2008 is $7,885 million higher than the contractual amount due at maturity. CMB are guaranteed by CMHC and CMHC’s liabilities are backed by the full faith and credit of the Government of Canada. Therefore, _there is no significant change in value that can be attributed to changes in
credit risk.'_

I intrepret this to mean that CMHC is actually showing debt of $7.8 billion _higher_ than it expects to have to pay in nominal value. I cannot confirm this interpretation right now, but if anyone can clarify, I would much appreciate. If true, this would effectively mean that CMHC will show $7.8 bln _extra_ net income over roughly the next five years. (This is an accounting artefact due to the fact that CMHC's debt is now priced so highly - another aspect of the flight to quality).

I won't bother going into details on another line item or two, because their interpretation is also arguable, but I believe there is another line item or two where CMHC is currently showing a much more conservative value than justified. In short, CMHC has the potential to basically show very large accounting profits over the next five years (basically because it borrows from government at a low cost and has bought assets with a bigger spread than ever before). It has the same potential to profit as the Federal Reserve does (print money and use it to buy financial assets), but with a much cleaner balance sheet.

If this were a stock, I would not be putting this down on paper, just buying it. My point is, contrary to some of the comments and the basic assumption that government always screws up, I'm willing to bet (socially) that CMHC will post very large profits/gains for the taxpayer over the next several years under almost any scenario.

I have no connection to CMHC, never even had a CMHC mortgage, could be completely wrong, have no qualifications that allow me to form opinions of any kind. Don't own property and personally have a negative scenario going forward - I think house prices in Canada will fall, but not much, and more likely through very slow price appreciation for eight to ten years. This is not investment advice, etc. This entire message has been machine generated in the future.

[Nick Rowe]

Thanks GA! I said I was sure that somebody reading this would understand it better than I do. And GA clearly does. I need time to read what you wrote more carefully, and digest it.

[casanova]

GA, I will not take issue with all your points, to long to debate it here, however you say :
"Given lower house price run-up in Canada,....."
let me remind you some facts(all figures are in USD):
avearage home price in US at peak : 230 200 USD
average home price Canada at peak : 278 362 UDD
Gross National Income per Capita (World Bank 2007 $US) USA: 46 040 USD
Gross National Income per Capita (World Bank 2007 $US) Canada: 39 650 USD
Ratio of Peak Home Price to Income USA: 5 times
Ratio of Peak Home Price to Income Canada: 7 times
Date of peak USA: Q3 2006
Date of peak Canada: Q2 2009 or maybe latter, the buuble is still inflating thanks to CHMC

You can spin anyway you want the CHMC balance sheet, but all those morgages have been insured by CHMC and I dont see how their balance sheet can be so sound when the asset they insure is a "geant avec des pieds d'argile"

[GA]

Casanova: the balance sheet is there, disagree where you like.

could you provide sources for these figures? If accurate, I'll give it some thought (although need to correct for US$ changes). Are World Bank figures PPP or straight?

Keep in mind that the US and Canadian housing markets are quite different. For example, as far as I recall, Canada is much more Urban than the US - which could easily lead to higher average house prices. By contrast, U.S. housing prices may have a much higher standard deviation (apart from Vancouver, fewer crazy high price markets, and biggest run-ups were in oil-bound markets, where at least there is some logic). I question whether the ratio of home price to income is the best, and whether it should be compared in absolute terms or changes over time.

And it is categorically NOT true that all those mortgages have been insured.

I'm not trying to spin anything, just my opinion.

[Jean Cooper]

GA - a couple of things, one you state: "Given lower house price run-up in Canada, some regional variance (like large housing markets more tied to commodities), stricter lending standards, very low prevalence in Canada of neutron bomb loan types (interest only and option ARMs in particular), and mortgages mostly remaining on bank balance sheets, I don't buy it. Plus, of course, Canadian government/Bank of Canada have monetary and fiscal policy tools remaining (unlike, say, Spain and Ireland)."
Take a look at this comparison of US and Canadian run up of housing prices, and tell me Canada did not run up prices as much as the US.
http://mjperry.blogspot.com/2009/05/banks-and-housing-markets-canada-vs-us.html
And if you look at the Teranet index, our housing prices have turned up again after an small drop.
http://www.housepriceindex.ca/Default.aspx
I know a couple personally who took out an interest only loan, and I'm sure they're not the only ones. CMHC brags about allowing people to put only 5% down for a house. I can't find any statistics in Canada that shows what the amortization is for various mortgages, i.e. how many are 35 year, 30 year, etc., nor can I find out the average down payment. I would bet that there are a lot more neutron bomb mortgages than we are led to believe, but as I say, can't find any hard data on them.

And at 1/4% overnight rate, Canada does not have many more fiscal tools in the toolshed.

Also, full fee is not necessarily paid up front, see:
http://www.cmhc-schl.gc.ca/en/co/moloin/moloin_010.cfm
I would imagine most people roll it into their mortgage, especially those that need insurance, i.e. they don't have a big downpayment.

You also state: "Short form is that sustained high unemployment is by far their biggest risk - consistent with full recourse leading to bankruptcy and low recoveries. Deflation would be bad in many ways for CMHC (not just house price deflation)."
Yes, unemployment is a big risk, but increasing interest rates I believe is even bigger. How long can rates stay this low? Weren't low rates in 2001-2003 in the US one of the main causes of the housing bubble? When the rates reset the bubble started to burst. Cheap money. The government has lowered rates to stimulate the economy, which means people take on more debt to consume. There is a limit to how much debt people can take on, and we are rapidly approaching that limit. The government may be able to control the short term interest rate, but longer term rates aren't under their control, but under investor's control. If investors come to believe that Canada is in a housing bubble, debt won't be as easy to sell, and rates will have to rise.

CMHC may be in good shape now, but if interest rates go up, I believe they, oops, I mean we, will be in big trouble.

[Mark]

Very good insight GA. I've bookmarked your comment for future reference.

Jean Cooper: I'll tell you that prices in Canada did not run up prices as much as the U.S. (tip: look at the y-axis scales).

[Steve]

Isn't there a problem with GAs analysis - using national totals, whereas the problems have proven to be very regional in nature, with foreclosures (both in terms of banks and households) being focused in a number of key states (e.g. Arizona, California, etc.)? Not an economist, a geographer by training and one of the things that has interested me is the strong spatial character of the collapse of the housing market.

[www.facebook.com/profile.php?id=611985302]

GDP per capita is not the right measure. Income distribution is very different in Canada than the US. Median family income is a much better indicator.

From StatsCan
http://www.statcan.gc.ca/daily-quotidien/090603/dq090603a-eng.htm

From US Census
http://www.census.gov/prod/2009pubs/acsbr08-2.pdf

They are not directly comparable, since the US measure appears to include unattached individuals which StatsCan separates out. But for families it appears that median income is HIGHER in Canada.

[Jean Cooper]

OK Mark, point taken. From looking at the raw data, US prices went up by 106% in 9 years, and Canadian prices "only" went up by 92%. Still in bubble territory as far as I'm concerned. US prices have corrected by about 31% since the peak, Canadian prices corrected by about 7%, and are now on their way back up. Who knows when our peak will come, but suffice to say that house prices cannot rise by 10% a year forever.
One thing that GA referred to, and that I don't really understand, is who is on the hook for the mortgage if it defaults? Let's say, as GA says, that after 5 years someone is below the 80% threshold, (they've paid off 20% of their mortgage). Then they default. If that person cannot pay, and goes bankrupt, is CMHC going to pay out to the banks, or are the banks going to take the loss? Does CMHC insure the entire life of the mortgage, or only til it reaches that 80% threshold?

[GA]

@Jean: I believe CMHC remains on the hook for the risk of a mortgage (remaining loan balance) over its full life. My point is that the vast majority of the risk CMHC bears (even in the posited house price run-up and subsequent fall) is in the first five years, so the age of the portfolio is a critical indicator. Once below 80% LTV (or 75%, or whatever level you prefer), the expected losses will be much lower. And remember, LTV will depend on market value at default - and in a house-price run-up, the older the portfolio, the more likely the LTV (risk remaining to CMHC) will have _fallen_.

@Steve: yes, geographic distribution is key, and you may have better tools to analyse this than I do. In the US/Canadian comparison, we have nothing comparable to Florida and California (in many ways, and specifically with respect to the housing bubble/lending disaster) as far as I am aware.

@Jean: I believe CMHC receives the fees up-front, and when the payment is spread out, it is added to the mortgage balance (essentially, the bank finances it). I also don't deny that there has been a price run-up (possibly excessive), that aggressive lending policies also exist in Canada, nor do I argue that Canada's economy is 'out of the woods' (the road through the woods is through the U.S.). My point is that the comparisons to the U.S. housing bubble and particularly the implicit comparisons to the problems faced by e.g. U.S. mortgage insurance companies or the housing GSEs are not exact. And specifically, my opinion (analysis) is that CMHC is not in as terrible shape as some think.

CMHC's analysis and audited statements show that interest rate risk is not a big driver for their losses on insurance. You can disagree, but I don't agree with your analysis, and CMHC's accounts - audited extensively and based on historical data - contradict you.

Apart from that, the only way in which short-term interest rates will go up significantly in the near term is if unemployment drops (we should have such problems). And you're actually mistaken about the long-term/short-term point: most mortgages in Canada are priced off of short-term interest rates, and so what long-term rates/investors think mostly doesn't matter: Bank of Canada can control this. (Read around for one of the conundrums of U.S. monetary policy, which is that since mortgages are mostly priced off ten year rates, monetary policy there has more lags and issues than in Canada, where short-term rates drive the market). The scenario you propose doesn't make sense, and doesn't fit the structure of the market - BoC sets short-term rates, and borrowers will go floating if price of fixing longer-term is too high.

The Volcker scenario - inflation gets much higher and BoC is forced to raise rates (and unemployment) to kill it - is many years away. While it would be bad for CMHC, it would also be after several more years of reserves build-up, and - very, very important - the inflation that creates the danger would erode the value of the debt and (all else being equal) support housing prices. Both would reduce losses to CMHC.

Nope, biggest risk to CMHC in immediate future is a great depression (massive unemployment and deflation). If you truly believe in that scenario, though, CMHC should be the least of your worries.

[Greg]

@Jean: 'Take a look at this comparison of US and Canadian run up of housing prices, and tell me Canada did not run up prices as much as the US.
http://mjperry.blogspot.com/2009/05/banks-and-housing-markets-canada-vs-us.html'

Canada did not run up prices as much as the US.

Apart from the fact that this graph uses different scales, as noted above, the article you reference says as much: 'it would appear so far that there was a much bigger housing bubble problem in the U.S. than in Canada.' US prices more than doubled, Canadian prices did not.

Even more importantly, you need to correct the run-up for the period of time that each took place in. I give the US run-up 6 years to peak, the Cdn 8 years. Correcting for this difference, US run-up was 12% annually, Canadian 8% annually. Since this increase is compounding (as is the underlying inflation, GDP growth and population changes that can also be expected to drive house prices), 4% annual difference is an enormously faster rate of home price appreciation over the time period considered.

Unfortunately I don't know enough about either index to comment on whether they are meaningfully comparable. Case-Shiller, I know, is based on tracking repeat sales of the same properties, so would be expected to possibly understate price appreciation IF a significant amount of price growth took place in newer properties (McMansions anyone?) or build-out of entire new towns (Inland Empire anyone?). Now, they are smart enough to attempt to correct for this, but don't know if they did or how effectively they might have done. So to the extent that Canada had less of this (if so), it is quite possible that the indices are biased in a direction that understates the price run-up in the U.S.

[Jean Cooper]

GA, not quite sure what you mean when you say "most mortgages in Canada are priced off of short-term interest rates". By short term rates do you mean prime? If so, then you're saying most people are taking the variable rate, and not locking in for 5 years? 5 year rates are 5.5% and 10 year rates are 6.9%, though I would guess not too many people are taking those.

You state:"CMHC's analysis and audited statements show that interest rate risk is not a big driver for their losses on insurance. You can disagree, but I don't agree with your analysis, and CMHC's accounts - audited extensively and based on historical data - contradict you."

Forgive me for my cynicism, but didn't Moody's and S&P give triple AAA ratings to $billions of CDOs based on historical data, which proved to be less than accurate? When a housing crisis hits, as it did in the states, historical data isn't much use. I'm pretty sure their historical data didn't include a 30-40% drop in housing prices. And the reason for the defaults in those CDOs was that interest rates went higher, and people could no longer afford their mortgages. Now interest rates in Canada cannot get much lower, so the only way to go is up. Will the BOC keep interest rates at .25 for 20 years like Japan? who knows? The fact that interest rates for a 5 year mortgage is 5.5% and 10 year is 6.9% leads me to believe that the banks think interest rates will go up in the future.

You say that rates will remain low in the near future, you're probably right. The thing is, mortgages are long term debts. I would imagine, given the price of houses, and the average Canadian salary, that a lot of people are taking 30 and 35 year terms to make their monthly payment affordable. So we may be OK for a few years, but when rates start to rise, and peoples' monthly payment start to go up, houses will become very unaffordable. (a shift from a variable rate of 2.55%, to 5% will cause monthly payments to go up by 40%) And that kind of rate increase is not beyond the realm of reality. At the beginning of 2008, the 1 year variable rate was 6%.

And I think lending here is just as risky as it was in the states. We may not call it subprime, but we are doing the same kind of lending. CMHC makes a big deal about how you can buy for as little as 5% down. There are a ton of mortgage brokers who can get you a no money down mortgage. TD will even give you 5% cashback on a mortgage. CMHC is insuring the riskiest mortgages, and probably buying the riskiest mortgages from the banks to package up for investors. Let's hope CMHC's historical data holds up in the future.

[GA]

@Jean: I grant that if indeed there is a bubble-to-end-all-bubbles the historical loss rates will not be much use. But (as for reasons above) I don't believe the analogy fits (although this is far from saying Cdn housing prices will not fall).

My point on currently low rates, inflation in the future, and priced off the short term is that a) in the US, most/many mortgages are 30-year fixed rates, but in Canada, generally priced off prime or max up to five years; b) that being the case, Bank of Canada can lower real mortgage rates much quicker; c) there is little likelihood BoC will have to raise mortgage rates in near future; and therefore, any losses CMHC will have to face due to sharply higher mortgage rates will be several years out and moderated by other factors and smaller compared to loss absorption capacity now.

Put another way, most scenarios in which BoC raises (mortgage) rates will be due to inflation (and likely linked to higher employment in the interim); any raise in rates due to inflation and higher employment reduces CMHC losses.

There are scenarios that would be worse, of course.

The existence of risky mortgages in Cda does not show lending was just as risky, because the proportion of high-risk ones is the important factor. You say you think that lending in Cda was just as risky as in the states. This is not true; if you have data, bring 'em on. I've looked at both, nothing comparable - the proportion of risky loans in Canada was lower.

[Jean Cooper]

@GA - I have tried to find data on mortgage lending, but haven't been able to find much specific information, such as: what length of mortgages are people taking out, how much do they put as a down payment, how many people have utilized HELOCs, etc., and I've looked around quite a bit. If you could point me in the direction where I can find some of these numbers, it would be greatly appreciated.

The only thing I have to go on is anectdotes and aggregate numbers. I won't go into anectdotes, but here are some aggregates...

From CMHC 2008 annual report (page 58), they issued insurance on $148 Billion worth of mortgages in 2008 and $125B in 2007 (for a total of $273B).

From Statistics Canada (http://www40.statcan.gc.ca/l01/cst01/fin21-eng.htm), the total residential mortgages outstanding at end of 2008 was $863 Billion. This includes mortgages issued over the past 35 years (or even longer, as 40 year mortgages were allowed for a short time).

So over the past 2 years alone, $273B worth of high ratio mortgages have been issued, which is almost 32% of total outstanding mortgage in Canada.

I think this points to risky lending, but again I don't have don't have specifics. I don't know if people are putting 15% down or 5% down (or 0 down, maybe getting that 5% cashback from TD for downpayment). CMHC isn't very specific, but if someone has to be insured, then it's a high risk mortgage. And I would bet money that downpayments with CMHC insured mortgages is closer to 5% than 15%.

I find it interesting that StatsCan shows total mortgage increase of $90.5B from 2007 to 2008, but CMHC insured $148B. It could be that $58B was paid off by mortgagees, maybe it was people utilizing HELOCs, and that sent them over the 80% mark, and they had to insure with CMHC. Without specifics, it's hard to see exactly what's happening.

I'd be interested in seeing any data that you might have (you've seen mine!)

[Nick Rowe]

Very high quality comments on this post.

[Stephen Gordon]

Indeed. I think the next time I've got a question I'm not sure how to answer, I'll do one of these kinds of posts instead of trying to thrash out the data on my own.

It's good to know that we now have an impressive array of expertise to call upon here at WCI.

[GA]

@Jean: thanks for the datapoints. It will be difficult to do any direct comparison, of course. I'll try to dig up what data I can, but travelling for a while (in places with not-so-good internet).

[Andrew F]

"So over the past 2 years alone, $273B worth of high ratio mortgages have been issued, which is almost 32% of total outstanding mortgage in Canada."

You're comparing a stock to a flow. At least some of those recent mortgages were repaid during the timeframe. Given that this is likely due to speculation (flipping pre-construction condos, as well as existing homes), I'm not sure we should find that encouraging.

[Derek Bruce]

I just did a search on google about CMHC and thankfully this thread came up. I am absolutely convinced that this ridiculous run-up in house prices in 2009 is mostly because of CMHC. We are in the middle of an awful recession and somehow miraculously we are hitting all time highs in real estate prices, seems a little weird, no??? Obviously interest rates are also playing a role but you can't tell me that banks would be making half of these new loans if they had to keep them on THEIR books or without insurance.

Anyways, I took a look at those CMHC financial statements and I wasn't surprised to see the huge increases in planned insurance increases and coverages planned for 2009. Could it be that CMHC was mandated by government to approve every loan in sight so that Canada didn't have to have a housing bubble burst too??? Nobody wants to see a bubble burst on their watch as everyone has seen how painful it's been down in the US. CMHC seems so well meaning on the exterior - we just want to make it affordable for all Canadians to own house.

As for regionality, of course that's at play. BC is our Florida/California/Vegas/Phoenix....

Anyways, the bubble is still flying high here for now. But as with all economics, things will come to roost and i'm off the mind that things will probably start to crash again this fall. And even worse next year when interest rates start to rise.