CMHC reserves revisited

There are two questions that Canadian taxpayers need to ask about the Canadian Mortgage and Housing Corporation:

1. Are CMHC fees for mortgage insurance high enough?

2. Does CMHC have a big enough reserve against potential losses?

The first question is the most important question. Because if the answer to that question is "yes", then CMHC mortgage insurance makes Canadian taxpayers better off, quite apart from any other benefits CMHC mortgage insurance may or may not have. It's also impossibly hard to answer. Because it is very hard to estimate the expected losses from low probability events. History is not a very good guide to the future for low probability events. Maybe we just got lucky in the past.

The second question doesn't really matter. Since the taxpayer owns CMHC, a $10 billion dollar loss to CMHC is a $10 billion dollar loss to the taxpayer, regardless of whether the CMHC can cover that loss out of reserves. Because the taxpayer owns those reserves. But setting money aside in a contingency fund forces us to recognise that bad things can happen, so we don't fool ourselves into thinking we are richer than we are.

I last looked at CMHC reserves in September 2009. At that time (as best I could figure it out), CMHC had $8.17 billion in reserves. (They seem to call it "Total Equity in Canada", which I guess is a nod to the idea that it is the Canadian taxpayer's CMHC-rainy-day-fund.) From the September 2013 quarterly report (pdf) that number has now increased by 85%, to $15.148 billion. That's good news.

But we need to compare that increase in reserves to the increase in the size of potential losses. If the size of potential losses had also increased by the same 85%, the good news would be cancelled out.

The 2008 to 2013 Financial Highlights (pdf) shows that "Insurance-in-force" has increased from $408 billion in 2008 to $573 billion (planned) in 2013. That's a 40% increase. (And the percentage of that due to "High ratio homeowner" has declined slightly from 57% to 53%). And "Securitization Guarantees-in-force" has increased from $234 billion in 2008 to $413 billion (planned) in 2013. That's a 76% increase.

So, as far as I can see (which isn't very far), that is good news. I do not know whether CMHC fees are high enough, and I do not know whether CMHC reserves are big enough. But the reserves do seem to be bigger, relative to maximum potential losses, than they were five years ago.

I wrote that post in 2009 because I was concerned about what would happen to CMHC if house prices crashed. According to Teranet National Bank data, Canadian house prices dropped about 10% from the peak in 2008 to the trough in 2009, and have increased about 30% since that trough. CMHC's "Net Insurance Claims expense" went from $372 million in 2008 to $1,112 million in 2009, when house prices fell and unemployment increased, then dropped to about half that amount since.

2009 was a bad year for CMHC losses. But despite that $1.1 billion in Net Insurance Claims expense, CMHC reserves increased by over $1 billion both in 2009 and in 2010. It would take something much worse than 2009, at least 10 times worse, to wipe out CMHC's reserves.

Which is good news. We have been more fiscally prudent over the last 5 years than I thought we were.

That's if I'm reading this right. Which I probably am not. Because I don't understand these financial statements very well. Which is why blogs have comments.

Comments

[honeyoak]

One thing that worried me about CMHC is the difficulty of figuring out how the insurance works. I once asked an economist who worked at CMHC and he told me that he did not know as he worked in the forecasting division. If I insure a mortgage with a 10% down-payment does CMHC cover 100% of the morgage outstanding? 90%? Par value? The difference between the recovery price and par? Who manages the defaulted assets for CMHC?

[Nick Rowe]

honeyoak: the way I have thought it works: your 10% downpayment doesn't matter (except the CMHC insurance premium is bigger with a smaller downpayment; if you borrow $100,000 from the bank, then CMHC is on the hook for $100,000, minus what you have already paid down, plus any interest payments you missed, minus the recovery price of the house. I don't know who manages the defaulted house.

So a perfect storm for CMHC is high loan to value mortgages, that have recently been taken out, so that little of the principal has been paid down, plus a fall in house prices, plus high unemployment, plus rising interest rates. (Empirically, IIRC, it is high unemployment rates that are the biggest factor in defaults.)

[rsj]

The first question is the most important question. Because if the answer to that question is "yes", then CMHC mortgage insurance makes Canadian taxpayers better off,

Then why not increase the fees by a factor of 50? Because it would make those with mortgages worse off, right? So the issue is not whether taxpayers are better off, since they are only part of the equation. The other side is whether it makes those buying a house worse off. And really as most taxpayers are also homeowners, the question is what is the net transfer from those who are not homeowners to those who are.

If the goal is to set fees in such a way so that no net transfers occur to homeowners, then why have a CMHC at all?

The goal of having a special federal insurance program for mortgages (and not, say, small business, auto, or credit card loans) is to reduce the risk premium charged to those purchasing a house as opposed to those purchasing a car, starting a business, or attempting to smooth consumption. Why is it a national policy to reduce this particular interest rate? The amount by which this interest rate is reduced is a subsidy to homeowners from the rest of the population, who would otherwise receive this foregone interest income. From that subsidy needs to be subtracted any fees that are paid, and added any defaults.

If all of these subsidies exactly summed to zero, then you might as well not have the CMHC. The CMHC exists only to not have these subsidies sum to zero -- i.e. to transfer income to homeowners from non-home owners.

Given that there is a policy to create net transfers, the question should be how much do you want to transfer, and whether a particular set of rates/fees effects the desired transfer.

[Phil Koop]

The CMHC is insuring losses; what they guarantee is timely payment of interest and principal. This is de facto the difference between the loan recovery and the market value of the loan.

The insurance premium is necessarily charged at the level of the individual mortgage. However, many (most?) CHMC-insured mortgages are destined to be pooled and packaged into negotiable bonds; in that case, it is the bond cashflows that the CMHC guarantees.

[Phil Koop]

I don't agree that the first question is "impossibly hard to answer" because I don't believe that mortgage defaults qualify as low-probability events relative to the amount of available data. Sure, the shape of the tail of the aggregate loss distribution (i.e., the maximum that the CMHC could lose in a bad year) is unknowable but that relates to your second question, not your first. It's not that difficult to measure the expectation.

[rsj]

Phil,

Yes, the bond cashflows are guaranteed, but I am assuming that the bonds are correctly priced so that the net effect is to provide lower rates to borrowers rather than arbitrage income to bond purchasers.

I am also sure that credit card, auto, and other business loans are also backed by bonds and often pooled together, yet these borrowers do not enjoy subsidized interest rates.

Of course if the goal is for the fees charged exactly match the subsidy provided, then why guarantee the loans at all?

[Determinant]

I don't know who manages the defaulted house.

AIUI the bank does. The bank is ensured for its equitable loss, but it still has to manage the defaulted assets. It's the one that exercises foreclosure or power-of-sale, depending on which province you live in.

[Nick Rowe]

rsj: insurance can make both sides better off. I think taxpayers could reasonably insist that they be made no worse off, as taxpayers. Because not everyone owns a house, or wants a high-ratio mortgage. Why the government should be involved in this particular insurance business is a good question, but one I am ducking.

Phil: There were two items in the financial statements: "Insurance-in-force" and "Securitization Guarantees-in-force". Should those two items be added together to get the maximimum possible losses, or would that be double-counting? Any idea?

The way I think about it is that the probability that an individual mortgage will default varies a lot year by year. It's not like car accidents, where you can estimate fairly accurately the aggregate number of crashes next year. 1982/3 was probably the worst year so far for CMHC (their web doesn't have financial statements going back that far). But years that bad are rare. I'm not sure how much the expectation is affected by what is happening out on that tail.

[Peter N]

What sort of creature is the reserve? Is it invested like a sovereign wealth fund, or is it a promise to tax, like the social security trust fund?

[jesse]

Two things to consider:
1) If prices drop, CMHC IIF will most likely increase as those whose equity has decreased will be asked to insure the lender.
2) To answer the reserve question you need to know recovery rates, and it gets very complex if CMHC insured loans are subordinate.

My estimates in the past have assumed a 50% recovery rate (surprise! Loans in default are often the loans with the least equity) and a default rate in line with past experience in the 1990s in Ontario but applied countrywide.

My guess was a government bailout (loan to CMHC) of about 30-40 billion dollars but spread over many years. Like in the early 1980s that will mostly be recovered, but in net taxpayers will be directly on the hook for 10-20 billion (wild guess but with precedents). CMHC would most likely increase its fees and could have much of its "good" book sold off to raise cash. That sucks but a relatively minor drop in the bucket as % of GDP.

If taxpayers are worried it should be on labour market disruptions if housing prices fall. That seems a much bigger impact on top line revenues than CMHC "shenanigans".

This is a stress test, so doesn't mean it's likely, but it is plausible and possible.

[honeyoak]

Thank you Nick, that was very informative. If CMHC takes on 100% of the credit risk for these loans, why on earth do the banks get to keep 3%-5% annually for themselves?

[Declan]

"It would take something much worse than 2009, at least 10 times worse, to wipe out CMHC's reserves."

A couple of notes of caution, a 20% drop in prices might be 10 times (or more, or less) worse than a 10% drop, as the relationship between price drops and losses is non-linear over that range.

Another thing to consider is that in 2009, most current prices were much higher than original purchase prices due to rapid price increases in previous years. Due to stagnation of prices across large market segments since 2008, there is much less cushion due to prior price increases than their was heading into the previous price downturn in 2008-2009.

The best approach to assess the risk you are talking about is to stress test the book using severe scenarios to see what happens (e.g. what if we had a downturn that matched the case-shiller 20 city decline (35%) over a number of years).

If you check the CMHC website and financial results, you'll see that they do in fact conduct this sort of stress testing on an annual basis and it influences their capital targets and other risk appetite measures, but until that sort of severe downturn actually comes, we'll never know how good their stress testing really was.

[Nick Rowe]

Declan: "A couple of notes of caution, a 20% drop in prices might be 10 times (or more, or less) worse than a 10% drop, as the relationship between price drops and losses is non-linear over that range."

Yep. It will definitely be non-linear. A very crude approximation would be a quadratic. The number of defaults will be proportional to the decline in house prices, and the loss in each default will be proportional to the drop in house prices, so expected losses will be proportional to the drop squared.

But, from my memory of empirical studies, it's the unemployment rate that matters most. If house prices drop 20%, but unemployment doesn't rise like it did in 2009, my guess is it won't be as bad. If people keep their jobs most of them will keep paying the mortgage, because you usually can't do "jingle mail" in Canada (mail the keys to the bank and walk away from the house and mortgage).

[Tom Brown]

O/T: Nick, if an NGDP futures market were off the table, do you think it's possible to devise an effective algorithm to replace the central bank decision makers and to implement NGDPLT (using traditional central bank tools, such as OMOs)? Would you publish the details of the algorithm so that it was public knowledge? Assume you had as much help from experts in any field that you needed for such a task.

[Tom Brown]

... I mean c'mon... John Taylor came up with a Taylor rule, right? Any MMist worth his salt ought be be able to do at LEAST that good! How come there's not a "Rowe Rule" for NGDPLT? :D

[Tom Brown]

... sorry, this just amuses me a bit: Is the idea of reducing "Chuck Norris" to a set of instructions distasteful somehow? But look at it this way: it'd get those pesky plodding "people of the concrete steppes" off your back! :D

[Tom Brown]

... OK, last bit on this I swear: What if we got Chuck Norris to write the algorithm, and called it the "Norris Rule?" Would that lend it the requisite cachet?

[Nick Rowe]

Tom: It could be done, but why bother? The Bank of Canada does not follow a Taylor Rule now (I don't know of any inflation targeting central bank that does). My own research (with David Tulk) showed that *any* simple rule (like a Taylor Rule) would do worse than the Bank of Canada actually did at inflation targeting, by using its models and judgment. It would probably be the same with NGDP targeting.

[Tom Brown]

Nick, thanks. I'll take a look at your link. I wasn't aware that you'd done that.

[Tom Brown]

Nick, Sadowski had this to say (in case you're interested):
http://www.themoneyillusion.com/?p=25832#comment-314215

[Majromax]

@Phil Koop:

> Sure, the shape of the tail of the aggregate loss distribution (i.e., the maximum that the CMHC could lose in a bad year) is unknowable but that relates to your second question, not your first. It's not that difficult to measure the expectation.

The shape of the tail is very important. CMHC doesn't need to insure against the average loss, it needs to maintain reserves against rare but large losses -- it's looking to have capital against the 5% or 1% or 0.1% event. That's related to the higher moments of the distribution, and that's precisely where the shape of the tail comes in.

The first question, of course, is the time derivative of the second. If reserves against losses will not be large enough in the future, that suggests the current premiums are too low.

[Avon Barksdale]

Nick,

I think that your discussion mischaracterizes how the Government of Canada actually creates mortgage insurance. CMHC together with Canada Housing Trust (CHT) create mortgage insurance through securitization. There really is no “money” set aside.

In effect, CMHC with CHT engineer a credit default swap. In a competitive market, a prospective homeowner would pay an interest rate commensurate with her default risk. The Canadian government (in its infinite wisdom) has decided long ago that the competitive market interest rate for homeowners is too high and thus the government subsidizes residential mortgages. The government accomplishes this subsidization by selling mortgage bonds and various mortgage backed securities to institutional investors (i.e., hedge funds, etc.) which swaps what would be the competitive market interest rate for our prospective homeowner with that of the Government of Canada's borrowing rate. Institutional investors are willing to hold CHT's debt instruments at low rates because if the the mortgage cash flow is interrupted by homeowner default, the Government of Canada will make up the difference. CMHC/CHT “insurance” is nothing more than a fancy swap - at the end of the complicated arrangements between the mortgage originator and CMHC/CHT, the institutional investor lends to the homeowner at a low interest rate with the proviso that if default occurs, the Government of Canada pays. There is no “reserve” sitting in a pile as such - this is finance after all - each dollar is rehypothecated many times over. And in the government's case, the "reserve" is really the power to tax, which is the reason the institutional investor is willing to hold the debt at a low rate in the first place. This subsidy forces the default risk on to the taxpayer giving banks (the mortgage originators) a free pass to operate with very little equity. Thus, we create the illusion that the Modigliani-Miller Theorem doesn't hold in banking.

What is so funny about the entire situation is that government's actions just resets the housing price level to a new equilibrium, making homes more expensive which requires further subsidization (wash, rinse, and repeat). I take issue with your suggestion that debt subsidization makes society better off if the CMHC premiums are set “correctly”. The elephant in the room is that the financial system becomes distorted from the subsidy by encouraging financing through run prone debt instead of equity (the origin of too big too fail). We could get around this mess by just using markets to set residential mortgage rates, but that doesn't buy votes in Ottawa or curry favour on Bay Street.

[Nick Rowe]

Avon: "...at the end of the complicated arrangements between the mortgage originator and CMHC/CHT, the institutional investor lends to the homeowner at a low interest rate with the proviso that if default occurs, the Government of Canada pays."

Yep. The taxpayer is on the hook if the homeowner defaults. Which is a subsidy. But in return for that subsidy, the homeowner must pay a fee (or a tax), and it's a non-trivial fee, equal to a few percent of the total mortgage, increasing with the loan to value ratio. If, over the long run, the taxpayer earns more revenue from that fee/tax than is paid out to cover defaults (with an adjustment for the taxpayer's risk-aversion), the taxpayer is better off, and it is more like a mutually beneficial trade between taxpayer and homeowner than it is a subsidy.

It is only a net subsidy if the fee is set too low. How exactly we estimate what would be "too low" is the tricky question.

[Avon Barksdale]

Nick,

I am afraid that the taxpayer is not better off. Financial economics is fundamentally about understanding risk premia, so let's trace the risk.

CHMC /CHT engineers a swap which effectively trades the comparative advantage of the Government of Canada's ability to borrow for that of the homeowner's. Without the swap, the competitive market would set a higher interest rate for residential mortgages. That risk has to be held by someone. If the homeowner pays the “fair” price (i.e., market price) for CMHC/CHT insurance, she would find that her interest rate would be the same as a competitive market and CMHC would have accomplished nothing. Swaps, which are derivatives, cannot turn a silk purse into a sow's ear – you can split the risk anyway you like, but someone still has to hold it. With CMHC/CHT, the Government of Canada is not just a market maker, but it is the counterparty to a credit default swap. The risk is held by the taxpayer; so if the taxpayer enjoys returns it will be for holding risk. This is not a responsible use for government – there is no market failure being addressed, just a vote buying exercise. Government sponsored mortgage insurance turns part of the government into a hedge fund while distorting the financial sector by explicitly favouring debt over equity. And we wonder where too big to fail comes from...

[Nick Rowe]

Avon: "I am afraid that the taxpayer is not better off."

You can not say that, without talking about CMHC insurance premiums.

Suppose the government made it compulsory to buy government-provided flood insurance. If the insurance premium is too low, relative to the risk of floods, the taxpayer will lose. But if the insurance premium is high enough, the taxpayer will gain.

Now, is there some sort of market failure, so that both the homeowner and the taxpayer would gain from compulsory government-provided flood insurance? That is a different question. One I do not address in my post.

One possible argument is that the taxpayer has a comparative advantage in selling insurance. Because we are talking about the pool of all past, current, and future taxpayers. The market cannot do deals with the dead and the unborn. Plus, the income tax system already exists, so transactions costs would be lower.

A second possible argument would be based on adverse selection.

A third possible argument would be based on the Samaritan's dilemma: since the government knows that it will be unable to precommit future governments to not bailing out homeowners (and their banks) if the houses flood, (or bailing out a private insurance company that defaults when there is a flood), the government charges a fee for the insurance it knows it will be paying anyway.

[Avon Barksdale]

Nick: “You can not say that, without talking about CMHC insurance premiums.”

Yes I can. Think about it. The government allows homeowners to borrow at REDUCED rates - including the costs of the mortgage insurance - relative to what they would have received in a competitive market. That means the government has assumed default risk that the mortgage originator refused to hold. This is an identity Nick - if you don't have to pay the market interest rate commensurate with your credit risk, you are subsidized. When you buy insurance for anything, you pay MORE than the expected loss. CMHC insured mortgages in Canada allow borrowers to get rates BELOW what would have been the competitive market rate. Make no mistake Nick, CMHC/CHT is a debt subsidy and that's why it is so insanely popular.

Nick: “One possible argument is that the taxpayer has a comparative advantage in selling insurance. Because we are talking about the pool of all past, current, and future taxpayers. The market cannot do deals with the dead and the unborn. Plus, the income tax system already exists, so transactions costs would be lower.”

Really? The government has a comparative advantage in selling insurance? Is that what we want government to be, an insurance company with an army? As for the unborn, you are talking about dynamic inefficiency arising from an infinite number of agents with overlapping generations (perhaps you use David Romer's example in his book for your undergraduates on this topic). There is absolutely no reason to think that the Canadian mortgage market is dynamically inefficient. And lower transaction costs? I seriously doubt that as a plausible argument for CMHC.

Adverse selection, etc, etc., this is reaching. We would have to carefully work the details of any alleged market failure of the Canadian mortgage market before making such claims. I find it alarming that we accept hand waving “adverse selection” as an excuse for a massive government program that erodes the price signal in credit allocation. Why even bother with asset markets if the government is so good at assuming risk?

What we do know is that the Modigliani-Miller Theorem doesn't appear to hold in banking. Canadian banks (American and European, too) have virtually no equity. Apparently capital structure matters very much in the financial industry. Why? It is precisely because of subsidization programs like CMHC that radically favour debt over equity. This is the origin of too big to fail - we make our financial system more susceptible to runs, more brittle, and with a lot more moral hazard. We build in the need for future bail outs with these type of programs.

[Nick Rowe]

Avon: "Yes I can. Think about it. The government allows homeowners to borrow at REDUCED rates - including the costs of the mortgage insurance - relative to what they would have received in a competitive market."

We do not know that. Maybe the rates, including the COMPULSORY mortgage insurance, are higher than they would be with a competitive market with insurance. It might be a TAX, not a subsidy. "You MUST buy this product from the government, at a price the government chooses" Is that a subsidy, or a tax? Depends on the price. Depends on the cost.

[Avon Barksdale]

Nick, you are about the only economist in the world who would suggest that programs like CMHC might be a tax on borrowing as opposed to a subsidy. The purpose of CMHC is to lower borrowing costs for people who would otherwise not be able to buy a home.

If you still think that CMHC might be a tax, try this thought experiment out: With $15,000 or less, I can walk in to any lender in this country and get a mortgage for a $300,000 home, CMHC insured, of course (if I'm clever, I can do it with nearly 0 down). Even with the extra cost from the mortgage insurance, I can get a rate of under 4%. Now, open a trading account with any of brokerage in Canada and hold only $15,000 in Government of Canada bonds. See if the brokerage will allow you to leverage an additional $285,000 for buying more Government of Canada bonds. GoC bonds are default-free as it gets. So why would the bank lend $285,000 to me at near prime brokerage rates to buy a house with 5% down or less, but the bank will never lend the equivalent to me to buy a default-free asset? CMHC is a subsidy.

[Nick Rowe]

Avon: do a back of the envelope calculation. IIRC, the average CMHC fee is around 5% of the value of the mortgage.

If more than 5% of the mortgages went bad, and the loss on each bad mortgage was 100%, the taxpayer would be worse off.

If more than 10% of the mortgages went bad, and the average loss was 50%, the taxpayer would be worse off.

If more than 20% went bad and the average loss was 25%, the taxpayer would be worse off.

Etc.

In the long run, for all CMHC mortgages, do those sort of numbers for expected losses sound plausible? I could see some sort of house price collapse and recession causing numbers like that *in one year*. But over the long run?

[Avon Barksdale]

Nick, derivatives do not work like pooled premium insurance. There is no 5% of a home's sales price for high ratio mortgages sitting in a pile somewhere. On a 25 year mortgage, the 5% insurance premium equates to about a half percent interest rate difference between the government insured mortgage and an uninsured mortgage. CMHC/CHT engineers a swap – institutional investors lend to the homeowner and the government hands the homeowner's mortgage payments to the institutional investor. That is all CMHC/CHT accomplishes – there is no money sitting around – it is a swap of cash flows. The interest rate differential that the institutional investor receives is not even 1% greater than holding an uninsured mortgage. So, a prospective homeowner who, in a free market, could never leverage 19 to 1 to buy a house - or any other asset for than matter - can do so with CMHC. Moreover, she can take on this much leverage at tiny spread relative to the Government of Canada's borrowing rates. The homeowner can accomplish this borrowing because the Government of Canada has promised to keep making the payments to the institutional investor even if the homeowner stops paying. We know for a fact that the tiny spread the sits between Canada Housing Trust issues and Government of Canada bonds does not reflect the default risk of the homeowner, because if it weren't for the government program no one at any interest rate would lend to her with 19 to 1 leverage. That's the point of CMHC – it provides a subsidy by transforming the low or non-existent credit rating of a prospective homeowner to that of the Government of Canada's.

[Avon Barksdale]

Nick, derivatives do not work like pooled premium insurance. There is no 5% of a home's sales price for high ratio mortgages sitting in a pile somewhere. On a 25 year mortgage, the 5% insurance premium equates to about a half percent interest rate difference between the government insured mortgage and an uninsured mortgage. CMHC/CHT engineers a swap – institutional investors lend to the homeowner and the government hands the homeowner's mortgage payments to the institutional investor. That is all CMHC/CHT accomplishes – there is no money sitting around – it is a swap of cash flows. The interest rate differential that the institutional investor receives is not even 1% greater than holding an uninsured mortgage. So, a prospective homeowner who, in a free market, could never leverage 19 to 1 to buy a house - or any other asset for than matter - can do so with CMHC. Moreover, she can take on this much leverage at tiny spread relative to the Government of Canada's borrowing rates. The homeowner can accomplish this borrowing because the Government of Canada has promised to keep making the payments to the institutional investor even if the homeowner stops paying. We know for a fact that the tiny spread the sits between Canada Housing Trust issues and Government of Canada bonds does not reflect the default risk of the homeowner, because if it weren't for the government program no one at any interest rate would lend to her with 19 to 1 leverage. That's the point of CMHC – it provides a subsidy by transforming the low or non-existent credit rating of a prospective homeowner to that of the Government of Canada's.

[Avon Barksdale]

Nick, derivatives do not work like pooled premium insurance - the naive calculation you suggest has nothing to do with how derivatives are priced.The CMHC 5% premium on a high ratio mortgage does not sit in a pile somewhere. On a 25 year mortgage, the 5% insurance premium equates to about a half percent interest rate difference between the government insured mortgage and an uninsured mortgage. The correct way to think about this problem is to examine credit risk implied by the interest rate spread. CMHC/CHT engineers a swap – institutional investors lend to the homeowner and the government hands the homeowner's mortgage payments to the institutional investor. That is all CMHC/CHT accomplishes – there is no money sitting around - it is a swap of cash flows. The interest rate differential that the institutional investor receives is not even 1% greater than holding an uninsured mortgage. So, a prospective homeowner who, in a free market, could never leverage 19 to 1 to buy a house - or any other asset for than matter - can do so with CMHC. Moreover, she can take on this much leverage at tiny spread relative to the Government of Canada's borrowing rates. The homeowner can accomplish this borrowing because the Government of Canada has promised to keep making the payments to the institutional investor even if the homeowner stops paying, thereby generating the contingent claim nature of the contract. We know for a fact that the tiny spread the sits between Canada Housing Trust issues and Government of Canada bonds does not reflect the default risk of the homeowner, because if it weren't for the government program no one at any interest rate would lend to her with 19 to 1 leverage. CMHC/CHT does not hold a replicating portfolio to hedge its obligations, it simple retains the default risk. That's the point of CMHC - it provides a subsidy by transforming the low or non-existent credit rating of a prospective homeowner to that of the Government of Canada's.

[Nick Rowe]

Avon: "The correct way to think about this problem is to examine credit risk implied by the interest rate spread."

No. That is the incorrect way to think about an insurance premium.

Suppose that 10% of all new CMHC mortgages will go bad, sometime over the life of the mortgage, and that the average loss will be 50% (in present value terms). If the taxpayer were risk-neutral, a 5% premium would leave the taxpayer indifferent. If we take that 5% premium, and put it in the CMHC reserve, and pay for losses out of that reserve, that reserve fund will neither rise nor fall, on average, over time.

This has got nothing to do with an interest rate spread. Interest rates have the units 1/time. We are not talking about percentage losses per year. We are talking about percentage losses over the life of the mortgage. And each mortgage has a finite life, and the risks to CMHC diminish over time, as the mortgage gets paid down.

If mortgages were perpetuities, then it might make sense to talk about risks per year. Because all new mortgages would default eventually, if they lasted forever.

[Avon Barksdale]

Wow. Honestly, I am shocked to see an economist reason the way you have, Nick. You continue to think that derivatives work like pooled premium insurance - there is no reserve sitting around. CMHC/CHT securitization is in effect a credit default swap in which the government remains the counterparty. Please re-read the logic behind dynamic replication and the origin of the Black-Scholes equation. Derivative prices are not premiums pooled together to pay out if things go bad - they are the initial capital required to set up a hedging portfolio and MBS contracts are notoriously difficult to hedge dynamically. Read the quant literature to understand how to price an interest rate swap, and how a tranche is set up. You will quickly learn that the comparative advantaged that is exchanged in an interest rate swap is a credit rating and that is why the swap rate is at the heart of the financial world. The MBS market is not like car insurance. Spreads to the swap rate tell you everything about the risk premium. CMHC/CHT interferes with this price signal by allowing people to borrow at rates they could not obtain in a competitive market - it is the entire reason the program exists. If my credit rating implies a borrowing rate of 10% in a competitive capital market (one with a competitive non-government backed MBS market), no amount of insurance can change that. Whatever rate reduction I get from the insurance, I will pay in premiums - it's just the logic of Modigliani-Miller. Someone always has to hold the risk, and since with CMHC/CHT the government holds most of the default risk, institutional investors are willing to lend to homeowners with little to no equity at very low rates.

[Nick Rowe]

Avon: I am not surprised, but I am saddened, to once again see finance people complicate something that is simple.

"CMHC/CHT interferes with this price signal by allowing people to borrow at rates they could not obtain in a competitive market - it is the entire reason the program exists."

"interferes with"?? Does the government "interfere with" the market for protection, by providing military, police, and firefighting services?

Look, I am certainly no raving lefty, but I do allow the possibility that in some cases the government might be able to provide a service more cheaply than the private sector, and that the taxes or insurance premiums we pay to the government might be less than the cost for getting similar service from the private sector.

"You continue to think that derivatives work like pooled premium insurance - there is no reserve sitting around."

So, what is that reserve sitting on the books of CMHC then? And might that reserve not have been bigger, if part of it had not been spent on other things?

BTW: sorry some of your previous comments got caught in our overactive spam filter. I fished them out.

[Avon Barksdale]

Nick, let's not confuse the issues of public goods (rivalry and excludability) with the issue as to whether the government is better at structuring mortgage derivatives than a fully private MBS market. Houses and credit are both excludable and rivalrous so there is no reason to think that the domestic mortgage market is akin to national defence. Asset markets provide the lifeblood of the economy by allowing society to trade and share risk. And, after all, mortgages and houses are just assets.

There are only so many dollars that can be borrowed and, thus like anything else, borrowing is a scarce resource. Programs like CMHC/CHT tip the equilibrium in favour of the homeowner because it is such a tremendous vote getter. Under no circumstance does the government want the credit market to discriminate between prospective homeowners like it does for cars or other consumer goods. This interference (and it is interference) removes the price signal in the economy as to the best place to lend dollars on a risk-return basis. The most distressing issue is that these programs represent yet another form of debt subsidy which means that our financial system ends up funding itself with runnable debt. Have you ever wondered why all the banks in Canada (and pretty much the Western world over) have virtually no equity? Why is banking so special that capital structure would matter so much? It is precisely because of explicit debt subsidies like CMHC/CHT and implicit ones like too big to fail. I am not interested in left vs right. I am interested in a financial system that is vibrant and innovative, and capable of handling the inevitable bankruptcies of market competition without leading to a financial crisis or a government bailout. Without a competitive financial industry, we move further along the road of crony capitalism.

[Nick Rowe]

Avon: OK. I accept your point that CMHC insurance is a subsidy for borrowing for houses *relative to borrowing to finance other assets*.

"Have you ever wondered why all the banks in Canada (and pretty much the Western world over) have virtually no equity?"

YES!

But remember Samaritan's Dilemma. The government cannot precommit not to bail out the banks. That precommitment is not credible, and banks know it.

All banks are a disaster waiting to happen. There has to be a better way. Maybe like this??

[Nick Rowe]

Though remember: banks do pay a fee for deposit insurance.

[Avon Barksdale]

Yes, I agree, we need much more bank equity. I find it a touch surprising that economists don't focus on the central lesson of the dot-com collapse vs the financial crisis: the dot-com collapse was all equity and thus it didn't freeze the payment system, but the run that we got on the shadow banking system in the US mortgage market threatened a full financial collapse. We need enough bank equity so that this doesn't happen again. Stockholders cannot trade at yesterday's prices, but with debt, we get a run. When the government monkeys with the capital structure to favour debt, we create serious harm. Financial regulation is a dog's breakfast: the government subsidizes debt, the financial industry responds to the incentive, and then the government tries to reduce the possible harm from the incentive by micro-managing the assets institutions hold with complicated regulation. Simple regulation based on more equity (and not the risk weighted nonsense we see now) would be much more effective. When Wall Street and Bay Street squeal that equity is more expensive than debt, what they mean to say is that they know where the government subsidies are. Deposit insurance and the like are just more forms of awful subsidies. The government would not need to pre-commit to anything if the regulation simply required more equity – and I mean a lot more.

I have enjoyed our blog conversation, I appreciate it.

[Avon Barksdale]

Yes, I agree, we need much more bank equity. I find it a touch surprising that economists don't focus on the central lesson of the dot-com collapse vs the financial crisis: the dot-com collapse was all equity and thus it didn't freeze the payment system, but the run that we got on the shadow banking system in the US mortgage market threatened a full financial collapse. We need enough bank equity so that this doesn't happen again. Stockholders cannot trade at yesterday's prices, but with debt, we get a run. When the government monkeys with the capital structure to favour debt, we create serious harm. Financial regulation is a dog's breakfast: the government subsidizes debt, the financial industry responds to the incentive, and then the government tries to reduce the possible harm from the incentive by micro-managing the assets institutions hold with complicated regulation. Simple regulation based on more equity (and not the risk weighted nonsense we see now) would be much more effective. When Wall Street and Bay Street squeal that equity is more expensive than debt, what they mean to say is that they know where the government subsidies are. Deposit insurance and the like are just more forms of awful subsidies. The government would not need to pre-commit to anything if the regulation simply required more equity - and I mean a lot more.

I have enjoyed our blog conversation, I appreciate it.

[Avon Barksdale]

I think your filter eliminated my last message. Not sure if you're aware - or if I've been banned :(

[Nick Rowe]

Avon: sorry. Our spam filter is playing up again.

"I find it a touch surprising that economists don't focus on the central lesson of the dot-com collapse vs the financial crisis: the dot-com collapse was all equity and thus it didn't freeze the payment system,..."

I focus on that. It's a very important lesson in my mind. (I think of the 1987 stock market crash, rather than dotcom, but it's the same.)

Economists talk about increasing banks' capital ratios, but it's peanuts.

And thanks for your comments!