« Why do economists assume tradeoffs? | Main | Should Greenpeace just give up? »

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451688169e2013484b8aac0970c

Listed below are links to weblogs that reference US fixed rate mortgages aren't fixed rate mortgages; they are weird, stupid, and dangerous:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Nick

The problem is that people move. If they have a mortgage with a 20 year duration and long term rates come down by 2% they will have to pay 40% of the notional amount of the mortgage to terminate. So if they are going to have 30 yr mortgages (which I'm not saying they do), then they probably need the embedded option.

I largely agree, but keep in mind that some businesses and producers might choose to take both futures and options positions on the same underlying. The futures to lock-in the price, and the option to provide a way to get the upside (or downside) potential if the market turns that way. For businesses and producers, I think it has less to do with risk aversion than with cash flows and debt/leverage.

I meant: "which I'm not saying they *need to*". I know that they "do" have them.

K. When you move, you take the mortgage with you to your new house. Or it stays with the house, if the new owner is a good credit risk. Or you can pay it off, if you are unexpectedly flush with cash. You have to pay a penalty for early repayment, of course, equal to the difference between your old fixed rate and current market rates. But that "penalty" is not really a penalty, or a risk. All it means is that you insured yourself against interest rates rising over the term of the mortgage, and they didn't rise, they fell, so your insurance wasn't needed ex post. Like when I take out fire insurance on my house, and then my house doesn't burn down. I don't say "Dammit, my house didn't burn down, so I lost that money I paid when I bet my house would burn down, so can I have my insurance premium back please?"

You don't need to write an option on interest rates to deal with people moving. In Canada, lenders offer portable and/or assumable mortgages - meaning you can transfer the mortgage from one property to another or change the borrower (no doubt subject to the lender agreeing).

The standard US mortgage is such a terrible deal for lenders, one wonders why they ever started doing it in the first place.

I'm sorry Nick but you were doing just fine in your description of the pre-payment option until you got to the part about if being weird, stupid and dangerous.

You've given no argument for why a portfolio of fixed rate bond and option weird, stupid or dangerous. Yes it's a different portfolio then the one without the option but you have given no explanation as to why it's worse.

> Lenders aren't always risk-neutral

This statement is a damaging distraction from your argument: nobody is every risk-neutral in that sense. Risk neutrality refers to probability measures, not people, and simply means that when calculating an expectation (average outcome) with respect to the measure, the market price will be recovered. When this language applies, the market price is determined by the price of the replicating hedge portfolio. In the case of MBS, that portfolio includes an option which has a positive price, but the market is not complete with respect to such options. They require a pre-payment model that amounts to price-of-risk fudging. And anyway, the buy-side to which you are referring is by definition not trading based on risk-neutral pricing.

> As I understand it, one of the main reasons behind the securitisation of mortgages in the US was because of the interest rate risk and liquidity risk I have talked about above.

You left out the most important part, which is the GSE's vital support for the securitization market. Along with US tax policy, this makes the world's largest market, in the self-proclaimed bastion of free-markets and capitalism, completely dependent on commie-pinko-socialist government meddling. Ha ha!

> the other weird and stupid feature of US mortgages that they are non-recourse

That is a state-by-state, not US feature.

> The problem is that people move.

That is not a difficult problem to solve theoretically, as there is no financial reason why mortgage collateral should not be substitutable (i.e. your mortgage "portable".) After all, it works for CDO pools. The obstacle in the US would be incompatible rules when moving between states; this could be overcome by the Federal government if it ruled that the state in which the mortgage originates is what counts. They already do this for CC issuers.

> Americans don't like foreigners interfering in their internal affairs

I agree that the Americans are especially wacky but when I had a Canadian mortgage there were lots of bundled options: I could switch from semi-monthly to bi-weekly payments, double up payments as I saw fit, and pay down 10% notional once per year. I didn't get paid for giving up these options so I exercised them - they happened to suit me because I was self-employed at the time and had an income that was relatively high but also relatively uncertain.

Patrick: good question. My guess is some legal restriction on closed mortgages, so the market isn't allowed to provide it. Or political pressure on Fanny and Freddy. Like the US legal system doesn't (in most states, I think) enforce recourse mortgages.

Adam: I thought I did give an argument. Not a formal math proof, but an argument nevertheless. I can understand a market in insurance. I can understand a market in lottery tickets. But I can't understand a market in which lottery tickets are bundled with insurance. So it's impossible to buy insurance unless you buy a lottery ticket at the same time. Presumably because of some legal restriction, so you cannot alienate your right to pay off a mortgage. And until someone comes up with a good explanation that explains not just why the two products are sometimes bundled, but why they are *always* bundled, I'm going to say it looks stupid, and it probably is stupid.

Phil: I don't understand your first paragraph, on risk neutrality. It went over my head. Good corrections and points in the rest. I don't understand why Canadian closed mortgages have those small "openings". But they are small. Maybe just a gimmick?

Apologies, Nick, as I obviously made a hash of my explanation. A danger whenever public commentary overlaps my professional area. On the material points of your post, I agree with you.

"I can't understand a market in which lottery tickets are bundled with insurance" is not an argument for it being weird, stupid or dangerous.

"Risk neutrality refers to probability measures, not people"

No Phil, risk neutrality can be used, correctly, to describe either. Most commonly it refers to people.

"simply means that when calculating an expectation (average outcome) with respect to the measure, the market price will be recovered."

and this is just wrong.

It is not really correct to say that there are no closed loans in the US. One of the primary features of subprime lending was the inclusion of prepayment penalties, which are indeed absent from 'prime,' ie Freddie and Fannie approved, mortgages. This was used by lenders to make the loans more explicitly bets on the collatoral rather than the credit worthiness of the borrower.

Interestingly there are almost no assumable mortgages in the US, which went out of fashion in the Eighties when Volcker was throttling inflation and the rest of the economy with 22% interest rates.

From your brief description it seems that mortgages in Canada are much less focused on the collatoral than the US structure, since mortgages are never portable here.

Here by the way is Mike Konzcal on the way pre-payment penalties fed into the sub-prime market in the US. Plus, I still don't get the danger part, it was, after all, the sub-prime prepayment mortgages that blew up first.

http://rortybomb.wordpress.com/2009/05/12/ban-prepayment-penalties-2-banks-gambling-on-real-estate/

Lenders don't operate in a monopoly, so they have to offer incentives to borrowers so that they don't take their business elsewhere.

Securitization is a separate issue. It becomes a problem if not properly regulated and rated, but otherwise is not that complicated or stupid.

Aren't you suffering from a social construction of interest rate risk? A closed, fixed rate 6% mortgage exposes you to the risk of interest rates declining. If nominal interest rates unexpectedly decline because of reduced inflation expectations, the real cost of your debt increases. Take someone who took a 15% closed FRM in 1982 expecting large nominal increases in income and home prices that didn't come to fruition, and finds they cannot meet their projected budget. Isn't this a form of "interest rate risk?" Why can't you call a closed, fixed mortgage an unusual and dangerous bundling of purchased insurance on overall interest rate volatility with written insurance on interest rate declines? Why should consumers take the debt-deflation risk (or debt-disinflation risk) when banks and other financial intermediaries might be better equipped to handle almost all kinds of interest rate risks? Because wages will be sticky downward?

Maybe you think that the risk of interest rates declining is unimportant, because if someone can make their first payment, they can make all future payments even if inflation expectations decline (i.e. because mortgage payment terms are inherently front-loaded in a world with inflation). Well, there would still be deflation risk. But also, a household might budget a big expense for year five (a big vacation) when they expect lower real mortgage payments, which they then would have to forfeit thanks to interest rate risk. Why is it more dangerous to offer a loan with full interest rate volatility insurance than a loan with partial interest rate volatility insurance?

> "I can't understand a market in which lottery tickets are bundled with insurance"

term variable annuities anyone?

OGT: Very interesting post by Mike Konzcal you link to there. The comment by "q" on that post is also interesting, and reflects more my way of thinking. US prepayment penalties seem very different from Canadian and UK, I think. A fixed amount, rather than the difference between the old and new interest rates. The latter means nobody has any incentive to re-finance when interest rates fall, and will only pay off a mortgage early if there's some significant unexpected change in their personal circumstances. And in the US it was the mortgages with prepayment penalties that had the higher rates, presumably because they were offered to borrowers with higher credit risk. So if their credit risk later improved, or if banks lowered standards, they sometimes paid the penalty to refinance at a lower rate, on a prime, open mortgage. Here, of course, closed mortgages have the lower interest rates.

I didn't know that US mortgages were never portable!! That maybe explains why people insist on open mortgages, in case they ever want to move house. But why aren't they portable? Is it because they are (in many states) non-recourse? That might relate back to your comment that Canadian mortgages seem less focused on collateral. In the event of default, you get the house, and can still go after the borrower (some exceptions in Alberta).

What's a "assumable" mortgage? Is that where I sell my house, and the new buyer assumes the mortgage? I wonder why they went out of fashion. But if they aren't assumable, and also not portable, that would definitely explain why Americans insist on open mortgages. Otherwise what do you do with the mortgage if you sell your house? You can't take it with you, and you can't sell it with the house!

The whole thing is more intertwined than I initially thought. Good comment.

asp: "Lenders don't operate in a monopoly, so they have to offer incentives to borrowers so that they don't take their business elsewhere."
Agreed. But why don't they compete on lower rates, rather than on providing open mortgages? Insurance companies don't compete by offing free lottery tickets with every insurance policy.

dlr: good question. But there's an answer.

First, it puzzles me that mortgages, and all loans, aren't indexed to inflation. But let's leave that aside.

The "risk" that matters, and hence the correct way to define "risk", is variance in your consumption stream, not your wealth at a point in time. For example, suppose I am about to retire and know I will live for exactly 30 years, and want to have constant known consumption for each of those 30 years. I buy a 30 year annuity. If interest rates rise or fall, the Present Value of my annuity will fall or rise. So an annuity is very risky in the Present Value sense, whereas leaving my money on deposit at a variable interest rate is perfectly safe. But I don't care about the present value. I care about having a risk-free rate of consumption for 30 years. And the 30 year annuity gives me that exactly, with perfect safety. Leaving my retirement funds on deposit, at a variable interest rate, exposes me to risk on my stream of consumption.

It's exactly the same, only with minus signs, for someone who wants to use a (they hope) constant stream of wages over the next 30 years to pay down a mortgage and consume the remainder.

adjacent: what's a "term variable annuity" (unless it's a life annuity)? And why would anyone buy and sell one (unless it's a life annuity)?

Don't forget that mortgage interest deductibility factors into higher interest rates and/or prices as well.

The US has a very distorted housing market.

Nick, I don't buy your response to dlr. Mortgages probably should be indexed to inflation, but in practice, they are not. The important risk for the borrower is not present value risk but deflation (or disinflation) risk. I'm presuming that most borrowers are in their working years, and their nominal income depends on the price level. Granted, a call option on interest rates doesn't fully protect the borrower (because real rates can rise, and because the effects of deflationary forces on an individual borrower are not a linear and predictable function of the expected inflation component), but it sure helps. As someone whose income has been reduced, but not enough to prevent my qualifying for a new mortgage, I can personally attest to the risk-reducing effect of the prepayment option on borrowers. I think that if I'd had a choice, I would have chosen to purchase the prepayment option even if it weren't bundled with my mortgage, and ex post that clearly would have been the right decision (even leaving aside the necessity for a prepayment option due to the possibility of relocation, given the non-assumability and non-portability of the mortgage).

I believe it comes down to an irrational fear of rising rates and an emotional need for certainty.

30-year fixed-coupon mortgages with payoff privileges used to be the norm both in Canada and the United States. I've spoken to friends (Canadians) who fondly remember this time. They think that a 25-year fixed mortgage is the height of certainty and security. They dread that uncontrollable or rising payments could force them out of their house.

The Canadian mortgage market took its present shape in the 1970's. The Canada Interest Act was amended to say that six months interest was the maximum penalty chargeable for breaking a mortgage, and rates started to rise into double digits. People started to refinance their mortgages at the earliest opportunity to try to get the rate down. Banks eventually started to offer the 1-5 year terms we see now in order to accommodate this. The fixed 25-year mortgage became a thing of the past. These rate terms also helped the banks because they could easily match their mortgage assets to their deposit liabilities. 5-year mortgages balanced off 5-year GIC's.

Thanks Nick. On assumable, you're right in guessing the meaning. When rates went sky it was common for sellers with lower rates to allow buyers to assume the mortgage in exchange for a higher selling price. Now that was a nice option! So lenders, who didn't want to keep getting paid 12% when the going rate was 18%, from someone they didn't underwite no less, started adding transfer clauses. Pretty much every mortgage I've ever seen in the US the balance is payable on transfer of the deed unless the transfer is approved by the lender.

The US brand of federalism certainly may play a part in non-portability, not only do states have different restrictions on recourse, they also have different foreclusore proceedures, and other regulations. Texas, for example, bans prepayment penalties and cash out refi's. In addition, remember national banks are a relatively recent development here, so one assumes a regional Boston bank wouldn't want to have an Arizona house over 2,000 miles away from their nearest branch as collatoral.

A few points on this:
-Nick, you asked why Canadian closed mortgages have these small and limited 'open' options (such as you can pay off a certain amount once a year, double-up payments, etc). The key is that the cost of the open prepayment (US mortgage) is very high repayment risk over the life of the 30-year fixed. For Canadian closed mortgages - typically fixed for no more than five years - the small 'options' make the additional risk of the closed mortgage small, both to the borrower and the lender. Basically, the term is short enough, and the options limited enough, that banks can price them reasonably finely - but not perfectly. And the borrower can accelerate their mortgage more or less at will - but not all at once, and not many actually do. (And the unwinding cost is smaller, due to shorter term, and really only arises when someone sells early. Plus portable.
-Why do these mortgages exist in the U.S.? Basically, only Fannie and Freddie can write 30-year open due to the risk. And they require them to be this way. (I believe there is also a legal requirement but can't 100% state this).
-When I learned this stuff, 'points' (up-front fees, like a front-end fee) were one way that US banks priced this risk in. Basically, you bought down your interest rate by paying higher points up-front; the pay-off structure of the option had different characteristics basically depending on whether you were likely to refinance. (As a sunk cost, it actually didn't effect borrower behaviour post facto - or at least much).

I work on emerging markets lending, esp mortgages. The general argument/line of discussion for years was that the U.S. could afford the expensive luxury of long-term open fixed mortgages due to deep capital markets - the explosion of the industry and Fannie/Freddie make this argument questionable.

But long-term fixed open are particularly weird, stupid and dangerous in most other markets with shorter yield curves/shallower capital markets. (There's an interesting corollary to this story, which is the proliferation of longer-term fixed, open mortgages IN FOREIGN CURRENCIES in much of Eastern Europe - these are not even weird, just stupid and dangerous).

The curve for e.g. Russian rubles goes out no more than ~10 years (and at that term, is available basically only to the government). How can you price a fixed-rate mortgage longer than that? (Hint: you can only do so by being willfully stupid, generally with someone else's money - typically the government's). And Russia has deeper capital markets in their own currency than most.

The other thing, Fannie in particular for years had a very active international program, basically promoting mortgages. Parts of it were good - Fannie knows, or used to at any rate, mortgage underwriting extremely well. Parts of it were absurd: in particular, the assumption that all mortgage markets would all develop to perfection over time, and ultimately resemble the U.S. market. It amounted to a religion: first, originate fixed rate mortgages; second, securitise them. I believe Fannie stopped doing active proselytisation after they got hit with their first big restatement/accounting scandal (2004?).

By the way: time to renew your analysis/data on CMHC. I still think CMHC is doing better than most think.

Now, I have a question that I would most like a "real" economist's answer to: the nature of inflation risk for 'individuals' as a whole.

My argument for floating rate mortgages includes the following:
-Floating rate loans are basically (over medium term) inflation plus some extra.
-For the class of individuals in an economy, inflation is basically the growth in their income.
-Hence, overall, for the class of individuals, it is collectively nonsense to hedge inflation risk - their income IS inflation for most intents and purposes (over time).
-Hence, for both lenders and borrowers (as classes), the risks they need to consider and insure against are temporary ups and downs in the floating rate (that in the short term may be out of line with inflation, which we could call basis risk for a major difference between inflation and the short-term lending rate), deflation risk (let's leave that to the central bank). This is pretty easily hedged against with caps on increases, rolling fixed rates, any number of modest add-ons. Five year fixing is probably enough to avoid the vast majority of 'interest shock' risk.
-Individuals still need to deal with the risk of their own wages growing substantially slower than inflation - which I would just call credit risk.

Thoughts?

The reason for the embedded call option is to allow mobility. If borrowers could prepay upon the transfer of a deed (but not because of falling interest rates), the mortgage would not inhibit mobility but would still be more attractive to investors. As mobility is more predictable across a pool than interest rates, this embedded risk could probably be priced reasonably well.

But from a credit perspective, the prepayable fixed has been safer than other mortgage types. According to the most recent MBAA data, "[t]he seasonally adjusted delinquency rate increased for all loan types with the exception of FHA loans. On a seasonally adjusted basis, the delinquency rate stood at 6.17 percent for prime fixed loans, 13.52 percent for prime ARM loans, 25.69 percent for subprime fixed loans, 29.09 percent for subprime ARM loans, 13.15 percent for FHA loans, and 7.96 percent for VA loans."

"And you will of course exercise that option at any time when the market interest rate for the remaining term falls below the rate you are currently paying."

It's not that bad because there are points and/or fees when you refinance (even if there's no explicit repayment fee, which sometimes there is). So, the gain from the lower interest rate will usually have to be fairly substantial to make it worth it to pay new points and/or fees plus the cost of your time and trouble. As a result, people aren't refinancing their mortgages every month, or every day, or every hour, so the price of these mortgages does not get extremely high relative to floating.

"The reason you didn't chose a variable rate mortgage is presumably because you wanted to insure against interest rate risk. So why buy a one-way bet on interest rate risk at the same time?? It's really stupid."

But it's not that easy to avoid. You might say why not unbundle. Just sell the insurance without the option. But you take away the option and you have a 30 year mortgage that can not be refinanced or paid off even if you sell the house. So what if you get a promotion and transfer after four years and have to sell the house and move? What if you lose your job and have to move for a new one? What if you get a divorce? What if you have children and now need a bigger house? What if your income permanently drops and you now need to buy a less expensive house?

The problem is it's too risky and/or problematic and/or complicated to ask many people to lock in to a 30 year mortgage that they can never payoff until 30 years, no matter how soon they might sell the house and move. True, if you have to move due to a transfer or promotion, you can sell your $200,000 home, put the funds in a government money market account, move to the new state, and then use the $200,000 to buy a new house for $240,000 plus perhaps a new small mortgage. Or, if the new house is $170,000, then they could invest the extra $30,000. There are ways to funge all this money, but now you are asking people to have a lot more expertise in finance and to spend a lot more time.

There's a real cost in time, complication, unpleasantness, and stress to many people in constantly asking them to learn all these things and deal with all this complication and risk. This stuff just piles up as you move more and more laissez faire in a world that's not 1810 anymore, and makes people spend more and more time learning finance and other areas of expertise and analyzing risks, and less and less time producing wealth or enjoying time with their families. These costs can get to be enormous even though they are assumed away and ignored in freshwater models.

So, you're not going to find any really smooth solution here.

And, a crucial concept to keep in mind is that it can be extremely efficient to shift risks from individuals to large groups that can greatly diversify away a lot of the risk, and that can pool risk in a very large entity to make it far less harmful per person. So, often it makes great sense to have an individual shift risk to a large organization, and in return pay a fee, or premium. Because given that the large organization is much less at risk from the thing in question than the individual, they can be more than compensated with a premium that is still a great deal for the individual.

With 30 year mortgages risk is shifted from individuals (for which interest rate risk can be really dangerous) to big companies (for which interest rate risk is a lot less dangerous) and to the government (for which interest rate risk is even far less dangerous – the government sometimes makes a great insurer because of its massive size, its massive life, its lack of an incentive to mislead for profit where asymmetric information is large, potentially gigantic economies of scale without having a for-profit monopoly, and other reasons).

Nonetheless, I think 30 year mortgages can be improved, perhaps with a larger and ubiquitous prepayment penalty to diminish socially expensive mortgage hopping which can have high transactions, time, and trouble costs, but do little, nothing, or negative to add to social utility. The prepayment penalty is less risky than a variable rate mortgage because it's capped at a fixed amount, while there's no limit to how high interest rates can go.

Richard:

Canada and the United States have almost identical home ownership rates, yet here it's the individual who bears the interest rate risk/reward. Same in the UK. In both countries a 25-year mortgage with a normal amortization curve consists of 5-25 individual mortgage terms with its own interest rate, set at the market's going rate.

You actually haven't proved anything about US 30-year fixed rate mortgages because other countries have achieved similar results without making US choices.

The "risk" that matters, and hence the correct way to define "risk", is variance in your consumption stream, not your wealth at a point in time. For example, suppose I am about to retire and know I will live for exactly 30 years, and want to have constant known consumption for each of those 30 years. I buy a 30 year annuity. If interest rates rise or fall, the Present Value of my annuity will fall or rise. So an annuity is very risky in the Present Value sense, whereas leaving my money on deposit at a variable interest rate is perfectly safe. But I don't care about the present value. I care about having a risk-free rate of consumption for 30 years. And the 30 year annuity gives me that exactly, with perfect safety. Leaving my retirement funds on deposit, at a variable interest rate, exposes me to risk on my stream of consumption.

If I didn't know better, I'd say you were falling for the money illusion. We care about variance in your real consumption stream, right? You don't want to take out a fixed annuity, see inflation rise unexpectedly, and be content with the smoothness of your nominal consumption (fixed, flat, annuity payments) while your real consumption sinks like an anchor. If you're talking about an inflation-indexed annuity (which I don't think you are), on the other hand, it is the wrong analog to a closed fixed rate mortgage (which is the whole point).

A consumer with a hypothetical budget for smooth, real consumption who experiences unexpected disinflation will suffer a negative nominal surprise in incomes and asset values. If they have taken on a closed FRM (presumably without owning off-setting fixed income instruments) their nominal payment obligations will remain unchanged and thus their expected real income will decline, reducing expected consumption in future periods relative to the smooth budget. Isn't this a consumption path risk?

@Richard Green: you are stating that the reason for the embedded call option is to allow for mobility. If this is the risk that is being mitigated, why not fix the rate for a shorter period of time? What's bizarre is the mismatch between the openness and the 30-year fixed rate.

Two points on the relative default rates between fixed/ARM:
-Some increased default is to be expected - the question is the trade-off between costs (including externalities).
-Most importantly: I have looked at some of these comparative figures, and for the most part, the data are NOT comparable. The securitizers may have SAID 'prime', but this does not mean they can be compared well - see the case brought by the SF Home Loan Bank about fraudulent reps and warranties. There is, ultimately, no common standard for prime.
-Keep in mind: the ONLY party securitizing fixed rate in any quantity was Fannie and Freddie. Almost literally the only game in town - anyone originating them was originating them for F&F.

If one really wanted to compare defaults on ARMs vs fixed, probably the only way to start would be ARMs from Freddie/Fannie (relatively few) to their fixed rate, and to be very careful about comparing like to like.

For a quick test to determine why there is something suspect about this data: can you think of any reason why, in a period when rates FELL, that the ARMs should have performed worse? All things being equal, rates on the ARMs should have fallen below their original payment levels, and hence defaults fallen. If this did not happen, there must have been some underlying other issue that was causing them to default - indicating that this comparison is probably not valid.

Wow! Lots of very good comments coming in. I'm not sure I can do them justice.

Christopher: mortgage interest deductibility should cause house prices to be higher, and cause people to pay down their mortgages more slowly. It's a big distortion. Not sure it would affect mortgage rates that much (very elastic supply?).

Andy: I think you have a valid point there. It works as long as falling interest rates are correlated with disinflation. It wouldn't have worked in (say) 1982. But outright indexation would presumably be better. And even floating rate mortgages would presumably be better, since then the bet works symmetrically, unlike open mortgages, where it only works in one direction.

Determinant: that's a very important bit of information. I hadn't known that. So Canadian "closed" mortgages aren't really as closed as they look, if there's a maximum 6 month penalty for prepayment. And that killed off the 25 year closed mortgage in Canada, understandably.

OGT: I think I can understand why lenders wouldn't want to make open mortgages assumable. It's a one-way bet for the seller of the house. But if it were a closed mortgage, I can't see why they would care (as long as the new buyer's credit was OK), and the borrower would like the feature, and be prepared to pay for it.

GA: I don't have anything to add to your very useful first comment. On your second comment, I tend to agree. Variable rate mortgages ought to give better indexation to inflation, and for that reason ought to be seen as less risky than fixed rate mortgages, both for lenders and borrowers. I think there are two arguments going the other way:

1. If inflation rises, and nominal interest rates rise by the same amount, the borrower is no worse off in real terms. But a variable rate mortgage becomes more "front-end-loaded". The effective average term to maturity falls. Take an interest-only mortgage for simplicity. With full indexation, you would only pay the real interest rate each year. But if inflation rises from 0% to 10%, and interest rates rise from 5% to 15%, your real payment has gone up in the first year, even though the remaining balance of the loan has fallen by 10% in real terms.

2. If inflation targeting is credible, and the target remains the same, all interest rate fluctuations should be real, not nominal, at least ex ante, in expected terms. And that seems to be close to the world we live in nowadays, at least in Canada. So there's no *expected* inflation for variable rate mortgages to insure against.

Richard Green: Welcome! Sorry to disagree with you so strongly!

But you can have mobility with closed mortgages as well. Mortgages can be portable (you transfer it to the new house you buy), assumable (the new buyer of your house assumes your mortgage, or you can even pay it off if you pay the interest rate differential.

But why has the "prepayable fixed" (i.e. "open") mortgage been a better credit risk? There must be some reason why borrowers with worse credit ratings are pushed into the mortgages with prepayment penalties, and yet actually pay higher interest rates than those who borrow in the "prime" market for open mortgages. I don't think you are comparing like with like.

Richard Serlin: As I said to the other Richard, there are 3 options for someone with a closed mortgage who wants to move house. Maybe it's complicated. And both open and closed mortgages exist in Canada, so they "meet the test of the market", but open mortgages are only readily available for one year terms (presumably often bought by people planning to sell), and don't meet the "test of the market" for longer terms. Canadians seem to manage to own and move houses, even with mostly closed mortgages.

Determinant: Yep. Which Richard are you addressing? Or maybe both!

dlr: "If I didn't know better, I'd say you were falling for the money illusion."

Yep! I was implicitly assuming zero inflation. I should have made that explicit. Point taken. See my comment above, in reply to GA, on variable rates as an inflation indexation device.

Before the 30s, mortgages were generally 5 year ballon loans. Imagine trying to find refinancing under deflation with plunging asset values, in no small part due to this. The 30 year amortization was introduced to stabilize this although deflation would still present a risk. A mortgage is really a confluence of a lender, a borrower, and a property. The size of the US mitigates against portability since lenders are often geographically limited. Inflation in the 70s ended assumability. Yes, 30 year fixed aren't really, but that is a good thing. It would be trying to eliminate risks that cannot be eliminated. Not only moving, but bankruptcy, death, and destruction. Not only interest rates, but inflation (deflation not so much). Rather than encouraging lending, borrows would view it as an anathema. How far would rates fall before walking away would be preferred. Creation of large rocks in the path of the economy would add rigidity. Most first time buyers are stretched to meet the payment which means they need fixed payments at least for a time. That option is valuable because interest rates do usually fall in recessions and is the primary way the Fed stimulates. Think of it like a first approximation to a five year fixed adjustable but having become accustomed to fixed most see no advantage to changing, not to mention the variability of terms on adjustables make them difficult to compare. There are, or were, 1, 2, 5, 7, 10 years, sometimes changing in 5 or 10 and sometimes not at all after that, using multiple indexes (some using the libor, can you imagine that after the crisis), caps in change and caps overall, with various teaser rates, and a notable propensity for calculation errors by both borrowers and lenders. All this makes adjustables more expensive than vanilla 30 years, so that option is free. Much of this is for historical reasons, but history is hard to change.

Non-recourse entered in the 30s as well to avoid interstate flight as a result. It takes two to make a bad loan, but the lender, having the money, deserves a higher level of accountability, especially at times like these that are the result of lender irresponsibility.

And most corporate debt is callable as well. Only treasuries are not.

Nick:

Lord mentions some key points when you talk about American and Canadian mortgages. An American looks at a Canadian mortgage and sees a "Balloon Mortgage", one where the bulk or whole of the principal is due at the end of the loan, usually 5 years or so. He shudders at the insecurity of renewing a mortgage.

A Canadian looks at an American mortgage and sees interest rates that are too high, too rigid and lacking in options on interest rate planning. Up here you get a menu of interest rate terms (which is NOT the same as amortization*) when you take out a mortgage. You can go really low on the yield curve and take a variable rate, or go higher up with a 5 year rate. You can switch back and forth. Your choice. Studies have consistently shown that 99% of the time, the shorter your term, the less you pay in interest. For a 25 year amortization, 50 6-month terms has always been cheaper by tens of thousands of dollars than 5 5-year terms. Basic economics because you are paying a rate that is lower when the yield curve is normal, which has been 99% of the time since WWII.

*In Canada the amortization is the period over which the loan will be retired, usually 25 years. The term is the period for which the interest rate is fixed, anywhere from 6 months to 5 years. You can pay off as much as you want when the term expires without paying a penalty.

In theory a variable rate mortgage works for deflation protection, but I have an intuitive feeling (which I think many will share) that it's actually a lot riskier than a prepayable fixed rate mortgage. For one thing, as you point out, the interest rate flexibility wouldn't have protected you from disinflation in 1982, and if you got the mortgage in the late 1970's, you would be totally screwed by a variable rate mortgage resetting upward just as the disinflation was about to start. Also, there is much to be said for having the security of constant payments in the case where the inflation rate remains roughly constant. It would be very tough to actually model all the various risks, but my feeling is that it's still much safer to pay for the option than to take a variable rate (and still worth paying for the option if it's not bundled). Of course, an inflation adjustment might be even better.

And I second the point Lord made about corporate debt. It's normal for debt to be callable. If you think callable mortgages are a puzzle, then you really have a much bigger puzzle to deal with.

Nick: to be clear, I disagree with a key part of your argument. It's the 30-year fixed rate that is weird and dangerous; making it open is stupid, but flows from it being weird and dangerous. You're focussing too much, I think, on the closed/open issue.

To state the big difference baldly: some floating element is normal and almost the worldwide standard; markets with very long fixed rates are the exception (US and Denmark usually cited, and Danish ones are weird and closed - but to be fair they seem to work). Those markets using floating rate mortgages do not have noticeably worse performance in virtually any element of home ownership/mortgage finance, or at least ones traced to this issue.

As for the closed/open, I think there's a bit of terminology that may be useful to readers. (probably dead obvious to others)

The red herring of 'what if you need to move and prepay?' keeps coming up, but a comparison to commercial finance might help. In a standard commercial (business) loan, the rate may be fixed or floating. Floating rate loans can generally be prepaid early; there may be a penalty for prepayment, but generally not onerous. This prepayment fee is basically to reimburse the bank for some of the up-front (fixed) costs of making the loan and a bit of foregone income, but it's mostly unrelated to the prevailing interest rates. (Borrower might be able to renegotiate the loan spread by threatening to prepay, the prepayment fee might decline in time, and other variations possible).

In a fixed rate loan, there will usually be 'unwinding costs' _in addition to_ the prepayment penalty. This departs from the assumption that the bank is fixing its costs on the liability side by getting matching funding (or equivalent using derivatives). If you prepay, the bank has to carry that fixed-rate liability or 'unwind' the derivatives related (or some equivalent). The unwinding cost is _directly_ related to the prevailing interest rates - it's the cost of reimbursing the bank for paying that higher, fixed interest rate on its borrowings over the life of the original contract when it can only make a new loan today at lower rates. (If rates have gone up, unwinding costs will be zero).

At any rate, this is why the shorter-term floating or modest-term fixed rate (hybrid) allows for small or modest prepayments. If it's floating, there's no unwinding cost. So if you need to move, you can - portability/assumability is a minor feature in this.

Someone will object that in complete markets there are a zillion options, puts, swaps, swaptions that can help manage this - which is true, but avoids the fact that markets are not complete and transaction cost or counterparty-risk free, and it's not easy or costless to manage this stuff carefully - complexity and risk goes up with the term of the contracts.

Likewise, US commenters often raise the hoary spectre of the depression, when borrowers with 'balloon' payments (e.g. five year mortgages with a bullet/balloon at end pending) that could not roll over because banks refused were foreclosed on en masse. (In a nice illustration of the fallacy of composition/herd mentality, the banks went bust too).

In other markets, they just required the lender to rollover at the new market rate. Banks end up taking liquidity risk - but, if you trust the Bagehot principle, with some certainty that they refinance their book. Since the mortgage will reset at market rates, the likelihood the mortgage will fall significantly below outstanding principle (par) should be small.

“I wouldn't recommend sex, drugs or insanity for everyone, but they've always worked for me.”

I would add 30 Yr Fixed Rate Mortgages as well to my list.

I believe the other reason for non-recourse was to clear court calendars. The bankruptcies were too widespread to be handled. Recourse after all, is of little practical use; normal times don't need it, abnormal times make it irrelevant.

Nick,

Ok so it's possible the complication is not that bad if you do things like making the mortgage portable.

I really should have used more "might"s and "may"s and "I think"s here because this is something I really haven't thought about. As an adjunct professor of personal finance for many years in the US, I've thought about what is the best mortgage type for an individual to take given his circumstances and the options that are given, but not which options are best for society.

But in Canada with these closed long term mortgages, what if an individual sells his home, moves somewhere else for work, and then just rents because he doesn't think he will live there for long. The mortgage company no longer has its collateral. Do they force you to repay the mortgage and the difference in interest rates? Sounds risky if you're mobile. Maybe these mortgages are so popular in Canada because the country is a lot less mobile than the US?

Nick: A term variable annuity is a term life annuity. I was thinking of something else -- the "variable" doesn't matter here -- but note that life annuities have embedded prepayment options that vary with the prevailing interest rate. The buyer locks in an effective interest rate for several decades at origination but makes payments spread out over time. If interest rates go up during the term it might make sense for the holder to exercise and then "refinance" (ie buy a new annuity). (I think I have that the right way.) So something that looks simple and is popular can be very complicated.

You also referred to the comments by the poster "q" on Konszcal's post. I'm the same person BTW. I would post here under "q" but most sites have policies about changing their name and I posted here once as "adjacent" before.

I think most of my good ideas are in the comment on the other site so I would copy it here if that's appropriate.

Easy Prof. Serlin.

A Canadian can easily move house one of four ways:

1) Move the mortgage along with the individual. Canadian banks are all national in scope and have branches everywhere. They will loan you money in any part of Canada.

2) Since moving normally requires a large amount of planning, a person will normally have converted their mortgage term into an "Open" term. Remember, amortization is not the same thing as the term in Canada. I hate to be a broken record but if we're not clear than we're going to talk at cross-purposes. "Term" is the period for which the rate is non-negotiable. An open mortgage can be paid off at any time without penalty.

A variation on this theme is to hold the mortgage until its term expires and it comes up for renewal, then pay off the balance. There is no penalty for this. Since terms are for 6 months to five years they are on average less than five years. If you listen to any financial planner in the country, you'll probably have a six-month fixed-rate term. Easy to bail out of. A person who knows they are going to move will likely arrange a short term for the convenience. Their banker would also likely persuade them to do so, for everybody's sake. See my next point.

3) A "closed" mortgage isn't truly closed. A "closed" mortgage means that penalties will be levied for paying off the mortgage and breaking it. According to the Canada Interest Act, this may be no more than three months interest or an interest rate differential. This:

http://www.vernonrealestate.com/blog-refinancing-penalty.html

contains a succinct example of what happens.

Banking law and interest is a federal matter in Canada. The Government of Canada has always taken an interest (no pun intended) in making sure that residential and farm borrowers (who vote) can bail out of unfavourable mortgages. Because the statutory penalty cap is very small in comparison to 30-year mortgages, they have become extinct. 6-month to 5-year terms mean the penalty is still a deterrent. They also match (not coincidentally) the terms offered on bank GIC's.

4) Many corporations who move their people around will have arrangements to buy their employees homes and sell it themselves. Banks do this with their managers. Managers can and do move across the country. The bank will buy the managers house. I have family friends who have done this.

also, just to point out what hasn't come to surface anywhere, if you actually look at where the losses are coming from in Fannie and Freddie's books, the vast majority come from two sources:

-- private label securities, which are mostly securitizations of subprime mortgages. as i understand it these were purchased to meet certain social / governmental goals. i hope the government has learned its lesson with these and doesn't allow any sort of government guarantee or funding on securitizations of subprime loans.

-- prime mortgages for the bubble years, mostly 2006 and 2007, mostly in nevada, florida, and california, which exposed the entities to considerable losses as the price of collateral plummeted.

ie during _this_ crisis, interest rate policy was not responsible for losses.

Great post. I hope that once we abolish Fannie and Freddie it brings to an end this unhappy chapter in our financial history.

Nick write:


Banks didn't want to hold risky open mortgages on their books. So they securitised them and sold them off. Then the default risk turned out to be higher than the buyers of those securities thought it would be.

So your position is that MBS arose because 'open mortgages' were too risky? Support for that claim? Why now? Why not years ago?

Consider an alternative explanation: certain very large investors (pension funds, insurance companies) were required to hold a certain amount of AAA securities. The Basel accords stepped it up as well by focusing on risk weighting.

In 1979 60 corporations had AAA ratings. Now there are 4, right before the bubble burst, only 7.

The push behind structure investment vehicles arose from the regulatory pressure to hold AAA graded debt, which JP Morgan cleverly concocted.

Nonetheless... sure the housing market took a big tumbled, but the victims have mostly been banks hold loans directly. Losses due to MBS have been small in absolute terms, but crippling to a handful of firms who made very leveraged bets.

AIG tumbled not because of actual losses in MBS but because the derivatives contracts they wrote permitted their counter-parties to demand collateral for losses in mark-to-market value not realized losses. AIG only incurred actual losses after the government ceased control bought out the contracts at par. Reckless stupidity to an extreme.

The GSEs have bled capital, but that appears to be a case of fraud. Sub-prime borrowers were classified as prime but loan terms reflect the lie, and in some cases these loans were repackaged and sold as MBS with the lie intact.

Nick: In response to my comment about market risk that is realized when people move you wrote:

"When you move, you take the mortgage with you to your new house. Or it stays with the house, if the new owner is a good credit risk. Or you can pay it off, if you are unexpectedly flush with cash. You have to pay a penalty for early repayment, of course, equal to the difference between your old fixed rate and current market rates. But that "penalty" is not really a penalty, or a risk. All it means is that you insured yourself against interest rates rising over the term of the mortgage, and they didn't rise, they fell, so your insurance wasn't needed ex post. Like when I take out fire insurance on my house, and then my house doesn't burn down. I don't say 'Dammit, my house didn't burn down, so I lost that money I paid when I bet my house would burn down, so can I have my insurance premium back please?' "

Imagine you had a very high risk of fire and you bought 30 year fire insurance at a very high fixed premium (I know, no such thing - but we have to imagine it to complete the analogy to 30yr fixed mortgages). Imagine then that your risk of fire was massively reduced and you then came to the insurance company to terminate your insurance. They aren't just going to let you cancel it for nothing. If they are sane, they are going to make you pay the PV of 30 years of premium difference. Maybe you will say "Dammit, I should have got an 'open rate' insurance policy" (or maybe "dammit, I'm burning down my house"). One big difference in reality is term: a real insurance contract is one year and so the cancelation cost is negligible. But there is another important real difference: The fire insurance hedged a very real problem, the potential loss of your house. If the contract was a good match for the risk, then the mark-to-market loss on the insurance contract would be equal to your mark-to-market gain on the expected loss of your house so you shouldn't be upset at all no matter what happens. The fixed rate mortgage unfortunately serves no such purpose: It "hedges" an exposure that you never had in the first place.

However, I disagree with those commenters who suggest that a mortgage should be used to hedge your future income/consumption stream. A more reasonable assumption is that your desired hedge/asset allocation versus your income/consumption stream has been achieved before you buy the house. Then you add the house and mortgage into the asset mix. The goal of mortgage configuration is then only to hedge the house since the rest is already optimal. Then the real question is: what is the correlation of house prices to interest rates. The answer, to first order, has to be that houses stay relatively fixed relative to other non-money assets, and so, they grow in nominal value at somewhere between the rate of inflation and the rate of nominal growth. So from that perspective, a floating rate loan with a notional amount somewhat less than the value of the house seems like a pretty good bet. (The only caveat being that floating rate mortgage rates dont go negative. So you still need some kind of additional hedge for the negative nominal growth scenario. Maybe one should receive fixed a bit - e.g. buy some bonds).

On a separate note... I think I (and many others) sometimes come across harsh and possibly arrogant when I post. In my case it stems from the fact that I get excited to post something on the rare occasion when I see that you have written something that I think is wrong about something I think I know something about. So the context of all my posts is "Hey look! Nick made a mistake". That seems horribly ungrateful in the face of the almost daily enjoyment and education I get from your thought-provoking posts. All of which is to say, thanks for the education, and if ever I come across pompous or impolite (which I promise to try to avoid), I truly apologize.

In the interests of fairness and evenhandedness, you really do next need to do a follow-up post "Canadian mortgage interest rate calculations are weird." Really, mortgage calculated half-yearly, not in advance?

It's neither stupid nor dangerous, but it is weird.

Some commenters have said that most corporate debt is callable, just like US mortgages. Effectively this is not really true. Most bonds are callable at "make whole + spread", which means that the call price is determined by valuing the bond on the treasury curve plus a (typically small) spread. The call is therefore only a credit spread option; the option does not get in-the-money just because rates go down. Only a small fraction of (typically junk) bonds are callable at a fixed price.

@adjacent/q

Exactly what social / governmental goals were the private label securities being purchased to meet?

K: Thanks, you make a good point about the practice for corporate debt/bonds. This is the point I was trying to make above, that in corporate lending, the practice is make whole (what I've called unwinding costs). When the note is floating rate, there are no unwinding costs - and the remainder is spread, possibly plus a small penalty to cover fixed costs. Same point - it's unrelated to interest rates.

For the floating rate (truly open) mortgage borrower, same principle; the fixed rate faces unwinding/make whole (with, in Canada, caps on the amount of the unwinding costs - but the more effective part of the cap is the limit of closed to, generally, five year). Hence usually the only reason to refinance a floating-rate is to either benefit from a spread change (usually limited for 'prime' borrowers, so only makes sense when you have truly moved credit classes), or to fix for whatever reason, such as to limit exposure to increasing rates.

For an example, about four years ago I fixed a mortgage on a rental property - at the time it seemed like rates could go no further down (I was very wrong on this), and the upside of even lower rates was limited. For unrelated reasons, sold the property later, and had to pay some unwinding costs - but was able to reduce them by paying off a partial balance before, doubling up some payments, etc. Final cost was, I think, less than 2% of the balance outstanding (probably 1%).

While annoyed, this was significantly less than other fees and costs, esp real estate agent fees.

If everyone is so worried about the consumer, why not focus on the high estate agent fees? (Oh wait, competition agency in Canada is doing this).

My cynical side believes strongly that the reasons underlying the structure of the industry in the states has very, very little to do with consumer protection, and much to do with the incentives (rents) of real estate industry, mortgage brokers, servicing companies, etc. They all collect fees at any change, and these are much more significant than unwinding costs.

I can't easily keep up with all these very good comments. But the discussion doesn't really need me.

K: Nothing to apologise for. You are very civil.

Your two paragraphs on whether a fixed long-term mortgage acts as a hedge are very good. I'm with you on your great 30 year fire insurance analogy. This is where I think I disagree:

"However, I disagree with those commenters who suggest that a mortgage should be used to hedge your future income/consumption stream. A more reasonable assumption is that your desired hedge/asset allocation versus your income/consumption stream has been achieved before you buy the house. Then you add the house and mortgage into the asset mix."

I think I see it differently. We are born with a short position in housing, since we will need somewhere to live. (I did a post on this a few months back http://worthwhile.typepad.com/worthwhile_canadian_initi/2009/07/we-are-born-with-a-short-position-in-housing.html ). And the only way to cover this short position is to buy a house. So assume I am born and have no assets, except a stream of future income from wages. And no liabilities, except a stream of future liabilities of needing somewhere to live. (Assuming no inflation) if I get a 100% fixed interest 30 year mortgage and buy a house I have hedged/covered? my short position in housing perfectly, by selling a portion of my stream of future wages forward to cover my stream of future rental liabilities.

I'm not sure I've got my head around this properly.

Jon: you may be right. I have heard people (who? I can't remember) say that banks really don't like holding those open mortgages, because they never know when the damned things will be paid off. But this may be only a very small part of the story of the financial crisis. I maybe overcooked it on "dangerous". I stick with weird and stupid.

adjacent/q: I'm glad to see you here. I liked your comment on the other site. Feel free to re-post it here, since it's relevant.

Scott: Thanks! Wish I understood all the details better though. But again, that's what comments are for, and why blogs are so great.

The FDIC seems to think that the mismatch played a role in the S&L crisis:
http://www.fdic.gov/bank/historical/history/211_234.pdf

"By law and regulation, MSB assets were permitted to be invested
primarily in fixed-rate mortgages and long-term bonds, but as short-term interest
rates rose to historically high levels between 1979 and 1982, the market value of these assets
plunged. At the same time, MSB liabilities were composed almost exclusively of shortterm
deposits paying rates of interest subject to deposit interest-rate ceilings—and as
market rates rose, even small savers began to think like investors. ... Yields on assets rose much more slowly, and net interest margins shrank and became negative. Operating losses were so great that capital levels built up over a century or more of profitable operations quickly eroded. MSB failures were predictable and, arguably, preventable."

This is not the only issue, but the mismatch between long-term fixed-rate mortgages (open, if you like) and deposits which reprice frequently is serious for banks. And yes, it is theoretically possible to manage, but it is not costless, and may be subject to tail risk.

Greg: there are really two issues:

1. If banks borrow short and lend long, that creates clear risk. That's presumably the main reason behind the S&L crisis.

2. In addition to that known risk, there is the additional risk created if your long loans are open mortgages rather than closed mortgages. Because you don't know when they will be re-financed.

Many people say that 30 year fixed rate mortgages would not meet the test of the market, absent some form of government intervention/interference. I'm not so sure about this.

1. Empirically, as one commenter (GA?, sorry) above notes, Canada *used* to have 25 year fixed closed mortgages, until the government put a limit of 6 months interest on re-finance penalties. 25 is very close to 30, and now people live longer, maybe 30 is the new 25.

2. Theoretically, there seems to be a natural source of supply of fixed rate closed mortgages. Not from banks, which borrow short, but from pension plans and people about to retire. But they want to supply closed, not open mortgages.

Does anyone have any immediate thoughts on this? Is this worthy of a separate short post? "30 year fixed mortgages *can* meet the test of the market, if they are closed, and the government doesn't screw up the right of free contract".

It was Determinant, sorry. Let me repeat the important bit from his comment above:

"30-year fixed-coupon mortgages with payoff privileges used to be the norm both in Canada and the United States. I've spoken to friends (Canadians) who fondly remember this time. They think that a 25-year fixed mortgage is the height of certainty and security. They dread that uncontrollable or rising payments could force them out of their house.

The Canadian mortgage market took its present shape in the 1970's. The Canada Interest Act was amended to say that six months interest was the maximum penalty chargeable for breaking a mortgage, and rates started to rise into double digits. People started to refinance their mortgages at the earliest opportunity to try to get the rate down. Banks eventually started to offer the 1-5 year terms we see now in order to accommodate this. The fixed 25-year mortgage became a thing of the past."

Wow!

Who/what killed the 30 (or 25) year mortgage? I used to think it was inflation. But that argument never really worked theoretically, since it's inflation uncertainty, rather than inflation, that would be expected theoretically to kill off long fixed rate mortgages. But then they didn't reappear when inflation and inflation uncertainty came down with inflation targeting. Now determinant comes up with a much better theory. The limit on prepayment penalties is what killed the 30 year mortgage. I like it.

Dammit. This really deserves a post.

Is determinant right? (And was it 25 or 30? I had better be exact, if I can.)

sorry Nick, again. I'm not reading the comments so maybe this has been covered but why is paying for your prepayment option at exercise time better than paying for it over the life of the bond (via a higher interest rate)?

Is it a behavioural argument you're making?

Adam: If you know in advance that you would want to exercise, because you will come into a wad of cash in 10 years' time, for example, then it would be better to just get a 10 year mortgage.

But if you don't know whether or not you will come into a wad of cash in 10 years' time, it would be better to wait until you have the cash before paying to exercise. Because the cost of exercising the option would be correlated with your coming into a wad of cash. Correlated across time, and across states of the world.

And if you exercise the option just because interest rates have fallen, there is no social benefit to your doing that ex post, since it's merely a transfer of resources from lender to borrower. And ex ante that option seems to just create additional risk for both parties, and so is socially inefficient.

"And ex ante that option seems to just create additional risk for both parties, and so is socially inefficient."

No, it shifts risk from borrower to lender. Usually the lender will be in a better position to hold the risk or hedge it (can hedge it far more cheaply).

Your middle paragraph is utter nonesense.

Also, what's wrong with non-recourse loans? Most credit is limited liability in some sense or other. Do you also think the limited liability company is weird, stupid and dangerous?

Nic,
I'm not sure why this matters. US mortgages are essentially callable bonds. Are corporate bonds with embedded call options weird stupid and dangerous? If I'm a bank or investor holding a mortgage, there are many ways that I can hedge the risk of declining interest rates and/or prepayments(swaps, options, ect).

The real problem was that borrowers, lenders, investors, rating agencies and regulators all assumed that a 30% decline in US home prices was impossible. And it turned out that it wasn't.

Adam: If you and I take a bet, that usually makes us both worse off ex ante, if we are risk averse. The bet increases the variance of consumption for both of us. It's inefficient. Only if the thing we are betting on is correlated with the consumption of one of us would it be efficient to bet. Like when I take out fire insurance on my house.

As far as I can see, the bet involved in an open mortgage just increases the variance of consumption of both parties. It's inefficient.

So, you might be asking, if it's inefficient, why does it exist? Why does the US market create these open mortgages? My guess is that there is some sort of legal restriction, or failure to enforce the right of free contract, or some sort of implicit subsidy at the root of it. And we are getting at that in the good comments above. The best explanation is that the government sets arbitrary limits on pre-payment penalties, effectively killing the market in closed mortgages. Though I wouldn't rule out some sort of behavioural explanation. People are just confused, or don't see the costs of the option.

Both recourse and non-recourse loans may be efficient, in different circumstances. But if the government and courts refuse to enforce recourse loans, we have a prime facie inefficiency. In effect, the government is forcing us to take non-recourse loans, so it's unsurprising that non-recourse loans sometimes create inefficiencies.

Good questions by the way. You are forcing me to make explicit something that was implicit in my argument.

Gregor: There's a good comment by K above that clears up our confusion regarding callable corporate bonds. Here is it:

"Some commenters have said that most corporate debt is callable, just like US mortgages. Effectively this is not really true. Most bonds are callable at "make whole + spread", which means that the call price is determined by valuing the bond on the treasury curve plus a (typically small) spread. The call is therefore only a credit spread option; the option does not get in-the-money just because rates go down. Only a small fraction of (typically junk) bonds are callable at a fixed price."

So, if I read K correctly, a corporate bond is no more callable than a closed mortgage. You can repay it, but have to pay the interest rate spread if you do so. So it's neutral.

"The bet increases the variance of consumption for both of us."

Why? Somewhere in here you might have an argument but you haven't got it out yet.

Let's keep in mind that the option hedges me against falls in interest rates (as opposed to low interest rates). To the extent that falls in rates tend to happen when income/consumption is falling (recessions, I got fired) the ability to refinance and lower my monthly payment reduces the variance of my consumption path.

Now, as for the lender they make their living insuring various risks so it appropriate for them to hold this risk. They should have sufficient access to captial markets and expertise to manage the risk.

Now their maybe an argment in here somewhere. As far as I see, behavioural is not how you want to go. It's going to have to be about market incompleteness/imperfections. But "I don't understand why they do this" says only that you don't understand, it does not imply that "this" is weird, stupid or dangerous.

"So, if I read K correctly, a corporate bond is no more callable than a closed mortgage. You can repay it, but have to pay the interest rate spread if you do so. So it's neutral."

No, you don't read him correctly. He said, that it is an option on credit spreads not on interest rates. This doesn't imply that the bond is no more callable than a closed mortage. Basically it means you don't call the bond by pre-paying (ie. buying back for par), you buy it back at a fixed spread to treasuries.


 Nick: "(Assuming no inflation) if I get a 100% fixed interest 30 year mortgage and buy a house I have hedged/covered? my short position in housing perfectly, by selling a portion of my stream of future wages forward to cover my stream of future rental liabilities."

Good question.  And I agree that your way of looking at problem is better.  Housing is covered if you own a house.  The problem is always wages.  So if we go with your no inflation hypothesis then rates will be determined by growth expectations alone.  So then it really depends how you think your wages are influenced by growth.  So if you are at or below the median wage, the answer, sadly, is no, you just get inflation.  No real raise for you.  But then, you are probably stuck paying rent anyways.  If you are at or above the mean, which is the category of people who get to make mortgage choices, then you participate significantly in growth (at least on the upside).  So you still ought to get a floating rate to protect you in a recession.  And because floating mortgage rates don't go negative you should probably buy some long bonds to get some protection against negative growth (note that this is the opposite position of a fixed rate mortgage).

So is there any reason to go fixed?  Maybe if you have some reason to expect that your wages will not participate in growth, e.g. because your income is relatively low (near the median).  But you would have to be sure that your wages will not participate in *negative* growth, which is probably a bad assumption if you are receiving a low wage.  Unemployment, for example, is likely to be probable for you in a recession.  So then what you need is a floating rate with a cap. And still you probably need to hold some bonds.  But going fixed would be the perfect disaster for you in a severe recession.

And if median income earners need interest rate capped mortgages, who should sell them the cap?  High income earners could since they seem to outperform nominal growth in the long run.

So as far as I can tell, even without inflation risk, the conclusion is largely similar:  fixed rates bad, floating rates (possibly capped) + long bonds good.


Maybe that last post was premature...

I seem to have implicitly assumed that annual income was equal to the size of the mortgage so that a nominal growth=rate increase would cause an increase in income equal to the increased interest cost.  Obviously flawed...  Maybe you do need the portion of your mortgage that exceeds your annual income to be fixed when the ratio of mortgage to income is greater as is typical.  Need to think some more, but maybe there are some useful ideas here.

@Nick:

I drew that fact from an excellent article that gave a great history of the mortgage market in Canada, but I can't find it on the Internet. I think it's by Freeman, published in 1996. I remembered the penaltyas six months, but it appears now from that Re/Max site that it's three months.

To be clear, in Canada the traditional limit was always 25 years, 30 in the US. Longer terms are permitted, but have never taken off here.

Here's an excellent primer on why Canadian mortgages look the way they do:

http://www.td.com/economics/special/el0610_cdn_mort_market.pdf

Per the article, the Canada Interest Act allows borrowers to break a mortgage at the five-year mark with only a three-month interest penalty. Therefore banks have little incentive to offer fixed terms beyond this mark.

Second, banks fund their loans out of their deposit book. The Canada Deposit Insurance Corporation only insures deposits up to a 5 year duration. Therefore banks' deposit books are heavily tilted to 5-year terms and under. Canadian banks have historically chosen not to pursue the same path as US Savings & Loans and limit their term matching risk accordingly.

Here's another overview of Canada's mortgage market history.

http://www.simonaheather.com/cgi-bin/realestate/mortgage_financial_history

@GA:

Per the Bank of Montreal (and any other chartered bank), variable mortgages compound monthly, fixed mortgages compound semi-annually. It's a legacy of the agricultural boom of the late 19th and early 20th Century. Farmers took loans for planting and paid them off at harvest. The restriction on compounding was designed to be a tradeoff between farms and banks. Anybody who spends any time with an interest rate table in Canada (as I have) knows that Canadian mortgage rates have to be converted to per-annum compounded monthly rates.

@Nick (and apologies, I have perhaps muddied the waters by posting here as GA and Greg):
-I still think you are fixating (pun intended) on the wrong aspect of Canadian mortgages. It's not the closed/open aspect that is fundamentally different, but the maximum term of fixed-rate. Essentially, the more important bit is that they are floating rate mortgages with a fixed rate portion that does not exceed five years.
-Corporate bonds are mostly callable. The part that I believe K and I were drawing attention to is that either a) they are floating with minimal penalties, or b) they are fixed with unwinding costs/make whole (cost of calling depends on rates at the time of calling) - essentially closed, and equivalent to (how I understand) Danish closed mortgages work - borrower buys out their own mortgage at prevailing interest rates. When floating, calling has little advantage unless there are other changes (spread compression, borrower does not need the borrowed funds, etc); when fixed and closed, calling does not provide gain either.

That is incorrect. If the risk premia does not change, there is no difference between corporate and mortgage refinancing with a small prepayment penalty. It is not even close to being closed. This is why yield to first call is more important than yield to maturity for pricing. Now corporates do generally have a set call schedule rather than at will, but they are very much like fixed rate mortgages.

@K: "A more reasonable assumption is that your desired hedge/asset allocation versus your income/consumption stream has been achieved before you buy the house. Then you add the house and mortgage into the asset mix. The goal of mortgage configuration is then only to hedge the house since the rest is already optimal. Then the real question is: what is the correlation of house prices to interest rates."

I think you're looking at that wrong. Asset markets are incomplete and (with respect to shelter) lumpy, and people need to live somewhere. If you don't own a house, you still need to live somewhere, so you have a liability equal to the present value of your future shelter needs: you are effectively short the housing market. If you have tradable assets, you might be able to offset that short position, but if (like most young people, and I imagine most first-time home buyers) your assets are largely in the form of untradable human capital, then you're forced to hold a second-best portfolio. When you buy a house, you settle out your short position in housing, but (if you don't have a lot of other tradable assets) that improvement in your portfolio is bundled with a requirement to lever up your human capital. (The first best would be to sell off part of your human capital to finance the house purchase, but you can't do that.) To minimize risk, you want the cash flows of your liability (mortgage) to match as closely as possible those of your asset (human capital). Of course, the distribution of possible cash flows from human capital is complicated, and no conventional kind of mortgage is going to be a very good match. But until someone convinces me otherwise, I will continue to guess that a prepayable fixed-rate mortgage is a better match than either a variable rate or closed fixed rate mortgage. You're protected against a deflationary depression, which would hit your human capital (in nominal terms, anyhow), and you are not vulnerable to a real interest rate spike on the liability side. Sure, there are still a lot of things that can go wrong, but that will be the case with any kind of mortgage.

Actually (continuing from my last comment), if one accepts my argument that a mortgage is actually being used to finance human capital, then I think I can make a slightly more elegant case than "this is what I think" when it comes to the risk-return characteristics. You need to start with the assumption that the next 30 years are a series of random draws (maybe, say, 2 years at a time) from a population of which the past 60 years have been representative. In that case, the worst way to finance human capital is with a variable rate mortgage, because you are risking drawing 1981-1982, when the nominal value of your human capital declines dramatically, while the nominal value of your mortgage liability rises dramatically. That's like Russian roulette. You don't have to assume a high degree of risk aversion before that option is clearly eliminated. A closed fixed rate mortgage would be better, and the remaining question is how much you would be willing to pay for the prepayment option. I would argue that you would attach substantial value to that option, because it helps in the second worst case scenario, 2008-2009, when the nominal value of your human capital falls significantly while the nominal present value of your liability also rises.

The indenture will set for the call provisions for the bond. It can uae any formula for the call value. It could include a make whole clause, but that would be a rarity. Most commonly used is par plus 1 years interest diminished by the expended life of the bond, the equivalent of a prepayment penalty but if rates drop significantly well worth paying. Corporations call bonds all the time when rates are low.

@Andy:

1981-1982 is a red herring and doesn't support your argument. Given that the average Canadian householder has to sit down and figure out what length of term they want when they take out their mortgage or renew, the study of which strategy (5 5-year rate terms over the length of a 25 year amortization or 25 1-year rate terms) is of great practical interest in Canada and has been extensively studied by economists.

The results: For any given time series INCLUDING the horrid 1981-1982 period, it was always better to take a shorter fixed term. The longer term had no net value, in fact for any given time series you would lose money on this strategy. Let me be very clear, Canadian economists have looked extensively for evidence to support your theory and they couldn't find it. Even if you tried to make a really big mistake and make a longer term look good, you can't try it. We just can't stack the deck enough to make it work.

Why? First, shorter terms carry lower rates, they are lower down the yield curve. They almost always are, and inversions of the kind that people notice are extremely rare. In 2008 and 2009 the yield curve in mortgages WAS slightly inverted due to the Asset Backed Paper Market fiasco. Still, it passed soon enough that the average mortgage holder could ignore it.

It turns out that banks charge an extra percentage for a five-year fixed term compared to an equivalent series of 1-year terms. This is the premium paid to the depositor who funded the mortgage, or it can be seen the net cost the bank paid to borrow on five-year terms instead of one-year terms. Again, this simply reflects a normal yield curve.

It also turns out that even if you do get caught in an inverted situation, the flexibility to renew in short order combined with the discount you get for taking a shorter term provides a net benefit to householders who pursue this strategy. The shorter renewal period means you get relief for your "mistake" at the earliest opportunity.

@Nick:

It wasn't inflation that killed the fixed 25-year mortgage in Canada, at least not directly, it was interest rates. Interest rates rose steadily in the late 1960's after being stable for the previous two decades. Two percentage point increases in two years got banks and borrowers worried. Banks looked backward and didn't want to be caught with low-rate loans on their books for 25 years. Borrowers looked forward and didn't want to be stuck with historically high rates for 25 years if rates dropped. They looked at rate history and did the rational thing: refinance as soon as possible. The federal government facilitated this with its 3-month penalty rule.

Since borrowers didn't want to borrow at long terms and lenders didn't want to lend at long terms (though for opposite reasons), the federal government amended the National Housing Act in 1969 to allow for 5-year renewable terms. Previously the NHA and therefore the Canada Mortgage and Housing Corporation nd its all-important insurance insisted on 25-year terms. They could not maintain this insistence if their terms caused the pool of deposits available for mortgage lending to dry up. So the government relented, and the 5-year Canadian rollover mortgage was born.

There is another subtle thing here about why Canada never when for an equivalent of Fannie Mae: our banks are national in scope. Fannie and Freddie were reformed in the 1960's to even out mortgage rates in the US. What was happening is that due to a fragmented banking market interest rates between cities were quite variable. 1.5% was not uncommon. Of course with national media people complained about uneven rates and their effects on affordability. The solution was to spread the funds and risk around through a secondary market.

Canada never had this particular problem. With national banks and trust companies (remember those?) interest rates were always even across the country. The "smoothing" function the secondary market in the US was naturally performed by the consolidated financial positions of our chartered banks and trustcos without need for a secondary market.

@Determinant:

Either we are misunderstanding each other, or interest rates behave very strangely in Canada. For the term from Dec 1977 through Dec 1982, the US Treasury paid 7.5% p.a. to borrow 5-year fixed. It paid an average 11.4% to borrow consecutive for 1-year terms over the same period.

Granted, I've chosen what may be the worst possible period, but it was also one of the worst periods for human capital, in nominal terms, which is my point: the variable rate borrower took the interest rate hit at the worst possible time. And of course, Treasury rates aren't mortgage rates, which data I don't have at my fingertips, but I don't see why there would have been a dramatic opposing change in the spread between the two over that period.

Was the situation so very different in Canada? In any case, the original post was about American mortgages, so the appropriate question would be about American interest rates.

Lord:

I don't know which corp market you are referring to but make whole at treasuries + spread is very common in the US market. The call spread is typically very low - much lower than the issue spread so the option isn't worth much. Calls at a fixed price are also done sometimes for high yield issues, but most are not callable. The reason for this is the same as what Nick has been pointing out all along: People (investors in this case) don't want random, especially short, option positions in their portfolios. They generally want bonds that are like treasuries but with some credit spread. Issuers, on the other hand, want all kinds of features. But when push comes to shove, they rarely want to pay the market price for them. So most bonds are fairly plain vanilla.

And a minor point. You seem to refer to credit spread as risk premium in your earlier comment. Credit spread is partly risk premium and partly compensation for expected credit loss. And maybe this is good time to comment on the debate over the meaning of "risk neutral". It generally (and surely originally) refers to an investor who is indifferent to investments with the same expected return, no matter how different the risk profile. The risk neutral measure is called that because it computes prices as expectations of payoffs divided by the value of a future bank account. That is, it is a pricing measure that is indifferent to a random payoff vs one earned in a bank account. Risk premium is the difference in expected return between the risk neutral measure and some investors subjective measure. (Or typically the oft discussed but rarely observed text book objective measure).

Andy Harless: "I think you're looking at that wrong." You're right. I was looking at it wrong (see my (also confused) comment at 12:33)

Reading this article, a word "ratchet" popped up in my mind. And by googling, I found that there really exists such thing as "The Ratchet Mortgage". It seems that "US fixed rate mortgage" is more close to this product than *true* fixed rate mortgage in the first place.

Determinant: That article by Eric Lascelles at TD you recommended is indeed excellent reading. http://www.td.com/economics/special/el0610_cdn_mort_market.pdf

I have done a bit of a Google search, and I can't find any references to Canadian mortgages ever having had the interest rate fixed for 25 years (or any longer than 10 years. As far as I can see, the reason is Section 10 of the 1886 Canada Interest Act. It has always allowed any borrower to pay off any loan after 5 years with a maximum of 3 months interest as penalty.

So closed mortgages, or almost all commercial loans (there may be some exceptions, but I couldn't understand the legalese) ate effectively prohibited in Canada, because that maximum 3 month interest penalty makes them unenforceable.

Andy:

Take a look at the Federal Funds rate, which topped out at 18.5% in 1982. According to Business Week, the US Dollar yield curve inverted by only 2% in 1980. So individual auction rates for long Treasuries should be considerably higher.

Here's the article: http://www.businessweek.com/bwdaily/dnflash/jan2006/nf2006019_5024.htm

In Canada mortgage rates are highly correlated to the Bank of Canada's Overnight Rate. The Overnight Rate controls bank prime.

Again, because Canadian mortgages adjust more rapidly than American ones do and with far fewer frictional costs to refinance (zero if you time it right, which can be easily done by the average person), according to the time series you would have come out ahead if you took a 1-year term in at 18% in 1982. The subsequent drops in interest rates made you come out ahead since you got to take advantage of those in short order. Plus you didn't pay the long-term premium.

My point is that 30-year rates don't protect you from anything and even if you choose the worst possible moment to get a Canadian-style rollover mortgage, the rollover still provides you a house at less total cost than the American one does. 1982 provides no evidence that longer terms are more advantageous, in fact the opposite is true. Therefore it does not support your argument, it contradicts it.

Nick, what about this point that I made and also Andy Harless made a bit later? (reproduced below):

Nick: "The bet increases the variance of consumption for both of us."

Why?

Let's keep in mind that the option hedges me against falls in interest rates (as opposed to low interest rates). To the extent that falls in rates tend to happen when income/consumption is falling (recessions, I got fired) the ability to refinance and lower my monthly payment reduces the variance of my consumption path.

Now, as for the lender they make their living insuring various risks so it appropriate for them to hold this risk. They should have sufficient access to captial markets and expertise to manage the risk.


So, I spent the past half hour looking for callable bonds (I would have done it last night, but didn't have access to my Bloomberg).  I have checked thousands, a few hurndred more carefully.  Almost all investment grade (IG) issued bonds are either vanilla or make whole + spread (about half of each).  The spread is about 1/6 or 1/7 of the issue spread.  Some have zero spread which seems funny since you'd think you could always buy back a bond at zero spread in the market.  But then when you think about the fact that USA CDS trades around 40 bps, and some US corps are tighter than that it makes sense.  

The only IG price callables I have come across are long term financials, e.g. JPM 30 yr or perpetuals.  That would make sense as a hedge for US 30 yr mortgages.  Most financial bonds are vanilla though. And many junk bonds have fixed price calls.  Assuming rates don't explode, these calls are struck so as to be deep in the money assuming all goes OK from a credit perspective.  It gives the company some hope of reissuing earlier at a lower spread.  So there is still not much of an interest rate component to these calls.  As the bond approaches maturity it is most likely to either roll down in spread or become distressed.  In either case, rates are unlikely to play a significant role in the exercise decision.  Most callable junk bonds also have a make whole provision.  Make whole is decidedly not "a rarity".

Other, much smaller categories of bonds include converts and funky financial bonds (e.g. I saw a Morgan Stanley bond that pays quarterly 4 times the difference between 30y swap and 2y swap rates???).  So apart from bonds that appear to be specifically designed to hedge mortgages, I dont see any evidence of a significant role for callability relevant to interest rates in the US corporate bond market.  I had a look at Europe too.  Even more vanilla.

Andy and Determinant:

You are having a well informed, intelligent debate about what historically was more successful: fixed or floating.  I think it misses the point a bit though.  First of all, there are not very many independent periods in your sample - business cycles are long.  Second, mortgage rates were astronomical over much of the period compared to today.  Then they fell to near zero.  Not bullish for fixed rate mortgages.  These periods aren't relevant to what might happen when the short rate is at zero. Third, the belief that fixed rates are safer causes excessive demand, causes fixed mortgage rates to be excessively high, causes fixed mortgages to underperform.  Give people enough time to see the pattern and they will switch away from fixed.  "Anyways, the Bank will never lose control of inflation", etc, etc... This could skew the market towards floating and balance the risk (or drive the car right into the other ditch).  Never underestimate the propensity for humans to 1) assume things will never change, 2) completely fail to hedge themselves against such change, and 3) try to hedge themselves at the last and worst possible moment, thereby bringing about the change they never thought could happen.

Adam: Your and Andy's point is potentially a valid one. But an open mortgage makes a one-way bet. Wouldn't a symmetric 2-way bet be better? Like with a Variable rate mortgage? And are all recessions associated with falling interest rates? 1982? But finally, I say that open mortgages fail the test of the market in Canada, except for 6-month and 1 year terms (people selling houses in the near future?). Given the choice, people choose closed. And closed mortgages longer than 5 years are banned by Section 10 of the Canada Interest Act. All Canadian mortgages longer than 5 years must by law be open. None exist. 5 year closed beat more than 5 year open. Doing a new post on this.

yes, but a one way bet that's entirely to the benefit of the borrower because the lender is in a position to deversify and/or hedge the risk!

the point here is that they're not ineffecient as you claimed. Borrower risk is reduced and if lenders diversify and/or hedge the risk then aggregate welfare is improved.

Determinant:

We are still clearly misunderstanding each other (plus it may also be true that Canadian rates behave differently than American). If you took out a 5-year fixed-rate mortgage in 1978 in the US, you surely did a lot worse than if you took out a 5-year variable rate mortgage. (The fixed rate I quoted for 5-year Treasuries was the reported average for December 1978. The floating was an average of 1-year rates for that and the next 4 Decembers. Technically I shouldn't have used an arithmetic average, but it's close enough.)

You can rightly argue that, if you're not liquidity constrained or at risk for becoming liquidity constrained, then, over any 25-year period, you would have done better (in the sense of paying less interest) with a 1-year than a 5-year reset. But my argument assumed a liquidity constraint as the motivation for getting the mortgage in the first place. (Because human capital is illiquid, you take out a mortgage instead of selling some of that asset to purchase the house.) In practical terms, if you got a mortgage during the late 1970's, you had (ex post, cross-sectionally) a very high risk of losing your job in the subsequent recession(s) and having difficulty making your mortgage payments. If you had a variable rate mortgage, the risk of losing your house was significantly higher.

I wonder, also, about the correlation with other assets. If one isn't liquidity constrained, the motivation for getting a mortgage must involve a desire to maintain some other asset that is expected to have a higher return than the rate on the mortgage. Do the studies take into account the possible need to liquidate stock at the bottom of a bear market in order to make a mortgage payment? A fixed rate taken out in the late 1970's would have left you with more money with which to buy stock in 1981-1982. Offhand, I would guess that the returns on that stock would be more than the extra interest you paid in subsequent years.

Nick:

Why would a 2-way bet be better? What you want is to do well in bad scenarios, which could involve either falling or rising rates. You want to maximize your flexibility when something weird happens, which means you want to buy options, not make a simple hedge bet.

And you'll have to forgive me if, after just having lived throught the first decade of the third millenium CE, I am a bit skeptical as to the validity of the test of the market, particularly where mortgages are concerned.

K:

I went with historical data as a way to given some quantitative substance to my intuition. I still think historical experience is a reasonably good guide to the type of patterns that might emerge. As they say, history doesn't repeat itself, but it rhymes. In particular, there still exist the two risks with which I am concerned.

On the one hand, when the economy recovers, there is a risk that the Fed will have to spike up interest rates and induce a new recession to prevent an actual or imagined resurgence of inflation: hence, for human capital and interest rates, a scenario similar to the early 80's, which would wreak havoc on people with variable rate mortgages and liquidity constraints who are at risk for job loss. Granted, there is not the history of ingrained inflation that existed in the early 80's, but there are high public debt and unconventional central bank assets, which might cause expected inflation once a recovery runs its course, plus there is plenty of room for a crash in the dollar, so it's not too hard to imagine a fairly drastic and persistent rise in short rates (to well above where long rates are now) coupled with a fairly serious recession at some time in the next 10-20 years.

The other danger scenario -- where interest rates fall dramatically along with the value of human capital, as in 2008-2009 -- is in my opinion still very much a risk too, even with interest rates already low in historical terms. One only needs to look to Japan's experience to see the possibility -- but note that Japan had the advantage of being only one country and therefore the possibility of increasing net exports to blunt the deflation. Today it's not too hard to imagine the entire developed world falling into a deflationary depression, which could conceivably reduce US rates even below where Japan's are today.

With either of my worst-case scenarios, you can certainly make arguments as to why it is unlikely, or not likely to be as bad as I imagine, but we are concerned here with what might happen, not with what is likely. For a typical first-time home buyer, the risks of a variable rate or closed fixed rate mortgage are still severe, and the amount of protection provided by a prepayable fixed rate mortgage, while far from complete, is still worth the cost. US 30-year mortgages are normal, smart, and safe.

Adam P: "Borrower risk is reduced and if lenders diversify and/or hedge the risk then aggregate welfare is improved"

You'd think that'd be more or less true ... but evidence of recent events doesn't lend support.

@Nick:

Prior to 1969, the CMHC under the authority of the National Housing Act insisted that mortgages be made by a single borrower at a fixed rate for 25 years. Essentially the same as what happened in the US at the time pre-Fannie. To wit, National Trust would lend you money at 7% for 25 years. The term and amortization where identical. The borrower could pay it off with a penalty, but the lender was stuck with the loan. A lender had to keep the loan on its books for 25 years and had no opportunity to change the rate.

With rising rates in the 1960's, this seemed a poor strategy to many banks and trust companies and they wanted to move to more flexible terms to keep their loan and deposit books more current. They did not want be stuck with 7% loans and 9% deposits. We didn't have a secondary market and with fewer institutions had little reason to develop one.

@Andy:

Variable mortgages in Canada don't work like that Andy. True variables have a fixed payment and a variable amortization. Interest rate increases cause the amortization to lengthen. Again, however you are compensated by the lower rate. Pursuant to this idea, Canada now has a rule that you must qualify at the highest five-year rate in order to get any mortgage. In a new rising-rate environment, this is sound.

You can also lengthen your amortization on a closed renewable mortgage if you want to, but banks generally frown on this.

Second, I an fundamentally unimpressed by your arguments because Canada experienced similar interest rate increases in the 1980's as the US but this didn't result in a huge wave of defaults. Banks adjusted to keep their customers and their income streams. Customers adjusted to pay off their mortgages as quickly as possible. We have achieved a similar (in fact now 1% greater) rate of home ownership by making almost the opposite choices the US has. You can't argue the benefits of US amortization structure when its absence has not had a demonstrable ill effect in Canada. There is no better experiment than that. I'm saying the facts don't fit your theory.

I'm also saying your interest rate evidence is faulty and I'm deeply sceptical of it. 8% mortgages when the Fed Funds rate is at 18%? No, you need better evidence. Interest rates at that time were extremely volatile and the arithmetic average does not give an accurate picture of what happened.

Also in Canada mortgage interest is not and has never been deductible from income tax. While much ink has been spilled on this subject, it does mean that Canadians have a strong incentive to pay off their mortgages. The choice of making retirement savings over paying down a mortgage is considerably weighted towards paying down a mortgage up here. In 1981 it would have been a no-brainer.

Patrick, well I did say "if":)

But the thing is, I don't think you can pin the bubble and subsequent crisis on the prepayment option. There've been many housing bubbles and bubbles in other assets that didn't require this feature of the debt contracts.

You can argue that it was a regulatory failure or just a combustible mixture of greed, denial and stupidity (or even give an explanation that I agree with) but I just don't see how you pin it on the specification of the mortgage contract. I mean, was this type of mortgage in any sense new?

It wasn't the prepayment option, it was lax lending standards as a result of greed combined with poor oversight.

It's clear that in the US many people were sold mortgages who couldn't afford it. You can blame it on mis-sold ARM's, but I blame it on poor standards. In Canada the CMHC is an insurer and as such has a clear and continuous interest in ensuring that banks maintain adequate lending standards so that defaults are minimized.

Lending standards in Canada are clear and didn't change through the 1990's or 2000's. Neither did our default rate.

Institutions in the US sold mortgages to anyone with a pulse and then bundled the results off to investors. The combination of poor underwriting and poor regulation to ensure that underwriting took place is what caused the crisis to occur.

Determinant:

"8% mortgages when the Fed Funds rate is at 18%?"

What are you talking about? The example I gave was a mortgage originated in December 1977, when the funds rate was 6.5%. (I said 1978 in my last comment, but that was wrong. I was thinking that because 1978 was the first year the mortgage would be outstanding.) Actually interest rates were not all that volatile in 1977, and I very much doubt that it would have been hard to get a mortgage near 8% (assuming that 5-year terms were available, which I'm not sure). A variable rate mortgage (which may have been hard to get, because I don't think they were common in the US at the time) would certainly have reset upward dramatically over the subsequent 5 years. I don't understand your skepticism about the data. You cite data from 1982, which is irrelevant to the fixed rate case: once you have a fixed rate, it stays fixed for the entire term of the mortgage; that's the definition of a fixed rate. Yes, bond buyers (and banks) in December 1977 were, in retrospect, clearly foolish to accept such low yields, but believe me, they did. I know people who saw the foolishness of the bond market first hand.

"You can't argue the benefits of US amortization structure when its absence has not had a demonstrable ill effect in Canada."

Well, I can if your evidence is only the rate of home ownership, since I'm talking about risks for the individual borrower. Average data don't give us much information about tail risks.

"True variables have a fixed payment and a variable amortization. Interest rate increases cause the amortization to lengthen."

You mean, a 5-year mortgage becomes a 6-year mortgage when the interest rate goes up and then a 4-year when the interest rate goes down? I've never heard of anything like that, but it does sound like a good idea. If such mortgages were available in the US, they would probably be a better alternative to fixed-term mortgages of any kind. So I'll accept the argument that US open fixed-rate mortgages are not the best possible vehicle, but I still think they are preferable both to closed fixed-rate mortgages and to any kind of mortgage that I am aware of being available in the US.

I am asserting that in Canada even during the volatile 1970's and 1980's, the person with the shorter term came out ahead. I will let you pick the worst possible combination of interest rates you can imagine using historical data. The shorter term still wins. It benefits from cuts more quickly.

Yes, that is exactly how variable mortgages work. The payment is fixed, but the amortization lengthens from say 20 to 23 years. Or, to look at it another way, the amount you have to roll over when the mortgage renews varies. You can make this up by taking advantage of the 15-20% no-penalty prepayment privileges that banks include as standard in their mortgage contracts now, but that is the borrower's choice.

Here, have a look at the Bank of Montreal's rate website to see what I mean: www.bmo.com/home/personal/banking/mortgages-loans/mortgages#


Determinant:

I can believe that the shorter term resets always win, in terms of total interest paid, when the full period of financing is 25 years or longer. I have a very hard time believing that the shorter term resets win over the particular 5-year period from 1978 through 1982.

A critical issue here is whether borrowers are guaranteed the opportunity to roll over the loan. If you got a loan starting in 1978 with a balloon payment in 1982, you could be in real trouble if you lost your job in the recession. If your prospects appear to have weakened dramatically relative to the value of the property, the bank might decide to foreclose instead of renegotiating.

A couple of things.

First, when interest rates rise, real estate tends to remain flat or even decline.

Second, rollovers have generally not been an issue at Canadian financial institutions. As long as you continue to make the payments, it is standard and customary practice to rollover an existing mortgage without question.

In Ontario at least, mortgage default is dealt with through contractual power-of-sale provisions which take effect after 50 days of default.

The only time that rollovers did become an issue during the early 1980's, the CHMC started a "renewal insurance" program to deal with the situation. It was more theoretical than actual and responded more to customer sticker shock on renewal rather than actual declined mortgages.

Being thrown out of your home because the bank fails to renew is almost unheard of. If your situation deteriorates you are far more likely to default first. Between two-income families and Employment Insurance, family income is not an issue on renewal.

As Canadian banks are large and diversified they have the flexibility to take a bit of risk and let the customer sort out his financial situation, so long as they are paid on time. It's all about simple principles and standard national rules, along with a bit of patience backed up by power-of-sale provisions.

Nick: "Adam: Your and Andy's point is potentially a valid one. But an open mortgage makes a one-way bet. Wouldn't a symmetric 2-way bet be better?"

Not for the borrower, the option to prepay if rates go up is worthless. So are you advocating that mortgages be callable by the lender, in addition to pre-payable???? (That would be just as much a 2-way bet as a closed mortgage)

And the whole point I'm trying to make is that intermediaries whole reason for existing, in large part, is to take on these risks and diversify/hedge them more cheaply than private agents can.

I agree that the US system is weaker than the Canadian. But forgetting the obvious, the US culture is high risk high reward, and they mostly bet to win. That's we why have these options.


Ron
mortgagebrokerstore.com

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad