Last week TD came out with a brief research report titled “Canadian Housing - How Bad Could This Get?” that looked at what the impact of a housing downturn would be on Canadian banks.
As a bit of background, banks are in the business of borrowing money from investors and lending that money out to borrowers (as well as other businesses like transaction facilitation and capital markets and what-not). A big component of the money lent out by a bank comes in the form of people taking out mortgages. If something happens in the housing market that affects borrowers and their mortgages, then that can affect the banks.
The two basic ways that banks can be affected is through the balance sheet or through the income statement.
If people stop repaying their mortgages entirely, as happened with many underwater borrowers in the US and Ireland (and was particularly severe in subprime pools), then the bank starts the foreclosure process to get their money back from the borrower, and will likely have a small loss due to the costs of foreclosing (particularly if the size of the mortgage is high relative to the value of the house). If the security that backs the loan (the house) is also worth less because housing prices have fallen, then even after foreclosing and auctioning off the house, a large part of the mortgage may be unrecoverable, and the bank takes a big loss on that. This is a balance sheet problem: the assets (mortgages) are worth less while the bank still has to pay the investors that lent it money. This is a crisis for a bank, and can lead to the bank having to raise more money, get a bailout, or go into bankruptcy.
The TD report says that Canadian banks are not at risk of a balance sheet crisis from a housing market slowdown, and I generally agree: even though we may see some severe price reductions and increases in mortgage defaults in certain Canadian cities, the banks here have been very good at getting rid of their exposure to that risk through insurance and securitization. The mortgages they do hold generally have low loan-to-value ratios, so that even with fairly severe price corrections they won’t have a critical imbalance between their own borrowings from investors and the value of the mortgages they’ve lent out.
The other impact can be felt in the income statement: since mortgages are a big part of a bank’s business, they also make up a large part of where the profits come from. If fewer people are buying houses, and taking out smaller mortgages to boot after prices come down, then the banks will be making less money because of the effect of writing fewer mortgages.
Jason Bilodeau says that there will be some impact of a correction there, but that it won’t be large: he estimates that for a 5% reduction in mortgage growth, the banks would report 7% lower profits. I found that rather surprising: that the impact to profits would be higher than the reduction in activity, though in hindsight it should have been expected of any business with fixed and variable costs.
Thing is, he calls this 5% reduction a “worst case”. That is not the worst case, not by a long-shot: hell, sales are already down 25% in Toronto and over 30% in Vancouver. Those two cities hold roughly 20% of Canada’s population, but the dollar-volume (i.e., the size of the mortgages for the banks) is even higher than that because house prices are also higher. With a bit of back-of-the-envelope math, it looks like the mortgage growth is already in negative territory by about 10% even if the rest of the country was perfectly stable.
One other important factor is that housing activity can change without necessarily changing the size of the banks’ books: a lot of housing activity consists of people who already have houses and mortgages selling to one another. In those cases, it’s possible that no new net mortgages will be written, so there is a case to be made for changes in first-time buyer activity being the important metric, not overall sales. But, I think it’s a reasonable to assume that first time buyers will move with the rest of the market, or if anything, to drop even more in times like these (in rising markets, there’s a sense of “buy now or be priced out forever” leading to more first time buyer activity, and larger mortgages at that as they jump in without downpayments).
So what would the worst-case-scenario look like? Well, in 1989, Toronto had about a year where housing activity was cut in half, and activity only picked up again after price declines set in. Even if the correction was limited to Vancouver and Toronto, with a sales decline of 50% (or some combination of sales declines and price correction) the overall national change in mortgage activity would be more like 17% lower.
At some point I may run through the financial statements of a bank to pull out how much of the increase in profitability over the past 10 years has been due to mortgage credit expansion, and what a correction of ~30% in prices with some decrease in volume would look like to earnings. But unfortunately dear readers, I simply don’t have the time right now. Instead, let’s just extrapolate from TD’s numbers in the report: if a 5% reduction in activity lead to a 7% decrease in EPS, a 17% decline in activity might lead to a 24% hit to EPS.
That’s not necessarily killer: I wouldn’t be shorting banks on that basis, but I wouldn’t be buying them either. Also keep in mind that this still isn’t the worst-case, since many other cities will also likely follow Toronto and Vancouver’s lead. First-time buyers might drop out of the market entirely for a few years, to correct the imbalance in the ownership rate. And people may start making downpayments if they aren’t rushing in to beat price increases.
Using this historical experience as a guide, our sensitivity analysis suggests that our 2013 estimates could be roughly 7% lower, all else the same, if the market were to experience what would be the worst housing correction on our record.
Part of where the low numbers come from is looking at the effect of past housing busts: 1989 was bad in Toronto (and indeed, many regional trusts collapsed), but the rest of the country was fairly unscathed, with Vancouver powering on for 5 more years. Nationally, the banks (and mortgage volume) came out ok. In 1994 Vancouver had a correction, but that was when Toronto was pulling out of its decline. These regional housing corrections have never lead to major problems for the banks before because they were regional and uncorrelated. This time around is going to be much worse than the historical record would suggest because there’s a lot of correlation. Alberta managed to blow off a bit of steam with a soft landing over the past 3 years, but could be sucked in for another 10% down-leg. Ottawa, Montreal, Halifax, Winterpeg and Regina are all equally affected by record-low interest rates and primed for a correction as well.
“In addition to the earnings risk, as is often the case, the cycle would likely be marked by poor sentiment and compressed multiples driving share underperformance. Typically, the group underperforms early in the decline, before recovering as the market looks through to the bottom and eventually improving conditions.”
There are enough investors worried about this, apparently, that multiple compression has already set in. Despite fairly good EPS numbers, most of the banks have gone nowhere since powering out of the 2009 crisis, more-or-less plateauing since the spring of 2010. Once earnings growth turns negative, there will likely be another round of share price decreases, at which point I may look at buying back into the sector.
As an aside, this report begins with the statement “We continue to believe that we are in the midst of a material slowdown in Canadian housing activity.” It’s a very devious statement: throwing in that “We continue” makes it seem like they forecasted this some time ago, rather than, say, just starting to say this last week. Indeed, back in the summer in a report on Genworth, they believed there would be no material slowdown.