He concludes that their financial health is substantially worse than conventional wisdom would have it.
The loan-to-value ratio in the Canadian market is similar to what prevailed in the United States just before the Great Recession, at 54 to 55 percent. The portion of loans with loan-to-value ratios greater than 80 percent is higher for Canada today than it was in the U.S. in 2007, just before everything fell apart.
With a large percentage of Canadian banks' mortgages in the 70%-80% loan/value bucket, it would take only a 10% decline in prices to cause those loans to exceed the standards currently allowed by the CMHC [Canadian Mortgage and Housing Corp.] on newly underwritten loans. If housing values were to fall precipitously, many of those loans would fall into the higher-loan/value categories. We also saw this with U.S. residential loans in 2009.
We have estimated the potential losses for CMHC coverage claims for each Canadian bank, on average, in its residential loan portfolio. While many Canadian mortgages have low loan/value ratios because of recent home price appreciation, 28% of insured Canadian mortgages have ratios of 80% or greater. We worry that this creates a risk that CMHC's liabilities could exceed its equity if Canadian home prices were to decline. We estimate that if home prices were to decline 20%, and if 20% of underwater loans defaulted and had 60% recovery rates, the resulting CAD 12 billion of losses would consume more than 90% of the insurance fund's CAD 13 billion of capital. If 100% of underwater loans were to default, we calculate that even a modest 10% decline in prices would more than exhaust CMHC's capital.
That is to say, CMHC may need to be recapitalized by the Canadian taxpayer, just as Fannie Mae and Freddie Mac were in the United States.
(Read More: Is Canada's Housing Bubble About to Burst?)
Note that these decline, default and recovery figures are not especially aggressive. They're well within the realm of probability. In March, Moody's warned that Canada's home prices could decline by 44 percent in the event of a severe economic shock.
The main path to damage in the banking sector will come from uninsured residential loans.
For the uninsured residential loan portfolio, we think the potential losses directly to the banks could be meaningful and impair their capital bases. Again, we assume a 60% recovery of the uninsured loan balance, assuming 100% losses of the uninsured balance with losses directly impairing tangible capital. If 20% of the uninsured underwater residential loans are losses, the impact on tangible capital levels becomes meaningful with a 30%-40% reduction in pricing. In a worst-case scenario, if all of the uninsured loans were losses and residential prices fell 30%, we think nearly half of most banks' tangible equity would be affected.
Werner thinks that National Bank of Canada and Canadian Imperial Bank of Commerce would be the most affected of Canada's large banks. Toronto Dominion Bank and Bank of Montreal the least affected.
It's important to note here that Werther doesn't see a crash coming until after interest rates are raised—something he doesn't expect to happen soon due to sluggishness in the Canadian economy.
_ By CNBC's John Carney. Follow him on Twitter @Carney