New Canadian mortgage rules have already unsettled borrowers and lenders, but banks can potentially mitigate the impact by overhauling models and taking advantage of enhanced analytics.
With some C$700 billion of mortgage balances maturing over the next 12 months, banks face an urgent need to act before portfolios become more risky and creditworthy borrowers pursue alternative lenders. Time is of the essence: RBC recently wrote that the housing market already has been “whipsawed” by the new rules even though they just took effect in January.
The new rules, imposed by the Office of the Superintendent of Financial Institutions (OSFI), require all uninsured mortgages that are funded by federally-regulated financial institutions to be stress-tested at the greater of two percentage points above the actual contracted rate or at the Bank of Canada five-year posted rate. The move comes more than a year after the regulator began requiring all insured mortgages to be qualified at the central bank’s five-year posted rate. It also coincides with a trend towards fewer insured and more uninsured mortgages in the portfolios of the top five banks (see Figure 1).
The tighter requirements are already impacting real-estate markets and mortgage pricing even though they don’t apply to borrowers who renew loans with their current lender. The Bank of Canada has reported that the new rules drove tighter lending conditions in the first quarter. “In the next quarter, respondents expect a decrease in the demand for low-ratio mortgages and HELOCs (home equity lines of credit),” the central bank said in its quarterly senior loan officer survey.
AN UNWELCOME SURPRISE
The situation is startling to both banks and borrowers. Years of falling interest rates in the aftermath of the last financial crisis taught a generation of home buyers that renewing a mortgage is an opportunity to reduce payments. Now, the first wave of post-crisis renewals is occurring at a time when mortgage rates are rising.
Most borrowers who are rolling off a five-year fixed term are finding their new rates to be substantially higher than their existing ones. Competitive discounted fixed five-year mortgage rates today range from 3.19% to 3.59% compared with 2.74% five years ago. This represents a 16% to 31% increase in monthly interest payments, equivalent to $75 to $142 a month incremental draw on average household income — a rate shock that may have long-term impact on historically benign Canadian provisions for credit losses (PCL). This shock will particularly be felt in Vancouver and Toronto where mortgage balances are significantly above average.
The biggest rate shocks will be felt by people who thought they were being prudent borrowers by originally putting down 20% or more, thus avoiding the cost of mortgage-default insurance. The competitive 3.19%, five-year fixed rate described above is for people who started with a so-called high-ratio mortgage, where the LTV is >80%, and/or for those who have a current LTV of <65% (with an original purchase price of below $1 million). Rates are currently in the 3.39%-to- 3.59% range, however, for mortgage renewals between 65% and 80% LTV and/or had an original purchase price of $1 million and higher. This asymmetry is due to the withdrawal of government support to banks for uninsured loans, making the >80% insured segment the most attractive because those loans are still readily securitisable.
The new mortgage-industry rules (known as B20) also are complicating the process of breaking with the current lender at product term maturity to take advantage of a lower rate available elsewhere. The new rules require uninsured borrowers (i.e. with LTV <80%) to undergo a stress test that ensures they could afford their mortgage payments at the greater of the Bank of Canada’s five-year benchmark rate (now 5.14%) or the actual rate being offered plus two percentage points. (People with LTV >80% already faced a stress test, but it was set at the five-year Bank of Canada rate and thus slightly less stringent.)
The potential problem for lenders is that that these stress tests only apply if borrowers move their mortgage to a new lender; there are no stress tests required if the renewal takes place with the current lender. That means mortgage lenders will experience a hitherto unsuspected stickiness with anyone who doesn’t pass the stress test (or think they may not pass). While that may result in higher renewal net interest margins, it also means that banks may find themselves with portfolios biased towards segments who may be deemed less creditworthy. This bias will only be accentuated by the falling off in new mortgage originations that accompanies the slowdown in real estate transactions.
Because the stress test also applies to those refinancing (i.e. drawing down equity) with the current lender, those refis at renewal may also slow down — reducing the balance increment presently resulting.
The ultimate impact of these changes on the dynamics and structure of banks’ mortgage portfolios is still largely unknown, but some broad indications can be inferred as the latest loans come up for renewal.
BORROWER BEHAVIOR LIKELY HEADED FOR A CHANGE
For one thing, the impact of a 45—85 bps hike as current terms mature is likely to induce certain borrowers who previously may have simply rolled into a new term to now take proactive action. That may include searching for better rates elsewhere — a research exercise that is easier to do today than in the previous cycle of rising rates. As a result, churn rates are likely to increase in those segments for whom switching remains an option.
Other borrowers are likely to want to re-mortgage with another lender to take advantage of lower rates, but will find themselves unable to because they fail the stress test (or think they will). Such people will then wind up staying with their current lender, most probably at a higher rate than is available on the open market. This will increase overall portfolio risk because the switch to a higher rate is itself likely to make these segments more prone to default and because their LTV is pinned to historically appreciating house price moves that may not apply going forward. Until now, house price index (HPI) growth has decisively outstripped the rate of growth of debt-to-disposable income but this may be about to change (see Figure 2).
MODEL OVERHAUL CAN HELP MITIGATE MORTGAGE IMPACT
To mitigate these repercussions, banks and other mortgage providers need to incorporate new forms of propensity models that address the rate shock and stress testing on renewal segments. Among other things, they should consider new pricing models that explicitly address how the price/LTV gradient translates into expected losses. This will help address the issue of adverse selection posed by upwardly pricing uninsured mortgage loans that are deemed to come closest to failing stress-test criteria.
Additionally, optimising renewals pricing for net economic profit (NEP) can capture the long-term cost of additional capital to cover the riskier mortgage loans that are retained in the portfolio.
Banks should also adopt more accurate and speedy automated valuation models (AVM) to capture the movements of real estate prices at a micro-geography and property segment level. They can also pursue more forward-looking loss-given-default (LGD) calculations based on latest house price indices and free of biases, including historical price appreciations and appraisal biases, and taking consideration of potential house damage.
Finally, banks can consider cognitive solutions for developing “contextual engagement” with in-flight renewal customers to understand how they are responding to the bewildering rate of change in the financial, economic and regulatory environment.
These may include optimization processes for converting deep learning insights from customer conversations into personalised communications treatments that address customer concerns and issues as and when they arise.
Banks should consider these actions as soon as possible, especially since current trends in real-estate demand and pricing don’t bode well for the months ahead.
RBC economists report Canadian home sales transactions slid by more than 19% in January and February and only eked out a small monthly gain of 1.9% in March. This is widely viewed as a reaction to the tighter mortgage requirements, as well as the interest-rate hike.
“Whether that impact (of tighter mortgage regulations) is largely psychological or reflects a substantial pool of buyers being shut out by stricter qualifying rules remains to be seen,” RBC wrote.
In addition, national benchmark pricing decelerated further to 4.6% in March from 7.1% in February with a negative year-over-year change in Toronto’s benchmark price for the first time since 2009.
RBC’s forecast for the rest of 2018 is similarly bearish.
The new measures and additional interest-rate hikes “will result in more subdued housing market activity this year, and much more moderate price increases overall in Canada,” the firm said. It forecasts resales to decline 3.7% to 497,400 units in 2018. We project price gains to be limited to just 2% at the national level, down significantly from nearly 11% in 2017.”
Banks and other mortgage providers that invest now in these new modelling, appraisal and deep- learning approaches are going to be better-equipped to deal with the approaching headwinds. The next generation of analytics are adaptive, so lenders can learn how to respond to this new landscape in-flight by making better pricing, risk and treatment decisions.