Although they sailed through the global financial crisis of 2008, Canadian banks’ early signs of stress due to the COVID-19 pandemic are developing into structural uncertainty. Their average LLRs/NPLs have already exceeded 2008 levels, but the Big 6’s provisions diverge significantly. (See Figure 1.)
Figure 1: Total Bank LLR/NPL
Source: Quarterly bank data
Even accounting for structural differences in risk-weighted asset portfolios, this variance suggests loss forecasts depend on divergent inputs to credit models that are themselves unpredictable in the current, disrupted environment. As a result, some of these banks may be underprovisioned for future shocks.
These challenges are underscored by the fact that Canadian consumers are already highly leveraged (debt service ratios of 15% are the highest in the developed world). An equity trap is likely to emerge as housing prices fall from the new highs of Summer 2020. With some of the highest RESL exposures among major Western banks, Canadian banks may be uniquely challenged in this environment.
Banks need to take steps now to prepare for a likely credit downturn that may be larger than any in recent Canadian market memory.
Given the indications that banks are underprovisioned for the scale of the possible losses, banks need to take steps now to prepare for a likely credit downturn that may be larger than any in recent Canadian market memory. That means adjusting credit models with new, advanced analytics and developing agile, quick response capabilities to recognize early changes to borrower risk profiles. It also means creating new, more efficient retention and renewal processes that are centrally driven and don’t rely on the whims of thousands of individual branch staff.
Banks must develop these strategies now so that they can be deployed before customers start to default. The timeline is extremely critical because, as the U.S. demonstrated 12 years ago, things happen very quickly once the first dominos start to fall — leaving little time to think.
This Time is Different
In 2008, careful mortgage and consumer lending practices at Canadian banks resulted in limited exposure to the types of products responsible for the origins of the crisis. The COVID-19 pandemic, however, is an entirely external shock to all markets. The complete shutdown of large sectors of the economy, closure of the U.S. border and a collapse in oil prices have all hit the Canadian economy — and the banks — hard.
In addition to the larger macro-economic shocks, there is evidence that Canadian banks may have the highest exposure to the RESL market among industrialized countries. Highly-indebted Canadian borrowers are experiencing more stress during the current crisis; 16% of residential mortgage holders have voluntarily deferred payments for six months, amounting to an industrywide $1 billion reduction in scheduled periodic repayments per month. Most deferrals expire this month, at around the same time that the 4.7 million CERB payment recipients will have lost that benefit (of whom only some will continue to receive income from Employment Insurance and the new Recovery Benefit programs.)
Coupled with forecasts of 7% to 22% negative HPI from the September high in the overheated Canadian housing market and a potential increase in 10% unemployment should a second pandemic wave hit, this represents a dangerous cocktail of unsustainable debt and negative equity that may not be adequately reflected in the banks’ risk provisions.
Unique Canadian Risks
Canadian mortgages have certain unique features that make them particularly prone to sudden shocks in the system:
Concentrated, over-leveraged LTV in large markets: Although reported mean LTVs aren’t onerous, the averaging hides significant extremes at the top end, including housing bubbles in Vancouver and Toronto that have led to over-leveraging.
Gaps in creditor insurance coverage: Loans with an LTV of more than 80% are required to have compulsory CMHC or equivalent insurance that is supposed to protect the lender in the event of default. But banks are still on the hook for the costs of repossession, maintenance and resale and are fully liable for the cost of negative equity. In addition, the proportion of uninsured loans in the 65% to 80% LTV range has increased, suggesting that banks will have more risk if HPI falls sharply.
Variable provincial regulations: If HPI falls enough to wipe out equity, the logical move is for the mortgage holder to hand back the keys. In theory banks are legally protected from this, but in practice the situation is regionally complex, with some provinces being more borrower-friendly than others (for example, as seen in Windsor, Ontario in 2008 where provincial jurisdiction enabled certain cases with negative equity to forsake their obligations).
COVID-19 Unbalances the RESL System
These structural fault lines have now been stressed further by the emerging repercussions of the COVID-19 crisis. Bank earnings for the second and third quarters were already hurt by unprecedented increases in loan-loss provisions. But the longer-lasting impacts are likely to be far more severe.
First, most expert industry forecasts are converging on a nationwide drop in Canadian home prices that exceeds the range built into most banks’ economic models. This presents a downside risk to capital as the book is revalued, requiring incremental RWAs and direct capital losses. Moody’s base case forecasts that the initial HPI declines will be highest in bubble markets of Calgary, Edmonton and Toronto, leading to a potential vicious circle of repossessions, forced resales and further HPI falls. (See Figure 2.)
Figure 2: RPS HPI, Forecast Peak-To-Trough Decline, %, September Vintage
Source: RPS, Moody’s Analytics
Additionally, the industry is likely to be hit hardest in Q2 2021 when the impact of mortgage-payment deferral programs that expire at the end of this financial year is most likely to hit the balance sheet. With 14% of the portfolio balances enrolled in deferral programs, the typical path of mitigation for stressed borrowers has been delayed, which will lead to a double shock now that OSFI has officially stopped allowing banks to recognize deferred loans as performing. Not only will a proportion of borrowers now deferring payments inevitably default, but the delinquency signals will have been unobserved, suggesting it may be too late to apply the normal mitigations.
Finally, banks that are most exposed to residential mortgages have rapidly grown their portfolios through aggressive originations over the past five years. This means they now have very large maturity cohorts bubbling up through the system with the capacity to further impair the balance sheet when the best customers leave for better deals. Left behind will be the less creditworthy borrowers who are further constrained by new stringent CMHC qualification criteria, putting more pressure on PCLs and RWAs.
Swift Action Needed
Banks can’t afford to wait to see what happens especially since decisions made under such circumstances are often haphazard and lead to their own set of unintended consequences. For example, extending deferrals to the whole portfolio is likely to increase the eventual default risk when the deferred interest payments are compounded into the remaining amortization period. And for every month that foreclosure is delayed, the final asset recovery value reduces.
Novantas sees three key steps that banks can make now to take charge of their portfolios even before the true size and scale of the risks are clear:
Use data to measure new dimensions of risk: While most banks have solid portfolio-level models that estimate sensitivities to stress and detailed loan-level models for predicting BAU risk, few have the capability to understand unique geographic and account-level sensitivity to the macroeconomic environment.
Doing this well requires banks to understand how detailed loan and borrower characteristics will be impacted by developments such as housing price declines and rising unemployment at the postcode level. It also should involve leveraging valuable, but untraditional, data sources such as deposit transaction data that profile individual cash management behavior, balance persistence and potential risk. With limited stress history in Canada, banks should overlay models with observations from other geographies to bolster understanding of stressed risk.
Create agile playbooks to de-risk borrowers in real time: Banks need to respond quickly to market and borrower conditions. This can include considering which clients should receive an extension of the payment deferral program based on broader relationship value. They can also create incentives for borrowers to lower their risk profiles (by providing prepayment penalty rebates or discounts for cash-in refinances) or better align resource allocation with expected risk to avoid building greater exposure in geographic hot spots. Another potential approach is to update risk concentration limits to be more dynamic and to incorporate new and different segmentation.
Develop proactive retention and renewal treatments driven centrally: A new approach to mortgage retention (both mid-term and end-of-term) would consider the advantageous return on capital represented by the most creditworthy borrowers who are the highest attrition risk. This would require laser-targeted customer retention communications via the deployment of advanced AI-based adaptive segmentation/messaging platforms to eliminate the costly false positives that occur in linear modelling approaches.
Banks that use this unique period to create new, innovative analytics, processes and treatments will have a competitive advantage over those that take a wait-and-see approach. Once the true scale of the crisis becomes evident in the portfolios, it will already be too late to alter the trajectory. In fact, being the first mover in credit crises always has advantages. The last mover, meanwhile, gets stuck with the leftovers.