Public consultation on guideline B-20: Residential Mortgage Underwriting

OSFI is seeking stakeholder feedback on the following debt serviceability measures, including impacts they may have on borrowers and lenders:

• Loan-to-income (LTI) and debt-to-income (DTI) restrictions – i.e., measures that restrict mortgage debt or total indebtedness as a multiple, or percentage, of borrower income.
• Debt service coverage restrictions – i.e., measures that restrict ongoing debt service (principal, interest and other related expenses) obligations as a percentage of borrower income.
• Interest rate affordability stress tests – i.e., a minimum interest rate that is applied in debt service coverage calculations to test a borrower’s ability to afford higher debt payments in the event of negative financial shocks.

It is worth noting that these measures could be implemented at the:

• Loan level – i.e., applied for each borrower’s mortgage application; or,
• Lender level – i.e., applied on the lender’s aggregate volume or portfolio of mortgage business.

At the lender level, limits can also be applied at different levels of measurement, restricting either the dollar value or count of the:

• Flow, or volume, of new mortgages underwritten; or,
• Stock of outstanding mortgages (portfolio).

Above all, any new measures would be intended to reinforce B-20 Principle 3 and the expectation that borrowers are appropriately assessed for varied financial and economic conditions and/or higher interest rates.

OSFI welcomes stakeholder views on these measures, how they might be implemented, and other measures that could address prudential risks arising from high household indebtedness. OSFI may choose to pursue one or more of these measures or others that meet OSFI’s policy objective.

Policy objective

To enhance the credit quality of residential mortgage assets and underwriting practices of federally regulated lenders.

1. Loan-to-income and debt-to-income restrictions

Some jurisdictions employ limits on either mortgage debt relative to borrower income (loan-to-income or “LTI”) or total indebtedness relative to borrower income (debt-to-income or “DTI”). The purpose of these limits is to control the extent of indebtedness. Lower indebtedness reduces the probability of borrower default by making ongoing debt payments more manageable and limits a lender’s potential losses in case of borrower default.

Although specific LTI or DTI limits have not been prescribed for FRFIs, an LTI of 4.5x (450%) or more is generally considered to be “high,” and in the past, OSFI has used a threshold of 3.5x (350%) LTI in its industry surveillance activities. That said, an appropriate level for LTI is informed by prevailing interest rates and can, therefore, shift over time. For example, as interest rates rise, a lower LTI limit may be more appropriate.

A lender-level volume restriction on high LTI lending, often referred to as a “flow limit” or “speed limit,” could support the policy objective outlined above. Such a limit would allow lenders discretion to underwrite a certain number of loans that exceed a prudent threshold but which may, based on other lending criteria, be permissible under their residential mortgage underwriting policies (RMUPs). Imposing such limits may also reduce the potential for policy leakage and migration of lending activity to the unregulated lending sphere.

A lender-level volume restriction on high LTI lending could complement existing measures. To operate effectively, OSFI is considering:

• A clear and consistent definition of “income”, for the purpose of calculating LTI;
• Appropriate “high” LTI thresholds (e.g., 3.5x, 4.5x), in view of different macroeconomic conditions, and through-the-cycle; and,
• A credible, industry-wide LTI volume limit (e.g., 20 – 30%);

OSFI is seeking stakeholder input on a specific proposal to adopt a “high” LTI threshold of 4.5x, with a 25% quarterly volume limit on originations that exceed this threshold. The proposed limit would apply on the total dollar value of quarterly originations. Compared to a restriction on loan count, this would limit incentives to concentrate high LTI lending in a small number of loans.

Chart 1 - Share of high loan-to-income (>4.5x) originations, industry-wide (volume)

High LTI lending is not evenly distributed across FRFI lenders, and OSFI is aware that smaller and medium-sized lenders tend to have a greater proportion of high LTI loans in their portfolios. Accordingly, any transition would consider the need for different FRFIs to adapt their business plans and RMUPs. High LTI-related risks could also be addressed through the capital framework, such that more capital is set aside for high LTI loans.

Chart 2 - Share of high loan-to-income (>4.5x) originations, Mid-Tier and Small and Medium-sized (SMS) Institutions (volume)

Finally, an LTI measure may require ongoing monitoring and calibration, similar to the current MQR, to reflect prevailing interest rate and other macroeconomic conditions.

2. Debt service coverage restrictions

Debt service coverage ratios are already employed by federally regulated lenders, with the Gross Debt Service (GDS) and Total Debt Service (TDS) ratios most commonly used. For insured mortgages in Canada, GDS and TDS limits are prescribed in law and are currently set at 39% and 44%, respectively.

Beyond those requirements, B-20 does not articulate limits on GDS and TDS for uninsured mortgages and generally permits FRFIs to establish debt serviceability metrics under their RMUPs that facilitate an accurate assessment of a borrower’s capacity to service the loan. Regardless of the metric used, OSFI still expects FRFIs to calculate the average GDS and TDS scores for all mortgages underwritten and/or acquired and ensure that they are less than the FRFI’s stated maximums, as articulated in its RMUP, and reflect a reasonable distribution across the portfolio.

Existing OSFI guidance could be strengthened by introducing a lender-level volume restriction on loans with high debt service ratios. For such a measure to operate effectively, OSFI is considering:

• The formulas and definitions for GDS and TDS and whether to adopt those that currently exist for insured mortgages;
• Appropriate GDS and TDS thresholds for uninsured mortgages, which could involve graduated or tiered limits; and,
• An explicit amortization limit used for qualification in B-20 that, in combination with debt service ratio restrictions, would limit excessive leverage.

GDS and TDS limits would need to be periodically reviewed to ensure they remain suitable for the risk environment. Similar to an LTI limit, OSFI believes that transitional implementation arrangements would need to accompany this measure since impacts may fall disproportionately on smaller and less complex FRFI lenders. OSFI is also exploring whether high debt service ratio lending could be addressed through the capital framework, such that more capital is set aside for high GDS and TDS loans.

3. Interest rate affordability stress tests

As outlined above, the MQR is a minimum interest rate that is applied in debt service coverage ratio calculations to test the borrower’s ability to afford higher debt payments in case of negative shocks to income, or increases in interest rates or expenses. The MQR is currently applied in the GDS calculation (i.e., mortgage-related debt service) at underwriting. OSFI maintains the MQR for uninsured mortgages, which is founded in B-20. The federal Minister of Finance maintains the MQR that applies to insured mortgages. As part of its annual review, on December 15, 2022, OSFI held the MQR steady, at the greater of the mortgage contractual rate plus 2% or 5.25%.

A minimum affordability test, while beneficial, can also reduce lenders’ incentives to establish more risk-sensitive affordability tests within their own RMUPs (i.e., tests that may be applied differently for various risk and product characteristics). For example, some loan types, such as variable rate mortgages  or mortgages with shorter terms, can carry heightened payment and renewal risk. A higher affordability test on such mortgages may, therefore, better respond to underlying risk. Conversely, loans with fixed rates and longer terms may pose less risk.

As such, OSFI is exploring design elements that could result in the adoption of more risk-sensitive tests of affordability, and could respond better to risks arising from high household indebtedness. Some of these design elements include:

• Creating an explicit principles-based expectation that lenders establish, monitor and report on different MQRs, in addition to the current MQR set in regulation, according to different risk characteristics and product types;
• An expectation that the MQR be applied to a borrower’s total debt service (i.e., for other existing mortgages and non-mortgage debt obligations), in addition to gross debt service; and, similarly,
• Tests of affordability and other debt serviceability measures for non-mortgage retail lending outside of B-20.