Can the characteristics of new mortgages predict borrowers’ financial stress? Insights from the 2014 oil price decline

Introduction

High household debt has been a vulnerability in the Canadian financial system for a long time. Given that most of the increase in household debt in Canada over time has come from mortgage debt, the Bank of Canada closely monitors the characteristics of newly issued mortgages.

Households with large mortgages relative to either their income or the value of their property may face hardship when confronted with financial disruptions, such as a drop in income or house prices. However, this hypothesis has not been verified—due in part to the limited data available but also to the rarity of episodes of financial stress in Canada.

In this note, we use a newly constructed dataset to explore the link between the characteristics of new mortgages and the likelihood that borrowers will experience financial stress. When many households simultaneously have difficulty paying their debts, this can have severe adverse effects on the financial system and the real economy. Therefore, a better understanding of the characteristics associated most frequently with financial stress is of paramount importance. It helps the Bank better focus its monitoring of the financial vulnerabilities of households.

Oil price decline as a trigger of borrowers’ financial stress

The collapse in oil prices in 2014 represents a unique opportunity to study how households kept up with their debt repayments as an adverse scenario unfolded. This event triggered an economic downturn in the energy-intensive regions of Alberta, Saskatchewan and Newfoundland and Labrador. Lower oil prices led to a decline in household income and a slump in housing markets in these regions, with house prices remaining below their peaks for years afterward.

The financial impact of the oil price decline is seen in the diminished ability of some households in energy-intensive regions to service their debt. The proportion of debtors behind on payments started to increase 2015 and remained high in later years compared with the rest of Canada (Chart 1).

Chart 1: The percentage of debtors in arrears in energy-intensive regions remains elevated after the oil price shock

Sources: TransUnion and Bank of Canada calculations

Tracking borrowers’ financial stress

The Bank typically tracks the quality of new mortgage lending using microdata from regulatory returns of federally regulated lenders. These data provide a good snapshot of borrowers’ financial position when they take on a mortgage—for instance, the size of their down payment or how large the mortgage is relative to their income.

By combining data on these new mortgages with credit bureau data, we can see whether borrowers stayed on track with the repayment schedule of their debt.1 Specifically, we analyze the repayment behaviour within the first three years of repayment of each household that obtained a mortgage in energy-intensive regions in the 2014–17 period.

We define our measure of financial stress as being behind on payments for at least 60 days in any credit category at any point within the first three years of obtaining a mortgage. These credit categories include mortgages, credit cards, loans and lines of credit. Because mortgages are typically the last product to go into arrears, missed payments on other types of debt can be early signs of financial distress.

Predicting borrowers’ financial stress

We consider mortgage characteristics that indicate the quality of both the borrower and the property at the start of the mortgage. The main indicators of interest are those the Bank tracks regularly to assess the evolution of household debt vulnerabilities:

  • the loan-to-value (LTV) ratio, a measure of the initial equity stake
  • the loan-to-income (LTI) ratio, a measure of initial affordability

We supplement the list with:

  • borrowers’ characteristics and indicators of creditworthiness—age, credit scores, the size of consumer debt payments relative to income
  • mortgage terms—interest rate, loan type, loan term, dummy for 30‑year amortization, mortgage purpose (purchase or refinance)
  • fixed-effect controls—for the lender, province and the month and year of the start of the mortgage

We conduct our analysis for insured and uninsured mortgages separately because they constitute two distinct segments of the mortgage market:

  • Insured mortgages allow households to borrow an amount of up to 95 percent of value of the property. To qualify for mortgage loan insurance, these households must satisfy explicit criteria about their ability to service their debt. Also, borrowers with insured mortgages generally lack sufficient home equity to refinance their mortgage.
  • Uninsured mortgages, with LTV ratios of up to 80 percent, are subject to somewhat more flexible approval rules depending on how much risk individual lenders are willing to take.2

Overall, 14 percent of insured and 10 percent of uninsured borrowers in energy-intensive regions experienced financial stress during the three‑year period of our study (Table 1). These borrowers tended on average to have modestly higher LTVs and lower credit scores than borrowers that did not experience financial stress, in both the insured and uninsured markets.

Table 1: Select summary statistics of new mortgages in energy‑intensive regions

Insured mortgage holders Uninsured mortgage holders
Experienced financial stress Did not experience financial stress Experienced financial stress Did not experience financial stress
Number of observations (share of insured/uninsured mortgages, %) 12,206
(14)
73,808
(86)
16,105
(10)
138,970
(90)
Average loan-to-value ratio (%) 96.3 95.8 68.2 66.4
Average loan-to-income ratio (%) 327 342 276 272
Average mortgage rate (%) 2.79 2.73 2.72 2.66
Average credit score 693 719 712 755
Average amortization (years) 25.0 25.0 27.5 27.4

Note: Financial stress refers to borrowers being behind on payments for at least 60 days in any credit category at any point within the first three years of obtaining a mortgage. For insured mortgage holders, the average loan-to-value ratio is above 95 percent because the numerator also includes the mortgage insurance premium.
Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculation

Regression results

To more rigorously assess which mortgage characteristics can best predict financial stress, we estimate a probit model. The model regresses a binary indicator of financial stress, as defined above, on various mortgage characteristics. Table 2 shows the estimation results for insured and uninsured mortgages (columns 1 and 2, respectively). Overall, the model explains well the share of mortgages in arrears.

Our findings show that the LTV ratio is the most economically significant predictor of future financial stress. For instance:

  • In the insured mortgage market, the probability of experiencing stress within the first three years of taking on a mortgage increases by 2.9 percentage points when comparing mortgages that have an LTV ratio of 95 percent with those that have an LTV ratio above 80 but below 95 percent.3
  • An uninsured mortgage borrower whose LTV ratio is between 65 and 80 percent is 1.8 percentage points more likely to experience financial stress within the first three years of taking on a mortgage than an uninsured borrower whose LTV ratio is 65 percent or less.

The importance of LTV can also be seen from the predicted three-year probabilities of financial stress obtained from our regression (Chart 2).

Table 2: Average marginal effects on the probability of financial stress

Insured mortgages Uninsured mortgages
Loan-to-value ratio: 65% or less Base group
Loan-to-value ratio: Above 65% and up to 80% 1.82***
Loan-to-value ratio: Above 80% but below 95% Base group
Loan-to-value ratio: 95% 2.89***
Loan-to-income ratio -0.07 0.13**
Consumer debt payments to income 0.22*** 0.06***
Mortgage rate 4.64*** 2.04***
Credit score -1.51*** -1.78***
Amortization = 30 years (dummy) 0.97***
Borrower’s age 0.19*** 0.07***
Refinance dummy   4.30***
Mortgage rate type and term Yes Yes
Year and month of the start of the mortgage Yes Yes
Lender Yes Yes
Province Yes Yes
Number of observations 86,014 155,075

Note: * p-value below 0.10; ** p-value below 0.05; *** p-value below 0.01
Numbers in the table are multiplied by 100 to produce a change in the dependent variable in percentage points. Financial stress refers to borrowers being behind on payments for at least 60 days in any credit category at any point within the first three years of obtaining a mortgage.
Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculations

Chart 2: A higher loan-to-value ratio is associated with a greater likelihood of future financial stress

Note: Financial stress refers to borrowers being behind on payments for at least 60 days in any credit category at any point within the first three years of obtaining a mortgage.
Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculations

Access to home equity allows borrowers to more easily smooth income disruptions, either through refinancing their mortgage or using a home equity line of credit (HELOC). Many borrowers in our sample did not have a sufficiently large down payment to access home equity when they first took out the mortgage.4 Declining house prices later slowed down equity buildup even as borrowers paid down their balances.

Turning to our measure of initial affordability, the LTI ratio, we find that it has a statistically significant effect on the likelihood of financial stress for uninsured mortgages: a 100 percentage point increment in the LTI ratio is associated with a 0.13 percentage point increase in the likelihood that the borrower will experience financial stress.5, 6

In contrast, the LTI ratio appears to be insignificant in the insured market. A potential explanation for this finding relates to the strong effect of the LTV ratio: even though many uninsured borrowers have enough equity to get themselves out of financial trouble by refinancing, those with high LTI ratios are less likely to meet the debt servicing criteria needed to qualify for a new mortgage or a HELOC. Most insured borrowers, in contrast, do not have any equity to access at the time of stress, which makes their income ratios less relevant.

Other variables yield expected results:

  • Higher credit scores, a proxy for the creditworthiness of a borrower, are associated with lower financial stress.
  • Refinances appear to be riskier than purchases, but this result could also indicate pre-existing financial stress before refinancing.
  • A higher ratio of consumer debt payments to income at the start of the mortgage is associated with potential repayment issues.
  • A longer amortization period (30 years) is associated with a higher probability of stress. Longer amortization can be a symptom of borrowers stretching themselves to meet debt obligations or to qualify for a mortgage. Even though a longer loan gives borrowers more payment flexibility in the short term, especially those with income constraints, it eliminates the flexibility to extend amortization if they face an adverse income shock.

Conclusion

Our study finds that mortgage borrowers more likely to experience financial stress include:

  • those with limited equity—especially those with down payments of 5 percent
  • those with a combination of at least 20 percent equity and large loans relative to income

These findings give us a fuller understanding of the extent to which households can be financially vulnerable to economic shocks.

Given the unique nature of the stress episode studied, however, we caution against applying these results to other episodes of economic disruptions. Bank staff will continue to study the links between debt repayment behaviour and the quality of new mortgages to assess evolving vulnerabilities arising from household indebtedness.

  1. 1. To protect the privacy of Canadians, TransUnion provided no personal information to the Bank. The TransUnion dataset was “anonymized,” meaning that it does not include information that identifies individual Canadians, such as names, social insurance numbers or addresses. In addition, the dataset has a panel structure, which uses fictitious account and consumer numbers assigned by TransUnion.[]
  2. 2. Federally regulated lenders must operate within principle-based approval guidelines established by the Office of the Superintendent of Financial Institutions.[]
  3. 3. Because the distribution of the LTV ratio clusters at 65, 80 and 95 percent, we split the LTV variable into four bins: 65 percent or less; above 65 percent and up to 80 percent; above 80 percent but below 95 percent; and 95 percent.[]
  4. 4. Even if a borrower has sufficient equity to refinance, refinancing may not be an option because it is costly and requires a borrower to meet certain criteria to qualify.[]
  5. 5. Our specification also includes a square term for the LTI ratio because the stress probability increases at a faster rate the higher the LTI ratio becomes.[]
  6. 6. Alternative measures of affordability, such as the debt service ratio (DSR), yield similar results. This is not surprising given the large overlap between households with high LTIs and those with high DSRs. Over time, however, the LTI is likely to be a better measure of vulnerabilities throughout the business cycle because it is not directly affected by changes in interest rates. Yet a decline in the equilibrium mortgage rate could lead to households being able to sustain higher LTIs without increasing their likelihood of future financial stress.[]

Disclaimer

Bank of Canada staff analytical notes are short articles that focus on topical issues relevant to the current economic and financial context, produced independently from the Bank’s Governing Council. This work may support or challenge prevailing policy orthodoxy. Therefore, the views expressed in this note are solely those of the authors and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.