The reliance of Canadians on credit card debt as a predictor of financial stress

Context and motivation

In this age of online shopping, having a credit card is nearly a must. In fact, in 2019, 90% of Canadian adults had at least one credit card.1 Roughly half of Canadian credit card users pay off their balance every month, while the other half pay only a portion of their balance and thus pay interest and carry over the remaining debt to the next month.2

Those who carry over debt could be the more concerning group for financial stability because an unpaid credit card balance is associated with a greater chance of being in arrears on any credit product. Unpaid balances also provide an early signal of rising household financial stress.

I assess household financial stress by looking at credit card data from TransUnion, one of two credit bureaus in Canada, to isolate those households who pay their credit card balance in full from those who carry a balance from one month to the next. I then analyze whether those carrying a credit card balance are more likely to miss debt payments in subsequent months.

I also examine how various characteristics associated with carrying a balance influence the likelihood of experiencing financial stress down the road. These characteristics include occurrence, duration and intensity of use. In this note, I define financial stress as falling behind on payments for at least 60 days on any credit product.

Keeping this context in mind, I show in my analysis that:

• In any given month, close to half of Canadians with a credit card carry a balance for at least two consecutive months.
• Canadians who carry a credit card balance will more likely experience financial stress within the next six months than those who pay off their credit card balances in full each month.
  • All else being equal, mortgagors who carry a balance on their credit card are more than twice as likely to fall into arrears on their debt within six months than those who pay their balance off in full.
  • However, non-mortgagors carrying a balance are five times more likely to fall into arrears within that time than those who pay in full.
• Canadians carrying a credit card balance for more than six consecutive months or those who owe the maximum balance on their credit cards become increasingly likely to experience financial stress in the near term. Here again, the probability of falling into arrears by missing a future debt payment is greater for non-mortgagors than it is for those with mortgages.

The dataset

To analyze how Canadians’ reliance on credit card debt relates to future financial stress, I use microdata from TransUnion. These data contain the anonymized credit history—including that of credit cards—for over 30 million Canadian residents.3 My sample ranges from January 2016 to February 2024 and covers nearly 58 million active credit card accounts.

For Canadians who have a mortgage from one of Canada’s federally regulated lenders, I also examine the influence of a broad set of household characteristics on their use of credit cards.4 For instance, I analyze whether factors such as income or the size of a mortgage in relation to the market value of the property—both of which are captured when the mortgage is first issued—affect the mortgagor’s likelihood of experiencing financial stress.

In this note, credit card holders are carrying an outstanding balance if they do so for at least two consecutive months on at least one of their credit cards. Table 1 shows various descriptive statistics, both for Canadians who pay their credit card balance in full each month and for those who do not.

Table 1: Descriptive statistics (for 2023)

Note: To protect the privacy of Canadians, TransUnion did not provide any personal information to the Bank. The TransUnion dataset was anonymized, meaning it does not include information that identifies individual Canadians, such as names, social insurance numbers or addresses.
Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculations

In any given month since January 2016, about 46% of Canadian credit card holders on average have carried a balance on their credit cards for at least two consecutive months (Chart 1).5 This share has fluctuated between 42% and 49%; after falling during the early stages of the COVID-19 pandemic, it has been rising since mid-2022 but remains below pre-pandemic levels. Among Canadians who carry a balance, those carrying a significant share (defined as unpaid balances representing 80% or more of the borrowing limit on their credit cards) has been trending up since mid-2020 and is now hovering near its pre-pandemic peak.

Chart 1: The reliance of Canadians on credit card debt has risen further over the past year

I also find that the share of mortgagors carrying an outstanding balance (53%) is higher than that of non-mortgagors (43%). However, as presented in Appendix 1:

• for mortgagors, 14% of those carrying a balance are borrowing 80% or more of their total credit card limit
• for non-mortgagors, 23% of those carrying a balance are borrowing 80% or more of their total limit

The empirical exercise

To investigate the relationship between the reliance on credit card debt and future financial stress, I estimate a set of equations using the large micro dataset. While the exact equations are found in Appendix 2, the intuition behind the empirical exercise is as follows.

I want to assess the marginal impact of credit card holders carrying a balance for at least two consecutive months on their probability of experiencing financial stress within the next six months, all else being equal. I control for various characteristics of the borrowers, such as income and loan-to-value ratios. I also estimate separate equations for Canadians with and without a mortgage. In all my estimations, the control group is Canadians who pay their credit card balance in full each month.

The empirical exercise focuses on the following three dimensions of Canadians’ reliance on credit card debt:

• Occurrence—Does a credit card holder carry an outstanding balance? The fact that some Canadians do not pay their balance in full each month may be a sign that they are financially stretched. As a result, they may have less financial flexibility to absorb a reduction of their income—even a temporary one—or an unplanned expense, such as a car repair bill.
• Duration—If a credit card holder has been carrying a balance, for how many consecutive months have they done so? Borrowers who carry over an unpaid balance month after month could be using their credit card to cover additional expenses due to cash flow issues or loss of income. Therefore, they may be more likely to miss a debt payment in the future.
• Intensity of use—If a credit card holder has been carrying an outstanding balance, what percentage of the borrowing limit on their credit cards have they been using? Although credit card debt may help temporarily bridge a period of budgetary pressure, an extreme case in which a borrower reaches their credit card limit leaves them with little financial flexibility for any unexpected financial shock.

The findings

In any given month, some Canadians who have paid their credit card balance in full in the prior six months fall into arrears on some of their debt products. During my sample period, this share was 1.3% for mortgagors and 1.7% for non-mortgagors.

In the results that follow, I focus on the estimated marginal impact for each of the three dimensions on the probability of experiencing financial stress at some point within the next six months, relative to that of the control group. This increase in probability assumes all other characteristics of the credit card holders stay the same as in the control group.

I control for one dimension at a time to avoid any collinearity issue. Each dimension is individually estimated in a model.

Occurrence of carrying a credit card balance

For credit card holders without a mortgage, carrying a balance increases their likelihood of financial stress over the next six months by 5.3 percentage points relative to that of those who pay their balance in full (Chart 2, yellow bars on the left side). Therefore, all else being equal, non-mortgagors who do not pay their credit cards in full are five times more likely to fall behind on debt payments than those who do.

Chart 2: Carrying an unpaid credit card balance increases the likelihood of experiencing financial stress in the future

While mortgagors carrying a balance are also more likely to miss a debt payment than those who are not carrying a balance, the impact is smaller, at 2.7 percentage points. Mortgagors carrying a balance are therefore only 2.5 times as likely to experience financial stress as those who pay their balance in full.

Duration of carrying a credit card balance

I find that carrying a balance for more than six consecutive months significantly increases the likelihood of missing a future debt payment (Chart 3). The impact is significantly more pronounced for credit card holders who do not have a mortgage. For instance, non-mortgagors who have been carrying a balance on at least one of their credit cards for 7 to 12 consecutive months are seven times more likely to experience financial stress than those paying in full each month. This scaling factor is more than two for mortgagors.

After 12 months, the likelihood of falling into arrears tends to rise but at a diminishing rate, peaking after two to three years.

Chart 3: Canadians carrying a balance on their credit cards for more than six consecutive months may face more future financial stress

Intensity of use of credit cards while carrying an unpaid balance

I find that the likelihood of both mortgagors and non- mortgagors missing future debt payments tends to rise at an increasing rate as borrowers carry higher shares of their credit limit as outstanding balances (Chart 4). Similar to results for the dimensions of occurrence and duration, this risk is more pronounced for borrowers without a mortgage than for those with one.

In particular, I find the likelihood of future financial stress increases when credit card holders borrow more than 80% of their credit limit. Compared with borrowers who pay their balances each month, Canadians who carry 80% of their credit limit are:

• more than seven times more likely to miss a future debt payment for mortgagors
• almost nine times more likely to miss a future debt payment for non-mortgagors

This risk is even more pronounced for those who carry balances greater than 90% of their limits.

Chart 4: The size of unpaid balances on credit cards affects the likelihood of Canadians experiencing financial stress in the future

Conclusion

The Bank of Canada regularly monitors the financial health of households. Their financial stress, if acute or widespread, can lead to significant credit losses for lenders and affect financial stability.

The share of borrowers carrying credit card balances has been rising since 2022, although it remains below pre-pandemic levels. If this upward trend were to persist, it could signal that increases in rates of financial stress are on the way: I find compelling evidence that outstanding credit card balances have an important and statistically significant relationship with the chance of missing future debt payments. My results also highlight that both the duration of carrying a balance and the share of credit card limits being carried forward are important early warning signs of borrower stress.

The Bank will continue to monitor these dimensions closely as part of its assessment of Canadians’ financial health.

References

Boutros, M. and A. Mijakovic. 2024. “The Macroeconomic Implications of Coholding.” Bank of Canada Staff Working Paper No. 2024-16.

Grodzicki, D. and S. Kulaev. 2019. “Data Point: Credit Card Revolvers.” Consumer Financial Protection Bureau’s Office of Research, July 2: 8–10.

Khan, M. and Y. Xu. 2022. “Housing demand in Canada: A novel approach to classifying mortgaged homebuyers.” Bank of Canada Staff Analytical Note No. 2022-1.

Timoneda, J. 2021. “Estimating Group Fixed Effects in Panel Data with a Binary Dependent Variable: How the LPM Outperforms Logistic Regression in Rare Events Data.” Social Science Research 93.

Tompkins, M. and V. Galociova. 2019. “Canadian Payment Methods and Trends: 2019.” Payments Canada discussion paper.

Appendix 1: Additional charts by mortgage-holding status

Chart A-1A: Mortgagors tend to carry an outstanding balance more often than non-mortgagors

Chart A-1B: Non-mortgagors tend to carry a bigger share than mortgagors of their borrowing limit

Appendix 2: Specification of the empirical exercise

I use two separate panel regression specifications for mortgagors and non-mortgagors. This allows me to include additional control variables for those with a mortgage, while for non-mortgagors, I use a simpler model:

\(\displaystyle\, Y_{T+6,it} \) \(\displaystyle=\, \theta CreditBal_{it} \) \(\displaystyle+\, T_{t} \) \(\displaystyle+\, \epsilon_{it} \) \(\displaystyle,\)

where

• The outcome variable \(\displaystyle\, Y_{T+6,it} \) is binary and indicates whether borrower\(\displaystyle\, i \) is at least 60 days late on a payment on any other credit product they hold, at any time over the subsequent six months
• \(\displaystyle\, CreditBal_{it} \) denotes the status of borrower\(\displaystyle\, i \) in month\(\displaystyle\, t \) with respect to carrying some outstanding balance on their credit cards
• \(\displaystyle\, T_{t} \) represents time-fixed effects, which help control for time trends and seasonal patterns

For mortgagors, I include three additional control variables:

\(\displaystyle\, Y_{T+6,it} \) \(\displaystyle=\, \theta CreditBal_{it} \) \(\displaystyle+\, D_{vy} \) \(\displaystyle+\, C_{c} \) \(\displaystyle+\, T_{t} \) \(\displaystyle+\, \epsilon_{it} \) \(\displaystyle,\)

where:6

• \(\displaystyle\, D_{vy} \) is a categorical variable grouping mortgagors by their loan-to-value (LTV) ratio and income at the time their mortgage is originated7
• \(\displaystyle\, C_{c} \) denotes fixed effects for the year in which a borrower’s mortgage was originated

I estimate the model using ordinary least squares. The literature finds that simple linear probability models with fixed effects can be superior to nonlinear models (such as logit or probit) for modelling rare events in a panel dataset, such as delinquencies.8

For each of the two specifications, I test three dimensions of credit card debt-carrying behaviour to examine their relationship with future delinquencies:

• a binary variable that indicates whether a credit card holder is currently carrying a balance that spans at least two consecutive months
• assuming a credit card holder has been carrying a credit card balance, the number of consecutive months this has been the case9
• assuming a credit card holder has been carrying a balance, the share of the total borrowing limit on credit cards currently being carried forward10

For mortgagors who carry a balance on their credit cards, I apply additional control variables related to LTV ratios and income. These variables show that borrowers with higher incomes and lower LTVs tend to have a lower probability than borrowers with less gross income and those with high LTVs of missing debt payments over the next six months (Chart A-2). This is particularly true for those with an LTV below 65%.

Chart A-2: Among mortgagors with unpaid credit card balances, those with a higher LTV and lower income are more likely to experience financial stress in the future

Endnotes