While many states have implemented shelter-in-place orders during the Covid-19 pandemic, some consumers have made changes to their personal finances. The changes might be temporary, but they have impacted consumer credit card balances. In this post, we examine that effect, which reflects changes in spending patterns and economic activity.
A Break with Traditional Spending Patterns
Balances exhibit a predictable seasonal pattern (Figure 1). We see expected spikes in balances at Christmas each year, then there are reductions in balance over the next three months. Finally, balances gradually build during the summer and fall months, until the holiday season repeats.
We start by looking at the trend in balances over the last three years. Every state has similar trends with different off-sets. Beginning in April 2020, non-typical values of average credit card balances are observed across all 50 states. Balances decreased significantly that month, whereas April is traditionally the first month in a sequence of monthly balance increases. As of September 2020, average credit card balances are still decreasing, though the trend has begun to level off.
Figure 1. Average credit card balance over time (Month/Year).
Average credit card balances by state were analyzed in March and September of 2020 to determine if lockdown orders have an effect on balances. March was the last month typical balances were observed, and September was the most recent comprehensive data available at the time of analysis. Historically, credit card balance is higher in September compared to March. However, we see consumers reducing credit card balances across all states this year (Figure 2).
This decrease in balance could be caused by any combination of economic uncertainty, loss of employment, reduced consumer spending and loan accommodations.
All states have experienced the same reduction in balances. Many states have reopened significant portions of their economies, with corresponding reductions in unemployment rates observed (Figure 3).
Figure 2. September 2020 versus March 2020 balance. The blue line corresponds to a 1-to-1 reference line.
Figure 3. Cumulative death per 100k versus unemployment rate for the month of August.
Three Lockdown Categories Yield Different Results
States were classified into one of three lockdown categories, according to the National Academy of State Health Policy*:
- No lockdown (six states)
- Short lockdown (ended by May 14; 20 states)
- Long or ongoing lockdown (ended May 15 or after; 24 states)
Using the 2020 data, the percent change in credit card balance (September – March)/March, indicated statistically significant difference among the three lockdown groups**. States with long or ongoing lockdown have reduced balances the most and states with no lockdowns the least (Table 1).
Table 1. Percent change (September – March)/March in 2020 credit card balances across lockdown types.
These trends can be visualized even further by analyzing county level trends across the U.S. The animation in Figure 4 shows a heat map of monthly balances for every U.S. county. Prior to April 2020, balances exhibit the expected trends of seasonal Christmas spikes, followed by balance reductions through March each year, and then followed by gradual increases throughout the remainder of the year. Beginning in April 2020, significant reductions in balances are observed for every county.
To learn more, view our resources for businesses impacted by COVID-19.
ANOVA on the percent change reduction in credit balances (Sep Bal – Mar Bal) / (Mar Bal) * 100. There is a significant difference in the percent change for each of the 3 lockdown levels (F statistic = 15.78, p-value < 0.0001). The average percent difference reduction for No Lockdown was 6.98%, Short Lockdown was 9.24%, and Long Lockdown 10.69%. Using Tukey-Kramer adjustments to analyze the pairwise differences shows that the most significant difference occurred between No Lockdown states versus Long Lockdown states (Adjusted p-value < 0.0001). Long Lockdown versus Short Lockdown (Adjusted p-value = 0.0075) and Short Lockdown versus No Lockdown (Adjusted p-value = 0.0066) were also statistically significant. The ANOVA analysis supports the hypothesis that the length of stay-at-home lockdown orders affect consumer credit card balances.
The post How COVID-19 Lockdowns Have Impacted Credit Card Debt appeared first on Equifax Insights Blog.