If you are anything like us, you have been surprised to witness delinquencies remain relatively low despite the pandemic. While this is good news, if you are anything like us, you are also asking yourself: will things change, and if so, when?
In his September 9th post, Scorenomics’ Chief Data Scientist Hutch Carpenter discussed some of the factors keeping delinquencies at bay. The availability of payment deferrals, the infusion of government cash, naturally suppressed spending, and consumer’s smart choices have kept delinquencies from creeping up in the past few months.
However, in the past few weeks, TransUnion has reported slight increases in delinquencies in two sensitive categories: auto and mortgage—the kinds of loans consumers are most likely to protect during a financial downturn. As seen in the table below, the number of car loans delinquent 30+ days has increased by 88 basis points between July and August.
While both Experian and Transunion reported similar increases, analysts make limited predictions about where delinquencies will head in coming months, and most importantly, if we’ll witness a tsunami. That’s because in addition to seeing increases, both Transunion and Experian have reported substantial decreases in the percentage of accounts in hardship programs, also seen in the table above for auto loans. Matt Komos, VP of Research and Consulting at TU, notes that the recent increases in delinquency might reflect sloppy bill paying routines after the end of forbearances.
New RC Data
A few weeks ago, our product team introduced a simple tool, available at no cost to financial institutions, that provides consumers with tailored advice for managing their finances during the pandemic. The COVID-19 Financial Health Check, like other Scorenomics products, gives consumers meaningful advice and reveals previously unknown data about consumer’s behavior, expectations, and attitudes.
One of the questions posed through the tool asks users to predict when they’ll run out of money. More specifically, the question asks: Some sources of support, such as payment assistance programs and government assistance, are due to end soon. If you received no further support from financial institutions or the government, when would you have trouble paying your bills?
As seen in the table below, 46% of respondents have thus far reported they’ll have trouble paying bills in October, November, December, or January.
These results should be tempered by the observation that people are inherently myopic about the long-term. This is to say, perhaps the relatively small distribution of people reporting running out of cash in February or March is due to people’s inability to make predictions that far out.
How much should we trust these results?
To check whether whether I could trust their validity, I stress tested these results against a different question. Within the COVID Financial Health Check, we also asked consumers how much of their government stimulus they’ve been able to save. I would feel more confident about the above results, I thought, if people going into delinquency soon also reported comparatively lower savings.
I ran a regression between the month consumers expect to go delinquent, and how much they have saved during the pandemic. Surprisingly, the correlation between savings and expected delinquency was minimal and my R squared unimpressive (.0051). This is reflected in surprisingly uniform saving averages (~$4000) regardless of consumers’ estimate of when they will run out of money. Strange!
I turned instead to another (trusted) question in our Financial Health Check, drawn from our BackOnTrack solution, that asks consumers to report: if you had to find $2,000 in the next month, what’s the main way you would come up with the money? Hutch wrote about this question in his July 22nd blog post. This question is helpful, because the way people answer has been externally proven to correlate tightly with financial fragility. The question can reveal an individual’s true financial vulnerability without asking them whether they are financially stable. As Hutch described when used in collections the responses can also indicate whether a delinquency is more likely to be serious or passing.
In the table below, sources of $2000 are prioritized from those that indicate financial security to those that indicate the greatest vulnerability. The averages on the right represent how many months the user has left before they go delinquent.
By the looks of it, consumers who expect they will go delinquent sooner are also more likely to demonstrate financial vulnerability (p value = 3.69E-07). I am thus more inclined to believe consumers may run out of money in the coming months since we also know that these consumer’s financial fragility is proven to be relatively high.
Of course, while these consumer responses are a helpful data point, the tsunami may look a number of different ways. It will be mediated by the government’s ability to provide additional help, whether COVID cases continue to rise during second waves, among many other factors. It still aligns with a slew of other analysts who, like me, expect that the exhaustion of government aid will inevitably mean an increase in delinquencies.
It’s interesting to note that both Experian and Transunion refrain from making predictions like the ones above. I wonder where this leaves financial institutions. Are they able to divine when the tsunami will hit? I imagine banks rely heavily on account activity, balances, and past delinquencies to make these predictions. I wonder how much more exact predictions might be when we solicit information like the $2,000 question as part of the collections or forbearance process. Certainly, the high proportion of consumers opting into payment arrangements posed a tremendous opportunity to solicit new data to help us better predict what will happen in coming months. My sense is few institutions had the digital tools to take advantage of that opportunity.
BackOnTrack, we hope, is a tool that will help financial institutions get better at this. Deployable digitally and without requiring any PII, banks can use the configurable tool to make the collections process more informative for both the consumer and their internal teams. Hit us up if you’d like to learn more!
Signup to be notified when we’ve posted a new article to the Scorenomics Blog!