What Can Natural Disasters Tell Us About a COVID-19 Economic Recovery?
The Past as a Window to the Future
In my last post, we spent time exploring methods to mitigate the loss of predictive power for risk scores in the era of COVID-19. This week I’d like to pivot and examine the past. More specifically I’d like to see if we can find historical events with parallels to today that possibly open a window into the future.
Rather than a slower paced recession, the impact of COVID-19 has been sudden and severe. As such, many economists find it far more similar to natural disasters like hurricanes and wildfires than a typical recession. When natural disasters strike, they disrupt business operations, interfere with supply chains and impact consumer finances. In response to natural disasters, we see high rates of consumer loan accommodations, deferrals and forbearance. This is much like we are seeing today.
For example, in the New York area 18% of consumers were granted loan accommodations as a result of Hurricane Sandy. In Northern California the 2018 wildfires resulted in 23% of consumers accepting loan accommodations. And in Houston, Hurricane Harvey resulted in 37% of consumers in the metro area reporting loan accommodations.
Similarly, as a result of COVID-19, we’re finding more than 15% of U.S. consumers in the credit file report having at least one loan accommodation. Excluding student loans, more than $1.1 trillion in loan balances are being reported as having accommodations. Given the sharp increase in loan accommodations, I think the data is telling us there are commonalities between COVID-19 and natural disasters.
WHAT SHOULD WE EXPECT?
The question then becomes, “So what can we expect to happen going forward?”
When we analyzed the Hurricane Harvey data, we found loan accommodations resulted in consumer delinquencies being far lower than expected. This occurred for almost six months after the initial strike. However, one year after Harvey, delinquency rates spiked and were significantly higher than expected.
When we shift our analysis to current trends, we’re finding a similar scenario might be underway. Like Harvey, delinquency rates for consumers with loan accommodations are currently lower than consumers without them. This is despite the fact consumers with accommodations have credit scores 50 points lower, on average, than consumers without accommodations. If Harvey is indicative of the COVID-19 road ahead, we may see delinquencies for consumers with accomodations spike in the next three to six months.
WHAT ACTION CAN LENDERS TAKE?
So if we can expect to see a surge in delinquencies, what can lenders do to reduce their exposure? First, we suggest lenders verify income and employment for new and existing customers. Analysis of Hurricane Harvey showed us that consumers with prolonged job losses experienced delinquency rates over 70% higher than consumers with stable jobs and steady pay. Similarly consumers with prolonged, significant pay decreases resulted in delinquency rates that were 50% higher than those with stable jobs and pay. Even less severe, shorter-term pay decreases or job losses were found to have delinquency rates 30% higher than consumers with stable jobs and pay.
Cumulative Delinquency Rates by The Work Number® Category
Let’s go deeper in our analysis. When controlling for risk score, we found that job losses and pay reductions had significant impacts on delinquency rates. This occurred across the spectrum of risk scores, from prime to subprime.
Those with prolonged job losses had delinquency rates 55% higher than prime consumers with stable jobs and pay. Prime consumers with prolonged, significant pay decreases had delinquency rates 50% higher than prime consumers with stable jobs and pay.
Near prime consumers
Near prime consumers with prolonged job losses had delinquency rates 30% higher than near prime consumers with stable jobs and pay, and near prime consumers with prolonged, significant pay decreases had delinquency rates 20% higher than near prime consumers with stable jobs and pay.
Those with prolonged job losses had delinquency rates 20% higher than subprime consumers with stable jobs and pay, and subprime consumers with prolonged, significant pay decreases had delinquency rates 10% higher than subprime consumers with stable jobs and pay.
24 Month Delinquency Rates by The Work Number Category
The Hurricane Harvey analysis also revealed that income verification provides much greater insight after natural disasters when loan accommodations are prevalent. Annualized income is always powerful when rank ordering consumer risk levels. But after Harvey, we discovered annualized income did a much better job of rank ordering risk within credit score bands than it did in other similar geographies not impacted by the natural disaster.
Lesson Learned from Canadian Wildfire
In addition to employment and income data, I’d like to highlight an additional data asset that proved quite predictive in a past natural disaster. In May 2016, a wildfire struck Fort McMurray in Alberta, Canada. With damages exceeding $10 billion, it was the costliest disaster in Canadian history.
Like Harvey, it struck suddenly and had immediate economic impact. Sixty-five thousand people had to evacuate and over 80% experienced job disruption. Some accepted payment deferrals to manage the impact but many of those who didn’t suffered from significant score decline.
To help lenders navigate which consumers were more likely to recover quickly and which ones would need more time, trended data attributes and scores using trended attributes proved to be the most predictive. Even better, they demonstrated the most value within the first six months after the wildfire. Additionally, the team in Canada added wealth data to the equation. It found it also helped to separate individuals who might recover more quickly from those who would take longer. In short, when you go beyond credit data alone and explore alternative data sources, you’re likely to discover insights and patterns previously undetected.
If we let the data tell the story, I think it’s safe to say the economic impact being felt from COVID-19, while unprecedented, does have parallels to the past, specifically in natural disasters. As a result we may find recovery takes a similar path. While we can never completely remove uncertainty, the good news is that there are tools to navigate this uncertain time. When using alternative data like income, employment, trended data, and wealth we’re better equipped to help consumers and predict risk.
Don’t miss the other posts in my series:
Recommended For You
Buying a home can be stressful and confusing — especially if a consumer is trying to learn about credit and […]
The rate of change in the auto industry is accelerating. And it’s not powered by just one factor; it’s a […]
This is a recap of our October 7 Market Pulse webinar, where we discussed the importance of credit unions in […]
Companies need data science to turn their data into insights that drive better business decisions. But how do they hire […]