top of page

Recession-Proof Credit Underwriting Using AI - A Credit Union Perspective

Author: Raabiya Singh

The consecutive negative GDP growth rate in the US confirms that the country is now in a recession. It seems, however, that the economy is still sending mixed signals. Home sales are tumbling but consumers are still spending. Despite the massive tech layoffs, the unemployment rate at 3.6% is near a half-century low.

On the other side, consumers are agitated as they’re reeling from painful prices at gasoline stations, grocery stores, and auto dealerships. To maintain spending, Americans have been using the heavy savings they built up during the pandemic. Aiming to douse the inflationary flames, the Federal Reserve is raising interest rates which means that it’s increasingly expensive to buy a house, a car, or a major appliance on credit.

Historically we have seen lending standards being tightened during recessions. Lending growth slowed to zero during the 1990-91 and 2001 recessions. Loan growth at commercial banks decreased substantially and remained negative for almost four years after the 2007-08 financial crisis. 70% of lenders reported tightening of standards in 2008. Despite these measures, consumer loan delinquencies were at an all-time high of 4.8 in 2009. Choking lending has never helped in the past.

Indicators of tough times ahead?

Early indicators such as credit balances, credit cards, and unsecured balances are showing some signs of stress. There is a surge in delinquency in the 11-29 day bucket. Personal loans, which were nonexistent for almost two years are now going up. As people’s budgets tighten up, previously unutilized credit lines are now being used.

Borrowers have started to feel some pain however credit unions report that this indicator has not translated to a rise in delinquencies in a significant way yet. Though they do expect that to tick up a little bit. There is debate on whether there is a serious recession ahead, risk leaders believe that it is certainly going to feel like one.

But there is a certain segment of the population that is being impacted more than others. A recent minimum wage increase may have little impact on businesses and employees going forward due to inflation. This uneasiness is expected to continue for a long time in 2023.

The AI way - Time for credit unions to shine

Credit unions were created to help people achieve financial well-being irrespective of the economic climate and they can facilitate responsible lending even in a recession. The key to navigating a recession well is to proactively reach out to members and offer them different solutions or products. The recession may be a mild one, but credit unions should certainly not wait for things to unravel themselves and start looking at certain indicators more closely.

In terms of modifying lending strategies to prepare for the recession, risk leaders are currently lending as they have been. But they are looking at levers and keeping them ready to minimize portfolio exposure should things change. The capabilities that AI brings to early adopters have made identifying risks and opportunities easy. For credit unions, one of the best indicators to look at is member behavior. Credit unions are now accessing their core to analyze members’ transactions and see patterns otherwise unaccounted for in traditional credit scoring.

Artificial intelligence is also transforming collections by predicting the best way to reach out to a member. For example, a collections team can make multiple phone calls to a member but sticking to only one strategy can be inefficient. AI can tell you that a member is more likely to pick up their cell phone at certain times.

AI is also being used to review images of documents and extract data for various workflows and processes. Currently, members may upload a utility bill or a W2, but the staff has to manually look at it. If the document is not correct, the technology can send alerts back to the members and the underwriters. Credit unions also want to make sure that the application process is simple, and that it is easy for members to communicate with them.

Currently, credit unions when managing the portfolio look at the descriptive analysis, such as the monthly view of the portfolio. Going forward they will need to get into predictive analysis to identify the trends for potential default and then start to offer products and collections strategies that better fit the members’ current and future financial situation, all possible with AI.

Debunking AI myths

Does automation replace human jobs in underwriting?

AI is about decision support rather than the decision-making itself. There’s a little bit of fear that we get out of what AI will do to us, humans. The reality is the combination works well. Automation of credit decisions enables approval or rejection of loan applications in large volumes reducing the turnaround time to a few seconds. The complex cases are still left to the underwriters. AI simplifies jobs, creates easier workflows, and allows financial institutions to grow without growing their expenses at the same rate.

Can AI-based underwriting increase approval rates even in recessions?

This is a challenging time for financial institutions globally and for some it could be hard to see substantial jumps in approval rates. Rejections are high globally. Those who embraced AI early conducted regular tests to assess their decisions, workflows, and various factors and expressed confidence that while rejections are likely to increase they would be much higher had they not implemented AI.

Does Scienaptic’s platform make compliant and fair decisions?

Yes. The platform uses adverse action reasons, and disparate income tests to make fair, compliant, and inclusive lending decisions.

Historically, even if credit quality is challenged for a year or two, credit unions have seen lower losses. But when we come out of this recessionary period do we have the tools, the resources, the people, and the automation in place to take advantage of the new market opportunities? Lenders often fail to reopen themselves quickly enough, which can be one of their biggest mistakes. Decisioning engines can help you test and deploy your lending strategies quickly and efficiently.

Using AI can help make sharper differentiation between creditworthy borrowers and high-default-risk applicants. But risk leaders need to think about the end-to-end lifecycle of loans and use AI for predictive decision-making. A credit union has always had the advantage of knowing its members better than any other financial institution. AI can provide you more time to have the conversation, to hear the story, and to dig into things that are behind a simple credit score.



bottom of page