Is Artificial Intelligence The New Debt Collector?
Debt collection may be among the more human and manual labor-intense activities when it comes to managing accounts receivables. Collections departments place calls, scores of them, send emails, and seek to work out payment plans — and very frequently none of the above translates into recovery of monies owed.
In fact, that happens only 20 percent of the time, at best. That’s in part tied to legislation that is decades old, such as was passed in 1991, which allows consumers to tell collectors to stop contacting them, and to the traditional methods of collection that are based on phone calls and letters and repetition.
The mismatch between efforts and success speaks to a looming — and significant — issue for corporations of all sizes and across all manner of verticals, but especially for financial institutions trying to manage credit card and student loan debt.
Consider the fact that earlier this month, data from the Federal Reserve Bank showed that the total debt carried by Americans now stands at about $13.5 trillion. That’s a record high, as noted by CNBC, and a trillion dollars higher than had been seen in the previous peak prior to the recession. Getting a bit more granular, the website noted that serious delinquencies, which includes debt that is past due 90 days or more, for student loans, rose to 9.1 percent from 8.6 percent in the previous quarter. Delinquency flows have been rising on auto-related debt through the past six years, and on credit card debt through the past year.
Consider too, that debt collection efforts are anything but well-received by consumers. A debt collection snapshot from the Consumer Financial Protection Bureau found that between July 2011 and May 2018, the Bureau received more than 400,500 debt collection complaints. That tally represented 27 percent of the total complaints received.
Want further proof positive that things could use a change — perhaps a bit of a tech-driven helping hand?
In an interview with PYMNTS’ Karen Webster, Dr. Akli Adjaoute, CEO of Brighterion, a Mastercard company, brought forth some salient points from the Fed that underscore the magnitude of delinquent debt. Of the $598 billion in debt that is delinquent, $403 billion is “seriously” delinquent, according to August stats, meaning at least 90 days past due.
He noted that 30 million Americans have at least one debt in collection. This translates into, as he termed it, “a big problem for the United States, for the economy and obviously for the financial institutions that are owed all of this money, and that not all the people will be able to pay.”
It’s an area in which “true AI” — unsupervised learning and smart agents that can profile behavior — can help, but not many financial institutions (FIs) are taking full advantage of. The AI Gap Report, a collaborative effort between PYMNTS and Brighterion, reports that only 5.5 percent of all FIs are using any kind of sophisticated true artificial intelligence (AI) technology to help improve their efforts.
True AI, Adjaoute said, can helps FIs take action at the first sign of trouble, through personalization and outreach. AI, he said, increases the odds that collection efforts will be successful, because trying to collect from someone who has hit some personal speed bumps but nonetheless has proven a good credit risk is a better strategy than trying to collect from those who are chronic delinquents. Separating and segmenting those borrowers can be done all the more effectively with true AI.
In fact, Adjaoute said, with AI, there would be no such thing as a chronic delinquent.
AI, said Adjaoute, can be used to examine past consumer behavior, monitor present consumer behavior and can help with onboarding efforts, especially in an age where competition among FIs to bring consumers onboard with speed is the name of the game. AI, in effect, can be helpful in raising red flags or green lights when it comes to onboarding consumers in the first place, and where credit is first extended. You can tell a lot about a consumer’s likelihood of paying down debt with insight as to how often they visit the ATM or bump up against account overdraft limits.
“It’s all about the monitoring of activities,” he said.
But after the credit is extended, should missed payments start to surface, the insight that machine learning offers, in real time, lets lenders tap into personalized methods of contact. “It’s a lot easier in the beginning [of collection efforts] to get back the money.”
Adjaoute offered the scenario of, say, text messages for Millennials versus phone calls — that can open the door to tailored conversations and repayment terms leading to markedly improved recovery rates. In fact, said Adjaoute, effective use of AI can help reduce delinquency rates by about 76 percent.
Prevention is the Best Cure
It is, of course, best to stop the problems of debt delinquency before they ever begin in earnest, said Adjaoute.
“I believe the prevention of delinquency is more important,” he told Webster, than efforts to eke out returns on processes that are still steeped in telephone calls and letters. He pointed to artificial intelligence as a conduit toward a successful, preventative mindset.
“There are a lot of things technology can bring to actually help companies,” he said. It’s a continuance of a theme in this space, where personalization of financial services can help cement relationships between firms and consumers and also optimize business efforts.
It is true AI, Adjaoute said, that can provide reliable information about who is in fact most likely to pay, whether nonpayment behavior is in fact an aberration, and when the best times and methods of contact might be. Context is key, he said, as is forming a 360-degree view of the borrower that could not be done without the aid of such methods and where personalization can boost recovery rates.
“It is important to remember that we are human … and that we do not use the same script or the same language,” when it comes to debt, he said. “We all react differently.”