By David LaRoche
I’m getting calls these days from consumer lenders concerned about surging delinquencies and struggles with traditional collection methods. The conversation normally pivots we can help them leverage Machine Learning Models (ML) and Robotic Process Automation (RPA) to tackle concerns over 90+ day delinquencies.
For context, Machine Learning Models spot patterns in vast amounts of data, while RPA handles repetitive tasks with precision and speed. RPA is being used for tasks such as inbound collections calls, while the benefit of Machine Learning Models is that they’re algorithm-based and learn over time but let you define the response.
Together, these technologies are redefining how financial institutions approach debt recovery.
The Challenge: A Perfect Storm of Delinquencies
As I write this, big banks are reporting 2Q 2024 earnings.
JPMorgan Chase reported 30-day delinquency rates on card loans at just under 2.1%, higher than the 1.7% seen in the same period last year but a bit lower than the 2.2% seen in the 1Q. Charge-offs stood at 3.5% in the 2Q, up from 3.3% in the 1Q and 2.4% fa year ago, And Chase’s 90-day delinquency rate was 1.1%, up from 0.8% a year ago but down a bit from the first-quarter’s 1.2%.
Wells Fargo’s 30+ day delinquency rate for the card segment was 2.7%, compared with 2.3% a year ago and down from 2.9% in the first quarter. The bank attributes the improvement to credit tightening.
Citigroup saw similar trends, although its delinquencies were significantly lower than Wells. The bank said lower FICO-band customers have sharply reduced their payment rates and are borrowing more as they are more acutely impacted by high inflation and interest rates. For additional context, 86% of Citi’s card loans are to customers with 660+ FICO scores.
On top of that. PYMNTS Intelligence reports that 65% of the U.S. population lives paycheck to paycheck, the highest share it has seen in two years. As consumers increasingly lean on credit to bridge financial gaps, traditional collection strategies are struggling to keep pace.
We all know the larger banks have the advantage of proprietary data and analytic platforms that help them optimize their collections efforts and address the volatility within FICO bands. Smaller financial institutions and fintechs rarely have those tools.
What’s you need is an approach that anticipates risks, adapts to individual circumstances, and navigates the ever-changing regulatory landscape. Enter GOBLIN, Predictive Analytics Group’s enterprise data platform.
The Game-Changers: Machine Learning and RPA in Collections
By leveraging vast amounts of data and advanced algorithms, ML and RPA are consistently outperforming traditional methods in reducing 90+ day delinquencies. Here’s how:
- Early Warning Systems: GOBLIN’s ML models analyze hundreds of variables to identify at-risk accounts long before they hit the critical 90-day mark. It’s like having a financial weather forecast, offering suggestions for proactive intervention.
- Hyper-Personalized Engagement: We’re leaving the days of one-size-fits-all collection notices. ML tailors communication strategies to individual borrowers, considering factors like preferred contact methods and optimal timing. RPA then executes these strategies with precision, significantly improving engagement rates.
- Adaptive Repayment Strategies: By continually analyzing borrower behavior and financial situations, systems like GOBLIN can create and adjust repayment plans in real time, increasing success rates.
- Regulatory Compliance Assurance: RPA automates compliance checks and adapts to regulatory changes, helping institutions navigate the compliance minefield while maintaining effective collection practices.
- Intelligent Prioritization: ML can prioritize collection efforts to focus on accounts with the highest probability of recovery. This targeted approach not only improves recovery rates but also allows human agents to focus on complex cases that truly require their expertise.
The Future of Collections
We’re not far from systems that engage in natural language conversations with borrowers, negotiating payment plans in real-time with empathy and understanding. Predictive economic modeling powered by ML will allow institutions to anticipate and prepare for market shifts that could impact delinquency rates.
Moreover, the integration of these technologies across all channels will create seamless, omnichannel collection strategies that meet borrowers where they are, whether through a mobile app, social media, or traditional communication methods.
Implementing Advanced Analytics: A Strategic Imperative
Embracing these technologies is a leadership decision. Successful implementation requires more than a technology investment. It demands a shift in organizational mindset, a commitment to data quality and integration, and a willingness to continuously learn and adapt.
Consider the experience of a major U.S. credit card issuer we recently worked with. By implementing new strategies powered by ML and RPA, they saw dramatic improvements across key metrics, including a significant reduction in 90+ day delinquencies and operational costs. This success story underscores the transformative potential of advanced analytics in collections.
The Road Ahead
The path forward is clear: Machine Learning and RPA offer a debt-recovery strategy that’s more proactive, personalized, and ultimately more effective.
For institutions ready to embrace this future, the rewards are substantial: reduced delinquencies, improved customer relationships, and a competitive edge in an increasingly challenging market. The question is no longer whether to adopt these technologies in collections, but how quickly you can harness their power to transform your operations.
Let’s set up a no-cost, no-obligation call to see how PAG can help you leverage the power of Machine Learning Models and Robotic Process Automation with our GOBLIN data and analytics platform.
David LaRoche is managing partner of U.S. Operations for Predictive Analytics Group. You can follow us on LinkedIn here.