The ongoing impacts of COVID-19 require lenders to continue to adapt their credit risk strategy
The vaccine rollout initially had a positive economic impact, however, the Delta variant put a damper on any ideas of a quick recovery, resulting in a growth in debt, delinquencies and uncertainty. The persistent impacts of COVID-19 require lenders to engage more proactively with their customers to predict and prevent delinquency as well as track who is doing well and who should be given access to more credit.
Leveraging real-time data with AI
Leslie Parrish, Aite-Novarica’s Strategic Advisor, Retail Banking and Payments, and Amyn Dhala, Vice President and Global Product Owner for Mastercard’s AI Express solution, recently participated in the LendIt webinar Leveraging real-time data with AI to improve the customer experience and mitigate credit risk. The webinar focused on how COVID-19 caused lenders to adapt their credit risk strategies and proactively manage debt to support both lenders and customers.
The combination of real-time data and AI allows lenders to take a more nuanced, personalized approach to assessing creditworthiness.
Real-time data enables more informed, real-time decisioning by creating a 360-degree view of the customer. According to International Finance, “by leveraging real-time analytics, banks are better equipped to create an enhanced customer experience, overcome constraints of legacy technology, drive greater value from customers and combat fraud.” Lenders can also improve the borrower experience with fast onboarding and loan approvals and even give credit to those previously declined thanks to real-time analytics.
In an article in BAI, Sudhir Jha, Senior Vice President and Head of Brighterion, wrote that “the benefits of AI for credit risk assessment will be enormous. With the right models in place, your organization can speed up credit applications and forecast delinquencies months in advance. And by leveraging available data, you can do all of this without adding more forms to fill out or other sources of friction to the customer experience.” AI can extract valuable insights from large volumes of real-time customer behavioral and financial data and continuously run different combinations of variables to learn from that data – and it’s much more adaptable than traditional rules-based decisioning systems.
Leveraging real-time data with AI allows lenders to ride the waves of economic uncertainty, quickly adapt to new products and improve the customer lifecycle while mitigating credit risk and reducing operational costs.
The ongoing economic impact of COVID-19
The economy has been a roller coaster over the last 18 months or so and the ride isn’t coming to an end anytime soon. Debt levels are at an all-time high and continue to rise, along with growth in delinquencies, and this uncertainty requires that lenders proactively engage with customers rather than playing catch up after the fact.
In addition, the pandemic has affected different households and industries unequally – with some thriving while others have lost significant ground – and those differences need to be taken into account when assessing credit risk.
These economic statistics give the big picture overview of where the economy is at in the second half of 2021:
- The unemployment rate was nearly 15% in April 2020, but it has since trended back down to just over 5%. The number of Americans who are counted among the long-term unemployed is still elevated compared to the recent past, but that’s also now headed back in the right direction.
- The GDP took a nosedive in Q2 2020 and then rose again in Q3 and Q4 of 2020. While economists were more bullish on their predictions for GDP growth early in 2021, they are still predicting a 5.9% growth rate for 2021 overall.
- While credit card debt has continued to decline, mortgage, auto and student loan debt are showing year-over-year increases. Mortgage lending is especially hot with the ongoing historically low rates.
Harnessing the power of AI and real-time data isn’t just a solution for the short term. According to McKinsey, “using real-time business data in decision making and advanced analytics to review credit-underwriting processes…will help banks cope with the present crisis but also serve as a rehearsal for the steep change that, in our view, credit risk management will have to make in the coming months and years. The best banks will keep and expand these practices even after the crisis, to manage credit risk more effectively while better serving clients and helping them return to growth more quickly.
5 key insights on leveraging real-time data with AI to mitigate credit risk
So what does all this mean for mitigating credit risk? Here are five key insights from the webinar:
Insight 01 – With increased uncertainty comes increased credit risk challenges
Financial institutions have been managing the unpredictable impacts of COVID-19 by adapting their credit risk strategies. They’ve been using tactics such as tightening credit standards during the decisioning process, more closely monitoring loan portfolios for signs of financial changes when that loan or line is open and upgrading collections capabilities and changing the mix of collection treatment strategies. And no matter where a customer is in the lifecycle, lenders are using different data and analytics to assess risk.
“Data and analytics capabilities are proving essential to the solution. Leading banks are accelerating digital transformation to enable real-time monitoring and effective mining of transaction data while automating the feeding of results into decision making,” reports McKinsey in an article on managing and monitoring credit risk after the COVID-19 pandemic.
Insight 02 – Buy now, pay later: new products need real-time solutions
Lenders aren’t just having to react to a volatile economic environment; they also need to track and adjust to new product categories. Buy now, pay later (BNPL) is a great example – it allows consumers to split the total cost of an item, or their shopping cart, into installments, and it’s growing in popularity. Lenders have the option of getting into this market themselves, but even if they don’t, they still have to take BNPL payment plans into account when assessing a customer’s credit risk. This can be a challenge when BNPL loans often aren’t classified as conventional credit and aren’t shared with the national credit bureaus, essentially hiding these loans and making it harder to build an overall picture of a consumer’s actual debt.
Real-time scoring, assessing credit risk at origination and instant underwriting help to mitigate the risk. And given that “customers financed about $96 billion of online purchases with installment loans last year, that could reach $300 billion globally by 2024,” according to Reuters. Tracking BNPL loans will continue to be a priority.
Insight 03 – Better credit risk management improves the customer lifecycle
Rising debt levels and increasingly complex credit risk challenge the need for being competitive and providing a good customer experience. Lenders are turning to AI to manage these competing factors. Recent studies show that banking executives have already made substantial investments in AI and are continuing to prioritize investing in it in the future. Looking at AI for credit risk specifically, 74% of executives say AI is extremely important for credit scoring.
AI is powerful across every stage of the customer lifecycle when it comes to credit products. Here’s how it’s used throughout the lifecycle:
Credit risk at origination:
- Assess credit risk for new customers and enables more informed underwriting and decisions
- Use AI to assist in determining initial credit limits and giving new customers a history
Credit delinquency and portfolio management:
- Catch early warning signs of delinquency and support developing more personalized strategies
- Detect customers who may be approaching delinquency and identify those who could use higher credit limits – improving the customer relationship by supporting those who may go delinquent and rewarding those who are doing well
- Use real-time data and intelligence to optimize customer outreach and collection strategies
Insight 04 – Better data means a better customer experience
Data is power in today’s world and having access to Credit Bureau data along with other account and transaction information makes AI models more accurate and effective. Important sources of data include:
- Credit Bureau information: this is currently used by lenders to make faster credit risk decisions and remains the cornerstone of assessing creditworthiness.
- Transaction behaviors: understanding how a customer interacts and uses their credit card helps predict future risk behavior. Transaction data can also be leveraged in real-time to continually update customer profiles.
- Account events: knowing how a customer interacts with lenders provides insights about future behaviors.
- Payment info: knowing exactly when a customer is making payments, how much they are paying off, and how they are paying it can be included in an AI model to help predict credit risk.
When AI models are developed, it’s important that they’re designed to be ethical which includes eliminating bias. Model governance and making fair lending decisions need to be adhered to and incorporated into the model, and credit policies and guidelines need to be built in.
Insight 05 – Scalability supports real-time analytics
What is scalability when it comes to AI and real-time analytics? It’s the model’s ability to grow and adapt to changes such as new customers, new products and increases in transactions and other information. Rather than slowing down when faced with an increase in data, for example, a scalable AI solution keeps up with whatever is thrown at it.
Brighterion’s AI is built with a powerful, distributed file system, specifically designed to store knowledge and behaviors. This architecture allows for unlimited scalability and resilience to disruption as it has no single point of failure. Distributed architecture also enables lightning‑speed response times (less than 10 milliseconds), and end‑to‑end encryption and traceability.
Developing an effective AI model is all well and good but developing a model that can deal with real customers, real transactions and real data is more challenging. According to Gartner, less than 50% of AI models make it to production. And a McKinsey survey concluded that “achieving (AI) impact at scale is still very elusive for many companies.” But because Mastercard has been leveraging AI for over a decade, they have a deep expertise in developing models that make decisions in real time with reliable uptime and unrivaled performance.
Marrying scalable AI with real-time analytics means lenders can be more responsive, make better decisions and have visibility into every stage of the customer lifecycle so they know what’s happening in the moment and aren’t left behind the eight ball.
How to get started with AI for credit risk
When lenders start the journey of using AI for credit risk, they’re not just facing a technology shift but also a mindset shift. Leaders have to understand their overall objectives and goals in implementing AI. This needs to begin with encouraging their teams to break out of siloed mindsets so they can look at the big picture of how models can support both them and their customers. The broad steps in the journey to using AI for credit risk include:
- Determining current challenges and goals
- Deciding on stakeholders
- Creating a road map for how AI can add value to a financial institution’s credit risk management strategy
- Developing a customized AI model
- Deploying the AI model at scale
Brighterion’s collaborative, six-step AI Express process provides organizations with the experience, evidence and plan to build and deploy their AI model. Brighterion’s team of AI experts work with lenders to determine their organization’s business challenge and desired outcomes. The team develops a fully working model in under two months, personalized to meet the organization’s goals while providing them with knowledge that leverages their internal business expertise and data analytics.
Upon completion, lenders receive a proof of concept that demonstrates ROI and a deployment plan to assist with decision making and moving to full production.
Learn more about leveraging real-time data with AI by watching the replay of the LendIt webinar.