Innovation in Lending
How To Win Over Stakeholders For AI Lending
April 1, 2021
There is a renewed sense of urgency around AI/ML investments in financial services. Many organizations view AI/ML as a necessity to remain competitive and grow portfolios safely in an uncertain and rapidly evolving economic landscape. A recent report revealed that one of the top ML use cases to emerge is in credit risk underwriting.
Machine learning-based models are more accurate at predicting default risk and credit eligibility by generating a more holistic view of an applicant. For lenders, ML models boost approvals with no added risk. With more confidence about whom to say yes to, ML also drives higher levels of auto-decisioning and inclusivity.
As more banks and credit unions consider AI underwriting, leaders are looking for the best strategies to secure buy-in across the organization. For many, AI elicits a range of responses from excitement about growth to confusion about its potential, to concerns about job displacement and potential biases. However, AI doesn’t have to be the boogeyman many fear it to be.
As a best practice, transparency, education, and speaking to business benefits such as competitiveness, revenue growth, and operational efficiency will be key to convincing stakeholders. To help you communicate the financial and non-financial benefits of investing in AI, we’ve put together a list of tips and stakeholder considerations to address.
Here are the influencers that you will want to engage and how to communicate AI’s impact:
The business case for ML is your biggest asset — so there’s plenty of common ground to go over when persuading executives and board members. Zest customers have seen approvals jump 15% with no added risk, or charge-offs drop by 30% while holding approvals constant when switching to ML underwriting. Calculating the business impact from such improvements form the backbone of your business case. Frame the business case using the goals set during AI strategy discussions, whether that’s growing market share, driving member inclusion, or increasing operational efficiency.
AI-driven underwriting has multiple benefits across lending objectives. Reaching new borrowers, cutting losses by reducing your exposure to bad loans and setting more accurate prices for better yield can all improve the bottom line while being seen as forward-thinking. Tie the benefits of AI adoption to the goals every business executive wants to achieve.
The Subject Matter Expert (Loan Originations)
For lending teams, AI provides better predictive abilities so they can avoid toxic borrowers and identify previously overlooked creditworthy buyers. For example, one lender achieved a 22% increase in portfolio approvals while holding risk constant. Lending teams will also benefit from automation productivity gains by expanding their auto-decisioning thresholds.
With AI improving risk assessment accuracy, underwriters can deliver faster decisions to help teams meet customer experience goals. Their day-to-day drastically improves by spending more time on a smaller percentage of applications that require a more personal touch. As Jenny Vipperman, the chief lending officer at VyStar Credit Union, said, “The role of the underwriter doesn’t go away. It just becomes something new, just like it became something new 12 years ago.”
In addition to communicating the benefits, you’ll also want to address explainability concerns. For lending teams, explainability is critical and they will be looking for answers to the following:
- How does the model make decisions?
- Why did an individual applicant receive a particular risk score?
- How much control does the organization retain over decisioning?
- How quickly can decisions be made?
- To what degree can/will decisions be automated?
The Credit Modeler & Model Validator
Overall, modelers can leverage ML to build more accurate, consistent, and efficient models with speed. Time-consuming manual tasks such as model documentation, validation, and monitoring will become automated, enabling teams to accelerate model development and production times.
Use these questions to help quantify the ROI for modelers:
- What is the current cycle time to create, validate and deploy a new model?
- How frequently are models updated, and how long does that process take?
The Regulatory Compliance Officer
For compliance teams, you’ll need to address explainability concerns. The compliance officer is worried about bias and upholding the bank’s Fair Credit Reporting (FCRA) and Equal Credit Opportunity (ECOA) obligations. These will be their top questions to address:
- How does the model make decisions?
- How do you ensure that models are fair?
- How much influence do they have over what data is used in the model and whether they can influence how the model makes decisions?
- How will they explain the model’s decisioning process to regulators and demonstrate that they are in compliance with all of the regulations and guidelines?
The Head of IT
Overall, IT is mostly concerned with the implementation logistics. You’ll want to be prepared and have answers to the following questions:
- How long will implementing an ML model take?
- What are the resource requirements for the project?
- What are the requirements for integrating with an LOS?
A Lender’s Roadmap to AI Adoption
COVID-19’s impact on the economy and lending industry has reinforced the urgent need to move to AI. In the current environment, lenders can’t rely on credit scores as they previously have and need a new way to evaluate and monitor credit risk with limited visibility and access to reliable data. ML models provide a more holistic picture of borrowers, enabling lenders to assess risk more accurately and make the right decisions faster, even during a pandemic.
Securing buy-in and addressing stakeholder concerns is a key milestone in the road to AI adoption. To help you succeed with the other key milestones, we’ve put together a guide provides a battle-tested approach and shares milestones and best practices from lenders who have moved to AI-powered underwriting to make more accurate risk assessments, accelerate modeling, and automate lending decisions. Read on to learn how lenders implemented machine learning models to deliver better results for every lending objective. Let’s get started!