Expanding market share is always challenging for businesses, and for lending institutions subject to CRA and fair lending regulations, it’s a tightrope walk. The sophistication of marketing technology has advanced steadily over the past few years, allowing marketers to target audiences with greater precision than ever. Combined with analytics and AI models, finding new target segments is a risky proposition for financial institutions. In fact, over the past 24 months, financial institutions have paid a minimum of $78 million in settlements as government regulators continue enforcing redlining and other discriminatory practices.1
In many organizations, marketers derive personas from customer data, or they create profiles of the targets. Typically, demographics come into play in these definitions, so you have a married white female, wealthy Boomer or Hispanic, college-educated, Gen Z male. When combined with targeted advertising or social media platforms with filters, the result is a marketing program with a built-in bias that either focuses on or excludes population groups. The very essence of 21st-century marketing centers on focusing on groups that provide the best basis for sales and not so much for fair lending compliance.
When we go one step further into the world of AI, bias is more challenging to detect. First, many financial institutions have fragmented back-end data systems that are not consolidated or harmonized. This data is the foundation to build or train AI models. Rumman Chowdhurty, Twitter’s former head of machine learning ethics, transparency, and accountability, states, “Lending is a prime example of how an AI system’s bias against marginalized communities can rear its head. Algorithmic discrimination is actually very tangible in lending.”2 As financial institutions mine their archived lending data in an effort to make better approval decisions, they unintentionally can create a host of fair lending compliance issues. She elaborates, “Fast forward a few decades later, and you are developing algorithms to determine the riskiness of different districts and individuals. And while you may not include the data point of someone’s race, it is implicitly picked up.”3
How can financial institutions get the benefits of marketing technology and AI without introducing bias and discrimination? The answer is to implement a compliance analytic solution(s). Knowing how your lending performance will look to examiners is a challenge. Investing in software designed to reduce the burden of data preparation, eliminate hours of manual clean-up, and staff retraining can provide the counterbalance needed to comply with regulations. Leveraging fair lending technology can help automate your collection, verification, and certification of HMDA data, eliminating manual clean-up hours.
Are you or do you know someone who is creatively using data to advance their institution’s compliance management goals? Submit them today for our Alfredo de Hass Excellence in Analytics award that illuminates the often-unseen heroes who wield data as a tool for transformative change. The submission deadline is October 18th, and the winner will be announced at our annual CRA & Fair Lending Colloquium in November.
