<img src="https://ws.zoominfo.com/pixel/pIUYSip8PKsGpxhxzC1V" width="1" height="1" style="display: none;">

How Regression Analysis Can Help Your Fair Lending Program

2 min read
Mar 11, 2020

Sometimes a fair lending analysis yields unexpected results. Pricing disparity or denial rates may seem unusually high. Origination or fallout rates might concern you.

These are potential signs of a lending compliance issue—but how can you know for sure?

You may want to consider regression analysis.

What Is Regression Analysis?

Regression analysis is a predictive statistical tool used to figure out if there is a relationship between variables. It takes in a wide range of variables and reveals which ones had the greatest impact and which ones don’t seem to make a big difference. It finds variables that correlate (i.e. variables that appear to be connected). It also lets you know how certain the results are using probability.

For instance, it can be used to see how rainfall correlates with crop growth or GDP correlates with employment rates.

Regression Analysis & Fair Lending

When used for fair lending analysis, regression analysis can take factors like age, race, gender, ethnicity, pricing, debt-to-income ratios, loan-to-value ratios, and credit score (among others) and determine if they have any impact on:

  • Credit decision
  • Loan pricing

It can even control for factors like special promotions, market-based pricing differences, loan programs, or how long the borrower has been a customer.

If a prohibited basis factor had any bearing on the credit decision or the price of the loan, regression analysis should flag it. 

Related: How to Build a Strong Fair Lending & Redlining Compliance Management System

What Goes into Regression Analysis?

Regression analysis is driven by data. The more data you have, the more insightful the results.

It can work with as little as 200 loans (the results won’t be statistically significant, but it may reveal issues that would otherwise go unnoticed) but does best with at least 1,000. The data should be cleaned up and should include a wide variety of factors. If every loan in the data set has the same pricing, same protected class of applicants, or if everyone was denied, it won’t produce helpful results.

One Regression Analysis Caveat

Regression analysis uncovers correlations, but it’s important to remember that correlation is not causation. Just because two variables are linked doesn’t mean that one causes the other.

The classic example is a notable correlation between ice cream sales and murder. Apparently, both ice cream sales and murder rates increase at the same time. That doesn’t mean eating ice cream makes people want to murder (In my experience, ice cream makes people happy).

Yet correlations are still extremely valuable, allowing a good data scientist to think about the relationships between variables in new ways, better scrutinize data, and uncover significant patterns.

Could Your FI Benefit from Regression Analysis?

Regression analysis is an incredibly useful tool, but many lenders are hesitant to embrace it because they think it’s too complicated. It’s a lost opportunity, but it doesn’t have to be that way.

Join Ncontracts for a one-hour webinar, Reducing Compliance Risk with Regression Analysis, on March 18, 2020 at 2 p.m. CST.

Our experts will take you through the basics of regression analysis and how you can leverage it to explain and defend disparities—without having to be a math wizard.

Whether it’s HMDA, auto lending, credit card, or consumer loan analysis, learn how regression analysis can help you promote proactive risk mitigation and fair lending compliance.


Subscribe to the Nsight Blog