Machine learning credit models help lenders solve an ever-elusive problem: how do you increase the size of your underwriting business without adding risk? The answer: By finding new borrowers overlooked by current credit scoring techniques. They’re out there, especially among populations with a thin credit file or a once-blemished credit history. Just because you have a low FICO score doesn’t mean you’re a bad risk. (The converse is also true: There are plenty of bad borrowers with “good” FICO scores.)
Traditional underwriting typically groups lenders into about ten or so risk categories based on a FICO credit score. For most borrowers, that’s a pretty good indicator of ability to pay back a loan or credit card debt. But in every category, there are borrowers who may be riskier or safer than their credit score suggests. Think of a 25-year old marketing manager who recently gave up his steady paycheck to pursue an acting career; or a doctor who recently emigrated to the U.S. to practice medicine, but lacks an established credit record.
Machine learning models excel at evaluating risk at the margins - which is where new customers reside, and losses lurk. ML models provide lenders with a more nuanced view by harnessing far more data and better math than traditional credit models use to classify borrowers. ML models score people using hundreds or thousands of factors instead of dozens. And capture the interactions among all these variables to produce a more accurate assessment.
As a result, ML models will swap in some people and swap out others. This “swap-set” of borrowers (the infographic above shows the reclassification of a middle group of borrowers) is then repeated again and again – until the lender’s loan portfolio is fully assessed. Result: More approvals with no added risk, or less risk with constant approval rates. Either way is a win.
ZestFinance’s ZAML software helps innovative financial firms deploy more profitable machine learning models in their underwriting. If you’re ready to take the next step, contact us at firstname.lastname@example.org to learn more.