Achieving Fairer Lending through Data Science

by Chris Conlan

I believe one of the biggest achievements in FinTech, since the term was coined, has been exposing bank account balance data, transaction data, and payment portals via APIs. I won’t name any specific APIs, but there are now an ample supply of well-supported products that allow users to easily share balance and transaction histories with 3rd parties.

Why is this good? Because credit scores are ineffective.

What do credit scores consider?

Credit scores are a black box, but we know that they primarily score on the following factors. This list is paraphrased from Experian.

  • Payment history on credit products.
  • Credit utilization (low to medium, but non-zero, utilization is better).
  • Credit mix (a mix of short-term, long-term, and a mix of assets is better).
  • Hard inquiries (when businesses access your credit file).
  • Negative public records (foreclosures, collection accounts, liens).

Readers may notice all of these factors revolve around the consumers’ interactions with credit products.

What should credit scores consider?

The ultimate goal of a credit score is to determine if a consumer is a reliable counterparty to the vendor in a lending, leasing, or renting relationship. It may be counterintuitive that the credit bureaus don’t consider your income or your balances when issuing a score. This is mostly because they don’t have access to it.

Most people would agree that a more fair and meritocratic system would consider the consumers’ balances and income in addition to relationships with past creditors. Some would even say balance and income information is much more important than what credit scores presently consider.

Queue the FinTech industry.

Who does this help?

Plenty of wealthy people have great credit scores. The people that can benefit from a more meritocratic system include the young, the underbanked, and those unfairly afflicted by the existing credit scoring system.

The FinTech industry is proposing a regime where each lender can easily ingest balance and transaction data from applicants’ bank accounts and make an informed decision, based on the lender’s own risk tolerance, whether or not to engage with the customer. A small percentage of the U.S. population is suspected to have the following three things ...

  1. A high savings ratio.
  2. Consistent income and employment.
  3. A bad credit score.

A small percentage of the U.S. population is still potentially dozens of millions of people. This system of greater financial transparency would serve to their benefit, by giving them access to a wider array of credit products.

The Logistics

There is one catch for lenders.

Someone needs to parse, clean, store, analyze, and model all of that data. FinTech isn’t offering a free alternative to credit scoring, but they are offering ambitious businesses the opportunity to supplant the function of credit scoring with data science. Those businesses that do so, like many of our clients, have found a competitive advantage and increased their return on assets.

But, remember the title of this article. Lenders are offering fairer opportunities under this regime, and they are increasing their return on assets, because they have found a way to serve trustworthy customers that have been overlooked by the existing credit system.

Improve your credit decisions.

Blog

Data science discussions and general updates from Conlan Scientific.

Events

Your guide to events we host and sponsor around the D.C. area.

Team

We believe good data scientists are great software developers. Our team of professionals is ready to solve your financial data challenges.