The Decision Maker

Building Member Risk Scores with Data Analytics

Posted by Nate Wentzlaff on Aug 12, 2015 10:59:00 AM

Many credit unions rely solely on FICO scores for their lending decisions.  Data analytics now gives them the ability to develop their own forward-looking member risk scores.


In 1956, Bill Fair and Earl Isaac founded FICO.  The premise of FICO was that data could be used to improve business decisions.  In 1981, FICO introduced their first credit score algorithm that would become known as “The FICO Score”.  Utilizing basic data elements about past behavior allowed FICO to develop a standard credit score to enhance relationships between lenders and borrowers.  Now, with the rise of Big Data and Analytics, credit unions can supplement the use of FICO scores by building their own risk scores for members.

At traditional credit scoring agencies, data is transformed by proprietary algorithms from databases that are kept secret from lenders who purchase the scores.  However, credit unions now store vast amounts of data on their servers (from many different sources) about their membership.  This data can be used to build their own member risk scores.

Relationship-Based Pricing

Credit unions should make decisions based on the relationship they have with each member.  Some members only joined the credit union for a low loan rate and intend to close their membership as soon as the loan is paid off.  Other members have their entire financial life invested at the credit union.  These two members, through the philosophy of the cooperative, should be rewarded differently.  Many programs currently exist to reward members based on behavior (i.e. - participation rewards and year-end dividends).  This philosophy should be implemented in lending.

Futuristic Credit Scoring

Traditional credit scores are based on past behavior.  Therefore, many members are being unfairly punished by their previous mistakes or age group (e.g. – college students).  Even if they have turned their financial life around and are now more trustworthy than members with higher traditional credit scores (those who have not had many chances to fail yet), these members are forced to pay higher rates in proportion to their actual risk.  This results in members being “priced out” of loans they should have qualified for based on in-house analytics.  Predictive analytics have opened up a window of opportunity to help underserved members.

Loan Pricing

Loan pricing is a critical strategy of credit unions.  Setting prices too high will result in lost opportunities to competitors.  Low prices will result in a loss for the credit union.  Setting prices is one of the most important decisions managers will make.  Many credit unions rely on the traditional credit score when deciding how to price loans for each member.  Figure 1 displays optimal rates for tiered member risk scores and LTV.  Using member risk scores (derived from in-house analytics), credit unions can capture opportunities for members that other institutions cannot see.


Figure 1: Utilizing LTV and tiered member risk scores allows credit unions to price loans dynamically.  Market opportunities can be captured through predictive analytics.

Data Mining

With advancements in data mining tools, credit unions now have the ability to build their own models using the “blocks” of data found throughout their own systems.  User-friendly data mining tools have been developed to bring predictive analytics to the masses.  After establishing a member-centric analytic data model (ADM), that collects as much transactional data as possible, credit unions will be able develop many different algorithms based on a 360 degree view of each member.  This will allow credit unions to build member risk scores that dynamically capture relationships with each individual member.

 Analytics-Driven Pricing

The days of single product pricing are coming to an end, especially in financial services.  With an ADM in place, credit unions can begin tailoring their offerings to each individual member.  Utilizing the thousands of data elements attributed to member transactions, every decision can be customized for each member based on data analytics.  Utilizing a traditional credit score, along with member risk scores, will allow credit unions to improve the financial lives of their members.

New Call-to-action

Topics: Big Data, Membership, Data Analytics