The best loan decisioning information may come from your own database.
Any way to increase lending to 30 to 50 million consumers with limited or no credit will get the industry’s attention.
As a result, there’s a lot of chatter in the industry about the potential value of alternative data for making lending decisions and finding ways to expand the credit box. Companies like LexisNexis, Factor Trust and even the mainstream credit bureaus have databases of alternative data that includes payments for rent, utilities and payday loans. Cell phone, email and device information also can assist in the verification of applicant identities.
Yet my credit union has looked into alternative data and frankly, “It’s got a long way to go, baby.” There are significant challenges to collecting rent data. For example, the cost and burden of reporting data on smaller complexes will certainly exceed the value to landlords of doing so. The integrity of the data may also come into question—how well would the information provided accurately reflect how people have paid their rent? We have seen an amazing increase in the amount of credit disputes in the last decade as it’s been the easy way to remove derogatory information from a (persistent) consumer’s credit bureau. All one has to do is repeatedly dispute negative information and wait for the lender to respond within 30 days. If the lender doesn’t respond in that window, the negative information is removed.
Another concern about the possibility of using alternative credit data is whether there truly will be increased lending to the underserved. For every consumer that pays rent well and has little credit, will there be another with a similar traditional credit file that is harmed even more by his or her history of utility, cell phone and rent payments?
There’s also some question about just how extensive the databases truly are. We talked to one alternative database provider that didn’t even have the data from Colorado Springs Utilities, serving a town of 600,000 people in our primary market. If our credit union is going to make better loan decisions on people with limited credit or past credit problems, we have to be able to access that data.
Recently I have seen alternative data discussed in regard to preventing identity fraud. The thought is that a database of information that isn’t in a traditional credit file and may not have been compromised in a massive data breach could help lenders catch credit criminals. A version of fraud using a synthetic identity (a credit persona that combines accurate identification information with fraudulent employment and contact information) is growing in leaps and bounds, yet it’s still a very small percentage of total losses. This new fraud is causing some lenders, mostly fintech companies that are trying to capitalize on the mobile movement by removing friction in the borrower’s experience, to struggle in managing application fraud. Yet for credit unions still trying to grow their digital presence, we mostly serve people we know—our members. The value of alternative data for us is still limited in this area.
I’d encourage credit unions to focus on their own alternative data, the checking and debit card history from their members with limited credit. In essence, it’s a gold mine just a few feet under the surface.
At Ent, we have a fair share of experience and data that suggests that a member relationship and the data contained within is valuable. At several points in time, we have done validations of Indirect loan performance and found that loans where borrowers who had just six months of membership before taking their loan had as much as 100 basis points of lower losses, specifically borrowers with low-600 FICO scores. A first-time auto buyer loan program a decade ago found that borrowers who were members for close to a year or more had about a tenth of the default rate of new members. Yes, there were other factors such as the quality of the collateral between the groups that contributed to the default rate. However, there was plenty of evidence that the loyalty created from a happy member and their checking account leads to a better payment history.
Finally, let’s imagine how we can make that type of data analysis actionable. How many year-old checking accounts with a certain amount of activity and minimal NSF history do you have? How many of these accounts also have direct deposit? If you also archive the initial credit score obtained when you open the account, you now have a database of underserved members who should be worthy of a smaller, short term loan (6-18 month amortization) that can help them build or rebuild credit. That small loan can then lead to a reasonable car loan!
Regardless of whether the member has had a prior small loan with your credit union, perhaps you can identify members who have had an outgoing ACH payment to existing creditors. We’ve had some success in identifying members who have made six consecutive ACH payments to sub-prime auto lenders and marketing a lower rate refinanced loan. Our lending staff was of assistance in identifying common sub-prime lenders from the credit bureau which was used to isolate prospects in our ACH data. In the majority of the cases, the member’s credit had improved dramatically since they obtained the original loan.
Your own alternative data is a gold mine. Get to work!
Bill Vogeney is the chief revenue officer and self-professed lending geek for $5.3 billion Ent Credit Union, Colorado Springs.