As credit unions seek to understand reasons for their lending problems, some need to look right in the mirror.
That's the contention of one analyst, who believes that many credit unions, as well as banks, are adding to their lending ills by turning off automated loan decision tools in favor of a manual process. Fear of poor loan decisions that have resulted in losses and delinquencies and not trusting technology is the reason, asserts Christine Pratt, senior analyst at the Aite Group. The move is resulting in a number of issues that are restricting lending at credit unions and banks.
Pratt said as financial institutions and regulators examined processes that may have caused the mortgage mess, it just became "more expedient for any lending institution to turn off the automated decision engines and just use those tools on denials."
In speaking with hundreds of credit unions and banks and delving through two financial institution surveys, Pratt came to her conclusion late last year. She estimates that at best, institutions that use automated tools for loan approval funnel only 20% of their business through that channel, while before the recession many were getting 35% of their loans granted electronically. "And I believe many have reverted completely back to judgmental decisioning."
The biggest issue besides not getting as many loans through the pipeline, said Pratt, is the credit union is not spending enough time with members to understand their loan needs since the manual approval process takes up much more administrative time. "They are having a tough time knowing and understanding their members at a time when they have to get much closer to them."
With margins tight and efficiencies at a premium, automated decisions lower the cost to lend, reminded Pratt. "When you change processes to do everything manually you have to hire more people, too."
In the case of banks, leaning more on a manual loan process runs the risk of being out of compliance with the Community Reinvestment Act, cautioned Pratt, and in the case of credit unions, not granting loans to members who deserve them. "Automated loan decisions make lending more consistent. People don't make the same decisions as a software program, which won't deny someone a loan because they have green eyes."
Pratt explained that credit unions and banks still will take borrower information online, such as the app, credit bureau score and debt-to-income ratio. "But that is the point at which the automated tools for loan approvals have been turned off."
Pratt argued that the decision to turn off the automated tools is illogical, since concerns about granting proper loans revolve mostly around mortgages. "This is silly because mortgages have been the least automated of all loan processes."
Another issue created by the move, according to Pratt, is that financial institutions are relying now on loan decisions from individuals who lack extensive lending experience. "There is a lack of experienced loan officers. For a very long time many more decisions were made in an automated fashion, so loan officers went on to do other things or they retried. Now, if you are having a loan looked at manually every day, the younger generation needs to be trained and that is expensive."
Bill Vogeney, SVP at the $3.1-billion Ent FCU in Colorado Springs, Colo., told Credit Union Journal that he is not aware of credit unions turning off automated decision tools. "But it does not surprise me. I've heard some stories that credit union decision models did not amount to much more than a score. We utilize FICO score, bankruptcy score, debt to income, income, age of oldest trade line, lack of public records, among other things, in our decision engine. In the past, we may have had a more liberal debt-to-income ratio for A+, but we never had much of an issue with losses in that range."
Vogeney, also vice chair of the CUNA Lending Council, argued that there is no reason why decision engines can't work. "You can't possibly expect to automate more than 40% to 50% of decisions. I think if credit unions failed with their decision engines, it was probably from trying to reach too low into the FICO score range. I think we've done best stopping at about a 660 FICO score with these other limitations. Let's face it, during the boom times most credit unions just wanted to make as many loans as they could and got sloppy both in automated and manual underwriting."
Pratt is not sure if credit unions will turn around the situation soon, gain greater comfort with automated decision engines, and use them more frequently. "When I ask credit unions about going back to automated decisioning they say, 'We can't yet. So we're training more loan officers. We just don't trust automation.' They don't trust it, and they don't understand it. They can get their arms around a person, they can't seem to get their arms around a computer."
