From asset returns to delinquency rates, underwriting has consequences. A lending strategy that’s too conservative shortchanges income, whereas an overly liberal policy jeopardizes a credit union’s financial stability. As credit unions strive to find the right balance between razor-thin yields and unacceptable risk, one key question keeps coming up: Should loan decisions be automated? The answer often depends on the community the credit union serves.
The two credit unions profiled here hail from the same state, but their underwriting systems are poles apart, reflecting vastly different memberships.
While Navy Army Community Credit Union ($1.9B, Corpus Christi, TX) scatters its underwriting authority among loan officers in 13 branches, Amplify Federal Credit Union ($626.5M, Austin, TX) relies on automated underwriting. Their lending strategies one aggressive, the other cautious influence each credit union’s underwriting style. Nevertheless, both have above average loan growth with modest delinquency rates and losses, proof that sound underwriting is possible whether it’s automated or not.
STRATEGY No. 1: An Army Of Underwriters
With just over half of its $1.6 billion portfolio in subprime loans, Navy Army Community understands the importance of solid underwriting. The credit union serves more than 115,000 members in south Texas, many of them immigrants, low-income families, or people with damaged credit. If Navy Army didn’t lend to these individuals, it wouldn’t be fulfilling its mission.
That mission certainly hasn’t hurt Navy Army’s financial performance, which includes a return on assets of nearly 2% and a loan portfolio that has grown 25.58% in just the past year. Meanwhile, at midyear, Navy Army had a 0.53% delinquency rate and 0.59% net charge-off ratio, the amount the credit union writes off as a loss after recovering what it can from delinquent loans. As of second quarter 2013, the average national delinquency rate was 1.03% with average net charge-offs at 0.58%, according to Callahan Associates’ Peer-to-Peer analytics.
Navy Army minimizes losses with an army of 80 lenders, all of whom have some level of underwriting authority. At the bottom of the lending hierarchy are member services representatives who can approve loan amounts up to $20,000 secured and $2,500 unsecured. Three more levels of lenders follow, each graduating to larger dollar amounts for secured and unsecured loans with the most senior lenders limited to $100,000 and $30,000, respectively. Lending executives co-authorize amounts that exceed the top limits.
The credit union trains its lenders to pick up on the subtle nuances that make some borrowers a better risk than others. An automated system, which relies heavily on credit scores, often misses those subtleties and rejects the loan.
The trouble with [automated] decisions is you don’t get enough intelligence to tell the story, says Gerry Morrow, Navy Army Community’s chief lending officer.
To get enough information, Navy Army’s lenders delve more deeply into the third c of credit character when capacity to repay and collateral aren’t enough to go by. Steady payment histories for past loans or bills can help a marginal applicant qualify for the loan, which the credit union prices higher for the risk. With most loan applicants applying in person, decisions are made within minutes while the member waits, and because lenders have credit-counseling training, they also advise rejected borrowers about which credit fixes they should make before reapplying.
The tiered underwriting hierarchy limits the damage any one lender can do, and additional practices address decentralized underwriting’s weakest points: consistency and fraud. Every loan decision includes a detailed account of why the loan was made or rejected.
I can go to any branch, pull that file up, and see the thought process and the comments of the person who made the loan decision, Morrow says. So the next person working with that member can see whether anything has changed.
To motivate lenders to make sound underwriting decisions, executives don’t share their targets for loan portfolio growth.
We’re not tweaking any underwriting to chase a number.
We’re not tweaking any underwriting to chase a number, Morrow says. Instead, the credit union evaluates its lenders based on their track record for consistency and loan performance, which includes monthly monitoring of delinquency rates and charge-offs for each loan officer’s book of business.
In some instances that book is pretty thick.
We have seasoned lenders who can’t take anybody else on because their book of business is so big, Morrow says. Instead, Navy Army passes the loan down to the next level of lender. The tiered hierarchy functions as a built-in training ground for junior-level employees to gain more sophisticated underwriting experience.
Although a centralized underwriting system would be cheaper, Morrow believes Navy Army would also pay a steep price for automating the process. Periodically, he gets calls from other financial institutions that are struggling to grow their loan portfolios, and all of them, he says, have centralized systems.
I ask them to go back and look at the stuff they’re turning down along with the review process for it, Morrow says. The loan growth may be coming through their doors, and they’re just not making the right call.
STRATEGY No. 2: Automation’s Many Advantages
Thanks to a membership that includes a large number of tech savvy, high-income professionals, Amplify Federal Credit Union rarely sees borrowers in person. Approximately 45% of loan applications are submitted online, while another 50% come from dealers. Amplify switched to automated underwriting in 2005 so its system could interact better with the portals the dealers used.
The shift gave the credit union the perfect opportunity to compare the performances of loans that originated before automated underwriting against loans that were processed through the centralized system.
We found that the delinquency rate and charge-offs were virtually identical, says Kendall Garrison, Amplify’s senior vice president for lending and marketing.
The credit union manages its risk well, with a 0.36% delinquency rate and a net charge-off ratio of 0.12% for a $500 million loan portfolio at midyear. Over the past year, that portfolio grew almost 9%, mostly from consumer real estate lending that includes first and second mortgages as well as home equity loans.
Garrison touts automation’s many advantages. Amplify did $13.5 million in consumer lending in July with just four credit analysts and six loan agents on staff.
Without automation, we would need almost twice the number of loan and credit analysts to handle the same volume of business, he says.
In addition, automation is consistent and impartial, helping credit unions comply with fair lending laws. The system is also fast, returning decisions to all applicants within 20 or 30 seconds. Human analysts review any rejected loans, a process that takes about 15 minutes if the member is waiting at a branch or dealership. Online applicants who are rejected wait up to four hours to get a second opinion. Garrison wants to halve that wait time, but to do so he has to fix another problem first: reducing the number of loans the automated system routinely rejects.
When Garrison joined Amplify in 2010, the credit union approved only 9% of submitted applications. That year, Garrison began reviewing in detail all of the loan constraints for automated underwriting with the goal of raising the acceptance rate to 40%. Because the system approved mostly applicants in the top credit tier, the margins for acceptable credit scores needed to be expanded. As a result, Garrison lowered top tier scores from 730 to 700 and then to 680.
Although Garrison met his goal of accepting more applicants, the system still kicks out 60% of loan requests, too many for human eyes to review quickly. Plus, the automated system continues to steer the portfolio to higher quality loans where interest-rate margins are narrow. As of second quarter 2013, 41% of the loans were for credit scores of 780 or higher, 22% fell between 730 and 780, and just 18% ranged between 680 and 730. Garrison wants to concentrate future growth in that lowest tier and raise total automated loan approvals to 60%.
As it happens, he has a good opportunity to do both. Amplify upgraded its automation system, and its new version went live in September. The new system delves more deeply into a borrower’s credit report so that payment history, for instance, is factored in to the automated decision-making process. Garrison hopes the broader approval criteria and the more in-depth automated analysis of borrowers will help Amplify accept more loans and improve interest margins. No matter the result, Garrison is certain the system will need further tweaking.
It’s always a nonstop adjustment to do automated approvals, he says.