You might think the growth in auto lending means it's time to hire another loan officer — but you might want to look at recruiting a data scientist, instead.
Bringing on such an expert to manage a credit union's decision engine who can work with a robust dataset with a custom default model can improve the accuracy of decisions and assess risk while making confident current Expected Credit Loss (CECL) calculations, according to Michael Cochrum, executive lending advisor for CU Direct.
"Even FICO tells lenders, if you are using [credit] score alone, it is insufficient to determine outlook for performance," Cochrum said during a recent CU Direct webcast on auto lending.
There is a greater probability of funding a loan if it is system approved compared to manually approved. "As the loan market changes, we want to sustain that market share, we need to look at how to get those additional loans," Cochrum said. Credit unions traditionally fund around 50% of the applications they manually approve, while funding 60% to 80% of applications approved by a system. Credit unions can potentially increase growth by 3% by moving towards a system-driven approval process, according to Cochrum.
As auto loan originations from credit unions increase, the need to efficiently and accurately consider applications becomes more important. Credit unions accounted for 25% of total auto loan originations in 2015, a 20% increase from 2014, according to Jose Torres, market research analyst for CU Direct.
Moreover, even as Ford Motor Credit reported an 18.6% deline from March 2015, credit unions continue to see strong growth. From 2010 to 2015, CU indirect loans have nearly doubled, Torres said, increasing from $70.2 billion to $137.3 billion.
Such growth is exactly why credit unions need to get a better handle on their underwriting and decisioning processes, Cochrum said.
Currently, the majority of financed auto loans fall into the super prime or prime category. But the most year-over-year growth has occurred in the near-prime and subprime categories. Near prime and subprime loan lending has grown 18.4% and 18.6% year-over-year, respectively. As these markets grow, a custom default model considering multiple factors becomes progressively necessary to assess risk.
While moving towards a more system-driven lending strategy may bring significant growth, member satisfaction is also crucial to growing as a lender. "Making our members wait for credit decisions is sometimes painful, which can cause them to choose another lending source," said Cochrum.
A system-approved lending system can increase the speed at which loans are processed while also freeing up underwriters to manually review credit applications that get kicked back by the system. This is still reliant on a decision engine that can simulate manual underwriting decision habits. Through utilization of data and a sound lending strategy, credit unions can make sure their members remain happy while also maintaining a healthy portfolio.
Cochrum outlined some misconceptions about underwriting:
- Credit unions cannot trust the credit score
"We've all been fooled by a credit score that appeared good on paper," said Cochrum. Credit scores are supposed to be a good judgment for default-risk over a 24-month period. While credit scores can signify the probability of a default, they are unable to foretell its severity. When looking at overall strength of loans there are multiple factors that need to be assessed. For example, a loan without a co-borrower is three times as likely to default than one with a co-borrower. Additionally, the change in a credit score over time is much more important than the original credit score. A borrower whose initial credit score is 670 with no change over a year is less of a risk than a borrower whose credit score goes from 730 to 680 over the same time period. Such a large change in credit score signifies a more serious financial issue for the member. The term of a borrowers loan also has an affect on the risk of default. Loans with a shorter term have a lesser probability of default, "There is somewhat of an exponential curve on terms of loans affecting default rate," Cochrum said. - Underwriters make better decisions than computers ever will
"Human beings are often distracted by unrelated information," Cochran declared. A multi-point review system is extremely beneficial in understanding the strength of a borrower/loan; credit score, loan-to-value (LTV), debt-to-income (DTI), payment-to-income (PTI), and net yield, are crucial data points to look at when developing underwriting strategies. Utilizing the data available to a credit union can give insight into how the parameters will affect loan performance.
According to data provided by CU Direct, 97% of lenders utilize internal customer data to determine things such as risk and profitability. Similarly, 78% of lenders analyze performance pools monthly - assessing how loans mature throughout their lifecycle — while 57% of lenders invest in historical customer data. Utilizing these three tactics can mitigate future plateaus in the market, offering up more opportunity for growth.
And key to all of it is being able to understand and use all that data to best advantage. "Data management is an area where the credit union industry does not do well…Sometimes hackers can get to info from a credit union better than the credit union itself," Cochran quipped.
