This week’s SBA announcement that the Agency will make available borrower’s credit score that takes into account both personal and business attributes has sent shock waves through the industry. The funny part is that these shock waves were not the ones intended. The intention was to let the Nation know that Maria Contreras-Sweet, the new head of the SBA, is ready to help minority lending and that will further stimulate the economy. However, despite the best intentions, that was not what bankers were talking about yesterday.
First some background. Starting in July, the SBA will provide its lending partners a score that is said to accurately predict credit worthiness for loans below $350,000. The score was designed to ensure that risk characteristics, rather than socio-economic factors, are used to determine credit. By using this system, the concept was to eliminate credit bias so that more minorities are able to obtain loans. All this sounds good except for a couple minor issues.
From a macro policy standpoint, the announcement drew an eerie parallel to the Clinton Administration’s 1995 authorization to allow FNMA and FHLMC to guarantee subprime home mortgages and the associated changes for banks on allowing greater “innovative and flexible” lending practices to low and moderate income borrowers. Homeownership indeed increased from 63% prior to the change to 68% by 2004, but so did the number of loans to borrowers with FICO scores below 650. Everyone knows how this story played out and so the SBA announcement raised the concern over whether we are making the same mistake again. Is the economy really still in need of stimulus? With asset prices at record highs, are we further inflating the bubble with cheap credit. If we put credit into borrower’s hands that would not normally qualify, no matter what race they are, is that a good thing?
While the SBA announcement kicked off a harsh partisan debate, it was the regulators that sounded even greater alarm bells. The announcement came as a surprise to many and the early July start date absolutely floored more. Setting credit economics aside, the credit scoring issue further opens up an ongoing debate with the Fair Lending Standards and intent. Will the SBA score be considered a “Bureau Score” not subject to oversight or a “Custom Score” that is subject to examiner oversight? Does the SBA credit scoring system contain any factors that are correlated with a “Prohibited Basis” so that the scoring system inadvertently treats minorities differently? Do banks have proper policies and procedures in place to rely on the SBA credit scores? What happens if bank staff overrides a credit score? Does that introduce indirect minority bias? Will a disparate treatment test be required? There are so many questions and so little time.
Then were was an outcry from underwriters at banks that said they are not sure if they could feel comfortable with a black box credit scoring methodology that eliminates the analysis on a company’s cash flow. Wasn’t it the regulators that have been driving home the point that banks need to focus on the cash flow of each transaction and pay less attention to third party rating agencies? Where does the SBA fit in?
Finally, there was a deep gulp from many SBA lenders that felt threatened to learn that the SBA will soon release “SBA One” an automated lending platform that automates much of the work for 7(a) lenders by automating uploads of documents, generating forms and the allowing of electronic signatures. Many lenders have made large investments in streaming their infrastructure to be competitive only to learn that the SBA may mitigate much of their competitive advantage. It was these lenders that have been screaming for greater automation for decades (such as the OK to use electronic signatures) and now find out that the whole industry will have this advantage come the first part of next year (we have our doubts as to timing).
In our opinion, this is all going to turn out OK and we understand that change always brings anxiety. We reviewed the SBA scoring methodology and did some back testing two years ago and found their model, at the time, to be fairly accurate. We never tested for inbred biases, but we will put our trust in the SBA that they have this figured out.
Thus, to us, the more important takeaway is that the SBA is forcing bankers to change more to automated underwriting for smaller sized loans which banks desperately need to embrace in order to reduce costs and make lending more efficient. Many banks had these initiatives already and while we understand the anger of these bankers over the SBA assisting those too lazy or too risk adverse to build their own systems, the SBA scoring and SBA One will democratize that technology while making it regulatory and culturally acceptable to rely more heavily on predictive analytics. We know this isn’t a popular stance, but artificial intelligence-based neural network with a continuous feedback loop that adjusts the weighting coefficients and the cutoff score can do a much better underwriting job than humans. This new SBA program will lay that ground work.
The SBA’s announcement will continue to be debated and will cause many parties to scramble, but we predict will pave the way for greater financial institution performance. The announcement has caused many to reassess their position on credit scoring and that alone is a good thing - at least in our humble opinion.
Submitted by Chris Nichols on June 12, 2014