Commercial Credit Scoring has been a topic that has often interested me. While it is not something that I deal with on a daily basis as the type of loans that are underwritten in my Department are too large or complex for this methodology, I have always wondered if one day the algorithms and code will get smart enough to replace me.
I am the first to admit that there are some areas where a piece of software has an advantage over a Credit Officer. For example if I were to count every single loan that has crossed my desk during my career, the number will be under a couple of thousand. If I was asked to recall the specifics for each loan, I might be able to give you partial answers in a couple of hundred of those loans even with the benefit of a pretty good memory. A good Commercial Credit Scoring model will have the benefit of a database of tens of thousands of loans, as well as their eventual outcomes.
Before we get any further, let’s define what I mean by Commercial Credit Scoring. A Credit Scoring system is a mathematical model that relies on information provided in your small business loan application, as well as other information collected from public records databases to automatically determine if you are a good enough risk. In essence, your bank is delegating the approval decision to the software. This process is called “auto-approval” or “auto-decisioning”.
Now, we can discuss the reasons why commercial banks put into service credit scoring systems, how cutoffs are determined, and tips for small business borrowers that intend to apply for a small business loan that will be scored.
There are several reasons why banks go with a credit scoring system for the approval of business loans:
Speed and Scalability. A fully integrated Small Business Credit Scoring system is a thing of beauty. The customer can either fill a hard copy commercial loan application to be entered by bank branch personnel or it can input the application directly into a web application. Once the application is in the system, the decision can be made instantaneously as soon as the external data associated with the borrower and guarantor is downloaded from external databases (business payment history, business registration, individual credit score, etc. The scalability is practically unlimited as a system designed to run 100 loan requests a week could easily run 1,000 loan requests.
Cost. Although the implementation costs can be significant, the long-term benefit for the bank is realized in the ongoing cost structure of the department. In other words, skilled people are expensive. If the bank has to analyze every single borrower using the regular underwriting process, it will take at a minimum a couple of hours per loan request even if we are talking about an abbreviated underwriting process.
Control and Consistency. Having the approval delegated to a smart decision system, avoids situations where one loan that is declined by one analyst could be approved by another. Furthermore, it allows you to instantly change the loan production of the bank by adjusting cutoffs. If for example the bank decided to reduce loan production by half because of temporary liquidity issues, you could very well look at the historical loan production and increase the cutoff to get the expected decrease in closings. Control over the model improves over time once there is a seasoned portfolio of small business loans, and you can test the assumptions in the model against the actual performance.
Risk based pricing. Being able to charge your riskiest borrowers more than your safer borrowers is a goal that every banker would like to reach. Being able to use the credit score as the support for the pricing helps in achieving this goal.
A cutoff is simply a credit score number that the loan request must meet to get an approval. Models are built based on historical performance. The process is better explained with the enclosed graph. Remember that in reality we are talking hundreds or thousands of loans, but for the sake of a clear graph the numbers have been reduced.
Taking into consideration that credit scoring does not work that well for large loans, lets concentrate on the loans under $200,000. According to the data reflected on the graph:
- A total of 34 loans under $200,000 were put in the books, 10 of those loans went bad. As we would like to decline those loans, we marked them with a “D”, loans that performed accordingly we would want to approve so we marked them with an “A”. The overall loan portfolio under $200,000 shows a default rate of 47% (16 / 34)
- The purpose of the cutoff is to determine the credit score floor that can be set that would allow for the greatest number of loans with a default rate that is manageable. In this example, we assume the bank decided to make the cutoff 675. This results in approving loans that have shown a historical default rate of 17% (2 / 12)
- The loans that would be automatically approved under the model are shown in the green zone, and the loans that would be automatically rejected are shown in the red zone. You might notice a buffer zone in between. That is an area where loans could be approved subject to individual review by an underwriter.
Of course these numbers and graph is just for show. If you are approving loans with an expected default rate of 16% you are in trouble.
What can a small business owner do to score better?
The single biggest thing a small business owner can do is to take care of his personal credit score. The truth is that the majority of business credit score decisions are based on the individual’s credit score. When one looks at small businesses, the line between the individual and the business entity is blurred. In fact, according to a survey prepared by the SBA Office of Advocacy in 2006 approximately 70% of commercial banks business score decisions were based on the credit score of the individual alone. An additional 25% used the individual credit score as a factor, and only about 5% used the business credit score alone.
Other steps that a small business owner can do to score better is to build up its business credit records of payments. If you do not have a DUNS number get one and use it to get small credit lines from vendors. Unfortunately, D&B is a pest at marketing and is always trying to get you to pay for their services by scaring you into buying their premium credit building packages. In my opinion, the premium service is not necessary specially since the weight of the decision is going to be based on your individual score.
Remember that each model is a little different. There is going to be some that assign some weight to certain factors that you might have not thought off. For example, certain banks give points on their formula for existing deposit customers so opening a checking account in the name of the business might be a good idea.