PCBB
BID Daily Newsletter
August 06, 2007

BID Daily Newsletter

August 06, 2007

CREDIT SCORING VS. TRADITIONAL LOAN APPROVAL

In bank strategic planning sessions all across the country, we repeatedly challenge management to architect their bank to live in a future world where margins hover around 3%. To be a survivor, banks need to improve in a number of areas, such as looking at how they underwrite loans. One item that we constantly study is credit scoring methodologies. At present, we have a body of data that suggest the following for community banks: 1) Credit scoring is no less accurate than traditional methods of loan approval and may be more accurate in many cases; and, 2) Regardless of credit accuracy, credit scoring is at least 30% more efficient. While we are still collecting data on accuracy, one thing we know for sure is that using a credit scoring methodology differs from traditional credit review and approval. Picking up on Small Business Administrative data and talking to a variety of bank underwriters that use both methods, we asked: how does credit scoring and traditional credit approval differ when it comes to approving loans? The largest difference is that traditional loan review places significantly more emphasis on the borrower's past relationship with the bank. A repeat borrower is almost 60% more likely to get their loan approved using traditional decisioning than credit scoring. On one hand, we agree with the non-credit scored approach in some respects, as a borrower's past history and relationship is one of those unquantifiable items in underwriting. However, that said, we have uncovered a variety of instances where this positive bias was clear despite the fact the borrower had not established a material repayment track record. That is, borrowers that took a loan out and had not yet repaid still have a higher probability of getting their loan approved than with credit scoring. In our eyes, without establishing a repayment track record, a borrower's tangential lending relationship with the bank may not necessarily indicate lower credit risk. In similar fashion, another major difference between traditional loan approval and credit scoring is location. Traditional loan review places a higher emphasis on the address of a property or business than does a credit scoring model. Where a model treats all locations as the same (all differences are taken into account through rents, property values, etc.), traditional methodology is influenced by "up and coming" or trendy lending areas. Credits in these areas are much more likely to get approved than loans in stagnate areas. On the other side of the coin, credit models place more emphasis on different aspects of underwriting. For instance, a 3rd party credit bureau score usually composes about 20% of the credit scoring decisioning. Compared to traditional methodology, as long as the score isn't below a certain threshold (which varies by bank), we found the 3rd party score gets very little weight under traditional underwriting. Another difference is the type of business. In traditional models, many types of businesses are categorized as the same credit risk, whereas credit scoring models treat most business types differently according to their historic default rate. Finally, the last major difference that we found between banks that use credit scoring models and those that don't are the weight of the size of the loan. Credit scoring models are more sensitive to the size of the loan relative to the quantitative loan fundamentals than traditional models. While we are fans of utilizing credit scoring models (often in conjunction with traditional credit approval), the evidence remains inconclusive. However, in the next 12 months, this is expected to radically change and over time, models will increase in accuracy. As such, we highly recommend banks gain experience and start understanding and working with credit scoring models in order to improve accuracy and efficiency over time.