Moving beyond out-of-the-box models

I’ve been employed in with credit unions for over 15 years as well as in financial services for more than twenty five. Based upon my experience, I have faith that there is one critical area where credit unions have lost ground when it comes to capturing the customer marketplace for loans and other lending options; that area is, making fact-based decisions. Years ago, banks and credit unions counseled me in the same boat if this came to business intelligence technology. In those days, regulation was the largest barrier for lending institutions to compete against commercial banks, because computers were non-existent on both sides. Many around the banking side would now argue, with few exceptions, that regulation is less intensive around the credit union side from the equation. But banks, unlike credit unions, have embraced the insights and efficiencies that can be found in business intelligence technology. Therefore, they have created a edge against your competitors by understanding consumer behavior much better than credit unions.

In a current overview of credit union portfolio performance data, it had been found that some of the risk factors widely used by credit unions to decision and value loans had very little effect on default risk using linear regression testing. Quite simply, when testing just one risk factor, keeping other risks constant, there wasn’t any linear correlation between the increase in the risk factor and an rise in default risk. That which was much more interesting and worth noting would be that the risk factors that impacted one credit union’s loan performance weren’t necessarily the standards that impacted another’s. So, what’s the point?

The point is, in case your bank is using risk factors and risk thresholds that you simply developed by cutting and pasting the other lenders have been doing in your market, and also have not re-examined those risks over time, you might be doing more harm than good for your lending program. Here’s a good example of default risk from two different credit unions in line with the Loan-to-Value (LTV) risk factor at origination.

In the example above, loan originations from 2012 were observed for defaults in the time since origination. It is clear, based on Bank A’s data, that there’s a linear relationship between default probability and LTV; default probability increases as the LTV increases. However, the same pattern is not noticed in Credit Union B, in which the default probability seems more random. Most credit unions don’t use LTV as a credit decision factor, but instead an issue in determining the utmost amount borrowed. However, if it is proven in the context of the credit union’s lending policies that LTV does have an affect on default risk, remodel which will it should be used in the choice criteria as well.

This reveals something related to risk modeling that needs to be considered. When the bank is applying a default or generic model of any sort, it should be understood the model is not considering the uniqueness of the credit union within the model and is, in most cases, statistically invalid for your bank. Quite simply, what’s true for the general population might not be true for any single credit union membership. This goes for credit bureau models as well as others for default prediction models. Even FICO? suggests that lenders introduce internally derived risk factors into credit risk models to be used in conjunction with a credit rating. Lending institutions should either do their own testing and analysis of data or contract having a vendor, like CU Direct’s Advisory Services, to conduct the testing while using lending institutions own data. If this doesn’t seem like good sense for you, consider the identical principle in a different context. If you have ever attempted to slim down, you will know different diets work with differing people. That is because our bodies metabolize food differently. Likewise, the uniqueness of every lender causes loans to “metabolize” within the portfolio differently.

In theory, all of this sounds simple, and it is if you possess the right data and technology in place. In truth, you probably already do — and have it open to you — however, you simply haven’t taken the steps to capitalize on it. One of the greatest challenges that Advisory Services, or other provider may have in aiding a credit union, would be that the applicable data hasn’t been collected or was not stored in an accessible format. The loan union ought to be collecting and storing any information that it uses to assess risk throughout the application. Common data points like credit rating, income, LTV, term, etc. should be stored along with credit agency details, for example inquiries and delinquent loan counts. This information will be invaluable when attempting to determine which factors impact loan performance to the highest degree.