Over Two Decades of Risk-Based Pricing, Are Credit Unions and Their Members Seeing the Benefits?

Twenty years ago, I stepped into the credit union movement after spending most of the first decade of my financial services career working for banks and finance companies. It was at this time that many credit unions were transitioning from a single-price lending strategy, where every approved member paid the same interest rate on a loan, to offering variable pricing to members based on risk. The theory posited at the time was, “we can offer loans to more members if we can properly price the loans for the relative risk.” Higher risk borrowers would pay more in interest than low-risk borrowers, as the additional interest earned would cover the presumed increased losses due to default. Low-risk borrowers would benefit from more competitive pricing.
Additionally, to overcome the sensitivity to treating members differently, many credit unions insisted in their lending policy that they were adopting a risk-based pricing, not a risk-based lending, strategy. What this meant was that no member’s access to loans would be curtailed because of risk, only that they may have to pay a higher interest rate to compensate for risk of loss. This alleviated the concerns of many members who felt that risk-based lending practices would cause the credit union to drift from its most important core value, serving the under-served. In 1995, the NCUA published guidance for Risk-Based Lending in 95-CU-174 (https://www.ncua.gov/Resources/Documents/LCU1995-174.pdf). This is an important read as it reveals some important facts about risk-based pricing. For example, we shouldn’t be pricing loans higher simply because borrowers have low credit scores. We should be pricing the loans higher if we know that the lower credit score will result in higher losses. Interesting, too, is that while pricing should be variable, net yields across the portfolio should be relatively consistent, meaning that the pricing variance strictly represents a risk premium, not increased profit opportunities.
So then, the question after twenty years is, how are we doing? As a lending consultant, having worked with hundreds of credit unions over the years, I have an insider’s view to loan production and cash-flows in credit union loan portfolios. Generally speaking, this is what I have observed: Credit union portfolios are highly concentrated in low risk loans, and variable pricing is not highly correlated to actual risk of loss. The result has been that low-risk loans often have thinly positive to negative net margins and high-risk loans have extremely high net margins. The issue is, when the portfolio is highly skewed toward the low-risk end of the spectrum, not enough profit is made from the high-risk spectrum to cover the thin margin at the low-risk end, resulting in extremely low ROAA. I’m not sure that this was the intended result of what was envisioned in 1995 and, in fact, may run contrary to the regulatory guidance.
Now, I don’t believe that the credit union philosophy has changed. On the contrary, anecdotally, I hear more and more credit unions espousing the desire to be more member-centric and increasing their services to the under-served. But, after over twenty years chasing the same goal, we often find ourselves looking over our shoulders for high losses from the past and regulatory scrutiny when we begin to originate more “risky” loans. Nobody wants to be that guy or gal that is part of their credit union's “troubled past” story. So, what are we missing and how can we more effectively price loans while remaining competitive in a highly competitive market place?
It all starts with the definition of risk. After all, the strategy is “risk”-based. Back in the mid-nineties, when powerful computers that could quickly run statistical analysis of borrower characteristics didn’t exist, especially in most credit unions, the FICO score model provided us with a reasonable and defend-able methodology to establish the relative differences in risk between borrowers. We created “pricing tiers” that would categorize borrowers in to pricing tranches based on their credit score alone in many cases. The credit score has a baked in default probability, so it is theoretically possible to calculate potential losses and factor those losses into loan pricing. The trouble is, factors that contribute to default, and more importantly loss, are often not even considered in the credit score model, unless you are using a custom score. Debt-to-income, payment-to-income and loan-to-value ratios often have a significant impact on losses and these ratios fall outside the credit score model.
What I have observed is that we tend to get it wrong on the poles of the portfolio. Low risk borrowers tend to under-perform, having higher default and loss rates than what was predicted by the score, and higher risk borrowers tend to over-perform, having lower default and loss rates that was predicted. Does this mean that the credit score model is faulty? No, more often than not, this can be caused by an over reliance on the credit score for risk of loss evaluation. We tend to make allowances for low-risk borrowers because of their high credit score that we wouldn’t make for someone with a lower credit score. For example, when a borrower with a 720 FICO score is requesting a 150% LTV on an auto loan and your maximum is 120%. Because of the borrower’s high credit score we begin to rationalize a decision to make the allowance. On the other end of the spectrum, we tend to mitigate low credit scores with stipulations we don’t put on lower risk borrowers. For example, we may require a cash down payment on an auto loan that we don’t require of our “A”-grade borrowers.
So, is this fixable? And, if it is, how? First, we need to properly understand what the credit score is intended to do and what it is not intended to do. I’ve conducted seminars across the country where I ask lenders to define the purpose of the credit score and there if very little understanding of its purpose other than to rank borrowers into pricing tiers. Many lenders have never actually seen a credit score model odds chart which defines the predicted default odds. For example, many are surprised to learn that he differences in default probability of a 680 credit score and a 720 could be as much as 300%. Second, understand what other factors are significant in predicting defaults and, more importantly, losses. This will take a little work, but it can pay-off big time in balancing portfolio risk and net income. For example, the mere existence of a co-borrower in most credit union portfolios reduces the probability of default and loss by 300%. Is this reflected in your current pricing model? Understand that while your competitors publish pricing similar to yours based on credit score, they are also considering other factors, as well and may ultimately adjust pricing in their final decision. This means that you can still publish competitive rates but leave open the door to adjust that pricing if necessary to cover additional risks discovered in the underwriting process.
If you would like to learn more about how to accurately price your loans and/or would like a review of your pricing strategy, please visit our website at www.cubi.pro, or email me at michael.cochrum@cubi.pro.


Popular posts from this blog

How to Estimate the Impact of CECL on ALLL

Measurements Matter

Are You a Coach or a Referee?