Why 720 is the new 680
By Bill Vogeney
April 28, 2009
CUES' Credit Union Management's Web-only "Loan Zone" columns run the fourth Tuesday of every month.
I know what you're thinking. Is saying "720 is the new 680" like the saying, "50 is the new 40?" Personally, I hope the whole "50 is the new 40" is true, because I'm gaining on 50 pretty quickly, especially now that the economy is aging me three times as fast.
The fact is I have been joking with my direct reports that in terms of credit scoring, a FICO score of 720 is going to wind up being the new 680. The only problem is that, in reality, I'm very serious. If there is one thing consumer credit executives can learn from this recession and the resulting problems we've experienced with our portfolios, it's this: We've all become too reliant on credit scoring for too wide a range of scores.
Credit scoring's value, the ability to separate the obviously good from the obviously bad applicants, still works quite nicely. Most borrowers with a score of 720 have credit reports that look pretty similar. Borrowers with a score of 520 also look pretty similar on paper. The time we save by automating those more obvious decisions should allow us to spend more time figuring out what to do with people whose scores fall in the middle of the spectrum.
Over the last decade, a score of 680 has been the standard on which credit managers:
1. Set their cutoffs for automated approvals. Obviously, some lenders go to a lower score but may combine some other data points from the bureau to make an immediate decision.
2. Offer pre-approved credit. Whether it's a car loan or credit card, offering pre-approved credit to consumers can be a cost-effective way of building a portfolio.
3. Offer the very best terms. Good credit should mean lower losses which should lead to the lowest possible rates.
However, my recent experience at Ent is that there is a big difference between 680 and 720 when it comes to loan performance, a finding that's only partially supported by the FICO odds chart. I know a lot of lenders who have an aversion to statistics, so for simplicity purposes, I'll tell you this: A borrower who has a FICO score of between 680 and 719 is four times as likely to cause a loss or default as a borrower who scores 720 or above.
As I pointed out in a previous
Zone article, our credit union has experienced a greater increase in losses
in the 680 to 719 tier than in any other portion of our portfolio over the last
I believe there are a lot of reasons to explain this variance, including:
1. Use of credit. As I've tried to educate our staff, "You can fake a 680. There's no faking a 720." This means to score 720, a consumer has to use credit really wisely. A 680 FICO score allows borrowers to carry a fair share of credit card debt, so long as they pay their bills on time. That's a big caveat. A former boss of mine early in my career (my finance company years) once told me in a rather jaundiced tone of voice, "Everyone is good until they go bad." Truly, there are a group of consumers who have been living on credit, and many of them score in that 680-719 range. And they've gone bad in the last year.
2. Loan terms. This issue is best explained by auto and home equity lending. Borrowers in the 680-719 range were more likely to finance a larger percentage of the automobile value because they didn't have a cash down payment. Members with home equity loans were more likely to finance a greater percentage of their home value because they needed a larger loan amount. This not only means these borrowers have less equity when a repossession or foreclosure occurs; I also believe it means a repossession or foreclosure is more likely to occur because these borrowers have fewer options if they lose their jobs. Basically, it's harder to sell the car or the home because of a lack of equity.
3. The "Home value ATM." Between 2003 and 2007, homes were appreciating so rapidly many consumers looked at their growing equity as cash in the bank. They were ready to spend, and they did. Home equity lines of credit covered up a multitude of problems, from credit card debt to buying a home that was too expensive. When home equity lending was quick and easy, consumers often increased their credit lines to help in paying their other bills. Borrowers took loans "knowing" their homes would keep increasing in value, giving them the ability to sell and walk away with equity.
4. Optimistic underwriting. I'm a firm believer in the dangers of risk layering, which is the practice of assuming several different types of risk in loan underwriting on top of credit (score) risk. Risk from higher loan-to-value ratios, risk from ability to pay (debt-to-income ratio) and loan term risk (longer terms) are all elements of risk layering. The optimism comes from believing a 680 will repay as well as a 720, and that they should be offered the same terms. The truth of the matter is that people with a FICO score of 680 won't repay as well, and that we should have slightly more conservative terms for this second tier of credit quality.
So, how bad is our A credit performance? First of all, the losses on A loans (680-719) are not four times what our A+ losses are. They're up to six times more for indirect loans on new cars, and five and a half times more for home equity loans.
Based on our loss ratios, this means that to cover our incurred losses, an A borrower needs to be charged as much as 1.25 percent more for an auto loan than an A+ borrower. On home equity loans, it means we need to charge .50 percent more. That's pretty significant, and probably a lot more than many of my friends in the credit union community are charging. Thankfully, I have all of the resources I need to tell me exactly what our losses are per loan type, per credit tier. I don't think that's the case in most credit unions.
My boss keeps asking me, "Bill, is the credit score really worth the paper it's printed on?" My response is simply this: "Yes, it still does a great job in rank ordering risk. A 720 in general is less risky than a 680, which is less risky than a 640." However, no credit score can ever guarantee to you that your losses won't ever exceed "X." There are too many variables, including the economy and the other underwriting criteria used by the lender.
I like telling stories, and I especially like this one as it pertains to credit scoring. About a decade ago, an NCUA examiner told me of a recent credit union visit. The chief lending officer tried to explain why his credit union had made a particular loan with a 640 score. The lender pulled out the FICO odds chart, and it showed that 95 percent of the loans granted at that level would perform. He stated something like this: "Well, for me to make these loans profitable, I've got to make every one at this score range."
My response was this: "No, that's not what it means. What that 95 percent good rate means is 95 percent of a sample group of loans with a 640 score at origination paid on time for a two-year period." The examiner paused to think about it. The industry did not approve every loan it could have with a 640 score. The industry used some additional criteria like loan amount, LTV, debt-to-income, etc., and made a final decision based on all of those factors. The industry said no to some of those loans. Loans with a 640 score but a currently past due credit card may have been turned down. A borrower with $2,000 per month in income probably was turned down for a $35,000 car loan at 120 percent of retail value.
The bottom line is this: How good your other credit standards are will determine how successful your risk-based lending and pricing systems will be. Risk-based lending means you make a practice of trying to make loans to people with lower credit scores. Risk-based pricing means you make a practice of charging higher rates for higher risks with the hope you'll make enough to account for the incremental losses.
Some final thoughts on the variance between results from borrowers with a 720 score and a 680 score. In general:
1. People with higher credit scores have these scores because they're more likely to pay their bills on time and they're less likely to carry revolving balances. I hope we agree on this point.
2. People who pay their bills on time, and have paid their bills on time over a long period of time, do so because their income is steady and they're more likely to have saved money for the proverbial rainy day. Steady income and a stable job mean steady payments.
3. When the economy takes a downturn, all groups of consumers will be impacted, regardless of whether their scores are 800, 700, 600 or 500. Consumers with higher scores will be more likely to survive a job loss or income reduction and potentially be less likely to lose their job. The assumption that a borrower with a higher FICO will be more likely to retain his or her job is open for debate in this recession, though. Regardless, while losses across the board are going to increase, your losses on lower score ranges will mostly likely increase faster as scores decline.
That, my friends, is what makes subprime lending so unpredictable and cyclical. Right now, there are quite a few subprime finance companies that aren't making loans for several reasons. For one, the financial markets are such that these companies, reliant on securitizations to fund loans, have no funding sources. Secondly, when the economy tanks, subprime loans tank even more. However, even 680 FICO score loans are causing us higher losses than we ever expected. When our 640 loans are performing almost as well as our 680 loans, this tells me we had good criteria and used good judgment when manually underwriting the 640 loans. A human set of eyes looked at the loan and made a loan that made sense, as opposed to criteria we used to automate the decisions on 680 loans. Those 680 loans clearly weren't as good as 720 loans, yet we priced them almost the same as a 720 and offered virtually the same terms. A mistake we won't make in the future.
As the economy improves, I truly believe the cost of consumer credit will remain somewhat "sticky," meaning that the losses incurred by consumer credit lenders will cause rates to be historically high in comparison to the prime rate or Treasury rates. I also anticipate that lenders will be much more competitive for the absolutely best loans-borrowers over 720 and above. Borrowers with scores below that level will pay higher rates and face tougher terms.
Like I said, 720 is the new 680.
CUES member Bill Vogeney is SVP/chief lending officer of Ent, Colorado Springs, Colo.
- Go to the current
issue of Credit Union Management