FICO score still rules
CEO explains why it remains the dominant metric for deciding whether a borrower is creditworthy
ON THE RECORD
For decades, the FICO score has been the dominant metric for deciding whether a borrower is creditworthy. When applying for a mortgage, a car loan or a credit card, chances are that a FICO score is one of the first things a lender looks at.
But over the last few years, a new breed of online lenders and credit scoring firms have started developing their own methods of judging whether a borrower is a good bet.
Some are focused on consumers who don’t have enough financial track record to have a FICO score. Others simply believe that they have a better method.
This year, for instance, San Francisco lender Social Finance said it had dropped FICO from its underwriting process in favor of its own system, which takes into account such factors as what a borrower does for a living or studied in college. Other firms look at even more unorthodox criteria, including Web search queries.
William Lansing, chief executive of San Jose’s Fair Isaac Corp., the company behind the FICO score, said some of these new ideas
might prove out, but others won’t — and some are likely to face opposition from federal regulators.
Lansing spoke with the Los Angeles Times about the new world of credit scoring and why he’s not worried that the FICO score will fall out of fashion. Here’s an edited excerpt of that conversation:
There’s been lots of talk over the last few years about scoring people based on social media data. Is that something you’ve looked at?
Along with credit scores, we also do predictive analytic software that lenders use to do [lending] and detect fraud. And we can incorporate anything the customer wants.
For example, what about a Facebook profile? Is there anything there that’s useful? It probably has more value than zero, but not a lot more. If you did a search for “wasted” in a person’s profile, would that have predictive value? It might, but there are so many other things that would have more predictive value.
Is that something customers have actually asked you for, using social media data?
We have that conversation. No customer has asked us to do it after a conversation about how little utility it would have.
What is predictive then? What are the factors that lenders ought to look at?
What we find is, anything that reflects personal responsibility tends to be highly correlated to your FICO score. If you’re a good student, a good driver. Those things tend to be correlated to being a good credit risk. One thing we don’t do: FICO doesn’t look at your income. It’s not focused on how wealthy you are.
Wouldn’t it make sense to look at how much someone makes if you’re planning on lending them money?
Of course you want to know they have the money to pay you back. You need capacity and you need intent to repay. If you focus on capacity exclusively, that’s a recipe for problems. You could have a high income and blow it all on coke and Ferraris.
So what do you look at?
The FICO score is designed to predict a consumer’s propensity to repay debt. A great thing to look at is what a consumer’s payment history has been in the past. If they’ve been a good payer in the past, it’s likely they’ll be a good payer in the future.
Credit card repayment history is a big factor for FICO, but what about people who don’t have credit cards?
There are high school graduates, immigrants, other people about whom less is known. What we do there is we ask, “Are there other data sets that provide a window into whether this person is creditworthy or not?” We designed our FICO XD score specifically for the purpose of scoring people who were previously unscorable.
What other data do you look at for FICO XD?
Are you paying utility bills on time? That’s predictive. Are you paying your phone bill on time? That’s predictive.
What do you make of firms using things like where you went to college or what you studied? Is that stuff predictive?
There’s a whole category of things that seem intuitive but don’t really help you. Let’s say Stanford graduates are a good credit risk — they get good jobs at a higher rate than others, they’ve done things that demonstrate good behavior. But does that scale? Even if you use that to look at thousands and thousands of students at prestige schools, it’s still a small number of people. The idea works, but it’s only useful for a tiny population. With FICO, you cover a lot of people.
Between FICO and FICO XD, how many American adults have some kind of credit score?
You’re in the 200 million range.
This month Hollywood company ZestFinance signed a deal to build credit scores for Chinese consumers based on their queries to leading Chinese search engine Baidu. I’m not sure that would fly in the U.S., but do you think that information could be useful?
What you use the money for is a little bit predictive as to whether you’ll repay. If you have knowledge of what the transaction is for, that can have some value. And if, through search results, you can have a deeper understanding of a consumer, it’s conceivable you could make a better credit decision.
The FICO score has been around for a long time and people understand it pretty well. Is that an advantage you have over some these new scoring systems?
When people say, “How do I improve my FICO?” the answer is, “Pay your bills on time.” That’s the lion’s share. This is an area where alternative lenders are going to run into issues. Regulators are going to say you have to be transparent with your consumers.