@sarge12 wrote:
@Anonymous wrote:Our current profiles are being compared to a group of profiles that had our profile data in the past. The dataset for this EQ-FICO Resilience Score is probably 1 to 2 years wide around the time of the Great Recession.
The more our profiles have in common with the ones that ended up with derogatories, the higher our Resilience Score. Same goes for all the rest of the usual FICO scores. This is the logistic regression part.
All we have to do is think about what people with derogatories might have done along the way. High utilization, inquiries, taking on more debt, missing payments, etc. Some of these are going to have a great effect on everyone's profile - even those with 850's.
Having a lot of money saved and a great job certainly isn't a good predictor: Biggest Defaulters on Mortgages Are the Rich
But since savings is not ever included in a credit report, neither could even come close to being predictive of credit issues. The credit score can because it is considering late payments and such.
It's really 'approximately predictive' with this sort of analysis. If our reported utilization goes to 60%, that puts us on a vector toward one of the derogatory scorecards and we lose points. Maybe we are going to PIF right after the statement, but the profile data won't show that until next month. There's no 100% guarantee we're going to default on anything, of course.
Somewhat real-time savings and paycheck data could definitely help us out with that, but that really opens up a can of worms.
If the credit report is clean in both my credit report and someone else who defaulted last recession had similar credit report data, it is not considering that the other card holder might have no savings.
You're not much of a risk if your report is clean with low utilization. Everyone had that at one point - it's basically 100% commonality and can't be used to predict anything.
Having emergency funds available has much more affect on what a persons ability to stay current on paying their bills than anything else. I agree that having a lot of money is not a great predictor, but if someone has for 15 years always payed in full because they care about their credit, they are not going to default on debt when the economy gets bad with 20 times their entire debts in liquid savings. They just aren't. Those rich mortgage defaults are not people with 15 years of a clean credit report.
There were an awful lot of executives and other high-earners who lost their job during the Great Recession. Credit scoring would have to know all of our outstanding debt obligations to make a determination against savings.
I bet there were a lot of people who chose to keep the emergency fund as high as possible and stop making some or all of their monthly payment obligations. Like those rich people that just walked away from debt. So what if they get a 500 FICO 8, right?
I really do understand what you're saying, and my credit profile would definitely benefit if more of that type of data was incorporated into the algorithm, but I'm very uncomfortable with any scoring model having access to that level of information. The new FICO 10 T with trended data doesn't bother me at all, and I think that's a good compromise.
But I also know how many fintech startups are out there incorporating that data right now while trying to convince everyone it's necessary and FICO is ancient technology. It seems inevitable at this point that artificial intelligence and even more invasive data will become the new way of life in credit risk analysis.
