New York | Did Goldman Sachs (GS) and Apple Inc. (AAPL) discriminate against women in the provision of credit on the new accounts for the Apple Card? Probably not deliberately, but credit scoring systems generally see women as being inferior credit risks – both internal and external to the actual underwriting process.
When you apply for consumer credit, your individual score and credit utilization history are the primary criteria considered. “Several husbands — including Apple co-founder Steve Wozniak — have complained that their wives were offered dramatically lower lines of credit than they were,” reports CBS.com. Sadly, the new reports did not disclose the FICO score and credit histories of the offended spouses.
A June 2018 paper by Federal Reserve Board researcher Geng Li, “Gender-Related Differences in Credit Use and Credit Scores,” outlines some of the difficulty faced when asking questions about gender bias in consumer lending. First and foremost, lenders are not permitted to gather or use gender related information, thus the public research in this area is limited. Li notes:
“The Equal Opportunity Credit Act largely prohibits the use of demographic information, including gender, in credit underwriting, pricing, reporting, and scoring.4 As a result, information on credit histories and demographic characteristics has rarely been collected in the same data source, making evaluation of gender-related differences in the credit market challenging.”
The mere fact that one spouse gets more credit than another does not necessarily indicate a problem in the credit scoring system. One spouse may be more active financially than the other, resulting in a higher rate of utilization of credit. Current usage of revolving credit lines such as a mortgage or a car lease is a major factor that drives credit scores and the availability of credit.
Also, each bank has an internal default rate score for the “ideal” customer for a given product. As we noted in our recent profile of GS, the bank currently has about 40bp of gross defaults or just inside the breakpoint for a “BBB” portfolio default rating. Applicants below the “ideal” credit profile for a bank credit card product, for example, are likely to see lower initial credit limits or be denied credit entirely.
But the real question raised by the consumer advocate wah-wah aimed at GS and AAPL is whether the lender reverse engineered the gender of the applicant during or after the underwriting process. Wall Street has for years wanted and tried to be able to track the obligor and the related assets and information in real time, particularly in areas such as mortgage lending. The early efforts backed by Buy Side hedge funds were massive, very expensive undertakings.
But whereas a decade ago and more it was too laborious to manually merge credit profiles with a given individual’s personal information, today that data is commonplace and ubiquitous. Merging third-party demographic information with a loan file is a very simple matter. Indeed, some of the major consumer data repositories reportedly have begun to offer permissioned clients such as lenders and servicers secure access to merged data sets.
Just imagine a confidential, non-public dashboard that merges credit, the value of the consumer’s home and behavioral data gleaned by Google and Facebook (FB). Once a lender or investor can determine whether or not you are a single parent and a biological male or female, the credit analysis changes. Why? Because there is a correlation between gender and default probabilities. Again Li:
“[S]ingle females tend to have higher installment loan balances, higher revolving credit utilization rates, and greater prevalence of delinquency and bankruptcy histories than otherwise comparable single males. Reflecting such differences in debt usage and credit history, on average, single female consumers have lower credit scores than comparable single male consumers.”
Now here’s the catch. GS and other lenders don’t necessarily even need to hack your complete profile because the FICO score largely captures the gender difference. This is one reason that consumer advocates have been pushing for “competition” among credit scores. Calling for “competition” in credit scores is another way of saying that we’ll “democratize credit” and stick the losses to bond investors and the taxpayer. Just because you pay your utilities on time does not mean that you can service a 30-year mortgage.
Of course, given the past track record of GS in other domains, it is easy to imagine the firm trying to limit default risk by reverse engineering the demographic attributes of their retail customers. As noted above, creating a comprehensive profile of an obligor is so easy in today’s market for behavioral data that it would almost be negligent for a lender or servicer not to have the information. And since there are any number of vendors happy to maintain such data remotely, away from the bank’s IT platform, it would be almost impossible to disprove such an allegation.
So ask not whether you should be soiling your nappies because Google is caching your search results or playing with the algorithm that determines what you see online. Ask instead how you feel about unregulated third-party data providers and consumer credit repositories maintaining profiles of all of your consumer spending behavior, by name, gender and other attributes, as well as the behavior of your family.