Everyone wins when lenders and borrowers are on a level playing field.
About that algorithm…
I wasn’t planning to write about the Apple Card, but then this happened:
And it led, rather quickly, to New York regulators opening an investigation into Apple and Goldman Sachs’ underwriting process.
The department will be conducting an investigation to determine whether New York law was violated and ensure all consumers are treated equally regardless of sex,” said a spokesman for Linda Lacewell, the superintendent of the New York Department of Financial Services. “Any algorithm, that intentionally or not results in discriminatory treatment of women or any other protected class of people violates New York law.
Three quick takes:
Without detailed information on the financial history and assets of the applicants and the specific underwriting process used by Goldman Sachs, judging this complaint on its merits is impossible. We’ll have to wait and see what the New York Department of Financial Services finds out.
Speaking of which, I’m not sure how I feel about regulatory investigations being triggered by same-day Twitter rants that go viral.
Also, making subsequent credit risk decisions in response to those same Twitter rants (instead of your established policies) isn’t the strongest statement in support of those policies. From the same Bloomberg article:
As soon as this became a PR issue, they immediately bumped up her credit limit without asking for any additional documentation,
And a slightly longer take:
In all the furor over this, a simple fact is getting a bit lost in the shuffle — neither Apple nor the applicant benefited from this outcome.
Information asymmetry in most cases leads to less efficient markets in the long run, but usually it also produces an advantage for either the seller (e.g. used car salesperson) or the buyer (e.g. life insurance applicant) in the short-term.
What’s interesting to me is that, in credit decisioning, there is no short-term advantage in asymmetric information for either the lender or the borrower. To return to our example, let’s assume that Hansson’s wife’s true credit worthiness justifies a significantly higher credit limit than the one she got. It’s not like Apple and Goldman Sachs wouldn’t have wanted to give it to her. The bigger a credit line that she can responsibly handle, the more money Apple/GS stand to make.
This is the good news — there is no rational economic reason for lenders like Apple/GS to give applicants less credit than they deserve.
However, this incident also highlights the bad news — despite the economic incentive, many applicants still seem to be getting less credit than they deserve (check out the replies to the original tweet thread for plenty of anecdotal data).
Bottom line — we have more work to do to close the information gap in credit decisioning.
Upstart Network, a fintech lender based in California, received a “no-action” letter from the CFPB, which gave them the regulatory green light to continue testing alternative credit data (as long as they reported their results to the agency). So far, so good (according to an article from @kateberry1):
This reported expansion of credit access reflected in the results provided occurs across all tested race, ethnicity, and sex segments resulting in the tested model increasing acceptance rates by 23-29% and decreasing average APRs by 15-17%
What types of alternative credit data you ask? Lenders and data providers are kicking the tires on a wide range of possibilities. One the most promising in the short-term is utility and phone bill payment data (according to reporting from @AAndriotis):
Equifax, one of the largest U.S. credit-reporting firms, is partnering with Urjanet Inc., a data aggregator that receives payment information from roughly 6,500 utility, phone and other companies. By around early next year, banks and other lenders will be able to ask consumers if they want to supplement their Equifax credit report with this data.
FinRegLab, which is the first to do an independent study on the idea, found that the use of cash-flow predicts loan performance, helps provide access to credit to borrowers who ordinarily wouldn’t be eligible for it (for instance, if they had no credit bureau file), and appears to meet the requirements of fair-lending rules.
More fun tidbits from this Apple Card meshugas:
I love this line from a CNN story:
In a response to a request for comment on this story, an Apple spokesperson directed CNN Business to Goldman Sachs.
Apple Card. Created by Apple, Not a Bank (unless there’s a problem, in which case it was totally created by a bank, we don’t know anything about it).
From under the bus, here’s Goldman Sachs’ statement on the issue:
@AaronSuplizio has been great on this all weekend (follow him on Twitter if you don’t already):Even if we eventually land on the fact that ECOA/Reg B compliant credit policies don’t discriminated based on gender, we still need to have a conversation about what needs to change. Here are a few good reads: 1/ HT@dhh @AppleCard It’s tough to address your specific concerns substantively without knowing more about your & your wife’s particular situation as well as GS’ underwriting process, but this may shed a bit of light on why women often get short shrift w/ credit limits: https://t.co/CxgDUYe3rRPeter @peter
And this @i2pi thread on the broader challenges (and opportunities) in risk management is worth your time as well:Credit risk programs are very highly regulated and frequently tested (as best as possible...) for disparate impact / bias against protected classes. However, there are few regs that prevent them from being outright stupid. 1/nThe @AppleCard is such a fucking sexist program. My wife and I filed joint tax returns, live in a community-property state, and have been married for a long time. Yet Apple’s black box algorithm thinks I deserve 20x the credit limit she does. No appeals work.DHH @dhh