 | |||||||||||
| |||||||||||
| | |||||||||||
 Happy Wednesday, Fintech Listeners! I hope your week is going well and your hard work has been paying off! Loosely related — my wife recently informed me that the term “elbow grease” literally refers to the sweat that collects in the crook of your elbow when using your arm for intense manual labor. It was first used as an idiom by English metaphysical poet Andrew Marvell in 1672. The more you know! — Alex Was this email forwarded to you? Sponsored by Method Financial Most personal finance apps stop at advice. 3 BIG IDEAS FROM THE PODCAST  This week on Facing Credit, Dave Wasik is back! He’s a Partner at 2nd Order Solutions, a deep well of expertise on all things credit risk, and one of my favorite people to nerd out with on data. Using 2nd Order Solutions’ latest research as our compass, the headline is: things are surprisingly OK, although there are plenty of yellow flags. We get into what I'll admit became a surprisingly philosophical conversation about consumer sentiment (Kyla Scanlon's term "vibesession" has never felt more apt). Plus, we close with a game Dave proposed that I love: The Non-AI Draft. We each picked two trends in credit and lending that would dominate every conversation if AI weren't the only thing anyone can talk about. It’s one of my favorite Facing Credit conversations to date. Tune in for the full conversation here And read below for my three big ideas... #1: Borrowing TimeDave’s research shows that no single category of consumer lending products has gone off a cliff. Lenders are still lending. And yet three data points taken together, are harder to dismiss than any one of them alone. Data point 1: Bankruptcies are up 14% year over year. That stat alone is notable, but Dave flagged a seasonal pattern underneath it. Bankruptcy filings usually follow a predictable rhythm: courts slow during the holidays, consumers make one last spending run, and filings tend to tick up in February and March. Yet this year, the pattern broke. Bankruptcy filings for February, March, and April all came in much, much higher than what we've seen historically. Dave's interpretation? Consumers didn't make bad decisions. They tried every workaround for higher prices on essential items and rising levels of indebtedness, but they ran out of room. Data point 2: Credit cards as asset class looks stable, but cards originated in Q1 and Q2 2025 are already delinquent at higher rates than the 2023 and 2024 cohorts, and it's getting worse quarter on quarter. In credit, early vintage performance is highly predictive. If they start bad, they stay bad. It could be that these customers are one tier above bankruptcy (in financial stress rather than distress), and most likely taking on new credit to buy themselves a little more time. And lastly, data point 3: Subprime auto delinquencies are at 20-year highs. Prime and super-prime auto are holding. But on the subprime side, delinquencies are rising and loan terms are stretching simultaneously, with more borrowers taking seven-year loans instead of five. A car loses value faster than a seven-year loan gets paid down. So the borrower ends up owing more than the car is worth, sometimes for years. This negative equity problem in subprime auto is not new, but the fact that more borrowers are stretching into longer terms just to make the monthly payment work tells you something about how thin the margins have gotten. None of these three data points form a crisis on their own. But now that bankruptcies have ticked up, new card vintages are underperforming right out of the gate, and subprime borrowers have stretched into auto loan terms they can barely service … well, that's not a coincidence. That's a system with less room for error than it had a year ago; one that shows up everywhere, a little, all at once. #2: The Sentiment GapCredit data tells you where consumers have been; consumer sentiment tells you where they're headed. Speaking of, the University of Michigan’s Index of Consumer Sentiment just hit its worst reading in the survey's long history. Perhaps unsurprisingly, though, the survey pointed to the gap between how consumers feel about their situation right now versus how they feel about where things are headed. Current sentiment is bad, but future sentiment is significantly worse. Here's the wrinkle Dave raised: consumer confidence has historically been an uncanny predictor of recession. But it has also become politicized. When a Democrat is in office, Republicans report low confidence about the economy, and vice versa. So the signal is noisier than it used to be, and the fact that confidence has been depressed for years without a recession materializing is significant. That said, Dave's framing of the K-shaped economy is something I keep returning to. Consumer sentiment is not uniform across income levels. The average American consumer is absolutely under strain, but the top 10-20% of earners are still in relatively ok shape, which is a segment that accounts for roughly 50-60% of consumer spending. Their confidence is more consequential for the overall economy than everyone else's combined. Meaning, if deterioration starts to show in that upper 10-20% branch of earners, then our current yellow flags would change to red. I drew a comparison to the 1970s, the only other period I could find in the survey's history with a similar gap between economic performance and consumer sentiment. There are real parallels: inflation, macro shocks, instability. But there's a key difference that I think matters for lenders specifically. In the 1970s, the personal savings rate was high. People were trying to build a buffer and largely succeeding, even under Regulation Q, which capped the interest rates banks could pay on deposits and pushed consumers toward money market funds, gold, and silver as stores of value. In our current moment, the impulse to save is present, but savings … just aren’t. In fact, the personal savings rate is near historic lows. Consumers say they want to save but can't find the spare dollars to set aside, despite having more and better savings options than at any prior point in the survey's history. #3: What Replaces Disparate Impact?On April 23 2025, President Trump signed an executive order instructing federal agencies to repeal or amend rules that rely on disparate-impact theory, including the implementing regulation for the Equal Credit Opportunity Act. Disparate impact is the legal principle covering unintentional discrimination in lending. The theory is elegant: even if a lender has no intent to discriminate, if their model or process produces discriminatory outcomes across protected classes, that's a violation. Under the Biden-era CFPB, Rohit Chopra made enforcement a priority. Dave's assessment (and I agree with him) was that enforcement overreached into something approaching "guilty until proven innocent," where lenders had to prove they weren't discriminating rather than enforcement agencies having to prove they were. That dynamic was frustrating for lenders who were committed to fair outcomes but found the standards unattainable. The overcorrection created the political conditions for the executive order, but that doesn't make the executive order a good idea. The order applies only to federal agencies, which means the states can fill the void, and they will. Dave noted that Chopra has already re-emerged, this time alongside Gavin Newsom, establishing a new business and consumer protection agency in California. Lenders who expected relief from federal rollback are now navigating a patchwork of state-level requirements that are inconsistent with each other and with whatever’s left of federal guidance. Dave's larger point is one I agree with. The problem with disparate impact enforcement was always the implementation, not the concept itself. The statistical methods used to infer race and ethnicity in lending data, because we don't collect it directly, are flawed. Fixing the methodology would’ve been the proportionate response; eliminating the requirement is not. And the timing is particularly strange. As models become more complex and less interpretable at the input level, examining outputs for discriminatory effects is precisely the tool credit risk professionals need most. Why we chose to remove it at the moment it was about to become more necessary, not less, beats me. Sponsored by Oscilar Regulators don't want to know what your agent decided; they want an audit trail before they ask. WHAT I'M LISTENING TO #1: Why Embedded Payments is a Retention Strategy for Vertical SaaS (Fintech One-on-One) 🎧I’ve been intrigued with Rainforest and the idea of infrastructure built specifically for vertical SaaS for a while now, so this was an interesting conversation. #2: How Modern Fatherhood is Changing Men’s Brains (Plain English) 🎧One of the most consequential changes in society over the last 60+ years has been the amount of time that fathers spend with their children. This is obviously a trend that I have a personal interest in, so I found this conversation particularly interesting! Thanks for the read! Let me know what you thought by replying back to this email. — Alex | |||||||||||
|
{beacon}
Workweek Newsletter
