Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
It’s Monday. Somebody’s won a T-shirt.
Here’s what you were tasked with identifying last week:
That’s the share price of Ryanair, which remained high enough for long enough that swear king Michael O’Leary’s unlocked Wirtz money.
That’s the exchange rate of the US dollar and the Swedish krona.
And that’s the market cap of NatWest, née Royal Bank of Scotland, which is now fully out of public hands.
These weren’t the hardest charts, and many of you guessed them. Specifically, the following people: Conor Murtagh, James Klikis, Henry Yates, Ben O’Dwyer, Erik Meyer, Rory Boath, Ethan Levine, Ed Roe, Jack Hodgkinson, George Boyd-Bowman, James Memon, Anthony Cheng, Alex Hymers, Sean Lightbown, Scott McNab, Thomas Fitzgerald, Ziyodulla Abdullaev, Gerco Goote, Matt Dawson, Oliver Spiby, Olly Wisking, Eden Gray, James Phillips, Rhett Wahlsten, Ignacio Navero, and Henri Besse de Laromiguière.
To the wheel:
Matt Dawson from Quilter Cheviot, you’re a winner. Everyone else, better luck this Friday (when Bryce will be in charge).
A New York Times article on how Americans have embraced buy-now-pay-later loans caused a minor online stir this week, with the sentence highlighted below causing some alarm:
“Food prices have skyrocketed,” said Mrs. Hodge, of Austell, Ga., who plans how much she’ll spend on each trip to the grocery store based on her cash flow and other expenses that month, including credit-card debt and student loans. Being able to spread out the payments for groceries has helped her family of four — soon to be five — budget better, she said.
Mrs. Hodge, 29, is hardly alone. Nearly a quarter of consumers using buy now, pay later loans finance groceries, up from 14 percent a year ago, according to a recent LendingTree survey. And it’s not just groceries; more Americans are using these loans to pay for recurring monthly bills, such as electricity, heat, internet and streaming services like Hulu.
Some people seemed to read this to mean that almost a quarter of all adult Americans are using BNPL loans to finance their grocery shopping, rather than that a quarter of BNPL loans are now used for groceries.
But even this is a little disconcerting. Given that the traditional FICO credit scores don’t pick up BNPL debts, the rising prevalence of BNPL even for small ticket items could be obscuring signs of consumer stress that would otherwise be picked up on by investors. We’ve already been making jokes about “burrito-now-pay-later”, so it made Alphaville wonder how alarmed we really should be.
Unfortunately, LendingTree’s website is glitching for us — it seems to have identified Alphaville’s curiosity as a DDOS attack and has blocked us — so we weren’t able to check out its survey’s methodology and raw data.
Fortunately, Morgan Stanley’s research team has been digging into the same subject, and found a similar if more nuanced picture.
The investment bank’s “AlphaWise” data science team ran its own survey in April, which showed that 28 per cent of Americans it polled had a BNPL loan. That’s shockingly high — at least to us — but these spanned the gamut from groceries and bigger-ticket items like pricey electronics. The most common items reportedly purchased through BNPL were clothes and shoes.
As you’d expect, the younger you are the more likely you are to use BNPL. However, Morgan Stanley found that wealthier people are more likely to have a BNPL loan than poorer ones. This probably isn’t what you’d expect, and runs counter to previous research by the NY Fed.
Alphaville suspects that this is because the well-heeled will tend to be more financially sophisticated, and simply treat BNPL as a short term zero-interest loan (which they often are, as long as you make all the payments on time).
With CD interest rates of about 4.5 per cent, this makes sense (even if it feels very late-stage-capitalism that wealthy Americans are possibly in practice getting free loans to buy air fryers thanks to poorer ones getting shafted on late fees).
Morgan Stanley’s analysts note that there does seem to be an increasing willingness — or need — to use BNPL for smaller everyday items like food. About 30 per cent of survey respondents that reported having used BNPL had used them for groceries, even higher than the LendingTree survey’s reading.
Moreover, data from asset-backed securities issued by Affirm — one of the big players in the BNPL space — indicates that the average loan size has decreased steadily from an average of $760 in 2020 to $372 in 2025.
This is obviously not ideal, as Morgan Stanley writes:
Increased use of BNPL on everyday items could be a sign of greater market penetration and outreach from the companies, but it also could be a sign of increased consumer stress. If BNPL usage were to grow rapidly later this year, when we expect consumers to be more stretched due to elevated inflation from tariffs and slow income growth, we would potentially take that as a warning sign of the latter.
However, there are other elements that make the bank’s analysts relatively relaxed, at least in the short term.
Firstly, at least some BNPL companies are working on integrating their data into credit reports, and over time most of them probably will. That should at least ameliorate the hidden-debts issue.
Secondly, BNPL remains tiny compared to credit card debt or student loans. A report by PYMTS last month estimated that the size of the US BNPL market had hit $175bn, but credit card balances stand at $1.18tn, car loans at $1.64tn, and student loans at $1.63tn at the end of the first quarter, according to the NY Fed.
Thirdly, the delinquency and default rates on the BNPL loans contained by Affirm’s ABSs are actually lower than in other unsecured loan ABS. This is probably mostly because of the extremely short-term nature of most BNPL loans, but it implies that the credit quality probably isn’t quite as shocking as commonly assumed.
Alphaville would also add that while Klarna last month reporting a 17 per cent year-on-year jump in credit impairments to $136mn is obviously not great, as a percentage of its overall payment volumes this only translated into an uptick from 0.51 per cent to 0.54 per cent.
Morgan Stanley therefore argues that “we do not see this as large enough yet to be a macro risk from an economic standpoint”.
However, that doesn’t mean that the BNPL industry itself isn’t heading for trouble. As the bank’s analysts conclude:
We expect consumer spending to slow this year and consumer loan performance to weaken as policy changes dampen income and wealth growth. Student loans pose another risk, and this risk is likely more pronounced for unsecured and BNPL compared to other types of consumer debt. The Trump administration announced last month that it would start to collect on defaulted student loans for the first time in five years. Meanwhile, consumer credit scores are taking hits from late student loan payments for the first time in five years as well.
We estimate that 10-15mn borrowers are late or defaulted on their federal student loans. If these borrowers start to pay again, this could be another ~$300-400 monthly payment (on average) that could eat at their ability to spend and to pay other debts. In our 2023 survey, we also found that BNPL had the most overlap in borrower base with student loans. At that time, 34% of consumers in our survey with BNPL loans also had federal student loans. This makes sense considering both debt types are concentrated in younger borrowers. Debts such as auto loans and mortgages are likely further up consumers’ payment priority, so we expect unsecured loans and BNPL to face the most risk from resumption of federal student loan payments.
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Mass surveillance. Pandemic tracking. Predictive policing. Insurrection detection. Drug testing. Political lobbying. AI deployment. NHS restructuring. Employee snooping. Vaccine rollouts. Spac investment. Referencing Tolkien. It’s hard to avoid priors with Palantir. To do no more than list its activities can make a person sound like a crank and/or co-founder.
That’s why a recent note from Trivariate Research caught our attention. A boutique founded by former hedge fund manager and Morgan Stanley’s chief US equity strategist Adam Parker, Trivariate rarely does single stocks and when it does, it just does data.
What most interests Parker is not what Palantir does, it’s how much it costs. According to his calculations, Palantir is one of the most expensive US large-cap stocks ever to exist.
Palantir’s $314bn market capitalisation makes it a top-30 constituent of the S&P 500, between Coca-Cola and Bank of America. Its PE ratios look nuts enough — 565x trailing, 228x forward — but they’re as nothing compared with sales multiples.
According to a Trivariate screen of 2000 established US non-financial stocks that goes back to the start of the century, only six have traded at a higher EV-to-forecast-sales ratio than Palantir does today:
Some content could not load. Check your internet connection or browser settings.
Five of the 15 stocks listed above are moonshot biotechs and one is MicroStrategy, which is in its own category of weird. Of those to peak before 2024, not many were good long-term investments.
Comverse Technology was dead by 2013 and its CEO went to jail in 2017, having been on the run for more than a decade. Bluebird Bio was taken private this week for less than a fifth of its 2019 valuation. Moderna is down 94 per cent from its mid-pandemic high. FuelCell Energy’s down 99 per cent in four years, as is Plug Power. Cheniere Energy, an LNG exporter, halved in 2015 then took five years to reclaim the level.
The broader point to make is that enterprise values of 70 times forecasted sales are never normal. Most years, the number of established US stocks hitting just 30 times sales is zero.
Toppy forward valuations only really happened during the dotcom boom and the free-money pandemic era, which makes last year an outlier: Soundhound AI and Astera Labs both joined the >30x sales club in December (and had lost 60 per cent of their value by early April).
Some of these examples are extreme outliers, but stocks on Trivariate’s screen that hit 30 times forecasted sales have underperformed the S&P 500 on average by 22.5 per cent over the next year and ended the period trading at just 18 times sales.
The derating is partly due to multiple compression and partly due to downgraded forecasts. Companies enter the bubble zone after growing annual revenue by 45 per cent on average. Analysts expect them to repeat the trick and are disappointed: year-two average sales growth slows to just 28 per cent.
To find anything comparable you need to go back to the year 2000, when the makers of network widgets and software routinely traded at three-digit sales multiples.
Using trailing revenue rather than forecasts, Parker and team list the top-five most expensive stocks of the dotcom bubble as Ariba, Brocade, Juniper, Versign and Tibco Software. Things didn’t work out great for them:
Palantir resembles a one-stock bubble. It’s over three times more expensive than the next most expensive company (excluding Strategy) and is forecast to deliver sales growth that’s unprecedented for a company of its size, yet it is no better than what’s available elsewhere for much cheaper:
Parker and team note that the end-June S&P 500 rebalance will raise Palantir’s index weighting over the large-cap threshold, and argue that this will compel active managers to apply some sanity checks:
Palantir may be fundamentally awesome, we don’t know. But we do know that no company has grown fast enough to justify this valuation, and that many companies are forecasted to grow faster according to consensus expectations. For large cap portfolio managers who are going to do new work on this before it hits their index, it’s hard to imagine this looks attractive.
That’s one take. Other analysis is available:
EXCLUSIVE: Alex Jones Responds To The Palantir Hysteria & Lays Out What’s Really Happening pic.twitter.com/dHUvPwH58k
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Here’s a paper:
We examine how selection bias in CEO promotion amplifies risk-taking, using prenatal exposure to pollution as an exogenous shock to individual risk preferences.
The idea is that reckless CEOs are born that way. The question it seeks to answer relates to whether companies with a high-risk-tolerance culture tend to choose a CEO with a similar character, or whether companies unintentionally promote risk-takers based on their luck:
If promotion systems reward executives whose risky decisions happened to succeed in low-stakes environments, firms may unknowingly elevate individuals whose leadership style reflects survivorship bias rather than skill. This concern is especially relevant for internal candidates, who account for over 70 per cent of CEO appointments
The study’s authors — P. Raghavendra Rau, YiLin Wu and Richard Lok-Si Ieong — choose as their measure the “biological shock” of prenatal exposure to pollution. It’s a different approach to previous research on risk tolerance and early life, which has concentrated on after the CEO is born.
Children of the Great Depression, for example, show heightened financial risk aversion on becoming CEOs except when their families were rich enough to shield them from hardships. Children exposed to fatal natural disasters in childhood become risk-taking CEOs, but that might be because they were raised by a family that chose to live in a disaster-prone region. Having rich or reckless parents will affect a person’s career path in countless hard-to-measure ways, so it’s hard to unpick behavioural effects from corporate selection effects. Taking it all the way back to conception removes these complicating factors, the authors argue.
There’s a good amount of evidence that if a baby is exposed to toxic chemicals in the womb, they’ll turn out more impulsive and aggressive, more prone to ADHD, and at a higher risk of adult criminality.
The paper’s method is to estimate a CEO’s likelihood to develop these traits based on whether they were born near US industrial pollution blackspots known as Superfund sites, which they grade by toxicity at the date of birth. There were more than 1800 such sites in the US as of December 2018.
Being born near a hazardous waste site makes a person less likely to become a CEO, the study finds, which might be for biological or economic reasons. However, being exposed to pollution makes a person more likely to be promoted to CEO.
Nearly all of the CEOs included in the dataset were born before such environmental dangers were widely recognised, so the data offers a clean-ish measure of innate risk appetite, the authors contend. The lines on the chart below mark the publication of Rachel Carson’s influential exposé of the chemicals industry, Silent Spring; the formation of the US Environmental Protection Agency; and the proposal of the Superfund Act that passed into law in 1980:
The study also finds that so-called Superfund CEOs are not disproportionately represented in high-risk sectors. The employees who rise to the top job in Superfund counties will tend to have been hired locally, whereas non-Superfund CEOs are more likely to be external appointments.
From this, the authors make some inferences about how prenatal toxic load affects management style:
Non-Superfund CEOs may advance through consistent, low-variance performance. Superfund CEOs, by contrast, are more likely to take high-variance internal risks — such as aggressive restructuring — that sometimes generate standout outcomes. When these gambles pay off, they are difficult to distinguish from skill. Firms, observing only results, may systematically promote risk-takers whose behavior is only revealed under the high-stakes pressure of exposed decision-making. This survivorship-driven selection bias offers a novel explanation for why some CEOs perform well in internal roles but falter once promoted.
Empirically, we find that Superfund CEOs are significantly more likely to be promoted internally rather than hired externally, suggesting that firms place considerable weight on prior internal success. Once in office, these CEOs pursue significantly riskier external financial policies — higher leverage, lower cash holdings, and more unrelated acquisitions. Their firms exhibit greater stock return volatility, lower credit ratings, and higher borrowing costs.
There’s quite a bit of guesswork underlying this claim. The study does not directly observe an executive’s decision-making before they became boss. The Peter principle is a convincing explanation, not a demonstrable trend.
Instead, the study measures consequences. A company whose CEO was born in a county with one Superfund site will add nearly 19 per cent to its bankruptcy risk score and make default more than 7 per cent more likely, it finds. Buy-and-hold stock returns are slightly worse, but probably not by enough to turn into a factor.
There is, however, also an unusually high number of Superfund CEOs leading companies on Fortune’s America’s Most Admired Companies top 50. It’d make sense that a cohort with a higher propensity to gamble will result in a few winners and a lot of losers.
The paper concludes:
Our findings help explain why firms may systematically overpromote risk-takers whose internal success reflects luck, not skill. In doing so, we show how performance-based promotion systems can inadvertently channel high-risk individuals into positions where their traits are misaligned with the demands of the role — creating a structural source of firm-level risk that is both hidden and difficult to correct.
Whatever you make of the argument, the dataset does throw up one odd quirk. America’s most toxic county, with 23 active Superfund sites, is Santa Clara County in California. Santa Clara is also where nearly all US tech companies have their headquarters.
Silicon Valley, the spiritual home of risk-taking and creative destruction, was built around the toxic legacies of local operators including Intel, HP, Applied Materials and National Semiconductor. We’re not saying there’s any connection. It’d be very wrong to suggest that US tech management culture is a byproduct of poisoning unborn babies mildly enough that they become CEOs and run companies into bankruptcy. That’d be ridiculous, reductive, and borderline offensive. All we’re doing is noting the correlation.
Further reading: — Narcissist CEOs like risky insider trading and are bad at it: study (FTAV)
— I used a Magic Eight-Ball as a life coach and it told me I was handsome but struggled with follow-up questions so I’m now questioning whether it’s really magic (Everything Is A Wave)
— Explaining overnight returns in the US (Klement on Investing)
— It takes money to make money (Lewis Enterprises)
— Back to basis: why the treasury basis trade is in focus (JPMorgan 🔊)
— What Happens If the United States Leaves the WTO? (Cato)
— Japan’s debt trap (Robin J. Brooks)
— Feudalism is our future (The Atlantic)
— A shark tale (The Fence)
— My AI Skeptic Friends Are All Nuts (Fly.io)
— For further further reading, Today In Tabs is back
Simply sign up to the UK inflation myFT Digest — delivered directly to your inbox.
A spectre is haunting the City of London — the spectre of VED.
Two weeks ago, Britain’s latest inflation statistics landed with a nasty shock, although which part was shocking was a matter of perspective.
Headline consumer prices came in 3.5 per cent higher year-on-year, a leap from 2.6pc that set social media managers’ mouths frothing.
But for the sell- and buy-siders — who were braced for a major uptick — the real shock lurked in the subcomponents.
Everyone knows April is the cruellest month, and that’s especially true for UK economic analysts, who are forced to make sense of a tricky assortment of tax changes, prices hikes, and the effects of Easter.
This time around, lots of attention was focused on energy, sewerage, and Vehicle Excise Duty. Two weeks on from the print, the last of those is still causing consternation.
VED is a complicated tax placed on every vehicle that uses the UK’s public roads. Various change to levy came into effect place last month as part one of its occasional overhauls:
First-year VED rates were doubled for most vehicles, with the level of charge now closely-linked to emissions.
Electric vehicles lost their VED exemptions, although new ones now pay a marginal first-year rate of £10
VED rates on hybrids got hiked
Overall, these made it hard predict the VED shift. But people are paid a lot of money to do that kind of thing, so they did.
Expectations across the street were pretty wide going in, reflecting significant uncertainty about the impact of the change. Based on conversations and our own inbox, predictions ranged from the mid single figures to the mid-teens for the monthly VED per cent change.
The ONS, we’re told, weren’t a lot of help. “The uncertainty was not made any easier since the ONS was very unresponsive around questions to the methodology,” Lucas Krishan, an analyst at Taula Capital Management, told FTAV:
A bit more of a back and forth would have been very helpful, ex ante, in gaining some certainty around what was likely going to happen.
Morgan Stanley’s preview said “we see large two-sided risks”, which is sell-sidese for ¯\_(ツ)_/¯.
Still, there was significant bamboozlement when Office for National Statistics reported a 26 per cent month-on-month increase.
Goldman Sachs’ James Moberly — who had called for 13.7 per cent jump — told clients “the rise was much larger than we had anticipated”.
Robert Wood of Pantheon Macroeconomics, who had predicted 16 per cent — said the gap was worth nearly 10 basis points on headline inflation, which is a lot amount of inflation if you’re trading the print. He told FTAV:
It could be a storm in a teacup, but it is the sort of thing that can shift markets a lot, and there is potential for something odd having happened here.
MS’s Bruna Skarica, who said “forecasting how the ONS would capture the VED reform was near-impossible”, wrote in a note (our emphasis):
It is not an exaggeration to say that the April inflation print is one of the most important data releases of the year in the UK. For three years now, it has surprised consensus meaningfully to the upside, although the drivers of the beat did vary. [This year] we think that consensus was caught out primarily by the strength in the ONS’ measure of the VED tax hike (our sense was that most analysts worked with an assumption of ~6-15%, where the actual figure came in at 26%), package holidays and air fares.
So how much of the April strength is likely to reverse in May, and what is the implication of the strong VED number? On this latter point, on our estimates, the VED hike added ~40bp to the uptick in headline services inflation today (~20-30bp more than we think consensus anticipated, and that we think seems plausible based on historical weights of vehicles on which inflation calculations are based). VED is car tax, with rates adjusted just once a year. Only in April 2026 will this boost to headline services inflation peter out from the numbers.
Krishan added:
So far, everyone I have talked to — that does not have an easily disprovable framework — also can’t make sense of the ONS’s VED number. I originally thought that either I would find the error I made or that I would find someone that managed to make sense of this, but we’re all confused about this still.
Now, clearly there’s a possibility that much of the City was just caught slipping this time around — although we see the argument that the ONS should be giving a clear steer on its methodologies and sources in advance.
But if the figure is wrong — which seems at least plausible — then it’s going to result in misleadingly-elevated services inflation for the next year, which isn’t much help for anyone setting interest rates, or force the ONS to issue a correction.
Asked about the VED number, a spokesperson for the ONS said it never speculates on the potential for revisions or corrections in any of the office’s statistics. The ONS only revises CPI and RPI numbers in exceptional circumstances, they added.
Further reading: — The ONS vs the Xbox — It’s possible that Pink broke UK hotel inflation. Has the ONS fixed it?
Your guide to what Trump’s second term means for Washington, business and the world
It’s Robert Armstrong’s world, and we’re just living in it. Borg:
President Donald Trump bristled at suggestions that Wall Street believes he’s ultimately unwilling to follow through on extreme tariff threats, saying his repeated retreats are instead part of a strategy to exert trade concessions.
“It’s called negotiation,” Trump said on Wednesday, adding that he intentionally would “set a number at a ridiculous high number” and then “go down a little bit” as part of talks.
Trump was asked during an Oval Office event to respond to reports of a so-called “TACO” trade, in which investors seize on market tumbles after the president makes tariff threats, predicting he will ultimately relent and equities will rebound.
The acronym — which stands for Trump Always Chickens Out — was coined by a Financial Times columnist and has since been adopted by traders attempting to navigate the dozens of changes to tariff policy Trump has announced over the early months of his presidency.
Inspired by Rob’s huge success in minting the coinage, FT Alphaville has tried to get in on the action by coining several of its own Mexican-inspired market adage acronyms.
MOLE
Macroeconomists Only List Events
QUESADILLA
Quants Used Excel Systematically And Don’t IRR Levels Look Appealing?
