Author: business

In the 18th and 19th centuries, a chess-playing automaton dubbed the Mechanical Turk mesmerised Europe’s ruling classes.

For the best part of a century, the Turk toured the continent’s palaces and salons, checkmating a roster of illustrious opponents that included Napoleon Bonaparte and Benjamin Franklin.

Two centuries before IBM’s Deep Blue bested Garry Kasparov across a gruelling six-match series, this robotic chess player put a slew of revered grandmasters through their paces, to the delight of well-heeled spectators.

But it was a lie.

Rather than a proto-form of artificial intelligence, the Mechanical Turk was nothing more than a sophisticated parlour trick. A human sequestered inside a cabinet operated the machine, with a succession of chess masters secretly powering the Turk to victory over the years.

In the telling of some commentators, the recently bankrupt Builder.ai bears the hallmarks of a modern-day Mechanical Turk.

Armed with funding from the likes of Microsoft, OpenAI’s chief backer, Builder.ai promised through the magic of AI to make the process of building apps and websites “as easy as ordering a pizza”. Instead, the tech company once valued at well over $1bn collapsed last month. It’s the first major corporate scandal of the AI boom.

Over on social media slop-swamp X.com, posts confidently asserting that the tech firm imploded after revelations that its AI was really just “700 engineers in India” have been widely shared (the specific claim around 700 Indian programmers seems to have originated in this post from a Swiss crypto-enthusiast called Bernhard Engelbrecht with no notable record of tech reporting).

In reality, however, Builder.ai’s extensive use of developers was no secret.

Sachin Dev Duggal, Builder.ai’s founder and self-proclaimed “chief wizard”, was upfront about the fact that his business made use of what he called “human-assisted AI”. And way back in 2019, when the company was still called Engineer.ai, the Wall Street Journal cast a sceptical eye at just how much of its coding work was attributable to AI, alleging “the company relies on human engineers in India and elsewhere to do most of that work” (Duggal’s company rebranded to Builder.ai after the WSJ’s exposé).

Illustrious backers including the Qatar Investment Authority and Insight Partners happily poured hundreds of millions of dollars into the business later on, apparently with full knowledge that human developers were doing at least some of the heavy lifting.

While serious question marks remain around the efficacy of Builder.ai’s artificial intelligence, the revelations that really prompted its nosedive into insolvency were instead around allegations of artificial revenue.

A run of stories from the Financial Times and Bloomberg has laid bare how Builder.ai imploded shortly after informing its backers that it had overstated sales by as much as four times. Critically, an internal investigation found evidence of potentially bogus sales, with the probe raising concerns around the legitimacy of sales booked through third-party intermediaries who sold Builder.ai’s products.

This also seems to be the tip of the iceberg. From mainFT’s recent dissection of the suspected revenue inflation techniques:

Now, interviews with former employees and documents seen by the Financial Times suggest that Builder.ai is suspected of using a broader range of methods to inflate some of its revenues, including improperly booked discounts, tiny upfront deposits and seemingly circular transactions with key customers.

“It’s an ocean,” said one former employee, referring to the breadth and depth of questionable revenue recognition practices in which the firm allegedly engaged. 

(Lawyers acting for Duggal, who stood down as CEO in February, said there were “serious inaccuracies” in the points the FT had put to him for comment.)

Builder.ai’s problems also stemmed from a more prosaic source common to any bankruptcy: debt.

On top of a $50mn loan from a syndicate of tech-focused lending firms — which promptly called a default after learning of the scale of the planned revenue restatements — Builder.ai owed even more money to operational creditors such as cloud providers.

Amazon Web Services alone was owed $88mn, while the AI firm had also racked up an unpaid bill with Microsoft in the region of $30mn (AWS actually filed a bankruptcy petition against Builder.ai in Indian courts back in January over its outstanding bill).

AWS and Microsoft both feature on the list of creditors Builder.ai’s main US holding company disclosed when filing for Chapter 7 in Delaware last week. The list also comprises lots of names you might expect to feature in the bankruptcy filings of a delinquent tech company, such as “Google Cloud EMEA Limited” and “Figma, Inc”.

Others are more eyebrow-raising. From mainFT again:

The creditor list also included: Tel Aviv-based private intelligence firm Shibumi Strategy; top US litigation law firm Quinn Emanuel; and Sitrick Group, a Los Angeles-based public relations firm specialising in so-called “crisis communications”.

All three firms were hired after the Financial Times reported last year that Builder.ai’s co-founders, including its chief executive Sachin Dev Duggal, were embroiled in criminal investigations in India, according to people with direct knowledge of the matter. 

Blimey.

If you were wondering how common it is for tech start-ups to turn to corporate spies and spinners when confronted with negative newspaper coverage, a senior former employee of Builder.ai assured the FT that “working with international professional advisers is perfectly normal practice for a successful billion-dollar technology company operating in multiple jurisdictions”.

For Alphaville readers who are blissfully unaware of the precise contours of what can broadly be termed the “reputational management industry”, here’s an introduction to the trio of specialists that Builder.ai brought on board after the FT began scrutinising the company last year.

First up: Quinn Emanuel. Founded by top American litigator John Quinn, Quinn Emanuel Urquhart & Sullivan LLP is one of the world’s pre-eminent litigation law firms. Entertainingly, its partners’ business cards and emails often carry the strapline: “The most feared law firm in the world”.

To give you an idea of how the firm markets itself, here’s an advert it placed in a certain salmon-pink newspaper a few years ago:

Quinn Emanuel sent a letter to the FT last year on behalf of Builder.ai and Duggal, alleging potential breaches of confidence in the course of the paper’s reporting on the tech company’s customer relations (it should be noted, however, that Quinn Emanuel is just one of several law firms that sent letters to the FT on behalf of Duggal and/or Builder.ai.)

On to Sitrick Group (which trades as Sitrick and Company). The PR firm’s founder Mike Sitrick is one of the pioneers of adversarial “crisis comms”, with this section from the bio on his firm’s website giving a flavour:

Fortune magazine called him “one of the most accomplished practitioners of the dark arts of public relations” and “The Winston Wolf of Public Relations.” “Wolf,” Fortune explained, was the fixer in Pulp Fiction. Played by Harvey Keitel, he washed away assassins’ splatter and gore. Sitrick cleans up the messes of companies, celebrities and others and he’s a strategist who isn’t averse to treating PR as combat.”

Sitrick and Company’s website helpfully breaks down the sorts of thorny issues it can deal with:

intellectual property matters, allegations of stock manipulation, wrongful termination, claims involving contract disputes, allegations of fraud and fraudulent inducement, wrongful death claims, allegations of illegal drug use and a variety of other white-collar crimes, criminal and civil cases against companies and their executives for such things as price fixing, insurance fraud, options backdating and antitrust violations, race and gender discrimination, sexual harassment, racism and even rape. We have done extensive work combating short sellers, handling product recalls, extremely sensitive environmental matters, racketeering cases, executive departures either through termination or otherwise, professional, college and high school sports issues, family disputes, and high-profile divorces.

Like Quinn Emanuel, Mike Sitrick contacted the FT in the run-up to publication of this May 2024 article on Builder.ai, raising concerns around the newspaper’s reporting process.

And finally, we have Shibumi, or “Mossad for hire” as The Times memorably dubbed it last year in an interview with one of its unnamed founders:

Mayfair on a crisp, clear morning. The demure woman I am meeting in the bar of an opulent hotel is dressed elegantly and entirely in black. Her nail polish is black, too. A flash of a gold bracelet when we shake hands hints at money.

To the waiter delivering her glass of still water, my companion might be a high-flying banker or even a minor European royal. Age? She could be 25 or 45. Appearances can be deceptive, and this woman knows all about deception.

For more than ten years she spied for Mossad, conducting undercover operations in more than 20 countries for Israel’s respected external intelligence service. These days, she runs Shibumi Strategy, a Tel Aviv-based agency she set up seven years ago to provide espionage services to businesses and high-net-worth individuals.

Saphia Fenton founded Shibumi in 2017 with fellow Israeli intelligence operative Ori Gur-Ari. In between leaving Mossad and setting up Shibumi, Fenton worked for a stint at the notorious Israeli private intelligence agency Black Cube.

This isn’t the first time Shibumi has graced the pages of the pink ‘un. Back in 2022, the FT revealed that the controversial financier (and longtime Alphaville favourite) Lars Windhorst had employed the Israeli firm to target the president of Hertha Berlin football club in a smear campaign.

The ensuing fallout saw Windhorst sell the club in disgrace, with German weekly Der Spiegel describing the affair as a “scandal that is unparalleled in the history of the Bundesliga”.

In terms of other interesting creditors, there’s at least one other corporate intelligence firm on the list (New York’s T&M USA), but we thought we would throw it open to FT Alphaville readers to see if anyone else piques your interest.

There’s the obvious proviso for any firm that collapsed in a state of chaos that its initial creditor list (which is “based on a review of the Debtor’s books and records”) is likely to be subject to change. Also, the US process will not capture all of the creditors to the wider Builder.ai group around the world.

Builder.ai should also soon file for administration in the UK, with local insolvency filings also expected at its key subsidiaries in countries such as India and Singapore. Other creditors without a claim on the US holding company should emerge in those processes.

With all that said, below is the full list of creditors disclosed in Delaware court last week. (Personal addresses are redacted.) Drop a comment below or emails us on [email protected] or [email protected] if you spot anyone interesting.

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Last year, Reform UK came up with a cunning wheeze to scrap the Bank of England’s ability to pay interest on reserves, saving £35bn. Readers may recall our big explainer on central bank reserve frameworks that kicked the tyres of the idea.

Over in America they do everything bigger. Four minutes and 47 seconds into this CNBC interview, Texas senator Ted Cruz drops his bomb:

The Federal Reserve pays banks interest on reserves. For most of the history of the Fed they never did that. But for a little over a decade they have. Just eliminating that saves a trillion dollars.

A trillion dollars 😮

Is this free money just sitting there for the taking? 

According to Barclays, the Fed paid $176bn in Interest on Reserve Balances and another $104bn in interest expense operating the Overnight Reverse Repo Facility in 2023. This $280bn total expense would — in more normal times — be more than covered by interest income from the asset side of their balance sheet.

But the asset side of their balance sheet consists largely of mortgage-backed securities and US Treasuries purchased when yields were ultra-low. So the interest income from these holdings doesn’t offset the cost of paying reserves. Net-net, the Fed has been running negative net interest income of $80-100bn over the past couple of years.

Let’s not be sniffy about these numbers. One hundred billion dollars is a lot of money. In fact, 10 years of $100bn gets us to a trillion dollars. But, as a new note from Samuel Earl and Anshul Pradhan at Barclays reminds us:

These losses the Fed generates from having interest expenses greater than interest income do not impact the deficit as they are recorded as deferred asset on its balance sheet. However, the deferred asset would need to be paid off before the Fed resumes its remittances to the Treasury.

Of course — the deferred asset.

The UK — somewhat inexplicably — flushes all negative net income on QE, as well as all realised losses attached to QT, through the current year’s fiscal accounts — so passing large fiscal decisions directly to the MPC. The US, in line with loads of other central banks, does not:

As such, it’s not that net interest and valuation losses are funded by the US Treasury. It’s more that seignorage — the profits from running the central bank for what is still the world’s reserve currency — are no longer remitted to the Treasury until all the losses generated by QE and QT have been paid off.

But Cruz is right — before the global financial crisis the Federal Reserve didn’t pay interest on reserves. Surely we can just rewind the clock?

Reserve scarcity

Today there are $3.3tn of central bank reserves in the Federal Reserve system. Back in 2006, there were only $10bn. The Fed required banks to meet minimum reserve requirements (since scrapped), and by fine tuning the supply of reserves in open market operations the Fed could pretty much control the policy rate.

What would happen if the Fed were to return to a scarce reserves framework? Barclays:

The problem with this framework is that it inextricably links the rate policy with the balance sheet policy. That means that the Fed can not provide emergency liquidity without impacting the supply of reserves and thereby potentially requiring a sterilization in order to keep policy at target. 

We think that to get from where we are to where Cruz seems to want us to be would require a very rapid pace of substantial quantitative tightening. Literally trillions of assets would need to be sold, and any realised losses on these sales would of course be added to the Fed’s deferred asset — precisely the thing that is preventing the resumption of seignorage remittances that appears to be Cruz’s end goal.

Furthermore, according to Barclays:

This essentially means the Fed can not run a large balance sheet at will, and would be constrained from providing emergency liquidity lest it loses control of its policy rates.

This all might suit liquidationistas just fine. And there is active support in Congress to simply abolish the Fed altogether. But to us, kneecapping the central bank’s ability to both control policy rates and serve as lender of last resort sounds … bad?

Reform UK are sometimes accused of aping the MAGA movement. Maybe the intellectual traffic is a bit more two-way than we’d appreciated.

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

It’s not every day that we spot a Sean Kingston reference in our inbox, but we’re fairly sure that’s what’s happening here:

Anyone not now whistling “Fire Burning”…

…might well have the same question we had: what’s this all about then?

Well, Citi’s Brazilian pharma team have been investigating a local crime wave. They write:

This week’s Citi Friday Charts analyses the cases of drugstore thefts/robberies in the state of São Paulo, which have continued to accelerate at a fast clip over recent months… we filter for the number of robberies involving GLP-1 products, which have risen by over 4x in the past year

Some people are presumably living small off the proceeds of these crimes, but the phenomenon of drugstore robberies has corporate victims. In its first-quarter report, Brazilian’s largest chemist, Raia Drogasil, blamed store thefts for pinching its margins (our emphasis):

Our gross profit totaled R$ 2,881.3 million in the 1Q25, with a gross margin of 26.6%, a contraction of 0.6 pp when compared to the same period of the previous year. This contraction stems from a sales mix effect, an investment into pricing & promotions, a 0.3 pp pressure from inventory losses, mitigated by a 0.2 pp NPV increase.

The group said it expects a turnaround this quarter as it makes efforts to mitigate losses, which Valor Econômico reports include stock reductions and putting products in acrylic display cases. The financial paper quotes Renato Raduan, RD’s chief executive, as saying:

It might seem minor, but the acrylic cover makes looting harder. You need one hand to lift it and the other to grab the product—it complicates the quick grab-and-hide tactic.

On last month’s post-results call, Raduan added:

And of course, here at an earnings call, I’m not going to give you details of the tactical plan, but I just want you to know that we are not accepting this level with 30 bps in losses. We are putting in place actions to address that.

Citi’s Leandro Bastos and Renan Prata seem satisfied that RD will sort this issue out, although they strike a note of concern about how new security measures may be received:

concerns about a worse in-store experience and/or incremental costs could be undesired side effects.

It’s a moderately interesting insight into the intersection of consumer pressures, pharmaceutical developments, economic agency and crime, in an emerging-market context.

None of this, however, answers the real question here: should Citi be suggesting — as their title does — that Brazilians call 911 if their local chemist is being robbed? After all, the standard emergency number from police in Brazil is 190.

The answer: it’s fine. Calls placed to 911 (and 112, the European emergency services number) in Brazil have directed to 190 since 2013. The change was rolled out in anticipation of an influx of international visitors for the 2014 football World Cup, which Brazil hosted (and where they were memorably demolished by Germany in a semi-final clash known locally as the Agony of Mineirão).

Did all this play on Citi’s minds as they apparently chose to frame this analysis around the United States’ 33rd-most-popular song of 2009? We doubt it. Still, it’s probably more suitable advice than “Just Dance” or “Boom Boom Pow”.

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

In the preamble of last week’s chart quiz we said:

The answers are loosely themed, though knowing that probably won’t help much.

Wrong! Spotting a theme could’ve helped you loads. In fact, for reasons that will become clear, it was probably your only available route to victory.

Answers first:

Chart One is CME live cattle futures, first contract generic. Nearly everyone got this right.

Chart Two is Puma shares. Nearly everyone got this right.

Chart three is Croatian kunas to dollars. Hardly anyone got this right.

A thing we hadn’t realised is that kuna has always traded almost one-for-one with the Danish krone. Why? Absolutely no idea. The only previous mention of the correlation we can find is an old Reddit thread that’s full of people arguing across each other.

Here they are on the same chart:

Yeah, OK. Unreasonably difficult. Sorry.

A person might’ve clocked the significance of the date range, which ends at Croatia’s switch to the euro at the end of 2022. Or they could’ve picked up on the animal theme: cow, puma, kuna (it means marten in Croatian).

Once again, sorry.

In spite of everything, six people got the right answers. Of them, the wheel picked . . .

. . . Luke Ashford, a quantitative researcher at Squarepoint Capital in London. Congratulations to him. Everyone else should come back on Friday for another go.

Morning! After some dabbling in the genre, FT Alphaville is starting an informal series of simple Q&A interviews.

The basic premise is that we’ll have a hopefully interesting chat with hopefully interesting people doing hopefully interesting things in and around finance, economics and business (sometimes they may be boring people doing interesting things, or interesting people doing boring things as well), and publish a lightly-edited Q&A afterwards. No promises, but we’ll try to do one a week.

We’re kicking things off with a conversation with Guiseppe “Gappy” Paleologo, head of quantitative research at Balyasny Asset Management. We have some ideas for future victims subjects, but feel free to make suggestions in the comments below. The transcript has been edited for clarity and length.

FTAV: Hi Gappy. You’re originally a physicist from Rome. How did you end up in finance?

I came to the US to study for my PhD at Stanford. In the last year of it I had a long internship at Enron, and that immunised me from joining finance for the next 10 years. So I went to IBM Research. After the 2008 crisis, with perfect timing, I decided to enter finance, mostly because of financial reasons. I had kids, the usual stuff. It was not a marriage of passion.

But I went to Axioma, which is a factor risk model provider, and was then hired by Citadel. And surprisingly I had a fantastic time at Citadel, contra expectations. So I came to like what I was doing, and have stuck around since then.

What was IBM Research like?

IBM Research was an incredible place for the first 40 years of its life. I had colleagues who spent, as part of their job, a year in Africa tracking the migration of elephants. I don’t even know why. It’s great to be a research lab attached to a near-monopoly.

There was plenty of research funding, you didn’t even need to ask, and you would just be evaluated once a year by your manager. As a result, there was a lot of really good research that was never monetised.

But in the 1990s, IBM started laying off of thousands because of the transition to personal computers, and that was the beginning of the end. We had to start to recover some of our salary through consulting. It became neither fish nor fowl. It was not fundamental research, it was not applied research, and over the years all the good people left for Google, Yahoo or Wall Street.

I found an interesting paper from your time there, with the title The optimality of the online greedy algorithm in carpool and chairman assignment problems.

That one was with much better mathematicians than me. The carpool problem is actually a beautiful problem that is really a dynamical system that arises in a variety of fields. The travelling salesman problem, the carpool problem, the news vendor problem — they are all beautiful names for classes of applied mathematical formulations. You’re not using the travelling salesman problem for sales, but for chip design.

You’ve worked at a lot of hot places, like Citadel, Millennium, Hudson River Trading and now Balyasny. What are their similarities and differences?

The key to understanding differences in hedge funds is the difference in personality of the founders, because hedge funds are incredibly mimetic organisations. There’s an absolute leader, and the people who report to them tend to mimic the leader. Over time they acquire the same tics and personality traits.

Citadel is in general extremely driven. Ken Griffin must be one of the most competitive people on earth. This obsession, this drive, is a constant at Citadel.

Ken is a big believer in technology, and is a persistent learner. So Citadel fails many times, but eventually gets things very right. And this kind of persistence is a beautiful thing.

How does it compare to Millennium?

In the words of another hedge fund manager, Citadel is like Singapore, and Millennium is like the US. Citadel is small and centralised, and Millennium is large and decentralised.

Someone once told me that Millennium achieved diversification by accident. I don’t know if that’s true, but it has become intrinsically diversified and extremely scalable. It has mastered the art of scale — perhaps at the cost of efficiency. That’s the key to understanding Millennium.

Instead of moulding the portfolio manager in the firm’s way — which is what Citadel does — it adapts the firm to the needs of PMs. And this has in some cases paid off spectacularly well.

I’d also add that Izzy [Englander, Millennium’s founder] is a humble person, and he has incredibly good risk instincts. It’s like some kind of Jedi power.

So what is Hudson River Trading like then? Canada?

HRT is a very different animal. It’s fundamentally a technology company. The technologists are true first-class citizens, which cannot be said about any hedge fund I have been at.

Algo engineers are probably a little more elevated, but a good core developer is highly valued at HRT. Jason Carroll, the one remaining founder at HRT who is still an active partner, was a core developer of legendary skill, by the way. If the most important person at a firm is a core dev it changes the equation.

It’s a very collaborative, friendly place, with a great respect for technical excellence. And they get the tech very right. They’ve tried many things and not everything has worked out. But their technical bets have paid off so spectacularly well. If you ignore Jane Street or Susquehanna — which are really more trading companies than technology companies — then I think HRT is the second-best company [of the big trading firms]. And it has a fair chance of overtaking Citadel Securities.

And Balyasny?

I’ve known BAM since forever. I almost came twice before, but third time’s the charm.

BAM also takes the characteristics of the founder. I once asked Dmitry [Balyasny] to describe the firm in three adjectives, and he said ‘humble, collegial and collaborative’. I think it’s actually true. It’s not into performative kumbaya collaboration. BAM doesn’t use buzzwords. But it is pretty open. And it’s a learning organisation.

Dmitry is at his heart a trader, with great instincts and humility, just like Izzy. As a result, BAM is also quite trading-oriented, and the managers tend to be quite humble, which I like.

I heard that when HRT first approached you that you thought it stood for hormone replacement therapy?

Yes! I had never heard of it, so I Googled it and that was the first thing that came up.

The different styles of quantitative investing and trading can be a bit mystifying to a lot of people. Do you have a rough taxonomy that can help?

The best way is to look at the sources of alpha, and there are really three ways of making money — very broadly speaking — that both quants and fundamental investors use.

The first is risk premia. This is by far the largest of the three. All endowments, all sovereign funds need to allocate to high-capacity fundamental indices or asset classes and seek to get compensated for taking some kind of risk, whether it’s credit risk, liquidity risk, whatever.

The second is informational advantage. It means looking at the same data as everyone else — hopefully, as long as no one is cheating — but by making more informed analysis I can come up with a better forecast of the future. This is at the core of long-short fundamental equities, statistical arbitrage or qualitative credit investing.

The third is arbitrage. At its core, arbitrage is violating the Law Of One Price. It’s a situation when you and I have different constraints or preferences, and someone can exploit this. A good example is exchange-for-physical transactions.

These three classes are not disjoined. There is some overlap. Large firms tend to be successful in more than one area, but most tend to be predominant in one of them.

AQR for example, is primarily a risk premium shop. Two Sigma is primarily a informational shop. Jane Street is an arbitrage shop. HRT is an informational shop, because most of their P&L is in high-frequency trading, and high-frequency trading is an informational strategy.

You have to consider the dominant strategy at each firm and then you have an idea of what the culture is like. Culturally, HRT is not like Jane Street — at all. One is a high-frequency, quantitative, informational strategy shop, and the other is an arbitrage shop.

There’s been a lot of talk about how multi-manager, multi-strategy firms like Balyasny, Citadel and Millennium are growing, while large parts of the rest of the industry stagnate. Do you think this will continue?

In all honesty I don’t actually think the trend is new. It’s just become more conspicuous now and has captured the attention of journalists. But it’s been a trend for something like 20 years. And a good, zero-information prediction is that if something has been going on for 20 years it will probably go on for another 20. I have structural reasons for believing that platforms will only grow in importance.

You’ve written two books on quantitative investing, both of them dedicated to Tofu. I have to ask, who is Tofu?

Tofu is my cat! It’s a British shorthair. Typically I wrote books during my non-competes and my cat keeps me company then.

So your first one is called Advanced Portfolio Management but that one is more of an introduction to quantitative investing, and your second one is called The Elements of Quantitative Investing, but that one is actually more hardcore for quant practitioners?

That’s right. I actually wanted to call the first book a Grimoire of Portfolio Management. But my editor didn’t know the term grimoire, and the rest of the editorial board there thought it had something to do with witchcraft. I thought “Advanced Portfolio Management” was bullshit, but my editor was right. He’s been doing this for 20 years. With the second book, I just liked the ‘Elements of . . . ‘ formulation.

I don’t know. I think grimoire has a bit of extra pizzazz. But why write books about quant investing?

Because I am not a finance person at heart. I don’t care particularly about money or finance, to be completely honest. I am motivated by people, and in my jobs I have had a lot of questions from portfolio managers. So I thought I’d write a short book to explain all the basic concepts to them. The first book is really a love letter to the portfolio managers.

The second book I had always wanted to write. I really believe in the exchange of ideas. I really think we are too obsessed with not sharing ideas. I think you can share a good amount of ideas without damaging anybody, and it’s good for society. I dislike writing. Writing is very painful. But I like to think about problems. And it’s a very basic book. It’s basically a letter to my past self, 10 years ago.

I’ve seen you say ‘boys use copulas, men use volatility and correlations’, in terms of risk management in investing. Can you explain that?

If you talk to a quantitative person who is enamoured with techniques — especially sophisticated techniques — it is a good heuristic to underweight them. The practices of a good quantitative modeller — in any industry, not just in finance — is not to apply the most sophisticated technique. It is to use the simplest technique that works for the problem at hand.

Volatility works. It’s not perfect, but this is not a good reason not to use it. Other metrics seem at face value to be more accurate, but they are often more fragile. Maybe vol is not beautiful, or the final word. But people need simple and common languages to progress.

You’ve mentioned that you’re not really a finance person at heart. If you were starting out again today, would you go into finance?

It’s hard to say. Firstly, it is a sector that is as a whole not growing. Secondly, there are problems that are incredibly interesting in biology, in computer science. Work in AI, work in drug design — even work in logistics. If I was young again I’d work in areas that people are not looking at now.

So if you were going to retire and do something completely different, what would you do?

That’s a good question . . . Unfortunately, this wouldn’t just mean retiring from finance but also retiring from my wife, but I would love to be a monk. Honestly, that’s my ideal job. I’d really love the solitude and contemplation.

Stay informed with free updates

Simply sign up to the Financial & markets regulation myFT Digest — delivered directly to your inbox.

It is not always easy to understand the appeal of stablecoins, particularly for European or British people.

Why would you want to try to pay with something that worked like a debit or credit card, but sometimes slower, with somewhat unpredictable fees, and which brought you into a confusing ecosystem in which consumer protection is poor and bad practice all too common? The only answer which really comes to mind is “because you’re doing something that you don’t want the regulated banking system to look at”. 

In other words:

But for Americans, there’s  a significantly more prosaic reason:

American payments fees are anomalously extortionate. As well as very high card fees, it has remittance costs that are closer to the developing world than to Europe:

Why is this? There’s no very obvious technological reason. The US retail payments network isn’t quite as modern and hi-spec as the European SEPA or the UK’s Faster Payments, but it’s not bad, and it’s a small part of the overall charges. The Mastercard and Visa networks are the same for everyone. The majority of US bank customers are with big institutions that can invest in the best systems.

But there’s a very obvious non-technological reason. As can be seen from the first chart, the biggest driver of card costs is “interchange fees” — the fee paid to the card issuer, which is meant to cover the costs of chargebacks, fraud detection, providing the physical card and so on. 

In most of the world, these fees are capped by regulation at somewhere near their actual cost. In the US, credit card interchange fees are completely unregulated, while debit card interchange fees have a relatively generous cap which the Fed has been trying to negotiate down.

Some of the excess cost is rebated to customers through free checking accounts and extremely generous rewards programmes. But quite a lot of it stays with the banks; there’s a lot of customer inertia created by the fact that it’s comparatively inconvenient to switch card providers, and there’s not much incentive to cut prices in a relatively comfortable oligopoly.

Whatever their many other nasty properties for the wider financial system, stablecoins are a potentially cheaper no-frills option for payment services. 

Find a way to “provide a worse service for less money” is the single best-proven fintech strategy of them all. It’s no surprise, then, that the biggest US banking groups are said to be exploring the idea of launching their own stablecoin solution. But even if they hang on to the market share, the dawn of price competition in US payments business is unlikely to be a pleasant experience for the incumbents.

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Having spent the last 20 years accumulating ever-greater piles of debt, governments now face tough choices/ difficult decisions.

What if there was one weird debt management trick that would allow governments to slash debt-to-GDP without spending cuts, tax raises, financial repression, inflation or default?

There is. But it’s terrible.

As a broad metric of fiscal health, debt-to-GDP has simplicity on its side. It means we can quickly summon charts like this one, showing quite how indebted governments are today compared to yesteryear:

And simplicity can be great for catalysing public policy.

While the US debt-to-GDP numbers don’t seem to be high enough to prevent the US Congress from pushing through monster tax cuts for the super-rich, they do appear high enough to force US Congress to slash spending on the poor.

But simplicity can also be . . . simplistic? So simple that it can be easily gamed.

All governments need to do is exchange their old bonds (trading sometimes far below par) for shiny new current-coupon bonds. Doing so would, in some cases, slash debt-to-GDP.

🥳 🥳 🥳 

To get a (very) rough estimate as to the potential impact that such a hypothetical Big Beautiful Bond Exchange™ might have on debt ratios, we cross-referenced average bond prices in the Bloomberg Global Treasury Index with the IMF’s estimates for debt-to-GDP.

The big winners of any massive debt swap look to be the United States (for sheer dollar debt reduction), Japan, and the UK (in cutting debt ratios).

True, looking more closely at the US, our working assumption that all US government debt is tradeable US Treasuries in the bond index turns out to be a bit of a stretch. According to the US Treasury, these make up less than half of total debt:

So sure, swapping old USTs trading at a discount for current coupons would cut US federal debt by $1.6tn — or around 50x itemised Doge savings — it’s still only five and half per cent of GDP, which is less than a single year of Trump’s planned deficit.

It hardly seems worth the bother.

How about a closer look at the UK, where gains could be greater relative to GDP?

The British government spent years selling shedloads of low-coupon gilts when yields were on the floor. In today’s higher-yielding environment, this means oodles of bonds trading at deep discounts.

Keen readers will recall that low-coupon gilts are pretty useful to UK taxpayers and hedge funds alike. And this has led to the gilt curve looking pretty kinky. 

We’ve drawn the curve below, colouring each bond according to its coupon:

We calculate that swapping all gilts into current coupon bonds in a bondholder-friendly exchange could wipe £355bn off the face value of government debt and cut the debt-to-GDP ratio by around 13 percentage points — from 101 per cent to a mere 89 per cent. And that’s before we even look at inflation-linked gilts.

But the eagle-eyed among you will have noticed that bonds with the lowest yields on the curve are all low-coupon gilts (in red). And bonds with the highest yields on the curve tend to be more recently issued gilts (in blue).

If swapped (on a non-coercive basis), our massive gilt exchange would see the richest bonds on the curve swapped into the cheapest ones — increasing the cost of debt. And so this is not an exchange that any government with an eye to actual economics of debt management — rather than mere optics — would think sensible.

So what?

Our mad exchange plan tells us something about debt-to-GDP. Bean-counters consider only principal payments to be debt; they just ignore coupons. The Big Beautiful Bond exchange reduces debt-to-GDP because the market would be nuts not to consider both.

Debt service as a percentage of GDP is — in general — a far more thoughtful way of considering debt-imposed constraints on current government finances. Whilst debt-to-GDP breaching the 100 per cent threshold may grab headlines, political appetite for fiscal consolidation reliably starts to pick up only when debt service rises. And eyeballing the chart below, when it breaches five per cent of GDP it has generally been met by a punchy response.

What does our one weird trick do to debt service? Nothing good.

Putting it altogether

The chart below shows where we are today on both debt-to-GDP and debt service across the G7. If you scroll down you can see where these ratios might head if we played OneWeirdTrickenomics without letting pesky facts as to the true composition of government debt get in the way.

Debt service ratios today are not particularly alarming across the G7. A massive bond exchange into current coupon bonds to cut the debt load would push interest bills immediately higher.

Of course, as debt matures and is refinanced at higher interest rates, debt service costs will tick up anyway.

But fiscal policymakers have time on their side. The average maturity of JGBs is almost ten years. The average maturity of gilts is over 12 years. If the UK and Japanese economies grow faster than their debts over these periods, the degree to which debt service rises will be limited.

Things look more challenging in America. The average maturity of US Treasuries is around seven-and-a-half years. And in practice US debt service is much more responsive to interest rate moves because of all the T-bills they print. In their long-term forecasts, the CBO reckons that debt service will rise to 5 per cent of GDP — although not until 2050. This is based on their expectation that the fed funds rate falls quickly to 3.25 per cent and that 10-year Treasuries stay below 4 per cent. Furthermore, the CBO long-term forecasts don’t yet incorporate the Big Beautiful Bill. So today their forecasts look optimistic.

What does it all mean?

Today’s government debt loads are high but not unprecedented. Cutting them down to size through our massive debt swap plan would clearly be preposterous (although swapping long expensive debt for shorter cheaper debt would not). Sure, it would lower the debt-to-GDP ratio in a number of countries. But only because the ratio is silly. Interest bills would be jacked up in the short run. And this is before we factor in any vibes-based premium that might be applied by the market for mucking about.

Sorry guys, difficult decisions it is.

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Following a bit of a wobble around tariff shenanigans, corporate bond spreads are tight once more. The average single-B US corporate bond yields around 3.4 per cent more than US Treasuries — around 1.8 per cent below its long-term average.

This is despite downgrades to economic growth projections, and interest rates remaining stubbornly high. Maybe this makes sense? After all, default rates have not been particularly high.

If we’re looking for signs that the credit cycle is turning, we might be looking in the wrong place. At least, that’s the gist of a new note from Jason Thomas, Head of Global Research & Investment Strategy at The Carlyle Group.

It’s not that he’s down on credit in general. Thomas reckons tight single-B spreads reflect solid business performance. But, he argues: “[w]hat may appear as credit market complacency may instead reflect bifurcation”.

And the place to see this bifurcation in action isn’t in public bond markets. It’s in private or semi-private credit markets.

Traditional default rates in the syndicated loan market are still low. But they’ve been kept low due to rising levels of “liability management exercises”, or opportunistic debt restructurings. If you add these to the mix, Carlyle believes that syndicated loan default rates are now above long-run averages.

Furthermore, while it’s hard to get decent numbers about the opaque world of private credit, Thomas reckons there is a similar dynamic in play in direct lending:

where default rates remain low, in part, because creditors have converted struggling borrowers’ cash-pay coupons into payment-in-kind (PIK) where foregone interest accrues to the principal balance

We’ve discussed PIKs before. And as S&P Global Ratings explains:

When a borrower opts to preserve cash flow by making PIK payments, it may signal an inability to meet cash interest demands.

Since late 2023, PIK loans have been proliferating, with S&P Global Ratings’ estimates of the share of loans making PIK payments being just over 10 per cent in the third quarter of 2024. This was actually down on the previous quarter, though S&P expects it to rise over the rest of 2025. Their estimate is based on their analysis of 165 BDCs — business development companies, which serve as wrappers for diversified pools of private credit.

As you can see in the chart below, the proportion of income accounted for by PIKs across the 16 publicly rated BDCs varies enormously:

[Pop]

Lincoln International, a global banking advisory firm that — among other things — assembles indices to track the performance of private assets, found that by the end of Q125, around 11 per cent of debt investments tracked had some element of PIK interest, up from 7 per cent in 2021. Six per cent of deals had what they call “bad PIK” attributes — meaning that PIK interest had been added to loans that did not get done originally with PIK interest.

They think “bad PIKs” offer a decent proxy for a shadow default index, because . . . 

[w]hile not all “bad PIK” borrowers are distressed … [loan to value ratios] for these deals increased from 49% at close of the investment to 86% in Q1.

There are lots of possible endings to this story.

In a good scenario, earnings grow, and interest rates fall. Creditmetrics improve, all firms come good with all their debt payments, and Forest Green Rovers win the FA Cup.

Carlyle’s Jason Thomas is less optimistic:

Creditors’ willingness to work with troubled borrowers over the past few years seems to have been influenced by the threat of companies stripping assets from lenders’ collateral pool, alongside rate cut expectations.

Moreover:

Such generosity is likely to fade as market participants come to terms with life in a new interest rate regime. As [liability management exercises] fall back into distress and equity evaporates in PIK conversions, expect default rates to move higher.

Of course, we guess everyone could just pretend that everything is fine, and kick the PIK can even further down the road.

Further reading:
— Is it PIK-up time for cash-strapped companies? (FTAV)