Welcome to another issue of Net Interest, where I distil 25+ years of experience analysing and investing in financial stocks into a weekly newsletter. This week the topic is FTX, which has just filed for bankruptcy. If you haven’t read my earlier piece on the crypto collapse, The Hedge Fund, the Bank and the Broker: How It All Fits Together, it may be worth reading that in parallel. Paying subscribers also get access to comments on Sculptor Capital Management, Goldman Sachs, Retail Trading. To join them, sign up here:

Banks go bust more often than you might think. In the five years between 2016 and 2020, 21 of them went bankrupt in the US. The most recent was Almena State Bank of Kansas, which collapsed at the end of October, 2020. The surprise is that there have been no bank failures since. We’re now tracking 749 days since the last bank in the US collapsed – the second longest streak in over 20 years.   

Banks fail for a variety of reasons. Usually it’s a slow process. Almena State Bank exhibited deteriorating asset quality for four years before it went bust. First City Bank of Florida, which failed one week earlier, had been unprofitable for 12 years before its demise. Slow failures usually happen because management makes poor investment decisions, often around credit. First City Bank of Florida had built up too high a concentration in commercial real estate, to a point where it represented 600% of the bank’s capital base just before the downturn. 

Sometimes, banks can fail very quickly, and these are the stories that hit the headlines. Failure stems from a “bank run” where those funding the bank’s assets want their money back more quickly than the bank is able to raise it. The quick and slow versions of bank failure are not unconnected: solvency fears often prompt liquidity runs, and liquidity runs may lead to fire-sales that push down asset prices, leading to solvency concerns.

To protect against the fallout of bank failures, a strict regulatory regime has emerged. There are regulations to ensure that banks hold sufficient capital as a shock absorber against the slow kind of failure. And there are regulations to mitigate the threat of the quicker kind: Banks are required to hold cash buffers to cover unexpected withdrawals, and a deposit guarantee scheme is in place to protect smaller depositors. 

In addition, there’s a tried-and-tested formula for resolving failing banks. Regulators walked into Almena State Bank after it closed its doors one Friday, took control of its assets and liabilities and transferred some of them to another bank – Equity Bank, headquartered 60 miles away in the city of Andover. On Monday morning, depositors woke up to find they were customers of a new bank. All deposits up to $250,000 were made good by the regulatory deposit insurance scheme, at a cost of $18 million to the fund.

Almena State Bank wasn’t a large bank – just $69 million of total assets – so the process was straightforward. But it works for much larger banks, too. In September 2008, regulators used the same formula to resolve Washington Mutual. With $307 billion in assets, it was the largest failure of an insured depository in history. After taking control, the regulator turned around and sold most of the bank’s assets and liabilities to JPMorgan. So quickly did it happen that Washington Mutual’s CEO missed it – he was midair between New York and Seattle when the deal was done. ¹

And so to FTX

For better in the eyes of some, for worse in the eyes of others, crypto does not have these regulatory safeguards. When a crypto firm goes bust, there is no playbook. This is the situation FTX now finds itself in. Technically, FTX is not a bank but in the world of crypto, the boundaries between bank, broker, exchange and clearing house are blurred. As I write, the company has just entered bankruptcy proceedings. With a recent valuation as high as $32 billion, it is the biggest crypto firm to fail so far in a year that has seen many other casualties (as we discussed in The Hedge Fund, the Bank and the Broker: How It All Fits Together). 

There has been a lot of speculation about what caused the collapse of FTX. Events began to unfold when Coindesk published excerpts from the balance sheet of Alameda Research. Alameda is a principal trading firm affiliated to FTX. According to the FT, it is 90% owned by Sam Bankman-Fried, who also owns 53% of FTX. Alameda was the vehicle through which Bankman-Fried made his first foray into crypto, exploiting arbitrage opportunities between bitcoin markets in 2017. His experience as a trading customer convinced him to build the kind of exchange he might want to use and so FTX – an exchange “by traders, for traders” – was launched shortly afterwards. 

There’s nothing unusual about a trading customer seeding liquidity on an exchange in return for some equity ownership. Intercontinental Exchange (ICE) was founded in 2000 with capital backing from Morgan Stanley, Goldman Sachs and several large energy trading companies. But the relationship between Alameda and FTX was opaque. Alameda made up around 6% of FTX’s trading volume, but the overlapping interests only became clearer with the Coindesk report. 

Turns out Alameda had $14.6 billion in assets, made up largely of crypto tokens. Its biggest holdings were a token called FTT ($5.8 billion) and one called SOL ($1.2 billion). Bankman-Fried has long been a fan of SOL; FTT we’ll come to. Other crypto valued at $3.4 billion, equity securities valued at $2 billion, cash of $134 million and remaining assets of $2.2 billion round out the balance sheet.

At first glance, the balance sheet looks solvent. On the other side of those assets are $8 billion of liabilities, leaving an equity cushion of $6.6 billion. But there’s quite a lot of concentration stacked on the asset side of the balance sheet, with FTT making up 88% of Alameda’s net equity. That amount of FTT was also an order of magnitude higher than the circulating supply, meaning it would have been very hard to sell and was likely overvalued on Alameda’s balance sheet. So what is FTT?

FTT is a token issued by Alameda’s sister company, FTX. Across the crypto universe, different tokens have different economic features – some look more like commodities, some more like equities. FTT was towards the equity end of the spectrum. The token gave FTX users discounts on their trading fees of up to 60%, making it popular among customers. FTX would use part of its commission income to buy and burn FTT tokens, analogous to a stock buyback process. And Sam Bankman-Fried would promote it regularly. 

Given that it represents a kind of equity interest in FTX, it may be surprising to see FTT sitting on sister company Alameda’s balance sheet, especially in such size. Banking regulators have strict rules about reciprocal cross-holdings in capital structures. The Basel Committee on Banking Supervision insists that “reciprocal cross-holdings of capital that are designed to artificially inflate the capital position of banks” are fully deducted from capital.

The situation gets worse when we look at how Alameda funded its FTT position. Within the $8 billion of liabilities on Alameda’s balance sheet are $7.4 billion of loans. According to the WSJ, these loans come from sister company FTX. Indeed, by this week, that loan had increased to $10 billion. Reports suggest that FTX initially made the loans to bail Alameda out of losses it incurred when crypto markets went into turmoil in May. Alameda’s FTT holdings served as collateral for the loans. ²

The problem is that FTX sourced the funds for the loan from its client accounts. Traditionally, brokers should keep client assets segregated. MF Global was fined for dipping into client funds to stave off its collapse in 2011, although they were all ultimately recovered. More regulated crypto brokers pay heed: Coinbase has $95.1 billion of customer crypto holdings stashed in a separate bucket on its balance sheet. “As you can review in our publicly filed, audited financial statements, we hold customer assets 1:1,” Coinbase said in a press release this week.

The practice of lending to another company to buy your stock has a precedent in traditional financial services. A few years ago, Barclays was charged with making a $3 billion loan to Qatar to allow it to buy shares in Barclays during the financial crisis in 2008. The charges failed to stick because the court found it difficult to assign corporate liability, but there’s no doubt the practice is highly irregular. It’s also very risky – as long as the stock is stable or going up, everything looks rosy, but as soon as it starts to fall, things can unwind rapidly.

That’s what happened to FTX. Allegedly spooked by the revelations around Alameda’s balance sheet, the founder of Binance, a competitor firm and initial backer of FTX, announced that he would sell $530 million of FTT sitting on his books. The price of FTT began to collapse. From a price of $22 at the beginning of the week, FTT is now trading between $2 and $3 per token. 

As it fell, a hole opened up on FTX’s balance sheet. Customers still had around $16 billion of assets on the platform. But the token that Alameda had pledged to borrow those funds was worth less. Initially, Sam Bankman-Fried tweeted that “FTX is fine. Assets are fine,” and followed up with another tweet: “FTX has enough to cover all client holdings. We don’t invest client assets (even in treasuries).” But those tweets were later deleted. Assets were clearly not fine, and FTX was left needing to find emergency funding – the exact amount a moving target linked to the pace of withdrawal requests and the valuation of assets. Estimates of how big the hole is range from $6 billion to $9 billion. 

In a mea culpa, Bankman-Fried admitted that he “fucked up twice”. First, his assessment of liquidity on hand was off. He thought he had sufficient dollar liquidity to meet 24x average daily withdrawals; in reality he had liquidity of only 0.8x Sunday’s peak withdrawal volume of $5 billion. Second, he underestimated customers’ margin usage. The errors were down to “poor internal labeling of bank-related accounts”. If so, it looks like internal operating systems were not equipped to keep pace with the growth of FTX. 

Watching FTX unravel, the founder of Binance tweeted:

CZ 🔶 Binance @cz_binance

Two big lessons: 1: Never use a token you created as collateral. 2: Don’t borrow if you run a crypto business. Don't use capital "efficiently". Have a large reserve. Binance has never used BNB for collateral, and we have never taken on debt. Stay #SAFU.🙏

10:05 PM ∙ Nov 8, 2022

---

98,805Likes16,192Retweets

Bankman-Fried’s reply: “At some point I might have more to say about a particular sparring partner, so to speak. But you know, glass houses. So for now, all I'll say is: well played; you won.”

Ironically, among leaders in the crypto industry, Bankman-Fried was more regulator-friendly than most. Some have suggested that differences in opinion over the role of regulators in the industry were at the root of the feud between him and CZ. Yet he didn’t practise many of the tenets of sound regulation. Even before dipping into client assets, his balance sheet remained a secret (reminding me of when hedge fund manager Richard Grubman told Enron management, “You’re the only financial institution that cannot produce a balance sheet.”) Corporate structure was equally opaque.

Instead, he leveraged his personal brand to instil confidence – something that happens a lot in crypto. Prior to its failure in May, the CEO of Voyager explained to investors why he waived taking much collateral from borrowers (as at March 2022, borrowers had posted collateral of $227 million on loans of $2.0 billion): “The people we lend to are some of the biggest names in the industry.”

There’s a saying, “Those who cannot learn from history are doomed to repeat it.” Financial services has a rich history – of personality cults, cross shareholdings, risk concentration, excessive growth, underestimation of tail risks. The conceit of crypto was to ignore it all.