Tuesday, May 23, 2023

Fractional Reserve Crypto-Banking

Michael Lewis is writing an eagerly awaited book on Sam Bankman-Fried and the collapse of FTX, and on his podcast series he has been interviewing sources he used in his research. The latest one is an entertaining interview with Molly White. The best part is their discussion of how the story justifying cryptocurrencies keeps changing.

Nakamoto's goal for Bitcoin was:
What is needed is an electronic payment system based on cryptographic proof instead of trust, allowing any two willing parties to transact directly with each other without the need for a trusted third party.
Below the fold I examine how this earliest cryptocurrency story changed.

Think of Nakamoto's as a world where dollar bills migrated from under one person's mattress to under another person's. Even Nakamoto understood that, in a world where mattresses had cost-free pseudonyms, this wouldn't work. He anticipated part of Amir Kafshdar Goharshady's Irrationality, Extortion, or Trusted Third-parties: Why it is Impossible to Buy and Sell Physical Goods Securely on the Blockchain by writing:
routine escrow mechanisms could easily be implemented to protect buyers
seeingly without noticing that "routine escrow mechanisms" had to be trusted third parties.

Even if we ignore the fact that virtual mattress-to-mattress transactions are unsafe, a fundamental problem remains. The thing about money is that more is always better and less is always worse. In Nakamoto's world there are only two ways to increase the size of the stash under your mattress; you either mine new Bitcoin, which is a hassle, or you buy low and sell high, which is even more of a hassle.

What Nakamoto's world needed is a way for the many people who don't want the hassles to lend to the few people who do want the hassle, and receive interest for doing so. It needed banks, which had to be trusted to pay the interest and pay back the loan.

So we come to Matt Levine's Crypto Had Its Bank Runs Too. He starts by describing the problem:
One possibility is that fractional reserve banking is deeply rooted in human nature. People have money, they would like to keep it somewhere safe, they would like it to grow, they would like to be able to get it back at any time. Other people need money, they are willing to pay to borrow it, but they want it for a long time — they want to be able to use it to buy a house or build a business; they don’t want their lender to be able to demand the money back at any time. The savers and the borrowers just want different things. Why would anyone want to lock their money up for a long time? Why would anyone want to borrow money for an uncertain time?
One (impractical) solution is full-reserve banking:
Full-reserve banking (also known as 100% reserve banking, or sovereign money system) is a system of banking where banks do not lend demand deposits and instead, only lend from time deposits. It differs from fractional-reserve banking, in which banks may lend funds on deposit, while fully reserved banks would be required to keep the full amount of each customer's demand deposits in cash, available for immediate withdrawal.
In practice, the "cash" would be in an account at the central bank, but central banks hate the idea, as Alex Harris reports:
The central bank has raised several concerns about narrow banks. The main one is that in times of stress they’d be too attractive as a haven. Money could pour out of Treasury bills, high-quality bonds or even accounts at conventional banks, amplifying risks to the broader financial system. Narrow banks could also make it harder for the central bank to manage short-term interest rates. And because conventional banks could end up holding few deposits, they might do less lending, making loans more expensive and credit harder to get.
The alternative is fractional-reserve banking, in which banks keep enough cash on hand to cover their expected worst-case withdrawals during the time they need to sell their longer-term, higher-interest assets. Levine explains the idea:
If you are an enterprising middleman, you can try to convince one side or the other to do something that doesn’t quite match their desires — “invest your money long-term, but if you need it back we can probably find a way to get it to you,” or “borrow short-term, but you can probably keep rolling your debt for a long time” — but it is easier and more appealing to just promise everyone exactly what they want. People who want to park their money give it to you and you promise to give it back whenever they want; people who want to borrow money borrow it from you and you tell them they can keep it for a long time; probably this all works out,
Except that sometimes, as with Silicon Valley Bank, it doesn't because if the "people" who lent you the money are (a) very rich and (b) very online you will have underestimated the rate of withdrawals once confidence in your promises erodes. Bank runs these days are extremely fast. Banks are regulated to force them to keep a lot more cash than they would like in order to reduce this risk.

Leviine tells the story much better than I did:
People in crypto did not trust the banks, in part for the good reason that the banks were doing something (maturity transformation) that is both risky and in some deep sense deceptive. But people in crypto did want the benefits of maturity transformation: People with crypto wanted to park it somewhere safe, earn interest and have access to it whenever they wanted; other people wanted to borrow crypto without the risk of having to give it back early. Crypto shadow banks — Celsius, Voyager, BlockFi, Genesis, Gemini Earn, FTX — sprung up to offer that service, to borrow short and lend long. Free of most regulation, they could offer the service efficiently, market it aggressively, and lose tons of their customers’ money.
Levine quotes Bank Runs During Crypto Winter by Gary Gorton and Jeffery Zhang:
First, despite mass marketing campaigns to the contrary, crypto lending platforms recreated banking all over again. Crypto lending platforms were vulnerable to runs because, like all banks, they borrowed short and lent long. This is the essence of banking, so we label these lending platforms “crypto banks.” Second, crypto space was largely circular. Once crypto banks obtained deposits and investments, these firms borrowed, lent, and traded mostly with themselves.
Gorton and Zhang warn regulators:
The next generation of crypto firms are linking up with the financial sector, which means their failures will spill over into the real economy. To contain the inevitable growth of systemic risk, regulators should use banking laws to address a banking problem.
Levine elaborates:
in the US, efforts at crypto regulation are largely about either securities law or commodities regulation, but the actual problem of 2022 was banking. Securities regulation is about giving investors full information so they can make informed decisions about what companies to invest in. Bank regulation is the opposite; it assumes that bank depositors should not have to care much about what their banks are up to; bankers and regulators and supervisors worry about a bank’s asset quality and liquidity and capital ratios so that depositors can be information-insensitive.
The essential part of making depositors "information-insensitive" is deposit insurance, which is feasible only if banks are strictly regulated. Applying banking regulation to crypto banks would definitely impose "regulatory clarity", but in a way that would render crypto banks unprofitable. They would be unable to rip off their customers or speculate in coins proceeding moonwards:

Levine points out that the crypto banks are fundamentally riskier:
Whereas traditional banking borrows short to lend long to people who want to build houses or start businesses, crypto shadow banking borrows crypto short to lend crypto long to people who want to speculate on cryptocurrencies. In some sense you’d expect this to create less of a maturity mismatch — how long does anyone really need to borrow for to day-trade cryptocurrencies? — but on the other hand the collateral is much riskier. If you borrow to buy a house and default, the bank gets the house. If you borrow to buy some magic beans and default, the bank gets some beans.
And in A Retrospective on the Crypto Runs of 2022, Radhika Patel and Jonathan Rose of the Chicago Fed point out that runs on crypto banks happen even faster than Silicon Valley Bank's:
customers withdrew a quarter of their investments from the platform FTX in just one day
FTX itself reported outflows of 37% of customer funds, almost all of which were withdrawn in just two days
And as with Silicon Valley Bank, it was the largest depositors who acted fastest:
The owners of large-sized accounts, with over $500,000 in investments, were the fastest to withdraw and withdrew proportionately more of their funding. In fact, during this run, 35% of all withdrawals at Celsius were by owners of accounts with more than $1 million in investments, according to our estimates.
All this focuses on trading platforms like FTX or Celsius. But there is another type of crypto bank, stablecoins. I wrote about the risk of runs against algorithmic stablecoins revealed by the Terra/Luna collapse in Metastablecoins, pointing out that the arbitrageurs who were supposed to keep UST trading at its peg had limited firepower that was overwhelmed in a run. The next week, after USDT had traded nearly 5% under its peg, I wrote More Metastablecoins, based on research from Barclays quoted by Bryce Elder in Barclays to tether: the test is yet to come. Elder described USDT's defenses against a run:
Tether’s closed-shop redemption mechanism means it cannot be viewed like a money-market fund. Processing delays can happen without explanation, there’s a 0.1 per cent conversion fee, and the facility is only available to verified customers cashing out at least $100,000.
And Barclays described how, even if USDT were backed 1-for-1 by dollars in FDIC-insured bank accounts (which it isn't) these defenses wouldn't actually prevent a run:
The only way to get immediate access to fiat is to sell the token on an exchange, regardless of the size of holding . . . [W]hile redemption is ‘guaranteed’ at par, the secondary market price of tether can trade lower, depending on the willingness of holders to accept a haircut in return for access to immediate liquidity. As last week’s price action suggests, some investors were willing to accept a nearly 5 per cent discount to liquidate their USDT holdings immediately.
We think that willingness to absorb losses, even though USDT is fully collateralized and has an overnight liquidity buffer that exceeds most prime funds, suggests the token might be prone pre-emptive runs. Holders with immediate liquidity demands have an incentive (or first-mover advantage) to rush to sell in the secondary market before the supply of tokens from other liquidity-seekers picks up. The fear that USDT might not be able to maintain the peg may drive runs regardless of its actual capacity to support redemptions based on the liquidity of its collateral.
Historically, and in the future as Anna Irrera and Emily Nicolle report in Tether to Purchase Bitcoin as Part of Reserves Strategy Shift, USDT has not been fully collateralized:
Tether Holdings Ltd., the operator of the largest stablecoin, will invest as much as 15% of profits on a regular basis in Bitcoin as part of a strategy to diversify its reserves.
Tether held around $1.5 billion of Bitcoin as part of the reserves backing its tokens at the end of March, according to a third-party attestation of its holdings.
The company said on Wednesday that it does not expect the value of its current and future Bitcoin holdings to exceed its shareholder capital cushion, referring to excess capital held by Tether to protect against heavy losses.

That figure now stands at more than $2.5 billion, Tether’s Chief Technology Officer Paolo Ardoino, said on Twitter.
This looks like an attempt to pump Bitcoin back over the important $30K mark. In Stablecoins Deliver on Their Promise: Disrupting Banks, Steven Kelley writes:
USD Coin, the second biggest stablecoin by market cap, received a government rescue in March—proving it really can compete with banks.
Pursuantly, over the course of three days in March, the backing assets of the “fully reserved” USDC became a portfolio enviable of a distressed credit investor. And, by extension, so did USDC itself. USDC started to wobble under the weight of the above disclosure (transparency!), and fell more sharply when Circle disclosed it in fact had $3.3 billion stuck at SVB despite attempts to withdraw
USDC traded at less than 90 cents on the dollar that weekend — until the government announced it would stand behind the uninsured deposits of the failed banks: ...
The “we don’t lend reserves” refrain was always nonsense, and now USDC has faced a 48-hour drill making that abundantly clear. To be truly “fully reserved” is to have all the reserves at the central bank.

Saying anything less is “fully reserved” is egregiously misleading. Uninsured dollars in banks—which USDC likely needs at least some of (and, in any case, had a lot of) as the on- and off-ramps to the blockchain—are loans to those banks. Circle is issuing demand liabilities and making risky loans; it’s a bank.
Levine summarizes the situation:
If your concern is that crypto shadow banks are becoming more interconnected with the real economy, and that therefore future runs on those shadow banks might be more destructive, there are two ways to go:
  1. Protect crypto shadow banks from runs, with deposit insurance and regulation; or
  2. Protect the real economy from crypto shadow banks, by making it really hard for the traditional financial system to connect with crypto firms.
US regulators seem to be choosing Option 2, which … seems … right … to me?
And to me, as I said to the regulators.


David. said...

Cryptadamus has an important thread entitled What #Tether Has Said Lately vs. Arithmetic: The Thread:

"Isn't it odd that somehow $1.5bn worth of bitcoin just appeared as the backing for #Tether few days ago?

➤ $1.50bn in bitcoin on March 31st
➤ $1.48bn in "profits" as of May 11th

Those numbers are ~1% different.
Are you seeing a pattern here?

#Tether's official document that definitely is not an audit claims $USDT is backed by:

a) $1.5 bn in bitcoin
b) $3.4 bn in gold
c) $2.1 bn in "other investments"
c) $5.3 bn in "secured loans" (not stated: what the loans are secured by)"

And about $69B in cash, treasury bills, etc. So:

"There's only 3 possibilities. Tether either:

1. Already had the BTC & their attestations were total lies or hid the bitcoin in "other investments"

2. Printed unbacked $USDT to buy BTC ("secured loans")

3. Used $ other than profits to buy BTC"

David. said...

As usual, Matt Levine nails it in Three Arrows Had a Fun Bubble:

"Instead 3AC invested using its partners’ capital and immense oceans of leverage from crypto lending platforms. This leverage was provided with very little in the way of due diligence or negotiation or often even collateral. And 3AC was very clear-eyed and thoughtful about identifying this part of the bubble. The bubble was not just “people buy crypto and it goes up”; it was specifically “people invest a lot of money in crypto lending platforms, which promise a high rate of interest but have nowhere good to lend their money, so we will just borrow all their money from them, pay them interest, and use the money to make insanely risky crypto bets. If they pay off, we buy yachts; if they go bust, we don’t pay them back and sail away on our yachts.” This worked for a while and made the 3AC guys rich, and then it stopped working and made 3AC’s lenders bankrupt. But the 3AC guys took money off the table along the way, and they will be wealthy for the rest of their lives. A bunch of crypto lending platforms had too much money and wanted to do something dumb with it. 3AC provided the dumb thing to do with it, and took a large fee for that service."

This was inspired by David Yaffe-Bellany's Their Crypto Company Collapsed. They Went to Bali.