This article is a response to a somewhat interesting argument that I saw: were the crypto exchanges any different than banks using fractional reserve lending? I explain the key differences — bank institutions and regulation are built upon centuries of responses to bad experiences, while the crypto bros reinvented the wheel. This text is a placeholder for a projection section wrapping up my manuscript on banking. The idea is that once we have looked at the actual practices of banks — and not popular Austrian fairy tales, or Economics 101 models — we can see how banking practices differ. I would later add references to earlier articles, which I am skipping due to time constraints.
In the initial flow of news from the failure of the FTX crypto exchange, there was a claim that FTX only had (liquid) assets that matched 10% of customer account valuations. This coincidentally matched up with the 10% reserve requirement universally used in Economics 101 textbooks (and actual reserve requirements for the United States at one point).
This coincidence led one malcontent to make an observation to the effect that FTX was just doing what banks already do.
For the purposes of this article, I am not interested in what really happened at FTX — that will be determined after lengthy court proceedings. Instead, I will just respond to how banking differs from a somewhat theoretical run on a crypto exchange.
For our purposes here, I am assuming that there is a hypothetical exchange that does the following things. Although similar allegations were made about FTX, they might not have happened. However, they could plausibly happen for a similar exchange.
- Less than 100% liquid assets versus customer account value.
- Assets matching customer accounts were lent to related entities, or used to make speculative investments.
- Customer holdings of cryptocurrencies — which have highly volatile prices — were not matched by actual holdings of those cryptocurrencies by the exchange.
- The potential use of “software backdoors” by exchange insiders to make transfers of the exchange’s cryptocurrencies away from the exchange to unknown “wallets,” with no immediate hope of reversal.
Similar to Non-Bank Collapses
The first thing to note is that non-bank financial entities have engaged in similar shenanigans since time immemorial. Of course, they similarly collapsed when faced with customer outflows. Although we like to think that non-bank finance is “unregulated,” it is in fact regulated — just with a different regulatory regime than banks. That regime is typically inadequate and is patched up after a collapse, which means that creative scammers need to come up with at least superficially different scams (even though the underlying principles remain the same).
As a result of the move towards deregulation in the 1970s-1980s, bank corporations are now knee deep in “non-traditional bank” finance. As such, banks have blown themselves up in securities markets (e.g., Barings in Japanese equities derivatives).
Since risk has largely been pushed into securities markets from traditional banking by the design of regulators and banks, it is unsurprising that banks dabbling in them face risks. I do not see such risks going away, but they are not an inherent part of fractional reserve banking.
The rest of the article just refers to fractional reserve banking.
Less Than 100% Liquid Assets
The main thing that bugs the “fractional reserve lending is fraud” crowd is that bank deposits are not 100% backed by reserves (or even notes and coins1). Sure, but since the banks do not claim 100% backing, this is not material. The question is: can they meet withdrawals in practice?
The banking industry and regulators have very long data sets of customer data deposit levels. Most customers with large bank deposits keep relatively stable deposit balances (although there will be large short-term blips such as when houses are bought and sold). Considerable effort is expended projecting potential drawdowns, and there are extra safety buffers beyond those projections.
Although runs happen, they tend to happen first in wholesale lending markets. By the time retail customers line up at the automated teller machines (ATM), the wholesale run is typically over.
Other holdings of short duration financial assets are less stable, since they are at the whims of portfolio allocation decisions. There is no reason to believe that a something like a cryptocurrency exchange can accurately model customer withdrawals. The absence of such information means that liquidity safety buffers would need to be very high.
Meanwhile, anything other than the smallest banks (which are rare outside the United States) will be heavily involved in wholesale funding markets. This is a necessity due to the structure of non-financial balance sheets: assets are weighted towards non-bank financial assets. This means that drains from the banking system are flowing to non-banks, and banks need to dip into those non-bank funding sources to make flows circular. Banks have the role of liquidity provision to the system at large — cryptocurrency exchanges do not.
If we put aside the securities arms of banks, traditional banking assets are normally either relatively liquid bonds/bills or loans. (A small portion of the balance sheet will be bank infrastructure, and unfortunately, assets seized in bankruptcy.) These lending assets have fixed par values, and if the bank appropriately books loan loss reserves, the balance sheet value will be close to the expected value of cash flows.
The behavioural conventions of bankers as well as bank capital rules would preclude banks holding large amounts of equity etc. as assets versus deposit liabilities.
That is not to say that every single bank loan is a legitimate asset. One can find examples of sketchy lending popping up whenever bank failures occur, with the Savings and Loan Crisis in the United States providing many examples. That said, the banks involved normally need to hide the dodgy loans within a larger loan book, and need to avoid loan concentration limits, etc.
No serious person claims that bank deposits are 100% backed by reserves — rather, they are supposed to be backed by a diversified mixture of fixed income and loan assets, with bank capital providing a loan loss buffer if the realised value of those assets ends up below their par value.
Well run banks do not have a currency mismatch between their assets and liabilities, even if they offer deposits in foreign currencies. For example, most Canadian banks offer U.S. dollar deposit accounts, yet they are not hit by foreign currency losses as exchange rates move.
An exchange that holds risky assets for customers needs to own those assets to hedge those liabilities. The ones that do not hedge are referred to as “bucket shops,” and being a bucket shop is one of the things that financial regulation is designed to avoid (given the experience with historical bucket shops).
Although I have heard of various illicit schemes to get large amounts of money out of banks in a non-recoverable way, those schemes tend to be fairly complex. There will be multiple people and entities involved that might face legal action to allow reversal of the flow in the courts — and therefore they have an incentive to block the transaction. Meanwhile, any bank that partakes in dubious transfers will have bank capital to absorb reversals imposed by the courts.
One of the selling points of crypto is that the courts (part of the dreaded government) cannot reverse transfers, so once the money goes walkies, that might be it. (Some protocols have reversed dubious transactions, but that is done at the whim of stakeholders of that protocol.)
The generally accepted accounting principles for banks are based upon historical experience, and dubious schemes should get unflattering accounting treatment that would raise the eyebrows of bank counterparties and regulators. Admittedly, a small bank might be able to find some sketchy accounting firm, but a large bank is going to need a large accounting firm with a good reputation. Signing off on obviously fraudulent financial statements is a good way to become an ex-accounting firm, so large accounting firms tend to at least follow the letter of accounting principles.
Conversely, the cryptocurrency industry generally avoids anything resembling a real accounting of their balance sheets. (Yes, there are some entities that run transparently, but things like exchanges and some of the larger so-called “stablecoins” are anything but transparent.)
Banks can and will fail. They intermediate credit risk to other firms and households, and some of those firms and households will go bankrupt. Furthermore, they are tied to non-bank wholesale markets, which are dens of speculation and silliness.
However, having less than 100% reserves on deposits is by itself not a mortal danger, nor does it provide it an excuse to float ridiculous scams upon rubes in the name of creating a non-bank alternative. The risks to traditional banking are well known, and considerable effort is made to contain those risks. The problem is that Economics 101 and the popular Austrian bank discussions completely ignore those efforts, and just focus on a few pet ideological points.
Demanding that a deposit be backed by both reserves (for wholesale transactions) as well as notes and coins (for withdrawals) would imply 200% backing.