Banks work in ways that are unintuitive for a lot of people.
And yet, understanding the banking system as a whole is really helpful for investing, because having a grasp of how the banking system is functioning helps you determine whether you should expect monetary inflation or disinflation, and whether the risks of widespread financial crisis are high or low.
A lot of people have been fearful to invest for many years because they are constantly afraid of 2008-style banking crises, while others keep buying bonds and underperforming most other assets for years-on-end due to misguided theories about how banking works. On the other end of the spectrum, many people have called for hyperinflation or other crises with a very misplaced sense of probabilities or timeframes.
I’ve written about bank dynamics numerous articles, and dedicated several chapters of my book Broken Money to it, but this article focuses on one specific area: fractional reserve banking vs full reserve banking and its implications on money itself.
Start here from the beginning or jump to the section you want:
Custodians vs Lenders
A pure custodian is a company that holds something of value on its customers’ behalf, and the customers usually pay a fee for that service.
Safe deposit boxes, gold vaulting services, exchange-traded funds that hold a bunch of stocks, and so forth are all examples of custodians. They hold and manage our assets, and we pay them for their services.
Some types of items are not fungible, meaning they are not interchangeable. For example, the contents of my safe deposit box are different than the contents of yours, and so the bank has to keep them separate.
On the other hand, if both of us own a 1-kilogram gold bar at the same gold-focused custodian, then perhaps the custodian doesn’t have to bother keeping track of which specific gold bar is mine or yours, since they are essentially the same anyway. In other words, gold coins and bars are pretty fungible, or interchangeable. The custodian can keep enough gold to cover all of their customers’ ownership of gold, and save money for themselves and for their customers by holding the gold in one big combined vault rather than dealing with the complexity of individual customer-to-bar logistical overhead.
When a custodian holds these types of fungible assets, which tend to be money or money-like, then it opens up a possibility: lending.
After all, most of us holding gold at a custodian are not intending to redeem it any time soon. And the gold bar I deposit doesn’t have to be the same gold bar I get back when I one day withdraw it, as long as it’s functionally identical.
The custodian can monitor statistics over time and determine that perhaps 10% of customers redeem their gold per year, or upwards of 20% during an unusual year. And they realize… they could put some of that gold to work for additional profits rather than just holding it there. They could lend some gold out, while keeping enough in reserves (let’s say 30% for this example) to cover all normal customer redemptions with a margin of safety. If they lend gold out, they could earn interest on it and boost their own profits, but then also reduce fees for customers so that customers will choose to hold their gold with them rather than their competitors, which makes them bigger.
The custodian is now a lender at this point, and thus is basically a bank. By putting that gold “to work” it gives the custodian/bank a new profit source and lowers costs for depositors, which at first impression seems like a win/win scenario. As long as they are holding an amount of reserves appropriate to meet all typical customer redemptions, they should be fine, right?
Right?
Nowadays gold is no longer legal tender and isn’t used for borrowing or lending at scale, and so gold vaulting is primarily about asset storage. But decades ago, gold was legal tender and/or was used to back legal tender money via a redemption process. Banknotes and bank deposits were fractionally-reserved custodial claims on gold, either held in someone’s name (in the case for deposits) or as tradeable bearer instruments (in the case for banknotes).
A solvent lender technically has enough assets to meet their liabilities like a custodian does, but unlike a pure custodian, some percentage of the lender’s assets (often a majority) are in the form of illiquid loans rather than held in the underlying asset that their liabilities are actually owed in. This fact introduces both institution-specific risk and more importantly introduces risk across the entire financial sector, as will be discussed in the next section.
Lending/banking can be done on top of any money or money-like fungible asset. Massive-scale lending generally only happens with the dominant money of a given era (i.e. society’s unit of account), but securities and commodities and similar things are regularly subject to lending to mild or moderate degrees as well, typically for the purpose of hedging or shorting and similar financial activities.
How Financial Crises Happen
Terms like “financial crisis” and “bank run” and “financial panic” and many others are terms that have been around for a while. They usually describe a situation where there is not just one bank experiencing a problem, but rather that there is a cascade of banking failures occurring together.
Much like how if you see a cockroach in a kitchen there are likely to be more, if you see one bank failure there are also likely to be more. In 2023 for example, the failure of Silicon Valley Bank triggered a cascade of other bank failures, up to and including Credit Suisse, which was a 167-year-old bank based half way around the world in Switzerland.
Why is that? When a chemical facility blows up, it doesn’t increase the odds of a bunch of other chemical facilities around the world blowing up. When a train derails, it doesn’t contribute to a series of other national or global train derailments within the same month. When a car company has a recall, it doesn’t trigger other car companies to have extra recalls as well. But in banking that usually seems to happen: one big failure immediately leads to a bunch of other failures.
That’s due to the fractional reserve structure of banks. They’re unstable and reliant on each other by design, more closely than companies in most other industries.
How Fractional Reserve Lending Works
Whether it’s gold, dollars, or bitcoin, any fractional reserve lending situation involves a base money and a broad money.
Base money describes how much actual money exists. For gold it’s determined by physics and mining. For dollars it’s determined by the Federal Reserve. For bitcoin it’s determined by the ruleset in a decentralized network of user-run nodes.
Broad money describes how many claims for liquid money exist across all banks. These deposit claims are different than bonds because they represent the ability to withdraw it immediately. At any given time, there are way more claims for base money than there is actual base money.
In a central banking arrangement, broad money and base money look like this in simplified form:
At the current time, there are approximately four broad dollars for every one base dollar in the United States. This is down from a peak of nearly 12-to-1 in 1985:
And if we zoom in and look specifically at banks, there are more than six bank deposits (claims for dollars) for every base dollar held by a bank. Right before the financial crisis in 2008, this ratio was 23-to-1. This higher ratio is due to the fact that a significant percentage of base dollars (i.e. most of the banknotes) are not held by banks and are floating around unlevered, held by people. The portion of base dollars owned by banks is levered quite a bit:
When a fractional reserve bank makes a loan, they create new broad money, new IOUs. For more information on that subject, I cover the mechanics of bank lending in this article and in Chapter 15 of Broken Money. Here’s an example from that linked article:
This is the simplest example to show how banks create deposits and broad money without reducing the amount of reserves in the system. It involves Mary buying a used car from Sara.
Here is the visual, with a beginning, intermediate, and ending state, and a description afterward, so you can go back and forth between the visual and the description:
Beginning State
Mary begins with “D” in assets, meaning a $1,000 deposit in her bank, and no liabilities. Her bank (which is very unlevered) starts with her deposit “D” as a $1,000 liability, and then has two reserve block assets “R”, representing $2,000 held at the Fed.
Sara begins with “DDDC” in assets, meaning $3,000 in bank deposits at her bank “DDD”, and a $1,000 used car “C”, and no liabilities. Her bank starts with her deposit “DDD” as liabilities, and has its assets primarily invested in Treasuries “TTT” and one reserve block at the Fed “R”.
The Federal Reserve holds the three blocks of reserves from the two banks as its liabilities, and has three blocks of Treasuries as its assets. The banks use the Federal Reserve as their bank, in a similar way that Mary and Sara use their banks. In other words, the two banks store their extra cash reserve assets in their accounts at the Fed, which are the Fed’s liabilities.
The U.S. Treasury Department, representing the financial arm of the overall U.S. Federal Government, has 6 blocks of Treasuries outstanding as its liabilities. For the sake of simplicity it doesn’t have any assets listed, but in reality, its assets would consist of working capital, various federal buildings and lands and military assets, and its ability to tax citizens. Its 6 Treasury liabilities are owned by the Fed and Sara’s bank.
Between Mary and Sara’s cash, there are 4 deposit “D” blocks in the total system, which are assets for them and liabilities for their banks. Likewise, there are 3 reserve “R” blocks in the system, which are assets for their banks and liabilities for the Fed.
Intermediate State
Now, for the intermediate state, Mary and Sara enter into negotiations, and Sara agrees to sell her car to Mary for $1,000. Mary, however, only has $1,000 in deposits, and although she needs the car, she doesn’t want to be completely cash-less. So, she goes to her bank, and takes out a $1,000 car loan “L”. Mary’s bank creates a $1,000 deposit “D” for Mary, and creates a $1,000 loan liability “L” for her as well. For the bank itself, Mary’s new deposit asset is its new liability, and Mary’s new loan liability is its new asset. No reserves moved, but a new deposit was created.
Mary’s net worth is unchanged at $1,000 in total, but she now has $2,000 in deposits and $1,000 in loan liabilities ,and thus is a bit more leveraged. Mary’s bank’s net worth is unchanged as well, but it also leveraged itself up a bit, by creating a new asset and a new liability, since it expects that Mary will be able to pay the loan back with interest.
Neither the Fed nor the U.S. Treasury are involved yet.
There are now 5 deposit “D” blocks in the system rather than 4, because Mary’s bank is more levered with an additional asset and liability. It created new broad money by loaning a new deposit into existence. However, there are still 3 reserve “R” blocks in the system.
Ending State
For the ending state, Mary writes Sara a $1,000 check for the car, and therefore gives her the new deposit “D” that she just received from her bank loan. Sara receives the check and deposits it in her bank account, and her bank credits this by giving her an extra $1,000 deposit asset “D”, which becomes a new liability for her bank. Behind the scenes, Mary’s bank sends a $1,000 reserve block “R” to Sara’s bank to honor the check. So, Sara’s bank now has a new liability “D” in the form of Sara’s new deposit, but also has a new reserve block “R” as its new asset. Sara’s bank doesn’t have any creditworthy clients asking for loans at the moment, so it keeps its new reserve block at its Fed account for now.
The Fed’s ending state is unchanged on net, except that it updated its book-keeping for its two client banks when Mary’s bank sent Sara’s bank a $1,000 reserve block “R”. The Fed used to attribute “RR” to Mary’s bank and “R” to Sara’s bank, but now it attributes “R” to Mary’s bank and “RR” to Sara’s bank. These reserve blocks are liabilities for the Fed, but assets for its client banks.
The U.S. Treasury’s ending state is also unchanged, and unlike the Fed, it wasn’t even aware of the transaction at all.
In the final ending state, just like the intermediate state, there are still three reserve blocks “R” in the system, and there are 5 deposit blocks “D”, which is one extra deposit block compared to the beginning state, created by Mary’s bank loan.
The point of this example is to show how, when a bank uses its reserves to lend money, the reserves aren’t destroyed. The money shows up in another bank, and the reserve amount is sent there. The overall amount of reserves or base money in the system is unchanged, but the system becomes slightly more levered, and has more consumer deposits and therefore more broad money. In other words, the money multiplier ratio (D-to-R, broad money to base money) increased from 4-to-3 to 5-to-3.
Any bank can increase or decrease its own amount of reserves by buying or selling assets, or making loans. However, those reserves get moved around to or from other banks rather than created or destroyed. Banks can, however, create or reduce the amount of deposits leveraged on those reserves, depending on how much risk it wants to take on and how many creditworthy opportunities it has to lend money for.
In an unregulated market, fractional reserve banks can determine how much of their reserves relative to deposits they should have on hand to cover the possibility of a lot of withdrawals happening at once, and customers can shop around for what they consider the best risk/reward. Of course, there are additional challenges as well, such as whether a bank can provide evidence to customers regarding its stated reserve ratio being true.
In a regulated market, fractional reserve banks are often given a minimum reserve ratio by the government or central bank. Alternatively, instead of being given a minimum reserve ratio, regulators often instead give banks a minimum capital ratio, meaning their assets have to exceed their liabilities by a certain amount, along with a requirement to hold a percentage of their assets in a list of defined “risk-free” assets, such as cash reserves or government bonds. This is known as the risk-weighted capital ratio.
To keep it simple, let’s focus on reserve ratios. If banks are told that they have to maintain at least 10% reserves relative to deposits at all times, then it means there can be 10x as many deposits in the system as there is base money. The other 90% of the deposits are backed by loans, which are less liquid, meaning they can’t be quickly sold for cash (selling a bank loan book to another bank is a complex process that requires a lot of analysis by the purchasing bank).
Even if a bank isn’t regulated by any specific reserve ratio, they have to keep some reasonable percentage of cash on hand, because they never know if there will be a higher than normal rate of withdrawals. Right before the 2008 financial crisis, U.S. banks in aggregate had less than 5% of deposits backed up by cash on hand:
Here’s the problem. When a bank makes a loan, it becomes a deposit in another bank, which allows that second bank to make a loan. When that second bank makes a loan, it becomes a deposit in a third bank, which allows that third back to make a loan. Over time, that’s how there is a lot more broad money in the system than there is base money. It’s a recursive series of lending. Each base dollar is counted, double-counted, triple-counted, quadruple-counted, and so forth as it is lent out and levered repeatedly throughout the system.
Most of the time, banks are solvent on an individual basis. They might only have 10% of customer deposits on hand in the form of liquid money, but the rest is covered by good loans, meaning that the bank has more assets than liabilities. Occasionally, however, a bank makes too many bad loans, and some of those loans are defaulted on by borrowers that are unable to pay. In this scenario, the bank’s total assets can fall below their total liabilities, and thus they become insolvent.
But even a solvent bank with no loan defaults can face a problem. If they have 10% of deposits on hand as reserves, and 15% of their depositors want to redeem their deposits at once, what do they do? In this scenario, they made no bad loans, and they have more assets than liabilities, but 90% of their assets are illiquid and they don’t have enough liquid cash reserves on hand to pay the depositors what they are demanding. Do they go bankrupt?
Usually the answer is no, the bank won’t go bankrupt. This is because a solvent bank has a backup option when they face a liquidity challenge: they can borrow from another bank!
If a bank is solvent but is facing a bank run, then it just has a liquidity problem. It has enough assets to meet all of its liabilities, but not all of its assets are in the right form to meet those liabilities right now. So, another bank that has plenty of liquidity on hand can make use of this situation. They can lend liquid cash to the troubled bank, in exchange for taking some of their safest loans as collateral. Or they could just outright buy some of their loans. They help the illiquid bank by temporarily or permanently transforming some of its illiquid assets into liquid reserves, so that it can continue to meet depositor withdrawals. Of course, the savior bank will charge interest for making these loans to the troubled bank, or will buy the assets from the troubled bank at a discount if it needs to be done fast. Liquid reserves therefore flow from one bank that doesn’t need them right now to the bank that is facing the liquidity drain and does need them right now.
That is how a highly-levered fractional reserve banking system can last for decades. The relatively small percentage of liquid reserves relative to total system-wide deposits can move around quickly to wherever there happens to be a need for them. The system might be levered 10-to-1 or more, meaning a lot of the depositors’ money is illusory, but wherever liquidity problems happen to be occurring, multiple banks that have excess liquidity can respond to address that liquidity shortfall somewhere in the system, for a profit.
The real trouble begins when the entire banking system is highly levered, and some sort of shock occurs. This is what happened in 2008, leading into the global financial crisis. The problem in that scenario is that very few banks have enough excess liquidity to rescue any significant portion of other banks, and just as importantly, banks don’t trust each others’ solvency enough to make loans to each other even if they do have some excess liquidity.
A solvent bank can fail in that scenario, because it can become unable to meet customer withdrawals, while also being unable to obtain enough liquidity from another bank. And this can happen on a broad scale. The more bank failures that begin to occur, the less likely that any remaining liquid banks will lend money to any other banks, and the more people there will be that want to pull their deposits out into cold hard cash.
This, fundamentally, is the heart of the issue with fractional reserve banking. Even if a fractional reserve bank does everything right in terms of responsible lending and maintaining solvency at all times, it’s still ultimately playing a game of probabilities, and if more depositors want their money back at once, the bank relies on the kindness of strangers (other commercial banks or the central bank) to bail them out.
Another way of putting it is that a collection of dozens of solvent well-run banks can still be broken as a group if they become greatly overleveraged compared to how many depositors want their money back at once. This is because each bank’s fractional reserve lending practices feeds another bank’s fractional reserve lending practices, and the system as a whole has 10x as many claims for money as there is base money.
Every deposit, every loan, and every bond, is a claim for a certain amount of money within a specified timeframe. And in a fractional reserve banking system, the number of claims greatly exceeds the base amount of money.
The well-known game of “musical chairs” involves children walking around a set of chairs while music plays, and there is one fewer chair than there are children. When the music stops, the children have to scramble for a seat, and one slow or unlucky child is left without one, and gets excluded from the next round. And then a chair is removed, and the game is repeated, with one fewer child and one fewer chair. These rounds continue until there are two children competing for one chair, and one child wins due to speed or luck.
Suppose there is a game of musical chairs with thirty children and only three chairs. How long would it last? Well, if the music keeps playing, it could last for hours. There are plenty of chairs relative to the number of children who need them, since the number of required chairs is technically zero when the music is playing. Realistically, maybe a child is allowed to sit down for one minute every half hour to take a break, and then has to get back up and keep going. So maybe a few chairs is enough to support a very large number of children for a very long time, while the music plays.
That’s how a fractional reserve banking system works. For as long as the music plays, a relative small number of chairs (bank cash and reserves) can meet requirements for the number of children needing to sit (bank deposits running into a liquidity problem). The chairs are basically allocated to whichever small percentage of children need to briefly sit down and take a break at any given time. It’s only when the music stops that the mayhem begins. Music is like trust in this analogy.
When the base layer itself can’t be printed by any centralized authority, and a financial crisis occurs and the music stops, then the number of claims begins defaulting down closer to the amount of base money. For example, throughout the first half of American history, there was no central bank. Base money was gold and silver. When banks failed and couldn’t secure a bailout from another bank, they actually failed. There were booms and busts, and cumulative price inflation was zero, and the United States went from a new-world emerging backwater in the late 1700s to being the world’s largest economy by the late 1800s.
When the base layer can be printed, the reverse happens. When the music stops, instead of claims collapsing down toward the amount of base money, the amount of base money can be expanded by the central bank to provide enough new liquidity so that most of the fractionally reserved claims can be made money-good. They can create new base money and either loan it out to banks for securities, or they can outright buy securities with it.
In the 1930s, this happened slowly. The amount of claims collapsed for three years, and broad money supply fell by a third as a significant number of banks failed and depositors were wiped out, but by 1933 the pain became enough and the government and central bank reduced the dollar-to-gold peg and printed a ton of new base dollars to start making the remaining two-thirds of bank claims money-good.
In the 2008 crisis, this happened quickly. Banks started to fail, and central authorities responded immediately with new base money creation. Virtually all of the claims were made money-good from the start.
This chart shows a century of fiscal and monetary policy, and shows the similarities between 1929 and 2008:
Another of my favorite charts looks at total debt (public and private combined) and its relation to base money. For decades, the total amount of debt increased smoothly, with no decreases ever. In 2008/2009, for the first time in the post-WWII era, total debt decreased slightly, by around one percent. The Federal Reserve cut interest rates to zero. This was a generational banking crisis, with virtually all banks highly leveraged and unable or unwilling to help each other. Banks started to fail. It would have caused cascading bank failures, depositor wipe-outs, massive defaults, etc. But instead they just printed more base money immediately, and centralized policymakers got to decide where that money went.
That’s the difference between a decentralized base money that can’t be printed, and a centralized base money that can be printed.
-With decentralized un-printable base money, excessive fractional reserve claims default and collapse down toward the amount of base money during a major crisis.
-With centralized printable base money, the amount of base money is expanded during major crises to make most claims money-good. Total base money and total broad money keep growing forever. Holders of that currency are losers relative to other assets via debasement. People and institutions who hold truly scarce assets and have select access to credit to short that currency on a long-term basis are the winners.
Most analysis occurs after the fact. Analysts say, “Look how important it was to increase the base money! If we hadn’t done that, even more defaults would have occurred!” And they’re not wrong in that specific assessment.
But a second-level thinker should go back a step prior to this and ask why such generational banking crises occur with such clear patterns. Why was there an order of magnitude more claims for money than there was base money prior to the crash? Who allowed this to occur? For what reason did it occur? Would it have been better to avoid the bubble before it inflated, rather than resolving the bubble after it popped by devaluing everyone’s money via central decree?
Fractional Reserve vs Full Reserve Lending
When people hear the term “full reserve banking”, many of them assume it means a bank that doesn’t lend and instead just acts as a custodian. But that’s not the case.
Ludwig von Mises used the term “commodity credit” and “circulation credit” to describe two different types of lending, the first being full reserve and the second being fractional reserve.
I think “duration-matched lending” and “duration-mismatched lending” is a more intuitive description today, even though the terms are a bit longer. That’s what describes the difference between full reserve banks and fractional reserve banks.
Full reserve banks can lend, but they do so with duration-matched liabilities.
There are two primary types of bank liabilities owed to customers: demand deposits and time deposits.
-Demand deposits can be withdrawn “on demand” at any time within banking hours. Your checking account and most types of savings accounts are types of demand deposits. You receive money, you pay money, and you don’t have to think about it because you assume (somewhat incorrectly) that the bank has it all there. If you’re a consumer, your ability to pay your mortgage bill or pay your electricity bill is often reliant on your bank having the money they say is there in your checking account at the time you need it. If you’re a small business, you pay your employees’ paychecks and other immediate expenses out of your demand deposits. And many business accounts are above the FDIC insurance limit. Your bank better be able to meet all your needs at any time.
-Time deposits or “certificates of deposit” are locked up for a specified amount of time, and are intended for savings rather than immediate liquidity. You don’t use this for your ability to pay bills or meet payroll; you use it to generate a bit of return on your longer-term savings.
In fractional reserve banking, banks can fund their illiquid long-term loans with demand deposits. They tell their demand depositors that they can withdraw their money at any time, but in reality they are playing a probability game as they put most of that money into longer-term illiquid loans. They hope that a lot of demand depositors won’t want their money back at once, or if they do that they’ll be able to borrow from another bank by shuffling base money around the system or get bailed out by the central bank with brand new base money. Fractional reserve banking is premised on making promises about liquidity to depositors that can’t always be kept.
In full reserve banking, banks only fund loans with duration-matched time deposits or similar investment contracts.
-Any demand deposits they have are held 100% in liquid money, fully available on demand even if every depositor comes and asks for it at once.
-Any loans they make are funded by time deposits of the same duration to the loan or longer, and with risks clearly outlined to the one that buys the certificate of deposit.
A full reserve bank can never run into a liquidity crisis; every asset is backed by a liability of similar or longer duration. It could still run into a solvency problem on its time deposits if they make bad loans, but then if the legal structure is set up with this in mind, it will only affect the savings/investments of people who chose to lock up capital for investment purposes; it won’t affect those who are holding fully-reserved demand deposits to meet their immediate needs.
More notably, the practice of using illiquid loans to back liquid deposits, and thus creating a liquidity mismatch, is the source of broad money creation. It’s what allows banks to print a lot more IOUs for money than the amount of base money that exists, because they are backing those IOUs by non-monetary things such as loans. Central banks create base money, and commercial banks create broad money by fractionally reserving that base money.
There’s an old phrase, “don’t eat your seed corn”. If you’re a subsistence farmer, you can’t eat everything you grow. You need to segment some of what you grow to be put back into the ground to seed the next year’s crop. That’s the “seed corn”.
Fractional reserve banking tells depositors that they can eat their seed corn and also plant it; that they can have their liquidity while also lending. The unspoken catch is that holders of money keep getting devalued by new credit creation, and after several decades the system blows up and a ton of new base money is created to make most of the credit claims money-good.
Full reserve banking tells you that your “eating corn” and your “seed corn” indeed need to be separate. In order to lend, you need to save. Deposits are fully backed by liquid reserves, and the amount of capital available for lending to borrowers is constrained by the amount of capital saved and intended for lending. New base money never needs to be created in this system; there is inherent preservation of money. It’s a more equity-based society overall rather than a debt-based society.
A debt-based society can grow more quickly at times than an equity-based society, but an equity-based society is more resilient.
Imagine for example that you run a debt-fueled business. You keep borrowing money, and reinvesting that money, and you are growing quickly, but you have constant stress regarding what happens if your debtors stop refinancing you? What if interest rates rise? What if your profit drops below your required interest expense due to a pandemic? You need to grow your die; you can’t just stagnate.
On the other hand if you are running an equity-based profitable business, you care less about consistent growth. You own your assets, and either you don’t have debt or you only make use of it judiciously, like you might borrow briefly to expand a business line or something, but overall you are not reliant on lenders and don’t carry large persistent debts. Your business might have up years and down years, and that’s okay. You’re solid, through those ups and downs. You seek to grow, but not at the cost of instability.
Do Any Full Reserve Banks Exist?
In the modern world, full reserve banks don’t really exist per se. The whole world operates on a fiat currency fractional reserve system, which is built upon central banks that can increase the monetary base and put out liquidity problems with debasement.
Some banks have attempted to be full reserve banks, but they have been denied access to reserve accounts at the Federal Reserve. The Narrow Bank and Custodia Bank are examples of banks that have applied for reserve accounts at the Federal Reserve with the intention of holding all demand deposits there, but have been denied.
However, many types of credit funds or “shadow banks” operate similarly to the lending side of how full reserve banks would operate. For example, the Blackstone Private Credit Fund, and many other funds like it, takes investor capital and uses it to lend to businesses. They aim to provide investors with occasional liquidity, meaning that they aim to let investors withdraw part of their investment where possible. However, they firmly say that this liquidity is not guaranteed:
- You should not expect to be able to sell your shares regardless of how we perform.
- You should consider that you may not have access to the money you invest for an extended period of time.
- We do not intend to list our shares on any securities exchange, and we do not expect a secondary market in our shares to develop prior to any listing.
- Because you may be unable to sell your shares, you will be unable to reduce your exposure in any market downturn.
- We have implemented a share repurchase program, but only a limited number of shares will be eligible for repurchase and repurchases will be subject to available liquidity and other significant restrictions.
-Blackstone Private Credit Fund summary of risk factors
The reason they cannot guarantee it is because their loans on average are longer than three months, and at any given time only a percentage of them are being paid back and creating liquidity for the fund. The fund managers can’t anticipate how many investors might want to pull their capital out in a given quarter. So under normal operating conditions where only a small percentage of investors are pulling out per quarter, they can typically meet investor redemptions when asked on a quarterly basis. But if too many investors want to pull out in one quarter, then they’ll need to wait longer until more of the loans mature and provide liquidity.
Various funds that take investor money which is intended for longer-term saving, clearly tell investors about the various the liquidity limitations of it, and then use that money for lending, are basically doing what Mises would refer to as commodity credit, or what we can simply refer to as duration-matched lending.
Technological Impacts on Lending
What we use as money, and how fast monetary claims move throughout the financial system, can affect the type of lending that we do.
In the gold standard era, a key problem was that gold was slow to transport and hard to authenticate down to its core, and so the gold monetary base moved very slowly while the claims for it continued to move very quickly. This was especially true after the invention and adoption of the telegraph, which allowed for instant communication (and thus instant transactions) across a country and then across the oceans.
In the 1875 book Money and the Mechanism of Exchange, William Stanley Jevons highlighted how leveraged the gold-backed system of his day was. Britain was the center of the Western financial world at the time, and the combination of the printing press and telegraph had allowed for ever-centralized finance. Transactions and claims could move around the developed world at the speed of light, while physical transport and verification of gold could only happen at the speed of matter. As a result, few people ever withdrew their gold, and claims for gold (meaning bank deposits and banknotes for various gold-backed currencies like dollars, pounds, and so forth) kept growing relative to the gold monetary base and generally netted against each other with high velocity.
But Jevons warned that, seeing as how the system was levered 20-to-1, if even 5% of people come and ask for their gold, it would all go insolvent:
It is requisite, too, that our bankers, financiers, and merchants should regulate their operations with a thorough comprehension of the immense system in which they play a part, and the risks of derangement and failure which they encounter by over-severe competition. No one doubts that alarming symptoms have during recent years presented themselves in the London money market. There is a tendency to frequent severe scarcities of loanable capital, causing sudden variations of the rate of interest almost unknown thirty years ago. I will therefore in the next chapter offer a few remarks intended to show that this is an evil naturally resulting from the excessive economy of the precious metals, which the increasing perfection of our banking system allows to be practised, but which may be carried too far and lead to extreme disaster.
[…]
The metals took the place of other commodities as currency, and delicate considerations began to enter concerning token and standard coins. From metallic representative money, we passed to paper representative money, and finally discovered that, by the cheque and clearing system, metallic money was almost eliminated from the internal exchanges of the country. Pecuniary transactions now present themselves in the form of a room full of accountants, hastily adding up sums of money. But we must never forget that all the figures in the books of a bank represent gold, and every creditor can demand the payment of the metal. In the ordinary state of trade no one cares to embarrass himself with a quantity of precious metal, which is both safer and more available in the vaults of a bank. But in international trade, gold and silver are still the media by which balances of indebtedness must be paid, and serious consequences may arise from any disproportion between the amount of transactions carried on, and the basis of gold upon which they are settled.
[…]
It is quite apparent, therefore, that the tendency is to carry on a greater and greater trade upon an amount of metallic currency which does not grow in anything like the same proportion. The system of banking, too, grows more perfect in the sense of increasing the economy with which money is used. The competition of many great banks, leads them to transact the largest possible business with the smallest reserves which they can venture to retain. Some of these banks pay dividends of from 20 to 25 per cent, which can only be possible by using large deposits in a very fearless manner. Even the reserves consist not so much of actual coins or bank-notes in the vaults, as of money employed at call in the Stock Exchange, or deposited in the Bank of England, which again lends the deposits out to a certain extent.
Now the larger the trade which is carried on, the larger will be the occasional demand for gold to make foreign payments; and if the stock of gold kept in London be growing comparatively smaller and smaller, the greater will be the difficulty in meeting the demand from time to time. Such is, I believe, the whole secret of the growing instability and delicacy of the money market in this country. There is a larger and larger quantity of claims for gold, and comparatively less gold to meet them, so that every now and then there is a natural difficulty in paying claims, and the rate of interest has to be suddenly raised to induce those who have gold to lend it, or to induce those who were demanding it to forego their claims for a time.
[…]
Mr. R. H. Inglis Palgrave, in his important “Notes on Banking,” published both in the Statistical Journal, for March, 1873 (Vol. xxxvi. p. 106), and as a separate book, has given the results of an inquiry into this subject, and states the amount of coin and Bank of England notes, held by the bankers of the United Kingdom, as not exceeding four or five per cent. of their liabilities, or from one twenty-fifth to one twentieth part. Mr. T. B. Moxon, of Stockport and Manchester, has subsequently made an elaborate inquiry into the same point, and finds that the cash reserve does not exceed about seven per cent. of the deposits and notes payable on demand. He remarks that even of this reserve a large proportion is absolutely indispensable for the daily transactions of the bankers’ business, and could not be parted with. Thus the whole fabric of our vast commerce is found to depend upon the improbability that the merchants and other customers of the banks will ever want, simultaneously and suddenly, so much as one-twentieth part of the gold money which they have a right to receive on demand at any moment during banking hours.
-Money and the Mechanism of Exchange, 1875
The leveraged system indeed broke a few decades later in World War I. The gold peg was defaulted upon, and claims that depositors thought they had for gold were instead inflated away in value. Thanks for playing, try again next time.
In the fiat currency era, it’s a different problem. There is no pretense about soundness anymore. The base layer of money is determined by the central bank, and the amount of claims or IOUs for base money keeps growing, and whenever the system breaks, more base money is created by the central bank. Monetary inflation happens more quickly, prices go up more quickly, and those who have privileged access to short/borrow the currency at low long term rates benefit the most. Broad money keeps expanding gradually, and base money expands in stepwise fashion when needed.
This happens on a much faster cycle throughout developing countries today. Many people within those countries just live with constant double-digit inflation, and occasional triple-digit inflation or more. They have to keep running as fast as they can on a treadmill of wage devaluation and savings devaluation, meaning people need to keep trying to get higher nominal wages to compensate for the devaluation of the unit of their wages, people need to invest their savings into something else because money itself keeps losing its value, and businesses need to keep re-adjusting their pricing contracts with other businesses and with their customers.
What would it look like if there was a hypothetical “bitcoin era” or something akin to it? In other words, what if there was scarce money that moved as fast as fiat currency domestically and internationally, and it was widely adopted by the world, and resistant to all sorts of threats like quantum computers and too large for governments to stop?
It’s hard to say for sure, but we can identify a few features.
Firstly, new bitcoin can’t be created by any central authority; their supply is limited to 21 million by the widely distributed node network. Any attempt to change the supply is a “hard fork”, meaning they instead create their own separate cryptocurrency that has little liquidity or security and that is not recognized by the existing nodes or miners.
Secondly, unlike gold, bitcoin is easy to transport and verify. Bitcoin can be moved around and fully audited globally by the Bitcoin network typically within an hour, 24 hours a day, 7 days a week, 365 days a year. Unlike gold, if you hold bitcoin without a bank/custodian, you can still send bitcoin transactions to other people globally peer-to-peer through the distributed mining network as long as you have internet access, so you are less reliant on banks/custodians in general. Even if you are without internet access for a period of time, your storage of coins is still safe.
These traits together mean that 1) there is less necessity for large savers to hold bitcoin in a custodian and 2) bitcoin can be swiftly withdrawn from a custodian and still used as money in a peer-to-peer manner. Therefore, fractional reserve banking with bitcoin as the base layer is incredibly dangerous, and will likely blow up whenever tried at scale (as it did in 2022 with the cascading failure of most of the cryptocurrency lenders).
That doesn’t mean that bitcoin loans can’t happen, but it means that trying to fund bitcoin loans with demand deposits is folly. To the extent that some entity wants to borrow bitcoin, the lenders should be using investor capital that is appropriately duration-matched, or that at least comes with significant qualifiers about access to liquidity.
Excessive amounts of fractionally reserved claims for bitcoin will quickly collapse down toward the base amount of bitcoin during major crises, rather than large multiples continuing for decades (in the case of gold), or the base layer being easily expanded by a centralized authority when needed (in the case for fiat currency).
The harder and faster a money is, the less likely it is to build a high ratio of broad money to base money on it. I refer to this as the “rekt cycle”, referring to how long or short the time period it will typically take a monetary system to get wrecked by trying to fractionally reserve it at scale depending on the qualities of the underlying base money that’s being fractionalized.
Gold has a long rekt cycle because it’s slow. Only a small percentage of participants tend to go through the hassle and expensive of withdrawing and verifying large amounts of gold, and so the system can operate on rehypothecation for decades before it eventually blows up.
Fiat currency also has a long rekt cycle because it’s flexible. The central bank can increase the amount of base money temporarily or permanently in order to put out any liquidity problems that happen before it starts to cascade. It takes a rather major destruction of productivity for a fiat currency system to hyperinflate. It’s designed to have a constant slow leak, which greatly delays any major breakage.
Bitcoin has a short rekt cycle because it’s as fast as fiat but unprintable like gold. It’s the worst possible type of money to fractionally reserve, since the time until such folly is tested and revealed will likely be rather short, and historically has been.
Productive vs Arbitrage Loans
“But without lending, wouldn’t the economy grow more slowly?” is a question I often receive when talking about fractional reserve banking.
Lending has existed since before the dawn of writing, and will continue to exist for the foreseeable future, regardless of what we use as money. Credit is a useful way to make transactions more efficient, credit smooths over a lot of monetary inefficiencies, and credit makes capital more flexible.
But the technology that we have available to us, and thus what we use as money, determines the nuances of how lending works and what types of things we use lending for.
In the 1800s, there was less emphasis on financial arbitrage. You made money from building things. Sometimes that meant borrowing money, like borrowing money for railroad development. And a lot of this was due to the type of money used; you don’t borrow gold unless you plan to outperform gold with what you invest in. And gold is pretty scarce, so you better have a high-ROI use case for borrowing it.
Today, there is a ton of emphasis on financial arbitrage because most reasonably scarce assets outperform fiat currency. The middle-men make a lot of the money. The most lucrative field for a reasonably smart school kid with a bachelor’s degree has been finance for the past few decades, not engineering or medicine. Unless you have the skills to be an outlier tech entrepreneur, the quickest path to becoming a multimillionaire within the past four decades has been to go into investment banking, private equity, and/or fund management.
Coca Cola (KO) was founded in 1886, is a conglomerate now, and has been profitable for decades straight, and yet they have nearly $50 billion in long term debt. Why? Because they choose to. They are arbitraging it. Borrowing money at low interest rates with five-year bonds, ten-year bonds, twenty-year bonds, and thirty-year bonds, and using it to buy back shares or make business acquisitions has been good for returns. Coca Cola shareholders are basically shorting the dollar for the long run at low interest rates, and using that money to buy attractive business equity that yields better returns. Most longstanding blue-chip corporations do this. The best products Coca Cola ever sold were their bonds.
I have two mortgages on two properties that collectively make up a relatively small percentage of my net worth. One in the United States and one in Egypt. I live in both of them, depending on the time of year, and the one in Egypt has many family members living in it year-round. I don’t need these mortgages, and could sell some stocks or save up a couple fiscal quarters of cash income to pay them off easily. Why don’t I? Because I have these loans primarily for arbitrage; the interest rate on the loans are so low (around 3.5% in both cases) that I can invest the money elsewhere at low risk for better returns. The American one is for a duration of thirty years and denominated in dollars, and the Egyptian one is for a duration of seven years and denominated in the Egyptian pound. Holding my 3.5% mortgages while holding assets that grow in value at more than 3.5% per year is a type of arbitrage that I choose not to pay off early.
In a world filled with weak money, the majority of debt is for the purpose of financial arbitrage or convenience. And wealthier individuals and larger corporations and governments have more access to cheaper and/or longer duration non-callable debt than poorer individuals and smaller businesses. Multi-decade mortgages, multi-decade corporate bonds, and permanent rolling government debts and deficits are mostly debts of arbitrage and convenience for the sake of shorting the currency and those who hold the currency, not debts of productivity.
Other types of loans can make sense regardless of whether money is strong or weak. A business owner taking out a two-year loan to expand her business for a high-ROI opportunity is productive. Someone taking out a loan for a degree that gives them a big earnings boost and then paying it back is productive. A business having a revolving credit liquidity facility on hand to meet occasional fluctuations in liquidity needs is productive. These types of high-ROI loans make sense regardless of what type of unit of account that a society uses.
Short duration high-ROI debt for productive purposes will always exist, whereas debt that is taken out primarily for the purposes of financial arbitrage (structurally shorting the underlying currency) is one that depends very much on the softness of the underlying unit of account that society uses.
When money is weak, we hold other scarcer assets instead, and we short (borrow) the money to do so wherever possible. When money is strong, we can more readily store wealth in it, and would only only dare borrow it under more thoughtful circumstances.
So the short answer is that productive lending exists in all types of monetary environments, but arbitrage lending only occurs at a massive scale when the conditions are right for it. Arbitrage lending is what significantly reduces when money is strong, not productive lending.
The Debt Bell-Curve
There is a key threshold and a key concept to be aware of when determining how much arbitrage lending is likely to occur in a given monetary system.
The threshold has to do with money supply growth. If a money supply grows more quickly than the typical growth rate of goods and services, then that currency is more likely to be used for arbitrage lending.
The key concept has to do with the lending/borrowing bell curve. I featured this chart in Broken Money:
-When a currency is very weak (the left side of the chart) then few lenders are going to want to lend long-term in it. This is because they don’t trust that they will receive enough value back over time. They’ll keep their lending short-term or they’ll offer loan terms using other currencies.
-When currency is very strong, then few borrowers are going to want to borrow long-term in it for the purpose of arbitrage and consumption. Would you want to take out a 30-year loan denominated in something scarce like gold or bitcoin to buy your house with? If you were Coca Cola, would you issue a 20-year bond denominated in either of those monies and use the proceeds to buy back your own stock to shift more of your capital ratio toward debt and away from equity? Probably not. However, you’d still borrow for a little while if you had a high-ROI business idea, or if you had near-term liquidity requirements and needed some flexibility (examples of productive lending).
-When a currency is of middling strength, that tends to attract the most activity from both lenders and borrowers, and thus results in the highest debt societies. This is where the most arbitrage debt will occur, meaning long duration debt that merely exists as a way to gradually short the fiat currency and/or to overconsume with. This typically occurs when the money supply is growing at a moderately higher pace than goods and services are growing (i.e. just above the aforementioned key threshold), which is what we see in most developed countries today with 6-8% average long-term broad money supply growth.
Final Thoughts
Banks come in all sorts of forms, and the ever-changing nature of technology impacts how banks work over time.
The modern banking system is extraordinarily complex with countless moving pieces, and whole books can (and have) been written on it. But understanding some core principles can cut to the heart of it.
Ultimately, duration matching or mismatching is one of the key things to understand when it comes to the broad topic of lending.
Are loans being made with money that supposedly can be redeemed at any time shorter than the length of those loans? If so, then the lender is relying not just on its own lending decisions (i.e. solvency) but is relying on statistical depositor behavior and the broader system around them, outside of their control. And by doing so, they’re expanding the broad money supply, and contributing to the fragility of the whole system as these fractional reserve practices feed each other.
This is ubiquitous in the current age, and is at the heart of why persistent monetary inflation occurs in every country in the world.
Certain monetary environments incentivize or disincentive that type of lending, and a key part of that monetary environment is the state of technology at the time.