Do Banks Create Money from Thin Air?

By Dan Kervick

It is sometimes said that commercial banks in our modern monetary system create money “from thin air”.  While there is truth in this metaphorical claim, the metaphor can also be seriously misleading, and leads some to attribute powers to commercial banks that are actually retained by the government alone under our system.  It is worth trying to get clear about all this.

Suppose you have some debt to pay; or suppose that there is some good or service that you wish to purchase on the spot.   How can you make the payment?   Clearly you are going to have to hand over something of positive value.  In order to pay off a debt you will have to possess some asset that you can transfer to your creditor in a way that discharges your obligation.  Similarly, in order to purchase some good or service on the spot, you will need to possess some asset that the seller is willing to accept in exchange for the good or service that is sold to you.  The asset you use might be a very specific, unique and particularized kind of thing, depending on the nature of whatever implicit or explicit contract you have with the creditor or seller.  But more often than not, you will pay with a generic and widely accepted type of asset, once which in used routinely to buy things and discharge debts, and that seems to exist mainly or solely for those very purposes.  Such payment assets have existed in many different forms historically, along with different kinds systems for generating, storing, transferring and regulating these assets.  We customarily call these assets “money”.

Payment assets obviously possess value, for if they didn’t they would not be accepted in exchange for other things of value.  As a result, it is rare that one can acquire these payment assets for nothing.  Usually you need to give something up in return.  The money you use to buy a car was most likely obtained either in exchange for some good you traded for the money, or for some labor service you provided to an employer.  You can also obtain payment assets in exchange for promises – even for promises to hand over an even greater quantity of the very same kind of asset as some point in the future.

Among the most common ways of paying one’s own debts or paying for goods and services is to pay with the debt of a third party.  For example, suppose I give a signed note to Pat Brown that says, “I agree to pay Pat Brown, or the bearer of this note, $100 on demand.”  And suppose Pat has a $100 debt to the corner grocer.  Pat might attempt to pay the grocery tab with that IOU.   If the grocer accepts the note, then Pat has paid a debt with a debt.

Once I have issued and signed the note, and Pat has accepted it, I have a liability.  And if that liability is of a kind that Pat is legally permitted to sign over or otherwise transfer to a third party, it is said to be “negotiable”.  If Pat does whatever is legally required to convey the IOU to the grocer to pay the grocery tab, the grocer becomes a “holder in due course” of the negotiable liability.  At that point, Pat no longer has a debt to the grocer.  But I still have a debt.  I previously had a debt to Pat; but now I have a debt to the grocer.  That’s what it means for a debt liability to be negotiable: the creditor who holds that debt as an asset can transfer it to a third party, so that the debtor ends up owing the same debt to a new creditor.

Where did the IOU come from?  Was it created from thin air?  More or less.  Yes, a certain amount of paper and ink and work might have been involved in producing it, so its production didn’t come with zero cost.  But typically the cost of making the promise will be so low in proportion to the amount promised, that we don’t go far wrong in thinking of the promise as having been produced from thin air, created ex nihilo as it were.  And it is not as though to make a promise I have to pull the promise out of my pre-existing promise stash.  The promise doesn’t really come from anywhere.  There is no effective limit on my ability to make promises beyond the length of my lifespan, and the number of people in the world to whom I might make them.  But the fact that my ability to make promises is virtually unlimited does not mean that my ability to get my promises accepted is virtually unlimited.  Some people and some commercial entities have a much easier time getting their promises accepted than do others.  Making promises is a lot easier than making credible promises; and accepting a promise that you personally find credible might be a lot easier than trading such a promise to someone else.  To trade away any promises you possess, you must be able to convince others that they should find the promise just as credible as you did when you first acquired it.  Also, once I have made a promise and had it accepted, I am bound in a way I wasn’t before.  The holder of the promise possesses a legally binding claim on my assets.

A bank deposit account is such a promise, and it therefore represents a debt or liability of a bank to the account holder.  If you have a bank deposit account then you are in possession of a promise by the bank to pay you a defined amount.  If it is a demand deposit account, then the promise is to pay on demand.  The agreement you have made with the bank specifies the conditions under which you can demand payment on the debt or make use of that debt – and those specified conditions usually include the right to transfer part or all of that debt to a third party.  Most banks have a good track record in paying the debts represented by their deposit accounts, and there are also reliable government guarantees in place to pay most of the deposit account debts that insolvent banks find themselves unable to pay.  Thus bank debt functions as a fairly reliable and very widely accepted payment asset.  Bank deposit liabilities are a form of money.

Of course, when you transfer the bank’s debt to a third party, you don’t literally transfer your bank account to that party.  Instead you give the party some financial instrument, or send some electronic payment instruction, that ultimately results in your bank having a smaller debt to you, but a correspondingly larger debt to the payee.  For example, you might have a demand deposit account with $10,000 in it and give someone a check drawn on that account for $1000.  If the recipient has an account at that bank, they can present the check – your payment order – to the bank, following which your bank reduces the amount in your account by $1000 and increases the amount in the recipient’s account by $1000.  In other words, the bank’s debt to you has been reduced by $1000 and its debt to the recipient has been increased by $1000.  You have paid the recipient with a debt.   However the debt you paid with was not your debt.  Rather the debt you paid with was the debt of some other debtor – the bank – and is a debt obligation of which you were the initial creditor, not the debtor.

Note that the person who took your check to the bank might not have been willing to accept payment from the bank in the form of further bank debt, that is, in the form of an amount credited to their account at that bank.  Instead they might have demanded cash.  Generally speaking, that’s up to them.  If they accept a deposit balance, then the bank still has a debt to them.  If they are paid in cash, then the bank’s debt has been discharged, but the bank has had to surrender a payment asset in order to discharge it – in this case money from its vault.

Now this raises a question.  As we have already seen, one way to pay a debt is with another debt – more specifically with a negotiable liability.  But if I pay a debt with my bank’s debt, how does my bank pay that debt when they are required to pay it off?  Or looked at slightly differently, given that my bank’s debt to me can serve as my payment asset, what kind of thing can serve as my bank’s payment asset?  And when do banks pay the debts represented by their depositors’ account balances?

For most banks in the US banking system there are two fundamental types of payment assets banks use to pay their debts.  But perhaps another way of putting it is that there are two main forms of one fundamental type of payment asset.  That payment asset consists in the negotiable liabilities of the central bank, i.e. the Fed.   These liabilities come in two forms: the deposit balances that commercial banks in the Fed system hold at the twelve Federal Reserve Banks, and the paper currency notes that the Fed also issues.   Just as you and I possess payment assets in the form of commercial bank account balances, commercial banks possess payment assets in the form of central bank account balances.  In each case, that balance is an asset of the holder of the account and a liability of the depository institution at which the account it is held.  You and I can pay our debts with commercial bank debt; commercial banks pay their debts with central bank debt.  Note, however, that one form of central bank debt is widely held by both commercial banks and the non-bank public: the currency notes that banks hold in their vaults and that you and I hold in our wallets.

But when do banks pay the debts represented by their depositors’ accounts?  We have already considered one occasion: someone presents a check at a bank and receives either cash or a positive increment to their account balance at that bank.   Another way in which bank’s pay their debts is by rolling them over into debts of the same kind or a different kind, as when the promise represented by a certificate of deposit is redeemed in the form of an increment of dollars to some demand account balance.

But banks also pay their debts when they are ordered by their depositors to pay someone who does not have an account at the same bank.  Suppose you pay $300 to Bob’s Propane with a check or electronic check card, and Bob’s Propane holds its accounts at Maple Valley Bank, while your account is held at Ridge Bank.  While the exact procedures for clearing and settling this payment differ according to the mechanisms used, the end result is the same.  Bob will end up with $300 more in his Maple Valley Bank account, and you will end up with $300 less in your Ridge Bank account.  But banks are not in the habit of giving away money for free, and Maple Valley Bank is not going to increase its deposit account liability to Bob’s Propane $300 without getting something in return.  In addition, Ridge Bank does not receive the benefit of reducing its liability to you by $300 without giving something up in return.   What Maple Valley Bank receives and Ridge Bank gives up is a $300 balance in their own deposit accounts at the Fed.  You paid Bob with Ridge Bank’s negotiable liability; Maple Valley Bank then gave Bob its liability in the form of a deposit balance in exchange for Ridge Bank’s liability, and demanded payment from Ridge Bank. Finally, Ridge Bank paid by directing its bank, the Fed, to reduce its own account balance by $300 and increase Maple Valley Bank’s account balance by $300.

But we often hear that banks create money “from thin air”.  Doesn’t that mean that a bank never has to obtain payment assets from some external source in order to pay its debts?  Can’t the bank simply create its own payment assets out of the thinness of air, so to speak, and pay its debts with those newly-created assets?  Aren’t banks in this sense self-funding?

No, banks are not self-funding, either individually or in the aggregate.  The “out of thin air” language, while containing elements of truth, can be extremely misleading, and people using this language sometimes woefully under-represent the significance of central bank liabilities and the government in the US financial system.  Banks can indeed create deposit account liabilities from thin air, just as you and I can create liabilities from thin air when we issue IOU’s and someone accepts them.  But those deposit liabilities are debts of the bank, just as the IOUs that you and I issue are our debts.  And these bank debts are not just so-called debts or pro forma debts.  They are real debts which banks must and do routinely pay off in the course of doing everyday business; and the assets a bank uses to pay these debts come from sources external to the bank.  A bank cannot simply manufacture its own payment assets from thin air.

Suppose our old friend Ridge Bank, for example, wishes to purchase a fleet of company cars. It might be able to pay for the cars by creating a deposit account for a car company and crediting that account with the total purchase price for the fleet.  But that account balance is itself a debt of the bank.  Yes, the bank can pay for the cars with this debt, but that is no different in principle than the fact that you and I can pay for a car with an IOU.  These debts are liabilities that can and will be extinguished over time by surrendering assets that the issuer of the liability doesn’t create or control.  It’s always possible that the car company will just allow the balance to sit in its account indefinitely.  But more likely, the company will begin to spend the money.   Some of the expenditures might be to people or companies who have accounts at the same bank, which means balances just move from one Ridge Bank account to another Ridge Bank account, without the deposit liability being discharged.  But over time a large proportion of that balance will either be withdrawn in the form of cash or used to pay people who bank elsewhere, and in each case the bank will have to surrender some externally created asset to meet its obligation.  And note that even if the liability just sits unredeemed in the car company’s account for an extended period of time, the existence of those liabilities reduces the bank’s equity, and thus reduces the degree to which the bank’s owners profit from the bank’s operations.

People who are fond of saying the banks create money “from thin air” often seem to suggest that banks are no different than the government in that regard, and can thus obtain valuable monetary assets simply by manufacturing them ex nihilo, in effect profiting from pure seigniorage in the way a currency-issuing government can.  But this picture is wildly inadequate.  If banks could simply summon their assets into existence out of the aether, then every bank in the country could be as rich as an Arabian Gulf emir, manufacturing money at will to purchase solid gold chandeliers, 100-story luxury high rises, Olympic swimming pools, indoor ski slopes, and a personal entourage of world-renowned chefs, attendants and masseuses.  The sky would be the limit.  But clearly this is clearly not the case.  There is a lot to complain about with regard to banking; lots of people in the banking system are making completely unwarranted profits from a massively bloated and exploitative financial system.  But the wrongness here comes from the banking system’s ability to suck, squeeze and swindle assets from others; not from its simply conjuring these assets out of nothing.

There are several features of the existing banking system that sometimes lead to confusion about the role of commercial banks liabilities in our existing system, and about their dependence on central bank liabilities.  We have space to consider just two of them: netting and government deposits at commercial banks.

Netting.   Suppose Cogswell Cogs owes $50,000 to Slate Quarry for a delivery of gravel, and Slate Quarry owes Cogswell Cogs $60,000 for a delivery of cogs.  The two companies might each issue separate payments of $50,000 and $60,000 respectively to settle their obligations.  A more efficient method of settling the obligations, however, would be for both companies to agree to use the Cogswell Cogs debt to reduce the Slate Quarry debt by $50,000.  Slate then pays Cogswell the net $10,000 balance and their business is terminated.

Banks can do the same thing.  In the US, registered banks can make use of CHIPS, the Clearing House Interbank Payment System.  CHIPS has its own account at the Fed, which participating banks pre-fund at the beginning of every business day by transferring money from their own Fed account to the CHIPS account.  Net daily payment balances are calculated as the resultant of all of the payment obligations the participating institutions owe to one another, and payments are made by CHIPS by the end of the day to banks that end up with a net positive closing position.  If a bank has a negative closing position – that is, if the amount pre-funded is insufficient to cover that days payments – then the bank pays CHIPS what it owes by making another Fedwire transfer from its Fed account to the CHIPS Fed account.  Because multiple payment obligations are incurred among those participating banks throughout the day, then just as in the case of Cogswell Cogs and Slate Quarry, the actual amount that needs to change hands in a given day is much less than it would be if each payment were processed separately by the gross settlement system Fedwire.

Notice, however, that the system is ultimately dependent on the Fedwire system, and CHIPS just inserts an efficiency-enhancing intermediary between the Fed and the banking system.  Participating banks can settle some of their less time sensitive interbank payments on the books of CHIPS, but they have to settle with CHIPS via the Fed.  And larger, more time-sensitive payments are still settled directly via Fedwire.

Also notice that even in the case of a netting system, financial debts are still settled with assets that are not internally manufactured from thin air by the debtor.  Consider, once again, Cogswell Cogs and Slate Quarry.   To settle their business, Slate paid Cogswell $10,000 and Cogswell paid nothing.  But Cogswell began by owing $50,000.  So does that mean that Cogswell somehow manufactured a $50,000 benefit out of nowhere?   Of course not.  Before they settled, Cogswell held a $60,000 debt from Slate, but at the end of the day it received only $10,000.  Cogswell received a cancellation of its own $50,000 debt in exchange for cancelling $50,000 of Slate’s debt.  In other words, it had to relinquish an asset.


Government deposits at commercial banks.  It is sometimes argued that the US government must be dependent on commercial bank money to fund its various activities and public enterprises, because the US Treasury holds some deposit balances at commercial banks.  But I believe this is a seriously misleading claim.  The government is certainly dependent on private sector economic activity and finance in a more general sense: if there were less private sector economic activity, there would be correspondingly fewer goods and services produced by our society, and thus fewer real assets that the government could make obtain and make use of to carry out its own activities.  But the government is not financially dependent in any fundamental way on commercial bank deposit liabilities to carry out government spending.

To see this, let’s first look at a simplified picture of Treasury taxing and spending, before moving to the more detailed and accurate picture.  The US Treasury has an account at the Fed called the “general account,” and that is the account from which it spends.  Suppose I have an account at Maple Valley Bank from which I pay a $2000 tax obligation to the US government.  Here’s the simplified picture: I send a check to the government, and as a result of the check being cleared $2000 is transferred from Maple Valley Bank’s Fed account to the Treasury general account.  At the same time, my deposit account balance at Maple Valley Bank is reduced by $2000 and so Maple Valley Bank’s debt to me is reduced by $2000.  Thus, Maple Valley Bank has lost both a $2000 asset and a $2000 liability, and experiences no net loss or gain.  But the US Treasury now has $2000 more and I have $2000 less.  The Treasury then spends that $2000 by buying $2000 worth of sticky note pads from Acme Office Supplies, a company which banks at Old Union Bank.   After the various payment operations are completed, Acme’s account at Old Union has $2000 more in it, and $2000 has been transferred from the Treasury general account to Old Union’s reserve account at the Fed.

Now here’s the more accurate picture:   In practice it has been found that conducting government operations in the way just described results in undesirable volatility in bank reserve balances, which interferes with the central bank’s ability to implement its target rate for interbank lending:   So the government has introduced Treasury Tax and Loan (TT&L) accounts.   TT&L accounts are US Treasury accounts at commercial banks designated as TT&L depositories.  Suppose Ridge Bank is such a depository.  Then when I send my $2000 check to the government, it may deposit it in its TT&L account at Ridge Bank.  As a result, $2000 is transferred from Maple Valley Bank’s Fed account to Ridge Bank’s Fed account.  At that point, no reserves have left the banking system.  But as the Treasury spends over time, it continually transfers money from its TT&L accounts to the general account, and then spends from the general account.  As that happens, central bank liabilities first leave commercial bank reserve accounts and then go back into those accounts after the Treasury spends.

Clearly there is no fundamental difference between the simplified system and the more complex system that uses the TT&L accounts as monetary way stations.  The TT&L accounts exist solely to smooth out the flow of central bank liabilities to and from the Treasury general account and commercial bank reserve accounts.   There is no sense in which the Treasury needs the commercial banks to “create” money in those accounts to carry out its taxing and spending operations.

In a broader sense it should be clear that, far from needing to acquire commercial bank liabilities in order to spend, the government doesn’t even need to obtain Federal Reserve liabilities from commercial bank reserve accounts in order to spend, and could alter the existing system if it so chose.  The central bank is itself an arm of the US government and thus liabilities of the Fed held as assets by the Treasury are just amounts owed by one government account to another government account.  That the US government chooses to operate in such a way that payments from one arm of the government are processed on the books of another arm of the government is an administrative and policy choice, not a deep feature of the monetary system.

What is true in the “from thin air” metaphor is that commercial banks are able to initiate the process of expanding deposit balances via lending without first obtaining any additional assets that might be needed to handle the added payment obligations and withdrawal claims that the additional deposit liabilities might impose on the bank.  It can expand the deposits first and acquire the additional assets, if necessary, afterwards.  And, of course, if the bank already possesses excess payment assets, it might not be able to expand its deposit liabilities without acquiring any more payment assets.   It is also important to recognize that while banks obtain some reserve payment assets by borrowing them – either from other banks or directly from the Fed – some of those reserve assets are acquired for “free”: as interest on loans the banks make to the Treasury and as interest on reserve balances they already hold.

But it is crucial to recognize that banks do not and cannot simply manufacture their own assets – whether from thin air or otherwise.  What they manufacture are liabilities; that is, debts.  And they obtain assets from external sources, mainly by trading debts for debts.