JP Morgan's Coffee Machine


It's now widely accepted, though still not universally, that banks create money when they lend. But it seems to be much less widely known that they also create money when they spend. I don't just mean when they buy securities, which is rightly regarded as simply another form of lending. I mean when they buy what is now colloquially known as "stuff". Computers, for example. Or coffee machines. 

Imagine that a major bank - JP Morgan, for example - wants to buy a new coffee machine for one of its New York offices (yes, it has more than one). It orders a top-of-the-range espresso machine worth $10,000 from the Goodlife Coffee Company, and pays for it by electronic funds transfer to the company's account. At the end of the transaction JP Morgan has a new coffee machine and Goodlife has $10,000 in its deposit account. 

What exactly is this money, and how is it created? I had a long argument with people on twitter who insisted that JP Morgan would pay for the coffee machine with reserves. JP Morgan can't create reserves, it can only borrow or buy them. So if JP Morgan pays for the machine with reserves, the reserves must already exist. No new reserves will be created in the course of this transaction. 

But only banks can accept reserves, and Goodlife is not a bank. So although JP Morgan may pay with reserves, what Goodlife actually receives is bank money. As I shall show, this money is newly created in the course of the transaction.  

In fact, JP Morgan does not necessarily pay with reserves anyway. Whether reserves are involved in this transaction at all depends on who Goodlife banks with.

Suppose Goodlife banks with JP Morgan. All JP Morgan has to do is put $10,000 into Goodlife's account and take delivery of the coffee machine. The balance sheet entries are as follows:

JP Morgan

Fixed assets                DR $10,000   (coffee machine)

Customer deposits      CR $10,000   (credit to  Goodlife business deposit account)


Inventory                    CR $10,000 (coffee machine)

Cash at bank               DR $10,000 (increase in balance at business deposit account)

There is no reserve transfer in this transaction. Reserves are used only to transfer funds between banks, so since Goodlife banks at JP Morgan, reserves are not needed. JP Morgan pays for the coffee machine with newly-created bank money, not with pre-existing reserves. 

A full balance sheet analysis would show that JP Morgan'e equity has reduced as a proportion of total assets. In effect, JP Morgan has paid for the coffee machine from its own equity. 

So why did my twitter friends say JP Morgan paid for the coffee machine with reserves? Well, they were assuming that Goodlife didn't bank with JP Morgan. In this scenario, JP Morgan must transfer reserves to Goodlife's bank. 

To show how this works, let's assume Goodlife banks with Citibank. The accounting entries for the two banks and Goodlife now look like this: 

JP Morgan

Fixed assets                DR $10,000 (coffee machine)

Reserves                     CR $10,000 (payment to Citibank)


Reserves                     DR $10,000 (payment from JP Morgan)

Customer deposits      CR $10,000  (credit to Goodlife business deposit account)


Inventory                    DR $10,000 (coffee machine)

Cash at bank               CR $10,000 (increase in balance at business deposit account)

 As far as Goodlife is concerned, JP Morgan has paid for the coffee machine with bank money. But in fact, JP Morgan has transferred reserves to Citibank, and Citibank has created a new deposit in Goodlife's account. So from JP Morgan's point of view, it has paid for the coffee machine with reserves. 

Note that in this case, the entries for JP Morgan, like those for Goodlife, are entirely on the asset side of the balance sheet. In effect, JP Morgan has exchanged reserves (asset) for a coffee machine (asset). Its balance sheet has not increased in size and its equity buffer is unchanged. 

But Citibank's balance sheet has increased in size. It has new reserves (asset) of $10,000 and has created a new customer deposit (liability) of $10,000. Its equity buffer is now smaller as a proportion of total assets. Citibank has increased its leverage to accommodate JP Morgan's purchase of a coffee machine. 

In both scenarios, new bank money is created when JP Morgan buys a coffee machine. But in the second scenario, the new money is created on behalf of JP Morgan by Citibank. Citibank in effect acts as JP Morgan's settlement agent. 

Now, lots of you will no doubt tell me that Citibank must create Goodlife's deposit when it receives reserves from JP Morgan, because otherwise its books won't balance. Sadly it's not quite so simple. The accounting above leaves out perhaps the most important part of all payment transactions - the messages that banks send to each other. 

Messaging is the untold story of payments. "JP Morgan pays for its coffee machine with reserves" is technically true, but it's only half the story. What JP Morgan actually does is send Citibank a message instructing it to pay for the coffee machine with its own money, and transfers reserves to Citibank in compensation. It's rather as if I ask you to go and buy me a pastrami sandwich and a latte. You pay out of your own pocket and I reimburse you when you deliver my lunch. 

So, JP Morgan instructs Citibank to credit Goodlife's account with $10,000. On receiving this message, Citibank increases the ledger balance on Goodlife's account by $10,000. At this point the reserve transfer has not been completed, and it might not complete for some hours or even days. But Goodlife can see the money in its Citibank account, and on that basis, delivers the coffee machine. 

However, banks are far more canny than people. If you were a bank, you wouldn't let me eat the lunch you bought me until I had paid you. Or if you did allow me to eat before paying, you would charge me interest on every mouthful. So Goodlife can't access its money until the reserve transfer from JP Morgan has completed. Citibank might allow Goodlife to "draw" on its uncleared balance, but this is actually lending. It's a short-term overdraft secured on the collateral of the anticipated payment. And Citibank would charge interest on it. Banks never knowingly pass up a chance to make money.

Messaging can go wrong, with unfortunate effects. This is more likely in international payments, which are by definition  complex and risky. But it's not unheard-of in domestic payments. Suppose the message from JP Morgan fails to arrive, perhaps because of a systems outage. Citibank receives the reserves, but it doesn't know that it has to put money in Goodlife's account. So it creates an entry in an "unidentified items" account. JP Morgan thinks it has paid, but Goodlife has not received the momey. There ensues a flurry of emails between an angry bank that thinks it has been defrauded of $10,000 and an equally angry coffee machine supplier that thinks its customer has failed to pay for goods it has ordered. 

Similarly, if the message contains wrong instructions - "pay $10,000 to Chris' Coffee", for example - the money goes astray even if the reserve transfer completes correctly. Transferring reserves is not by itself sufficient to complete the coffee machine transaction. 

Whenever a bank buys something from a non-bank, new money is created. That is true whether or not the transaction involves a reserve transfer between banks. Focusing only on reserve transfers, and ignoring the vitally important role of payment messages and bank deposit creation, entirely misses the point. 

Related reading:

Money creation in the modern economy - Bank of England

Image: By Takeaway - Own work, CC BY-SA 4.0,

Many thanks to Goodlife Coffee Company and Chris' Coffee for allowing me to take their names in vain. I hope this piece drives some genuine business to their long-suffering sites. 


  1. Hi Francis,

    I am a little disappointed, a mere coffee machine?

    A couple of years ago I was wandering through the business district and looked at all the massive buildings owned by banks.

    Surely not? I thought.

    But yes indeed, they are all funded by the banks simply making balance sheet entries.

    Josiah Stamp knew banks....

  2. So, banks don't loan out deposits. Deposits are the result of lending. So all $15 trillion in bank-held savings are un-used and un-spent, lost to both consumption and investment, indeed to any type of payment or expenditure. That is the source of Secular Stagnation, not robotics, not globalization, not monopolization, not demographics.

  3. Contrary to Bankrupt-u-Bernanke’s claim that: “Money is fungible”…“One dollar is like any other”, pg. 357 in "The Courage to Act", the utilization of savings is a catalyst, it is not a matching of economic accounts, not a 1-2-1 economic transaction (correlation between two sets). This is aptly demonstrated by debits to particular deposit accounts.

    The regulatory release (yes it is up to Congress), of savings invokes a spontaneous chain reaction, an expanding sequence of reactions, a self-propelling and amplifying chain of events. In other words, savings’ products have a positive economic multiplier (and $15 trillion are frozen), whereas new money products (QE forever) have a negative economic multiplier.

    How could this be? It is because lending by the commercial banks is inflationary, whereas lending by the nonbank public is non-inflationary (if savings are not expeditiously activated, put back to work, then a dampening economic impact is generated).

  4. The U.S. Golden Era in Capitalism was where small savings were gov't insured in the nonbanks and invested in targeted real investment outlets (residential real estate). That resulted in a demand for both labor and materials. I.e., 2/3 of economic output was financed by velocity - not money.

    An increase in money products reduces the real rate of interest. An increase is savings' products raises the real rate of interest.

    The solution to subpar economic growth is to gradually drive the banks out of the savings business (which doesn't reduce the size of the payment's system).

    1. Or we just hope that crypto derails and replaces the entire system. I think it will.

  5. Great stuff. Keep going. What happens when Greenfield uses its $10,000 to buy a treasury security?

  6. When a bank (or clearing house) tallies all the incoming checks (for deposit) each day, it can easily decide to how much it's reserves have increased. All the checks deposited from other banks are in the form of reserves. All the checks deposited from within the bank itself are transfers within the confines of the bank without reserve altering effect. Not to be forgotten, each deposit is destined for an owner's account.

    We can believe this statement is correct for private banks but what about a central bank? Well, private banks deposit reserves into CB accounts, as does the government. In both cases, the reserves have account names attached. So who creates the reserves themselves?

    It seems to me that it could be only the CB or a government authorized agency. This would not include a private bank.

    So what happens when a private bank makes a loan? The private bank is giving the borrower access to the reserves held by the bank. So long as the borrower spends into other accounts wholly residing within the bank envelope, there is no impact on reserves. If the borrower spends outside the envelope, reserves are impacted placing the bank at financial risk.

    If we think macroeconomically, the entire banking system is but one bank within a national economy. Therefore, the national reserves are never impacted by private bank loans. Instead, these private bank loans are simple additional claims on existing reserves.

    We can think of reserves as being real fiat money and bank account deposits as being claims on existing reserves no matter what the actual moment-to-moment size of the reserve pool might be.

    1. I don't think it is remotely helpful to think of the banking system as one bank. How banks interact with each other and with their customers matters. By ignoring not only the movements of funds between banks, but also the fact that bank customers use bank-created money to make purchases, not reserves, you've managed to eliminate pretty much the entire private sector.

    2. The question is whether private banks create money or merely create claims on existing money.

      I notice that private banks can disrupt their financial health by making too many or unwise loans. If they truly created money, their financial health would never be a concern.

    3. Banks do create money. But they can't create the money that they use. They can only create money for their customers to use. So the fact that banks create money does not mean their financial health is never a concern.

    4. Alphonse: If Goodlife buys a Treasury, it's merely a Debit to Bank Deposit, Credit Treasury Security. Citibank Debits GL's Deposit and Credits its Reserve Deposit. Fed Debits Citibanks Reserve Deposit and Credits its Treasury Security Account.

    5. Frances: Unless there is something prohibiting it in their charter/license, the problem with creating money for say a nice office building or a coffeemaker is that the seller can deposit the check at another bank and the first bank better have the reserves to cover it. If they created a Loan Receivable to offset the Deposit, it would be alot easier to borrow reserves against it (as collateral) from another bank or as a last resort - The Central Bank.

      I do like to think of the banking sector as one consolidated entity - since this sectoral way of thinking reflects the realities of the interbank money market. ("Fed Funds" in the US.) It helps to clear the brain of fractional banking nonsense.

    6. I've already pointed out that the purchasing bank needs to have, or be credibly able to obtain, the reserves to pay the settlement bank. So I'm really not sure what point you are trying to make here. You seem to be repeating what I have already said.

      Re thinking of the banking sector as a single entity, - yes, I've noticed there is a fashion for eliminating all private sector transactions when thinking about the financial system. It is however fundamentally wrong and very misleading. Transactions between private sector actors are the lifeblood of the economy.

      The interbank money market doesn't matter any more. It died in 2008. Policy rates now are deposit rates (IOR in the US), not lending rates (Fed Funds).

    7. Also, DoDeals, you've only done half the accounting for Goodlife's purchase of a Treasury security. We need the accounting for the seller of the Treasury too, please. Goodlife is not a broker-dealer, so the purchase is in the secondary market and there is no change to the TGA or total reserves. But the accounting and the effect on broad money (M2) depend on whether the seller is a bank or a non-bank.

  7. Just to be nitpicky:

    RE: "... A full balance sheet analysis would show that JP Morgan's equity has reduced as a proportion of total assets. In effect, JP Morgan has paid for the coffee machine from its own equity. ... ..."
    • Actually not. JPM Debited Fixed Assets and Credited: Deposits. Equity not touched.

    RE: "... What JP Morgan actually does is send Citibank a message instructing it to pay for the coffee machine with its own money, and transfers reserves to Citibank in compensation. ... ..."
    • Actually it’s the other way around. What actually happens is JPM sends GL a check, credit card transaction or wire, and GL instructs it to deposit at Citibank. It’s Citibank that then deposits at its reserve account at the Fed, which then debits/credits the reserve accounts of each bank.

    1. Unfortunately your nitpicking is wrong.

      1. "A full balance sheet analysis would show that JP Morgan's equity has reduced AS A PROPORTION OF TOTAL ASSETS". Both fixed assets and deposits increase, and as a result equity reduces as a proportion of total assets. This is basic maths.

      2. You have the messaging wrong. JPM instructs the settlement bank, not the settlement bank's customer. Yes, if JPM sends GL a cheque, GL must physically present it at Citibank. But GL does not have to instruct its settlement bank to receive an electronic payment. That is done by JPM.

    2. You are............correct.

  8. Jerred Seisyll, I have deleted your comment because once again it consisted mainly of personal attacks on me. You are now banned from this blog.

  9. It seems to me that we need to follow the ownership of real fiat money at all times.

    When a person deposits his government sourced check at a local bank, the bank takes funds into local bank ownership. The depositor gets a positive entry in his account balance but relinquishes ownership of the actual real fiat money (assuming government has no account at local bank). Depositor now has remaining only a claim on the resources of local bank in the form of a positive account entry.

    The reader can see that I have assumed that government always pays using real-fiat-money.

    1. Since banks are licensed by government to create fiat money, bank-created money is real fiat money. So I profoundly disagree with your comment, Roger.

    2. In my view it is plain ridiculous to say that the reserves that banks use only for transfers between themselves are "real money", but the money the private sector uses for real transactions isn't "real money".

    3. The banksters should not have the sovereign right to create public money. Lending by the banks is inflationary. Lending by the nonbanks is noninflationary.

      Banking systems should be nationalized. Lending should be outsourced.

      Reserves were never a "tax". They are clearing balances. They are "Manna from Heaven", showered on the payment's system and costless to the banks.

      You can't properly control the money stock by any attempt to control the cost of credit.

  10. I think you have the concept. It is a radical way of thinking about money but has parallels with the storage/ownership of grain stockpiles.

    I am thinking that only a central bank has government authorization to create real fiat money.

    1. It's not a radical way of thinking about money. It is straight up monetarism.

      I'm afraid it is not true that only a central bank has government authorization to create fiat money. Licensed banks do too.

    2. I am thinking that private banks have a limited license to print money, limited by the amount of real fiat money they control. It follows that we might be more accurate to say that banks can only grant limited access to real fiat money. access that is available until it's not available (perhaps made unavailable by higher authority).

      The very top authority would come from government, probably acting through a central bank.

    3. Roger, private banks' ability to create money is not limited by the amount of central bank created money they control. This is the "money multiplier" fallacy that has been resoundingly debunked by the Bank of England - see the link in Related Reading at the end of the post. Banks' ability to create money is limited by the amount of lending they can do, which in turn is limited by capital requirements and availability of creditworthy borrowers.

    4. The "money multiplier" is no fallacy. The FED lost control of legal reserves in 1995, after Greenspan reduced reserve requirements by 40%. That caused the GFC.

    5. The effect of tying open market policy to a repurchase agreement (or some interest rate policy peg) was to supply additional (and excessive) reserves to the banking system whenever loan/investment demand increased.

      Today the money multiplier is determined by the trading desk's purchases between the banks and the nonbanks (which is undisclosed by the authorities). I.e., the trading desk used to target "RPDs" in the 60's and 70's, or "reserves for private deposits".

  11. This comment has been removed by the author.

  12. This comment has been removed by a blog administrator.

  13. Hi Frances!

    Great article (again), so thanks for sharing!

    I've a question regarding:

    > A full balance sheet analysis would show that JP Morgan'e equity
    > has reduced as a proportion of total assets. In effect, JP Morgan has
    > paid for the coffee machine from its own equity.

    Since equity didn't change in absolute terms it's implied that, for a bank, solvency is (also) defined as their equity ratio. Does this ratio stem from the capital requirements of Basel III? What would be the legal/regulatory limit for a banks "stuff" purchases?

    Thank you!

  14. If a central bank buys a coffee machine is it paid for with newly created Central Bank money?

  15. Hello Frances. Very interesting. I've long understood that banks create money when they make loans and also how messaging and payment works. What I had not thought about was the banks own spending. While I see how it may work for appreciating assets (land and to some extent buildings), the cost of expenses and the write down of depreciating assets (like coffee machines) must be charged against profit, so a bank cannot simply create money to pay all its expenses

  16. "As I shall show, this money is newly created in the course of the transaction."

    Wow, this article was illuminating, scary, and even a little depressing. I feel like if I tried to sum this up at a party, I'd be labeled a conspiracy-theorist.

    "JP Morgan pays for the coffee machine with newly-created bank money, not with pre-existing reserves."

    JP Morgan is playing by different rules in the same game. If corporations are people like you and me, why are you and I not allowed to do the same? How unfair!


Post a Comment

Popular posts from this blog

WASPI Campaign's legal action is morally wrong

The foolish Samaritan