This post describes the nitty gritty of what happens when the Fed purchases Treasuries. I will go into detail on the balance sheet implications for each participant, which will vary depending on whether the market participant is a bank or a non-bank. The bank/non-bank distinction matters because non-banks do not have Fed accounts and thus cannot hold reserves.
The Fed only does QE trades through Primary Dealers, who generally are not banks (they are broker-dealers) and do not have Fed accounts. (The exception is few U.S. branches of foreign banks who house their broker-dealer business in the bank entity, which do have reserve accounts). In practice, Primary Dealers tend to bank with custodian banks like Bank of New York Mellon, who specialize in collateral management services.
But the focus should not be the primary dealers as they are merely conduits. The newly created reserves ultimately end up in the account of whoever sold the Primary Dealer the Treasuries. If the seller is a Bank, then it will end up in the Bank’s Fed account. If the seller is a non-bank, it will end up in the Fed account of the bank that the Non-Bank banks with. The bank’s new reserve asset will be balanced against new bank deposit liabilities owed to the Non-Bank.
Below I walk through four scenarios of QE sales: Non-Bank Investor to Non-Bank Primary Dealer, Bank Investors to Bank Primary Dealer, Bank Investor to Non-Bank Primary Dealer, and Non-Bank Investor to Bank Primary Dealer. This should offer insight into the plumbing of QE.
Non-Bank Sells Treasuries to Non-Bank Primary Dealers
In this common scenario the non-bank Primary Dealer buys Treasuries from a Non-Bank (could be a corporation, hedge fund, mutual fund, another dealer etc.). The Primary Dealer borrows money from a money market fund via the repo market to make the purchase. The Treasury purchased is immediately pledged as collateral for the money market fund’s repo loan. The Primary Dealer will repay the repo loan using the proceeds it receives from selling the Treasury to the Fed.
Note that none of these entities have reserve accounts. The QE reserve impact happens behind the scenes when the commercial banks that the participants have accounts with settle payments with each other. Recall, reserves are basically a special type of money used for transactions between commercial banks. Reserves will move from the money market fund’s bank to the Non-Bank’s Bank, via the Primary Dealer’s Bank.
The Primary Dealer then sells the Treasury to the Fed, and repays the repo loan to the money market fund.
At the end of the day, there are $100 more reserves in the banking system. They end up in the Fed account of the non-Bank’s bank. The Non-Bank itself essentially exchanges Treasuries for bank deposits.
Bank Sells Treasuries to Bank Primary Dealers
In this scenario a Bank sells Treasuries to a Primary Dealer who is also a Bank, who then sells it to the Fed. Because all entities have Fed accounts, all transactions are settled in reserves. (The Primary Dealer can also fund the transaction using a repo loan, but for simplicity I construct the example assuming it has enough reserves to not need funding.)
Bank Sells to Non-Bank Primary Dealer
In this scenario things are bit more complicated because one participant has a reserve account, while others do not. At the end of the day the Fed increases its balance sheet, and the Bank Seller swaps Treasuries for reserves.
The Primary Dealer intermediates this by buying the Treasuries with funding from a repo loan and then sells the Treasuries to the Fed. Behind the scenes, the Primary Dealer’s Bank receives reserves from the Money Market Fund’s Bank, then sends them to the Bank Seller. Then the Primary Dealer’s bank receives reserves from the Fed when the Primary Dealer sells the Treasuries to the Fed. And finally, the Primary Dealer’s bank sends reserves to the Money Market Fund’s bank when the Primary Dealer repays the repo loan. Note all these transactions take place within a day.
Non-Bank Sells to Bank Primary Dealer
In this example a Non-Bank sells Treasuries to a Primary Dealer who is also a bank. The Non-Bank cannot hold reserves, so the Treasuries are sold for bank deposits, which are created by the Non-Bank’s bank. The Non-Bank’s bank holds the reserves, while increasing its bank deposit liabilities to the Non-Bank. The Bank Primary Dealer funds the purchase with a repo loan from a money market fund.
Thank you for the information. I now understand why people just call QE ‘money printing’ because its easier.
When Congress passes a $1.9T “stimulus bill” and the Treasury sells bonds to Primary Dealers, from where do the primary dealers get the $1.9T before selling the bonds to The Fed? Is there enough money in the MMF/Repo market to finance this type of transaction?
They use repo transaction with the Fed to buy bonds in the first place.They buy bond with the loaned money from the Fed,then pledge a bond as collateral and buy back the next day or two the same bond back.
money market fund can also lend them reserves to buy bonds with them and buy them back the next day.
read 101 central banking from fedguy
If it’s money printing, why we haven’t seen hyper-inflation yet, after almost doubling FED’s balance sheet? Why Japan, after decades of QE, is struggling with deflation? (Honest questions – I’m trying to understand this).
QE is more about changing the composition of money – fewer Treasuries, more bank deposits – than the quantity. The impact of that is lower interest rates, which pushes investors into riskier assets (the ‘portfolio rebalancing’ effect). Thus QE pushes asset prices higher, rather than the prices of goods and services, which is what indexes like CPI measure.
Many thanks for clarifying this.
Excellent article. Exactly its a duration swap. It hasnt provided any overall lending growth. If that cash isnt being spent by the recipient into the real economy why expect inflation? The fed isnt forcing the secondary market to sell treasuries or preventing the secondary from purchasing them. Its just supporting the market for securities that are likely unproductive assets… ie gov debt.
Hello Fed Guy,
Hope you are well. Very much enjoying your content.
I hope you pick this question up. I’m trying to learn about this, so bear with me on what I hope is a sensible question.
When the Fed creates new reserves to pay for the bond already held by an asset manager, would I be right in saying that that bond is being bought and paid for again a second time? New money in the form of reserves is being created to pay for the same asset again that the original bond investor has already lent the bond issuer the money for.
Forgive me if my explanation isn’t completely clear.
When an asset manager buys a bond, he is using money already existing in the system. But when the Fed buys a bond, the Fed is purchasing with newly created money. I’ll walk through this to be more concrete.
1)I buy a $100 bond issued by a corporation (lending them money). At the end of the day the corporation has $100 more money, and I have $100 less (I have a bond instead). So the total money in the system is not changed.
2)I sell $100 bond to Fed. To purchase the bond, the Fed creates money out of thin air. At the end of the day then there is $100 more money in the system (technically, it is $100 more reserves balance by $100 more in bank deposits. non-banks don’t have reserve accounts and must transact through the banking system with trading with the Fed see – two tiered monetary system).
Thank you for the clear explanation, I understand.
So, the bond is actually being purchased twice, with newly created money in the form of reserves to pay for the second purchase by the Fed when they purchase the bond from the asset manager. It’s feels like the asset manager is getting free money on the second purchase of the bond by the Fed when the new reserves are created. As you say, there’s $100 more money in the system. Who does that actually belong to ?
I’ve read all of your blog posts which are excellent to say the least. Sincere thanks for taking the time to explain such an arcane subject to the non-practitioners out there.
Thanks for your support. The asset manager doesn’t get free money because he must give up the bond – it’s an exchange. The Fed gains the bond, but now owes someone $100. Another way to think about this is that the composition of assets held by the private sector changes. Now instead of holding a $100 bond, it holds $100. In practice the Fed is buying Treasuries, which are credit risk free assets. So the composition of safe assets changes, rather than the quantity.
Thank you again for taking the time to reply.
Okay, I understand that’s perfectly clear.
Just one more question though if I may please. The original proceeds that were invested in the bond the first time, would I be correct in saying that they are returned to the bond holder, which in this case, is the asset manager such as a pension fund or sovereign wealth fund etc? That would then straighten out the composition of the assets on the balance sheet.
My apologies for the continual questions. I’m going to read your posts thoroughly again.
Thanks again for your time. Have a great weekend.
In a simplified example you can think of it that way. In practice an institutional investor will be talking with securities dealers to sell and buy throughout the day. The central bank’s activity is just another market participant among an ocean of others.
For example – Investor A thought the bond was a good deal, so he purchased the bond. But then he changed his mind so he called up his dealer, who bought it from Investor A and re-sold it to Investor B. Investor B held on to it for a month, but wanted some cash to buy a new house. So Investor B called up his dealer, who bought it from Investor B and then sold it to the central bank.
Thank you for the explanation and for taking the time.
Very much appreciate all of your help and insight.
You’re saying Primary Dealers are just conduits? They interact with the Fed via OMO. The Fed is not purchasing directly from investors. PD’s purchase T’s at auction, warehouse the T’s, and sell them off to the public over time. No Fed needed here. This idea that the Fed creates new bank deposits with commercial by purchasing T’s from investors seems to be inaccurate. This comes from the paper on money creation by the BoE, which I read. I found that paper to have a lot of holes, so did Jeff Snider.
Fed pays for its QE purchases using reserves it creates. This increases the level of reserve assets held by banks, which must be balanced by a liability. That is usually a bank deposit. To be concrete – an investor sells a Treasury to the Fed and receives bank deposits in its account. The investor’s bank receives reserves. So at the end of the day the banking sector has more reserve assets and deposit liabilities. For more details see this post.
Not sure where to post this question (I’m new to your website)… Please excuse me if this is not the correct place to post it.
I am an accountant, and I work for a small (<$2 billion) community bank on the East Coast in their accounting department. When we generate a loan, we actually wire out cash to another bank. But, from what I have read, Money Center Banks don't wire out the cash, they just create an offsetting deposit when they originate a loan. Is my understanding correct regarding Money Center Banks creating deposits and not wiring out cash when generating a loan?
I am really enjoying your website. This is great information.
Loans create off-setting deposits, but if those deposits are then sent to someone who banks at another bank then cash will be wired out. One of a bank’s primary job is to manage that liquidity risk. Many people have accounts at money center banks, so when a money center bank create an off-setting deposit it is often used to pay someone who also banks at the money center bank. So it would just be a shuffling of liabilities on the bank’s balance sheet from account A to account B (no need to wire out money). I recommend this brief pamphlet from the Bank of England for more details.
Do banks, who have reserves with the Fed buy treasuries at auction using excess reserves, meaning that the treasury receives money that was never in the system, which in turn leads to an increase in the money supply?
Reserves held by banks are already in the money supply. The money supply increases when the Fed buys Treasuries. The Fed creates reserves to buy them, which increases the reserve assets and bank deposit liabilities in the banking system.
But money supply doesn’t increase all the time when Fed does QE. Japan has been doing QE yet money supply hasn’t increased dramatically. And not all deposits(liabilities of com. bank) are backed by bank reserves. If all reserves were in the money supply, perhaps the Fed’s balance sheet would be equal to M2?
I am unsure if I understand some details of the “Non-Bank Sells Treasuries to Non-Bank Primary Dealers” example.
1. Do the 100 $ MMF shares exist before any transactions occur, and are they constant throughout the example?
2. Shouldn’t all transactions of the MMF’s Bank cancel out at the end? The 4th figure says it still has 100 Reserves and 100 Deposits, but the Non-Banks’ Bank also has them.
1. Yes MMF shares exist before the transactions occur – they represent money already in the system. These shares were bought by a MMF investor who purchased the shares of the MMF in exchange for cash (investor’s deposits in the investor’s bank). MMFs are considered to be extremely low risk investments. Emphasis on “considered to be”.
2. The MMF bank’s reserves and deposits merely reverted to the same amount before the repo transaction – once the repo loan was paid off by the Primary Dealer. The new reserves and deposits were created by the stroke of a keyboard at the Fed and ended up with the non-bank’s bank and bank respectively.
In Bank Sells Treasuries to Bank Primary Dealer, it looks like the Bank Primary Dealer does NOT use a repo loan from a Money Market Fund. Is that the actual case? Or is it a matter of swapping reserves entirely, without Money Market Funded loans?
“The Primary Dealer borrows money from a money market fund via the repo market to make the purchase. The Treasury purchased is immediately pledged as collateral for the money market fund’s repo loan. The Primary Dealer will repay the repo loan using the proceeds it receives from selling the Treasury to the Fed.”
I think there is generally a lot of confusion surrounding Primary Dealers and what else they do. For example, how can a PD pledge a security that it hasn’t bought yet with the MMF’s loan? Doesn’t it need to pledge an existing security immediately upon entering the repo agreement? Or is this a result of the delay in settlement period of repo so that the collateral can be delivered later on?
Would you also be able to explain if the above description is the same thing as a “matched book repo”? Or does matched book refer to re-hypothecating within the same day? It would be helpful to understand how primary dealers run these matched books. Is this the market-making function of PDs, or the provide clients funding part?
Hi Fed Guy,
This is the first article (out of DOZENS over the years) that actually and finally helped me understand how QE works. Thank you. My new fear is that you will one day stop producing content, so i really hope you keep up the good work!!
Thanks! Your comment made my day
I actually have a follow-up question.
I read the BoE .pdf pamplet that you’ve referenced a bunch in the comments, and it talks extensively about the actual way money is created which runs counter to how most textbooks teach it.
In this article you say: “At the end of the day, there are $100 more reserves in the banking system. They end up in the Fed account of the non-Bank’s bank. The Non-Bank itself essentially exchanges Treasuries for bank deposits.”
Based on what I’ve learned in the BoE pamplet, don’t you need a lending event to create “fountain pen money”? When the non-bank receives reserves, how do they become bank deposits? Under whose account (or accounting) do they go under? Are they just assets on the balance sheet, waiting to be called into action in a future loan and bank deposit?
Government can also indirectly create bank deposits. So in the example you note the non-bank sells Treasuries to the Fed but cannot hold reserves (it doesn’t have a Fed account), yet the Fed only pays in reserves. So the Fed sends a payment to the non-bank’s bank. That payment increases the level of reserves the bank holds. The bank then balances that asset with a deposit liability created out of thin air that is due to the non-bank Treasury seller. See this link for more details.
That makes sense; it all links back to the non-bank’s account, who now has +$100 in assets as seen in the first picture. I was missing that piece for some reason which is now obvious in hindsight.
Thanks again. BTW reading this webiste prompted me to buy your book and I must say that the chapter on interest rates is (similar to my experience reading this article on QE) the first resource out of sooooo many others read over the years that finally helped me understand the relationship between the federal funds rate and short-term interest rates.
Thanks a lot for the content. Appreciate it.
Just to follow up on the above conversation:
So when a bank sells treasuries to the Fed via PDs (bank or non-bank), the bank seller’s balance sheet remains unchanged in size, only the composition of assets shifts from treasuries to reserves. But in case of all non-bank sellers of treasuries, deposits are created out of thin air, thus the non-bank sellers’s bank’s balance sheet expands. Thus in the former case monetary base expands but in the latter case the M2 does as well right? So if the composition of UST sellers during QE is skewed towards non-banks then the impact on M2 growth can be larger than otherwise?
My disclaimer… I’m just an engineer, I don’t know anything… but I’m curious to understand this thing from a systems engineering perspective.
I’ve been trying to get my head round this debate about “moneyprinting” v “it’s just an asset swap”.
I think I’m getting closer to the core of the confusion.
One the main problems is that words like “money” and “cash” are thrown around a bit too liberally, and this creates confusion or the illusion of understanding.
On one side you’ve got those who see the graphs at american feds showing a sudden increase in something that some people call “money”, so the conclusion is jumped to, that this is an increase in “money”, and we can be a bit more precise and use terms like “increase in money supply” or “credit expansion” perhaps.
On the other side, you’ve got people saying, the feds and the commercial banks are just swapping some junk for some reserves. When I say junk, I mean I’m skipping through the details of what kind of debt it is and where it originates to indicate that it’s something that banks who hold it want to get rid of it (perhaps because it includes non-performing loans etc…). I say banks, for simplicity, perhaps some are, some aren’t, some are primary dealers, or whatever, but that part can be clarified later.
Let’s get to “This is where the trouble starts…”, because people seem to leap to calling the reserves that get put on the banks’ balance sheets “cash” or “money”, but as your examples illustrate, all of these reserves stay as reserves.
the bit you’ve snagged me on is this notion that somehow deposits are increased by this asset swap… but how can they be? deposits are apples, and reserves are oranges, so to speak.
as I understand it, deposits are only created by banks creating credit in the form of loans, (or a tiny amount of physical cash minted by mints), and that being “deposited” into a bank account.
as I understand it, all that reserves do is act like a shadow or mirror to these deposits… that sounds like they’re interchangeable, but from what I read they absolutely are not.
perhaps some institution could take out a loan, and buy reserves with it – I don’t know why they would, and I don’t know why a bank would create credit for an organisation to do that.
so I’m puzzled what all these excess reserves do… whilst they might be created to nudge banks to create credit by issuing loans… that’s not what seems to happen. loans seem to have decline precipitously over the past couple of years that these reserves have been created- so they don’t seem to have that effect.
the purchasing of debt by the central bank seems to have coincided with the continuing increase in bond prices/fall in bond yields, and the increase in asset prices in what to many looks like a bubble.
it’s not entirely clear whether and what the causal relationship is.
from graphs I’ve seen none of the rates the fed sets seems to have any influence on bond yields… not since about 2000, and in fact the fed seems to lag behind where bond yields go… so the fed in some respects seems a bit irrelevant.
so what is my question? I suppose it’s what do all these excess reserves actually do?
they seem to be just being passed around in parallel with the real economy and the real monetary system. as if the fed swapped a bag of rotting garbage for an inert bag of nothing that doesn’t do anything. the garbage still needs to be dealt with somehow somewhere surely… it surely can’t be just magicked away as when debt is paid off. there has to be cost somewhere for it being taken onto the fed’s account.
if the reserves don’t do anthing at the banks… if they aren’t counterbalanced by new loans – new money… are the banks somehow “lopsided”? is everything seizing up because the reserves are an irrelevant solution to a perceived problem that needs a different kind of solution?
my notion rests on the idea that reserves are not cash, are not really money, are not “usable” in terms of entering the economy and prompting inflation, they operate between banks like their own parallel currency and parallel economy, or maybe they have some role with the infamous offshore eurodollar system? whilst it seems like the current “inflation” is not real inflation, but supply chain disruption, it remains a mystery as to what impact to excess reserves actually have on anything at all.
sorry for the rambling question.
In short: do banks use the virtual currency reserves they recieve from QE to buy stocks?
Apparently not… FedGuy says “Yup!” but then goes on to say:
“Note: Although banks can spend reserves on anything, bank are heavily regulated in what they can buy. They cannot go out and load up on equities and high yield or other risky assets without violating risk and regulatory limits. Regulation and profitability, not reserves, is what constrains a bank’s balance sheet.”
…which sounds more like “not really” than “yup”.
Right now, the reserves rules has been set to zero… and that seems like something that could – in theory – pump stocks… but surely you would see those purchases by the top 25 Primary Dealer Banks, of major stocks like Apple or commodities like gold or financials like bonds or esoterics like bitcoin or anything else…
It’s amazing that something so important is so lacking in front and centre clear explanations online
” A prospective dealer must be able to demonstrate an ability to provide sizable, sustained performance in operations in, at a minimum, the U.S. Treasury repo and cash markets, and preferably also in one or more other markets in which the New York Fed is eligible to transact. The New York Fed also prefers that a prospective primary dealer show underwriting capability in all such markets. Because Treasury auction participation is not limited to primary dealers, the New York Fed expects a prospective primary dealer to participate in Treasury auctions at levels expected of primary dealers (described below) prior to becoming a primary dealer. ”
” The primary dealer must meet certain minimum capital requirements:
A registered broker-dealer must have at least $150 million in regulatory net capital as computed in accordance with the SEC’s net capital rule.5 The broker-dealer must also be in compliance with all capital and other regulatory requirements imposed by the SEC or its self-regulatory organization (SRO).
A bank must meet the minimum capital standards to be considered ‘Well Capitalized’ or its equivalent by the applicable U.S. or foreign regulator, and must have at least $150 million of Tier I capital as defined in the applicable Basel Accord.”
it doesn’t sound like they can just take Santa’s sack of reserves and blow it all on memestocks, bubbletech, or chinacoin.
March 14, 2021 at 5:11 pm
Can the Commercial Banks use their Fed created Bank Reserves to buy non treasury things like bonds and stocks?
Fed Guy (Post author)
March 15, 2021 at 6:32 pm
Yup! see https://fedguy.com/can-banks-spend-their-reserves/
Hi Joseph ,
Thank you for these really insightful posts !! I had a question on one of your scenarios above :
“Non-Bank Sells Treasuries to Non-Bank Primary Dealers:…..The Primary Dealer borrows money from a money market fund via the repo market to make the purchase. The Treasury purchased is immediately pledged as collateral for the money market fund’s repo loan.
How can the treasury be pledged to the MMF when the Primary dealer has not yet purchased the treasury in the first place ? Don’t all purchases happen on a DvP basis ?
This is such a great site – clears the fog. When the statement is often made that the Fed is “printing money”, some infer this to mean that the Fed is funding the Treasury. It’s my understanding that this is not accurate. The Fed may be injecting money into the economy but it is not in support of government spending. Would you expound on this?
My question is about Treasury purchases that happen with foreign banks. In your book, you consider two scenarios: (1) the foreign bank is large enough and has a US presence and has a Fed account; (2) the foreign bank does not have a Fed account but holds USD assets with a bank that does have a Fed account.
What about the following scenario? Suppose a foreign bank created a USD loan (asset) to a non-bank foreign entity and corresponding deposit (liability). Suppose the foreign bank has no Fed account and no other USD holdings. In other words, the USD loan and deposit are the only USD entries on the balance sheet and there are no USD assets held in any accounts of this foreign bank with another bank. Could the non-bank foreign entity purchase a US Treasury with its deposit at the foreign bank? If so, how would this transaction be settled? The deposit would need to be transferred to another bank…
I wish the Fed Guy had picked up on my comment but I will restate. While the Fed can create reserves to purchase treasuries, the purpose is to adjust liquidity and interest rates. The Fed does not purchase bonds to fund government spending. It’s not printing money you and I use. Easy to see – total bonds around $26T Fed holdings around $6T. As Joseph Huber points out, reserves are an artifact of our current banking system.
Question for all: Does QT go through primary dealers first too?
Hi Fed Guy,
Data Guy here. I think there might be a mistake in the section Non-Bank Sells Treasuries to Non-Bank Primary Dealers.
The final state of the balance sheet of the Money Market Fund’s Bank should be empty if I’m not mistaken.
When the primary dealer “repays the repo loan to the money market fund,” the balance sheet of the Money Market Fund’s Bank should cancel out the previous state.
I might be wrong here. Looking forward to hearing back from you.
Thanks for your readership and your careful analysis. I don’t believe there is an error. The MMF’s bank started with $100 in reserves. The extra $100 created by QE sits in the Non-Bank’s Bank.
Hi Fedguy, sorry just a doctor with a random question
I thought primary dealers purchased treasuries directly from the US treasury department. Is this not the case?