There is still $1 trillion in Fed liquidity that will gradually flow into the private sector after QE stops. A large chunk of liquidity created by QE over the past two years never entered the banking system, but instead sat first in the Treasury’s Fed account and later in the RRP Facility. In the coming months Treasury will restart bill issuance and draw those funds out of the RRP into the TGA, and then spend those funds into the banking sector. Over time that will leave the banking sector with about $1t more in reserves, and the non-banks with a $1t more in deposits. If the past is any guide, that suggests more portfolio rebalancing where banks will purchase more Treasuries and non-banks more risk assets. In this post we trace the recent flows of Fed liquidity, show why more liquidity will soon flow into the banking system, and suggest that it will be a tailwind for all asset classes.
Shuffling Money
Fed pays for QE purchases by printing reserves, which are essentially digital cash issued by the Fed that only entities with an account at the Fed can hold. Those reserves usually end up in one of three account types at the Fed: banks, Treasury’s Fed account (“TGA”) or the RRP. The tremendous amount of reserves created by the Fed last year largely ended up in the Fed accounts of commercial banks, but a sizable chunk also found its way into the two other account types. Note that increases in reserve assets held by banks are also mirrored by an increase in deposit liabilities due to our two tiered monetary system. QE increases reserves (money for banks) and deposits (money for non-banks).

Around $1.5t of the QE liquidity created last year was held by Treasury and never made it into the banking system. At the time, the Treasury was anticipating potential stimulus legislation so it kept a large precautionary TGA balance. As that became unnecessary and with a looming debt ceiling, Treasury began to spend down its TGA balance by repaying Treasury bills (short-term debt). Those repayment flows largely went to money market funds (“MMFs”), who are the largest investors in Treasury bills. The MMFs had no where to reinvest those funds so they just deposited them into the RRP facility, which currently has a balance of about $1.5t.

The $1.5t of QE liquidity that has been sitting in either the TGA or the RRP for over almost two years will eventually enter the banking system. The mechanism for that entry is through increased bill issuance and a lower TGA target.
Bill Issuance Drains the RRP
Bill issuance trends higher because the Federal deficit only goes higher, but there are also two developments next year that will increase bill issuance. First, the debt ceiling will (eventually) be resolved and free up space for more issuance. Second, Treasury will likely shift its funding mix next year towards a higher percentage of bills even as it cuts coupon issuance. An anticipated rapid taper of the Fed’s sizable $80b monthly purchases amidst perceived Treasury market fragility may compel the shift. Although Treasury was advised to issue 15 to 20% of debt in bills, Treasury also recently commissioned a study that suggested flexibility in that target given the high level of cash in the RRP. In this context the prudent and lower cost path is likely to issue more bills at the current rock bottom rates.

The bulk of the issued bills will likely be purchased by MMFs, who are the dominant investors in bills. MMFs have around $1.5t deposited in the RRP just waiting to be invested in any safe asset yielding more than the RRP offering rate of 5bps. Treasury will use some of the issuance proceeds to build up the TGA account from its debt ceiling lows. But with no emergency stimulus plans in sight the TGA will stay far below its 2020 level of ~$1.5t and instead hover around a few hundred billion. Treasury appears to be modeling a $650b target next year, leaving around $1t of reserves to flow into the banking system.
Into the Banking System
A short series of transactions moves money out of the RRP and ultimately into the banking system. From the Fed’s perspective, money moves out of the RRP and into the TGA when a MMF buys a Treasury bill (lends Treasury money), and then the moves from the TGA into a bank’s Fed account when the Treasury spends it. All these flows are just changes in the Fed’s accounting software, where numbers are moved from one account to another.

From the private sector’s perspective, the Treasury’s spending increases the level of reserves held by banks and deposits held by non-banks. In essence, QE liquidity that was already created is making it way into the private sector after a delay. The timing of the flows depends largely on the path of bill issuance, but persistent deficits mean eventually the entire $1.5t in the RRP will flow into the banking sector (minus a few hundred billion to top up the TGA).
QE Eternity
The financial and economic impact of QE is intensely discussed, but few would dispute that it boosts financial assets. One channel through which this occurs is via the mechanical rebalancing that happens when a chunk of low yielding assets is pushed onto the private sector. Banks rebalance out of low yielding reserves into Treasuries, and non-banks rebalance out of 0% bank deposits into riskier or higher duration assets. The end result is a rally in all assets that market observers have come to appreciate.
To the extent this mechanism describes the impact of QE, then there is effectively $1t more of QE coming down the pipeline.
42 comments On The QE Afterparty
What I found out:
The Federal Reserve did the tipping, which is about $ 1.5 trillion in treasury securities.
The Treasury transferred the $ 1.5 trillion to RRP accounts by repurchasing its securities from the money market.
And now, after a year, the treasury wants to sell the bonds back to the money market.
But I do not understand why you said at the end of the article that there would be an extra trillion dollars in the money market?
The lack of sound money has absolutely destroyed this once great country and, clearly, is at the root of massive inequality. The result (and purpose of the Federal Reserve” is that the inefficient, wasteful and coercive monopolist known as government grows with no checks on it whatsoever, at the expense of liberty…How else are you going to fund this horrendous foreign policy, a welfare state, 17 spy agencies and all the vacation homes of politicians and lobbyists…. Instead of being a FED apologist, it would be nice if Mr. Wang discussed a strategy for doing away with the Federal Reserve Notes as money system and bringing back sound money and capitalism (which does not exist, since capitalism requires the rule of law, free markets, and sound money (none of which we have, hence the current mess…) The natural state of actual Capitalism with sound money is increasing purchasing power, deflation (the good kind), and increasing living standards…. So what is the oath forward? Or do we just continue submitting to this fiat standard until…what? Where does Mr. Wang see the trends heading 5 to 10 years from now? How do we throw the Vampire Squid off the back of the USA?
How can you have a fie money supply with a growing population size though? Sound money doesn’t work. Also, if there ever was another major war countries will just revert to fiat to fund it just like previous WW’s. Fiat is an evolution of a failed gold standard.
Fixed money supply*
The problem is not changes in population. The problem is paying back the massive debt upon which the system depends.
Once the system crashes, as it will inevitably will, replace the fed with a currency board. Replacing it now would cause a crash.
“Instead of being a FED apologist, it would be nice if Mr. Wang discussed a strategy for doing away with….”
I for one appreciate that this blog focuses on how things work based on his expertise. We don’t get enough of this type of information in our daily dose of opinion/news feeds. Plenty of other sources for political debates.
Aren’t reserves all tapped out though? Banking system can’t really support another $1T inflow due to leverage ratios (that’s why TGA paydown of bills didn’t flow into banking system first time, but went to RPP), so bank treasurers will continue to push depositors out to MMF, who will continue to rely on RRP once bill issuance is exhausted. Basically, RRP is a plug for TGA as reserves are full.
What am I missing?
There is definitely a bank capacity issue, but also a couple points worth noting. 1) As the money is entering via fiscal spending then the recipients are going to be retail/corporations, which banks are more willing to accommodate. 2) Over time balance sheet capacity grows with bank earnings/equity
Can you explain mechanistically how this ~1T of QE « money » ends up pushing up all assets? I understand the treasury / fixed income side as banks will bid these up. But does this mechanistically translate into stocks going up (other than argument that lower discount rate = higher P)? In other words, how exactly will this increase in bank reserves actually translate into higher stock / commodity prices? Thanks! appreciate your work btw
I struggle with this as well. Higher deposits in the non-bank sector used to purchase stocks simply changes ownership of each asset. Is it simply the ‘hot potato’ James Tobin argument often cited by John Hussman that creates some sort of higher ‘velocity’ of those deposits? Jeff Snider often argues that anyone who wants to buy stocks could simply borrow as banks have plenty of balance sheet capacity and the main effect QE has on stocks is the psychological impact it has on money managers – money is being printed therefore must buy stocks!
Yes, the only plausible reason I gather is the psychology argument. Thanks for responding. But that correlation between CB balance sheets and equity markets is just so strong it makes you wonder if mechanically there is something going on that is not just chalked up to animal spirits.
Thought you might find this Twitter Thread interesting.
https://twitter.com/MacroAlf/status/1467938612641730570?s=20
“Banks rebalance out of low yielding reserves into Treasuries, and non-banks rebalance out of 0% bank deposits into riskier or higher duration assets. The end result is a rally in all assets that market observers have come to appreciate.”
Is this the bottom-line? The rebalancing tilts to more risk thus pushing up asset prices. So it’s not the volume of money so much as the allocation of this money to riskier assets.
Yes but isn’t the owner of the riskier or higher duration asset who sells it themselves rebalancing (or reallocating) back into 0% deposits. Why are they doing that? Taking profits? As I understand it all financial assets must be held by someone until there is reflux (contraction/destruction) in which say the deposits created no longer exist, e.g., by repayment of debt or QT for example. Maybe Joseph can write about this!
When you are forced by the Fed to hold a higher quantity of low yielding assets, then you are also forced to do more rebalancing.
Thanks for that Joseph. I think what I continue to get confused by though is that the forced rebalance just moves the fed created deposits from one entity to another. The receiver of the deposit from the ‘rebalancing entity’ then has a choice, hold cash or buy assets which just moves the deposit again. Holding cash seems an issue due to FDIC limits but some entity still ends up with the deposits until they are destroyed do they not? Is there a merry-go-round (hot potato dynamic) happening here which
causes constant rebalancing again forcing up prices?
Thanks for this! I’ve grown to appreciate the significance of how the Fed, Treasury & commercial banks operate which seems to truly dictate markets. I’m relatively new in this arena but it seems like it’s easy to just say “Fed prints more money –> immediate inflation” but where those reserves go & whether (or in this case, when) they become M2 is crucial.
I’m not sure if a reply can be written here but I did have a question somewhat unrelated. It seems to be that there are many reserves w/ the primary dealer banks & a dog fight over collateral in the Repo market. I may be wrong here, but if, 1) QE continues, this means that more assets (treasuries) are on the Fed’s balance sheet and therefore dries up collateral even more 2) I’m not sure what counts as collateral but presumably to get more of it, banks would have to go up the risk curve (?)
So the fight for collateral continues but is this only relative to the number of reserves they have? My question therefore is, would it be in the banks best interest to reduce the number of reserves (to acquire a better equilibrium & ultimately reduce risk)?
& a follow up question, if an individual (Joe) with 50K in assets and 30K in liability passes away. Assuming no estate planning was present, do the banks use his 50K to dissolve the 30K in liability (that is, remove dollars from the system)?
As I said, relatively new at this & surely i’m missing details or over simplifying but thanks again! I’ve shared your article
I don’t follow your explanation for why this $1.5T of RRP matters, why would it impact fiscal spending decisions/amounts?
Seems like the only expansion of liquidity in this model comes from fiscal spending, a political decision, and the rest is just swaps of money types.
Thanks for the informative work.
Appreciate the explanations. So many questions, but I’ll limit myself to two:
1) “A large chunk of liquidity created by QE over the past two years…sat first in the Treasury’s Fed account ”
Does “created by QE” in this context mean the Fed bought Treasury securities from some source and credited the TGA? Or does it mean they simply made a few computer keystrokes and increased the balance of the TGA without receiving anything in return?
2) When the doctor receives the +100 in deposits, the bank receives +100 in reserves in their Fed account, right? So, when the Doctor spends those deposits (on supplies or shares of SPX) does the bank lose 100 in reserves from their Fed account?
Vozmo,
On question 1, the Fed buys treasuries from primary dealers who then deposit proceeds in banks, the primary dealer owns a bank deposit, and the bank gets their reserve account credited at the Fed.
The TGA gets credited when the treasury sells securities. The funds usually come from the primary dealers, so bank reserves are reduced when the primary dealers purchase treasuries direct from the govt.
On question 2, yes, the bank loses the reserves. They are transferred to the bank receiving the funds. It has to be this way – the receiving bank now has a liability (the deposit), and an offsetting asset (reserves at Fed).
What I’m ultimately trying to understand here is whether money is being created or destroyed in this process, and if so, where exactly. How did that $1.5 trillion in the TGA get there and what was its genesis? Did it start with the Treasury issuing securities which were sold to primary dealers? The other day Joseph said primary dealers are not banks, so then are they buying Treasurys with deposits? Is the money going from deposits in the commercial banking system into the TGA at the Fed somehow? Surely new money is being created at some point in this process, but where exactly?
The new money is initially created when the Fed buys treasuries and MBS from the primary dealers. The asset side of the Fed balance sheet increases with treasuries and MBS, and the liability side also increases. This is the money creation event. The question is, which liability accounts increase on the Fed’s balance sheet? And how does the newly created money flow between the liability accounts?
Besides currency in circulation, there are 3 major liability accounts on the Fed Balance Sheet: Bank Reserves (commercial banks’ deposit accounts at the Fed), RRPs (mainly collateralized deposits of money market funds at the Fed), and the Treasury General Account (the “TGA” is the Treasury’s deposit account at the Fed).
There are lots of things happening all at once, so presented below is a reasonable scenario.
1. When the Fed buys the securities from the primary dealers, it credits the Fed accounts of the primary dealers’ commercial banks. Bank reserves increase. Net effect: Bank Reserves up. This is the new money – it’s now “in the system”, but it can end up in different accounts per 2 and 3 below.
2. The Treasury may then sell some treasury debt – this will be purchased by investors with bank accounts – so their bank balances will decrease and the reserves will move to the TGA. Net effect: Bank reserves down; TGA account up.
3. Some money may leave the banking system and go to money market funds. Those Money market funds then invest in RRP. So, Bank reserves down; RRP up.
The point is that after step 1, the money can then move between the banks, RRP, and TGA – but now you are just reshuffling the newly created money between entities that all have deposit accounts with the Fed. Depending on what is happening, the flows can change. If the treasury starts spending money like crazy that flows to the bank accounts of the recipients of the treasury’s money, then TGA goes down, and Bank Reserves go up. These flows and changes are constant.
BTW, Joseph’s book is excellent and goes through this in more detail.
Hi. Thanks for another great and insightful piece. Just wanted to check if I understand correctly. That is, is the following correct in terms of the mechanism via which extra $1 trillion that is set to flow into the banking system:
1) MMF (bank) buys Treasury securities, which to need be issued in addition to the planned $650bn in bills to fund the budget deficit. This leads to a fall in the RRP and a rise in the TGA
2) When the Treasury spends proceeds from the MMF purchase, the amount goes from the TGA to the bank’s Fed account
3) This raises the level of reserves held by the banks
4) This is mirrored in a rise in non-bank deposit
5) Non-banks rebalance out of 0% deposits into riskier or higher duration assets.
6) do banks deposit extra reserves in reverse repo, depositing interest proceeds in equity?
rsm,
Banks deposit extra reserves in their own reserve accounts at the Fed. They earn a current interest rate of 0.15% from the Fed. Although commercial banks can do RRPs with the Fed, the RRP rate is currently only 0.05%. So, the banks earn more on their reserve accounts than RRP – hence, no RRP.
A more precise answer is that “excess reserves” are a math calculation based on applying different % requirements to the bank’s liability structure. Excess reserves are Actual Reserves minus Required Reserves. There is no separate excess reserves account that I know of.
Anything invested in RRP would be outside the Excess Reserve calculation (I think but I’m not familiar with the exact calculation).
Agree it would be nice to have Joseph weigh in.
Yes you are right – if a bank deposited into the RRP it would be a reverse repo asset, and not a reserve asset. Like you mentioned, banks are not going to deposit into the RRP because it yields less than IOR.
However – also note that the ‘excess reserve’ idea is no longer applicable as reserve requirements have been abolished. This is largely an acknowledgement of how ‘excess reserves’ is an obsolete concept post GFC/QE.
Hi Joseph. Any clarification on the above would be much appreciated. Thanks
Dan S said in a comment above: “Some money may leave the banking system and go to money market funds. Those Money market funds then invest in RRP. So, Bank reserves down; RRP up.”
How exactly do reserves get transferred out of TGA and into RRP, without MMFs having Fed accounts?
Do reserves go from TGA to RRP? Are banks and MMFs intermediaries, each taking a cut, of this self-dealing operation? Does Treasury sell IOUs for reserves used to buy back its IOUs? And do both banks and MMFs profit on these trades?
Also, why does everyone leave out specials?
《To express a short Treasury view, a trader needs to first borrow a Treasury security and then sell it. If the price of the security falls more than the price of the purchase and associated financing cost, the strategy can be profitable. For a variety of technical reasons, the market may prefer to short a particular Treasury coupon. This can lead to localized supply and demand imbalances for a particular Treasury issuance that is reflected in repo pricing.》
https://www.cmegroup.com/openmarkets/interest-rates/why-the-market-should-watch-special-u-s–repo.html
Can bank trading desks short Treasuries using RRP?
Dan S: how do we know who RRP counterparties are? When will the data be available (see https://www.newyorkfed.org/markets/desk-operations/reverse-repo )?
How do you know banks and dealers aren’t depositing reserves in repo because they can get a higher return than IOR by shorting Treasuries, or lending Treasuries special, or whatever?
rsm: Thanks for link – good question on counter-party data. The website says the data on counter-parties is released with a delay of 1 calendar quarter – but when I search dates 6 months ago I still don’t see the data. That said, we do know it will not include banks as they earn more on reserves (0.15%) than RRP (0.05%). Joseph and other sources say it is mostly MMFs since they cannot earn IOR of 0.15% (they are not banks).
Not sure I understand your second question(s). Did you mean depositing in RRP? IOR only relevant for banks – not dealers. Shorting treasuries means borrowing. You are paying the treasury interest plus any interest to borrow the treasury security shorted. You can short treasuries and invest proceeds in longer duration treasuries and/or increase credit risk (buy corporates, MBS, etc.) to gain profits – but both entail more risk.
Actually, just found these historical files that list counter parties. But they are more than 2 years old – can’t find anything more recent. Nevertheless, data confirms mostly MMFs doing RRP.
https://www.newyorkfed.org/markets/omo_transaction_data#rrp
Dan S: is your model too cut-and-dried, too pat, a little glib? From your link:
《The New York Fed typically settles the repo and reverse repo transactions it conducts through a tri-party arrangement. In a tri-party arrangement, a third party (the tri-party bank) acts as custodian and agent for the buyer and seller. The tri-party agent is responsible for screening and approving eligible securities, as identified by type by the buyer and seller, from the seller’s pool of available securities, determining the current market value of the eligible securities, and ensuring every day that a transaction is outstanding that the buyer receives, in its account at the tri-party bank, eligible securities having a market value (based on the market value and agreed-upon margins) at least equal to each outstanding repo trade amount.》
Is there a lot of room there for dealers to get paid for “borrowing” securities from the Fed via RRP, then short those while using “cheapest to deliver” pools to send other securities on to MMFs?
May I ask how you know dealers don’t get IORB? Goldman Sachs is listed as a primary dealer; is it also a depository institution, receiving IORB? (Why have I been unable to find a list of banks that receive IORB?)
Why can’t GS get paid to borrow Treasuries from the RRP, short them to make more than the IORB rate, while delivering other cheaper-to-deliver Treasuries to MMF counterparties?
As the rapid rise in RRP volume has occurred since the most recent data on RRP counterparties that you helpfully provided, is it still relevant? When will we be able to check?
Is there a lot more going on than the simple story that MMFs are driving RRP growth?
For example, why wouldn’t an MMF’s bank pay the MMF 6 bps to keep their cash with them, so they can still make 9 bps from IORB on those reserves instead of transferring them to the Fed? Can the bank make more by having its dealer subsidiary short Treasuries even as it intermediates RRP with MMF counterparties?
The key fact is that non-banks have bank accounts with banks.
Assume you bank with Bank X and send $10K to a MMF who has an account at Bank Y. Bank X debits your account for $10K and sends $10K of reserves to Bank Y. Bank Y credits the MMF deposit account for $10K and receives the $10K in reserves from Bank X. The MMF now holds your $10K in a deposit account at Bank Y and simultaneous credits your new MMF account with 10K “shares” (par value $1.00). The Fed Balance sheet entry is to reduce Bank X and increase Bank Y Reserves at the Fed for $10K each. This is the basic payment mechanism between parties with bank accounts and no one is taking a cut (unless there is a fee for the EFT or check).
If the MMF wants to convert their new $10K deposit at Bank Y to an RRP at the Fed, then they instruct Bank Y to send the 10K to the MMF’s RRP account at the Fed. Bank Y reduces the MMF deposit account by 10K and Bank Ys reserves decrease by 10k. On the Fed balance sheet, the Fed reduces Bank Y’s reserve account and increases the MMF’s RRP account by 10K each.
If the MMF fund bought T-bills direct from the US Treasury, it is the same basic mechanism as above. The end result is the Fed reduces Bank Y’s reserve account and increases the Treasury’s TGA account.
Would love to have Joseph confirm paragraphs 2 and 3 above. I’ve assumed the RRP is not a true repurchase agreement since Joseph says in his book: “The Fed’s RRP facility offers MMFs a place to park their money at a set interest rate.” So I’m assuming that the Fed just debits Bank Y reserves and either credits the MMF’s RRP or the Treasury TGA depending on scenario above.
Would greatly appreciate your comments Joseph. Would love to know if the above text has correctly reflected the points in your book. Thank you.
Yes you’re right. Thinking of the RRP account as another deposit account (with interest) is how I think of it. Technically it is a repo – so the MMF receives Treasury collateral – but that doesn’t really matter. MMF just holds that (no rehypothecation). Note that Gov MMFs are not allowed by their mandate to do unsecured deposits, but can do Treasury backed repo.
It is not clear to me why, necessarily, the money will flow from RRP to Bills/TSY’s (it is taking 5bps for granted). FED should decrease the ONRRP or we should see some higher short term rates before, right?
What do you think?
So:
government issues a treasury -> bank buys treasury with a deposit -> government spends x money -> FED swaps treasury with a bank reserve -> deposit is withdrawed -> customer spends x money
The result is the circulating supply of currency chasing goods and services increased from x to 2x (money spent by the customer and the government). The amount of money created by QE is capped by deposits. It creates not only supply but also demand for the currency as it needs to be repaid with interest.
Is the above correct?
Hi Joseph, I caught your recent interview with Blockworks from last Friday, so where do you stand now?
Do you stand closer to “Rate Hikes Will Crash Markets” or “The end result is a rally in all assets that market observers have come to appreciate” going into the start of 2022?
Or is there a timeframe where both of these events come to fruition?
Thanks for your insight and much appreciated.
In my view it comes down to how the rates hikes are executed. If they are slow and shallow (as they are currently forecast to be) then that is risk positive. The concern is that inflation is not transitory and the Fed becomes more aggressive. This matters because faster hikes mean market has less time to adjust, and higher rate path makes it more difficult to adjust to. Recent political commentary is slightly increasing the risk of an aggressive Fed, but too early to say
Looking forward to your next article, especially if you have further clarification on which way it’s going to go.
Everyone’s nervous and trying to read tea leaves and we’re looking for any sort of guidance.
Hi,
Can the Treasury repay the Fed after issuing more T-bills to MMFs? in theory get $1.5tn back and repay the Fed
and then issue more coupons to banks?
Wouldn’t this alleviate the QT impact?
wont banks just push these excess reserves right back onto the mmfs like they did before?