The Fed’s new domestic Standing Repo Facility (“SRF”) makes Treasuries more fungible with reserves and will thus slightly impact the composition of GSIB liquidity portfolios. Post-Basel III GSIBs are required to hold large High Quality Liquid Asset (“HQLA”) portfolios that in practice largely consist of reserves. Although reserves and Treasuries are equal under the letter of Basel III, regulators prefer banks to hold reserves because they are more liquid. This distinction was further highlighted last March when many investors had trouble liquidating their Treasuries. An SRF addresses this concern by allowing GSIBs to instantly convert HQLA securities to reserves. The primary mechanism through which the SRF impacts markets is thus through GSIB HQLA portfolios. An SRF means GSIBs can hold fewer reserves, and more Treasuries. In this post we review the SRF, show that reserve demand is not currently constraining GSIB HQLA portfolios, and suggest the SRF’s market impact will be slight.
The Domestic SRF
The SRF announcement turns the “temporary repo facility” the Fed has already been running each day into a permanent facility. The only major change is it will now also be accessible to banks and not just primary dealers. Recall, an SRF is essentially a stigma free discount window that allows borrowers to instantly convert high quality securities to cash at an above market rate. Its primary function is thus rate control and liquidity provision.
From a rate control perspective, the SRF helps put a ceiling on repo rates by offering to lend almost unlimited cash at a rate equal to the top of the Fed’s target range. The facility may improve sentiment, but will probably never be used. There is ~$1t sitting in the RRP waiting for a chance to earn anything more than the RRP rate, and a few trillion more in banks waiting to earn anything more than IOR. There will be no upward pressure in repo rates for the foreseeable future.
The SRF’s big incremental change is the inclusion of banks as counterparties. From a liquidity perspective, that is unnecessary. Basel III mandates huge liquidity buffers and makes runs on GSIBs virtually impossible. In the event that there are liquidity needs, GSIBs can always borrow from the FHLBs or the discount window. But the SRF will allow regulators to treat Treasuries as more fungible with reserves. A GSIB could now conceivably have a Treasury heavy HQLA portfolio on the claim that Treasuries can readily be converted to reserves at the SRF.
But this only matters if a GSIB’s desire to hold Treasuries is being constrained by its need to hold reserves.
Reserve Demand Is Much Lower Than Believed
Legend has it that the Fed shrank its balance sheet too much and blew up the repo market in September 2019. That is not strictly inaccurate, but it obscures a key dynamic of the episode. It was not the banks who needed money, but the repo dealers. This nuance is important because it shows that banks were not reserve constrained even when system reserves were around $1.3t (vs $4t today).
Heading into September 2019, Treasury repo volumes and rates had been steadily increasing for months. Banks had become the marginal lenders in Treasury repo, which had the same regulatory treatment as reserves but were higher yielding. This was an unstable equilibrium because demand for repo financing was growing even as QT was shrinking the cash available for lending. Eventually, something broke and the repo borrowers scrambled for cash by bidding up repo rates.
Banks were net lenders in the repo market, not net borrowers. They lent hundreds of billions of extra reserves into repo when rates are above IOR, but even that was not enough. If banks had more reserves could keep on lending, but that is very different from saying that the banks themselves were short of reserves. If banks were short of reserves, then the first thing they could do is reduce their lending into repo. They did not do that. In fact, a later survey revealed that the great majority of surveyed banks maintained or increased repo lending during the September repo spike.
During QT the Fed conducted periodic surveys of all the major banks to understand their demand for reserves. The surveys showed that banks largely held reserves to make payments and satisfy liquidity regulations. The surveyed domestic banks (which included all US GSIBS) consistently reported that on aggregate the lowest level of reserves they would be comfortable holding was around $400b. For context, JPM alone held $465b in reserves on the March 2021 quarter-end.
Balance sheet normalization could have continued much longer without any impact to the banking sector, though it would have hurt the repo dealers (and their hedge fund clients). If GSIBs were not constrained by the need to hold reserves back in 2019 during QT, they certainly are not constrained today.
SRF’s Impact Will Be Slight
Liquidity metrics are just one of many factors that determine the composition of GSIB HQLA portfolios. An easy way to see this is how differently JPM and Bank of America, two large universal bank GSIBs, manage their HQLA portfolios. BoA sees that deposit rates are ~0% (and sticky), so its strategy is to hold a securities heavy HQLA portfolio, swap them into floating and benefit from future rate hikes. In contrast, JPM’s strategy is to maintain large reserve holdings until rates rise.
The SRF lowers the minimal level of reserves GSIBs are required to hold by making Treasuries (and other HQLA securities) more liquid. But the need to hold reserves does not to appear to be constraining GSIB Treasury holdings at the moment, so the SRF’s ultimate impact on Treasury demand (and thus rates) will be slight.
10 comments On Reserve Demand Post-SRF
Could you explain a little more about what happened in September 2019? How exactly did the repo dealers run out of money? Was it a matter of Balance Sheet constraint, or did lenders (to these dealers) pull back?
Lenders didn’t pull back, they just didn’t have any more money to lend any more. Repo demand kept growing, leverage kept building, and then one day repo dealer went to the market and realized there wasn’t enough money for everyone. Then the panic bidding started.
That was a great show on Gammon! You’re a brilliant writer but this long-form conversational format really suits you too- plus it helps with some of these more finicky details, and Gammon‘s a great host. Will you be doing any more long form stuff? Maybe with Emil and Snider? Snider has some different opinions, and it would be great to hear the two of you discuss bilateral repo, t bills, dealers and Qe. Again, thanks for sharing your knowledge!
Would the SFR help commercial banks soak up more treasuries ( issued from deficit spending) as a reserve drain?
I have a question regarding bank loans vs capital market lending and money creation. You addressed this from the banks as lenders perspective of money creation by loans. But in the capital market process you addressed this from the perspective of the borrower and not the lender.
Isn’t the lender ( i.e. a fixed income investor) when purchasing a bond on margin through a bank that is providing the leverage of the bank’s balance sheet in fact creating or adding to the money supply?
Banks are leveraging their balance sheets when they create deposits backed by an asset i.e. a loan. But deposits are not created to sit idle, they are created to complete a transaction, so reserves are required to complete the transactions at the interbank level. ( your tier 1)This process is the same when a banks provides margin to a capital market investor.
When a financial institution expand it’s balance sheet through leverage for a client when making a purchasing of a bond, aren’t they in fact providing elasticity in the money supply? The banks are in fact creating a loan. Isn’t this how the wholesale funding markets grew and was eventually exposed by the financial crisis. The institutions were if fact leveraging their own balance sheets creating labilities and by default creating mismatch on their books at the commercial banks level( your tier 2) When the source of wholesale funding dried up this exposed their lack of reserves to match the build up of liabilities, then the Fed had to come in after the fact to create the reserves (QE) that were exposed by this mismatch. I guess my questions is……isn’t it the leverage that is creating the elasticity and only when the discipline of settlement is required, because you can not put it off by continuing to borrow, that the completion of the transaction requires the additional reserves resulting in the expansion of the money supply?
I think a loose analogy may be (not sure) the difference between the thinking of….. banks need reserves to create deposits or the bank creates deposits then acquire the reserves, either way you think about it…. the money supply expands.
The reason i ask this is because if in fact the capital market path of lending is only the transfer of existing money supply and not creating additional liabilities then this system as a whole would never fail, the GFC exposes that as a false construct.
I look forward to your response.
PS Your work is a breath of fresh air is a space that is filled with people that have no experience in these matters, keep it up!
If a bank makes a loan, that creates a deposit and adds to the money supply.
If an investor makes a loan (by buying a bond), they just transfer existing reserves to the borrower, and no additional money is created.
If an investor makes a loan (by buying a bond) that and they themselves borrowed from the bank, then their bank borrowing created new money.
Dear Joseph, Thank you very much for your blog and your book. It is a very good read. Quick question on the value of the “Treasury Securities Sold by the Federal Reserve in the Temporary Open Market Operations” versus the value of the SOMA holdings. Is it correct to compare as of August 18, 2021: 1115.656 billion (Reverse repurchase agreements – Series ID RRPTSYD) to 4559.562 (US Treasury Notes and Bonds as – NY Fed)? Is it correct that as of today MBS are not used? The links are: https://www.newyorkfed.org/markets/soma-holdings
Those are different but related items on the Fed’s balance sheet. SOMA holdings are the Treasuries the Fed owns outright. When the Fed engages in RRP what it is technically doing is taking some of those Treasuries it owns outright and then selling them with a simultaneous contract to buy them back the next day. This is functionally a secured loan. So when a money market fund wants to invest in the RRP, it is technically buying a Treasury from the Fed, then selling it back the next day at a slightly higher price. That slightly higher price is the “interest” from the secured loan. SOMA holdings are a theoretical limit as to how high the RRP can go – as Fed needs to own Treasury securities to offer as collateral in the transaction.
Hi, bit of broad question
Firstly am I correct on these definitions at a high level:
SRF – banks borrowing reserves overnight by pledging treasuries as collateral
RRP – Banks lending excess reserves overnight in return for treasuries
The news over the past few months on RRP usage being greater than 1trillion – am I right in saying banks earn higher rates on ‘Interest on Reserves’ (0.15%) compared to RRP (0.05%)?
If so what could be reasons for the high demand in RRP – Is it just because of QE and too much cash in the system?
Banks don’t participate in the RRP because – as you noted – the RRP offering rate is lower than IOR. It is the money market funds, who do not receive IOR, that participate. See here.