Today I celebrated my birthday and Jerome Powell celebrated an FOMC meeting. I reckon my cake was better than his – see below. So, here’s a gift: some comments on US liquidity. You can’t complain about too much of a good thing. I last looked at system liquidity in September 2023. Since then, ONRRP has declined much faster than I estimated, and, surprisingly, System Reserves have risen. Which goes to prove that projecting system liquidity demand is a mug’s game. So, let me risk looking a mug and explain why US banks are rebuilding reserves. It is a tale of simultaneous fear and greed, which is unusual even in finance.
Banks’ demand for reserves is driven by: the interest paid on reserves and the size of their deposit base – the main liability against which reserves serve as a protection.
Reserves earning 5.40% is truly a gift for banks, especially as Jerome Powell today suggested this level may remain in place for a bit longer than some had expected. It’s better than cake.
For, in addition, reserves perform a risk-insurance function. Indeed, this function the main long-term driver of reserve accumulation. Banks need to feel safe, especially when a good part of their securities portfolio show deep losses. A good ‘ball-park’ estimate of the ‘safe’ floor of total liquidity demanded by the banking system, (provided by Fed researchers Lopez-Salido & Vissing-Jorgensen) is ~US$3.5 trillion – you can follow their arguments in this link.
Their paper helpfully provides the following formula:
Reserves = Securities-Autonomous Factors + Loans to banks – non-bank deposit facility (mainly ONRRP)
From which we can derive:
Total Liquidity = Reserves+ONRRP = Net Securities (SOMA-Autonomous factors(mainly TGA)) + loans to banks
ONRRP is overwhelmingly dominated by flows from money-market funds, and has been plummeting since Q2 2023. It is the biggest contributor to change in Total Liquidity.
ONRRP adds to Total Liquidity because if reserves falls below the ‘safe’ level demanded by system, banks can bid for additional liquidity from money-market funds by offering higher rates in repo than is available from ONRRP. However, the valuable opportunity and credit option for ONRRP holders (money market funds) means the interest rate offered needs to be high; high enough even to breach the Standing Repurchase Facility backstop offered by the Fed – as happened at year-end 2023.
While a large stock of ONRRP exists, overall liquidity will remain above the ‘safe’ lower bound, even though QT is removing liquidity steadily by running down the Fed’s securities portfolio. As Lopez-Salido & Vissing-Jorgensen write:
“From an interest-rate volatility perspective, it is prudent to run down securities only to the point that fluctuations in autonomous factors will not result in reserves+ONRRP below the value assessed to be feasible value (e.g., below $3.495T…)”
Anything below this level may cause tension in overnight money markets as reserve demand rises above available supply.
As ONRRP holdings fall, the ‘lower bound’ of system liquidity gets closer, banks have a growing incentive to hold more precautionary reserves. They also get paid for the privilege. This is what we see happening.
The most important driver of the decline in ONRRP has been a rise in Treasury General Account (TGA) the main autonomous factor in the equation above. In other words, fiscal policy is an important recent determinant of system liquidity. The relationship is clear since late 2020 by comparing Total Liquidity (Reserves + ONRRP) to TGA using the same y-axis scale, as below.
It is worth noting, though I’ll save you the chart, that Total Deposits have been stable since Q2 2023 and are expected to remain so. This alleviates some pressure on liquidity.
If Lopez-Salido & Vissing-Jorgensen are correct in their estimate of a safe lower bound of Total Liquidity at $3.5 trillion, then a simple application of the Merton Model suggests the possibility of breaching that lower bound in the next 12 months is high.
The following chart shows the impact of announced QT on the probability of breaching various lower bounds of liquidity. At current Total Liquidity levels of US$ 4.2 trillion, there is a 78% chance of hitting the ‘safe’ lower bound of US$ 3.5 trillion. The calculation assumes volatility of liquidity in the last 12 months continues at the same level for the next 12 months.
You can also read across to see how lower levels of of current Total Liquidity affect the probability of breaching the lower bound in 12 months time. I’ve also included other probabilities based on other, lower assumptions of required liquidity (US$ 3.0 trillion and US$ 2.5 trillion).
The maths tell us there is a high probability the ‘safe’ lower bound of Total Liquidity will be breached in the next 12 months, assuming Lopez-Salido & Vissing-Jorgensen are correct. That is not a huge concern for the system – the Fed can always add temporary extra liquidity, or halt/reverse QT. However, for individual commercial banks, it may prove at least inconvenient and costly and potentially dangerous to find themselves short of liquidity. So, yes 5.40% return is wonderful, but a good buffer of reserves is probably a necessary condition for bank executives sleeping soundly.
If Lopez-Salido & Vissing-Jorgensen are right, then it makes sense for banks to build their reserve holdings. It is also a high return investment; much better than most Treasuries, except T.Bills and a better return than most other investments, combined with the highest credit possible.
Bankers need cake and are eating it too.