It’s likely that bills will trade negative in the coming days as Treasury pays down existing bills, thus reducing their supply by a few hundred billion (see this previous post for more context). In this post I will discuss why large paydowns would push bill yields negative, the mechanics behind how IOR/ON RRP adjustments work and whether they make sense now, and other potential policy options to keep bill yields positive. It does not offer any view on which option, if any, may be taken. This post is meant to educate on a obscure corner of the market.
Why Bills are Going Negative
Broadly speaking, the purpose of markets is to allocate resources that are scarce. Scarce resources move according to price signals to those who are willing to pay the most for them. Resources that are not scarce – like air – have no prices. Forever QE has made money super abundant to banks and sovereigns, so much so that it has no value to them. Banks have more deposits than they want (which is why LIBOR is heading towards 0), and the Treasury has more money than it wants. Yet, investors still need a place to park their money. Investors who are only willing to park their money in liquid risk free assets must buy assets like Treasury bills, and if there are not enough bills their demand will push the yields negative.
In the current Fed framework, the Fed sets a floor on interest rates by adjusting the offering rate of the Overnight Reverse Repo Facility (“ON RRP”), which is essentially a risk free investment option offered to money market funds, an investment complex with almost $5 trillion in assets. Given that these funds could lend to the Fed risk free at on the ON RRP offering rate, the funds should never be willing to invest in anything below that offering rate. Their portfolio allocation decisions are the mechanism that transmits the ON RRP rate throughout money markets. The ON RRP has proven to be a sturdy floor for repo transactions backed by Treasury collateral, where SOFR has never printed below the ON RRP offering rate.
However, the ON RRP has not been a strong floor for short term Treasuries or Agency debt. Prior to 2018, yields on those assets frequently traded below the ON RRP offering rate. This is because many investors demand risk free assets but do not have access to the ON RRP (or Treasury backed repo). They may also view Treasury backed repo as insufficiently liquid (bills can be sold intra-day, but overnight repo investments are locked in the tri-party repo platform until the next afternoon), or do not want to take any repo counterparty risk.
In early 2018 the U.S. Treasury flooded the market with $300b in net bill issuance, which pushed all money market rates above the ON RRP floor. Since then net issuance has steadily grown, until now. The Treasury estimates it will lower the supply of bills by a few hundred billion in the coming weeks. (GSE discount note issuance, a comparable risk free investment, has been declining as well).
The investors who will only (or can only) hold bills will be competing for a lower supply of bills, and may drive yields negative. It’s hard to know the buying power of these investors, except that they drove bills below the ON RRP floor pre-2018 and likely have much more money today. At the moment short dated bills are trading at around 2 basis points, so a sizable decline in bill supply could push bills negative. If the Fed is concerned, here are some options it could undertake.
Raise Interest on Reserves
IOR affects money market rates by adjusting the opportunity cost of commercial banks, who are borrowers in the fed funds market (very rarely lenders) and lenders in the repo market (rarely borrowers). The Fed has made several 5 basis point (0.05%) adjustments (“technical adjustments”) to IOR over the past few years to move the fed funds rate (“EFFR”) closer towards the center of the target range. This tool as been very effective in moving the EFFR, and somewhat effective in influencing repo rates. Banks don’t usually own bills (their Treasury holdings tend to be in the belly of the curve), so any impact from a technical adjustment on bill yields would be indirect.
Foreign banks Borrow Fed Funds to Earn the IOR-EFFR Spread
The Federal funds market is a market where banks lend each other Fed reserves on an overnight basis. In the pre-Crisis world it was a dynamic market that set the overnight opportunity cost of banks. Basel III effectively killed the market by strongly discouraging banks from engaging in unsecured, short-term inter-bank borrowing. Fed funds has been a zombie market for years, with EFFR printing at the same rate virtually every day. The entire market today is essentially made up of a few foreign banks borrowing from a few Federal Home Loan Banks (“FHLBs”).
Some foreign banks are active borrowers in the federal funds market because they borrow at EFFR, and then deposit that money at the Fed to to earn the spread between IOR and EFFR (currently around 3bps). Almost all lenders in fed funds are FHLBs, who have Fed accounts but don’t receive IOR (see this speech by Simon Potter for more details). The FHLBs want to receive some return on their money, so they lend it to the foreign banks. Foreign banks are able to do this trade because they are under a slightly different regulatory structure that allows the trade to be profitable. (Specifically, they don’t pay FDIC insurance fees and their leverage ratios are calculated on period-end snapshots rather than daily averages.)
The foreign banks seek to the maintain the spread of their IOR-EFFR trade, which is already barely profitable. A 5bps move in IOR directly impacts the profitability of this trade, so they correspondingly adjust the rate they are willing to pay the FHLB to maintain their spread. This mechanism tightly links the EFFR to IOR.
Big Banks Lend in Repo to Maximize HQLA Returns
IOR affect the repo market through a very different mechanism – a lending channel. When repo rates rise above IOR, some banks will lend their own treasury cash into repo to earn a little extra return. This trade is commonly done by large U.S. banks as it is neutral from a regulatory perspective – repo lending against Treasury collateral and Fed reserves are both considered HQLA (high quality liquid assets). When SOFR (the Fed’s index for overnight repo loans backed by Treasuries) rose to trade consistently above IOR in late 2018, banks increased their collective investments in repo by $300b. JP Morgan notably increased its repo lending by over a $100 billion in late 2018 as repo rates moved above IOR (this was discussed by its CFO on its 2018 Q4 earnings call).
Fast forward to today, where SOFR is several basis points below IOR, we see bank lending in repo dropping steadily. Adjusting IOR affects repo rates by influencing the amount of bank lending in the repo market, but becomes ineffective when repo is already below IOR (as it is today).
Raising IOR will thus move the EFFR but have very limited impact on repo. Since fed funds is a dead market, changes in EFFR would not have any impact on other money market rates. There are no mechanisms that link EFFR to bill yields, as the participants in the funds market (banks and FHLBs) have limited bill holdings.
Raise ON RRP
The ON RRP rate is the opportunity cost of money market funds (2a-7 funds), a gigantic class of investors that have a bit under $5 trillion in assets. These funds have no reason to invest in anything below the ON RRP offering rate, and through their investments link broader money market rates to the ON RRP offering rate. For example, money funds invest $1 trillion in the repo markets. It is difficult for repo rates to trade below the ON RRP rate if vast swaths of the market will only lend at or above the ON RRP rate.
Raising the ON RRP rate when overnight repo rates are already bumping along the floor would pull repo rates higher. The arbitrage relationships between bills and repo should be strong enough to also pull bill yields higher. In addition to $1 trillion in repo, money funds also hold over $2 trillion in bills. Right now short-dated bills and overnight repo are both around similar levels, so higher repo rates would cause some funds to allocate from bills to repo. This mechanism should put some upward pressure on bill yields.
The problem is that it that it’s very hard to know A) how much below the ON RRP bills are going to trade as bill supply decreases, and B) how much of the ON RRP hike would pass-thru to bills. The bills yields to ON RRP chart earlier suggests that in 2016 bills have traded over 15 basis points below the ON RRP floor. The crossing into negative yields may prevent bills from trading too much below the ON RRP, but how much below is just guesswork. A small 5bps ON RRP hike may not be enough to keep bills positive, and a larger adjustment would look very much like a rate hike.
Sell SOMA Bill Holdings
The Fed holds about $320b in bills that it purchased in late 2019 in response to the September 2019 meltdown in the repo market. The Fed perceived the meltdown to be connected to a low level of aggregate reserves, so it added reserves into the system by purchasing Treasury bills. Those bills have been reinvested as they matured, though their original purpose of increasing aggregate reserves seems more and more quaint as the Fed balance sheet continues to grow rapidly.
If the Fed were to sell those $320b in bills, that would largely neutralize the decrease in bill supply due to Treasury paydowns. This should counteract the downward pressure on bill yields.
Maybe Nothing Needs to be Done
Of course, there is also the option of doing nothing. The Fed’s policy rate is EFFR, and there is in no danger of EFFR trading negative. All participants in the fed funds market have access to a Fed account offering 0% (GSEs) or 0.1% (Banks). It would never make sense for them to lend at rates below 0%.
Political developments are very difficult to predict, but it appears that a $1.9 trillion stimulus plan is coming in the following weeks. There also whispers of trillions more on a new infrastructure bill. All that spending needs to be funded, and that means more bill issuance. Probably much more. All this talk of not enough bills may be forgotten very soon.