Earlier this week (on Tuesday, February 27, 2018)
researchers at the Fed published a paper, inauspiciously titled
A Model of the
Federal Funds Market: Yesterday, Today and Tomorrow. The authors develop a quantitative
model of the fed funds market –honestly one of the best and by far the most
comprehensive that I’ve seen – and project changes in future trading volumes
and levels of the fed funds rate. The dynamics are not based on the pace of fed
funds rate hikes themselves, though they do assume hikes proceed along the Fed’s
baseline. Instead, the model projects the level of the effective Fed funds rate
relative to the target as the Fed undertakes policy normalization and aggregate
outstanding reserves – excess cash in the banking system – are drained by the
Fed, declining from roughly $2 trillion to $200 billion.
This reduction – or normalization - of the level of
reserves is expected to gradually reshape the interbank market, raising the trading
level of the effective fed funds rate above the upper bound of the fed funds
target, due to increased trading volume and an increase in the number of
participating banks. The return of bank trading activity will supplant the
current domination of the fed funds market by the GSEs. These projections, if
realized and supported by the FOMC as prudent policy, obviously have important
implications for a variety of short term interest rates and spreads, including those
for repo, overnight index swaps (OIS), Libor and its descendants, the overnight
bank funding rate and commercial paper.
Key Takeaways
·
The authors project that large banks will return
to lending in the fed funds market when the level of reserve balances falls to
$900 billion.
·
At the current pace of policy normalization,
reserve balances are expected to reach $900 billion in January of 2021. Fed
projections of SOMA holdings and the level of reserve balances can be found
here.
·
The baseline projection is that as total
reserves pass through the $850 to $800 billion range, trading volume in fed
funds increases and the effective fed funds (EFF) rate begins trading above the
interest on reserves (IOR) rate, which is set at the upper end of the fed funds
target range. This rise in the EFF rate relative to the target should happen in
February / March of 2021 timeframe.
·
By December 2021, when aggregate reserves fall
by another $400 billion, the model projects that the EFF rate will be sloping
upwards to the midpoint between the upper bound of the FF target range and the
discount window lending rate, which is set 50 basis points above the upper
bound of the target.
·
That implies the EFF rate will float from 8 bps
below the FF target to 25 bps above it, tightening short term interest rates by
an additional ~33 bps as the Fed proceeds with policy normalization and reserve
balances decline. This additional move higher in the EFF rate would presumably
occur gradually throughout 2021.
Excerpts from the paper
Caveat: I would
strongly recommend reading the entire paper, as it is a terrific analysis and
includes much more detail and nuance, as well as confidence intervals around
their model projections, than I can cover here. Emphasis added.
For our main policy exercise we
trace the evolution of the FF market as we reduce the aggregate supply of
reserves from its current levels down to $200 billion Doing so requires us to
specify the complete dynamics of the distribution of excess reserves across
banks along the path, for which we have little more than an educated guess. For
our baseline case we find that the banks
with the largest balances return to lending funds at around $900 billion in
total reserves, driving a resurgence in FF volume. Quite quickly the EFFR drifts above the IOR—somewhere
between $800 billion and $850 billion—as bank-to-bank trades necessarily
execute at rates above the IOR. This is an important event, as it marks the end
of the current implementation framework that equates the IOR to the top of the
target range for the EFFR. However, it
takes about $400 billion less aggregate excess reserves for the EFFR to start
sloping upward to the midpoint between the IOR and the discount-window rate—what
would be the hallmark of a classic corridor system.
We find that the evolution of the FF market is quite
sensitive to the dynamics of the distribution of excess reserves across banks.
In particular, the extent to which the largest banks hoard reserves is crucial
to determine when the EFFR crosses above the IOR. By varying the rate at which
the banks with higher balances reduce them relative to those banks with lower
balances, the EFFR first drifts above the IOR anywhere between $500 billion and
$1.1 trillion. Of particular interest is the possibility that the largest banks
plan to rely on reserves as high-quality liquid assets to satisfy regulations.
If so, the distribution of reserves becomes more concentrated at the top, and mechanically
there is a larger fraction of banks with low balances. This combination increases the gains of trade and leads the largest
banks to lend funds at a higher level of aggregate reserves, driving the EFFR above the IOR much sooner
than under the baseline.
Refresher on Fed’s
Monetary Policy Rates
Secured lending to the
Fed
Ø
Safest counterparty (the Fed) + safest
collateral (Treasuries) = Lowest return on investment
Overnight Reverse Repurchase rate (ON RRP) – Interest
rate offered by the Fed for overnight loans, with borrowed funds secured by
Treasury collateral from the Fed’s SOMA portfolio. These borrowings of cash in
exchange for collateral are categorized as open market operations by the Fed,
and they temporarily drain reserves from the financial system. Eligible counterparties
include primary dealers, banks, money market funds and the GSEs. The offering
rate is set at the lower bound of the FF target range, though it can fluctuate
in certain circumstances based on demand (see additional notes); currently 1.25%.
Borrowing unsecured
in the US based interbank market, e.g. the Fed funds market
Ø
A US bank borrowing from another US bank or GSE,
to meet the level of cash reserves depository institutions are required to hold
in their accounts at the Fed. Roughly speaking, these required reserves are 10%
of a bank’s transaction deposits.
Ø
Prior to QE and LSAPs the level of excess cash
in the system was structurally low, so the Fed funds market was very active and
liquid. The Fed conducted daily open market operations as a lender of cash to keep the effective
interbank borrowing rate close to their Fed funds target rate. These were
overnight and short term repurchase (repo) operations.
Ø
As the Fed normalizes monetary policy they will
gradually reduce their portfolio of Treasury and agency securities, and will conduct
reverse repo operations, both of which will drain excess cash reserves from the
banking system.
Interest on Reserves (IOR) rate – Interest rate paid
by the Fed to depository institutions for reserves held in their account at the
Fed. Set at the upper bound of the target FF range; currently 1.50%.
Ø
Pointing out the obvious (because I’m great at
it) – since a depository institution can leave their cash in their reserve
account at the Fed and earn the IOR rate, there is no incentive for that bank
to lend money to anyone at a lower rate.
Ø
The reverse is also true. If a bank can borrow
funds from another institution at a rate below the IOR, they can park those
funds in their account at the Fed and earn the difference. A nice risk-free
arbitrage.
Ø
In theory the IOR sets a lower bound for
unsecured interbank lending rates. In practice it’s imperfect, due to (a) regulatory
charges imposed by the FDIC; and (b) the participation of the GSEs in the
interbank market. The GSEs are not required to hold reserves at the Fed, so cannot
earn the IOR rate on their excess cash.
Ø
The GSEs will therefore lend cash in the interbank
market at rates below IOR, and banks will execute the arbitrage. In times of
abundant reserves – post QE and LSAPs – these lending operations dominate the
interbank market, pushing the effective Fed funds rate below the upper bound of
the target.
Target Fed funds rate (FF) – Fed’s monetary policy
target range for overnight interbank lending. Difference between lower and
upper bound set at 25 basis points; target adjusted in 25 basis point increments
(e.g. +/- 25 bps, +/- 50 bps); current FF target range is 1.25% to 1.50%.
Effective Fed funds rate (EFF) – Calculated as the
volume weighted daily median borrowing rate where reserves actually trade in the
US based interbank market. This trading data is reported by depository
institutions, and collected and published by the Fed.
Ø
The FF target range is currently 1.25 – 1.50%,
and the effective FF rate – with the exception of days which are on month-end –
typically trades 8 bps below the upper end of the target. Right now that means the
EFF rate is trading steadily at 1.42% with daily trading volume in the range of
$80 to $100 billion. You can find complete details
here.
Financial
institutions borrowing funds from the Fed, secured by a wide range of
collateral
Discount window lending rate – The rate the Fed
charges eligible institutions to borrow money from the Fed on a secured basis,
where a very wide range of collateral is accepted. Borrowing from the Fed at
the so-called discount window for large institutions is generally seen, and
meant to be, a measure of last resort. There are smaller banks and financial
institutions which engage in typically seasonal borrowing from the discount window,
which is not an indication of stress. The discount window rate is set at the
upper bound of the target Fed funds rate + 50 basis points; currently 2.00%.
Technical note:
The GSEs (Fannie Mae, Freddie Mac, the Federal Home Loan Banks, et al) are not
eligible to earn IOR on their excess cash because these organizations are not depository
institutions. Their participation in the Fed funds market is one-sided as they
enter only as lenders of funds to depository institutions (banks), not as
borrowers. Because the GSEs (a) cannot earn IOR from the Fed; and (b) do not
owe depository institution insurance fees to the FDIC, they are willing to lend
their excess funds at rates below IOR.