This is the second of two posts focusing on “Overnight RRP Operations as a Monetary Policy Tool: Some Design Considerations”, Federal Reserve Bank of New York Staff Reports, Staff Report No. 712 February 2015.
So what sort of design features could mitigate the potential problems in the facility? Will the RRP permanently change financial plumbing? One answer is to make it clear the RRP is a temporary measure. From the article:
“…A temporary facility is less likely than a permanent one to cause lenders to curtail relationships with money-market borrowers, to abandon money-market infrastructure, and to rely exclusively on ON RRPs as a place to invest cash…”
If the repo market is shrinking due to regulatory pressure, yet there is still high demand for safe assets (that repo provides cash investors), we wonder if making the RRP temporary really helps? In a crisis, money will seek a safe port. Better with the Fed than in a mattress. Perhaps the answer is making it clear that the RRP facility will be available should it be necessary will comfort investors that there is a place to land when they need it and slow down pre-emptive runs. Think of it as the Fed becoming the “borrower of last resort”.
The pivot toward policy “normalization” is not when the Fed raises rates, but when there is a meaningful decline in excess reserves. As the Fed’s balance sheet declines, so will excess reserves. But that will take years and years. Only then will IOER become an effective tool to control short-term rates and the RRP facility won’t be necessary. But the Fed knows this already:
“…the Committee intends to reduce its securities holdings in a gradual and predictable manner primarily by ceasing to reinvest repayments of principal. This process will result in a reduction in excess reserves, and thus gradually improve the extent to which IOER exerts upward pressure on money market rates. For example, a lower amount of reserves in the banking system should reduce banks’ balance sheet costs and the costs associated with arbitrage activities, and hence foster a tighter link between the IOER and other market rates…”
We wonder about the link between lower reserves and banks’ balance sheet costs. Will the spreads banks earn simply borrowing short-term cash and lending either as IOER or Fed Funds really get to where they cover the regulatory costs and justify blowing up the balance sheet? Is it reasonable to assume that linkage still will behave the way it once did (given the pressure to keep balance sheet down)? The banks may not take the bait.
How about a shift toward term RRPs?
“…during the normalization process, the Committee could complement ON RRPs with term RRPs. The Federal Reserve could also use a Term Deposit Facility (TDF), which would allow banks to maintain term deposits with the Federal Reserve at a higher rate than the IOER rate. These tools could provide additional arbitrage opportunities for banks and nonbanks and help to exert upward pressures on money market rates…”
Is selling some of the Fed’s holdings as a way to reduce system reserves completely off the table? Maybe not.
“…Finally, the Committee could also choose to sell assets, which all else equal, would reduce the amount of reserves in the banking system and diminish the aggregate size of the banking system’s balance sheet…”
How about widening the spread between RRP and IOER? This would inflict some pain discipline on cash investors.
“…the spread between the IOER rate and the ON RRP rate could be set wide enough to discourage very large average ON RRP take-up and reduce the facility’s usage in normal times. If the ON RRP rate is significantly below the IOER rate, money market investors are more likely to invest (directly or indirectly) in reserves and less likely to invest in ON RRP—that is, changes in the spread should cause substitution between these two forms of Federal Reserve liabilities…”
But:
“…if investors become rate-insensitive during such episodes—demanding safe assets at essentially any rate—reducing the rate may be ineffective in discouraging usage. Moreover, demand for safe assets can increase abruptly in a crisis, so a surge in ON RRP take-up could occur before policymakers can react…”
The widening of the spread could easily be done as the Fed raises rates – simply leave the RRP rates alone while IOER goes up. But this assumes that the transmission mechanisms continue to work. Will banks bid up cash from clients and recycle it, bloating their balance sheets for 10 or 20 basis points? Will cash investors care that the RRP rate is so much lower? Probably not.
Another modification discussed in the article was capping the program, either by participant or overall. The Fed introduced these in September when they limited the overall program to $300 billion per day and individual participation to $30 billion. Also included was an auction process to clear the market if demand went above $300 billion. The article also raised the idea of awarding repo on a pro-rata basis when demand exceeds the cap. The Fed is experimenting with these constraints, but they are likely to be binding only over statement dates (when window dressing exercises take over). Based on the limited data points, extrapolating about how the market will behave in a crisis is difficult.
From the report:
“…A key advantage of having only individual caps is that they allow each counterparty to know in advance the maximum amount that it can invest at the facility, since the amount would not depend on the behavior of other ON RRP counterparties (as it would with an aggregate cap if it was binding)…”
“…An aggregate cap on its own could provide a more flexible allocation of usage among counterparties than a system of individual caps. Hence, an aggregate cap may be better suited to allocating ON RRPs to the counterparties that value them most…”
Caps might be dynamic or static – or some combination.
“…Static caps are set in advance and do not vary unless policymakers choose to adjust them. In contrast, dynamic caps would be adjusted at regular intervals (such as daily or weekly), perhaps using a pre-specified rule, which might be based on recent or projected usage. For example, an aggregate dynamic cap could be reset every week at an amount equal to average take-up during the previous week plus a specified amount. Such a cap could be used as a “circuit breaker” that is calibrated to contain potentially disruptive surges in take-up (Dudley 2014)…”
The level of the caps is important.
“…Very low limits that often bind could contain the Federal Reserve’s footprint in short-term funding markets and virtually eliminate potentially disruptive surges into an ON RRP facility, but such caps also would diminish the effectiveness of ON RRPs in achieving their primary purpose, that is, in helping to establish a floor on short-term interest rates…”
Too high a cap and….
“…very high caps set at levels that are never close to binding could allow an ON RRP facility to set a firmer floor on interest rates. But such high caps may not do enough to address concerns that a large ON RRP facility could expand the Federal Reserve’s role in short-term intermediation and potentially pose a threat to financial stability. Figuring out which constraints to use (and how much) is a tough exercise all about tradeoffs. The Fed is trying to insure predictability yet not sacrifice the banks in the process.
And of course the tradeoffs between the levels of the caps is itself dynamic. As banks change their business models in response to regulatory requirements, their appetite for short term funding will change.
“…the full implementation of new capital and liquidity rules in coming years may drive up financial institutions’ costs of short-term funding and diminish their demand for receiving this type of financing. This could reduce the availability of money market instruments and boost demand for ON RRPs. If so, all else equal, a larger ON RRP facility may be needed to prevent a widening of spreads between the IOER rate and other short-term interest rates…”
and
“…Once the bottom end of the target range is raised above zero, monetary policy tools that are intended to set a floor for short-term interest rates, such as an ON RRP facility, may need to be relied upon more heavily…”
The RRP facility is intended to be temporary and used in conjunction with IOER to control interest rates from going below the Fed’s target. As and when monetary policy becomes normalized (whatever that really means), the RRP is expected to sunset. But the article suggested that when policy is initially “normalized”, the RRP might be boosted (at least temporarily) to insure that stress created by an increase in rates doesn’t blossom into a destabilizing flight-to-quality.
We suspect that as long as banks are holding large excess reserves (which in turn is driven by the Fed’s securities portfolio), that RRP will be an important tool and isn’t going anywhere. The tweaking of the program specifics — individual and program-wide caps, auction versus pro-rate, rates relative to IOER, etc. — are important to get right and we applaud the Fed for experimenting. But in a crisis, cash will run from the banks and seek safe havens. And all bets will be off.