"If you put the federal government in charge of the Sahara Desert, in five years there'd be a shortage of sand."
(Attributed to Milton Friedman)
While my fellow Fed-watchers have had their eyes firmly glued to the repo-market for several weeks now, I've been preoccupied with some other Fed monkey business: its plans to launch a new real-time retail payments service. Having recently testified before the Senate on that topic, I finally have some time to devote to adding my two cents to what others have had to say about the Fed's repo-market tribulations.
A Broken Floor
By the "repo-market imbroglio," I mean, for those of you who haven't been watching the repo-market at all, the Fed's struggles of late to keep short-term private-market interest rates, including both the fed funds rate and rates on private repurchase agreements (or "repos"), from rising above—if not well above—the top of its fed funds rate target range.
The basic problem is that the IOER rate has been steadily losing gravity, meaning its power to keep money market rates orbiting around it, for some time. The following chart illustrates the point with reference to the effective fed funds rate:
The blue line shows how the gap between the fed funds rate and the IOER rate, which was originally negative, has grown since mid-2018, while the red line shows the four five-point "technical adjustments" to which the Fed has resorted in order to keep the funds rate from creeping above the upper limit of its target range. On September 17th, those adjustments proved inadequate, with the effective funds rate briefly bursting through the target upper limit. An even sharper spike in repo rates that day had some borrowers paying 10 percent for secured overnight funds.
In response to these rate spikes, the Fed began injecting fresh reserves into the banking system through its own repo offers. So far, as the next chart shows, banks have borrowed and repaid several hundred billion dollars; and they are likely to keep on borrowing for some time.
What all this means is that, for the time being at least, the Fed's "floor" operating system, which it decided to make permanent only this January, is broken. Under a proper floor system, temporary open-market operations shouldn't be necessary. Instead, banks earn enough interest on their excess reserves to be content to hoard such reserves, and are in turn supplied with so many reserves that they never have to borrow them on the private market, let alone directly from the Fed. Because reserves are no longer scarce in this setup, marginal changes to the stock of reserves no longer influence market interest rates. Instead, the central bank adjusts the stance of monetary policy by changing the interest rate it pays on bank reserves, triggering arbitrage that leads to corresponding changes to money market rates.
Simplicity is supposed to be the hallmark of this arrangement. As New York Fed Vice President and acting System Open Market Account (SOMA) manager Lorie Logan put it before things went sour, compared to arrangements that depend on routine open-market operations, a floor system
is simple and efficient to operate. The interest rate the Fed pays on excess reserves serves as the primary policy implementation tool, with support from a standing facility that offers overnight reverse repos (ON RRPs) at an administered rate. There is no need to forecast specific factors affecting reserves or to conduct discretionary open market operations each day… Market forces keep the federal funds rate in the FOMC’s target range by allowing a wide range of counterparties to price trades against the alternative option of investing with the Federal Reserve.
So much for the theory. In fact, as Stephen Williamson points out in his own excellent post on the repo-market mess, the scale of the Fed's recent repo-market operations far eclipses that of its routine pre-2008 temporary open-market operations. It also overshadows the Fed's unwind, which at its peak had the Fed withdrawing $50 billion in reserves each month.
A Dearth of Reserves
What went wrong? In brief, Fed officials found out that keeping the banking system flush with reserves, as a floor system requires, is harder than it sounds. More precisely, they discovered the hard way that even $1.4 trillion in excess reserves wasn't sufficient to avoid a mad scramble for cash.
That Fed officials were caught off guard by this discovery isn't altogether surprising. After all, before the Fed switched to a floor operating system, banks managed with a grand total of less than $2 billion in excess reserves, or 1/700th of their recent holdings! Yet under that old arrangement, with all its pesky forecasting requirements, the fed funds rate never wandered far from the Fed's target, which was, moreover, a precise value rather than the generous 25-point bullseye the Fed gets to aim at today. Of course, individual banks often found themselves short of reserves in that arrangement. But when they did they could readily make up for those shortages, and do so without paying a premium, by borrowing on the fed funds market.
Fed officials understood, of course, that the new system called for substantially more reserves than the pre-crisis set-up did. They just didn't know precisely what "substantially more" meant. Surely, they reasoned, banks wouldn't need all the reserves provided to them through the Fed's crisis-era QE programs. So, some unwinding seemed appropriate. But just how much was unclear. Hence the Fed's decision to proceed gingerly and, ultimately, to suspend its unwind sooner than expected in response to evidence that reserves were getting scarce again.
That those symptoms of reserve scarcity started to show while there were still almost $2 trillion in reserves outstanding made it natural to suppose that, instead of an aggregate reserve shortage, what was being witnessed was a distributional problem only, reflecting a breakdown of some kind in the market's ability to get reserves where they were needed. "Apparently," Williamson observes,
banks holding reserves were foregoing large profit opportunities to lending in the repo market and fed funds market, and fed funds market lenders were similarly lending unsecured overnight and foregoing large profit opportunities to secured lending in the repo market. … The problem is that overnight markets – particularly in the United States – are gummed up with various frictions.
But it turns out that most of these "frictions" consist of regulations that substantially boosted the demand for excess reserves, particularly among the largest banks. Five of those banks hold about 90 percent of all outstanding excess reserves; and once one allows for all the liquidity regulations to which those banks are subject, the possibility that they can't afford to part with any of them becomes increasingly plausible.
What regulations? Basel III's Liquidity Coverage Ratio (LCR) requirement is the most obvious culprit. Owing to it, large banks, meaning those with over $250 billion in total assets or over $10 billion in on-balance sheet foreign exposure, must hold enough High Quality Liquid Assets (HQLAs) to fund their anticipated cash outflows for 30 days. Under Basel's rules, both U.S. government and agency securities and excess reserves qualify as level 1 HQLAs. But when reserves yield more than Treasuries banks have every reason to prefer them. Thanks to an inverted yield curve for which the Fed itself bears some responsibility, using excess reserves to meet LCR requirements has been a no-brainer since June:
Shadow Liquidity Regulation
But this still doesn't explain why the big banks didn't fund the private repo market as repo rates rose sharply above the Fed's IOER rate. After all, since the repos we're talking about are temporary swaps of reserves for Treasury securities, banks that engage in them don't reduce their HQLAs. Yet they stand to profit when repo rates spike.
But LCR requirements were far from the only liquidity requirements big banks had to meet. "Banks," JPMorgan CEO Jamie Dimon said at the peak of the turmoil, "have a tremendous amount of liquidity, but also have a tremendous amount of restraints on how they use that liquidity.” As Francisco Covas and Bill Nelson explain in a recent BPI post, besides Basel's LCR requirements the very largest U.S. banks are also "subject to non-public liquidity stress tests, non-public liquidity requirements associated with their resolution plans, and non-public ad hoc examiner mandates."
Bill Nelson first drew attention to those ad-hoc examiner mandates almost a year ago, causing yours truly to complain in turn that Fed supervisors were, perhaps unwittingly, placing artificial limits on the Fed's ability to shrink its balance sheet, as it was endeavoring to do at the time. In particular, as Zoltan Pozsar explained in a remarkably detailed and prescient May 2019 essay, to which I can hardly do justice here, "increased supervisory focus on intraday liquidity increased banks' demand for reserves just as the taper was destroying them" (my emphasis). Among other things Pozsar delves into liquidity requirements connected to banks' resolution plans, which grew out of Dodd-Frank's "living will" provisions. Suffice to say that those requirements, besides being much stiffer than ordinary LCR requirements, can only be satisfied by excess reserves. Finally, as if all this weren't enough, some big banks, including J.P. Morgan, are subject to a capital surcharge that depends on the extent of their repo lending, which can prevent them from coming to the rescue of a distressed repo market even when they have excess reserves to spare.
The bottom line is that regulators have managed to raise the biggest banks' liquidity needs enough to compel them to sit on most of the banking system's seemingly huge stock of excess reserves, and to do so even as repo markets present them with an opportunity to earn five times what those reserves are yielding just by lending them out overnight.
And Shadow Reserve Drains
While regulatory requirements, assisted by an inverted yield curve, have been ramping-up banks' demand for excess reserves, other factors have caused the supply of excess reserves to shrink considerably more than Fed officials bargained for. As the next FRED chart shows, between October 2017, when the Fed began its unwind, and September 17th, when the repo market started to go haywire, the Fed had shed about $600 billion in assets. Excess reserves, on the other hand, fell by almost $780 billion.
Two causes, also shown in the chart, account for the overall discrepancy. The first and better known has been an accumulation of Treasury General Account (TGA) balances. The less well known contributor to the reserve drain has been an increase, practically all of it coming from Japanese banks, in the size of the New York Fed's "foreign repo pool." The basic story is the same for each: unlike most of us, but like depository institutions, both foreign central banks and the U.S. Treasury can have accounts with the Federal Reserve. But when they accumulate balances in those accounts, they remove a corresponding amount of reserves from the banking system. (For more details, see this excellent just-released essay by BPI's Bill Nelson.)
A final factor that helped set the stage for last month's reserve shortage was the Treasury's need to fund the Trump administration's $1 trillion fiscal 2019 deficit. That meant having primary dealers underwrite a like amount of newly-issued Treasuries. Because the Primary Dealers Act of 1988 obliges those dealers to get a "reasonable price" for Treasuries they underwrite, they typically submit bids at a range of prices on their own accounts. They can therefore find themselves acquiring substantial security inventories to dispose of gradually following an initial offering.
But dealers have only so much money on hand, in so-called "clearing" accounts they keep at the Bank of New York Mellon, with which to finance their Treasury inventories. Because they only intend to carry such inventories for short periods, they naturally turn to the repo market for additional funding or to replenish their depleted clearing accounts. Thanks to trade war jitters and also to the inverted state of the yield curve, the demand for Treasury securities has been far from consistently brisk. Dealers' inventories and their demand for short-term funds by which to finance them have had to grow accordingly. On the very day of the repo spike, the Treasury placed about $50 billion in securities on the market, causing primary dealers in turn to purchase and then repo a substantial portion of those securities.
So there you have it: a host of developments adding to banks' demand for excess reserves, while others gradually chipped away at the stock of such reserves. Add a spike in primary dealers' demand for short-term funding, a coinciding round of tax payments that transferred as many reserves to the TGA, and binding intraday liquidity requirements at the banks holding a large share of total system excess reserves, and you have the makings of last month's perfect repo-market storm.
Solutions: Shallow and Deep
On second thought, "perfect storm" is not the right metaphor, since such a storm is a freak event that's unlikely to be repeated any time soon. What happened in the repo market is, in contrast, quite likely to repeat itself, and might be repeating itself as I write this were the Fed not repeatedly offering fresh reserves to the money market. Perhaps "climate change" is more like it.
In any case, so far as Fed officials are concerned, adding still more excess reserves to a U.S. banking system that once seemed to be drowning in them is the only way to fix things. In place of ongoing ad-hoc repo operations, the Fed plans to establish the Standing Repo Facility (SRF) it has been contemplating for a while now. It will also start expanding its balance sheet once again. Word on the street has it that the SRF will be up and running sometime early next year. As for renewed Fed balance sheet growth, Chairman Powell has already suggested that it's coming, and odds are the FOMC will announce new purchases at this month's meeting. Joe Gagnon expects the Fed to purchase about $250 billion in securities over the coming six months; but I won't be surprised if it buys even more.*
But are these really the only solutions? The answer is that they are so only assuming that the Fed must stick to a floor system of monetary control. But a floor system is only one of many alternative operating systems the Fed could rely upon; and one need only recognize this to see, not only that there are other ways for the Fed to regain control of interest rates, but that some of those other ways make a lot more sense than the Fed's preferred solution.
Two options in particular come to mind. One is to leave the banking system with a systematic liquidity deficit that makes it systematically dependent upon the proposed Standing Repo Facility. Implementing monetary policy would then be a simple matter of adjusting the overnight rate implicit in the terms of the SRF's full-allotment (that is, unlimited) collateral purchases and sales. According to Ulrich Bindseil (p. 51), this sort of arrangement, which was long employed by the Reichsbank, and is similar to the way the Fed operated when it was first established, is actually "the simplest way to steer short term interest rates." Besides making further outright security purchases unnecessary, the set-up would dispense with the need for a rate target "range" and the Overnight Reverse Repo Repurchase arrangement used to define and policy that range's lower limit. Under it, excess reserves could still bear interest, though at a rate set below the SRF rate. Finally, to deal with financial crises, the SRF could be supplemented with a broader facility that would offer liquidity to banks in exchange for non-Treasury collateral at rates set using "product-mix" auctions designed to price risk and allocate emergency funds efficiently.
Alternatively, the Fed could replace its floor system with a corridor system. That means keeping the SRF rate above the Fed's target, while setting the interest rate on reserves below it, and resorting to daily open-market operations to limit target rate variations within those boundaries. This solution is more challenging, owing to the very large open-market operations that might be needed to accommodate "autonomous" changes in the stock of reserves connected to changes in either the TGA balance or the size of the foreign repo pool. But the Fed could reduce that difficulty by placing caps on both the foreign repo program and the TGA, forcing the Treasury, through the latter change, to once again take advantage of TT&L (Treasury Tax and Loan) program for its originally intended purpose, namely, "reducing uncertainty about the supply of reserves in the banking system and simplifying the Fed's implementation of monetary policy."
*In fact, by the time the COVID-19 crisis struck in mid-March 2020, the Fed had purchased over $500 billion in securities, of which only $40 billion or so served to accommodate growth in the demand for currency. [Note added 12-5-2020.]