You’ve always had the power to go back to Kansas.
–the Good Witch of the East, to Dorothy, in The Wizard of Oz
(This is the conclusion of a two-part essay. For Part 1 click here.)
Equipped with some historical background, we can now consider ways in which the Fed might get the Treasury and Foreign Official Institutions (FOIs) to revive their pre-crisis practice of parking surplus dollars somewhere other than at the Fed. In fact, most of the necessary means are already at hand. Fed officials only need to recognize and take advantage of them. They need, as it were, to click the heels of their ruby slippers together, and repeat three times that, so far as foreign dollar pools and big Treasury balances are concerned, "There's no place like somewhere else."
Just how can the Fed get the Treasury and FOIs to take their business elsewhere? Broadly speaking, their options are: capping their Fed deposits, or making those deposits less remunerative, or taking steps to restore the attractiveness of alternative investments that don't gobble-up bank reserves.
Draining the Foreign Repo Pool
In principle, limiting FOIs' use of the Foreign Repo Pool (FRP) is very easy, because the Fed is free to set any rules it likes for that facility. Specifically, as Matthew Klein observes in his Barron's piece, "The New York Fed has the right to 'manage the overall size of the foreign repo pool' to 'maintain orderly market or reserve management conditions'." The Fed could, therefore, restore the limits it once placed on the pool. It could even scrap the facility altogether, since it operates the FRP as a courtesy to FOIs, and not because the Federal Reserve Act obliges it to do so.
Why shouldn't the Fed take one of these steps? According to Zoltan Pozsar, it may be reluctant to cap the FRP because other central banks offer similar facilities without capping them. Consequently, he says, the "optics" of the Fed either limiting or discouraging the use of the FRP would be bad. Caps would deprive FOIs of an unlimited and relatively remunerative safe haven for their dollars when private banks come under stress.
These arguments aren't all that convincing. Optics? Those of the Fed's recent repo-market tribulations, of its deciding to renew its asset purchases, and of the huge balance sheet it must possess by continuing along its present course, are themselves far from appealing. A "safe haven" for FOI dollars? Should the Fed make it easy, during times of financial stress, for FOIs to yank their money, not just from particular U.S. banks, but from the U.S. banking system as a whole? Isn't it supposed to protect U.S. commercial banks, not Foreign Official Entities?
Still, the Fed has reasons to avoid ruffling foreign feathers. So, it's important for it to be able to take the sting, or apparent inequity, out of any plan for limiting FOIs' use of the FRP. For starters, instead of capping (let alone abolishing) the FRP, it could discourage its use by the less Draconian step of adjusting the FRP rate formula to make the facility less attractive. At the very least, it could quit making its own rate better than corresponding private market lending rates, as it appears to have been doing lately by basing that rate on the SOFR (Secured Overnight Financing Rate). Since the Fed is a uniquely risk-free counterparty, it arguably should make its FRP rate not higher but lower than corresponding private market rates. For example, it might set a rate equal to SOFR minus 10 basis points.
But even if they must pay a (reasonable) price to do so, FOIs aware of the risk of another financial crisis may nonetheless prefer to keep substantial dollar reserves at the Fed than to either lend them on the private market or to invest them in safe but less than perfectly liquid dollar assets. Fortunately, there's a way for the Fed to overcome this final incentive for FOIs to employ the repo pool, to which I'll come after I discuss ways to limit the TGA balance.
Taming the TGA
While the Fed alone can take steps to limit use of the FRP, it will need the Treasury's help to discourage use of the TGA, both because the Treasury has the right to put as much money in the TGA as it likes, and because the Treasury sets the interest return on Treasury Tax and Loan (TT&L) notes, one of its main alternative investments.
The good news, if past experience is any guide, is that the Treasury is likely to cooperate with the Fed, provided that doing so isn't too costly. The bad news is that the Treasury itself may be powerless to revive the TT&L program, making that avenue for reducing the use of the TGA a dead-end. Under present rules, were the Treasury to open a TT&L note account, it would pay the Treasury just 1.58 percent–the current (November 1) fed funds rate minus 25 basis points. But the Treasury would indirectly "pay" its depository 2.40 percent–the current IOER rate–for the privilege of keeping its money there. In other words, the transfer would leave the Treasury with a net return of minus 82 basis points!
Having the Treasury revise its TT&L rate formula so that TT&L notes themselves pay the IOER rate might seem an obvious solution. But things aren't so easy. Depository institutions are likely to oppose the plan, as they opposed a similar effort in 1999. And were the Treasury to proceed with the change anyway, some of them might withdraw from the program altogether, leaving it even more concentrated and capacity-constrained than it was on the eve of the financial crises. Consequently, rather than try and revive the TT&L program once again, the Treasury would have every reason to let that sleeping dog lie.
But there's another solution, and banks don't have to agree with it. Better still, it has already been successfully tested, thereby winning the Treasury's approval. Finally, it couldn't be easier: it consists of simply allowing the Treasury to enter directly into repo agreements with the same primary dealers the Fed itself relies upon for its own domestic repo operations. Because they "cut out the middleman" by dispensing with commercial depositories, Treasury repos would allow it to earn an overnight rate roughly equal to, if not higher than, the IOER rate.
The Treasury first started negotiating overnight repos with depository institutions in March 2006, in a one year pilot the GAO encouraged it to try. As the chart below (from a subsequent, September 2007 GAO report) shows, that program was extremely limited in scope and scale, with the Treasury's average of $4 billion repos a day comprising but a miniscule share of what was then a $1.8 trillion per day repo-market.
Yet the program was successful, yielding rates considerably above TT&L rates. The rate was in fact very close to the Fed's own repo rates, and therefore better even than the term investment option (TIO) rate. In its September 2007 report evaluating the program, the GAO found that, had the Treasury been able to repo all save a $2 billion of its surplus cash, it might have earned another $12.6 million in fiscal '06–no small amount then–while relaxing TT&L and TIO program concentration and capacity constraints. Unsurprisingly, the GAO went on to recommend expanding the Treasury repo program further, to include the full complement of primary dealers, and also to allow the Treasury to repo a broader set of Treasury securities. Finally, the GAO urged the Treasury to consider switching to a triparty repo arrangement which would be both less costly and safer.
In a comment included in the report, the Treasury's Fiscal Assistant Secretary, Kenneth Carlson, agreed with all of these suggestions. Unfortunately, the Treasury never got around to taking them. Less than a year after the GAO issued its report, the Great Financial Crisis was in full swing. Soon afterward, the Fed began paying interest on bank reserves at rates that for some time exceeded not only repo rates but returns on many Treasury securities. The momentum behind the repo program was lost.
Today, with repo rates well above the IOER rate, Treasury repos could once again offer the Treasury an attractive alternative to the TGA. The Fed has only to encourage Congress to take whatever steps are needed to revive and expand the program, thereby presenting the Treasury with an attractive means for reducing its TGA balance to a modest and stable level.
A Standing Repo Facility Would Help
Only one hitch remains. Even a triparty repo program entails some risk. So, the Treasury might still want to keep a large balance in its TGA account to assure itself access to cash even in the event of another major financial crisis.
But there's a solution to this problem as well–and it's the same one that can help overcome FOIs' reluctance to hold Treasury securities or rely on the private repo-market. That solution consists of a modest refinement to an idea the Fed has already been considering, namely, creating a Standing Repo Facility, or SRF, for short.
As I explained here some months ago, the main purpose of the SRF is to reduce depository institutions demand for excess reserves, by allowing them to convert Treasury securities and other "High Quality Liquid Assets" (as defined by Basel III) into reserves at a moment's notice, even when private repo-markets are under stress. To avoid undermining the private repo-market, the facility would charge a spread above market repo rates. According to Bill Nelson, because the facility would, at the very least, have to admit all depository institutions as counterparties, "the spread over the target fed funds rate range would have to be relatively wide, probably at least 25 basis points." Granting large non-bank repo borrowers access to the facility would then tend to keep spikes in market repo rates from exceeding that 25-point spread.
The "modest refinement" I have in mind for the SRF is one that would allow it not just to reduce banks' demand for reserves, but to help increase the reserve supply. It consists, simply, of including the U.S. Treasury and FOIs, or at least some key foreign central banks, on the list of SRF-eligible counterparties. Doing that would assure them that, so long as they have ample amounts of Treasury securities or other SRF collateral on hand, they need never fear being short of cash, or paying more than a modest premium to acquire it. Because the yield on short-term SRF-eligible dollar-denominated collateral would be close to what the FRP offers, this arrangement would approximate the risk-return profile of that facility without sucking reserves from the banking system.
If allowing FOIs, and important foreign central banks especially, to borrow at the SRF, using good Treasury collateral, serves to limit use of the FRP, the decision should be a no-brainer. After all, as Zoltan Poszar has observed, the alternative of "adding reserves through a standing repo facility or asset purchases while having an uncapped foreign RRP facility sterilizing reserves would be odd." "Odd" is saying the least. "Futile" is more like it!
Some may object to the idea of allowing the Fed to lend to FOIs, even though the lending involves no risk to the Fed, and would be done only at a spread above the Fed's rate target upper limit, which is to say, only during periods of extreme credit stringency. But they should keep in mind that the Fed already has the authority to lend to foreign central banks through the "dollar liquidity swap lines" it authorized in May 2010 and made permanent in 2013. The chief difference is that the swap lines allow foreign central banks to borrow dollars against their own currency, while the FRP would allow them to borrow dollars against U.S. Treasury securities. To some extent, the arrangements would be substitutes. But the proposed facility alone would assist the Fed in minimizing the use of the Foreign Repo Pool.
So Would Reinstating the Treasury's Security Draw Authority
The same might be said of not taking advantage of the SRF to assure the Treasury that it doesn't need a big-fat TGA balance. Only in this case the "obvious" solution isn't so obvious, because direct Fed lending to the Treasury happens to be illegal. But here also, experience points to a solution, in the shape of the "security draw authority" Congress granted the Treasury for almost four decades starting in 1942. That authority permitted the Treasury to repo securities with the Fed to meet its short-term cash requirements until it was allowed to expire in 1981.
As Kenneth Garbade informs us, when the security draw authority's postwar continuation was being debated, then-Fed chair Marriner Eccles explained that having that exception to direct Fed lending to the Treasury helped to prevent Treasury activity from reducing the stock of bank reserves around tax collection dates–precisely the sort of reduction that caused trouble this September. "If," Eccles said,
the Treasury could not borrow from the Federal Reserve banks by what is, in effect, an overdraft at these tax payment periods, and in this way avoid withdrawals from its war loan accounts to pay off maturing obligations, money conditions would unduly tighten and tend to [destabilize] the money market and the Government securities market.
Eccles added that, were it not for its security draw authority, "the Treasury would feel obliged to carry much larger cash balances."
Despite its advantages, the security draw authority was never made permanent. Instead, having initially granted it for just two years, the Treasury proceeded to renew it again and again. Finally, in 1981, the authority was allowed to lapse, mainly because new Treasury cash-management options, including the introduction of TT&L note accounts and of special Treasury "Cash Management" bills, made it seem redundant. That the Treasury had taken advantage of the special draw authority in 1977 to borrow from the Fed in anticipation of a debt ceiling constraint also helped to seal the privilege's fate, prompting then representative Ron Paul of Texas to complain, with reason, that it had "become a device that is used by the Treasury to ignore Congress and the law." Although, despite complaints like Paul's, the authority was extended for once more in 1979, that final renewal limited its use to "unusual and exigent circumstances… pursuant to an affirmative vote of not less than five members" of the Federal Reserve Board.
Fear of such "unusual and exigent circumstances" is, of course, one of the reasons why the Treasury might be reluctant to forego a hefty TGA balance even were it able to take advantage of a revived and expanded Treasury repo program to lend most of its surplus funds to private borrowers. Its concern is that, during a major crisis, it might only be able to get cash immediately from the one U.S. lender that can't fail. Reinstating the Treasury's special draw authority, or developing a new version of it specifically meant to allow the Treasury to be among the Fed's SRF counterparties, at least under certain circumstances, would overcome that reluctance. Just as the previous special draw authority imposed a $5 billion borrowing limit, the new one might impose a limit of $100-150 billion, depending on the Treasuries preferred new TGA balance target, the goal being to still allow the Treasury certain access to cash sufficient to cover one week's expenditures.
Needed Changes: A Recapitulation
Let me now gather together the various steps I'm proposing for substantially reducing use of the TGA and the RTP. These are as follows:
- Congress should (1) revive the Treasury's direct repo borrowing program, expanded (as per the GAO's September 2007 recommendations) to allow all primary dealers to take part and also to allow the Treasury to offer any of its securities as collateral, and (2) restore the Treasury's "security draw authority," or otherwise have Congress authorize it to borrow directly from the Federal Reserve;
- The Fed should (1) set its FRP rate at 10 basis points below the SOFR rate, (2) establish the Standing Repo Facility that it has already been considering, and (3) allow Foreign Official Entities, and foreign central banks especially, as well as the U.S. Treasury (as authorized by Congress) access to the Standing Repo Facility.
Floor versus Corridor (Again)
Besides allowing the Fed to maintain its floor operating system with a considerably smaller balance sheet, the reforms I've recommended would also make it possible for the Fed to slim down considerably further by allowing it to switch from a floor system to a corridor system.
When the Fed first switched to a floor operating system in October 2008, it had no intention of doing so permanently; and as recently as 2018, the Fed claimed that it was still entertaining the possibility of switching to a corridor system. Nevertheless, this January, Jay Powell announced that the Fed had decided to make the Fed's floor system, with its dependence upon an abundant supply of excess bank reserves, permanent.
Although a number of factors informed the Fed's decision, one of the more important was the increased importance and volatility of the Fed's non-reserve liabilities. As Lorie Logan explained in May 2017, that change meant that, to operate a corridor system, the Fed would have to routinely engage in open market operations "that are larger, more variable, or even very different from those used before the crisis."
This April Logan pursued this same theme, specifically pointing to the behavior of the TGA balance as a factor justifying the Fed's decision to keep a floor system. "There have been times," she said, "when the TGA has increased by around $100 billion over the course of a week and there has been a similarly large decline in reserves." By maintaining its floor system, the Fed could avoid having to compensate for such fluctuations. Instead "a buffer of reserves above the banking system’s demand for reserves" would simply absorb them. Logan went on to declare "that the floor system has proven to be highly effective at controlling the effective federal funds rate and other money market rates, is resilient to significant shifts in market structure, and is efficient to operate."
Although it's now painfully obvious that Logan's assessment of the Fed's floor system was premature, so long as the TGA balance and FRP remain both large and volatile, Logan's arguments for retaining that system continue to have weight. But I hope I've succeeded in showing that, far from being "given" to the Fed, the size and behavior of the TGA balance and FRP are functions of government policy, and especially of the Fed's own policies. And Fed officials surely know it. Were they to follow the suggestions offered here, they could rein-in those pesky reserve-absorbing liabilities once and for all. And then the Fed would no longer have any good reason for not switching to a corridor system, and to the truly lean balance sheet that would then be all it needed "to implement monetary policy efficiently and effectively."
I've tried to show how, with the Treasury and Congress's help, the Fed might reduce the demand for its non-reserve liabilities, and thereby add to the stock of bank reserves, while also making a large excess reserve cushion unnecessary. Perhaps I've overlooked some details that render my plan incomplete. Or perhaps there's an even simpler plan that would work at least as well. No matter: my goal is only to convince readers that the objectives I have in mind are both achievable and worth pursuing. If anyone can come up with some better ways to achieve the same ends, believe me, I'll be the first to embrace them.