Last month, in what will presumably be one of his last public speeches before he retires this June, New York Fed President William Dudley defended the Fed's current "floor" system of monetary control against the alternative of a corridor arrangement. Because Dudley's is one of the more complete arguments for retaining the current system to be publicly aired by a high-ranking Fed official, it's worth quoting it at length:
In my view, the case for retaining the current floor system is very compelling for a number of reasons. First, it is operationally much less complex than a corridor system. In the current regime, the setting of IOER is largely sufficient to maintain the federal funds rate within the FOMC's target range, as we have seen over the past few years. In contrast, a corridor system requires forecasting the many exogenous factors that affect the amount of bank reserves outstanding, and then engaging in open market operations on a near daily basis to keep reserves at a level consistent with the FOMC's target range. This task would likely be more difficult now because of greater fluctuation in these exogenous factors relative to when the corridor regime was last in place.
Second, a corridor system constrains the Federal Reserve's ability to provide the types of lender-of-last-resort backstops that can help support financial stability. Prior to gaining the authority to pay interest on reserves in the fall of 2008, Federal Reserve officials — myself included — were faced with a question: if the take-down on a proposed liquidity facility were large, how would we be able to drain the added reserves on a timely basis to maintain control of the federal funds rate? This was not just a theoretical issue during the financial crisis. The initial auctions of the Term Auction Facility, or TAF, were kept relatively small, in part, because of concerns that larger programs would make it more difficult for the manager of the System Open Market Account (me, in that particular case) to drain sufficient reserves to offset the TAF reserve additions. Similarly, the Term Securities Lending Facility (TSLF) — which involved the swap of Treasury collateral from the Federal Reserve against lower-quality collateral held by primary dealers — was introduced in large part because collateral swaps, unlike cash loans, did not affect the amount of reserves in the banking system. My overall point is that broad-based, open-ended lender-of-last-resort facilities are more difficult to accommodate in a corridor system because of the need to drain any reserve additions to keep the federal funds rate close to the FOMC's target.
I see this as an important shortcoming of the pre-crisis corridor regime that does not get sufficient attention. Having the ability to introduce broad and credible lender-of-last-resort backstops — well secured by pledged collateral — can be critical when confidence falters and financial stability is at risk. Private sector participants are more likely to continue to engage with their counterparties — and to borrow and lend — when they know a central bank backstop will be available if economic and financial conditions deteriorate. A corridor regime could cause the Federal Reserve to delay or avoid providing such backstops, which could increase financial stability risks by causing lenders and borrowers to disengage with one another earlier.
Dudley added that his opinions were shared by most other members of the FOMC.
For the sake of clarity, let's label Dudley's two arguments for a floor system the "Unrestrained Last-Resort Lending" argument and the "Simplified Monetary Control" argument.
The Floor System and Last-Resort Lending
How compelling are those arguments? Let's start with the "Unrestrained Last Resort Lending" argument. According to Dudley, the Fed's floor system allows it to avoid the "constraints" a corridor system would place on its ability to engage in last-resort lending. Specifically, unlike a corridor system, the Fed's floor system allows it to expand its balance sheet, whether through last resort lending or by means of open-market security purchases, to any desirable extent, without altering its monetary policy stance. In particular, by switching to a floor system after October 6, 2008, the Fed was able to prevent its emergency lending from contributing to a general loosening of credit without having to continue "sterilizing" its loans by disposing of securities from an already depleted Treasuries portfolio.
What Dudley neglects to point out, which anyone conversant with the history of the Great Recession knows, is that the last thing the U.S. economy needed in the fall of 2008 was a new device for keeping a tight reign on the supply of credit! Rather than helping the Fed to maintain a desirable monetary policy stance, its new floor system contributed to what proved, in retrospect, to be a severe over-tightening of monetary policy. Consequently, on that occasion at least, the U.S. economy would have been better off had the Fed stuck to its old, corridor-type operating system.
Might the advantage Dudley claims for a floor system come in handy on some future occasion? It's highly unlikely, for two reasons. The first is that occasions on which a central bank is warranted in making last-resort loans on a large scale are ipso facto ones on which there's an aggregate liquidity shortage. The second is that, to address an aggregate liquidity shortage, a central bank has not only to expand its total lending, but to see to it that its lending adds to the general flow of credit, and not just to the piling-up of excess reserve balances. Put another way, in economies with otherwise well-organized private credit markets, solvent private financial institutions will have recourse to central banks' emergency credit facilities only when those private markets have themselves begun to run dry. The central bank in that case has a positive duty to see to it that those markets are replenished. It was this duty that the Fed so conspicuously failed to perform in late 2008.
Indeed, the overarching purpose of liberal last-resort lending has always been, according to thoughtful authorities going back to Bagehot (and probably beyond), not to engage in what Marvin Goodfriend calls "credit policy," meaning the re-shuffling of existing credit among particular firms and markets, but to combat overall liquidity shortages. In taking the steps it took in the fall of 2008, first by sterilizing its emergency loans and then by encouraging banks to hoard the proceeds, the Fed seemed to have forgotten this. As Thomas Humphrey explains,
To classical writers, injections of base money were the essence of LLR operations: The LLR existed solely to expand the base temporarily in amounts sufficient to offset panic-induced falls in both the money multiplier and the circulation velocity of money thus preserving monetary supply-demand equilibrium and price-level stability in the face of bank runs.
In short, what Dudley sees as an advantage of the floor system in dealing with financial crises is no advantage at all. Although a floor system would make it easier for the Fed to engage in pure "credit policy," such policy alone doesn't usually suffice to avoid a credit crunch. On the other hand, a floor system undermines the capacity of either last-resort lending or open-market purchases to revive flagging credit markets. Indeed, because it encourages banks to lend to the Fed instead of funding private money markets, a floor system actually contributes to the drying-up of those markets.
The "Simplicity" of a Floor System
Now let's turn to the "Simplifies Monetary Control" argument. Dudley's basic claim here — that a floor system "is operationally much less complex than a corridor system" — is correct so far as it goes. Under its floor system, the Fed can stay within a very narrow Fed funds target range without having to resort to any open-market operations save the reverse repos it engages in for the sake of fixing the range's lower bound.
But this gain in simplicity, far from being free, comes at the cost of (1) a substantial decline in interbank borrowing and lending, particularly on the (unsecured) federal funds market; and (2) a related increase in the Fed's absolute and relative involvement in financial intermediation. As Ulrich Bindseil and Jukliusz Jablecki have compellingly argued, to arrive at a sound decision regarding the best operating framework for a central bank, one has to take all of the relevant trade-offs — that is, all of the costs and benefits of alternative arrangements — into account. One must recognize, furthermore, that instead of a simple choice between a floor system on the one hand and a corridor system on the other, central banks face a much larger set of options. These include systems with corridors of varying widths, where a narrower corridor makes conducting monetary policy less "complicated" than a broader one, though still more complicated than a floor system, while a broad corridor may be more conducive to active (overnight) interbank markets and maintaining a "lean" central bank balance sheet. Among these many alternatives, a floor system represents a corner solution, which ought to be reason enough to doubt that it's optimal.
But does it really matter whether there's an active interbank lending market or not? As Bindseil explains elsewhere, it does, for several reasons:
First, [an active interbank market] allows measuring easily the short term interest rate and thereby the degree of achievement of the operational target of monetary policy… . Second, it has been argued that interbank money markets are useful because they incentivize banks to cross‐monitor each other, which strengthens the market mechanism (e.g. Blasques et al, 2016). Of course, incentives for monitoring banks should also be to some extent effective through other bank funding markets (e.g. short term paper, bond and equity markets). But it seems plausible that interbank markets, as a peer‐to‐peer market, have something particular in this respect. If monitoring is effective and long term relationships (and sufficient limits) are in place, this can also support financial stability as temporary capital market funding stress or deposit outflows affecting a single banks that are not due to fundamentals could be compensated by interbank money markets. Finally, an active interbank money market provides evidence of netting of the liquidity needs of the banking system before recourse is taken to central bank credit.
The study by Blasques et al, which looks at the Dutch interbank market, finds that encouraging banks to take part in that market results in more intense peer monitoring, strengthened relationships among the lending banks, reduced bank-to-bank credit risk, and an increase in the overall market liquidity associated with a given level of reserves. It's one of several such studies inspired by the pioneering work of Jean-Charles Rochet and Jean Tirole (1996), who first drew attention to the advantages of interbank peer monitoring, while warning that, insofar as it discourages such monitoring, "government intervention…destroys the very benefit of a decentralizing [banking] system."
Another recent study, by Marie Hoerva and Cyrol Monnet (2016) asks, "If liquidity provision is centralized during crisis times, with a central bank acting as the intermediary, why isn't it centralized also in normal times? What is the role of decentralized money markets? What is the function of unsecured and secured markets, respectively?" These are excellent questions, and one only wishes that all regulators would ask them. Hoerva and Monnet's answer is that money markets, and unsecured interbank markets especially,
can provide market discipline, which we define as an ex ante provision of incentives to banks to conduct business in a safe and sound manner. The bilateral interaction between a borrower and its lender in the money market ensures that incentives to take on excessive risk are tamed.
The authors find, furthermore, that
In the absence of aggregate liquidity shocks, a central bank subject to the same informational friction as lenders in the money market cannot improve on the money market allocation. In fact, we show that unsecured financing by the central bank is harmful since it hurts market discipline. When there is an aggregate liquidity shortage, however, we show that secured lending by the central bank is beneficial. In this case, some borrowers with valuable collateral who do not meet lenders in the secured money market can get financing from the central bank. This prevents inefficient liquidation of projects, while preserving market discipline in the unsecured market.
All of which sounds to me a lot like what I just said about the "classical" purpose of last-resort lending.
Finally, there's Craig Furfine's (2001) study, "Banks as Monitors of Other Banks," where he looks specifically at lending on the federal funds market and finds
that the interest rate charged on federal funds transactions reflects, in part, the credit risk of the borrowing institution. In particular, borrowing banks with higher profitability, higher capital ratios, and fewer problem loans pay lower interest rates on federal funds loans. Thus, the empirical evidence suggests that banks identify risk in their peers and effectively monitor other banks.
It was mainly owing to their desire to reinvigorate their economies' unsecured interbank lending markets that all save three of the monetary authorities that adopted floor systems either before or during the financial crisis have since abandoned them. Although most have reverted to corridor systems, several have instead opted for "tiered" systems in which a fixed quantity of reserves only earns the going policy rate, which reserves beyond that amount earning a lower, and in some cases a negative, rate. If Fed officials have some reasons for thinking it O.K. to ignore the considerations that caused so many other central banks to resist keeping their floor systems in place, I hope they will eventually share them with us.
Apart from discouraging unsecured interbank lending, a floor system can also allow central bank intermediation to crowd-out other sorts of private intermediation, including bank lending to businesses and consumers. The extent of such crowding-out will, of course, depend on the extent of banks' excess reserve holdings. In principle, one can have a floor system where the crowding-out effect isn't all that large. But that's far from the sort of floor system Dudley has in mind: instead, he'd rather the Fed's balance sheet got no lower than $2.9 trillion, or roughly three times its pre-floor-system level, and would prefer to have it stay even larger. Such a big balance sheet will keep life simple for the staff at the New York Fed's trading desk, all right. But don't expect it to do the U.S. economy any good.