Wednesday, February 3, 2016

Does the Fed lack the technical means to dive into negative rate waters?

The Federal Reserve may be in a bit of a bind.

With the Bank of Japan reducing rates to -0.1%, many commentators are calling on the Fed to reverse its policy of rate normalization and follow Japan into negative territory.  The problem is this. Thanks to the rules laid out in the Federal Reserve Act, the Fed may lack the technical means to dive into negative rate waters. Let me restate this in different terms. If the Federal Reserve were to reduce the rate at which it pays interest on reserves (IOR) to -0.25% or so, the overnight rate may not follow very far. Monetary policy is useless, or at least less effective than it would otherwise be.

Not only would monetary policy lose some of its potency when IOR falls below 0%, but an unapproved fiscal transfer from the Fed to another set of government institutions could occur. This is because negative IOR has the potential to provide a large subsidy to a narrow range of governments sponsored-entities that are allowed to keep 0%-yielding deposits at the Fed. At deeply negative rates, this subsidy could get very large, turning the Fed's substantial profits into large losses. The result, a steady fall in its share capital, might undermine Fed independence and the credibility of monetary policy.**

Why these odd effects? As Nick Rowe says in this delightful post, in a world with negative interest rates everything old is new again, only it's a mirror image. And in a world of negative U.S. rates, the mirror image of the Fed's leaky floor is a sticky ceiling. And just like the leaky floor (more on this later) dragged the overnight rate down to 0% when IOR was positive, the sticky ceiling effectively pulls the overnight rates back up to 0% when IOR is negative. To see why this would happen, we need to take a quick tour of the legal and technical history behind IOR.

Until October 2008 the Fed was legally prohibited from paying interest to banks. Any bank manager who left reserves on deposit at the Fed earned 0%. This changed with the passage of the  Financial Services Regulatory Relief Act of 2006 which bolted Section 19(b)(12) onto the Federal Reserve Act allowing banks to "receive earnings to be paid by the Federal Reserve Bank." As a direct result of 19(b)(12), the Fed has been paying interest of 0.25% for a number of years, a rate that was recently increased to 0.5%.

Not only did Section 19(b)(12) provide the Fed with the ability to pay interest on reserves, it also gives it the technical means to pay negative rates on reserves. Now there is some controversy whether the wording in Section 19(b)(12) authorizes a negative rate; for instance, it mentions the paying of earnings to banks, not the payment of negative earnings:
...and here is Stephen Williamson on the issue.

Ex-Fed official Narayana Kocherlakota points out that various members of the Fed Board of Governors, including William Dudley and Stanley Fischer, have already hinted at negative rates as a potential tool, the implication being that the Fed has high confidence in a certain interpretation of the Act that would legally justify the move, otherwise they would not have spoken. Don't forget that Fed general counsel Scott Alvarez will have to sign off on the question of legality. Alvarez is the one who allowed a truck to be driven through Section 13.3 during the credit crisis, legalizing the bailout of Bear Stearns and AIG via Maiden Lane.

Let's assume that there is no legal controversy and the Fed sets negative IOR of -0.25% tomorrow. That's where the Fed's real problems start. Because as I said at the outset, the overnight rate won't follow IOR down in lock-step, it may even stick to the 0% ceiling. On top of this is the aforementioned subsidy provided to those institutions that have access to interest-free deposits at the Fed, namely the government-sponsored entities Freddie Mac, Fannie Mae, and the Federal Home Loan Banks (I'll describe this subsidy later).

As I said at the outset, the sticky ceiling problem is the opposite of the well known leaky floor problem. Over the last few years the various U.S. overnight rates have traded below IOR. These leaks are odd since IOR is supposed to set a lower bound to the overnight rate. After all, why invest your funds at 0.05% overnight when you can get 0.25% at the Fed? We can blame the leakage on our three government-sponsored entities (GSEs): Freddie, Fannie, and the Federal Home Loan Banks. These GSEs are allowed to keep accounts at the Fed but are prohibited from collecting interest. Rather than earn nothing, the GSEs have been lending their reserves overnight to banks who have access to IOR. These banks have been engaging in arbitrage whereby they borrow GSE funds at 0.05%- 0.15% and invest them at 0.25%, a nice gig if you can get it.

From whence this prohibition on interest? 19(b)(12), the same bit of legalese that authorizes IOR, emphasizes that only "depository institutions" can receive interest, and since Freddie, Fannie, and the Federal Home Loan Banks are not depositories, they do not qualify (this article speculates why).

Which means that if the Fed is going to stretch Section 19 authority for payment of interest  to allow for the payment of negative interest rates, then the GSE prohibition on receiving positive interest means that they will likewise be exempt from paying negative interest.

The ultimate effect of all this is that if and when the Fed decides to brings IOR to -0.25%%, the overnight rate is going to do a flip—rather than trading below IOR it will start to trade above it. Why the inversion? In positive rate land the IOR exemption meant that it really sucked to be the GSEs. But in negative rate land it's great to be a GSE. When all other banks must endure a -0.25% penalty, an exemption from IOR becomes an asset, not a liability. Banks, sensing a way to improve their returns, will compete to purchase 0% shelter from the GSEs. Say they offer to lend funds to the GSEs in the overnight market at  -0.17%, an 8 basis point improvement to IOR of -0.25%. GSEs would be hard-pressed to not accept this gift. After all, they'd be borrowing at -0.17% in order to lend to the Fed at 0%, which means 17 basis points in risk free profit. Taken to the extreme, a profit-maximizing Fannie, Freddie, and the Federal Home Loan Banks will compete to borrow the entire $3.3 trillion in reserves held at the Fed, driving overnight rates back up towards 0%. No matter how deep the Fed brings IOR into negative territory, the overnight rate stays stuck at a shallow level thanks to the GSEs.

You may recognize that the sticky ceiling problem is just another (perhaps weaker) version of the good old zero-lower bound problem, but in the role of 0% cash as antagonist, 0% deposits held at Fannie, Freddie, and the Federal Home Loan Banks have been substituted.

Now it could be argued that GSE arbitrage will force the overnight rate within a hair of negative IOR. For instance, if  IOR is at -0.25%  the overnight rate might trade at -0.24%. If so, the problem is less about monetary policy impotence and more about the massive subsidy being granted to the GSEs. These institutions have a large incentive to hoover all the reserves held by non-GSEs in order to earn risk free returns. Rather than earning 0.25% on reserves held by banks, the Fed will get 0%. And since the Fed's profits--what economists call seigniorage--typically flows to the taxpayer via the Treasury, taxpayers would be subsidizing Fannie, Freddie, and the Federal Home Loan Banks to the tune of millions, a fiscal transfer lacking the appropriate Congressional approvals.*

If the GSEs take the free gift, this will have very real effects on the Fed's financial health.  With IOR at -0.25% the steady rolling-over the Fed's assets will result in a decline in their yields, maybe even negative returns. As the spread between the return on assets and cost of liabilities shrinks, the Fed's seigniorage will drop dramatically. In  normal times, a certain portion of this seigniorage goes to rebuilding any impairments to the Fed's capital base while the rest flows through the Treasury to the taxpayer, either in the form of reduced taxes or services. With a 0% loophole being offered to the GSEs, the Fed risks running permanent operating deficits which in turn will lead to a steady erosion in its capital base. If the equity writedowns gets too large this may reduce the Fed's credibility in the eyes of the market, thus reducing the effectiveness of monetary policy.**

The Fed will have to plug this hole if it doesn't want to find itself neutered and/or subsidizing the GSEs. Maybe Fed officials can cap the amount of reserves that the GSEs are allowed to borrow, or use moral suasion to get the GSEs to drop out of the market. Better yet, maybe Fed Chair Janet Yellen can ask Congress to change 19(b)(12) so that, like all other banks, GSEs get to receive IOR (and are forced to pay it). Alternatively, maybe the laws can be altered so that the GSEs are prevented from keeping deposits at the Fed to begin with.

The legal option is a good idea for two reasons. First, it means that negative IOR effectively sets the overnight rate, restoring the pass-through from negative rates into financial markets and the real economy. It also removes the subsidy to the GSEs and the potential for large hits to share capital. Secondly, it fixes the Fed's leaky floor problem at positive rates of IOR. The Fed has already set up an overnight reverse repo agreement (ON RRP) facility to borrow from those ineligible to receive IOR, the idea being to keep a sterner floor under overnight rates. But as Marvin Goodfriend writes, ON RRP
violates the minimal intervention principle of central banking by turning the Fed into a financial intermediary operating directly on a large scale beyond the banking system with the potential to distort short term credit allocation and enable disruptive flight-to-quality flows during periods of financial distress.
Far less obtrusive to fix the leak with a quick change to legislation than fully arming a new battle station.

In closing, if the Fed wants to go negative, it may have some technical details to worry about. Fed officials no doubt already know this. If they don't, the blame can be attributed to their all-out focus on so-called normalization. By throwing all their organizational capital into the creation of a mechanism for keeping rates on an upward trajectory (the ON RRP), the Fed has diverted resources and attention away from the design of a complementary mechanism necessary for a downward rate trajectory. In hindsight, perhaps the Fed should have been tasking half its lawyers, accountants, financial architects, and media personnel to the rate normalization project, and the other to the negative rate project. Time to play catch-up?

Due to the difficulty of this subject matter, I've altered this post a few times since initially publishing it. The general idea has remained the same.

* In my original post, this paragraph didn't appear but was in the footnotes. I think it's important enough to bring up to the body of the text.
** Added on January 4.
*** Since originally posting this, I've softened my position on the sticky ceiling. I originally thought that the overnight rate would stay stuck at 0%, but I now feel it is likely to move closer to negative IOR. It would remain above IOR, however, so the idea of a sticky ceiling, while muted, is still relevant.


  1. I suppose signing them up as depositories has other unintended consequences.

    1. Yep, it would. That would mean they could compete with banks for deposits.

  2. OMG. This mirror-image thing is getting even weirder.

    Great post JP.

    1. Thanks Nick. It's like an upside down universe--it takes a while to figure out how to walk around.

  3. Something bothers me about the GSE story. If a free lunch arbitrage can persist with GSEs, why can't it just as easily persist without GSEs in the picture? The no-free-lunches argument for why IOR should put a floor on rates also "proves" that the current situation is impossible. As far as I know, absolutely nobody foresaw that IOR would create an arbitrage, but we're supposed to accept the revised story (with GSEs as culprit) without question. I'm not so sure!

    (I do agree that GSEs earning 0% is a problem for negative rates. I'm just talking about the positive rate story).

    1. Max, what do you think about this?

      "Financially astute readers may ask why the spread is so large. Shouldn't competition between banks force up the rate bid on F&F federal funds sold until it reached the point where banks profits were almost zero? The answer to that question is that F&F face credit risk in selling funds, so they limit both the set of banks to whom they will sell funds and the amount they sell to any individual bank. The result has been that F&F have had more funds to sell than the sum of the bank credit lines. Thus, the banks can bid for the funds at rates far below the interest rate they earn without fear that another bank will bid the funds away at a higher rate."


  4. Really good post

    Quick questions:

    - Sometime back I recall seeing data to the effect that a disproportionate amount of the GSE related arbitrage was flowing through the foreign banks, as reflected in their disproportionate share of system reserves. Any info on that? I think the reason may be that the domestics consider themselves more constrained in the sense of nominal capital ratios, and don’t want to intermediate a lot of this stuff. That, plus central bank moral suasion against it can be a legitimate force that impinges more on the domestics.

    - Is the RP facility being used by the Fed in part to cannibalize GSE deposits?

    1. - Yeah, I remember that too. Here is FT Alphaville:

      - I'm not sure exactly what you mean by cannibalize. My understanding is that the RP facility will accept GSE deposits overnight (as well as non-bank funds) so that these institutions needn't deal exclusively with private counterparties in the various overnight markets... and that in turn should help support the fed funds rate. Maybe that's what cannibalize means? ;)

    2. On your fist point, here is Stephen Williamson:

    3. "cannibalize" is a term used in retail banking (or at least it used to be) to refer to a new deposit category offering that anticipates increasing general market share of deposits, but at the cost of some intra-bank movement from an already existing deposit category that may in fact be less expensive than the new offering

      hence, by analogy, the Fed is offering a better rate to GSEs on RRP than what it pays on existing GSE deposits with the Fed

  5. Very good. For example, if it were legally possible to pay negative interest on reserve balances at, say, -0.25%, how close fed funds trade to 0% depends on the "balance sheet costs" of GSEs. They don't have to worry about deposit insurance premia, but maybe there are other implications - capital requirements?

    Though I don't think negative nominal interest rates are a fix for anything, going back to Congress would be useful. No one seems to know whose idea it was to leave GSEs out of earning interest on their reserve accounts.

    1. I suppose one difference here is that in the positive IOR case, the GSEs have to worry that they are lending in the overnight market to a counterparty that is not risk-free, private banks.

      But in the negative IOR case, banks don't have to worry that the counterparty to which they are lending overnight is not risk-free, since the GSEs have the same credit quality as the Fed. That may also affect how close fed funds trade to 0%, or -0.25%.

  6. JP, I'd be interested in any discussion of "illegal exactation" or "takings" when applied to negative interest rates in the US. It's a out-of-scope legal question - but one that has appeared before the court vis the Treasury's taking of Federal Reserve surpluses (courts held that this was a sovereign immunity question - Texas State Bank).

    But how the Fed can legally expropriate cash from privately-held banks is a bit of a mystery to me. It's a 5th amendment takings clause issue -- a hurdle that simply does not apply in Japan or parts of Europe.

    What exactly is "just compensation" if bank reserves get eminent domained? Thanks!

    1. I'm not sure. Negative interest is really just a fee. Does the levying of a fee qualify as expropriation?

    2. Well, a "fee" is usually voluntarily paid for a service rendered. The banks had always received zero fees on reserves (and complained bitterly at getting zero). Fed funds was the voluntary means where interest/fees could be obtained for these zero-fee reserves.

      When the Fed was set up, banks voluntarily joined the Federal Reserve system under the contract of zero-paying reserves. Nevertheless, this zero-agreement does not rule out voluntarily accepting positive interest payments on reserves -- you want to pay me, that's your business. You want to take money away from me, that's another story entirely.

      Involuntary negative rates could easily drain away hundreds of bank's shareholder equity. That's expropriation. If I am a shareholder, I go to court. If it's a fee, what service is the bank receiving? If its a penalty, what contract did they sign with the Fed? If the Fed is sovereign, what tax is NIOR?

      Would banks leave the FR System when faced with negative rates? Maybe it would make more sense to become a quasi money market fund -- they can now also hold Fed liabilities , anyway.

      If anything, JP, the Fed lacks the legal means to dive into negative rates. The only thing that will achieve negative rates in the US is a nasty financial crisis. NIOR will never be accepted lawfully or peacefully.

    3. How did Sweden, Switzerland, the ECB, Japan, and Denmark slide so easily into negative land?e What makes the U.S. unique?

    4. The U.S. has particular institutional restrictions on direct taxes and expropriation, alongside a quasi-private-public voluntary membership banking system overseen by both state and federal regulations. It's a very unique system. Involuntary negative rates won't work here without brutal institutional changes.

    5. Interesting.

      But isn't the "takings" argument an indictment of monetary policy loosening in general?

      Say the fed funds rate is at +4% and the Fed loosens so it falls to +3%. The Fed has effectively constricted the flow of income that a bank can earn overnight on its reserves, this in turn reducing bank shareholder equity. Why would a reduction from 4% to 3% be different from a reduction from +0.5% to -0.5%? They both take away bank income.

    6. JP, Fed Funds is a market - it's all voluntary. The fed simply controlled the scarcity of reserves. Now, IOR is set by statute. It's involuntary. By law, one cannot expropriate private bank assets by involuntary statutes. It's an exceedingly dodgy notion at least. The federal government in the United States has to regularly contort itself to make certain participation in its projects is voluntary. Now it could be that the Fed is legally private in this dimension; or it could be a public institution.

      Let's just say that the injection of reserves into the banking system in QE was nothing less than a capital injection - it handed equity to the banking system. You can voluntarily accept equity. Negative IOR can expropriate bank equity capital. You'll find that shareholders will not easily roll over and voluntarily hand over equity - which would likely end up at the Treasury via the Fed.

    7. FYI, more material on negative IOR, newly declassified.

  7. I'm guessing that a Japanese style tiered system under negative rates would mute the concern to some degree.

    It would be an interesting post to combine those two effects and simulate bank/GSE behavior under these conditions.

  8. "If the GSEs take the free gift, this will have very real effects on the Fed's financial health. With IOR at -0.25% the steady rolling-over the Fed's assets will result in a decline in their yields, maybe even negative returns. As the spread between the return on assets and cost of liabilities shrinks, the Fed's seigniorage will drop dramatically."

    So that kind of central bank margin squeeze risk is built into the Japanese system now, right?

    1. Yes, great points (also the one @5:30). You're getting ahead of me here but I think you're right.

      At negative Fed IOR, tiering allows banks to share in the same subsidy that GSEs enjoy, safe harbour at 0%. But that doesn't change the fact that as negative rates fall, say to -0.75%, the Fed is getting squeezed since its short term assets are earnings -0.75% and its liabilities (at least the tiered ones) 0%.

      And as you point out, that's the risk built in the the Japanese system and the Swiss one as well. They can't let the 0% tier get too big. We can think of the Fed's 0% subsidy to the GSEs as a teir without a cap.

  9. You asked some interesting questions in comments at Williamson’s 2010 post.

    I’m guessing your thinking has advanced up a personal learning curve since then.

    However, for what it’s worth, maybe you’d agree - or disagree - with the following:

    “Put differently, if fed funds are simply transfers of excess reserves, why does the end of day size of the fed funds market seem to dwarf the actual amount of reserves?”

    The “end of day size” of the fed funds market is a volume/flow statistic – not a stock statistic.

    I’d analogize it loosely to the FX market, where daily volume is in the trillions, but net long FX bank balances at end of day are no doubt a fraction of that.

    Also, in the case of pre-QE Fed funds, banks were constantly adjusting their positions throughout the day – in response to both net inflows and outflows from other banking activity and trading opportunities in the funds market. Excess reserve balances were relatively tiny, providing the impetus for extremely high sensitivity and reaction to changing reserve positions in fed funds velocity. Conversely, QE has stomped down on the velocity of fed funds simply because of the over-insurance provided by the quantity of reserve balances now provided to the system.

    “Is the relationship of a Fed Funds loan to reserves the same as a chequing deposit's relationship to paper dollars?”

    I’d analogize to the relationship between a non-bank financial institution’s loan to a customer, and the money balance it holds in its own bank checking account.

    But this is all pretty old stuff now.