Friday, January 4, 2019

Selgin on IOER and TNB

George Selgin has a nice piece on TNB and IOER, which I missed when it came out in September, but it's still relevant.

(HT a correspondent. TNB is "The Narrow Bank" which I wrote about here; IOER is interest on excess reserves. The Fed pays banks interest on reserves, which are accounts that banks hold at the Fed.) 

As George points out, TNB's model is to take money from, large corporations or money market funds, invest that money at the Fed as interest-paying reserves, and give as large an interest rate back to the depositors as possible. (Well, that's what their model will be if their suit against the Fed  winds through the US legal system before the next crash, which is unlikely, These customers can't get large enough insured deposits at regular banks; that TNB invests entirely in reserves make it impossible for TNB to fail so its customers don't need insurance. TNB doesn't want to let you or me give them money because that opens them to an immense amount of costly regulation.

The puzzling question is, how can TNB make money at that.?TNB takes money, invests it with the Fed, and the Fed in turn buys US treasuries. How is that better than TNB simply operating a money market mutual fund that invests directly in Treasurys?

The answer is, that for most of the last decade, the Fed has paid more interest on reserves than comparable treasury rates. Yes, "money" pays higher interest than "bonds," an inversion of classic monetary theory. Since money is more liquid, how can this survive? The answer is, because only banks can access this kind of "money." TNB was going to upend that.

Just why does the Fed pay more interest on reserves than comparable treasuries?  This is, like it or not, a nice little subsidy to banks, who get about 0.2% more on their reserves than anyone else can get.

Where does that 0.2% come from? You and me. George explains vividly
Just how is it that the Fed's IOER payments could allow MMMFs to earn more than they might by investing money directly into securities themselves? Because the Fed has less overhead? Don't make me laugh. Because Fed bureaucrats are more astute investors? I told you not to make me laugh! No, sir: it's because the Fed can fob-off risk — like the duration risk it assumed by investing in so many longer-term securities — on third parties, meaning taxpayers, who bear it in the form of reduced Fed remittances to the Treasury. That means in turn that any gain the MMMFs would realize by having a bank that's basically nothing but a shell operation designed to let them bank with the Fed would really amount to an implicit taxpayer subsidy. There Ain't No Such Thing As A Free Lunch... As it stands, of course, ordinary banks are already taking advantage of that same subsidy.
This is good, and I conclude that the Fed should keep a large balance sheet, flood the economy with liquidity as Friedman said it should, and run a tight corridor system paying no more on excess reserves than comparable Treasury rates.  Here we part company.

George seems to agree with the Fed though, that this subsidy is an integral part of the interest on reserves scheme, and that TNB will undermine the whole project of a large balance sheet and targeting interest rates directly via interest on reserves and later, the discount rate. I disagree.

The explanation, in a phrase, is that, were it to gain a charter, TNB could cause the Fed's present operating system, or a substantial part of it, to unravel. 
What would happen, then, if TNB, and perhaps some other firms like it, had their way? That would be the end, first of all, of the Fed's ON-RRP facility and, therefore, of the lower limit of the Fed's interest rate target range that that facility is designed to maintain.
Let me explain a bit more slowly. The Fed does not just want to peg the short term interest rate. For obscure reasons, it wants the Federal funds rate (the rate at which, these days, about $10 gets lent from one bank to another overnight) to wander between an upper and lower band. 

OK, normally central banks who want a "corridor" offer to pay a low rate, say 1% on any amount of money,  and lend any amount of money at a higher rate 2%.  Nobody will bother borrowing at more than 2%, when they can get 2% from the Fed, and nobody will lend at less than 1%, when they can get that from the Fed. Presto, the central bank now controls the interest rate, between 1% and 2%. Or, 1.499% and 1.5001%, if it wishes. 

Our Fed, again for obscure reasons, wants the amount it pays to be the upper bound, not the lower bound. So it pays more than a market rate to banks, and pays money market funds about 0.2% less (money market funds can invest at the Fed, that's what ON-RRP is, but they get a lower rate than banks), and the Fed keeps the whole thing 20-25 basis points above treasuries.

 (BTW,  When the Fed wants to push rates down under current procedures, it's going to have to invert this whole business, and lend money at lower rates than markets offer. Get ready for screaming opeds of rage.)

Whew. OK, George is right. This lower bound business will come falling down if TNB is allowed to operate, and will take over the entire ON-RRP business, since it can offer the same service and 20 more basis points. But so what?

So what's wrong with viewing this as a small ( 0.2% x $2 trillion = $4 billion) subsidy to banks, easily remedied by paying the same rate on IOER as treasuries?  As George asks:
So what if the Fed's leaky "floor-type" operating system lacks a "subfloor" to limit the extent to which the effective fed funds rate can wander below the IOER rate? Why not have the Fed pay IOER to the money funds, and to the GSEs while it's at it, and have a leak-free floor instead? Besides, many of us have money in money funds, so that we stand to earn a little more from those funds once they can help themselves to the Fed's interest payments. What's not to like about that?
My question exactly. Here George really seems to endorse the Fed's fears: 
... the Fed would face a massive increase in the real demand for excess reserve balances that would complicate both its monetary control efforts and its plan to shrink its balance sheet.
But the Fed has total control over its balance sheet. It can simply refuse to buy or sell, and then reserves are what they are. Banks have to get them from other banks. When supply is vertical, "massive demand" means that prices have to change. OK, either Treasury rates go up 20bp, or IOER goes down 20bp. I don't see the explosion.

Moreover, answering "What's not to like about that?''
Plenty, actually. Consider, first of all, what the change means. The Fed would find itself playing surrogate to a large chunk of the money market fund industry: instead of investing their clients' funds in some portfolio of Treasury securities, money market funds would leave the investing to the Fed, for a return — the IOER rate — which, instead of depending directly upon the yield on the Fed's own asset portfolio, is chosen by Fed bureaucrats.
The first just isn't true. Once IOER is no more than Treasury rates, money market funds that invest in Treasuries pay the same as IOER. Again, quantities don't have to change because prices can move.

The last sentence is the one that I really think underlies it all. George doesn't like interest rates being "chosen by Fed bureaucrats."

But George,  The point of an interest rate target is for the central bank to set interest rates.  

Yes ,in good old monetarist times, one thought the Fed should set the money growth rate at 4%, and leave interest rates to markets.  I harbor some related thoughts -- I would like the Fed to set the price level, and leave interest rates to markets. (I'll explain how to do that some other day). 

But nothing under discussion here gets the central bank out of setting interest rates. Today, the Fed is targeting interest rates.  The interest on excess reserves is set by Fed bureaucrats, and in a corridor the "market" only gets to pick up the last 20 basis points. In a corridor system, the Fed is targeting interest rates, just 20 basis points lower. In the old (1982-2008) operating procedures, paying no interest on reserves, but setting interest rates by rationing reserves, Fed bureaucrats were setting interest rates.

If we want to talk about the Fed getting out of interest rate setting altogether... well, that is an entirely different issue that these little changes in operating procedure say nothing about.


  1. This is helpful but I’m still confused--perhaps because I don’t understand these “obscure reasons” of which John speaks.

    I hate subsidies (unless, of course, I’m the direct beneficiary) and so setting IOER to a comparable treasury rate seems like a good idea. But if that happened, would TNB’s have any reason to exist? And if not, do we care?

    I guess you can tell a story about how TNB’s will provide a bit more stability—tall tales about about scary “shadow banks” are always fun to tell your kids around a campfire (and a few MMMF’s did, after all, briefly break the buck.)

    It sounds like George just wants the Fed to do less stuff, not more stuff. Maybe his real worry is that TNB’s are a kind of Hotel California—once you check in, you can never leave. But would that preclude a different kind of Central Bank? Suppose that in 2021 President Warren appoints Scott Sumner as Fed Chair. Scott immediately sets up an NGDP futures market and starts targeting that price . I don’t see how the existence of TNB’s with or without an interest rate subsidy would complicate that. But like I say, I’m still confused.

  2. Put me down as completely baffled by this post. I am pretty sure that in 2008 and the years soon after that, the IOER did function as a subsidy to banks, and that it was intended to allow them to rebuild their capital, when some of them (Citibank, BoA) were in really dodgy shape.

    Fast forward to today. I went to find out how much is being paid on IOER right now and I found this:

    "Implementation Note issued December 19, 2018: Decisions Regarding Monetary Policy Implementation"

    "The Board of Governors of the Federal Reserve System voted unanimously to raise the interest rate paid on required and excess reserve balances to 2.40 percent, effective December 20, 2018. Setting the interest rate paid on required and excess reserve balances 10 basis points below the top of the target range for the federal funds rate is intended to foster trading in the federal funds market at rates well within the FOMC's target range."

    Incidentally, on Friday 4 Jan, 1 month T-Bills were at 2.40 and 2 and 3 month bills were at 2.42.

    I am no longer sure if the Fed is the man with the baton in front of the parade or the one with the push broom and wheeled trash can at the back.

    1. They're always pushing on a string. It's why the recognition lag and effectiveness lag vex them so greatly.

  3. John,

    “Our Fed, again for obscure reasons, wants the amount it pays to be the upper bound, not the lower bound. So it pays more than a market rate to banks, and pays money market funds about 0.2% less (money market funds can invest at the Fed, that's what ON-RRP is, but they get a lower rate than banks)“

    While the Fed effective has been for the most part below or on par with the IOER, let’s not forget that the Fed did not originally intend for the IOER to be the upper limit. Instead, this was the original floor system, but which as you have pointed out, was ineffective due to the fact that it shut out GSEs and money funds. The more inclusive subfloor of ON-RRP fixed this. Now, why can’t they set the two rates to be the same? Well, they could, but on paper, the ON-RRP is a collateralized rate, so it conceivably makes sense that it should be set lower than the “unsecured” rate on reserves.

    In any case, the relevance of fed funds has been increasingly called into question, what with the ever more segmented short term funding market. Moreover, as policy tightens, one should expect to see upward pressure on market rates, eventually pushing fed effective above the IOER and returning the latter to a true floor, in the sense of the pre 2008 corridor system.

    - WW

  4. wanted to share that almost half of consumption expenditures are purchased electronically. From FT today In 2017, according to the Federal Reserve, US consumers made $6.6tn in payments on credit, debit, and prepaid cards, 8 per cent more than the year before — and a little less than half of total consumer expenditures on goods and services. That proportion could be even higher, analysts think, if the US did not lack cutting-edge card payments networks.

  5. John, I very much appreciate your thoughtful reply to my post. However, I'm afraid you're mistaken in assuming that I'm against the Fed's targeting of interest rates as such.

    In fact I have often complained myself about those who argue that the Fed should "just let the market set interest rates" and that sort of thing; and as for monetary targeting I gave up on it around the time most monetarists did, that is, in the early 1980s. In fact, outside of the lower bound, whatever that may be, I share the consensus view that an intermediate interest rate target is the best practical option. Indeed, although I have criticized the Fed's floor system of interest rate control, I have recommended instead a corridor system, which I obviously wouldn't do were I against interest-rate targeting per se.

    So just what am I objecting to in the passage you quote toward the end of your post? It's not having an interest rate target, but having an interest yield on Fed balances, accessible not just to banks but to others, and to MMMFs especially, were that yield is not necessarily the same as the risk-adjusted yield on the Fed's actual asset portfolio.

    In an orthodox floor system, the IOER rate would be approximately the same as the Treasury bill rate; moreover the Fed might be expected to hold a portfolio consisting of Treasury bills only. In that case the IOER rate would also typically reflect the yield on the Fed's portfolio. Extending that rate to MMMFs would then just be a way of having those institutions earn the same return on Fed balances that they might earn by holding Treasury bills. It would really make no difference whether the funds bought Treasuries or held reserves.

    But in our unorthodox system, apart from other departures, the Fed's portfolio consists of long-term Treasuries and MBS; consequently two things will tend to be true if MMMFs can all hold IOER-earning reserves instead of Treasury securities. First, the MMMFs are now investing in assets they could not otherwise invest in. Second, the return they realize on those investments, rather than being simply the return (allowing for risk) that the Fed realizes on its portfolio, is instead the IOER rate, which is generally NOT the same. Among other things, instead of the MMMFs themselves bearing the risk associated with the assets backing their reserve holdings, that risk is borne by the Treasury, and thus ultimately by taxpayers.

    Now you may say that that's all well and good. But it seems to me that there are real distortions, or at least real changes, involved in such a set-up compared to the one in which MMMFs don't have access to IOER-earning Fed balances, and must instead invest in Treasury bills.

    Note finally that, in the actual Fed floor system until quite recently T-bills routinely earned rather less than IOER. That's why TNB looked like a good idea in the first place. In that case, the distortion from having MMMFs hold IOER-earning reserves would have been still greater than the distortions I've pointed out.

    In general, I think it is a good idea that investors should earn a return that reflects the true risk-adjusted yield of the portfolio of assets into which they savings have been invested. That would certainly not be the case in the TNB plan; and it is with regard to that particular problem that I object to having MMMFs earn bureacratically-regulated interest rates. It has nothing to do with not wanting the Fed to engage in any sort of interest-rate targeting.

    1. George:
      Thanks for the clarification. I think we now agree! Yes, the Fed should not offer more than short-term treasuries, and ideally the balance sheet would all be short term treasuries.

    2. Just wanted to point out that when only a segment of the market holding large sums of cash has access to interest paying reserves, it creates a situation where the “haves” borrow cheaply (below IOER) from the “have nots” to arb the system. This is what happened until the Fed granted “backdoor” access to the GSEs and MMMFs through the ON-RRP to contain the floor from leaking.

  6. The purpose of the IOER policy is to set the minimum lending rate and so it would be a bit perverse to make the facility available to a Bank that is not going to be doing any lending.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.