Recent Posts

Tuesday, November 24, 2020

Comments On Scott Sumner's Questions About MMT

Scott Sumner asked a few questions about Modern Monetary Theory (MMT) (link), and this article is a short response to some of them. The issue at hand is the effect of open market operations, in particular, in the pre-2008 U.S. system (where reserves paid not interest). 

The Issue

Sumner had looked at the text Macroeconomics by Mitchell, Wray, and Watts, and had some concerns.

Sumner writes:
2. The second paragraph seems to apply to all open market purchases, not just those that occur when market interest rates equal the interest rate on excess reserves (IOER). That’s clear from the rest of the book, which focuses heavily on historical examples from the 1960s, 1970s and 1980s. So for the rest of the post I’ll consider OMOs that occur in a world where nominal interest rates are positive and there is no IOER, i.e. the pre-2008 world. To be sure, I don’t think their claim is even true in a world with IOER, but it’s at least more defensible in today’s world.

3. Most mainstream economists (including John Maynard Keynes) would not agree with the claim that a purchase of interest-bearing bonds with zero interest base money is “irrelevant”.
He then poses a question about a hypothetical large balance sheet expansion.
OK, so why do MMTers believe that OMOs don’t matter? Suppose we go back to the economy of the 1990s, when risk-free interest rates were often about 5%. Assume the monetary base is $500 billion (as in 1998). Now consider this thought experiment. The Fed buys another $500 billion in Treasury bonds, paid for with zero interest base money. Is that action actually “irrelevant” because the sum of base money and publicly held debt doesn’t change?
This is in response to (page 342 of Macroeconomics):
Second, they would challenge the theory of inflation based on QTM, and argue that if a fiscal deficit gives rise to demand pull inflation, then the ex post composition of  ΔB +  ΔMb in Equation (21.1) is irrelevant. Overall spending in the economy is the driver of the inflation process, and not the ex post distribution of net financial assets created between bonds and base money.
The passage from Macroeconomics does not completely cover this response, so we need to look further into the literature. 

The first thing to note that Sumner's assertion "OK, so why do MMTers believe that OMOs don’t matter?" is a misunderstanding of the passage from Macroeconomics. The passage says that the ex post decomposition does not matter for inflation, but it does not say that that open market operations have no effect on anything. Open market operations are related to how the central bank determined the short-term rate of interest in the pre-2008 system -- but the determination process bears little relationship to what Monetarist models say. Open market operations "mattered" to the extent that nominal interest rates matter. (In the post-2008 Fed system, or in places where central banks use a corridor system, open market operations have a very tenuous relationship to interest rates, solely based on very questionable beliefs about supply/demand effects on term premia.)

Returning to the question, the MMT analysis -- which now seems to be accepted by non-Monetarist neoclassicals -- is that the Fed had almost no discretion in setting the amount of reserves in the system (pre-2008). 

The chart above shows the percentage of reserves that were excess to requirements from 1984-2000 (I truncated data 2000 since the Y2K spike obscured the scale.) We see that excess reserves were generally sub-5% of the total before 1995, with a spike in the early 1990s. After 1995, the percentage drifted higher (with a few spikes), and then the Large Scale Asset Purchases drove them to nearly 100% of the total post-2008.

The Fed's open market desk had almost no day-to-day discretion in setting the level of aggregate reserves. (The modern Monetarist theory has evolved to discussing expectations about the money supply, which is another issue.)
  • If they tried to drive the amount of reserves below the required amount, they would technically cause solvency breaches for some banks, which would be destabilising. (Sine banks did not want to go to the discount window, Fed Funds rates could become arbitrarily large as banks bid up Fed Funds. And if they went to the discount window, the missing reserves would be created.) As such, the excess reserve amount could not drop below the required amount -- and the required amount is determined with a lag, so there is no story about banks being able shrink their loan books. 
  • In practice, the excess reserves could not get too large, or else the Fed Funds rate would go to zero. As the data above show, outside crises, excess reserves were around 5% of the total amount, and that was stable across widely changing interest rate levels. Sumner's thought experiment of injecting $500 billion of reserves would almost certainly have been enough to drive Fed Funds to zero, and the Fed would lose control over interest rates until excess reserves dropped to a more sensible level.
This is of course what happened in 2008, but the addition of paying interest on excess reserves ended the relationship between short-term rates and the size of the Fed's balance sheet. (At most, bond yields could allegedly be lowered by open market purchases.)

I pointed this out in the comments section, but Sumner objected, pointing to the Monetarist experiment under Volcker. However, the MMT literature argues that the Monetarist interpretation of that episode is incorrect. 
  • There is a paper by Kliesen and Wheelock of the St. Louis Fed that notes that the operating procedure widened the band for the Fed Funds rate, and thus was not solely a money growth target. That is, the open market desk would just allow Fed Funds to be at the top end of the allowed band if money growth was above target.
  • The MMT literature notes that policymakers knew that they were setting a policy rate, and hoping that money growth would align with estimated money demand functions. In Chapter 6 of Central Banking, Asset Prices, And Financial Fragility, Éric Tymoigne. quotes Fed Chair Volcker from the October 1979 FOMC transcript, page 28: "We are not - at least I am not - proposing that we go to a purely mechanical reserve targeting approach. [... I cut the quote further] There are all these elements of judgement that enter into the process. That, combined with the constraint on the federal funds rate [emphasis mine], brings us to something of a hybrid."

Concluding Remarks

The MMT argument is straightforward.
  • In the pre-2008 Fed operating procedure, the Fed had almost no discretion in setting the level of reserves in the system -- even during the Monetarist experiment. If they engaged in hypothetical QE operations, they would drive Fed Funds to zero, and they would lost control over interest rates. Once they entered into such a regime, short-term bonds and reserves are fungible, and so changes in the mix have no effect on the economy. (The effect of the zero interest rate policy is another question.)
  • In the current Fed operating procedure (and among the developed countries that I am aware of), settlement balances at the central bank pay policy interest rates, and are fungible with short-term government bonds. The sole effect of balance sheet expansion is the alleged effect of reducing term premia.
Once we look at the above statements, it is not that hard to see that the decomposition between monetary expansion and bond issuance does not matter for inflation. For the pre-2008 Fed, money growth was almost entirely driven by private sector demand. Otherwise, central bank balance sheet expansion is just swapping fungible governmental liabilities, which has tenuous effects on the economy.


  • Kliesen, Kevin L., and David C. Wheelock. "Managing a New Policy Framework: Paul Volcker, the St. Louis Fed, and the 1979-82 War on Inflation." FRB St. Louis Working Paper 2020-022 (2020).
  • Tymoigne, Éric. Central banking, asset prices and financial fragility. Routledge, 2008.
Note: I was given a number of references for this discussion on Twitter. @Wonkmonk gave me the Kleisen/Wheelock reference. A few people noted the transcripts, and rather than go through them, I used the Tymoigne reference. Scott Fullwiler wrote "Meulendyke says they were targeting interest rates in her response to Basil Moore 30 years ago in the JPKE" - a reference that I did not have time to track down. Also, Stephanie Kelton offered this reference by Papadimtriou and Wray:

(c) Brian Romanchuk 2020


  1. I find it very uncomfortable to try to explain MMT. Even when I say look, this is my personal understanding of what MMT says, I am always worried that I am going to screw it up somehow. Lord knows I argue with MMTers all the time even within that school of economics. Do you ever feel the same way Brian? Trying to explain something to the best of your knowledge but worrying the people who came up with what you're trying to explain won't agree?

    1. I’m writing a MMT primer, and concerned about my characterisation. That’s why I previously avoided writing “MMT says...” unless I am working with an actual reference.

      In the case of the Sumner discussion, my input didn’t help. The key thing is that he didn’t parse the statement about “irrelevant” properly, and the whole thing devolved into a snark hunt.


Note: Posts may be moderated, and there may be a considerable delay before they appear.

Although I welcome people who disagree with me, please be civil.

Please note that my spam comment filter appears to dislike long "anonymous" posts.

Note: if you want to post comments from Apple devices (iPhone, iPad), you apparently need to turn off "prevent cross-site tracking" in Safari privacy settings. (The reason presumably is that another URL handles comments, and so the user session needs to be preserved when redirected to that site. I don't like this, but this is not enough to make me switch my hosting service.)