49 Comments

Prof. Sumner,

Has anyone tried to encourage the people behind Polymarket to create a giant, highly-liquid NGDP futures market? If not, are there any likelier potential candidates out there? Couldn't Tyler Cowen, Alex Tabarrok, Yglesias, or Noah Smith get someone's attention in order to do that?

Expand full comment

I don't think there'd be much interest in trading the contract.

Expand full comment

That's true right now and that's too bad.

But we do have agency (albeit limited) in this situation. And Tyler Cowen, Alex Tabarrok, Yglesias, or Noah Smith might have even more influence!

Don't. Stop. Thinkin' About Tomorrow.......

:)

Expand full comment

Polymarket smart contracts were hacked twice, and you want those programmers (all of whom are subpar) to build the smart contracts for a futures market? lol.

Every programmer has already thought of a futures market, but it's a lot more complicated than a simple prediction market. And no entreprenuer is going to build one for just NGDP fans, which is fewer than 10,000 worldwide.

Expand full comment

I agree. The best solution would not be a conventional prediction market, it would be a Fed commitment to takes NGDP futures positions in support of its policy target. (Say defending NGDP growth in the 3% to 5% range.)

Expand full comment

On Yglesias’s blog (or was it Noah Smith’s) an economist smarter than me with decades of experience wrote the following:

“The US Treasury coud just issue a 5 and 10 year Trillionth of the NGDP security and sell it”

Expand full comment

I think I first heard that from Sumner. But not recently.

Of course that is an instrument. The instrument does not determine the target and I still have not seen a point by point comparison of why Sumner or any other NGDP targeting person prefers that over FAIT.

Expand full comment

I thought, but pls correct me, that the gist of Cochrane’s argument is that money supply is endogenous. In other words, it is just that money supply is not controlled by the central bank under the current regime, but it cannot be controlled under the current technology.

Expand full comment

Which technology are your referring to? I have a hard time imagining how technology could prevent the control of the money supply.

It is true that if you target interest rates then money is endogenous, just as if you target money then interest rates are endogenous.

And if you target NGDP, then both are endogenous.

Expand full comment

I used the term ‘technology’ to refer to the proliferation of close substitutes for narrow money, which Cochrane describes. I confess that I have a hard time in imagining a world in which inflation can be dictated by the supply and demand of cash, most of which, as you note, is held in places like Russia and Latin America

Expand full comment

I'd make several points. "Technology" is not the right term for your argument. Even if you are correct about the close substitutes, it would still be true that the monetary base is highly controllable.

I'd push back very strongly on the close substitute argument. I strongly believe that there are no close substitutes for currency. Almost no one says "I can't decide whether to bring cash or T-bills with me when I go shopping today". And the fact that some of the currency demand is overseas has no bearing on the question of whether the Fed can control the purchasing power of currency.

You may find it hard to imagine a monetary regime that worked that way, but it's roughly the policy in place in the US until 2008. The base was more than 98% currency. The Fed controlled inflation through open market operations. The 2% average inflation from 1991-2008 wasn't just coincidence. We can and should go back to that policy.

Expand full comment

Interesting stuff leading to many thoughts, but let me make a couple points.

The biggest elephant in the room when it comes to targeting the money supply is "What is it?" Beyond choosing between M1, M2 and Dvisia M4, the assets that are highly liquid change over time. In particular, in a financial crisis assets (think commercial paper when Lehman failed) can go from being money-like to illiquid in a day. So what exactly should we target?

Second, its not fair to compare velocity and the real natural rate of interest in terms of uncertainty and volatility. Statistically, we cannot reject that velocity contains a random walk, meaning forecasting it is a very tall order. Any real natural rate is hard to estimate and changes over time, but it changes slowly due to factors like world demography.

Putting these ideas together means its very difficult to target the money supply with the goal of compensating for changes in velocity. Furthermore, the actions the Fed took in the Great Recession, which includes targeting the fed funds rate, accomplished that to a degree, depending on which version of M you use. The money supply rose just as velocity fell. Causality probably goes both ways.

So interest rates themselves are much more transparent as is policy targeting them.

Expand full comment

"Any real natural rate is hard to estimate and changes over time, but it changes slowly due to factors like world demography."

I strongly disagree with this. The natural interest rate is highly volatile over the business cycle. One cause of business cycles is that central banks are often too slow to adjust to sharp changes in the natural interest rate, as we saw in 2008, and also 2021.

Expand full comment

We must have a different interpretation of “natural.” Here’s the work I have in mind

https://www.newyorkfed.org/research/policy/rstar

Statistically, its not a slam dunk. There may be unit root in rstar, but velocity sure looks more volatile to me. I’ll post some graphs soon.

Expand full comment

Here's their definition:

"Their approach defines r-star as the real short-term interest rate expected to prevail when an economy is at full strength and inflation is stable."

That's not Wicksell's concept, and I don't see how it's useful. Under interest rate targeting, what we need to know is not where interest rates would be if we were in macroeconomic equilibrium, the Fed would need to know where interest rates need to be to get to equilibrium. That's the sense in which the equilibrium interest rate is highly unstable and hard to estimate.

If the Fed actually knew where interest rates ought to be set then we would not have had the huge NGDP overshoot in 2021-23.

Expand full comment

Is the Fed targeting interest rates or are they targeting inflation?

Both.... neither? The interest rate target is a short-term mechanism employed in the big-picture goal of price stability. We don't have timely data on inflation. So, if inflation appears to be too high, we tighten money. How much? Enough to move interest rates for now, see what that does, and maybe do it again next quarter.

I would say, that the Fed isn't even targeting interest rates. They have a vague and ad-hoc idea of economic direction with indicators of GDP, CPI, Unemployment, etc. but no firm target for any one indicator.

On the other topic of interest on reserves -- yes, it is time to go. It is incredibly distressing that ECON 101 books still discuss monetary policy based on the pre-2008 mechanisms. They are 15 years out of date! But, more to the point, the spread between the cost of holding idle cash vs. the potential return on investments drives growth. If you get the same return, in either case, why take the risk? Furthermore

Expand full comment

I'd say they target interest rates over the short term (say 6 weeks) and inflation over the long run (a period of years.)

Expand full comment

Not sure if this is directly related to your post, but I was thinking about some similar themes recently.

I greatly enjoyed the recent Econtalk where Russ Roberts talks to J. Doyne Farmer about agent-based modelling (https://www.econtalk.org/chaos-and-complexity-economics-with-j-doyne-farmer/). Enough to buy his book. It's an interesting methodology, but one point he makes in the book is that they try to create the simplest model they can to reproduce the behavior we see in the real world. For instance, they can replicate the distribution off asset price returns (like fat tails and volatility clustering) by including agents that are value investors and trend followers and allowing them to use leverage.

Wanting to have the simplest model that explains things is kind of how John Cochrane starts building up his fiscal theory of the price level approach. He then adds a more-or-less standard New Keynesian DSGE framework on top of that.

I don't think Farmer and his colleagues are necessarily wedded to the agent-based approach if a model with representative agents does a good enough job. However, I wonder if it makes sense to go the route he describes in his book off building a complex agent-based model of the economy if you don't get the underlying fundamentals right. You can get a more and more realistic model for usual times, but will it be able to capture events like our recent bout of inflation? If a model can't get these big events right, then it's not that good.

Expand full comment

I have a simpler model than John---inflation is all about the supply and demand for base money.

Expand full comment

It's a model, but it's not the same kind of model that John is working with.

Expand full comment

Yes, it's a very different kind of model.

Expand full comment

There have been new-Keynesian types in the recent past that have supported the focus of Federal Reserve on NGDP indices. Has that approach losing currency

Expand full comment

I believe there's still substantial support.

Expand full comment

I thought the issue with targeting monetary aggregates in the 1980s was not so much that they led to recession, but that the aggregates became unstable due to financial innovation. This article examines the international experience and shows that targets were maintained well into the mid-late 1980s in many countries: https://www.rba.gov.au/publications/confs/1989/argy-brennan-stevens.html

Expand full comment

Monetarism has never been tried. Monetarism involves controlling total legal reserves and reserve ratios.

Volcker didn’t understand Bob Roosa’s “little red book”. In 1980, Paul Volcker, Past chairman of the Board of Governors of the Federal Reserve System, appeared before the House Domestic Monetary Policy Subcommittee. In response to a question as to why the Fed had supplied an excessive volume of legal reserves to the member banks in the third quarter 1980 (annual rate of increase 13.2%), Volcker’s defense was that there are two types of legal reserves: 1) borrowed (reserves obtained by the banks through the Federal Reserve Bank discount windows), and 2) non-borrowed (reserves supplied the banking system consequent to open market purchases).

He advised the congressmen to watch the non-borrowed reserves — “Watch what we do on our own initiative.” The Chairman further added — “Relatively large borrowing (by the banks from the Fed) exerts a lot of restraint.”

This is of course, economic nonsense. One dollar of borrowed reserves provides the same legal-economic base for the expansion of money as one dollar of non-borrowed reserves. The fact that advances had to be repaid in 15 days was immaterial. A new advance can be obtained, or the borrowing bank replaced by other borrowing banks.

The importance of controlling borrowed reserves was indicated by the fact that Volcker targeted non-borrowed reserves (@$18.174b 4/1/1980) when total reserves were (@$44.88b).

Expand full comment

re: "the Fed had supplied an excessive volume of legal reserves"

Note that Dr. Richard G. Anderson's "reconstruction" of reserves at the Federal Reserve Bank of St. Louis's database showed little change. Anderson's construction was an inaccurate historical correction related to the DIDMCA of March 31st 1980.

Expand full comment

This other RBA article (https://www.rba.gov.au/publications/confs/1997/grenville.html) comments as follows, citing Goodhart, C. (1989), ‘The Conduct of Monetary Policy’, Economic Journal, 99(396), pp. 293–346., which is gated:

"By the time Australia ‘suspended’ monetary targets in February 1985, many other countries had abandoned or downgraded them. ‘By the latter part of the 1980s the technical elements (of monetary targets) were deemed by the generality of policymakers to have comprehensively failed’ (Goodhart 1989, p. 296). The main reason was, as in Australia, a breakdown of the relationship between money and nominal income. As Governor Bouey of the Bank of Canada said: ‘We didn't abandon monetary targets, they abandoned us’."

Expand full comment

I’m somewhat skeptical of your assessment of the monetarist experiment. In an important sense it was successful: Inflation was defeated, a strong recovery eventually ensued, and monetary dominance was proven. Yet these successes being a result of targeting monetary aggregates seems hard to prove. Money growth rates exceeding the target range were associated with falling stock and commodity prices. On many occasions Volker allowed M-1 to exceed its target ranges, yet prices and exchange rates remained stable. And so the monetarist prescription was not really truly followed in the 1980s, with money growth targets not strictly adhered to, but that’s what shows the failure of it. The inflationary predictions of the monetarists did not come true, and not adhering to money growth targets arguably avoided a lot of unnecessary tightening in the 1980s. Seems like Monetarists got the sign wrong, and while this looks like an identification problem, I think theres issues with using monetary aggregates even as an intermediate target, related to criticisms David Glasner has frequently made on applying quantity theoretic reasoning to inside money.

Expand full comment

I don't disagree with those points, but I'd frame it differently. Monetarists often complain that money targeting was never tried, for the reasons that you discuss. But if that is true, then obvious 1980-82 cannot be a failed monetarist experiment, because it means the experiment was never actually conducted. That supports the broader point of my post.

Keynesian critics tend to view that argument as a sort of "excuse making". For any policy, one can always say it was not done perfectly, and that that excuses any flaws.

So let's say it was done to some extent, but not faithfully (which is my view). What can we learn from that? In that case, I'd say we learned that it is a powerful tool for reducing inflation, but it's dangerous to target M because of unstable velocity. In other words, a flawed experiment.

You are right that the inflation predictions of the monetarists did not come true, although one could add that the inflation predictions of the Keynesians also did not come true. They predicted that Reagan's deficits would lead to high inflation. So I'd stand by my claim that there is nothing in the 1966-82 experience that points to interest rate targeting being better. I'd say the overall record showed a mediocre performance from money targeting and an even worse performance from interest rate targeting.

Now of course interest rate targeting did much better after 1982, but if we'd adopted a flexible money supply targeting regime based on what we learned from 1980-82, that also would have done better after 1982. And NGDP forecast targeting would do better than either of them.

You said:

"Money growth rates exceeding the target range were associated with falling stock and commodity prices."

Isn't that because investors expected the Fed to respond by tightening money, to bring it back within target?

Expand full comment

re: "t means the experiment was never actually conducted"

That is right. The money stock exploded after Volcker told us to watch the money stock.

Paul Volcker’s version of monetarism (along with credit controls: the Emergency Credit Controls program of March 14, 1980), was limited to Feb, Mar, & Apr of 1980. With the intro of the DIDMCA, total legal reserves increased at a 17% annual rate of change, & M1 exploded at a 20% annual rate (until 1980 year’s-end).

Why did Volcker fail? For two reasons. The first was the failure to recognize monetary lags. Contrary to Nobel Laureate Dr. Milton Friedman, lags are not “long and variable”. The distributed lag effects for both real output and inflation have been mathematical constants for over 100 years.

Volcker in his book "Keeping At It": Pg. 105: “We needed a new approach. To have more direct impact, we could strictly limit growth in the reserves that commercial bans held at the Federal Reserve against their deposits. That would effectively curb growth in deposits and the overall money supply. Put simply, we would control the quantity of money (the money supply) rather than the price of money (interest rates). “

Paul Volcker lied.

Expand full comment

Velocity became endogenous when the Fed started paying interest on reserves. QT will keep it grinding higher until the Fed responds to another liquidity panic with QE and it lurches lower. Abundant reserves required paying interest on them, turning reserves into substitutes for repo and other short-term investments. The several liquidity squeezes since September 2019 indicate that, far from reducing market panics, IOR and abundant reserves makes them more likely. Imho much good would come if the Fed could merely “get rid of IOR and go back to using open market operations as the primary policy tool.”

Expand full comment

Yes, and haven't bank regulators also responded to QE by demanding that banks hold much larger reserve balances than before 2008?

Expand full comment

Contrary to Sumner, banks don't lend deposits. Deposits are the result of lending.

See: BANKS DON’T LEND MONEY (youtube.com) Dr. Richard Werner

Link: George Garvey:

Deposit Velocity and Its Significance (stlouisfed.org)

“Obviously, velocity of total deposits, including time deposits, is considerably

lower than that computed for demand deposits alone. The precise difference

between the two sets of ratios would depend on the relative share of time deposits

in the total as well as on the respective turnover rates of the two types of

deposits.”

But Powell sums up all deposits together and claims M2 is worthless: “We have had big growth of monetary aggregates at various times without inflation, so something we have to unlearn.”

The FED’s Ph.Ds. don’t know a debit from a credit. Banks are not intermediaries. Funds do not leave the payment’s system. Indeed, as evidenced the existence of “float”, or the “check is in the mail”, (largely eliminated by “The Check 21 Act” which introduced electronic substitutes on Oct 28, 2004), the payment system’s reserve credits tend, on the average, to precede the reserve debits.

This being so it is a delusion to assume that savings can be “attracted” from the commercial banks, for the funds sever leave the payment’s system.

Secular stagnation was the impoundment of savings, the bottling up of savings, in the banks. Only 16% of bank deposits represented transaction deposits at the beginning of the GFC. With C-19, 56% of bank deposits have grown to represent transaction deposits. That’s what has changed and is propelling the economy.

Expand full comment

Greenspan discontinued the longest running time series that the FED had in September 1996, the G.6 Debit and Deposit Turnover release:

https://fraser.stlouisfed.org/files/docs/releases/g6comm/g6_19961023.pdf

Note the difference between the turnover ratios for bank deposits on this release. Demand deposits turnover was at a radically faster rate than other deposits. Ergo, demand deposits are our means-of-payment money supply.

Lorie Logan gives a prime example of this delusion: “But in an efficient system, the costs of liquidity should be similar for banks and non-banks.”

Shadow Stats “extraordinary flight to liquidity” in Basic M1 (Currency plus Demand Deposits) reflects this "sea change".

Expand full comment

Interest rate targeting is seen as accommodating Treasury issuance. The time horizon of the trading desk’s policy has been 24 hours rather than 24 months. The FED would have to "wash out" any adjustments to reserves.

Expand full comment

On September 17, 2008, the effective fed funds rate was 2.80%. It was no longer possible to target the fed funds rate while pegging the interest rate on reserves at zero. Similarly, NGDP futures target would be more stable in practice with IOR, especially when it really matters during a financial crisis.

Expand full comment

"It was no longer possible to target the fed funds rate while pegging the interest rate on reserves at zero."

You say that like it's a bad thing. :)

Expand full comment

It was a very bad thing, the effective fed funds rate of 2.80% was

very contractionary (the target was 2%). As a result, 1 month LIBOR

increased by 2 percentage points or so.

Imagine NGDP futures trading all over the place in a range between 3% and 7% during the same week.

Expand full comment

You don't solve that problem with IOR, you solve it with QE (or better yet NGDPLT.)

Expand full comment

Targeting N-gDp mitigates lag effects.

Expand full comment

Targeting N-gDp encompasses both money and velocity fluctuations.

Expand full comment

IOR helps you do QE even when equilibrium interest rates are above zero. A good example where that was needed was Poland in 2008. I discussed the Polish situation in 2008 in a comment on the 'Moving Toward NGDP Targeting' post (financial crisis, healthy NGDP growth and 3% interest rates).

Level targeting does mitigate the damage and is very important.

Expand full comment

I don't see any role for monetary policy beyond stabilizing NGDP. I'm skeptical of bank bailouts, but if they are necessary then I'd rather let another institution like the Treasury get involved.

And it's hard for me to imagine a severe banking crisis in the US where NGDP does not fall. Do enough QE to keep NGDP from falling and the banking system should be OK.

Expand full comment

This is not about bailing out any specific financial institution; it is about ensuring that broad measures of credit risk do not explode, destabilizing the market for NGDP expectations. Fiscal deficits in the US might raise the neutral interest rate significantly over the next 100 years, requiring QE to keep NGDP from falling while interest rates never drop to zero.

Expand full comment

Monetary policy works with lags. The EFFR would never require such an adjustment if N-gDp was targeted.

Expand full comment

Scott Sumner asks, Why does almost everyone assume that interest rate targeting is best?

I would ask, who _does_ think interest rate targeting is best? The targets on the table are inflation targeting, specifically Flexible Average Inflation Targeting, the Fed’s actual target, and NGDP targeting favored by Sumner and others. Sumner does not cite a single economist who advocates for interest rate targeting, which the Fed actually had from 1945-51. And for good reason. Targeting the short-term interest rate (at what level?) would in effect turn macroeconomic management of inflation and NGDP over to fiscal policy. This policy change, although favored it appears by ex-President Trump, has little support elsewhere. Interest rate targeting is surely NOT favored as the only alternative to _money supply_ “targeting,” which is almost universally rejected. “Monetary supply” management is yet another possible _instrument_ for achieving some target.

Sumner seems to be referring to the use of short-term interest rates, like the Effective Federal Fund Rate as the main _instrument_ for the achievement of its inflation _target._ I would agree with Sumner that over-emphasis on that one instrument compared to IOR and purchases/sales of longer-term instruments is a communications error and led during the Great Recession to concerns that the Fed was “out of ammunition” or that its ability to manage inflation was constrained by the “Zero Lower Bound” of short-term interest rates. But none of this bears on what outcome the Fed should target.

Much of Sumner’s post is puzzling as he recognizes that both interest rate targeting and money supply targeting lead to unstable outcomes of variables of fundamental concern like inflation and output. In this he shares the view of John Cochrane, whom he cites, in rejecting “interest rate targeting,” although he has many long quibbles with Cochrane’s analysis.

Most puzzling of all is that in rejecting “interest rate targeting,” assuming that nowadays it is even a “thing,” Sumner does not in fact advocate for anything different, for NGDP targeting, for example, which he has advocated in the past. [See: https://thomaslhutcheson.substack.com/p/why-target-ngdp]

Expand full comment

I would call 1945-51 "interest rate pegging". I'd call current policy interest rate targeting.

Expand full comment

21.yrs ago, the Fed did a very successful monetary policy. It does not mention interest rates but calls for "policy accomodation". It could be succesful in the future if by "monetary policy accomodation" it explicitly mentioned an NGDP target level!

https://thefaintofheart.wordpress.com/2012/11/13/forward-guidance-then-now/

Expand full comment

One idea mentioned by Fed people in that post would be to target long term interest rates instead of short term rates. But I don't see that as solving the problem.

Expand full comment
Error