Scott Sumner  

There is nothing tautological about market monetarism

PRINT
Saving Money with One Income... Response to Krugman on My Cana...

Bill Woolsey directed me to a post by Kevin Grier (commenting on Lars Christensen):

OK, so the first graph is the path of Nominal income (PY) relative to trend. The second is the path of real income (Y) and the third is the path of prices (P). Nothing objectionable about the graphs in themselves.

You can see NGDP has fallen a lot (relative to trend), mostly due to lower real GDP. Since we are dealing with accounting here, we really only need two of these graphs. the third one is implied by the other two.

But people, what just sets my teeth on edge and puts a bee in my bonnet is the idea that, and I quote:

The message from the graphs above is clear - the Italian economy is suffering from a massive demand short-fall due to overly tight monetary conditions (a collapse in nominal GDP).

Nominal GDP IS nothing more than the product of prices and output. To say that a fall in nominal GDP relative to trend "caused" the fall in the path of prices and output relative to trend is just gibberish.

Try it in the abstract without the sacred labels. "The fall in XY caused the fall in X and the fall in Y".

Ummm, maybe the fall in Y caused the fall in X and as a result XY also fell?? Or the fall in X? Or some third factor caused both X and Y to fall and as an unavoidable consequence of arithmetic, XY also fell?

Labeling PY as "Monetary conditions" and then saying Y fell because PY fell and blaming that on monetary conditions is not an economic theory. It's not even an un-economic theory.


First let me remind readers that although their names sound similar, real and nominal GDP are completely unrelated. Just as the product of an odd and even number is always 100% even, and not sort of even and sort of odd, nominal GDP is a 100% nominal variable. Real GDP is more similar to the output of the kiwi fruit industry than it is to nominal GDP. The fact that RGDP and M*V are strongly correlated in the US is a stunning, mind-boggling fact that is not predicted by classical economic theory. There are many countries where the two variables are not correlated.

Nominal GDP can be compared to other nominal variables like the money supply and inflation. Is there anything controversial about saying that a fall in the money supply reduced output? How about claiming that deflationary monetary policies reduced output? Those are the sorts of things you'd hear from old monetarists and Keynesians. Conversely, a real business cycle economist would predict that a fall in NGDP (or M or P) would not reduce output.

Grier is right that Italian output might have fallen for other reasons. Normally if that were to occur (a negative supply shock) you'd expected Italian prices to rise while Italian output fell. But Italy is now part of a monetary union. In that case you might expect a negative supply shock to produce falling output and stable prices. The fact that prices also fell is indicative of the fact that the monetary union it belongs to was itself creating a negative demand shock, for Italy and for the union as a whole.

To be more specific, the fall in RGDP suggests that Italy was not just experiencing a negative demand shock (which we know from falling NGDP) but was "suffering" from it. If only prices had fallen, Italy would not be "suffering."

Of course a set of graphs does not "prove" anything. But I doubt Grier would have reacted the way he did if the first graph had shown M2 growth falling below trend, and being correlated with deflation and depression. And yet that would be a very similar type of claim. And it would be equally true that a set of 3 graphs showing falling M2, RGDP, and prices did not "prove" anything. RBC economists would have argued for reverse causation.

Outsiders frequently argue that there is something "tautological" about the claim that falling NGDP causes falling output. It's no tautology, and as we saw in Zimbabwe it's not even always true that falling output is associated with falling NGDP. The real issue is this: Which nominal variable should we target to get the best macro outcomes. Old monetarists said the money supply. Mundell might say exchange rates. Some Austrians might say the price of gold. Conservative Keynesians might say inflation. Liberal Keynesians might say a weighted average of inflation and the output gap. Market monetarists say the level of NGDP.

If it were a tautology it could not be tested. Our claim is that if the Fed pegs the price of NGDP futures contracts along a 3%, 4%, or 5% growth path, then RGDP will be more stable in the future than in the past.

Bill Woolsey also has some good comments. This caught my attention:

Finally, we can imagine a Walrasian Auctioneer determining the real output of all the products using an arbitrary numeraire. And all output and resource use is determined. Presumably, we could measure real GDP using any good as numeraire.
We could measure nominal GDP in terms of kilobytes of computer memory. Oddly, this measure of NGDP would not be correlated with RGDP in the way that money NGDP is correlated with RGDP. Ask yourself why. After all, the quantity of kilobytes times kilobyte "velocity" equals NGDP in kilobyte terms.

Comments and Sharing






COMMENTS (17 to date)
James writes:

There is nothing "market" about about market monetarism either, or for any production target recommendation for any nationalized industry.

In the planned economies of the twentieth century, people recommended all kinds of targets for different industries. Whatever the relative merits of those targets, at least no one recommended the shipbuilding industry produce a certain tonnage of ships and then tried to pass the tonnage target off as "market maritime."

As long as the target is determined by some economist rather than by consumer preferences via profits and losses, it's just another central plan.

AD writes:

"The fact that RGDP and M*V are strongly correlated in the US is a stunning, mind-boggling fact that is not predicted by classical economic theory."

Can you expand on this? I'm not seeing why this is mind-boggling if inflation is relatively controlled.

Gordon writes:

The term "NGDP target" seems to distort people's thinking and cause a knee jerk reaction to market monetarism. Instead, what if economists talked about the need to avoid a monetary disequilibrium by adjusting the money supply to offset changes in money demand. I'm no economist so I'll let Scott explain why market monetarism and an NGDP level target IS a plan that uses the market to adjust the money supply so as to avoid a monetary disequilibrium. The free banking concept by Austrian economists George Selgin and Larry White is also a plan that uses the market to adjust the money supply. As far as I know, market monetarism and free banking are the only two plans that use the market to adjust an important macroeconomic variable.

Michael Byrnes writes:

James wrote:

"There is nothing "market" about about market monetarism either, or for any production target recommendation for any nationalized industry.

In the planned economies of the twentieth century, people recommended all kinds of targets for different industries. Whatever the relative merits of those targets, at least no one recommended the shipbuilding industry produce a certain tonnage of ships and then tried to pass the tonnage target off as "market maritime."

As long as the target is determined by some economist rather than by consumer preferences via profits and losses, it's just another central plan."

This would make sense if NGDP was a "production target". But it is not. If NGDP rises 5% next year, that fact alone tells you nothing about the level of real production (RGDP) in the economy, never mind which (and how many) goods are actually produced and sold.

marcus nunes writes:

The Italian French comparison nails it, although Cochrane argues that the fall in inflation in Italy is a supply-side phenomenon!
http://thefaintofheart.wordpress.com/2014/09/02/in-italy-its-supply-induced-deflation/

James writes:

Michael Byrnes:

What part doesn't make sense? I didn't claim that NGDP is a measure of production. I also don't agree with the idea that the weight of shipbuilding output is a measure of economic output in the shipbuilding industry.

My point is that market monetarism is a misnomer. Whatever the target, if it's not established in the market via price signals, profits, losses, etc., it's not "market" anything.

James writes:

Gordon,

You wrote:

"As far as I know, market monetarism and free banking are the only two plans that use the market to adjust an important macroeconomic variable."

All proponents of central banking favor the use of markets in some form. Keynesians favor using open market operations to exchange Federal Reserve Notes for goverment debt. Shall we call them "market Keynesians" now?

What shall we call the wild eyed ones who favor using markets, rather than economists, to determine the targets for interest rates, the unemployment rate, the price level and level of output?

Scott Sumner writes:

James, You don't seem to understand where the term "market monetarism" came from. Google my Mercatus papers on NGDP targeting.

AD. Good point, but the correlation holds even during periods where inflation is not being "controlled," like the 1960s, or 1981-84.

Gordon, That's right. AFAIK, market monetarism is the only proposal that let's market forces set interest rates, the money supply, exchange rates, and the price of gold. All other proposals contemplate the monetary authority controlling one or more of those variables, as a way of controlling broader macro aggregates.

Andrew_FL writes:
Some Austrians might say the price of gold.

Who are these Austrians? It doesn't sound particularly Austrian to me to propose that "We" do anything. At least not as a first best solution. And as far as I'm aware, the ideal Rothbardian monetary "rule" would be fixed M (though I would single out the Rothbardians on this point specifically) with the best approximation being a gold backed dollar with 100% reserves. On the other hand, the Free Bankers would have as an ideal "rule" something more like fixed MV but again this is not a real monetary rule, it's just expected to emerge as spontaneous order.

I believe Hayek talked about targeting an index of producer prices at some point. Then there is the productivity norm-something else Hayek either advocated or came close to advocating (in fact, Keynes came close to this too, advocating at one point targeting an index of wages) with a modern advocate of this idea being George Selgin (although I'm not clear if he thinks this is a first best solution? At any rate, for a constant population it would reduce to the "rule" of Free Banking)

I am not really aware of any Austrian who advocates targeting Gold Prices, especially since the vast majority of them favor defining the price of gold as relates to the dollar.

Michael Byrnes writes:

Andrew, I think that by "targeting the price of gold", Scott means a fixed exchange rate between gold and dollars. There are definitely some Austrians who support that, though none AFAIK who support a central bank.

Andrew_FL writes:

But that is not a target, that's a definition of the monetary unit. targeting implies aiming for a particular value of some index, or a particular growth rate. Fixing the exchange rate of gold to dollars, or more accurately, defining the dollar in terms of gold, is something altogether different from any sort of targeting.

I am trying to wrap my head around who the Austrians Scott had in mind are here, who presumably think that either a decentralized banking system or a monetary authority should manipulate money in such a way as to try and achieve either a particular price for gold or a particular growth rate for gold.

If what he meant to refer to was defining the unit of account in terms of gold, he phrased it very poorly, mis-framing it as a form of gold price targeting. It's a much stricker "rule" than that.

To be clear here, I'm probably more in Selgin's camp than the Rothbardians on this, I just was baffled by Scott's statement and wondered if there was some other schism in Austrian thinking I was unaware of.

Michael Byrnes writes:

Andrew wrote:

"But that is not a target, that's a definition of the monetary unit. targeting implies aiming for a particular value of some index, or a particular growth rate. Fixing the exchange rate of gold to dollars, or more accurately, defining the dollar in terms of gold, is something altogether different from any sort of targeting."

I would disagree.

If you issue money under a gold standard, you want users of that money (i.e. buyers and sellers) to treat that money as if it were equivalent to gold. You achieve this equivalence first by promising to exchange it for the amount of gold specified on the note, and them by following through on your promise. In other words, it's not enough to just state the value of currency you issue, you also must defend that target with gold.

This is similar to the Fed's open market operations. In effect, you are offering to sell unlimited amounts of gold in exchange for the notes you have issued. If you don't do this, then your notes will start to trade at a discount to their face value.

In practice, this would be a major constraint on your ability to issue gold-backed notes.

So to me it is the same thing - a target that must be defended.

Andrew_FL writes:

It's different because, provided there is no breach of contract, one is not aiming for a "target" one is "on" target. Always.

But one is only offering to sell unlimited amounts of gold if one has not limited ones note issues. For a central bank, of course, no such strong limit exists, at least in the short run. But as you said, the Austrians don't want there to be a central bank.

A failure to exactly hit a target is a failure to exactly follow a monetary rule. A failure to redeem notes for gold is a breach of contract. The former is merely to be expected, and cannot be helped. The latter is a matter for adjudication by courts.

Michael Byrnes writes:

Andrew wrote:

"It's different because, provided there is no breach of contract, one is not aiming for a "target" one is "on" target. Always."

I don't think so. If there is some doubt about the credibility of a bank issuing gold backed currency, that currency might trade at a discount in markets. How "on target" a bank's currency is would depend entirely on its credibility.

And of course, if it is a government/central bank maintaining a gold standard, it can do what FDR did in 1933 - devalue. The less credible the government's commitment to maintaining the peg, the more of a discount its currency would trade at - even without any devaluation/breach of contract.

Obviously targeting gold is very different from targeting inflation or NGDP in some ways. But it is still a target.

James writes:

Scott:

I never claimed to know anything about the history or origins of the term market monetarism. So what?

Whatever the history of the idea, it's still a misnomer to use the word "market" to describe a form of central planning where a state run monopoly tries to reach a target chosen by an economist.

Andrew_FL writes:

If we're still talking about what Austrians would want, I think they'd want to prohibit the government from devaluing, or otherwise do things to lose credibility.

That being said, I guess I have a better idea where you are coming from. Yes, under something more like the Gold-Exchange Standard and with a highly untrustworthy Government like under FDR, Gold may not continue to trade at par. This is, however, an unacceptable deviation from the "Austrian" rule, whereas a slight failure to perfectly meet a target is usually acceptable as long as it's not too large.

George Selgin writes:

Kevin Grier has things quite backwards. Nominal income (Y = Py) is a direct measure of nominal dollar flows, that statisticians decompose, perhaps accurately, perhaps not, into P and y components. And the magnitude of nominal income is what's behind every AD schedule ever drawn on a blackboard. In practice, when one speaks of stabilizing AD, one means, or ought to mean, stabilizing Py, that is, the product of P and y, RATHER THAN either P or y itself.

It is those who, implicitly or explicitly, imagine that P and y ought to be introduced individually and additively, albeit after some seat-of-the-pants assignment of weights, into a central bank "loss" function, who have got things badly jabberwockied, losing track in the process of the crucial distinction between fluctuations in "natural" and in cyclical output. The central bank should avoid the latter--which it does by keeping Py (=AD) stable. But it should let "natural" output changes take their (natural) course.


Similar logic holds, pari passu, if one substitutes for y a measure of the unemployment rate.

Comments for this entry have been closed
Return to top