Scott Sumner  

Macro events (in big economies) don't have micro causes

In Defense of Dudley... Law Professor Glenn Reynolds D...

A new paper by Vipin Veetil and Richard Wagner discusses a heterodox theory of business cycles, and then uses this theory to criticize NGDP targeting. Here is the abstract:

This paper argues that NGDP targeting is unlikely to produce macroeconomic stability. Contrary to the policy objective, NGDP targeting may increase macroeconomic turbulence. DSGE models that prove the effectiveness of NGDP stabilization policy rest on two assumptions. The first assumption is that macroeconomic volatility is a consequence of exogenous shocks to an otherwise stable system. The second assumption is that central banks can act upon aggregate variables. Neither of these assumptions is accurate. An economy is an ecology of interacting agents, some of whom have rivalrous plans. Macroeconomic volatility is an endogenous and intrinsic feature of such an economic system. Furthermore, central banks act upon some agents in the economic system, not on aggregate variables. The percolation of central bank actions through production networks can impede coordination efforts of economic agents during recessions, thereby increasing macroeconomic volatility.
I read the entire paper and was not able to discover any clear link between the underlying theory and the critique of NGDP targeting. More specifically, it's not clear what would happen if you had replaced "NGDP" with RGDP, or inflation, or M2, or the price of gold, or indeed any other possible monetary policy framework. In other words, when they criticize NGDP targeting I'm constantly asking myself, "Compared to what?" Which monetary regime is better?

For instance, they suggest that adjusting M to offset changes in V will inhibit the necessary re-coordination of economic activity after some part of the economy experiences turbulence. That rules out NGDP targeting. That led me to expect them to advocate money supply targeting, where you don't offset changes in V. But that doesn't appear to be their preferred solution either:

NGDP stabilization, in other words, will impede creative evolution within the ecology of plans of a society. Lavoie (1983) recognized this problem in his paper "Economic Calculation and Monetary Stability". Lavoie argues that a rule of fixed rate growth of money supply is not sufficient to solve problems associated with inflation because inflation necessarily diverts resources from one use to another. Such diversions are problematic because they will tend to produce readjustments in so far as the diversions are inconsistent with the plans that economic agents would have formed in the absence of inflation. Lavoie's view of the economic system is very different from that of the DSGE view. Lavoie (1983, p. 164) sees an economy as a "complex network of production relations"
How can one argue that NGDP targeting is a bad idea without suggesting a benchmark alternative?

I'm also a bit confused as to why they think NGDP targeting would harm an economy:

Macroeconomic turbulence is reflection of microeconomic coordination problems. The decline in velocity of money during recessions does not indicate an increase in the demand for money per se, but merely the postponement of spending as economic agents make new plans to create coordination. Increasing the quantity of money to accommodate the decline in velocity of money as prescribed by NGDP targeting will do little to solve microeconomic coordination problems. DSGE models present a picture of an economy where central banks can directly act upon aggregate variables. From an ecological perspective, central banks act upon some agents within an economic system, who in turn pass the effects of central bank actions to other agents through production relations.
Here it seems like words are getting in the way of meaning. I don't doubt that economists sometimes talk about monetary policy directly impacting NGDP, but surely they always have in the back of their minds the idea that it must first impact the behavior of individuals. NGDP aggregates prices and output, and thus NGDP cannot change unless at least some individuals adjust their output or pricing decisions. Does anyone disagree?

Similarly, what's the point of denying that a decline in V represents an increase decline in money demand? It may occur for various reasons, but it is what it is.

Proponents of NGDP stabilization point to such abrupt downturns in the growth of nominal GDP as occurred in 2008 as something that could be eliminated by having the Fed stabilize nominal GDP. We do not doubt that a sufficient monetary expansion could offset any decline in nominal GDP, provided only that V remains positive. To conclude that nominal stabilization implies real stabilization, however, is possible only if monetary change is always neutral.
This is exactly backwards. People who favored a more expansionary monetary policy in 2008 did so precisely because we believed money is non-neutral in the short run. Back in late 2008, I recall speaking with a new classical economist who opposed monetary expansion precisely because he thought it would merely lead to higher inflation. Economists who believe money is neutral typically oppose NGDP targeting.

The primary problem I have with this paper is that their theory of the business cycle doesn't seem consistent with the empirical evidence. They suggest that economic dislocation in one part of the economy can have a sort of "cascading effect" on the overall economy, because the actions of economic agents are interrelated. They criticize the standard view, which is that in a large diversified economy the impact of micro level disturbances tend to balance out:

Economists for many years believed in the dictum that small changes at the micro level cancel out to produce little or no change at the macro level. In Lucas's (1977, p. 20) words the cancelling of small changes in a large system is "the most important reason why one cannot seek an explanation of the general movements we call business cycle in the mere presence, per se, of unpredictability of conditions in individual markets".
In contrast, I see the business cycle as being (in Fisher's words) a "dance of the dollar". Unstable monetary policy shows up as unstable NGDP. Since wages are sticky, employment tends to move with NGDP in the short run, and unemployment is countercyclical. Recessions occur when a sharp decline in NGDP growth leads to a rise in unemployment:

Screen Shot 2017-09-09 at 3.19.59 PM.png

Screen Shot 2017-09-09 at 3.19.46 PM.png
Now of course this doesn't prove that stabilizing the growth rate of NGDP would help to smooth out the business cycle, but it's certainly better than an alternative theory that seems completely at variance with the empirical data. Thus the March 2011 Japanese tsunami is often considered the single most disruptive real shock to hit a major economy in decades. It disturbed all sorts of supply lines in Japanese manufacturing, and closed down the entire nuclear power industry for years. The media were full of reports that Japan would be driven into recession, and indeed industrial production did decline for a few months. But unemployment was unaffected because it's NGDP, not real shocks that drives the unemployment rate:

Screen Shot 2017-09-09 at 3.27.49 PM.png
In contrast, unemployment spiked when the BOJ allowed NGDP to decline in 2009.

A large diversified economy is going to have a relative stable labor market, unless:

1. The central bank has an unstable monetary policy (unstable NGDP).
2. The national government disrupts the aggregate labor market with unwise regulations, such as an extremely high minimum wage.

HT: Ben Klutsey, Patrick Horan

Comments and Sharing

COMMENTS (11 to date)
Jerry Brown writes:

I read a lot of 'heterodox' economists, they just are from the opposite side (if I can characterize it that way) being as they are labelled as post-Keynesian or MMT types. My guess is that the ones I read don't have any problem with NGDP level targeting, just that they doubt it will always be effective. But most of them doubt that ANY monetary policy will always be effective enough by itself. Which is fine with me because that is no reason not to try NGDPLT and see if it works better.

I think your question- "How can one argue that NGDP targeting is a bad idea without suggesting a benchmark alternative?"- is a very good one. Bad compared to what?

Trevor Adcock writes:

Works like this are ridiculous. Monetary shocks can only distort relative prices if some prices are stickier than others. However, price stickiness is all we really need to assume to get nominal shocks to have real effects as the vast New Keynesian literature shows.

Rajat writes:

Just a typo:

Similarly, what's the point of denying that a decline in V represents a decline in money demand?
I presume you mean " increase in money demand".

Jon Murphy writes:

@Scott Sumner

In other words, when they criticize NGDP targeting I'm constantly asking myself, "Compared to what?" Which monetary regime is better?

I'm wondering if you could expand on this point. Why would Vipin and Dr. Wagner need to propose an alternative? If their goal is to evaluate NGDP targeting, isn't it sufficient to only critique NGDP targeting? By way of metaphor: assume someone says "minimum wage legislation improves employment prospects for the poor." I present all kinds of evidence showing that it is not true; that minimum wage harms employment prospects for the poor. He responds "well, what would you do instead?" My goal was to critique his argument, not make one of my own, wasn't it?

Also, as a Chicago-style economist with Austrian sympathies, your title intrigued me. But that's just my microeconomist bias coming out :-)

Tiago Irineu writes:

@Jon Murphy

I don't think that your point is a valid one because minimum wage laws have a default comparison option, that is, the absence of these laws. In the case of NGPD targeting this is true. You always need to compare it with another approach to monetary policy, if you don't do this the criticism would be meaningless because I don't think that the point of whom advocates NGPD targeting is that it is the perfect monetary policy, but that in comparison to the other policies, it is the best choice.

PS. Sorry for my poor English, I'm still learning how to write in English.

Jon Murphy writes:


Thanks for your response. I understand your point. It is a good point I'll have to consider.

Scott Sumner writes:

Rajat, Thanks, I fixed it.

Jon, If you criticize a $15/hour minimum wage, it would be helpful to know whether you prefer a $30/hour minimum wage, or no minimum wage at all.

Presumably they don't favor barter, so if they want to have a monetary regime, which sort of monetary regime do they think is superior to NGDP targeting?

Jesper writes:

@Scott Sumner: I don't disagree with your main point that NGDP (level) targeting would likely be better at stabilizing the macroeconomy than the current regime, but I do think the headline of this post is misleading.

A lot of micro events have aggregate consequences. Lehman was a micro-event that sparked the Great Recession, through global interconnections in the financial system (yes - the Fed could have helped offset the adverse shock, but the point here is that the trigger could have micro foundations). On the real side, there are several studies that illustrate this, see for example Gabaix (2011, Econometrica) or Acemoglu et al (2012, Econometrica).

Jon Murphy writes:

@Scott Sumner:

Thank you. I see your point now

Scott Sumner writes:

Jesper, Lehman occurred 9 months after the Great Recession began and 3 months after it began to get much more intense. I think you've reversed causation.

SaveyourSelf writes:

Scott Sumner,

I read your rebuttal and I think it is your best defense of NGDP targeting to date, at least of those I’ve read. Also, it seems as if you’ve thrown down a gauntlet: ‘if you don’t like it, offer something better.’ I’ll take you up on that challenge. Here’s my best effort:

To begin with, your choice of title—“Macro events (in big economies) don’t have micro causes.”—is interesting. It’s intriguing because it is, I think, simultaneously, true and false.

As a matter of fact, “macro events don’t have micro causes,” is false. All macro events are the result of micro transactions—by definition. And that simple fact has important implications. From what I can tell, THE major error we all make when considering the economy is the way we consider the macro without simultaneously considering all the micro transactions that make it up. The reason we all make this mistake is simple and humbling—because we have too.

We are designed, as human beings, to absorb information from our local universe and process it. We are also blessed with a limited capacity to absorb information through other individuals using speech and text and pictures and symbols and empathy, etc. But, as the number of layers of people encircling a person expands beyond about two or three, the amount of information in the system becomes, literally, infinite compared to the sensory and processing power of any individual in the system. That’s a tremendous problem when an individual—like a macro-economist or a politician—is trying to make decisions for the whole system and everyone in it. He can’t.

Given such staggering limitations, how can we hope to manage large quantities of information? Well, the quick answer is, we can’t. But we try. We summarize, compartmentalize, generalize, model, and guess—all reductionist activities. The idea being, if we can systematically, mathematically reduce the amount or complexity of the information, then perhaps we can understand and manipulate it reliably. Unfortunately, all those reductions are mathematical exercises that attempt to reduce infinity. Divide the economy into three parts so it is more manageable—like lower, middle, and upper—and you will have accomplished nothing because infinity ÷ 3 = infinity. It’s not until you decrease the size of the society back down to an individual and a rather small number of people around him that infinity drops out of the equation and we can actually gather and work with the information available using the tools at our disposal.

Which leads nicely into the topic of NGDP targeting. My interpretation of the blogpost is that Veetil and Wagner are simply pointing out the pattern I’ve described above. What they didn’t point out, apparently, to anyone’s satisfaction, is what would work better. I can think of only one conclusion that makes sense given the model described above: monetary policy would have to be managed at very local levels. In such a de-centralized case, no single currency would exist and competition would ensue such that not all currencies would survive. Those surviving currencies would likely have the greatest collection of features that individuals desired in a currency medium. So if we are looking for the optimal central bank policy for minimizing the size and number of monetary-corrections, it is to abolish central bank monetary policy in favor of independent, competitive, micro-level monetary policies.

Unfortunately, central bank errors are not the only cause of massive monetary value fluctuations. It is now, in retrospect, obvious that regulations, subsidies, and tax policies can have enormous influence on the value of money, especially through their influence on banking practices. In short, the best you can hope for from any central bank policy is to not exacerbate problems. Such a hope is futile, however, given that the information the central bank would need to process to satisfy that “modest” objective—or any objective—is infinitely large. We could stop there. But you don’t. I get the impression that you think NGDP targeting by the central bank could also counteract errors made by the other branches of the government—branches whose members are also effectively blind due to the volume of information they are attempting to process and the tools they have available to process it. The politicians, like the monetarists, are blind to their blindness. It’s the blind leading the blind, no matter which way you look at it.

But back to the title: Could people using Macro instruments—read: monopoly, monetary tools—cause macro events like recessions or prevent them? Obviously they can cause recessions, and if they avoided causing them they are, in a fashion, preventing them. But could the Fed, say, ameliorate a mess started by the legislature? I think the answer, given the limits of knowledge detailed above, is no. The Fed and the legislature both produce macro outcomes by introducing coercion [violence] in to micro transactions. Don’t think Fed policy is an exercise of violence? Try printing your own money. See who comes to visit. Is it possible that a small amount of surgical violence added to every micro transaction could make the whole system work better? Probably not. Is it likely that the whole system would work better with less violence in each micro transaction? Probably, which is why I think the title, “Macro events (in big economies) don’t have micro causes,” is—also—true. If you remove macro policy, you remove the major and most common cause of economic instability.

The key to effective monetary policy—and nearly every other macro policy for that matter—is to avoid attempting to operate at the macro level. It is a fool’s errand.


Return to top