In the latest installment in his series, "Understanding Money Mechanics," Bob Murphy takes on Market Monetarism, and Scott Sumner's case for having central banks practice NGDP level targeting in particular. A commentator there writes, "I hope George S. pipes up to defend MM! Seeing the other side can helps [sic] me to understand the theory better."
Far be it from me to refuse such a request!
Murphy devotes much of his post to distinguishing Market Monetarism from both old-time Monetarism and Austrian monetary economics. Much of what I have to say also concerns those distinctions. I hope to persuade readers that Market Monetarism is more consistent with both old-fashioned Monetarism and Austrian economics than Murphy allows, and that, to the extent that it differs from versions of either, it does so in ways that improve upon them.
Velocity: The Elephant in the Room
Murphy's essay begins with a generally accurate summary of the many points of resemblance between Sumner's Market Monetarism and Monetarism of the old-fashioned Milton Friedman sort, from which Sumner draws much of his inspiration. Among other things Murphy notes, correctly, that both Sumner and Friedman reject the naive view that the level of nominal interest rates is a reliable indicator of the stance of monetary policy, with low rates serving as evidence that money is "easy," and high ones suggesting that it's "tight."
Murphy's summary falls short, however, when it comes to explaining the difference between Sumner's framework and Friedman's. He recognizes that Sumner's "proposal to use a futures market in NGDP contracts to effectively automate Fed policy" supplies the "Market" aspect of Market Monetarism. And he of course understands that Sumner differs from Friedman in regarding the value of nominal GDP as a better indicator of the stance of monetary policy, and hence as a more fitting object of the Fed's stabilization efforts, than any money-stock measure. But he never considers the rationale for this second, more crucial difference, and his failure to do this vitiates a substantial part of his critique.
That rationale couldn't be simpler. It's that the velocity of money—the ratio of nominal GDP to the money stock, which is an inverse measure of the intensity of people's desire for "real" (that is, purchasing-power-adjusted) money balances—is unstable. There are times when people want to stock up on real balances, just as there are times when, for reasons good or bad, they want to stock up on toilet paper. And there are others when they'd rather not hold onto money at all, as happens when hyperinflation burns a hole in their pockets.
Friedman and other old-fashioned Monetarists long took the predictability of the velocity of money for granted, treating it as a stable function of a relatively small set of variables. But the merit of that procedure fell into doubt during the 1970s, and since then velocity has become notoriously unstable. Consider, for example, how the behavior of the M2 velocity has changed since 1960:
The story for M1 velocity is similar: although it was never constant, until the early 80s it seemed to grow at a modest, steady rate. But since then, its movements resemble those of a roller coaster:
So long as the velocity of money was itself stable, it made sense to suppose that some correspondingly stable money stock growth pattern would prevent monetary policy from becoming either too easy or too tight, and to regard any substantial deviation from the same pattern as evidence of over-tightening or its opposite. But when, by the 1980s, velocity had ceased to be stable, this was no longer so. From then on, whether changes in the level or growth rate of some money measure signified over- or under-tightening depended on whether they compensated for an opposite velocity change. In terms of the famous equation of exchange, MV = Py, what mattered wasn't whether M grew at a modest and steady rate, but whether MV or, equivalently, Py, did so. Nominal GDP is, of course, a popular measure of Py. Nominal GDP targeting can thus be understood as nothing more than a velocity-compensated version of Friedman's famous money growth rate rule.
Money Growth Measures Mislead
That the velocity of money ceased being stable has for some decades now been recognized by most self-described Monetarists, who stopped recommending constant money growth rate rules for that reason. Prominent early examples included Ben McCallum and John Taylor. So it's only natural that Sumner and other "Market Monetarists" today would no longer treat the behavior of any money-stock measure as a guide to the stance of monetary policy. Nor of course would most non-Monetarists wish to treat it so.
Still, there are exceptions; and Murphy is one of them. He writes as if it were still 1960, and as if the pace of money growth alone could be treated, as old-fashioned Monetarists once treated it, as a sufficient indicator of the tightness or looseness of monetary policy. Remarkably, the word "velocity" never even comes up in his essay.
Thus, of four "problems" Murphy has with Market Monetarism, the first consists of its alleged failure to recognize that "Monetary Growth Accelerated after the 2008 Crisis." According to Murphy, that growth proves that Sumner and other Market Monetarists are wrong to think that money was tight during the recession and subsequent, slow recovery.
But as we've seen, Murphy's claim makes sense only assuming that the velocity of money hasn't decelerated enough to offset, or more than offset, the more rapid money growth he documents. And that assumption is clearly wrong, for otherwise the level and growth rate of nominal GDP themselves would not have declined! In short, Murphy faults Market Monetarists for failing to treat a stable money growth rate as the key to sound monetary policy, when the whole point of Market Monetarism is to improve upon that long outmoded perspective!
It gets worse. For Murphy also treats the rapid post-October 2008 growth of the monetary base as ipso facto evidence of easy monetary policy. That he does so makes one wonder, not only whether he appreciates the need to account for velocity changes, but whether he's aware of the important changes to the Fed's operating procedures that anticipated its resort to quantitative easing. In particular, he seems to overlook how, in October 2008, the Fed began paying interest on bank reserves with the deliberate aim of encouraging banks to hoard newly-created reserves that came their way, thereby severing the traditional link between base money growth and growth in broader money measures. Just how well this strategy worked isn't evident from Murphy's charts. I trust this one will make it so:
Despite the change just described, both the M2 and the M1 money stock did grow somewhat more rapidly after 2008 than they had beforehand. But owing to a concurrent decline in velocity, the increase fell short of what was needed to avoid lackluster growth of total spending on final goods, let alone make up for an initial, absolute spending drop. Here's the relevant chart from M1. (The one for M2 tells the same story.)
Murphy's second complaint about Market Monetarism is that its NGDP criterion for judging the stance of monetary policy is "vacuous and (almost) non-falsifiable." "Sumner argues that by definition," he says, the 2008-9 decline in NGDP represented "a 'tight' central bank policy." "The fundamental problem with this definition," he adds, "is that it assumes Sumner's conclusion."
I confess that I can't make much sense out of Murphy's complaint. Yes, Sumner and other Market Monetarists regard the behavior of NGDP as a reliable indicator of the stance of monetary policy. Accordingly, when that value falls significantly below trend, they conclude that monetary policy is too tight; and when it rises above the same trend, they consider it too easy. It's also true that, were there nothing more to Market Monetarism, it would indeed be "vacuous." But that's hardly the case: Market Monetarists' treatment of NGDP as a reliable indicator of the stance of monetary policy isn't a mere article of faith, as Murphy suggests. It's a position they've arrived at on the basis of a substantial body of theory and evidence.
Much of that theory and evidence can be found in works by Market Monetarists themselves, including several by Sumner and David Beckworth that Murphy himself refers to. More can be found in various other works, including several I surveyed here not long ago. Being myself a Market Monetarist fellow traveler, and even a pioneer of sorts, I've contributed to its foundations myself, especially in my book Less than Zero, which argues for a particular stable-NGDP norm that would typically allow for modest, secular deflation.
In light of all these writings, it's absurd for Murphy to portray Sumner's case for treating NGDP as an indicator of the stance of monetary policy as a mere case of circular reasoning. Still, he tries, not by seriously considering the arguments and evidence to be found in the aforementioned writings, but by means of the following "medical analogy":
Suppose a patient is suffering from fever, running a temperature of 103 degrees. One group of doctors recommends injecting the patient with substance M, in order to cure the fever. Yet another group of doctors argues that past injections of substance M are what made the patient sick in the first place.
Now, if they are to have any hope of resolving this dispute, how should the doctors measure the amount of substance M being injected into the patient? Most people would argue that the doctors should look at absolute physical measurements, involving the volume and/or rate of injection. And so, for example, if they injected the patient with more M than had ever been administered to any patient in the history of that hospital, it would be odd if the patient's chart read, "Received a very restrictive treatment of M."
Indeed, imagine if the doctors who think that substance M is a helpful medicine wanted to define the M treatment in terms of the fever. That is, if after they had injected the patient with unprecedented amounts of M, whether the fever stayed the same or went up, the doctors were to declare, "We just made the patient sicker with our shift to restrict the M treatment." This would be Orwellian and obviously would make it virtually impossible to figure out whether more or less M was what the patient needed.
Analogies are necessarily imperfect. But some are worse than that; and this one is just plain foolish, for all sorts of reasons.
The issue, first of all, is not how to define "M." Market Monetarists define it the same way, or ways, as other economists do. They deny, however, that a fixed M regimen is always the right treatment for every possible condition. Some conditions call for more M, while others call for less.
To understand the Modern Monetarist perspective, consider a slightly different—but I dare say less labored—version of Murphy's analogy. A doctor specializing in diabetes treats every patient who comes to his clinic with a fever to the same regimen of 100 units of insulin per day, without considering other symptoms or test results. Such a doctor would surely count as a quack, for not every patient with a fever (which can be a symptom of either hyper- or hypoglycemia, or of any number of other things) requires such a treatment: on the contrary, a hypoglycemic might well die from it. A responsible doctor, in contrast, would first determine a patient's blood sugar level, and would only then decide how much insulin to give him, assuming he needs any. In this analogy, insulin is like money, a 100-unit regimen is like Friedman's naive k-percent money growth rule, and a patient's blood sugar level is like an economy's NGDP level: it depends both on preexisting insulin input and the patient's particular insulin needs.
We thus return yet again to the subject of velocity. Like a naive Monetarist, our quack assumes that it's constant. And Murphy sides unabashedly with the quack.
Austrian and Monetarist Busts
If the assumption that an economy always requires the same dose of monetary medicine is dangerously naive, the view that any dose of monetary medicine is harmful is downright puerile.
Yet that view is implicit in Murphy's final criticism of Sumner, which is that his recommended solution to a collapse of NGDP, namely, a Fed-engineered boost to money growth, would "simply perpetuate the boom-bust cycle."
Although I stopped calling myself an Austrian economist long ago, I'm pretty familiar with the Austrian business cycle theory to which Murphy's complaint alludes. What's more, I agree that it can help explain the boom-bust cycle of the last half of the 2000s. Still, I reject Murphy's suggestion that a central bank following Sumner's advice would be bound to trigger an Austrian-type cycle.
Why wouldn't it? According to the Austrian theory, excessive money growth distorts relative prices, in part by leading to artificially low-interest rates, and this gives rise to booms. But the theory also holds that relative prices, and the rate of interest in particular, are bound to eventually return to their undistorted or "natural" levels, and that as they do, booms give way to busts. In other words, once the bust begins, the economy starts with a clean slate, so far as the efficient working of its price system is concerned: there's absolutely no need for any contraction of either the money stock or total spending (money times velocity), or for deflation, to restore relative prices to their proper levels. The opposite view—that a bout of deflation somehow helps the economy to rid itself of prior malinvestment—was quite properly scorned by von Mises as akin to the view that one might make up for running over a neighbor's cat by backing over it as well.
So what money growth rate—what dose of "money medicine"—does the economy require going forward? I agree with Murphy that the last thing the economy needs is another boom-generating dose. But monetary expansion is warranted so long as it only serves to maintain a stable and modest level of spending. Because it's common for money's velocity to decline, even drastically, during busts, that needed dose might be substantial.
Avoiding a Depression Double-Whammy
Monetarists, "Market" ones included, tend to downplay the importance of Austrian-style boom-bust cycles, while some Austrians dismiss the Monetarist theory that busts are caused by money shortages. This is a shame, because it often takes both theories, and then some, to explain any actual cycle. I'm pretty sure that's so for both the 1927-33 and the 2005-10 cycles. In each of these episodes, a period of overly-loose money was followed by one in which money was too tight, adding the insult of deflation to the injury of malinvestment.
When I used to teach undergraduates about the Great Depression, I made this point using yet another (semi) medical analogy, after having first covered both the Austrian and the Monetarist explanations.
Suppose that a man has the bad habit of coming home very late, and badly hungover, every Friday night. His wife has had to put up with this for years. One Friday, he stumbles through their apartment door, and immediately starts to throw up. His wife has had enough: she wallops him with a frying pan. A neighbor, hearing the ruckus, runs in from across the hall, to find the man lying unconscious, in a pool of his own vomit. "What happened!?," she asks.
Must the answer be either "he's hungover" or "he's been hit by a frying pan"? Can't it be both?
Yes, it can. And the correct answer to the question, "What caused the Great Depression?," I said to my students, is "both monetary excess in the late 20s, and a severe monetary shortage in the early 30s.
The correct answer to the question, what was ailing the economy in late 2008 and 2009, is likewise that it was suffering from both the after-effects of easy monetary policy between 2004 and 2007 and a money shortage afterward.
Hair of the Dog?
Murphy, however, argues that even monetary expansion aimed at preventing a collapse in NGDP is harmful, because it distorts relative prices, inviting another boom-bust cycle:
if the Austrians are correct, then if the Fed reacts to a downturn (which normally would go hand-in-hand with a fall in NGDP growth) with monetary expansion, then, besides the impact on aggregate nominal variables, this action will also distort relative prices. In particular, short-term interest rates will typically be pushed below their "natural" levels, giving the wrong signal to entrepreneurs and setting in motion another unsustainable boom.
This claim has now become commonplace among certain Austrians. Yet far from being sound, it shows a very poor understanding of basic monetary theory.
Suppose, for example, that I want to increase my money holdings. To do so, I just have to spend less on goods and on non-money assets. Then, provided the monetary system is working properly, my nominal money holdings can accumulate.
The rub is that, for any given stock of money, I can only increase my money holdings if others reduce theirs by the same amount. It follows that, if the overall demand for money goes up, either prices have to fall to make every unit of money worth more, so that we all can once again rest content with what we've got, or the total stock of money must increase. In the second case, the Fed generally has to help, either by supplying more currency and bank reserves, or by otherwise lowering interest rates to encourage more commercial bank lending and borrowing. Of course, the new money thus created doesn't itself go to those who wish to increase their money holdings. But it doesn't have to: the mere fact that it is created at all allows those persons to collectively gather to themselves the extra money they want, as they wouldn't be able to do otherwise. Nor is there anything perverse about this. On the contrary: it's just what one would wish to see in any system in which money consists of claims on financial intermediaries. Those who accumulate such claims—like me if I let my bank deposit grow—supply that many more savings to the banks that issue them; and in an ideal system those banks will in turn generate more claims. The Fed's job is to see to it that this ideal is achieved, at least approximately.
It follows that some monetary expansion and the lowering of interest rates that may accompany it needn't imply any "artificial" lowering of rates. Instead, it can reflect a genuine increase in savings. This doesn't mean that interest rates always need to decline when the demand for money goes up, for people might choose to hold more money, not as an alternative to spending it on final goods and services, but as an alternative to other sorts of saving. Either way, monetary expansion isn't like a "hair of the dog" remedy for a hangover. It's more like making sure that the hangover isn't compounded by dehydration. Money may resemble alcohol in being able to spur reckless behavior that eventually results in a painful reckoning, but it also resembles water in being absolutely essential to an economy's health.
There is, by the way, nothing particularly un-Austrian about the arguments I've just made. Similar ones can be found in Hayek's writings of the mid-1930s, and especially in his 1933 essay on "Saving," reprinted as chapter five of Profits, Interest and Investment. They occur as well in chapter twelve of Fritz Machlup's 1940 work, The Stock Market, Credit, and Capital Formation. The same ideas take up a large chunk of chapter four of The Theory of Free Banking, which I wrote while I was still a very Austrian grad student, and a similarly large chunk of Steve Horwitz's Microfoundations of Macroeconomics.
The case of von Mises is more complicated. While certain passages of his appear to treat any growth in the quantity of "fiduciary" money as harmful, others suggest on the contrary that such growth can help to keep relative prices where they belong. Thus he observes, in The Theory of Money and Credit, that insofar as banks "increase and decrease their circulation pari passu with the variations in the demand for money… they make an essential contribution to stabilizing the inner objective exchange value of money." As Jörg Guido Hülsmann explains, Mises regards "a stable inner objective exchange value of money" as a monetary policy ideal, albeit one that's unobtainable in practice.
1920 and 2020
In fact, Murray Rothbard and those who cleave to his teachings, including Murphy, are the only Austrians—if not the only economists of any school—who deny that monetary expansion can be beneficial even when it matches corresponding growth in the demand for real money balances. Their dissidence has several roots, one of which is their belief that, instead of being rigid or "sticky," prices are perfectly capable of quickly adjusting downward in response to such shortages. If prices respond quickly enough, people can always have all the real money balances they want, without delay, even if the money stock never grows.
Bob Murphy is nothing if not forthright when it comes to pleading this Rothbardian case, as he does in his essay's closing paragraph. "The entire 'sticky prices' boogeyman," he says,
is a red herring. During the 1920–21 depression, consumer prices collapsed more rapidly than in any twelve-month stretch during the Great Depression. Yet the 1920s were not a decade of economic stagnation. Blaming the worst economic crises in US history on "deflation" and "sticky prices" doesn't fit the facts.
Let's set aside the burning question of whether a herring of any color can qualify as a boogeyman. Instead, let's just ask whether the 1920-21 episode really shows what Murphy thinks it shows, namely, that prices aren't really sticky, so that there's never anything to be gained from an expanding money stock.
The answer is that it shows no such thing. First of all, although prices fell relatively quickly during the 1920-21 recession, it was a recession nonetheless, and a serious one. It lasted a year and a half, which was longer than most post-WWI recessions, and it involved a substantial decline (6.9 percent) in output—the largest between the Panic of 1873 and the Great Depression. Unemployment rose sharply, to somewhere between 8.7 and 11.7 percent, depending on which statistics one consults, to roughly twice its level for the remainder of the decade. Although it's true that the relatively rapid decline in prices prevented a still longer recession, it's also obvious that they weren't flexible enough to prevent a serious, albeit temporary, shortage of money with its inevitable counterpart: a glut of goods and labor.
Although the U.S. was on a gold standard when it entered the World War, a gold export embargo soon put that standard in abeyance, allowing both the U.S. money stock and U.S. prices to rise substantially. After the war, as the U.S. and other belligerent nations resumed gold payments, it was understood that prices would have to come down again. This supplies another reason for doubting the general relevance of the 1920-21 episode, for actual prices are more likely to keep up with their equilibrium counterparts, when equilibrium price changes, instead of coming as a surprise, are expected. The rapid downward adjustment of prices in 1920-21 was in this respect at least atypical, for in other episodes, including both the Great Depression and the slump that began in 2007, no one had been anticipating a major decline in prices or the rate of inflation.
Finally, although it should go without saying, things have changed since 1921, including the extent of price and wage stickiness. Indeed, they'd already changed substantially by the end of the 1930s—an era that witnessed numerous initiatives aimed at discouraging or prohibiting downward changes to prices and wage rates. These initiatives began with Hoover's "high wages" campaign, by which he sought to convince businessmen to stick to paying high wage rates on the dubious grounds that doing so would allow workers to spend more. They continued with the various price- and wage-control programs of FDR's first New Deal, and also with the 1935 Wagner Act. Finally, they culminated, in 1938, in the nation's first minimum wage law.
As those last two measures mentioned illustrate, some of those depression-era measures remain in place to this day; and as a vast empirical literature suggests, they together with a host of other factors—whether legislative, contractual, or psychological—continue to limit the extent of downward nominal price and (especially) wage adjustment that occur in response to spending downturns in modern economies.
Murphy has himself written eloquently of the destructive effects of minimum wage increases; and he presumably understands that the same lawmakers responsible for such increases have never so much as considered reducing the minimum wage in response to an economic downturn. In light of this, and of the aforementioned empirical evidence, can he really be capable of such a stroke of cognitive dissonance as would allow him seriously to maintain that wage rates and prices are perfectly flexible?
Alas, I suspect he can. But that's no reason why anyone else should.
 So, back in 2008, were David Henderson and Jeff Hummel, whose outdated Monetarist metrics I criticized at the time. Using it they concluded that Greenspan's Fed contributed little if anything to that era's housing boom. I can't help wondering now what Bob Murphy thinks of this old exchange!
 These depression-era impediments to downward price and wage adjustments are nowhere better described than in the historical chapters of Rothbard's America's Great Depression. That makes it particularly perplexing that, in its opening chapters, the same work treats concerns about downward wage rigidity as a question-begging obsession of "Keynesian" economics, dismissing them—and arguments for preserving a stable level of overall spending on final goods—accordingly. "The Keynesian linkage of total employment with total monetary demand for products," Rothbard writes, "implicitly assumes rigid wage rates downward; it therefore cannot be used to criticize the policy of freely-falling wage rates." "Policy"? The question isn't whether it would be best if wages fell as needed. It's whether they can fall that far in fact. Generally speaking, they can't.