Money is a veil, but when the veil flutters, output sputters.
Leland Yeager
A bit ago, I posted about what money is and why people use it, in the context of video games. A tangentially related idea emerged: is monetary policy neutral? By “neutral” I don’t mean useless; what I do mean is that monetary policy (or changes in the supply of money overall) cannot change real quantities. So if neutrality holds, then money can affect, for example, prices or wages, but it can’t affect the real amount of stuff produced in an economy.
Note: throughout the post the term “real” will be used as narrowly as possible to refer to variables that aren’t nominal.
What are the facts?
I’m surprised by how many people struggle with the fact that you can’t print your way to prosperity, but you can print your way back to prosperity.
Let’s start with a definition: monetary neutrality refers to the idea that changes in the quantity of money can only impact nominal variables, such as prices and (nominal) wages, but not any real variables. We won’t touch on it much, but there’s also superneutrality, which refers to the idea that, whilst money directly doesn’t impact real variables, inflation does. The separation between nominal and real variables is known as the classical dichotomy, and it’s summarized best as “money is a veil”. If monetary neutrality held, what you’d see is that monetary policy is (strongly) correlated with prices but not with output or employment.
What are the facts of monetary neutrality, then? In general, economists agree on three main things concerning neutrality over the long term:
There is a strong correlation between money supply growth and inflation.
There is very little correlation between money growth and real output growth.
There is very little correlation between inflation and real output growth.
This is all over the long term - meaning that printing more money doesn’t actually increase the amount of things the economy produces, which are agreed to depend on the factors of production and on technology rather than on aggregate demand. This is not an uncontroversial view, especially outside of mainstream economics, but in general there’s not a whole lot of evidence against the idea that money is neutral in the long run, and also that (in most cases) it is also superneutral1.
Actually, the first claim is somewhat contested in monetary economics circles, since the evidence for inflation being “always and everywhere a monetary phenomenon” is fairly weak for countries with low, stable inflation - although it is worth pointing out that, definitionally, countries that have inflation under control have very small variations of it, so there would appear to be no correlation between money and prices even if it actually very strongly existed, since changes in policy wouldn’t be reflected as changes in prices since the changes in policy are designed to prevent changes in prices. Using a thermostat to keep your house’s temperature constant versus the weather would mean that interior temperatures are uncorrelated with weather.
However, the long-run neutrality proposition collides face-first with another, seemingly obvious issue: monetary policy actually is very relevant and effective in the short term. Milton Friedman and Anna Schwartz’s now classic 1963 book “A Monetary History of the United States” finds that all major recessions between 1867 and 1960 were associated with a large drop in the supply of money, and that large drops in the supply of money were always associated with a major recession. The fact that loose monetary policy can jump-start the economy and tight monetary policy can pull it under seems to fly straight in the face of the long-run neutrality claims.
Neutrality is a four letter word
Were all the gold in England annihilated at once, and one and twenty shillings substituted in the place of every guinea, would money be more plentiful or interest lower? No surely: We should only use silver instead of gold. Were gold rendered as common as silver, and silver as common as copper, would money be more plentiful or interest lower? We may assuredly give the same answer. (…) No other difference would ever be observed, no alteration on commerce, manufactures, navigation, or interest, unless we imagine that the color of money is of any consequence.
David Hume (1752), “Of Interest”
In his 1752 essay “Of Money”, David Hume proposed a rough version of what we now call the Quantity Theory of Money: an increase in the amount of gold in an economy would only increase prices, but not increase the amount of stuff produced, because all prices (including wages) would adjust equally so there’d be no changes to real variables. He also gives examples straight out of Milton Friedman’s playbook, like “suppose that, by miracle, every man in Great Britain should have five pounds slipped into his pocket in one night”. Then he gets all New Keynesian and introduces sticky prices, and because relative prices adjust at uneven rates, then money isn’t neutral.
Let’s move on to Milton Friedman, the father of Monetarism2. I’ve posted about the Quantity Theory of Money before, but TL;DR: “Inflation is always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of money than in output.” This can be summarized as MV = Py, where M equals the supply of money, V is money’s velocity of circulation, P is the price level, and y is real output. This is always true because, by definition, both sides equal nominal output (or income), though this doesn’t tell you a lot about anything, because M could increase and V could decrease at the same time.
Traditionally, the equation was turned into an actual theory by bounding it through assumptions: V is fixed in the short term but not the long term (because of things like preferences or technology), and M couldn’t affect y in equilibrium (or full employment). Then, in equilibrium, any change in M equals a change in prices, but not otherwise. But hold on - you’re just assuming long term neutrality and allowing short term non neutrality! This is not good scientific procedure.
Friedman had a lot more to say about money, prices, and output, so I’ll keep it brief by mostly focusing on two topics. One way you can (not) get neutrality is through an old friend, the Phillips Curve. Back in Friedman’s day, whether or not the tradeoff between inflation and unemployment was stable or unstable, and over the short or long term, was a huge item of controversy. The “Old Keynesian” view was that it was stable and short term, so it was possible to reduce unemployment through monetary policy - i.e. non neutrality3. The “Old Monetarist” view was that it was unstable over the short term and non-existent in the long term because there was a single long-run equilibrium, resulting in any increases in unemployment going away (what good is an equilibrium otherwise) but higher inflation remaining. So Friedman’s iconic 1968 speech “The Role of Monetary Policy” concludes that, if the “monetarist” Phillips Curve is the true one, then monetary policy is neutral and therefore the Federal Reserve should mostly focus on price stability outside of recessions.
In addition, there’s another (less convoluted) way in which to derive neutrality, which has everything to do with the structure of business cycles. In the 1993 paper “The Plucking Model of Business Fluctuations Revisited”, which reconsiders a hypothesis he proposed in the 1960s, Friedman wonders whether the business cycle can best be understood as a cycle going over and under potential (in real terms), or if it can only go under potential4. There’s major differences between the two models, but the “cyclical” view concludes that monetary neutrality is a tossup (without something like sticky wages or prices, anyways), but the plucking model supports the short term non-neutrality - long term neutrality issue. Exactly whether or not the plucking model is a better representation of the economy is a complicated question, but some evidence seems to point in that direction.
Stick in the mud
... the alterations in the quantity of money, either on one side or the other, are not immediately attended with proportionable alterations in the price of commodities. There is always an interval before matters be adjusted to their new situation; and this interval is as pernicious to industry, when gold and silver are diminishing, as it is advantageous when these metals are encreasing.5
David Hume (1752), “Of Money”
As mentioned several times above, there is a way to get around the messiness of the business cycle: stickiness (also known as “nominal rigidity”, which is both more informative and more boring). This simply means that adjustment in prices or wages is imperfect - people, for some reason or another, can’t raise or cut prices instantly to the optimal value. It’s commonly that downwards stickiness is the one that matters, which means it’s not easy to lower prices or lower wages.
Sticky prices are usually described as “Calvo prices”6, where a random fixed fraction of suppliers in a given market aren’t able to adjust their prices in a given time period. This results in some people losing out in real earnings because it’s not their turn to adjust prices, which means they have to either shut down or fire workers. Thus, a random exogenous shock to demand results in some goods being “too expensive” for consumers, which in turn causes some businesses to run losses, which forces reductions in employment, wages, or both. If prices are sticky, then the distortions they cause results in downturns being worsened, and therefore the central bank should focus on keeping prices stable, so that sticky prices don’t ever adjust anything. Money is neutral in the long run, because eventually all prices adjust, but non-neutral in the short term because it allows for stable price growth that prevents the “no sales, losses, disemployment” channel mentioned above.
The converse view is that wages, not prices, are sticky. When recessions happen, for whatever reason, labor becomes “too expensive” to sustain in the present business environment, leading to people becoming primarily involuntarily unemployed. Unlike in sticky prices models, unemployment doesn’t rise because you were “too stupid and too stubborn” to accept lower wages, but rather because your employer couldn’t afford to pay you the present real wage. Wait - higher real wages? The sticky wages view entails that real wages (at least for new hires) should increase during recessions, which obviously isn’t true, but it does matter where the recession comes from. If the recession starts from a decrease in aggregate demand (normally because of monetary policy), real wages increase; if it starts from a decrease in supply, wages decrease. Obviously, in turn, this means that looser monetary policy can raise nominal wages (like all other prices) and therefore keep the economy stabilized during the short term, but it cannot really increase employment in the same manner over the long term.
You can also believe that all markets function perfectly and still end up with a stickiness of sorts. Imagine a world where people don’t directly observe the average price level, but instead the prices of individual goods. Prices, as we’ve touched upon, convey information about scarcity. If the prices of the goods you make increase - how can you know if it’s because all prices are increasing, or because something specific to your industry is making it more expensive?. Answer is, you can’t, and this reduces real output because firms make bad decisions. Taking out the noise from price signals is very difficult. Normally, this model (called the Lucas Islands Model) is applied to inflation, because the higher inflation is, the more volatile it is, and the more distorted information is, but it very obviously applies to recessions as well: did sales go down because the economy is doing badly, or because people don’t like our product anymore? This results in stabilization policy having short-term benefits.
It’s worth pointing out that, whatever the source of stickiness, it both introduces monetary non-neutrality over the short term and monetary neutrality in the long term, and it also provides a justification for the Phillips Curve, in broad strokes:
Superbad
Now, the elephant in the room: superneutrality. I didn’t mention it before because it’s a very specific can of worms: does inflation have negative real consequences?
Over the short term, superneutrality somewhat controversial, because most models economists use assume that inflation has very small welfare losses, but people really dislike inflation (see Joe Biden’s approval ratings) and countries with higher inflation tend to also do badly in general. Even accounting for inflation distorting real incomes, real wages, and/or relative prices, thus causing real fluctuations, superneutrality is commonly assumed to be a consequence of short term non-neutrality overall.
The guy to look at for long term superneutrality is James Tobin, particularly because he proposed a model where money wasn’t superneutral. In it, there are two investment assets: money, and physical capital. When monetary policy raises inflation, it gives money a negative real return but doesn’t change that of capital (which depends wholly on its marginal productivity), therefore diverting investment away from money and towards capital - i.e. raising long-term output. If you include financial assets, you can probably get non superneutrality too, since the real return of financial assets is also temporarily lower. The effect of inflation on growth is known as the Tobin Effect, and whether or not it exists is a first question - is it also positive?
The first issue with the Tobin effect is that there’s not really any empirical evidence showing that inflation is positively correlated with growth and investment. The second is that its perfectly plausible for inflation to just reduce overall savings to either offset the amount of money that would have been saved, or to even reduce aggregate savings, resulting in no or even negative effects. Thirdly, think back to the Lucas Islands Model: it’s likely that higher, more volatile inflation rates result in higher dispersion of information, and therefore shift money away from investment simply because of uncertainty over returns7. And this is a major nitpick, but the Tobin Effect could be 100% real and still be bad, since the quality, not quantity, of investment seems to be what matters (just ask the World Bank).
The only evidence I’m aware of regarding the Tobin Effect comes from Argentina: if you model asset demand as between financial assets in pesos, assets in dollars, “productive capital”, or real estate, then the optimal portfolio composition under higher inflation is… real estate and US dollars. Not very growth-inducing. Read more here!. But regardless, absence of evidence is not evidence of absence.
Conclusions
What have we learned?
Money is neutral in the long run, but non-neutral in the short term. Money can influence inflation over the long (but not short) run, and vice versa for real variables such as output and unemployment.
The idea that money is neutral is very old, and integral to “Monetarist” beliefs. The Quantity Theory starts as a way to justify the “empirical record” (which didn’t exist back then), but it has to be backed by a more robust theory of macroeconomic fluctuations, such as the plucking model.
Stickiness, either of prices or wages, and noisy signals can both explain the stylized facts of neutrality. If adjustment of prices/wages, or interpretation of price signals, is imperfect, then short term fluctuations can be corrected through monetary policy.
Inflation could be negatively, positively, or not at all correlated with growth. Superneutrality emerges from many models, but the empirical evidence is weaker than for neutrality, and theoretically it is possible for nonsuperneutrality to emerge - in either direction.
Sources
Previous posts on money and the Quantity Theory of Money
The facts
Scott Sumner, “Money neutrality, super-neutrality, and non-neutrality”, 2021
McCandless & Weber (1995), “Some Monetary Facts”
Bullard, J. (1999): “Testing Long-Run Monetary Neutrality Propositions”
De Grawe & Polan (2001), “Is Inflation Always and Everywhere a Monetary Phenomenon?”
The quantity theory view
David Hume (1752), “Of Money”, “Of Commerce” and “Of Interest”
Robert Lucas (1995), “Nobel Lecture: Monetary Neutrality”
Milton Friedman (1976), “Nobel Lecture: Inflation and Unemployment”
Friedman (1968), “The Role of Monetary Policy”
Friedman (1993), “The Plucking Model of Business Fluctuations Revisited”
The complicated theories
Hicks (1935), “A Suggestion for Simplifying the Theory of Money”
Wallace (1998), “A Dictum for Monetary Theory”
Kiyotaki & Wright (1989), “On Money as a Medium of Exchange”
Wallace (1990), “A Suggestion for Oversimplifying the Theory of Money”
Stickiness
Basil Halperin, “It was a mistake to switch to sticky price models from sticky wage models”, 2021
Rotemberg (1987), “The New Keynesian Microfoundations”
Ball & Mankiw (1994), “A Sticky-Price Manifesto”
Superneutrality
Tobin (1967), “The Neutrality of Money in Growth Models: A Comment”
Orphanides & Solow (1990), “Chapter 6 Money, inflation and growth” in the Handbook of Monetary Economics, Volume 1
Burdisso & Corso (2011), “Incertidumbre y dolarización de cartera: el caso argentino en el último medio siglo” (in Spanish)
Burdisso, Corso, & Katz (2013), “Un efecto Tobin perverso: disrupciones monetarias y financieras y composición óptima del portafolio en Argentina” (in Spanish)
You won’t believe which country has strong evidence for non-superneutrality.
Technically it was Irving Fisher but I’m not as familiar with him. And there’s even a “Keynesian” Quantity Theory that has exactly the same implications but substitutes velocity for a money demand function that depends on income.
In more detail, increases in monetary policy would increase inflation but decrease unemployment, and the overall stance of monetary policy was determined by how low the non-inflation accelerating rate of unemployment was, plus the steepness of the tradeoff itself. Not surprisingly, estimates landed on “very low” and “very flat”.
This is a very complicated subject so this post by Bruegel is pretty good at summarizing it. Friedman’s conclusion is, pretty naturally, that his original paper was generally right, and that there is no correlation between the size of an economic expansion and the size of a subsequent recession. He might have even been right!
Ironically enough, New Keynesian David Hume forgets that the same real decreases caused by monetary contractions (i.e. deflation) happens when the price level increases.
There’s a “rival” theory known as Rotemberg pricing, which focuses on the cost of adjusting prices rather than the possibility of doing so. It’s very interesting, though it should be noted they have nearly identical implications for the Phillips Curve, policy, and welfare.
The long shadow of Rognlie (2015) shows up here: “physical capital” can imply things that are actually productive for the economy at large, or just bidding up real estate. The latter is not a positive for the economy (especially if housing faces supply side constraints).
Footnote 1 does link to Argentinean Association of Political Economy but page not found. That may have been the intention rather some paper proving non-superneutrality in Argentina and published by the Association.