In December 1967, Milton Friedman delivered his Presidential Address to the American Economic Association in Washington DC. In those days the AEA met in the week between Christmas and New Years, in contrast to the more recent practice of holding the convention in the week after New Years. That’s why the anniversary of Friedman’s 1967 address was celebrated at the 2018 AEA convention. A special session was dedicated to commemoration of that famous address, published in the March 1968 American Economic Review, and fittingly one of the papers at the session as presented by the outgoing AEA president Olivier Blanchard, who also wrote one of the papers discussed at the session. Other papers were written by Thomas Sargent and Robert Hall, and by Greg Mankiw and Ricardo Reis. The papers were discussed by Lawrence Summers, Eric Nakamura, and Stanley Fischer. An all-star cast.
Maybe in a future post, I will comment on the papers presented in the Friedman session, but in this post I want to discuss a point that has been generally overlooked, not only in the three “golden” anniversary papers on Friedman and the Phillips Curve, but, as best as I can recall, in all the commentaries I’ve seen about Friedman and the Phillips Curve. The key point to understand about Friedman’s address is that his argument was basically an extension of the idea of monetary neutrality, which says that the real equilibrium of an economy corresponds to a set of relative prices that allows all agents simultaneously to execute their optimal desired purchases and sales conditioned on those relative prices. So it is only relative prices, not absolute prices, that matter. Taking an economy in equilibrium, if you were suddenly to double all prices, relative prices remaining unchanged, the equilibrium would be preserved and the economy would proceed exactly – and optimally – as before as if nothing had changed. (There are some complications about what is happening to the quantity of money in this thought experiment that I am skipping over.) On the other hand, if you change just a single price, not only would the market in which that price is determined be disequilibrated, at least one, and potentially more than one, other market would be disequilibrated. The point here is that the real economy rules, and equilibrium in the real economy depends on relative, not absolute, prices.
What Friedman did was to argue that if money is neutral with respect to changes in the price level, it should also be neutral with respect to changes in the rate of inflation. The idea that you can wring some extra output and employment out of the economy just by choosing to increase the rate of inflation goes against the grain of two basic principles: (1) monetary neutrality (i.e., the real equilibrium of the economy is determined solely by real factors) and (2) Friedman’s famous non-existence (of a free lunch) theorem. In other words, you can’t make the economy as a whole better off just by printing money.
Or can you?
Actually you can, and Friedman himself understood that you can, but he argued that the possibility of making the economy as a whole better of (in the sense of increasing total output and employment) depends crucially on whether inflation is expected or unexpected. Only if inflation is not expected does it serve to increase output and employment. If inflation is correctly expected, the neutrality principle reasserts itself so that output and employment are no different from what they would have been had prices not changed.
What that means is that policy makers (monetary authorities) can cause output and employment to increase by inflating the currency, as implied by the downward-sloping Phillips Curve, but that simply reflects that actual inflation exceeds expected inflation. And, sure, the monetary authorities can always surprise the public by raising the rate of inflation above the rate expected by the public , but that doesn’t mean that the public can be perpetually fooled by a monetary authority determined to keep inflation higher than expected. If that is the strategy of the monetary authorities, it will lead, sooner or later, to a very unpleasant outcome.
So, in any time period – the length of the time period corresponding to the time during which expectations are given – the short-run Phillips Curve for that time period is downward-sloping. But given the futility of perpetually delivering higher than expected inflation, the long-run Phillips Curve from the point of view of the monetary authorities trying to devise a sustainable policy must be essentially vertical.
Two quick parenthetical remarks. Friedman’s argument was far from original. Many critics of Keynesian policies had made similar arguments; the names Hayek, Haberler, Mises and Viner come immediately to mind, but the list could easily be lengthened. But the earliest version of the argument of which I am aware is Hayek’s 1934 reply in Econometrica to a discussion of Prices and Production by Alvin Hansen and Herbert Tout in their 1933 article reviewing recent business-cycle literature in Econometrica in which they criticized Hayek’s assertion that a monetary expansion that financed investment spending in excess of voluntary savings would be unsustainable. They pointed out that there was nothing to prevent the monetary authority from continuing to create money, thereby continually financing investment in excess of voluntary savings. Hayek’s reply was that a permanent constant rate of monetary expansion would not suffice to permanently finance investment in excess of savings, because once that monetary expansion was expected, prices would adjust so that in real terms the constant flow of monetary expansion would correspond to the same amount of investment that had been undertaken prior to the first and unexpected round of monetary expansion. To maintain a rate of investment permanently in excess of voluntary savings would require progressively increasing rates of monetary expansion over and above the expected rate of monetary expansion, which would sooner or later prove unsustainable. The gist of the argument, more than three decades before Friedman’s 1967 Presidential address, was exactly the same as Friedman’s.
A further aside. But what Hayek failed to see in making this argument was that, in so doing, he was refuting his own argument in Prices and Production that only a constant rate of total expenditure and total income is consistent with maintenance of a real equilibrium in which voluntary saving and planned investment are equal. Obviously, any rate of monetary expansion, if correctly foreseen, would be consistent with a real equilibrium with saving equal to investment.
My second remark is to note the ambiguous meaning of the short-run Phillips Curve relationship. The underlying causal relationship reflected in the negative correlation between inflation and unemployment can be understood either as increases in inflation causing unemployment to go down, or as increases in unemployment causing inflation to go down. Undoubtedly the causality runs in both directions, but subtle differences in the understanding of the causal mechanism can lead to very different policy implications. Usually the Keynesian understanding of the causality is that it runs from unemployment to inflation, while a more monetarist understanding treats inflation as a policy instrument that determines (with expected inflation treated as a parameter) at least directionally the short-run change in the rate of unemployment.
Now here is the main point that I want to make in this post. The standard interpretation of the Friedman argument is that since attempts to increase output and employment by monetary expansion are futile, the best policy for a monetary authority to pursue is a stable and predictable one that keeps the economy at or near the optimal long-run growth path that is determined by real – not monetary – factors. Thus, the best policy is to find a clear and predictable rule for how the monetary authority will behave, so that monetary mismanagement doesn’t inadvertently become a destabilizing force causing the economy to deviate from its optimal growth path. In the 50 years since Friedman’s address, this message has been taken to heart by monetary economists and monetary authorities, leading to a broad consensus in favor of inflation targeting with the target now almost always set at 2% annual inflation. (I leave aside for now the tricky question of what a clear and predictable monetary rule would look like.)
But this interpretation, clearly the one that Friedman himself drew from his argument, doesn’t actually follow from the argument that monetary expansion can’t affect the long-run equilibrium growth path of an economy. The monetary neutrality argument, being a pure comparative-statics exercise, assumes that an economy, starting from a position of equilibrium, is subjected to a parametric change (either in the quantity of money or in the price level) and then asks what will the new equilibrium of the economy look like? The answer is: it will look exactly like the prior equilibrium, except that the price level will be twice as high with twice as much money as previously, but with relative prices unchanged. The same sort of reasoning, with appropriate adjustments, can show that changing the expected rate of inflation will have no effect on the real equilibrium of the economy, with only the rate of inflation and the rate of monetary expansion affected.
This comparative statics exercise teaches us something, but not as much as Friedman and his followers thought. True, you can’t get more out of the economy – at least not for very long – than its real equilibrium will generate. But what if the economy is not operating at its real equilibrium? Even Friedman didn’t believe that the economy always operates at its real equilibrium. Just read his Monetary History of the United States. Real-business cycle theorists do believe that the economy always operates at its real equilibrium, but they, unlike Friedman, think monetary policy is useless, so we can forget about them — at least for purposes of this discussion. So if we have reason to think that the economy is falling short of its real equilibrium, as almost all of us believe that it sometimes does, why should we assume that monetary policy might not nudge the economy in the direction of its real equilibrium?
The answer to that question is not so obvious, but one answer might be that if you use monetary policy to move the economy toward its real equilibrium, you might make mistakes sometimes and overshoot the real equilibrium and then bad stuff would happen and inflation would run out of control, and confidence in the currency would be shattered, and you would find yourself in a re-run of the horrible 1970s. I get that argument, and it is not totally without merit, but I wouldn’t characterize it as overly compelling. On a list of compelling arguments, I would put it just above, or possibly just below, the domino theory on the basis of which the US fought the Vietnam War.
But even if the argument is not overly compelling, it should not be dismissed entirely, so here is a way of taking it into account. Just for fun, I will call it a Taylor Rule for the Inflation Target (IT). Let us assume that the long-run inflation target is 2% and let us say that (Y – Y*) is the output gap between current real GDP and potential GDP (i.e., the GDP corresponding to the real equilibrium of the economy). We could then define the following Taylor Rule for the inflation target:
IT = α(2%) + β((Y – Y*)/ Y*).
This says that the inflation target in any period would be a linear combination of the default Inflation Target of 2% times an adjustment coefficient α designed to keep successively chosen Inflation targets from deviating from the long-term price-level-path corresponding to 2% annual inflation and some fraction β of the output gap expressed as a percentage of potential GDP. Thus, for example, if the output gap was -0.5% and β was 0.5, the short-term Inflation Target would be raised to 4.5% if α were 1.
However, if on average output gaps are expected to be negative, then α would have to be chosen to be less than 1 in order for the actual time path of the price level to revert back to a target price-level corresponding to a 2% annual rate.
Such a procedure would fit well with the current dual inflation and employment mandate of the Federal Reserve. The long-term price level path would correspond to the price-stability mandate, while the adjustable short-term choice of the IT would correspond to and promote the goal of maximum employment by raising the inflation target when unemployment was high as a countercyclical policy for promoting recovery. But short-term changes in the IT would not be allowed to cause a long-term deviation of the price level from its target path. The dual mandate would ensure that relatively higher inflation in periods of high unemployment would be compensated for by periods of relatively low inflation in periods of low unemployment.
Alternatively, you could just target nominal GDP at a rate consistent with a long-run average 2% inflation target for the price level, with the target for nominal GDP adjusted over time as needed to ensure that the 2% average inflation target for the price level was also maintained.