Friedman as a critic of Keynesian economics
By Scott Sumner
I’ve finished reading the first volume of Edward Nelson’s impressive new study of Milton Friedman, and highly recommend the book to anyone interested in macroeconomics. This remark on page 380 caught my eye:
In the half century or more after 1951, Friedman would receive a lot of criticism from monetary researchers and practitioners for his alleged neglect of interest rates and his supposed reluctance, or outright unwillingness, to describe monetary policy actions or their effects in terms of interest rates. However, it was concluded in chapter 5 that the record of Friedman’s statements is more nuanced than this criticism suggests.
[As an aside, that criticism would be valid for my own views on money and interest rates, which are not nuanced. I have a recent paper that criticizes the interest rate approach to monetary policy.]
While Friedman did accept that monetary policy had an effect on interest rates, and that this could impact the broader economy, it’s also undeniable that he was often quite critical of the interest rate approach to monetary policy used by Keynesian economists. Keynesians argue that an expansion in the money supply will reduce interest rates, which boosts aggregate demand. Friedman argued that an increase in the money supply would boost aggregate demand, and interest rates might rise or fall depending on the relative strength of the liquidity, income and Fisher effects. The extra money would directly push rates lower, but if the policy led to faster growth in real income or inflation, this would tend to push nominal interest rates higher. The overall effect was ambiguous, especially after a few months or years.
In a recent post, I quoted Friedman on the Phillips Curve, an area where he was also skeptical of Keynesian orthodoxy:
There was, however, a crucial difference between Fisher’s analysis and Phillips’s, between the truth of 1926 and the error of 1958, which had to do with the direction of causation. Fisher took the rate of change of prices to be the independent variable that set the process going. [From p. 273 of Nelson’s book.]
This requires a bit of history. In 1926, Irving Fisher had argued that inflationary shocks impact real output. In the 1960s, Keynesian economists used
Edmund A.W. Phillip’s (1958) empirical study to argue that changes in the unemployment rate impact inflation. Put simply, Fisher thought that rising inflation could cause a boom, whereas Keynesians thought a boom could cause rising inflation.
Is there any connection between Friedman’s heterodox views on these two separate issues? Let’s begin with the simplest possible monetary model of nominal GDP:
M*V(i) = P*Y
Here I assume that velocity is positively related to the nominal interest rate, an assumption that is supported by a wealth of empirical evidence. That means that a lower interest rate will tend to reduce velocity, and other things equal that will tend to reduce nominal spending.
That claim might seem surprising, but it is pretty obvious when you think about it. Consider a simple thought experiment. The Fed suddenly boosts the money supply by 20%. (For simplicity we’ll use the monetary base, which is directly under the Fed’s control.) Obviously the level of NGDP (i.e. P*Y) doesn’t change overnight. So at the moment the money supply rises by 20%, the velocity of circulation must fall by an equal amount, leaving M*V and P*Y unchanged. The extra money lowers the nominal interest rate, which reduces the opportunity cost of holding money enough to induce people to hold 20% larger cash balances, at least in the very short run (say the same day). On that point, both Keynesians and monetarists agree. The more difficult question is what happens next. Before considering what happens next, let’s think about the very long run equilibrium.
In the very long run, money is neutral. A one-time 20% boost in the money supply has no long run effect on interest rates, no long run effect on velocity, and no long run effect on real output. Instead, prices rise by 20% and nothing changes in real terms. The more difficult question is what comes before the long run and after the very short run. It’s the intermediate period between these extremes that is so difficult to model.
We know the Keynesian explanation. More money leads to lower interest rates. Lower interest rates lead to more spending. Because prices are sticky in the short run, the extra spending increases real output. When increased spending pushes output past its natural rate, the economy overheats and inflation results. More money causes lower interest rate which causes more output which eventually leads to higher prices.
For monetarists like Friedman, that’s not a satisfactory explanation. Start with the impact of interest rates on nominal spending. If lower interest rates lead to lower velocity, it’s hard to see how low interest rates are expansionary. Indeed if money growth is extremely rapid, then interest rates will often rise as lenders begin to expect higher inflation and demand to be compensated with higher nominal interest rates. But higher nominal interest rates will increase the velocity of circulation, causing inflation to rise by even more than the money supply. This happened in Germany in the 1920s and America in the late 1970s. So it’s too simple to say that easy money has an expansionary effect because it leads to low interest rates. Easy money has an even more expansionary effect when it leads to higher interest rates.
Now think about the impact of economic growth on inflation. It seems obvious that a booming economy is inflationary, right? But if you look at the equation above, it’s clear that (for any given level of nominal spending) the faster the rate of economic growth, the slower the rate of inflation. For any given P*Y, more Y means less P.
In the monetarist model, more money causes more nominal spending (P*Y) because the public has a well-defined demand for money, and will try to get rid of excess cash balances by spending the new money that is injected into the economy. Velocity is not absolutely fixed, but it’s stable enough that the new money will tend to boost NGDP. (At least when interest rates are above zero.) And higher nominal spending will tend to boost real output in the short run, because wages and prices are sticky, or slow to adjust. In the long run, only prices will increase and output will return to the natural rate.
Put even more simply:
In the Keynesian model, real shocks have nominal effects.
In the monetarist model, nominal shocks have real effects.
Of course both claims are true to some extent, which Friedman acknowledged. But it’s fair to say that he put more weight on the nominal shock –> real effect perspective, at least compared to his Keynesian critics. In a 2013 article, Nelson contrasts the two views:
[T]he Bank of England sees monetary policy as exerting effects on spending in the first instance through effects on nominal spending. This view of policy transmission lines up with older expositions of monetary policy transmission—both by monetarists and, it should be stressed, Keynesians (like Tobin, 1981)—and seems to be implicit in past Bank of England discussions of the connections between monetary policy and the economy (see King, 1997).
This view is not consistent with standard models of monetary policy transmission that have become prevalent in recent decades; indeed, this view has been treated caustically by Svensson (1999, p.642). Svensson refers to ‘a previous, somewhat simplistic, view of the transmission mechanism for monetary policy … [under which] monetary policy only determines nominal GDP, but cannot affect the distribution of nominal GDP between inflation and output growth’. To Svensson, nominal GDP is not useful in monetary policy analysis: while monetary policy certainly affects nominal GDP, nothing is gained by thinking of monetary policy as working via nominal GDP; the reaction of nominal GDP to monetary policy reflects the dependence of prices and output, individually, on variables that are affected by monetary policy. The behavior of nominal GDP is then derived recursively, by the behavior of its two definitional components; nominal GDP does not itself appear in the structure of the model.
Svensson is espousing a widely held (modern Keynesian) view.
To be clear, it is not a question of whether monetary policy affects P and Y individually, or nominal GDP (P*Y) as a whole. Both claims are clearly true. Rather the issue is which modeling approach is the most useful.
The Keynesian approach depends on the estimation of all sorts of difficult to observe variables, such as the “natural rate of interest” and the “natural rate of output” (or unemployment.) Policy is appropriate when the output gap is zero and the interest rate equals the natural rate. But how can we estimate the various natural rates?
In contrast, Friedman advocated the broad money supply (M2) as an indicator of the stance of monetary policy, and some market monetarists favor NGDP futures prices. There is no natural rate of M2 or NGDP to estimate, just targets to set.
Friedman won a lot of debates during the 1960s and 1970s because his critique of Keynesian economics was mostly correct. Keynesians did make the mistake of assuming that rising interest rates meant that money was getting tighter and falling rates meant money was getting easier (some still do). Keynesians did make the mistake of assuming that there was a long run trade-off between employment and inflation.
You might think that the model mentioned above (M*V(i) = P*Y) is “dumb”. But this seemingly naive and simplistic view of the world won a lot of debates in the 1960s and 1970s. Unfortunately, Friedman was probably wrong about M2 being the appropriate indicator of the stance of monetary policy; the expected rate of growth in NGDP is a more reliable indicator. So Friedman lost some debates in the 1980s.
If a macroeconomist lives long enough, they’ll eventually end up on the losing side. I’ve been advocating NGDP targeting for decades, and then as soon as people started paying attention we get a once in 100-year recession where NGDP targeting would not have been appropriate. But perhaps targeting 2 year forward expected NGDP would still have been fine.
Back in 2009, some people seemed to think I had something useful to say about the economy. I’m in no position to know if that’s correct, but if it is then it is largely due to what I learned from Milton Friedman.