Tuesday, April 4, 2017

Scott Sumner Disputes Robert Shiller's Narrative Theory Of The 1920-21 Recession

See Heterodoxy at the AER. Excerpts:
"The depression of 1920-21 is probably the most well understood business cycle in world history. Until today, I couldn't even imagine anyone contesting the standard view, which is that it was caused by deflationary monetary policy. If we are wrong about this, we have no reason to have any confidence in anything we teach in the macro half of our EC101 textbooks.

If Shiller is right, then macroeconomics is back to square one. Not back to 1967, before the natural rate hypothesis. Not back to 1935, before the General Theory. More like 1751, before Hume's writing on money, velocity and business cycles. Is Shiller actually that much of a nihilist? I very much doubt it. More likely, Shiller's a relatively normal Keynesian, who thinks we do know certain things, such as that a highly contractionary monetary/fiscal policy can cause the unemployment rate to rise. (Please correct me if I am wrong.) Rather the problem seems to be his understanding of macroeconomic history.

Consider Shiller's description of the 1920-21 depression:
In looking for the narrative basis of economic recessions, which might be hard to see since narratives are not easy to measure, it would appear that we would have the most luck looking at really big ones: 1920-1921 was the sharpest US recession since modern statistics are available. The US Consumer Price Index switched suddenly from inflation to deflation: between June 1920 and June 1921, during the Depression, it fell 16 percent, the sharpest one-year deflation ever experienced in the United States. The Index of Wholesale Prices fell much more: 45 percent over the same time interval, its sharpest decline ever.
This sort of implies that there is a puzzle to be explained. Why would the price level have suddenly plunged by 16%, the largest decrease ever? Perhaps because the Fed reduced the supply of high-powered money by 15.2% between the fall of 1920 and the fall of 1921, by far the sharpest one-year decline ever experienced in the US? But Shiller doesn't even mention that fact. Instead he says the following:
Surprisingly, the online NBER Working Paper Series, almost a hundred years later, when searched, has virtually nothing to say about what caused this spectacular depression. Why, after all, did it happen? Milton Friedman and Anna J. Schwartz, in their Monetary History of the United States, have given the most influential account. According to them, the 1920-1921 contraction has a single identifiable cause: an error made by the fledgling Federal Reserve to raise the discount rate to trim out-of-bounds inflation in 1919 caused by their carelessly over-expansionary policy right after World War I, leading to a necessity to take strong measures against inflation in 1920. Benjamin Strong, the president of the New York Fed, was on a long cruise starting December 1919, and was unable to prevent Federal Reserve Banks (which did not coordinate their policies with each other so much back then) from raising the discount rate as much as a full percentage point in one shot in January 1920.

This is misleading in all sorts of ways. First, the 1920-21 depression has been studied by a number of researchers. Much less than the Great Depression, but that's partly because the cause of the 1920-21 recession is so obvious. It's a simple problem. Second, Friedman and Schwartz's views are mischaracterized in a way that makes them seem Keynesian. They certainly did not argue that higher interest rates caused the recession, in the ordinary Keynesian sense of causation. Rather they suggested that higher discount rates led to a massive decline in the money supply, and that this is what caused the sharp deflation and depression.
All of these events--World War, the influenza epidemic, the race riots, the Big Red Scare, the oil shock--were associated with hugely unsettling narratives that could have led to a sense of economic uncertainty that might have discouraged discretionary spending of households and slowed down hiring decisions of firms around the world. These certainly sound like more significant potential causes than New York Fed President Benjamin Strong's decision to take a cruise when he was needed.
This is a weak argument. Start with the fact that it mixes up two issues, F&S's incidental conjecture that Gov. Strong's cruise led to the Fed policy error, and the much more important question of whether the Fed's highly deflationary monetary policy could have caused a depression. Shiller makes it seem like someone rejecting the cruise ship conjecture must also reject the entire orthodox monetary explanation for the 1920-21 depression. That's just silly.

But it's even worse. Shiller offers a series of alternative explanations where no alternatives are needed. We know what happened in 1920-21, a deflationary monetary policy. You'd at least think he would have picked an example where the issue to be explained was a mysterious drop in velocity, not a drop in the money supply. I still wouldn't agree with him, but I could imagine someone arguing that mood shifts among the public might impact velocity. Instead he picks the one depression where velocity seems to add very little to the simple money supply contraction story.

And it's even worse. After telling us that the CPI fell by 16% and the WPI by 45% in one year, the most in US history, he speculates about possible psychological causes that have no plausible link to deflation. Take his oil shock example. We also had massive oil shocks in 1974 and 1979-80. Does anyone seriously believe those had a deflationary effect? We had very serious race riots in 1965, 1967 and 1968. Hmmm, those were the first years of the Great Inflation. Red scares? Did the 1979 Russian invasion of Afghanistan trigger a deflationary mood among the public? What happened to the WPI in 1979-80? How about the McCarthyism period? The Cuban Missile Crisis?

He also mentions the 1919 influenza epidemic, which may have killed more people than the Black Death. But the Black Death was inflationary, just as the AS/AD model predicts. Indeed all five shocks that he speculates might have cause a big deflation are actually known to have inflationary consequences.

Maybe Shiller's theory is "it depends". Sometimes oil shocks are inflationary, and sometimes they are deflationary. But if you are going to make your psychological theory of cycles that elastic, what use is it? I could invent 1000 narrative "theories" to fit any set of time series data."

No comments:

Post a Comment