Bayesian Investor Blog

Book review: The Midas Paradox: Financial Markets, Government Policy Shocks, and the Great Depression, by Scott B Sumner.

This is mostly a history of the two depressions that hit the U.S. in the 1930s: one international depression lasting from late 1929 to early 1933, due almost entirely to problems with an unstable gold exchange standard; quickly followed by a more U.S.-centered depression that was mainly caused by bad labor market policies.

It also contains some valuable history of macroeconomic thought, doing a fairly good job of explaining the popularity of theories that are designed for special cases (such as monetarism and Keynes’ “general” theory).

I was surprised at how much Sumner makes the other books on this subject that I’ve read seem inadequate.

Gold

Sumner presents a good argument that previous attempts at analyzing monetary causes of the depression failed because they looked at the wrong evidence. The existence of a gold exchange standard means that looking at money supply figures from one or two countries can miss important changes in the supply of and demand for money. France’s increase in gold holdings was large enough to create significant deflation in gold standard countries. Other histories have little more than obscure hints about that problem.

Now that I’ve absorbed Sumner’s evidence, it’s pretty clear that the depression originated in a combination of (1) stress on the gold exchange standard from WWI-related inflation, (2) ordinary business fluctuations, and (3) a moderate amount of mismanagement by central banks.

Sumner is rather hostile to a pure gold standard, but doesn’t say much about it. He argues fairly convincingly that the gold exchange standard which the world had in the 1920s was less stable (and therefore more harmful) than a pure gold standard.

Liquidity Traps

Sumner sheds some light on the mystery of what causes people to believe in liquidity traps, but still leaves me somewhat puzzled.

Part of the explanation is that people have developed a habit of measuring Fed policy by looking at interest rates. If your One True Equation of macroeconomics says the Fed can only affect the economy via altering interest rates, then it follows that the Fed can’t do more to inflate than to force interest rates to zero [1] [2].

Another problem is that when the Fed is constrained by a gold standard, it can’t produce as much inflation as it sometimes wants. (That’s the point of holding an unreliable central bank to a supposedly rigid standard). Were economists careless enough that they effectively didn’t notice that we abandoned the gold standard?

Another strange factor dates back to 1933, when the US temporarily left the gold standard in order to inflate back to conditions of the mid 1920s. Historians looking at quarterly (or less frequent) data see that as a failure. Sumner looked instead at monthly data [4], and saw a 3-4 month period of inflation in which economic activity recovered more than half of what it lost in the prior 3.5 years (probably the most dramatic expansion in US history).

U.S. Wage problems

A second reason that most historians see a liquidity trap where Sumner sees a dramatic recovery is that the recovery suddenly stopped in July 1933. Sumner provides evidence that this coincides with when markets realized the extent to which wages would be forced up by the NRA.

Many commentators have noted that the NRA was harmful, but Sumner is the first [5] to argue that the harm was large enough to halt a recovery that would have otherwise been complete by the end of 1934!

Sumner has some moderately good reasons for that claim, including a good comparison to the 1921 downturn. But I’m disappointed at how little evidence he uses. He glosses over evidence that the end of the NRA in 1935 caused only a mild improvement. And he hardly says anything about the international comparisons that should confirm his guess. [6]

Low Interest Rates

The Friedman and Schwartz history had led me to think that high real interest rates were an important problem. I was quite surprised when Sumner caused me to doubt that real interest rates were high in any relevant sense. He claims, quite plausibly, that (under a gold exchange standard) inflation was not predictably different from zero. It is only hindsight bias that makes us think deflation could have been predicted to continue at any stage of the depression. That means that interest rates weren’t much of a deterrent to economic activity.

Efficient Markets

I find it refreshing that Sumner takes the efficient market hypothesis seriously, and puts more weight than other historians on evidence of how markets reacted to relevant news. He may go slightly overboard about rejecting the long and variable lags that other authors see, but it seems safer to err in that direction (authors who err in the other direction seem more able to rationalize away evidence that doesn’t support their theory).

I’m a bit puzzled that Sumner is willing to describe 1936-1937 as a commodity bubble, but is reluctant to call the 1929 stock market peak a bubble. He gives plausible reasons for believing that market couldn’t foresee the specific mistakes that caused the ensuing crash. But why should I believe that the market was sensible to be confident in the competence of governments?

Conclusion

The book is not as well organized as I’d hoped. Sumner encourages “specialists” to read the last chapter immediately after chapter 1. Anyone who has read another economic history covering that period, or a good book on macroeconomics, would be better off reading the last chapter before chapter 2.

Sumner only succeeding in making modest parts of the book readable by a wider audience. If you aren’t willing to read the whole book, it’s still valuable to read chapters 1 and the summaries at the ends of each chapter.

The book is a surprisingly large improvement on previous histories, and has some important criticisms of modern monetary economics.

Sumner’s macroeconomic theories are somewhat more complex than those he attacks, but they better fit the details of the various downturns and recoveries.

Footnotes

[1] – It’s often assumed to be obvious that the Fed can’t set interest rates below zero. I’m pretty sure the Fed can cause negative interest rates, but I expect it always has better options.

[2] – But any respected economist will apparently admit (if pressed) that the Fed can do more. Maybe there’s some secret rule saying that if the Fed drops money from helicopters, it has to drop enough money to create hyperinflation [3]?

[3] – Ok, technically the Fed can’t drop money from helicopters without first begging congress for permission, and the Fed doesn’t know how to beg. So if you want to be picky, assume I meant buy t-bonds or gold.

[4] – Economists have a fetish for using GNP/GDP data, which is only available on a quarterly frequency. Sumner uses industrial production data instead. After seeing the difference made by monthly data, I won’t have much respect for people who ignore it.

[5] – Sure, I’ve seen some people who complain that FDR’s policies were the main reason the economy didn’t recover until World War 2. But Sumner is the first author I’ve seen who (1) connects the timing of halts in the recovery to specific policy changes, and (2) distinguishes the good FDR policies from the bad FDR policies. Sumner is somewhat pro-FDR, noting that FDR’s worst policy was popular enough that most others in his place would have tried it, and his best policies had much less popular support.

[6] – Sumner has since cited a paper which (according to my quick skim) suggests, based on evidence from a similar French mistake in 1936, that Sumner is only exaggerating a little bit.