My recent book on the Great Depression is not easy to review. The arguments are complex, and not always easy to follow. I did my best, but it’s a very complicated story. Given my low expectations, I was pleasantly surprised by a couple of recent reviews that described my views fairly accurately (I’d have a few very small quibbles).

Arnold Kling
does a nice job summarizing the key arguments of the book, and then ends with a few reservations:

Although I highly recommend The Midas Paradox, I did not find it entirely persuasive. I worry that it relies too much on Sumner’s reading of the link between wages, prices, and output. My concerns are these:

1. The price index that Sumner uses is the Wholesale Price Index. This is a volatile index that largely excludes finished goods and instead tracks goods that are intermediate inputs to other producers. From the standpoint of those final-goods producers, an increase in the WPI indicates not a positive demand shock but an adverse supply shock. Sumner did not succeed in convincing me that the causality runs from increases (decreases) in the WPI to increases (decreases) in output, rather than the other way around.

2. For the past 60 years or so, the cyclical behavior of real wages has not been consistent. In fact, on average, real wages have risen at least as rapidly during recoveries as during recessions.

The comment on the WPI is a reasonable complaint. Ideally, I should have used monthly data on NGDP, but of course that data was not available (and indeed even today is only available from Macroeconomics Advisers.) Oddly, I believe that the WPI tracked NGDP fairly well during much of this period. That’s because of two offsetting errors. The WPI looks at prices and ignores output, which means you’d expect it to underestimate the fall in NGDP during a depression. However the WPI falls much more sharply than the GDP deflator, as it over-samples highly cyclical commodity prices. Those two factors roughly offset, and I seem to recall that both NGDP and the WPI fell roughly in half during the 1929-33 contraction.

The real wage cyclicality question is quite complex, and I have a 1989 JPE paper on the subject if you are interested in more detail. The final chapter (which is sort of a technical appendix) covers this issue in lots of detail, including the reasons why the model used in this book is ad hoc, and does not apply to the post-war period. I do believe that a model replacing real wages with W/NGDP would hold up during the post-war period.

As far as the question of causality is concerned, you can think of a monetary shock such as the 1933 devaluation as boosting both prices and output directly, or as boosting prices (or NGDP) directly, and then output rises because real wages fall. I prefer the latter framing.

Clark Johnson wrote a book on the Depression back in the 1990s, which anticipated some of my themes, especially French gold hoarding. I highly recommend the book to readers interested in the Great Depression.

Clark also did a good job of summarizing my arguments in a review over at Lars Christensen’s blog, but had a few objections:

In proposing a hypothetical increase in the gold price, perhaps at the time of the Genoa Conference in 1922, Mundell and Johnson intended a counterfactual through which subsequent deflation might have been prevented. Almost no one suggested changing the gold price at the time – in my research the only advocacy I found for a price increase came from a gold producers’ association. In 1934, of course, the US raised the price it would pay for gold – which removed weak systemic demand as a cause of the international depression.

Sumner raises the objection that increasing the price of gold in the early 1920s would have risked significant inflation unless central banks raised their demand for gold in the short run. I believe he overstates the threat of inflation. For one thing (as Sumner acknowledges in his theoretical chapter), prewar gold reserve ratios fluctuated considerably; central banks did not generally act as though bound to monetize new gold to satisfy “rules of the game” – nor did central banks of the US, France, or Germany show much inclination to monetize excess reserves a few years later. Also, only the US among major economies was on a gold standard during the early 1920s, so there would have been no central bank coordination requirement had the price then been raised.

Just to be clear, I did say that the Johnson-Mundell proposal would have been better than what was actually done (as would have any other alternative, including Austrian, monetarist and Keynesian policy counterfactuals.) But I worry that the global increase in gold demand that occurred in the late 1920s and early 1930s would have been deflationary even with a devaluation in 1922, albeit from a higher price level. For example, the gold hoarding of late 1937 was deflationary from a much higher price level (in early 1937) than in 1933. That problem could have been avoided if the interwar gold standard had been properly managed (and here Clark correctly points out that the real issue is the counterfactual path of gold/currency ratios.) But if central banks were capable of managing the interwar gold standard then we never would have had a Great Depression. In that case even a 1922 devaluation would not have been needed.

Clark also suggests (probably correctly) that I underestimate the subtlety of Keynes’s argument:

Keynes’ deeper concern in the General Theory, much more than in his earlier writings, was with what he saw as the tendency of “present day capitalist individualism” to lead into stagnation. He put forth such concepts as that of an “average marginal efficiency of capital” falling to zero, and the “euthanasia of the rentier, of the functionless investor.” Such conclusions are only incidentally related to Keynes’ understanding of monetary economics.

Perhaps I’m missing something, but I can’t see Keynes’s worry about the marginal efficiency of capital falling to zero as anything other than a worry about a liquidity trap.

Clark continues:

Private gold movements, as Sumner describes them, were baffling and somewhat contradictory – first driven by fear of a revaluation of the dollar gold dishoarding, then by fear of a devaluation and gold hoarding. The second makes little sense: with new gold piling up at the Fed, and no deflationary pressure coming from abroad, why would US monetary authorities have wished to devalue in 1937?

It does look a bit puzzling in retrospect, but the 1933 devaluation was not caused by international factors either (the US had the world’s largest gold stocks at the time)—it was an attempt to quickly reflate the economy by lowering the value of the medium of account. Given that late 1937 saw a similar slump to 1930, investors might have expected FDR to respond in much the same way he did in 1933. Instead he took a slower and more conservative path in 1937-38.

Industrial production rose by about 40 percent from the pre-NIRA-shock peak in July 1933 to the precrash peak in July 1937 – at which time it was higher than it had been at its peak in 1929.6 This was a disappointing rate of growth for a period shortly after the worst depression in US history, but growth it was; it is not convincing to roll this four-year period into a longer “supply-side depression.”

It certainly sounds unconvincing, but I do marshal a lot of evidence that the level of output was far lower than it would have been without New Deal labor policies, during the 7 years after July 1933. I argue that industrial production would have recovered quickly after July 1933, if not for the high wage policy. How quickly? Perhaps at the pace of the 1921-22 recovery. If so, the Depression would have ended in just a couple years. Instead, industrial production in July 1935 was about the same as two years earlier.

The 2007 downturn, in contrast, as Sumner observes, began with a financial crisis, the heart of which was widespread and often hidden exposure to low-quality mortgage debt. US monetary policy was not an initial trigger, and probably did not become contractionary until the dollar started to rise sharply in July 2008, at which point the recession entered a harsh, and unnecessary, new phase.

I slightly disagree. I certainly agree that the financial crisis had something to do with the Great Recession. But I believe the mechanism was that it lowered that Wicksellian equilibrium interest rate, and because the Fed did not cut its target rate fast enough, monetary policy became contractionary at the end of 2007, even though (as Clark points out) the dollar did not get strong in the forex markets until later in 2008. I focus on the NGDP growth rate.

I’ve focused on a few areas where I at least slightly disagree. But I’d encourage people to read the entire review, which contains lots of interesting observations, including the following:

Paul Krugman has argued that the recovery of profits and stock prices since 2009 owes much to wage compression. – if he is correct, it is the opposite of the pattern Sumner describes post-1933. It is reasonable to expect that an incoming US administration in 2017 might want to use administrative measures to boost compensation – possibly through a higher minimum wage, mandatory home leave provisions, or obligatory profit sharing. If monetary expansion slows while wages and salaries grow faster, the recovery will face an uncertain future, and perhaps a short one.