Today is the official launch date of my book on the Great Depression, entitled . . . well . . . do you want the official title (The Midas Paradox) or my preferred title (The Midas Curse)? The publisher thought the latter title sounded too depressing, and wouldn’t sell. But in my own mind it will always be The Midas Curse. The metaphor reflects the fact that attempts to hoard lots of gold ended up impoverishing the countries that got too greedy. And I’d add that the second theme of the book is that attempts to artificially drive up wages ended up impoverishing workers through high unemployment.
A few observations about the book:
I began research on this subject in 1986. I was interested in McCloskey and Zecher’s critique of Friedman and Schwartz, and tried to reconcile the two studies on my own terms, by developing a model of discretionary monetary policy under a gold standard.
Then I tried to get empirical estimates of both national and global monetary policy. At that time I defined monetary policy not as expected NGDP growth, but rather changes in the gold currency ratio. Even today, I would defend that seeming inconsistency as follows: Under a fiat money regime there are no significant constraints on monetary policy, hence it makes sense to think of the stance of policy in terms of the goal variable (NGDP, or something similar). Under a gold standard, policy is sharply constrained; hence it makes sense to think about the stance of policy in terms of deviations from the rules of the game, i.e. changes in the gold reserve ratio. That’s what central banks actually controlled, not the money supply or interest rates.
Then I started noticing that private gold hoarding was also an issue. And finally, I added changes in the price of gold (1933-34) to complete the demand-side model. I noticed that real wages seemed highly (negatively) correlated with output, and thus added New Deal policies aimed at boosting wages as the supply-side of the model. (Based on research with Steve Silver.) Obviously other factors matter, but I still think these are the key drivers of high frequency fluctuations in output from 1929-39.
I had a lot of frustrations along the way. Most of my empirical data (on “Minitab”) was eventually lost when Bentley changed it’s computer system. I once lost a notebook with almost a year’s worth of notes on the New York Times I had read from the 1930s. I read virtually the entire NYT from that decade, on microfilm. I also read many old dusty volumes at Bentley, Brandeis and Harvard Baker Libraries.
I think I worked far too carefully, trying too hard to be accurate. At the time I was working on my own (I did not use research assistants) and didn’t know how other economists did research. I was very careful collecting all sorts of obscure data like the stock of Estonian kroons (monthly) in the 1930s, to make data sets as complete as possible. (To give you an idea how old this research is, Estonia didn’t exist at the time.) Later I learned that most researchers cut lots of corners, and I doubt that any tiny improvement in accuracy from my painstaking approach was worth all the time I spent. (I wasn’t thinking like an economist.) If I had known how difficult the project would be, I probably never would have done it. And even with all that effort, the book undoubtedly has some mistakes.
By around 2005 the book was done. After a long period at Cambridge University Press it was rejected. The reviewer was not at all helpful. Then I set it aside, before sending it to Princeton after I started blogging in 2009. Another rejection (at least the reviewer was acceptable—just wasn’t their cup of tea.) Then Independent Institute, and again more delays. Unfortunately that means the book is somewhat out of date in terms of the literature review. Once I started blogging I had no time to keep up with my research. But the good news is that my work is so out of the mainstream that I doubt anyone else was doing the same sort of thing. It’s a throwback to the approach taken by Friedman and Schwartz. The differences from F&S include:
1. Defining monetary policy in terms of the global gold/currency ratio, not the domestic money supply.
2. Focusing on both supply and demand shocks, instead of mostly demand shocks as F&S did.
3. Looking at market reactions to policy shocks, which led me to abandon the F&S “long and variable lags” approach. I noticed that policy shocks, the market reaction, changes in the WPI, and changes in industrial production all tended to occur at roughly the same time, or perhaps a month lag, which is impossible if there are long and variable lags. If there actually are long and variable monetary policy lags then my entire book is utterly worthless, and should be thrown in the trash.
Ideally, my book should have been done using NGDP instead of the price level. But I had better price level data at the monthly frequency, and it turns out that during the 1930s you get roughly the same results either way, both the WPI and NGDP fell roughly in half during 1929-33. (Both the WPI and NGDP were also highly correlated, as output was highly correlated with prices during this period.) But if I were doing a similar study with postwar data I’d definitely use NGDP.
I sort of regret having to be so critical of the Fed’s performance during the Great Recession. This material was added after the book was finished, and I would still defend everything I say. But when I wrote the book I always envisioned Bernanke as the “ideal reader.” I think he might have liked the book when it was finished in 2005, but would now hate it, due to just a half dozen pages out of 500.
I looked briefly at the 1920-21 recession, and it seems to have had the same cause—central bank gold hoarding. I seem to recall that the Fed alone hoarded enough gold in 1920-21 to reduce the entire world price level by about 17%. I did not study the 1893 depression, but I suspect that gold hoarding (perhaps private) associated with devaluation fears explain some of that depression too. I also suspect there were a few other factors as well. Someone could do the same sort of study of 1893-96 as I did for 1929-33, and it would be interesting to compare results.
For better or worse this is my life’s work.
PS. Scott Alexander (who I had the good fortune of meeting last week in Boston) said the new title sounds like an airport thriller. (I’ll bet Robert Ludlum could have made it into a real page-turner.) Speaking of Alexander, I hope everyone saw his list of suggested hardball questions for the next debate. Funniest post of the year.
PPS. I also have some comments over at Econlog.