The industrious James Grant recently published a book about the 1920-21 Depression. It has received enthusiastic reviews in the Wall Street Journal and Barron’s, was the subject of an admiring column by Washington Post columnist Robert J. Samuelson, and was celebrated at a Cato Institute panel discussion, luncheon, and book-signing event. The Cato extravaganza elicited a dismissive blog post by Barkley Rosser which was linked to by Paul Krugman on his blog. The Rosser/Krugman tandem provoked an unhappy reply on the Free Banking blog from George Selgin who chaired the Cato panel discussion. And the 1920-21 Depression is now the latest hot topic in the econblogosphere.
I am afraid that there are multiple layers of errors and confusion that are being mixed up and compounded in this discussion, errors and confusion derived from basic misunderstandings about how the gold standard operated that have been plaguing the economics profession and the financial world for about two and a half centuries. If you want to understand how the gold standard worked, what you have to read is the book by Ralph Hawtrey entitled – drum roll, please – The Gold Standard.
Here are the basic things you need to know about the gold standard.
1 The gold standard operates by creating an equivalence between a currency unit and a fixed amount of gold.
2 The gold standard does not require gold to circulate as money in the form of coins. That was historically the case, but a gold standard can function with no gold coins or even gold certificates.
3 The value of a currency unit and the value of a corresponding weight of gold are necessarily equalized by arbitrage.
4 Equality between a currency unit and a corresponding weight of gold does not necessarily show the direction of causality; the currency unit may determine the value of gold, not the other way around. In other words, making gold the standard of value for currency affects the demand for gold which affects the value of gold. Decisions made by monetary authorities under the gold standard necessarily affect the value of gold, so a gold standard does not somehow make the value of money independent of monetary policy.
5 When more than one country is on a gold standard, the countries share a common price level, because the value of gold is determined in an international market.
Keeping those basics in mind, let’s quickly try to understand what was going on in 1920 when the Fed decided to raise its discount rate to the then unprecedented level of 7 percent. But the situation in 1920 was the outcome of the previous six years of World War I that effectively destroyed the gold standard as a functioning institution, even though its existence was in some sense still legally recognized.
Under the gold standard, gold was the ultimate way of discharging international debts. In World War I, belligerents had to pay for imports with gold, thus governments amassed all available gold with which to pay for the imports required to support the war effort. Gold coins were melted down and converted to bullion so the gold could be exported. For a private citizen in a belligerent country to demand that the national currency unit be converted to gold would be considered an unpatriotic if not a treasonous act. So the gold standard ceased to function in belligerent countries. In non-belligerent countries, which were busy exporting to the belligerents, the result was a massive inflow of gold, causing a spectacular increase in the amount of gold held by the US Treasury between 1914 and 1917. Other non-belligerents like Sweden and Switzerland experienced similar inflows.
Quantity theorists and Monetarists like Milton Friedman habitually misinterpret the wartime inflation, and attributing the inflation to an inflow of gold that increased the money supply, thereby perpetrating the price-specie-flow-mechanism fallacy. What actually happened was that the huge demonetization of gold coins by the belligerents and their export of large quantities of gold to non-belligerent countries in which a free market in gold continued to operate drove down the value of gold. A falling value of gold under a gold standard logically implies rising prices for all other goods and services. Rising prices increased the nominal demand for money, which more or less automatically caused a corresponding adjustment in the quantity of money. A rising price level caused the quantity of money to increase, not the other way around.
In 1917, just before the US entered the war, the US, still effectively on a gold standard as gold flowed into the Treasury, had experienced a drastic inflation, like all other gold standard countries, because gold was rapidly losing value, as it was being demonetized and exported by the belligerent countries. But when the US entered the war in 1917, the US, like other belligerents, suspended operation of the gold standard, thereby accelerating the depreciation of gold, forcing the few remaining countries on the gold standard to suspend the gold standard to avoid runaway inflation. Inflationary pressure in the US did increase after entry into the war, but the war-induced fiat inflation, to some extent suppressed or disguised by price controls, was actually slower than inflation in terms of gold.
When the war ended, the US went back on the gold standard by again making the dollar convertible into gold at the legal parity. Doing so meant that the US price level in terms of dollars was below the notional (no currency any longer being convertible into gold) world price level in terms of gold. In other belligerent countries, notably Britain, France and Germany, inflation in terms of their national currencies exceeded gold inflation, requiring them to deflate even to restore the legal parity in terms of gold. Thus, the US was the only country in the world that was both willing and able to return to the gold standard at the prewar parity. Sweden and Switzerland could have done so, but preferred to avoid the inflationary consequences of a return to the gold standard.
Once the dollar convertibility into gold was restored, arbitrage forced the US price level to rise to so that it would equal the gold price level. The excess of the gold price level over the US price level level explains the anomalous post-war inflation – everyone knows that prices are supposed to fall, not rise, when a war ends — in the US. The rest of the world, then, had to choose between accepting US inflation, by keeping their currencies pegged to the dollar, or allowing their currencies to appreciate against the dollar. The anomalous post-war inflation was caused by the reequilibration of the US price level to the gold price levels, not, as commonly supposed, by Fed inexperience or incompetence.
To stop the post-war inflation, the Fed could have simply abandoned the gold standard, or it could have revalued the dollar in terms of gold, by reducing the official dollar price of gold. (I ignore the minor detail that the official dollar price of gold was then determined by statute.) Instead, the Fed — whether knowingly or not I can’t say – chose to increase the value of gold. The method by which it did so was to raise its discount rate, thereby making it easier to obtain dollars by selling gold to the Treasury than to borrow from the Fed. The flood of gold into the Treasury in 1920-21 succeeded in taking a huge amount of gold out of private and public hands, thus driving up the real value of gold, and forcing down the gold price level. That’s when the brutal deflation of 1920-21 started. At some point, the Fed and the Treasury decided that they had had enough, having amassed about 40% of the world’s gold reserves, and began reducing the discount rate, thereby slowing the inflow of gold into the US, and stopping its appreciation. And that’s when and how the dearly beloved, but quite dreadful, depression of 1920-21 came to an end.
Uneasy Money 
David, your observations about gold flow are very much welcome. In fact I had been tempted to make some remarks on the subject, only to decide that it was beside the main point, in response to the following statement, which occurs toward the end of The Economist’s review of Jim Grant’s book: “The wrinkle in 1920-21 was that because of the gold standard, the influx of gold into America—a prerequisite for easy money—required tight money in countries that were losing gold, such as Britain. Deflation worked, but only after inflicting widespread unemployment on America. Both the gold standard and deflation would wreak even greater havoc during the Great Depression.”
It seems to me that this is a far more egregious misunderstanding of what gold flows were about at the time in question than anything to be read in Grant’s book itself. Indeed, Grant’s discussion of the gold situation, (e.g., pp,. 180 ff.) seems to me largely consistent with your own understanding. Because you don”t say which of the participants in the recent discussion is responsible for the particular “layers of confusion” to which your post constitutes a reply, I think you may give the wrong impression concerning the extent to which each of us has contradicted what you (and Hawtrey) have to say about the matter.
On a related point, market rates in the U.S. fell ahead of the Fed’s own discount rate. There is every appearance, indeed (as I also argued in my post), that the Fed was merely playing catch-up, and that reluctantly, with the rest of the loan market.
LikeLike
“Rising prices increased the nominal demand for money, which more or less automatically caused a corresponding adjustment in the quantity of money. A rising price level caused the quantity of money to increase, not the other way around.”
Here I must beg to differ. Yes, other things equal rising prices caused the nominal demand for money to increase. But there is no reason why that should itself have led “automatically” to any increase in the nominal money stock. The nominal money stock response was strictly a function of Fed asset-purchasing and discount policies, which kept increasing well after the end of the war. Total Fed interest-earning assets, consisting mainly in the end of war paper, were $.5 billion at the time of U.S. entry into the war, about $2.5 billion at the time of the armistice, and about $3.5 billion in mid 1921. The Fed was never “compelled” to purchase all those assets, by changes in the price level (which depended in fact on how many it purchased) or otherwise. The proof, indeed, consists precisely of the fact that, once it determined to do so, it could reduce its holdings and, by doing so, increase the price level. It makes no sense, in my opinion, to treat the Fed as having been a passive agent so long as P was rising, and an active one when P was falling. Here I agree with Friedman in regarding the Fed as having held the reins both going up and coming down, even if gold flows were informing its decisions.
LikeLike
Sorry: please read “decrease the price level” in the 5th line from bottom of my last comment.
LikeLike
David, Nice post. I agree that your way of looking at things is best (naturally, as it is also my way of looking at things.) But I don’t think that makes Friedman’s money-oriented approach wrong, just less useful. Gold and currency notes were dual media of account, so the price level can be defined in terms of changes in the supply and demand of either. You and I think the gold market is a more useful way to look at the picture, as we see that market as being the dog, and the money market as being the tail. But changes in Fed policy regarding the base can impact the demand for gold, and hence Friedman could analyze the situation from a money market perspective.
Suppose the Fed had gone on its gold accumulation binge in 1919-20, instead of 1920-21? Might the price level have been fairly steady throughout 1919-21, instead of rising sharply then falling sharply?
LikeLike
Don’t know how I overlooked George’s comment, but I think he’s making a similar point.
LikeLike
Very nice to start the weekend by reading such an interesting post.
I haven’t been following very closely the recent blogosphere discussion on the 1920-21 depression as it seems the same old arguments just get rehashed every year or so by people with preconceived ideas on what must have happened around which they shoehorn the facts to fit.
My main takeaway from this article is that this depression:
– Was caused by tight money. The fed raised the discount rate to a very high level. This was both deflationary in itself and led to people selling gold to the fed which had the effect of increasing the price of gold which is itself deflationary under a gold standard.
– Was ended when the fed stopped this policy and reduced the discount rate again.
Is that correct , or is that an oversimplification ?
LikeLike
Rob, the gist of my post, to which David supplies a link, is that the last point on your list is not correct, in that it gives a misleadingly exaggerates impression of the Fed’s role in the recovery.
LikeLike
I just read George’s also interesting article.
He says “The Fed’s contribution to recovery, in short, consisted, not of any actual monetary stimulus, but of a mere cessation of what had been a precipitous decline in its interest-earning asset holdings “. This is not totally inconsistent with what David is saying.
He then goes on to explain the significance to the recovery of monetary expansion due to gold inflows, driven by a purely market mechanisms.
My interpretations now is:
– The fed abandoned a highly deflationary policy that was causing a market breakdown
– This allowed the market to work efficiently again, allowing appropriate adjustments to the money supply driven by the “normal working of the price mechanism”, which allowed recovery to take place.
LikeLike
George, Glad to hear that you found what I wrote useful. I saw that review in the Economist and cringed when I read the passage you quote from, but the main problem I have with it is that no country but the US was on the gold standard then. The correct statement would have been that deflation in the US required other countries to accept deflation unless they allowed their currencies to depreciate against the dollar. As I think you know I agree that the gold standard and deflation would wreak even greater havoc during the Great Depression, though, in theory, it would have been possible to retain the gold standard without undergoing deflation and depression if the spirit of the Genoa Agreement of 1922 could have been maintained. But I gather from your comments at the Cato book event that you still hold the old Austrian view that the Genoa Agreement and the idea of trying to stabilize the post-1921 gold price level were mistaken. I would be really glad to hear that you actually think otherwise. (I don’t own a copy of Grant’s book, so I can’t comment on the passage that you cite.)
To be honest, when I started thinking about writing this post, I had intended to be more explicitly critical of Grant, and you and Larry. But as I started writing, it actually became clear to me that just about everything I was reading on both sides of the debate about 1920-21 was in some important respects missing an important aspect of what was happening in 1920-21 and that the sources of the confusion go back to misunderstandings about how the gold standard operated and the nature of the breakdown in the gold standard during World War I. Writing the post in that way saved me from having to criticize anyone in particular (except for Milton Friedman, but he’s special) and allowed me to just explain my own thinking in what I thought was a pretty simple and direct way.
I agree that the Fed delayed in reducing its interest rate. I guess the question is whether by keeping the discount rate at 7% and then at 6.5% as long as it did, the Fed prevented the recovery from starting sooner that it did.
About the cause of the monetary expansion in 1920-21, I wrote the passage you are quoting in haste, so the mechanism I was referring to may not have been as automatic as I was thinking. Under an international gold standard, if there is not enough domestically supplied money to meet demand, an export surplus causes an inflow of gold to make up the difference. The mechanism under situation when the US was the only country on the gold standard requires a bit more thinking on my part. If the Fed had not been supplying enough money domestically to meet the demand for money, then the US would have been importing gold and driving the value of gold toward the value of the dollar. US monetary policy in 1919-20 was driving down the value gold, thereby increasing the inflation in terms of dollars needed to equalize the value of the dollar and the value of the legal equivalent amount of gold. So I think that I now agree that the nominal quantity of money was not endogenously determined in 1919-20.
Scott, As I just responded to George, I now think that I did not state the point about the US nominal money supply in 1919-20 correctly. In the peculiar circumstances of 1919-20, the US money supply was determined by the decisions of the Fed even though the US was legally back on the gold standard. The peculiarity of the situation was that the US was simultaneously determining both the value of gold and the nominal quantity of money. The Fed could have avoided inflation in 1919 by adopting a policy that would have increased the value of gold. I think your defense of Friedman is valid for 1919-20, but not when the gold standard was international.
Rob, Just a sematic point: it is confusing to say that people selling gold to the Fed increased the price of gold, because the price of gold (in terms of dollars) was legally fixed. It is better to say the Fed increased the real value of gold by its policy. I think that otherwise you have stated my main point correctly.
Though I didn’t raise the point in my reply to George’s comment above, I would observe that I am uncomfortable with the idea that monetary expansion due to gold inflows had anything to with the recovery to the 1920-21 Depression. To suggest gold inflows played a role in the 1920-21 recovery is to reason from a quantity change.
LikeLike
David, what is the role of a central bank’s discount rate policy under a gold standard? Adjusting the rate can’t affect the price level since the gold peg does the job. A large central bank like the Fed might increase the world purchasing power of gold by increasing its discount rate, but what about a central bank that is too small to affect the purchasing power of gold by its discount rate policy? Does adjusting its discount rate do anything?
LikeLike
JP, I’ve mentioned this in a couple of posts — I can’t quite remember which ones. The main effect of a central bank’s discount rate is on its holdings of gold or other foreign-exchange reserves. Raising the rate induces a gold inflow; reducing the rate induces an outflow.
LikeLike
David: I’ve been thinking about this good post. I finally collected my thoughts: http://worthwhile.typepad.com/worthwhile_canadian_initi/2014/12/alpha-beta-and-gold.html
Did I get it roughly right?
LikeLike
Nick, Yes, I would say that you did. My follow-up post shows that in 1922 D. H. Robertson figured out that a large country could control its own price level even while on the gold standard by altering the real value of gold, accumulating gold to increase its value and exporting gold to reduce its value.
LikeLike