Scaling to New Depths* with Scott Sumner
by Mike Kimel
Scaling to New Depths* with Scott Sumner
I’ve been having a bit of back and forth with Scott Sumner. Here is his latest post, helpfully entitled: “A suggestion for Mike Kimel.”
His key suggestion:
“Please take a close look at the data from the Great Depression, before doing more posts claiming I don’t know the facts.”
He then goes on to point out he’s been studying the 1933 period for 20 years. From there he goes on to explain my first mistake:
He insists that FDR’s dollar depreciation program began in October 1933, even though all economic historians agree in began in mid-April 1933, when the exchange rate for the dollar began declining (against gold and against other currencies.) He insists prices began rising before FDR took office off, which is not true. He presents a graph that he claims shows prices rising before FDR took office, but his graph shows inflation rates, not the price level. In fact, the graph actually supports my argument that inflation didn’t turn positive until after FDR took office. There’s a difference between the rate of inflation and the price level.
OK. Let’s redo the graph showing not inflation but rather the price level. And I’ll keep it very simple… I will limit it to two points. Well, three, though the third is not exactly on the curve so to speak. As before, I’m still using PPI because its the publicly available source most closely related to the prices Sumner seems to be discussing, and I’ll use the graphics tool at the Federal Reserve Economic Database (FRED)
The graph shows the PPI for February and March of 1933. FDR took office in March 1933.
As I noted in my previous post,
You can see the decline in prices halt and start reversing even before he took office.
Now, I don’t remember arguing that inflation didn’t turn positive before then. To me, its a big deal that PPI hit rock bottom and reversed itself. Getting out of free-fall was in itself a big deal. Here’s a graph for 1929 to 1934 to give you an idea:
Note that February 1933 happened to be the low point for PPI during its entire history, and the PPI had been calculated since 1913.
But there’s another important point in the quote I provided above, namely this:
He insists that FDR’s dollar depreciation program began in October 1933, even though all economic historians agree in began in mid-April 1933, when the exchange rate for the dollar began declining (against gold and against other currencies.)
This isn’t quite right. As I’ll make clear, I don’t think the dollar actually depreciated against gold until January 1934. Sumner was so insistent on this depreciation occurring before then that I spent a bit of time on google and found a story by Jesse Jones, head of the Reconstruction Finance Corporation, about how FDR had him and soon to be Treasury Secretary Morgenthau help him (FDR) revalue the price of gold.
Now, I am not an economic historian, and I’m not sure I know any these days, so for all I know, Sumner is correct about what all economic historians agree happened. I am, instead a data guy. I like data. Scratch that. I love data. I go through data in my spare time. Most of the stuff I do at this blog, for instance, has absolutely nothing to do with my day job. Nothing. But its an opportunity to play with data. My wife usually scratches her head wondering why I do this kind of thing, but everyone needs a hobby and I don’t watch tv.
One thing I’ve learned with data is that its generally important to go back as close to the original source of data as possible. Another is to know something about your sources. Go through the data. Read footnotes.
So in that spirit, I decided to try see what I can learn by looking for data from the era or thereabouts, ideally coming directly from the folks who collect it. I have not succeeded in finding a series that shows what Sumner claims. In fact, data from around that era, particularly on gold prices, isn’t easy to come by. But I have found a few examples.
For instance, Table Number 230 of the 1936 Statistical Abstract of the United States shows the supply of gold in the United States on June 30 of each year (going back annually to 1887, and with selected years before then). The data seems to originate with the Treasury and the Fed, though I haven’t been able to locate the contemporaneous originals.
Footnote 1 reads in part:
By a proclamation of the President dated Jan. 31, 1934 the weight of the gold dollar was reduced from 25.8 to 15 5/21 grains of gold, 0.9 fine. The value of gold is therefore based on $35 per fine ounce beginning June 1934; theretofore it is based on $20.67 per fine ounce.
In other words a couple months after Sumner and other economic historians believe the dollar had started losing value against gold, the Fed and/or the Treasury were reporting to the Census (which publishes the Statistical Abstract) that the price of gold was still exactly the same as it had been.)
Now, its possible the Census or the Fed or the Treasury made a mistake and it went uncorrected by the time of the 1936 Statistical Abstract. So one source is not enough, especially when Sumner and “all economic historians” agree it is wrong.
Which leads to a Fed document called Banking and Monetary Statistics 1914 – 1941. This is from the section on gold (bottom paragraph, left hand column, page 522)
All figures are in dollars, calculated at the rate of $20.67 per fine ounce of gold through January 1934 and $35 per fine ounce thereafter (except that the figures for the year 1934 in Table 159 are based upon the $35 gold price). The change in rate results from the fact that on January 31, 1934, the dollar was devalued by 40.94 per cent in terms of gold in accordance with a proclamation issued by the President.
If you’re curious, $35 – $20.67 = $14.33. $14.33 happens to be 40.94% of $35.
The document is chock full of tables that show, including other things, the monthly value of US gold holdings. Where dollar figures are involved, those tables also carry a helpful note indicating the price as $20.67 an ounce through January 1934, and $35 an ounce thereafter. Note that the Fed valued monthly holdings at $20.67 an ounce in April, May, June, July, August, September, October, November and December of 1933 when, all along, according to Scott Sumner who spent 20 years studying the era and “all economic historians,” insist the price of gold had been rising at the time.
I’ve stumbled on a few other sources as well but they don’t look any different. I’m just not seeing the series that shows the dollar price of gold rising during the months from April 1933 to January 1934.
So what is going on? I’m going to split the baby here and suggest that both Scott Sumner and “all economic historians” are right that there was a devaluation, and the Fed and the Treasury and the Statistical Abstract of the United States were (and are) right that there wasn’t. But the way in which they are right is very definitely not a good thing for Scott Sumner and “all economic historians.”
See, as I said above, I’m not an economic historian, but I did spend my formative years in South American in the 1970s and 1980s. As anyone who spent roughly the same years in the region as I did could tell you, or as any Zimbabwean can do today, during times of turmoil (which can last decades) the official exchange rate can come to bear no relationship with the actual price at which a currency trades against something that is considered more stable and more desirable to hold. Heck, you don’t have to track down someone from Arrgentina or Zimbabwe – ask any European who ever visited the Soviet Block and traded in some Western currency at the airport or the border about how unrealistic official exchange rates could be. In many an economic basket case, the likelihood that a transaction takes place at anything resembling the official exchange rate is similar to the probability that someone walks into a Chevrolet dealership and pays the MSRP, in cash.
And like the MSRP, the official exchange rate has a purpose. Yes, there’s always someone clueless or coerced enough to pay that price. But for the most part, its a fiction that either serves as a baseline for something or papers over something the government wants to really do, usually printing money. Its a handy excuse to get from point A to point B, and if the excuse doesn’t fly, another one will do.
My guess, and I’ll repeat that I’m not an economic historian, is that when FDR and Jones and Morgenthau were picking prices out of the air, it was in that vein. The country was in turmoil when FDR took office, and there were fears that if things got worse there would be an armed insurrection. It wasn’t a time for half measures. My guess is the mood in the White House at the time was best summarized by a quote decades later from the immortal John Candy, “There’s a time to think, and a time to act. And this, gentlemen, is no time to think.”
So what did the fiction of changing the price gold accomplish if nobody else believe that the price had actually changed? I suspect it meant, in practice, that the Reconstruction Finance Corporation could pay more than $20.67 an ounce for gold. And why would the RFC (which, I note, could borrow outside the budget) want to pay more than $20.67 an ounce for gold if that was the price everyone was accepting?
Think of the RFC the way you think of the Fed trying to bail out banks in recent years – loaning money at below market rates to banks who then used the money to buy Treasuries which paid higher rates. In effect, paying more than $20.67 an ounce was a way to funnel riskless profits to banks. (Of course, the RFC often replaced management, but things have gotten permissive as well as more sophisticated in recent decades.)
Which brings us back to Sumner and “all economic historians” being right, at least technically. Yes, the currency was being devalued throughout much of 1933, but no, it wasn’t. Not really. There were a series of fictional devaluations that served a specific purpose, but which nobody else made believe was real (and its possible which almost nobody else was aware were happening – don’t ask me, I’m not an economic historian). Pretending otherwise, and using that fictional data to do an analysis is the equivalent of trying to understand the East German economy in 1974 using the exchange rates a traveler would have received at Checkpoint Charlie during that year.
* The title comes from a book put out by Mad Magazine in the 1970s or 1980s. Sorry I can’t be more specific – it has been a while
Urgh. I sent in the wrong graphs to Dan.
I just resent the new ones. Hopefully he will get them and have a chance to switch the old ones out. Apologies.
I resent it, too!
Hey Mike, you and Sumner’s debate has given me the gist for another post-appreciate! http://diaryofarepublicanhater.blogspot.com/2011/11/i-weigh-in-on-scott-sumner-vs-mike.html
Your post seems to indicate neither of us has completely made our point in a convincing fashion. I think you’re probably right. Overnight I realized there are a few problems with this post, for example… to me this looks a lot like when Argentina had to break its peg with the dollar. There is a difference, and its not a point in this post… its not just FDR’s pronouncements that were artificial, the gold standard itself is artificial. So you had two artificial prices at the same time, each intended to serve a different purpose.
All that said, Sumner wrote me yesterday and gave me permission to post the index he painstakingly collected manually. In fairness to him, I intend to graph it and also put up a table with the numbers but otherwise not comment on it beyond explaining what Sumner said about it. I hope to get to it today, though things that put food on the table take a priority.
I guess next week or sometime soon I’ll put up a post showing a simple function of the federal gov’t’s nondefense spending and its influence on real GDP growth through 1938. (Admittedly the relationship falls apart after 1938. I suspect it has to do with the start of the militarization of the economy.) The numbers should go a long way toward answering your question about how much the New Deal affected growth.
Hey Mike, I again try to get to the bottom of you and Sumner’s debate
In your latest comment you sound closer to getting at a point I can grasp-lol. As far as how much we can assess the benefit of FDR’s fiscal policies vs his monetary policies Eggertsson does actually give a very specific answer.
He estimates that New Deal-fiscally stimulative-policies where 55% responsible for the recovery in output between 1933-37 and 70% responsible for the recovery in inflation during the same period.
FDR issued executive order 6102, which required citizens to deliver gold to the Fed by May 1, 1933 at a rate of $20.67/oz. That order, after being reissued over Magenthau’s signature rather than FDR’s, remained in effect till January 30 of 1934. In the interim, as part of FDR’s overall effort to manipulate commodity prices, the government purchased newly mined gold at prices other than the official $20.67/oz in what Keynes in a letter to the NYT described as “a gold standard on booze”. There was, as you suggest, more than one price. Citizens got $20.67, mines got more. (The vig created by the January 30, 1934 gold price reset was used to pay for some of FDR’s other programs.)
Max Sax – Your reference to Eggertsson is where this debate needs to go. Sumner is arguing the details of gold manipulation and ignoring anything having to do with fiscal policy, in a debate that began with him denying that FDR’s fiscal policies were effective. As if that is all one needs to do to make the case against the success of FDR’s fiscal efforts. Got a link or a reference?
By all means K. First of all, for the record, I listed the links in my post which you can find here
Independently in this link Eggertsson shows that the cumulative effect of FDR’s poicies both monetary and fiscal raised expectations which got us out of the Depression.
Here he specificaly disputes Sumner’s canard that the New Deal-in particular the NIRA-was contractionary, in a piece aptly called “Was the New Deal Contractionary?”
blogger just ate my response to you. Here’s a short version.
Between what you and I seem to think, there’s still one thing missing in the story. Over time, it became more and more evident that there were imbalances due to the price of gold and the dollar was going to be revalued. Heck, at one point Father Coughlin testified before Congress on whether 20.67 an ounce made sense. So at some point expectations started to have an effect.
A recent example of something very similar was the year leading to Argentina’s breaking the peg with the dollar. The banks were the first to notice that Argentina’s stash of FX was asymptotically approaching zero and something had to change (i.e., either that flow of FX had to cease, or the peg had to break). The banks were also connected. The guy taking a bus to work, on the other hand, probably got an inkling that something was wrong a month or so before the end, and certainly had no idea what to do about it. In the middle, some industries were winners and some were losers, and some of that could be tracked very reliably with some indices that were being affected by the very same forces. That doesn’t make any of that policy, nor were the ups and downs in the Argentine economy leading up to the break best explained by the government’s behavior vis a vis the floundering peg.
as far as Sumner’s argument goes, I don’t think he needs the price response to be to policy. Rather, he only needs the price response to be to public perceptions of likely future policy. (That is not necessarily a good selling point for “rational expectations” in general, because there is a falsifiability problem. We are required to assume a change in expectations as an intermediate step between supposed causes and effects. Nowadays, we have inflation expectations surveys and TIPS and such. Then, we did not.)
However, beyond that,…
We have Eggertsson saying that a regime change accounts for the sudden change in output and prices trends from 1933 to 1934. His efforts involve both fiscal and monetary (gold standard) regime changes. Sumner says that he and Gauti believe “rational expectations” explain the sudden change in price and output trends. However, he seems to differ from Gauti in that Gauti “focuses on how the regime change led to higher inflation expectations, and thus reduced real interest rates” while Sumner prefers “to think in terms of specific policy signals sent as rising gold prices changed the future expected gold price, and hence the future expected money supply. I don’t see any non-Ratex explanation that can account for the extraordinary rise in prices and output during March-July 1933. Nominal interest rates didn’t change much, and open market purchases in 1932 (under the constraint of the gold standard) had accomplished little or nothing.”
We are not treated to what Gauti takes into account in his regime change thinking. It would be nice to know. What we can infer is that Gauti has at least some disagreement with Sumner’s view, but aren’t told what that is. Otherwise, there no reason for Sumner to draw a distinction. So, there are folks other than Kimel who don’t buy Sumner’s analysis in full. At a good many points along the way, we are told that Sumner knows lots of important things, and that Sumner sees no good explanation other than his own, and that none of the monetary phenomenna he lists account for the change in prices under examination (he leaves out fiscal policy as if it does not exist – a point with which a capable deconstructionists could run wild).
What we also know is that, in response to a very polite initial post by Kimel, Sumner says he has once again been “attacked” by an Angry Bear. Kimel mentions Keynes, but points to the data. Sumner responds by trying to saddle Kimel with Keynes, requiring Kimel to accept demerits because there is an output gap, and so there must be a liquidity trap, and so increased demand cannot increase prices in a Keynesian world. But Kimel’s argument was based on observations of data, not adherence to dogma. It’s Sumner who is relying on dogma, as well as on a gold price series not widely available, the reliability of which cannot be immediately assessed. At the same time, Sumner repeatedly asserts that he just knows stuff and that the rest of us need to accept that the stuff he says he knows is true.
I really don’t know whether FDR’s gold policy or his fiscal policy or solar flares caused the price and output changes that are under investigation here. What I am pretty sure of is that Sumner’s argument, as provided to us, is very leaky. And apparently, takes any expression of doubt toward his leaky argument as a bear attack.
And K in my view I believe it’s a combination as Gauti clearly thinks it is too. He agrees with Sumner about getting off the gold standard but disagrees that fiscal stimulus has any part to play. I do think your right that “as far as Sumner’s argument goes, I don’t think he needs the price response to be to policy. Rather, he only needs the price response to be to public perceptions of likely future policy.”
The reason I mention Eggertsson is because Sumner appealed to him to deny Sidelsky. So I figured it would be in order to see what Gauti says about fiscal stimulus, especially NIRA which Sumner says was a disaster.
Again as I said above, Eggertsson “estimates that New Deal-fiscally stimulative-policies where 55% responsible for the recovery in output between 1933-37 and 70% responsible for the recovery in inflation during the same period
You can see this on pg 24 of