Tuesday, October 29, 2013

Did nominal interest rates fail to spike during the run on the Franc?

Paul Krugman has a new paper investigating the role of currency regimes on the nature of sudden stops, where he argues that a sudden stop would not be contractionary for a country that has a floating exchange rate. The big chunk of the paper is theoretical, but towards the end of the paper, Krugman digs out an interesting case study – France in the 1920s – that helps put his argument into historical perspective.
“The closest one can come to historical situations at all resembling the situation of today’s floating-rate debtors is that of France in the 1920s, which emerged from World War I with a large debt burden that it had great political difficulty dealing with, and did in fact experience a run on the franc.”
“The basic insight is that France grew quite strongly as the franc slid due to loss of confidence […] What actually happened was a sharp fall in the franc, substantial inflation, but no interest rate spike and a quite good performance in terms of real output. Nothing in that story validates the conventional wisdom [about the vulnerability of the United States and other nations to a loss of capital inflows].”
I happened to be particularly interested in that part of the paper, as I’ve recently been trying to understand why nominal interest rates did not spike following the election of the Popular Front in 1936 and the country’s departure from the Gold Standard a few months after. Krugman seemed to point to a similar situation: despite the Franc losing two thirds of its value between 1922 and 1926, he notes that there was “no spike in interest rates”.

Source: Paul Krugman
In fact, Krugman’s diagram shows that “rates actually fell during the first leg of franc depreciation”, which is even more puzzling. After a bit of research, I figured out that Krugman is using the average yield for a portfolio of 180 securities of varying returns from the NBER macro history database (series 13027). One would have to turn to the original sources to understand exactly how that series was constructed, but it strikes me as a choice that would warrant a bit of discussion.
As I show below, the more standard measure of nominal interest rates for that period – the rates of returns calculated on the French consols or rentes – gives a different picture. According to this series, nominal interest rates did increase over this time period. The spike, although seemingly not huge, is an understatement of the increase in nominal interest rates as these returns “are averages of the rates during these troubled times and of the rates that were expected to prevail afterwards, once stability was reached again” (Hautcoeur and Sicsic (1999)).  Assuming a long-term interest rate of 3.75 per cent, Hautcoeur and Sicsic (1999) show that the 2 percentage point spike in the nominal interest rates of the consols that I document in the diagram below correspond to an increase in medium-term nominal interest rates of about 8 percentage points.
Note: Note: The first vertical line indicates the start of the depreciation according to NBER series 14004b. The second vertical line indicates the Poincare stabilization

So nominal interest rates did spike, although clearly not enough to make real interest rates increase. Overall, the Krugman story holds, since he only needs a decrease in real interest rates for the sudden stop to be expansionary. But unless I missed something, that small part of his paper on nominal interest rates needs to be revised.
Jeremie Cohen-Setton (@JCSBruegel) is a PhD candidate in economics at UC Berkeley and an Affiliate Fellow at Bruegel. He specializes in Macroeconomic Policies and Macroeconomic History and worked previously as an economist at HM Treasury and at Goldman Sachs. Jeremie blogs at ecbwatchers.org and is the main author of the blogs review at bruegel.org.


  1. Anonymous10:39 AM

    Good catch, Krugman should revise his paper and either justify his choice or use the more commonly used benchmark.

  2. Would the consol be influenced by the inflation Krugman mentioned?

  3. "What actually happened was a sharp fall in the franc, substantial inflation, but no interest rate spike and a quite good performance in terms of real output."

    According to Michael Bordo's dataset nominal GDP increased at an average annual rate of 19.3% from 1921 to 1926. According to Angus Maddison's dataset real GDP increased at an average annual rate of 7.6%. Thus the GDP implicit price deflator rose at an average annual rate of 11.0%.

    It seems to me that Krugman is right. Any increase in nominal interest rates is too trivial to be considered a "spike" given the steep rise in prices.

  4. France suffered 6 million dead and wounded in World War One. That's more than one male out of three. Or more than one male of working age, out of two.
    To use a country that has suffered such a trauma and economic shock in any general theory is fraught with a massive, intrinsic error. This is exactly the sort of outlier that ought to be eliminated from general consideration.

  5. Anonymous2:46 PM

    Krugman is revising his paper.


  6. And notice how Krugman reacted instantly to this suggested correction and is not, like Alan Greensoan and various other VSPs insisting that he got it perfectly right from the get go.

    This little incident, almost as much as anything else, tells you who is your real Daddy when it comes to Econ.

    And kudos to Cohen-Setton for setting up the chain of events that made it happen.

  7. John S1:59 AM

    Well, if Krugman is so amenable to correct his errors, will he correct the misleading impression he gave that it was the gold standard, rather than restrictive US banking regulations, which was the cause of the 19th century US banking panics?

    Krugman wrote:

    "Now, the gold bugs will no doubt reply that under a gold standard... there wouldn’t be major financial crises. And it’s true: under the gold standard America had no major financial panics other than in 1873, 1884, 1890, 1893, 1907, 1930, 1931, 1932, and 1933. Oh, wait."


    Steve Horwitz's reply:

    "The federal and state governments played a huge role in the banking industry and it was those regulations that were responsible for the pre-Fed panics.

    The two most relevant regulations were: 1) the prohibition on interstate banking, which created overly small and undiversified banks that were highly prone to failure; and 2) the requirement that federally chartered banks back their currency with purchases of US government bonds, which made it prohibitively expensive to issue more currency when the demand rose, leading to the currency shortages and resulting panics that culminated in the Panic of 1907."


  8. The French economic system at the time was a sleepy backwater. Most Frenchmen kept their francs in their mattress. Americans were amazed to find that the French still crossed their checks. I had to look it up in a 19th century business reference book. It was dying out in the US in the 1880s as best I can tell. Crossing a check was writing a sort of endorsement across the front at a right angle to the main body of the check (with the "pay to the order of") that stated that the payment would only be made at a particular bank branch. Yup, not every Frenchman accepted new fangled ideas like having a check clearinghouse.

    If you think finance and the economy are two different animals today, you haven't studied early 20th century France.

  9. And that is how scholars correct their work and modify their conclusions when presented with new information. Thank you, gentlemen, for a brilliant example of economics as a not-so-dismal science.