Friday, January 20, 2012

Me and the "Mussa Puzzle"

The international economist, and former chief economist of the IMF, Michael Mussa, died earlier this week.  I never met the man, and reading these remembrances from the Peterson Institute makes me sorry I didn't.  But we have done some work in the same area: one of my own favorite papers confirms one of Mussa's most well-known findings, that the behavior of real exchange rates depends on whether nominal exchange rates are fixed or floating.  Paul Krugman explains:
Probably his most influential paper — certainly the one that had the biggest impact on me — was his 1986 paper (pdf) on currency regimes and the behavior of real exchange rates. This bore on the question of whether exchange rate changes make adjustments in relative costs and prices easier; it bore more broadly on the question of whether prices are flexible, as fresh-water economists like to assume, or instead sticky in nominal terms.

Mussa had a simple but powerful insight: if prices were flexible, then all relative prices should be determined by “real” factors, and their behavior shouldn’t change if, say, a country goes from a fixed exchange rate to a flexible rate or vice versa. As he pointed out, this proposition could be tested using a natural experiment, the breakdown of Bretton Woods and the move to floating rates. Did the behavior of real exchange rates — relative price levels expressed in a common currency — change?
I've seen the large increase in real exchange rate volatility under floats referred to as "the Mussa puzzle", though, as Krugman points out, "sticky" prices are a straightforward explanation.

I was able to address one significant limitation of Mussa's analysis. His paper relied heavily on a comparison of the periods immediately before and after exchange rates began to float in the early 1970's.  That left open the possibility that some other changes at around the same time led to the difference in real exchange rate behavior.  Vittorio Grilli and Graciela Kaminsky made that argument in a 1991 Journal of Monetary Economics paper:
In particular, the high volatility of the real exchange rate since the breakdown of the Bretton Woods system in the early 1970s may have arisen as a consequence of factors unrelated to the nominal exchange rate regime - such as the two oil price shocks of the 1970s or the wide fluctuations in interest rates during the late 1970s and early 1980s.
However, in "Across Time and Regimes: 212 Years of the US-UK Real Exchange Rate" (Economic Inquiry 48:4 [October 2010]), I exploited the fact that the exchange rate between the US and Britain has alternated a number of times between fixed and floating.  One of the tables in the paper shows the average monthly change in the real exchange rate during different regimes.
  1. Jan. 1794-Apr. 1821 (floating): 2.66%
  2. May 1821-Dec. 1861 (fixed): 2.02%
  3. Jan. 1862-Dec. 1878 (floating): 2.52%
  4. Jan. 1879-June 1914 (fixed): 1.17%
  5. July 1914-Mar. 1919 (wartime controls): 1.75%
  6. Apr. 1919-Apr. 1925 (floating): 2.03%
  7. May 1925-Aug. 1931 (fixed): 0.98%
  8. Sept. 1931-Aug. 1939 (floating): 1.77%
  9. Sept. 1939-Sept. 1949 (wartime controls): 1.52%
  10. Oct. 1949-July 1971 (fixed): 0.49%
  11. Aug. 1971-Dec. 2005 (floating): 1.97%
The pattern of higher volatility in floating periods is a robust one (and there is more statistical evidence in the paper). However, the differences between fixed and floating are smaller in earlier periods, which suggests that perhaps price stickiness has become more important over time.

No comments: