Page 176 - DECO503_INTERNATIONAL_TRADE_AND_FINANCE_ENGLISH

International Trade and Finance

Notes Interpretations
Various explanations of the results are suggested by the existing literature and by readers of earlier
versions of this paper. Some of these explanations are discussed now. No explanation is necessarily
complete, and they are not mutually exclusive. Moreover, generous readers of earlier drafts are not
responsible for my paraphrasing of their comments.
An inefficient foreign exchange market
The interpretation of the results above is based on the hypothesis that the assessment of E(S t + 1 – S ) in
t
F – S is efficient or rational. An alternative hypothesis is that the negative slope coefficients in the
t
t
regressions of S – S on F – S reflect assessments of E(S S )that are consistently perverse realtive
t + 1 t t t t + 1 t
to the true expected value of the change in the spot rate. The large positive coefficients in the F – S t +
t
regressions are then a simple consequence of the complementarity of the F – S and S – S
1 t t + 1 t + 1 t
regressions rather than manifestation of movement in rationally determined premiums. Under this
interpretation, the similarity of the regression results for the two subperiods indicates that market
irrationality in forecasting exchange rates is not cured by continued experience with flexible exchange
rates.
Government intervention in the spot exchange market
A kind of ‘market inefficiency’, suggested by Richard Roll, can result from government intervention
in the spot foreign exchange market. For example, suppose forward rates are determined by the
interest rate parity condition (7) and interest rates in different countries rationally reflect their expected
inflation rates. Left to the open market forces suggested by purchasing power parity, spot exchange
rates would tend to move in the direction implied by the forward-spot differential F – S . Government
t
t
logic and obstinacy, however, may be inversely related to natural market forces. Governments may
support their currencies more vigorously (through open market operations, trade restrictions, and
restrictions on capital flows) the stronger are the market forces, like differential expected inflation rates,
which indicate that the currency should depreciate. They may try to move back toward a free market
equilibrium by changing the direction of the underlying factors pressuring the exchange rate, like
differential inflation rates, rather than by letting adjustments take place through the exchange rate.
The doomsday theory
Michael Mussa suggests that there are episodes, often brief, during which the distribution of anticipated
changes in exchange rates is highly skewed. For example, market participants may assess a small
probability that a country will change its monetary policy so that its inflation rate will rise dramatically
relative to other countries. The result may be a highly skewed distribution of anticipated inflation
rates, which in turn increases interest rate differentials and forward-spot exchange rate differentials
between this country and other countries. Since the phenomenon centers on skewness that exists for
brief periods, the ex post drawings from the distributions of anticipated inflation rates and changes in
exchange rates are likely to be below the ex ante means. This creates negative sample correlations
between changes in exchange rates and forward-spot differentials which would not be observed if
the skewed distributions were sampled over longer periods.
Stochastic deviations from purchasing power parity
Stockman (1980) and Lucas (1982) develop international models in which shocks to real activity
work in part through money demand functions to drive changes in inflation and exchange rates.
Fama (1982) also argues that through the workings of a standard money demand function and inertia
in money supply, variation in anticipated real activity in the U.S. leads to variation in expected inflation
of the opposite sign. Fama and Gibbons (1982) argue that expected real returns on U.S. nominal
bonds are also driven by and move in the same direction as anticipated real activity. With a somewhat
different story in which monetary shocks cause changes in real variables, Tobin (1965) and Mundell
(1963) likewise conclude that the expected real and expected inflation components of nominal interest
rates are negatively correlated.

170 LOVELY PROFESSIONAL UNIVERSITY