Lecture Twenty Two
Expectations and Exchange Rate Dynamics; the British
Devaluation of 1992; the Mexican Devaluation of 1994
(Economics 100b; Spring 1996)
Professor of Economics J. Bradford DeLong
601 Evans, University of California at Berkeley
Berkeley, CA 94720
(510) 643-4027 phone (510) 642-6615 fax
delong@econ.berkeley.edu
http://www.j-bradford-delong.net
March 18, 1996
Administration
Mundell-Fleming in (Y, r) Diagram
Uncovered Interest Parity
Exchange Rate Volatility
Collapse of the European "Exchange Rate Mechanism": increase in
r*
Collapse of the Mexican Peso in 1994
What Is To Be DoneWhen the Peso Collapses?
Administration
Try hard to finish open economy issues by spring break
Begin by running through the topic list for this lecture:
Mundell-Fleming in (Y, r) Diagram
Note that the Mundell-Fleming model is, in sense, just our old IS-LM
model with an extra NX(e) term in the IS equation, and an extra
restriction--r=r*--imposed by world capital markets.
So what happens to force r = r*? Suppose that, at current exchange
rates, the IS curve (augmented by NX(e), but note: no *: IS, not IS*)
and the LM curve intersect at r > r*?
U.S. interest rates exceed world interest rates; money flows in;
people bid up the price of U.S. assets--which is the same thing as
the exchange rate rising. And as the exchange rate rises, the IS
curve shifts in and to the left because the trade balance
deteriorates. Equilibrium? When the exchange rate has risen enough to
move IS back far enough that there is no desired short-term
speculative capital inflow. (Distinguish between "speculative" and
"fundamental" capital inflows).
Uncovered Interest Parity
Exchange rates, since the early-1970s breakdown of the Bretton Woods
fixed exchange rate system, have been much more volatile than anyone
had predicted. Why? A bunch of reasons--one that governments were let
loose to pursue independent monetary policies. But here I want to
focus on expectations and exchange rate dynamics.
Begin by noting that so far we have assumed a fixed exchange rate
that sometimes jumps in response to shocks. This seems inconsistent.
Suppose that people have a pretty good idea about what the long-run
exchange rate is going to be--call it e*--and notice differences
beween e, the current exchange rate, and e*.expected exchange rate
movements:
Invest in the foreign country, and earn r*
Invest in the home country, and earn r plus or minus the expected
exchange rate movement: e*- e
So that the right equation to go alongside of:
Y = C(Y-T) + I(r) + G + NX(e)
and
M/P = L(i, Y)
isn't r=r*, but is instead:
r = r* + (e - e*)
When e is above e*, the exchange rate is expected to
depreciate, and so domestic real (and nominal) interest rates
will be higher than world market values. When e is below e*,
the exchange rate is expected to appreciate.
Implications?
The higher is the exchange rate e, the higher are domestic interest
rates--and so the LM* curve is upward sloping... It rotates clockwise
around its intersection with the e=e* line... Call this
curve--allowing for expected exchange rate changes and their effects
on domestic interest rates--LM**
The higher is the exchange rate e, the higher are domestic interest
rates--and the lower is investment. So the IS* curve also rotates, in
this case counterclockwise, around its intersection with the e=e*
line. Call this curve--allowing for expected exchange rate changes
and their effects on domestic interest rates--IS**
If the current exchange rate is above e*, then domestic
interest rates will be somewhat higher--and the exchange rate
somewhat lower--than in the static expectations case. If the current
exchange rate is below e*, then the exchange rate will be
somewhat higher--and the domestic interest rate somewhat lower--than
in the static expectations case.
Exchange Rate Volatility
So let's take a look back at the case we considered at the end of
last week: sudden increase in foreign interest rates r*; giving the
government a choice between recession (as they pull LM curve back to
keep exchange rate from falling) or decline in the exchange rate. And
suppose everyone expects that in the long run the exchange rate will
decline.
So the rise in foreign interest rates pulls the IS* curve down and to
the left, and also pushes the e* curve, the equilibrium long-run
exchange rate line, downward. Then we move to the IS** curves and
LM** curves by rotating the IS* and LM* curves counterclockwise and
clockwise, respectively, around their intersections with the new e*
line.
We find that the current exchange rate e falls, and falls by a
lot--by more than the decline in e*. Why? Well look back at your new
foreign exchange market equilibrium condition:
r = r* + (e - e*)
If r were equal to or greater than r*--that would mean that the
government had responded to the increase in r* by inflicting equal or
greater degrees of monetary contraction on the domestic economy. In
which case the LM* curve would have shifted in by enough (or more
than enough) to maintain the previous exchange rate. Hence no
expectations of LR devaluation--and no fall in the e* line.
The whole point of letting e fall was to keep r from having to rise
as much as r*--and to avoid the recession that would come when it
did.
Hence the exchange rate "overshoots" its long-run value: falls more
than its long-run value falls, and then slowly climbs back up.
Under a floating rate system, not only do exchange
rates shift because governments follow inconsistent monetary
policies, but exchange rates fluctuate much more than do "LR
equilibrium" rates because of overshooting.
Rudiger Dornbusch saw this in the early 1970s, before our current
experience with flexible exchange rates more than a couple of years
old. Rudiger Dornbusch really smart...
Collapse of the European "Exchange Rate Mechanism": increase in
r*
All this applies very neatly to the collapse of the ERM in Britain in
late 1992: Britain wound up with (a) an undervalued exchange rate
(expected to appreciate), low domestic interest rates, and something
of a small boom...
All this does not apply neatly to Mexico at the end of 1994--even
though the analogies with Britain appear quite close...
Collapse of the Mexican Peso in 1994
But things can get more complicated and more difficult very
quickly. Suppose, for example, that we have:
r = r* + (e - e*) + (risk premium); and suppose that the risk premium
= s(e(1) - e(2))--that people are scared of the fall in the exchange
rate
possibly because they are irrational "positive feedback" traders
possibly because large moves in exchange rates bankrupt lots of
people in the domestic economy
possibly because they fear that--even if it is irrational--that
others will stampede out and make it impossible to get their
money...
If the risk premium turns out to move faster and stronger in response
to devaluation, then you have Stanley Fischer's nightmare...
There is No Exit
Stanley Fischer's moment of terror
What Is To Be DoneWhen the Peso Collapses?
So what do you do? You calm people down. You point out that there is
a perfectly good equilibrium out there in which the economy is
fine--that in large part the risk premium has risen so far for no
reason other than the fact of its rise.
When worse comes to worst, you provide liquidity--and hope that it is
a liquidity crisis, not a solvency crisis, and that the high real
interest rates reflect fear of the crisis itself and panic and not
knowledge by the market of bad news that you do not yet know.
A link to our (Sherman Robinson's, Chris DeLong's, and my) draft
paper on The
Mexican Peso Crisis
When it works, it is one of the few free lunches out there that
economists can actually help the world obtain...
>
