Professional Documents
Culture Documents
(IMF Staff Papers) The Impact of Monetary and Fiscal Policy Under Flexible Exchange Rates and Alternative Expectations Structures
(IMF Staff Papers) The Impact of Monetary and Fiscal Policy Under Flexible Exchange Rates and Alternative Expectations Structures
2
The presence of stochastic variables means that the forward rate could differ
from the actual future exchange rate whenever there are "surprise," or unantici-
pated,
3
occurrences between time / and / + 1.
The Box and Jenkins time series methodology assumes that the current value
of any variable (e.g., the rate of inflation—Pi) can be represented as a function
of past values of the variable and a stochastic term. One could thus write the
current rate of inflation as
P, = iri P,_i + ir2Pi-2 + . . . + at
or as ir(B) P, = at
The demand for real balances is taken as a function of the level of real
income (y) and the domestic nominal interest rate (r). In log-linear form,
4
See Appendix I for a summary of notation. The symbols are generally the
same as those used by Dornbusch (1976). The exchange rate is defined as
domestic currency/foreign currency.
(2)
where m, p, p*, e, and y denote, respectively, the logs of the nominal
quantity of money, the price of the domestic good, the price of the
imported good, the exchange rate, and real income.5 The supply of real
balances equals the level of nominal money divided by the "price" level.
Since there is a domestic good and an imported good, the price level that
is relevant for portfolio owners must reflect the effects of changes in the
prices of both goods. For simplicity, it has been assumed that the true
price level can be represented by a log-linear, fixed-weights index that
depends on the price of the domestic good and the domestic equivalent
of the price of the imported good (which will equal the exchange rate
multiplied by the price of the imported good). The price index is thus
given by ap + (1 — «)e + (1 — a)p* where a represents the weight that
the price of the domestic good receives in the general price index.
The demand for the domestic good (D) is taken as a function of the
relative price of the domestic good, the level of real income, and the real
interest rate. Thus,
(3)
e
where p is the expected rate of change in the price of the domestic good.
The real interest rate has been defined to equal the nominal interest rate
less the expected rate of inflation as measured by our price index. Since
the foreign price of the imported good is fixed, the expected rate of price
change depends only on the expected rates of change in the price of the
domestic good and the exchange rate.
The rate of increase in the price of the domestic good (p) is assumed to
depend on the excess demand for the domestic good. Thus,
(4)
EXPECTATIONS
To examine the influence of expectations on the adjustment process,
5
It is assumed that real income is fixed at the full employment level.
(7)
(long-run expected change in p" owing to change in m or g)
(8)
(9)
(medium-term relationship between p" and p)
(10)
(long-run expected change in «" owing to change in m or g)
with
(ID
(12)
(medium-term relationship between ee and «)
The adjustment processes generated under these different expectations
structures can be compared if we first consider an increase in the money
supply.
6
Footnotes 8 and 23 explain why the price of the domestic good rises with
an increase in the money supply and falls with an increase in government
spending.
7
Footnotes 8 and 23 explain why the exchange rate rises with an increase in
the money supply and falls with an increase in government spending.
the domestic good and a depreciated exchange rate.5 But let us examine
how the economy will move from the initial equilibrium to its new long-
run equilibrium by first considering rational expectations.
RATIONAL EXPECTATIONS
(14)
The adjustment process implicit in the preceding relationships can be
described with the aid of Figure 1. The p = 0 curve represents the combina-
tions of p and t that will yield a stable price for the domestic good. Any
point to the right (left) of this line will yield a rising (falling) price of the
domestic good. The k = 0 curve portrays the combinations of p and e
that will generate a stable exchange rate. Any point to the right (left) of
8
In the long run, the conditions for steady equilibrium are given by
Thus,
These results also explain wh> (in equation (8)) and (in equation (11)).
9
In Appendix II, it is shown that the adjustment process described by equations (13)
and (14) will be stable only if 1 — ainr>0. This condition is required to ensure that the
goods market is stable. Any disturbance in the goods market that generates an increase
in the price of the domestic good (p>0) will also reduce the real interest rate (by — ap).
A decline in the real interest rate, however, will stimulate spending on goods and will
lead to a further increase in the price of the domestic good (by arairp). If the price level
is ultimately to converge to a stable steady state value (i.e., the goods market is to be
stable), then the induced change in p must be smaller than the initial change, which
means that 1 —trour must be greater than zero.
It is also assumed that dp/de=Tr[5\—(1 — a)<ra]/(l — oW)>0. This essentially requires
that the relative price effect generated by a change in the exchange rate outweigh the
real interest rate effect. One could reverse the sign of this expression and still derive the
same qualitative conclusions as described in the text.
The arrows in Figure 1 describe the movement of p and e when they are off the p—0 and
i=0 curves.
11
This reflects the assumption that there is perfect capital mobility, which
implies that the yields on all bonds will be equalized across countries (after
allowance is made for expected exchange rate movements).
12
When combined with the adaptive expectations mechanisms described in equa-
tions (5) and (6), the adjustment processes in the money and goods markets are
given by
m-ap-(l-a)(-(l-a)p*=-\r+<t,y=-\(r*+6e')+^y (15)
and
P=v[t>(i+p*-p}-<,(r-ap<-(\-a}i'}+(y-\}y]
or, using r*+'te=r,
i>=*[!>(t+P*-p-)-v(r*+at°-ap°-)+(y-V)y} (16)
These four relationships describe the adjustment process for the actual and expected
price of the domestic good (p andp') and the actual and expected exchange rate (e and «•)•
As shown in Appendix III, this set of relationships can be reduced to a system of three
differential equations in p, pe, and <?. A necessary condition for this system to be stable
is that 1— a—|3X>0. This is equivalent to the condition that a depreciation of the ex-
change rate will work to reduce any excess supply of real money. An increase in e will
reduce the real money supply by —(1 — a)de (where 1 —a is the weight that the price of
the imported good receives in the overall price index). At the same time, however, an
increase in e will raise the expected rate of depreciation of the exchange rate by dt°=0d<-
(equation (6)), which will thereby increase the domestic interest rate by an equivalent
amount. As the domestic interest rate rises, however, the demand for money will de-
cline by —Xrfe e = — X(3rfe. Thus, a depreciation of the exchange rate will reduce any initial
supply of real balances as long as 1— <*>/3X or 1— a—/3X>0.
p=0
14
If i'=Q initially, then continuous equilibrium in the money market will be main-
tained if «=[Xr*—<t>y—ap]/(l— a). The EE curve thus has a negative slope equal to
-(!-«)/«.
the domestic interest rate is initially tied to the world interest rate, the
demand for real money will not be affected by the increase in the money
supply. Money market equilibrium can therefore be maintained only if
exchange rate movements increase the general price level sufficiently to
offset the increase in the nominal money supply and to leave the real
money supply unchanged. Since the price of the imported good carries
the weight 1 — a in the aggregate price index, the exchange rate must
depreciate by de — dm/(\ — a) > dm.
In terms of the analysis developed in Figure 3, the increase in the money
supply will initially shift the EE curve to the right but will leave the p = 0
curve unchanged. (See Figure 4.) The initial increase in e that is required to
maintain money market equilibrium is given by the move from A to B in
Figure 4. Thus, just as under rational expectations, the initial exchange rate
response to an increase in the money supply will involve an overshooting
of the ultimate exchange rate level.
While the initial overshooting of the exchange rate will therefore occur
under both adaptive and rational expectations, the amount by which the
exchange rate overshoots its long-run value will be different under the
two alternative expectations structures. As has been shown, the overshoot-
ing of the exchange rate is part of the process by which instantaneous
FIGURE 4. INITIAL IMPACT OF AN INCREASE IN THE MONEY
SUPPLY UNDER ADAPTIVE EXPECTATIONS
E'
15
One would usually expect an increase in the money supply to drive down
the domestic interest rate initially. This cannot happen in our model as long as
we maintain the assumption of perfect capital mobility. In this situation, the
domestic interest rate can differ from the world rate only by the expected rate
of depreciation. If one were to assume that there was less than perfect capital
mobility, then it is possible that one could see both a rise in the expected rate
of 16depreciation and a fall in the domestic interest rate.
In the long run, the percentage increases in the exchange rate and the price
of the domestic good will exactly equal the percentage increase in m. The cycli-
cal nature of the adjustment paths will be determined by the characteristic roots
of the three differential equations that describe the behavior of the economy under
adaptive expectations. (See Appendix III.)
A. RATIONAL EXPECTATIONS
SEMIRATIONAL EXPECTATIONS
17
The learning component of the semirational structure will initially be fixed.
exchange rate, the domestic interest rate will rise above the world interest
rate by the amount of the anticipated depreciation. This is represented by
the movement of the interest rate from A to B in Figure 6.18 This initial
increase in the domestic interest rate will reduce the demand for money
(by —Arfee), which will work to further increase the initial excess supply
of money. To restore equilibrium in the money market, the exchange rate
must therefore initially undergo a discrete depreciation (which will raise
the price level and reduce the supply of real money) by an amount that is
larger than the initial depreciations under either adaptive or rational ex-
pectations. The initial depreciation must be larger under semirational
expectations because it must offset the excess supply of money created by
both the increase in the nominal money supply and the decline in the
demand for money induced by the rise in the domestic interest rate. In
contrast, the initial depreciation (under adaptive expectations) has to
offset only the increase in the nominal money supply, since the domestic
interest rate will initially be tied to the world interest rate. And under
rational expectations, the realization that the exchange rate will initially
overshoot its long-run value creates the anticipation of a long-run apprecia-
tion of the exchange rate that drives down the domestic interest rate. Since
a lower interest rate will increase the demand for money, however, a smaller
initial depreciation is required to restore money market equilibrium.19
18
Figure 6 portrays both the initial and the medium-term movements in r, e,
and19 p. Appendix IV examines the stability of the actual adjustment path.
Using equations (10)-(12), it can be seen that the initial response of semi-
rational expectations to an increase in m equals
This increase in the expected exchange rate will raise the domestic interest rate by an
equivalent amount. In the money market, equilibrium will be maintained if the exchange
rate depreciates (de > 0) by
20
This analysis is based on the assumption that expectations are formulated
at the beginning of the trading "day" and are revised only at the beginning of
the next "day," once market participants have observed the trading prices that
have prevailed during the current "day."
interest rate reflect the assumption that asset markets clear instantaneously.
If asset markets adjusted more gradually, then one would see a gradual
upswing of the interest rate followed by a gradual decline, instead of the
discrete changes described in this paper.
As shown in Figure 6, these price, exchange rate, and interest rate
movements will interact to produce a smooth asymptotic path to the long-
run equilibrium. (See Appendix IV for an analysis of this path.) During
this part of the adjustment process, the interest rate will gradually rise to
equality with the world interest rate, the price of the domestic good will
increase to its long-run value, and the exchange rate will appreciate to
its new steady value.
What then are the basic differences and similarities between the adjust-
ment processes under the three different expectations structures following
an increase in the money supply? The major difference between the various
adjustment processes lies in the behavior of the domestic interest rate.
Under rational expectations, a sharp initial decline is followed by a gradual
return to equality with the world interest rate. Under adaptive expecta-
tions, in contrast, the interest rate may gradually rise above the world
level and then return to its initial level, or there may be a series of cycles
with a rising and falling interest rate. And under semirational expecta-
tions, the interest rate will undergo a sharp rise, then a sharp decline, and
finally a gradual recovery to the initial value. A second difference between
the adjustment processes is that only the adaptive expectations structure
can possibly involve a cyclical adjustment process. Under either rational
or semirational expectations, the initial discrete adjustments will be fol-
lowed by an asymptotic movement to the long-run equilibrium. However,
one element that is shared by all three adjustment processes is an initial
exchange rate depreciation that is larger than the increase in the money
supply. This discrete depreciation is required to restore money market
equilibrium, and the existence (but not the size) of this overshooting is
independent of the expectations structure.
To incorporate fiscal policy into this analysis, we must allow for the
impact of government spending and taxation on asset and goods market
behavior. To simplify the analysis, it is assumed (1) that government
spending can be expressed as a proportion (g) of private spending (/)),
(2) that government tax revenues are raised via a lump-sum tax and a tax
on income, and (3) that all government deficits or surpluses are financed
In the asset markets, the impact of fiscal policy will be generated via the
issuance of bonds and the effects of changes in taxes on disposable income
and thereby on asset demands, whereas the effects of fiscal policy on the
goods market will be produced by changes in government spending and
the impact of changes in taxes on disposable income and thereby on
consumption demands. The conditions for money market equilibrium
will be identical with those described in equation (2). We are thus assum-
ing that the level of output (rather than disposable income) is still the best
indicator of the transactions demand for money. Since it has been assumed
that government spending (G) can be expressed as a proportion of private
spending on the domestic good, the total demand for the domestic good
will be G + D = gD + D = (1 + g)D. Equation (4) must therefore be
rewritten as
RATIONAL EXPECTATIONS
continuously reduce the real supply of money. During this period, money
market equilibrium can therefore be maintained only if the exchange rate
appreciates and thereby reduces the price of the imported good. This con-
tinuous exchange rate appreciation will keep the domestic interest rate
below the world rate during the move to the long-run equilibrium.
Figure 8 portrays the movements of e, p, and r during the adjustment
FIGURE 8. ADJUSTMENT PROCESS GENERATED BY AN INCREASE IN
GOVERNMENT SPENDING UNDER RATIONAL EXPECTATIONS
ADAPTIVE EXPECTATIONS
the feedback between changes in the actual and expected levels of both the
exchange rate and the price of the domestic good. The initial excess demand
for the domestic good will drive up the price of that good and will thereby
reduce the real supply of money. To maintain money market equilibrium,
the exchange rate must therefore begin to appreciate to lower the domestic
price of the imported good. This appreciation of the exchange rate will
also lower the expected value of the exchange rate and will lead to a de-
cline of the domestic interest rate below the world interest rate. From this
point on, however, exactly what path the domestic variables will follow
to the new steady state will depend on whether the adjustment path is
direct or cyclical. In Figure 10, we have illustrated the paths that will be
followed by e, /?, and r if there is a direct approach to equilibrium. A com-
parison with the rational expectations paths (Figure 8) indicates that the
economy's response to an increase in government spending will be quite
similar under both adaptive and rational expectations.25 The initial
changes in r and e provide the only differences between the two adjust-
ment processes. The behavior of the domestic interest rate following an
increase in government spending may once again provide the best evidence
of whether expectations are rational or adaptive. Under rational expecta-
25
This assumes that there is a direct rather than a cyclical adjustment process
under adaptive expectations.
tions, one will see a sharp initial decline in the interest rate followed by a
gradual recovery to equality with the world interest rate, whereas under
adaptive expectations, one will see a gradual decline and a gradual re-
covery of the domestic interest rate.
SEMIRATIONAL EXPECTATIONS
(Confd on p. 562.)
below the world interest rate. The only really distinguishing difference
between the various adjustment paths is that the domestic interest rate
will not undergo an initial discrete decline under adaptive expectations.
It is clear that our model of the effects of monetary and fiscal policy
under floating exchange rates is based on a number of assumptions that
not only simplify the analysis but also limit its generality. The most
important restrictions are imposed by the assumptions of a small country
and perfect capital mobility. The small-country assumption implies that
the economy faces a fixed world price for the imported good and a given
international interest rate. And the perfect capital mobility assumption
means that the domestic interest rate can differ from the world interest
rate only by the expected rate of depreciation of the exchange rate. Thus,
apart from exchange risks, domestic bonds are viewed as perfect sub-
stitutes for international bonds. As has been shown, these assumptions
are especially important for analyzing the effects of expansionary fiscal
policy, for they imply that the authorities can issue any quantity of bonds
that they desire without affecting the international interest rate. A more
complete analysis would make the international interest rate sensitive to
the quantity of bonds issued. In this situation, a higher level of debt-
financed government spending would raise the international interest rate
applied to domestic bonds. This means that one would see a smaller
appreciation than in the small-country case, or possibly even a deprecia-
tion of the exchange rate. The incorporation of these effects will be an
important part of future work.
It is usually argued that the manner in which the private sector formu-
lates the expectations will play a crucial role in determining the economy's
response to monetary and fiscal policy. Table 1 summarizes the nature of
the adjustment processes generated by either an increase in the nominal
money supply or an increase in government spending for a small, open
economy with a floating exchange rate. What is surprising about Table 1,
however, is the rather similar qualitative nature of the adjustment processes
under the different expectations structures. Perhaps the only way to estab-
lish which expectations structure best describes the economy's actual
expectations structure is to examine the behavior of nominal interest
rates following an increase in the money supply. An increase in the nominal
money supply will initially force the interest rate to decline if expectations
are rational, and to first rise sharply and then decline sharply if expecta-
tions are semirational; it will have only a gradual impact on the interest
Expectations Structure
Rational Adaptive 1 Semirational
Effects of increase in
money supply
Initial impact
Price of domestic good None None None
Interest rate Decline None Sharp rise fol-
lowed by
sharp decline
Exchange rate Depreciation Depreciation Depreciation
Medium-term impact
Price of domestic good Gradual rise Gradual rise Gradual rise
Interest rate Gradual rise Gradual rise Gradual rise
and then
decline
Exchange rate Gradual ap- Gradual ap- Gradual appre-
preciation preciation ciation
Effects of increase in
government spending
Initial impact
Price of domestic good None None None
Interest rate Decline None Decline
Exchange rate Appreciation None Appreciation
Medium-term impact
Price of domestic good Gradual rise Gradual rise Gradual rise
Interest rate Gradual rise Gradual de- Gradual rise
cline fol-
lowed by
gradual
rise
Exchange rate Gradual ap- Gradual ap- Gradual appre-
preciation preciation ciation
1
If direct approach to equilibrium.
APPENDICES
I. Notation
Unless otherwise noted, each variable represents the log of the original variable
(a capital letter denoting the nonlog value):
m = nominal money supply
p = price of domestic good
p* = international price of imported good
c = exchange rate (domestic currency/foreign currency)
y = real output
r = domestic interest rate (not as log)
r* — world interest rate (not as log)
pe = expected price of domestic good
ee — expected exchange rate
E€ = long-run expected exchange rate (under semirational expectations)
BG = stock of outstanding government bonds (not as log)
to = lump-sum taxes (not as log)
h = income tax rate (not as log)
YD = nominal disposable income (not as log)
g = government spending as proportion of private spending on the domestic good
(not as log)
D
B - private holdings of domestic government bonds (not as log)
D = domestic demand for the domestic good (not as log)
For this system to have a stable saddle point path, one must have
(21)
(22)
The saddle point path that will allow the economy to return to a stable equilibrium
and still maintain perfect foresight will be given by
Since both the numerator and denominator go to zero as the steady state equi-
librium is approached, one can evaluate the slope of the preceding expression by using
L'Hopital's rule. Thus,
then this system Jias two roots—one positive and one negative. One can show
that the positive root implies a path that has a slope greater than the p = 0 curve—an
unstable path. The negative root is thus the stable path.
with
Sufficient conditions for this system to be stable are
(28)
(29)
BIBLIOGRAPHY
Box, George E. P., and Gwilym M. Jenkins, Time Series Analysis: Forecasting
and Control (San Francisco, 1976).
Daniel, Betty C., "Analysis of Stabilization Policies in Open Economies under
Conditions of Inflation and Inflationary Expectations" (unpublished doctoral
dissertation, University of North Carolina, 1976).
Dornbusch, Rudiger, "Expectations and Exchange Rate Dynamics," Journal of
Political Economy, Vol. 84 (December 1976), pp. 1161-76.
Feige, Edgar L., and Douglas K. Pearce, "Economically Rational Expectations:
Are Innovations in the Rate of Inflation Independent of Innovations in
Measures of Monetary and Fiscal Policy?" Journal of Political Economy,
Vol. 84 (June 1976), pp. 499-522.
Friedman, Milton, "The Case for Flexible Exchange Rates," in his Essays in
Positive Economics (University of Chicago Press, 1953), pp. 157-203.
Isard, Peter, "The Process of Exchange-Rate Determination: A Survey of Impor-
tant Models and Major Issues," Board of Governors of the Federal Reserve
System, International Finance Discussion Paper No. 101 (January 1977).
Johnson, Harry G., "The Case for Flexible Exchange Rates, 1969," in his Further
Essays in Monetary Economics (Harvard University Press, 1973), pp. 198-228.
Knight, Malcolm, "Output, Prices, and the Floating Exchange Rate in Canada:
A Monetary Approach" (unpublished, International Monetary Fund,
December 30, 1976).
Kouri, Pentti J. K., "The Exchange Rate and the Balance of Payments in the
Short Run and in the Long Run: A Monetary Approach" (unpublished,
International Monetary Fund, December 17, 1975).
Machlup, Fritz, The Alignment of Foreign Exchange Rates: The First Horowitz
Lectures (New York, 1972).
Mathieson, Donald J., "Monetary Interdependence Under Flexible Exchange
Rates" (unpublished, International Monetary Fund, March 2, 1976).
McKinnon, Ronald L, review of Fritz Machlup, The Alignment of Foreign Ex-
change Rates: The First Horowitz Lectures, in Journal of International Eco-
nomics, Vol. 5 (February 1975), pp. 99-101.
Shafer, Jeffrey Richard, "The Macroeconomic Behavior of a Large Open Econ-
omy with a Floating Exchange Rate" (unpublished doctoral dissertation,
Yale University, 1976).
Sohmen, Egon, Flexible Exchange Rates: Theory and Controversy (University
of Chicago Press, 1961).