Abstract
THE EVOLUTION of the international monetary system into a regime of flexible exchange rates and the large volatility of these rates during the 1970s have led to a renewed interest in studying the principal determinants of equilibrium exchange rates. The large fluctuations stimulated theories of exchange rate dynamics and have led to the development of various versions of the overshooting hypothesis.1 The explanations of the overshooting phenomenon vary, but they all rely on the short-run fixity of some nominal quantity. Some account for it by the assumption that in the short run commodity prices are slow to adjust relative to asset prices (Dornbusch, 1976); some attribute it to the lack of sufficient speculation in the markets for foreign exchange (McKinnon, 1976); some attribute it to the differential effects of new information on commodity and asset markets (Dornbusch, 1979; Frenkel, 1981 a, 1981 b; Frenkel and Mussa, 1980; Mussa, 1979); and some attribute it to the implications of the process by which asset holders restore portfolio balance in the face of disturbances (Kouri, 1976; Calvo and Rodriguez, 1977; Branson, 1979; Ethier, 1979).2
THE EVOLUTION of the international monetary system into a regime of flexible exchange rates and the large volatility of these rates during the 1970s have led to a renewed interest in studying the principal determinants of equilibrium exchange rates. The large fluctuations stimulated theories of exchange rate dynamics and have led to the development of various versions of the overshooting hypothesis.1 The explanations of the overshooting phenomenon vary, but they all rely on the short-run fixity of some nominal quantity. Some account for it by the assumption that in the short run commodity prices are slow to adjust relative to asset prices (Dornbusch, 1976); some attribute it to the lack of sufficient speculation in the markets for foreign exchange (McKinnon, 1976); some attribute it to the differential effects of new information on commodity and asset markets (Dornbusch, 1979; Frenkel, 1981 a, 1981 b; Frenkel and Mussa, 1980; Mussa, 1979); and some attribute it to the implications of the process by which asset holders restore portfolio balance in the face of disturbances (Kouri, 1976; Calvo and Rodriguez, 1977; Branson, 1979; Ethier, 1979).2
In this paper the implications of two classes of models for the dynamics of exchange rates are analyzed. It is assumed throughout that expectations are formed rationally and that individuals are fully informed. Therefore, the possibility that the dynamics of exchange rates are due to systematic expectational errors is precluded.
In Section I, the dynamics of exchange rates in a model in which commodity prices are slow to adjust is analyzed. The basic structure of the model is due to Dornbusch (1976). The extension here allows for a finite rate of capital mobility, and it is shown that the speed of adjustment in capital markets plays a critical role in determining whether or not following a monetary disturbance the exchange rate overshoots its equilibrium value. It is shown that, when the degree of capital mobility is low, the exchange rate is likely to undershoot its equilibrium value.
In Section II, the dynamics of exchange rates within the framework of the portfolio-balance model with complete flexibility of prices is analyzed. In this context, earlier results of Calvo and Rodriguez (1977) concerning the effect of monetary growth on the real exchange rate are interpreted. It is shown that the fundamental factors determining the dynamics of the real exchange rate are the specification of the portfolios of assets, the degree of capital mobility, and the relative qualities of the various assets as hedges against inflation. The study of the dynamics of the real exchange rate shows that throughout the adjustment process the nominal exchange rate is changing at a rate that differs from the rate of inflation. It follows that a policy that ties the exchange rate to the rate of inflation would be inconsistent with the equilibrium self-fulfilling expectations adjustment path.
I. Exchange Rate Dynamics and the Speed of Adjustment in Commodity and Asset Markets
In this section, the dynamics of exchange rates and the overshooting hypothesis are analyzed from the perspective that emphasizes the speeds of adjustment in commodity and asset markets. This perspective was developed by Dornbusch (1976), who assumed that, as a first approximation, asset markets clear instantaneously while adjustment in commodity markets is sluggish. Dornbusch showed that, when these assumptions are coupled with the assumption that expectations are rational, a monetary expansion induces an immediate depreciation of the currency in excess of its long-run equilibrium value; that is, a monetary expansion results in an overshooting of the exchange rate. To gain further understanding of the relationship between the dynamics of exchange rates and the speeds of adjustment in goods and asset markets, the Dornbusch model is modified and allowance has been made for a finite speed of adjustment in asset markets. It is shown that under these circumstances the key factors determining whether the exchange rate overshoots or undershoots its equilibrium value are the relationship between the speed of adjustment in asset markets, the interest elasticity of the demand for money, and the effects of relative prices on the balance of trade. The speed of adjustment in commodity markets does not seem to be a fundamental factor.
The analysis draws on the analytical framework developed by Dornbusch (1976) and, in order to highlight the key point, the model has been simplified by abstracting from many of the inconsequential details. It is assumed that the economy faces a given price of foreign output and a given world rate of interest, that domestically produced goods differ from foreign goods, and that the supply of output is fixed.
the money market
Let the demand for real balances depend on real income, Y, and on the rate of interest, and let the logarithm of the demand be linear in the logarithm of income, y, and in the rate of interest, i. Equilibrium in the money market attains when
where m and p denote, respectively, the logarithms of the nominal quantity of money and the price level. Thus, the equilibrium rate of interest can be written as
the goods market
The demand for domestic output, D, is composed of domestic demand and foreign demand. This demand can be expressed as the sum of total domestic absorption (domestic demand for domestic and foreign goods) and the excess of exports over imports (the trade balance surplus). Absorption is assumed to depend on real income, while the trade balance is assumed to depend on the relative price of domestic and foreign goods.3 Total demand for domestic output can therefore be written as
where A denotes domestic absorption, T denotes the balance of trade, S denotes the exchange rate (the price of foreign exchange in terms of domestic currency), P* the fixed foreign price level (in terms of foreign currency), and P the price of domestic output. For convenience, units are defined so as to equate the foreign price level to unity; therefore, the trade balance may be viewed as depending on the real exchange rate, s = S/P. Furthermore, since real output is given, absorption is fixed and the demand for domestic output varies with the real exchange rate.
Long-run equilibrium obtains when the demand for domestic output equals the fixed supply, that is, when D = Y. The real exchange rate that is associated with this long-run equilibrium is defined by
Proceeding with the log-linear specification, the trade balance is written as
or equivalently as
where e, p, and k are the logarithms of S, P, and
The percentage change in the price level, ṗ, is assumed to be proportional to excess demand (D – Y):
where the parameter π measures the speed of adjustment in the goods market.4 Substituting equation (6) into equation (7) yields
where α = πδ.
At each moment in time, the price level is given, and its evolution is described by equation (8). The coefficient α is the product of two factors: π—the speed of adjustment in the goods market, and δ—the sensitivity of the balance of trade to the real exchange rate. As will be seen below, of these two factors only the latter plays a role in determining whether or not monetary changes result in exchange rate overshooting.
the capital account and the balance of payments
Equilibrium in the world asset market attains when the difference between the rates of interest on domestic and foreign securities, which are identical in all respects except for the currency of denomination, just equals the expected rate of change in the exchange rate. For example, when the domestic currency is expected to depreciate at the percentage rate x, long-run equilibrium requires that
where i* denotes the rate of interest on assets denominated in foreign currency. It is assumed that expectations concerning the percentage rate of depreciation depend on the relationship between the equilibrium long-run exchange rate,
where ē is the logarithm of
The equilibrium that is described in equation (9) is attained through the mechanism of arbitrage that is effected through the international mobility of capital. It is assumed that capital flows are proportional to (i – i* – x)—the discrepancy between the net rates of return on the various securities. Substituting equation (10) for the expected depreciation of the currency, the international flow of capital can be specified as
where C denotes net capital inflow (the surplus in the capital account) and where β denotes the speed of adjustment in asset markets.5 When capital is perfectly mobile, β = ∞, and the mobility of capital ensures that equation (9) holds all the time. At the other extreme, when β = 0, capital is completely immobile, and the mechanism of arbitrage in asset markets is completely inoperative. It is shown below that the magnitude of β is a key factor determining the dynamics of exchange rates.
Equilibrium in the balance of payments in the absence of central bank intervention is attained when the sum of the trade balance and the capital account is zero. Adding equations (5) and (11) and substituting equation (2) for the (equilibrium) domestic rate of interest yields
as the equilibrium condition for the balance of payments. It should be noted that the equilibrium condition for the balance of payments (which is a flow relationship) is necessary, since the speed of adjustment in asset markets is assumed to be finite. When asset markets clear instantaneously, equation (12) always holds as an identity, and the equilibrium condition is replaced by the interest parity condition as in equation (9) above.6
equilibrium exchange rate, speed of adjustment, and the price level
An analysis of the equilibrium exchange rate and the relationship between the exchange rate, the price level, and the speed of adjustment in the goods and asset markets follows. It is noted, first, that in the long run, given the quantity of money, the exchange rate equals its long-run value7 and i = i*. Substituting i* for i in equation (1)—the condition for money market equilibrium—the long-run price level,
To obtain the relationship between
As may be seen from equations (13) and (14), the system satisfies the homogeneity postulate: a given change in the money supply results in an equiproportional change in the long-run equilibrium price level and the exchange rate. Using equations (13) and (14) in equation (12), the equilibrium in the balance of payments can be written as
Equation (15) expresses the various accounts in the balance of payments as functions of the discrepancies between current and long-run values of the price level and the exchange rate. Since at each moment in time the price level is given, equilibrium in the balance of payments is obtained only when the exchange rate is at a level that satisfies equation (15). By collecting terms in equation (15), the equilibrium exchange rate can be written as
where
Equation (16), which is central to the analysis, relates the equilibrium exchange rate to its long-run value and to the discrepancy between current and long-run prices. This is a reduced form relationship that holds at each moment in time. It is pertinent to note that, depending on the sign of the parameter ε, the relationship between the price level and the exchange rate may be positive or negative. As may be seen, the sign of ε depends on the degree of capital market integration, which has been characterized here in terms of the speed of adjustment β. When the speed of adjustment is low, δ > βb and ε > 0. In that case, given ē and
the effect of monetary expansion: overshooting and undershooting
In the previous section the relationship between the exchange rate, the price level, and the speed of asset market adjustment was characterized. This relationship can be illustrated with the aid of Figures 1 and 2, which will then be used to analyze the effects of monetary changes. In these figures, the ṗ = 0 schedule shows a combination of exchange rates and price levels for which there is no excess demand for domestic output. The schedule plots equation (8), and its slope is unity.9 The intercept of the schedule corresponds to the (logarithm of the) equilibrium long-run real exchange rate, k. Also, along this schedule trade must be balanced, so that T = 0. All points to the right of the ṗ = 0 locus correspond to an excess demand for goods and to a trade balance surplus. Consider next the schedule along which the capital account is balanced. From equation (15), the capital account can be written as




Thus, the slope of the schedule along which the capital account is balanced (C = 0) is –θ/b < 0, as is shown in Figures 1 and 2. Points below the C = 0 schedule correspond to a deficit in the capital account.
The equilibrium relationship between the price level and the exchange rate (which must hold at each moment in time) is summarized by equation (16), plotted as the QQ schedule in Figures 1 and 2. The slope of the schedule is 1/ε, which may be positive or negative depending on the sign of ε, which in turn depends on whether β is smaller or larger than δ/b. When the degree of capital mobility is low, β < δ/b, and the slope of the schedule is positive and larger than unity since ε < 1. When asset markets adjust relatively fast, the QQ schedule is negatively sloped. Since along the equilibrium path the balance of payments is balanced, the QQ schedule must pass in a region that is characterized by a surplus in the balance of trade and a deficit in the capital account (T > 0, C < 0) or by a deficit in the balance of trade and a surplus in the capital account (T < 0, C > 0). Thus, when the QQ schedule is negatively sloped as in Figure 1, it must be steeper than the C = 0 locus.10
Consider point B in Figures 1 and 2. At this point, there is an excess demand for goods, and ṗ > 0. Also at point B there is a trade balance surplus (T > 0) and a capital account deficit (C < 0). The path of adjustment is described by the arrows along the equilibrium schedule QQ, and long-run equilibrium is reached at point D, where prices and exchange rates reach their equilibrium values
To analyze the effects of a monetary expansion, consider an initial long-run equilibrium at point A with
The impact effect of the monetary change can be analyzed in terms of equation (16). By the homogeneity postulate, dm = dē = d
and substituting equation (16) for ε, the elasticity can be written as
When the speed of adjustment is relatively high, so that β > δ/b, ε < 0 and the short-run elasticity exceeds unity as in Figure 1. This is the overshooting phenomenon that corresponds to the Dornbusch case. On the other hand, when the speed of adjustment is relatively low, so that β < δ/b, ε is positive but smaller than unity, the short-run elasticity is smaller than unity, and the exchange rate undershoots its new long-run equilibrium value. In the border case for which β = δ/b, the QQ schedule is vertical and the exchange rate reaches immediately its long-run value, as e1 coincides with ē1. It may also be noted from equation (18) that the extent of overshooting or undershooting depends on the magnitude of θ—the speed of adjustment of expectations, the determinants of which are analyzed below. As is clear from equation (18), other things being equal, the short-run elasticity gets closer to unity as the value of θ increases, thereby reducing the extent to which the current exchange rate differs from its long-run value. It is also evident that as long as θ is not negative its magnitude is irrelevant for determining whether the short-run elasticity is larger or smaller than unity, and, therefore, the analysis is consistent with a variety of assumptions concerning the formation of expectations, ranging from the assumption of static expectations (for which θ = 0) to the assumption of perfect foresight, which is analyzed below.
As is evident, the key factor determining whether the exchange rate overshoots or undershoots its long-run equilibrium value is the relationship between β and δ/b. These parameters characterize the speed of adjustment in asset markets, the sensitivity of aggregate demand (the balance of trade) to relative prices, and the interest (semi) elasticity of the demand for money. As long as the speeds of adjustment are finite, the speed of commodity price adjustment, π, does not determine whether or not there is overshooting.
Turning to the effect of the monetary expansion on the international accounts, it is noted that, at point B, independent of whether the exchange rate overshoots or undershoots, the rise in e improves the balance of trade and the rise in m deteriorates the capital account by lowering the rate of interest. The deterioration in the capital account due to the interest rate effect is mitigated when the exchange rate overshoots its long-run equilibrium value, since, in that case, expectations are for an appreciation of the currency (a decline in e). In the case of undershooting, the expectations for a further depreciation reinforce the interest rate effect in deteriorating the capital account. The transition toward the long run (the path between points B and D) is characterized by a rising price level and by a decline in the real exchange rate. The decline in (e – p) results in a deterioration of the balance of trade and, therefore, equilibrium in the balance of payments implies that during the transition the capital account improves.
The above analysis implies that the qualitative characteristics of the dynamics of the price level, the trade balance, the capital account, and the real exchange rate are independent of whether there is overshooting or undershooting of the exchange rate. These given qualitative paths may be associated with a path along which the nominal exchange rate is rising, as well as with a path along which the nominal exchange rate is falling. The ambiguous relationship between the path of the nominal exchange rate and the other paths should not be taken to imply that the exchange rate does not exert a definite effect on the trade balance, the capital account, and the path of prices. Rather, it implies that one should not expect to observe, independent of the speed of adjustment, a unique qualitative relationship between the equilibrium paths of the exchange rate and those of prices and the international accounts. This lack of a unique general relationship between the exchange rate and the various balance of payments accounts may be responsible for the view that in recent years exchange rates have shown erratic and unpredictable movements.11
In this section the effects of a once-and-for-all unanticipated change in the money supply have been analyzed. The analysis can be easily extended to examine the effects of other parametric changes such as changes in output, the foreign price level, and the foreign rate of interest. Similarly, the analysis can be extended along the lines suggested by Wilson (1979) and Gray and Turnovsky (1979) to examine the effects of an anticipated future change in the supply of money or in another parameter. For example, it can be shown that an anticipated future rise in the money supply induces an immediate adjustment of the exchange rate, which jumps, for example, to point E in Figures 1 and 2. The extent of the instantaneous jump in the exchange rate is smaller than the change that would have taken place had the money supply been expected to rise at the present (in which case the exchange rate would have adjusted to point B). Following the initial jump in e, both the exchange rate and prices proceed to rise gradually, and their path converges to the new QQ schedule (corresponding to the new quantity of money) at the point in time at which the rise in the money supply actually occurs. Thereafter, the convergence proceeds along the (new) QQ schedule toward the new long-run equilibrium.12
perfect foresight and the coefficient of expectations
So far it has been assumed (in equation (10)) that the expected percentage change in the exchange rate is proportional to the discrepancy (ē—e), with θ > 0 being the proportionality factor. Dornbusch (1976) showed that in a model of perfect foresight the coefficient θ cannot be chosen arbitrarily but, rather, that it must be consistent with the structure of the entire model. An analysis of the determinants of θ follows.
Using equation (14) in equation (8), the rate of inflation may be expressed as
and using equation (16), the rate of inflation can be written as
The relationship between the equilibrium exchange rate and the price level that is described by equation (16) must always be satisfied. Therefore, given the long-run values of ē and
Substituting equation (20) for ṗ yields
Equation (22) describes the actual change in the exchange rate, while equation (10) describes the expected change. It is clear that, under perfect foresight, consistency requires that the two be equal and, therefore,
Substituting for ε from equation (16) results in
from which it follows that the coefficient of expectations can be obtained by solving the quadratic equation
From equation (25), the solution for 0 (obtained by taking the positive root) is
which expresses the coefficient of expectations as a function of the various parameters of the model.
As may be verified, the coefficient of expectations decreases with 1/b, the interest (semi) elasticity of the demand for money, while it increases with π and β, the speeds of adjustment in the goods and asset markets, respectively. These propositions are independent of the degree of capital mobility. In contrast, however, it is noteworthy that the dependence of the coefficient of expectations on the parameter δ (the sensitivity of aggregate demand to relative prices) is ambiguous and depends on whether πβ—the product of the speeds of adjustment in the goods and asset markets—is smaller or larger than a critical value that is equal to 1/(1 + b/θ). For a given value of π, a high value of β, such that πβ exceeds the critical value, yields a positive relationship between the coefficient of expectations and the parameter δ. On the other hand, when β is small, such that πβ is smaller than the critical value, a high value of δ reduces the size of the coefficient of expectations. Since the relationship between the speed of adjustment of expectations and δ is ambiguous, it follows that the effect of a higher value of δ on the extent of the overshooting or undershooting is also ambiguous. Finally, it might be noted that in the extreme case for which asset markets clear instantaneously, so that β = ∞, equation (24) becomes
which corresponds to equation (14) in Dornbusch (1976, p. 1167) and, as noted above, in that case δ and θ are positively related.
The dynamics of exchange rates within a rational expectations model in which commodity prices adjust slowly has been analyzed in this section. The analysis was based on a modified version of a model due to Dornbusch (1976). The model was modified to allow for a finite speed of adjustment in asset markets, and it was shown that the short-run effects of a monetary expansion depend on the degree of capital mobility. When capital is highly mobile the exchange rate must overshoot its long-run value, 13 but when capital is relatively immobile the exchange rate undershoots its long-run value. The key reason for the nonneutrality of money in the short run in this model arises from the assumption that prices are not fully flexible. The dynamics of exchange rates in a portfolio-balance model in which all prices are assumed to be fully flexible also in the short run is analyzed in the next section.
II. Exchange Rate Dynamics and Portfolio Balance
The development of a variety of portfolio-balance models that emphasize the effects of asset substitution on exchange rate determination are the subject of numerous recent articles (e.g., Kouri (1976), Branson (1979), Dornbusch (1979), Girton and Roper (1981), Frenkel and Clements (1981), and the references therein). The purpose here is to highlight the implications of alternative assumptions concerning the degree of substitution among assets on the dynamics of exchange rates. It will be shown that the effects of a monetary expansion on the dynamics of exchange rates and in particular on whether exchange rates overshoot or undershoot their equilibrium path depend critically on the specification of asset choice. To contrast the analysis with the one in the previous section, it will be assumed that goods markets clear instantaneously and that all prices are perfectly flexible.
the currency substitution model
The Calvo-Rodriguez (1977) model of currency substitution analyzes a fully employed small, open economy in which residents are assumed to hold portfolios of domestic and foreign currencies. The key building blocks of the model are the specifications of the markets for assets and goods.
The Asset Market
Asset holders are assumed to hold portfolios of domestic money, M, and foreign money, F. The value of assets in terms of foreign currency is denoted by a:
where M' denotes the value of domestic currency holdings in terms of foreign exchange, that is, M' ≡ M/S.
The desired ratio of domestic to foreign money holdings is assumed to depend on the expected percentage change in the exchange rate, which measures the expected difference between the rates of return on the two assets. The assumption of rational expectations (which in this model amounts to perfect foresight) permits us to identify the expected change in the exchange rate with the actual change. Using a circumflex (^) to denote the percentage change in a variable, portfolio equilibrium can be written as
which indicates that the desired ratio of domestic money to foreign money declines when the domestic currency is expected to depreciate. It will prove useful to express the portfolio-balance relationship in terms of its inverse:
The Goods Market
The economy is assumed to produce two classes of goods, tradable and nontradable. For a given state of technology and factor endowment, the rates of production of the two composite commodities depend on their relative price. Denoting the domestic currency price of traded goods by PT and of nontraded goods by PN, the relative price that is relevant for production decisions is PT/PN. The domestic price of traded goods is linked to the foreign price of that good,
where QT and QN denote, respectively, the output of traded and nontraded goods.
The demand for the two goods is assumed to depend on their relative price and on the value of assets according to
where CT and CN denote, respectively, the demand for traded and nontraded goods.
At each point in time the stock of domestic holdings of foreign assets, F, is given, and the assumption that the small country’s currency is not held by foreigners ensures that F cannot be adjusted instantaneously. Asset holders can, however, alter the stock of foreign assets gradually by running a surplus or a deficit in the balance of trade. Thus,
where Ḟ ≡ dF/dt denotes the rate of change in F. Equilibrium in the market for nontraded goods requires that the rate of domestic production equal the domestic demand, so that
Equation (34) implies that there is a specific relationship between the real exchange rate and the value of assets that is consistent with equilibrium in the market for nontraded goods. A rise in the value of assets must be accompanied by a decline in the real exchange rate, since the former creates an excess demand for nontraded goods while the latter induces an excess supply. This relationship is summarized by
Since equation (35) must hold at each moment in time because of the assumed flexibility of prices, it can be substituted into equation (33) to yield a relationship between the rate of change in F and the value of assets:
The value of assets is uniquely related to the equilibrium real exchange rate and thereby to the rates of production and consumption of goods and to the rate of accumulation of foreign currency. Therefore, knowledge of the time path of assets is necessary for determining the time path of these variables.
Dynamics
Changes in asset holdings arise from changes in the domestic and the foreign asset components of the portfolio:
Recalling that M' ≡ M/S, it follows that Ṁ ' = M'(μ - Ŝ), where μ denotes the percentage change in the nominal money supply, that is, μ ≡ Ṁ/M. Since M' = a – F, the change in M' can be expressed as Ṁ' = (a – F) (μ – Ŝ). Thus, equation (37) can be written as
where, using equations (30) and (36),
Equations (36) and (38) characterize the dynamics of the system. As is clear, in the steady state, when Ḟ = ȧ = 0, f(a) = 0 (equation (36)) and μ = Ŝ (equation (38)). The steady-state values of a and F are denoted by ā and
Figure 3 describes the dynamics of the system. In the panel on the right, the Ḟ = 0 and the ȧ = 0 schedules describe combinations of a and F that satisfy equations (36) and (38), respectively. The slope of the ȧ = 0 schedule is drawn on the assumption that (around the steady state) ∂ȧ/∂a < 0. As is evident, the system exhibits a saddle-path stability, and the motion of the variables is described by the arrows that are implied by the signs of the partial derivatives of equations (36) and (38) around the steady state. None of the qualitative conclusions is altered in the case for which ∂ȧ/∂a > 0. In that case, the ȧ = 0 schedule is positively sloped and is steeper than the saddle path. In general, along the perfect-foresight path (which is the unique path that converges to the steady state and satisfies the laws of motion and the initial conditions) a higher value of F is associated with a higher value of a. The panel on the left in Figure 3 presents the combinations of the value of assets and the real exchange rate that satisfy equation (35). These are the combinations that are consistent with equilibrium in the nontraded goods market for which a rise in the value of assets must be associated with a decline in s.

Exchange Rate Dynamics in the Currency Substitution Model
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001

Exchange Rate Dynamics in the Currency Substitution Model
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001
Exchange Rate Dynamics in the Currency Substitution Model
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001
Consider a rise in the rate of monetary expansion. From equation (38) it is seen that the higher value of μ shifts the ȧ = 0 schedule, and hence shifts the saddle path to the right, and results in new, higher steady-state holdings of foreign assets. In the new steady state, the real value of assets remains unchanged so as to ensure that Ḟ = 0 (equation (36)). Since the equilibrium real exchange rate is uniquely related to the value of assets, also the steady-state real exchange rate remains unchanged and, thus, production and consumption of both goods remain unchanged. Finally, since in the new steady state the new rate of monetary expansion equals the rate of depreciation, Ŝ, individuals will lower the desired ratio of domestic to foreign currency holdings, which, given the unchanged value of assets, implies that F rises and M' falls.
The dynamics of adjustment are shown in Figure 3, where it is assumed that the initial steady-state position was at point A, with
The transition toward the new steady state is characterized by the path between B and C, along which s declines (so that
The key feature of the Calvo-Rodriguez model is the specification of the portfolio of assets, where it is assumed that the two alternative assets are domestic and foreign currencies. Since the foreign currency price of traded goods is assumed to be given, the accumulation of foreign currency is equivalent to an accumulation of claims on stocks of traded goods. With this perspective, the currency substitution model can be specified in terms of choice and substitution between domestic money and traded goods. It is intuitively clear, therefore, that following the rise in the rate of monetary expansion asset holders wish to hedge against the expected inflation by shifting the composition of portfolios toward the inflation hedge (traded goods), a shift which results, in the short run, in a rise in s—the relative price of traded goods.
This interpretation of the currency substitution model suggests that, when the inflation hedges are stocks of nontraded goods instead of traded goods, the rise in the rate of monetary expansion might result in an undershooting of the exchange rate as the relative price of nontraded goods rises. A formal analysis of this possibility follows.
nontraded goods as inflation hedge
The structure of the model in which nontraded goods are used as the inflation hedge is very similar to the one described in the currency substitution model. In what follows the minimal modifications are introduced.
The Asset Market
To sharpen the contrast with the currency substitution model, it is assumed that the portfolios of assets are composed of domestic money and stocks of nontraded goods, N. The value of assets in terms of foreign exchange is
where N/s = PNN/S. As before, the desired ratio of money to inventories of goods is assumed to depend on the difference between their rates of return, that is, on the expected percentage change in the price of nontraded goods. Assuming that expectations are always realized, the desired portfolio composition depends on
The Goods Market
The specification of production and consumption is assumed to be the same as in the previous section and is summarized by equations (31) and (32). In the present case, however, an allowance is made for an accumulation of inventories of nontraded goods that must equal the excess of production over consumption of these goods.
where N denotes the rate of accumulation of nontraded goods.
Since in this case no allowance is made for capital mobility, equilibrium in the market for traded goods requires that
equation (42) implies that there is a specific relationship between the real exchange rate and the value of assets that guarantees trade balance equilibrium. A rise in the value of assets raises the demand for traded goods and, therefore, to eliminate the excess demand, it must be accompanied by a rise in the real exchange rate, s. This relationship, which must be satisfied at all times, can be written as
For later use, it is noted that
where γ denotes the elasticity of the real exchange rate with respect to the value of assets. Substituting equation (43) into equation (41) yields a relationship between the rate of accumulation of nontraded goods and the value of assets:
Dynamics
The rate of change in the value of assets can be written, using equation (39), as
Noting that
where (ν(a)a/N) – 1 was substituted for M'/(N/s) by using equations (39) and (43) and where γâ was substituted for ŝ by using equation (44).
Using equations (45) and (47) in equation (46) yields
Equations (45) and (48) characterize the dynamics of the system. In the steady state, Ṅ = ȧ = 0, the value of assets is ā0, the holdings of nontraded goods are
The dynamics of the system are described in Figure 4. In the panel on the right, the Ṅ = 0 and the ȧ = 0 schedules show combinations of a and N that satisfy equations (45) and (48), respectively. The slope of the ȧ = 0 schedule is drawn on the assumption that around the steady state ∂ȧ/∂a < 0. As in the currency substitution model, the system exhibits a saddle-path stability and the motion of the variables is described by the arrows, which reflect the signs of the partial derivatives of equations (45) and (48) around the steady state. When ∂ȧ/∂a > 0, the ȧ = 0 schedule is positively sloped and is steeper than the saddle path. As before, none of the qualitative conclusions are altered in that case. The key difference between the analysis in Figure 4 and that in Figure 3 lies in the panel on the left, which shows the equilibrium relationship between the real exchange rate and the value of assets. In contrast with the currency substitution model, here s and a must be positively related so as to ensure trade balance equilibrium (equation (42)).

Exchange Rate Dynamics When Portfolios Contain Money and Nontraded Goods.
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001

Exchange Rate Dynamics When Portfolios Contain Money and Nontraded Goods.
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001
Exchange Rate Dynamics When Portfolios Contain Money and Nontraded Goods.
Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A001
A rise in the rate of monetary expansion results (from equation (48)) in a rightward shift of the ȧ = 0 schedule and hence of the saddle path. The new steady state is associated with larger holdings of nontraded goods, whose attractiveness has risen with the accelerated inflation. Suppose that the initial equilibrium is described in Figure 4 by point A and that the schedules drawn correspond to the new higher rate of monetary expansion. Since at the initial point in time the stocks of nontraded goods are given at
The higher expected rate of inflation lowers the desired ratio of money to inventories of nontraded goods,
The transition toward the steady state is described by the path from B to C, along which the value of assets and the holdings of nontraded goods increase and the real exchange rate rises toward its initial level. Since along the path s rises, the nominal exchange rate must rise faster than the price of nontraded goods.
The fact that the rise in the rate of monetary expansion results in an instantaneous rise in the price level that exceeds the rise in the depreciation of the currency is consistent with the general principles that were outlined at the end of the discussion of the currency substitution model. In this class of the portfolio-balance model, the key factor determining whether following a monetary disturbance the exchange rate overshoots or undershoots the domestic price level is the specification of the inflation hedge. When the domestic rate of inflation accelerates, asset holders substitute away from domestic money into alternative assets. To the extent that the alternative assets are traded goods, their relative price will rise and the nominal exchange rate will overshoot the domestic price level. To the extent that the alternative assets are domestic nontraded goods, their relative price will rise and the exchange rate will undershoot the domestic price.
In this section the two polar cases in which the substitutes for domestic money were either foreign currency (or equivalently traded goods) or stocks of domestic nontraded goods were analyzed. A more general model would allow for portfolios that consist of domestic and foreign monies as well as of domestic goods. The general principles, however, are likely to be the same. Whether exchange rates overshoot or undershoot the domestic price level will depend on the relative degree of substitution among domestic money, traded goods, and nontraded goods. The relative degrees of substitution will reflect the degree of capital mobility and the qualities of the various assets as hedges against inflation.
III. Concluding Remarks
This paper has analyzed the determinants of the evolution of exchange rates within the context of alternative models of exchange rate dynamics. It has examined the overshooting hypothesis in models that emphasize differential speeds of adjustment in asset and goods markets as well as in models that emphasize portfolio-balance considerations. It has been seen that exchange rate overshooting is not an intrinsic characteristic of the foreign exchange market and that it depends on a set of specific assumptions. It has been shown that the overshooting is not a characteristic of the assumption of perfect foresight, nor does it depend in general on the assumption that goods and asset markets clear at different speeds. As long as the speeds of adjustment in the various markets are less than infinite (which is obviously the case), the key factor determining the short-run effects of a monetary expansion is the degree of capital mobility. When capital is highly mobile, the exchange rate overshoots its long-run value, and when capital is relatively immobile, the exchange rate undershoots its long-run value.
Within the context of the portfolio-balance model, it has been shown that the effects of a monetary expansion on the dynamics of exchange rates and in particular on whether exchange rates overshoot or undershoot their equilibrium path depend critically on the specification of asset choice, on the degree of substitution among assets, and on the quality of the various assets as a hedge against inflation.
REFERENCES
Bilson, John F. O., “The ‘Vicious Circle’ Hypothesis,” Staff Papers, Vol. 26 (March 1979), pp. 1–37.
Boyer, Russel S., and Robert J. Hodrick, “Perfect Foresight, Financial Policies, and Exchange Rate Dynamics.” (This is scheduled for publication in the Canadian Journal of Economics, May 1982.)
Branson, William H., “Exchange Rate Dynamics and Monetary Policy,” in Inflation and Employment in Open Economies, ed. by Assar Lindbeck (Amsterdam, 1979).
Calvo, Guillermo A., and Carlos Alfredo Rodriguez, “A Model of Exchange Rate Determination under Currency Substitution and Rational Expectations,” Journal of Political Economy, Vol. 85 (June 1977), pp. 617–25.
Dornbusch, Rudiger (1976), “Expectations and Exchange Rate Dynamics,” Journal of Political Economy, Vol. 84 (December 1976), pp. 1161–76.
Dornbusch, Rudiger (1979), “Monetary Policy Under Exchange Rate Flexibility,” in Managed Exchange-Rate Flexibility: The Recent Experience, Proceedings of a Conference held at Melvin Village, New Hampshire, October 1978, Federal Reserve Bank of Boston, Conference Series, No. 20 (1979).
Dornbusch, Rudiger, and Stanley Fischer, “Exchange Rates and the Current Account,” American Economic Review, Vol. 70 (December 1980), pp. 960–71.
Ethier, Wilfred, “Expectations and the Asset-Market Approach to the Exchange Rate,” Journal of Monetary Economics, Vol. 5 (April 1979), pp. 259–82.
Flood, Robert P., “An Example of Exchange Rate Overshooting,” Southern Economic Journal, Vol. 46 (July 1979), pp. 168–78.
Frenkel, Jacob A. (1981 a), “The Collapse of Purchasing Power Parities During the 1970s,” European Economic Review, Vol. 16 (May 1981), pp. 145–65.
Frenkel, Jacob A. (1981 b), “Flexible Exchange Rates, Prices, and the Role of ‘News’: Lessons from the 1970s,” Journal of Political Economy, Vol. 89 (August 1981), pp. 665–705.
Frenkel, Jacob A., and Kenneth W. Clements, “Exchange Rates in the 1920s: A Monetary Approach,” in Development in an Inflationary World, ed. by M. June Flanders and Assaf Razin (New York, 1981), pp. 283–318.
Frenkel, Jacob A., and Michael L. Mussa, “The Efficiency of Foreign Exchange Markets and Measures of Turbulence,” American Economic Review, Vol. 70 (May 1980), pp. 374–81.
Girton, Lance, and Don Roper, “Theory and Implications of Currency Substitution,” Journal of Money, Credit and Banking, Vol. 13 (February 1981), pp. 12–30.
Gray, Malcolm R., and Stephen J. Turnovsky, “The Stability of Exchange Rate Dynamics Under Perfect Myopic Foresight,” International Economic Review, Vol. 20 (October 1979), pp. 643–60.
Henderson, Dale W., “The Dynamic Effects of Exchange Market Intervention Policy: Two Extreme Views and a Synthesis,” in The Economics of Flexible Exchange Rates, Proceedings of a Conference at the Institute for Advanced Studies, Vienna, March 29–31, 1978, ed. by Helmut Frisch and Gerhard Schwödiauer; supplement to Kredit und Kapital, Heft 6 (Berlin, 1980), pp. 156–209.
Kimbrough, Kent P., “Real Aspects of the Monetary Approach to the Exchange Rate” (unpublished doctoral dissertation, University of Chicago, 1980).
Kouri, Pentti J. K., “The Exchange Rate and the Balance of Payments in the Short Run and in the Long Run: A Monetary Approach,” Scandinavian Journal of Economics, Vol. 78 (May 1976), pp. 280–304.
Levich, Richard M., Overshooting in the Foreign Exchange Market, Group of Thirty, Occasional Papers, No. 5 (New York, 1981).
Liviatan, Nissan, “Monetary Expansion and Real Exchange Rate Dynamics,” Journal of Political Economy, Vol. 89 (December 1981), pp. 1218–27.
McKinnon, Ronald I., “Floating Foreign Exchange Rates, 1973–74: The Emperor’s New Clothes,” in Institutional Arrangements and the Inflation Problem, ed. by Karl Brunner and Allan H. Meltzer, Carnegie-Rochester Conference Series on Public Policy, Vol. 3 (New York, 1976), pp. 79–114; supplementary series to the Journal of Monetary Economics.
Mussa, Michael L. (1976), A Study in Macroeconomics (New York, 1976).
Mussa, Michael L. (1979), “Empirical Regularities in the Behavior of Exchange Rates and Theories of the Foreign Exchange Market,” in Policies for Employment, Prices, and Exchange Rates, ed. by Karl Brunner and Allan H. Meltzer, Carnegie-Rochester Conference Series on Public Policy, Vol. 11 (New York, 1979), pp. 9–57; supplementary series to the Journal of Monetary Economics.
Mussa, Michael L. (1980), “The Role of the Current Account in Exchange Rate Dynamics” (unpublished, University of Chicago, 1980).
Mussa, Michael L. (1981), “Sticky Prices and Disequilibrium Adjustment in a Rational Expectations Model of the Inflationary Process,” American Economic Review, Vol. 71 (December 1981), pp. 1020–27.
Mussa, Michael L. (1982), “A Model of Exchange Rate Dynamics,” Journal of Political Economy, Vol. 90 (February 1982), pp. 74–104.
Niehans, Jürg, “Exchange Rate Dynamics with Stock/Flow Interaction,” Journal of Political Economy, Vol. 85 (December 1977), pp. 1245–57.
Rodriguez, Carlos Alfredo, “The Role of Trade Flows in Exchange Rate Determination: A Rational Expectations Approach,” Journal of Political Economy, Vol. 88 (December 1980), pp. 1148–58.
Shafer, Jeffrey R., “Flexible Exchange Rates, Capital Flows and Current Account Adjustment” (unpublished, Board of Governors of the Federal Reserve System, Washington, 1980).
Wilson, Charles A., “Anticipated Shocks and Exchange Rate Dynamics,” Journal of Political Economy, Vol. 87 (June 1979), pp. 639–47.
Mr. Frenkel is a Professor of Economics at the University of Chicago, an editor of the Journal of Political Economy, and a Research Associate of the National Bureau of Economic Research. He is a graduate of the Hebrew University of Jerusalem and of the University of Chicago.
Mr. Rodriguez is a Professor of Economics at the Center for Macroeconomic Studies, Buenos Aires, and was a member of the faculty of Columbia University. He is a graduate of the University of Chicago.
This paper is a revised version of an earlier paper written in the fall of 1977 and entitled “The Anatomy of the Overshooting Hypothesis in the Market for Foreign Exchange.” The authors are indebted to Joshua Aizenman, Kent Kim-brough, and Michael Mussa, and to participants of the joint international economics workshop of Tel Aviv University and the Hebrew University of Jerusalem for useful comments. Mr. Frenkel acknowledges research support from the National Science Foundation, grant SOC 78-14480. This research is part of the National Bureau of Economic Research’s Program in International Studies.
Work on this paper was done while the authors were consultants with the Fund. The views expressed are not necessarily those of the sponsoring organizations.
For a classification of the various versions of the overshooting hypothesis, see Levich (1981).
In addition to these studies, a partial list of recent studies includes Bilson (1979), Boyer and Hodrick (1982), Flood (1979), Henderson (1980), Kimbrough (1980), Liviatan (1981), Mussa (1982), and Shafer (1980). Further references are included in the following sections of this paper.
This specification abstracts from the effects of the rate of interest on absorption, the effects of income on the trade balance, and the distinction between gross national product and gross domestic product that results from interest income on net ownership of foreign securities. These abstractions are made for simplicity, since they do not affect the nature of the argument. Allowance is made for some of these factors by Dornbusch (1976).
The assumption that prices change at a finite speed is taken as a stylized fact and there is no attempt to rationalize it here. For an interesting theory of sticky prices, see Mussa (1981). The specification of the rate of inflation as a function of excess demand is analogous to the specification in Dornbusch (1976), equation (8).
The theoretical deficiencies of the capital flow equation (equation (11)) are well known. Since the purpose here is to highlight the role of alternative assumptions concerning the speed of adjustment in asset markets, it has been chosen to specify the simplest formulation and to abstract from many other issues.
For a discussion of the role of flow equilibrium when asset markets adjust slowly, see Kouri (1976) and Niehans (1977). For a discussion of the conceptual issues involved in the formulation of flow equilibrium in asset markets, see Mussa (1976).
In the present framework, the long-run value of the exchange rate is fixed. In general, the equilibrium value may be specified in terms of a movement along an equilibrium path; see Mussa (1982).
The case for which β = ∞ is the Dornbusch case. equation (16) here coincides with equation (6) in Dornbusch (1976, p. 1164).
The ṗ = 0 schedule would be flatter than a 45° line if the demand for goods were to depend negatively on the rate of interest as in Dornbusch (1976). In that case, a rise in the price level lowers the demand via the higher relative price and via the rate of interest that must be higher in order to restore money market equilibrium.
The two schedules tend to coincide in the limit, when the magnitude of β tends to infinity.
The relationship between the exchange rate and the trade balance is analyzed in detail in many recent contributions. See, for example, Kouri (1976), Dornbusch and Fischer (1980), Mussa (1980), and Rodriguez (1980).
It follows that when capital is highly mobile, so that the QQ schedule is negatively sloped, the path of prices will be monotonic while the path of the exchange rate will exhibit a turning point, that is, e will initially rise and then decline; when capital is less mobile, so that the QQ schedule is positively sloped, both prices and the exchange rate will approach their new, higher values monotonically.
This conclusion is based on the assumption that output is fixed. Dornbusch (1976) shows that, when the monetary expansion induces a short-run rise in output, the exchange rate may undershoot its equilibrium level even if capital is highly mobile. This will occur if the rise in output raises the demand for money sufficiently so as to result in a rise in the rate of interest.
This characteristic may not be necessary in a model with intertemporal utility maximization; see Liviatan (1981).