Macroeconomic Stability Effect of Intermediate Exchange Rate Regimes: Target Zone vs. Managed Float *1

Transcription

1 Asia-Pacific Journal of EU Stuies Vo. 8 No. / 9 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies: Target Zone vs. Manage Float * YEONGSEOP RHEE * Seoul National University, Korea This paper copares alternative intereiate exchange rate regies instea of coparing an extree regie vs. an intereiate one, an investigates whether the stabilizing effect on the exchange rate of a regie can be extene to other econoic variables. Central results fro the coparison of two intervention rules between a target zone an a anage float are: if the sensitivity of aggregate ean to the real exchange rate is very large or the onetary isturbance is preoinant copare to the real isturbance, then the stabilizing effect on the exchange rate of a tighter intervention in an intereiate regie than the other is positively associate with the stabilizing effect on the price; if the sensitivity of aggregate ean to the real exchange rate is very sall or the real isturbance is preoinant, the stabilizing effect on the exchange rate of a tighter intervention can be achieve only at the expense of price stability. Keywors: Foreign Exchange Market Intervention, Target Zone, Manage Float, Exchange Rate Stability, Price Stability I. INTRODUCTION Along with the progress of the European onetary integration an also after the Asian crisis in 997, there have been long an recurrent iscussions on regional onetary cooperation an integration in East Asia. Many ieas have been floating aroun fro an extree free float to an extree har peg syste. Aong various alternatives to the current exchange rate regies in East Asian countries, soe intereiate regies have attracte special attention than others: in particular, a target zone an a anage float. A coon proble of the literature on the choice of an optial exchange rate * This work was supporte by 00 Research Settleent Fun for the new faculty of SNU. ** Professor, Grauate School of International Stuies, Seoul National University, Korea; E-ail: ysrhee@snu.ac.kr For exaple, see Moon et al. (000), Rhee (004).

2 0 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies regie is that they assue the sae rule of central bank intervention an copare an intereiate regie with an extree one, a free float or a fixe rate regie. In fact, this kin of research is not to fin an optial regie but an optial egree of intervention within a given exchange rate regie. Consequently a usual conclusion is that the optial egree of exchange rate flexibility epens on the structural characteristics of the country an on the preferences of onetary authorities (Benassy-Quere, 00). If we follow this approach, the choice of an optial exchange rate regie shoul be very liite. When Frankel (999) sai that no single currency regie is right for all countries or at all ties, it oes not ean a iversity only of intervention egrees but that of intervention rules. To avoi the proble, this paper copares alternative intereiate exchange rate regies. In other wors, instea of coparing an intereiate regie with an extree one, we copare intereiate regies with each other. Through a ifferent iension of coparison, we want to investigate whether an intereiate regie leas to the stability of exchange rates an whether the stabilizing effect on the exchange rate of a regie can be extene to other econoic variables. In the paper, an exchange rate regie is classifie accoring to the rule of foreign exchange arket intervention. A free float is efine as the case with no intervention at all in the arket, while a fixe rate regie as that with perfect intervention so that the exchange rate oes not ove at all. In between, there are any kins of intervention rules an within each rule continuous egrees of intervention. Aong so any kins of intervention rules for intereiate regies, we have a particular interest in those for a target zone an a anage float. Accoring to the IMF s classification of exchange rate arrangeents, a target zone is ubbe as a pegge exchange rate within horizontal bans. Uner this regie, the value of the currency is aintaine within certain argins of fluctuation of aroun a fixe central rate (Harvereier et al., 009). A notable exaple was the Exchange Rate Mechanis (ERM) of the European Monetary Syste (EMS). The intervention rule of a target zone is that within the ban, there is no intervention, but if the exchange rate approaches a argin, the central bank intervenes in the arket by increasing or ecreasing oney supply so that noinal exchange rate will not be beyon a liit. In a anage float with no preeterine path for the exchange rate, the onetary authority attepts to influence the exchange rate without having a specific exchange rate path or target. Inicators for anaging the rate are broaly jugental (e.g., balance of payents position, international reserves, parallel arket evelopents), an ajustents ay not be autoatic. Intervention ay be irect or inirect (IMF, 005). But in this paper, the intervention rule of a anage float is siplifie so that oney supply is ajuste to the changes in the exchange rate accoring to the policy of leaning against the win. To exaine behaviors of exchange rates an other econoic variables uner alternative exchange rate regies, in particular uner a target zone, this paper will See Harvereier et al. (009) an IMF (005). A note is that while the IMF classifies a anage float as a ore flexible one than a target zone, this paper oes not consier their egrees of flexibility but inclues both in the intereiate regie with ifferent intervention rules.

3 YEONGSEOP RHEE follow the approach aopte in the literature on target zones by Krugan (989, 99), Froot an Obstfel (989a, 989b), Miller an Weller (989), an Floo an Garber (989) aong others. Those stuies share the technique of regulate Brownian otion, an are universally base on the onetary oel of the exchange rate. A coon liitation of these stuies is that they ene up with a single-equation oel of the exchange rate an other econoic variables such as the price or output are assue to be preeterine. It is not persuasive that only the exchange rate is affecte by the intervention policy an all isturbances are anage by that single policy. To avoi the proble, this paper uses a ultiple-equation oel of the exchange rate an the price instea of a single-equation oel of the exchange rate, which inclues both expecte epreciation an expecte inflation. This ultipleequation oel allows us to exaine behaviors of other econoic variables besies the exchange rate. 3 It also consiers a ore realistic case where the onetary authority has only one instruent an cannot anage all isturbances with that single policy intervention. The rest of the paper is organize as follows. Section II escribes the oel an erives solutions uner various exchange rate regies incluing a free float as the benchark, a target zone an a anage float. Section III copares acroeconoic stability effects on the exchange rate an the price of those alternative regies: a free float vs. intereiate regies an a target zone vs. a anage float. Section IV conclues the paper with a brief suary. II. BASIC MODEL AND SOLUTIONS. Basic oel To copare various arrangeents of exchange rates, this paper uses a siple open-acroeconoic oel consiste of the following equations. 4 * y = b( i E CPI ) + b( s+ p p) + u () = c y c i+ u () * i i E() s = 0 (3) y = y (4) = CPI (5) x In the above oel, x an E enotes the expectation operator. CPI = t * ( α) p+ α( s+ p ), 0 < α <, is the consuer price inex where α is the share 3 4 Klein (99) also uses a two-equation oel but inclues two policy instruents so that all isturbances are controllable. For siplicity, the notation for tie, t, is roppe.

4 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies of foreign goos in the inex. y, s, p, p * an are the logariths of output, the exchange rate, the price level of oestic output, the price level of foreign output, an oney supply. i an i * are the noinal oestic an foreign interest rates. u an u are utually inepenent stochastic shocks to aggregate ean an oney ean, respectively. y, i * an p * are assue to be constant with their natural logariths noralize to zero. () represents an IS-type expeniture function. Since the purchasing power parity is not assue to hol continuously, the real exchange rate enters in orer to capture international relative price changes an consequent shifts in aggregate ean. () is a usual oney ean scheule, which epens on output an the interest rate. (3) represents the uncovere interest rate parity. Assuing risk neutral agents an perfect capital obility, we take foreign bons to be perfect substitutes for oestic bons. Equilibriu in the international bon arket requires that the iscrepancy between oestic an foreign interest rates shoul be equal to the expecte epreciation of the oestic currency. Finally, (4) an (5) require continuous clearing of both goos an oney arkets, where (4) entails full flexibility of the oestic price. Soe anipulation an siplification lea to the following syste of siultaneous ifferential equations: u q Eq s Es b γ 0 = + α γ( α) c u u b (6) where q = s p an γ ( α). (6) is the low of otion for the exchange rate = b b an the price. The first bracket of the right han sie is the funaental, which is copose of acroeconoic variable (e.g., ) an isturbances. It is assue that u an u evolve accoring to the process: u = μ t + σ z, j =, (7) j j j j where u j is the expecte change in u j, z j is a stanar Wiener process, an σ j is a constant.. Solutions As (6) shows, the syste has a block recursive syste. 5 We first solve (6) for the rational expectations path of the real exchange rate (q) an then, in conjunction with the solution for q, solve it to obtain the path of the noinal exchange 5 If an aggregate supply function were introuce in the oel or the real balance were an aitional arguent in the aggregate ean function, the syste woul not be block recursive.

5 YEONGSEOP RHEE 3 rate (s). Assuing no speculative bubbles, 6 path for q is then given by: the rational expectations solution γ exp ( γ q = ( r t) ) E u ( τ) Ω( t) t b (8) where Ω() t is the inforation set at tie t. Uner rational expectations, the equilibriu real exchange rate epens on the future expecte path of the funaental u. Note that there is a coplete ichotoy between real an onetary phenoena an that q epens exclusively on the funaental u which shifts aggregate ean, an not on onetary variables. Therefore, whether or not the arket expects the onetary authority to intervene in the future, the funaental is inepenent of onetary behavior an is expecte to follow (7). Thus, the real exchange rate q will be: γ exp ( γ q = ( τ t) ) u ( τ) τ = u ( t) t b b (9) Unlike the real exchange rate, the equilibriu noinal exchange rate epens on the future expecte paths of oney (), oney arket isturbance ( u ), an aggregate ean isturbance ( u ), which epen on in turn the intervention of the onetary authority. But, uner a free float without any intervention, the oney supply (t) is kept constant. As a consequence, the funaental is siply equal to the su of isturbances. This iplies that the exchange rate uner a free float is also a Brownian otion. Therefore, changes in the funaental will translate into equal changes in the exchange rate, an the noinal exchange (s) will be: t ( )[ ] s = c exp c ( τ t) f( τ) τ = f( t ) (0) where α f() t = u() t u() t b. Since p = s q, fro (9) an (0): α α p = u u u = u () t + u () t b b b () In a target zone, it is assue that the intervention rule is base on a specific ban for the funaental an that the funaental will be regulate to reain within the ban, f f() t f, where f is the lower liit an f is the upper 6 This assuption is require to obtain a unique stable sale-path to the long run equilibriu. See Sargent an Wallace (976) an Obstfel an Rogoff (984).

6 4 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies liit. This iplies that the funaental ( f ) follows a regulate Brownian otion with a constant rift ( μ ) an an instantaneous stanar eviation ( σ ): f (t) = μ t +σ z(t) + L(t) U(t) () where L(t) an U(t) are the lower an upper regulators, efine as continuous an increasing functions of t. L(t) an U(t) are zero within the ban, an L(t)>0 only if f()= t f an U(t)>0 only if f() t = f. As the funaental reaches the lower (upper) ban, the onetary authority increases (ecreases) oney supply so that noinal exchange rate will not be beyon the lower (upper) liit. If the with of the ban, f f, is zero, it correspons to a fixe exchange rate regie, while the infinite ban with iplies a free float. An intereiate regie can be efine as a case of a finite ban with: the larger the ban with, the closer to a free float. When the funaental follows a regulate Brownian otion, the solution proceure is not the sae as that for (0). Borrowing the result of Froot an Obstfel (989a) an applying the sooth-pasting conitions 7 at the bounary ipose by Krugan (99), we can obtain: 8 s = f + cμ + λexp( λ f + λ f) λexp( λ f + λ f) + λexp( λ f + λf ) ( ) ( ) ( ) λexp λ f + λ f + λλ exp λ + λ + exp λ + λ f f f f (3) where λ > 0 an λ < 0 are the roots to the quaratic equation in λ : σ λ c + λc μ + = 0. (4) For the case with a syetric zone an no rift, 9 (4) becoes: 0 exp( λf) exp( λf) s= f λ exp( λf) + exp( λf) (5) This solution for the exchange rate uner a target zone is copose two parts, one linear an the aitional non-linear part In the funaental. Since p = s q, the solution for the price of oestic output is reaily erive fro (9) an (5): 7 Those conitions are expresse as G ( f ) = G ( f ) = 0 where G( f) is the solution for the noinal exchange rate an assue a twice continuously ifferentiable function of the funaental ( f ). Krugan (99) shows the institution behin the conitions. Froot an Obstfel (989a, 989b) show other types of bounary conitions as well. 8 The solution in (3) is true only when C > 0. If C < 0, λ s are iaginary roots an the stabilizing effect on exchange rate in a target zone is gone. See Appenix. 9 For convenience, we choose units so that the ban of the exchange rate is syetric aroun zero ( s = s) so that the bounary for the funaental also becoes syetric aroun zero ( f = f). 0 We let λ = λ for ease of notation.

7 YEONGSEOP RHEE 5 α exp( λf) exp( λf) p= ( u + u). (6) b λ exp( λf) + exp( λf) Also fro (3) an (6), the interest rate ifferential is: [] * = = = i i E s s f exp( λf) exp( λf). (7) c c λ exp( λf) + exp( λf) In a anage float, since exchange rate volatility itself is viewe as a ajor proble by the onetary authority, the ost iportant objective of central bank intervention is to liit isruptive short ter oveents of exchange rates. A coon rule of exchange rate anageent for the purpose is the policy of leaning against the win as follows: = β s, β 0 (8) (8) iplies that if the exchange rate suenly increases, the onetary authority ecreases oney supply to apen the rise of the exchange rate, while it follows the opposite action in the case of suen exchange rate ecrease. If β = 0, it correspons to a policy that targets oney supply growth uner a free float, an an infinite value of β correspons to a policy uner a fixe rate regie. An intereiate value of β correspons to a anage float: Fro (8): = β s+ c (9) where c is a constant. Since the onetary authority is peranently coitte to the policy of (8), there is no regie shift in a anage float an the solution is irectly obtainable. Using (9) an following the sae proceure as the above to get the solutions uner a free float, we obtain: α s = ( u + u) + β b α + β p = ( u + u ) + β b (0) () III. MACROECONOMIC STABILITY UNDER INTERMEDIATE REGIMES Using the above solutions for the exchange rate, the price an the interest rate A negative value of β iplies the opposite policy of leaning with the win.

8 6 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies uner a free float, a target zone an a anage float, we can escribe an copare the behaviors of econoic variables uner alternative regies.. Target zone vs. Free float Fro (0) an (5), it is apparent that a target zone has a stabilizing effect on exchange rates within the ban. This can be irectly shown. Since the exchange rate is a function of the funaental, s = G( f ) in (5), it will be an Ito process. Then, by Ito s lea, the instantaneous stanar eviation of the exchange rate is siply the prouct of the erivative an the instantaneous stanar eviation of the funaental. Thus, the variance of the exchange rate in a target zone is: s exp( λf) + exp( λf) Vtz = [ G'( f) ] σ = σ. exp( λf) + exp( λf) () Since f f f, the secon ter in the parenthesis always resies between 0 an s. Therefore, 0 G'( f ) an V tz of a target zone is always less than σ, which is the variance of a free float. This stabilization effect on the exchange rate takes place even when the onetary authority is not actively efening the ban because the arket s expectations of a future intervention at the liits are reflecte in the current exchange rate. (5) also shows that the stabilizing effect in a target zone epens on the with of the ban. With a large ban, the exchange rate woul be approxiately linear in the funaental while, with a sall ban, it is ore non-linear in the funaental. The arket s regressive expectations ay be represente as the curvature of the exchange rate near the ege of the ban. As the ban grows wier, the exchange rate gets closer to the free floating exchange rate except aroun the ege an the curvature of the exchange rate near the ege becoes sharper, which eans the stabilizing effect eclines. 3 As Floo an Garber (989) pointe out, however, while a target zone stabilizes exchange rates, it estabilizes interest rates. The uncovere interest parity tells us that the interest rate ifferential equals the expecte epreciation of the oestic currency. 4 For both a fixe rate regie an a free float, the interest ifferential will be constant (zero in this oel). In contrast, since s f in (7) inclues the aitional rano eleent, the interest ifferential is not constant in a target zone. Thus, interest rates are estabilize while exchange rates are stabilize: i.e., exchange rates are stabilize at the expense of the interest rate stability in a target zone. Although the target zone s effect on the stability of exchange rates an interest rates are apparent, its effect on other econoic variables such as prices is not. Accoring to (0), the variance of the exchange rate uner a free float is that of the funaental, i.e., σ. 3 See Svensson (989, 990). 4 We assue no risk preiu. For the case with an enogenous risk preiu, see Svensson (989).

9 YEONGSEOP RHEE 7 When we copare () an (6), it is reveale that uner a free float the price is given by the first bracket of (6) while the secon ter is ae in a target zone. This ter reflects the effect of non-zero conitional expectations cause by aitional constraints on the funaental. In our oel, when the onetary authority reacts to shocks to keep the funaental fro oving outsie the liits, it α ajusts oney supply () in response to the su of isturbances, v = u + u b containe in the funaental ( f ). This intervention will control the secon ter entering into the eterination of the price. Although the su of isturbances is controllable, none of the is separately controllable. Money supply which controls the su cannot separately control u an u an therefore the price which inclues both are not fully controlle through the oney supply ajustent. Fro (6) the ifference of price volatility between a target zone an a free float is given by: α V V = ( G')( + G') σ ( G') + ( G') σ p p fl tz b (3) where p V fl an p V tz are the instantaneous variances of the price in a target zone an a free float respectively. Accoring to (3), prices can be either stabilize or estabilize in a target zone, epening on the econoic structure an the characteristics of isturbances. If the sensitivity of aggregate ean to the real exchange rate ( b ) is very high or the onetary isturbance ( u ) is very large copare to the real one ( u ), () is likely to be positive an a target zone leas to price stability. But if b is sall or the real isturbance is preoinant, () is likely to be negative an a target zone oes not result in price stability but instability. It iplies that the stabilizing effect on the exchange rate of a target zone ay not be extene to other econoic variables an that a target zone ay not be a goo hybri of a fixe an a floating exchange rate regies for econoic stability. The intuition behin this observation is the following. Suppose that the onetary isturbance is very large. The arket knows that the exchange rate an the price are ostly influence by onetary factors an believes that onetary policies to control the exchange rate through oney supply change will be effective an can affect price oveent as well. In this case, the effect of regressive expectations on the exchange rate in a target zone also applies to the price, an prices will be stabilize. In contrast, when the real isturbance is preoinant, the arket s response to onetary policies woul be ifferent. Now the arket knows that the exchange rate an the price are ostly affecte by real factors an will question the effectiveness of onetary policies. If the onetary authority intervenes in the foreign exchange arket to control the exchange rate, the arket woul rather believe that it ay be effective to ove the exchange rate but the policy woul istort the goos arket ajustent an also price behavior. Thus

10 8 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies the effort to stabilize exchange rates ay estabilize prices.. Manage Float vs. Free Float Fro (0) an (0), it is also apparent that a anage float has a stabilizing effect on exchange rates, not because of regressive expectations but rather because of continual intervention. The ifference of exchange rate volatility between a anage float an a free float is: s α V f = σ + σ = σ β + b + β (4) where s V f is the instantaneous variance of exchange rate in a anage float. Since is always less than one, this iplies that a anage float leas to exchange rate stability. + β Interest rates will be neither stabilize nor estabilize in a anage float. Since expecte epreciation is zero in (0), the interest ifferential fro the uncovere interest parity will be zero in a anage float as in a free float. On the other han, the price can be either stabilize or estabilize in a anage float. The ifference of price volatility between a anage float an a free float is: V β( + β) β( + β)( α + β + αβ) V = σ σ p p fl f ( + β) b ( + β) (5) This equation iplies that the stabilization of prices in a anage float also epens on the econoic structure an the characteristics of isturbances. If the sensitivity of aggregate ean to the real exchange rate ( b ) is large enough or the onetary isturbance is preoinant, prices can be stabilize in a anage float. In contrast, if b is low or the real isturbance is huge copare to the onetary one, prices ay be estabilize rather than stabilize. Again, the intuition behin this is the sae as the above. If the onetary isturbance is preoinant, the exchange rate an the price woul show siilar behaviors because they are ostly influence by onetary factors. In this case, the policy to control the oveent of exchange rates woul effectively control that of prices as well, an the policy of leaning against the win to stabilize exchange rates will lea to the stabilization of prices. If the real isturbance is very large copare to the onetary one, on the other han, the behavior of the exchange rate an that of the price woul be quite ifferent fro each other. Therefore, the policy to control exchange rates woul be no longer effective for stabilizing prices.

11 YEONGSEOP RHEE 9 3. Target zone vs. Manage float Now, let s copare two alternative intereiate regies, a target zone an a anage float. By coparing (5) an (0), we can obtain: s s V [ '( )] tz Vf = G f σ = G'( f) G'( f ) + σ (6) + β + β + β To erive soe eaningful iplication, let s change the expression in (6) a exp( λf) + exp( λf) little bit by letting A =. Then G'( f) = A an substituting exp( λf) + exp( λf) this into (6) yiels: β + β + β + β + β s s Vtz Vf = A G'( f) + σ = A G'( f ) + σ (7) Now we can interpret A= G'( f ) as the intervention inex in a target zone. As the liit f increases, G'( f ) onotonically increases an the regie gets closer to a free float. As the foreign exchange arket tightens, A= G'( f ) increases while G'( f ) eclines. Therefore, A= G'( f ) can be regare as the egree of foreign exchange arket intervention: A = 0 eans no intervention in a free float an A = perfect intervention in a fixe rate regie. A value in between shows β an intereiate target zone. Now, let B = =. Again we can interpret this as the intervention inex in a anage float. As β increases, the inter- + β + β vention increases an the regie gets closer to a fixe rate regie: B = 0 eans no intervention in a free float an B = perfect intervention in a fixe rate regie. A value in between shows an intereiate anage float. Using A an B, (7) can be expresse as: s s Vtz Vf = [ B A] G'( f ) + σ (8) + β The sign of this equation only epens on B A. That iplies that if the intervention in the arket is tighter 5 in a regie than in the other, the tighter intervention clearly leas to a greater stabilization of exchange rates in that regie. For exaple, if the intervention in a target zone is relatively tight to that in a anage gloat, B A < 0, then (8) becoes negative an a target zone leas to 5 Here a tighter or looser intervention eans that A or B, the intervention inex, is higher than the other.

12 30 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies ore stability of exchange rates than a anage float. However, this avantage of tighter intervention oes not always lea to ore stability of other econoic variables. 6 Whether the greater tenency of one intervention rule in a regie to stabilize exchange rates is associate with that in the other regie epens on the econoic structure an the source of isturbances. Fro (6) an (), we can erive: α V V = ( B A)( A B) σ ( B A) ( α)( A+ B) + α σ [ ] p p tz f b α = ( B A) ( A B) σ [( α)( A+ B) + α] σ b (9) Suppose the intervention in a target zone is tighter than in a anage float. Even though B A < 0, this oes not guarantee a saller variance in a target zone because the sign of the secon bracket of the right han sie is uncertain. If the sensitivity of aggregate ean to the real exchange rate ( b ) is very large or the onetary isturbance is preoinant ( σ >> σ ), then the secon bracket becoes positive an (9) becoes negative. I.e., the stabilization effect on exchange rates of a tighter intervention is positively associate with that on prices. In contrast, if b is very sall or the real isturbance is preoinant ( σ << σ ), then the secon bracket becoes negative an (9) becoes positive. Thus the stabilization effect on exchange rates of a tighter intervention is not extene to that on prices. Both a target zone an a anage float require foreign exchange arket intervention by the onetary authority to constrain exchange rate oveents. The effect of the intervention rules on the stability of econoic variables iffers fro each other. In a target zone, they intervene infrequently; but when they o intervene (when the exchange rate reaches the ege of the ban), a big ajustent in oney supply is neee to fully offset the effect of isturbances on the exchange rate. On the other han, in a anage float the authority intervenes in the arket continuously. The necessary ajustent of oney supply is saller, because the authority oes not nee to fully offset the effect of isturbances on the exchange rate but only to apen the effect so as to reuce an sooth suen changes of exchange rates. As we showe in the above, the relative effectiveness between infrequent big ajustent an frequent sall ajustent epens on the econoics structure an the characteristics of isturbances. IV. CONCLUSION This paper copare alternative intereiate exchange rate regies instea of coparing an extree regie vs. an intereiate one. Particularly focusing on a 6 Regaring the stability of interest rates, a target zone is always worse than a anage float.

13 YEONGSEOP RHEE 3 target zone an a anage float, we wante to investigate whether an intereiate regie leas to the stability of exchange rates an whether the stabilizing effect on the exchange rate can be extene to other econoic variables. In the paper, an exchange rate regie is classifie accoring to the rule of foreign exchange arket intervention. A free float is efine as the case with no intervention at all in the arket, while a fixe rate regie as that with perfect intervention so that the exchange rate oes not ove at all. In between, there are any kins of intervention rules an within each rule continuous egrees of intervention. The intervention rule of a target zone is that within the ban, there is no intervention; but if the exchange rate approaches a argin, the central bank intervenes in the arket by increasing or ecreasing oney supply so that the noinal exchange rate will not be beyon the lower (upper) liit. The intervention rule of a anage float is siplifie; oney supply is ajuste to the changes in the exchange rate accoring to the policy of leaning against the win. Base on the onetary oel of the exchange rate which inclues both expecte epreciation an expecte inflation an yiels a ultiple-equation syste of the exchange rate an the price instea of a single-equation syste of the exchange rate, we foun the followings. First, the stabilizing effect on exchange rates takes place in an intereiate regie, whether the regie follows a target zone intervention rule or a anage float one. Secon, however, the stabilizing effect on exchange rates is not extene to prices: it epens on the econoic structure an the source of isturbances. Thir, between a target zone an a anage float, a tighter intervention in a regie leas to ore stability of exchange rates, copare to the other. Fourth, again whether this stability effect leas to the stability of other econoic variables epens on the econoic structure an isturbances. If the sensitivity of aggregate ean to the real exchange rate is very large or the onetary isturbance is preoinant copare to the real isturbance, then the stabilizing effect on exchange rates of a tighter intervention in an intereiate regie is positively associate with the stabilizing effect on prices; if the sensitivity of aggregate ean to the real exchange rate is very sall or the real isturbance is preoinant, the stabilizing effect on exchange rates of a tighter intervention can be achieve only at the expense of price stability. REFERENCES Benassy-Quere, A. an B. Coeur, The Survival of Intereiate Exchange Rate Regies, CEPII, France, 00. Floo, R. P. an P. M. Garber, A Moel of Stochastic Process Switching, Econoetrica, Vol.5 No.3, 983, pp Floo, R. P. an P. M. Garber, The Linkage between Speculative Attack an Target Zone Moels of Exchange Rates, NBER Working Paper # 98, 989. Frankel, J., No Single Currency Regie is Right for All Countries or at All

14 3 Macroeconoic Stability Effect of Intereiate Exchange Rate Regies Ties, Essays in International Finance, Princeton University, # 5, 999. Froot, K. A. an M. Obstfel, Exchange Rate Dynaics uner Stochastic Regie Shifts: A Unifie Approach, NBER Working Paper # 835, 989a. Froot, K. A. an M. Obstfel, Stochastic Process Switching: Soe Siple Solutions, NBER Working Paper # 998, 989b. Habereier, K., A. Kokenyne, R. Veyrune, an H. Anerson, Revise Syste for the Classification of Exchange Rate Arrangeents, IMF Working Paper # 09/, 009. IMF, De Facto Classification of Exchange Rate Regies an Monetary Policy Fraework, 005. Klein, M. W., Playing with the Ban: Dynaic Effects of Target Zones in an Open Econoy, International Econoic Review, Vol.3, 990, pp Krugan, P., Target Zones with Liite Reserves, ieo, 989. Krugan, P., Target Zones an Exchange Rate Dynaics, Quarterly Journal of Econoics, 99. Miller, M. H. an P. Weller, Exchange Rate Bans an Realignents in a Stationary Stochastic Setting, in M. H. Miller, B. Eichengreen, an R. Portes (es.), Blueprints for Exchange-Rate Manageent, Acaeic Press, 989, pp Moon, W., Y. Rhee an D. Yoon, Asian Monetary Cooperation: A Search for Regional Monetary Stability in the Post-euro an the Post Asian Crisis Era, Econoic Papers, Bank of Korea, Vol.3, No., 000. Rhee, Y., East Asian Econoic Integration: Destine To Fail?, Social Science Japan Journal, 004. Sargent, T. an N. Wallace, Rational Expectations an the Theory of Econoic Policy, Journal of Monetary Econoics, Vol., No., 976, pp Svensson, L., Target Zones an Interest Rate Volatility, Institute for International Econoic Stuies Seinar Paper, Vol.4, 989. Svensson, L., The Siplest Test of Target Zone Creibility, NBER Working Paper # 3466, 990. Williason, J., The Exchange Rate Syste, Institute for International Econoics, Policy Analysis in International Econoics, Vol.5, 985.

15 YEONGSEOP RHEE 33 <APPENDIX> If c < 0, then λ an λ in (4) are iaginary roots. Let λ = λi an λ = λi. The general solution to (4) is: G( f) = f + A exp( λ f) + A exp( λ f ) = f + A exp( λ fi) + A exp( λ fi ) cos( λ ) sin( λ ) [ cos( λ ) sin( λ )] cos( λ ) sin( λ )] = f + A f + i f + A f i f = f + B f + B f (A-) where B = A + A an B = ( A A ) i. Sooth-pasting conitions now becoe: G'( f) = λbsin( λ f) + λbcos( λ f ) = 0 (A-) G'( f) = λ B sin( λ f) + λb cos( λ f ) = 0 (A-3) Since f = f, (A-3) changes into: G'( f) = + λb sin ( λf) + λb cos( λf ) = 0 (A-4) Fro (A-) an (A-4) we obtain: Therefore: B = 0 (A-5) B = λ cos( λ f ) (A-6) G( f ) = f sin( λ f) λcos( λf ) (A-7) G( f ) in (A-7) oes not converge to a level but fluctuates. Therefore, when C < 0 an (4) has iaginary roots, the stabilizing effect of expectations is gone.