Monetary Model of Exchange Rate for Thailand: Long-run Relationship and Monetary Restrictions

Transcription

1 MPRA Munich Personal RePEc Archive Moneary Model of Exchange Rae for Thailand: Long-run Relaionship and Moneary Resricions Venus Khim-Sen Liew and Ahmad Zubaidi Baharumshah and Chin-Hong Puah Universii Malaysia Sarawak, Universii Pura Malaysia, Universii Malaysia Sarawak 009 Online a hp://mpra.ub.uni-muenchen.de/17715/ MPRA Paper No , posed 8. Ocober 009 1:44 UTC

2 Moneary Model of Exchange Rae for Thailand: Long-run Relaionship and Moneary Resricions Venus Khim-Sen Liew a,*, Ahmad Zubaidi Baharumshah b and Chin-Hong Puah a a Faculy of Economics and Business, Universii Malaysia Sarawak, 9400 Koa Samarahan, Sarawak, Malaysia. b Faculy of Economics and Managemen, Universii Pura Malaysia, 4400 UPM Serdang, Selangor, Malaysia. Absrac This paper examines he long-run relaionship beween exchange rae and is deerminans based on he flexible-price moneary model. Mulivariae coinegraion approach (Johansan 1988, 1989 and Johansen-Juselius 1990) is adoped o aain our objecive of sudy. The empirical resuls provide evidence favoring he moneary approach o exchange rae for a small and open emerging economy, namely Thailand. In addiion, he validiy of he underlying assumpions of he moneary approach o he deerminaion of exchange rae is esablished. The above findings sugges ha exchange rae players may effecively monior and forecas he exchange rae movemen via he money supplies, incomes, and ineres raes variables of boh Thailand and Japan. Besides, one has o follow he economic developmen of Thailand s major rading parner, Japan, o undersanding he movemen of exchange rae for Thailand. Moreover, our findings add new insighs o accompanimen previous sudies ha documened he imporan influence of US in he emerging Asian economies. Keywords: Exchange rae, moneary model, Thailand, coinegraion This paper is forhcoming in Global Economic Review. * Corresponding auhor.

3 1. Inroducion The validiy of he moneary model exchange rae has been of paricular ineres in he exchange rae sudy in he pas 40 years or so. This is due o he fac ha exchange raes have been nooriously volaile ever since he saring of floaing exchange rae regime in he beginning of 197. This volaile naure renders exchange rae movemen somewha difficul o be racked. Subsequenly, being able o undersand or even predic he movemen of exchange rae is crucial o foreign raders and invesors. Policy makers also find i imporan o monior and manage he movemen of exchange rae so ha i will no depar oo far away from economic fundamenals. Oherwise, i would be harmful o he rade and disrup economic progress. Few models have been pu forward in he lieraure o help undersand he exchange rae movemen. Among ohers, he flexible-price moneary model, which posulaes ha exchange rae may be deermined by he money supplies, aggregae incomes and ineres raes of domesic and foreign counries, has received much aenion in he lieraure; see among ohers, MacDonald and Taylor (1991), Choudhry and Lawler (1997), Du and Ghosh (000), Miyakoshi (000), Abbo and de Via (00) and he more recen work Long and Samreh (008). MacDonald and Taylor (1991) examine he long-run validiy of he his model of exchange rae deerminaion by employing he Johansen (1988, 1989) and Johansen and Juselius (1990) coinegraion procedure. They provide supporive evidence for wo major currencies, namely he USD and Briish pound. Meanwhile, Choudhry and Lawler (1997) apply he Johansen-Juselius and Engle-Granger approaches o es he validiy of moneary model for he case of Canada. The oucome of heir ess resuls reveals he exisence of long-run relaionship beween Canadian Dollar- US Dollar and he variables of moneary models including

4 money socks, incomes and ineres raes over he period of Canadian floa spanning from Ocober 1950 o May 196. Du and Ghosh (000) applied KPSS and Johansen-Juselius approaches o deermine he coinegraion relaionship beween nominal Japanese Yen-US Dollar exchange raes and moneary fundamenals (money supply, ineres raes and income), in he fixed exchange rae regime (1959M1 o 197M1). Empirical evidence in favor of moneary model is obained in his sudy. Miyakoshi (000) examines he flexible-price moneary model in he case of Korea. Based on he Johansen-Juselius coinegraion echnique, his sudy finds ha he Korean Won exchange raes wih US Dollar, German Mark and Japanese Yen as numeraires are all coinegraed wih money supplies, incomes and ineres raes during he sample period 1980M1 o 1996M1. The auhor hus concludes ha he flexible-price moneary model has long-run validiy (in Korea). The focus of previous sudies on he moneary model is on he developed counries, however. Laely, Lee e al. (007), Long and Samreh (008) and Liew (009) resusciae he sudy of moneary model in he conex of emerging economy. These sudies are able o provide evidence favoring he long-run validiy of his moneary model for he case of he Philippines 1. Baharumshah e al. (009) also find he predicive power of moneary model ouperformed well. Specifically he ou-sample forecas of he moneary model ouperforms he naïve random walk model a four o eigh quarers horizon. The overall findings from hese sudies have moivaed us o apply he model o he Thai bah. 1 Lee e al. (007) adop he mulivariae coinegraion echniques and he VECM approach in heir analysis, while he Auoregressive Disribued Lag (ARDL) approach is employed in he sudy of Long and Samreh (008). Liew (009) conribues o he lieraure by employing he linear and nonlinear moneary approaches.

5 This sudy aemps o conribue o he lieraure by adding empirical evidence for Thailand, a small and open emerging economy, which has no received much aenion before he oubreak of he Asian Currency Crisis. In paricular, his sudy aims o examine he long-run validiy of he flexible-price moneary model for he bah-yen exchange rae. The Japanese yen is chosen as he base currency, since i is he radiional major rading parner of Thailand. In he year 006, Thailand and Japan regisered a bilaeral rade value of 4416 million of USD rendering Japan as he leading rading parner (followed by US, China, Malaysia and Singapore in ha order) of Thailand (ASEAN-Japan Cenre, 007). Various popular moneary resricions are also esed in his sudy o provide furher insigh on he relaionship among exchange rae and is deerminan. For his purpose, he commonly adoped mulivariae coinegraion echniques in Vecor Error Correcion (VEC) framework are uilized o achieve hese asks. The remainder of his sudy is organized as follows. Secion offers an overview of he Thailand exchange rae regime since 196. Secion describes he model and Secion 4 explains he esing procedures. Secion 5 conains he daa descripion and discussions on he esimaed resuls. The final secion spells ou he implicaions of sudy and concludes.. Overview of Thailand exchange rae regime Thailand was said o have sared off a series of failures in he financial and exchange rae markes in he Asian region hrough conagion effec in he second half of year Before ha, Thailand has achieved a remarkable average growh rae of 8% per year for wo decades, due largely o he opening of he economy o inernaional rade (Agbola and Kunanoppara, 005).

6 Thailand adoped a floaing exchange rae regime before 196. However, Bank of Thailand (BOT) abolished his regime on 0 Ocober 196 and bah (B) was pegged agains USD a a rae of B0.80 per USD. In order o hang on o he pariy, he gold conen of Thai Bah had been reduced by 7.89% in December To avoid he coninuous reducion of gold reserves, he BOT revised he exchange rae policy by allowing Bah o floa wihin a 4.5% flucuaion range in May 197. However, on he 14 February 197, he gold conen of bah was furher reduced by 10% o accommodae for he USD devaluaion. The implemenaion of he limied flucuaion range was aken ino effec on 15 July 197 when he gold conen was raised by 4% and he official rae was upgraded o B0.00 per USD. The pegged susained for few years unil 8 March 1978, when an effecive rae was esablished based on a weighed baske of currencies of U.S., Wes German, Swiss and Japan as major rading parners. This marked he end of bah s link o USD. On 0 Ocober 1978, BOT placed he Effecive Rae under conrolled (Conrolled Floaing Rae) and bah was allowed o floa wihin a limied range. Since hen, bah devalued several imes o improve expor compeiiveness. On 5 November 1984, he baske of currencies was revised o include currencies of U.S., Japanese, Wes German, Unied Kingdom (U.K.), Malaysia, Hong Kong and Singapore. To sop capial flow, a managed floa was adoped o conrol he currency on December A he same ime Bah was depreciaed % o B6.69 per USD o promoe expors. In 1990, more currencies (Brunei ringgi, Indonesian rupiah and Philippine peso) were added o he currencies baske. In he oubreak of he Asian Currency Crisis, bah fell by % in one monh, from 5.80 bah per USD in June 1997 o 4.57 bah per USD in he following monh. Since July 1997, Thailand had given up in defending is currency and adoped he independen floa exchange rae regime, of which he value of he Bah is deermined by marke

7 forces. The bah fell by 60% in jus few monhs, o is hisorical low of 7.89 bah per USD in January 1998 before i reversed is course o appreciae. The exchange rae was as high as bah per USD in April 006, bah had no resumed is value prior o he crisis, as was he case for he currencies of he res of ASEAN-5 counries. As of 006, he de faco classificaion for bah exchange rae according o IMF was a managed floaing regime wih no pre-deermined pah for he exchange rae, bu basically is guided by he inflaion forecass, which ac as he inermediae arge (IMF, 006). As of 008, Thailand is sill pursuing his inflaion argeing framework (IMF, 008).. The moneary model The flexible-price moneary model of exchange rae may be represened by: * e = ( m - m ) + (y * 1 - y ) + (i * - i ) + μ (1) where e is nominal exchange rae expressed as domesic price of foreign currency. In he presen case, bah-yen exchange rae is considered. Meanwhile, m, y and i sand for domesic money supply, aggregae naional oupu and nominal ineres rae respecively. The corresponding variables for he foreign counerpar are marked wih aserisk. μ is he whie noise. This model posulaes ha he movemen of exchange rae may be deermined by he * differenials of money, income and ineres rae [ ( m - m ), (y * 1 - y ), (i * - i ) ]. According o his model, a rise in he domesic money supply leads o a proporional

8 rise in he price level via he quaniy heory of money and o a proporional rise in exchange rae via he purchasing power pariy and vice versa. The same is rue for foreign money supply. Besides, income differenial and ineres rae differenial are expeced o have negaive and posiive impacs respecively on he exchange rae movemen. Moreover, i is assumed ha he elasiciies for domesic and foreign money are idenical. Tha is, hese variables have effec of opposie direcion bu of same magniude on exchange rae. The same assumpion is apprehended for domesic and foreign ineres rae variables; for more deails, see, for insance, Frenkel (1976), MacDonald and Taylor (1991) and Du and Ghosh (000). 4. Tesing procedures and moneary resricions Empirically, he model is esed in he less resriced form, in which he proporionaliy beween money differenial and exchange rae, he idenical elasiciies of incomes and also of ineress, are relaxed. The esable version can be wrien in he following reduced form: e 1m + * m + y * + 4 y + i 5 + i * 6 + μ () Convenionally, he long-run validiy of his model has been esed using he Johansan (1988, 1989) and Johansen-Juselius (1990) coinegraion esing procedure in he mulivariae framework. This procedure requires he variables in he model o be inegraed in he same order, in paricular, order one (MacDonald and Taylor, 1994; Rapach and Wohar, 00). If all variables are inegraed of order one, I(1), hen here is a possibiliy ha he linear combinaion of his variables are saionary. This in urn may be aken as evidence of long run (coinegraing) relaionship among he variables

9 under examinaion. As such, he following Vecor Auoregression (VAR) of order k, denoed as VAR(k), in he Vecor Error Correcion (VEC) framework exiss and (Engle and Granger, 1987, Johansan 1991) may be represened as: k 1 X X-k-1 i X-i D, () i 1 * * * where X =( e, m, m, y, y, i, i )' is a 7 1 vecor of moneary variables, D conains a se of condiional variables which may include consan, rend and dummies. ε ~ iid (0,Ω) is a 7 1 vecor of independenly and idenically discibued (iid) Gaussian error erms. X k 1 is a 7 1 vecor of he error correcion erms (ECTs), ha measures he long-run equilibrium relaionships. Π = αβ ' is a 7 7 marix ha conains informaion abou he of long-run relaionships among he variables in he VAR sysem, where α and β are each of dimension 7 r, where r is he number of coinegraing vecors conains in marix β. α describes he speed of adjusmen of each of he 7 individual variables in he sysem o deviaions from he coinegraion relaionships. In our case, if 0 r 6, hen β' X is saionary even hough X iself is no. This may be inerpreed as he exisence of long-run relaionship among he variables. The value of r can be deermined by he race (Johansen, 1989) and maximum eigenvalue ( λ -max) (Johansen, 1988) saisics. The race saisic is used o es he null hypohesis of here are a mos r coinegraing vecors agains he alernaive hypohesis of more han r coinegraing vecors, and i is given by: Trace=-T k i= r+1 ln(1- λˆ i ), (4)

10 where T is he sample size, λ ˆr + 1,, λˆ k are k - r smalles esimaed eigenvalues. The maximum eigenvalue saisic ess he null hypohesis of r coinegraing vecors agains he alernaive hypohesis of r+1 coinegraing vecors and i is compued as: λ -max=-t(1- λ ˆr + 1 ) (5) This sudy uilized he race saisics as i is more robus o boh skewness and excess kurosis in he residuals han he λ -max saisic (Cheung and Lai, 199). In he lieraure, some moneary resricions are commonly performed (MacDonald and Taylor, 1991; Long and Samreh, 008). The resricions o be imposed in his sudy are lised in he firs column of Table. Briefly, wih reference o Equaion (), he null hypohesis of H 1 : β1 =- β =1 ess wheher here is proporionally beween exchange rae and he money differenial. The resricions of opposie coefficiens on income and ineres rae differenial are esed in H β + = 0 and H β + = 0 : β4 : 5 β6 respecively. The oher null hypoheses, which include H 4 : H1, H 5 : H1, H : H and H 7 : H1, are o es he join significance of he some of 6 H he combined effecs of he firs hree resricions. The likelihood raio (LR) es of resricions is applied for hese hypoheses esing. 5. Daa, empirical findings and discussions

11 Monhly daa uilized in his sudy is obained from Inernaional Financial Saisics published by Inernaional Moneary Fund. Income, money supply and ineres rae are, in ha order, measured by gross domesic produc (GDP), M, consumer price index and money marke rae. The bah-yen exchange rae is employed in his sudy. The sample period runs from January 1977 (1977: M1) hrough March 006 (006: M). All variables are log-ransformed. In his sudy, he semi-parameric esing procedure of Phillips and Perron (PP, 1988) are deployed o es he inegraion order of he variables. The resuls are repored in Table 1. I is eviden from Table 1 ha he null hypohesis of non-saionary series can be rejeced in none of variables in heir levels, even a 10% significance level. However, all variables are found o be saionary afer firs-differencing, a 1% significan level. As hese variables achieved saionary only in he firs differences, here are said o be inegraed order one, I(1). Having idenified ha all variables are I(1) variables, he VAR model is hen esimaed in he firs difference according o Equaion (). In order o esimae Equaion (), he opimal lag lengh k mus be deermined in advance. Following MacDonald and Taylor (1991), k is se a a maximum value of 1 and for lag selecion, he general o specific approach is conduced wih he help of likelihood raio (LR) saisic. The VAR residuals are checked for whieness. If he residuals are found o be non-whie noise, a higher lag srucure is added unil hey are whiened. MacDonald and Taylor (1991) examine he same moneary model for he US dollar based currency of Germany, Japan and UK using monhly daa. The auhors provide empirical evidence of long-run validiy for he moneary model in hese counries.

12 Table 1. Order of Inegraion Tes Resuls. Variable Level Firs difference Decision Consan + Trend Consan Lag a Saisic Lag Saisic b e * I(1) m * I(1) i * I(1) g * I(1) * m * I(1) * i * I(1) * g * I(1) Criical value 1% % % Noes: a A maximum of 1 lag order is considered in he PP es and he opimal lag repored is seleced based on he Newey-Wes bandwidh crierion. b The null hypohesis of saionary series is rejeced in favor of he alernaive of non-saionary series if he es saisic is smaller han he criical value. Aserisk indicaes rejecion of he null hypohesis a 1% significance level. The LR saisic suggess ha VAR (1) model is opimal. However, i is observed ha he residuals of VAR (1) model are no whie noise. Higher lags are herefore added one by one unil he residuals are whiened. This resuled in he final selecion of VAR (16) model o be esimaed 4. Thus, he coinegraion es is hen performed and he race es saisic is hen compued. A his sage, here is a concern on he specificaion of deerminisic componen in he coinegraion relaion, as i can influence he oucome of he es. In his sudy, hree pracical model specificaions are esimaed. They are Model which permis inercep in he coinegraion relaion; Model which includes deerminisic rends in he levels, and Model 4 which allows for rend in he coinegraion relaion. 4 The Lagrange Muliplier (LM) es resuls indicae ha here is no serial correlaion in he VAR (16) model s residuals up o 4 lag order, implying he fulfillmen of he requiremen of whieness (Du and Ghosh, 000). Resuls are available upon reques.

13 The esimaed resuls are presen in Table, in which he repored race saisic has been adjused for small finie biased by a correcion facor of (T-nk)/T, where n=7 in his sudy, o avoid over rejecion of no coinegraion (Cheung and Lai, 199; Oserholm, 00). The opimum model may be chosen based on he Panula principle (Johansen 1995). The selecion sraegy by his principle begins wih looking a he mos resricive model (Model ) for he null hypohesis of r=0 from Table and comparing he es saisic wih he corresponding criical values given o he righ of he es saisics. If he es saisic exceeds he criical value, he model is rejeced, as is he case here, we proceed on o Model under he same null hypohesis of r=0. As Model is also rejeced here, we nex consider Model 4 in he same row. Since Model 4 is also rejeced, we move o he nex row wih null hypohesis of r 1 and sar over anoher round of inspecion. I is clear ha all Models in his row are rejeced. So he process is coninued wih he nex row wih he null hypohesis null hypohesis of r, unil he null hypohesis is no rejeced for he firs ime, in his case, in Model. Subsequenly, Model has been seleced by he Panula principle. Trace saisic suggess ha here are a mos coinegraion relaionships among he 7 variables considered 5. In oher words, we have obained evidence supporing he long-run validiy of he moneary model under scruinized for Thailand. The in-sample forecass of he bah-yen exchange rae obained from he fied model is ploed in Figure 1, alongside wih he acual values. Figure 1 shows ha he forecased value is able o follow he movemen of he acual exchange rae closely 6. In addiion, he Pearson correlaion coefficien of he wo series is 0.99 and i is significan a 1% 5 Ou of curiosiy, he Panula principle is also applied o he maximum eigenvalue saisic. Model is consisenly seleced bu his saisic suggess ha here is only 1 coinegraing relaionship. Resuls are no repored bu are available upon reques. 6 The forecas error as measured in roo mean squared error (RMSE) is 0.05.

14 level. This suggess ha he wo series are almos perfecly and posiively correlaed, hereby cross-validaing our earlier conclusion ha he forecased values can closely mimic he acual exchange rae movemen. Moreover, he R value obained from regressing he acual values on a consan and he forecased values is 0.986, implying ha 98.6% of he variaion in he exchange rae can be explained by he variaion in he moneary variables in he fied model 7. All-in-all, hese findings indicae ha he esimaed VAR(16) model fi he daa adequaely. Table. Model Selecion and Trace Saisic Hypohesis Model Model Model 4 Null Alernaive Trace 1% CV Trace 1% CV Trace 1% CV r = 0 r > * * * r 1 r > * * * r r > 9.7* r r > * r 4 r > r 5 r > r 6 r = Noes: The 1% criical values are included in parenheses. Aserisk indicaes rejecion of he null hypohesis a 1% significance level. 7 The inercep is insignificanly differen from zero, whereas he slope coefficien is exacly one and i is significan a 1% level.

15 The finding of long-run validiy of he moneary model as specified in Equaion () allows us o proceed o some ess of moneary resricions which are commonly imposed in he lieraure (MacDonald and Taylor, 1991). The resuls as shown Table reveal ha he oucomes depend very much on he number of coinegraing equaions (r) we specified in he es. For r=1, all he resricions can be rejeced a 5% significance level or beer. On he oher hand, for r=, hree hypoheses, namely H 1 : β 1 =- β =1, H 4 : H1 and H 5 : H1, can no be rejeced even a 10% significance level. However, he oher wo hypoheses, ha is, H 6 : H and H : H can be rejeced a 5% significance level. Meanwhile, LR es is no 7 H1 available for H β + = 0 and H β + = 0 due o no binding resricions in he esimaion. : β4 : 5 β acual forecas Figure 1: Forecased and acual values of bah-yen exchange rae

16 As he number of coinegraing equaion has been idenified as previously by he race es, we should resor o he resuls for he case r =. Accordingly, from he nonrejecion of H 1, we can conclude he bah-yen exchange rae response proporionally o changes in money differenial beween Thailand and Japan. This will leave he real exchange rae unchanged and hus neuraliy of money is binding for he case of Thailand. Besides, failing o rejec H 4 implies ha he domesic and foreign incomes do have same impac on he movemen of exchange rae bu hey ac in differen direcion, apar from he fac ha money is neural. By he same principle, he nonrejecion of H 5 indicaes ha domesic and foreign ineres raes move of he exchange rae in opposie direcion by he same amoun. Table. Moneary resricions ess Hypohesis r =1 r = χ Probabiliy χ Probabiliy H 1 : β 1 =- β =1 8.05** H : β + β4 = *** 0.00 n.a. n.a. H : β5 + β6 = ** 0.05 n.a. n.a. H 4 : H ** H 5 : H ** H 6 : H 9.75** ** 0.0 H : H H 8.16*** ** Noes: χ of he LR es has a r m degrees of freedom, where r refers o he number of coinegraing vecors and m is he number of resricion in he null hypohesis. n.a. sands for unavailabiliy of LR es due o no binding resricions in he esimaion. ** and *** indicaes he rejecion of null hypohesis a 5 and 1% significance levels respecively.

17 6. Conclusion The moneary model of exchange rae for developed economies has received much aenion from researchers. Sudies using daa from emerging economies are relaively limied. Long and Samreh (008) resusciae he sudy of moneary model in he conex of he Philippines, an emerging economy. In he spiri of Long and Samreh (008) and ohers, his sudy examines he validiy of he flexible-price moneary model for he case of Thailand using he Johansen mulivariae coinegraion esing framework. Several resricions are imposed o he esimaed moneary model for furher analysis. The key findings and implicaions of his sudy include: Firs, here exiss of wo coinegraing vecors in he esimaed VAR model, indicaing he presence of long-run relaionship among exchange rae and he moneary variables for Thailand. Moreover, i is observed ha he forecass of exchange rae generaed by he fied model end o mimic he acual daa closely. Therefore, exchange rae players may monior and forecas he fuure exchange rae movemen via he money supplies, incomes, and ineres raes variables of boh Thailand and Japan. Besides, he finding indicaes ha one has o follow he economic developmen of Japan, o undersanding he movemen of exchange rae for Thailand, an emerging economy. This finding adds new insighs o accompanimen he majoriy of previous sudies ha documened he imporan influence of US in he emerging Asian economies. Second, i has been shown in his sudy ha he oucomes of moneary resricions ess are sensiive o he number of coinegraing equaions enered in he esimaion. Hence, i is imporan o appropriaely idenify he correc number of coinegraing equaions.

18 In his regards, his sudy adops he Panula principle and he race saisic o conclude ha here are wo coinegraing vecors in he exchange rae under invesigaion. Third, he proporionaliy of exchange rae and money differenial canno be rejeced by he resricion es. This implies ha money is found o be neural in Thailand, in he sense ha i influences he nominal exchange rae proporionaely bu he real exchange rae remains unalered. Fourh, equal and opposie effecs of income differenial on he exchange rae is also found in his sudy, indicaing ha foreign and domesic economic growhs are imporan deerminans ha canno be excluded from he moneary model. Fifh, ineres raes variables also have equal and opposie impac on he movemen of exchange rae. This finding signifies ha influences of domesic and foreign moneary policy on exchange rae of Thailand can hardly be negleced. This sudy conribues o he lieraure by providing empirical evidence supporive of he flexible-price moneary model from Thailand, a small and open emerging economy which has no received much aenion in he exchange rae lieraure before. This sudy also offers overview of Thailand exchange rae regimes. Noe ha, his sudy fails o rejec he null hypoheses of major moneary resricions and his finding is conrasing o majoriy of he findings documened in he lieraure (see, for insance, MacDonald and Taylor, 1991; Long and Samreh, 008). While his finding implies he validiy of various underlying assumpions of he moneary approach o exchange rae deerminaion, furher researches may be conduced o

19 explain why hese assumpions are mainain in he case of Thailand exchange rae bu no in ohers. References Abbo, A. and De Via, G. (00), Tesing long-run srucural validiy of he moneary exchange rae model, Economic Leers,75, Agbola, F.W. and Kunanoppara, C. (005) Deerminans of exchange rae pracices: some empirical evidence from Thailand, Applied Economics, 7, ASEAN-Japan Cenre (007) Japan-ASEAN and China s Major Trading Parners. Available online a: hp:// Accessed on December 5, 008. Baharumshah, A.Z. Mohd, S.H. and Ahn, S.K. (009) On he predicabiliy of he moneary model: he case of he Malaysian ringgi/us dollar, Applied Economics, 41, Cheung, Y.-W. and Lai, K.S. (199) A fracional coinegraion analysis of purchasing power pariy, Journal of Business and Economic Saisics, 11, Choudhry, T. and Lawler, P. (1997) The moneary model of exchange raes: evidence from he Canadian floa of he 1950s, Journal of Macroeconomics, 19, Du, S.D. and Ghosh, D. (000) An empirical noe on he moneary exchange rae model, Applied Economics Leers, 7, Engle, R.F. and Granger, C.W.J. (1987) Coinegraion and error correcion represenaion, esimaion, and esing, Economerica, 55, Frenkel J. A. (1976) A moneary approach o he exchange rae: docrinal aspecs and empirical evidence, Scandinavian Journal of Economics, 78, IMF (006). Classificaion of Exchange Rae Arrangemens and Moneary Frameworks. Inernaional Moneary Fund. Available online a hp:// [Accessed on December, 16, 008]. IMF (008). Classificaion of Exchange Rae Arrangemens and Moneary Frameworks. Inernaional Moneary Fund. Available online a hp:// [Accessed on July 1, 009]. Johansen, S. (1988) Saisical analysis and coinegraing vecors, Journal of Dynamic and Conrol, 1, Johansen, S. (1989) Esimaion and hypohesis esing of coinegraion vecors in gaussian vecor auorregresive models, Economerica, 59,

20 Johansen, S. (1991) Esimaion and hypohesis esing of coinegraion vecors in Gaussian vecor auoregressive models, Economerica, 59, Johansen, S. (1995) Likelihood-based inference in coinegraed vecor auoregressive models. Oxford: Oxford Universiy Press. Johansen, S. and Juselius, K. (1990) Maximum likelihood esimaion and inference on coinegraion, wih applicaions o he demand for money, Oxford Bullein of Economics and Saisics, 5, Lee, C., Azali, M., Zulkornain, B., and M. B. Yusoff (007) The moneary model of exchange rae: evidence from he Philippines, Applied Economics Leers, 14, Liew, V.K.S. (009) Linear and nonlinear moneary approaches o he exchange rae of he philippines peso-japanese yen. Economics Bullein, 9(), Long, D. and Samreh, S. (008) The moneary model of exchange rae: evidence from he Philippines using ARDL approach, Economics Bullein, 6, 1-1. MacDonald, R. and Taylor, M.P. (1991) The moneary approach o he exchange rae: long-run relaionships and coefficien resricions, Economics Leers, 7, MacDonald, R. and Taylor, M.P. (1994) Re-examining he moneary approach o he exchange rae: he dollar-franc, , Applied Financial Economics, 4, Miyakoshi, T. (000) The moneary approach o he exchange rae: empirical observaions from Korea, Applied Economics Leers, 7, Öserholm, P. (00) Tesing for coinegraion in misspecified sysems A mone carlo sudy of size disorions, Working Paper Series, 00:1, Uppsala Universiy, Deparmen of Economics. Phillips, P.C.B. and Perron, P. (1988) Tesing for a uni roo in ime series regression, Biomerika, 75, Rapach, D.E. and Wohar, M.E. (00) Tesing he moneary model of exchange rae deerminaion: new evidence from a cenury daa, Journal of Inernaional Economics, 58,