Is Malaysia exchange rate misalignment before the 1997 crisis?

Transcription

1 MPRA Munch Personal RePEc Archve Is Malaysa exchange rate msalgnment before the 1997 crss? Chn Lee and Azal M. and Zulkornan Yusop and Mohammed Yusoff Department of Economcs, Faculty of Economcs and Management, Unverst Putra Malaysa 2008 Onlne at MPRA Paper No , posted 2. August :50 UTC

2 IS MALAYSIA EXCHANGE RATE MISALIGNMENT BEFORE THE 1997 CRISIS? Lee Chn a*, M. Azal a, Zulkornan B. Yusop a, Mohammed B. Yusoff b a Department of Economcs, Unverst Putra Malaysa, UPM, Serdang, Selangor Darul Ehsan, Malaysa b Kullyah of Economcs and Management Scences, Internatonal Islamc Unversty Malaysa, Jalan Gombak, Kuala Lumpur Abstract Ths paper seeks to use the flexble-prce monetary model n the contegraton and vector error correcton model (VECM) contexts to determne whether there was msalgnment n the Malaysan rnggt - U.S. dollar before the 1997 currency crss. Unt roots, contegraton and weak exogenety are tested to valdate the monetary exchange rate model. Generally, t s found that all the seres are I(1) process and there exsts sgnfcant contegratng vectors. Usng the contegratng vector and the fnal parsmonous VECM, out of sample predctons for Rnggt exchange rate are generated. The resultng resduals between the actual and the ftted values of exchange rate are the estmated msalgnments. From contegraton, our results suggest that the Malaysan rnggt was overvalued from 1995Q2-1997Q2. Based on VECM, our results suggest that rnggt was overvalued from 1995Q2-1996Q2 and slghtly undervalued from 1996Q3-1997Q2. Keywords: Monetary Model, Exchange Rate, Msalgnment, Malaysa. * Correspondng author. Lee Chn; Department of Economcs, Unverst Putra Malaysa, UPM, Serdang, Selangor Darul Ehsan, Malaysa; Tel: ; Fax: ; e-mal: leechn@putra.upm.edu.my

3 1. Introducton The most mportant event n the world economy n 1997 perhaps n the decade was fnancal crss that beseged much of the Asan countres. The crss began n Thaland n July 1997 and t quckly splled over to engulf Indonesa, Korea and Malaysa by the end of The stock prces of these countres fall sharply together wth the massve deprecaton of ther currences. The outcome was that many Asan economes experenced drastc slowdown n ther economc growth and a loss of confdence by foregn nvestors. The speed and severty of the crss took everybody by surprse. One of the key questons surroundng the 1997 Asan crses s what caused the fnancal crss? A number of papers have ponted to exchange rate overvaluaton as a promnent determnant of currency crses (Frankel and Rose, 1996; Sachs et al., 1996; Kamnsky and Renhart, 1999; Goldfajn and Valdes, 1999; and Chnn, 2000). Hence, the presence of overvaluaton s potentally mportant for polcy purposes because of ts role as a component of an early warnng system (see e.g. Berg et al., 2000). Ths study examnes the degree to whch the Malaysan rnggt - U.S. dollar exchange rate s msalgned by usng a theoretcal baselne model. Frstly, we formulate and test the valdty of the monetary exchange rate model for Malaysa usng the contegraton and vector-error correcton technques. Then the models obtaned are used to determne the exchange rate 1

4 for the country before the currency crss to see whether there s any currency msalgnment. Ths paper extends the exstng lteratures n three drectons. Frst, ths paper adds to the pool of very few studes that have nvestgated whether the rnggt s msalgnment before the 1997 fnancal crss. Those have came to our notce are Furman and Stglz (1998), Sazanam and Yoshmura (1999), Chnn (2000), Kwek and Yoong (2002), Lee and Azal (2005) and Husted and MacDonald (1999). Second, we advance the above mentoned lteratures n some ways. Compared to Furman and Stglz (1998), Sazanam and Yoshmura (1999), and Chnn (2000) who used PPP to measure the exchange rate msalgnment, we use a more structural model. The unrealstc assumpton underlyng the PPP based approach s that t assumes an unchanged equlbrum real exchange rate throughout the perod. We employed a theoretcal baselne model whch ncorporates the changes n economy fundamentals to estmate equlbrum exchange rate and to derve exchange rate msalgnment. Although we employed the same monetary exchange rate model as Husted and MacDonald (1999) and Lee and Azal (2005), our study s dfferent from thers n the followng aspects. Husted and MacDonald (1999) measured long-run exchange rate msalgnment whle Lee and Azal (2005) derved short-run msalgnment from VECM monetary model, but we estmate both long-run and short-run msalgnments. In addton, we had observed that most of the estmated 2

5 coeffcents of monetary model n the study of Lee and Azal (2005) carred an unexpected sgn. Ths mght due to the model s estmated under a restrcted form; hence we relaxed the assumptons of equalty n money and ncome elastctes across countres and tested the monetary model n an unrestrcted form. Thrd, n the lterature, the monetary model s wdely used for testng the valdty of the approach and ts out-of-sample forecastng performance. However, ths model s not wdely used for assessment purposes. Notable exceptons are Husted and MacDonald (1999), Chnn (2000), Cvcr (2004) and La Cour and MacDonald (2000). The structure of the paper s as follows: Secton 2 dscusses the concept of exchange rate msalgnment, undervaluaton and overvaluaton. Secton 3 revews the emprcal studes of estmatng exchange rate msalgnment. Secton 4 outlnes the monetary exchange rate model to estmate the exchange rate behavor and then to derve exchange rate msalgnment. Secton 5 descrbes the methodology and data set used. Secton 6 presents the emprcal results. Concludng remarks appear n secton Msalgnment, Undervaluaton and Overvaluaton To understand the concept of exchange rate msalgnment and the equlbrum exchange rate, one must begn wth a smple defnton of the equlbrum 3

6 exchange rate. Conventonally, the long-run equlbrum exchange rate s defned as the exchange rate level that, for a gven set of structural fundamentals s compatble wth smultaneous achevement of nternal and external balances of economy 1. And the exchange rate msalgnment s defned as a persstent departure of the exchange rate from ts long-run equlbrum level 2. Hence, the lterature on exchange rate msalgnment has been mostly confned to estmaton from long-run parameter. However, wth the development of VECM, some recent studes had employed VECM to estmate exchange rate msalgnment 3. Regardless of the specfc approach, to measure exchange rate msalgnment, frst, the equlbrum exchange rate must be ascertaned. Then, the absolute dfference between the actual spot exchange rate and the estmated equlbrum exchange rate s the estmated msalgnment. For ths study, we quantfy equlbrum value usng avalable tme seres data on the varables consttutng the fundamentals that underpn the exchange rate. The equlbrum exchange rate s modelled as beng dependent on these fundamentals and any devaton of the actual from the predcted value s nterpreted as msalgnment. If the value of actual spot exchange rate s above the value of equlbrum exchange rate, t s called exchange rate overvaluaton. Whle exchange rate undervaluaton descrbes the stuaton where the market value s below equlbrum rate. 1 See Edwards (2001). 2 See Wllamson (1985). 4

7 3. Emprcal Studes on Exchange Rate Msalgnment A number of alternatve methods have been used to calculate an equlbrum exchange rate. A comprehensve and detaled survey s offered n ÉGert et al. (2006) 4. Nevertheless, two man approaches that are most popular found n the emprcal lteratures for measurng currency msalgnment for developng countres are the prce based purchasng power party (PPP) approach and the model based approach 5. The frst approach s based on a smple no-arbtrage condton. Ths approach uses devatons of the actual real exchange rate from some base year n whch the real exchange rate s beleved to be n equlbrum to calculate msalgnment. Ths approach s relatvely easy to mplement, but does not address the economcally nterestng queston of whether a partcular exchange rate s at an optmal level, besdes that defned by a no-arbtrage condton (Chnn, 2000). In addton, nadequate consderaton s gven to changes n the equlbrum real exchange rate caused by fundamentals because ths approach assumes an 3 For nstance, Chnn (2000); Kemme and Teng (2000); and Chand (2001). 4 ÉGert, Halpern and MacDonald (2006) present a crtcal overvew of the varous methods avalable for calculatng equlbrum exchange rates such as Purchasng Power Party (PPP), ts trend-adjusted varants, the nternal external approach and ts varants [the Fundamental Equlbrum Exchange Rate (FEER), the Macroeconomc Balance Approach and the Natural Real Exchange Rate (NATREX)], the Behavoural Equlbrum Exchange Rate (BEER), the Permanent Equlbrum Exchange Rate (PEER), the Captal Enhanced Equlbrum Exchange Rate (CHEER) and the New Open Economy Macroeconomcs (NOEM) approach to the determnaton of the equlbrum exchange rate. 5

8 unchanged equlbrum real exchange rate throughout the perod (Zhang, 2001) and ssues on the choce of a relevant prce ndex and a proper base year reman. On the other hand, the second approach uses a formal model for determnng the equlbrum real exchange rate. Its prncpal advantage s the capablty of ncorporatng changes n the equlbrum real exchange rate (Zhang, 2001). The man contrbuton of ths approach s to capture explctly economc factors n estmatng equlbrum real exchange rate. Prce based PPP model have been employed by Furman and Stglz (1998), Sazanam and Yoshmura (1999) and Chnn (2000) to estmate the rnggt exchange rate msalgnment before the 1997 currency crss. Employng monthly data from January 1989 to December 1991 n long-run averagng stylzed facts base perod (where average real exchange rate over as the base year), Furman and Stglz (1998) found that Malaysa rnggt were overvalued from January to June On the other hand, Sazanam and Yoshmura (1999) used mean revertng as base perod to measure currences overvaluaton. Usng monthly data spannng from 1986 January to 1996 December, they found that RM/USD and RM/yen were overvalued on May Chnn (2000) estmated Malaysan currences overvaluaton respectve to USD and yen usng long-run averagng PPP model. They tested the model usng deflated producer prce ndces (PPI) and deflated consumer prce ndces (CPI) 5 For examples: Furman and Stglz (1998); Sazanam and Yoshmura (1999); Baffes et. al. (1999); Husted and MacDonald (1999); Chnn and Dooley (1999); Chnn (2000); Lm (2000); 6

9 over the data of January 1978 to December Both models provde consstent results of overvaluaton for RM/USD and RM/yen at May Lee and Azal (2005), Husted and MacDonald (1999) and Kwek and Yoong (2002) used a formal model to determne the equlbrum exchange rate for Malaysa before the currency crss. Lee and Azal (2005) used a restrcted verson of stcky-prce monetary model n the envronment of vector errorcorrecton to estmate the short run RM/USD exchange rate msalgnment before the currency crss. Frst, the authors formulated the exchange rate model for Malaysa by usng quarterly data of 1980Q1 to 1995Q1. The model was checked for the dagnostcs tests as well as n sample and out of sample forecastng performances. Next, usng the model obtaned, the equlbrum RM/USD exchange rates are generated. Fnally, the exchange rate msalgnments were derved from the resduals between the actual and the predcted equlbrum exchange rates. Ther results showed that the Malaysan rnggt was slghtly overvalued n the second quarter of However, t s statstcally nsgnfcant. Instead of RM/USD, Husted and MacDonald (1999) employed monetary model n panel OLS to examne the RM/yen exchange rates msalgnments before the crses. The unrestrcted verson of flexble prce monetary exchange rate model together wth annual data rangng from s used to produce estmates of equlbrum exchange rates and a plot of equlbrum or ftted values derved from the OLS fxed effects panel estmates Zhang (2001); Kakkar (2001); and Saxena (2002). 7

10 of the monetary model was compared wth the actual values. The results suggestng the RM/yen was overvalued at end of Kwek and Yoong (2002) assessed the RM/USD exchange rate valuaton by employng an equlbrum exchange rate model. Quarterly data from 1991Q1 to 2001Q1 s used to estmate the equlbrum exchange rate model. Ther results showed that the RM/USD was undervalued by 11% n 1997Q3. For the perod just pror to 1998Q3, the real exchange rate s lower than the equlbrum real exchange rate, ths suggests that rnggt Malaysa s pegged at an overvalued level as compared to equlbrum rate. However after 1999Q3, the rnggt Malaysa started to be undervalued after the peggng at RM3.80 to USD. Nevertheless, they concluded that the RM/USD exchange rate has not been msalgned wth excessve overvaluaton or undervaluaton even after the government s polcy on peggng of RM/USD. Except for Kwek and Yoong (2002), t had been observed that all of the above menton emprcal studes both PPP and model based approach found that the rnggt was overvalued on the eve of the currency crss. 4. Exchange Rate Model The monetary model of exchange rate determnaton serves as the basc construct for equlbrum nomnal exchange rate n varety of macroeconomc models 6. All monetary models rely on the twn assumptons of purchasng 6 See, for nstance, Frenkel (1976), Blson (1978), Hodrck (1978), Dornbusch (1976) and Frankel (1979). 8

11 power party (PPP) holds contnuously (Equaton 1) and the exstence of stable money demand functons for the domestc and foregn economes (Equatons 2 and 3): e t = p t p t * (1) m t = γ 1 p t + γ 2 y t + γ 3 r t (2) m t * = γ 1 p* t + γ 2 y* t + γ 3 r* t (3) where e t s the spot exchange rate (defned as the prce of a unt of foregn money n terms of domestc money), m t s the domestc money supply, p t s the domestc prce level, y t s the domestc real ncome, r t s the domestc nterest rate, whle an astersk denotes the correspondng foregn varables, and all varables except for nterest rate, are expressed n natural logarthms 7. Solvng Equatons (2) and (3) for the relatve prce level, and substtutng nto Equaton (1) yelds the basc flexble-prce monetary model: e t = (m t - m * t )- β (y t - y * t) + λ (r t r * t) (4) Note that Equaton (4) assumes the equalty of dentcal demand for money coeffcents across countres. 7 In ths study, home country s Malaysa whle foregn country s the Unted States. 9

12 Relaxng these assumptons, gves Equaton (4) n an unrestrcted form as: e t = α 0 m t + α 1 m * t + β 0 y t + β 1 y * t + λ 0 r t + λ 1 r * t + ε t (5) Theoretcally, α 0 and α 1 should equal 1 and 1 respectvely, β 0 and β 1 should be negatve and postve, respectvely, wth values equal to ncome elastctes from domestc and foregn money demand functons, and λ 0 and λ 1 should be postve and negatve wth values smlar to those of sem-nterest rate elastctes n money demand functons. 5. Methodology and Data As an ntal step n our analyss, the order of ntegraton for the tme seres were determned usng the standard Augmented Dckey-Fuller (ADF) and Phllps- Perron (PP) unt root tests. If the seres are of same order, then we may proceed to test the exstence of contegratng relatons among the exchange rate and ts fundamentals usng Johansen multvarate contegraton technques. The test used s the trace statstc, whch tests for at most r contegratng vectors among a system of n tme seres (where r = 0, 1, 2,..., n 2, n 1) (Johansen, 1988). If we are able to reject the null hypothess of no contegratng vectors based on the 10

13 11 trace statstc, ths ndcates the monetary model have some long-run valdty (Enders, 2004). Accordng to the Granger Representaton Theorem, f a contegratng relatonshp exsts between a seres of I(1) varables, then an error-correcton model (ECM) also exsts. An ECM s a model, whch uses the lagged resdual from the contegratng regresson n combnaton wth short-run dynamcs to adjust the model towards long-run equlbrum (Tawadros, 2001). Ths suggests that there should exst an exchange rate equaton of the form: (6) 1 0 * * * t t n t n t n t n t n t n t n t t Z r r y y m m e C e µ + Π + Γ + Γ + Γ + Γ + Γ + Γ + Γ + = = = = = = = = where C denotes a constant, µ t denotes a error term, Z t represents the contegratng vector normalzed on e t and Π-matrx captures the adjustment of the exchange rate towards ts long-run equlbrum value. Π = αβ, where α represents the speed of adjustment to dsequlbrum whle β s a matrx of longrun coeffcents such that the term β Z t embedded n equaton (6) represents up to (n - 1) contegraton relatonshps n the multvarate model whch ensure that the Z t converge to ther long-run steady-state solutons. Equaton (6) can be used to test the Granger causalty for weakly exogenety. The Granger causalty test must be conducted n the envronment of vector

14 error-correcton model (VECM) where the relevant error-correcton terms (ECTs) were ncluded n the model to avod msspecfcaton and omsson of the mportant varables. Granger causalty s testng the null hypothess that the lagged values of the ndependent varables are jontly sgnfcant n the equaton of the dependent varable. Ths can be done by runnng a VECM on the system of equatons and testng for zero restrctons on the lagged values of the ndependent varables coeffcents. Fnally, followng the Hendry s general-to-specfc methodology, the fnal parsmonous specfcaton of Equaton (6) s used to forecast the exchange rate before the currency crss. Ths fnal parsmonous specfcaton can be acheved by sequentally removng the nsgnfcant regressors from the general model f the t-statstc of the coeffcent s less that one n absolute terms. The data for ths study are collected from varous ssues of the Internatonal Monetary Fund s Internatonal Fnancal Statstcs yearbook spannng from 1973Q2 to 2003Q1. Data durng the flexble exchange rate perod and before any evdence of currency msalgnment.e. 1973Q2 to 1995Q1 were used to formulate the model, whle the data from 1995Q2 onwards were set asde for comparson and for out-sample forecastng exercses 8. Exchange rates are quarterly averages n terms of RM/USD. The chosen monetary aggregates are 8 Sazanam and Yoshmura (1999) found that the msalgnment of the East Asan currences started snce Aprl

15 broad money stock (M2) 9. The ndustral product ndces (IPI) were utlzed as proxes for domestc ncome. The nterest rates are the quarterly averages of three-month treasury bll rates. 6. Emprcal Results The order of ntegraton of the seres was determned usng the standard Augmented Dckey-Fuller (ADF) and Phllps-Perron (PP) unt root tests. Table 1 reports that all of the seres are I(1) processes 10. Table 1 The Johansen-Juselus lkelhood contegraton test was reported n Table 2. The result ndcates that the null hypothess of one contegratng vector s rejected usng 5% crtcal value. Ths mples that the varables n ths model are contegrated wth two contegratng vectors. Our fndng of at least one contegratng vector ndcates that the monetary model would seem to have some long-run valdty. Table 2 9 Data for monetary aggregates are seasonally adjusted. 10 Except for two cases,.e. constant wth trend PP test shows that the exchange rate s level statonary and constant wth trend ADF test shows that the foregn nterest rate s nonstatonary at both levels. We treat them as I(1) process snce the other three tests show that they are I(1). 13

16 Havng determned how many contegratng vectors there are, t s necessary to consder whether these are unque and consequently whether they tell us anythng about the structural economc relatonshps underlyng the long run model. The estmated contegratng vectors are gven economc meanng by means of normalzng. The vector that makes economc sense s that the estmated coeffcents are close to and have the same sgns as those predcted by economc theory. The values of the coeffcents n these two contegratng vectors are reported n Table 3. The upper panel shows the coeffcents n the estmated contegratng vectors, whch normalzed on the exchange rate and the lower panel shows the results of some popular homogenety restrctons on the monetary model. The contegratng vectors seem represent the monetary model where eght out of twelve coeffcents carred the expected sgns. In partcular, the long-run parameters for domestc money and ncome, as well as foregn nterest rate are correctly sgned and consstent n both vectors. These suggested that an ncrease n Malaysan money supply nduces a deprecaton to the rnggt; rapd domestc growth wll strengthen the rnggt; and a rse n US nterest rate leads to deprecaton of rnggt. Our fndng of contegraton allows us to proceed to test some of the popular mposed monetary restrctons. The estmated values of the χ 2 statstcs, whch test the equalty n money, ncome and nterest rate elastctes across countres, are reported n the lower panel of Table 3. The statstcs results rejected the homogenety n money and nterest 14

17 rate elastctes across countres. Only the restrcton of dentcal ncome elastcty cannot be rejected. These suggested that we should proceed wth unrestrcted verson of monetary exchange rate model snce most of the restrctons do not hold. Table 3 After obtanng the long-run contegraton relatons usng the Johansen approach, we can estmate the short-run behavour n error correcton form wth the contegraton relatonshps beng ncluded. Table 4 reports the summary results of Granger-causalty test n the envronment of VECM 11. The results clearly ndcate that the domestc money stock, domestc ncome and domestc nterest rate are weakly exogenous. The results also show that the foregn money stock, foregn ncome and foregn nterest rate granger cause the Malaysan exchange rate. Ths s reasonable snce Malaysan s external sector s closely related to the US. Table 4 Followng the general to specfc approach to modellng, parsmonous errorcorrecton regresson s obtaned by removng the nsgnfcant regressors 12. The 11 The result of F-statstcs for the Granger-causalty test s reported n Appendx. 12 The full estmates of the regressors are avalable upon request from the authors. 15

18 fnal parsmonous estmated error-correcton regresson for flexble prce monetary model s: e t = ( ) ( ) (2.222 ) ( ) * ECT ECT m t 1 * y t rt ( ) (2.111) * 2 2 (7) where ECT s normalzed contegratng equaton beng ncluded n errorcorrecton form. All the varables (except for nterest rate) are expressed n natural logarthms and the values n parentheses below the coeffcents are the t- values. It shows that all the coeffcents are statstcally sgnfcant. Equaton (7) shows the error-correcton terms are statstcally sgnfcant and negatve, except the second vector 13. The speed of adjustment coeffcent suggests that approxmately 18% of the change n the exchange rate per quarter can be attrbuted to the dsequlbrum between actual and equlbrum levels. It also shows that changes n some of the lagged varables have sgnfcant short-run effects on exchange rate. The foregn money enters n wth a negatve sgn, ndcatng that an ncrease n the US money supply lead to an apprecaton to the rnggt. Sgnfcant negatve foregn ncome varable mples that the US economc growth tends to strengthen the RM/USD rate. And postve foregn nterest rate coeffcent suggests that US nterest rate rse nduces rnggt to deprecate. 16

19 Usng the contegratng vector and Equaton (7), out of sample predctons of rnggt exchange rate for the perod of 1995Q2 to 1997Q2 are generated. The resultng resduals between the actual and the ftted values of exchange rate are the estmated msalgnment (Table 5 and Fgure 1). The resduals represent the devatons from short-run and long-run equlbrum. The short-run msalgnments are expected to be smaller than the long-run msalgnments snce there are opportuntes for adjustments n exchange rate through short-run dynamcs. From contegraton, our results suggest that the RM/USD was overvalued from 1995Q2-1997Q2, n partcular the overvaluaton was persstently hgh at 13% 18% over the perod 1996Q2-1997Q2. Based on VECM, our results suggest that RM/USD was overvalued snce 1995Q2, however, t turns to become slghtly undervalued after 1996Q3. The turnng sgn n short-run msalgnment mght due to overshootng of self-correcton mechansm. As expeceted the short-run msalgnments derved from VECM are smaller than the long-run msalgnments of contegraton. Table 5 Fgure 1 13 ECT2 s postve, however, ts magntude s relatvely small. 17

20 7. Concludng Remarks Although there has been much emprcal work on the valdty of the monetary model, there have been only a few that utlzed ths model for other purposes. Ths paper s sought to use the flexble-prce monetary model n the contegraton and VECM contexts to determne whether the Malaysan rnggt - U.S. dollar exchange rate was msalgned before the currency crss. The estmates from contegratng vector suggest that the Malaysan rnggt was substantally overvalued on the eve of the fnancal crss. And the results of VECM ndcate that the Malaysan rnggt was overvalued from 1995Q2 1996Q2. Thus, evdence do support that the exchange rate overvaluaton mght be one of the causes contrbuted to the 1997 fnancal crss. The relatonshps among exchange rate and the economc fundamentals as depcted n our model may provde some nsghts to the deprecaton of rnggt. The estmated long-run parameters of monetary exchange rate model ndcated that an ncrease n Malaysan money supply and a reducton n domestc ncome lead to a deprecaton of rnggt. Malaysan expansonary monetary polcy durng the 1990s may contrbute to the weakenng of rnggt. Malaysa has been mantanng hgh monetary growth n response to a decade of rapd economc growth and to mnmse the dsruptons of captal nflows n Although Malaysa had experence rapd economc growth before the currency 18

21 crss, the export growth was declne sharply n After recordng strong export growth of 26% n 1995, Malaysan export growth falls to 7% n 1996 due to the downturn n the global electronc ndustry, rsng wage costs and competton posed by the low-wage countres. Hence, the demand for rnggt n nternatonal market may tampered by the sharp declne n exports. In addton, the Granger-causalty tests also show that foregn money stock, ncome and nterest rate granger cause Malaysan exchange rate n the short-run. Ths s reasonable snce Malaysan s external sector s closely related to the US. References Baffes, J.; Elbadaw, I. A. and O Connell, S. A. (1999) Sngle estmaton of the equlbrum real exchange rate. In L. E. Hnkle and P. J. Montel (Eds.). Exchange rate msalgnment: Concepts and measurement for developng countres. Washngton: Oxford Unversty Press, Inc. Berg, A., Borenszten, E., Mles-Ferrett, G.M., Pattllo, C. (2000) Antcpatng balance of payments crses the role of early warnng systems. IMF Occasonal Paper 186. IMF, Washngton, DC. Blson, J. F. O. (1978) Ratonal expectatons and the exchange rate. In J. A. Frankel and H. G. Johnson (Eds.). The economcs of exchange rates. Mass: Addson-Wesley. Chand, S. (2001) How msalgned s the Australan real exchange rate? Techncal report economcs of development workng papers. APSEG: The Australan Natonal Unversty. Chnn. M. D. (2000) Before the fall: Were East Asan currences overvalued? Emergng Market Revew, 1,

22 Chnn, M. D. and Dooley, M. P. (1999) Internatonal monetary arrangements n the Asa-Pacfc before and after. Journal of Asan Economcs, 10, Cvcr, I. (2004) The long-run valdty of monetary exchange rate model for a hgh nflaton country and msalgnment: The case of Turkey. Emergng Markets Fnance and Trade, 40, Dckey, D. A and Fuller, W. A. (1979) Dstrbuton of the estmators for autoregressve tme seres wth a unt root. Journal of Amercan Statstcal Assocaton, 74, Dckey, D. A. and Fuller, W. A. (1981) Lkelhood rato statstcs for autoregressve tme seres wth a unt root. Econometrca, 49, Dornbusch, R. (1976) Expectatons and exchange rate dynamcs. Journal of Poltcal Economy, 84, Edwards, S. (2001) Exchange rate regmes, captal flows and crss preventon. NBER Workng Paper No ÉGert, B.; Halpern, L. and MacDonald, R. (2006) Equlbrum exchange rates n transton economes: Takng stock of the ssues. Journal of Economc Surveys, 20, Enders, W. (2004) Appled econometrc tme seres. (2nd ed.). Unted States of Amerca: John Wley and Sons. Engle, R. F. and Granger, C. W. J. (1987) Contegraton and error correcton: Representaton, estmaton, and testng. Econometrca, 55, Frankel, J. A. (1979) On the Mark: A theory of floatng exchange rates based on real nterest dfferental. Amercan Economc Revew, 69, Frankel, J.A., Rose, A.K. (1996) Currency crashes n emergng markets: An emprcal treatment. Journal of Internatonal Economcs, 41, Frenkel, J. A. (1976) A monetary approach to the exchange rate: Doctrnal aspects and emprcal evdence. In R. MacDonald and M. P. Taylor (Eds.). Exchange rate economcs Volume I. Cambrdge: Cambrdge Unversty Press. 20

23 Furman, J. and Stglz, J. (1998) Economc crses: Evdence and nsghts from East Asa. Brookngs Papers on Economc Actvty, 2, Goldfajn, H. and Valdes, R. (1999) The aftermath of apprecatons. Quarterly Journal of Economcs, 114, Hodrck, R. J. (1978) An emprcal analyss of the monetary approach to the determnaton of the exchange rate. In R. MacDonald and M. P. Taylor (Eds.), Exchange Rate Economcs Volume I. Cambrdge: Cambrdge Unversty Press. Husted, S. and MacDonald, R. (1999) The Asan currency crash: Were badly drven fundamentals to blame? Journal of Asan Economcs, 10, Internatonal Monetary Fund (varous ssues). Internatonal Monetary Fund s Internatonal Fnancal Statstcs Year Book. Washngton D.C.: IMF. Johansen, S. (1988) Statstcal analyss of contegraton vectors. Journal of Economc Dynamcs and Control, 12, Johansen, S. (1991) Estmaton and hypothess testng of contegraton vectors n Gaussan vector autoregressve models. Econometrca, 59, Johansen, S. and Juselus, K. (1990) Maxmum lkelhood estmaton and nference on contegraton, wth applcatons to the demand for money. Oxford Bulletn of Economcs and Statstcs, 52, Kakkar, V. (2001) Long run real exchange rates: Evdence from Mexco. Economcs Letters, 72, Kamnsky, G. and Renhart, C. (1999) Currency and bankng crses: The early warnngs of dstress. Amercan Economc Revew, 89, Kemme, D. M. and Teng, W. Y. (2000) Determnants of the real exchange rate, msalgnment and mplcatons for growth n Poland. Economc Systems, 24, Kwek, K. T. and Yoong, F. Y. (2002) Is Rnggt Malaysa over-valued aganst the USD? Paper presented at 8 th Conventon of the East Asan Economc Assocaton, 4-5 November 2002, Kuala Lumpur. 21

24 La Cour, L. and MacDonald, R. (2000) Modelng the ECU aganst the U.S. Dollar: A structural monetary nterpretaton. Journal of Busness and Economc Statstcs, 18, Lee, C. and Azal, M. (2005) Exchange rate msalgnments n ASEAN-5 countres. Labuan Bulletn of Internatonal Busness & Fnance, 3, Lm, G. C. (2000) Msalgnment and managed exchange rates: An applcaton to the Tha Baht. IMF Workng Paper (WP/00/63). Internatonal Monetary Fund. Phllps, P. C. B. and Perron, P. (1988) Testng for a unt root n tme seres regresson. Bometrka, 75, Sachs, J., Tornell, A., Velasco, A. (1996) Fnancal crses n emergng markets: the lessons from Brookngs Papers on Economc Actvty, 1, Saxena, S. C. (2002) Exchange rate dynamcs n Indonesa: Journal of Asan Economcs, 13, Sazanam, Y. and Yoshmura, S. (1999) Restructurng East Asan exchange rate regmes. Journal of Asan Economcs, 10, Tawadros, G. B. (2001) The predctve power of the monetary model of exchange rate determnaton. Appled Fnancal Economcs, 11, Wllamson, J. (1985) The Exchange Rate System. Washngton, DC: Insttute for Internatonal Economcs. Zhang, Z. C. (2001) Real exchange rate msalgnment n Chna: An emprcal nvestgaton. Journal of Comparatve Economcs, 29,

25 Table 1: Unt Root Tests Augmented Dckey-Fuller Phllps-Perron constant wthout trend constant wth trend constant wthout trend constant wth trend Level Frst Dfference Level Frst Dfference Level Frst Dfference Level Frst Dfference e *** *** *** -3.51** *** m ** * *** *** m* *** ** *** *** y *** *** *** *** y* *** *** *** *** r *** ** *** *** r* * *** *** Notes: Fgures are the t-statstcs for testng the null hypothess of ADF (PP) that the seres s nonstatonary (statonary). ***, ** and * denotes sgnfcance at 1%, 5% and 10% level, respectvely. For constant wthout trend, the crtcal values for rejecton are -3.50, and at 1%, 5% and 10%, respectvely. For constant wth trend, the crtcal values for rejecton are 4.07, and 3.16 at 1%, 5% and 10%, respectvely. Lag length used n all seres s 4. 23

26 Table 2: Johansen-Juselus Lkelhood Contegraton Test Null Hypotheses Maxmum Egenvalue Trace Crtcal Value (5%) Crtcal Value (1%) (r = 0)*** (r 1)** (r 2) (r 3) (r 4) (r 5) (r 6) Notes: r ndcates the number of contegratng vectors. *** and ** denotes rejecton of the hypothess at 1% and 5% crtcal value. Seasonal dummes are not ncluded n ths test snce they had been dropped n the parsmonous model although they had been consdered n the prelmnary analyses. 24

27 Table 3: Normalzed Contegratng Vectors and Restrctons Test Normalzed contegratng vectors: m m* y y* r r* (Predcted sgn) (+) (-) (-) (+) (+) (-) Test of homogenety restrctons: m =-m* = 1 y = -y* r = -r* χ 2 (p-value) (0.008) (0.936) (0.052) Notes: The upper panel shows the coeffcents n the estmated contegratng vectors, whch normalzed on the exchange rate. Coeffcents n shade ndcate correctly sgned. The lower panel shows the results of mpostons of varous homogenety restrctons on the monetary model. 25

28 Table 4: Summary of the of Granger Causalty Results based on VECM From varable Drecton of causalty To varable Remarks m > e No causal relatonshp y > e No causal relatonshp r > e No causal relatonshp m* => e The presence of causal relatonshp y* => e The presence of causal relatonshp r* => e The presence of causal relatonshp 26

29 Table 5: Malaysan Exchange Rate Msalgnment Perod Actual Values Ftted Values Msalgnment (%) COINT VECM COINT VECM 1995: : : : : : : : : Notes: Fgures are n log. Ftted values are estmated from contegratng vector (COINT) and VECM. Msalgnment s the resdual between actual and ftted values of exchange rate. Postve (negatve) value for resdual denotes an undervaluaton (overvaluaton) of the RM/USD. 27

30 1.2 Fgure 1: Actual and Ftted Values of Malaysan Exchange Rate Log of Exchange Rate (RM/USD) :2 95:3 95:4 96:1 96:2 96:3 96:4 97:1 97:2 Year Actual Value Ftted Value from Contegraton Ftted Value from VECM 28

31 Appendx Table A1: Granger Causalty Results based on VECM Dependent Independent Varable Varable e m m* y y* r r* e ** *** * m m* 4.115** * y y* *** r r* *** Notes: Fgures are the F-statstcs for testng the null hypothess that the jont sgnfcance of the lagged values of the ndependent varables n the equaton of the dependent varable. ***, ** and * denotes sgnfcance at 1%,5% and 10% level, respectvely. All varables are n ther frst-dfferences. 29