The Demand for Money in Asia: Some Further Evidence

Transcription

1 Inernaonal Journal of Economcs and Fnance; Vol. 4, No. 8; 1 ISSN X E-ISSN Publshed by Canadan Cener of Scence and Educaon The Demand for Money n Asa: Some Furher Evdence Augusne C. Arze 1 & Kseok Nam 1 College of Busness and Technology, Texas A&M Unversy, Commerce, Texas 749, U.S.A Syms School of Busness, Yeshva Unversy, New York, NY , U.S.A Correspondence: Augusne C. Arze, College of Busness and Technology, Texas A&M Unversy, Commerce, Texas 749, U.S.A. E-mal: Chuck_Arze@amu-commerce.edu Receved: May 1, 1 Acceped: June 4, 1 Onlne Publshed: July, 1 do:1.39/jef.v4n8p9 URL: hp://dx.do.org/1.39/jef.v4n8p9 Absrac Ths paper nvesgaes emprcally he mpac of exchange rae changes on he money demand of seven Asan counres over he quarerly perod, Esmaes of he conegrang relaons are obaned usng dfferen esmaors and he error-correcon echnque was used o oban he esmaes of he shor-run dynamcs. The major resuls show ha ncreases n he exchange rae, exer a sgnfcan posve effec upon money demand n boh he long-run and he shor-run n each of he seven counres. Furher, domesc neres raes are found o have sgnfcan negave effec on he demand for money. These effecs may resul n sgnfcan reallocaon of resources by moneary auhores and marke parcpans. Our resuls provde jusfcaons for he moneary auhorzaon o pay aenon o broad money. Keywords: money, demand, model, exchange rae, conegraon JEL: E41 1. Inroducon Durng he las wo decades, exchange-rae changes have played an mporan role n he behavor of marke parcpans n Asa wness he dscussons abou he fnancal urmol of 1997 and 1998 whch led o currency devaluaons and even, ssuance of balou packages by he Inernaonal Moneary Fund (IMF) for some counres. Also, as he housng bubble n he Uned Saes ha sared n Augus 7 resuled no a wors world-wde fnancal and economc upheaval beween 7 and 9 (Das, 1), forced Asan economes no sgnfcan deprecaons. Furher, afer he G- Summ n June 1, hese economes sared a buldup of foregn exchange reserves. They held more han enough precauonary foregn exchange reserves n order o have greaer flexbly and o mnmze currency apprecaons. A cenral queson has been he effecs of such movemens n he exchange rae on he demand for money behavor. Ths paper res o undersand emprcally he possble conrbuon of exchange-rae changes n explanng he shor-and long-run money demand behavor of he Asan economes. In convenonal money demand sudes for less-developed counres (LDCs), real money balances are usually modeled as a funcon of real domesc ncome and expeced nflaon rae; ha s, he neres rae varable s generally excluded. I s ofen clamed ha he nflaon rae should be preferred because for LDCs he choce of asse holdngs s lmed o eher real goods or money n he absence of neres bearng fnancal asses -- see, for nsance, Bahman-OsKooee and Rehman (). Ohers have argued ha he neres raes daa when avalable have shown very lle dsperson over me due o governmen conrol. In hs paper, we sugges ha falure o nclude neres raes wll lead o omed varable bas because over he pas wo decades, Asan economes have nsued connung polces of fnancal marke lberalzaon and economc reforms whch have led o greaer neres rae lberalzaon as well as varyng neres rae daa (Arze & Shwff, 1993). Furhermore, can be argued ha even f he depos raes n LDCs are se by he moneary auhores, s sll a relevan alernave asse o money ha he publc can hold (Arze, Malndreos & Grvoyanns, ). Thus, he emprcal mporance of he neres rae varable should be reaed as an emprcal maer. Concernng he mpac of exchange rae on money demand, we noe ha wh he adven of he floang exchange-rae era n 1973, a number of analyss have denfed he exchange rae as a possble omed varable n he money demand funcon of developed and developng counres; see, for example, Arango and Nadr (1979, 1981), Arze (1989), Arze and Ndubzu (199), Arze and Shwff (1993) and Bahman-OsKooee and 9

2 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Rhee (1994). Economerc heory suggess ha omsson of such a varable may bas a model owards oversang he nfluence of he ncluded varables. Also, McNown and Wallace (199:17), who examned he demand for money n he U.S. have concluded ha "nonsaonary n he demand for money can be resolved by ncluson of he exchange rae." Few emprcal sudes have nvesgaed he mpac of exchange rae on he demand for money n LDCs, and he resuls have been mxed. Lee and Chung (199) fnd ha exchange rae exers a negave and sascally sgnfcan effec on Korea's demand for money. Tan (1997) consders he demand for money n Malaysa and repors ha exchange rae has a sascally sgnfcan effec on real M1 balances and no effec on real M balances. Ibrahm (1) reexamnes Malaysa and fnds ha exchange rae has a negave effec on demand for money. Welwaa and Ekanayake (1998) sudy Sr Lanka's demand for money and fnd ha exchange rae has a sgnfcanly negave effec on real M1 balances. They faled o oban a meanngful real M relaon. Asan economes are largely que open o he res of he world and exchange-rae movemens have had a bearng upon domesc money demand; see Rajan (1) for more on hs. An emprcal sudy of he money demand funcon can provde moneary auhores wh useful nformaon for desgnng approprae sablzaon polces. The prmary purpose of hs paper s o examne he long-run relaons and shor-run dynamcs of he demand for broad money n Asan economes (Inda, Korea, Malaysa, Paksan, he Phlppnes, Sr Lanka and Thaland) over he quarerly perod 1973:1 hrough 9:4. (Noe 1) Our objecves are fourfold. The frs objecve s o check he daa for un roos. The second objecve s o deermne wheher here exss a saonary long-run relaonshp beween real money balances, domesc neres rae and exchange rae. The conegraon mehod used s he Johansen (199) mehod, whch s a Full Informaon Maxmum Lkelhood (FIML) esmaor. The choce of hs echnque follows from he Mone Carlo sudy by Haug (1996), whch shows ha he Johansen esmaor has he leas sze dsoron. The hrd objecve s o deermne, provded he long-run equlbrum exss, he sgn, magnude and sascal sgnfcance of he long-run effecs of real ncome, neres rae and exchange raes on money demand. Recen specfcaon suggesons (Noe ) have shown ha s beer o employ sngle equaon esmaor once r =1 (conegraon) has been confrmed. We repor he esmaed long-run elasces obaned afer normalzng hem on real money balances and our analyss s based on Fully Modfed Leas Squares (FIML) esmaor of Phllps and Hansen (199). The fourh objecve s o examne he shor-run dynamcs of he money demand model for each counry usng he general o-specfc paradgm (see Hendry, 1987). The feaures of hs paper whch dsngush from he oher research n hs area are: (1) he ncluson of neres rae daa raher han he nflaon rae; () he focus on he movemens of exchange rae as an mporan deermnan of demand for money; (3) he use of recen quarerly daa nsead of daa before he fnancal crses noed above; (4) evdence from more han one counry; and () esng for nonlneary n he shor-run dynamcs of he model. The res of he paper s organzed as follows. Secon descrbes he money demand model. The emprcal resuls are presened n Secon 3, and concludng remarks close he paper n Secon 4.. Model Specfcaon and Theorecal Consderaons A sandard smple money demand specfcaon s as follows: 1 m a y ; a 1 > and a < (1) where m refers o real money balances a me, y s real ncome and a me, and s domesc neres rae a me. Drawng on he emprcal leraure n hs area (Arze, 1989; Arze & Shwff, 1993) and he mplcaons for dynamc specfcaon of he possble exsence of error-correcng mechansms n he daa-generang process, we esmae a smple, convenonal model, augmened wh exchange rae as he long-run equlbrum (equaon ) and he shor-run relaonshp (equaon 3) for he (desred) real money balances: m 1 y 3 X 6 e m L (3) 1 where m s he logarhm of desred holdngs of real money balances (real M); (Noe 3) real M consss of currency ousde he banks and demand deposs plus quas-money dvded by he consumer prce ndex; (Noe 4) y s he logarhm of real GDP; s he level of domesc neres rae whch s proxed by money marke rae or lendng rae; e s he exchange rae varable; and he sochasc dsurbance s ε. ()

3 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Accordng o equaon (), real money balances are assumed o be an ncreasng funcon of real ncome (.e., real GDP) as he usual budge condons dcae; ha s, α 1 s expeced o be posve. On he oher hand, he opporuny cos of holdng money relave o fnancal asses () s expeced o yeld a negave nfluence on money demand, so α s expeced o be negave. To provde a background on how varaons n exchange raes could affec he demand for money n an open economy, we sar by nong ha foregn secor consderaons, such as varaons n exchange raes (Arango & Nadr, 1981:7) play mporan roles n deermnng domesc money demand. The currency subsuon leraure (Agenor & Khan, 1996) suggess ha porfolo shfs beween domesc and foregn money provde a role for a foregn exchange varable. (Noe ) A leas wo effecs on he demand for domesc currency wll resul from varaons n he foregn exchange rae. On he one hand, here s a wealh effec as currency deprecaon leads o ncreases n he domesc prce of foregn secures whch hen causes porfolo-adjusmen effecs; see he dscussons, for nsance, by Arango & Nadr (1979, 1981), Arze and Shwff (1993) and Bahman-Oskooee and Techaraanacha (1). Assume ha f wealh holders evaluae her asse porfolos n erms of her domesc currency, (Noe 6) a deprecaon of he domesc exchange rae ha ncreases he value of foregn asses held by domesc resdens would boos wealh (.e., a wealh-enhancng effec). Ths n urn leads o a rebalancng effec because, n order o manan a fxed share of her wealh nvesed n domesc asses, hey wll reparae par of her foregn asses o domesc asses, ncludng domesc currency (.e., rebalancng). (Noe 7) As Arango and Nadr (1981:79) have argued, s because of he rebalancng effec brough abou by changes n he exchange rae, ha an exchange rae deprecaon has a posve effec on he demand for money. Hence, exchange rae deprecaon would ncrease he demand for domesc currency. On he oher hand, varaons n he exchange rae can generae a currency subsuon effec n whch a key role s played by nvesors' expecaons. Accordng o he currency subsuon leraure, as a weak domesc currency develops expecaons for furher weakenng, asse holders wll respond by shfng some of her porfolos away from domesc currency no foregn asses. So, f an ncrease n exchange rae (.e., deprecaon) nduces a declne n money holdng by domesc resdens, he esmae of α 3 should be negave. To summarze, based on he above dscussons, an ncrease n he exchange (.e., deprecaon) could have a posve or negave effec on he demand for money; herefore, whch effec domnaes s an emprcal ssue. As s cusomary, equaon () has assumed ha n he long run, any devaon of acual (observable) real money balances from desred (unobservable) should dsappear so ha may be vewed as conegrang model (Ercsson, 1998). The basc dea of conegraon s ha wo or more nonsaonary me seres may be regarded as defnng a long-run equlbrum relaonshp f a lnear combnaon of he varables n he model s saonary (converges o an equlbrum over me). (Noe 8) Because he money demand funcon n equaon () descrbes a saonary long-run relaonshp among he varables, hs can be nerpreed o mean ha he sochasc rend n real money balances s relaed o he sochasc rends n he explanaory varables. In oher words, even f devaons from he equlbrum should occur, hey are mean- reverng (Arze, 1997). 3. Esmaon Resuls The emprcal analyss uses quarerly frequency daa for seven Asan counes, namely Inda, Korea, Malaysa, Paksan, he Phlppnes, Sr Lanka and Thaland. The daa are aken from he Inernaonal Moneary Fund s Inernaonal Fnancal Sascs (IFS) laes CD-ROM (9). The daa used are he logarhmc value of real broad money, real GDP and exchange rae. In general, he daa for he perod, 1973:1 hrough 9:4 were used for our analyses. The frs sep n esng for conegraon n a se of varables s o es for sochasc rends n he auoregressve represenaon of each ndvdual me seres usng augmened Dckey-Fuller and Johansen ess. For space consderaon, he emprcal resuls are no presened here, bu hey sugges ha all he varables n equaon () are nonsaonary. For any se of I (1) varables, Johansen (199) has developed a sysem-based conegraon procedure o es he absence or presence of long-run equlbra among he varables n equaon (). The es ulzes wo lkelhood rao (LR) es sascs, he maxmal egenvalue (λ-max), and race (Tr) o es he presence or absence of long-run equlbra beween he varables n equaon (). For he λ-max, he null hypohess of r s esed agans r+1 conegrang vecor, whereas for he race, he null hypohess s ha here are a mos r conegrang vecors and he alernave hypohess s of a general form. The presence of a sgnfcan conegrang vecor or vecors ndcaes a sable relaonshp among he relevan varables. The number of lags used n he Vecor Auoregresson (VAR) s based on evdences provded by Schwarz s creron; however, n he case of seral correlaon suffcen numbers of lags are nroduced o elmnae he seral correlaon of he resdual. For he 61

4 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 four-varable sysem, we employ he lag order of wo for Inda, Korea and Malaysa, hree for he Phlppnes, and Sr Lanka, one for Paksan, and fve for Thaland, respecvely. Table 1. Resuls from conegraon ess Maxmum Egenvalue Trace sascs H: r= r 1 r r 3 r= r 1 r r 3 H1: r=1 r= r=3 r=4 r 1 r r 3 r 4 Inda Korea Malaysa Paksan Phlppne Sr Lanka Thaland CV (%) Descrpon: The crcal values are from Table 1 of Oserwald-Lenum (199). The % shown above represens he crcal values from able of Oserwald-Lenum (199). I s used because Souh Afrca s model allows for a rend and consan n he conegraon space; see Johansen (199). Table 1 presens he conegraon es resuls. The resuls answer o wo quesons: wheher a long-run equlbrum relaonshp exss and wheher s unque. Sarng wh he λ-max es resul, he null hypohess esed s ha here s no conegrang vecor (r = ) agans he alernave of conegraon of order one (r =1). Snce all es values n column 1 of Table 1 are sgnfcanly larger han he crcal value of 7.7 a he fve per cen level, we easly rejec he null hypohess (r = ) of no conegraon. For space reasons, we do no repor he daa for comparng he null hypohess of conegraon wh r 1 o he alernave of conegraon wh r = because none of he es sascs exceeds he crcal value for hs es (.97 a he fve per cen level) so we accep he null of conegraon wh a sngle conegrang vecor. A smlar procedure apples o he race es sasc. We frs es he null hypohess of r = agans he alernave hypohess of r 1 n each counry. Snce all emprcal es sascs exceed he crcal value of 47.1 for he race sasc, we rejec he null of r = for all counres. Nex, we es r 1 agans he r alernave hypohess. Ths me, none of es sasc exceeds he crcal value of 9.68 and we hus accep he null of conegraon wh a sngle conegrang vecor. Noe ha, had we excluded he exchange rae varable from our sysem, we would no have found evdence of conegraon a he percen level. Ths fndng mples ha he effec of exchange rae s crucal o he long-run sably of he sysem. 3.1 Long-run Equlbrum Esmaes Nex, we are neresed n he long-run mpac of he hree explanaory varables (real ncome, domesc neres rae and exchange rae) on money demand. In parcular, we wan o know wheher he sgns of he esmaed coeffcens are n lne wh her predced values, wheher hese esmaes are sascally sgnfcan, and how he explanaory varables perform n relave erms. Excep he neres rae, our sem-log specfcaon of he esmaon model (see equaon ()), yeld long-run elasces n he case of real ncome and exchange rae. In Table, we show he emprcal resuls of Equaon () usng he fully modfed leas squares (FMLS) esmaor of Phllps and Hansen (Noe 9). To ncrease confdence n he esmaes obaned, we provde emprcal resuls from he dynamc leas squares (DLS) esmaor of Sock and Wason. From hese esmaes, we gaher ha our conclusons are no parcularly nfluenced by he mehod of esmaon. 6

5 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Table. Long-run Elasces and Hypohess ess Counry Inda Korea Malaysa Paksan Phlppne Sr Lanka Thaland Phllps-Hansen Esmaor Sock-Wason Esmaor (FMOLS) (DOLS) y e y e (1.39) (.6) (3.88) (11.6) (.) (1.81) (6.11) (.1) (7.) (.6) (.4) (7.11) (.4) (1.7) (1.37) (11.66) (.6) (4.46) (.41) (4.74) (7.18) (1.86) (1.78) (4.1) (7.64) (3.4) (1.97) (9.7) (.73) (3.38) (9.66) (1.43) (18.96) (9.3) (1.4) (1.79) Descrpon: The absolue -values are n he parenheses beneah each esmaed coeffcen. The crcal value a 1 percen s 1.3 and 1.67 a percen level. Focusng on he resuls obaned from he FMLS esmaor, we gaher ha he demand relaon s esmaed o be posve for real ncome (y) and exchange rae (e), bu negave for he neres rae (). These hold for all Asan counres n he sample. Furhermore, all coeffcen esmaes are sascally sgnfcan a he convenonal levels. For he domesc neres rae varable, gnored n several LDCs sudes, he equlbrum elasces (calculaed by mulplyng he mpac elasces by he mean of he varable) are -.64, -.17, -.9, -.17, -.448, -.184, and -.4 for Inda, Korea, Malaysa, Paksan, he Phlppnes, Sr Lanka and Thaland, respecvely. These resuls provde srong suppor for he heorecal predcons regardng he mpac of real ncome, neres rae and exchange rae on real money balances. As far as relave srengh of he varous deermnans s concerned, we fnd ha he magnude of he exchange-rae effec s subsanally smaller han he real ncome effec bu s generally (n absolue erms) larger han of neres-rae elascy. Ths mples ha he economc mporance of boh ncome levels and exchange rae effecs far ouweghs he mporance of neres rae varaons for money demand. Table 3A. Long-run parameers usng nflaon rae (nf). Counry Inda Korea Malaysa Paksan Phlppne Sr Lanka Thaland Phllps-Hansen Esmaor (FMOLS) y nf e (1.69) (.31) (.6) (8.96) (3.3) (-6.) (68.16) (1.6) (.37) (7.37) (1.76) (3.) (8.1) (1.49) (4.7) (.98) (.7) (6.16) (36.3) (.9) (4.9) 63

6 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Table 3B. Bahman-Oskooee and Rehman () Long-run Coeffcen Esmaes Counry Parameers Esmaes Y nf e Inda (.18) (.6) (.3) Malaysa (7.4) (1.83) (3.38) Paksan (6.86) (.31) (4.64) Phlppne (.11) (.4) (.6) Thaland (1.9) (1.63) (1.3) Descrpon: See Table for deals, however, noe ha he daa repored n Table 3b are aken from Bahaman-Oskooee and Rehman() whose sample perod s from 197:1 hrough :4. For comparson, we repor esmaes of he conegrang equaon usng he nflaon rae n place of he domesc neres rae n Panel A of Table 3. Also, n Panel B of he same able, we repor hose aken from Bahman-Oskooee and Rehman (). Whou dscussng each panel n deal, s noeworhy ha n Panel A, he nflaon rae varable has a posve sgn and sgnfcan n Korea. The same s rue for Sr Lanka. For he Phlppnes and Malaysa, he coeffcen on nflaon has he correc sgn bu he -values are generally small and sascally nsgnfcan a he percen level. Also, noe ha for Sr Lanka, our real ncome s only.19, and for Malaysa, he exchange rae s sascally non-sgnfcan a he convenonal levels. How do hese resuls compare o hose of Bahman-Oskooee and Rehman ()? As can be seen n Panel B, coeffcen on real ncome n Inda s wrongly sgned and nflaon and exchange rae are sascally nsgnfcan. In a smlar ven, he exchange rae varable s nsgnfcan n Thaland. Furher, he esmaed coeffcens on nflaon varable, alhough correcly sgned, are of an unusual magnude. Neverheless, our real ncome coeffcens are generally conssen wh hers. All n all, he resuls sugges ha he nflaon varable s no longer adequae for hese counres because her use can lead o msleadng nference. Furher, we noe ha we expermened wh he ncluson of foregn neres rae (he rae of U.S. cerfcae of depos or U.S. Treasury bll rae) n he conegraon model, ha s, esng for he exsence of conegraon among real money balances, real ncome, domesc neres rae, foregn neres rae whou much success. Before urnng o ssue of error-correcon modelng, seems pruden o check wheher he varables n Equaon () are weakly exogenous (Arze, Osang and Sloge, ). The resuls are repored n Table 4. Table 4. Resuls of weak exogeney es Counres Real Balance Real Income Ineres rae Exchange rae Inda -.4(1.81).(1.6) (4.78).(1.1) Korea -.7(3.1).1(.98) -.36(4.6).4(1.33) Malaysa -.96(.4).3(.88) -1.9(3.1).4(1.31) Paksan -.9(.61).3(.8) -1.49(1.3).7(1.3) The Phlppnes -.64(.6).4(.31) -.94(1.3).1(1.8) Sr lanka -.163(3.39).9(4.41) -6.46(3.88).(.6) Thaland -.4(1.7).1(.3) -.17(4.79).4(.1) Descrpon: The absolue -values are n parenheses and he crcal value s The daa n Table 4 sugges ha may be approprae o ulze an nsrumenal varable (IV) esmaor when esmang our error-correcon models. Ths s so because he null of weak exogeney can be rejeced n four ou of seven cases for he real ncome varable and neres rae for fve ou of seven cases. The resuls also confrm ha he real money demand varable should be consdered endogenous. 64

7 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 3. Error-correcon Model The Granger represenaon heorem (GRT) proves ha, f a congregang relaonshp exss among a se of nonsaonary seres, hen a dynamc error-correcon represenaon of he daa also exss. The mehod used o fnd hs represenaon follows he general-o-specfc paradgm. For hs purpose; he followng error-correcon model (ECM) exss for he varable n equaon (): m k EC 1 D jy j j j 1 je j jm j1, j1 j (4) where EC -1 s he error-correcon (one-lagged error) erm generaed from he FMLS conegrang esmaes, µ s he error erm, and D are counry-specfc me dummes ha accoun for observable money demand shocks oherwse no ncluded n he model. The exsence of he error-correcon erm (EC -1 ) n equaon (4) reflecs he belef ha acual real money balances do no adjus promply o her long-run deermnans; see Arze and Malndreos, (1) for a dealed dscusson. The error-correcon model resuls are summarzed n Table. The ls of nsrumenal varables consss of he consan erm, he lagged EC erm, fve lags n he dfference of he varables ncluded n he long-run soluon. In he case of Paksan, we dd no use any IV procedure. Noe ha he resuls from Sargan s es for legmacy of he nsrumen se (no repored here) suppor he valdy of he nsrumen se. 6

8 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Table. Error- Correcon Model Regresson Resuls (198:1-1:1) Inda m.1.4ec 1.34 m 4.18 y e R.38, R Korea (.9) (1.81) (4.79) (3.3) (1.9) (1.7).36, F(,13) 16.7, DW 1.93, BG F 1. (.3) m (.) (3.1) (4.9) (3.) (.63) (3.) (.) R.14.7EC m 4. y.1 y 4.. 7e.1, R Malaysa m R.49, F (6,14 ) 1.4, DW.4, BG F 1. (.)..96EC 1.16m 1.146y e. 143e 1 (8.3) (.4) (.14) (.94) (4.14) (.78) (.) (.8).33, R Paksan.3, F(7,13) 9.3, DW 1.89, BG F 1.3 (.7) m.9.9ec 1.8m.43y e. 74D1 (.938) (.61) (1.8) (.39) (1.63) (3.) (9.81) R.4, R The Phlppnes., F(6,138) 6., DW 1.84, BG F 19. (.) m..64ec D R.89, R Sr Lanka 1.777m 4.y.149y e.34D (1.11) (.6) (18.7) (.4) (.7) (.14) (4.83) (7.36) (8.8).89, F (7,16) , DW 1.97, BG F.67 (.678) m.6.163ec m m 4.4 y.6.91 e 1. 14D 4 (1.91) (3.39) (1.7) (4.9) (1.1) (.33) (.46) (4.4) R.34, R Thaland.3, F(7,114) 8., DW 1.9, BG F.9 (.4) m.7.4ec 1.84m 1.67m 4.93y.98e D (.68) (1.7) (3.) (7.79) (.) (1.81) (43.) R.9, R.94, F(7,114) , DW 1.86, BG F.96 (.43) Descrpon: The number n parenheses repor absolue -sascs. The crcal value a 1 per cen s 1.67 and 1.96 a per cen level. Tess: DW=Durbn-Wason es sasc, BG =Breusch-Godfrey es. D 1 =1 8:1,-1 n 8:3 and zero oherwse. D =1 n 1984:1, 1 n 1984:3 and zero oherwse. D 3 =1 n 8:1, -1 n 9:1 and zero oherwse. D 4 =1 n 198:1, -1 n 1989:1 and zero oherwse and D =1 n 9:3 and zero oherwse. s +. for Malaysa s y-. for The Phlppnes s y+. for Sr Langa s y+. for Thaland s ( - )-( y-. for Inda s ( -. for Korea s e -. for Paksan s e+ e -1. for Sr Lanka s e - + e -4. for Thaland s e -1 + e The regress and n Table s he change n real M balances as he dependen varable. Focusng on hese resuls, we conclude ha he shor-run mpac of any explanaory varable on money demand s posve (negave) f he sum of he sascally sgnfcan pon esmaes of he covarae n queson n equaon (4) s posve 66

9 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 (negave). For each counry, a parsmonous and sascally accepable model s obaned hrough he smplfcaon of a ffh-order ECM. The smplfcaon process nvolves deleng successvely he frs-dfferenced varable wh he lowes -raos. Consderng ha each regress and n he able s cas n he frs dfference, he emprcal resuls sugges ha he sascal f of he each model o he daa s sasfacory, as ndcaed by he value of adjused coeffcen of deermnaon, whch ranges from.3 (Sr Lanka and Malaysa) o.94 (Thaland). The sascal appropraeness of he models s furher suppored by a number of dagnosc ess. In parcular, we repor he Durbn-Wason and he Breusch-Godfrey es o es for non-ndependence of he error dsrbuon. Based on hese wo sascs, we can rejec he null of non-ndependence for all counres. Parameer sably ess reveal no serous volaon of sable parameers. For example, we expermened wh a -es of he null hypohess ha he mean of he recursve resduals for each of our esmaed models s no sascally sgnfcan. A smlar fndng was obaned by usng he Dufour es. Furhermore, o ensure ha no serous volaon of he lneary assumpon n he srucure of he model, we mplemened wo nonlnear conegraon ess-- Kapeanos, Shn and Snell (3) es as well as he Solls (9) along he lnes suggesed by Ghoshray (1). The resuls are repored n he Appendx. Table 6. Speed of adjusmens and mean me lags for adjusmens of desred real mpors. Speed of Adjusmens Response of Desred Real Impors o Each Regressor Counres Desred Real Impors Half-Lfe Adjusmen Mean Tme Lag m ψ y e Inda -.4 (.13) Korea -.7 (.18) Malaysa -.96 (.19) Paksan -.9 ( The Phllpnes -.64 (.4) Sr Lanka (.48) Thaland -.4 (.) Average Descrpon: The values n parenheses besde he speed of adjusmens are he sand errors. Boh he half-lfe and he mean me lag are n absolue erms and n quarers Havng provded evdence supporng he adequacy of he esmaed equaons, several feaures of he resuls n Table deserve menon. Frs and foremos, he coeffcen of he error-correcon erm s sascally sgnfcan n each of he seven cases and s always negave. Therefore, he valdy of an equlbrum relaonshp among he varables n he conegrang equaon s suppored. Ths mples ha overlookng he conegraedness of he varables would have nroduced msspecfcaon n he underlyng dynamc srucure and noe ha here s no evdence of possble nonlnear conegraon. In he case of he Phlppnes, he possbly of asymmerc adjusmen exss gven he sgnfcance of Solls es. Second, he change n real money balancer per quarer ha s arbued o he dsequlbrum beween he acual and equlbrum levels s measured by he absolue values of he EC erm of each equaon. There s consderable nercounry varaon n he adjusmen speed o he las perod s dsequbrum, wh Sr Lanka havng he larges and Inda, he smalles. The average coeffcen s -.77 per quarer. Ths mples ha he adjusmen of real balance o changes n he regressors may ake abou sx quarers n Sr Lanka o slghly more han hry-egh quarers n Inda. In economc erms, hs means ha when real balances exceed her long-run relaonshp wh her regressors, hey adjus downwards a a rae of 6 o 38 percen n each quarer. Our resuls concernng he coeffcen of he EC erm for Inda, conflc wh hose of Bahman-Oskooee and Rehman (), because hey oban -.6 whch s sascally nsgnfcan wh a -value of.6. However, her resuls of -.1, -.1, -.1 and -.3 for Malaysa, Paksan he Phlppnes and Thaland, respecvely, are smlar o hose repored n Table and are (n absolue erms) whn he un nerval. Taken ogeher, hese resuls pon o he exsence of marke forces n he money marke ha operaes o resore long-run equlbrum afer a shor-run dsurbance. Fnally, he dynamcs of he equaon show ha changes n real ncome, neres rae and exchange rae have shor-run effecs on real balances n addon o her long-run effecs esablshed earler. Therefore, hey are conssen wh he heory. 67

10 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 In sum, our fndngs mply (a) ha no nocng he conegraon of he varables would have nroduced a msspecfcaon n he underlyng dynamc srucure; (b) ha use of neres rae as he opporuny cos of holdng money performs beer han he expeced nflaon n he esmaed Asan money demand relaons; (c) ha here exs marke forces n he money marke secor ha operae o resore long-run equlbrum afer a shor-run devaon; (d) ha, on average, akes more han hree years for per cen of he devaons from long-run equlbrum (he conegraon equaon) o be correced, alhough much longer for Inda (see Table 6); (e) ha he mean me lags sugges ha real balances reac faser o changes n domesc ncome han o changes n exchange raes; (f) ha real balances reac faser o exchange rae changes han o neres rae changes, herefore, had we focused only on he radonal varables (ncome and neres rae), we would have obaned more nference bas and mssed he key role ha exchange raes play n he demand for money funcon of Asan economes; (g) ha lnear error correcon model s que lkely a good approxmaon of he shor-run dynamcs because boh he KSS and Sols mehods revealed nsgnfcan -values and (f) ha he full adjusmen of real balances o changes n he regressors may ake beween sx o hry-egh quarers, dependng on he counry. 4. Concludng Remarks As se ou n he nroducon, he prmary am of hs paper has been o nvesgae emprcally he fundamenal ssue of wheher exchange rae changes nfluence he demand for money n Asan counres by usng recen breakhrough me seres economercs. Unlke prevous research, hs paper uses neres rae as he opporuny cos of holdng money. Therefore, he bass of our analyss s a money demand funcon n whch real money balances depend on real ncome, shor-erm neres rae and exchange rae. Our resuls sugges ha here s a unque, sascally sgnfcan long-run equlbrum relaonshp among he real M balance, real ncome, neres raes and exchange rae n each of he seven counres. The addon of exchange rae o he money demand model was found o be quanavely mporan n mprovng money demand formulaon. Exchange rae was posvely sgned n all cases and mples ha an ncrease n exchange rae (.e., deprecaon of, for example Tha bah), would ncrease he demand for real money balances ha s, he value of foregn secures owned by domesc resdens rses; also, he value of domesc secures n he hands of foregners s lowered. Ths, n urn, ncreases he demand for real balances. Our resuls are drecly relevan o concerns abou whch moneary aggregaes bes deermnes he long run effecs of moneary polcy acon n Asa. Acknowledgemen The auhor would lke o hank Hal Langford, Don Englsh, Ed Manon, and Allan Headley for helpful commens on an earler draf. Specal hanks o Duanporn Boonsrkamcha and Pacharaporn Sanglua for excellen research asssance. Ths research s funded by a GSRF-TAMU-C gran. References Agenor, P. R., & Khan, M. S. (1996). Foregn Currency Deposs and Demand for Money n Developng Counres. Journal of Developmen Economcs,, hp://dx.do.org/1.116/ (96)-3 Ambler, S., & McKnnon, R. (198). U.S. Moneary Polcy and he Exchange Rae: Commen. Amercan Economc Revew, 7, 7-9. Arango, S., & Nadr I. M. (1979, June). Prce Expecaons, Foregn Exchange and Ineres Raes, and Demand for Money n an Open Economy (Workng Paper no.39). Rereved from Naonl Bureau of Economc Research webse: hp:// Arango, S., & Nadr I. M. (1981, January). Demand for Money n Open Economes. Journal of Moneary Economcs, 7, hp://dx.do.org/1.116/34-393(81)9- Arze, A. C. (1989). An Economerc Invesgaon of Money Demand Behavor n Four Asan Developng Economes. Inernaonal Economc Journal, 3, Wner, hp://dx.do.org/1.18/ Arze, A. C. (1997). Condonal Exchange-Rae Volaly and he Volume of Foregn Trade: Evdence from Seven Indusralzed Counres. Souhern Economc Journal, 64, 3-4. hp://dx.do.org/1.37/16149 Arze, A. C., & Malndreos, J. (1). Foregn Exchange Reserve n Asa and s Impac on Impor Demand. Journal of Economcs and Fnance, 4(3), 1-3. Arze, A. C., Malndreos, J., & Grvoyanns, E. C. (). Inflaon-rae Volaly and Money Demand: 68

11 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Evdence from Less Developed Counres. Inernaonal Revew of Economcs & Fnance, 14(1), 7-8. hp://dx.do.org/1.116/j.ref.3.4.1, PMd: Arze, A. C., & Ndubuz, G. (199). Modelng Money Demand Funcons wh Exchange Raes: Regresson Esmaes and Dagnosc Tess. The Inernaonal Journal of Fnance, (), Arze, A. C., Osang, T., & Soje, D. (, January). Exchange-Rae Volaly and Foregn Trade: Evdence from Threen LDC's. Journal of Busness and Economc Sascs, 18(1), hp://dx.do.org/1.37/13913 Arze, A. C., & Shwff, S. S. (1993, June). Conegraon, Real Exchange Rae and Modelng n he Demand for Broad Money n Japan. Appled Economcs, (6), hp://dx.do.org/1.18/ Bahman-Oskooee, M., & Techaraancha A. (1). Currency Subsuon n Thaland. Journal of Polcy Modelng, 3, hp://dx.do.org/1.116/s ()3- Bahman-Oskooee, M., & Rhee H. J. (1994). Long Run Elasces of he Demand for Money n Korea: Evdence from Conegraon Analyss. Inernaonal Economc Journal, 8, hp://dx.do.org/1.18/ Bahman-Oskooee, M., & Rehman, H. (). Sably of he Money Demand Funcon n Asan Developng Counres. Appled Economcs, 37, hp://dx.do.org/1.18/ Das, D. D. (1). How dd he Asan economy cope wh he global fnancal crss and recesson? A Revaluaon and Revew. Asa Pacfc Busness Revew, 18(1), 7-. hp://dx.do.org/1.18/ Domowz, I., & Crag S. H. (199). January-March. Inerpreng an Error Correcon Model: Paral Adjusmen, Forward Lookng Behavor, and Dynamc, Inernaonal Money Demand. Journal of Appled Economercs,, hp://dx.do.org/1.1/jae.3913 Ercsson, N. R. (1998). Emprcal Modelng of Money Demand. Emprcal Economcs, 3, hp://dx.do.org/1.17/bf19449 Engle, R. F., & Granger, C. W. J. (1987, March). Conegraon and Error-Correcon: Represenaon, Esmaon and Tesng. Economerca,, hp://dx.do.org/1.37/ Funke, M. (1). Money Demand n Euroland. Journal of Inernaonal Money and Fnance,, hp://dx.do.org/1.116/s61-66(1)17-1 Ghoshray, A. (1). Smooh Transon Effecs n Prce Transmsson: he Case of Inernaonal Whea Expor Prces. Economcs Leers, 16, hp://dx.do.org/1.116/j.econle Goldfeld, S. M. (1976). The Demand for Money Revsed. Brookngs Papers on Economc Acvy, 3, Haug, A. A. (1996). Tess of Conegraon: A Mone Carlo Comparson. Journal of Economercs, 71, hp://dx.do.org/1.116/34-476(94) Haug, A. A., & Rober F. L. (1996). Long-Run Money Demand n Canada: In Search of Sably. Revew of Economcs and Sascs, 78, hp://dx.do.org/1.37/19938 Ibrahm, M. H. (1). Fnancal Facors and The Emprcal Behavor of Money Demand: A Case Sudy of Malaysa. Inernaonal Economc Journal, 1, Auumn, -7. hp://dx.do.org/1.18/ Johansen, S. (199). Lkelhood-Based Inference n Conegrang Vecor Auoregressve Models. New York: Oxford Unversy Press. hp://dx.do.org/1.193/ Kapeanos, G., Shn, Y., & Snell, A. (3). Tesng for a Un Roo n he Nonlnear STAR Framework. Journal of Economercs, 11, hp://dx.do.org/1.116/s34-476()-6 Lee, T. H., & Chung K. J. (199). Furher Resuls on he Long-Run Demand for Money n Korea. Inernaonal Economc Journal, 9, Auumn, hp://dx.do.org/1.18/ McKnnon, R. I. (1983). Currency Subsuon and Insably n he World dollar Sandard. Amercan Economc Revew, 7, McNown, R., & Wallace, M. S. (199, February). Conegraon Tess of a Long-Run Relaonshp beween Money Demand and he Effecve Exchange Rae. Journal of Inernaonal Money and Fnance, 11, hp://dx.do.org/1.116/61-66(9)94-r 69

12 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 Mundell, A. R. (1963). Capal Mobly and Sablzaon Polcy under Fxed and Flexble Exchange Raes. Canadan Journal of Economcs and Polcal Scence, 9, hp://dx.do.org/1.37/ Oserwald-Lenum, M. (199, Augus). A Noe wh Quanles of he Asympoc Dsrbuons of he Maxmum Lkelhood Conegraon Ranks Tes Sascs: Four Cases. Oxford Bullen of Economcs and Sascs, hp://dx.do.org/1.1111/j b13.x Phlps, P. C. B., & Hansen B. E. (199). Sascal Inference n Insrumenal Varables Regresson wh I (1) Processes. The Revew of Economc Sudes, 7, hp://dx.do.org/1.37/974 Rajan, R. S. (1, July). The Evoluon and Impac of Asan Exchange-Rae Regmes. (Economcs Workng Seres No 8). Rereved from Asa Developmen Bank webse: hp:// Solls, F. (9). A Smple Un roo Tes Agans Asymmerc STAR Nonlneary wh An Applcaon o Real Exchange Raes n Nordc Counres. Economc Modellng, 6, hp://dx.do.org/1.116/j.econmod.8.6. Tan, E. C. (1997). Money Demand Amd Fnancal Secor Developmens n Malaysa. Appled Economcs, 9, hp://dx.do.org/1.18/ Welwa, A., & Ekanayake, E. M. (1998). Demand for Money n Sr Lanka Durng he Pos-1977 Perod: Conegraon and Error-Correcon Analyss. Appled Economcs, 3, hp://dx.do.org/1.18/ Noes Noe 1. Specfcally, our sample perod for each counry ends n 9:4, however, for Inda, Malaysa and Paksan he sarng quarer s 1973:1. In he case of he Phlppnes and Thaland, s 1977:1; for Sr Lanka, s 1978:1, and for Korea, s 1976:1. Avalably of neres rae daa was a major facor n deermnng our sarng quarerly perod. Noe. See he dscussons n he Revew of Economcs and Sascs (RESTAT), 79, May (1997: ); see RESTAT, 8, Augus (1998: 4-43; Journal of Busness and Economc Sascs, 16, Ocober 1998; Arze, Osange and Sloje (); Funke (1) and Arze (1). Noe 3. I s mperave ha we esmae he demand for money n real erms, because we wan o undersand how he real demands for money balances behave over me, so ha changes n he nomnal sock of money are adaped o he growng real demand a consan prces. See also Goldfeld (1976: 64). Noe 4. See Domowz and Hakko (199:3) for a dealed explanaon of he mporance of deflang by consumer prce ndex and usng real GDP as scale measure. Noe. Mundell (1963) has argued ha he demand for money s lkely o depend upon he exchange rae, n addon o he neres rae and he level of ncome. A porfolo model of he fnancal marke explored by Arango and Nadr (1981) provde a role for he exchange rae varable n he money demand funcon; however hey pon ou ha he effec of he exchange s more dffcul o sgn. See, also McKnnon (1963). Noe 6. The exposon draws from Arango and Nadr (1979, 1981), Arze (1989), Bahman-OsKooee and Rhee (1994) and Bahman-OsKooee and Techaraanacha (1). Noe 7. Rebalancng s he process of adjusng he allocaon of varous asses n a porfolo o reflec he new scenaro or o acheve or manan a desred mx. Noe 8. The consequence of nonsaonary s he napplcably of he sandard samplng heory. The conegraon approach s aracve n ha can properly accoun for he nonsaonary seres. Noe 9. Followng he specfcaon suggeson n Funke (1) we base our coeffcen analyss on FMLS esmaor.as Funke (1) pons ou one of he key advanages of hs approved over alernave esmaors, such as Johansen (1988) and Sock and Wason (1988), s ha faclaes a complee analyss of he ncluson of deermnsc rends n he conegraon se. An addonal benef of he above framework s ha faclaes a es of conegraon, where conegraon s aken o be he null hypohess. In he sascs leraure hs would be he naural way o es for conegraon 7

13 Inernaonal Journal of Economcs and Fnance Vol. 4, No. 8; 1 71 Appendx. Nonlnear Tess of he Resdens of he Long-run Equaon. Counry Sample Perod KSS Solls : : Inda 198:1-11: (.13) Korea 198:1-11: (.8) Malaysa 198:1-11: (.6) Paksan 198:1-11: (.83) Phlppnes 198:1-11: (.3) Sr Lanka 198:1-11: (.14) Thaland 198:1-11: (.1) Crcal Value (1%) Descrpon: The values n parenheses are he p-values. a. Kapeanos, G., Shn,Y., and Snell, A. (3) was obaned from H : λ = and H a : λ < b. Solls (9) was obaned from v m e y EC m 3 1 w m e y EC EC m