The Changing Malaysian Financial Environment and the Effects on Its Monetary Policy Transmission Mechanism

Transcription

1 The Changng Malaysan Fnancal Envronmen and he Effecs on Is Moneary Polcy Transmsson Mechansm Mala Vallamma Raghavan School of Economcs and Fnance, RMIT Unversy GPO Box 2467V, Melbourne, Vcora 3001, Ausrala In recen mes, economss concur ha economy s response o moneary polcy s somewha weaker hen hey were n he pas. However, he cause of such change remans an open ssue. One plausble reason for hs change could be arbued o he fnancal reform processes ha have brough sgnfcan srucural changes n he fnancal sysems hrough fnancal nnovaon, negraon and marke developmen. As a resul, hese changes n he fnancal sysems have promped a reassessmen of he ransmsson mechansm hrough whch moneary polcy affecs he fnal varables of ncome and prces. In he Malaysan experence, major reforms were underaken n he 90s ha have opened up new avenues and opporunes for s fnancal marke developmen. However, he changng fnancal envronmen posed grea challenge o Bank Negara n he formulaon and mplemenaon of s moneary polcy. Ths paper nvesgaes o wha exen he lberalsaon processes n Malaysa have affeced he ransmsson channels of moneary polcy and her ably o acheve he ulmae goals of susanable real ncome growh and prce sably. The Johansen co-negrang echnque and he error correcon model are used o analyse he long run and he shor run properes of he fnancal varables such as moneary aggregaes, cred aggregaes and neres raes wh respec o ulmae goal varables. The emprcal fndngs sugges ha he cred and moneary aggregaes are no longer relable as he man nermedae arge. Conversely, neres raes seem o have ganed a sgnfcan role n he pos reform perod. JEL classfcaon: C22; E52; E58; G18; O53 Keywords: Malaysa; Moneary polcy; Transmsson Mechansm; Fnancal Lberalsaon; Johansen co-negraon

2 1. Inroducon In economc leraure, moneary polcy s seen as a powerful polcy nsrumen o acheve he ulmae economc goals of prce sably and susanable economc growh. There are several channels hrough whch he moneary polcy changes affec economc acvy and nflaon and hey are collecvely known as he ransmsson mechansm of moneary polcy. The major channels ha have been denfed n he leraure are neres rae channel, monears channel, exchange rae channel, wealh channel, asse prce channel and bank lendng channel. In recen mes, economss concur ha economy s response o moneary polcy s somewha weaker hen hey were n he pas (Kuner and Mosser, 2002; Clnon and Enger, 2000; Bovn and Gannon, 2002). However, he cause of such change remans an open ssue. As saed n Gordon and Sellon (2002), one reason for he change n he moneary ransmsson mechansm could be due o he sgnfcan srucural changes n he fnancal sysem. Snce, he moneary ransmsson mechansm depends on banks and fnancal markes o channel moneary polcy acons, changes n he srucure of he fnancal sysem could aler he moneary ransmsson mechansm. The fnancal lberalsaon processes have brough many srucural changes n he fnancal sysems hrough fnancal marke developmen, fnancal nnovaon, fnancal negraon and ec. In hs envronmen of lberalsed fnancal sysem and srong capal flows, he effecveness of moneary polcy has ofen been quesoned. I has acually alered he channels of moneary polcy manly affecng he relaonshp beween moneary aggregaes, cred aggregaes, exchange raes and neres raes on ncome and prces. As a resul, he reform processes promped a reassessmen of he ransmsson mechansm 2

3 hrough whch moneary polcy affecs he aggregae demand and ulmaely he fnal varables of prces and oupu (Raghavan, 2000). Ths paper examnes o wha exen he fnancal marke developmen and he lberalsaon processes n Malaysa have affeced he ransmsson channels of moneary polcy and her ably o acheve he ulmae goals of susanable real ncome growh and prce sably. Secon wo of hs paper denfes he mporan changes and major developmen ha has aken place n he Malaysan fnancal srucure. Secon hree provdes he breakdown of he perod of sudy and sources of daa used n hs work. Secon four descrbes he economerc models used o analyse he relaonshp beween he seleced fnancal varables and he ulmae goal varables. In parcular, he Johansen conegraon echnque and he error correcon model are used o nvesgae he long run and shor run properes of he fnancal varables such as moneary aggregaes, cred aggregaes and neres raes wh respec o ncome and prces. In secon fve, he emprcal evdence on he long run properes of he fnancal varables are analysed o asceran her appropraeness as arge varables n he process of formulang a polcy goal. The shor run properes are also examned manly o denfy her nformaon conen n he moneary ransmsson mechansm durng he pre and pos lberalsaon perods. Fnally, secon sx concludes hs paper. 3

4 2. The Fnancal Lberalsaon Process n Malaysa The modern economc analyss of fnancal polcy sared wh he works of Ronald Mcknnon (1973) and Edward Shaw (1973). Boh auhors drew aenon o he prevalen fnancal represson and argued ha such represson was mposng a major cos on counres ha pracced. A repressed fnancal sysem can be descrbed as a sysem wh neres rae resrcons, domesc cred conrols, hgh reserve requremens, and segmened fnancal markes, under developed money and capal markes and conrols on nernaonal capal flows. As me passed by, became apparen ha severe fnancal represson had large coss n erms of growh and dsrbuon on counres ha pracced. Wh regards o he above menoned problems, fnancal lberalzaon can be vewed as a se of polcy measures, whch are predomnanly desgned o deregulae and ransform a hampered fnancal sysem n order o acheve a lberalzed marke orened sysem wh an approprae regulaory framework. Fnancal lberalzaon can be perceved as a se of operaonal reforms ha can lead o greaer flexbly n neres raes, enhance he role for cred and foregn exchange allocaons and ncrease he auonomy of he commercal banks. In addon, s expeced o provde greaer deph for money, secures and foregn exchange markes, and ncrease he cross border flow of capal. On he whole, he broad hrus of fnancal lberalzaon was prmarly o enhance effcency hrough greaer relance on marke forces and o reduce governmen nervenon n he fnancal secor. In he early 1980s, he waves of fnancal nnovaon and reforms, whch were sweepng he advanced counres, also sared spreadng o he developng counres. Thereafer, fnancal lberalzaon became a worldwde phenomenon. Along wh he scope of fnancal lberalzaon varous aspecs of reform processes emerged and hey nclude fnancal secor reform, deregulaon, re-regulaon and fnancal nnovaon. 4

5 For Malaysa, he fnancal secor reforms have been a gradual, phased and connung process. I was mplemened n sages so as o allow fnancal markes and consumers o adjus o he new envronmen. There were enave seps aken owards lberalzaon n he early 1970s. Neverheless, as saed n Awang (1992), he major phase of lberalzaon commenced n Ocober 1978 when he Cenral Bank of Malaysa, Bank Negara nroduced a package of measures as a concree sep owards a more marke-orened fnancal sysem. Among he measures nroduced were freeng he neres rae conrols and reformng he lqudy requremens of he fnancal nsuons. The lberalzaon of neres raes n 1978 ndeed represened a conscous polcy measure by Bank Negara o promoe a more lberal and compeve fnancal sysem. Though Malaysa assumed he lead n neresrae deregulaon n he regon, s reforms were no compleed ll he lae 1980s. In fac here were some emporary polcy reversals along he way. For example, he marke deermnaon of neres raes was suspended durng he gh lqudy perod from Ocober 1985 o January 1987 and n Sepember Anoher major phase of fnancal reforms was underaken n January 1989, whereby Bank Negara nroduced a package of reforms o broaden, deepen and modernze he fnancal sysem. The reforms ncluded he apponmen of prncpal dealers, he ssung of Malaysan Governmen Secures by aucon, conducng Bank Negara s open marke operaons hrough prncpal dealers and freeng of dscoun house operaons. Ths perod also wnessed several sgnfcan developmens wh far reachng mpac on he developmen of capal marke. As par of Bank Negara s connung effors o develop he prvae deb secures marke, he counry s frs cred rang agency, he Rang Agency Malaysa Berhad was esablshed n 1990 o rae deb ssues by corporaons. Anoher sgnfcan developmen n he capal marke was he seng up of he marke wach dog n he form of Secures Commsson n March In 1997, he fnancal fuures and 5

6 opons markes began her operaon, namely he Kuala Lumpur Fuures Marke (KLFM) and he Kuala Lumpur Opons and Fnancal Fuures Exchange (KLOFFE). The fnancal lberalzaon and reforms underaken n Malaysa have opened up new avenues and ncreased opporunes for fnancal marke developmen. However, n he new envronmen wh closer fnancal negraon and srong capal flows, he effecveness of moneary polcy has ofen been quesoned. Fnancal lberalzaon had mporan mplcaons for boh he ransmsson mechansm, and he operang procedure of moneary polcy. I has acually alered he channels of moneary polcy manly affecng he relaonshp beween moneary aggregaes, cred aggregaes and neres raes on ncome and prces. These changes posed a major challenge n he formulaon and mplemenaon of moneary polcy. As a resul, he reform processes promped a reassessmen of he approprae nermedae arges of moneary polcy. In he 1980s, Malaysan moneary polcy was based on argeng moneary aggregaes; however he polcy was no formally announced o he publc. Ths sraegy was seleced based on evdence ha moneary aggregaes were closely lnked o he ulmae goals of moneary polcy. By conrollng he supply of money, Bank Negara was able o nfluence he major marke varables of aggregae demand and nflaon. In he early 1980s, he emphass was on M1, as had been emprcally found o have a sable and conssen relaonshp wh aggregae ncome (BNM, 1999). The subsanal fnancal lberalzaon and nnovaon n he 1980s has however rendered M1 less relable for polcy argeng. The changes n he lqudy preferences of he publc and perssenly hgh neres raes led o srucural changes n he demand for money. Rsng sophscaon n he fnancal sysem and he demand for money by he publc resuled n savers becomng ncreasngly sensve o neres rae movemens (BNM, 1994). Consequenly, he radonal relaonshp of M1 o aggregae ncome was compromsed and he focus nevably shfed o he broader moneary aggregaes. 6

7 Accordng o Bank Negara repor, a correlaon es conduced usng quarerly daa from , showed a posve and hgh correlaon beween broad moneary (M3) growh and nflaon (BNM, 1999). Snce prce sably was an ulmae goal of moneary polcy, broad moneary argeng was seen as a suable arge. In md-1980s, Bank Negara seleced M3 as he polcy arge. In he 1990s, he far-reachng changes n he economy and he fnancal sysem weakened he relaonshp beween moneary aggregaes and he ulmae goal varables. The annual growh of money supply as measured by M3 was exremely volale durng he perod of large capal flows ( ) and he large swngs n he moneary aggregaes reduced he vably of M3 as nermedae arge. The growh n money supply was largely conrbued by he expanson of fnancal neworks, and he wdenng of he range of bankng nsrumens and servces, he ncrease n he role of money n he economy, and he rapd monezaon process. Hence, durng hs perod, moneary veloces and raos of nomnal GDP o varous moneary aggregaes have shown frequen and marked deparures from her hsorcal paerns makng he moneary aggregaes unrelable as ndcaors of economc acvy and as gudes for sablzng prces. Ths hghlghed he problems assocaed wh usng moneary aggregaes as polcy arge. The lberalsaon processes and consequenly he changes n he fnancal sysem posed a major challenge o Bank Negara n he formulaon and mplemenaon of moneary polcy. In vew of he changng fnancal envronmen, he moneary polcy should adhere o a suable polcy framework so ha can reman as an effecve polcy n promong economc growh and mananng prce sably. 7

8 3. Perod of Sudy and Sources of Daa In order o see he mpac of fnancal lberalzaon on moneary polcy, he perod of sudy, whch covers 1973:4 o 2000:1, wll be broken down as follows: 1973:4 o 1989:4 Represens he perod before he major shf n he mplemenaon of moneary polcy n Malaysa. Durng hs perod, he Malaysan Fnancal Sysem had been subjeced o he wdely pracced mechansm of fnancal represson such as neres rae regulaon and preferenal cred scheme. 1979:1 o 2000:1 Represens he pos-lberalzaon perod n Malaysa. A he end of 1978 neres raes were deregulaed and hus represenng he begnnng of he era of fnancal reforms. The nernaonal economc and fnancal envronmen n he pos lberalzaon perod posed new challenges for naonal economc managemen and he operaon of moneary polcy. 1990:1 o 2000:1 Represens he perod afer he major shf n he mplemenaon of moneary polcy, whereby Bank Negara embarked on a seres of fnancal reforms o mprove and modernze he fnancal sysem. Quarerly seres were ulzed for he emprcal analyss. A oal of nne fnancal aggregaes are employed and hey are hree moneary aggregaes, hree cred aggregaes and hree neres raes. The quarerly gross domesc produc s used o proxy economc acvy, he consumer prce ndex s used o represen he prce level and Federal governmen expendure s used o represen he fscal varable. Varables, descrpons and sources are provded n Table A n he Appendx. 8

9 4. Mehodology The co-negraon echnque advocaed by Johansen (1988) was ulzed o deermne he long run relaonshp beween he fnancal varables, ncome and prce. As demonsraed n Greene (1997) and Tan (1996) he advanage of usng he Johansen echnque s ha allows for he possble exsence of mulple co-negrang vecors and her denfcaon parcularly n regressons nvolvng more han wo varables. The am of he co-negraon es s manly o denfy f he varables n queson are drfng apar or ogeher. To carry ou he Johansen co-negraon es, he followng VAR mus frs be esmaed. Y = Γ1Y 1 Γ2Y 2... ΓpY p e The order of he model p mus be deermned n advance. If he above model nvolves N varables hen Y = [y 1, y 2,. yn], whch are ndvdually I(1) and as a resul N x 1 vecor can be formed and Γ s an N x N marx of parameers. Re-parameersng he sysem of equaons above n an error correcon represenaon yelds he followng: (1) or = Π py p Π Y 1 Π p 1Y p 1 Y 1... e (2) Y = Π p py p Π Y = 1 where = 1, 2, (p 1). The long run relaonshp among he varables n he VAR s emboded n he marx Π p and hus hs marx conrols he co-negraon properes. As menoned before, n he above model, Y s a vecor of I (1) varables whle Y, Σ =1 Π Y - and e are saonary. All he possble combnaons of he levels of Y ha yeld hgh correlaon wh he I (0) elemens n equaon (2) are hen esmaed and hese combnaons are referred as he conegrang vecors. e (3) 9

10 The rank of he marx Π p s deermned by he number of co-negrang vecors r among he elemens of Y. There are hree possble cases: Case 1: Π p s a full rank N, hen any lnear combnaon of Y -1 s saonary and by mplcaon, he elemens n vecor Y are no I (1). Case 2: Π p has rank beween 0 and N, hen here exs r co-negraed vecors whch are denfable and ncorporaable no error correcon model. Snce he maxmum number of conegrang vecors can only be N 1, r mus be smaller han N. Case 3: Π p has rank zero and hus no lnear combnaons of Y -1 are saonary mplyng ha he elemens n vecor Y are I(1) bu no co-negraed. If Π p has less han full rank, we can express as follows: Π p = γα (4) The above equaon nvolves esmang he marx α (.e. an N x r marx) ha conans all he possble co-negrang vecors and he γ marx conanng he correspondng se of error correcon coeffcens. If here are r co-negrang vecors, hen α and γ each have r columns (Cochrane, 1997). Rewrng equaon (2) wh α and γ, shows he error correcon represenaon: γα (5) Y Y Y Y e = ' p Π Π p 1 p 1 α Y -p mus be saonary so ha γα Y -p wll also be saonary. Thus α s he marx of co-negrang vecors. If elemens n Y do co-negrae hen a leas one of he α vecors 10

11 wll be sascally sgnfcan. Thus by vrue of Granger represenaon heorem, γ mus also conan a leas one non-zero elemen. The rank of he marx Π p can be deermned by referrng o he egenvalues λ I derved from he maxmzaon of he concenraed lkelhood funcon of equaon (5). Hence, he number of co-negrang vecors r can be deermned by usng he followng maxmal egenvalue sascs. ς = T r n = r 1 where r = 0,1,2,3,.,n-2, n-1 ln( 1 λ ) (6) The null hypohess s ha here are a mos r co-negrang vecors agans he alernave hypohess of r1 co-negrang vecors. The error correcon model (ECM) noed by Engle & Granger (1987) s a model whch forces gradual adjusmens of he dependen varable owards some long run values wh explc allowances made for he shor run dynamcs. In hs sudy, he ECM s used o es he shor run relaonshp connecng he growh rae of fnancal varables o he growh rae of ncome and prces. In order o nvesgae he predcve relaonshp of moneary aggregaes, cred aggregaes and neres raes o economc acvy, he followng ECM equaons were esmaed. Equaons (7) and (8) represen he real ncome equaon whle equaons (9) and (10) represen he prce equaon. The y, p, x, and g are respecvely he dfference of log of real ncome, prce ndex, fnancal varable and governmen expendure. Meanwhle, EC and S represen he error correcon erm and he seasonal dummy varables respecvely. The α, β, γ, λ, δ and θ are coeffcens o be esmaed whle he ε are dsurbance erms. 11

12 12 = = = = n n n S y p x EC y ε θ δ γ β α (7) = = = = = n n n n S y g p x EC y ε θ δ λ γ β α (8) = = = = n n n S y p x EC p ε θ δ γ β α (9) = = = = = n n n n S y g p x EC p ε θ δ λ γ β α (10) The error correcon erm EC -1 s consruced by usng he coeffcens from he conegraon regressons. The frs dfference erms n he above equaons capures he shor run dynamcs whle he error correcon erms capures he adjusmens owards long run equlbrum. In hs sudy, he F-sascs ess he null hypohess ha he coeffcens of he fnancal varables ndcaed are zero. Tesng he sgnfcance of coeffcens of he seleced fnancal varables assesses he nformaon conen of hese varables.

13 5. Emprcal Resuls and Analyss Pror o assessng he relaonshp amongs varables based upon he noon of conegraon and vecor auoregresson, her unvarae me seres properes have o be examned. Tes for he saonary of a me seres was carred ou usng he Augmened Dckey-Fuller (ADF) es. The Hylleberg, Engel, Granger and Yoo (HEGY) es was carred ou o asceran wheher each of he seres n queson possess a un roo a some frequency oher han he usual zero frequency such as bannual and or annual frequences. Boh he ess sugges ha he me seres saonares can be acheved smply by frs order dfferencng. The HEGY es mples ha here s no seasonal un roo problem for he varables and hus no need o worry abou he applcaon of he seasonal co-negraon echnque. The long run relaonshp beween he seleced varables and he economc acves s examned by usng he co-negraon es. Frs, a smple bvarae co-negraon es manly nvolvng resduals was carred ou usng he ADF es. The resuls n Table 1 show ha n he case of fnancal varable-ncome relaonshp, he ADF es sascs of he resduals for all hree moneary aggregaes, ALR and IBR3 are rejeced a 5% sgnfcance level. The emprcal resuls hus ndcae ha hese varables are co-negraed wh ncome and a large par of he movemens n ncome are anchored by he long run movemens of hese fnancal varables. On he oher hand, he cred aggregaes and TBR3 faled he es and herefore are no co-negraed wh ncome. As n he case of he fnancal varableprce relaonshp, he ADF es sascs are only sgnfcan for he IBR3 (5% sgnfcance level) and ALR (10% sgnfcance level). The oher fnancal varables faled he sgnfcance es and hus have no endency o reurn o an equlbrum relaonshp. 13

14 Tables 2 and 3 provde he resuls of he applcaons of he Johansen echnques o he denfcaons of long run relaonshps of he varous fnancal varables wh ncome, prce level and fscal varable along wh a dummy varable D78Q4. The dummy varable D78Q4 s nended o reflec a swch n he Malaysan neres rae regme naed n Ocober The esmaon resuls presened n Tables 2 and 3, sugges ha, wh he excepon of he TBR3, for all he oher fnancal varables, wh or whou he ncluson of he fscal varable, here exs one co-negrang vecor a he 5% sgnfcance level. A necessary condon for he arge ndcaor o be effecve n he mplemenaon of he moneary polcy s ha he fnancal varables should co-negrae wh he ncome and prce. Overall, he Johansen co-negrang resuls provded srong evdence of a sable long run relaonshp amongs mos of he fnancal varables wh ncome and prces. Usng he co-negraon echnque, an error correcon (EC) erm was obaned and s normalzed o real ncome and he equaon s as follows: I. Three varable sysem: EC = C LRY α LNP α LRX α D78 4 (11) II. Four varable sysem: Q EC = C LRY α LNP α LRG α LRX α D78 4 (12) Q Table 4 shows he error correcon erms for all he fnancal varables n he hree and four varables sysem. These error correcon erms represen he co-negraon relaonshp and are nerpreed as devaons from he long run equlbrum. The error correcon models (ECM) as noed by Engle and Granger (1987), s a model whch forces gradual adjusmens of he dependen varables owards some long run value wh explc allowance made for he shor run dynamcs. The ECM model s esmaed based on he general auoregressve dsrbuon wh an error correcon erm formed by he 14

15 relevan esmaed co-negrang vecor and seasonal dummes. To nvesgae he predcve relaonshp of he fnancal varables, equaons (7) o (10) were esmaed nne mes for each sub-perod (.e. he hree moneary aggregaes, he hree cred aggregaes and he hree neres raes). The lag lengh for he models was deermned afer subjecng each model o a seres of ARCH correlaon LM es rangng from he frs order o fourh order. A hree-lag perod seems o be approprae and was herefore used unformly n he mulvarae ECM models. Tables 5 and 6 shows he -sascs of he coeffcens of he error correcon erm (α) and he coeffcen of deermnaon (R 2 ) of he real ncome equaon and prce equaon respecvely for each sub-perod. In all esmaons of real ncome equaon, he -sascs of he coeffcen of he error correcon (EC) erm shows a sgnfcan level, hus mplyng ha, here s a hgh endency for ncome o adjus owards some long run values wh explc allowance made on he shor run dynamcs. Moreover, he coeffcen of deermnaon R 2 s also reasonably hgh, where more han 80% of he varaon n he dependen varable real ncome can be explaned hrough he models. As for prce equaon, he -sascs of he coeffcen of he error correcon (EC) erms are nsgnfcan and hus ndcang ha here s low endency for he prce o adjus owards an equlbrum relaonshp. The coeffcen of deermnaon R 2 s also no very hgh whereby he prce equaon models can only explan abou 40 o 50% of he varaon n prce. The F-sascs are obaned from he Wald es and Tables 7 and 8 summarze he F- sascs for he sgnfcance of he fnancal varables n real ncome equaon. As shown n he wo ables, real M1 seems o be sgnfcanly relaed even a 1% level wh real ncome hroughou he sample perods. Though M2 and M3 are nherenly co-negraed wh economc acvy n he long run, n he shor run hey faled o provde he necessary nformaon abou he fuure ncome movemens. Therefore he money-ncome relaonshp 15

16 does no sasfy he srngen condon ha would be requred o render he src use of broad money as an nermedae arge. The null hypohess of β = 0 for all hree cred aggregaes are no rejeced even a 10% level hroughou he perod of sudy. Even wh he ncluson of a fscal varable as a conrol varable n he error correcon model, dd no have any effec on he sgnfcance level. Boh ALR and IBR3 conaned sascally sgnfcan nformaon abou he fuure flucuaons n ncome especally n he pos lberalzaon perods. The move owards a lberalzed fnancal sysem has acually enhanced he role of neres raes n he moneary ransmsson mechansm. As for TBR3, he F-sascs are nsgnfcan hroughou he sample perod because he developmen of he marke for hese blls s sll shallow and hus could no provde he necessary nformaon needed o predc ncome. Tables 9 and 10 provde he summary of he F-sascs for he sgnfcance of he fnancal varables n he prce equaons. Overall, he hree moneary aggregaes seem o be nsgnfcanly relaed o prce. The ncluson of fscal varable also dd no mprove he predcve power of he moneary aggregaes. As for he cred aggregaes, wh he excepon of he sub-perod 1990:1 o 2000:1, n all he oher sub-perods, he real CR1 and CR3 are sascally sgnfcan a 5% level and hus upheld her role n forecasng fuure prce movemens. On he conrary, he F-sascs of he ALR and IBR3 are sascally sgnfcan n he 1990s, hus hghlghng he growng mporance of he neres raes n he moneary ransmsson mechansm. 16

17 6. Concluson The Malaysan experence n movng owards a lberalzed fnancal sysem brough many new challenges for Bank Negara and hese challenges manly relae o he way n whch moneary polcy s formulaed and mplemened. Based on he emprcal evdence, would no be unreasonable o conclude ha he evoluon of he economy and he fnancal sysem had a grea mpac on boh he ransmsson mechansm and he operang procedures of moneary polcy. Wh respec o operang procedure of moneary polcy, fnancal lberalzaon has also affeced he seng of polcy arges. Afer he major reform n he lae 1980s, was found ha he relaonshps beween broad moneary aggregaes and economc acves have changed. Though he conegraon es suppors he exsence of he long-erm relaonshp beween he fnancal varables wh ncome and prce level, he error correcon model (ECM) whch manly capures he shor run dynamcs, found he moneary aggregae M1, average lendng raes and hree monh ner-bank raes o be sgnfcanly affecng ncome. As for prce level, only neres raes are found o be sgnfcan n he pos lberalzaon perod. The broad moneary aggregaes M2 and M3 faled o provde he necessary nformaon abou he fuure ncome and prce movemens and hus rased some pernen quesons abou he ably of moneary arge o serve as a communcaon devce. These resuls bear srong negave mplcaons for many famlar moneary polcy frameworks ha cenred he desgn and mplemenaon of polcy on broad money. Ths s because, he money ncome or money-prce relaonshp does no sasfy he srngen condons ha would be requred o render he src use of money as an nermedae arge. Ths sudy s mporan fndng wh poenally posve mplcaons for moneary polcy s he relance on neres raes. The average lendng raes and he hree-monh ner-bank 17

18 raes conaned he nformaon abou subsequen movemens n real ncome and prce level and s hghly sgnfcan n he 1990s. In he pos lberalsaon perod, neres rae seems o play a sgnfcan role n he moneary polcy framework. The major developmens n he Malaysan fnancal srucure have ncreased he effecveness of neres rae as a moneary polcy varable and hus appeared o be an approprae and necessary arge for an effecve mplemenaon of moneary polcy. Acknowledgemen The auhor s graeful o Professor Rober Brooks and Assocae Professor Snclar Davdson of he School of Economcs and Fnance, RMIT Unversy for her consrucve commens on earler versons of hs paper. References Awang, A.H., Ng T.H. & Raz A Fnancal Lberalsaon and Ineres Rae Deermnaon n Malaysa. Kuala Lumpur: Bank Negara Malaysa Dscusson Paper. Bank Negara Malaysa Annual Repors. Kuala Lumpur. Bank Negara Malaysa Quarerly Economc Bullens. Kuala Lumpur. Bank Negara Malaysa Monhly Sascal Bullens. Kuala Lumpur. Bank Negara Malaysa Money and Bankng n Malaysa. Kuala Lumpur: 35 h Annversary Edon Bank Negara Malaysa The Cenral Bank and he Fnancal Sysem n Malaysa: A Decade of Change. Kuala Lumpur: 40 h Annversary Edon. Bovn, J., Gannon, M Assessng Changes n he Moneary Transmsson Mechansm: A VAR Approach. Federal Reserve Bank of New York: Economc Polcy Revew. pp Clnon, K., Enger, W Conference summary: Money, Moneary Polcy, and Transmsson Mechansm. Bank of Canada Revew. pp Cochrane, J.H Tme Seres for Macroeconomcs and Fnance. Graduae School of Busness - Unversy of Chcago. 18

19 Dckey, D.A., Fuller, W.A Dsrbuon of he Esmaors for Auoregressve Tme Seres wh a Un Roo. Journal of he Amercan Sascal Assocaon 74: pp Engle, R.F., Granger, C.W.J Conegraon and Error Correcon: Represenaon, Esmaon and Tesng. Economerca 55(2): pp Gordon, H.S The Changng U.S. Fnancal Sysem: Some Implcaons for he Moneary Transmsson Mechansm. Federal Reserve Bank of Kansas Cy: Economc Revew. 87(1): pp Greene, W.H Economerc Analyss, New Jersey: Thrd Edon, Prence Hall. Inernaonal Moneary Fund Inernaonal Fnancal Sascs Year Books.Washngon D.C. Johansen, S Sascal Analyss of Conegrang Vecors. Journal of Economc Dynamc and Conrol. 12: pp Kuner, K.N., Mosser, P.C The Moneary Transmsson Mechansm: Some Answers and Furher Quesons, Federal Reserve Bank of New York: Economc Polcy Revew. pp McKnnon, R.I Money and Capal n Economc Developmen. Washngon D.C: The Brookngs Insuon. pp Raghavan, M.V The Effecs of Fnancal Lberalzaon on Moneary Polcy: The Malaysan Experence. M.Ec Dsseraon. The Faculy of Economcs and Admnsraon. Unversy Malaya, Kuala Lumpur. Shaw, E.S Fnancal Deepenng n Economc Developmen. New York: Oxford Unversy Press. pp Tan E.C Money Demand Amd Fnancal Secor Developmens n Malaysa. In Proceedngs of he Inernaonal Conference on Globalzaon and Developmen. Kuala Lumpur. Tlak, A., Lee, C Bes Lnear Unbased Inerpolaon of Quarerly GDP: The Case for Malaysa In Proceedngs of he Fourh Malaysan Economerc Conference. Kuala Lumpur. 19

20 Appendx Table A. Varables, Descrpons And Sources Varables Descrpons Sources Y Real Income Real Gross Domesc Produc by expendure componens n consan prces (1987=100). BNM & Tlak P Prce Level Consumer Prce Index (1994=100) BNM G Real Fscal Varable Federal Governmens curren expendure plus developmen expendure. BNM M1 Moneary Aggregae Real M1 Currency n crculaon plus demand deposs. BNM M2 Moneary Aggregae Real M2 M1 plus savngs deposs plus fxed deposs plus NIDs plus Repos plus foregn currency deposs. BNM M3 Moneary Aggregae Real M3 M2 plus deposs placed wh oher bankng nsuons. BNM CR1 Toal Real Domesc Cred Moneary Survey s clams on cenral Governmen, sae & local Governmens, non-fnancal publc enerprses, prvae secors, oher bankng nsuons and non-bank fnancal nsuons. Moneary Auhores and Depos Money Bank s daa consoldaed no a Moneary Survey. IFS CR2 Real Moneary Survey s Clams On Prvae Secors I s he sum of Moneary Auhores clams on prvae secors plus Depos Money Banks clams on prvae secor. IFS CR3 Real Depos Money Bank s Clams On Prvae Secors Depos Money Banks comprse commercal banks and oher fnancal nsuons. IFS ALR Average Lendng Raes Commercal banks average lendng raes. BNM IBR3 Three Monh Inerbank Raes Iner-bank money marke raes. BNM TBR3 Three Monh Treasury Bll Raes Average dscoun rae on Treasury blls. BNM Noes: 1. BNM s referred o Quarerly and Monhly Bullen, Cenral Bank of Malaysa. 2. IFS s referred o Inernaonal Fnancal Sascs of Inernaonal Moneary Fund. 3. Tlak s referred o Tlak & Lee (1996). 20

21 Tables Table 1. Tess For Saonary Fnancal Varable-Income And Fnancal Varable-Prce Relaonshps (Augmened Dckey- Fuller Tes) Fnancal Varables Income Prce ADF; τ ADF; τ LRMI * LRM * LRM * LRCR *** LRCR LRCR ALR ** IBR ** *** TBR ** Noes: 1. Un roo ess summary of sascs for resduals for sample perod coverng from 1973:4-2000:1. 2. There are 106 observaons. 3. All varables are n naural logarhm and real erms excep for neres raes. 4. The lag lengh was deermned afer subjecng each regresson o a seres of LM ess for seral correlaon a fve-percen level rangng from frs order o fourh order. One lag perod seems o be approprae and was used unformly for all he varables. 5. (***), (**) And (*) denoes rejecon of he null hypohess of he un roo a he 10%, 5% and 1% level respecvely. 21

22 Table 2. Summary Of Tes Sascs For The Number Of Co-negrang Vecors In A Three Varable Sysem (Johansen Co-negraon Tes) Fnancal Varables r = 0 r 1 r 2 Number of CVs LRM ** LRM ** LRM * LRCR * LRCR ** LRCR * ALR * IBR * TBR Noes: 1. CV represens co-negrang vecors and r represens number of co-negrang vecors. 2. Three-varable sysem ncludes real ncome, prce ndex and fnancal varable along wh he dummy varable D78Q4. 3. Tes assumpon: Lnear deermnsc rend n he daa seres. 4. The co-negrang vecors were esmaed wh a provson for hree lags and he lag lengh was deermned afer subjecng each varable sysem o a seres of ARCH seral correlaon LM ess. 5. Sample perod coverng from 1973: o 2000:1 and oal number of observaons are (**) and (*) denoes rejecon of null hypohess a 5% and 1% respecvely. 22

23 Table 3. Summary Of Tes Sascs For The Number Of Co-negrang Vecors In A Four Varable Sysem (Johansen Co-negraon Tes) Fnancal Varables r = 0 r 1 r 2 r 3 Number of CVs LRM * LRM ** LRM * LRCR * LRCR * LRCR * * ALR * IBR * TBR ** Noes: 1. CV represens co-negrang vecors and r represens number of co-negrang vecors. 2. Four-varable sysem ncludes real ncome, prce ndex, fnancal varable and fscal varable wh D78Q4. 3. Tes assumpon: Lnear deermnsc rend n he daa seres. 4. The co-negrang vecors were esmaed wh a provson for hree lags and he lag lengh was deermned afer subjecng each varable sysem o a seres of ARCH seral correlaon LM ess. 5. Sample perod coverng from 1973: o 2000:1 and oal number of observaons are (**) and (*) denoes rejecon of null hypohess a 5% and 1% respecvely. 23

24 Table 4. Error Correcon Terms For The Three And Four Varables Sysem Fnancal Varable M1 M2 M3 CR1 CR2 CR3 ALR IBR3 Error Correcon Equaon Three Varable Sysem EC=1.909 LRY 0.591LNP 1.060LRM D78Q4 EC = LRY LNP 0.002LRM D78Q4 EC = LRY 0.867LNP 0.387LRM D78Q4 EC = LRY 0.447LNP LRCR D78Q4 EC = LRY 0.851LNP 0.389LRCR D78Q4 EC = LRY 0.505LNP 0.491LRCR D78Q4 EC = LRY 0.284LNP 0.794ALR 1.033D78Q4 EC = LRY LNP IBR 0.761D78Q4 Four Varable Sysem M1 M2 M3 CR1 CR2 CR3 ALR IBR3 TBR3 EC = LRY 1.341LNP 0.483LRG LRM D78Q4 EC = LRY 1.021LNP 0.437LRG LRM D78Q4 EC = LRY 0.674LNP 0.057LRG LRM D78Q4 EC = LRY 2.226LNP 0.193LRG LRCR D78Q4 EC = LRY 1.286LNP 0.319LRG LRCR D78Q4 EC 1 = LRY 2.128LRG LRCR D78Q4 EC 2 = LNP 2.817LRG LRCR D78Q4 EC = LRY 6.122LNP 1.828LRG ALR 0.419D78Q4 EC = LRY 4.375LNP 1.236LRG IBR 0.280D78Q4 EC = LRY 2.932LNP 0.452LRG TBR 0.207D78Q4 24

25 Table 5. Summary Of The -sascs Of The Coeffcen of Error Correcon Term (α) And The Coeffcen Of Deermnaon (R 2 ) In Real Income Equaon Sample Perod 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 Fnancal Varables LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R * * * * * * ** * * * * * * * * ** * * * * * * * ** * * ** * * * * ** * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Noes: 1. Three-varable sysem and four-varable sysem are represened by real ncome equaon 7 and 8 respecvely. 2. (**) and (*) denoes rejecon of he null hypohess a 5% and 1% respecvely. 25

26 Table 6. Summary Of The -sascs Of The Error Correcon Term (α) And The Coeffcen Of Deermnaon (R 2 ) In Prce Equaon Sample Perod 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 Fnancal Varables LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem 3 Varable Sysem 4 Varable Sysem -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R 2 -sas R ** ** ** ** ** ** ** ** * ** * * ** ** ** ** * ** ** ** ** * ** ** Noes: 1. Three-varable sysem and four-varable sysem are represened by prce equaon 9 and 10 respecvely. 2. (**) and (*) denoes rejecon of he null hypohess a 5% and 1% respecvely. 26

27 Table 7. F-Sascs For The Sgnfcance Of The Fnancal Varables In The Three Varable Real Income Equaon Sample Perod Fnancal Varables LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 F-sas Prob F-sas Prob F-sas Prob F-sas Prob *** *** *** *** * *** *** ** ** * *** ** Noes: 1. The Error Correcon Models were esmaed wh a provson of hree lags and he lag lengh was deermned afer subjecng each varable o a seres of ARCH seral correlaon LM ess. 2. The F-sascs was obaned from he Wald Tes and was esed for he null hypohess ha all he coeffcens of he fnancal varable ndcaed are zero. 3. (***), (**) and (*) denoes he rejecon of he null hypohess a 1%, 5% and 10% respecvely. 27

28 Table 8. F-Sascs For The Sgnfcance Of The Fnancal Varables In The Four Varable Real Income Equaon Sample Perod Fnancal Varables LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 F-sas Prob F-sas Prob F-sas Prob F-sas Prob *** *** *** *** ** *** ** ** *** ** Noes: 1. The Error Correcon Models were esmaed wh a provson of hree lags and he lag lengh was deermned afer subjecng each varable o a seres of ARCH seral correlaon LM ess. 2. The F-sascs was obaned from he Wald Tes and was esed for he null hypohess ha all he coeffcens of he fnancal varable ndcaed are zero. 3. (***), (**) and (*) denoes he rejecon of he null hypohess a 1%, 5% and 10% respecvely. 28

29 Table 9. F-Sascs For The Sgnfcance Of The Fnancal Varables In The Three Varable Prce Equaon Sample Perod Fnancal Varable LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 F-sas Prob F-sas Prob F-sas Prob F-sas Prob * * ** ** ** *** Noes: 1. The Error Correcon Models were esmaed wh a provson of hree lags and he lag lengh was deermned afer subjecng each varable o a seres of ARCH seral correlaon LM ess. 2. The F-sascs was obaned from he Wald Tes and was esed for he null hypohess ha all he coeffcens of he fnancal varable ndcaed are zero. 3. (***), (**) and (*) denoes he rejecon of he null hypohess a 1%, 5% and 10% respecvely. 29

30 Table 10. F-Sascs For The Sgnfcance Of The Fnancal Varables In The Four Varable Prce Equaon Sample Perod Fnancal Varable LRM1 LRM2 LRM3 LRCR1 LRCR2 LRCR3 ALR IBR3 TBR3 1973:4 o 2000:1 1973:4 o 1989:4 1978:4 o 2000:1 1990:1 o 2000:1 F-sas Prob F-sas Prob F-sas Prob F-sas Prob * * * ** ** * * ** *** Noes: 1. The Error Correcon Models were esmaed wh a provson of hree lags and he lag lengh was deermned afer subjecng each varable o a seres of ARCH seral correlaon LM ess. 2. The F-sascs was obaned from he Wald Tes and was esed for he null hypohess ha all he coeffcens of he fnancal varable ndcaed are zero. 3. (***), (**) and (*) denoes he rejecon of he null hypohess a 1%, 5% and 10% respecvely. 30