TAI KUI GUAN AIA Berhad. HOOY CHEE WOOI Universiti Sains Malaysia ABSTRACT

Transcription

1 7 Jurnal Ekonom Malaysa 51(), Does Marke Inegraon Promoe o Frm Informaon Effcency? Emprcal Evdence for Malaysa Lsed Frms (Adakah Inegras Pasaran Menggalakkan Kecekapan Makluma Frma? Buk Emprk unuk FIrma-Frma Tersenara) TAI KUI GUAN AIA Berhad HOOY CHEE WOOI Unvers Sans Malaysa ABSTRACT Global nformaon ransms fas no share prce f a company s hghly negraed wh he world marke. On he conrary, a segmened frm ha has hgher negraon wh local marke s expeced o have slow response o global nformaon. Ths paper esed wheher he above s rue for lsed frms n Malaysa, a fas-emergng marke ha consss of large pool of boh negraed and segmened frms. The degree of marke negraon s measured by he varance raos approach proposed by Akdogan (1997) whle nformaonal effcency s capured by he prce delay measure suggesed by Hou and Moskowz (005). We analyzed a panel of 65 frms from 1995 o 010 and found ha frms wh hgher negraon wh world marke have beer global nformaon effcency whle segmened frms were sgnfcanly slow o mpound global news. Our resuls are robus o he choce of dfferen proxes of nformaonal effcency as well as dfferen conrol varables and a varey of alernave regresson specfcaons. Our fndngs mply ha he polcymakers should consder furher lberalze he Malaysan sock marke o ncrease s ransparency and reduce nformaon barrers so ha nformaon effcency can be mproved and conrbue o economc growh agenda. Keywords: Informaonal effcency; Marke negraon; Prce delay; Varance raos; fnancal crss ABSTRAK Makluma duna melnas panas ke dalam harga saham jka sesebuah syarka sanga ernegras dengan pasaran duna. D sebalknya, frma bersegmenas yang mempunya negras yang lebh ngg dengan pasaran empaan djangka membalas lebh perlahan erhadap makluma duna. Keras n menguj sama ada perkara d aas adalah benar bag frma-frma ersenara d Malaysa, sebuah pasaran yang pesa berkembang yang erdr darpada sekumpulan besar frma bernegras dan frma bersegmenas. Tahap negras pasaran dukur dengan pendekaan nsbah varans yang dcadangkan oleh Akdogan (1997) manakala kecekapan makluma dukur dengan kelewaan harga yang dcadangkan oleh Hou dan Moskowz (005). Kam elah menganalsa sau panel 65 frma dar ahun 1995 hngga 010 dan mendapa bahawa frma yang mempunya negras yang lebh ngg dengan pasaran duna mempunya kecekapan makluma duna yang lebh bak manakala frma-frma bersegmenas ama lamba dalam penyerapan makluma duna. Kepuusan kam adalah manap dalam plhan proks kecekapan makluma dan pembolehubah kawalan yang berbeza sera pelbaga spesfkas regres alernaf. Penemuan kam mengmplkaskan bahawa pembua dasar perlu mempermbangkan unuk melberalsaskan lag pasaran saham Malaysa bag menngkakan keelusan dan mengurangkan halangan makluma supaya kecekapan makluma dapa dperngkakan dan menyumbang kepada agenda perumbuhan ekonom. Kaa kunc: kecekapan makluma; negrase pasaran; kelewaan harga; nsbah varans; krss kewangan INTRODUCTION Over he pas decades, fas pace of marke deregulaon, lberalzaon, and globalzaon, couple wh advanced nformaon echnology have led o ncreasngly negraon of emergng marke no he world marke. Ths s refleced n remendous grow n drec and porfolo nvesmens no emergng marke, as well as acve foregn lsng and acquson acves. Emprcal evdence suggess ha when frms are hghly negraed wh he world marke, hey end o enjoy

2 8 Jurnal Ekonom Malaysa 51(), lower prce volaly, lower cos of capal, and ncreasng real nvesmen opporunes. All hs has spurred local producvy and conrbue o economc growh (Bekaer e al. 005; Bekaer & Harvey 000; Henry 000; Mon 006; Ta 007). The speed of nformaonal ransmssons s expeced o mprove under a more globally negraed sock marke. When a marke s more negraed wh he world, here wll be a hgher complance o nernaonal norms n erms of corporae governance, more lberalzed n foregn equy ownershp, srngen dsclosure rules, and more compeve prcng (Hooy & Lm 013). Smlarly, a he frm level, he dssemnaon of global nformaon s expeced o be faser for he more globally negraed frm han hose segmened frms. If a frm s hghly negraed wh he world marke, he qualy and speed of nformaon arrval s expeced o become beer and faser, respecvely. Ths wll leads o hgher degree of nformaonal effcency a frm level. Bascally, frm level nformaon effcency s sensve o he exen how frm has dversfed s global sysemac rsk. On he oher hand, f frm s no negraed wh he world marke, one would expec he frm o be sluggsh n global nformaon effcency and response slower o global news bu response relavely faser o local news. Snce mos of he emergng markes are less lberalzed, he absence of foregn equy ownershp and parcpaon n local sock marke mgh dmnshes he benefs from nernaonal nformaon sharng, hence, some of he local frm are exposed more o local marke shocks. Albuquerque e al. (009) has developed a heorecal model o hghlgh he nformaon asymmery beween local and global nvesors n acqurng global nformaon causes local nvesors o underreac o he movemens of global facors. Socks ha are no accessble o global nvesors are unlkely o ncorporae global nformaon no s prces n a mely and accurae way. Emprcally, Bae e al. (01) has esed and renforced he emprcal valdy of hs nformaon asymmery hypohess on marke lberalzaon process. Ther paper examned he mpac of marke lberalzaon (proxy by degree of nvesbly) on prce delay wh respec o local and global markes nformaon, 1 and found ha marke frcons ncludng resrcon on foregn equy ownershp n emergng counres are lkely o mpede swf processng of global marke nformaon. The paper concludes ha removal of ownershp resrcons could mproves he degree of nformaonal effcency of emergng sock markes, n oher words, marke negraon can promoe nformaon effcency. Ths paper focuses on a small and open emergng counry, Malaysa. Bursa Malaysa, he Malaysan sock marke, s one of he rapdly growng and leadng equy marke n emergng counres. Indeed, Malaysa was clamed as one of he hghly negraed marke wh he world (see Bekaer & Harvey 1995; Ta 007). As Malaysa s a small and open economy ha based heavly on exernal rade clmae, he Malaysan publc lsed frms hence are hghly expose o global nformaon. Wh he erupon of he 1997 Asan fnancal crss and he mplemenaon of he capal conrol and fxed exchange rae polcy n 1998, he Malaysan equy marke has had a dynamc negraon wh he world marke over he one and a half decade. The peggng was lfed n July 005, bu hen here was he oubreak of he US subprme crss and European soveregn deb crss n he lae 000s. Hence, Malaysa s frms can serve as good plaforms o examne how global nformaon was mpounded a he frm level gven he varaon n her marke negraon over me. To es frm level nformaon effcency, we used he prce delay proposed by Hou and Moskowz (005). We also followed he approach of Akdogan (1997) o measure he degree of frm level negraon wh local and global sock markes. Fnally, we employed panel regresson model o deermne he assocaon beween hese wo conceps. We documened emprcal evdence o suppor he negraon-effec hypohess where he delay o global marke nformaon was negavely relaed o frm negraon wh he world marke; whle frm negraon wh he local marke s posvely assocaed wh he prce delay. The resuls were robus across dfferen panel esmaons, alernave proxes for prce delay proposed by Hou (007), and added conrol varables on fnancal crses. Overall, our sudy suggesed ha polcy effors am o remove boh drec and ndrec barrers ad o enhance nformaonal effcency of Malaysan publc lsed frms. Our conrbuon s wofold. Frs, we provde he frs emprcal evdence a frm level o esablsh he mpac of marke negraon on nformaon effcency. To dae, here s no such emprcal work a frm level albe Hooy and Lm (013) has provdes he frs emprcal evdences on he lnk beween marke nformaon effcency wh marke negraon a he aggregae marke level. Second, we adoped he negraon measure a marke level proposed by Akdogan (1997) o consruc frm level measures ha dsngushes he marke effcency process of globally negraed frms versus he locally negraed frms. As Hooy and Lm (013) was usng counry level daa, hey only measure world marke negraon on marke effcency. The remander of he paper s organzed as follows. Secon demonsraes leraure revew of hs sudy. Secon 3 mehodology and he sample of hs sudy. Secon 4 addresses he resuls and man fndngs. Secon 5 presens robusness analyss and Secon 5 concludes. LITERATURE REVIEW The rrelevan assumpons of he convenonal asse prcng heory assumed ha he marke s frconless and nvesors are well dversfed. These assumpons had lmed he ably of researchers o nvesgae he relave effcency of sock markes. A breakhrough was acheved by Hou and Moskowz (005), who proposed he prce delay measure o

3 9 Jurnal Ekonom Malaysa 51(), measure he effec of marke frcons on cross-seconal reurn predcably. Hou and Moskowz (005) sugges ha ransacon coss, nformaon coss and nsuonal force have delayed he sock prce adjusmen process. In he exan leraure, he prce delay has been appled o nvesgae he assocaon beween speed of sock prce adjusmen and marke frcons (Hou & Moskowz 005), accounng qualy (Callen e al., 010), opon prcng (Blau & Brough 011; Phllps 010), shor sales (Boehmer & Wu 010; Chen & Rhee 010; Saff & Sgursson 011), marke lberalzaon (Bae e al., 01; Chang e al. 008) and marke negraon (Hooy & Lm 01). Many sudes also examne he deermnans of prce delay. For example, Bae e al. (01) nvesgaes he effec of marke lberalzaon on asymmery nformaon marke effcency for 1 emergng sock markes. The degree of asymmerc nformaon effcency s proxed by he prce delay of Hou and Moskowz (005) whle he marke lberalzaon (or nvesbly) s proxed by degree open facor. They furher classfed he socks no hree nvesable groups based on he nvesbly wegh, e.g. non-nvesable sock (foregners own zero percen of local socks), parally nvesable socks (foregners ownershp up o 50 percen) and hghly nvesable socks (foregners ownershp more han 50 percen). They documened a posve assocaon beween nvesable socks and he speed of sock prce adjusmen o new global marke nformaon afer conrol for frm sze, urnover, volaly and analys coverage. Bae e al. (01) concluded ha he parcpaon of global nvesors n local sock markes have faclaed he dffuson of global marke nformaon no local sock prces and hs s conssen wh he heory of asymmerc nformaon of Albuquerque e al. (009). Overall, he fndngs from Bae e al. (01) research ndcae ha he openness of he local marke o foregn nvesors or lberalzaon had mproved he overall nformaonal effcency of local markes. Nex, prce delay also has been appled o examne he relaonshp beween shor sales consrans and prce dscovery process a he frm-level. Among he popular sudes nclude Chen and Rhee (010), Boehmer and Wu (010) and Saff and Sgursson (011). On op of ha, Chen and Rhee (010) examned he effec of shor sales on Hong Kong Sock Exchange (HKSE) effcency by usng he prce delay measure. They found ha shorable sock prces were faser o adjus o new marke-wde nformaon han nonshorable socks. The esmaed resuls were sgnfcan even he sock marke s n bull me and afer conrol for frm characerscs lke radng volume, frm sze, lqudy and share prces. 3 Smlar o Chen and Rhee (010) sudy, Boehmer and Wu (010) nvesgaed he effec of shor sellng on prce delay. The sample sudy consss of socks lsed on New York Sock Exchange (NYSE) and he analyss was based on hgh frequency daa, daly daa se from January 005 o June 007. The esmaed resuls were suppored by Chen and Rhee (010) research, suggesng ha shor sellng has conrbued o more effcen sock prcng n developed marke. Saff and Sgursson (011) also found smlar fndngs where here was a posve relaonshp beween shor sales and prce delay measure. Besdes, he prce delay also has been exended o nvesgae he ssue of accounng qualy and nformaonal effcency. Callen e al. (010) has examned he effec of accounng qualy on prce delay processes for socks lsed n CRSP across he perod of 1981 o 006. They pos ha poor accounng qualy has a posve relaonshp wh prce delay measure and hus, nvesors demand hgher fuure reurns o compensae for poor accounng qualy. The accounng qualy was proxed by accrual qualy, earnngs surprses, loss frequency and specal ems. Followng Hou and Moskowz (005) model, Callen e al. (010) documened ha poor accrual qualy, more frequen loss socks and large negave specal ems end o exhb he posve relaonshp wh he prce delay measures. Conssen wh he prce delay leraure, fndngs from Callen e al. (010) sudy has suggesed ha he level of accounng qualy sgnfcanly affecs he degree of nformaonal effcency n any frms. Mos of he emprcal sudes on prce delay dscussed so far had focused on frm-level analyss. However, Lm and Hooy (010) s he frs sudy o examne he speed of sock prce adjusmen o new marke-wde nformaon by usng he aggregae counry level daase. Ther fndngs were conssen wh prce delay leraure where emergng counres end o exhb he hgher values of delay measures han developed counres. The larger of prce delay ndces n emergng markes ndcae ha hese markes are slower n ncorporang global news no local sock prces. Moreover, hey also documened ha small marke sze, resrcons on foregn equy ownershp, lower radng volume and shor sales resrcon had mpeded he sock prce adjusmen processes. Subsequenly, Hooy and Lm (013) nvesgae wheher a more negraed sock marke s assocaed wh he prce delay measure as a proxy for nformaon effcency. They employed he adjused prcng errors from an equlbrum nernaonal asse prcng model as a proxy for marke negraon o explan he aggregae counry-level prce delay as nformaonal effcency measures. Usng daa from 49 counres, hey documened robus evdence supporng he hypohess ha markes more negraed wh he world are also more effcen, bu hs posve assocaon s only sgnfcan n he sub-sample of emergng sock markes. The paper do no provdes frm level analyss on he ssue. Ths s he research gap ha our paper s conrbued o fll up. METHODOLOGY AND DATA ASYMMETRY INTEGRATION AND THE SPEED OF STOCK PRICE ADJUSTMENT

4 30 Jurnal Ekonom Malaysa 51(), We es he effec of asymmery marke negraon on nformaonal effcency va he panel regresson model. In lne wh he leraure n prce delay, we conrol he effecs of oher deermnan varables. We nroduce four common deermnans for prce delay whch are he degree of foregn nvesbly, frm sze, frm urnover, and frm volaly. Adoped from Hooy and Lm (013), our emprcal model s shown below: DELAY 1INTEGM, INTEGW, 3Conrol, (1) where DELAY s he frm specfc prce delay a me. INTEGM, and INTEGW, are frm negraon ndexes wh local and world marke, respecvely. Followng he leraure n prce delay research our conrol varables are marke nvesably, frm sze, frm lqudy proxy by urnover and frm volaly. Marke nvesably s calculae by he proporon of foregn nvesors shareholdngs; frm sze s he naural log f marke capalzaon; urnover s number of shares raded over he year scaled by he number of shares ousandng; and frm volaly s he naural logarhms of sandard devaon of weekly sock reurns n a year. The symbol s frm fxed-effec whch capure all he menvaran frm level characerscs; s he me-effec whch represen common shocks; whle, denoes he whe nose resduals. PRICE DELAY We use a wo-facor marke model o gauge he speed of sock prces adjusmen o global nformaon so ha we can conrol for local marke reurns. The annual prce delay ndex of each frm s obaned by regressng he frm weekly reurns wh he resrced four weeks lagged reurns on boh he local and world marke porfolos over he fscal year. Four weeks lagged marke reurns are chosen because s consdered a far amoun of me for a frm o respond o new marke-wde nformaon (Hou & Moskowz 005). The reurns generang model s shown n he followng equaon: 4 4 ( n) ( n) r, rm, rm, n rw, rw, n, n1 n1 () where r, denoes he reurn of frm a week. r M, and r W, represen he KLCI and MSCI weekly value-weghed ndex reurns n week, respecvely. rm, n and rw, n are KLCI and MSCI weekly value-weghed lagged ndex reurns, respecvely for n= 1,, 3 and 4. he projecon of local marke porfolo on world porfolo. So denoes he whe nose error erms. Local marke porfolo ), M ( r M, s obaned from r, represens he par of reurns varaon n local marke ha s no able o capure by r W,. Hence, r, s orhogonal o r W,. M We follow Hou and Moskowz (005) o consruc he prce delay ndexes, akng one mnus he rao of ( n) resrced model 0, n [1,4 ] over R from unresrced model gven below: R DELAY 1 ( n ) 0, n[1,4] R (3) R from The larger he value of he DELAY mples a greaer delay n he response of frm o world and local marke-wde news. Bascally, he delay seres s an nverse measure of marke effcency. ASYMMETRY MARKET INTEGRATION- THE VARIANCE RATIO MODEL We follow Akdogan (1997) o decompose he sysemac rsk no componens o measure he dfferen negraon process of frms wh local marke (KLCI), and negraon wh world marke (MSCI). 4 Hence, he ssue of asymmery negraon of local versus global markes s addressed va he wo-ndex reurn generang process: r [ r r ] B [ r r (4) r, F, M M, F, W W, F, ],

5 31 Jurnal Ekonom Malaysa 51(), r rf, r r, r r, where, s he frm excess reurns a week, M, F and W, F denoe he local and global marke rsk premums a me. The epslon (, ) represens he whe nose error erms. The varance of sock can be decomposed no hree componens by dvdng boh sdes of Equaon (4) wh he varance of r,. A B C 1 (5) where A B C M var( rm ) INTEGM var( r ) (6) W var( rw ) INTEGW var( r ) (7) var( ) var( r ) (8) A and B measure he proporon of local and world sysemac rsks conrbuon o frm s varaon whle measures he proporon of frm s unsysemac rsk. Inuvely, he greaer he value of A ndcaes ha frm s drven by local sysemac rsk. In conras, f frm s more negraed wh world sock marke, one should expec he value for B o be greaer han A. DATA DESCRIPTION The weekly closng prces for ndvdual frms are rereved form Thomson DaaSream, whle daa for foregn equy ownershp are obaned from Bursa Malaysa. The analyss s based on weekly US dollar reurns. We oban weekly closng prces (Wednesday) of ndvdual sock, annual seres of marke capalzaon, and radng volume from DaaSream. Frm volaly s proxed by he sandard devaon of sock reurns whn a year. The sample sudy ncludes socks ha are lsed snce January 1995 and sll survved unl December 010 (16 years n oal). Ths perod covered all he epsodes of fnancal crses of he las decade, from he Asan Fnancal crss o Do-Com Bubble and US Subprme crss. We resrced our sample o acve frms only and excluded hose suspended and delsed. We also excluded observaons whch closng prces are eher zero or are mssng (Bae, e al. 004). We merged hese frms-level daa from DaaSream wh foregn ownershp annual daa obaned from Bursa Malaysa. Afer hese flraon, we were lef wh 65 frms. World sock marke reurn s proxed by he value-weghed Morgan Sanley Capal Inernaonal (MSCI) All Counry World Index. Local sock marke reurns are capured by he value-weghed FTSE Bursa Malaysa Kuala Lumpur Compose Index (KLCI). Moreover, he world and local markes rsk free raes are proxy by he US 3-monh Treasury bll raes and Malaysa Inerbank 3-monh rae (quoed by Maybank), respecvely. C RESULTS AND DISCUSSIONS Table 1 presens he descrpve sascs. The mean and medan of DELAY are and , respecvely, mplyng ha Malaysan frms are n general, sluggsh n response o global news, as he average delay value for world marke level was esmaed wh a mean of only, whle for Malaysa marke level he mean was abou 0.5. In erms of he marke negraon, Malaysan frms are more negraed wh local marke nsead of world marke where he mean for INTEGM s greaer han INTEGW. Nex, he percenage of foregn equy ownershp (INVEST) vares consderably across frms and he range s zero o 87.19% wh a mean of 18.3%. Ths suggess ha Malaysa s a favourable desnaon for global nvesors for her porfolo dversfcaon. Smlarly, he values of marke capalzaon (SIZE) also vary sgnfcanly across frms, rangng from USD 0.57 mllon o USD 1, mllon wh mean USD mllon. Tradng volume (VOLUME) and volaly of sock reurns (VOLATILITY) also show wde cross-seconal varaon. We check he correlaon marx and fnd ha he parwse correlaon of our explanaory varables level are less han 0.5, safely exclude us from mulcollneary ssue. The resul s as repored n Table.

6 3 Jurnal Ekonom Malaysa 51(), TABLE 1. Descrpve sascs of man varables Sandard Mean Medan Devaon Maxmum Mnmum DELAY INTEGM INTEGW INVEST (%) SIZE (mllon of USD) TURNOVER (mllon of USD) VOLATILITY (%) Noe: Ths able presens descrpve sascs of man varables over he perod of The oal number of observaon s 440. DELAY s consruced based on Hou and Moskowz (005) model. INTEGM and IMTEGW are negraon wh local and world sock markes, respecvely. INVEST s he proporon of foregn equy ownershp on local socks. SIZE s he marke capalzaon of local sock n mllons of USD. VOLUME represens he number of shares raded (n mllon) n a year. VOLATILITY denoes he sandard devaon of frm weekly reurns n a year. DELAY INTEGM TABLE. Correlaon marx beween delay measure and s deermnan varables DELAY INTEGRM INTEGRW INVEST SIZE TURNOVER VOLATILITY INTEGW INVEST SIZE TURNOVER VOLATILITY Noe: Ths able repor he correlaon coeffcens among he varables. The oal number of observaon s 440. Fgure 1 shows he rend of average frm DELAY, INTEGM and INTEGW over he sample perod. I s clear ha he degree of nformaonal effcency for frm lsed n Bursa Malaysa has mproved sgnfcanly over he years afer he Bursa Malaysa embarked on a seres of lberalzaon reforms n he early 000s as a resul of he 1997/98 Asan fnancal crss. We can see also INTEGM has sared o declne afer 1998 bu was dsruped by he 001 US do.com bubble burs, bu he downrend connues afer ha. In fac, pror o 009, INTEGM s hgher han INTEGW, suggess ha Malaysan frms are generally more segmened and less negraed wh he world marke. However, over me, he gap has becomng narrow snce 001 unl 009 when more frms are acually negraed wh he world han he local marke. Perhaps he 001 US do.com crss has nroduced a srucure change n world sock marke negraon. We wll accoun for hs possbly n our robusness secon. Anoher neresng observaon s durng fnancal crses of , 001, and , INTEGM reaches s peaks whle INTEGW suffered a sgnfcan drop (excep he 001 case). Ths provdes us some emprcal ground o nvesgae furher n robusness es o conrol for fnancal crses n our model n he laer secon of hs paper.

7 33 Jurnal Ekonom Malaysa 51(), FIGURE 1. Evoluon of Marke Inegraon and Prce Delay over Tme. Table 3 repors he deermnans of prce delay usng hree dfferen esmaon mehods;.e. pooled OLS, wo-way fxed effecs and wo-way random effecs models. INTEGW has a negave sgn and s sascally sgnfcan wh prce delay n hese models. The negave relaon suggess ha when Malaysan frms are negraed wh he world, he degree of nformaonal effcency wh respec o new global marke nformaon have mproved where he frms prces are faser n response o world news, or a lower value of prce delay. On he oher hand, INTEGM shows a posve and srong sgnfcan relaonshp wh delay measure across hese dfferen models. Ths ndcaes ha Malaysan frms are sluggsh n response o new global marke nformaon when s negraed wh local marke. Thereby, he fndngs are conssen wh pror evdence ha he removal of legal resrcons and reduce counry-level barrers wll mproved foregn drec nvesmen n local capal marke (Hooy and Lm, 013). In erms of conrol varables, INVEST, SIZE, TURNOVER, and VOLATILITY have showed negave relaonshp wh prce delay under he pooled OLS model. In he wo-way fxed effecs seng, hough all he conrol varables have negave sgns wh prce delay, hey are sascally nsgnfcan a all levels. Nex, INVEST and TURNOVER reman negavely and sgnfcanly assocaed wh prce delay n he wo-way random effecs model. We check he model specfcaon ess usng redundan fxed-effec and Hausman ess. Based on he es sascs, he pooled OLS and wo-way random effecs models were rejeced. Ths mples ha afer consders he possbly of heerogeney among Malaysan frms and possbly omed me-dependen varables, he wo-way fxed effec model s he mos effcen n explanng he deermnan varables of prce delay n Malaysan sock marke. TABLE 3. Deermnans of Prce Delay Pooled OLS Fxed Effecs Random Effecs INTERCEPT *** *** 0.538*** (0.0000) INTEGM *** 0.597*** *** (0.0000) INTEGW *** *** *** (0.000) INVEST *** ** (0.0000) (0.4457) (0.07) SIZE *** (0.0000) (0.9330) (0.393) TURNOVER *** * (0.0013) (0.8538) (0.0408) VOLATILITY *** (0.0000) (0.581) (0.986) Frm Dummes No Yes Yes

8 34 Jurnal Ekonom Malaysa 51(), Year Dummes No Yes Yes No. of Frms Observaons R Adjused R Redundan Fxed-Effec Tess Cross-Secon/Perod F *** (0.0000) Hausman Tess Cross-secon and perod random ** (0.010) Noe: Ths able presens he deermnans of prce delay. The dependen varable s based on Hou and Moskowz (005) model. The resuls are esed across hree dfferen models, namely pooled OLS model, wo-way fxed effecs model, and wo-way random effecs model. INTEGM and INTEGW are negraon wh local and world sock markes. NVEST s he proporon of foregn equy ownershp on local socks. SIZE s measured by naural logarhm of marke capalzaon of local sock n mllons of USD n a year. VOLUME s he logarhm of number of shares raded n a year. VOLATILITY s he sandard devaon of frm weekly reurns n a year. Numbers n (.) are p-value. H0 of he Redundan Fxed-Effec Tess s Pooled regresson s preferable han a Fxed Effec model; whle H0 of Hausman Tes s Random Effec s preferable han a Fxed Effec model. ROBUSTNESS CHECKS ALTERNATIVE PRICE DELAY The frs robusness check nvolves wheher our core fndngs are conssen across alernave proxy for dependen varable. For hs purpose, we use a logsc ransformaon of orgnal prce delay (LDELAY) nroduced by Hou (007). Accordng o Hou (007), he ransformed verson of prce delay ad o remove he excess skewness and kuross of he orgnal prce delay whle preservng s monooncy. Besdes, he delay ndexes also no beng resrced beween zero and one. Table 4 repors he esmaed coeffcens based on pooled OLS, wo-way fxed effecs and wo-way random effecs models. Agan, INTEGW and INTEGM show he expeced sgns and boh are sascally sgnfcan wh LDELAY across he hree models. Moreover, he model specfcaon ess have srongly rejeced he pooled OLS and wo-way random effecs models. Overall, hs robusness es confrms he fndng n prevous secon where he speed of sock prce adjusmen o global nformaon s explaned by boh negraon measures. TABLE 4. The Logsc Transformaon of Prce Delay Pooled OLS Fxed Effecs Random Effecs INTERCEPT 1.53*** 0.076* * (0.0000) (0.066) (0.060) INTEGM.7497***.9610***.961*** (0.0000) INTEGW *** *** *** (0.0000) INVEST *** * (0.0000) (0.53) (0.0507) SIZE *** (0.0000) (0.886) (0.3551) TURNOVER *** ** (0.0011) (0.7714) (0.0409) VOLATILITY *** (0.0000) (0.638) (0.6636) Frm Dummes No Yes Yes Year Dummes No Yes Yes No. of Frms Observaons R Adjused R Redundan Fxed-Effec Tess Cross-Secon/Perod F (0.0000) Hausman Tess Cross-secon and perod random (0.0000)

9 35 Jurnal Ekonom Malaysa 51(), Noe: Ths able presens he resuls from frs robusness ess. The orgnal prce delay s replaced wh he logsc ransformaon of prce delay (LDELAY) proposed by Hou (007). The esmaons are esed across pooled OLS model, wo-way fxed effecs model, and wo-way random effecs model. INTEGM and INTEGW are negraed wh local and world sock markes respecvely. NVEST s he proporon of foregn equy ownershp n local socks markes. SIZE s measured by naural logarhm of marke capalzaon of local sock n mllons of USD n a year. VOLUME s he logarhm of number of shares raded n a year. VOLATILITY s he sandard devaon of frm weekly reurns n a year. Numbers n [.) are Whe perod sandard error whle (.) are p-value. H0 of he Redundan Fxed-Effec Tess s Pooled regresson s preferable han a Fxed Effec model; whle H0 of Hausman Tes s Random Effec s preferable han a Fxed Effec model. CONTROLLING FOR THE IMPACT OF FINANCIAL CRISIS In he second robusness check, we conrolled he mpac of fnancal crss va dummy varable. Based on he leraure survey, he dummy varables for fnancal crss nclude Asan Fnancal Crss ( ), Do-Com Bubble ( ), and US Subprme Crss ( ). Table 5 repors he esmaed resuls usng he pooled OLS model. The resuls from column (1) hrough (4) conssenly show ha all he compeng varables ncludng INTEGW and INTEGM are sascally sgnfcanly wh DELAY and LDELAY. In addon, hese varables also exhbed he expeced sgns afer conrollng for he crses perods. The negave relaon beween Asan Fnancal Crss and US Subprme Crss wh prce delay mply ha unfavourable news durng he crss perods are spread nsananeously across Malaysan sock marke. TABLE 5. Conrollng for Fnancal Crses DELAY LDELAY CONSTANT INTEGM INTEGW INVEST (0.0003) (0.0006) SIZE (0.0000) (0.0001) TURNOVER (0.0051) (0.0059) VOLATILITY (0.0003) (0.0001) D D D R Adjused R Noe: Ths able presens he resuls from second robusness ess where we conrolled he effec of fnancal crss n pooled OLS model. DELAY s orgnal prce delay developed by Hou and Moskowz (005) whle LDELAY s he alernave delay by Hou (007). INTEGM and INTEGW are negraon wh local and world sock markes respecvely. NVEST s he proporon of foregn equy ownershp n local socks. SIZE s measured by naural logarhm of marke capalzaon of local sock n mllons of USD n a year. VOLUME s he logarhm of number of shares raded n a year. VOLATILITY s he sandard devaon of frms weekly reurns n a year. D9798 s he dummy varable for Asan Fnancal Crss. D0001 s he dummy varable for Do-Com Bubble. D0809 s he dummy varable for US Subprme Crss. Numbers n (.) are p-value.

10 36 Jurnal Ekonom Malaysa 51(), CONCLUSION In hs paper, we examned he effec of marke negraon on nformaonal effcency a he frm level. We conduced a seres of panel regresson modellng focusng on he degree of marke negraon of 65 frms lsed n Bursa Malaysa and he resuls changed n erms of sock prce effcency. We found ha nformaonal effcency was lkely o mprove for more globally negraed frm. The fndng was complemen o Hooy and Lm (013) whch documened smlar evdence a he aggregae counry level. A frms level, our resuls show ha more segmened or locally negraed frms have less world nformaon effcency. Overall, we found a sold emprcal evdence o clam ha marke negraon and nformaonal effcency are posvely assocaed, and so Malaysa sock marke was generally ncreasngly negraed wh he world. Our fndngs mply ha he polcymakers lke he Secures Commsson and he Mnsry of Fnance, should consder o furher lberalze he Malaysan sock marke o ncrease s ransparency and reduce nformaon barrers. Ths would ensure a hgher level of frm negraon can be acheved o mprove her nformaon effcency. Wh hgher nformaon effcency, hs mples beer resource allocaon and hence hgher producvy for he naon. For example, he removal of sauory foregn resrcons on nvesmen wll lkely o arac more foregn nsuonal nvesors. Ths wll surely mprove he corporae governance level of he lsed frms. Wh more foregn parcpans, local nvesors wll also have more access o hgher-qualy and more relable nformaon n he marke. On he oher hand, however, when marke s vulnerable o foregn news he regulaor should also need o ghen her survellance sysem and closely monor he radng acves n he marke n order o conrol for unhealhy speculaon ha s no based on fundamenal value. NOTES 1 Bae e al. (01) used he degree open facor o proxy for he degree of nvesbly. These daa were obaned from Sandard & Poor s Emergng Markes Daabase (EMDB). The degree of open faco s rereved from Sandard & Poor EMDB daabase. I s used o measures he amoun of a company s marke capalzaon ha a foregn eny can legally own. 3 Chen and Rhee (010) s he frs sudy o explore he role of shor sales on speed of sock prces adjusmen o marke-wde nformaon durng he up (bull) marke perod. 4 Barar (004) modfed he wo-facor model and appled o address he ssues on regonal versus global negraon. REFERENCES Akdogan, H Inernaonal secury selecon under segmenaon: Theory and applcaon. Journal of Porfolo Managemen: 8-9. Albuquerque, R.A., Bauer, G.H. and Schneder, M Global Prvae Informaon n nernaonal equy markes. Journal of Fnancal Economcs 94: Bae, K.H., Ozoguz, A., Tan, H. & Wrjano, T.S. 01. Do foregners faclae nformaon ransmsson n emergng markes? Journal of Fnancal Economcs 105 (1): Barar, M Equy marke negraon n Lan Amerca: A me-varyng negraon score analyss. Inernaonal Revew of Fnancal Analyss 13: Bekaer, G, Campbell, R. H. & Lundblad, C Does fnancal lberalzaon spur growh? Journal of Fnancal Economcs 77: Bekaer, G. & Harvey, C Tme-varyng world marke negraon. Journal of Fnance 50: Bekaer, G. & Harvey, C Foregn speculaors and emergng equy markes. Journal of Fnance 55: Blau, B.M. & Brough, T.J Opons and Marke Frcon. SSRN workng paper, avalable a: hp://ssrn.com/absrac= Boehmer, E. & Wu, J Shor sellng and he prce dscovery process. SSRN workng paper, avalable a: hp://ssrn.com/absrac=9760. Callen, J.L., Khan, M. & Lu, H Accounng Qualy, Sock Prce Delay and Fuure Sock Reurns. SSRN workng paper avalable a hp://ssrn.com/absrac= Carrer, F., Chaeb, I. & Errunza, V.R Do mplc barrers maer for globalzaon? SSRN Workng Paper avalable a hp://ssrn.com/absrac= Chen, C.X. & Rhee, S.G Shor sales and speed of prce adjusmen: Evdence from he Hong Kong sock marke. Journal of Bankng and Fnance 34: Chang, C.T., Nellng, E. & Tan, L The speed of adjusmen o nformaon: evdence from he Chnese sock marke. Inernaonal Revew of Economcs and Fnance 17: Henry, P.B Do sock marke lberalzaon cause nvesmen booms? Journal of Fnancal Economcs 58:

11 37 Jurnal Ekonom Malaysa 51(), Hooy, C.W. & Lm, K.P Is marke negraon assocaed wh nformaonal effcency of sock markes? Journal of Polcy Modellng 35: Hou, K Indusry nformaon dffuson and he lead-lag effec n sock reurns. Revew of Fnancal Sudes, 0: Hou, K. & Moskowz, T.J Marke frcons, prce delay, and he cross-secon of expeced reurns. Revew of Fnancal Sudes 18: Lm, K.P. & Hooy, C.W The delay of sock prce adjusmen o nformaon: A counry-level analyss. Economcs Bullen 30: Mon, T Sock marke lberalzaon and operang performance a he frm level. Journal of Fnancal Economcs 81: Phllps, B Opons, Shor-Sale Consrans and Marke Effcency: A New Perspecve. SSRN workng paper, avalable a: hp://ssrn.com/absrac= Saff, P.A.C. & Sgurdsson, K Prce effcency and shor sellng. Revew of Fnancal Sudes 4: Ta, C.S Marke negraon and conagon: evdence from Asan emergng sock and foregn exchange markes. Emergng Marke Revew 8: Ta Ku Guan AIA Berhad Senor Assocae Parnershp Dsrbuon No 19, Jalan Kelawa, 1050 Penang MALAYSIA Emal: kuguan.a@aa.com Hooy Chee Woo* School of Managemen Unvers Sans Malaysa USM Pulau Pnang MALAYSIA Emal: cwhooy@usm.my * Correspondng Auhor