Linkage between Stock Market Prices and Exchange Rate: A Causality Analysis for Pakistan

Transcription

1 The Pakisan Develomen Review 4 : 4 Par II (Winer 2004) Linkage beween Sock Marke Prices and Exchange Rae: A Causaliy Analysis for Pakisan MOHAMMAD TAHIR FAROOQ and WONG WING KEUNG * 1. INTRODUCTION Globalisaion and financial secor reforms in develoing economies have ushered in a sea change in he financial archiecure of he economies. In he conemorary scenario, he aciviies in he financial markes and heir relaionshis wih he real secor have assumed significan imorance. Corresondingly, researches are also being conduced o undersand he curren working of he economic and he financial sysem in he new scenario. Ineresing resuls are emerging aricularly for he develoing counries where he markes are exeriencing new relaionshis which are no erceived earlier. The analysis on sock markes has come o he fore since his is he mos sensiive segmen of he economy. The sock markes of emerging economies have been of vial imorance o he global invesmen communiy. Since emerging markes are more volaile han he well develoed sock marke, herefore he emerging markes end o be unrelaed o one anoher and wih he develoed markes. Numerous invesors worldwide selec o diversify heir funds across he emerging markes. Sock markes of emerging economies have recenly been of vial imorance o global invesmen communiy. The caialisaions, reurns and volailiy have increased dramaically in hese markes. Since emerging markes are more volaile han he well-develoed sock markes, herefore, he emerging markes end o be unrelaed wih one anoher and wih he develoed markes. Numerous invesors worldwide selec o diversify heir funds across he emerging markes in order o minimise orfolio risk. In recen years financial crises semming from sudden and unexeced oscillaion in sock and foreign exchange markes has become a common henomenon in emerging economies. This realisaion rods he researches o invesigae he relaionshi beween sock marke erformance and exchange rae. Mohammad Tahir Farooq is Senior Subjec Secialis in Economics, Dearmen of Educaion, Governmen of he Norh Wes Fronier Province, Peshawar. Wong Wing Keung is Associae Professor, Dearmen of Economics, Naional Universiy of Singaore, Singaore.

2 640 Farooq and Keung Bahmani-Oskooee and Sohrabian (1992) is among he firs o use coinegraion and Granger causaliy o exlain direcion of movemen beween exchange rae and sock rices. Abdalla and Murinde (1997) finds ou ha for India, Korea and Pakisan, exchange rae Granger cause sock rices. Yu (1997) analyses his relaionshi for Hong Kong, Tokyo and Singaore. Amare and Mohsin (2000) examines his associaion beween exchange raes and sock rices for nine Asian counries. Mukherjee and Naka (1995) and Ajayi and Mougoue (1996) show ha sock marke rices are coinegraed wih exchange raes in eigh indusrial economies. Kououlas and Kryzanowski (1996) furnishes evidence ha sock marke volailiy significanly resonds o exchange rae volailiy in Canada. Kearney (1988) discovers similar resuls for Ireland. Bahmani-Oskooee and Sohrabian (1992) oins ou ha here is wo-way relaionshi beween US sock marke and exchange rae. Smih (1992) demonsraes ha sock reurns have significan influence on exchange rae in Germany. Fang (2000) ges a negaive dereciaion effec in he sock reurn rocess in Taiwan over Asian crisis. Mos of research on sock rices and exchange rae concenraes on develoed markes. Our aim in his aer is o comlemen he exising lieraure by examining he linkage beween sock rices and exchange raes in emerging Pakisan sock marke. Pakisan rovides an ineresing arena o invesigae inerrelaions beween sock and exchange markes for hree reasons. Firs Karachi Sock Exchange () is one of he faser growing emerging markes in he region as well as in he emerging economies. Marke caialisaion has increased in recen years. In January 199, according o, he marke caialisaion value was Rs millions while a he end of 200; on he oher hand he marke caialisaion has increased o Rs millions. An increase of 47 ercen is winessed in marke caialisaion during he eriod of his sudy. Secondly, Pakisan s economy observed financial crises in recen years. The sock and foreign exchange markes suffered flucuaions from he financial crisis, which broke ou in 1998 due o nuclear deonaion and Eas Asian financial crises. For examle, sock general index droed from a he beginning of July 1997 o a he end of Augus Surrisingly rare emirical research has examined he ineracion beween sock and exchange markes in Pakisan. Alhough a few sudies invesigae Karachi sock marke under he ambi of cross counry analysis. None of hem exclusively examined he causaliy relaionshi beween sock rices and exchange raes for Karachi Sock Exchange. Thirdly, his aer invesigae, for he firs ime, he relaionshi beween sock rices and exchange rae by decomosing Karachi Sock Exchange ino hree sub-secors. The henomenon of sock marke and exchange rae nexus received subsanial aenion afer Eas Asian financial crises. Theoreically, if here is any linkage beween boh of he variables hen he crises can be avered eiher by managing exchange rae or by adoing indigenous olicies o sable he sock

3 Linkage beween Sock Marke Prices and Exchange Rae 641 marke. Moreover he invesors can uilise his relaionshi beween sock rices and exchange rae o redic he behaviour of hese variables. There is lack of heoreical consensus on he relaionshi beween exchange rae and sock rices. For insance, orfolio balance model aroach of exchange rae deerminaion and sock rices, furnishes boh osiive and negaive relaionshis beween sock rices and exchange rae and ha causaion runs from sock rices o exchange raes. In hese models individuals holds foreign and domesic asses, including currencies, in heir orfolio. Exchange rae lay key role in balancing he demand and suly of asses. To buy more domesic asses local invesors would sell foreign asses (hey are relaively less aracive now), causing local currency areciaion. An increase in wealh due o rise in domesic asse rices will also lead invesors o increase heir demand for money, which inern rises domesic ineres rae. This again leads o areciaion of domesic currency by aracing foreign caial. Anoher channel of his relaionshi is an increase in he foreign demand of domesic asses due o sock rice increase. This will also cause a domesic currency areciaion. In conras, a osiive relaionshi beween sock rices and exchange rae wih direcion of causaion running from exchange rae o sock rices is because when domesic currency dereciaion makes local firms more comeiive, leading o an increase in heir exors, his in urn rises he sock rices. A weak or no associaion beween sock rices and exchange rae can also be osulaed. The asse marke aroach o exchange rae deerminaion reas exchange rae o be he rice of an asse (rice of one uni of currency). Therefore, like he rices of oher asses he exchange rae is deermined by execed fuure exchange raes. Any news/facors ha effec fuure values of exchange raes will effec oday s exchange rae. The facors/news ha causes changes in exchange rae may be differen from he facors ha causes change in sock rices. Under such scenario, here should be no link beween sock rices and exchange rae. The causaion beween he wo variables may be uni-direcional or bi-direcional. So i is clear ha here is lack of heoreical and emirical consensus on his relaionshi and he direcion of causaion. However, he linkage beween he wo markes is imoran. This aer rovides emirical evidence beween sock indices and exchange rae for Karachi Sock Marke, using monhly daa by emloying coinegraion aroach o examine he relaionshi. The res of he aer is organised as follow: in he nex secion he ime series mehod is briefly resened. Secion hree describes he daa and rovides he emirical analysis. Las secion conain conclusion. 2. TIME SERIES METHODS To analyse he relaionshi beween sock indices and exchange rae, his aer focuses on causaliy among hese variables using he mehod adaed by

4 642 Farooq and Keung Granger (1969). Formally a ime series x Granger causes anoher ime series y can be rediced beer by using as values of (X, Y ) han by using only he lag values of Y. In oher words, X fails o Granger-cause Y if all m>0 he condiional robabiliy disribuion of Y +m given (Y,Y 1 ) is he same as he condiional robabiliy disribuion of Y +m given boh (Y,Y 1 ) and (X, X 1 ). Tha is, X does no Granger-cause if Pr ( Y + m ψ = Pr ( Y + m Ω ) (1) where Pr( ) denoes condiional robabiliy, Ψ is he informaion se a ime on as values of Y and Ω is he informaion se conaining values of boh X and Y uo ime. Tesing causaliy beween wo saionary series X and Y can be based on he following bivariae auoregression: Y αky k + βk X k + k = 1 k = 1 =α0 + u (2) ϕky k + Φk X k + k = 1 k = 1 X =ϕ0 + v () where is a suiably chosen osiive ineger; α k s and β k s, k = 0,1,.., are consans; and u and v is usual disurbance erms wih zero means and finie variances. The null hyohesis ha X does no Granger-cause Y is rejeced if he β k s, k>0 in Equaion 2 are joinly significanly differen from zero using a sandard join es (e. g., an F es). Similarly Y Granger-causes X if he ϕ k s, k>0 coefficien in Equaion are joinly differen from zero. A bi-direcional causaliy relaion exis if boh β k s and ϕ k s, k>0 are joinly differen from zero. I may be menioned ha above es is alicable o saionary series. So, in realiy he series under analysis may no be saionary. In such cases one has o ransform he original series ino saionary series and causaliy es would be erformed based on ransformed-saionary series. A secial case of non-saionary rocess is he I (1) rocess, ha is, ossessing a uni roo. An I (1) may be convered ino saionary one by aking firs order differencing. Thus dealing wih wo I (1) rocess for causaliy, Equaion 2 and Equaion mus be exressed in erm of differenced-series. However, if underlying I (1) rocesses are coinegraed, he secificaions so obained mus be modified by insering he lagged value of coinegraion relaion (error correcion erm) as an exra exlanaory variable, ha is, Equaion 1 and Equaion 2 should be modified as follow: Y = α 0 + αk Y k + βk x k + δ ECT 1 u (4) k = 1 k = 1

5 Linkage beween Sock Marke Prices and Exchange Rae ϕk Y k + Φk X k + η ECT 1 + v (5) k = 1 k = 1 X =ϕ where is he difference oeraor and ECT 1 reresen an error correcion erm derived from he long run coinegraion relaionshi beween he I(1) rocesses X and Y. This erm can be esimaed by using residual from a coinegraion regression.. DATA DESCRIPTION AND EMPIRICAL ANALYSIS Our daa consis of monhly rices of four aggregae indices, General 100 index, Financial Secor, Indusrial Secor and Services Secor. General 100 index is he main index of Karachi Sock Exchange. Inclusion of Karachi Sock Exchange is because of he fac ha i ossesses he lion share of marke caialisaion (around 80 ercen) of he oal sock marke business in Pakisan. The oher hree indices are secoral indices. Our monhly daa on General are obained from Saisical Bullein and Annual Reor ublished by Sae Bank of Pakisan, Economic Survey ublished by Governmen of Pakisan, Finance Division, and Karachi Sock Exchange Publicaions. Due o non-availabiliy of daa on secoral level, he daa on secoral indices have been deermined and calculaed by auhors. The Karachi Sock Marke 100 index is divided ino hree main sub-secors on he basis of marke caialisaion of hese secors. Daa on all indices sar from January 1994 o December 200. Exchange rae is exressed in erm of US dollar. Two subses of his daa se are analysed: he full daa se consising all monhly values of marke indices and Rs/US exchange raes: he subse covering eriod rior o sock marke financial crises in 1998: and subse covering laer eriod of he financial crises u o December 200. We obain similar resuls in wo subses, herefore, we only reor he resuls obained from he full daa se. Before analysing he relaionshi beween exchange rae and sock indices i is imoran o carry ou univariae analysis. The saionariy of each variable is checked. Oherwise inference from he F-saisic migh be surious because he es saisic will have non-sandard disribuions. The saionariy of each series was invesigaed by emloying Augmened Dickey-Fuller uni roo es. The es consiss of regressing each series on is lagged value and lagged difference erms. The number of lagged differences included is deermined by he Akaike informaion crierion. Following he lieraure all daa series have been ransformed ino naural logarihms. The equaion used for conducing Augmened Dickey Fuller es has he following srucure: X =α +β +ρ X + δk X k + k = 1 1 e (6) In Equaion 6, if (i) β = 0 and ρ <1, he series X is saionary; (ii) β = 0 and ρ = 1 hen he series is an I (1) rocess; (iii) β 0 and ρ < 1 hen he series is rend saionary.

6 644 Farooq and Keung Table 1 reresens he Augmened Dickey Fuller es saisics under he null hyohesis of uni roo. Number of lagged difference erms included in he regression is also reored in Table 1. Null hyohesis of uni roo agains he saionary alernaive is no rejeced a 5 ercen level of significance for exchange raes and sock indices wih or wih ou deerminisic rends. However he firs differences of heses variables are saionary under he es, which is reored in Table 2. Hence, we conclude ha hese variables are inegraed of order 1. Table 1 Consan ADF Tes Saisic Exchange Rae Uni Roo Tess (Levels) General Financial Indusrial Services Lag Consan and Trend ADF Tes Saisic Lag Noe: 5 ercen Criical value for ADF es wih consan is 2.89 while i is.45 for he es wih consan and rend. Table 2 Consan ADF Tes Saisic Uni Roo Tess (Firs Difference) Exchange Rae General Financial Indusrial 6.27 Services Lag Consan and Trend ADF Tes Saisic Lag 1 2 Noe: 5 ercen Criical value for ADF es wih consan is 2.89 while i is.45 for he es wih consan and rend. On he basis of above uni roo ess, we erformed Johansen s coinegraion es o see wheher any combinaions of he variables are coinegraed. This aroach uses a maximum likelihood rocedure ha ess for he number of coinegraion relaionshis and esimae he arameers of hose coinegraing relaionshis. Likelihood raio (LR) es saisics and 5 ercen criical values are reored in Table. The resuls show ha

7 Linkage beween Sock Marke Prices and Exchange Rae 645 Table Johansen Maximum Likelihood Coinegraion Tess Max Tes Trace Tes Null Alernaive Saisic 95% Criical Value General Null Alernaive Saisic 95% Criical Value r = 0 r = r = 0 r = r < 1 r = r < 1 r = Finance r = 0 r = r = 0 r = r < 1 r = r < 1 r = Indusry r = 0 r = r = 0 r = r < 1 r = r < 1 r = Services r = 0 r = r = 0 r = r < 1 r = r < 1 r = here is no long run relaionshi beween exchange rae and sock indices for Pakisan. Consequenly, an error correcion erm need no be included in he Granger causaliy es equaions. Since hese resuls are robus and highly sensiive o he choice of he lag lengh. The Engle and Granger (1987) confirms his finding. 12 Our his resul suors he sudy of Abdallah and Murinde (1997). In ha sudy no long relaionshi is found for Pakisan and Korea, while for India and Philiines hey find long run relaionshi beween sock rices and exchange raes. As coinegraion is found o be sensiive o he choice of lag lengh herefore we find causal associaion beween sock rices and exchange rae. For his urose we emloy sandard Granger causaliy es o examine wheher variables under consideraion are causally relaed a leas in one direcion, ha is, wheher change in sock rices causing change in exchange rae or change in exchange rae causing change in sock indices? Taking ino accoun ha he resuls derived from his es may be sensiive o he selecion of lag lengh, he minimal final redicion error suggesed by Akaike (1969) has been used. Table 4 reors he F-saisics and robabiliy values consruced under he null hyohesis of non-causaliy. I can be observed ha general index effecs exchange rae. This finding is in conras o he finding of Abdallah and Murinde who found exchange rae Granger cause sock rices. This roves ha direcion of causaliy can vary according o he eriod of sudy. In he analysis we have an ineresing observaion ha exchange rae Granger cause services secor. These analysis exlicily show ha uni-direcional causaliy exis from sock rices o exchange, while exchange rae effec services secor index. 1 The resuls are no reored here bu are available on reques.

8 646 Farooq and Keung Table 4 Granger Causaliy Tes beween Sock Indices and Exchange Rae Null Hyohesis Exchange Rae does no Granger Sock Indices Sock Indices does no Granger Cause Exchange Rae Lags General.675 (0.0577) (0.0212) 1 Finance (0.1777) (01164) 1 Indusry (0.177) (0.1172) 1 Services (0.022) (0.7267) 1 4. CONCLUSION This aer analyses emirically he relaionshi beween four sock indices and exchange rae in Karachi Sock Exchange (). Since he variables in his aer are non-saionary a level and saionary a firs difference, herefore, Johanson s coinegraion echnique has been alied. The resuls obained by using his mehodology rovided no coinegraing relaionshi among he variables. (Since, mehodology of coinegraion rovides evidence of unique coinegraing vecor, herefore, a long run relaionshi beween sock indices and exchange rae does no exis in our analysis. This means ha sock indices and exchange do no move ogeher in he long run. We nex erform Granger non-causaliy es. The resuls resen some ineresing evidence. We find ha causaliy run from general sock rices o exchange rae. We also find ha causaliy runs from exchange rae o services indices. So according o our analysis general sock index affecs exchange rae in he shor run, herefore he invesors can use informaion obain from sock marke o redic he behaviour of exchange rae. Moreover auhoriies in Pakisan can use he sock rices as a olicy ool o arac he foreign orfolio invesmen by aking sabilising measures for sock marke. Our resuls rovide evidence in favour of orfolio balance models of exchange rae deerminaion ha osulaes a uni-direcional causaion ha runs from sock rices o exchange rae. However, our resuls are in conras o radiional models ha hyohesised causaion from exchange rae o sock rices. In our analysis, ineresingly exchange rae Granger cause services secor, which, sugges ha currency flucuaions effec services index. Hence exchange rae movemens can affec he rices of services secor. As services secor is a rerequisie for aaining economic growh and imroving counry s roducive caaciy by reducing roducion cos and i has also been widely recognised ha economies wih efficien services secor are osiioned more advanageously in erm of over all comeiiveness. So exchange rae can be used as olicy ool o imrove he services secor and sabilisaion of sock marke.

9 Linkage beween Sock Marke Prices and Exchange Rae 647 REFERENCES Abdalla, I. S. A., and V. Murinde (1997) Exchange Rae and Sock Prices Ineracions in Emerging Financial Markes: Evidence on India, Korea, Pakisan and Philiines. Alied Financial Economics 7, Ajayi, R. A., and M. Mougoue (1996) On he Dynamics he Dynamics Relaion beween Sock Prices and Exchange Raes. Journal of Financial Research 21, Akaike, H. (1969) Fiing Auoregressive Model for Predicion. Annals of Inernaional Saisics and Mahemaics 21, Amare, T., and M. Mohsin (2000) Sock Prices and Exchange Raes in Leading Asian Economies: Shor Run versus Long Run Dynamics. Singaore Economic Review 45: 2, Bahmani-Oskooee, M., Mohan, and Liker Dumas (1997) Turkish Sock Prices and Value of Turkish Lira. Canadian Journal of Develomen Sudies 18, Engle, R. F., and C. W. Granger (1987) Coinegraion and Error Correcion: Reresenaion, Esimaion, and Tesing. Economerical 55, Fang, W. S. (2001) Sock Marke and Execed Dereciaion over he Asian Financial Crises. Alied Economics, Granger, C. W. J. (1969) Invesigaing Causal Relaionshi by Economeric Model and Cross-secral Mehods. Economerical 7, Kearney, C. (1998) The Causes of Volailiy in a Small Inernaionally Inegraed Sock Marke: Ireland, July 1975 June Journal of Financial Research 21, Kououlas, G., and L. Kryzanowski (1996) Macro facor Condiional Volailiy, Time Varying Risk Premium and Sock Reurn Behaviour. Financial Review 1, Mukherjee, T. K., and A. Naka (1995) Dynamic Relaions beween Macroeconomic variables and he Jaanese Sock Marke: Alicaion of a Vecor Error Correcion Model. Journal of Financial Research 18, Smih, C. (1992) Sock Marke and Exchange Rae: A Muli-counry Aroach. Journal of Macroeconomics 14, Yu, Ciao (1997) Sock Prices and Exchange Raes: Exerience in Leading Eas Asian Financial Cenres: Tokyo, Hong Kong and Singaore. Singaore Economic Review 41,

10 Commens The aer resened by Mr Mohammad Tahir Farooq and Mr Wong Wing Keung relaes o an area of research which is relaively new. As oined ou by he auhors, he earlies aer on he subjec was by Bahmani-Oskooee (1992) which rovided he basic simulus o furher research. The resen aer hus sulemens he research iniiaives aken in he field of linkages beween sock marke rices and exchange raes in he conex of a develoing economy like Pakisan. The basic roblem wih he aer is ha i is excessively mechanical and lacks in diagnosic conens. For ha reason, i rovides limied undersanding of he basic deerminans of sock marke rices and exchange raes. By adoing a mechanical aroach, he aer has failed o rovide any fresh insigh ino he behaviour of hese imoran variables of he economy. The aer rovides divergen resuls wih regard o he long-run and he shorrun framework. This oucome is naural because he auhors fail o exlain wha is he long run and wha is shor run in he conex of he behaviour of he sock rices and exchange raes. The aer is no based on any heoreical foundaions. In he absence of heory, much of he analysis in his aer aears o be raher surious. I seems ha he auhors have used economeric ools such as uni roo ess and co-inegraion echniques wihou conceualising any basic hyohesis abou he underlying deerminans of hese variables and heir iner-relaionshi. The auhors allude o his missing link when hey sugges ha he henomenon of sock marke and exchange rae nexus received subsanial aenion afer Eas Asian financial crises. Theoreically, if here is any linkage beween boh he variables hen he crises can be avered eiher by managing exchange rae or by adoing indigenous olicies o sable he sock marke. Moreover he invesors can uilise his relaionshi beween sock rices and exchange rae o redic he behaviour of hese variables. The auhors hen add: There is a lack of heoreical consensus on he relaionshi beween exchange rae and sock rices. The orfolio balance aroach of exchange rae deerminaion and sock rices, furnishes boh osiive and negaive relaionshis beween sock rices and exchange rae. The causaion beween he wo variables may be uni-direcional or bi-direcional. I is clear from hese saemens ha he emirical work has been underaken by he auhors wihou a sound heoreical framework and for ha reason i loses much of is usefulness in he real world siuaion. The eriod covered in his sudy is from January 1994 o December 200. During his eriod, he exchange rae of Pakisan moved from Rs er dollar o

11 Commens 649 Rs er dollar indicaing he devaluaion of Pak ruee o he exen of 57.9 ercen. Pakisan had mainained a managed floa exchange rae during his eriod. This was characerised by inervenion by he Sae Bank of Pakisan in he foreign exchange marke on he one hand and conscious devaluaions of he ruee by he governmen from ime o ime wihou allowing a free floa of he currency. Juxaosed o he managed exchange rae regime, however, he behaviour of he sock marke rices was rimarily deermined by marke forces which relae o invesors confidence, execaions of he marke layers abou fundamenal macroeconomic and microeconomic arameers such as rice sabiliy, growh and invesmen levels in he counry. During his eriod, he sock rices have shown immense volailiy. This volaile behaviour of sock rices can be analysed by looking ino he underlying micro and macroeconomic facors. This level of analysis is alogeher missing in he aer. The smooh and gradual dereciaion of ruee ook lace during his eriod wihou receiving any simuli from he sock marke. In oher words, he behaviour of sock marke rices and ha of exchange rae in case of Pakisan have been broadly auonomous and indeenden of each oher. Therefore, o aly echniques of coinegraion and Granger-causaliy wihou looking ino he acual behaviour of hese variables is aarenly a mislaced exercise. In his very Conference, a few aers are being conribued boh on he behaviour of sock marke rices and exchange raes. These aers clearly esablish ha in case of Pakisan, sock marke rices and exchange rae movemens have been behaving indeenden of each oher. Therefore here is no jusificaion for alying Granger ess o hese wo imoran variables of he economy o check heir inerrelaionshi. Besides he fac, ha he aer carries numerous conradicory saemens, one does no find any serious conclusion, which could be helful in he formulaion of olicy framework for he exchange rae managemen as well as sock marke rices. Planning Commission, Governmen of Pakisan, Islamabad. Aqdas Ali Kazmi