Stock Prices as a leading Indicatator of the East Asian Financial Crisis.

S ock P ices as a Leading Indica o o he Eas Asian Financial C isis Simon B oome Economics Depa men Na ional Uni e si y o I eland, Maynoo h and B uce Mo ley* Economics G oup Uni e si y o Wales, Abe ys wy h Sep embe 2003 Abs ac : Using a basic cu ency c isis model, we assess he e ec i eness o s ock p ices as a leading indica o o he Eas Asian cu ency c isis in 1997 and 1998. S ock p ices a e inco po a ed in o a basic mone a y model, h ough he weal h e ec pos ula ed by F iedman (1988). In addi ion o he domes ic s ock p ice, we also inco po a e he s ock p ices o Hong Kong, China and Japan o de e mine hei abili y o p edic he c isis. Using mon hly da a, he esul s indica e ha he domes ic s ock p ice is a signi ican leading indica o , howe e he main s ock p ices indica o o he c isis is he Hong Kong s ock p ice. In addi ion he US p ice le el is also a highly signi ican p edic o o he c isis. Causali y es s sugges e idence o bi-causali y be ween he s ock ma ke s and o eign exchange ma ke s. JEL. F30, E44 Key Wo ds: Cu ency C isis, S ock P ices, Mone a y Model. * Add ess o co espondence: D B.Mo ley, Economics G oup, SMB, UW Abe ys wy h, Abe ys wy h, SY23 3DD, UK. Tel. +(01970) 622522, E-mail: [email p o ec ed], Fax + (01970) 622740. I In oduc ion The aim o his pape is o p ima ily de e mine whe he he domes ic s ock ma ke can be used as a leading indica o du ing a egional cu ency c isis, as in Eas Asia du ing 1997 and 1998. Fu he we in es iga e whe he he majo s ock ma ke s wi hin Eas Asia had any e ec on he cu encies su e ing he c isis and can also be used as an ea ly wa ning sys em. A pa icula ea u e o he Eas Asian c isis was he almos simul aneous decline in asse p ices and cu encies, as in e na ional in es o s mo ed hei capi al ou o he espec i e domes ic ma ke s. This esul ed in a subsequen dep ecia ion o he exchange a e as he domes ic cu ency was sold. Howe e he deg ee o se e i y o he c isis a ied ac oss Eas Asia, as did he ex en o he decline in he domes ic s ock ma ke s. To da e mos models o p edic ing cu ency c ises ha e concen a ed on leading indica o s om ei he he banking sec o o he cu en accoun (e.g., Kaminsky e al., 1998, Kwack, 2001). When s ock p ices ha e been included, i has been limi ed o only domes ic s ock p ices. In his pape we sugges a simple mone a y based model, in which s ock p ices can be inco po a ed as a leading indica o o cu ency c ises. We also in es iga e whe he o eign s ock p ices a e a signi ican leading indica o , based on h ee heo ies conce ning he o igins o he Sou h Eas Asian cu ency c isis. When in es iga ing he ela ionship be ween s ock p ices and exchange a es, he main conce n is usually o e he di ec ion o causali y be ween hese a iables. The e a e heo e ical easons o causali y o un in bo h di ec ions, as sugges ed by Bahmani-Oskooee and Soh abian (1992) and G ange e al (2000). The empi ical 2 e idence is equally ambiguous, al hough G ange e al (2000) inds e idence o causali y unning om s ock ma ke s o exchange a es du ing a cu ency c isis, bu hei s udy uses daily da a a he han mon hly da a. In his s udy we oo assume causali y is om s ock p ices o exchange a es. Following he in oduc ion, we assess he in e ela ionship be ween s ock p ices and exchange a es du ing a cu ency c isis. We hen de i e a simple model o cu ency c ises, in which s ock ma ke s a e inco po a ed h ough he demand o money unc ion. The nex sec ion desc ibes he da a and assesses he esul s om he empi ical models. Finally we gi e ou conclusions and sugges some implica ions o u u e policies. II Cu ency C ises and S ock P ices The main a emp o inco po a e he domes ic s ock ma ke in o empi ically based cu ency c isis models has been Kaminsky and Reinha (1996) and Kaminsky e al (1998). These models inco po a e a numbe o eal, inancial and poli ical a iables in pu ely empi ical models, o iden i y which a e signi ican and also he leng h o he signalling ho izon. They use he “signals” app oach, which in gene al in ol es a one s ep ahead p obabili y o a de alua ion, in he con ex o a mul i a ia e p obi o logi ype model. The domes ic s ock ma ke s a e gene ally ound o be a signi ican leading indica o o cu ency c ises, o e a numbe o di e en cu ency c ises in he ecen pas . In he s udy by Kaminsky e al. (1998), s ock p ices a e ound o be he ou h bes p edic o o cu ency c ises. 3 The e has been a ce ain amoun o deba e in he li e a u e as o he di ec ion o causali y be ween s ock p ices and exchange a es. G ange e al. (2000) show ha causali y uns om domes ic s ock p ices o exchange a es. They o e some heo e ical suppo o a bi-di ec ional ela ionship be ween he s ock ma ke and o eign exchange ma ke , bu only o e he sho un. They a gue ha a change in exchange a es would change he ma ke alue o all i ms ha ade in e na ionally. This would depend on whe he he i ms a e ne impo e s o expo e s, such ha i in agg ega e mos i ms we e ne expo e s, a de alua ion would ha e a bene icial a ec on hose i ms p o i abili y and he e o e s ock ma ke alue. This causal ela ionship is e med he adi ional app oach, al hough i does no speci y he sign o he e ec . In con as o his app oach, bo h Bahmani-Oskooee and Soh abian (1992) and G ange e al. (2000) s ess he impo ance o he po olio app oach o analysing he ela ionship be ween s ock p ices and exchange a es. This sugges s ha a ise in s ock p ices, inc eases he domes ic weal h o in es o s, acili a ing a ise in he demand o money. Following he consequen ise in in e es a es, capi al is a ac ed in o he domes ic economy app ecia ing he domes ic cu ency. This app oach assumes he e is a nega i e ela ionship be ween s ock p ices and exchange a es, wi h causali y unning om he s ock ma ke o he o eign exchange ma ke . Wu (2001) p o ides e idence o he nega i e ela ionship be ween equi ies and exchange a es in Sou h Eas Asia. This explana ion is he mos ele an o his ela ionship du ing a cu ency c isis. We ha e conduc ed a se o G ange causali y es s be ween domes ic s ock p ices and exchange a es, as wi h o he s udies, he e is e idence o bicausali y be ween he s ock ma ke and o eign exchange ma ke s. When causali y 4 uns om s ock p ices o he exchange a e, he e is a nega i e ela ionship, which sugges s he po olio app oach is mos impo an in Sou h Eas Asia. Apa om he US s ock ma ke , we ha e in es iga ed whe he he c isis could ha e o igina ed om any o he main neighbou ing s ock ma ke s, in Japan, Hong Kong and China. Al hough he c isis began in Thailand in 1997, o he s ha e sugges ed he c isis began ea lie han his, wi h bo h China/ Hong Kong and Japan being sugges ed as con ende s. Fe nald e al (1999) ha e sugges ed ha he ise in China’s economic s eng h du ing he 1990’s added o he c isis. They ocused on he 1994 de alua ion and subsequen s ong expo pe o mance as he cause o he oubles, as China cap u ed expo ma ke s, which had p e iously been domina ed by he Associa ion o Sou heas Asian Na ions (ASEAN) coun ies. This acili a ed cu en accoun p oblems, educed company p o i abili y and culmina ed in he gene al inancial demise o he a ea. I has also been a gued ha a u he sou ce o he c isis could ha e been Hong Kong, which in 1997 was o icially e u ned o China om he UK. Ini ially his had a posi i e e ec on Hong Kong’s sha e p ices, bu as he da e o he hando e app oached, so conce n abou in es men in Hong Kong inc eased. Some a gue ha ano he sou ce o he c isis was Japan. Following apid ises in he s ock ma ke du ing he 1980’s, he ea ly and mid 1990’s saw a e e sal o his end, wi h some sha p alls. A he same ime he Japanese economy expe ienced an e a o s agnan g ow h and lack o demand. This acili a ed a decline in demand o ASEAN impo s and a all in in es men lows o hese coun ies. The inancial c isis in Japan culmina ed in he ailu e o Yamaichi co po a ion in 1997, he ou h la ges inancial ins i u ion. This in u n caused ano he la ge all in he Japanese s ock ma ke . 5 I ei he Japan o China/ Hong Kong we e he ins iga o s o his c isis, hen a measu e o hei inancial oubles could be a use ul leading indica o o he cu ency c ises in Eas Asia. To de e mine i his is he case, we ha e inco po a ed he s ock ma ke indices o China, Hong Kong and Japan in o he basic mone a y based model, along wi h he domes ic s ock ma ke index. III Model The ollowing model is based on he K ugman (1979) model o cu ency c isis, wi h s ock ma ke e ec s inco po a ed h ough he money demand speci ica ion. As wi h Edin and V edin (1993), his is only used as a basis o he empi ical in es iga ion and as wi h he leading indica o li e a u e in gene al, o he leading indica o s a e also inco po a ed in o he empi ical es s, wi hou speci ic heo e ic modeling. The s ock ma ke is included in he money demand unc ion2 o h ee easons, as sugges ed by F iedman (1988). I p ima ily ac s as a weal h e ec , as a ise in s ock p ices inc eases nominal weal h. Addi ionally a ise in s ock p ices e lec s a ise in expec ed e u ns om isky asse s. To o se his ise in isk, agen s swi ch away om long- e m bonds o sa e mone a y asse s. Thi dly a ise in s ock p ices implies a ise in inancial ansac ions and hus ansac ional demand o money. These imply a posi i e ela ionship be ween money and s ock p ices. The e could also be a nega i e ela ionship h ough a subs i u ion e ec . As s ock p ices ise, agen s subs i u e he 2 The s ock ma ke could ha e been ela ed o o he mac oeconomic a iables, such as consump ion, ou pu and in es men . Howe e gi en he mone a y basis o mos cu ency c isis models, we ha e inco po a ed i in o he money demand unc ion. 6 mo e a ac i e equi ies o money. Howe e as wi h F iedman (1988) we assume he posi i e e ec domina es. As wi h he con en ional mone a y model, we assume pu chasing powe pa i y (PPP) and unco e ed in e es pa i y (UIP) 3hold; (1) epp  * Whe e, e is he exchange a e, p a e domes ic p ices and p* a e o eign p ices. Ede d ii () *  (2) Whe e i a e domes ic in e es a es and i*a e o eign a es. Money demand akes he ollowing o m: mp y i s       (3) Whe e m is domes ic money balances, y is domes ic income and s is a domes ic s ock ma ke index. We ha e assumed ha he domes ic money supply is pu ely accommoda ing. By ea anging he abo e equa ions, we ge : 3 As wi h cu ency c isis models in gene al based on he mone a y model, we ha e assumed PPP and UIP, al hough empi ical s udies ques ion whe he he o me holds due o ade impedimen s and anspo cos s and he la e due o he p esence o a isk p emium. 7 d de sypime    ** (4) To es o he e ec s o s ock p ices on he exchange a e, we ha e chosen a s anda d empi ical model, based on equa ion4 (4). This sugges s he c isis is a unc ion o : ),,,*,*,,,( xs es e dspiym cc      (5) Whe e cc is he a e o change in he exchange a e o he mon h when he exchange a e is in a c isis condi ion,Δ m is he domes ic money supply (M1), y is ou pu , i* is he US in e es a e, p* is he US p ice le el, ds is he domes ic s ock ma ke index,  e is he change in he eal exchange a e, es a e o eign cu ency ese es and xs a e he ele an o eign s ock ma ke indexes (All a iables a e in logs and in change o m). As wi h Edin and V edin (1993) we ha e also included he eal exchange a e, a he han he nominal exchange a e and added o eign exchange ese es. The ele an o eign s ock ma ke indexes a e he US, Japan, Hong Kong and China. The US s ock ma ke is included because he Eas Asian cu encies a e pegged o he dolla , he Japanese ma ke s ha e adi ionally had a s ong e ec on he egion as a whole. China has ecen ly unde gone impo an poli ical changes which ha e a ec ed 4 An al e na i e app oach would ha e been o include a a ie y o o he ele an a iables ela ing o he banking sec o , cu en accoun and poli ical ac o s, as in Kaminsky e al. (1998). Howe e in o de o p o ide a heo e ical basis o he empi ical es s, he emphasis is on a iables ela ing o he F iedman based mone a y model. In addi ion i would no be possibl o inco po a e all he po en ial leading indica o s in o he model, so we ha e concen a ed on he s ock ma ke a ea. 8 hei inancial ma ke s and Hong Kong has been ans e ed om UK o Chinese con ol. In addi ion o he main leading indica o es s, we ha e also included some G ange Causali y es s, be ween he exchange a es and domes ic s ock p ices. In addi ion we ha e included some causali y es s be ween he exchange a es and o eign s ock p ice a iables. The basic es s is: i i i i i i ess usee        12110 12110     (6) Whe e e is he exchange a e, s is he s ock p ice a iable and u and a e e o e ms. As wi h Bloms om e al. (1996) we ha e included coun y dummy a iables and use he -s a is ic on he lagged explana o y a iable o de e mine i he e is e idence o causali y. IV Da a and Resul s The coun ies included in he in es iga ion a e hose ha expe ienced he wo s p oblems du ing he c isis. This includes Thailand, Malaysia, Sou h Ko ea, Indonesia and he Philippines. The da a is all mon hly, unning om Janua y 1996 o Decembe 1999, so including mon hs whe e a c isis was e iden as well as mon hs in which he exchange a e was ela i ely s able. The exchange a e is exp essed as he domes ic cu ency in e ms o he US dolla , he o eign explana o y a iables a e he espec i e US a iables. This e lec s he ac ha he ele an cu encies we e 9 Kaminsky, G. and. Reinha , C.M., 1996. The win c ises: The causes o banking and balance o paymen s p oblems. In e na ional Finance Discussion Pape 544, Washing on: Boa d o he Go e no s o he Fede al Rese e Sys em (Ma ch). Kaminsky, G., Lizondo, S. and Reinha , C.M., 1998. Leading indica o s o cu ency c ises. IMF S a Pape s, 1-47. K ugman, P., 1979. A model o balance paymen s c isis. Jou nal o Money, C edi and Banking, 11, 311-325. Kwack, S. Y., 2000. An empi ical analysis o he ac o s de e mining he inancial c isis in Asia. Jou nal o Asian Economics, 11, 195-206. Loungani, P., 2000. Com ades o compe i o s?: T ade links be ween China and o he Eas Asian coun ies. Finance and De elopmen , 37, 1-5. Radele , S. and Sachs, J.D., 1998 The Eas Asian inancial c isis: Diagnosis, emedies, p ospec s. B ookings Pape s on Economic Ac i i y, 1-89. Wu, Y. 2001., Exchange a es, s ock p ices, and money ma ke s: e idence om Singapo e. Jou nal o Asian Economics, 12, 445-458. Table 1. Causali y es s be ween S ock p ices and Exchange a es. Δe Δds 16 Δe(-1) Δds(-1) ΔHKS(-1) -0.087 (1.674)** -0.074 (1.159) 0.182 (1.916)** ΔUSS(-1) ΔCHS(-1) ΔJPS(-1) 0.070 (0.499) -0.076 (1.683)** -0.209 (1.927)** No es: T-s a is ics a e in pa en heses. E and ds a e he domes ic exchange a e and s ock p ice espec i ely. HKS, USS, CHS and JPS and Hong Kong, US, China and Japan’s s ock p ices espec i ely. ** indica es signi icance a he 10% le el o signi icance. Only lagged explana o y a iable included. Table 2. P obi Models using panel da a om Thailand, Malaysia, Sou h Ko ea, Philippines and Indonesia 17 Va iable 1 2 3 4 ΔM (-1) ΔDS (-1) ΔUSP (-1) -1.157 (1.709) -0.620* (2.363) -107.181* (11.195) -1.136 (1.681) -0.585* (2.186) -1.3.874* (9.488) -1.095 (1.596) -0.572* (2.170) -102.311* (9.916) -1.131 (1.442) -0.320 (0.847) -110.289* (10.234) ΔUSS (-1) USi (-1) ΔY (-1) 0.064 (0.308) -0.608 (1.312) -0.393 (0.615) 0.073 (0.351) -0.629 (1.363) 0.007 (0.032) -0.708 (1.514) 0.057 (0.240) -0.636 (1.253) ΔRER (-1) ΔRES (-1) ΔTHS(-1) 0.283 (0.807) 0.281 (0.807) 0.206 (0.577) -0.796 (1.453) 0.090 (0.232) -0.471 (1.651) L-L P obabili y -122.031 70% -121.841 71% -120.956 73% -97.229 68% No es: -s a is ics a e in pa en heses. P obabili y is he p obabili y o a p edic ed ou come occu ing om he 115 possible ou comes, ac ing as a pseudo-R2. M is he money supply, DS is he domes ic s ock p ice, USP, Usi and USS a e US p ices, in e es a es and s ock p ices espec i ely. Y is ou pu and THS a e Thailand’s s ock p ices, RER is he eal exchange a e and RES a e ese es. L-L is he es ic ed log likelihood A * indica es signi icance a he 5% le el. Table 3. P obi Models using panel da a om Thailand, Malaysia, Sou h Ko ea, Philippines and Indonesia, including o eign s ock p ices. 18 19 Va iable 1 2 3 4 ΔM (-1) ΔDS (-1) ΔUSP (-1) -1.061 (1.596) -0.445 (1.668) -107.643* (12.306) -1.129 (1.665) -0.500 (1.860) -109.676* (11.680) -1.116 (1.655) -0.594 (2.265) -106.075* (11.008) -1.071 (1.609) -0.476 (1.747) -116.063* (10.558) ΔUSS (-1) USi (-1) ΔY (-1) 0.041 (0.188) -0.573 (1.266) 0.082 (0.391) -0.535 (1.150) 0.110 (0.519) -0.683 (1.450) 1.069* (1.224) 0.064 (0.278) -0.578 (1.234) ΔRER (-1) ΔJPS (-1) ΔCHS (-1) 0.040 (0.114) 0.175 (0.500) -0.493* (2.292) 0.253 (0.718) -0.602* (1.093) -0.035 (0.097) -0.512 (0.820) -0.047 (0.164) ΔHKS (-1) L-L P ob -0.974* (3.072) -117.412 72% -119.494 72% -121.431 71% -1.141* (2.234) -116.428 73% No es: See Table 1 and 2.