On the Relationship between stock return and exchange rate: evidence on China

Transcription

1 O he Relaoshp bewee sock reur ad exchage rae: evdece o Cha Yaqog L a b, Lhog Huag a The Busess School, Loughborough Uversy,UK b College of Mahemacs ad Ecoomercs, Hua Uversy, Chagsha,Hua,Cha b Absrac The purpose of hs paper s o vesgae he relaoshp bewee RMB exchage rae ad A-share sock reurs Cha, parcular Shagha sock marke. We fd ha boh sock reurs ad RMB omal exchage rae are egraed of order. The Egle Grager coegrao es s he performed, suggesg ha here s o a log-ru equlbrum relaoshp bewee sock reurs ad RMB exchage raes a 5% sgfcace level. However, here s srog evdece suggesg ha here s a shor-ru u-drecoal causaly relaoshp from he omal exchage rae o he sock reurs. Keywords: coegrao; Grager causaly; RMB exchage rae; sock reur; u roo es.. Iroduco The Cha s exchage rae polcy has recely emerged as oe of major ssues he rade bewee he PR of Cha ad he Ued Saes of Amerca. The coroversy s fuelled by Cha s peggg of RMB o USD. Sce a major devaluao of he RMB 994, he Chese currecy s exchage rae vs-a-vs USD remaed more or less uchaged ul July 5, ad has flucuaed from RMB 8. o 8. per dollar sce he. The Chese Auhory has recely aouced ha RMB wll be o loger pegged o he US dollar ad ha Cha wll reform he exchage rae regme by movg o a maaged floag exchage rae regme based o marke supply ad demad wh referece o a baske of curreces (People s Bak of Cha.Publc aouceme( Wh he apprecao of he RMB exchage rae, sock reurs crease he Chese A-share markes of Shezhe ad Shagha. The queso s: does he apprecao of he RMB exchage rae lead o he crease of sock reurs? I he paper, he relaoshp bewee sock reurs ad RMB exchage raes s vesgaed. Ths paper employs daly daa ad cosders a more rece sample perod afer Chese exchage rae reform. The relaoshp bewee sock reurs ad foreg exchage raes has draw much aeo of ecoomss, for heorecal ad emprcal reasos, because hey boh play crucal roles fluecg he developme of a coury s ecoomy. I addo, he relaoshp bewee sock reurs ad foreg exchage raes has frequely bee ulzed predcg he fuure reds for each oher by vesors. Macroecoomc fudameals are see by ecoomss as provdg he robus meda o lk sock reurs ad foreg exchage raes. Amog all he moeary models of exchage rae deermao, moey supply, eres rae, reur level, ad flao are ake o accou o

2 predc he exchage rae moveme. See e.g. Frekel (976), Blso (978), Dorbusch (976), Frakal (979), Dorbusch ad Fsher (98), ad Hooper ad Moo (98). Braso, e al. s (977) porfolo-balace model furher corporaes asses of porfolos o descrbe he sock-oreed exchage rae moveme. Amog emprcal sudes, Meese ad Rogoff (983), Wolff (988), Balle ad Selover (987), ad Gharey (998) have foud cera relaoshps amog macro-fudameals ad exchage raes, whereas he emprcal evdece for he relaoshps amog macro-fudames ad sock reurs are foud Balay (99), Sadegh (99), ad Kwo ad Sh (999). The exsg face leraure s coclusve o he relaoshps bewee exchage rae movemes ad macroecoomc varables (see Aggarwal, 98; Edso, 99; Frekel, 983; Gadolfo, Padoa, & Palado, 99; Gova & Joro, 987; Hardouvels, 988; Wassserfalle, 99). Solk (987) coeds ha he major reaso s he poor qualy of he macroecoomc daa used. Mos of such daa, he argues, suffer from sgfca measureme errors whle some are o eve measurable drecly. He furher suggess sock reurs as surrogaes for macroecoomc daa, coedg ha sock reurs reflec expeced chages fuure ecoomc acvy, ad ha reurs ca be measured drecly ad accuraely. He employs mohly ad quarerly daa for egh dusral coures from o exame he relaoshp bewee real sock reur dffereals (RSR) ad chages real exchage raes (RER) ad repors a egave relao bewee RSR ad RER for mohly ad quarerly daa over hs sample space. However, usg mohly daa over he sub-perods, he observes a weak posve relaoshp bewee he wo varables. I arrvg a he above coclusos, Solk modeled he exchage rae as a fuco of eres rae ad sock reur dffereal. Soee ad Aggarwal (989) re-assess he Solk model usg daa for he same egh dusral coures. They repor a posve correlao bewee sock reurs ad exchage raes for hree coures ad egave correlao for fve. Oher researchers, however, exame sock reurs as a fuco of exchage raes (see Aggarwal, 98; Ma & Kao, 99; Seoe. & Hegar, 988). Employg mohly daa, Aggarwal (98) exames he relao bewee U.S. sock marke dexes ad a rade-weghed value of he dollar for he perod He fds ha he sock reurs ad exchage raes are posvely correlaed. I coras, Soee ad Hegar (988), also usg mohly daa, repor a srog egave correlao bewee U.S. sock dexes ad a ffee currecy-weghed value of he dollar for he perod Che-Chug Neh & Cheg-Few Lee () explore he dyamc relaoshps bewee he sock reurs ad he exchage raes for each of he G-7 coures. Boh he Egle-Grager (EG) wo seps ad he Johase maxmum lkelhood coegrao ess are employed. Ther emprcal work rejecs mos of he prevous sudes ha sugges a sgfca relaoshp bewee sock reurs ad exchage raes. The resul suppors Bahma-Oskooee ad Sohraba s (99) fdg ha here s o log-ru equlbrum relaoshp bewee hese wo facal varables. Noeheless, from vecor error correco model, sgfca shor-ru fdgs show he oe-day predcg power of he wo facal asses for cera coures. Issam; Abdalla ad Vcor (997) vesgae eracos bewee exchage raes ad sock reurs he emergg facal markes of Ida, Korea, Paksa ad he Phlppes o esablsh he causal lkages bewee leadg reurs he foreg exchage marke ad he sock marke. The resuls show u-drecoal causaly from exchage raes

3 o sock reurs all he sample coures, excep for he Phlppes. So far, he research hus far has exeded he qury o he sock marke ad exchage rae relaos across developed ad emergg ecoomes o brg a broadly-based sgh o he ssue ad o fll he gap he leraure. However, he Cha s marke s rarely cluded o he exsg leraure. The purpose of hs paper s o vesgae he relaoshp bewee sock reurs ad exchage raes Cha. The res of he paper s orgazed as follows. Seco s devoed o he daa ad varables used he aalyss. I Seco 3 we perform aalyss ad he repors he emprcal resuls. Fally Seco 4 provdes dscusso ad coclusos.. Daa ad varables We colleced he daly omal RMB exchage rae agas he USD. The exchage rae seres were he daly mddle exchage rae obaed from he Sae Admsrao of Foreg Exchage Cha. Daly closg sock marke dex Cha s sock marke dex, Shagha Sock Exchage Compose, ad daly closg sock marke dex, US s S&P5, were obaed from hp://face.c.yahoo.com. All daa ses were from July 5 o 8 Jauary 8 whou coverg he observaos o Saurday ad Suday, as well as some major holdays Cha ad US. The reaso sar dae s o July 5 ha Cha revalued he Remb ad offcally modfed he exchage rae regme from ha me. The followg varables are used he model. H represes omal Exchage rae ( s defed as domesc currecy us per u of U.S. dollar) day. R represes Cha s Sock reur day, where we use daly closg sock marke dex Cha s sock marke dex, Shagha Sock Exchage Compose hs paper. RM represes U.S. sock reur day, where we use daly closg sock marke dex: US s S&P5. Le S = R RM be he dfferece of he sock reurs bewee Cha ad U.S. Fg. ad Fg. show me seres plo of RMB omal exchage rae ad Dfferece of he sock reurs bewee Cha ad U.S //5 4/7/6 //7 /8/7 H //5 4/7/6 //7 /8/7 S Fg.. RMB omal exchage rae Fg.. Dfferece of he sock reurs bewee Cha ad U.S. 3. Mehodologes ad aalyss Ths seco s devoed o he mehodologes hs paper corporaes o explore he

4 relaoshp bewee sock reurs ad exchage raes Cha ad he releva daa aalyss. I 969 Grager proposed he Grager causaly o es wheher oe ecoomc varable ca help forecas aoher ecoomc varable (Grager, 969). Specfcally, Grager causaly from X o Y s esablshed whe he coeffces of he lagged dfferece of X are foud o be joly sascally sgfca ad herefore help expla ad predc Y, over ad above wha he lagged dffereces of Y ca predc. To cofrm ha he Grager causaly resuls wll o geerae ay spurous fereces, he u roo ad coegrao ess are performed before coducg he Grager causaly es o measure he chages he exsece ad drecos of causaly (see Eders, 995; Phllps, 986, 987). 3.. U roo ess To deerme he saoary of each varable for each subsample, we employ he augmeed Dckey Fuller (ADF) es. The ADF model used s gve as follows: p ΔH = α + γh + β ΔH + ε, () = where H represes omal exchage raes agas USD, α s he ercep erm, γ s he coeffce of eres he u roo es, β s he parameer of he lagged frs dfferece of H o beer represe he ph-order auoregressve process, ad ε s he whe ose error erm. Tme red s o cluded he model sce exchage raes do o exhb me red (Mlls & Taylor, 989). To esure ha he exchage rae seres are o egraed of more ha, he ADF u roo es s also ru usg he frs dfferece exchage rae seres as follows: p = Δ H = α + γδh + β Δ H + ε. () I boh cases, he ull hypohess of he u roo es s γ =,.e., he { } sequece coas a u roo process (osaoary) whle he alerave hypohess dcaes ha he seres s a saoary process. We rejec he ull hypohess of he u roo f he -sasc of γ s smaller ha he 95% Dckey Fuller crcal value, gve by MacKo (99). We also cosder he ADF model for sock reur as follows: p ΔS = α + α + γs + β ΔS + ε, (3) = p = Δ S = α + α + γδs + β Δ S + ε, (4) where S represes sock reur dffereals, α s he red erm. Tme red s cluded he model. Noe ha sock reur dffereals are expressed as he dfferece bewee sock reurs of Cha ad he U.S. sock reurs. Ths defo s based o he fac ha he U.S. marke accous for a subsaal share of he world s equy marke capalzao ad s cosse wh our defo of exchage raes as domesc currecy us per u of U.S. dollar (Rchard A. Ajay, Joseph Fredma ad Seyed M. Mehda. 998). The ADF es resuls Table clearly show ha boh he varables are o saoary a he 5% level of sgfcace; however, ADF sascs rejec he ull hypohess of o-saoary a he 5% level of sgfcace afer he varable have bee frs dffereced. Thus, he varables are egraed of order. S H

5 Table ADF u roo es Varable lag ADF I (d) H I () Δ H I ( ) S I () Δ S I ( ) * Sgfca a he 5% level for ADF es. Noe: Lag legh p s seleced based o he smalles umber of AIC (Akake s Iformao Creror) ad SC (Schwarz crero) 3.. Coegrao es (Egle Grager coegrao es) The Egle Grager procedure cosss of esmag he coegrag regresso by ordary leas square, obag her resduals, ad applyg u roo ess for he resdual (Egle & Grager, 987). Thus, f he me seres s foud o be osaoary ad egraed of he same order from he ADF es, he Egle Grager coegrao es s performed. To oba he resdual, he followg coegrag regressos are performed: H β S + e, (5) = + β h S = β + βh + es, (6) ad he ADF es s as follows: k + = Δeˆ = α eˆ + α Δeˆ + ε, (7) where Δê clude eh or es sequece ad wh he ull hypohess of H : α = (o coegrao). The value of opmal lag legh k s seleced by he smalles Akake formao crero (AIC) ad Schwarz crero (SC). Sce he resdual seres s calculaed from a coegrag equao, a ercep or me red s omed from he equao (Eders, 995). Table shows ha he exchage rae ad sock reur are o foud o be coegraed a 5% sgfcace level. Ths meas ha here s o log-ru equlbrum relaoshp bewee sock reurs ad RMB exchage raes a 5% sgfcace level (Eders, 995). Table Egle Grager coegrao es Coegrag regressos Resdual ADF 5% Crcal Value Hˆ 57 = S eh H H ˆ = ˆ Sˆ 474 = H es S S ˆ = ˆ Grager causaly es

6 Nex, he parwse Grager causaly es s used o exame wheher he pas value of a varable seres, X, wll help o predc he value of aoher varable seres a prese, Y, akg o accou he pas value of he Y (Grager, 988). If he me seres of a varable s saoary or I () from he ADF es, he Grager causaly es s performed as follows: H = α + β, jh j + γ, js j + ε, S = α + β, js j + γ, jh j + ε,, (8). (9) If he me seres of a varable s osaoary, I() from he ADF es, ad o coegraed, hey wll be covered o a I () by frs dfferecg, ad he followg Grager causaly es ca be appled: Δ Δ H = α + β, jδh j + γ, jδs j + ε, j = S = α + β, jδs j + γ, jδh j + ε,, (), () where ΔH s he frs dfferece a me of omal exchage rae where he seres s osaoary ad Δ s he frs dfferece a me of sock reurs where he seres s S osaoary. Because he exchage rae ad sock reur are o foud o be coegraed a 5% sgfcace level, we apply he equaos () ad () o es relaoshps of Grager-causaly. The resuls from Grager-causaly es s sesve o he seleco of lag legh, so resuls are preseed from VAR lag legh seleced by AIC. We esed for Grager-causaly usg he sadard F-es compued from resrced ad uresrced versos of he varas of Equaos () ad (). The resuls are repored Table 3. The F -es resuls sugges we rejec he ull hypohess ha exchage rae does o Grager-cause sock reur. Table 3 Parwse Grager Causaly Tess (Lags: 8) Null Hypohess ( H ): F-Sasc Probably Δ H does o Grager Cause Δ S Δ S does o Grager Cause Δ H The emprcal resuls reveal ha a a 5% sgfcace level here s he u-drecoal causaly from he frs dfferece of omal exchage rae o he frs dfferece of sock reurs our sample. Hece, here are shor-ru u-drecoal causaly relaoshps from he omal exchage rae o he sock reurs a 5% sgfcace level. 4. Coclusos

7 Ths paper sheds lghs o he lkages bewee RMB exchage raes ad A-share sock reurs Cha s sock marke. Some ecoomerc echques are appled o a bvarae VAR model of sock reurs ad exchage raes order o es for Grager-causaly bewee he wo varables. Esseally, we have performed he ADF ess wh dffere auoregressve orders ul we oba dvdual seres for H ad S ha are cosse wh whe ose error erms. Havg esablshed he order of egrao of he dvdual seres, we move o es for coegrao usg he wo-sep Egle-Grager procedure. The resuls sugges ha we have o proceed wh a VAR. Takg he VAR model, we deerme he opmum lag legh of he bvarae VAR, he carry ou dagoscs o he resulg equaos, proceedg o es for ferece of Grager-causaly wh hese equaos. The eresg fdg s ha hs paper rejecs mos of he prevous sudes ha sugges a sgfca relaoshp bewee sock prces ad exchage raes. Our me-seres esmaos suppor Bahma-Oskooee & Sohraba s (99) ad Che-Chug Neh & Cheg-Few Lee s () fdg ha here s o log-ru equlbrum relaoshp bewee hese wo facal varables for G-7 coury. Ths fdg s obaed from he EG wo-seps. Aoher fdg s ha, based o he resuls from he parwse Grager causaly es, here are shor-ru u-drecoal causaly relaoshps from he omal exchage rae o he sock reurs.the cocluso supor Issam; Abdalla ad Vcor s (997) fdg here are shor-ru u-drecoal causaly relaoshps from he exchage rae o he sock prce dex for Korea ad Paksa. We fd ha here s o log-ru equlbrum relaoshp bewee hese wo facal varables ad here are shor-ru u-drecoal causaly relaoshps from he omal exchage rae o he sock reurs a 5% sgfcace level. The ma mplcaos are ha, he case of Grager-causaly from exchage raes o sock reurs, chages exchage raes affec sock reurs shor-ru. Thus, erms of polcy relevace, he fdgs of hs paper sugges ha he goverme should be cauous her mplemeao of exchage rae polces sce hey affec sock markes shor-ru. Bu I s o suppored sascs ha he chages exchage raes affec sock reurs log-ru. As oed Grager (969), he daly daa may be more desrable because a samplg frequecy less ha oe day may roduce spurous sascal sgfcace o he ess. For comparave purposes, we should ru he ess o weekly daa ad mohly daa, herefore wll be explored our fuure research. Ackowledgemes We would lke o express our graude o Babg L for valuable commes ad suggesos o he mauscrp. Ay vews expressed ad all remag errors are he resposbly of he auhors. Refereces Aggarwal, R. (98). Exchage raes ad sock prces: A sudy of U.S. capal markes uder floag exchage

8 raes. Akro Busess ad Ecoomcs Revew, (), 7-. Bahma-Oskooee, M., & Sohraba, A. (Aprl 99). Sock prces ad he effecve exchage rae of he dollar. Appled Ecoomcs, 4(4), Baley, W. (Sepember 99). U.S. moey supply aoucemes ad Pacfc rm sock markes: evdece ad mplcaos. Joural of Ieraoal Moey ad Face, 9(3), Balle, R. T., & Selover, D. D. (987). Coegrao ad model of exchage rae deermao. Ieraoal Joural of Forecasg, 3, Blso, J. (978). The moeary approach o he exchage rae: some emprcal evdece. IMF Saff Paper, 5, Braso, W, Halue, H., & Masso, P. (Dec. 977). Exchage rae he shor ru: he dollar Deusche mark rae. Europea Ecoomc Revew,, Che, C.N., & Cheg, F.L. (). Dyamc relaoshp bewee sock prces ad exchage raes for G-7 coures. The Quarerly Revew of Ecoomcs ad Face,4, Dorbusch, R. (976). Expecaos ad exchage rae dyamcs, Joural of Polcal Ecoomy, 84, Dorbusch, R., & Fscher, S. (98). Exchage raes ad he curre accou. Amerca Ecoomc Revew, 7, Edso, H.J. (99). Forecasg performaces of exchage rae models revsed. Appled Ecoomcs, 3(B), Eders, W. (995). Appled ecoomerc me seres. USA: Wley, New York. Frakel, J. A. (979). O he Mark: a heory of floag exchage raes based o real eres dffereals. The Amerca Ecoomcs Revew, 69(4), 6 6. Frekel, J. (976). (976). A moeary approach o he exchage rae: docral aspecs ad emprcal evdece. Scadava Joural of Ecoomcs, 78(), 4. Frekel, J. (ed.), (983). Exchage raes ad eraoal macroecoomcs. Chcago: Uversy of Chcago Press. Gadolfo, G., Padoa, P.C., & Palado, G. (99). Exchage rae deermao: Sgle equao or ecoomy-wde models: A es agas he radom walk. Joural of bakg ad Face, 4(5), Gharey, E. E. (998). Moeary dyamcs Ghaa: evdece from coegrao, error correco modellg, ad exogeey. Joural of Developme Ecoomcs, 57(), Gova, A. & Joro, I. (987). The fscal ad moeary lkage bewee sock reurs ad flao. Joura of Face, 38, March, -34. Grager, C. W. J. (969). Ivesgag causal relaos by ecoomerc models ad cross-specral mehods. Ecoomerca, 37, HardouveIs, G.A., (988). Ecoomc ews, exchage raes ad eres raes. Joural of Ieraoal Moey ad Face, 7(l), Hooper, P., & Moro, J. E. (98). Flucuaos he dollar: a model ad real exchage rae deermao. Joural of Ieraoal Moey ad Face,, Issam S.A. Abdalla & Vcor, M.(997). Exchage rae ad sock prce eracos emergg Þ acal markes: evdece o Ida, Korea, Paksa ad he Phlppes. Appled Facal Ecoomcs, 7, Kwo, C. S., & Sh, T. S. (999). Coegrao ad causaly bewee macroecoomc varables ad sock marke reurs. Global Face Joural, (), 7 8. Ma, C.K. & Kao, G.W. (99). O exchage rae chages ad sock prce reacos. Joural of Busess Face ad Accoug, 7(), MacKo, J. (99). Crcal values for co-egrao ess. I R. F. Egle, & C. W. J. Grager (Eds.),

9 Log-Ru Ecoomc Relaoshps ( pp ). Ued Kgdom: Oxford Uversy. Meese, R. A., & Rogoff, K. (983). Emprcal exchage rae models of he sevees: are ay f o survve? Joural of Ieraoal Ecoomcs, 4, 3 4. Mlls, T. C., & Taylor, M. P. (989). Radom walk compoes oupu ad exchage raes: Some robus ess o UK daa. Bulle of Ecoomc Research, 4, Phllps, P. C. B. (986). Udersadg spurous regresso ecoomercs. Joural of Ecoomercs, 33, Phllps, P. C. B. (987). Tme seres regresso wh a u roo. Ecoomerca, 55, Rchard, A.A., Joseph, F. & Seyed, M.M. (998). O he relaoshp bewee sock reurs ad exchage raes:ess of Grager causaly. Global face joural, 9(), 4-5. Sadegh, M. (99). Sock marke respose o uexpeced macroecoomc ews: he Ausrala evdece. IMF Workg Paper, 6, 9. Seoe, L.A. & Aggarwal, R. (989). Facal prces as deermas of chages currecy values. Paper Preseed a he 5h Aual Meegs of Easer Seoe, L.A. & Hegar, ES. (988). A aalyss of exchage raes ad sock prces: The US experece bewee 98 ad 986. Akro Busess ad Ecoomc Revew, 9(4), wer, Solk, B. (987). Usg facal prces o es exchage rae models: A oe. Joural of Face, 4(l), Wassserfalle, W. (99). Expeced ad uexpeced chages omal ad real varables: Evdece from he capal markes. Joourul of Ieraoal Moey ad Face, 9(l), 9-7.