Munich Personal RePEc Archive. Ebaad Momin and Mansur Masih. INCEIF, Malaysia, INCEIF, Malaysia. 18 June 2015

Transcription

1 MPRA Munch Personal RePEc Archve Do US polcy uncerany, leveragng coss and global rsk averson mpac emergng marke eques? An applcaon of bounds esng approach o he BRICS Ebaad Momn and Mansur Mash INCEIF, Malaysa, INCEIF, Malaysa 8 June 205 Onlne a hps://mpra.ub.un-muenchen.de/65834/ MPRA Paper No , posed 30 July :23 UTC

2 Do US polcy uncerany, leveragng coss and global rsk averson mpac emergng marke eques? An applcaon of bounds esng approach o he BRICS Ebaad Momn and Mansur Mash 2 Absrac: When he Uned Saes sneezes, he world caches a cold. And when Amerca recovers, he plane has a sprng n s sep For decades ogeher, hs meaphor has seemed an accurae descrpon of he global economy. Through hs paper we have red o examne he shor and long erm dependence srucure beween he sock markes of emergng markes and nfluenal global facors (US economc polcy uncerany, he global rsk averson and he cheap borrowng coss n he US) usng he BRICS counres (Brazl, Russa, Inda, Chna and Souh Afrca) as a case sudy. The sudy apples he Auo-Regressve Dsrbued Lag (ARDL) echnque (Pesaran, Shn, &Smh, Journal of Appled Economercs, 200) whch has aken care of a major lmaon of he convenonal conegrang ess, n ha hey suffer from he prees bases. Based on he above rgorous mehodology, our evdence ends o sugges ha alhough here have been sudes whch ndcae he mpac of he dsurbances semmng from he developed world, n he long- run here s a lmed mpac of hese on he BRICS equy markes. These fndngs are plausble and have srong polcy mplcaons for porfolo nvesng and dversfcaons by nvesng n he emergng markes as he BRICS eques could funcon as a hedge agans negave shocks from he developed economes. Keywords: US Polcy Uncerany, Rsk Averson, Leverage, BRICS Ebaad Momn PhD Canddae a INCEIF, Lorong Unvers A, 5900 Kuala Lumpur, Malaysa. 2 Correspondng auhor, Professor of Fnance and Economercs, INCEIF, Lorong Unvers A, 5900 Kuala Lumpur, Malaysa. Phone: Emal: mansurmash@ncef.org

3 Do US polcy uncerany, leveragng coss and global rsk averson mpac emergng marke eques? An applcaon of bounds esng approach o he BRICS Inroducon: The 'BRIC acronym was coned by Jm O'Nell n 200, Goldman Sachs chef global economs, who emphaszed on he specacular economc growh prospecs of he Brazlan, Russan, Indan and Chnese economes. Souh Afrca has more recenly joned he BRIC economes o now form he BRICS group. Based on recen economc forecass, Brazl, Russa, Inda, Chna and Souh Afrca (BRICS) are ancpaed o exhb exceponally hgh economc growh raes over he nex 50 years. Ths wll resul o BRICS jonly growng larger han he G-6 n US dollar erms (Wlson &Purushohaman, 2003). The BRICS cover 25% of he world's land mass, 40% of he world's populaon and run ncreasngly as global marke economes (Frank & Frank, 200). The BRICS share n world GDP and global expors s expeced o grow by 205 from 4% o 2.6% and from 2.4% o 20.% respecvely (a he same me, he US expor share s ancpaed o declne from 25 o 22%) (Wlson &Purushohaman. 2003). The susanably of BRICS mpressve growh pah s subjec o furher srucural and nsuonal reforms and fnancal lberalzaon, foregn nvesmen nflows and nernaonal compeon (Aye e al., 204; Chkl& Nguyen, 204; De Vres e al., 202; Manamper, 204; Pradhan e al., 203; Sarwar,20I2). As global nvesors perssenly pursue aracve asse classes o allocae her porfolos on alernave syle nvesng. BRICS capal markes receve ncreasng nernaonal fund nflows(chenge al., 2007;Chkl&Nguyen, 204; Ghoshe al., 2009; Sledzk, 202).

4 Undersandng he funconng ofbrics equy markes, her dynamc rsk-reurn properes, poenal volaly spllover effecs, ner- relaonshps and reacons o shocks, evens or news, relave o leadng global maure markes, such as he US, remans a crucal ssue for nernaonal nvesors, porfolo managers and polcy makers. By hs paper, we examne how economc facors n he US such as changes he U.S. economc polcy uncerany, he sock marke uncerany as defned by CBOE VIX a recognzed proxy o rsk averson and he cheap borrowng coss nfluence he performance of BRICS sock markes. Our analyss s movaed by he fac ha he BRICS counres are he major recpens of global nvesmen flows and are among he man global consumers of commodes. Therefore, changes n he global economc facors could be a channel hrough whch flucuaons n he world s economc and fnancal condons are ransmed o he BRICS sock markes and affec her economc growh. The recen global spllover and conagon effecs nduced by he US subprme morgage fnancal crss, llusrae hs sor of a dynamc neracon beween maure and emergng capal markes (Berger & Turle, 20). Moreover, nernaonal nvesors are especally neresed n he BRICS sock markes co-movemens wh heseglobal facors, gven ha nvesmen, speculaon and rsk dversfcaon opporunes mayarse.shor and long-run sock marke dynamcs can have crcal mplcaons for asse valuaon, porfolo allocaon, effcen dversfcaon, hedgng, and rsk conrol. If, for nsance,reurn and volaly spllover effecs are seen o spread from one marke o anoher a mes of marke crashes, adverse evens or fnancal crses,porfolo dversfcaon benefs should be expeced o reman lmed. In hs case, global nvesors would have o adjus her asse allocaon

5 decsons n order o mgae conagon rsks (Alou e al. 20; Celk, 202; Kenourgos e al. 20; Syropoulos, 203; Syropoulos&Roumps, 2009). Below s a graphcal represenaon of he movemen of he varous BRICS sock marke ndces over he perod under consderaon. BOVESPA - BRAZIL MICEX - RUSSIA NIFTY - INDIA SCOMP - CHINA JSE - SOUTH AFRICA

6 Despe growng global aenon on he BRICS capal markes, he relevan body of emprcal research remans surprsngly lmed and furher nsgh would be useful. Ths sudy aemps o fll some of he gaps n he opc and conrbues a range of nnovave and fruful emprcal conclusons. The man objecves of hs paper are: To examne f economc polcy uncerany n he U.S. has any effec on he reurns on sock markes n he BRIC (Brazl, Russa, Inda and Chna) counres. The curren sudy also nvesgaes how sock marke reurns n he four counres respond o he U.S. economc polcy uncerany shock To acess f he Rsk Averson rades have a sgnfcan mpac on he BRICS eques To undersand f favorable leveragng opporunes lead o he movemen of capal o he BRICequy markes, drvng hem up. Ths paper res o nvesgae f he above facors affec he BRICS equy markes n he long and shor-run. We employ monhly daa over he perod 2000: - 205:3 by usng he Auoregressve Dsrbued Lag (ARDL) conegraon. Ths paper s organzed as follows: Secon II revews on he relevan emprcal leraure. The heorecal specfcaon, daa and he preference for he ARDL conegraon mehodology are explaned n secon III. The emprcal resuls and dscussons are presened n secon IV. The las secon ends wh he concludng remarks and polcy mplcaons of he paper. II. Leraure Revew: Gven he above ncreased economc negraon of he BRICs wh he world economy, shocks orgnaed from advanced economes such as he Uned Saes can have a sgnfcan mpac on

7 he BRICs economes. Trade and fnancal lnkage beween counres play an mporan role n explanng nernaonal spllovers (Forbes & Chnn, 2004). Many sudes have emprcally documened he nernaonal spllovers from he US o oher counres. (Ehrmann and Frazscher, 2009) repor ha he US moneary polcy shocks spll over o oher equy markes around he world. Km (200) shows ha long-erm yelds and oupu of oher counres and oupu are affeced by he US moneary polcy shocks; smlar fndngs are repored n oher sudes (Awad& Goodwn, 998; Chnn & Frankel, 2004; Ehrmanne al, 20). Numerous oher sudes repor he nernaonal ransmsson of fnancal markes around he world (Ammer e al, 2008; Ehrmann&Frazscher, 2006; Hausmann&Wongswan, 20; Wongswan, 2006). In parcular, because of he sze of he US economy, shocks o he US economy and fnancal markes can spll over o oher counres fnancal markes (Bayoum&Swson, 2007; Ehrmann&Frazscher, 2005; Goldberg & Leonard, 2003). Kng and Wadhwan (990) argue he correlaon beween fnancal markes around he world exss snce raonal marke parcpans observe and analyze prce movemens n oher sock markes. Moreover, many oher sudes have examned f macroeconomc varables can predcs he performance of varous fnancal secures (Cooper & Presley 2005; Menzly e al, 2004; Pazzese al, 2005). In addon, sudes such as (Bansal e al, 2005; Dzelnsk, 20; Ozoguz, 2009) have documened he mpac of uncerany relaed o he economy and oher polces on he performance of he sock markes. Paser and Verones (20) assocae he decreased sock prces o he ncrease n governmen polcy uncerany. Furhermore, negave sock reurns are assocaed wh ncreased changes n economc polcy uncerany n he Uned Saes (Sum, 202a) Europe (Sum, 202b), and fve ASEAN counres (Sum, 202c).

8 Deparng from he aforemenoned sudes ha have a man focus on markes across he world, here have been some noable sudes for he BRICS counres. In hs emprcal leraure, he mpac of varous global facors on hese economes sock markes have been consdered. These nclude developed markes eques, ol, cred spreads ec. Hammoudeh e al. (203) have examned he nerrelaonshp beween he fve BRICScounres equy marke ndces, and her relaonshp wh he Inernaonal Counry RskGude (ICRG) s hree counry rsk rang facors (economc, fnancal and polcal), hes&p500 ndex and he Wes Texas Inermedae (WTI) ol prce. Ono (20) on smlar lnes has examned he sysemc mpac of ol prces on he sock marke reurns for he four BRIC counres and fnds ha ncreases n ol prces pull up he sock marke ndcesfor all hese counres excep Brazl. Alou e al. (20) examned hefnancal nerdependences of he BRICemergng markes wh he U.S. markes and provde srong evdence of me-varyng dependence beween hem. Ths dependency s sronger for he commody-prce dependen markeshan for he fnshed produc expor-orened markes of he BRIC counres. Moreover, hey observe hgh levels of dependence perssence for all marke pars durng boh bullsh and bearsh markes.dmrou e al. (203) however fnd an ncreasng co-movemen beween he BRICS and U.S. markes durng he pos-crss perod(from early 2009 onwards), mplyng ha he dependence s larger n bullsh han n bearshmarkes. Hwang e al. (203) n a wder sudy, examned he dynamc condonal correlaons beween heu.s. and en emergng sock markes (.e., he fve BRICS markes, Souh Korea,

9 Thaland,Phlppnes, Tawan, and Malaysa). They show ha dfferen paerns of he U.S. fnancal crss spllovers exs among emergng economes. They also conclude ha ncreasesn he cred TED spread (.e., he yeld dfference beween he hree-monh LIBOR rae andhe U.S. hree-monh Treasury blls) and soveregn CDS spread, boh represenng hgherrsks, decrease he esmaed condonal correlaons. Zhang e al. (203) provde srong evdence ha he recen global fnancal crss has changed he condonal correlaons beween he developed (U.S. and Europe) markes and he BRICS sock markes. Also Bekros (203) by usng lnear and nonlnear causal lnkages o analyze he volaly spllovers among he U.S., he EU and hebric markes - fnd ha he BRICs have become more nernaonally negraed snce he U.S. fnancal crss. Whle hese sudes add furher evdence o he facors affecng he BRICS sock markes, hey brng up a noable dmenson on he subjec. Namely, he effec of US based facors on he BRICS. Through he paper we ry o examne f he meaphor of he US sneezng used earler sands rue. Alhough as dscussed above here have been sudes o undersand he effecs of several facors on he BRICS equy markes, o our knowledge here seem o be few parallels whch can be drawn heorecally or emprcally o he sudy underaken. Ths sudy conrbues o he exsng leraure by makng a humble aemp a examnng he long and he shor run relaonshp beween he BRICS sock marke, he polcy uncerany n he US, rsk averson and he neres raes.

10 III. Underpnnngs, Daa and Mehodology: Underpnnngs based on he above leraure: The emergng markes over he years have been buldng up he srengh of her equy markes (lqudy and deph), however hey reman heavly dependen on he foregn money flow. Consderng hs s majorly n he form of ho/speculave money, nvesors fnd avenues o borrow cheapand nves n emergng economy eques whch offer consderable hgher reurns. However durng mes of he rsk-off rades (As seen durng he US subprme crss ec.) hs money also quckly fnds s way back, whereby hs leads o a negave mpac on hese sock markes. Also he BRICS equy markes also mpac each oher as he money flows a mos mes move n andem and mes are subsuve (due o relave srengh of he economes). Through hs sudy we would lke o examne f he BRICS equy markes (proxed by he BRICS ndces), neres / borrowng cos (proxed by 3MLbor) paerns, rsk-off rades (rsk averson - proxed by he VIX ndex) and he polcy uncerany n he US (proxed by he US polcy uncerany ndex) have a long erm relaonshp. Daa: The monhly reurn daa over 2000: o 205:3, peranng o he sudy has been colleced from four dfferen sources. The daa on he sock marke ndces of Brazl (Bovespa Toal Reurn Index), Russa (Russa MICEX-0 Index), Inda (NSE CNX NIFTY Index), Chna (Shangha SE Compose Index), Souh Afrca (FTSE/JSE All Share Index) and he 3monh Lbor are obaned from he Thomson Reuers Daasreamdaabase. Daa of economc polcy uncerany ndex n

11 Uned Saes and CBOE VIX s obaned from he Economc PolcyUncerany Index websewww.polcyuncerany.comconsruced by Baker, Bloom, and Davs (202) and he CBOE webse respecvely. Insead of opng o ake one sngle ndex such as he S&P BRIC 40 or MSCI BRIC as a proxy for all he BRICS sock ndces, we have ncluded each of he BRICS sock ndex separaely as we ancpae ha movemen of capal no one marke would also effec he ohers. Mehodology: Ths sudy employs a me seres echnque, n parcular, Auoregressve Dsrbued Lag (ARDL) conegraon mehod, n order o fnd emprcal evdence of he naure of relaons beween BRICS equy markes and he facors as alluded o n he nroducory paragraphs. Ths mehod has been preferred over radonal regresson mehod for he followng reasons: Sock markes ndces lke mos oher fnance varables are non-saonary. Ths would enal ha performng an ordnary regresson on he varables wll render he resuls msleadng as when sascal ess lke -raos and F sascs are no sascally vald when appled o non-saonary varables. Performng regressons on he dfferenced form of hese varables wll solve he above problem, however hs would lead o an even graver msake. When varables are regressed n her dfferenced form, he long erm rend s effecvely removed. Thus, he regresson only capures shor erm, cyclcal or seasonal effecs. Under hs suaon, he regresson s no really esng long erm (heorecal) relaonshps Under radonal regresson, he endogeney and exogeney of varables s predeermned by he researcher, usually on he bass of heory. Consderng he above sudy

12 and as seen n he leraure revew here s noable absence of esablshed heores apar from probably rsk averson. Conegraon echnques are advanageous n a way ha does no presume varable endogeney and exogeney. The daa deermnes whch varables are exogenous, and whch are exogenous. Conegraon echnques for he lack of words, embrace he dynamc neracon beween varables whereas radonal regresson mehods, exclude or dscrmnae agans neracon beween varables. Even hough convenonal conegrang procedure has made an mporan advance on regresson analyss, he conegrang esmaes also are subjec o a number of lmaons(mash e al, 2008). The esmaes derved from he conegrang ess (such as he Johansen es) and he un roo ess (such as, he Augmened Dcky-Fuller, Phllps-Peron, Kwakowsk-Phllps- Schmd-Shn ec. whch precede he conegrang ess), are found o be based. The ess lack power and are based n favor of accepng he null hypohess. The conegraon ess requre he varables o be I() bu he order of negraon of a varable, wheher I() or I(0), may depend on he number of lags ncluded or wheher he nercep and/or he rend are ncluded or excluded n he un roo ess. Moreover, he Johansen conegrang ess have small sample bas and smulaney bas among he regressors. To ge around he above lmaons of he un roo and conegraon ess, hs sudy uses he Auo Regressve Dsrbuve Lag (ARDL) mehod (bounds esng approach), proposed by Pesaran-Shn-Smh (200). Ths approach also does no requre he resrcon mposed by

13 conegraon echnque ha he varables are I() or I(0), whch s he case wh he daa n he sudy. (Ths s seen when he varables have been esed o ensure ha hey are no I(2) - Appendx) The exsence of long-run relaonshp among varables s done by consrucng an unresrced error correcon model (UECM) wh each varable n urn as a dependen varable and hen esng wheher or no he lagged levels of he varables n each of he error correcon equaons are sascally sgnfcan (.e., wheher he null of no long run relaonshp s acceped or rejeced ).The es consss of compung an F-sasc esng he jon sgnfcance of he lagged levels of he varables n each of he above error-correcon form of he equaon. The compued F-sasc s hen compared o wo asympoc crcal values. If he es sasc s above an upper crcal value, he null hypohess of no long-run relaonshp can be rejeced regardless of wheher he varables are I(0) or I(). When he es sasc falls below a lower crcal value, he null hypohess of no long-run relaonshp s acceped regardless of wheher he varables are I(0) or (). If he es sasc falls beween hese wo bounds, he resul s nconclusve. If all he F-sascs n all equaons happen o be nsgnfcan, hen ha mples he accepance of he null of no long run relaonshp among he varables. However, f a leas one of he F- sascs n he error-correcon equaons s sgnfcan, hen he null of no long-run relaonshp among he varables s rejeced. In ha case here s a long run relaonshp among he varables. When he F-sasc s sgnfcan, he correspondng dependen varable s endogenous and when he F-sasc s nsgnfcan, he correspondng dependen varable s

14 exogenous or called long-run forcng varable. (For he daa under consderaon he resulsare par of he Appendx) Afer demonsraed of he long run relaonshp, we can move on o he nex sage of he analyss nvolvng he long rung coeffcens esmaon(afer selecng he opmum order of he varables hrough AIC or SBC crera) and hen esmae he assocaed error correcon model n order o esmae he adjusmen coeffcens of he error-correcon erm. As he daa used by us s monhly, and consderng he varables are equy ndces we expec relavely faser adjusmen and hence have chosen four for he maxmum order of he lags n ARDL model. The error correcon verson of he ARDL (4, 4,4, 4, 4, 4, 4, 4) ha we have esmaed s: u a LVIX LLBR LPUI LJSE LSHC LNIF LMIC LBOV DVIX h DLBR DPUI g DJSE c DSHC f DNIF e DMIC d b DBOV DBOV (e-) - lagged error correcon erm whch would be derved from he ECM model would ell us how long wll ake o ge back o long erm equlbrum gven a devaon. The coeffcen represens proporon of mbalance correced n each perod. The lag srucure approprae o he ECM s deermned by Schwarz Bayesan Crera (SBC), Akake Informaon Crera (AIC), and Adjused LR Tes. IV. Emprcal Resuls and Dscussons:. Un Roo Tess:

15 We begn our emprcal esng by deermnng ha he varables used n he sudy aren I(2) Saonary only n he second dfferenced form and no n he level or frs dfferenced form. In order o proceed wh he ARDL echnque our varables can be eher I(0) or I() saonery n her level form or saonary n her frs dfferenced form. The dfferenced form for each varable used s creaed by akng he dfference of her log forms. For example, DBOV = LBOV LBOV-.We hen conduced he Augmened Dckey-Fuller (ADF) he Phlps Perron (PP) and he Kwakowsk Phllps Schmd Shn(KPSS) es on each varable (n boh level and dfferenced form). The able below summarzes he resuls. Below s a summary of he ADF es for he resuls of he PP & KPSS kndly refer o he Appendx. Table : Summary of he ADF es: Varable Tes Sasc Crcal Value Implcaon Varables n Level Form LBOV Varable s non-saonary LJSE Varable s non-saonary LLBR SBC Varable s non-saonary AIC Varable s non-saonary LMIC SBC Varable s non-saonary AIC Varable s non-saonary LNIF Varable s non-saonary LPUI SBC Varable s non-saonary AIC Varable s non-saonary LSHC SBC Varable s non-saonary AIC Varable s non-saonary LVIX Varable s non-saonary Varables n Dfferenced Form DBOV Varable s saonary DJSE Varable s saonary DLBR Varable s saonary Varable s saonary DMIC Varable s saonary DNIF Varable s saonary DPUI SBC Varable s saonary AIC Varable s saonary

16 Varable Tes Sasc Crcal Value Implcaon DSHC SBC Varable s saonary AIC Varable s saonary DVIX Varable s saonary Relyng prmarly on he AIC and SBC crera, he concluson ha can be made from he above resuls s ha all varablesbeng used for hs analyss are I() (apar from PUI whch s I(0) as per he PP). Also KPSS has conflcng resuls o he saonary of many varables n he level form hs s ye anoher reason for opng for he ARDL approach raher han he sandard me seres approach. Noe ha n deermnng whch es sasc o compare wh he 95% crcal value for he ADF sasc, we have seleced he ADF regresson order based on he hghes compued value for AIC and SBC. In some nsances, AIC and SBC gve dfferen orders and n ha case, we have aken dfferen orders and compared boh (for example, hs apples o he varable LPUI, LLBR and LMIC, see he able above). Ths s no an ssue as n all cases, he mplcaons are conssen. 2. Selecng he lag lengh: In order o esmae he ARDL regresson, selecon of he lag lengh s mporan. The es runs over 4 lags lengh of,2,3 and 4 for he opmum lags. Based on he AIC, SBC and he Adjused LR es as per Table, lag lengh of has been deermned. Thus lag has been furher used. Table 2: Tes Sascs and Choce Crera for Selecng he Order of he VAR Model Order LL AIC SBC LR es Adjused LR es CHSQ(64)= [.050] [.604]

17 CHSQ(28)= [.000] [.277] CHSQ(92)= [.000] [.334] CHSQ(256)= [.000] [.486] * 433.5* CHSQ(320)= [.000] [.77]* CHSQ(384)= [.000] [.000] 3. Tesng long run relaonshp beween he varables: F-sascs for each equaon: F ( LBOV LMIC, LNIF, LSHC, LJSE, LPUI, LLBR, LVIX ) = F ( LMIC LBOV, LNIF, LSHC, LJSE, LPUI, LLBR, LVIX ) = F ( LNIF LBOV, LMIC, LSHC, LJSE, LPUI, LLBR, LVIX ) =.738 F ( LSHC LBOV, LMIC, LNIF, LJSE, LPUI, LLBR, LVIX ) =.637 F ( LJSE LBOV, LMIC, LNIF, LSHC, LPUI, LLBR, LVIX ) = F ( LPUI LBOV, LMIC, LNIF, LSHC, LJSE, LLBR, LVIX ) = F ( LLBR LBOV, LMIC, LNIF, LSHC, LJSE, LPUI, LVIX ) = F ( LVIX LBOV, LMIC, LNIF, LSHC, LJSE, LPUI, LLBR ) = TABLE 3: F-Sascs for Tesng he Exsence of Long-Run Relaonshp Compued F-Sasc LBOV * Crcal Values a 5 percen level Lower; upper 2.604; The crcal values are aken from Pesaran e al. (200), unresrced nercep and rend wh egh regressors. * denoes rejecng he null a 5 percen level. The range of he crcal value a percen and 0 percen are and respecvely.

18 As per he Table 3he calculaed F-sascs s hgher han he upper bound crcal value of a he 5% sgnfcance level, aleas for one equaon (LBOV). Ths mples ha he null hypohess of no conegrang long-run relaonshp can be rejeced. These resuls reveal ha a long-run relaonshp exss beween Polcy Uncerany n he US, he Rsk Averson, he Ineres Raes and he BRICS equy ndces. The evdence of long run relaonshp rules ou he possbly of any spurous relaonshp exsng beween he varables. In oher words, here s a heorecal relaonshp exsng beween he varables. 4. Esmang long run coeffcens: The Error Correcon Model s represenaon of he ARDL model s seleced usng he Akake Informaon Creron. Followng ables provde he esmaes of he ARDL long run coeffcen for he model. As we are ryng o undersand he mpac of he varables on each of he BRICS markes, Table represen he resulsofesmaedlong-runcoeffcensusnghe ARDL Approach wh each marke as a dependen varable. TABLE 4.: ResulsofEsmaedLong-RunCoeffcensusnghe ARDL Approach LBOV (DEP) Independen Sandard Coeffcen Varable Error P-Value LJSE * LLBR LMIC * LNIF * LPUI LSHC LVIX INPT * Noe: * denoes sgnfcan a 5 percen level

19 The esmaed long run coeffcens of he long run relaonshp above show ha he Johannesburg Sock Exchange - JSE, Mcex and Nfy have sgnfcan effecs on he performance of he Bovespa. The coeffcen of he Nfy mples ha a % ncrease n reurns on he Nfy on an average leads o a.2% ncrease n he Bovespa, all hngs beng equal. Ths effec s also smlar o he Mcex, whereby a % ncrease n he Mcex would lead o a 0.4% ncrease n he Bovespa. Ths suggess ha hese markes complemen each oher, whereas s he oppose for he Johannesberg sock exchange whereby a % ncrease n he JSE leads o he Bovespa o drop by 0.78%. Wha s however seen s ha he US Polcy uncerany, he VIX and he Lbor are no sascally sgnfcan and hus do no mpac he Bovespa. TABLE 4.2: ResulsofEsmaedLong-RunCoeffcensusnghe ARDL Approach LMIC (DEP) Independen Varable Coeffcen Sandard Error P-Value LNIF LBOV LJSE LSHC LLBR LPUI LVIX INPT Table 4.2 suggess ha none of he varables n he model are sgnfcan and hus have no mpac on he MICEX. Ths pons ou o oher facors whch drve he MICEX such as s growh raher han ncluded varables. As hs ndex and he Russan economy s a commody drven economy, could be he case ha facors such as ol and oher commody markes drve.

20 TABLE 4.3: ResulsofEsmaedLong-RunCoeffcensusnghe ARDL Approach LNIF (DEP) Independen Sandard Coeffcen Varable Error P-Value LBOV * LJSE * LMIC LSHC LLBR LPUI ** LVIX INPT Noe: * denoes sgnfcan a 5 percen level **denoes sgnfcan a 0 percen level The Nfy as per Table 4.3 n he long erm s mpaced by he Bovespa and he Johannesburg Sock Exchange. I could be he case ha foregn porfolo nvesmens no and ou of hese hree counres happens n andem, whch s depced by he coeffcens of LBOV and LJSE. A % ncrease n he Bovespa and he JSE leads o an app 0.45% and 0.47% ncrease respecvely n he Nfy. Besdes he US Polcy uncerany ndex s sgnfcan and negave, hs mples ha he Indan sock markes do observe economc polcy condons n he US and a % ncrease n he uncerany leads o he marke o go down by 0.47%. Ths could also ndcae ha he Nfy s negraed wh he US and consderng he Foregn Insuonal flows orgnang from he US funds no Inda, s no a surprse If here s uncerany n he home counry, funds and people would wan o ge ou from emergng markes lke Inda. TABLE 4.4: ResulsofEsmaedLong-RunCoeffcensusnghe ARDL Approach SHC (DEP) Independen Sandard Coeffcen Varable Error P-Value LNIF ** LBOV

21 LJSE LMIC LLBR LPUI LVIX INPT Noe: * denoes sgnfcan a 5 percen level **denoes sgnfcan a 0 percen level The above able shows ha none of he varables apar from he Nfy are sgnfcan. Consderng Chna and Inda have been wo economes whch have ouclassed he ohers n he BRICS, hey have been major benefcares of Foregn Insuonal mones. Ths could be he reason whereby a % ncrease n he Nfy would lead o a 3% ncrease n he Shangha Compose and veceversaa fall as well. The Chnese markes have also been sufferng from he lack of ransparency, whch has led o money movemen o oher markes lke Inda. TABLE 4.5: ResulsofEsmaedLong-RunCoeffcensusnghe ARDL Approach JSE (DEP) Independen Sandard Coeffcen Varable Error P-Value LMIC LNIF LBOV LSHC LLBR LPUI LVIX INPT Souh Afrca has been a recen addon o he BRICS and probably ha s one reason none of he varables are sgnfcan. Beng a par of hs group would mean ha a number of Exchange Traded Funds (ETF s) and Emergng Marke funds would make Souh Afrcan eques par of her porfolo, however as dscussed beng a recen enry hs may no reflec n he long run

22 equaon. Thus he major deermnan of hs ndex would be he counry s nernal facors, GDP, macro-economc performance ec. 5. Error Correcon Models: A long run relaonshp beween he varables s ndcaed by conegraon, however here could be a shor-run devaon from he long-run equlbrum. Conegraon does no unfold he process of shor-run adjusmen o brng abou he long-run equlbrum. The error correcon model n Tables help us o undersand hs. The p value of he error-correcon coeffcen ndcaes f he devaon from equlbrum (represened by he error-correcon erm) has a sgnfcan feedback effec on he dependen varable (.e. each of he BRICS equy ndces)..e. If he dependen varable s endogenous or exogenous. The error-correcon coeffcen beng sgnfcan confrms he sgnfcan long-run conegrang relaonshp beween he varables. Also he speed of shor-run adjusmen of he dependen varable o brng abou he long-run equlbrum s ndcaed by he sze of he coeffcen of he errorcorrecon erm. The sze of he coeffcen of he error-correcon erm s also ndcave of he nensy of he arbrage acvy o brng abou he long-run equlbrum. Table 5. ResulsofErrorCorreconModels ΔLBOV (DEP) Independen Sandard Coeffcen Varable Error P-Value ΔLJSE * ΔLLBR ΔLMIC * ΔLNIF ΔLPUI ΔLSHC ΔLVIX *

23 Ecm(-) * Noe: * denoes sgnfcan a 5 percen level Table 5.2 ResulsofErrorCorreconModels ΔLMIC (DEP) Independen Sandard Coeffcen Varable Error P-Value ΔLNIF ΔLBOV * ΔLJSE * ΔLSHC ΔLLBR ΔLPUI ΔLVIX ecm(-) * Noe: * denoes sgnfcan a 5 percen level

24 Table 5.3 ResulsofErrorCorreconModels ΔLNIF (DEP) Independen Sandard Coeffcen Varable Error P-Value ΔLBOV ΔLJSE ΔLMIC ΔLSHC ΔLLBR ΔLPUI ΔLVIX * ecm(-) * Noe: * denoes sgnfcan a 5 percen level Table 5.4 ResulsofErrorCorreconModels ΔLSHC (DEP) Independen Sandard Coeffcen Varable Error P-Value ΔLNIF ΔLBOV * ΔLJSE ΔLMIC ΔLLBR ΔLPUI ΔLVIX ecm(-) * Noe: * denoes sgnfcan a 5 percen level Table 5.5 ResulsofErrorCorreconModels ΔLJSE (DEP) Independen Sandard Coeffcen Varable Error P-Value ΔLMIC * ΔLNIF ΔLBOV * ΔLSHC ΔLLBR * ΔLPUI ΔLVIX ecm(-) Noe: * denoes sgnfcan a 5 percen level

25 The error correcon erms of ΔLBOV -> (0.000), ΔLMIC -> (.004), ΔLNIF -> (0.05), and ΔLSHC -> (0.020)are sgnfcan and also have he correc sgn, hs mples a moderae speed of adjusmen afer a shock. In he above cases 5.2%, 7%, 2.5% and 6% of he prevous perod s (monhs) shocks adjuss o he long run equlbrum n he curren quarer. Also he p values of he coeffcens of he dfferenced varables ndcae f he effecs of hese varables on he ndvdual BRICS markes are sgnfcan. We broadly fnd smlar sgnfcan effecs of he oher BRICS markes as seen n he long run, however n he ΔLJSE ->ΔLLBR and ΔLBOV / ΔLNIF -> ΔVIX sgnfcan n he shor run. These ndcae ha n he shor run he rsk off rade does affec he Brazlan and he Indan sock markes and he leveragng n he case of he Souh Afrcan equy marke. 6. Varance Decomposon: Varance decomposon (VDC) helps us asceran relave endogeney and exogeney. VDC decomposes he varance of forecas error of each varable no proporons arbuable o shocks from each varable n he sysem, ncludng s own. The leas endogenous varable s hus he varable whose varaon s explaned mosly by s own pas varaons. I frs apply orhogonalzed VDCs and obaned he followng resuls. Consderng he daa s on sock marke ndces, we forecas for a me horzon of 2 (monhs).e. a year.

26 DBOV DJSE DLBR DMIC DNIF DPUI DSHC DVIX DBOV DJSE DLBR DMIC DNIF DPUI DSHC DVIX For he above able, rows read as he percenage of he varance of forecas error of each varable no proporons arbuable o shocks from all varables (n columns), ncludng s own. The columns read as he percenage n whch ha varable conrbues o oher varables n explanng observed changes. The dagonal lne of he marx (hghlghed) represens he relave exogeney. Accordng o hese resuls, he rankng of ndces by degree of exogeney (exen o whch varaon s explaned by s own pas varaons) s as per he able below: No. Varable DPUI 2 DNIF 3 DLBR 4 DSHC 5 DMIC 6 DBOV 7 DJSE 8 DVIX However he resuls above gve conradcory resuls o he VECM. Thus we need o recognze wo mporan lmaons of orhogonalzed VDCs. I assumes ha when a parcular varable s shocked, all oher varables are swched off

27 More mporanly, n orhogonalzed VDCs he generaed numbers are dependen upon he orderng of varables n he VAR. Thus, he frs varable would repor he hghes percenage ands lkely o be specfed as he mos exogenous varable. Consderng hs lmaon, we decded o rely nsead on Generalzed VDCs, whch are nvaran o he orderng of varables. In nerpreng he numbers generaed by he Generalzed VDCs, we needed o perform addonal compuaons. Ths s because he numbers do no add up o 00% or as n he case of orhogonalzed VDCs. For a gven varable, a a specfed horzon, we oaled up he numbers of he gven row and we hen dvde he number for ha varable (represenng magnude of varance explaned by s own pas) by he compued oal. In hs way, he numbers n a row wll now add up o.0 or 00%. The ables below show he resul, we forecas for a me horzon of 2 (monhs).e. a year. DBOV DJSE DLBR DMIC DNIF DPUI DSHC DVIX DBOV DJSE DLBR DMIC DNIF DPUI DSHC DVIX Accordng o hese resuls, he rankng of ndces by degree of exogeney (exen o whch he varaon s explaned by s own pas varaons) s as per he able below:

28 No. Varable DLBR 2 DSHC 3 DPUI 4 DNIF 5 DVIX 6 DMIC 7 DJSE 8 DBOV The above resuls are slghly off wh he resuls as per he VECM, whereby as per he VDC he Shangha Compose s he second mos exogenous varable and he VIX ndex s he second mos endogenous varable. However hese resuls by hemselves may no be relable as all he varable are forced wh he same number of lags whch s no he case wh ARDL, where he opmum number of lags are assgned o each varable. Thus usng he frs approach o fnd relave endogeney/exogeney may no be approprae. 7. Impulse Response: The mpulse response funcons (IRFs) essenally produces he same nformaon as he VDCs, excep ha hey can be presened n graphcal form. Raher han shockng all he varables, n order o make he exercse meanngful below we shock only he exogenous varables of Polcy Uncerany, Lbor and he VIX (As per he VECM) and observe he effecs on he oher varables. Wha can be seen from he graphs below s ha all he varables rever back he equlbrum whn a perod rangng from wo o seven monhs.

29 Orhogonalsed Impulse Responses o one SE shock n he equaon for DVIX DBOV DJS E DLBR DMIC DNIF DPUI DS HC DV IX Orhogonalsed Impulse Responses o one SE shock n he equaon for DPUI DBOV DJS E DLBR DMIC DNIF DPUI DS HC DV IX Orhogonalsed Impulse Responses o one SE shock n he equaon for DLBR DBOV DJS E DLBR DMIC DNIF DPUI DS HC DV IX

30 Generalsed Impulse Responses o one SE shock n he equaon for DVIX DBOV DJS E DLBR DMIC DNIF DPUI DS HC DV IX Generalsed Impulse Responses o one SE shock n he equaon for DPUI DBOV DJSE DLBR DMIC DNIF DPUI DSHC DVIX Generalsed Impulse Responses o one SE shock n he equaon for DLBR DBOV DJS E DLBR DMIC DNIF DPUI DS HC DV IX

31 8. Concludng Remarks and Polcy Implcaons: Brazl, Russa, Inda, Chna and Inda, known as he BRIC counres form a sgnfcan par and play an mporan role n he world economy. Due o he ncreased economc and fnancal negraon whn he world economy shocks orgnang from he advanced world such as he US can have a sgnfcan mpac on he BRIC s economes. Based on he daa and he resul analyss, seems ha alhough here s co-negraon amongs hese equy markes and he varables depcng polcy uncerany and rsk averson, hey only sgnfcanly mpac few of he ndces n he shor run. Mos of hese markes, over he years have been ryng o develop nsuons and domesc real nvesors o form a back-up o he ho money movng n and ou of hese markes. I could be he case ha hey are succeedng n dong so, also could be he case ha consderng he nuances of he lack of developmen, ransparency, lqudy ec. has been keepng away large nsuonal sources of money away from he BRICS markes. Ths sudy conrbues o furher he undersandng of global ransmsson of economc and fnancal shocks. The fndng suggess ha he sock marke performance n Brazl, Russa, Chna and Souh Afrca are no lnked o he polcy uncerany and rsk averson rades n he U.S. However he fndngs mply ha marke parcpans n he Indan sock markes do observe economc polcy condons n he US. Anoher vew could be he case ha n he long run he facors exernal o he economes do no affec he BRICS markes much, whch could have mplcaon o he nvesors n he developng world. Thus he BRICS equy markes can be looked a as a grea dversfcaon sraegy o he developed world.

32 References: Ahmad, A. H., Nurazra, M. D. S., &Anulashkn, M. (2008). Soveregn Cred Rangs and Macroeconomc Varables: An Emprcal Analyss on Dynamc Lnkages n Malaysa Usng Bound Tes Approach. The IUP Journal of Appled Economcs, 7(6), Al Awad, M., & Goodwn, B. K. (998). Dynamc lnkages among real neres raes n nernaonal capal markes. Journal of Inernaonal Money and Fnance, 7(6), Alou, R., Aïssa, M. S. B., & Nguyen, D. K. (20). Global fnancal crss, exreme nerdependences, and conagon effecs: The role of economc srucure?.journal of Bankng & Fnance, 35(), Aye, G. C., Balclar, M., Gupa, R., Klman, N., Nakumuryango, A., & Redford, S. (204). Predcng BRICS sock reurns usng ARFIMA models. Appled Fnancal Economcs, 24(7), Bansal, R., Khacharan, V., &Yaron, A. (2005). Inerpreable asse markes?.european Economc Revew, 49(3), Bekros, S. D. (203). Conagon, decouplng and he spllover effecs of he US fnancal crss: Evdence from he BRIC markes. Inernaonal Revew of Fnancal Analyss, 33, Berger, D., & Turle, H. J. (20). Emergng marke crses and US equy marke reurns. Global Fnance Journal, 22(), Celık, S. (202). The more conagon effec on emergng markes: The evdence of DCC-GARCH model. Economc Modellng, 29(5), Cheng, H. F., Guerrez, M., Mahajan, A., Shachmurove, Y., & Shahrokh, M. (2007). A fuure global economy o be bul by BRICs. Global Fnance Journal, 8(2), Chnn, M., & Frankel, J. A. (2007). Wll he euro evenually surpass he dollar as leadng nernaonal reserve currency?. In G7 Curren accoun mbalances: susanably and adjusmen (pp ). Unversy of Chcago Press. Chkl, W., & Nguyen, D. K. (204). Exchange rae movemens and sock marke reurns n a regme-swchng envronmen: Evdence for BRICS counres. Research n Inernaonal Busness and Fnance, 3, Cooper, I., & Presley, R. (2009). Tme-varyng rsk premums and he oupu gap. Revew of Fnancal Sudes, 22(7),

33 De Vres, G. J., Erumban, A. A., Tmmer, M. P., Voskoboynkov, I., & Wu, H. X. (202). Deconsrucng he BRICs: Srucural ransformaon and aggregae producvy growh. Journal of Comparave Economcs, 40(2), Dmrou, D., Kenourgos, D., & Smos, T. (203). Global fnancal crss and emergng sock marke conagon: A mulvarae FIAPARCH DCC approach.inernaonal Revew of Fnancal Analyss, 30, Dzelnsk, M. (20). Measurng economc uncerany and s mpac on he sock marke. Fnance Research Leers, 9(3), Ehrmann, M., Frazscher, M., &Rgobon, R. (20). Socks, bonds, money markes and exchange raes: measurng nernaonal fnancal ransmsson. Journal of Appled Economercs, 26(6), Ehrmann, M., &Frazscher, M. (2006). Global fnancal ransmsson of moneary polcy shocks*. Oxford Bullen of Economcs and Sascs, 7(6), Ehrmann, M., &Frazscher, M. (2005). Equal Sze, Equal Role? Ineres Rae Inerdependence Beween he Euro Area and he Uned Saes*. The Economc Journal, 5(506), Forbes, K. J., & Chnn, M. D. (2004). A decomposon of global lnkages n fnancal markes over me. Revew of Economcs and Sascs, 86(3), Frank, W. P., & Frank, E. C. (200). Inernaonal busness challenge: can he BRIC counres ake world economc leadershp away from he radonal leadershp n he near fuure. Inernaonal Journal of Ars and Scences, 3(3), Hammoudeh, S., Sar, R., Uzunkaya, M., & Lu, T. (203). The dynamcs of BRICS's counry rsk rangs and domesc sock markes, US sock marke and ol prce. Mahemacs and Compuers n Smulaon, 94, Hausman, J., &Wongswan, J. (20). Global asse prces and FOMC announcemens. Journal of Inernaonal Money and Fnance, 30(3), Hwang, E., Mn, H. G., Km, B. H., & Km, H. (203). Deermnans of sock marke comovemens among US and emergng economes durng he US fnancal crss. Economc Modellng, 35, Kenourgos, D., Samas, A., &Palalds, N. (20). Fnancal crses and sock marke conagon n a mulvarae me-varyng asymmerc framework. Journal of Inernaonal Fnancal Markes, Insuons and Money, 2(),

34 Kng, M. A., &Wadhwan, S. (990). Transmsson of volaly beween sock markes. Revew of Fnancal sudes, 3(), Manamper, N. (204). The shor and long-run dynamcs beween nflaon and economc growh n BRICS. Appled Economcs Leers, 2(2), Mash, M., &Algahan, I. (2008). Esmaon of long-run demand for money: An applcaon of long-run srucural modellng o Saud Araba. EconomaInernazonale/Inernaonal Economcs, 6(), Menzly, L., Sanos, T., & Verones, P. (2004). Undersandng predcably. Journal of Polcal Economy, 2(), -47. Ono, S. (20). Ol prce shocks and sock markes n BRICs. The European Journal of Comparave Economcs, 8(), Alou e al. (20) Ozoguz, A. (2009). Good mes or bad mes? Invesors' uncerany and sock reurns. Revew of Fnancal Sudes, 22(), Pesaran, M. H., Shn, Y., & Smh, R. J. (200). Bounds esng approaches o he analyss of level relaonshps. Journal of appled economercs, 6(3), Pradhan, R. P., Dasgupa, P. R. A. T. E. E. K., &Samadhan, B. (203). Fnance developmen and economc growh n BRICS: A panel daa analyss. Journal of Quanave Economcs, (&2), Sarwar, G. (202). Is VIX an nvesor fear gauge n BRIC equy markes?.journal of Mulnaonal Fnancal Managemen, 22(3), Śledzk, K. (202). BRICS, MIST (MIKT) or G5-Whch'Club'of Economes Has he Bgges Growh Poenal?.Avalable a SSRN Sollnger, R., Pndyuk, O., Hunya, G., Dachs, B., Lennon, C., Poplawsk Rbero, M.,...& Chrse, E. (2009). EU and BRICs: Challenges and opporunes for European compeveness and cooperaon. EUR-OP. Sum, V. (202). Economc polcy uncerany and sock marke performance: evdence from he European Unon, Croaa, Norway, Russa, Swzerland, Turkey and Ukrane. Journal of Money, Invesmen and Bankng, 25, Syropoulos, T., &Roumps, E. (2009). Asse allocaon and value a rsk n shppng equy porfolos. Marme Polcy & Managemen, 36(),

35 Syropoulos, T., Makram, B., &Boubaker, A. (203). Sock marke volaly spllovers and porfolo hedgng: BRICS and he fnancal crss. Inernaonal Revew of Fnancal Analyss, 39, 7-8. Wongswan, J. (2006). Transmsson of nformaon across nernaonal equy markes. Revew of Fnancal Sudes, 9(4), Wlson, D., &Purushohaman, R. (2003). Dreamng wh brcs: The pah o 2050 (goldman&sachs, global economcs paper no. 99). Goldman Sachs. Zhang, B., L, X., & Yu, H. (203). Has recen fnancal crss changed permanenly he correlaons beween BRICS and developed sock markes?.the Norh Amercan Journal of Economcs and Fnance, 26,