Country Risk Analysis in Emerging Markets: The Indian Example BY Sankarshan Basu Associate Professor Finance & Control Indian Institute of Management Bangalore Bannerghatta Road, Bangalore 5600 76 Ph. 080-26993078 sankarshanbiimb.ernet.in D. Deepthi Student, IIMB & Jyothsni Reddy Student, IIMB Year oI Publication 2011 2 CCUN1k kISk ANALIS IN LMLkGING MAkkL1S: 1HL INDIAN LkAMLL Abstract: 1he 8eta Country k|sk Mode|, as oescrlbeo by Lrb, Parvey ano vlskanLa (1996) ano useo by Anoraoe ano 1eles (2004) for 8razll, ls useo Lo esLlmaLe Lhe counLry rlsk of lnola baseo on several macroeconomlc lnolcaLors. Crolnary leasL squares regresslon ls run on Lhe whlLe nolse (unexpecLeo componenL) of Lhese varlables Lo explaln Lhe varlaLlon ln counLry rlsk Lo loenLlfy Lhe mosL relevanL of Lhese varlables. 1he sLuoy shows LhaL Lhe varlaLlon ln counLry rlsk of lnola ls hlghly correlaLeo wlLh changes ln lul flows, lnLeresL raLes (moneLary pollcy), exchange raLes ano Lhe unemploymenL raLe. 1he effecL of pollLlcal rlsk on overall counLry rlsk ls also sLuoleo. key words: counLry rlsk, counLry beLa mooel, rlsk mooellng 3 1. Introduct|on CloballzaLlon ano lncreaslng flnanclal unlflcaLlon has leo Lo a raplo growLh of lnLernaLlonal lenolng, forelgn olrecL ano lnsLlLuLlonal lnvesLmenL. WlLh Lhls, economles across Lhe globe are lncreaslngly becomlng lnLeroepenoenL ano oevelopmenLs ln one parL of Lhe worlo affecL reLurns ln anoLher. Clven Lhls, counLry rlsk analysls provloes lnslghLs lnLo LhaL parL of Lhe rlsk of an lnvesLmenL speclflc Lo a cerLaln counLry. Country k|sk", ln general refers Lo Lhe rlsk assoclaLeo wlLh Lhose facLors LhaL oeLermlne or affecL Lhe ablllLy ano wllllngness of a soverelgn sLaLe or borrower from a parLlcular counLry Lo fulflll Lhelr obllgaLlons Lowaros one or more forelgn lenoers ano / or lnvesLors, Lhls ls Lhe approach ano Lhe oeflnlLlon useo by 8aLes ano Salnl (1984) as well as by Abassl ano 1affler (1982). 1hls shall also be Lhe oeflnlLlon useo ln Lhls paper. 1he analysls of counLry rlsk conslsLs of Lhe assessmenL of Lhe pollLlcal, economlc ano flnanclal facLors of a borrowlng counLry or lul 1 hosL. 1hese facLors glve an lnolcaLlon of Lhe sLablllLy ano proflLablllLy ln an economy. As Parvey ano vlskanLa (1996) polnL ouL, non - olverslflable sysLemlc rlsk" arlses ouL of Lhe facLors over whlch borrowers have llLLle conLrol, ano counLry rlsk may also represenL such non-olverslflable sysLemaLlc rlsk". Lmerglng MarkeLs counLry rlsk analysls provloes a challenge for researchers, accorolng Lo Luler Permes 2 , slnce calculaLlon of sLaLlsLlcal properLles of Lhe varlous parameLers baseo on hlsLorlcal reLurns coulo be mlsleaolng. ln aoolLlon, rellable oaLa ls noL avallable for several perloos, especlally far back lnLo Lhe pasL. Such oaLa mlghL noL even be relevanL as, by Lhelr very naLure, Lhe pasL ln emerglng economles rarely reflecLs Lhe presenL ano Lo a lesser exLenL, Lhe fuLure. 1he Country 8eta Approach ls a quanLlLaLlve meLhoo of counLry rlsk analysls ln whlch Lhe olfference beLween Lhe reLurns of a counLry's equlLy markeL ano Lhe worlo equlLy markeL ls aLLrlbuLeo Lo Lhe counLry rlsk. 1hls olfference lnolcaLes Lhe reLurns ln a counLry speclflc Lo lL ano olfferenL from Lhe resL of Lhe worlo. 1hls mooel has flrsL been oescrlbeo ln Lhe semlnal paper by Lrb, Parvey ano vlskanLa (1996b). 1hls mooel has been applleo Lo AusLralla by Cangeml, 8rooks ano laff (2000) Lo examlne Lhe effecLs of forelgn oebL on counLry rlsk, Lo LaLln Amerlca by verma ano Syoermlr (2006) Lo sLuoy Lhe economlc oeLermlnanLs of a Llme-varylng counLry beLa ano Lo 8razll by Anoraoe ano 1eles (2004) Lo sLuoy Lhe effecL of lnLeresL raLes. 8uL such an analysls has noL been prevlously oone for lnola. lnola provloeo an lnLeresLlng case for counLry rlsk ano sLuoylng Lhose facLors affecLlng counLry rlsk ln an emerglng economy, Lhrough Lhe llberallzaLlon phase ln Lhe early 1990s. 1hls sLuoy examlnes Lhe relaLlonshlp beLween counLry rlsk ano macroeconomlc varlables ano loenLlfles Lhose varlables LhaL affecL counLry rlsk Lhe mosL, uslng Crolnary LeasL Squares (CLS) regresslon on Lhe whlLe nolse of Lhe varlables. ln aoolLlon, Lhe lmpacL of pollLlcal rlsk ls also sLuoleo. lL can been seen LhaL lul lnflows, lnLeresL raLes (moneLary pollcy), exchange raLes ano Lhe unemploymenL raLe lmpacL counLry rlsk Lhe mosL. SecLlon 2 glves a brlef hlsLory of Lhe 1 lorelgn olrecL lnvesLmenL ls LhaL lnvesLmenL, whlch ls maoe Lo serve Lhe buslness lnLeresLs of Lhe lnvesLor ln a company, whlch ls ln a olfferenL naLlon olsLlncL from Lhe lnvesLor's counLry of orlgln. 2 Luler Permes ls a lrench creolL lnsurance company. 1he arLlcle can be accesses aL hLLp://www.eulerhermes.com/en/press/press_20090112_00100060.hLml 4 sLuoles oone ln Lhls flelo. SecLlon 3 oescrlbes Lhe counLry beLa mooel, Lhe meLhooology useo Lo whlLe Lhe Llme serles of Lhe varlables ano Lhe flnal regresslon. SecLlon 4 glves an analysls of Lhe resulLs obLalneo uslng Lhls mooel. CerLaln llmlLaLlons ano fuLure scope for Lhls sLuoy are presenLeo ln secLlon S. 2. L|terature kev|ew CounLry rlsk analysls has been oeflneo ano sLuoleo ln several olfferenL ways slnce Lhe laLLer parL of Lhe prevlous cenLury. 8lbelro (2006) caLegorlzeo some sLanoaro economlc varlables LhaL ofLen coulo be founo ln mosL of Lhe olverse approaches aoopLeo by flnanclal lnsLlLuLlons ano raLlng agencles (such as Coloman Sachs, Merrlll Lynch, S& ano llLch 8aLlngs) lnLo Lxterna| sector (exporLs, lmporLs, oebL servlces, olrecL lnvesLmenLs, loans, repaymenL of loans, exLernal oebL ano flow of forelgn reserves), Interna| sector (lnLeresL raLe, publlc oebL ano lLs servlce, level of lnvesLmenLs, buogeL equlllbrlum, lnLernal savlngs, consumpLlon, Cu/Cn, lnflaLlon raLe, money supply, eLc) ano Cther var|ab|es (populaLlon, llfe expecLancy, raLe of unemploymenL, level of llLeracy, eLc). 1elxelra, kloLzle ano ness (2008), loenLlfleo Lhe oeLermlnanL facLors of Lhe counLry rlsk for selecLeo emerglng markeLs. 1hree mooels were useo Lo esLlmaLe counLry rlsk - ln Lhe flrsL mooel Lhe relaLlon beLween counLry rlsk ano funoamenLal economlc varlables was LesLeo, ln Lhe secono mooel Lhe exLernal componenL was be aooeo Lo Lhe group of explanaLory varlables, ano Lhe Lhlro mooel LesLeo Lhe relaLlon beLween speclflc counLry rlsk ano Lhe economlc funoamenLals. 1he resulLs founo for emerglng markeLs lnolcaLeo LhaL four oomesLlc facLors are conslsLenL oeLermlnanLs of counLry rlsk ano speclflc counLry rlsk - growLh raLe, exLernal oebL, publlc oebL ano lnLernaLlonal reserves. varlous meLhoos useo for counLry rlsk appralsal may be caLegorlzeo lnLo one of four Lypes - lully CuallLaLlve MeLhoo, SLrucLureo CuallLaLlve MeLhoo, CheckllsL MeLhoo ano CLher CuanLlLaLlve MeLhoos. 1he popular quanLlLaLlve meLhoos useo for counLry rlsk analysls are llsLeo by naLh (2008). ArLlflclal neural neLworks are exLenslvely useo for counLry rlsk analysls. ?lm ano MlLchell (2004) lnvesLlgaLeo Lhe posslblllLy of ouLperformance of LraolLlonal sLaLlsLlcal mooels by Lwo arLlflclal neural neLworks, mulLllayer percepLron ano hybrlo neLworks, for preolcLlng counLry rlsk raLlng. 1he resulLs ln sample lnolcaLe LhaL Lhe hybrlo Ann - Ann-LoglL- loglL - proouceo Lhe besL resulLs. 1hls supporLs Lhe concluslon LhaL for researchers, pollcymakers ano oLhers lnLeresLeo ln early warnlng sysLems, hybrlo neLworks woulo be useful. AnoLher novel mooel useo for counLry rlsk analysls ls Lhe counLry beLa mooel oescrlbeo by Lrb, Parvey ano vlskanLa (1996). 1hls mooel was applleo Lo esLlmaLe Lhe counLry rlsk of 8razll from 1991 Lo 2002, by Anoraoe ano 1eles (2004). 1he four varlables useo for Lhe mooel are forelgn reserves, worlo oll prlces, nomlnal lnLeresL raLe ano publlc oebL. 1hree olfferenL speclflcaLlons of Lhe mooel were analyzeo - one lncluolng all Lhe 4 varlables, one wlLhouL publlc oebL ano one wlLhouL lnLeresL raLe. 1he followlng observaLlons were maoe - one, Lhe effecLs of forex reserves ls very small slnce Lhe aoopLlon of Lhe floaLlng exchange raLe reglme, ano Lwo, unanLlclpaLeo lncrease ln lnLeresL raLes reouces counLry rlsk. 1hls paper uses Lhe same mooel, as has been useo by Lrb, Parvey ano vlskanLa (1996) ano Anoraoe ano 1eles (2004) for esLlmaLlng 8razlllan counLry rlsk, for analyzlng lnola's counLry rlsk. S 3. 1he Mode| Country 8eta Mode| of Lrb, Parvey ano vlskanLa (1996) ls oescrlbeo below. As sLaLeo earller, Lhls ls Lhe mooel useo ln Lhe sLuoy of 8razlllan counLry rlsk ano ls also useo Lo esLlmaLe lnola's counLry rlsk. 1he oaLa perloo for Lhe sLuoy ln Lhe lnolan conLexL ls beLween 1984 ano 2008. 3.1. 1he Country 8eta Mode| Lrb, Parvey ano vlskanLa (1996) have shown LhaL Lhe olfference beLween Lhe reLurns of a counLry's equlLy markeL ano Lhe worlo equlLy markeL may be aLLrlbuLeo Lo Lhe counLry rlsk. 1hls relaLlon may be expresseo as follows: 8 LqulLy_CounLry = o + 8 LqulLy_Worlo + e L (1) ls Lhe baslc measure of counLry rlsk, slnce lL lnolcaLes Lhe reLurns ln a counLry speclflc Lo lL ano olfferenL from Lhe resL of Lhe worlo. As lncreases, counLry rlsk oecreases, LhaL ls, Lhe reLurns ln Lhe counLry are affecLeo only by facLors common Lo Lhe resL of Lhe worlo, whlch ls essenLlally a non-olverslflable rlsk for a parLlcular counLry. CounLry rlsk woulo be a varlable affecLeo by cerLaln macroeconomlc varlables speclflc Lo Lhe counLry. 1hus, beLa ls mooeleo as a llnear comblnaLlon of Lhose varlables: = b 0 + b.k (2) where k represenLs a vecLor of macroeconomlc lnolcaLors. 1hls was applleo Lo Lhe lnolan conLexL ano Lhe followlng mooel was useo Lo esLlmaLe counLry rlsk: 8 lnola = o + 8 Worlo + e L (3) where 8 lnola ls Lhe reLurn on Lhe lnolan equlLy markeL ano 8 Worlo ls Lhe reLurn on Lhe worlo equlLy markeL. ls an lnolcaLor of lnola's counLry rlsk. As lncreases, counLry rlsk oecreases. 1he varlables LhaL go lnLo Lhe vecLor of macroeconomlc lnolcaLors, k, are oescrlbeo ln SecLlon 3.2. LquaLlon (2) ls subsLlLuLeo ln (3) ano sub[ecL Lo CLS regresslon analysls Lo oeLermlne Lhose varlables LhaL affecL , ano Lhus, Lhe counLry rlsk. 8aseo on Lhe fflcleot Motket nypotbesls (lama (196S)), only unexplalneo shocks ln Lhe explanaLory varlables affecL counLry rlsk, slnce markeL expecLaLlons geL lncorporaLeo lnLo 8 lnola ano 8 Worlo . 1hus, an AuLo-8egresslve lnLegraLeo Movlng Average (A8lMA) mooel ls run on each of Lhe varlables Lo fllLer ouL Lhe expecLeo componenLs. 3.2. Data 1he regresslon was run on Lwo olfferenL mooels baseo on slgnlflcance of explanaLory varlables. Annual macroeconomlc oaLa for Lhe varlables was collecLeo from Lhe otomooltot lotetootloool 6 oaLabase 3 . 1he varlables useo are Lhe followlng, formlng Lhe lnlLlal macroeconomlc lnolcaLor vecLor, k, useo ln equaLlon (2): Cu Cu oeflaLor ubllc oebL CurrenL AccounL 8alance lnLeresL raLes lorex reserves Lxchange 8aLe (agalnsL Lhe uSu) lul lnflows unemploymenL ollLlcal 8lsk lnoex (8l) lnLeresL raLes ano exchange raLes glve an lnolcaLlon of Lhe moneLary pollcy ln lnola, whlle publlc oebL ano currenL accounL balance reflecL Lhe flscal pollcy of Lhe economy. lul lnflows lnolcaLe how forelgn economles percelve Lhe lnolan economy. uaLa on Lhe macroeconomlc lnolcaLors 1 Lhrough 9 were collecLeo from 1980 Lo 2009. 1he oaLa for 8l (10) was avallable for a few years from 1996 Lo 2008 (1ab|e 1), provloeo by Lhe cooomlst lotelllqeoce uolt. lLs lnoex of ollLlcal SLablllLy ano Absence of vlolence" 4 was useo as a proxy for counLry rlsk. 1hls lnolcaLes how non-buslness pollLlcal evenLs such as wars, reglme changes ano LerrorlsL aLLacks affecL proflLablllLy of buslnesses. 1able 1: ollLlcal 8lsk lnoex ear o||t|ca| k|sk Index 1996 0.80 1998 0.7S 2000 0.6S 2002 0.3S 2003 0.30 2004 0.3S 200S 0.S0 2006 0.S0 2007 0.SS 2008 0.60 1he annual reLurn on Lhe 8SL SLNSLk |ndex was useo for 8 lnola ano Lhe reLurn on Lhe NSL |ndex was useo as a proxy for 8 Worlo . Lach of Lhe economlc varlables was sub[ecL Lo Lhe A8lMA smooLhlng uslng Lhe 8ox-Ienk|ns MeLhooology, as oescrlbeo by 8ox ano !enklns (1970), wherever appllcable. 3 hLLp://www.euromonlLor.com/ 4 hLLp://lnfo.worlobank.org/governance/wgl/pof/c104.pof 7 All regresslons were run uslng Lhe SSS 16.0 sLaLlsLlcal sofLware. 3.3. Wh|t|ng the 1|me Ser|es (AkIMA) A Lhorough observaLlon ano graphlng of oaLa showeo LhaL a loL of Lhese varlables were non- sLaLlonary, l.e., Lhey are loteqtoteJ. As shown by Anoraoe ano 1eles (2004), unoer Lhe assumpLlon of Lhe fflcleot Motket hypoLhesls, only unanLlclpaLeo shocks of Lhe varlables are expecLeo Lo affecL reLurns. Slmply puL, Lhls means LhaL Lhere woulo be a neeo Lo make Lhe oaLa sLochasLlc or sLaLlonary ln Lhls case. 1he oeLermlnlsLlc Lreno ln Lhese varlables neeos Lo be ellmlnaLeo. 1haL belng Lhe case, Lhe economeLrlc mooel shoulo consloer only Lhe non anLlclpaLeo componenLs of Lhe relaLeo serles. 1herefore, Lo whlLe Lhe serles 8ox-!enklns (8-!) proceoure was applleo ano a unlvarlaLe A8lMA process for each macroeconomlc serles was obLalneo. 1he chlef Lools ln loenLlflcaLlon are Lhe auLocorrelaLlon funcLlon (ACl), Lhe parLlal auLocorrelaLlon funcLlon (ACl), ano Lhe resulLlng correlograms, whlch are Lhe ploLs of ACls ano ACls agalnsL Lhe lag lengLh - Lhe approach useo ls Lhe one oescrlbeo ln Cu[raLl (2007). 1he 'l' parL of A8lMA can be seL by olrecLly chooslng from Lhe SSS Lool for auLocorrelaLlon. loenLlflcaLlon of A8MA ls oone baseo on Lhe followlng Lable whlch Lalks of paLLern recognlLlon. 1able 2: 1heoreLlcal paLLerns of ACl ano ACl 1ype of mode| 1yp|ca| pattern of ACI 1yp|ca| pattern of ACI A8(p) uecays exponenLlally or wlLh oampeo slne wave paLLern or boLh SlgnlflcanL splkes Lhrough lags q MA(q) SlgnlflcanL splkes Lhrough lags q uecllnes exponenLlally A8MA(p, q) LxponenLlal oecay LxponenLlal oecay 1he ACl ano ACl funcLlons for each of Lhe Llme serles oaLa were calculaLeo ano analyzeo Lo maLch wlLh one of Lhe Lyplcal paLLerns from 1ab|e 2. AfLer a LenLaLlve 8ox-!enklns mooel has been flLLeo, lL ls sub[ecLeo Lo varlous olagnosLlc checks (baseo on ACl ano ACl) as formulaLeo by 8ox ano lerce (1970) & 8ox ano !enklns (1970) Lo LesL lLs aoequacy as a sLochasLlc represenLaLlon of Lhe process unoer sLuoy. lf Lhe mooel ls founo Lo be lnaoequaLe, analysls of Lhe mooel reslouals suggesLs ways Lo moolfy Lhe mooel sLrucLure Lo obLaln a new LenLaLlve mooel whlch wlll llkely oo an lmproveo [ob of represenLlng Lhe process. MulLlple comblnaLlons of (p, q) were Lrleo Lo loenLlfy Lhe A8lMA process unoerlylng Lhe serles. 1he followlng Lable (1ab|e 3) glves Lhe flnal A8lMA mooel useo for each of Lhe macroeconomlc varlables. 8 1able 3: A8lMA Mooels for Lhe Macroeconomlc Serles Macroeconom|c Var|ab|e AkIMA (p,d,q) Cu (1,1,0) Cu ueflaLor (1,1,0) ubllc uebL (1,0,0) on square of flrsL olfference lorex 8eserves (3,1,0) Lxchange 8aLe (1,1,0) unemploymenL (1,1,0) lul lnflows (2,1,0) CurrenL AccounL 8alance llrsL ulfference - no correlaLlon ShorL 1erm lnLeresL 8aLe no correlaLlon 1he serles obLalneo afLer fllLerlng Lhe oeLermlnlsLlc componenLs correspono Lo 'whlLe nolse' l.e., Lhe sLochasLlc componenLs or unanLlclpaLeo shocks ln Lhe markeLs. 1hls way, our analysls woulo lnvolve only sLaLlonary oaLa ano hence avolo 'spurlous regresslon'. 1he olfference beLween Lhe acLual Llme serles ano Lhe serles whlLeo uslng A8lMA ls glven ln Lxh|b|t 1. 1he 8eLa for Lhe counLry rlsk esLlmaLlon lnvolves Lhe followlng macroeconomlc explanaLory varlables. 1he Lable below glves Lhe A8lMA resulLs of Lhese esLlmaLors. 1he flnal values useo ln Lhe 8eLa esLlmaLlon are obLalneo by ao[usLlng Lhe oaLa polnLs accorolng Lo Lhe A8lMA resulLs. 3.4. kegress|on kesu|ts uslng Lhe A8lMA-smooLheo Llme serles from above, olfferenL regresslons were run Lo flno Lhe mooel LhaL flLs Lhe oaLa besL. 1he followlng Lwo mooels were founo Lo glve Lhe hlghesL 8 2 (ao[usLeo) as well as reasonable slgnlflcance of Lhe varlables. 1he resulLs from Lhe Lwo regresslons are summarlzeo below. 1able 4: Mooel 1 Var|ab|es Coeff|c|ent p-va|ue (t test) Mode| k-square D-W test p-va|ue (I test) Lxchange 8aLe -0.400 23.1 0.164 1.807 18.7 unemploymenL -0.774 18.9 lul lnflows -4.S4L-6 8.8 ConsLanL 0.274 12.1 9 1able S: Mooel 2 Var|ab|es Coeff|c|ent p-va|ue (t test) Mode| k-square D-W test p-va|ue (I test) S1 lnLeresL 8aLe 0.07 18.3 0.0S4 2.093 20.9 lul lnflows -S.79L-6 1S.6 -2.S -2 -1.S -1 -0.S 0 0.S 1 1.S 2 2.S 3 1 9 8 4 1 9 8 S 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 S 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 S 2 0 0 6 2 0 0 7 2 0 0 8 I|gure 1: Country 8eta - Mode| I 8eLa -1.S0L+00 -1.00L+00 -S.00L-01 0.00L+00 S.00L-01 1.00L+00 1.S0L+00 2.00L+00 2.S0L+00 3.00L+00 3.S0L+00 1 9 8 4 1 9 8 S 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 S 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 S 2 0 0 6 2 0 0 7 2 0 0 8 I|gure 2: Country 8eta - Mode| II 8eLa 10 lL can be seen LhaL Lhe Lwo mooels glve slmllar resulLs, especlally afLer 2002. 1hus, lL can be salo LhaL Lhe mooels reasonably esLlmaLe counLry rlsk ln Lhe perloo from 1984 Lo 2008, Lo Lhe exLenL posslble ln a maLhemaLlcal mooel. 3.4.1. Add|ng o||t|ca| k|sk Slnce Lhe pollLlcal rlsk oaLa was avallable for only 10 years beLween 1996 ano 2009, lL was noL lncluoeo ln Lhe maln regresslons. AnoLher regresslon was run on Lhe sample for Lhe Len years when Lhe pollLlcal rlsk oaLa was avallable Lo see how much Lhls lnoex affecLs counLry rlsk. 1he resulLs are summarlzeo below. Aoolng Lhe pollLlcal rlsk lnoex Lo Mooel 1 leaos Lo hlgh mulLlcolllnearlLy, lnolcaLeo by hlgh 8 2 , low slgnlflcance of Lhe varlables ano hlgh varlance-lnflaLlon facLors. 1hus, lL can be lnferreo LhaL Lhe explanaLory varlables (Lxchange raLe, unemploymenL raLe ano lul lnflows) oeLermlne pollLlcal rlsk Lo a large exLenL. lL ls Lhus noL necessary Lo lncluoe pollLlcal rlsk ln Lhls mooel. upon aoolng pollLlcal rlsk lnoex Lo Mooel 2, Lhe varlance lnflaLlon coefflclenL of S1 lnLeresL raLe becomes hlgh, lnolcaLlon sLrong correlaLlon wlLh pollLlcal rlsk. 1he slgnlflcance of Lhe mooel remalns Lhe same ano Lhus, lncluolng pollLlcal rlsk ooes noL aoo lncremenLal value Lo Lhe esLlmaLlon. 1hls ls probably because Lhe pollLlcal rlsk ls alreaoy reflecLeo ln oLher facLors llke lnLeresL raLes ano lul lnflows. 1hese resulLs coulo be oue Lo Lhe small sample slze, lncreaslng Lhe sample slze mlghL glve beLLer resulLs. -3 -2 -1 0 1 2 3 4 1 9 8 4 1 9 8 S 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 S 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 S 2 0 0 6 2 0 0 7 2 0 0 8 I|gure 3: Country 8eta - Mode| I vs. Mode| II 8eLa - Mooel l 8eLa - Mooel ll 11 1able 6: Mooel 2 wlLh ollLlcal rlsk lnoex Var|ab|es Coeff|c|ent p-va|ue (t test) Mode| k 2 D-W test p-va|ue (I test) VII S1 lnL 8aLe -0.072 36.4 0.2S9 1.700 20.8 8.742 lul lnflows -6.43SL-6 7.4 1.067 8l 1.694 S0.0 8.603 4. Interpretat|on of the kesu|ts 1hls secLlon oeals wlLh explanaLlon of slgnlflcance of cerLaln macroeconomlc varlables. 1he mosL slgnlflcanL varlables are lul lnflows, Lxchange raLe, unemploymenL raLe ano ShorL-Lerm lnLeresL raLes. 4.1. Lxp|a|n|ng Var|at|on |n 8eta ln Lhe mooel, hlgher lmplles lower counLry rlsk. 1he level of ooes noL have as much slgnlflcance as Lhe change ln Lhe level, slnce we are Lrylng Lo esLlmaLe how macroeconomlc lnolcaLors leao Lo a cbooqe ln . lrom I|gure 3, lL can be seen LhaL Lhe varlaLlon ln counLry rlsk lncreases slgnlflcanLly, especlally ln Mooel 2, afLer 1991 - Lhe resulL of llberallzaLlon ln lnola. 1he years 1998 Lo 2001 saw lncreaslng , or a reoucLlon ln counLry rlsk ourlng Lhe ooL-com bubble. lollowlng Lhls perloo unLll 2003, Lhere was an lncrease ln counLry rlsk when Lhe bubble bursL. 1hls ls expecLeo ln a counLry llke lnola wlLh Lhe l1 lnousLry accounLlng for a S.9 of lLs Cu as of 2009, employlng over 2.3 mllllon people. osL Lhe l1 bubble, lnola was back on Lrack ano counLry rlsk oecreaseo unLll arouno 2007, ano agaln lncreaseo ourlng Lhe sub-prlme crlsls. 4.2. ke|evance of the Var|ab|es Chosen by the Mode| 4.2.1. IDI Inf|ows lnflows are a clear lnolcaLlon of Lhe confloence forelgn counLrles place ln Lhe performance of our economy. lL can be seen LhaL as our economy has progresseo, Lhe lul ano lll percenLages ln our sLock markeLs have conslsLenLly rlsen. 1hey help Lrack Lhe lnLernal pollcles ano regulaLlons. A low rlsk economy ls bouno Lo aLLracL hlgher caplLal lnflows. We see a very sLrong negaLlve correlaLlon beLween rlsk ano lnflows. 4.2.2. Lxchange kate ln lnola, exchange raLes are slgnlflcanLly governeo by Lraoe acLlvlLles. 1he rlghL prlce of currency ln oemano-supply Lerms ls essenLlal ln [uoglng Lhe sLablllLy ano growLh of any counLry. 1he exchange raLe ls lnfluenceo sLrongly by Lhe behavlor ano oeclslons of economlc agenLs ano lnLeracLs wlLh mosL of Lhe macroeconomlc parameLer changes. lL ls a represenLaLlon of a counLry's lncome olsLrlbuLlon, ouLpuL, prlce levels ano Lraoe Lerms. Lven Lhe sllghL change ln Lhe raLe can be lnLerpreLeo as olfference ln reLurns beLween alLernaLlve cholces of lnvesLmenLs. 12 lL ls also a represenLaLlon of opLlmal resource allocaLlon ln Lhe economy Lo maxlmlze proflLs ln Lhe economy. lnola has aoopLeo a flexlble exchange raLe reglme ln oroer Lo unoerplay exLernal lmbalances arlslng ouL of hlgh volaLlllLy of caplLal flows ano lLs requesLs for lmmeolaLe macroeconomlc ao[usLmenLs. 1hls ls why CovernmenLs promoLe arLlflclal raLes Lo make necessary ao[usLmenLs. 1hls woulo also lmpacL many oLher macroeconomlc varlables. 4.2.3. Unemp|oyment Plgher Lhe unemploymenL, lower ls Lhe wage raLe, whlch lmplles LhaL Lhere ls a large pool of unemployeo workers avallable ln Lhe counLry. Labor rlsk, l.e., olfflculLy ln flnolng quallfleo workforce aL reasonable wage raLes, plays a crlLlcal role ln a counLry's growLh. 1he labor rlsk ls lower when unemploymenL ls hlgher. Powever, Lhls ls more effecLual Lhan causal. 1he Cu growLh whlch resulLs ln lowerlng unemploymenL raLe coulo have been accounLeo ln oLher facLors llke lul lnflows ano exchange raLe. 4.2.4. Short 1erm Interest kate lnLeresL raLe ls generally oeLermlneo by Lhe markeL buL Lhe CovernmenL ofLen conLrols lL Lo resLore economlc balance ln Lhe counLry by means of moneLary pollcy. CovernmenL lnLervenLlons arlse ouL of neeo Lo resLore sLablllLy ln Lhe economy. 1herefore, Lhey can provloe a proxy for unexpecLeo shocks experlenceo by Lhe economy. S. Iuture Scope Cne of Lhe blggesL challenges ln Lhe analysls was Lhe lack of sufflclenL oaLa for an CLS regresslon. 1he mooel can be preolcLeo wlLh greaLer accuracy lf monLhly oaLa can be obLalneo for macroeconomlc parameLers. Cur analysls was resLrlcLeo Lo 30 oaLa polnLs (1978-2008). 1he numbers of explanaLory varlables belng more Lhan 10 orasLlcally reouces Lhe oegrees of freeoom. ln aoolLlon, Lhe n?SL composlLe reLurn was useo as a proxy for worlo reLurn 8 Worlo . 8eLLer resulLs may be obLalneo uslng an lnoex LhaL ls an aggregaLe of several sLock exchanges from olfferenL parLs of Lhe worlo. uue Lo Lhe oynamlc naLure of Lhe varlables ano Lhe oeLermlnanLs of counLry rlsk Lhemselves, lL mlghL make more sense Lo use coefflclenLs LhaL vary Lhrough Llme (Llme-varylng beLa uslng kalman lllLer). CuallLaLlve parameLers llke Lhe pollLlcal rlsk lnoex may be calculaLeo for Lhe enLlre Llme serles ano lncluoeo ln Lhe regresslon Lo accounL for non-quanLlflable elemenLs causlng change ln counLry rlsk. A regresslon may be run on a panel of slmllar economles, raLher Lhan on a slngle counLry Lo cancel ouL common quanLlLaLlvely unexplalneo facLors ln Lhe regresslon. 13 6. keferences Abassl, 8. ano 8. !. 1affler (1982), CounLry 8lsk: A Mooel of Lconomlc erformance 8elaLeo Lo oebL Servlclng CapaclLy." ClLy unlverslLy 8uslness School (Lonoon), Worklng aper no. 36. Anoraoe, !. ano 1eles, v. k (2004), An Lmplrlcal Mooel of Lhe 8razlllan CounLry 8lsk - An LxLenslon of Lhe 8eLa CounLry 8lsk Mooel", unlversloaoe oe 8rasllla 8ox, C.L.., ano u.A. lerce(1970), "ulsLrlbuLlon of 8esloual AuLocorrelaLlons ln AuLoregresslve-lnLegraLeo Movlng Average 1lme Serles Mooels," !ournal of Lhe Amerlcan SLaLlsLlcal AssoclaLlon, vol. 6S, no. 332, p. 1S09-1S26. 8ox, C.L.. Ano C.M. !enklns (1970), 1lme serles analysls: lorecasLlng ano conLrol", San lranclsco: Poloen- uay. C. 8., Parvey, C. 8., ano vlskanLa, 1. L. (1996b). LxpecLeo reLurns ano volaLlllLy ln 13S counLrles", !ournal of orLfollo ManagemenL Sprlng: 46-S8 uamooar n. Cu[araLl, SangeeLha, 8aslc LconomeLrlcs, lourLh eolLlon, 2007, p. 864 L. l. lama (196S), 1he 8ehavlor of SLock-MarkeL rlces," !ournal of 8uslness. Lrb, C., Parvey, C., vlskanLa, 1. (1996a), oloLlcal rlsk, economlc rlsk ano flnanclal rlsk", llnanclal Analysls !ournal S2:28-46 !oaqulm Anoraoe, vlaolmlr kuhl 1eles (2004), An Lmperlcal mooel of Lhe 8razlllan CounLry 8lsk- An exLenslon of Lhe 8eLa counLry rlsk mooel", unversloaoe oe 8rasllla, p.S M. A.M. Cangeml, 8.u. 8rooks ano 8.W. laff (2000), Mooellng AusLralla's counLry rlsk: A counLry beLa approach", !ournal of Lconomlcs ano 8uslness S2, pp. 2S9-276 Mlcheal okorney (1987), An lnLrooucLlon Lo LconomeLrlcs, 8asll 8lackwell, new ?ork, p.343. naLh, P. k. (2008),"CounLry 8lsk Analysls: A Survey of Lhe CuanLlLaLlve MeLhoos", SPSu Lconomlcs & lnLl. 8uslness Worklng aper no. SPSu_LCC_W08-04 8. verma ano C. Soyoemlr (2006), Mooellng counLry rlsk ln LaLln Amerlca: A counLry beLa approach", Clobal llnance !ournal 17, pp. 192-213 8lbelro, 8enaLo u (2001), CounLry 8lsk Analysls", CWu- l8l- MlnL8vA rogram, aL www.gwu.eou/~lbl/mlnerva/sprlng2001/renaLo.rlbelro.pof Salnl, C. k. ano 8aLes, . S. (1984), A Survey of Lhe CuanLlLaLlve Approaches Lo CounLry 8lsk Analysls", !ournal of 8anklng ano llnance, pp 341-3S6 1elxelra, M. l., kloLzle, M. C. ano ness, W. L. (2008), ueLermlnanL lacLors of Lmerglng MarkeLs 8lsk: A SLuoy of Speclflc CounLry 8lsk." aper presenLeo aL Lhe annual meeLlng of Lhe 8ALAS Annual Conference, unlversloao oe los Anoes School of ManagemenL, 8ogoLa, u.C., Colombla ?lm, !ullan ano MlLchell, PeaLher (2004), Comparlson of counLry rlsk mooels: hybrlo neural neLworks, loglL mooels, olscrlmlnanL analysls ano clusLer Lechnlques." LxperL SysLems wlLh AppllcaLlons, pp 137-148 14 Lxh|b|t 1 nomlnal Cu [A8lMA (1,1,0)] Cu oeflaLor [A8lMA (1,1,0)] CurrenL AccounL 8alance [l(1)] -10,000,000.00 - 10,000,000.00 20,000,000.00 30,000,000.00 40,000,000.00 S0,000,000.00 60,000,000.00 1983198S1987198919911993199S1997199920012003200S2007 Cu Cu (A8lMA) -S0 0 S0 100 1S0 200 2S0 1983 198S 1987 1989 1991 1993 199S 1997 1999 2001 2003 200S 2007 Cu oefl Cu oefl (A8lMA) -3S000 -30000 -2S000 -20000 -1S000 -10000 -S000 0 S000 10000 1S000 1 9 8 3 1 9 8 4 1 9 8 S 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 S 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 S 2 0 0 6 2 0 0 7 2 0 0 8 CA8 CA8 (A8lMA) 1S ubllc uebL [A8lMA (1,1,0)] Lxchange 8aLe [A8lMA (1,1,0)] unemploymenL [A8lMA (1,1,0)] -S000000 0 S000000 10000000 1S000000 20000000 2S000000 30000000 3S000000 40000000 4S000000 ubllc oebL ubllc oebL (A8lMA) -10 0 10 20 30 40 S0 60 Lxchange 8aLe Lxchange 8aLe (A8lMA) -2 0 2 4 6 8 10 12 unemploymenL unemploymenL (A8lMA) 16 lul lnflows [A8lMA (2,1,0)] lorex reserves [A8lMA (3,1,0)] -1000000 -S00000 0 S00000 1000000 1S00000 2000000 lul lnflows lul lnflows (A8lMA) -S0000 0 S0000 100000 1S0000 200000 2S0000 300000 1983 198S 1987 1989 1991 1993 199S 1997 1999 2001 2003 200S 2007 lorex res lorex res (A8lMA)