DETERMINANTSOFEXCHANGERATES:THECASEOFTHE

CHILEANPESO

SebastianZwanzger

AThesisSubmittedtothe
UniversityofNorthCarolinaWilmingtoninPartialFulfillment
oftheRequirementsfortheDegreeof
MastersofBusinessAdministration

CameronSchoolofBusiness
UniversityofNorthCarolinaWilmington
2008

Approvedby

AdvisoryCommittee

PeterSchuhmann

LutherLawson

.
.

CetCiner
Chair

Acceptedby

DN: cn=Robert D. Roer,

o=UNCW, ou=Dean of the
Graduate School & Research,
email=roer@uncw.edu, c=US
Date: 2009.05.19 15:47:36 -04'00'

Dean,GraduateSchool

TABLEOFCONTENTS
ABSTRACT ........................................................................................................................ iii
LISTOFTABLES ............................................................................................................... iv
LISTOFFIGURES.............................................................................................................. v
INTRODUCTION ................................................................................................................ 1
THEORIESBEHINDEXCHANGERATEDETERMINATION,BRIEFREVIEW............. 3
BACKGROUNDONCHILEANDTHEUNITEDSTATES ECONOMYAND
CURRENCIES...................................................................................................................... 4
TheU.S.Dollar ..................................................................................................................... 4
ExchangeRateMovementsandEconomicPerformance........................................................ 4
EconomicBackgroundandexchangerateregimeinChile.................................................... 5
COMMODITYCURRENCIES............................................................................................ 7
ChileandCopper .................................................................................................................. 9
DATASOURCESANDCOLLECTIONMETHODS......................................................... 10
DataDescription ................................................................................................................. 10
MoneyStock(M2) .............................................................................................................. 11
MonetaryPolicyInterestRate ............................................................................................. 12
DataSplit ............................................................................................................................ 12
EMPIRICALANALYSIS................................................................................................... 13
CorrelationAnalysis............................................................................................................ 14
CONCLUSION................................................................................................................... 22
LITERATURECITED ....................................................................................................... 24

ii

ABSTRACT
Inthisstudy,weanalyzedtherelationshipbetweentheexchangeratemovementsof
twocountries,ChileandtheUnitedStates,bystudyingtheunderlyingfundamentalsgivenby
themodernexchangeratetheory.Inthiscontext,weincludedinourregressionanalysisthree
maineconomicfactors.Monetarypolicyinterestrate,moneysupplyandinflationrateswere
consideredforbothcountriessinceJanuary1990.
WealsoincludedafourthvariableinourmodelCopperPrice.Theevolutionofthis
commodityspriceplayedanimportantroleinourstudyaswewilldiscoverthatasignificant
portioninexchangeratesvariationisexplainedbythisvariable.Copperwasconsidered
becauseoftheincreasingimportanceofthiscommodityintheChileaneconomy.
Theresultsshowthatthedeterminantsoftheexchangeratemayvaryovertime.The
independentvariablesthathaveaneffectontheexchangeratemaylosetheirexplanatory
powerwheneconomicconditionschangeoraswitchintheforeignexchangeratepolicy
dictatedbycentralbanksor,asweproved,whenvariationsoncertainmarketstakesplace.

iii

LISTOFTABLES

Table

Page

1.CorrelationAnalysis................................................................................................. 14
2. LaggedRegressionCoefficients............................................................................... 21

iv

LISTOFFIGURES
Figure

Page

1.EvolutionCHP/USDcomparedtotheChileanInflation ............................................. 6

2. EvolutionCHP/USDcomparedtotheEvolutionofCopperPrice ............................... 9
3.CHP/USDInflationUSandChileCopperPriceEvolution ................................. 12

INTRODUCTION
Sincetheearlyseventies,severaltheoriesconcerningexchangeratesdetermination
haveevolved.Atthistime,itwascommontoemphasizeinternationaltradeflowsasprimary
determinantsofexchangerates.Thiswasdue,inpart,tothefactthatgovernments
maintainedtightrestrictionsoninternationalflowsoffinancialcapital.Theroleofexchange
ratechangesineliminatinginternationaltradeimbalancesuggestthatwewouldexpect
countrieswithcurrenttradesurplusestohaveanappreciatingcurrency,whereascountries
withtradedeficitsshouldhavedepreciatingcurrencies.
Thesedifferentmodelsdevelopedmainlyinthe70sledtotestablepropositionsin
whichthechangesintheexchangeratearelinearlyrelatedtothenewsinthefundamental
variables,suchasmoneystocks,prices,output,currentaccounts,etc.Amongthesemodels,
wefindthemonetarymodel,Dornbushmodel,andtheportfoliobalancemodel.
Wecanconsiderthisasthefirststageinexchangeratedeterminationtheory.
Unfortunately,thesefirstgenerationmodelswererejectedbythedata,atleastforcountries
wherethelevelsofinflationwhererelativelylow.Threeanomaliesweredetected.
First,therandomwalkforecasttypicallyoutperformsaforecastbasedonthefirst
generationmodels.ThestudyconductedbyMeeseandRegoff(1983 1988)
demonstratedthatthetheorybehindthesemodelscantbeconfirmedbytheresultsof
thestudy.Randomwalktheorystatesthatcurrenciestakearandomandunpredictable
path.
Second,sincethestartofthefloatingexchangerateregime,variabilityofexchange
ratesincreaseddramatically.Atthesametime,thereisnoevidencetobefoundthat
thevariabilityofthefundamentalsidentifiedbythetheoreticalmodelshasincreased
comparedtothefixexchangeratesperiod.Firstgenerationmodelsimpliedthatthe
variabilityofexchangeratescanonlyincreasewhenthevariabilityoftheunderlying
fundamentalsincreases
Third,thefirststagemodelsassumethatexchangeratescanonlychangeatacertain
pointintimeasaresultof newsinthefundamentals.AstudydirectedbyDeBoeck
(2000)andAltavilla(2000)findthatunanticipatedshocksinthefundamental
variablesexplainonlyasmallfractionoftheunanticipatedchangesinexchange rates.

Typicallytheyfindthatnewson inflation,outputandinterestratesexplainabout5%
inthevariationofexchangerates.
Atthispoint,wheretheoryhasbeenprovenwrong,researchersaredriventoalternatives
tothisfirststagemodels.Somestudiesintroducenonlinearitiesintothemodel(seeDe
GrauweandDewatcher,1993FrankelandFrot,1990KilianandTaylor,2003Kurzand
Motolese,1999).Thesemodelsarecharacterizedbytheexistenceofseveralagentsusing
differentinformationsetsand/orbytheexistenceoftransactioncosts.Theinsightprovided
bythesemodelsisthattheypredictfrequentstructuralbreaksinlinearexchangerate
equations,andthattheyallowforchangesintheexchangeratesthatareunrelatedtonews
abouttheunderlyingfundamentals.
ItisalsoworthmentioningandconsideringthecontributionofNelsonC.Mark(1995)
whereheinvestigatestheextenttowhichdeviationsoftheexchangeratefromafundamental
valuesuggestedbyeconomictheoryareusefulinpredictingexchangeratesoverlong
horizons.Thestudypresentsevidencethatthereisaneconomicallysignificantcomponentin
longhorizonchangesinlognominalexchangerates.Thiswasmadebytestingregressionsof
multipleperiodchangesinthelogexchangerateonthedeviationofthelogexchangerate
fromitsfundamentalvalue.Theresultsofthestudyshowthatwhileshorthorizonschanges
tendtobedominatedbynoise,thisnoisetendtobeaveragedoutovertime.Thesocalled
Noisereferstoexchangeratesfluctuationsinthemarketthatcanconfuseones
interpretationofmarketdirection.Thestudyrevealssystematicexchangeratemovements
thataredeterminedbyeconomicfundamentals.Thiswasoneofthefirststudiesthatproved
withempiricalresultsthatexchangeratechangescanbeforecasted.
Theobjectiveofthisprojectis,consideringpreviousstudiesonthefield,toanalyzethe
natureoftherelationshipbetweenexchangeratechangesandthevariationsintheunderlying
fundamentals,lookingatthehistoricdevelopmentoftwocurrencies,theUSdollarand
ChileanPeso.
BesidestestingtheChileancurrencyagainstthefundamentalvariables,anewtestwillbe
performedtakingintoconsiderationtheimpactthatcommoditypricesmayhaveinforeign
exchangedetermination,inviewofthehighimpactofcopperintheChileaneconomy.

THEORIESBEHINDEXCHANGERATEDETERMINATION,BRIEFREVIEW
1. PurchasingPowerParity(PPP): Basedonnoarbitrageargumentorlawofoneprice,
PPPisaflowmodeloftheinflationtheoryofexchangeratesvisavisthebalanceof
trade.OnlyrelativePPPseemstoholdinthelongrun.Shiftsintechnology,tastes,
commercialpoliciesorlaborforcegrowthwillstructurallychangenationalproductivity
andhencewillpermanentlychangetherealexchangerate.
2. MonetaryApproach:ThesestockmodelsarebasedonIS/LM/PhillipCurveparadigm.
Basicallythetheoriesarebasedonfindingtheexchangeratewhichtheavailableamount
ofcurrencysupplyisequaltothedemandtoholdthecurrency.
2.1 MundellFleming Model:Thetheoryconsidersthreemarkets: money,assetandgoods
marketsunderperfect priceflexibilityinlongrun.Oneimplicationisdevaluationmay
leadtofurtherdevaluationiffiscaldiscipline,inflationandbalanceofpaymentsarenot
wellmanaged.Anotheristhatthehigherthedegreeofreexportprocessingindustrythe
countryhas,thelowertheimpactofdevaluationforcurrentaccountimprovement.
2.2 MonetaristModel: Thisconceptimpliesthattheexchangeratelevelisperfectly
correlatedwiththeleveloftherelativemoneysupplyinlongrun.Ina stationary
economy,therelativemoneygrowthratewouldbezeroandtheexchangerate
expectationwouldplayatrivialrole.Postulatedinaninflationaryand/orhighgrowth
economy,thismodelexplainswhyaforeignexchangeratemarketmaybe
characterizedbya selffulfillingprophecy.Whenthemoneysupplybecomesstochastic,
rationalexpectationandaccuracyofmarketinformation playanimportantrolein inter
temporal analysis.
2.3 StickyPriceModel :Whenacurrencyisdevaluedandthepriceofgoodsremains
fixedintheshortrun,butnotinthelongrun,thecurrencyvaluemayovershoot.A
balanceofpaymentcrisis,extendedfromthemodel,istheequilibriumoutcomeof
maximizingbehaviorbyrationalagentsfacedwithafundamental inconsistency
betweenmonetaryandexchangeratepolicies.Oneimplicationcanbeinagame
theoretic perspectiveinpolicyimplementation.Inanoncooperativeanduncoordinated
gamewithmanyplayers,eachindividualplayerhaslittleinformationonother players
payoffs.Thereisnoreasontobelieveeveryonewouldacttogethertoreachthedesired
outcomeinasinglestep.Goodgovernmentcoordinationasasignal tothefinancial

marketbecomesessentialtoachieveeffectivemonetarypolicyforcurrency
stabilization.
2.4. PortfolioBalanceApproach:Thistheorydeterminestheexchangerateastherelative
priceofmoniesinshortrun.Theassetsubstitutioneffectsandthenatureof
expectationsformationplacesmoreemphasisonshortruncapitalflowsratherthanthe
tradebalance.Ingeneral,theshortrunimpactofpoliciescanbequitedifferentfromthe
longrunimpact,dependingonthenatureoftheexpectations.Exchangeratesshould
implicitlybehavelikethebehaviorofassetpricesinspeculativemarkets.

BACKGROUNDONCHILEANDTHEUNITEDSTATES ECONOMYAND
CURRENCIES
TheU.S. Dollar
TheUSdollaraccountsforabout70%ofallforeignexchangereservesandoverone
halfofglobalprivatefinancialwealth.AbouttwothirdsofworldtradeisinvoicedinUS
dollarsandthreequartersofinternationalbanklendingisdenominatedindollars.1 The
dollarsimportancehas,however,decreasedsincethemid1970s,whenitwasresponsiblefor
80%ofofficialreservesinforeigncountries.Investorswouldstartdiversifyingtheir
portfoliosastheworldmovestowardsamultiplecurrencysystem.Alegitimaterivalofthe
USDistheEuro,especiallyafteritsintroductionin20002004,andinrecentyearsmany
countrieshaveannouncedthattheyareincreasingtheireuroreservesasithasanincreasing
roleintheworldseconomy.
ExchangeRateMovements andEconomicPerformance
The1990swasaperiodofunprecedentedeconomicgrowthintheUS,andin
February2000theeconomysurpassedthepreviousrecordforthelongestsustained
expansionof107months.Themainfeatureoftheexpansionwasasteadyimprovementin
productivity,owingpartlytowideruseofcomputersandotheradvancedtechnology.Real
GDPgrewbyalmost40%between1991and2001,equivalenttoanannualaverageof3.4%.
ThiseconomicprospectallowedtheUSdollartoappreciatethroughouttheninetiesand
beginningofthenewcenturyagainstmosttradingpartners.TheFederalReservesindexof

www.eiu.com,countryprofileUSA

theUSdollarseffectiveexchangerateagainstothermajorcurrenciesrosefromanaverage
of94.06in1998(March1973=100)to107.66in2001.Thelongappreciationofthedollar
reflectedthestrongunderlyingperformanceoftheUSeconomyandtheprospectofhigh
investmentreturns.
Thefirstindicationthatthelongexpansionwascomingtoanendcamewiththe
collapseinthevalueofhightechstocksthatbeganinMarch2000.Investmentsslowed down
dramatically,whileslowgrowthintheUS.smajortradingpartnerscontributedtoa
deterioratingtradebalance.Growthslowedfurtherduring2001,especiallyintheimmediate
aftermathoftheSeptember11thterroristattacks.Thiswasalsoabadtimeforfinancial
markets,whichwererockedbyanumberofmajorscandals.Mostofthesecanbetraced back
totheexcessesofthestockmarketboomofthe1990s,whensomehighprofilecompanies
resortedtodubiousaccountancypracticestoboosttheirprofitsandshareprices.
EvenastheUSeconomyweakenedandUSinterestratesfellduring2001,theUS
dollarhelditsvalueagainstmostcurrencies.Poorinvestmentsreturnselsewhereintheworld
contributedtotheUSdollartomaintainitsvalueforthattime.
However,theUSdollardroppedinvaluesignificantlyagainstmostmajorcurrencies
during2002,withthepacequickeningin200304,primarilyagainsttheeuroandtheChinese
yen. ThiswasalsothecasefortheChileanpeso.
Thecentralbanksexchangerateindexagainstmajorcurrenciesaveraged93.o4in
2003and85.42in2004,ornearly21%belowits2001average.2 Thedepreciationreflected
concernaboutthewideningcurrentaccountdeficitandtherapiddeteriorationinthefederal
governmentsfiscalposition.In2005theUSdollarregainedsomestrength,propelledbya
combinationofrisingshortterminterestratesintheUSandtheweaknessofother
economies,notably thosein theeurozone.
TheUSdollarreachedaoneyearlowofUS$1.27againsttheeuroinMay2006,a
28yearlowofUS$1.12againsttheCanadiandollar,andalsoretreatedsignificantlyagainst
theyen.G7financeministersmadeitclearatameetinginApril2006thattheywould
welcomeanappreciationofcurrenciesofemergingeconomieswithbigtradesurpluseswith
theUS,suchasChinaandSouthKorea.
Economic Background andexchangerateregime inChile

www.eiu.com,countryprofileUSA

TheChileanpesotrendhadaremarkableinflexionpointinthebeginningofyear
2003,whereitreacheditsmaximumvalueof745.21pesosperUSdollar.Theappreciationof
theCLPisevidentafterthatdate.
Startingin1990,theCentralChileanBank(CCB)adoptedaninflationtargetregime.
Aninflationobjectivewasseteveryyearandgraduallyadjustedsoastoallowforanongoing
reductionofinflation.Infact,in1990,thetargetforthefollowingyearwassetat27%,
whereasin2001thetargetwas3%.
Figure 1. EvolutionCHP/USDcomparedtotheChileanInflation

FromthemideightiestoSeptember7, 1999,theCCBadoptedacrawlingexchange
rateband,wheretherateisallowedtofluctuateinabandaroundacentralvalue,whichis
adjustedperiodically.Themainobjectivesofthebandweretomaintaininternational
competitivenessandreduceexcessiveexchangeratevolatility.
However,sincethestartoftheband,manyofitsfeatures,includingitswidth,rateof
crawl,referencecurrencybasket,degreeofsymmetryandcentralparityweremodified.Since
September2,1999,theCCBhasembracedafullyflexibleexchangerateregime,withthe
6

possibilityoftheCCBinterveninginthemarketonly inthecasethattheexchangeratedoes
notreflecttherealvalueoftheforeigncurrency.
However,theeliminationofthebandinSeptember1999didnotimplytheabsenceof
foreignexchangeinterventions.Infact,inresponsetolargeexchangeratedepreciationand
volatility,theCCBannouncedandcarriedoutatemporarypolicyofsterilizedinterventions
betweenJuly2001andJanuary2002.Thestatedobjectivesoftheinterventionswereto
reduceexcessiveexchangeratevolatilityandprovideahedgeagainstfuturedevaluations,
withoutaffectingexchangeratetrends.

COMMODITYCURRENCIES
Commoditycurrenciesarethoseof countrieswhichdependheavilyontheexportof
certainrawmaterialsorcommoditiesforincome.
Differentstudieshavebeenperformedinthisareatodetermineifsuchcommodities
exportingcountriesseetheircurrenciesaffectedbyvariationsoninternationallytraded
commodityprices.ThestudybyYuChinYenandRogoff(2002)analyzedthe relationship
betweentheexchangeratemovementsandcommoditypricesforCanada,NewZealandand
Australia.Thesecountrieswerechosenbecausetheyconstituteopeneconomiesandare
highlyintegratedtoglobalcapitalmarketsandaredynamicparticipantsininternationaltrade.
Allofthemhavebeenoperatingunderafreelyfloatingexchangerateregime.Commodity
exportsaccountfor60%ofAustraliastotalexportsand50%forNewZealand.Canadahasa
strongerindustrialbase,butstillcontinuestorelymorethanaquarterofitsexportson
commodities.Theauthorsfindthatworldpricesofcommodityexports,expressedin
American dollars,haveafirmandstableinfluenceontheexchangeratesforAustraliaand
NewZealand.ForCanadatheinfluenceisnotasstrong,butfairlysignificant.Eventhough
thesecountrieshaveopencapitalmarketsandfreefloatingexchangerateregimes,areal
explanatoryvariablewasfoundfordeterminingexchangeratemovements.
Indeed,thestudyperformedbyCashin,Cspedes,andSahay(2002)basedonthe
assumptionthatfluctuationsinrealcommoditypriceshavethepotentialtoexplainchangesin
realexchangerates,giventhatmany countriesarehighlydependentoncommoditiesfortheir
exportsrevenues.Theyalsostudiedtherelationshipbetweencommoditypricesandcurrency
determination,butexpandedtheirstudytoallthecommoditiesexportingcountries.58
7

commodityexportingcountrieswereincludedinthesample.53ofthosearedeveloping
countries3.Commodityexportstypicallyexceeded50percentofthetotalexportsofseveral
subSaharanAfricancountries,especiallyBurundi(97percent),Madagascar(90percent),
andZambia(88percent).Theshareofprimarycommodityexportsintotalexportswasquite
highevenfortheindustrialcountries(Australia,54percentIceland,56percent).4
Theresultsobtainedbytheirstudyshowedthat22outofthese58countriespresenta
surprisinglystrongrelationshipthatremainsovertimebetweencommoditypriceshiftsand
thecurrencydetermination.Halfofthesecommoditycurrencycountriesbelongtothesub
SaharanAfricancountries,duetotheirstrongdependencetocommodities.In manycases
theyrelyontheexportsofonlyonecommodity.Forthesecountriestheircommodities
exportsusuallyaccountformorethan90%oftheirtotalexports.Thatisthecasefor
Dominica(bananas),Ethiopia(coffee),Mauritius(sugar),Niger(uranium),andZambia
(copper).
Over80%ofthevariationintherealexchangeratecan,onaverage,beexplainedby
movementsinrealcommoditypricesalone.Measuringtheimpactofpricevariations,they
foundthattheelasticitytypicallyrangedbetween0.2and0.4,withamedianof0.38.Thus,a
10percentdropintherealpriceofthecommodityexportsofcountrieswithcommodity
currenciesistypicallyassociatedwitha3.8percentdepreciationoftheirrealexchangerate.
Furtheranalysisalsoindicatedthat,whendeviationsfromtherelationshipbetweenexchange
ratesandcommoditypricesoccurredincountrieswithcommoditycurrencies,theywere
causedprimarilybychangesinrealcommodityprices.Followingamovementin commodity
prices,itistypically therealexchangeratethatthenadjuststorestoreitslongrun
relationshipwithrealcommodityprices.5

(classifiedbytheIMF'sWorldEconomicOutlookasexportersofnonfuelprimaryproductsandthosewith
diversifiedexportearnings)
4

PaulCashin,LuisCspedes,andRatnaSahay(2002)

PaulCashin,LuisCspedes,andRatnaSahay(2002)

ChileandCopper
Chileisthelargestproducerandexporterofcopperintheworld,followedbyPeru,
USAandIndonesia.Utilizedprimarilyin theconstructionindustry,itisalsowidelyusedasa
conductorforelectronicdevises,householdproducts,coinage,biomedicalapplications,and
chemicalapplicationsamongothers.Copperstandsfor45%ofChileanexportsandaccounts
for13.9%ofChilesNominalGDP.Chileproduces35.6%oftheworldscopperproduction
andrepresents30%oftheworldsreserveofcopper.6
ThestudybyCashin,Cspedes,andSahay(2002)doesnotconsiderChiletobea
commoditycurrencycountry,butthismaybeexplainedbecausehistoricpricesforcopper
remainedstablefortheperiodstudied.After2003,copperpricesstartedincreasingatavery
quickpaceduetogrowing worlddemandandalargedropininternationalinventorieslevels.
Ourlatteranalysisisfocusedtodetermineiftherelationshipbetweencopperpriceand
currencydeterminationisstableandsignificant.
ThefollowinggraphshowstheevolutionofCLPagainsttheincreaseinpriceofapoundof
copper.
Figure 2. EvolutionCHP/USDcomparedtotheEvolutionofCopperPrice
800

450

700

400

600

350
300

500

250

400

CHP/USD

200

300

150

200

100

100

50
ene04

ene03

ene02

ene01

ene00

ene99

ene98

ene97

ene96

ene95

ene94

ene93

ene92

ene91

ene90

ene89

ene88

ene87

0
dic85

COPPER
(US$/lb.)

Threecommodityexchangesprovidethefacilitiestotradecopper:TheLondonMetal
Exchange(LME),theCommodityExchangeDivisionoftheNewYorkMercantileExchange
(COMEX/NYMEX)andtheShanghaiMetalExchange(SHME).Intheseexchanges,prices

COCHILCO(ChileanCopperCommission)

aresettledbybidandoffer,reflectingthemarketsperceptionofsupplyanddemandofa
commodityonaparticularday.OntheLME,copperistradedin25 tonslotsandquotedin
USdollarspertononCOMEX,copperistradedinlotsof25,000poundsandquotedinUS
centsperpoundandontheSHME,copperistradedinlotsof5tonsandquotedinRenminbi
perton.Morerecently,minicontractsofsmallerlotssizeshavebeenintroducedatthe
exchanges.Exchangesalsoprovideforthetradingoffuturesandoptionscontracts.These
allowproducersandconsumerstofixapriceinthefuture,thusprovidingahedgeagainst
pricevariations.Inthisprocesstheparticipationofspeculators,whoarereadytobuytherisk
ofpricevariationinexchangeformonetaryreward,givesliquiditytothemarket.Afuturesor
optionscontractdefinesthequalityoftheproduct,thesizeofthelot,deliverydates,delivery
warehousesandotheraspectsrelatedtothetradingprocess.Contractsareuniqueforeach
exchange.Theexistenceoffuturescontractsalsoallowsproducersandtheirclientstoagree
ondifferentpricesettlingschemestoaccommodatedifferentinterests.Exchangesalso
provideforwarehousingfacilitiesthatenablemarketparticipantstomakeortakephysical
deliveryofcopperinaccordancewitheachexchangescriteria.7

DATASOURCESANDCOLLECTIONMETHODS
DataDescription
DataonmonthlyanddailyexchangerateswasgatheredfromtheCentralBankof
Chilesdatabase.Thedailydollarpricedenominatedin Chileanpesosiscalculatedbasedon
transactionsoffinancialinstitutionsthattookplacethepreviousworkingday.Weekendsand
holidaysareexcluded.
Thesampleconsideredinthisstudycontemplatesinformationonexchangeratesstartingin
January1990untilinMarch2008.
Theexplanatory variablesincludedinthemodelare:

http://www.icsg.org/

10

MoneyStock(M2)
Consistsofthetotalamountofmoneyheldbythenonbankpublicatapointintimein
aneconomy.
M0=Cashorassetsthatcouldquicklybeconvertedintocurrency.
M1=M0+Demanddeposits,whicharecheckingaccounts.
M2=M1+smalltimedeposits(lessthan$100,000),savingsdeposits,andnoninstitutional
moneymarketfunds.
Money supplymonthlyobservationsforChilewerealsotakenfromthecentralbank
ofChilesdatabase,whereastheUSmoneysupplyfiguresweretakenfromtheFederal
Reservesdatabase.
Inflation
MonthlychangesintheConsumerPriceIndexrepresenttherateofinflation.CPIisa
measureoftheaveragepricelevelofafixedbasketofgoodsandservicespurchasedby
consumersasdeterminedbytheBureauofLaborStatisticsfortheUSandtheNational
InstituteofStatistics(INE)forChile.
TheConsumerPriceIndex(CPIU)iscompiledbytheBureauofLaborStatisticsand
isbasedupona 1982Baseof100.8 ForChile,CPIvariationsaregatheredbytheNational
InstituteofStatistics(INE).
Thebasketofgoods,andtheweightofeveryiteminit,maydifferfromcountryto
country.Forexample,indevelopingcountries,theweightthatisgiventofoodproductstends
tobehigherthanforthedevelopedcountries,wheretechnologicalgoodshaveahigher
portionofthebasket.Thisisoneofthedistortionsofcomparinginflationbetweentwo
countries,butforouranalysiswewillnotaddressthisproblemsincethosedistortionsarenot
extraordinarilyconsiderable.
MonthlyobservationsontheChileanCPIweretakenfromtheChileannational
instituteofstatisticsdatabase.FortheUSconsumerpriceindex,thisinformationwas
gatheredfromtheFederalReservesdatabase.Aftercompilingthehistoricalconsumers

www.inflationdata.com

11

priceindexforbothcountries,Inflationwascalculatedbyestablishingthedifferential
betweenoneperiodandthepreviousone.
MonetaryPolicy InterestRate
AlsocalledFederalFundsRate,correspondstotheinterestrateatwhichadepository
institutionlendsimmediatelyavailablefunds(balancesattheFederalReserve)toanother
depositoryinstitutionovernight.
Dataonthefederalfundsratewascollectedfromthefederalfundsdatabase.The
ChileanmonetarypolicyinterestratewastakenfromtheCCBdatabase.Observationsare
againonamonthlybasis.
DataSplit
Figure 3. CHP/USDInflationUsandChileCopperPriceEvolution

Thedatawerefirstanalyzedasonecompletesample,andlaterseparatedintotwo
differentgroupstobetterexplaintheexchangeratesdeterminants.Lookingatthegraphs
shownbelow,wecanseethattherearetwopossiblebreakingpoints.Thefirstone occurred
inSeptember1999whentheCentralBankofChilechangedtheforeignexchangeregimetoa
12

completefloatingpolicy.ThesecondbreaknoticedinthedatawaswhentheChileanpeso
reacheditslowerlevelinMarch2003.Thiseventbrokealongtrendofadepreciating
currency.Thecompletesetofregressionswereperformedtobothsetsofsamples,meaning
thatwefirstseparatedthedataintheobservationnumber117(Sept1999)andthenwe
separatedthedataintheobservationnumber158(March2003).Theresultsobtainedby
splittingthedatawhentheexchangerateregimewasmodified(Sept1999) gaveusbetter
resultsinoursearchforthedeterminantsoftheexchangeratethereforewefocusedour
analysisin thatevent.Summarystatisticsfortheregressionsperformedsplittingthedatain
year2003areprovidedintables10,11,12and13.
9

EMPIRICALANALYSIS
VariablesandModels
Thevariablesconsidered are:

MoneyStockdifferential(M2).

(MSUS$ - MSCLP)
Inflation differential

(I US - ICL)
FederalFundsInterestRatedifferential.

(r US -rCL)
Thesearethemainfundamentalvariablesusedinmanyregressionmodelsandweretestedin
aconventionalmultipleregressionmodelshownbelow.

Det = a + b 1(MSUS$ - MSCLP )+ b 2(IUS - ICL )+ b 3(rUS - rCL )+ e

WhereDet =

DCLP
,MSrepresentMoneyStocksdifferential,theerrortermisnotedas e .
DUSD

Theperpendicularlinetotheleft(orange)denotesaSplitinthedatawhereachangeintheChileanforeign
currencypolicytookplace.Theperpendicularlineontheright(green)denotesabreakinthedepreciating
tendencyoftheCLP.

13

Also,thismodelwasusedinthestudyconductedbyPauldeGrawkeandIsabel
Vansteenkiste(2007)wheretheauthorsdevelopedanonlinearmodelbasedonthe existence
oftransactioncosts.Theyperformedtheiranalysisusingsamplesofbothhighandlow
inflationcountries.Theempiricalanalysisshowsthatforthehighinflationcountriesthe
relationshipbetweennewsinthefundamentalsandtheexchangeratechangesisstableand
significant.Thisisnotthecase,howeverforthelowinflationcountries,wherefrequent
regimeswitchesoccur.
CorrelationAnalysis
Apreliminarytestwasmadetoestablishthecorrelationbetweenthevariables.Asin
allthefollowingtests,theanalysiscontemplatesatestusingthefullsample,andtwo
consecutivesonesutilizingthefirstandsecondperiodsseparately.Aswementionedearlier,
thesplitpointwasestablishedwhenthecentralbankofChiledecidedtomodifytheexchange
ratepolicyinSeptember1999.Theresultsofthecorrelationtestareshowninthefollowing
table.
Table 1. CorrelationAnalysis
FullSample
et
et
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%ChangeCOPPER

(IUSICh)

1
0.21720268
1
0.006527779 0.000337447
0.012682794 0.085201824
0.26398336 0.189467543

(rUSrCh)

%Change
COPPER

1
0.0017249
1
0.03996015 0.107563922

(MSUSMSCh)

FirstPeriod
et
et
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%ChangeCOPPER

(IUSICh)

1
0.22101799
1
0.089704954 0.022637297
0.05990003 0.088942881
0.0984901 0.096864671

(rUSrCh)

%Change
COPPER

1
0.03288607
1
0.0046928 0.175329971

(MSUSMSCh)

SecondPeriod
et
et
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%ChangeCOPPER

(IUSICh)

1
0.1890756
1
0.12711475 0.04958071
0.02133903 0.239124111
0.34262285 0.299720935

Whereet= Det

14

(rUSrCh)

%Change
COPPER

1
0.034159938
1
0.155370445 0.091477354

(MSUSMSCh)

Aswecanseeinthetable,forthefullsample,thestrongestcorrelationbetweenour
dependantandindependentvariablescanbefoundintheinflationdifferentialandthe copper
pricechange.Theotherindependentvariablesdonotshowastrongcorrelation.Letsnotice
thatforbothvariables,(IUSICH)and%ChangeCopper,thecorrelationshowsanegative
sign,meaningthatifanyofthesevariablesincreases,theChileanpesowillappreciate.
Forthefirstperiod,inflationshowedthehighestcorrelation,0.22,andthecopper
pricechangewasweakerthaninthefullsampleanalysis.Thisvariableshowedamuch
strongercorrelationwhenanalyzing thesecondperiod.Copperpriceinfluenceinexchange
ratedeterminationseemstobemuchhigherafter1999,reaching 0.34correlationcoefficient.
Next,wedescribethetestsperformedutilizingsimplemultipleregressionanalysis.
Thefollowingregressionswereperformed:
I.

The first sets of regressions were performed utilizing the model described earlier that
includes the fundamental variables. The first test considers the complete sample
consistingof220monthlyobservations.Thesecondregressionwasperformedutilizing
the first period, from January 1990 until September 1999, where the exchange rate
regime changed.The third regression consists of the secondperiod, from Sept1999to
March2008.
Theresultsforthefirsttestareshownin Table1intheappendix.
Inthiscase,forboth changeinmonetarypolicyinterestratesandchangeinmoneystock

coefficients,thenullhypothesishastobeaccepted.Thereisnostatisticalevidencethatthe
slopeofthesecoefficientsissignificantlydifferentfromzero.However,forthedifferentialin
Inflation,thePvalueis0.001129498ata5%confidencelevel.
Thecoefficientsare:

Det = 0.0002521170.604220554(IUS$ - ICLP )

IftheAmericaninflationdifferentialishigherthantheChileaninflationdifferential,the
Chileancurrencywillappreciate.TheadjustedRsquareforthistestis3.4981737%
Forthesecondtestperformedweused115observations,wherewedonothavea
changeinthemonetarypolicy.LetsrememberthatforthisperiodtheregimeoftheCCB
wasinflationtargeting.
15

Thesummarystatisticsforthistestcanbefoundin Table2 intheappendix.

Onceagain,forboth changeinmonetarypolicyinterestratesandmoneystock
change,thereisnostatisticalevidencethattheslopeofthesecoefficientsissignificantly
differentfromzerothereforethenullhypothesishastobeaccepted.
ThePvaluefor (I

US$

-ICLP) is0.007445252,meaningthatitsstatisticallysignificant

ata5%confidencelevel.
Inthiscase,thecoefficientsare:

Det = 0.0019948570.442222329(IUS$ - ICLP )

TheadjustedRsquareforthisregressionis4.9446156%anditisthehighestinthis
firstsetoftestsperformed.
Thethirdtestperformedconsidersthesecondperiod,afterSeptember99andconsists
of105observations.Thesummarystatisticsforthistestarein Table3.
Westilldonotfindevidencethateitherthechangeininterestratesormoneystocks
aresignificantdeterminantsoftheexchangerate,butMShasimprovedasanexplanatory
variable.
Thecoefficientsare:

De t = 0.0005393751.289662897(IUS$ - ICLP )
Evenifthecoefficientfor (I

US$

-ICLP) isthehighest,thepvalueistheless

significant(0.051925028,significantata6%confidencelevel)andtheadjustedRsquareis
alsothelowest(2.4737736%).
Theresultsofthesetestsshowsusthatforthefirsthalfoftheperiodstudied,where
theobjectiveofthecentralbankofChilewastotargetandlowerinflation,theinflation
differentialhadthegreatestimpactintheexchangeratedetermination,outofallvariables
considered.Inanyoftheseregressionstheothervariablesshowedstatisticalsignificancein
theanalysis.

16

AfinalregressionwasperformedutilizingaDummyvariableforrepresentingtheregime
changeoccurredin1999,butitdoesnotaddanystatisticalsignificancetothemodel.The
resultsforthisregressioncanbefoundin Table7intheappendix.
II.

For the second set of regressions, testing the second period, the change in copper
prices was added to the model todeterminethe influencethat the commodity price
may have in foreign exchange determination. Copper is traded in the London
Mercantile Exchange. The data also consists in monthly observations and will be
separated just as in the tests performed previously. As mentioned earlier, copper
prices increased dramatically due to international inventory drops and increasing
international demand.
Themodelusedthistimeisthefollowing.

Det = a + b 1(MSUS$ - MSCLP )+ b 2(IUS$ - ICLP )+ b 3(rUS$ - rCLP )+ b 4(D %CopperPrice)

Thesummarystatisticsforthefirstsetoftestscanbefoundin Table 4in the

appendix.
Inthiscase,aswefoundinthefirstsetofregressions,changesinthemonetarypolicy
interestrateandchangesinthemoneystockarenotsignificantata5%confidencelevel.
Thereisnostatisticalevidencethattheslopeofthesecoefficientsissignificantlydifferent
fromzero,thereforethenullhypothesishastobeaccepted.
Thecoefficientsareasfollows,

Det = 0.0012631610.48602805(IUS$ - ICLP)0.08042013(D %CopperPrice)

ThePvaluefor (I

US$

-ICLP) is0.00791446ata5%confidencelevel,whilefor

(D%Copper Price)thePvalueis0.000426143,makingitstatisticallysignificantata5%level.
Anincreasingcopperpricewouldresultinanappreciatingcurrency,whichisconsistentwith
ourexpectations.
Lookingattheinflationeffect,theresultsaresimilartotheonesobtainedinthe
previouslytestedregressions.IfthechangeintheAmericaninflationishigherthanthe
changeintheChileaninflation,theChileancurrencywillappreciate.
17

Thesecondregressiontestperformedusingthismodelconsidersthefirstperiodofthe
sample,consistingof115monthlyobservations.Thesummarystatisticsforthisregression
canbefoundin Table5intheappendix.
Forthisperiodoftime,copperpriceremainedfairlystableandwithoutgreat
volatility.ThePValueforthisvariableis0.330662929thereforeitisnotstatistically
significantata5%confidencelevel.Sincethisvariabledoesnotaddevidenceofexchange
ratesdetermination,theresultsareverysimilartotheonesobtainedinthesecondregression
performedusingtheoriginalmodel.
Forthethirdregression,weutilizedtheobservationsforthesecondperiodoftime.At
thispoint,copperpricessufferedarapidincreaseduetogrowinginternationaldemandanda

Det startedappreciating

dropininventorylevelsworldwide.Alsoourdependantvariable
afteraprolongeddepreciatingtrend.

Thesummarystatisticsforthisregressioncanbefoundin Table6intheappendix.
Thecoefficientsare

Det = 0.0012518870.11440637(D%CopperPr ice)

Inthiscase,andforthefirsttimeinourstudy,thedifferentialoninflationlevelsisnot
significantata5%confidencelevel,withaPValueof0.258064303.Ontheotherhand,the
PValueforthechangeincopperpriceis0.003354111makingithighlysignificant.Once
again,resultsshowusthatanincreasingcopperpricewouldresultinanappreciating
currency.TheadjustedRsquarecoefficientis10.2890058%,thehighestobtainedsofar,
meaningthat10.29%ofexchangeratesmovementsareexplainedbyvariationsincopper
price.
Anewregressionwasperformedusingonlythechangeincopperpricevariableand
excludingallotherindependentvariablesfromthemodel.Summarystatisticscanbefoundin
table 8forthefullsampleandtable9 forthesecondperiod.
Herewecanseethattheinfluenceofcopperforthesecondperiodstudiedis
considerablyhigherthanforthefullsample.
Thecoefficientsfortheregressionusingonlythesecondperiodareshownbelow.

Det = 0.0014920640.132153835 (D%CopperPr ice)

18

TheadjustedRsquarevalue fortheanalysisusingthefullsampleis6.54%,and
increasedto11.08%whenanalyzingonlythesecondperiod.Theincreasedexplanatory
powerofthisvariableisclear.
Summarizingtheresultsobtainedsofar,wecanconcludethataftersplittingthedata
intwoperiods,wefinddifferentexchangeratedeterminantsineachoneofthem.Thebreak
inthedatawasmadeinSeptember1999,wherethecentralbankofChiledecidedtomodify
itsforeignexchangeratepolicy.Beforethisperiod,thegoaloftheCCBwastotargetand
lowerinflation,andpreciselythisisthevariablethatshowedusthegreatestroleinthe
exchangeratedetermination.Theresultsshowsusthatall otherindependentvariable
consideredwerenotstatisticallysignificantinanyoftheregressionsperformed.
IfthedifferentialbetweenthechangeintheUSinflationandthechangeinthe
Chileaninflation(I

US$

-ICLP) isanegativevalue,thentheCLPwilldepreciate.Inother

words,iftheChileaninflationishigherthantheUSinflationinacertainperiodoftime,we
canexpecttheChileancurrencytodepreciate.Thatwasthecaseforalmostthecomplete
periodfrom 1990to1999.
Afteraddingafourthvariabletothemodel,theresultsshowedusthattheroleof
copperintheexchangeratedeterminationincreaseddramaticallyafteryear2003.Beforethat
year,itsaveragepricewas94.62US$/Lb.Inthesecondhalfofthesample,itsaverageprice
was218US$/Lb,reachingitstoppriceof393US$/Lb.
Forthisperiod(99 08),regressionresultsshowedthatallothervariableswerenot
statisticallysignificant.
Apositivevariationincopperpricewillresultinan appreciatingcurrency.
AlsotheadjustedRsquarecoefficientforthisregressionis10.29%,beingthehighest
obtainedsofar.
Theanalysisshowedusthatmoneysupplywasneversignificantinanyofthe
regressionsperformed.Thisresultwassomehowexpectedbecausethismodelwasalsoused
inthestudybyGrauweandVansteenkiste(2007)andtheycouldnotfindsignificanceforthis
coefficienteither.
Itisinterestingtonoticethatinoneregimethevariablesinquestionmayhavea
significanteffectontheexchangerate,whileinotherregimetheireffectisnotsignificantly
different from zero.Inthiscase,forthefirstperiod studiedthesignificanceofinflationis
19

evident,whileforthesecondperiodthisexplanatorypowerisdisplacedbytheinfluenceof
copper prices.
TheresultsaresimilartotheonesobtainedbytheearliermentionedstudybyGrauwe
andVansteenkiste(2007),wheretheyperformedsimilarteststohighandlowinflation
countries.Letsrecallthatintheirstudytheycouldnotfindarelationbetweenthe
fundamentalsandthecurrenciesoflowinflationnations,whereasthislinkwasstableand
significantforhighinflationcountries.
Thisphenomenonisexplainedasfollows.Transactioncostsarepresentintraded
goodsanddefineabandwherearbitrageopportunitiesdonothold.Thisisthesameforboth,
highandlowinflationcountries.Ifexogenousshocksareintroducedintheunderlying
fundamentalsforthelowinflationcountries,theseshockswillusuallytendtobesmall
relativetothetransactioncostsband(differentialinflationshocksaretypically1%or2%per
year),hence,arbitragewillnotbeprofitableandwillnottakeplace.Theconnectionbetween
thenewsinthefundamentalsandthecurrenciesforthesecountrieswillnotbestrong.
Moreover,ifashockislargercomparedtothetransactioncostband,arbitragewilltakeplace,
butasaconsequencetherelationshipbetweenthefundamentalsandtheexchangeratewillbe
unstable.
Ontheotherhand,forhighinflationcountries,shocksinthefundamentalvaluestendto
belargecomparedtothetransactioncostsband,allowingforstrongarbitragerelations.This
impliesastrongerandmorestablerelationshipbetweenthefundamentalvaluesandthe
foreignexchangerate.
III.

The following analysis will consider daily exchange rates information and daily
copperpricesstartingJan1,1990toJul31,2008.Theobjectiveforthisnewanalysis
istodeterminethetimethatittakestothemarkettoreacttonewcopperpricesandto
reflectthesechangesinthecurrencydetermination.Themodelusedinthisanalysisis
derivedfromtheGrangerCasualtytestandisbasicallyalaggedlinearregression.The
Granger causality is a statistical concept of causality that is based on prediction.
According to Granger causality, if a signal X Grangercauses (or Gcauses) a
signalY,thenpastvaluesofXshouldcontaininformationthathelpspredictYabove
andbeyondtheinformationcontainedinpastvaluesofYalone.Inourstudy,weare
trying to find evidence in past values of copper prices and how long it takes to be
reflectedinthecurrencies.
20

Themodelcanberepresentedasfollows

D e t = a + b t (D %Copper Price )+ b t-1(D %Copper Price )+ b t- 2 (D %Copper Price )+ b t- 3 (D %Copper Price )

WhereDet =

DCLP
bt representsthechangeincopperpricesonthesameday, bt -1
DUSD,

representschangesincopperpricesthepreviousday and bt-2 & bt-3 representtwoandthree

dayslagrespectively.
CompleteSummarystatisticsforthistestcanbefoundontable16.
Thecoefficientsareshownbelow.
Table 2. LaggedRegressionCoefficients
Coefficients
Intercept

Standard
Error

tStat

Pvalue

6.3873E05 0.000119192 0.535882148

0.592097026

DeltaCPPR
DeltaCPPR 1

0.008757582 0.007557047 1.158862886

0.078383687 0.007570042 10.35445843

0.246644469
1.54794E24

DeltaCPPR 2

0.031023761 0.007570987 4.097717127

4.33174E05

DeltaCPPR 3

0.013199982 0.007558277 1.746427518

0.080883586

DeltaCPPR(changeincopperprice)

Theseresultsshowusthatourvariable bt isnotsignificantata5%confidencelevel,
while bt -1 and bt -2 aresignificant.Thenullhypothesishastobacceptedforbt .

bt-3 issignificantata10%confidencelevel.
Thesignsof bt -1 and bt -2 areexpected,consideringtheresultsobtainedinprevious
setsofregressionsusingmonthlydata.Anincreaseincopperpricewillleadtoan
appreciatingcurrency.Thesignof bt-3 maycorrespondtomarketadjustmentsafter
variationsincopperprice.
TheadjustedRsquareforthismodelis5.6%.
Themarketwouldrespondonaveragebetweenoneandtwodaysafterthevariationin
copperpricetoreflectthenewexchangerate.

21

CONCLUSION
Ourresearchobjectivewastoanalyzetherelationshipbetweenexchangeratechanges
andthevariationsintheunderlyingfundamentals.WeconsideredtheChileanpeso,and
comparedittotheUSdollar.Wetestedthe performanceofthiscurrencyforthelasteighteen
years.
Theresultsshowthatthedeterminantsoftheexchangeratemayvaryovertime.The
independentvariablesthathaveaneffectontheexchangeratemaylosetheirexplanatory
powerwheneconomicconditionschangeoraswitchintheforeign exchangeratepolicy
dictatedbycentralbanksor,asweproved,whenvariationsoncertainmarketstakesplace.
Themodelthatweusedincludesthreedependablevariables.Theorysupportsthese
variablesandincorporatesthemintheirexplanationonhowexchangeratesmoveovertime.
Thesevariablesarethemoneysupply(M2),changesin domesticpricelevelsandmonetary
policyinterestrate.
Thedatacorrespondedtomonthlyobservationsfortheindependentvariablesforboth
countries,TheUnitedStates ofAmericaandChile.Thisdatasetwasanalyzedasawholeina
firststageandseparatedintotwogroupsforamoredetailedstudy.Bytakingapreliminary
lookatthedata,twonaturalbreakpointsappear.Thefirstonetakesplacewhenthecentral
bank ofChiledecidedtochangetheforeign monetarypolicyregime.BeforeSeptember1999
theobjectiveoftheCCBwastotargetinflation.Afterthat,theyadoptedafreelyfloating
exchangerateregime.ThesecondbreakingpointtakesplaceinMarch2003,wherethe
Chileanpesoreachedhishistoricallowestlevelof745.21pesosperUSDollar.Afterrunning
simplelinearregressionanalysisandcomparingtheresultsonbothsamplegroups,wefind
thatthefirstdatadivisionexplainedinabetterwaythedeterminantsoftheChileanpeso
movements.ThebreakinthelongdepreciatingtrendoftheChileanpesocomparedtotheUS
Dollarhasdifferentexplanationsthatarenotcoveredinthisstudy,forexample,theeuro
communitystartedutilizing astheircurrency.Thatperiod(2003)wastheinitiation ofa
processwheretheUSDollarlostvalueagainstmanycurrencies,principallythe, and
Chinese.
Oncewedefinedourdatasplit,weperformedsimplelinearregressionsutilizingthe
modelderivedfromtheMarkovswitchingautoregressive(MSAR).
Wefindthatforthefirstperiod(1990 1999)changesinpricelevelsplayeda
significantroleintheexchangeratedetermination.Bothmoneysupplyandmonetarypolicy
22

interestratewerefoundtobenotsignificantintheanalysis.Forthesecondperiod(afterthe
switchintheexchangeratepolicy)wefoundthatInflationdoesnothavethesame
explanatorypowerasinthefirstperiod.
Ouranalysiscontinuedincludingafourth variableintoourmodel.Copperplaysa
greatroleintheChileanEconomy,asthiscountryisthebiggestproducerandexporterof
copperworldwide.Exchangeratemovementsseemtomoveinconcordancewithvariations
ininternationalcopperprices.Theresultsofourregressionanalysisshowedusthatforthe
firstperiod,inflationcontinuestobethestrongestvariableinthedeterminationofthe
exchangerate.However,thestrongaugmentinpricethatcopperexperiencedstartinginyear
2000showsusanincreasingexplanatorypowerofthisvariable.Ascopperpricesboosted,
theinfluenceofcopperbecomesmoreandmoreimportantindeterminingthevalueofthe
Chileanpeso.Ourstudyshowedusthatforthesecondperiod,inflationisnotsignificant
anymoreandvariationsininternationalcopperpricesbecomeourmaindependablevariable.
Ourconcludinganalysisconsistedinmeasuringthetimethatthemarketreactsto
variationsincopperprices.Forthis,weperformedasimplelaggedregressionderivedfrom
theGrangercausesconcept.Onaverage,themarketreflectsthecurrencyadjustmentsone
andtwodaysafterthepricevariations.

23

LITERATURECITED

EconometricApproachestoempiricalmodelsofexchangeratedetermination,Blackwell
Synergy
ForeignExchangeMarket,theoryandeconometricevidence,byRichardT.Baillie
FxTradingandexchangeratesdynamics,JSTOR
ExchangeRatesandfundamentals:ANonlinearRelationship?byPauldeGrawkeand
IsabelVansteenkiste
EconometricmodelsintheUK,P.G.Fisher,S.K.Tanna,D.S.Turner,K.F.Wallis,J.
D.Whitley,JSTOR
ExchangeRateTheory:AReviewByDr.PongsakHoontrakul[Dec99]
Recent Evolution on Exchange Rates, (LA EVOLUCIN RECIENTE DEL TIPO DE
CAMBIO)ByJosdeGregorio,presidentoftheCentralBankofChile
Exchange Rates and Fundamentals: Evidence on LongHorizon Predictability by
NelsonC.Mark,march1995.
TheEconomistIntelligenceUnitLimited2007

APPENDIX

FundamentalModel ChangeinMoneyStock,FedInterestRateandInflation.Complete
Sample
RegressionStatistics
MultipleR
0.219547637
RSquare
0.048201165
AdjustedRSquare
0.034981737
StandardError
0.019878344
Observations
220
Coefficients
Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

0.000252117
0.604220554
0.001837492
0.006746988

Standard
Error

tStat

0.001487929 0.169441595
0.183076367 3.300374396
0.018325092 0.100271904
0.01430251
0.47173456

Pvalue

Lower95%

Upper95%

0.865607888 0.002680602 0.003184836

0.001129498 0.965065433 0.24337568
0.920221469
0.0342814 0.037956384
0.637592431 0.021443365 0.034937342

FundamentalModel ChangeinMoneyStock,FedInterestRateandInflation FirstPeriod

1990 1999
RegressionStatistics
MultipleR
0.272874936
RSquare
0.074460731
AdjustedRSquare
0.049446156
StandardError
0.015059647
Observations
115
Coefficients
Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

0.001994857
0.442222329
0.01716942
0.01860858

Standard
Error

tStat

0.001731549
1.15206536
0.162206046 2.726299918
0.015319625 1.120746763
0.02079971 0.894655699

Pvalue
0.251769465
0.007445252
0.264814603
0.372907026

Lower95%

Upper95%

0.001436322
0.763644428
0.013187439
0.022607434

0.005426037
0.12080023
0.047526279
0.059824593

FundamentalModel ChangeinMoneyStock,FedInterestRateandInflation Second

Period1999 2008
RegressionStatistics
MultipleR
0.229935429
RSquare
0.052870302
AdjustedRSquare
0.024737736
StandardError
0.023888785
Observations
105

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

Coefficients
0.000539375
1.289662897
0.074231854
0.007306387

Standard
Error
tStat
0.00238137 0.226497956
0.655645047 1.967013863
0.052527454 1.413201048
0.020752335 0.352075398

Pvalue
Lower95%
Upper95%
0.821271917 0.005263372 0.004184622
0.051925028
2.59028627 0.010960475
0.160671117 0.178432191 0.029968483
0.725515264 0.033860663 0.048473436

FundamentalModel+CopperPriceVariationsFullSample
RegressionStatistics
MultipleR
0.318937299
RSquare
0.101721001
AdjustedRSquare
0.085008833
StandardError
0.019356235
Observations
220

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%Change
COPPER

Coefficients
0.001263161
0.486028054
0.004456595
0.011432884

Standard
Error
tStat
0.00147613 0.855724552
0.181300713 2.680784013
0.017858778 0.249546453
0.013988257 0.817320122

Pvalue
Lower95%
Upper95%
0.393102801 0.001646379
0.0041727
0.00791446
0.84338247 0.12867364
0.803176367 0.030744112 0.039657302
0.41465029 0.016138797 0.039004564

0.080420134

0.02246952 3.579076586

0.000426143 0.124708885 0.03613138

FundamentalModel+CopperPriceVariationsFirstPeriod1990 1999
RegressionStatistics
MultipleR
0.279334794
RSquare
0.078027927
AdjustedRSquare
0.044803708
StandardError
0.015114873
Observations
116

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%Change
COPPER

Coefficients
0.002331346
0.409838736
0.017789744
0.023025809

Standard
Error
tStat
0.001725694 1.350961786
0.162746808 2.518259755
0.015367875 1.157592953
0.021163861 1.087977733

Pvalue
Lower95%
Upper95%
0.179455078 0.001088231 0.005750923
0.01321942
0.73233239 0.08734508
0.249515036 0.012662725 0.048242213
0.278961637 0.018911792 0.06496341

0.025548833

0.026148621 0.977062324

0.330662929 0.077364069 0.026266404

26

FundamentalModel+CopperPriceVariationsSecondPeriod1999 2008
RegressionStatistics
MultipleR
0.371118956
RSquare
0.137729279
AdjustedRSquare
0.102890058
StandardError
0.02292137
Observations
104

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%Change
COPPER

Coefficients
0.001251887
0.752811767
0.051734457
0.007737537

Standard
Error
tStat
0.0023965 0.522381164
0.661797739 1.137525444
0.051432206
1.00587669
0.019915125
0.38852564

Pvalue
Lower95%
Upper95%
0.602571956
0.00350329 0.006007063
0.258064303 2.065962025 0.560338491
0.316927118 0.153787109 0.050318195
0.698461373 0.031778391 0.047253464

0.114406366

0.038058948 3.006030668

0.003354111 0.189923574 0.03888916

FundamentalModel+Copper+DummyVariableforregimechangeFullSample
RegressionStatistics
MultipleR
0.32004048
RSquare
0.102425909
AdjustedRSquare
0.081454551
StandardError
0.019393794
Observations
220

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%Change
COPPER
RegDum

Coefficients
0.001919104
0.455930382
0.004188149
0.010779153

Standard
Error
tStat
0.002178882 0.880774833
0.195927656 2.327034329
0.017905409 0.233904104
0.014105824 0.764163276

Pvalue
0.3794278
0.020898258
0.815283038
0.445611688

0.079409514
0.001177357

0.022647686 3.506297012
0.002871907 0.409956654

0.000553507 0.124050622 0.03476841

0.682247839 0.006838206 0.004483491

Lower95%
Upper95%
0.002375714 0.006213923
0.842125581 0.06973518
0.031105403
0.0394817
0.017024995
0.0385833

CopperModelFullSample
RegressionStatistics
MultipleR
0.263983363
RSquare
0.069687216
AdjustedRSquare
0.065419726
StandardError
0.019562338
Observations
220

Intercept
%Change
COPPER

Coefficients
0.002968622

Standard
Error
0.001329311

0.08963467

0.022181234 4.041013575

tStat
2.233203532

27

Pvalue
0.026551637

Lower95%
Upper95%
0.000348676 0.005588567

7.37842E05 0.133351788 0.04591755

CopperModelSecondPeriod1999 2008
RegressionStatistics
MultipleR
0.345601829
RSquare
0.119440625
AdjustedRSquare
0.110807689
StandardError
0.022819997
Observations
104

Intercept
%Change
COPPER

Coefficients
0.001492064

Standard
Error
0.002318382

tStat
0.643579668

Pvalue
Lower95%
Upper95%
0.52129243 0.003106436 0.006090563

0.132153835

0.035528999 3.719604769

0.000326584 0.202625435 0.06168224

28

TheFollowingfourtablesshowtheresultsobtainedwhen splittingthedataintheobservation
number158(February2003)intheinflexionpointofthecurrencies.

FundamentalModel ChangeinMoneyStock,FedInterestRateandInflation FirstPeriod

1990 2003
RegressionStatistics
MultipleR
0.188739591
RSquare
0.035622633
AdjustedRSquare
0.016713273
StandardError
0.017752725
Observations
157

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

Coefficients
0.004165401
0.40019354
0.009127144
0.010593976

Standard
Error
0.001648171
0.177860292
0.017618362
0.014103314

tStat
2.527287265
2.25004431
0.518047267
0.751169214

Pvalue
Lower95%
Upper95%
0.012510028 0.000909291 0.007421511
0.025872265 0.7515726 0.04881447
0.605173597 0.02567952 0.043933809
0.453704644 0.0172684 0.038456346

FundamentalModel ChangeinMoneyStock,FedInterestRateandInflation Second

Period2003 2008
RegressionStatistics
MultipleR
0.15245748
RSquare
0.023243283
AdjustedRSquare
0.02642231
StandardError
0.023013459
Observations
63

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

Coefficients
0.00690401
0.63486799
0.04117006
0.01232468

Standard
Error
tStat
Pvalue
Lower95%
Upper95%
0.002963606 2.32959753 0.023269662 0.01283417 0.00097385
0.733919522 0.86503761 0.390521842 2.10343755 0.833701571
0.057515706 0.71580547 0.476934981 0.15625872 0.073918604
0.039370278
0.3130452 0.755350259 0.09110442 0.066455067

29

FundamentalModel+CopperPriceVariationsFirstPeriod1990 2003
RegressionStatistics
MultipleR
0.22732449
RSquare
0.051676424
AdjustedRSquare
0.02672054
StandardError
0.017662156
Observations
157

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)
%ChangeCOPPER

Coefficients

Standard
Error

tStat

Pvalue

Lower95%

Upper95%

0.004263683
0.37283555
0.009800577
0.013333015
0.04557351

0.001640906
0.177772897
0.017533505
0.014134879
0.028410583

2.598370651
2.09725753
0.558962805
0.943270577
1.60410324

0.010288394
0.037626114
0.577009695
0.347039391
0.110767126

0.001021754
0.72406037
0.02484026
0.01459318
0.10170412

0.007505612
0.02161072
0.044441416
0.041259208
0.010557107

FundamentalModel+CopperPriceVariationsSecondPeriod2003 2008
RegressionStatistics
MultipleR
0.326640577
RSquare
0.106694066
AdjustedRSquare
0.045086761
StandardError
0.022197334
Observations

63
Standard
Error
tStat
Pvalue
Lower95%
Upper95%
0.003086521 1.35877872 0.179477674 0.01037224 0.001984444
0.746879139 0.10780597 0.914521681 1.57555905 1.414522993
0.055896106 0.45170656 0.65316506 0.13713685 0.086639575
0.037986658
0.26458 0.792270997 0.08608907 0.065988047

Intercept
(IUSICh)
(MSUSMSCh)
(rUSrCh)

Coefficients
0.0041939
0.08051803
0.02524864
0.01005051

%ChangeCOPPER

0.09385988 0.040322822

2.3277112 0.023437755 0.17457478 0.01314499

LaggedRegressionderivedfromtheGrangerCasualtytestDailydataJan2000 Jul2008
RegressionStatistics
MultipleR
0.240508217
RSquare
0.057844202
AdjustedRSquare
0.056041033
StandardError
0.005422683
Observations

Intercept
DeltaCPPR
DeltaCPPR1
DeltaCPPR2
DeltaCPPR3

2095
Coefficients
6.3873E05
0.008757582
0.07838369
0.03102376
0.013199982

Standard
Error
0.000119192
0.007557047
0.007570042
0.007570987
0.007558277

tStat
0.535882148
1.158862886
10.3544584
4.09771713
1.746427518

30

Pvalue
0.592097026
0.246644469
1.54794E24
4.33174E05
0.080883586

Lower95%
0.00016987
0.00606254
0.09322929
0.04587122
0.00162255

Upper95%
0.000297621
0.023577705
0.06353808
0.0161763
0.028022516

1990
1990
1991
1991
1992
1993
1993
1994
1994
1995
1995
1996
1997
1997
1998
1998
1999
2000
2000
2001
2001
2002
2002
2003
2004
2004
2005
2005
2006
2007
2007
2008

1990
1990
1991
1991
1992
1993
1993
1994
1994
1995
1995
1996
1997
1997
1998
1998
1999
2000
2000
2001
2001
2002
2002
2003
2004
2004
2005
2005
2006
2007
2007
2008
1990
1990
1991
1991
1992
1993
1993
1994
1994
1995
1995
1996
1997
1997
1998
1998
1999
2000
2000
2001
2001
2002
2002
2003
2004
2004
2005
2005
2006
2007
2007
2008

ExchangeRateVolatility
ExchangeRateVolatility

0.08
0.06
0.04
0.02
0
0.02
0.04
0.06
0.08

CopperPriceVolatility
CopperPriceVolatility

0.3
0.2
0.1
0
0.1
0.2
0.3

InflationVolatility

InflationVolatility

0.04
0.03
0.02
0.01
0
0.01
0.02
0.03
0.04
0.05

31