Professional Documents
Culture Documents
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
Page1
Economics
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
Page2
Economics
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
Page3
Economics
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
Page4
Economics
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
Page5
Economics
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
Page6
Economics
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
Page7
Economics
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
Page8
Economics
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
Page9
Economics
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
Page10
Economics
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
Page11
Economics
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Page12
Economics
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
Page13
Economics
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
Page14
Economics
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
Page15
Economics
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
Page16
Economics
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
Page17
Economics
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
Page18
Economics
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
Page19
Economics
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Page20
Economics
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
Page21
Economics
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Page22
Economics
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
Page23
Economics
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Page24
Economics
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
Page25
Economics
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
Page26
Economics
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
Page27
Economics
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
Page28
Economics
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
Page29
Economics
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
Page30
Economics
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
Page31
Economics
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Page32
Economics
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
Page33
Economics
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
Page34
Economics
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
Page35
Economics
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
Page36
Economics
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
Page37
Economics
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
Page38
Economics
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Page39
Economics
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingan
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
Page40
Economics
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Page41
Economics
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Page42
Economics
Itishopedthatthereaderfindsthematerialinthisbookhelpfultounderstand
why
suchnewmethodscanbeusefulforforecasting.
Althoughthisbookamountstoanintroductiontothefieldoftimeseriesanalysis
and
forecasting,itisnecessarythatthereaderhasknowledgeofintroductory
econometrics.
Specifically,regressionanalysis,matrixalgebraandvariousconceptsin
estimation
shouldbeincludedinthatknowledge.Thisbookshouldthenbeusefultoadvanced
undergraduatestudentsandgraduatestudentsinbusinessandeconomics,butalsoto
practitionersandappliedeconomistswhowishtoobtainafirst,butnottoo
technical,
impressionoftimeseriesforecasting.Infact,mostofthematerialhasalready
been
usedinTimeSeriesAnalysiscoursesforthirdyearundergraduatestudentsatthe
EconometricInstituteinRotterdameversince1996.
Thefirsteditionofthisbook(Franses(1998))containedmaterialthathasnow
beendeleted.Periodicmodelsforseasonaldataarenotincludedanymoreandalso
an
extensivediscussionofcommonfeatureshasbeendeleted.Ontheotherhand,more
detailsonARCHmodelsandonnonlinearmodelshavebeenincluded,wherewe
drawfromthematerialina2000CambridgeUniversityPresstextbookbythefirst
twoauthors.Moreimportantly,thisfullyrevisedsecondeditioncontains
exercises.
Answerstoselectedexerciseswillbemadeavailableonaspecialwebsite.These
exercisesmatchwiththosepresentedatpastexamstoourstudents.
ThisbookwaswrittenduringouraffiliationwiththeEconometricInstituteat
theErasmusUniversityRotterdam.ThisInstituteisaverystimulatingteachingand
researchenvironment.Wewishtoexpressourgratitudetoour(then)colleagues
Teun
Kloek,ChristiaanHeij,HermanvanDijk,DennisFok,RichardPaap,AndreLucas,
MariusOomsandanonymousreviewersfortheirkindwillingnesstocommenton
someorallchapters
Theeconometricanalysisofeconomicandbusinesstimeseriesisamajorfieldof
researchandapplication.Thelastfewdecadeshavewitnessedanincreasing
interest
inboththeoreticalandempiricaldevelopmentsinconstructingtimeseriesmodels
andintheirimportantapplicationinforecasting.Thisbookaimsatreviewing
several
importantdevelopmentswithinthecontextofforecastingbusinessandeconomictime
series.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
Page43
Economics
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Afullblowntextbookonallaspectsoftimeseriesanalysiswillcoverthousands
ofpages.Forexample,thefieldofunitrootanalysishasexpandedinthelast
three
decadeswithsuchapaceandvariationthatabookonlyonthistopicwouldtake
morepagesthanthecurrentbookdoes.Thisbookisthereforenotintendedtobea
surveyofallthatisavailableandthatcanbedoneintimeseriesanalysis.
Obviously,
suchaselectioncomeswithacost,thatis,thediscussionwillsometimesnotbe
as
theoreticallypreciseassomereaderswouldhaveliked.Merely,itisourpurpose
that
thereadersshouldbeabletogeneratetheirownforecastsfromtimeseriesmodels
that
Page44
Economics
adequatelydescribethekeyfeaturesofthedata,toevaluatetheseforecastsand
tocome
upwithsuggestionsforpossiblemodificationsifnecessary.Insomeinteresting
cases,
though,wealsorecommendfurtherreading.Toattainthis,wemakeaselection
between
allthepossibleroutestoconstructingandevaluatingtimeseriesmodels,between
all
thepossibleestimationmethods,andbetweenallthevariousteststhatcanbe
used.
Basically,ourchoiceisalsomotivatedbytheavailabilityofmethodsinsuch
statistical
packagesasEviews,whilesometimesalittlebitofGauss,RorMatlabprogramming
isneeded.Infact,allempiricalresultsinthisbookarethusobtained.An
additional
motivationforourchoiceisgivenbyourownpracticalexperienceinforecasting
businessandeconomictimeseries.Thisexperienceisalsobasedonsupervising
projectsofoureconometricsundergraduatestudentsduringtheirinternshipsat
banks,
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletoget
investmentcompanies,andconsultancyagencies.
Thesecondpurposewiththisbookisthatthereaderwillbeabletogetsome
understandingofnovelapproachesreportedinrecentandfutureissuesof,say,the
JournalofTimeSeriesAnalysis,JournalofEconometrics,JournalofBusinessand
EconomicStatistics,JournalofForecasting,InternationalJournalofForecasting,
Journal
ofAppliedEconometricsandtheJournaloftheAmericanStatisticalAssociation.
Page45