Full Te ms & Condi ions o access and use can be ound a
h ps://www. and online.com/ac ion/jou nalIn o ma ion?jou nalCode=ces 20
Economic Sys ems Resea ch
ISSN: (P in ) (Online) Jou nal homepage: h ps://www. and online.com/loi/ces 20
Linking mul isec o al economic models and
consump ion su eys o he Eu opean Union
Ignacio Cazca o , An onio F. Amo es , Inaki A o & Ku K a ena
To ci e his a icle: Ignacio Cazca o , An onio F. Amo es , Inaki A o & Ku K a ena (2020):
Linking mul isec o al economic models and consump ion su eys o he Eu opean Union,
Economic Sys ems Resea ch, DOI: 10.1080/09535314.2020.1856044
To link o his a icle: h ps://doi.o g/10.1080/09535314.2020.1856044
© 2021 The Au ho (s). Published by In o ma
UK Limi ed, ading as Taylo & F ancis
G oup
View supplemen a y ma e ial
Published online: 24 Dec 2020.
Submi you a icle o his jou nal
A icle iews: 2163
View ela ed a icles
View C ossma k da a
ECONOMIC SYSTEMS RESEARCH
h ps://doi.o g/10.1080/09535314.2020.1856044
Linking mul isec o al economic models and consump ion
su eys o he Eu opean Union
Ignacio Cazca o a,b, An onio F. Amo es c,d, Inaki A o band Ku K a ena e
aAgencia A agonesa pa a la In es igacion y Desa ollo (ARAID), Ag i ood Ins i u e o A agon (IA2). Dep .
Economic Analysis, Za agoza, Spain; bBasque Cen e o Clima e Change, Bilbao, Spain; cEu opean
Commission Join Resea ch Cen e, Ci cula Economy and Indus ial Leade ship, Se ille, Spain; dDep .
Economics, Quan i a i e Me hods and Economics His o y, Pablo de Ola ide Uni e si y, Se illa, Spain; eWIFO,
Vienna, Aus ia
ABSTRACT
Mul isec o al models usually ha e a single ep esen a i e household.
Howe e , mo e di e si y o household ypes is needed o analyse
he e ec s o mul iple phenomena (i.e. ageing, gende inequali y,
dis ibu ional income impac , e c.). Household consump ion su eys’
mic oda a is a ich da a sou ce o hese ypes o analysis. Howe e ,
eeding mul isec o al models wi h his ype o in o ma ion is no sim-
ple and ecen s udies show how e en sligh ly inaccu a e p ocedu es
migh esul in signi ican ly biased esul s. This pape p esen s he
ull p ocedu e o eeding household consump ion mic oda a in o
mac oeconomic models and o he i s ime p o ides in a sys ema ic
way an es ima ion o he b idge ma ices needed o link Eu opean
Union Household Budge Su eys’ mic oda a wi h he mos popula
mul i- egional inpu –ou pu amewo ks (e.g. Eu os a , WIOD, EORA,
OECD).
ARTICLE HISTORY
Recei ed 14 Janua y 2019
KEYWORDS
Household budge su eys;
CPA-COICOP; con ingency
b idge ma ices; alua ion
ma ices; EU28
1. In oduc ion
Mos economic models, such as Inpu –Ou pu (IO) models o Compu able Gene al Equi-
lib ium (CGE) models, ep esen p i a e consump ion using a single ep esen a i e agen .
Howe e , his assump ion o he ep esen a i e agen has been widely c i icised (B ock
& Du lau , 2001;Hoo e ,2008;Ki man,1992; Leh inen & Kuo ikoski, 2007;Sa a d,
2004;S igli z,2018) and he inclusion o he e ogeneous household p o iles in economic
modelling is inc easingly encou aged.
Fo example, he ecommenda ions o S igli z e al. (2009) included ‘2. Emphasis-
ing he household pe spec i e’ and ‘4. Gi e mo e p ominence o he dis ibu ion o
income, consump ion and weal h’. The G-20 ‘Da a Gaps Ini ia i e’ also ecommended
CONTACT An onio F. Amo es an onio [email p o ec ed]opa.eu Eu opean Commission Join Resea ch Cen e,
Ci cula Economy and Indus ial Leade ship, c/ Inca Ga cilaso, Ed. EXPO, Se ille, 41092 Spain; Economics, Quan i a i e
Me hods and Economics His o y, Pablo de Ola ide Uni e si y, C a. U e a Km.1, Se illa, 41013 Spain
Supplemen al da a o his a icle can be accessed he e. h ps://doi.o g/10.1080/09535314.2020.1856044
© 2021 The Au ho (s). Published by In o ma UK Limi ed, ading as Taylo & F ancis G oup
This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.
o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly
ci ed.
2I. CAZCARRO ET AL.
es ablishing a link be ween Na ional Accoun s (NA) da a and dis ibu ional in o ma ion
as a concep ual/s a is ical amewo k.
As –among o he s– Kim e al. (2015) ha e shown, ea ing he household agg ega e
and i s consump ion s uc u es in a he e ogeneous way (di e en ia ing household ypes)
e eals impo an insigh s in o s uc u al change d i en by socio-economic changes. The e
is a con inuum be ween in oducing many (hund eds o ) household g oups in a model and
linking a mul isec o al (IO o CGE) model wi h a mic o-simula ion model. Recen ly he e
has been some de elopmen in his a ea (see, among o he s, Colombo, 2008)andi has
been shown how his app oach ou pe o ms he ep esen a i e agen app oach o many
policy issues ha in ol e income dis ibu ion (Sa a d, 2004).
Ne e heless, a la ge pa o IO and CGE models do no ake ull ad an age o
he in o ma ion a ailable on consump ion s uc u es by household ypes in inc eas-
ingly publicly a ailable o icial da a. Indeed, se e al ac o s con ibu e o his s a e o
‘unde - esea ch’.
One impo an issue is ha , in all cases, he s uc u al in o ma ion in household su -
eys (Household Budge Su eys, HBS) needs o be b idged o he consump ion s uc u e
in o ma ion in IO s a is ics, which is no s aigh o wa d gi en he lack o publicly a ailable
‘con ingencyma ices’(commonlyknownas‘b idgema ices’
1) and he di e en alua-
ionme hodso he woda ase sin ol ed.Bo hissuescomplica eandhinde analyseso
he implica ions o household he e ogenei y o he e ec i eness o public policies o he
dis ibu ional e ec s o policies.
Con ingency/b idge ma ices a e pa o he na ional accoun s (NA) bu na ional s a is-
ical ins i u es (NSI) do no usually publish hem. In consequence, mos o b idge ma ices
used in he li e a u e a e ad-hoc es ima es and he me hods used o hei cons uc ion a e
no always explained in dep h. Among o he wo ks, some examples in he pas we e Kehoe
e al. (1988a,1988b) o he cons uc ion o social accoun ing ma ices (SAMs) and CGE
models;o Wie e al.(2001), Flo es & Maina (2009), S een-Olsen e al. (2016), o anal-
yse en i onmen al impac s. Se ano and Fe nandez-Vázquez (2017) ecen lyemphasised
he impo ance o using accu a e con ingency ma ices by showing how signi ican ly con-
sis ency can a ec he esul s in e ms o impac analysis. No ice ha in demand-d i en
models, he size o he impac is o en mo e due o he size o he shock han o he ech-
nological s uc u e (see o example A o & Die zenbache , 2014). The e o e, an e o in
he size o he shock is especially ele an .
On heo he hand, he ea e someaccoun ing di e ences be ween hehousehold expen-
di u es epo ed in su eys and he consump ion da a in he IO amewo k. Fo example,
expendi u e da a om su eys is epo ed in pu chase s’ p ices while consump ion da a in
IO ables is epo ed in basic p ices. Also, he da a ha e o be adjus ed due o di e ences
in he geog aphical scope (expendi u e da a ollows he esidence p inciple while IO da a
ollows he e i o ial p inciple) and, when he IO ables a e indus y by indus y, expendi-
u e da a ha e o be ans o med om p oduc o indus y. Thus, he linkages be ween he
wo da ase s equi es a numbe o ans o ma ions ha a e no always known by modelle s
1B idge ma ices a e made o ans o ma ion coefficien s o a ios. Con ingency ma ices a e he ma ices wi h alues ha
unde lie he coefficien s o he b idge ma ices. The ad an age o publishing con ingency ma ices is ha he use can
agg ega e ows and columns o adap hem o he agg ega ion needed.
ECONOMIC SYSTEMS RESEARCH 3
and p ac i ione s. Rele an s eps o ollow when linking consump ion (e.g. household) su -
eys o mul isec o al models we e explained in Mongelli e al. (2010) and Min and Rao
(2017).
In his con ex , his a icle in oduces a sys ema ic me hod o linking consump ion
da a om he IO amewo k and expendi u e su eys and p o ides es ima es o he b idge
ma ices o he 28 coun ies ha made up he Eu opean Union (EU-28) in 2016 (when
he consump ion su ey mic oda a cu en ly a ailable was eleased).
The emainde o he a icle is o ganised as ollows. Sec ion 2p esen s he da a and
me hods o b idging in o ma ion om consump ion su eys and mac oeconomic mod-
els, including a b ie desc ip ion o he di e en s eps o linking he wo da ase s and he
me hod o es ima ing he b idge ma ices, which a e a cen al elemen o he p ocedu e.
Sec ion 3shows he esul s o he es ima ion o con ingency ma ices o he EU-28 and
es s he main assump ions. Sec ion 4concludes and discusses he esul s.
2. Me hods
A i s sigh ,linkingexpendi u eda a omsu eysandconsump ionda a om heIO
amewo k could be seen as a simple p oblem ha can be sol ed wi h a di ec con e -
siono classi ica ionsandanadjus men me hodsuchasaRAS( hisis ep esen edby he
dashed a ow in Figu e 1). Howe e , as we show in Sec ion 2.1, hisapp oachiscomple ely
w ong and esul s in biased ou comes. P oducing esul s a he household le el sounds
e y appealing; howe e , be o e s a ing he calcula ions, i is impo an o unde s and and
apply he igh me hod in o de o p oduce obus esul s. As shown in he lowe panel o
Figu e 1, he p ocedu e in ol es a numbe o s eps and ele an concep s o he Na ional
Accoun s (NA).
Figu e 1. Linking mic oda a om consump ion su eys and mul isec o al models.
Sou ce: Own elabo a ion.
4I. CAZCARRO ET AL.
The e o e, he ollowing s eps should be conside ed be o e using consump ion-su ey-
based p o iles in mac oeconomic models:
(1) Align consump ion mic oda a o NA p inciples.
On he one hand, su ey da a does no ollow he accoun ing p inciples o NA which
a e a compendium o mul iple da a sou ces. On he o he hand, mul isec o al models
a e mainly based on NA; hus, i is necessa y o adap he da a om su eys o he
na ional accoun ing p inciples o be able o use i co ec ly in mac oeconomic mod-
elling (mo e de ails o he easons o his adap a ion and he p ocedu e a e gi en in
Sec ion 2.1).
(2) Con e consump ion mic oda a aligned o NA p inciples o p oduc ion-based
classi ica ions.
Consump ion da a is usually a ailable in classi ica ions o expendi u e acco ding o
pu pose such as COICOP (Classi ica ion o Indi idual Consump ion by Pu pose).
Howe e , mul isec o al mul i-indus y models ollow a classi ica ion o p oduc s
aligned wi h an indus y classi ica ion such as CPA (Classi ica ion o P oduc s by
Ac i i y, Eu os a , 2019a). Thus, i is necessa y o b idge he wo. Fo such b idging,
es ima ions o b idge ma ices based on public o icial da a o COICOP s CPA a e
used. The sub-s eps o his cen al s ep a e as ollows:
(a) P epa e he a ailable o icial con ingency ables o be used as p io s: hey should
ha e he p ope agg ega ion and e i o ial o esiden ial adjus men .
(b) P epa e he consump ion da a om he use ables o be used as inpu o he
es ima ion p ocess: his da a should be in pu chase s’ p ices and ha e he p ope
agg ega ion.
(c) Iden i y simila coun ies o selec he mos sui able p oxies o he coun ies o
which o icial con ingency ables a e no a ailable, o be used as p io s.
Es ima e he b idge ables consis en ly o each da abase, linking he consump-
ion in he use ables and in he inal consump ion expendi u e o households by
consump ion pu pose (COICOP 3-digi le el) (pa o he Na ional Accoun s).
In Sec ion 2.2we will in oduce u he de ails o his s ep and in Sec ion 3we will
show he esul s o he es ima ion p ocedu e.
(3) Change he alua ion o consump ion mic oda a in NA p inciples and p oduc ion-
based classi ica ions o basic p ices (bp).
Consump ion mic oda a and Household Final Consump ion Expendi u e (HFCE) in
NA a e in pu chase s’ p ices (pp). Howe e , mul isec o al models usually wo k a basic
p ices (p oduce p ices). The di e ence is he ne axes (i.e. axes less subsides) paid by
heuse bu no pe cei edby hep oduce and he adeand anspo a ionma gins
ha need o be eloca ed o he ‘ma gin’ indus ies(e.g. ade and anspo indus ies).
In Sec ion 2.3,weexplainhow ocon e hep ices om‘pp’ o‘bp’.
(4) Adjus he da a om he p oduc classi ica ion (i.e. CPA, see EC, 2008) o heindus y
classi ica ion (i.e. he S a is ical Classi ica ion o Economic Ac i i ies NACE, see EC,
2008,2010), i he model is based on an indus y classi ica ion.
Finally, i he IO able is p oduc by p oduc , he Household Final Consump ion in
basic p ices (HHFC) esul ing om he p e ious s ep can be connec ed o he IO able.
Howe e , wi h an indus y-by-indus y able like he OECD o WIOD, he ec o o
HHFC mus be con e ed in o an indus y classi ica ion (see Sec ion 2.4).

ECONOMIC SYSTEMS RESEARCH 5
The emainde o his sec ion explains he me hod.
2.1. Alignmen o consump ion mic oda a o NA p inciples
Mul isec o almodelssuchasIOandCGEmodelsa ebasedonSupplyandUse ables
and IO ables which a e a co e elemen o he NA (see Eu os a , 2019b). The e o e, he
accoun ing p inciples o NA a e e lec ed in he da a and consequen ly in he model. How-
e e , his is no he case o consump ion su eys. Thus, i is c ucial o align he accoun ing
p inciples o all he da ase s in ol ed.
Agg ega e HFCE da a o he whole economy is o en epo ed by NSIs using COICOP.
This da a a e pa o he NA and hei o als a e consis en (excluding in ages issues) wi h
he o alHHFCin heCPA epo edin heSupplyandUse ablesandIO ablesando he
mainagg ega eso heNA,suchas hespli o heGDP om heexpendi u eside.Howe e ,
he o al HFCE esul ing om summing up he HFCE o indi idual households epo ed in
consump ion su eys does no ma ch he agg ega e HFCE o NA, wi h di e ences anging
om 50% o 97% o EU Membe S a es in 2010 (Eu os a , 2018a).
Fu he mo e, no only he agg ega es, bu , mo e impo an ly, he s uc u es o he con-
sump ion su eys a e inconsis en wi h he HFCE igu es in NA (see Table 16 in Eu os a ,
2015). The co e age a e o he consump ion su eys wi h espec o he NA consump ion
a ied ac oss he di e en COICOP ca ego ies om 6% o 119% o EU membe s a es in
2010 (Eu os a , 2018a). This is due o he ac ha in he compila ion o HFCE o NA, con-
sump ion su eys a e a majo da a sou ce bu no he only one. Da a om su ey esul s is
complemen ed wi h addi ional in o ma ion such as ax s a is ics o anspo a ion su eys
o p oduce he HFCE o he whole economy. Amo es (2018) de ails some o he sou ces o
he di e ences such as concep ual and classi ica ion di e ences, measu emen e o s (e.g.
unde /o e -es ima ion o some ca ego ies like alcohol, obacco, housing, wa e , ene gy o
ood) and es ima ion e o s (e.g. high-income households being unde ep esen ed).
To make he mic oda a o he consump ion su eys consis en wi h he HFCE in NA, we
sugges he ollowing p ocedu e. Fi s , o e e y COICOP ca ego y, we calcula e he a io
be ween he HFCE in NA and he o al o he whole popula ion in he consump ion su -
ey. These a ios can be in e p e ed as scaling coe icien s o he a e age ep esen a i e
household o he NA. Second, we use hese coe icien s o align he consump ion p o-
ile ex ac ed om he su ey o he NA accoun ing s anda d. To do so, each ca ego y o
consume p o iles om he su ey is up a ed/down a ed by mul iplying i ca ego y-wise
agains such coe icien s. The exis ence o unma ched COICOP ca ego ies be ween HFCE
in NA and consump ion su eys makes a s epwise p ocedu e necessa y. Fu he de ails on
such p ocedu e can be ound in Sec ion A3.1 in he Supplemen a y Ma e ial.
2.2. Con e sion om COICOP o CPA
Once he consump ion da a in COICOP has been adap ed o he NA p inciples, hey mus
becon e edin oCPA.Thisisdoneusing heso-calledcon ingencyo b idgema ices.In
he case o he Eu opean Union, NSIs build hese b idge ables o compile NA. Howe e ,
hese ma ices a e no pa o he da ase s compulso ily submi ed o Eu os a (2014)and
NSIs do o do no publish hem depending on di e en conside a ions ( anspa ency, qual-
i y, con iden iali y, esou ces, e c.). In he case o he Eu opean Union, only eigh coun ies
6I. CAZCARRO ET AL.
(Aus ia, Czechia, Denma k, Es onia, Finland, Slo akia, Sweden and he Uni ed Kingdom)
make hem a ailable, and i is e y unlikely ha he majo i y o NSIs will publish such da a
in he sho un.
In his con ex , we sugges a sys ema ic p ocedu e o es ima e b idge ma ices ha we
apply o he es ima ion o he 20 coun ies o he Eu opean Union o which da a is no
a ailable.Thes a ingpoin sa e he ec o o consump ioninCPA(64p oduc s)and
COICOP (47 ca ego ies) om he NA o Eu os a . In he cases in which we ound di e -
ences be ween he wo ec o s, we p ese e he ec o o CPA, in o de o keep i consis en
wi h he IO ables. We combine his in o ma ion wi h he se o eigh a ailable o icial con-
ingency ma ices o 2010 ha will be used as a benchma k o es ima e he ma ices o he
emaining 20 coun ies.
The con ingency ables a e ma ices wi h a dimension o 64 (CPA p oduc s) x 47
(COICOPca ego ies).The elemen xi,jo hema ix ep esen s he o alquan i yo p oduc
i(e.g. chemical p oduc s) ha is used o he pu pose j(e.g. ou ine household main e-
nance); hesum ow-wisegi es he o alHHFCo heuse ableinpu chase s’p ices(CPA)
and he sum column-wise gi es he o al HFCE o he NA (COICOP).
One key elemen o he es ima ion o he con ingency ables is he selec ion o he
benchma k coun y whose s uc u e will be used as a p io . In ou p ocedu e, we y o
iden i y he mos sui able p oxies o he coun ies o which o icial con ingency ables
a e no a ailable by compa ing a se o mac o indica o s wi h hose o he eigh po en ial
benchma ks. In pa icula , we compa e he s uc u e o he HHFC om he use able in pp
(CPA classi ica ion); he s uc u e o he HFCE om he NA (COICOP classi ica ion); he
GDP pe capi a as an indica o o de elopmen s age, and he sociocul u al dis ance.
As Supplemen a y ma e ial, da a o he 28 ma ices can be ound in Annex A1 ( o use
be o e Annex A2 wi h he ans o ma ion o da a om pu chase ’s p ices o basic p ices),
and u he de ails on hei es ima es in Annex A3.
2.3. T ans o ma ion o da a om pu chase ’s p ices o basic p ices
Once he da a a e ma ched wi h he p ope classi ica ion (CPA s COICOP) and wi h he
accoun ing p inciples (NA s su eys), he alua ion o he da a mus be aligned. De ini-
ions o ypes o p ices can be ound in he UN Sys ems o Na ional Accoun s 2008 and
he Eu opean Sys ems o Accoun s 2010 (UN, 2009 and EC, 2013 espec i ely). As shown
in Figu e 1,consump ionda a,bo h omsu eysand omNA,isbasedonpu chase s’
p ices. In con as , he mul isec o al models usually wo k wi h basic p ices.
This s ep is equi ed o wo easons. Fi s , he ec o o consump ion in basic p ices
doesno include axesand,inconsequence,i islowe han he ec o inpp.Fo example,
jus looking a he o als o households’ consump ion om he Eu os a use ables, he alue
inpu chase s’p icesisona e age o heEU-28a ound13%highe han heoneinbasic
p ices, eaching in some cases mo e han 20%. The e o e, apa om all issues in ol ing
he s uc u e o consump ion, in gene al o each eu o o expendi u e om a consump ion
p o ile om su eys, 13% o he shock es ed wi h an IO able would be an o e es ima-
ion. Second, he ec o o consump ion in basic p ices eco ds all ade and anspo a ion
ma gins associa ed wi h inal consump ion in he so-called ma gin indus ies while in he
ec o o consump ion in pu chase s’ p ices (i.e. coming om consump ion su eys) hese
ma gins a e epo ed as pa o he alue o he p oduc consumed. Thus, he di ec link o
ECONOMIC SYSTEMS RESEARCH 7
consump ion p o iles in pu chase s’ p ices o an IO able in basic p ices would esul in an
unde es ima ion o he impac linked o he demand o ade and anspo a ion se ices
(hidden in he pu chase s’ p ices) and an o e es ima ion in he es o all o he indus ies.
Da a in pu chase s’ p ices is ans o med in o basic p ices using he ollowing in o ma-
ion: he ec o s o HHFC in basic p ices and pu chase s’ p ices, he ec o s o ma gins
associa ed wi h HHFC om he able o ade and anspo a ion ma gins, and he ne
axes associa ed wi h HHFC om he able o axes less subsidies on p oduc s. We use his
in o ma ion ocalcula e he implici ne ax a esandma gin a espe p oduc asdesc ibed
by Mongelli e al. (2010) who also explain how o use hem, as well as he limi a ions o he
app oach.
The Supplemen a y ma e ial includes a ool o con e consump ion p o iles om HBS
in pu chase s’ p ices (al eady in CPA and aligned o NA) in o basic p ices (Annex A2) and
u he de ails (Annex A3.3).
2.4. Adap a ion o he da a o he ype o IO able (p oduc s s indus ies)
The inal s ep consis s o linking he consump ion p o iles adjus ed o NA, in CPA and basic
p ices, o he IO ables. I he IO able is p oduc by p oduc , his can be done in a s aigh -
o wa dway.Howe e ,i heIO ableo CGEmodelisindus ybyindus y(e.g.WIODo
OECD-ICIO), an addi ional ans o ma ion is equi ed in o de o ans o m he p o iles
om p oduc s (CPA) o indus ies (NACE classi ica ion). This should be done ollowing
he same app oach ollowed o ans o ming he supply-use able in o he symme ical IO
able.
In he case o MRIO ables, de elope s o en use he ixed p oduc sales s uc u e
assump ion (Model D, see Box 12.3 in UN, 2018). The e o e, i consump ion p o iles a e
o be applied o an MRIO, he ecommenda ion is o use Model D o ans o ming he
consume p o iles as well.
On he o he hand, NSIs o en apply hyb id models (wi h ex ensi e in e nal da a) ha
canno be exac ly ep oduced by p ac i ione s. Howe e , hese hyb id models a e o en
close o he commodi y echnology assump ion han o he al e na i e indus y ech-
nology assump ion. On he o he hand, he commodi y echnology assump ion ends o
p oduce nega i es ha should be sol ed (i.e. applying Almon, 1970).
The omission o his s ep can also gene a e impo an e o s. As an example, we com-
pa ed he ec o o domes ic demand o households in he In e na ional Use ables 2010
om WIOD2(Timme e al., 2015) wi h he WIOT 2010 (cons uc ed ollowing Model D,
see UN, 2018) domes ic households ec o (Table 1). This gi es us a measu e o he size o
he po en ial e o i his s ep is omi ed.
We calcula ed he ela i e di e ences be ween he wo ables by p oduc o each coun-
y, and looked a he median and he maximum ac oss coun ies o each p oduc (uppe
pa o Table 1). The ela i e di e ences in each p oduc can be huge: 9% median ac oss
coun ies o he median ac oss p oduc s. Howe e , he median ac oss coun ies o he
maximum ac oss p oduc s is 100% and he ela i e di e ence can be as high as 3189%
(Manu ac u e o coke and e ined pe oleum p oduc s in C oa ia). Then, we ook he
2h p://www.wiod.o g/da abase/in _su s13 We look a he domes ic pa only because he impo ed consump ion is ob in
he In e na ional Use Tables and he e o e canno be di ec ly compa ed wi h he WIOT. The sha e o domes ic consump ion
o e o al consump ion o households’ accoun s is be ween 58% and 97% ac oss coun ies, wi h a median o 84%.
8I. CAZCARRO ET AL.
Table 1. Rela i e diffe ences and WAPE be ween he ‘Final consump ion expen-
di u e by households’ ec o s o each coun y in he WIOD In e na ional Supply-
Use Tables and he WIOT.
Median
Min. ac oss coun ies Max.
Summa y ac oss p oduc s
Median % diffe ence by p oduc 0% 9% 100%
Max. % diffe ence by p oduc 0% 100% 3189%
Summa y ac oss coun ies
Median % diffe ence by p oduc 0% 10% 86%
Max. % diffe ence by p oduc 0% 100% 3189%
WAPE 0% 9% 63%
Sou ce: Own elabo a ion based on he WIOD Release 2016.
median and maximum ac oss he p oduc s o each coun y independen ly, which, we sum-
ma ised ac oss coun ies (bo om pa o Table 1). The di e ences a e abo e 10% in hal
o he coun ies. To unde s and he implica ions, we need some con ex because he sha e
consumed o e e y p oduc is no he same, bu he e a e p oduc s ha a e much mo e
ele an in he consump ion baske han o he s. Tha is why we calcula ed he a e age
weigh ed de ia ions o each coun y (WAPE), which can be e y signi ican as well (9%
median ac oss coun ies, bu up o 63% in C oa ia and 58% in Luxembou g).
3. Resul s o es ima ion o con ingency ma ices o he EU-28
This sec ion shows he esul s o he es ima ion o he con ingency ma ices o he EU-
28.Sincewewan oachie eanapp op ia eme hod,whichwillbeinlinewi husingan
o icial con ingency ma ix o app oach he desi ed one, we i s examine he simila i ies
ac oss o icial con ingency ma ices ha a e a ailable and depa om his compa ison.
Then we iden i y benchma k ables by explaining assump ions o c i e ia which may lead
o good choices o benchma k ables. Finally, we es hese assump ions and p esen he
inal esul s.
3.1. Simila i ies ac oss o icial con ingency ma ices
The compa ison o a ailable o icial con ingency ma ices p o ides an idea o possible
sou ces and sizes o e o s when using he s uc u e o he ma ix o one coun y o es ima e
he ma ix o a second coun y. In o de o compa e he simila i y o di e ence among o i-
cial con ingency ables, we compa e he s uc u e o coe icien s (o sha es) o how each o
he COICOP ca ego ies is dis ibu ed ac oss CPA. We use he Weighed A e age Pe cen age
E o (WAPE) o he sha es o each cell o he con ingency ma ix in he o al o he coun-
y objec i e Owi h espec o hose o a benchma k coun y B,WAPE_sO,B(Annex A4
p esen s al e na i e measu es). The WAPE is de ined as:
WAPE_sO,B=
m

i=1
n

j=1|sO
ij −sB
ij|
m
i=1n
j=1sB
ij ×100 =
m

i=1
n

j=1
|sO
ij −sB
ij|×100 (1)
ECONOMIC SYSTEMS RESEARCH 15
As indica ed abo e, Annex A1 p o ides hese con ingency ables. The ables a e p o-
ided a he Eu os a s anda d o NA o CPA (64 ca ego ies) and 3-digi le el o COICOP
(wi h 47 ca ego ies, om CP011 o CP127).
Wi hsimpleagg ega ions,onecanob ain hecon ingency ables omwhichonede i es
he b idge a he le els o 56 CPA ca ego ies o he in e na ional WIOD ables Release 2016
and a he le els o 61 ca ego ies o EU coun ies in EORA. OECD agg ega ion in ISIC 4
(36 indus ies) can also be ob ained by agg ega ion, gi en ha he only indus y ha is
mo e disagg ega ed in OECD IICOIs han in Eu os a ables is CPA_B (mining) which is
no di ec lyconsumedbyhouseholds.
Simila ly, COICOP ca ego ies o he con ingency ables can be agg ega ed o accommo-
da e hem o he le el o de ail desi ed in he model used.
4. Conclusions
Using he de ailed in o ma ion con ained in consump ion su eys ha is becoming
inc easingly publicly a ailable opens up esea ch a enues o be e unde s anding
changes in consump ion s uc u e, d i en by indi idual beha iou al changes and socio-
demog aphic ends. In eg a ing ha in o economic modelling cons i u es a majo s ep
o wa d in analysing how changes in consump ion s uc u es d i e agg ega e s uc u al
change in an economy.
The consis en in oduc ion o in o ma ion om household consump ion su eys in o
he IO amewo k also opens up a whole a ea o esea ch on he impac o household
cha ac e is ics on economic and en i onmen al a iables. Tha includes issues like he a -
ious links (in bo h di ec ions) be ween he labou ma ke and income dis ibu ion, ca bon
oo p in s by income/age g oup and o he household cha ac e is ics and many mo e.
Howe e , he p ope in eg a ion o he in o ma ion o consump ion su eys in o he
s uc u e o economic models is no s aigh o wa d. The p ocess equi es a numbe o da a
manipula ions,whicha eno alwayswellknownbymodelle sbecausei equi eshighly
specialised expe ise in Na ional Accoun s and su eys. Fu he mo e, i also equi es addi-
ional da a in he o m o con ingency ables o de i e he sui able b idge ma ices ha , in
mos cases, a e no publicly a ailable. Indeed, al hough NSIs build hese b idge ables o
compile he Na ional Accoun s and he Supply-Use ables o he IO amewo k, i is e y
unlikely ha he majo i y o hem will publish his kind o da a in he nea u u e.
We b ie ly p esen ed a me hod, which summa ises he main asks o making his link:
(1) align su ey da a wi h Na ional Accoun s, (2) con e su ey da a om he expen-
di u e/consump ion classi ica ion (COICOP) o he p oduc classi ica ion (CPA), and (3)
change he alua ion o consump ion da a om pu chase s’ p ices o basic p ices. All hese
asks a e equi ed in o de o make his link. Howe e , hey a e o en neglec ed, unknown
o non- anspa en in many o he wo ks using da a om consump ion su eys.
The me hod and he se o con ingency ables p esen ed in his pape aim o p o ide
a comp ehensi e oolbox o make he link be ween consump ion su eys and economic
models in a igo ous manne . In his ega d, we conside ha he linking me hod and
con ingency ables p esen ed in his wo k, oge he wi h he ools in he Annexes, can
acili a e he use o mic oda a om consump ion su eys in IO analysis and economic
modelling.

16 I. CAZCARRO ET AL.
The main con ibu ion o his a icle, apa om cla i ying he linking p ocess, is he
se o con ingency ables ha p ac i ione s can use o de i e he b idge ma ices a hei
desi ed disagg ega ion o he 28 coun ies ha made up he Eu opean Union in 2016
(when he consump ion su ey mic oda a cu en ly a ailable was eleased). These con-
ingency ables cons i u e a key elemen o his da a in eg a ion. As analysed by Se ano
and Fe nandez-Vázquez (2017), inaccu a e b idges cause impo an biases in he analy -
ical s udies o consump ion. In addi ion, as he p ope es ima ion o hese ables is e y
ime-consuming and equi es in-dep h knowledge and expe ise on Na ional Accoun s,
hispape p o idesa eady- o-useda abase ha willbeuse ul o manyp ac i ione s.
We depa ed om 8 o icial publicly a ailable con ingency ables, ea anged hem
consis en ly o common classi ica ions, and es ima ed he emaining 20 ables.
We also ound ha while using an o icial able as a p io , e en i andomly selec ed, i
ou pe o ms ecen me hods such as he coun -seed RAS (Cai & Vandyck, 2020 and Cai &
Rueda-Can uche, 2019) and signi ican ly ou pe o ms he naï e p io .
We also de eloped a me hod o iden i ying which o he se en selec ed a ailable ables
is he bes o use as a p io o es ima e he con ingency able o each missing coun y, inally
wo king wi h all se en o hem as p io s. We success ully es ed he obus ness o his
benchma k iden i ica ion me hod. All in all, as mo e o less summa ised by a e iewe ,
om his wo k i eme ges ha ‘i you wan o ob ain a able o coun y ‘D’, make a RAS o
he indica ed b idge ma ix o coun y ‘A’ and you will ge a good app oxima ion.’
The limi a ion o he es is he concen a ion o a ailable o icial ables used as bench-
ma ks in he no he n Eu opean coun ies. Ob iously, ha ing some benchma k con in-
gency ables a ailable o sou he n Eu opean coun ies would imp o e ou es ima es o
his se o coun ies.
Toge he wi h he limi ed o a ailabili y o con ingency ma ices o e ime, he a icle
has a geog aphical ocus on he Eu opean Union since i has mic oda a on consump ion
a ailable o esea ch pu poses, which is no common wo ldwide. Howe e , he me hod
is gene al and alid whene e he e is simila da a and simila i ies be ween a ge ed and
benchma k coun ies.
Fu he mo e, we ound s abili y in he choice o p io coun ies acco ding o he s uc-
u es o he sea ched o al ec o s o he con ingency ma ices. Consequen ly, ou es ima es
a e alid o he Eu opean Union coun ies o he WIOD and OECD and we conside
he applica ion o he me hod o o he coun ies wo hy o u he esea ch, as well as
o ele an da abases such as EXIOBASE o GTAP. Finally, we ecommend ha u u e
upda es o all hese da abases also include in o ma ion b idging he da a on expendi u e
wi h classi ica ions like COICOP. This should comp ise mo e han he b idge COICOP /
p oduc classi ica ion (speci ic o he da abase), bu also he su ey o NA adap a ion and
he pu chase s’ p ices o basic p ices ans o ma ion.
I should be s essed ha , due o he impo ance o hese con ingency ables o eco-
nomicanalysis,ideally hisexe ciseshouldbeca iedou wi hino icials a is icalp o-
g ammes o NSIs. This would educe he numbe o assump ions and inc ease he quali y
o heda a.Fo example,a hele elo Eu os a o heOECD,beingable oga he o icial
con ingency ma ices in a consis en manne , wi h di e en links o s anda d classi ica ions
ando he pa s o he Na ional Accoun s, wouldeno mously educe unce ain y,biases and
assump ions made by esea che s and use s.
ECONOMIC SYSTEMS RESEARCH 17
Acknowledgemen s
The au ho s a e g a e ul o he e iewe s and o he edi o Man ed LENZEN, whose insigh ul
con ibu ions helped o imp o e he a icle. The au ho s a e also g a e ul o Sanji MAHAJAN
(O ice o Na ional S a is ics, Uni ed Kingdom), Pa KILDUFF (Cen al S a is ics O ice, I eland),
Yl a PETERSSON STRID (S a is ics Sweden), Isabelle REMOND-TIEDREZ, Pille DEFENSE-
PALOJARV, Sigi a GRUNDIZA, F ide ike OEHLER and E ika TAIDRE (Eu os a ) o hei
cla i ica ions ega ding o icial da a. Anna ATKINSON (Eu opean Commission’s Join Resea ch
Cen e) kindly p oo ead he manusc ip .
Ignacio Cazca o and Iñaki A o hank he Spanish Minis y o Science, Inno a ion and Uni e si-
ies o he RTI2018-099858-A-I00 g an (p ojec “MALCON”) and o he PID2019-106822RB-I00
g an (in which Ignacio Cazca o pa icipa es) unde he p og amme o R&D P ojec s Resea ch
Challenges; he Spanish S a e Resea ch Agency h ough Ma ía de Maez u Excellence Uni acc edi a-
ion 2018–2022 (Re . MDM-2017-0714) and he p ojec LOCOMOTION H2020-LC-CLA-2018-2
(No 821105).
Disclaime
The iewsexp essedin hispape belong o heau ho sandshouldno bea ibu ed o he
ins i u ions o which he au ho s a e a ilia ed.
Disclosu e s a emen
No po en ial con lic o in e es was epo ed by he au ho (s).
Funding
This wo k was suppo ed by Minis e io de Ciencia, Inno ación y Uni e sidades: [g an numbe s
RTI2018-099858-A-I00 and PID2019-106822RB-I00]; he Spanish S a e Resea ch Agency h ough
Ma ía de Maez u Excellence Uni acc edi a ion 2018-2022 [g an numbe MDM-2017-0714] and
he p ojec LOCOMOTION H2020-LC-CLA-2018-2 [g an numbe 821105].
ORCID
Ignacio Cazca o h p://o cid.o g/0000-0001-7517-0053
An onio F. Amo es h p://o cid.o g/0000-0002-0216-4098
Inaki A o h p://o cid.o g/0000-0001-6427-7437
Ku K a ena h p://o cid.o g/0000-0002-0257-4169
Re e ences
Almon, C. (1970). In es men in inpu –ou pu models and he ea men o seconda y p oduc s.
In A. P. Ca e , & A. B ôdy (Eds.), Applica ions o Inpu –Ou pu Analysis (pp. 103–116). No h-
Holland.
Amo es,A.F.(2018, Sep embe 27–28). The challenge o using consump ion su eys o eed mac oe-
conomic models. 6 h SHAIO Wo kshop, Mad id.
A o, I., & Die zenbache , E. (2014). D i e s o he g ow h in global g eenhouse gas emissions.
En i onmen al Science & Technology, 48(10), 5388–5394. h ps://doi.o g/10.1021/es5005347.
B ock, W. A., & Du lau , S. N. (2001). In e ac ions-based models. In J. Heckman, & E. Leame (Eds.),
Handbook o Econome ics olume 5, chap e 54 (pp. 3297–3380). No h Holland.
Cai,M.,&Rueda-Can uche,J.M.(2019). B idging mac oeconomic da a be ween s a is ical
classi ica ions: he coun -seed RAS app oach. Economic Sys ems Resea ch,31(3), 382–403.
h ps://doi.o g/10.1080/09535314.2018.1540404.
18 I. CAZCARRO ET AL.
Cai, M., & Vandyck, T. (2020). B idging be ween economy-wide ac i i y and household-
le el consump ion da a: ma ices o Eu opean Coun ies. Da a in B ie ,30, 105395.
h ps://doi.o g/10.1016/j.dib.2020.105395
Colombo, G. (2008). Linking CGE and mic osimula ion models: A compa ison o di e en
app oaches, ZEW discussion pape , No. 08-054, ZEW, Mannheim, July 2008.
Eu opean Commission. (2008). Regula ion (EC) No 451/2008 o he Eu opean Pa liamen and o
he Council o 23 Ap il 2008 es ablishing a new s a is ical classi ica ion o p oduc s by ac i i y
(CPA) and epealing Council Regula ion (EEC) No 3696/93 (O ice o O icial Publica ions o
he Eu opean Communi ies, Luxembou g).
Eu opean Commission. (2010). Eu opean Commission Regula ion (EU) No 715/2010 o 10 Augus
2010amending CouncilRegula ion(EC) No2223/96 as ega dsadap a ions ollowing he e ision
o he s a is ical classi ica ion o economic ac i i ies NACE Re ision 2 and he s a is ical classi i-
ca ion o p oduc s by ac i i y (CPA) in na ional accoun s (O ice o O icial Publica ions o he
Eu opean Communi ies, Luxembou g).
Eu opean Commission. (2013). Eu opean Pa liamen and Eu opean Council Regula ion (EU) o 21
May 2013 on he Eu opean Sys em o Accoun s 2010.
Eu os a . (2014). Eu opean sys em o accoun s - ESA 2010 - T ansmission p og amme o da a.Eu os a
Di ec o a e-Gene al o he Eu opean Commission. h ps://ec.eu opa.eu/eu os a /web/p oduc s-
manuals-and-guidelines/-/KS-01-13-429-3A-C.
Eu os a . (2015). Household budge su ey 2010 wa e, EU Quali y epo , Ve sion 2 July 2015.
Eu os a Di ec o a e-Gene al o he Eu opean Commission.
Eu os a . (2018a). Concep s o households’ consump ion - compa ison be ween Household Budge
Su ey andNa ionalAccoun s.S a is icsExplained.Eu os a Di ec o a e-Gene alo he Eu opean
Commission. h ps://ec.eu opa.eu/eu os a /s a is ics-explained/index.php? i le=Concep s_ o _
household_consump ion_-_compa ison_be ween_mic o_and_mac o_app oach&oldid=409044.
Eu os a . (2018b). ‘COICOP 1999 - CPA 2008’. Re e ence and Managemen O Nomencla u es.
Eu os a Di ec o a e-Gene al o he Eu opean Commission.
Eu os a . (2019a). ‘CPA 2008 s uc u e’/ ‘CPA 2008 in oduc o y guidelines’. Eu os a Di ec o a e-
Gene al o he Eu opean Commission. h ps://ec.eu opa.eu/eu os a /web/cpa-2008.
Eu os a . (2019b). Re iew o na ional supply, use and inpu -ou pu ables compila ion.Eu o-
s a Di ec o a e-Gene al o he Eu opean Commission. h ps://ec.eu opa.eu/eu os a /s a is ics-
explained/index.php? i le=Re iew_o _na ional_supply,_use_and_inpu -ou pu _ ables_
compila ion.
Fe nández Vazquez, E., Rubie a Mo ollón, F., & Apon e Ja amillo, E. (2013). Es ima ing spa ially
disagg ega ed da a by en opy econome ics: An exe cise o ecological in e ence o income in
Spain. Resea ch in Applied Economics,5(4), 80. h ps://doi.o g/10.5296/ ae. 5i4.4373
Flo es-Ga cía, M., & Maina -Causapé, A. J. (2009). Ma iz de con abilidad social y mul iplicado es
con ables pa a la economía a agonesa. Es adís ica Española,51(172), 431–469.
Hoo e ,K.D.(2008). Idealizing educ ion: The mic o ounda ions o mac oeconomics (27 May
2008).
Kaasa, A., Vadi, M., & Va blane, U. (2016). A new Da ase o Cul u al Dis ances o Eu o-
pean Coun ies and Regions. Resea ch in In e na ional Business and Finance,37, 231–241.
h ps://doi.o g/10.1016/j. iba .2015.11.014
Kehoe, T., Man esa, A., Noyola, P., Polo, C., & Sancho, F. (1988b). A Gene al Equilib ium
Analysis o he 1986 ax Re o m in Spain. Eu opean Economic Re iew,32(2-3), 334–342.
h ps://doi.o g/10.1016/0014-2921(88)90177-8
Kehoe, T., Man esa, A., Polo, C., & Sancho, F. (1988a).Unama izdecon abilidadsocialdela
economía española. Es adís ica Española,30(117), 5–33. h ps://www.ine.es/ss/Sa elli e?blobcol=
u lda a&blobheade =applica ion%2Fpd &blobheade name1=Con en -Disposi ion&blobheade
alue1=a achmen %3B+ ilename%3D117_1.pd &blobkey=u lda a&blob able=MungoBlobs&
blobwhe e=397%2F963%2F117_1%2C0.pd &ssbina y= ue
Kim,K.,K a ena,K.,&Hewings,G.J.D.(2015). The ex ended econome ic inpu -ou pu model
wi h he e ogeneous household demand sys em. Economic Sys ems Resea ch,27(2), 257–285.
h ps://doi.o g/10.1080/09535314.2014.991778
ECONOMIC SYSTEMS RESEARCH 19
Ki man,A.(1992). Whom o wha does he ep esen a i e indi idual ep esen . Jou nal o Economic
Pe spec i es,6(2), 117–136. h ps://doi.o g/10.1257/jep.6.2.117
Kullback, S. (1997). In o ma ion heo y and s a is ics. Do e Publica ions.
Kullback, S., & Leible , R. (1951). On in o ma ion and su iciency. Annals o Ma hema ical S a is ics,
22(1), 79–86. h ps://doi.o g/10.1214/aoms/1177729694
Leh inen, A., & Kuo ikoski, J. (2007). Un ealis ic assump ions in a ional choice heo y. Philosophy
o he Social Sciences,37(2), 115–138. h ps://doi.o g/10.1177/0048393107299684
Min, J., & Rao, N. D. (2017). Es ima ing unce ain y in household ene gy oo p in s. Jou nal o
Indus ial Ecology,22(6), 1307–1317. h ps://doi.o g/10.1111/jiec.12670
Mongelli,I.,Neuwahl,F.,&Rueda-Can uche,J.M.(2010). In eg a ing a Household Demand Sys em
in he Inpu –Ou pu F amewo k. Me hodological Aspec s and Modelling Implica ions. Economic
Sys ems Resea ch,22(3), 201–222. h ps://doi.o g/10.1080/09535314.2010.501428
Robinson,S.,Ca aneo,A.,&El-Said,M.(2001). Upda ing and es ima ing a social accoun ing ma ix
using c oss en opy me hods. Economic Sys ems Resea ch,13(1), 47–64. h ps://doi.o g/10.1080/
09535310120026247
Sa a d, L. (2004). Po e y and inequali y analysis wi hin a CGE amewo k: A compa a i e analysis
o he ep esen a i e agen and mic o-simula ion app oaches, CIRPÉE, Wo king Pape 04-12,
Uni e si é La al.
Se ano, M., & Fe nandez-Vázquez, E. (2017). Technology o he p e e ences: linking consump ion
expendi u es o alueaddedwi h minimalin o ma ion.UBEconomicsWo kingPape s2017/365.
h ps://www.ub.edu/school-economics/wo kingpape s/ echnology-p e e ences-linking-
consump ion-expendi u es- alue-added-minimal-in o ma ion/.
S een-Olsen, K., Wood, R., & He wich, E. G. (2016). The ca bon oo p in o no we-
gian household consump ion 1999-2012. Jou nal o Indus ial Ecology,20(3), 582–592.
h ps://doi.o g/10.1111/jiec.12405
S igli z, J. E. (2018). Whe e Mode n Mac oeconomics Wen W ong. Ox o d Re iew o Economic
Policy,34(4), 70–106. h ps://doi.o g/10.1093/ox ep/g x040
S igli z, J. E., Sen, A., & Fi oussi, J. P. (2009). Repo by he commission on he measu emen o
economic pe o mance and social p og ess.
Timme ,M.P.,Die zenbache ,E.,Los,B.,S eh e ,R.,&deV ies,G.J.(2015). An illus a ed use
guide o he wo ld inpu –ou pu da abase: The case o global au omo i e p oduc ion. Re iew o
In e na ional Economics,23(3), 575–605. h ps://doi.o g/10.1111/ oie.12178
UN. (2018). Handbook on Supply, use and Inpu -Ou pu Tables wi h Ex ensions and Applica ions.
S udies in Me hods Handbook o Na ional Accoun ing Se ies F No.74, New Yo k, NY, USA.
UN. (2009). Sys em o Na ional Accoun s 2008, ST/ESA/STAT/SER.F/2/Re .5. Eu opean Communi-
ies, In e na ional Mone a y Fund, O ganisa ion o Economic Co-ope a ion and De elopmen ,
Uni ed Na ions and Wo ld Bank, New Yo k, USA.
Wie , M., Lenzen, M., & Munksgaa d, J. (2001). E ec s o Household Consump ion Pa e ns on CO2
Requi emen s. Economic Sys ems Resea ch,13(3), 259–274. h ps://doi.o g/10.1080/095373201
20070149