Does the World Cup get the economic ball rolling? Evidence from a synthetic control approach

Viana, Jo ge H. N.; Ba bosa, An onio Vinicius; Sampaio, B eno A icle Does he Wo ld Cup ge he economic ball olling? E idence om a syn he ic con ol app oach EconomiA P o ided in Coope a ion wi h: The B azilian Associa ion o Pos g adua e P og ams in Economics (ANPEC), Rio de Janei o Sugges ed Ci a ion: Viana, Jo ge H. N.; Ba bosa, An onio Vinicius; Sampaio, B eno (2018) : Does he Wo ld Cup ge he economic ball olling? E idence om a syn he ic con ol app oach, EconomiA, ISSN 1517-7580, Else ie , Ams e dam, Vol. 19, Iss. 3, pp. 330-349, h ps://doi.o g/10.1016/j.econ.2018.05.001 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/266927 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ A ailable online a www.sciencedi ec .com ScienceDi ec HOSTED BY EconomiA 19 (2018) 330–349 Does he Wo ld Cup ge he economic ball olling? E idence om a syn he ic con ol app oach Jo ge H. N. Vianaa,∗, An onio Vinicius Ba bosaa, B eno Sampaiob aDepa men o Economics, Uni e sidade Fede al da Pa aíba, Ja dim Cidade Uni e si a ia, 58059-356 João Pessoa, Pa aíba, B azil bDepa men o Economics, Uni e sidade Fede al de Pe nambuco, Rua dos Economis as, Cidade Uni e si a ia, 50740-590 Reci e, Pe nambuco, B azil Recei ed 30 No embe 2015; ecei ed in e ised o m 29 Ma ch 2018; accep ed 9 May 2018 A ailable online 16 May 2018 Abs ac In his pape we analyze he impac o hos ing he FIFA Socce Wo ld Cup on GDP pe capi a in a wo ldwide sample o coun ies using a anspa en s a is ical me hodology o da a-d i en case s udies – he syn he ic con ol me hod. Using coun y le el annual- da a co e ing all e en s occu ing in he pe iod be ween 1978 (A gen ina) and 2006 (Ge many), we show ha he es ima ed a e age ea men e ec was ei he ze o o nega i e o all bu one o he coun ies analyzed. Ou esul s, he e o e, suppo he gene al claim ha Wo ld Cups a e no s a is ically associa ed o de elopmen and economic g ow h. JEL classifica ion: O4; O5; C50 Keywo ds: Economic g ow h; Wo ld Cup; Syn he ic con ol me hod P oduc ion and hos ing by Else ie B.V. on behal o Na ional Associa ion o Pos g adua e Cen e s in Economics, ANPEC. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). 1. In oduc ion In his pape we use coun y-le el annual da a on GDP pe capi a o analyze he impac o hos ing he FIFA Socce Wo ld Cup on g ow h. Fo ha we use he syn he ic con ol me hod o cons uc adequa e coun e ac uals o all coun ies hos ing he e en in he pe iod be ween 1978 and 2006. The FIFA (Fédé a ion In e na ionale de Foo ball Associa ion) Socce Wo ld Cup anks among he h ee la ges e en s in he wo ld, oge he wi h he Olympic Games and he Wo ld Expo. These e en s no only a ec he in lux o people/ ou ism in he hos coun y, bu also in ol es publicly inanced capi al imp o emen p ojec s ha a e unde aken o imp o e in as uc u e, such as he cons uc ion o new and he imp o emen o old s adiums ( o he pa icula case o he Wo ld Cup), oad and ai po cons uc ion and imp o emen , among many o he s. In U uguay, o ins ance, which held he i s edi ion o he Cup in 1930, o al a endance was li le abo e hal million spec a o s, while his ∗Co esponding au ho . E-mail add esses: jo ge[email p o ec ed].b (J.H.N. Viana), [email p o ec ed] (A.V. Ba bosa), [email p o ec ed] (B. Sam- paio). Pee e iew unde esponsibili y o Na ional Associa ion o Pos g adua e Cen e s in Economics, ANPEC. h ps://doi.o g/10.1016/j.econ.2018.05.001 1517-7580 P oduc ion and hos ing by Else ie B.V. on behal o Na ional Associa ion o Pos g adua e Cen e s in Economics, ANPEC. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 331 Table 1 FIFA Wo ld Cups. Cup yea Yea o decision o hos Hos Numbe o eams To al a endance 1930 1929 U uguai 13 590,549 1934 1932 I aly 16 363,000 1938 1936 F ance 16 375,700 1950 1946 B azil 13 1,045,246 1954 1946 Swi ze land 16 768,607 1958 1950 Sweden 16 819,810 1962 1956 Chile 16 893,172 1966 1960 England 16 563,135 1970 1964 Mexico 16 1,603,975 1974 1966 Ge many 16 1,865,753 1978 1966 A gen ina 16 1,545,791 1982 1966 Spain 24 2,109,723 1986 1983 Mexico 24 2,394,031 1990 1984 I aly 24 2,516,215 1994 1988 USA 24 3,587,538 1998 1992 F ance 32 2,785,100 2002 1996 Ko a/Japan 32 2,705,197 2006 2000 Ge many 32 3,359,439 2010 2004 Sou h A ica 32 3,178,856 Sou ce: FIFA. numbe su passed h ee millions in wo ecen edi ions held in Sou h A ica, wi h 3.178 millions spec a o s in 2010, and in Ge many, wi h 3.359 millions in 2006 (see Table 1). The amoun o in es men s has also inc eased subs an ially, eaching es ima es in he o de o D 6 billion o bo h Ge many 2006 and Sou h A ica 2010. Besides he isks and high cos s in ol ed in hos ing an e en o his magni ude, all p e ious Wo ld Cup edi ions ha e had la ge compe i ion be ween po en ial hos coun ies (FIFA, 2012). Fo example, 1930 Wo ld Cup had six candida es (Hunga y, I aly, Ne he lands, Spain, Sweden and U uguay), while 2018 and 2022 had six and i e candida es, espec i ely. This beha io clea ly indica e ha coun ies unde s and ha he e a e g ea bene i s in hos ing such an e en . Common a gumen s ci ed in a o o such decision include highe economic g ow h a es, educ ion on unemploymen a es, inc ease in ou is ic ac i i ies and go e nmen income, inc ease in capi al in low and an imp o emen o he image o he coun y wo ldwide. The a ionale behind hese a gumen s is ha he Wo ld Cup en ails no only in es men s o build and upda e s adiums, bu also a lo o in es men in secu i y, u ban mobili y and ai po in as uc u e. Cu en li e a u e showing empi ically he connec ion be ween he Wo ld Cup and he a iables ci ed abo e is sca ce, wi h almos all pape s published on he subjec ocusing on es ima ing he e ec o he e en on economic g ow h. These s udies may be di ided in o wo b oad ca ego ies: ex-an e and ex-pos s udies. The i s o hese ca ego ies ocus on using inpu -ou pu ma ices (ex-an e analysis), while he second ocus mainly on ime se ies o di e ences-in-di e ences/ ixed e ec s models (ex-pos analysis) o es ima e he e ec o in e es . The esul s p esen ed o bo h ex-an e and ex-pos s udies could no be mo e di e se. While some ex-pos pape s ind posi i e, insigni ican o e en nega i e e ec s o Wo ld Cups on a iables such as GDP, unemploymen , go e nmen ax income and ou is in low, ex-an e pape s p edic la ge posi i e e ec s o hese same a iables when analyzing he same coun ies conside ed on ex-pos pape s. Fo example, Bohlmann and an Hee den (2005), using a Compu able Gene al Equilib ium (CGE) model de eloped speci ically o he Sou h A ican economy, p edic ed ha he 2010 Wo ld Cup held in Sou h A ica would a ec eal GDP in excess o R10 billion (abou US$1.18 billion), wi h mo e han 50,000 jobs being c ea ed by he cons uc ion o new enues and upg ading o exis ing in as uc u e. Addi ionally, he au ho s concluded ha he ‘[e]xpec ed imp o emen o he in as uc u e o he coun y, especially he anspo sec o , would g ea ly bene i p oduc i i y in he longe e m and u he inc ease GDP.’ Mo ing o pape s ha pe o m ex-pos analysis, Hagn and Maennig (2008, 2009) ound insigni ican o nega i e e ec s o Wo ld Cups on unemploymen using, espec i ely, da a o he e en s held in Ge many in he yea s o 1974 and 2006. Fo he e en held in he US, Baade and Ma heson (2004), using da a a he Me opoli an S a is ical A ea (MSA) le el, ound insigni ican o nega i e e ec s o he Wo ld Cup on GDP g ow h. Finally, Allme s and Maennig (2009) analyzed he e ec s o he Wo ld Cups held in Ge many 2006 and F ance 1998 on o e nigh s ays a ho els, 332 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 na ional income om ou ism, and e ail sales and ound ha none o he esul s we e s a is ically signi ican when using F ench da a. When using Ge man da a, howe e , hey es ima ed an addi ional 700,000 o e nigh s ays and US$900 million in ne na ional ou ism income. One o he main easons why ex-an e s udies end o epo in la ed numbe s when compa ed o ex-pos s udies is ha hey do no ake in o accoun ha mos o he in es men in ol ed in a Wo ld Cup is no new money en e ing he economy, bu esou ces aken om o he sec o s in which hey could ha e o he wise induced la ge economic bene i s. In addi ion, a g ea pe cen age o he e en a endance include local esiden s, and his is pa icula ly ue o Wo ld Cups in coun ies wi h g ea adi ion in oo ball. Also, a endance include people who would ha e isi ed he hos coun y e en i he e was no e en . The e o e, as a gued by C omp on (1995), one ough o expec he boos obse ed in ou ism o be no as la ge as he numbe s p edic ed by ex-an e models. Mos pape s in his li e a u e, howe e , migh su e om he consequences o he well known ‘ undamen al p oblem o causal in e ence’ (Holland, 1986), which imposes an addi ional challenge o empi ical esea che s when es ima ing causal e ec s o policy changes. This p oblem exis s due o he ac ha compa isons o wo ou comes o in e es o he same uni when exposed, and when no exposed, o a ea men is an in easible ask, gi en he same uni can ei he pa icipa e o no in a p og am in he same pe iod (Imbens and Woold idge 2009). In o he wo ds, one can ne e obse e a speci ic coun y when unde and when no unde he in luence o a Wo ld Cup a he same poin in ime (Sampaio, 2014). Hence, es ima es based on inpu /ou pu ma ices o imes se ies models (o e en di e ences- in-di e ences models unde he p esence o ime- a ying unobse able con ounde s) a e only alid unde s ong assump ions ega ding shocks ha a e co ela ed o he policy being e alua ed. In his pape , we mo e away om he me hods p e iously used o s udy he subjec , which o he mos pa ely on he assump ion o selec ion on obse ables, and p opose o use he syn he ic con ol me hod (SCM), a echnique which gained popula i y ecen ly and is well sui ed o s udy he p oblem add essed in his pape . This me hod, which was de eloped by Abadie and Ga deazabal (2003) and ex ended by Abadie e al. (2010), uses da a-d i en p ocedu es o cons uc adequa e compa ison g oups/coun e ac uals gi en ha , in p ac ice, i is a di icul ask o ind a single uni /coun y unexposed o he policy change o in e es ha app oxima es he mos , ele an cha ac e is ics o he ea ed uni ( he coun y ha had a Wo ld Cup, o example) and ha would p o ide a good con ol g oup. In o he wo ds, he me hod will p o ide he esea che wi h an op imal weigh o each coun y such ha he weigh ed a e age o he a iable one is in e es ed in explaining (in ou case, GDP pe capi a) bes app oxima es he alue o his a iable o he coun y ha had he policy change (in ou case, hose coun ies hos ing a Wo ld Cup). Acco ding o Belo and Vandenbe ghe (2009), he basic in ui ion behind he SCM is ha a combina ion o coun ies – a syn he ic con ol – o e s a be e compa ison han any single coun y alone. A second ad an age o he p oposed me hod, as highligh ed by Billmeie and Nannicini (2013), is ha , unlike mos o he es ima o s used in he li e a u e o p og am e alua ion and in he li e a u e analyzing e ec s o wo ld cups, i can deal wi h endogenei y om omi ed a iable bias by accoun ing o he p esence o ime- a ying unobse able con ounde s. This is a signi ican imp o emen conside ing p e ious analysis using imes se ies and ixed e ec s models, which can only accoun o ime-in a ian unobse able con ounde s. This me hodology was ecen ly used by Abadie e al. (2010) o analyze he e ec s o P oposi ion 99, a la ge-scale obacco con ol p og am ha Cali o nia implemen ed in 1988, on obacco consump ion using annual s a e-le el panel da a o he pe iod 1970–2000; by Belo and Vandenbe ghe (2009) o analyze he e ec s o g ade e en ion on a ainmen using a e o m in oduced in 2001 in he F ench-Speaking Communi y o Belgium whe eby he possibili y o g ade e en ion in g ade se en was ein oduced; by Billmeie and Nannicini (2013) o analyze he impac o economic libe aliza ion on eal GDP pe capi a in a wo ldwide sample o coun ies using annual da a co e ing abou 180 coun ies o e he pe iod 1963–2000; and by Sampaio (2014) o analyze he e ec s o New Yo k s a e’s law p ohibi ing handheld cell phone use while d i ing on a ali y a es using annual s a e-le el panel da a o he pe iod o 1995–2006. The e o e, gi en he ad an ages o he SCM ega ding he s uc u e o he da a and he ins i u ional amewo k o he p oblem being analyzed in his pape , we see he SCM as a p omising s a egy o o e come some o he sho comings o p e iously p oposed me hods and as a decen ins umen o analyze he impac o la ge spo ing e en s on mac oeconomic a iables. Ano he con ibu ion o ou pape ela es o he numbe o e en s conside ed in he analysis. Wi h only one excep ion (Allme s and Maennig, 2009), all pape s es ima ing he e ec s o wo ld cups look only a one e en . In his a icle, we expand p e ious esea ch and o e a se o empi ical coun y s udies o bes analyze he ela ionship be ween he e en s and he pa e n o income pe capi a. We conside a o al o 8 e en s held in 9 coun ies, co e ing all Wo ld Cups occu ing in he pe iod be ween 1978 (A gen ina) and 2006 (Ge many). J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 333 Ou empi ical indings show ha o all he coun ies conside ed in he analysis (Ge many, Japan, Ko ea, F ance, Uni ed S a es, I aly, Mexico, Spain and A gen ina), he Wo ld Cup had a null o a nega i e e ec on income pe capi a. We should emphasize ha p e- ea men adjus men be ween eal and syn he ic con ol o he majo i y o coun ies ci ed abo e was qui e good, wi h eal GDP ime se ies almos o e lapping ha o syn he ic GDP ime se ies, which alida es de exe cise being ca ied ou . Fo Spain, howe e , he syn he ic con ol p esen ed a poo p e- ea men i , which implies ha a sui able coun e ac ual was no ound and his educes he in e en ial alue o his speci ic expe imen . The e o e, ou esul s suppo he gene al conclusion ha Wo ld Cups a e no s a is ically associa ed o economic g ow h. Looking a i s mos ecen edi ion (2014 Wo ld Cup in B azil), C espo (2014) es ima ed ha o al o eseen cos o he Cup would be abou R$ 26 billion, wi h ‘only’ R$ 8 billion spen in s adiums (31% o he o al expenses) and R$ 13 billion (50% o he o al) spen in u ban mobili y and ai po in as uc u e. Ne e heless, despi e hese la ge absolu e numbe s, o al o eseen cos o he 2014 Wo ld Cup is s ill a d op on he ocean when compa ed o he huge public budge o B azil ( o example, i ep esen s only 9% o o al annual cos s on public educa ion, as no ed by Pa u e al., 2014). The e o e, besides he Cup being a huge spo ing e en , i appea s no o ha e he economic impo ance as many a gue o ha e. Mo eo e , as inpu -ou pu s udies end o o ge , mos o hese so called in es men s ha a ise once he coun y is assigned as hos a e no new esou ces, bu a e in ac ealloca ed esou ces. The e o e, hese wo aspec s aken oge he cas se ious doub on he possible bene i s o hos ing such an e en . A e his in oduc ion, he es o he pape is o ganized as ollows. Sec ion 2 desc ibes in de ail he syn he ic con ol app oach o compa a i e case s udies o agg ega e e en s. In Sec ion 3 he da a used is p esen ed and in Sec ion 4 esul s a e discussed. Finally, a ew concluding ema ks a e d awn in Sec ion 5. 2. Me hodology In his sec ion we p esen he empi ical s a egy used o iden i y he causal e ec o hos ing a Wo ld Cup on he ou comes o in e es . Le Yc be he ou come o coun y ca ime , WCc be a dummy a iable ha assumes alue equal o 1 o he yea s ollowing he occu ence o a Wo ld Cup (o ollowing he announcemen o he hos ing coun y), and εc be unobse ed de e minan s o he ou come a iable. The pa ame e o in e es , β1, which ep esen s he e ec o he Wo ld Cup on he ou come, may be es ima ed ia he ollowing model Yc = β0+ β1∗ WCc + εc (1) One can easily e i y ha by es ima ing equa ion 1 using da a only o he coun y ha had a Wo ld Cup, he pa ame e o in e es would equal he a e age o he ou come a iable a e he Wo ld Cup (when WCc = 1) minus he a e age o he ou come a iable be o e he Wo ld Cup (when WCc = 0). I is ha d o a gue, howe e , ha such di e ence ep esen he causal e ec o he Wo ld Cup, gi en ha o he con ounding ac o s no con olled o ha migh comp omise iden i ica ion, ha is, i migh be ha COV (WCc , εc )/= 0. To o e come he p oblems desc ibed abo e, he usual p ac ice in his li e a u e has been o use da a on ano he coun y (o many o he coun ies) ha did no hos any Wo ld Cup du ing he yea s be o e o a e he coun y cu en ly hos ing he Wo ld Cup. These coun ies would hen be used as coun e ac uals o he coun y being analyzed and he pa ame e o in e es would be iden i ied ia a di e ence-in-di e ences (DID) se up. This s a egy would emo e bias ha migh esul om pe manen di e ences be ween he coun y hos ing he Wo ld Cup and o he coun ies used as coun e ac uals, as well as bias om compa ison o e ime in he coun y ha had he Wo ld Cup ha could be he esul o ime ends un ela ed o he Wo ld Cup i sel (Imbens and Woold idge, 2009). In his case, he equa ion o be es ima ed is gi en by Yc = α0+ α1∗ WCc + ΘXc + λc+ λ + μc (2) whe e Xc is a ec o o con ols, and λcand λ a e, espec i ely, coun y and yea ixed e ec s o con ol o coun y ime-in a ian unobse able cha ac e is ics and o yea ly di e ences be ween he ou come o in e es . The pa ame e o in e es , α1, equals he a e age gain o e ime in he coun ies no hos ing a Wo ld Cup minus he a e age gain o e ime in he coun y hos ing he Wo ld Cup. One main hypo hesis equi ed o he alidi y o his app oach in iden i ying he Wo ld Cup e ec , is ha bo h ea ed and con ol coun ies mus ha e exac ly he same ime end in he absence o he Wo ld Cup, and i is no clea why his should be he case. I , o example, he coun ies no hos ing 334 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 a Wo ld Cup ha e di e en ends compa ed o he coun y hos ing he Wo ld Cup, he esea che will be unable o di e en ia e be ween he Wo ld Cup e ec and he end di e ence. This sho coming is exac ly wha we aim o o e come in he p esen pape by using he syn he ic con ol me hod o cons uc a combina ion o coun ies ha bes desc ibes p e- ea men a iables o he coun y hos ing he Wo ld Cup, i.e., his a i icially cons uc ed g oup is simila o he ea ed coun y in he p e- ea men pe iods han any o he con ol coun y on hei own. 2.1. The syn he ic con ol me hod (SCM) In his sec ion we desc ibe he syn he ic con ol me hod (SCM) de eloped by Abadie and Ga deazabal (2003) and ex ended in Abadie e al. (2010). We also discuss i s ad an ages and limi a ions when compa ed o o he me hodologies used in he li e a u e, paying pa icula a en ion o DID s a egies. Suppose he e a e J + 1 egions/coun ies and ha only he i s egion is exposed o he policy change ( he coun y hos ing he Wo ld Cup), so ha he e a e J emaining egions as po en ial con ols (all o he coun ies no hos ing Wo ld Cups in he pe iod nea he one being analyzed). Le YN c be he ou come ha would be obse ed o egion c a ime in he absence o he in e en ion, o uni s c = 1, . . ., J + 1, and ime pe iods = 1, . . ., T. Le YI c be he ou come ha would be obse ed o uni c a ime i uni c is exposed o he in e en ion in pe iods T0+ 1 o T, whe e T0is he numbe o p e-in e en ion pe iods such ha 1 ≤ T0< T. I is assumed ha he in e en ion has no e ec on he ou come o in e es be o e he implemen a ion pe iod, such ha o ∈ 1, . . ., T0and all c ∈ 1, . . ., N we ha e ha YI c = YN i . Now le αc = YI c − YN c he e ec o he in e en ion o uni c a ime , and le Dc be an indica o ha akes alue one i uni c is exposed o he in e en ion a ime , and ze o o he wise. In his case, he obse ed ou come o uni c a ime is gi en by Yc = YN c − αc Dc . Fo egion one, which is he only egion exposed o he in e en ion a e pe iod T0, i ollows ha Dc = 1 o > T0and ze o o he wise. Ou objec i e is o es ima e (α1T0 + 1), which is gi en by α1 = YI 1 − YN 1 = Y1 − YN 1 . The p oblem in es ima ing α’s in his case is ha YN c is ne e obse ed o he ea ed egion once > T0. Thus, one mus es ima e i s alue. To see how a con ol g oup migh be ob ained om he se o un ea ed egions, suppose as in Abadie e al. (2010) ha YN c is gi en by he ollowing model YN c = δ + θ Zc+ λ μc+ εc (3) whe e δ is an unknown common ac o wi h cons an ac o loadings ac oss uni s, Zcis a ec o o obse ed co a ia es (no a ec ed by he in e en ion), θ is a ec o o unknown pa ame e s, λ is a ec o o unobse ed common ac o s, μcis an ec o o unknown ac o loadings, and he e o e ms εc a e unobse ed ansi o y shocks a he egion le el wi h ze o mean. Now conside a(J × 1) ec o o weigh s W =(w2, . . ., wJ+1)such ha wj≥ 0 o j = 2, . . ., J + 1 and w2+ . . . + wJ + 1= 1. Each alue ha W migh ake ep esen s a syn he ic con ol g oup o egion one. Fo exam- ple, i w2= 1 and wj= 0 o 3, . . ., J + 1, hen egion 2 wo ks as con ol o egion one ( he ea ed one). I , on he o he hand, one se s a subse J⊂ J o ha e equal weigh s, such ha w j=1 J o j ∈ J and 0 o he wise, he compa ison would be be ween he ea ed egion and he a e age o all o he egions ha belong o he g oup J. Using W as weigh s o cons uc a weigh ed a e age o Eq. (3), one ob ains he ollowing exp ession j+1  J=2 wjYj = δ + θ j+1  J=2 wjZj+ λ j+1  J=2 wjμc+ j+1  J=2 wjεj (4) J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 335 I one assumes ha exis s weigh s w∗ 2, . . .w∗ J+1such ha he ollowing holds, j+1 J=2wjYj = Y11, . . ., j+1 J=2w∗ jYjT0= Y1T0and j+1 J=2w∗ jZj= Z1 hen Abadie e al. (2010) p o e ha he ollowing equa ion is ue YN 1 − j+1  J=2 w∗ jYj = j+1  J=2 wj T0  s=1 λ T0  n=1 λ nλn−1 λ sεjs − ε1s− j+1  J=2 w∗ jεj − ε1 (5) and ha i s igh hand side will be close o ze o i he numbe o p e-in e en ion pe iods is la ge ela i e o he scale o he ansi o y shocks. This implies ha YN 1 =j+1 J=2w∗ jYj which sugges s he ollowing es ima o o he α ec o : ˆ α = Y1 − j+1  J=2 w∗ jYj (6) To ob ain he ec o o op imal weigh s W, le X1=Z 1, Y11, . . ., Y1T0be a ec o o p e-in e en ion cha - ac e is ics o he ea ed egion an X0be a ma ix ha con ains he same a iables o he un ea ed egions, such ha he j h column o X0is Z j, Yij , . . ., YjT0. Then W* is chosen o minimize he dis ance X1− X0WV= (X1− X0W)V(X1− X0W)be ween X1and X0W subjec o wj≥ 0 and w2+ . . . + wJ+1 = 1, whe e V is symme ic and posi i e semide ini e ma ix chosen in a way ha he esul ing syn he ic con ol egion app oxima es he ajec o y o he ou come a iable o he a ec ed egion in he p e-in e en ion pe iods. The model desc ibed abo e has se e al ad an ages when compa ed o o he app oaches used in he li e a u e. As poin ed ou by Billmeie and Nannicini (2013), he model is anspa en , gi en he weigh s w∗ 2, . . ., w∗ J+1iden i y he egions ha a e used o cons uc coun e ac uals o he ea ed egion, and he model is lexible, as he se o po en ial con ol egions can be app op ia ely es ic ed o make he compa isons sensible. Also, he model elaxes he assump ion ha con ounding ac o s a e ime in a ian ( ixed e ec s) o sha e a common end (di e ences-in-di e ences), gi en he e ec o unobse able con ounding ac o s is allowed o a y wi h ime. On he o he hand, his app oach has he limi a ion ha i does no allow one o assess he signi icance o he esul s using s anda d in e en ial echniques, gi en he numbe o un ea ed egions and he numbe o pe iods conside ed a e small. Abadie e al. (2010) sugges ha in e ence should be ca ied ou by implemen ing placebo expe imen s. In his case, in e ence is based on compa isons be ween he magni ude o he gaps gene a ed by he placebo s udies and he magni ude o he gap gene a ed o he ea ed s a e. Thus, i he gap es ima ed o he ea ed s a e is la ge compa ed o he gap es ima ed o he placebo expe imen s, hen he analysis would sugges ha he ea men had an e ec on he ou come o in e es and is no d i en by chance. 3. Da a We use coun y-le el da a co e ing mos FIFA Wo ld Cups occu ing in he pe iod be ween 1978 (A gen ina) and 2006 (Ge many).1Ou da a comes om h ee di e en sou ces. The main a iable o in e es (GDP pe capi a) and h ee o he co a ia es included in he ec o o p e-in e en ion cha ac e is ics (popula ion, in es men (g oss o al in es men /GDP), go e nmen inal consump ion expendi u e (% o GDP) and consume spending (% o GDP)), we e ob ained om he Penn Wo ld Tables 7.1. Addi ionally, we ob ained da a om Ma shall e al. (2014) and om Ba o and Lee (2010) on a poli ics index and on seconda y school en ollmen , espec i ely, o use as addi ional co a ia es in he ec o o p e-in e en ion cha ac e is ics.2We should emphasize ha , ollowing Abadie e al. (2010), we augmen ed 1Selec ion o Wo ld Cups o be analyzed was based on da a a ailabili y. 2We hank an anonymous e e ee o sugges ing hese addi ional co a ia es. 336 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 Table 2 Da a desc ip ion and sou ces. Va iable Desc ip ion Sou ce RGDPL Real g oss domes ic p oduc pe capi a in PPP 2005 dolla s Penn Wo ld Table 7.1 ( gdpl) Pop. G o. Popula ion g ow h Penn Wo ld Table 7.1 (POP) In ./GDP In es men sha e o PPP con e ed GDP pe capi a a 2005 cons an p ices Penn Wo ld Table 7.1 (ki) Go ./GDP Go e nmen consump ion sha e o PPP con e ed GDP pe capi a a 2005 cons an p ices Penn Wo ld Table 7.1 (kg) Con./GDP Consump ion sha e o PPP con e ed GDP pe capi a a 2005 cons an p ices Penn Wo ld Table 7.1 (kc) Sec.En oll Senconda y en ollmen in housands Ba o and Lee (2010) Poli ics Index om −10 (Au oc acy) o 10 (Full Democ acy) Ma shall e al. (2014) Table 3 Desc ip i e s a is ics. G oup o coun ies Va iables Mean S.D. Min Max Median All coun ies RGDPC 8.45 11.01 0.16 136.31 4.03 Pop. G o. 0.02 0.04 −0.56 3.58 0.02 In ./GDP 22.93 11.24 −11.50 93.64 21.68 Go ./GDP 11.86 8.97 0.31 67.19 8.89 Con./GDP 70.67 19.18 3.56 219.12 70.86 Sec.En oll 5.33 27.88 0.00 492.74 0.47 Poli ics 0.25 7.47 −10.00 10.00 −1.00 OECDaRGDPC 19.17 11.79 1.24 80.22 17.27 Pop. G o. 0.01 0.10 −0.02 3.58 0.01 In ./GDP 24.25 8.12 1.75 80.39 23.10 Go ./GDP 8.07 3.71 1.66 28.40 7.55 Con./GDP 66.69 9.59 29.27 96.05 67.62 Sec.En oll 2.75 3.70 0.00 19.65 1.28 Poli ics 6.79 6.33 −10.00 10.00 10.00 La in Ame ica RGDPC 4.97 2.53 1.17 12.53 4.57 Pop. G o. 0.02 0.01 −0.10 0.05 0.02 In ./GDP 19.96 7.73 −11.50 52.13 19.52 Go ./GDP 9.72 6.28 1.18 49.37 8.10 Con./GDP 74.21 10.60 29.22 110.98 75.51 Sec. En oll 1.75 4.22 0.01 44.96 0.49 Poli ics 1.64 6.69 −10.00 10.00 5.00 Sou ce: esea ch da a. aI also includes Taiwan and Singapo e, while excludes Czech Republic, Es onia, Iceland, Is ael, Slo akia, Slo enia, besides he coun ies ha hos ed he Wo ld Cup. ou ec o o p e-in e en ion cha ac e is ics o allow o he inclusion o lagged GPD pe capi a o imp o e p e- in e en ion adjus men . Tables 2 and 3 p o ides a b ie explana ion and desc ip i e s a is ics o all he a iables used in ou analysis. As we b ie ly explain abo e, he e a e wo possible de ini ions o he ea men assignmen pe iod (T0). One possibili y is o use he yea in which he wo ld cup eally occu ed. In his case, we would conside he yea s p esen ed in column 1 o Table 1. The second possibili y is o use he yea in which FIFA announced he hos ing coun y o he ollowing e en (p esen ed in column 2 o Table 1). Fo his pape , we decided o use in ou main speci ica ion he yea o announcemen , because mos in es men s in in as uc u e (new cons uc ion p ojec s like s adiums, e c.) happen be o e he ac ual cup happens. We do, howe e , conside using he yea in which he Wo ld Cup is ealized. As i will be shown in he nex sec ion, quan i a i e esul s a e qui e simila ega dless o wha de ini ion is used. Ano he impo an poin ega ding he applica ion o he syn he ic con ol me hod is he choice o coun ies ha will be included as po en ial con ol uni s o he ea ed coun ies. In ha sense, a i s limi a ion comes di ec ly om he a ailable da a. Al hough Penn Wo ld Tables con ain da a on 188 coun ies since 1950, o he da a sou ces ha e una ailable da a o some coun ies ( o example, s a ing a e 1950) o a e incomple e, such as Ba o and Lee’s J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 337 (2010) da a, which a e only a ailable o e i e-yea in e als.3Also, he s a ing yea T0is di e en o each o he coun ies analyzed, he e o e, we mus exclude om he lis o po en ial con ol uni s ecen hos ing coun ies which migh s ill be unde he e ec o he e en . Ou inal se o po en ial con ols, he e o e, a y wi hin hos ing coun ies. Fo ins ance, Ge many 2006 has mo e han 70 po en ial con ols while A gen ina 1978 has only 34.4 In Table 4 we p esen summa y s a is ics o he 9 coun ies ha hos ed a Wo ld Cup in he pe iod analyzed. No e ha o each coun y we p esen wo se s o s a is ics, labeled as 1 and 2. Label 1 e e s o he case in which ea men assignmen is based on he da e he hos ing coun y was announced, while label 2 e e s o he case in which ea men assignmen is he yea he e en occu ed. Fo example, A gen ina be o e announcemen (A gen ina 1) had an a e age g oss o al in es men /GDP o 22.31, while A gen ina be o e he Wo ld Cup occu ed (A gen ina 2) had an a e age g oss o al in es men /GDP o 23.91. 4. Resul s We s a by p esen ing esul s ha conside in ou dono pool all coun ies wi h a ailable da a (uncons ained dono pool). Then, acco ding o he usual p ac ice in his li e a u e, we es ic ou dono pool o be composed o coun ies wi h simila cha ac e is ics ( o ins ance, only OECD coun ies a e conside ed as dono s o USA and Eu opean coun ies). 4.1. Uncons ained dono pool Be o e looking a he es ima ed e ec s o he Wo ld Cups, le us i s look a he coun ies ha compose he syn he ic coun y o each o he ea ed coun ies and how hei p e- ea men cha ac e is ics compa e o he p e- ea men cha ac e is ics o he eal hos ing coun y. Table 5 p esen s he es ima ed weigh s o each coun y in he se o po en ial con ol coun ies.5Syn he ic A gen ina and syn he ic F ance, o example, a e con ex combina ions o many o he coun ies, while syn he ic Ge many is composed o only Aus ia, Belgium, Bulga ia and Swi ze land. Hence, as we highligh abo e, he model is anspa en as he weigh s clea ly iden i y he coun ies ha a e used o cons uc he coun e ac uals. In Table 4 we p o ide he nume ical compa ison by explana o y a iable be ween each ea ed coun y and he cons uc ed syn he ic con ol. Again, no e ha we ha e wo syn he ic con ols o wo di e en coun ies in each case s udy. Also, he ma ix V was chosen o minimize he mean squa ed p edic ion e o p oduced by he weigh s W∗(V) du ing he alida ion pe iod (eigh yea s be o e ea men yea , T0). A i s poin we look a is how di e en a e coun ies and hei syn he ic con ol cha ac e is ics when using label 1 (announcemen yea ) o label 2 (Wo ld Cup yea ). In mos coun ies, he e a e signi ican di e ences be ween he wo. Syn he ic A gen ina 1 has an a e age in es men /GDP a io o 21.11, while o Syn he ic A gen ina 2 his numbe equals 22.83. A gene al conclusion, howe e , is ha he syn he ic coun ies seem o p o ide a be e con ol g oup han only compa ing he ea ed coun y wi h he a e age cha ac e is ics o all o he coun ies in he dono pool o wi h a single coun y. This ad an age will become clea e in he g aphs below. As s a ed abo e, he con ex se o weigh s ha composes he syn he ic con ol uni is chosen by he minimiza ion o he ec o s V and W(V). Reliable causal syn he ic con ol es ima o s equi e ha he coun e ac ual uni esemble he main cha ac e is ics o he ea ed coun y in he pe iod be o e he in e en ion. In pa icula , he choice o coun ies ha compose he syn he ic uni s highligh an impo an ea u e o he SCM since he ea ed and syn he ic uni s a e easonably compa able, by excluding hose uni s ha do no p esen simila cha ac e is ics in he p e- ea men pe iod. The g aph on he le -hand side o Figs. 1 and 2 ep esen he ime se ies o he ou come a iable, eal GDP pe capi a, o he ea ed uni (solid line) and he syn he ic con ol uni (dashed line), bo h in he en i e p e- ea men pe iod and o en yea s a e he pe iod he e en occu ed. The do ed e ical line ep esen he yea FIFA announced he hos ing coun y. The compa ison be ween he solid and dashed line be o e ea men shows he quali y o adjus men in he ime se ies o he ou come a iable o he coun y hos ing he Wo ld Cup and he ime se ies o he ou come 3Da a o in-be ween yea s we e ob ained ia in e pola ion, as is common in he li e a u e. 4We emphasize ha he lis o po en ial con ols do no include any o he coun ies ha hos ed a Wo ld Cup du ing he sample we analyze. 5In his able we epo only coun ies ha ecei ed weigh s la ge han 0.005. Fo ins ance, he syn he ic con ol o Ge many is composed o 78 coun ies, mos wi h e y small weigh s. Fo he sake o space limi a ion, we only epo hose wi h ep esen a i e weigh s. 344 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 a e no clea jumps in he ou come a iable be o e announcemen . The e o e, ou esul s appea no o su e om p e- ea men mac oeconomic decisions. 4.2. Cons ained dono pool The esul s p esen ed on he p e ious sec ion conside ed all coun ies wi h a ailable da a as po en ial dono s o each ea ed coun y conside ed in he analysis. Howe e , Abadie e al. (2010) and Billmeie and Nannicini (2013) d aw a en ion ha his huge and e y di e se pool could esul in o e i ing o he syn he ic con ol, he e o e assigning la ge weigh s o coun ies ha di e subs an ially om he coun y one is ying o cons uc he syn he ic con ol. To check i ou esul s a e a ec ed by his so o p oblem, we ollow he li e a u e and es ima e all he syn he ic con ols using es ic ed pools o coun ies. Speci ically, o USA and Eu opean hos s, we es ic ed he pool o coun ies o hose belonging o he OECD. Fo Japan and Ko ea, gi en hey we e he only Asian coun ies belonging o he OECD, we es ic ed hei dono pool o OECD coun ies plus Taiwan and Singapo e. Finally, o he Wo ld Cups hos ed in La in Ame ica (A gen ina and Mexico) we es ic ed he dono pool o only La in Ame ican coun ies. Table 3 p esen s he desc ip i e s a is ics o hese subg oups. The esul s o he es ima ions wi h es ic ed pool o dono s a e shown in Figs. 4 and 5. Fo mos coun ies, esul s a e p ecisely he same as hose ob ained using he un es ic ed dono pool. The only wo excep ions a e Mexico and Spain. Fo he Mexican case, he e ec appea s o be sligh ly posi i e bu p e- ea men adjus men was so poo ha any conclusions de i ed om his empi ical exe cise a e weak. On he o he hand, o Spain he (doub ul) posi i e e ec ob ained conside ing he uncons ained case is now con incingly null, he e o e sus aining ou main conclusion ha wo ld cups a e no s a is ically associa ed o posi i e economic g ow h. 4.3. Robus ness checks Al hough o iginal placebo p ocedu es like RMSE p o ide use ul in o ma ion abou he likelihood o es ima ed e ec s unde he null hypo hesis o no e ec , plausible p- alues a ea no eadily a ailable due o he limi ed numbe o placebo uni s in some case s udies and he exis ence o uni -speci ic ansi o y shocks ha should be ze o mean bu no a e aged ou in SCM es ima o s. Ando (2015) p o ides wo es s a is ics aimed o access he signi icance o he in e en ion impac . We use such es s o access he e ec o he Wo d Cup on hos coun ies’ GDP pe capi a. The i s es consis s in compa ing he a e age ea men e ec s sizes ela i e o he dis ibu ion o a e age placebo e ec s o all placebo uni s, in all case s udies, whe e he o al numbe o a e age placebo e ec s a e inc eased by summing placebo ials in di e en case s udies. I consis s in es ima ing all ea men e ec s, ˆ αg , and placebo e ec s, ˆηgi , o all case s udies, wi h g indica ing one speci ic Wo ld Cup edi ion, i ep esen ing a con ol uni in dono pool and ∈{T0+1, . . ., T}. Then, we calcula e he a e age ea men e ec s o he ea ed uni s, ˆ αg, and o he placebo uni s, ˆηgi, o e ime o all case s udies in he pos -in e en ion pe iod. We use he dis ibu ion o ˆηgi o signi icance es s on ˆ αg, assuming ha hey ollow common dis ibu ion unde he null hypo hesis. The second es p oposed by Ando (2015) checks whe he he o e all a e age ea men e ec is su icien ly la ge in compa ison o he dis ibu ion o co esponding placebo es ima es ha a e calcula ed wi h andomly chosen placebo uni s. The i s s ep consis s in es ima ing he o e all ea men e ec , ˜ α =G g=1ˆ αg/G, and he o e all placebo e ec s, ˜ γ =G g=1ˆ γgi/G, by andomly choosing a con ol uni i in each g, whe e G is he numbe o Wo ld Cup edi ions. Then, we eplica e he p e ious s ep M imes o c ea e a dis ibu ion o o e all a e age placebo e ec s, ˜ γm, wi h m = 1, . . ., M. Finally, he dis ibu ion o ˜ γmis used o signi icance es s on ˜ α. The esul s o bo h es s a e p esen ed in Figs. 6 and 7, espec i ely. In Fig. 6 we obse e ha he dis ibu ion o he a e age e ec o all uni s ha e mean a ound ze o, indica ing ha he ea men assignmen has no sys ema ic e ec s on he con ol uni s. Rega ding he magni ude o he a e age e ec s, all he Wo ld Cup edi ions ail o ejec he null hypo hesis o no impac on GDP pe capi a. Fig. 7 p esen s he esul o he o e all a e age ea men e ec es . I J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 345 Fig. 4. Syn he ic con ol analysis ( es ic ed pool o dono s) o he 1994–2006 FIFA’s Wo ld Cup. 346 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 Fig. 5. Syn he ic con ol analysis ( es ic ed pool o dono s) o he 1978–1990 FIFA’s Wo ld Cup. J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 347 Fig. 6. Densi y o a e age placebo e ec s. No e: CDF(x) is he empi ical cumula i e dis ibu ion unc ion o a e age placebo e ec s. Fig. 7. Densi y o o e all placebo e ec s. No e: numbe o Mon e Ca lo epe i ions M = 100,000. CDF(x) is he empi ical cumula i e dis ibu ion unc ion o o e all a e age placebo e ec s. shows ha he o e all e ec is lowe han he h eshold line a CDF = 99,5% when we un M = 100,000 Mon e Ca lo simula ions. In his sense, i is unlike ha he magni ude o his o e all e ec is gene a ed by andom e o s. 5. Concluding ema ks The FIFA (Fédé a ion In e na ionale de Foo ball Associa ion) Socce Wo ld Cup anks among he h ee la ges e en s in he wo ld. I no only a ec he in lux o people/ ou ism in he hos coun y, bu also in ol es publicly inanced capi al imp o emen p ojec s ha a e unde aken o imp o e in as uc u e, such as he cons uc ion o new 348 J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 s adiums and he imp o emen o old ones, oad and ai po cons uc ion and imp o emen , among many o he s. The isks and cos s in ol ed in hos ing an e en o his magni ude a e la ge; howe e , in all p e ious Wo ld Cup edi ions he e has been la ge compe i ion o hos he e en . This, in pa , is due o he common belie ha hos coun ies will expe ience highe economic g ow h a es, educ ion on unemploymen a es, inc eases in ou is ic ac i i ies and go e nmen income, inc ease in capi al in low and an imp o emen o he image o he coun y wo ldwide. Cu en li e a u e on his subjec is sca ce and inconclusi e, specially conside ing he me hods used so a , which migh su e om se e e iden i ica ion p oblems. In his pape we mo e away om he me hods p e iously used o s udy he subjec and p opose o use he syn he ic con ol me hod (SCM), a echnique which gained popula i y ecen ly and is well sui ed o s udy he p oblem add essed in his pape . The main ad an age lies on he ac ha , unlike mos o he es ima o s used in he li e a u e o p og am e alua ion and specially in he li e a u e analyzing he e ec s o wo ld cups, he SCM deals in some deg ee wi h ime- a ying unobse able con ounde s. A second con ibu ion o ou pape ela es o he numbe o e en s conside ed in he analysis. Unlike mos pape s in he li e a u e which conside only one wo ld cup in hei analysis, ou pape expands p e ious esea ch and o e a se o empi ical case s udies o bes analyze he ela ionship be ween he e en s and he pa e n o income. We conside a o al o 8 e en s held in 9 coun ies, co e ing all Wo ld Cups occu ing in he pe iod be ween 1978 (A gen ina) and 2006 (Ge many). Ou empi ical indings show ha o he majo i y o he coun ies conside ed in he analysis (Ge many, Japan, Ko ea, F ance, Uni ed S a es, I aly, Mexico and A gen ina), he Wo ld Cup had a null o a nega i e e ec on income pe capi a. Wi h he excep ion o Spain and Mexico, in which he syn he ic con ol p esen ed a poo p e- ea men i wi h eal GDP ime se ies, all o he coun ies p esen ed a qui e good p e- ea men adjus men be ween eal and syn he ic con ol se ies. This is qui e com o ing because he in e en ial alue o his expe imen inc eases when he me hod deli e s a sui able coun e ac ual o he analysis. The gene al conclusion o he pape poin s in he di ec ion ha hos ing a Wo ld Cup leads o no economic bene i . We emphasize, howe e , ha ou pape looks only a GDP pe capi a. Hence, any o he bene i s ela ed o economic well-being o he popula ion, ade (see, o example, he wo k o Rose and Spiegel, 2011), o gains ela ed o he image o he coun y and u u e ou is ic ac i i ies a e no cap u ed in ou analysis. Howe e , despi e he po en ial posi i e bene i s in e ms o well being in hos ing he Wo ld Cup, ou pape sugges s ha coun ies unde economic u moil migh su e om hos ing a Wo ld cup, ins ead o boos ing hei economy, as mos belie e. Acknowledgemen s We a e indeb ed o wo anonymous e e ees o hei help ul commen s. Any emaining e o s a e ou esponsibili y only. This esea ch did no ecei e any speci ic g an om unding agencies in he public, comme cial, o no - o -p o i sec o s. Re e ences Abadie, A., Ga deazabal, J., 2003. The economic cos s o con lic : a case s udy o he Basque Coun y. Am. Econ. Re . 93, 113–132. Abadie, A., Diamond, A., Hainmuelle , J., 2010. Syn he ic con ol me hods o ompa a i e case s udies: es ima ing he e ec o Cali o nia’s obacco con ol p og am. J. Am. S a . Assoc. 104, 493–505. Ahle , G., 2001. The economic e ec s o he Socce Wo ld Cup 2006 in Ge many wi h ega d o di e en inancing. Econ. Sys . Res. 13, 109–127. Allme s, S., Maennig, W., 2009. Economic impac s o he FIFA socce wo ld cups in F ance 1998, Ge many 2006, and ou look o Sou h A ica 2010. Eas . Econ. J. 35, 500–519. Ando, M., 2015. D eams o u baniza ion: quan i a i e case s udies on he local impac s o nuclea powe acili ies using he syn he ic con ol me hod. J. U ban Econ. 85, 68–85. Baade, R.A., Ma heson, V.A., 2004. The ques o he cup: assessing he economic impac o he wo ld cup. Reg. S ud. 38, 343–354. Ba o, R.J., Lee, J-W., 2010. A new da a se o educa ional a ainmen in he wo ld, 1950–2010. NBER Wo king Pape 15902. Belo , M., Vandenbe ghe, C., 2009. G ade e en ion and educa ional a ainmen : exploi ing he 2001 Re o m by he F ench-Speaking Communi y o Belgium and Syn he ic Con ol Me hods. Discussion Pape 2009-22, mimeo, Ca holic Uni e si y o Lou ain, Lou ain. Billmeie , A., Nannicini, T., 2013. Assessing economic libe aliza ion episodes: a syn he ic con ol app oach. Re . Econ. S a . 95, 983–1001. Bohlmann, H.R., an Hee den, J.H., 2005. The Impac o Hos ing a Majo Spo E en on he Sou h A ican Economy. Uni e si y o P e o ia Wo king Pape 2005-09, Uni e si y o P e o ia, P e o ia. J.H.N. Viana e al. / EconomiA 19 (2018) 330–349 349 C espo, Sil io Guedes, 2014. Copa cus a R$ 8 bi em es ádios e R$ 16 bi em ou as ob as; al a acaba 18%. A ailable a h p://achadoseconomicos.blogos e a.uol.com.b /2014/07/14/copa-cus ou- -8-bi-em-es adios-e- -16-em-ou as-ob as- al a- e mina -18/. (Accessed 10 Janua y 2015). C omp on, John L., 1995. Economic impac analysis o spo s acili ies and e en s: ele en sou ces o misapplica ion. J. Spo Manage. 9, 14–35. Edwa ds, S., 1996. A Tale o wo C ises: Chile and Mexico. NBER Wo king Pape 5794. Hagn, F., Maennig, W., 2008. Employmen e ec s o he oo ball wo ld cup 1974 in Ge many. Labou Econ. 15, 1062–1075. Hagn, F., Maennig, W., 2009. La ge spo e en s and unemploymen : he case o he 2006 socce Wo ld Cup in Ge many. Appl. Econ. 41, 3295–3302. Holland, P., 1986. S a is ics and causal in e ence. J. Am. S a . Assoc. 81, 945–970. Imbens, G., Woold idge, J.M., 2009. Recen de elopmen s in he econome ics o p og am e alua ion. J. Econ. Li . 47, 5–86. Ma shall, M.G., Gu , T.R., Jagge s, K., 2014. Poli y IV P ojec . Poli ical Regime Cha ac e is ics and T ansi ions 1800–2013. h p://www.sys emicpeace.o g/insc da a.h ml. Pa u, G., Amo a, D., Cou inho, F., 2014 Cus o da Copa Equi ale a um mês de gas os com educac¸ão. A ailable a h p://www1. olha.uol.com.b / sp/pode /167243-cus o-da-copa-equi ale-a-um-mes-de-gas os-com-educacao.sh ml. (Accessed 10 Janua y 2015). Rose, A.K., Spiegel, M.M., 2011. The olympic e ec . Econ. J. 121, 652–677. Sampaio, B., 2014. Iden i ying pee s a es o anspo a ion policy analysis wi h an applica ion o New Yo k’s handheld cell phone ban. T anspo - me ica A: T ansp. Sci. 10, 1–14.