What role for aid for trade in (deep) PTA relations? Empirical evidence from gravity model estimations

S ende , F ede ik; Vogel, Tim Wo king Pape Wha ole o aid o ade in (deep) PTA ela ions? Empi ical e idence om g a i y model es ima ions IDOS Discussion Pape , No. 13/2023 P o ided in Coope a ion wi h: Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS), Bonn Sugges ed Ci a ion: S ende , F ede ik; Vogel, Tim (2023) : Wha ole o aid o ade in (deep) PTA ela ions? Empi ical e idence om g a i y model es ima ions, IDOS Discussion Pape , No. 13/2023, ISBN 978-3-96021-216-4, Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS), Bonn, h ps://doi.o g/10.23661/idp13.2023 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/279528 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/4.0/ Wha Role o Aid o T ade in (Deep) PTA Rela ions? Empi ical E idence om G a i y Model Es ima ions F ede ik S ende Tim Vogel IDOS DISCUSSION PAPER 13/2023 Wha ole o aid o ade in (deep) PTA ela ions? Empi ical e idence om g a i y model es ima ions F ede ik S ende & Tim Vogel Bonn 2023 D F ede ik S ende is a senio esea che in he “T ans o ma ion o Economic and Social Sys ems” p og amme a he Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS). Email: ede ik.s ende @idos- esea ch.de D Tim Vogel is a esea che in he “T ans o ma ion o Economic and Social Sys ems” p og amme a he Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS). Email: im. ogel@idos- esea ch.de Published wi h inancial suppo om he Fede al Minis y o Economic Coope a ion and De elopmen (BMZ). Sugges ed ci a ion: S ende , F., & Vogel, T. (2023). Wha ole o aid o ade in (deep) PTA ela ions? Empi ical e idence om g a i y model es ima ions (IDOS Discussion Pape 13/2023). Bonn: Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS). h ps://doi.o g/10.23661/idp13.2023 Disclaime : The iews exp essed in his pape a e hose o he au ho (s) and do no necessa ily e lec he iews o policies o he Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS). Excep o he wise no ed, his publica ion is licensed unde C ea i e Commons A ibu ion (CC BY 4.0). You a e ee o copy, communica e and adap his wo k, as long as you a ibu e he Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS) gGmbH and he au ho (s). IDOS Discussion Pape / Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS) gGmbH ISSN 2751-4439 (P in ) ISSN 2751-4447 (Online) ISBN 978-3-96021-216-4 (P in ) DOI: h ps://doi.o g/10.23661/idp13.2023 © Ge man Ins i u e o De elopmen and Sus ainabili y (IDOS) gGmbH Tulpen eld 6, 53113 Bonn Email: publica ions@idos- esea ch.de h ps://www.idos- esea ch.de P in ed on eco- iendly, ce i ied pape . IDOS Discussion Pape 13/2023 III Abs ac While ecen p e e en ial ade ag eemen s (PTAs) co e an inc easingly b oad ange o policy a eas beyond hei adi ional compe ence o educing bila e al a i s, li le is known abou he implica ions o his new emphasis on in e ac ions wi h o he ade- ela ed policy measu es. We app oach his gap by examining he e ec i eness o bila e al aid o ade (A T) in deep No h– Sou h PTA ela ions. To his end, we use a s uc u al g a i y model o bila e al panel da a o 29 OECD DAC coun ies and 144 de eloping coun ies om 2002–2015 and ind ha he ma ginal e ec o A T dec eases as he policy a eas o a PTA expand. Fu he in es iga ion o he unde lying mechanisms sugges s ha he obse ed ade-o be ween PTA dep h and A T e ec i eness may be due o compliance wi h he non- a i p o isions con ained in deep PTAs. We ind wo lines o easoning plausible. Fi s , compliance e o s appea o consume la ge ac ions o A T and hus educe A T a ailable o po en ially mo e e ec i e p ojec s, as we do no obse e an alignmen o A T in deep PTAs. Second, since we also obse e he e ogenei y in in e ac ions ac oss dono s, depending on hei speci ic p ojec po olios, A T p o ided by high- income PTA pa ne s could well be used o edi ec expo s o hi d coun ies wi h compa a i ely ewe bila e al obliga ions. Dono coun ies should he e o e ca e ully weigh compliance cos s o de eloping coun ies agains he non- ade bene i s o common deep PTAs, and accu a ely iden i y inancial and echnical assis ance needs oge he wi h hei PTA pa ne s. Keywo ds: Aid o ade e ec i eness, g a i y model o ade, non- a i p o isions, p e e en ial ade ag eemen s, PTA dep h, Sou h-No h ade JEL codes: F14, F15, F35 IDOS Discussion Pape 13/2023 IV Acknowledgemen s We g a e ully acknowledge aluable commen s and sugges ions om Cla a B andi, Ma hias Busse, Niels Keijze , A min on Schille , Yabibal Walle, and pa icipan s o he Eu opean T ade S udy G oup (ETSG) 2023 con e ence held in Su ey. The iews exp essed in his pape a e ou own. Any emaining e o s a e he au ho s’ esponsibili y. IDOS Discussion Pape 13/2023 V Con en s Abs ac III Acknowledgemen s IV Abb e ia ions VI 1 In oduc ion 1 2 Rela ed li e a u e and analy ical conside a ions 3 3 Empi ical s a egy 5 3.1 Da a 5 3.2 Es ima ion model 8 4 Main esul s and discussion 9 4.1 Baseline es ima ion esul s 9 4.2 Robus ness checks 12 5 Explo a o y ex ensions 14 6 Conclusion 21 Re e ences 22 Appendix 26 Table A1: Dono coun ies in he sample 26 Table A2: Recipien coun ies in he sample 26 Figu es Figu e 1: Dis ibu ion o A T disbu semen s by PTA dep h 8 Tables Table 1: PTA dep h and p o isions co e age in sample PTAs 7 Table 2: Baseline es ima ion esul s 10 Table 3: Es ima ion esul s o a ying lags o A T & con ol unc ion app oach 12 Table 4: Es ima ion esul s o a e aged A T da a & al e na i e PTA dep h indica o 13 Table 5: Es ima ion esul s o di e en p e- and pos -en y in o o ce pe iods 15 Table 6: Es ima ion esul s o dono expo s 16 Table 7: Es ima ion esul s o commodi y clus e s 17 Table 8: Es ima ion esul s con olling o o he PTA p o isions 18 Table 9: Es ima ion esul s o dono -speci ic in e ac ions 20 IDOS Discussion Pape 13/2023 VI Abb e ia ions ACP A ican, Ca ibbean and Paci ic A T aid o ade CRS C edi Repo ing Sys em [da abase] DESTA Design o T ade Ag eemen s [da abase] DAC De elopmen Assis ance Commi ee ESR economic and social igh s EPA economic pa ne ship ag eemen EP en i onmen al p o ec ion EU Eu opean Union FDI o eign di ec in es men IPRs in ellec ual p ope y igh s ODA o icial de elopmen assis ance OECD O ganisa ion o Economic Co-ope a ion and De elopmen PPML poisson pseudo maximum likelihood PTA p e e en ial ade ag eemen SPS sani a y and phy osani a y SITC S anda d In e na ional T ade Classi ica ion TBT echnical ba ie s o ade TPP T ans-Paci ic Pa ne ship WTO Wo ld T ade O ganiza ion IDOS Discussion Pape 13/2023 1 1 In oduc ion The 2030 Agenda o Sus ainable De elopmen emphasises he ele ance o in e na ional ade o p omo ing inclusi e economic g ow h and educing po e y. Despi e his pe cep ion, apping in o in e na ional ade channels is no s aigh o wa d. Indeed, many di e en app oaches o (be e ) in eg a ing de eloping coun ies in o he global economy ha e been applied, he mos p ominen being he emo al o isible ade ba ie s. A he mul ila e al le el, de eloping coun y membe s o he Wo ld T ade O ganiza ion (WTO) a e g an ed special and di e en ia ed ea men compa ed o he Mos Fa ou ed Na ion (MFN) clause, and commonly quali y o unila e al a i p e e ences o hei expo s o majo ma ke s. In pa allel, he e is a s eady inc ease o de eloping coun ies’ pa icipa ion in ecip ocal p e e en ial ade ag eemen s (PTAs), bo h among hemsel es and wi h high-income coun ies. A he end o 2022, he WTO eco ded a o al o 355 PTAs in o ce, he majo i y o which we e signed be ween coun ies a di e en s ages o economic de elopmen . No h–Sou h PTAs ypically elimina e he unce ain y o de eloping coun ies abou hei quali ica ion o non- ecip ocal a i p e e ences and he eby help o c ea e a mo e a ou able en i onmen o hei expo s (He z & Wagne , 2011; Mange & Shadlen, 2014; Limão, 2016), bu hey also en ail inc eased compe i i e p essu e ac oss he boa d om No he n in eg a ion pa ne s. Mo eo e , despi e he posi i e e ec s o PTAs on educing a i s (e.g. Hayakawa & Kimu a, 2015; S ende , 2019), i was ecognised, no leas h ough he 2005 WTO Aid o T ade (A T) ini ia i e, ha ma ke access alone is o en insu icien o s imula e engagemen in in e na ional ade o , mo e speci ically, o adequa ely exploi he ull po en ial o ade a angemen s. Ins ead, he expo s o de eloping coun ies a e equen ly cons ained by a a ie y o supply-side limi a ions, including he lack o p oduc i e capaci ies, in as uc u e de ici s, and non-compliance wi h s anda ds and egula ion in pa icula ly No he n des ina ion ma ke s (Suwa-Eisenmann & Ve die , 2007). A T aims o help de eloping coun ies de elop he ade capaci y and in as uc u e needed o eap he bene i s o ade libe alisa ion. I is a componen o b oade o icial de elopmen assis ance (ODA) ha includes g an s and concessional loans speci ically ea ma ked o ade- ela ed ini ia i es and e o s (WTO, 2023). While e idence on A T mobilisa ion in he wake o he WTO ini ia i e emains mixed (e.g. Gambe oni & New a me , 2014; Lee, Pa k, & Shin, 2015; Gnangnon, 2019a), de elopmen assis ance dedica ed o add essing hese behind- he-bo de ade cos s is also inc easingly inco po a ed in o PTAs. A p ominen example o linking economic in eg a ion and ade- ela ed assis ance is ha o he Economic Pa ne ship Ag eemen s (EPAs) be ween he Eu opean Union (EU) and a numbe o A ican, Ca ibbean and Paci ic (ACP) s a es. These ag eemen s a e buil on asymme ic ecip ocal a i libe alisa ion bu , acco ding o hei legal ex s, come wi h he p omise o accompany (mo e) A T (see also Holden, 2014; Kilolo, 2018). An addi ional laye o ade- ela ed assis ance in No h–Sou h PTAs a ises om he ecen end o expand he scope o PTAs beyond he me e emo al o a i s and o selec om a menu o addi ional policy a eas ha o en in e e e hea ily wi h na ional policy sphe es (Shadlen, 2005; Ma oo, Rocha, & Ru a, 2020). Mode n PTAs equen ly a ge ha monisa ion o ade- ela ed s anda ds among in eg a ion pa ne s, demand be e p o ec ion o in ellec ual p ope y, es ic domes ic policy-making au onomy in he a eas o compe i ion and in es men policy, o igh en en i onmen al law and egula ion in he ields o ag icul u e o ex ac i e indus ies (Dü , Baccini, & Elsig, 2014; Ho mann, Osnago, & Ru a, 2019). Such “deep” PTAs hus inc easingly aim o achie e goals beyond he p omo ion o ade. Howe e , he new emphasis on hese non- adi ional dimensions in PTAs is no only la gely d i en by high-income coun ies, bu hese coun ies also se he scene in PTA nego ia ions wi h de eloping coun ies (e.g. Allee & Elsig, 2019; Peacock, Milewicz, & Snidal, 2019). De eloping coun ies pa icipa ing in No h–Sou h PTAs he e o e inc easingly ace he need o implemen e o m equi emen s and o comply wi h a g owing and di e se se o PTA p o isions. In his con ex , ade- ela ed assis ance can help IDOS Discussion Pape 13/2023 8 Figu e 1: Dis ibu ion o A T disbu semen s by PTA dep h Sou ce: Own illus a ion based on sample used o empi ical analysis. See sec ion 3.1 o de ails. Epanechniko ke nel used. 3.2 Es ima ion model To es he combined e ec o A T and PTA dep h on bila e al ade, we use Poisson pseudo maximum likelihood (PPML), p oposed by San os Sil a and Ten ey o (2006), and es ima e an augmen ed g a i y equa ion o he ollowing baseline o m:6 𝑀𝑀 𝑖𝑖𝑖𝑖𝑖𝑖 =exp�𝛽𝛽 ln 𝐴𝐴𝐴𝐴𝐴𝐴 𝑖𝑖𝑖𝑖𝑖𝑖 +𝛾𝛾 𝑃𝑃𝐴𝐴𝐴𝐴 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ 𝑖𝑖𝑖𝑖𝑖𝑖 +𝛿𝛿ln 𝐴𝐴𝐴𝐴𝐴𝐴 𝑖𝑖𝑖𝑖𝑖𝑖 ∗𝑃𝑃𝐴𝐴𝐴𝐴 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ 𝑖𝑖𝑖𝑖𝑖𝑖 +𝜋𝜋 𝑖𝑖𝑖𝑖 +𝜂𝜂 𝑖𝑖𝑖𝑖 +𝜇𝜇 𝑖𝑖𝑖𝑖 � +𝜀𝜀 𝑖𝑖𝑖𝑖𝑖𝑖 (1) whe e 𝑀𝑀𝑖𝑖𝑖𝑖𝑖𝑖 deno es o al impo s o OECD DAC coun y 𝑖𝑖 om de eloping coun y 𝑗𝑗 in pe iod 𝑑𝑑. We ely on dono impo s ins ead o ecipien expo s because he o me a e ypically be e epo ed. On he igh -hand-side, A T disbu semen s a e inco po a ed in loga i hmic o m o p ima ily educe skewness in he A T dis ibu ion.7 Bo h a iables a e exp essed in cu en US dolla s. 6 PPML yields consis en es ima es unde esidual he e oscedas ici y and is compa ible wi h he exis ence o ze os in ade low da a. 7 In ou sample, abou 70 pe cen cons i u e ze o-A T obse a ions. To a oid losing hese obse a ions in he es ima ion o equa ion (1) by aking hei loga i hm, we applied he in e se hype bolic sine ans o ma ion o ou A T da a, o mally exp essed as ln (𝑥𝑥+�(𝑥𝑥2+ 1)), wi h 𝑥𝑥 being he alue o A T disbu semen s. We also inco po a ed a no-A T dummy a iable in equa ion (1), ollowing Wagne (2003). Fo bo h o hese al e na i e app oaches, es ima ion esul s a e, howe e , nea ly iden ical o hose epo ed below and a ailable upon eques . IDOS Discussion Pape 13/2023 9 𝑃𝑃𝐴𝐴𝐴𝐴 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ𝑖𝑖𝑖𝑖𝑖𝑖 is a coun a iable based on DESTA’s PTA dep h index, as desc ibed abo e. I coun ies 𝑖𝑖 and 𝑗𝑗 ha e a common PTA in pe iod 𝑑𝑑, i anges om wo o se en, depending on he scope o key policy a eas included in he PTA, and ze o o he wise.8 No e ha he join inco po a ion o A T and he PTA dep h a iable inciden ally o se s a po en ial omi ed a iables bias in p e ious g a i y applica ions in he li e a u e due o he ac ha A T could de e mine he willingness o ecipien s o accep non-economic p o isions in hei PTAs wi h dono s (see Hoekman e al., 2023). Ou p ima y in e es is in he in e ac ion o he A T and PTA dep h a iable. Recalling om ou discussion in Sec ion 2, a p io i i s coe icien , 𝛿𝛿, could be bo h posi i ely and nega i ely signed. Following he in e ac ion speci ica ion be ween bila e al o e all ODA and economic in eg a ion in Ma ínez-Za zoso (2019), we addi ionally use a plain dummy a iable, signalling he exis ence o a PTA be ween coun ies 𝑖𝑖 and 𝑗𝑗 in pe iod 𝑑𝑑, ze o o he wise, in exchange o he PTA dep h a iable. Gi en he wide a ia ion in PTA dep h in he sample (see Table 1), we delibe a ely do no include he dep h o PTAs as explici ca ego ies in equa ion (1), as his would likely cap u e he e ec s o speci ic PTAs a he han allow iden i ica ion o a gene alised ela ionship. Simila ly, because o s ong mul icollinea i y among he se en PTA policy a eas, i is no possible o p ope ly isola e he e ec o each p o ision sepa a ely.9 As sugges ed by Baie and Be gs and (2007), we include ime-in a ian coun y-pai (𝜋𝜋𝑖𝑖𝑖𝑖), and wo-way impo e -yea and expo e -yea ixed e ec s (𝜂𝜂𝑖𝑖𝑖𝑖 and 𝜇𝜇𝑖𝑖𝑖𝑖, espec i ely) in equa ion (1). The o me con ol o a numbe o s anda d g a i y a iables, such as he dis ance o cul u al and his o ical commonali ies be ween ading pa ne s, as well as he endogenei y o ade policies, including he o ma ion o PTAs. The la e wo ixed e ec s con ol o ime- a ying mul ila e al ade esis ance (Baldwin & Taglioni, 2007) bu also cap u e any coun y-yea speci ic supply o demand shock a ec ing bila e al ade pa e ns, including FDI lows. Mo e speci ically, he coun y-yea ixed e ec s addi ionally p oxy ading pa ne s’ in e es and ba gaining powe o shape he con en o PTAs. Due o using coun y-pai ixed e ec s, howe e , no e ha we only conside coun y pai s wi h ime- a ying PTA (dep h) obse a ions du ing he pe iod o analysis in ou es ima ions. 4 Main esul s and discussion 4.1 Baseline es ima ion esul s The esul s o ou baseline es ima ion a e p esen ed in Table 2. Columns (1)-(3) conside he yea s o signa u e o he PTA a iables and hei espec i e in e ac ions wi h A T, while columns (4)–(6) conside he yea s he PTA en e ed in o o ce. The a e age ime be ween signa u e and en y in o o ce o he PTAs in he sample is oughly 1.6 yea s. While some PTAs s a (p o isional) implemen a ion immedia ely a e signa u e (e.g. he China–New Zealand FTA o he CARIFORUM–EU EPA), o he PTAs did no en e in o o ce un il se e al yea s a e signa u e (e.g. he Uni ed S a es–Colombia FTA o he Bosnia and He zego ina–EU 8 The bila e al coun y sample used o empi ical analysis does no include any ecip ocal PTA wi h a dep h le el o one du ing he s udy pe iod (also see Sec ion 3). 9 Wi hou p ejudice o he esul s below, no wi hs anding hese conce ns, we an eg essions in which we included dummies o he se en PTA policy a eas and hei espec i e in e ac ions wi h A T sepa a ely. Howe e , all in e ac ions a e s a is ically signi ican when we include hem one by one, bu all a e insigni ican when we include hem join ly. In ou poin o iew, hen, he e is no he e ogenei y ac oss PTA policy a eas ha could be o in e es o deepe analysis. IDOS Discussion Pape 13/2023 10 S abilisa ion and Associa ion Ag eemen s), o did no ye en e in o o ce du ing he pe iod o ou analysis ( he EU–Uk aine Deep and Comp ehensi e FTA, he Japan–Mongolia EPA, and he Republic o Ko ea–Colombia FTA). While he e may occasionally be a need o e o m wi h espec o implemen a ion o compliance wi h he p o isions included in a PTA p io o i s en y in o o ce, we sugges ha bo h ha e implica ions, in pa icula once a PTA akes e ec . The e o e, we ea speci ica ions ha ake in o accoun he yea s in which he PTA en e s in o o ce as ou p e e ed model, e en hough he esul s o bo h ways o coding a e nea ly iden ical. Table 2: Baseline es ima ion esul s PTA signa u e PTA en y in o o ce VARIABLES (1) (2) (3) (4) (5) (6) ln A T 0.0146*** 0.0144*** 0.0143*** 0.0145*** 0.0145*** 0.0144*** (0.00508) (0.00497) (0.00497) (0.00512) (0.00501) (0.00500) PTA dummy 0.0599 -0.148 0.0777 0.0161 (0.0555) (0.209) (0.0544) (0.157) ln A T ∗ PTA dummy -0.0178** -0.0189** (0.00856) (0.00808) PTA dep h 0.0118 0.0362 0.0150 0.0123 (0.00918) (0.0337) (0.00980) (0.0284) ln A T ∗ PTA dep h -0.00347** -0.00342** -0.00380*** -0.00380*** (0.00152) (0.00152) (0.00147) (0.00147) Obse a ions 19,588 19,588 19,588 19,588 19,588 19,588 Pseudo R2 0.996 0.996 0.996 0.996 0.996 0.996 Coun y pai s 2,119 2,119 2,119 2,119 2,119 2,119 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses. As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. Ac oss columns, ou es ima ions hin a a posi i e and s a is ically highly signi ican gene al e ec o A T on he expo s om ecipien s o dono s.10 By con as , economic in eg a ion does no appea o in luence de eloping coun y expo s o No he n ma ke s pe se, a inding consis en wi h all o he speci ica ions in Table 2. Al hough su p ising, he insigni ican coe icien es ima e o bo h he PTA dummy and he PTA dep h a iable could e lec inc eased sec o al egula o y bu dens o de eloping coun ies unde deep PTAs due o compliance obliga ions wi h some o he non- a i p o isions included in he PTAs (compa e, e.g., Timini e al., 2022; Hoekman e al., 2023), bu could also be owing o he PTA obse a ions conside ed o es ima ion. Mo e speci ically, no only does ou sample include only OECD DAC dono s as impo e s, bu ou es ima ions conside only hose PTA ela ions wi h a con empo aneous posi i e A T-obse a ion. In o he wo ds, ou esul s o he PTA dummy and he PTA dep h a iable could su e om a sample selec ion bias in compa ison o p e ious indings in he li e a u e. Al e na i ely, he non-signi icance o he PTA dummy coe icien and PTA dep h coe icien es ima es could simply be due o he limi ed imp o emen in he a i ea men o 10 This inding is con i med by eg essions in which he in e ac ion e m is no used. IDOS Discussion Pape 13/2023 11 de eloping coun ies unde No h–Sou h PTAs, as o he p e e en ial a angemen s we e o en p e-exis ing. Tu ning o he in e ac ion be ween A T and economic in eg a ion, columns (1) and (4) i s epo es ima ion esul s o he in e ac ion be ween A T and he plain PTA dummy. In line wi h he indings by Ma ínez-Za zoso (2019) on o e all ODA, we ind ha he in e ac ion be ween A T and economic in eg a ion is nega i e and s a is ically signi ican a he i e-pe cen le el. When in es iga ing he nega i e in e ac ion mo e closely, columns (2)-(3) as well as (5)-(6) sugges ha he nega i e in e ac ion is d i en by he con en o a PTA, a he han by i s me e exis ence. Mo e speci ically, ins ead o being complemen a y wi h deep PTA p o isions o p omo e bila e al ade, we ind A T e ec i eness in No h–Sou h PTA ela ions o be condi ional on he ex en o he policy a eas included he ein. Gi en ou discussion in Sec ion 2, he nega i e in e ac ion be ween A T and he PTA dep h a iable could be due o compliance wi h he non- a i p o isions con ained in deep PTAs. In Sec ion 5, we explo e his conjec u e mo e deeply by b eaking down he esul s by he iming a ound PTA implemen a ion, he di ec ion o ade lows, economic sec o , and A T dono p o iles. Impo an ly, he nega i e in e ac ion does no mean ha bila e al A T hampe s ecipien expo s in deep PTA ela ions. Gi en he posi i e and s a is ically signi ican coe icien es ima e o uncondi ional A T e ec i eness, and he non-signi ican coe icien es ima e o he PTA dep h coe icien , ou in e p e a ion is ha he gene ally posi i e e ec s o A T, howe e , g adually diminish as he dep h o PTAs inc eases. The e o e, he o e all e ec o A T on ecipien expo s unde PTAs may emain posi i e, depending on he scope o he p o isions included in he PTA. We in en ionally do no compu e a h eshold PTA dep h le el a which he posi i e base e ec o A T is ully abso bed. This is due o he coun -da a na u e o he PTA dep h a iable, whe e an inc ease could imply he inclusion o any addi ional policy a ea, and his expansion is no subjec o any s ic o de ing (see Sec ion 3.1). The e o e, he composi ion o policy a eas can a y signi ican ly e en o PTAs wi hin he same dep h le el.11 Since he simul aneous inclusion o he PTA dummy and he PTA dep h a iable (columns 3 and 5) does no al e es ima ion esul s quali a i ely o quan i a i ely, we e ain om including bo h a iables simul aneously in u he es ima ions. Fo a mo e echnical conside a ion, omi ing he plain PTA dummy also a oids collinea i y p oblems wi h he PTA dep h a iable.12 Howe e , a i libe alisa ion e o s be ween PTA membe s may be co ela ed wi h he inclusion o o he non- a i PTA p o isions. While he plain PTA dummy picks up he a i libe alisa ion componen o PTAs (as a esidual) in he speci ica ion wi h he PTA dummy and he PTA dep h a iable, we he e o e also es ed speci ica ions ha explici ly inco po a e annual e ec i ely applied a i s be ween ading pa ne s. Howe e , hei inclusion does no change he esul s epo ed, and he coe icien s on he a i s hemsel es is no s a is ically signi ican ( esul s a e a ailable upon eques ). 11 We a e also eluc an o epo esul s o bila e al A T b oken down by he sub-ca ego ies “aid o economic in as uc u e”, “aid o building p oduc i e capaci ies”, and “aid ela ed o ade policy and egula ion”, because dono s may epo aid o se e al main objec i es in he OECD CRS. In addi ion, he b eakdown o o al A T by ca ego y pa ly leads o a d as ic educ ion in he obse a ions used o es ima ion, as some ecipien s ne e ecei ed aid in any o he abo e ca ego ies. Howe e , aking hese ca ea s in o accoun , in e ac ions wi h each A T ca ego y in sepa a e eg essions a e h oughou nega i e and s a is ically signi ican ( esul s a e a ailable upon eques ). 12 The PTA dummy and he PTA dep h a iable a e co ela ed wi h a co ela ion coe icien o 0.96. IDOS Discussion Pape 13/2023 12 4.2 Robus ness checks Endogenei y o A T While he coun y-yea and coun y-pai ixed e ec s in equa ion (1) con ol o an omi ed a iable bias wi h espec o he endogenei y o PTA o ma ion, hei inclusion does no add ess a po en ial es ima ion bias s emming om e e se causali y issues. Mo e speci ically, while A T migh in luence ecipien expo pe o mance, i migh likewise be de e mined by i . As a i s emedy, we he e o e lag he A T a iable by a ious yea s, also because we expec a posi i e impac o A T on ecipien expo s o ma e ialise only wi h a ce ain ime delay. Co esponding es ima ion esul s a e shown ac oss columns (1)–(3) in Table 3, and lagging he A T a iable by one, wo o h ee yea s does no al e ou baseline esul s wi h espec o he nega i e in e ac ion be ween A T and PTA dep h. Howe e , we also no e ha he es ima ed coe icien o he gene al e ec o A T u ns s a is ically non-signi ican when lagging he A T a iable by h ee yea s. This inding would sugges ha A T gene ally does no help inc ease ecipien expo s o dono s, bu could also sugges ha A T e ec i eness declines a e a pe iod o ime a e disbu semen . Al e na i ely, his obse a ion could be caused by di e ences in es ima ion samples due o he educ ion o obse a ions compa ed o ou baseline esul s and hose epo ed in columns (1)–(2) in Table 3. Mo eo e , we ind a posi i ely signed and only ma ginally, bu s ill s a is ically signi ican , PTA dep h coe icien es ima e when we lag he A T a iable in column (1) by one yea . Apa om he ac ha his e ec is no obus ac oss all model speci ica ions, no e ha his does no imply a change in ou in e p e a ion o he nega i e in e ac ion abo e. Table 3: Es ima ion esul s o a ying lags o A T & con ol unc ion app oach Va ying lags o A T Con ol unc ion 1 yea 2 yea s 3 yea s app oach VARIABLES (1) (2) (3) (4) ln A T 0.0114** 0.0104** 0.00468 0.0205* (0.00486) (0.00500) (0.00398) (0.0111) PTA dep h 0.0146* 0.00876 0.00749 0.0175 (0.00847) (0.00752) (0.00661) (0.0139) ln A T ∗ PTA dep h -0.00377*** -0.00261** -0.00317*** -0.00457** (0.00129) (0.00129) (0.00123) (0.00208) Reduced o m esiduals -0.00971 (0.00665) Obse a ions 18,112 16,647 15,208 13,850 Pseudo R2 0.996 0.996 0.996 0.996 Coun y pai s 2,074 2,034 1,999 1,564 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses in columns (1)–(3). Column (4) shows he second-s age PPML es ima ion esul s o a wo-s age con ol unc ion app oach. Boo s apped s anda d e o s in pa en heses (1,000 eplica ions). As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. We complemen he use o lagged alues o A T by a wo-s age con ol unc ion app oach o es o possible endogenei y o A T in equa ion (1). The con ol unc ion app oach sha es iden i ica ion condi ions wi h ins umen al a iables es ima ion, bu p o ides g ea e IDOS Discussion Pape 13/2023 13 compu a ional lexibili y, pa icula ly in he con ex o high-dimensional h ee-way ixed e ec s es ima ion. Mo eo e , i is also sui able in non-linea models such as PPML. Fo implemen a ion, we i s es ima e a linea eg ession model o selec ion in o A T disbu semen s. This model includes s anda d g a i y a iables ob ained om he Cen e d’E udes P ospec i es e d’In o ma ions In e na ionales (CEPII) g a i y da ase (Head, Maye , & Ries, 2010), which include ime- a ying dono and ecipien g oss domes ic p oduc s and popula ion sizes, ime- in a ian bila e al dis ance, language commonali y and colonial ies, as well as he second lag o A T commi men s as exclusion a iable. The ob ained eg ession esiduals hen se e as an addi ional con ol a iable in equa ion (1). Es ima ion esul s o he second-s age equa ion a e p esen ed in column (4) in Table 3 and con i m ou baseline esul s. No ably, he esiduals included in he second-s age a e no s a is ically signi ican a any o he s anda d le els, indica ing ha endogenei y does no appea o be a p essing issue o ou analysis. Da a a ia ion & o he PTA dep h measu es Table 4: Es ima ion esul s o a e aged A T da a & al e na i e PTA dep h indica o A e aged A T da a Wo ld Bank ho izon al dep h da a VARIABLES 2 yea s mo ing a g (1) 3 yea s mo ing a g (2) All p o isions (3) Co e p o isions (4) ln A T 0.0165*** 0.0180*** 0.0136** 0.0137*** (0.00635) (0.00677) (0.00552) (0.00521) PTA dep h 0.0153* 0.0178* 0.00248 0.00414 (0.00919) (0.00992) (0.00233) (0.00398) ln A T ∗ PTA dep h -0.00381** -0.00385** -0.000629* -0.00122** (0.00155) (0.00167) (0.000338) (0.000577) Obse a ions 22,207 25,456 19,588 19,588 Pseudo R2 0.995 0.995 0.996 0.996 Coun y pai s 2,426 2,479 2,119 2,119 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses. As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. Ou baseline esul s u he wi hs and he a e aging o A T da a o conside ing o he PTA dep h indica o s in Table 4. Mo e speci ically, we use mo ing-a e aged A T da a o e wo and h ee yea s o accoun o ola ili y and he missing da a poin s in he annual A T obse a ions (columns 1 and 2), and an al e na i e sou ce o PTA dep h (columns 3 and 4). Fo he la e , we use da a om he Wo ld Bank’s “Ho izon al Dep h” da abase (Ho mann, Osnago, & Ru a, 2017), which co e s a wide ange o 52 WTO+ and WTO-X policy a eas in 279 PTAs no i ied o he WTO. While he Wo ld Bank’s dep h indica o co e s ewe PTAs han he DESTA dep h index, i also includes p o isions ela ed o he en i onmen and labou s anda ds. Co e age o a policy a ea does no , howe e , necessa ily imply i s ele ance in a PTA. Thus, unlike he DESTA dep h indica o , he Wo ld Bank da a do no p o ide in o ma ion on he ope a ional dep h o PTAs. As a emedy, he Wo ld Bank’s dep h indica o also allows a di ision in o “co e” and “non-co e” p o isions. Co e p o isions a e de ined as p o isions unde he WTO manda e (WTO+), such as cus oms egula ions o ade- ela ed s anda ds, and addi ionally ou o he WTO-X p o isions conce ning compe i ion policy, in es men , mo emen o capi al and in ellec ual p ope y igh s p o ec ion. The e o e, “co e” p o isions a e mo e ele an ega ding o e all ade IDOS Discussion Pape 13/2023 14 impac s. We use he simple coun o p o isions co e ed as an al e na i e dep h measu e compa ed o he one p o ided by DESTA. No e ha , unlike he DESTA index, he Wo ld Bank da a only include non- a i p o isions o PTAs. As can be seen in columns (1)–(2), ou esul s o base A T e ec i eness and i s in e ac ion wi h PTA dep h emain unchanged when using a e aged A T da a. Howe e , a no able di e ence om ou baseline es ima ion esul s is he posi i ely signed and sligh ly s a is ically signi ican coe icien es ima e o PTA dep h. One explana ion o he now signi ican e ec o PTA dep h may be ound in he unde lying di e ence in he PTA-obse a ions conside ed o es ima ions. Mo e speci ically, a e aging he A T da a o e se e al yea s educes ze o-A T obse a ions, which has a posi i e e ec on he PTA ela ions conside ed o es ima ion. In o he wo ds, a e aging he A T da a o e se e al yea s educes he sample selec ion bias po en ially p esen in he es ima ions o Table 2. Tu ning o he esul s using he Wo ld Bank da a, we ind quali a i ely simila indings as when using he DESTA dep h index. Howe e , a di ec compa ison o he e ec size using he wo di e en indica o s is no possible because hei maximum alues (52 o he Wo ld Bank da a agains se en o he DESTA da a) and s uc u e a e di e en . Al hough di e ences in magni ude be ween he co e and non-co e p o isions may indica e ha he mo e economically ele an p o isions a e esponsible o he obse ed nega i e in e ac ion, a causal compa ison be ween he wo is no possible o simila easons. 5 Explo a o y ex ensions Following ou conjec u e ha compliance wi h he non- a i p o isions con ained in deep PTAs educes bila e al A T e ec i eness, in Table 5 we p esen he esul s o ou examina ion o a ious ime e ec s in mo e de ail o he pe iods be o e and a e a PTA akes e ec . To his end, o he coun y pai s wi h a PTA ela ion a any ime du ing ou sample pe iod, we i s add o columns (1) and (2) empo al placebo PTA dep h a iables and espec i e in e ac ion e ms wi h A T speci ied o h ee and i e yea s espec i ely be o e he ag eemen en e s in o o ce. Thus, o he a e age PTA in ou sample, he placebo in e ac ion cap u es a phase much u he back in ime be o e implemen a ion han he signa u e speci ica ion in Table 2. While he e could be an icipa o y e ec s be ween signa u e and he PTA aking e ec we would no gene ally expec a s a is ically signi ican in e ac ion be ween A T and PTA dep h p io o he implemen a ion o he ag eemen i bila e al A T e ec i eness is mainly educed by he implica ions s emming om compliance wi h he non- a i p o isions con ained in deep PTAs. As can be seen in columns (1)–(2), ou baseline esul s ega ding he nega i e in e ac ion be ween A T and PTA dep h a e obus o his exe cise. Mo eo e , he e is no e idence ha an icipa o y compliance seems o play a ole in educing bila e al A T e ec i eness in deep PTA ela ions. Also, he esul s om column (1)–(2) may explain he somewha smalle size o he es ima ed coe icien on he in e ac ion e ec upon PTA signa u e in Table 2 o signa u e. In ac , he inclusion o he yea s be ween a PTA’s signa u e and en y in o o ce could e en lead o a downwa d bias in he in e ac ion e ec i he e a e no an icipa ion cos s in hese yea s. An al e na i e in e p e a ion o he es o an icipa o y e ec s is o examine whe he coun ies wi h deepe PTA ela ions and hose wi h no o shallowe ag eemen s exhibi a pa allel end p io o he c ea ion o a (deep) PTA. Coun y pai s wi h shallow PTAs could di e sys ema ically om hose wi h deepe PTA ela ions, and his o m o he e ogenei y could bias he es ima ion esul s i coun ies o ming deepe PTAs a e al eady inc easing hei ade wi h each o he mo e han o he coun y pai s. While we explici ly es o endogenei y in PTA o ma ion in equa ion (1) by means o he coun y-pai ixed e ec s, he e we can es he hypo hesis ha he esponse o coun y pai s o A T and deep PTAs does no s em om uncap u ed he e ogenei ies ha IDOS Discussion Pape 13/2023 15 exis ed be o e he ac ual PTA o ma ion. Ou esul s show ha ea ed and un ea ed coun y pai s, hose wi h and wi hou deep PTAs espec i ely, do no di e in hei esponse o A T p io o a o hcoming PTA. This gi es us u he con idence in he obus ness o ou esul s. Table 5: Es ima ion esul s o di e en p e- and pos -en y in o o ce pe iods Placebo pe iod Implemen a ion pe iod 3 yea s be o e PTA en y in o o ce 5 yea s be o e PTA en y in o o ce 3 yea s a e PTA en y in o o ce 5 yea s a e PTA en y in o o ce VARIABLES (1) (2) (3) (4) ln A T 0.0137*** 0.0136*** 0.0133** 0.0128** (0.00500) (0.00522) (0.00529) (0.00522) PTA dep h 0.0141 0.00906 0.0151 0.0182 (0.0127) (0.0145) (0.0128) (0.0120) ln A T ∗ PTA dep h -0.00337* -0.00336* -0.00346** -0.00354** (0.00173) (0.00182) (0.00165) (0.00158) P e-en y-in o- o ce pe iod: Fu u e PTA dep h -0.00317 -0.00716 (0.00930) (0.0113) ln A T ∗ u u e PTA dep h 0.00143 0.00101 (0.00173) (0.00196) Obse a ions 19,588 19,588 19,588 19,588 Pseudo R2 0.996 0.996 0.996 0.996 Coun y pai s 2,118 2,118 2,118 2,118 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses. As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. We also conside PTA dep h and i s in e ac ion wi h A T wi h a lag o h ee o i e yea s, espec i ely, om he yea in which he PTA en e s in o o ce. Co esponding es ima ion esul s p esen ed in columns (3) and (4) o Table 5, howe e , do no show signi ican di e ences om he esul s o ou baseline es ima ions. Thus, he ade-o be ween deep PTAs and he e ec i eness o bila e al A T does no appea o be a one- ime adjus men p ocess, bu a he o pe sis o e he medium e m, a inding which co esponds wi h he ac ha he cos s o compliance associa ed wi h deep PTAs appea o linge o e ime (compa e Chau ou & Mau , 2011). The esul s in Table 5 also make i less likely ha he nega i e in e ac ion is due o A T being p o ided as a pu ely s a egic side-paymen o mo i a e de eloping coun ies o sign deepe PTAs (see, e.g. Hoekman e al. 2023). Indeed, aid p o ided based on economic and poli ical conside a ions o dono s, a he han based on ecipien needs, o en has nega i e implica ions on aid e ec i eness (Bea ce & Ti one, 2010; Kilby & D ehe , 2010). Howe e , while he e may well be a empo al o e lap be ween A T disbu semen s as a side-paymen and he en y in o o ce o co esponding PTAs, we would no expec he nega i e in e ac ion o occu se e al yea s a e en y in o o ce, gi en ha aid side-paymen s associa ed wi h PTAs end o be pa icula ly sho -li ed (e.g. Baccini & U pelainen, 2012; B andi, Mo in, & S ende , 2022). IDOS Discussion Pape 13/2023 16 The es ima ion esul s p esen ed in Table 6 lend addi ional suppo o he assump ion ha compliance wi h deep PTAs has eal e ec s pa icula ly on he side o de eloping PTA membe s. Mo e speci ically, eg essing dono expo s o ecipien s on A T, PTA dep h and hei in e ac ion e eals a posi i e and s a is ically signi ican e ec o he o me wo policies indi idually, bu ende s he in e ac ion s a is ically insigni ican . While A T could s imula e dono expo s o ecipien s no only h ough logis ics imp o emen s in ecipien coun ies bu also h ough ied aid e ec s, no ably, PTA dep h does no seem o comp omise bila e al A T e ec i eness. The signi ican and posi i e PTA (dep h) e ec o high-income coun y expo s compa ed o he p edominan ly non-signi ican coun e pa on he side o de eloping coun ies may also indica e a ela i ely biased imp o emen in a i -based ma ke access in a ou o de eloped coun ies in No h–Sou h PTAs. Table 6: Es ima ion esul s o dono expo s PTA dummy PTA dep h VARIABLES (1) (2) ln A T 0.00728* 0.00808* (0.00442) (0.00424) PTA dummy 0.195*** (0.0407) ln A T ∗ PTA dummy 0.00708 (0.00628) PTA dep h 0.0317*** (0.00754) ln A T ∗ PTA dep h 0.00135 (0.00119) Obse a ions 19,629 19,629 Pseudo R2 0.996 0.996 Coun y pai s 2,125 2,125 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses. As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. Nex , we examine he in e ac ion be ween A T and PTA dep h on ade in wo sec o al clus e s, ha is, ag icul u e (sum o S anda d In e na ional T ade Classi ica ion [SITC] Re ision 3 sec ions 0, 1, 2, excluding 27 and 28, and 4) and manu ac u es (sum o SITC Re ision 3 sec ions 5, 60, 61, 62, 63, 64, 65, 66, 67, 69, 7, and 8). While we ecognise ha he ade p o iles o indi idual de eloping coun ies can a y widely, de eloping coun ies gene ally end o ha e a compa a i e ad an age in ag icul u al o e manu ac u ing ade, pa icula ly in compe i ion wi h high-income coun ies. Indeed, he esul s p esen ed in Table 7 show signi ican di e ences in ou comes o he wo sec o s. No only do we ind ha A T gene ally does no seem o s imula e expo s om ecipien s o dono s in manu ac u ing, bu also he in e ac ion be ween PTA dep h and A T o manu ac u ing ade is no s a is ically signi ican . This con as s wi h he coe icien es ima e o he in e ac ion e m o ag icul u al ade, which emains quali a i ely unchanged compa ed o ou baseline es ima e, al hough o smalle magni ude. A i s unc i ical glance, his migh gi e he imp ession ha he cos s o complying wi h he p o isions included in deep PTAs a e highe o ag icul u al han manu ac u ed goods, po en ially jeopa dising exis ing compa a i e ad an age o de eloping coun ies in he o me and hus making A T less e ec i e. IDOS Discussion Pape 13/2023 17 This imp ession is ein o ced by he di e ence in s a is ical signi icance o he es ima ed coe icien on PTA dep h in ag icul u e compa ed o manu ac u es. Compliance cos s incu ed in ag icul u e may a ise om implemen ing mu ual ag eemen s on, in pa icula , SPS measu es o o he ade- ela ed s anda ds be ween in eg a ion pa ne s. SPS clauses a e among he mos common non- a i p o isions in PTAs oday and p ima ily a ec ade in ag icul u al sec o s. SPS measu es a e ules and p ocedu es designed o ensu e ood sa e y o consume s and p o ec animals and plan s om pes s and diseases. Implemen ing hese measu es could, he e o e, ha e a posi i e impac on ade. Howe e , ad hoc ha monisa ion o SPS measu es among in eg a ion pa ne s could imply no only di ec cos s o es ablishing and main aining he necessa y quali y in as uc u e, bu also indi ec cos s h ough mo e expensi e p oduc ion due o ecu ing ce i ica ion equi emen s o , o example, adhe ence o maximum pes icide le els. These ac o s po en ially unde mine he exis ing compa a i e ad an age o de eloping PTA membe s and ha m hei ade. Ou esul s would hin a he cos o s anda d ha monisa ion ou weighing po en ial bene i s o bila e al ade alues. Table 7: Es ima ion esul s o commodi y clus e s Ag icul u e Manu ac u es VARIABLES (1) (2) ln A T 0.0148*** 0.00493 (0.00408) (0.00437) PTA dep h 0.00515 0.0191*** (0.00667) (0.00628) ln A T ∗ PTA dep h -0.00248** 7.35e-05 (0.00117) (0.00118) Obse a ions 19,322 19,584 Pseudo R2 0.995 0.999 Coun y pai s 2,071 2,117 No es: Robus , clus e ed (a he coun y-pai le el) s anda d e o s in pa en heses. As e isks deno e he le el o s a is ical signi icance wi h *** p<0.01, ** p<0.05, * p<0.1. Coun y-pai and coun y-yea ixed e ec s always included bu no epo ed. Howe e , mos shallow PTAs also con ain SPS p o isions (see Sec ion 3). I hese p o isions alone a ec ed he compa a i e ad an age and hus educed e ec i eness o bila e al A T, his would no necessa ily ela e o he dep h o he PTA, and he in e ac ion e ms should he e o e no be nega i e. The nega i e in e ac ion, pa icula ly in ag icul u e, may he e o e ha e ano he cause. Namely, while non-compliance wi h s anda ds and egula ion in No he n des ina ion ma ke s is seen as a majo ba ie o ag icul u al expo s om de eloping o high-income coun ies, and ha monisa ion o SPS measu es could he e o e be an app op ia e way o add essing his p oblem, compliance wi h o he non- a i p o isions con ained in deep PTAs could also a ec he necessa y A T unding in his a ea. This would sugges ha he A T p o ided is no necessa ily a ge ed a he mos p omising a eas. Es ima ion esul s p esen ed in Table 8 addi ionally a gue agains he hypo hesis ha he nega i e in e ac ion is mainly due o a educ ion o exis ing compa a i e ad an age. He e we include p e iously uncon olled ea u es o PTAs which a e no cap u ed by he di e en componen s o PTA dep h, namely labou s anda ds o en i onmen al egula ion. To his end, we include mul idimensional indices ha assess he scope o PTA p o isions on economic and social IDOS Discussion Pape 13/2023 24 Lee, H.-H., Pa k, D., & Shin, M. (2015). Do de eloping-coun y WTO membe s ecei e mo e aid o ade (A T)? The Wo ld Economy, 38(9), 1462-1485. Lee, S., & Kim, C. (2021). The impac o deep p e e en ial ade ag eemen s on (global alue chain) ade: Who signs hem ma e s. Eme ging Ma ke s Finance and T ade, 58(6), 1629-1638. Limão, N. (2016). P e e en ial T ade Ag eemen s. In K: Bagwell, & R.W. S aige (Eds.), Handbook o Comme cial Policy (pp. 279-367). No h-Holland. Mange , M.S. and Shadlen, K.C. (2014). Poli ical ade dependence and No h–Sou h ade ag eemen s. In e na ional S udies Qua e ly, 58(1), 79-91. Ma ínez-Za zoso, I. (2019). E ec s o o eign aid on income h ough in e na ional ade. Poli ics and Go e nance, 7(2), 29-52. Ma ínez-Za zoso, I., Nowak-Lehmann D., & Rehwald, K. (2017). Is aid o ade e ec i e? A panel quan ile eg ession app oach. Re iew o De elopmen Economics, 21(4), e175-e203. Ma oo, A., Mulabdic, A., & Ru a, M. (2022). T ade c ea ion and ade di e sion in deep ag eemen s. Canadian Jou nal o Economics, 55(3), 1598-1637. Ma oo, A., Rocha, N., & Ru a, M. (2020). The e olu ion o deep ade ag eemen s. In A. Ma oo, N. Rocha, & M. Ru a (Eds.), Handbook o deep ade ag eemen s (pp 1-43). Washing on, DC: Wo ld Bank. Nishi a eno, S., & Ume ani, H. (2023). He e ogeneous e ec s o Aid‐ o ‐T ade on dono expo s: Why is Japan di e en ? Re iew o In e na ional Economics, 31(3), 1117-1145. OECD. (2022). De elopmen da abase. h ps://s a s.oecd.o g/ O e ice, G., & Rocha, N. (2014). Deep in eg a ion and p oduc ion ne wo ks: An empi ical analysis. The Wo ld Economy, 37(1), 106-136. Peacock, C., Milewicz, K., & Snidal, D. (2019). Boile pla es in in e na ional ade ag eemen s. In e na ional S udies Qua e ly, 63(4), 923-937. Pe e sson, J., & Johansson, L. (2013). Aid, aid o ade, and bila e al ade: An empi ical s udy. The Jou nal o In e na ional T ade & Economic De elopmen , 22(6), 866-894. Pie ma ini, R., & Bude a, M. (2009). A mapping o egional ules on echnical ba ie s o ade. In A. Es e adeo dal, K. Suominen, & R. The (Eds.), Regional ules in he global ading sys em (pp. 250- 315). Camb idge, MA: Camb idge Uni e si y P ess. San e amo, F.G., & Lamonaca, E. (2022). S anda ds and egula o y coope a ion in egional ade ag eemen s: Wha he e ec s on ade? Applied Economic Pe spec i es and Policy, 44(4), 1682- 1701. San os Sil a, J.M.C., & Ten ey o, S. (2006). The log o g a i y. Re iew o Economics and S a is ics, 88(4), 641-658. Shadlen, K.C. (2005). Exchanging de elopmen o ma ke access? Deep in eg a ion and indus ial policy unde mul ila e al and egional-bila e al ade ag eemen s. Re iew o In e na ional Poli ical Economy, 12(5), 750-775. S ende , F. (2019). Sign he e, please! On de eloping coun ies’ ma ke access condi ions in egional ade ag eemen s. Re iew o De elopmen Economics, 23(3), 1347-1367. Suwa-Eisenmann, A., & Ve die , T. (2007). Aid and ade. Ox o d Re iew o Economic Policy, 23(3), 481- 507. Tadesse, B., Shuk alla, E., & Fayissa, B. (2019). Ins i u ional quali y and he e ec i eness o aid o ade. Applied Economics, 51(39), 4233-4254. T ös e , B., on A nim, R., Raza, W., Chandoul, J, & Ben Rouine, C. (2023). Non- a i measu es and egula o y alignmen in a No h–Sou h con ex : Assessing compliance cos s o Tunisian ag icul u e unde he EU-Tunisia F ee T ade Ag eemen (ALECA). Jou nal o Globaliza ion and De elopmen , o hcoming: h ps://doi.o g/10.1515/jgd-2021-0044 Timini, J., Co ino is, N., & López Vicen e, F. (2022). The he e ogeneous e ec s o ade ag eemen s wi h labou p o isions. The Wo ld Economy, 45(9), 2820-2853. IDOS Discussion Pape 13/2023 25 Uni ed Na ions. (2022). UN Com ade da abase. h ps://com ade.un.o g/ Vijil, M. (2014). Aid o ade e ec i eness: Complemen a i ies wi h economic in eg a ion. The Wo ld Economy, 37(4), 555-566. Vijil, M., & Wagne , L. (2012). Does aid o ade enhance expo pe o mance? In es iga ing he in as uc u e channel. The Wo ld Economy, 35(7), 838-868. Wagne , D. (2003). Aid and ade: An empi ical s udy. Jou nal o he Japanese and In e na ional Economies, 17(2), 153–173. Wang, Z., & Xu, M. (2018). Aid o ade and he quali y o expo s. Applied Economics Le e s, 25(10), 668-673. WTO (Wo ld T ade O ganiza ion). (2022). Aid o ade global e iew 2022 – Empowe ing connec ed, sus ainable ade. Gene a: Au ho . WTO. (2023). Aid o ade ac shee . h ps://www.w o.o g/english/ a op_e/de el_e/a4 _e/a4 _ ac shee _e.h m IDOS Discussion Pape 13/2023 26 Appendix Table A1: Dono coun ies in he sample Aus alia Ge many Luxembou g Spain Aus ia G eece Ne he lands Sweden Belgium Hunga y New Zealand Swi ze land Canada Iceland No way Uni ed Kingdom Czech Republic I eland Poland Uni ed S a es Denma k I aly Po ugal Finland Japan Slo ak Republic F ance Ko ea, Republic o Slo enia Table A2: Recipien coun ies in he sample A ghanis an Democ a ic Republic o he Congo Libe ia Sain Lucia Albania Djibou i Libya Sain Vincen and he G enadines Alge ia Dominica Madagasca Samoa Angola Dominican Republic Malawi Sao Tome and P incipe An igua and Ba buda Ecuado Malaysia Saudi A abia A gen ina Egyp Maldi es Senegal A menia El Sal ado Mali Se bia Aze baijan Equa o ial Guinea Ma shall Islands Seychelles Bah ain E i ea Mau i ania Sie a Leone Bangladesh Eswa ini Mau i ius Solomon Islands Ba bados E hiopia Mexico Somalia Bela us Fiji Mic onesia Sou h A ica Belize Gabon Moldo a S i Lanka Benin Gambia Mongolia Su iname Bhu an Geo gia Mon eneg o Sy ian A ab Republic Boli ia Ghana Mo occo Tajikis an Bosnia and He zego ina G enada Mozambique Tanzania Bo swana Gua emala Myanma Thailand B azil Guinea Namibia Timo -Les e Bu kina Faso Guinea-Bissau Nau u Togo Bu undi Guyana Nepal Tonga Cabo Ve de Hai i Nica agua T inidad and Tobago Cambodia Hondu as Nige Tunisia Came oon India Nige ia Tu key Cen al A ican Republic Indonesia Niue Tu kmenis an Chad I an No h Macedonia Tu alu Chile I aq Oman Uganda China (People's Republic o ) Jamaica Pakis an Uk aine Colombia Jo dan Palau U uguay Como os Kazakhs an Panama Uzbekis an Congo Kenya Papua New Guinea Vanua u Cook Islands Ki iba i Pa aguay Venezuela Cos a Rica Ky gyzs an Pe u Vie Nam Cuba Lao People's Democ a ic Republic Philippines Yemen Cô e d'I oi e Lebanon Rwanda Zambia Democ a ic People's Republic o Ko ea Leso ho Sain Ki s and Ne is Zimbabwe