Technology, identification, and access to social programs: Experimental evidence from Panama

Reyes, Angela; Rose h, Benjamin; Ve a Cossío, Diego Alejand o Wo king Pape Technology, iden i ica ion, and access o social p og ams: Expe imen al e idence om Panama IDB Wo king Pape Se ies, No. IDB-WP-1261 P o ided in Coope a ion wi h: In e -Ame ican De elopmen Bank (IDB), Washing on, DC Sugges ed Ci a ion: Reyes, Angela; Rose h, Benjamin; Ve a Cossío, Diego Alejand o (2021) : Technology, iden i ica ion, and access o social p og ams: Expe imen al e idence om Panama, IDB Wo king Pape Se ies, No. IDB-WP-1261, In e -Ame ican De elopmen Bank (IDB), Washing on, DC, h ps://doi.o g/10.18235/0003485 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/245878 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/3.0/igo/legalcode Technology, Iden i ica ion, and Access o Social P og ams: Expe imen al E idence om Panama A ngela Reyes Benjamin Rose h Diego A. Ve a-Cossio IDB WORKING PAPER SERIES Nº IDB-WP-1261 A ugus 2021 Depa men o Resea ch and Chie Economis In e -Ame ican De elopmen Bank A ugus 2021 Technology, Iden i ica ion, and Access o Social P og ams: Expe imen al E idence om Panama A ngela Reyes Benjamin Rose h Diego A. Ve a-Cossio In e -Ame ican De elopmen Bank Ca aloging-in-Publica ion da a p o ided by he In e -Ame ican De elopmen Bank Felipe He e a Lib a y Reyes, Angela Ma ía. Technology, iden i ica ion, and access o social p og ams: expe imen al e idence om Panama / Angela Reyes, Benjamin Rose h, Diego A. Ve a-Cossio. p. cm. — (IDB Wo king Pape Se ies ; 1261) Includes bibliog aphic e e ences. 1. Iden i ica ion ca ds-Panama-Econome ic models. 2. In e ne in public adminis a ion-Panama-Econome ic models. 3. Panama-Social policy. I. Rose h, Benjamin. II. Ve a-Cossio, Diego A. III. In e -Ame ican De elopmen Bank. Depa men o Resea ch and Chie Economis . IV. Ti le. V. Se ies. IDB-WP-1261 Copy igh © In e -Ame ican De elopmen Bank. This wo k is licensed unde a C ea i e Commons IGO 3.0 A ibu ion- NonComme cial-NoDe i a i es (CC-IGO BY-NC-ND 3.0 IGO) license (h p://c ea i ecommons.o g/licenses/by-nc-nd/3.0/igo/ legalcode) and may be ep oduced wi h a ibu ion o he IDB and o any non-comme cial pu pose, as p o ided below. No de i a i e wo k is allowed. Any dispu e ela ed o he use o he wo ks o he IDB ha canno be se led amicably shall be submi ed o a bi a ion pu suan o he UNCITRAL ules. The use o he IDB's name o any pu pose o he han o a ibu ion, and he use o IDB's logo shall be subjec o a sepa a e w i en license ag eemen be ween he IDB and he use and is no au ho ized as pa o his CC-IGO license. Following a pee e iew p ocess, and wi h p e ious w i en consen by he In e -Ame ican De elopmen Bank (IDB), a e ised e sion o his wo k may also be ep oduced in any academic jou nal, including hose indexed by he Ame ican Economic Associa ion's EconLi , p o ided ha he IDB is c edi ed and ha he au ho (s) ecei e no income om he publica ion. The e o e, he es ic ion o ecei e income om such publica ion shall only ex end o he publica ion's au ho (s). Wi h ega d o such es ic ion, in case o any inconsis ency be ween he C ea i e Commons IGO 3.0 A ibu ion-NonComme cial-NoDe i a i es license and hese s a emen s, he la e shall p e ail. No e ha link p o ided abo e includes addi ional e ms and condi ions o he license. The opinions exp essed in his publica ion a e hose o he au ho s and do no necessa ily e lec he iews o he In e -Ame ican De elopmen Bank, i s Boa d o Di ec o s, o he coun ies hey ep esen . h p://www.iadb.o g 2021 Abs ac Access o iden i ica ion ca ds (IDs) is o en equi ed o claim go - e nmen bene i s. Howe e , i is unclea which policies o inc ease ID owne ship a e mo e e ec i e. We expe imen ally analyze he e ec o wo policy in e en ions o induce he imely enewal o iden i ica ion ca ds on access o a go e nmen social p og am in Panama. Sending eminde s abou expi a ion da es inc eased he p obabili y o on- ime enewals and o accessing o bene i s om a social p og am by 12 and 4.3 pe cen age poin s, espec i ely, ela i e o a con ol g oup. In con as , allowing indi iduals o enew hei ID online only inc eased enewals and access o bene i s by 8 and 2.9 pe cen age poin s, espec i ely. This esul was d i en by lowe -income indi iduals. The esul s sugges ha policies o inc ease owne ship o alid iden i y documen a ion can e-duce inclusion e o s in go e nmen p og ams and ha simply g an ing access o digi al ools may no be enough o unlock impo an e ec s.1 JEL classi ica ions:D90,H53,I38 Keywo ds: Digi iza ion, Nudges, Social p o ec ion 1Acknowledgemen s: We would like o hank Ma ´ıa In´es Vasquez o he cons an suppo h oughou his p ojec . We a e g a e ul o Sa a Espino, Edua do Valdez, Jose Luis Jim´enez, Vic o Rod ´ıguez, Ma ´ıa Tu˜n´on, John Palm and he s a o T ibunal Elec- o al and AIG (Au o idad Nacional pa a la Inno aci´on Gube namen al) o hei suppo in g an ing access o he adminis a i e eco ds. Sebas i´an Espinoza and Da id Va gas p o ided supe b esea ch assis ance. We would also like o hank Raz an Vlaicu, Mau i- cio Rome o, Sebas i´an Gallegos, Juli´an C is ia, and Julie h San ama ´ıa o aluable eed- back. All e o s a e ou own. Opinions, indings, conclusions, and ecommenda ions ex- p essed he e a e hose o he au ho s and do no necessa ily e lec he iews o he In e - Ame ican De elopmen Bank. A andomized con olled ials egis y en y is a ailable a h ps://www.socialscience egis y.o g/ ials/5371 I In oduc ion Go e nmen s commi an impo an sha e o hei budge s o edis ibu i e policies (Wo ld Bank, 2018), bu such policies o en ail o be de ac o p o- g essi e. While social p og ams end o be designed o a ge hose in need, indi iduals may no ha e access o he mos basic inpu s, such as o icial iden- i y documen a ion, o in e ac wi h he go e nmen and claim hei bene i s (Wo ld Bank, 2019). Inc easing he owne ship o alid iden i ica ion ca ds (IDs) is none heless challenging. Beyond supply-side cons ain s, issues o limi ed a en ion o misin o ma ion and ime-consuming adminis a i e p ocesses may de e indi- iduals om ob aining a alid ID. Technology o e s he p omise o ackling hese issues, ei he by easing he low o in o ma ion be ween he go e nmen and ci izens o by educing ansac ion cos s. Howe e , i indi iduals can- no ully ake ad an age o i , elying on echnology can gene a e exclusion, especially among he mos ulne able indi iduals (Mu alidha an e al., 2020). Despi e he impo ance o ID owne ship, he e is li le causal e idence on he e ec o digi al ini ia i es o inc ease alid ID owne ship and on hei e ec s on access o go e nmen bene i s and, subsequen ly, on ci izens’ wel a e. This pape ills hese gaps by expe imen ally s udying he e ec o wo popula digi al policy ools: he p o ision o eminde s h ough SMSs and access o an online pla o m o ID enewals. We pa ne ed wi h Panama’s T ibunal Elec o al (TE), he go e nmen o - ice ha o e sees he issuing o IDs, o e alua e an ini ia i e o acili a e he on- ime enewal o go e nmen IDs. Fo his pu pose, we andomly alloca ed all ci izens wi h a ailable con ac in o ma ion and whose IDs we e se o ex- pi e be ween Janua y 2020 and Augus 2020 in o wo ea men g oups and 2 a con ol g oup. Indi iduals assigned o he i s ea men a m we e gi en he op ion o enew hei IDs using an online pla o m and hus a oid one in-pe son isi o TE’s o ices ( he online ea men ). Indi iduals we e no i- ied h ough ex messages (SMSs) which included a eminde o hei ID’s expi a ion da e and a pe sonalized link o he online pla o m. They also e- cei ed weekly ollow-up messages wi h simila con en ei he un il ou weeks a e hei ID’s expi a ion da e o un il hey enewed hei ID. To isola e he e ec s o he eminde s om he online enewal op ion, a subse o indi idu- als ecei ed SMS eminde s o hei ID’s expi a ion da e bu wi hou access o he online enewal p ocess ( he eminde ea men ). Finally, a subse o indi iduals was alloca ed o a con ol g oup ha did no ecei e any eminde and could only enew hei IDs in pe son. The in e en ion was o iginally se o end in Augus 2020, bu i was suspended on Ma ch 20, 2020 due o he ou b eak o he COVID-19 pandemic. Thus, ou analysis uses he obse a ions o ci izens in he ea men g oups who, based on hei IDs’ expi a ion da e, ecei ed an SMS om T ibunal Elec o al, and obse a ions o ci izens in he con ol g oup who, based on hei IDs’ expi a ion da e, would ha e ecei ed an SMS had hey been in he ea men g oups. We ind ha bo h ea men s we e e ec i e a inducing ID enewals. Two mon hs a e he in e en ion was suspended, indi iduals in ei he ea men g oup we e 10 pe cen age poin s mo e likely o ha e enewed hei IDs, ela i e o indi iduals in he con ol g oup. These e ec s we e la gely d i en by he eminde ea men . Simply sending a eminde inc eased he p obabili y o enewal by 12.3 pe cen age poin s, a 21 pe cen inc ease ela i e o he con ol g oup mean. In con as , sending a eminde wi h access o he online pla o m only inc eased he p obabili y o enewal by 8.7 pe cen age poin s. These 3 e ec s a e s a is ically di e en om hose achie ed by simply sending he eminde message (p- alue=0.07). The ela i ely poo pe o mance o he online ea men does no seem o be ela ed o low ake-up o he online enewal pla o m. Ra he , i seems ela ed o ba ie s o comple ing he p ocess online. Twen y i e pe cen o he ci izens in he online ea men a m s a ed he online enewal p ocess, bu only 6% comple ed i (a success a e o 23%). Olde (age below he sample median) indi iduals and indi iduals wi hou a college deg ee we e less likely o comple e he enewal p ocess online. This sugges s ha he di icul ies expe ienced du ing he online enewal p ocess may ha e dissuaded indi iduals om enewing hei IDs a all, e en hough he in-pe son enewal p ocess was s ill a ailable o hem. This, in u n, o se he posi i e impac s o eminde s. We nex analyze i he in e en ion also inc eased he access and usage o eme gency go e nmen bene i s. S a ing Ap il 2020, a ew weeks a e he in e en ion was suspended, he go e nmen olled ou Vale Digi al, an in-kind ans e p og am ha deli e ed digi al ouche s o $PBA 100 ($USD 100) on a mon hly basis by linking hem o he ba o QR codes o he ID ca ds o ecipien s.2Thus, ID ca ds u ned in o debi ca ds o he pu chase o ood, medicines and cleaning supplies om pa icipa ing endo s. En ollmen was au oma ic, and indi iduals we e eligible as long as hey li ed in locali ies (co egimien os) wi h pa icipa ing endo s and we e no o mally employed. Howe e , bene icia ies could lose eligibili y i hey ailed o use hei bene i s a all o h ee consecu i e mon hs. Using adminis a i e eco ds o p og am disbu semen s and ouche usage, we ind ha , i e mon hs a e he ini ial ollou o he p og am, indi iduals 2The mon hly ouche amoun ep esen s oughly 69 and 93% o he u ban and u al po e y line in 2019, espec i ely. 4 in ei he ea men g oup we e 3.6 pe cen age poin s mo e likely o ecei e he ans e s han indi iduals in he con ol g oup; o al ouche spending was 11.3 pe cen highe ela i e o he con ol g oup. In bo h cases, he e - ec s we e d i en by he eminde ea men —a 4.3 pe cen age-poin inc ease in ouche ecep ion (p- alue<0.05). In addi ion, he in e en ion may ha e p e en ed some exclusion e o s, making he p og am de ac o mo e p og es- si e; he e ec s on access and spending a e d i en by households in low-income locali ies. These e ec s a ise despi e he ac ha he go e nmen ex ended he alid- i y o expi ed IDs weeks a e he in e en ion ended. By inc easing he chance o ha ing upda ed esidence in o ma ion, he in e en ion may ha e inc eased access o bene i s o indi iduals in p og am locali ies. Al e na i ely, ea ed indi iduals could ha e subs i u ed a los o damaged ID ca d by a new one— educing he isk o expe iencing p oblems o access hei esou ces h ough he digi al ouche —o could simply be mo e awa e o go e nmen se ices. Using ollow-up su eys, we ind sugges i e e idence ha , among indi iduals in lowe -income locali ies, he in e en ion educed he p obabili y o expe ienc- ing p oblems edeeming he ouche s and inc eased he use o online pla o ms o check p og am eligibili y, ouche balance, and p og am in o ma ion. O e all, he esul s sugges s ha in e en ions o inc ease ID enewals can inc ease access o downs eam go e nmen bene i s and imp o e a ge ing. They also sugges ha elying on digi al pla o ms o conduc go e nmen ansac ions may no always deli e . A bad use expe ience wi h he online enewal p ocess ended up de e ing use s om enewing hei IDs, e en hough he adi ional in-pe son me hod was s ill a ailable, and a enua ed he posi- i e impac s o eminde s, bo h in e ms o ID enewals and access o he digi al ouche s. A simple coun e ac ual analysis sugges s ha , had he indi iduals 5 numbe , and ma i al s a us, among o he i ems). I he in o ma ion was co - ec , hey could mo e o wa d o he second s ep. I hey a emp ed o modi y i , he sys em would p omp hem o isi an o ice in pe son, as he online applica ion did no allow o such changes, pe TE equi emen s. Second, pa - icipan s we e asked o e i y o modi y hei add ess and phone numbe .10 Thi d, pa icipan s we e p omp ed o upda e hei pho o ei he by aking a pic u e wi h hei de ice o by uploading an exis ing pic u e.11 Fou h, pa icipan s we e asked o e i y o upda e hei signa u e.12 Finally, he pa - icipan s could selec hei p e e ed ID pick-up loca ion, e iew and submi hei applica ion. A e each submission, he s a om TE alida ed and conduc ed quali y checks on he da a, pho og aph and signa u es p o ided in he online appli- ca ion. I he applica ion ailed o comply wi h quali y equi emen s, he TE con ac ed pa icipan s ia email asking hem o eplace hei pho o o signa- u e. In he case o h ee ailed a emp s o sol e he issues, he pa icipan was p omp ed o isi a TE o ice o comple e he ansac ion ollowing he s a us quo p ocess. In he case o success ul online enewals, pa icipan s we e 10This in o ma ion could be modi ied online, bu i he use decided o do so hey could no choose a di e en pick-up o ice han he one co esponding o he co egimien o o he modi ied add ess. 11The applica ion was equipped wi h a powe ul biome ic moni o ha could e i y ha he pic u e ollowed he TE quali y and echnical pho o equi emen s such as a whi e, uni o m backg ound and being in colo . In addi ion, he pa icipan s we e equi ed o be acing he on wi h a neu al acial exp ession, wi hou wea ing accesso ies o sunglasses and wi h hei hai behind he ea s. The pho og aph could no be aken wi h he on al came a o he phone. 12To upda e he signa u e, use s had o sign on a blank pape and ake a pic u e o hei signa u e wi h hei mobile de ice o webcam, which would be au oma ically digi ized by he applica ion. 12 au oma ically no i ied when hei ID was eady o pickup. A dedica ed cha line was made a ailable o p o ide cus ome suppo . III.C Randomiza ion We g ouped indi iduals in ou s udy popula ion in o andomiza ion s a a based on hei geog aphical loca ion (co egimien os) and wo g oups based on expi a ion da es (be o e and a e Feb ua y 20, 2020).13 On a e age, he e we e 81 indi iduals pe s a um. Wi hin each s a um, indi iduals we e andomized wi h equal p obabili y in o he h ee a o emen ioned expe imen al g oups. III.D Implemen a ion The i s message was sen on Janua y 29 , 2020. The in e en ion was sched- uled o las un il Sep embe 2020, bu i was suspended on Ma ch 17, 2020 due o he COVID-19 pandemic. Amid he pandemic, he go e nmen manda ed mobili y limi a ions and es ic ed in-pe son se ices p o ided by go e nmen agencies, including in-pe son and online ID enewals. In addi ion, o ensu e ha he dis up ion o se ices did no d ama ically a ec ci izens wi h expi - ing IDs he go e nmen ex ended he alidi y o expi ed IDs un il Augus 31, 2020. This ex ension was announced in Ap il 2020, a e he in e en ion was suspended. Figu e 1 p esen s a imeline o impo an e en s ela ed o his s udy. 13The i s g oup included indi iduals whose IDs we e se o expi e be ween Janua y 20,2020 o Feb ua y 19, 2020. The second g oup included indi iduals wi h ID expi a ion da es be ween Feb ua y 21, 2020 and Augus 20, 2020. The i s g oup was in ended o be a s udy pilo g oup. Howe e , as he in e en ion was discon inued on Ma ch 17, 2020, we decided o epo esul s using all indi iduals wi h IDs expi ing be ween Janua y 20, 2020 and Ma ch 20, 2020 o make up o he loss in obse a ions. 13 Th oughou he pape we epo esul s co esponding o 3,459 indi iduals wi h IDs ha expi ed be ween Janua y 20, 2020 and Ma ch 20, 2020. This sample mos ly includes pa icipan s om bo h ea men a ms who we e a leas po en ially exposed o ou eminde s be o e hei expi a ion da es and indi iduals in he con ol g oup who, based on hei IDs expi a ion da e, would ha e been exposed o eminde s had hey been in ei he ea men g oup. How- e e , a small subse o pa icipan s we e only pa ially ea ed as, based on hei ID expi a ion da e, hey only ecei ed no i ica ions a e hei ID was ex- pi ed (477 pa icipan s wi h ID expi a ion da es be ween Janua y 20-29, 2020) o only ecei ed no i ica ions be o e he expi a ion da e (445 pa icipan s wi h IDs expi a ion da e be ween Ma ch 13-20, 2020). This subsample excludes indi iduals whose IDs we e se o expi e be ween Ma ch 20 and Ap il 20, 2020 whose esponse o he in e en ion (i ea ed) would ha e been limi ed by o ice closu es and he suspension o he online pla o m. We show ha he implemen a ion dis up ion did no comp omise he a- lidi y o he expe imen al design. Fi s , we show ha , a leas in e ms o obse able cha ac e is ics, he indi iduals in he educed sample do no seem o di e om hose in he ini ial s udy popula ion. Second, we show ha ea men assignmen is balanced among se e al obse able cha ac e is ics in he educed sample. We discuss hese esul s in mo e de ail in Sec ions III.E and IV. III.E Da a We use adminis a i e eco ds co esponding o indi iduals in ou s udy popu- la ion. The adminis a i e eco ds include baseline in o ma ion ega ding he ID’s expi a ion da e, whe he he ID was enewed, he da e o enewal and he 14 p ocess h ough which i was enewed (in-pe son and online). The da ase also eco ds in o ma ion on demog aphic cha ac e is ics such as gende , age, ci il s a us, numbe o child en and home loca ion. We use hese da a o conduc balance es s and o con ol o demog aphic cha ac e is ics o inc ease powe . We also ob ained adminis a i e da a on ID enewals up o May 2020 (accessed in June 2020), wo mon hs a e he in e en ion was discon inued. These da a include in o ma ion on enewals, he mode o enewal (in-pe son o online), he da e o enewal, and on he las s ep comple ed in he online p ocess. Because only indi iduals wi h ID expi a ion da es be ween Janua y 20 and Ma ch 20, 2020 we e po en ially exposed o he ea men ,14 he adminis a i e eco ds enable us o use da a om he subse o po en ially exposed indi id- uals o compu e ea men e ec s and use da a om un ea ed indi iduals, who we e ini ially andomized in o ea men and con ol g oups, o compu e placebo e ec s and alida e ou esea ch design. We complemen he adminis a i e da a wi h phone su ey da a collec ed be ween la e Augus and ea ly No embe 2020. We collec ed a andom sample o 1,109 esponses om a o al o 3,459 pa icipan s wi h ID expi a ion da es be ween Janua y 20 and Ma ch 20, 2020, dis ibu ed equally ac oss s udy a ms. The su ey collec ed in o ma ion abou ime and mone a y cos s ela ed o he ID enewal, usage o online go e nmen and comme cial pla o ms, us in he implemen ing agency and access o eme gency-aid p og ams. Table 1 epo s summa y s a is ics based on adminis a i e and phone- su ey da a. Columns 1 and 2 epo means co esponding o all indi iduals 14As men ioned be o e, his g oup includes pa icipan s in he Reminde and Online ea men as well as pa icipan s in he con ol g oup who, based on he expi a ion da e o hei IDs, would ha e ecei ed SMSs had hey been in he ea men g oup. 15 in he ini ial s udy popula ion and o indi iduals whose IDs we e se o ex- pi e be ween Janua y 20 and Ma ch 20, 2020. Column 2 shows ha 39% o pa icipan s we e males, 36% we e ma ied, 77% had child en and he a e age age was 42 yea s old.15 These means a e e y simila o hose co esponding o Column 1 and sugges ha he indi iduals in he educed sample do no sys ema ically di e om hose in he ini ial s udy popula ion. In addi ion, Column 3 shows ha esponden s om he phone su eys did no subs an ially di e in e ms o obse able demog aphic cha ac e is ics. Fi- nally, Column 4 epo s summa y s a is ics om a na ionally ep esen a i e su ey in Panama collec ed in 2018 (Encues a de P op´osi os Mul iples). S udy pa icipan s do no appea o di e sys ema ically in e ms o obse able cha - ac e is ics, bu hey seem o ha e highe access o an in e ne connec ion. IV E ec s on ID Renewals We ob ain es ima es o he causal e ec o he in e en ion on ID enewals h ough he ollowing equa ion: Renewali=β0+β1T ea men i+XiΣ + θs+i(1) whe e Renewali akes he alue o 1 i indi idual i enewed he ID by May 31, 2020. T ea men iis an indica o ha akes he alue o 1 i he ci izen is in ei he ea men ( eminde o eminde and access o online enewal), 15Because IDs a e se o expi e e e y 10 yea s and adul ID ca ds a e only gi en ou o indi iduals when hey each he legal age o 18 yea s old, he age o he indi iduals in he s udy is bounded om below a 28 yea s old. In addi ion, he age o he indi iduals in he s udy is bounded om abo e a 70 yea s old, as IDs gi en o people 70 yea s old o olde a e no se o expi e. 16 and ze o in he case o ci izens in he con ol g oup. To inc ease p ecision, we also include a ec o o p ede e mined demog aphic cha ac e is ics (Xi) such as age, gende , numbe o child en, ci il s a us and ID expi a ion da e. When we use su ey da a, we also con ol o le els o educa ion. Finally, we include s a a ixed e ec s o accoun o he s udy design. Following Abadie e al. (2017), we use obus s anda d e o s—i.e., s anda d e o s clus e ed a he indi idual le el—as he uni o ea men is he indi idual. The coe icien o in e es is β1which cap u es he e ec o he in e en ion on ID enewals. To dis inguish he e ec s o eminde s (Reminde ea men ) om hose o he inc eased access o online enewals (Online ea men ), we also epo es ima es o he ollowing equa ion: Renewali=α0+α1Reminde i+α2Onlinei+XiΣ + θs+i(2) He e, Reminde iis an indica o o whe he he ci izen ecei ed a eminde o he expi a ion da a h ough ex messages and Onlineiis an indica o o whe he he ci izen also was allowed o enew hei ID h ough T ibunal Elec- o al websi e. In his case we a e in e es ed in h ee coe icien s: α1which cap u es he causal e ec o eminde s, α2which cap u es he e ec o e- minde s plus access o online enewal, and α2−α1which cap u es he e ec o access o he online pla o m, ne o he e ec o eminde s unde he as- sump ion o no complemen a i ies be ween he online enewal and eminde ea men s. Balance and A i ion Online Appendix Table A1 epo s means co esponding o a ba e y o demo- g aphic cha ac e is ics o con ol ci izens in Column 1 and o ci izens in he 17 Reminde and Online g oups in Columns 2 and 3, o pa icipan s whose IDs we e se o expi e be ween Janua y 20 and Ma ch 20, 2020. Columns 4 epo s di e ences in means be ween ea ed indi iduals ( ega dless o he ea men ) and con ol indi iduals, a e adjus ing o s a a ixed e ec s. Columns 5 and 6 epo di e ences in means o indi iduals in he Reminde and On- line g oups wi h espec o he con ol g oup, espec i ely. Finally, Column 7 epo s di e ences in means be ween ci izens in he Reminde and Online ea men s. Panel A uses adminis a i e da a co esponding o indi iduals whose IDs we e se o expi e be ween Janua y 20 and Ma ch 20, 2020. I shows ha he e a e no signi ican di e ences in e ms o demog aphic cha ac e is ics o expi a ion da es be ween ea men and con ol g oups. T ea ed indi iduals, howe e , seem less likely o ha e only a cell phone numbe egis e ed in T i- bunal Elec o al eco ds, as opposed o bo h a cell phone and an email add ess. These di e ences a e small ela i e o he sample mean (0.6) and a e likely ound by chance; we a e no able o ejec he null o no di e ences be ween ea men and con ol g oups in he eigh co a ia es included in Panel A. This sugges ha despi e he dis up ion in he in e en ion due o he pandemic, he alidi y o he ini ial expe imen al design was no comp omised. Table A2 in he Online Appendix shows a balance able using he ini ial s udy popula ion o compa ison. Reassu ingly, he e a e nei he subs an ial no signi ican di e ences be- ween ea men and con ol g oups o be ween ea men a ms when we ana- lyze he subsample o su ey esponden s (Panel B). Finally, o es o di e - en ial a i ion by ea men g oup, Panel C epo s su ey esponse a es by ea men s a us. We ind no e idence o di e en ial a i ion in he ollow-up su ey. 18 IV.A Resul s We begin by p o iding g aphical e idence o he e ec s o he in e en ion. Figu e 2 epo s he cumula i e sha e o indi iduals who enewed hei IDs wi hin each ea men g oup as a unc ion o weeks o/ om he expi a ion da e. Fi e o mo e weeks be o e he expi a ion da e, he enewal a es a e small and simila ac oss expe imen al g oups. As messages s a ed being sen ou weeks be o e he expi a ion da e, he enewal a es s a o di e ge a in- c easing a es un il he expi a ion da e (week 0), a e which he gaps be ween ea men and con ol g oups emain ai ly cons an . Table 2 con i ms hese pa e ns and shows ha he in e en ion was suc- cess ul a inc easing enewals on ime. Column 1 shows ha , ela i e o he con ol g oup, he p obabili y ha an indi idual enewed he o his ID in- c eased by 10.5 pe cen age poin s. The e ec on enewals is la ge, as i ep- esen s a 17 pe cen inc ease ela i e o he con ol g oup mean.16 Column 3 shows ha he e was an inc ease o 12 pe cen age poin s in he p obabili y o enewing he ID on ime (be o e o on he expi a ion da e). The small di e ence be ween he wo poin es ima es sugges s ha he e ec s o he in- e en ion we e no d i en by p oc as ina o s who would ha e enewed hei IDs anyway bu a e he expi a ion da e. To dis inguish be ween he e ec o eminde s and access o he online pla - o m, we analyze he e ec s o he in e en ion by ea men a m. Rela i e o he con ol g oup, bo h he p obabili y o enewing an ID inc eased in bo h 16DellaVigna and Linos (2020) on a comp ehensi e e iew o he li e a u e documen ing he e ec s o in e en ions p o iding “nudges” ind ha highly e ec i e in e en ions pub- lished in academic jou nals end o inc ease ake-up o a ge ed ou comes by 8.7 pe cen age poin s, and ha in e en ions implemen ed a scale by p ac i ione s in he Uni ed S a es and he Uni ed Kingdom lead o inc eases in ake-up o 1.4 pe cen age poin s, on a e age. 19 ea men a ms bu by di e en magni udes. Column 2 epo s es ima es o he e ec s o he in e en ion by ea men a m. Simply p o iding eminde s inc eased he enewal a e by 12.3 pe cen age poin s, while he Online ea - men only inc eased he p obabili y o enewal by 8.7 pe cen age poin s. This di e ence is signi ican a he 10% le el (p- alue <0.07). Columns 1 o 4 o Table A3 in he Online Appendix show simila poin es ima es among he indi iduals who pa icipa ed in he phone su ey sample. In addi ion, o alida e hese esul s, columns 5 o 8 epo esul s om a placebo es using in o ma ion o indi iduals who we e o iginally assigned o ea men and con ol g oups bu , because o he suspension o he in e en ion and he expi a ion da es o hei IDs (Ap il o Augus 2020), we e no in ac a ec ed by he in e en ion. Reassu ingly, he e a e nei he subs an ial no signi ican di e ences. The p e ious esul s a e consis en wi h models o limi ed a en ion o excep ional asks as in Ka lan e al. (2016). As ID enewals a e due e e y 10 yea s, he idea o enewing he IDs may no be a he op o he indi iduals’ minds, and he in e en ion appea s o ha e educed issues ela ed o limi ed memo y.17 IV.A.i Digi iza ion and Use Expe ience The esul s discussed abo e also sugges ha p o iding access o he online applica ion pa ially unde mined he posi i e e ec s o he eminde s. One explana ion is ha indi iduals wi h access o he applica ion we e discou aged by po en ial di icul ies wi h he online enewal p ocess. Indeed, only a small 17Ano he in e p e a ion is ha he message p o ided aluable in o ma ion, as he enewal loca ions and p ocesses could ha e changed since he las ime; his is unlikely, howe e , as he in o ma ion ha was p o ided was al eady a ailable e en be o e he in e en ion. 20 sha e o indi iduals we e able o comple e he enewal p ocess online. Column 6 in Panel A epo s ake-up a es o he online pla o m o 0.25 in he case o indi iduals in he online ea men a m, bu Column 8 shows much lowe online comple ion a es (0.06) among indi iduals in he online ea men , implying a success a e o 24%. Al hough mos o he s eps o he online enewal p ocess we e simple (e.g., con i ming and upda ing in o ma ion), o he s we e mo e complex. Panel A o Table A4 in he Online Appendix shows ha mos use s epo ha he online pla o m was easy o na iga e (82%). Howe e , bo h su ey and adminis a- i e da a sugges ha he key bo leneck was aking he pho o: many use s we e unable o mee he pla o m’s s ingen pho o quali y equi emen s. Fo y pe cen o su ey esponden s epo ed p oblems aking o uploading he pho- og aphs, which needed o comply wi h speci ic quali y equi emen s o ensu e secu i y.18 Indeed, using adminis a i e da a on he online p ocess we ound ha 60% o he indi iduals who s a ed he online p ocess s opped he p ocess a e expe iencing p oblems wi h he pho og aph (see Panel B). One explana ion o hese esul s is ha hey may be la gely d i en by s uc u al ac o s such as in e ne access and educa ion le el, which may be e en mo e ele an when he pla o ms a e no necessa ily use - iendly. Fo ins ance, ecen e idence om 13 La in Ame ican coun ies sugges s ha ech- nical p oblems wi h he online pla o ms and lack o cla i y in he ins uc ions a e among he op wo di icul ies ci ed by use s o go e nmen online se ices (Rose h e al., 2021). 18Recall ha he pho og aph had o be in colo and aken using a whi e, uni o m back- g ound. In addi ion, he pa icipan s we e equi ed o be acing he on wi h a neu al acial exp ession, wi hou wea ing accesso ies o sunglasses and wi h hei hai behind he ea s. 21 a mon hly basis, he go e nmen eplica ed he alida ion p ocess, included wo ke s whose con ac was suspended o e mina ed o he pool o bene icia- ies and emo ed bene icia ies whose con ac s we e eac i a ed o who s a ed a new con ac . Finally, al hough bene icia ies could ca y o e he balance in hei ouche s o he nex mon h, hose bene icia ies who did no use hei ouche s a all o h ee consecu i e mon hs we e au oma ically emo ed om he p og am, and hei ouche balance was e u ned o he go e nmen . Deli e ing bene i s h ough he ID ca ds could lead o highe cos s o ac- cessing he unds o people who did no ha e a alid ID ca d o o people wi h olde o damaged ca ds. To a enua e hese issues, s a ing Ap il 2020 (a e he enewal in e en ion was suspended), he go e nmen ex ended he expi- a ion da es o ID ca ds un il Augus 2020 and scaled up he online enewal pla o m o eques duplica es o los ID ca ds.22 Despi e hese solu ions, he in e en ion could s ill inc ease access o he digi al ouche s. Fi s , ea ed bene icia ies we e mo e likely o ha e a ecen ID ca d wi h upda ed in o ma- ion. This may ha e acili a ed da a e i ica ion ac oss go e nmen da abases. Second, QR and ba codes in new IDs a e likely mo e legible han hose in IDs eaching he end o hei 10-yea alidi y pe iod. This may also ha e e- duced he issues ela ed o scanning p oblems a cashie s and educed he isk o being sc eened ou om he p og am due o lack o use o he ouche s. Thi d, he in e en ion inc eased awa eness abou he digi al se ices o e ed by he go e nmen and could ha e inc eased he use o go e nmen pla o ms o e i y eligibili y, upda e o egis e hei in o ma ion, eplace a missing o damaged ID ca d, and sol e ouche usage p oblems.23 (bille e os). 22In Augus , he expi a ion da es was ex ended again un il Janua y, 2021. 23The go e nmen made a ailable a p og am websi e o consul a ions, and li e-cha assis- 28 V.B E ec s on Access o Social Bene i s du ing a C isis To analyze he impac o he in e en ion on access o eme gency assis ance, we use adminis a i e eco ds o disbu semen s and usage o Vale Digi al o ea ed and con ol pa icipan s. Fo each pa icipan , we obse e whe he hey ecei ed he ans e s du ing he i s mon h in which he p og am was olled ou in hei locali y un il Augus 2020. We es ima e equa ion (1) using an indica o o whe he he indi idual has ecei ed a leas one paymen o Vale Digi al be ween Ap il and Augus 2020 as he dependen a iable. We complemen hese da a wi h da a on o al disbu semen s and ouche spending up o Augus 2020, which we accessed in Sep embe 2020 (see Figu e 1).24 Columns 1 and 2 in Table 5 shows ha he in e en ion inc eased he p obabili y o ecei ing unds du ing Augus 2020 by 3.6 pe cen age poin s (p- alue<0.05); an 11% inc ease ela i e o he mean o he con ol g oup. Consis en wi h he e idence in Sec ion IV.A, Column 2 shows ha he poin es ima es o he e ec s o he Reminde ea men (4.3 pp.) a e la ge han hose o he Online ea men (2.9 pp), al hough he di e ence is no s a is- ically signi ican . Columns 3 and 5 show ha he o al amoun o ans e s ecei ed by each indi idual and o al ouche spending a e highe o ea ed indi iduals. Bo h poin es ima es a e simila , sugges ing ha bene icia ies we e indeed able o use hei ouche s. Columns 4 and 6 show ha he e ec s on o al disbu semen s and spending a e d i en by he Reminde ea men . To ensu e he alidi y o he esul s, Table A5 in he Online Appendix ance se ice, he A en ion and Relie Response Sys em (S.A.R.A. by i s Spanish ac onym), which was a ailable on he web and h ough he mos popula online messaging pla o ms. In addi ion, he go e nmen also made a ailable a oll- ee line so ha bene icia ies could ob ain assis ance by phone. 24Un o una ely, da a on mon hly disbu semen s and spending a e no a ailable o us. 29 epo s di e ences in ou comes be ween indi iduals assigned o he ea men and con ol g oups bu who, based on he expi a ion da es o hei IDs and due o he suspension o he in e en ion, we e no ac ually ea ed. Reassu ingly, he e a e no subs an ial o signi ican e ec s in he placebo sample. We analyze how he e ec s o he in e en ion a y by quin iles o baseline pe capi a income measu ed a he locali y-o - esidence-le el using da a om Panama’s 2011 po e y maps (Minis y o Finance, 2015).25 Figu e 3a shows ha he e ec on he p obabili y o ecei ing digi al ouche s is subs an ially highe o indi iduals li ing in a eas wi h he lowes pe capi a income. Speci - ically, he 12.9 pe cen age poin inc ease in ouche ecep ion o indi iduals in he lowes quin ile ep esen s oughly 39% o he con ol g oup mean (0.32, see column 2 in Table A6 in he Online Appendix). Likewise, Figu e 3b shows ha he e ec s on ouche spending a e d i en by indi iduals li ing in lowe - income a eas, al hough we canno de ec sign ican di e ences wi h espec o indi iduals in highe income quin iles (see Table A6 in he Online Appendix). Appendix Figu e A1 shows a simila g adien when we use sel - epo ed da a on income ca ego ies o he subsample o indi iduals who esponded o he ollow-up su ey. The esul s sugges ha he in e en ion may ha e p e en ed he exclusion o needy indi iduals. V.B.i Mechanisms Al hough he in e en ion may ha e p o ided upda ed in o ma ion o he go - e nmen , i may ha e also inc eased ouche access and usage by o he channels 25Al hough we do no ha e p e-in e en ion da a on income a he indi idual le el, he locali y-le el da a coincides wi h he andomiza ion s a a and hus using i minimizes he isk ha he esul s a e d i en by lack o balance be ween ea men and con ol g oup by income g oups as opposed o he e ogenei y. 30 oo. We analyze some o hese mechanisms using he subsample o su ey e- sponden s in Table 6 Column 1 in Panel B eplica es he pa e ns desc ibed in Figu e 3a. Column 2 shows simila pa e ns using a sel - epo ed measu e o ecep ion o go e nmen eme gency aid (Vale Digi al bu also o he bene i s). Column 3 in panel B shows ha he in e en ion educed he p obabili y o e- po ing p oblems using go e nmen bene i s among indi iduals in low-income a eas.26 This sugges s ha ea ed indi iduals wi h newe ID ca ds may ha e been less p one o expe ience issues scanning hei IDs when making pu chases. In addi ion, Column 4 shows ha , among indi iduals wi h lowe income, he in e en ion inc eased an index o usage o digi al pla o ms ela ed o eme - gency go e nmen bene i s.27 This sugges s ha he in e en ion may ha e inc eased awa eness o o he digi al se ices p o ided by he go e nmen . Ta- ble A7 in he Online Appendix shows simila esul s using sel - epo ed income ca ego ies. O e all, by inducing he enewal o ID ca ds, he in e en ion inc eased ac- cess o go e nmen bene i s, pa icula ly among lowe -income indi iduals. One implica ion is ha policies o inc ease access o alid iden i y documen a ion can unlock impo an e ec s on access o downs eam go e nmen se ices and bene i s. 26Speci ically, we asked whe he indi iduals o a amily membe had p oblems using o cashing ou go e nmen bene i s du ing he h ee mon hs p eceding he ollow-up su ey. 27The index is made up o indica o s on he usage o online pla o ms o e i y eligibili y o go e nmen bene i s, upda e da a o egis e on eme gency go e nmen bene i s and o check he balance on digi al ouche s du ing he h ee mon hs p eceding he su ey. The index is compu ed ollowing (Ande son, 2008). 31 VI Wel a e Implica ions In his sec ion, we adap Finkels ein and No owidigdo (2019)’s amewo k o compu e he social wel a e changes due o he in e en ion s udied in his pape . We hen use hese wel a e changes as benchma ks o quan i y he wel a e consequences o a coun e ac ual in e en ion ha sol es he usabili y issues ela ed o he online enewal pla o m. Such compa ison would yield es ima es o he wel a e gains ha can be achie ed i he usabili y issues ha discou aged use s om enewing hei IDs online we e sol ed. We model access o go e nmen bene i s as a unc ion o owne ship o a alid ID and allow o mispe cep ions abou he bene i s o ha ing a alid ID and ansac ion cos s o de e mine whe he an indi idual enews he o his ID. Le y0and u(y0) deno e an indi idual’s income and u ili y i she does no o enew he ID. I an indi idual owns a alid ID, she would ecei e a o al o $PBA Bin bene i s om he go e nmen wi h p obabili y π. The indi idual’s cos o enewing an ID is a unc ion o he ime needed o comple e he p ocess τ alued a ma ke hou ly wages W, o he ees p(i.e., anspo a ion, enewal ees, e c.) and o he cos s c(wi h p obabili y dis ibu ion (c)), such ha he o al enewal cos is: τW +p+c. Indi iduals may o e o unde es ima e he e u ns o owning a alid ID by and hus choose o enew hei ID i he pe cei ed u ili y o doing so (ne o cos s) is highe han he u ili y o no enewing hei ID: u(y0+ (1 + )πB)−τW −p−c > u(y0). Le c∗deno e he cos s a which an indi idual would be indi e en be ween enewing hei ID o no , he sha e o indi iduals choosing o enew hei ID Awill be de ined by F(c∗).28 28No e ha ou amewo k is indeed e y simila o he applica ion o Finkels ein and No owidigdo (2019)’s amewo k o SNAP applica ions. The main di e ences is ha , as 32 No e ha i indi iduals unde es ima e he bene i s o a alid ID by  < 0, e.g., due o issues o limi ed a en ion o misin o ma ion, he enewal a e A will be lowe han enewal a e in absence o mispe cep ions (i.e., = 0). In addi ion, no e ha al hough an indi idual’s choice is based on he pe cei ed gains om enewing an ID ((1 + )πB), he ac ual wel a e will depend on he ac ual payo (πB). We ollow Finkels ein and No owidigdo (2019) o de i e an exp ession o he a e age p i a e wel a e change due o an in e en ion ha inc eases e- newal a es by dA/dT— he willingness o pay o he in e en ion. We also conside he cos s aced by he go e nmen due o he in e en ion composed o a di ec cos : he cos o sending eminde s (D), and he indi ec iscal cos s o p ocessing an ID enewal (g) and o p o iding bene i s o each indi idual who enews he o his ID (πB). We hen compu e he ma ginal alue o public unds (MVPF) associa ed wi h he in e en ion— he wel a e inc ease in $PBA achie ed by an addi ional $PBA spen by he go e nmen .29 WTPj=−jπjBsjdA dT (4) MV PF =W TPL+WT PM+W TPH ((πLsL+πMsM+πHsh)B+(sL+sM+sH)g)dA/dT+D(5) In his case, we dis inguish be ween h ee ypes o indi iduals (low, medium and high-income) o cap u e di e ences in access o social bene i s (π), and wages (W) ha may a y by income ype. He e sjdeno es he sha e o ea ed complie s in each g oup. Ou app oach makes h ee simpli ying assump ions. Fi s , i does no opposed o modeling he choice o applying o SNAP, we model he choice o ob aining a alid ID. 29See Online Appendix D o a de i a ion o he exp essions 33 conside labo supply esponses due o social bene i s. This is consis en wi h e idence om de eloping coun ies showing ha cash ans e s do no dec ease labo supply (Bane jee e al., 2017). Second, i igno es posi i e ex e nali ies ela ed o ID owne ship o go e nmen s and indi iduals (e.g., highe ax e - enue o ecep ion o u he bene i s). Thus, ou calcula ions a e likely o yield lowe bounds o he MVPF. Thi d, we assume ha all he e ec s o he in- e en ion a e d i en by he eminde s and no by educ ions in ansac ion cos s, which is consis en wi h he e idence discussed in Sec ion IV.A. Table A8 epo s he calib a ed pa ame e s. As Vale Digi al p o ides a la bene i ega dless o income, we calib a e Bas he alue o he mon hly ans e pe pe son ($PBA 100).30 We also compu e he ime and money spen du ing he enewal p ocess (i.e., anspo a ion, pho ocopies, e c.) as a e ages among indi iduals who enewed hei ID in he con ol g oup, espec i ely.31 We se dA/dT=0.105; he a e age ea men e ec o he in e en ion on enewals (see column 1 in Table 2). Panel B epo s calib a ed pa ame e s by income g oup j∈ {L, M, H}. Fo each g oup, we calib a ed πby compu ing he sha e o indi iduals in he con ol g oup who ecei ed Vale Digi al. We compu e sj as he sha e o ea ed indi iduals ha did enew hei IDs (complie s). Wj ep esen s he a e age hou ly income pe adul in each income ca ego y, which we calcula e using da a om he ollow-up su ey.32 We calib a e jsuch ha , 30No e ha al hough Vale Digi al was implemen ed a e he in e en ion, he mon hly amoun is oughly he midpoin o he mon hly paymen s o he wo main social p o ec ion p og ams in Panama: a noncon ibu o y pension o $PBA 120 pe mon h paid o indi iduals olde han 60, and Red de Opo unidades, a condi ional cash ans e p og am deli e ing mon hly paymen s o $PBA 50 and a ge ed a households wi h child en. 31We no malize hese pa ame e s o $PBA pe mon h. 32A e age mon hly household income di ided by he numbe o adul s in he household and ans o med o hou s. We compu e he a e age household income in each ca ego y as 34 gi en he cos s o enewing an ID, indi iduals a e indi e en be ween enewing i o no (see Sec ion D in he Online Appendix o de ails). The a e age willingness o pay o he in e en ion is $PBA 3.2 (see Panel C o Table A8 in he Online Appendix). This is la ge han he di ec cos o sending he eminde s ($PBA 0.04 pe mon h).33 O cou se, he in e en ion also inc eased he numbe o people enewing hei ID and he numbe o people ecei ing social bene i s, which in u n indi ec ly inc eased go e nmen spending. A e we inco po a e hese indi ec cos s, he MVPF associa ed wi h he in e en ion is $PBA 0.92. No e ha , as discussed abo e, his migh be a lowe bound as i igno es o he bene i s ela ed o owning a alid ID. In e es ingly, lowe and middle-income indi iduals exhibi highe MVPFs ($PBA 0.92 and 0.93, espec i ely) han high-income indi iduals ($PBA 0.82). This is d i en by he ac ha , ela i e o high-income indi iduals, lowe - income indi iduals unde es ima e he bene i s o owning a alid ID by mo e (i.e., L< H). This is consis en wi h sugges i e e idence ha he in e en- ion yielded la ge e ec s on enewals o low-income indi iduals (see Table 3). One implica ion is ha i would ha e been mo e cos -e ec i e o a ge he eminde s o lowe -income indi iduals. The esul s in Sec ion IV.A showed no e ec s o p o iding access o an online pla o m o enew IDs mainly due o p oblems wi h he use expe ience wi h he pla o m. We epo a coun e ac ual analysis o quan i y he wel a e gains om sol ing hese issues. We begin by compu ing he e ec o he in e en ion on he p obabili y o he mid-poin be ween he lowe and uppe limi s o each income ca ego y 33The cos pe eminde was $PBA 0.1 and, on a e age, 5 eminde s we e sen o each indi idual. Di iding he o al spending by indi idual by 12 mon hs yields he mon hly equi alen o he cos pe indi idual. 35 ei he s a ing he applica ion online o enewing he ID, which equals o 0.17 (p- alue <0.01). This e ec is equal o he e ec on enewals had all he indi- iduals who s a ed he online p ocess comple ed he p ocess (dA/dTC=0.17)— a coun e ac ual ea men TC. In addi ion, we compu e he change in he ime spen in he enewal p ocess had he online pla o m been e ec i e (d(−τ)). The online pla o m was designed o enable indi iduals o comple e he p o- cess in 30 minu es; a 66% decline wi h espec o he a e age in-pe son enewal ime o 90 minu es. Thus, we se d(−τ) = 1 hou o -0.083 in mon hly e ms. Appendix Sec ion D shows ha he MVPF associa ed wi h he coun e ac ual in e en ion can be w i en as: MV PFC=−BdA dT CPj∈{L,M,H}jπjsj+d(−τ)Pj∈{L,M,H}Wj(Aj+sjdA dT C) ((πLsL+πMsM+πHsh)B+(sL+sM+sH)g)dAC dT +D(6) In his case, he wel a e gains a ise om educing mispe cep ions h ough eminde s ( i s e m in he nume a o ) and educing he ansac ion cos s o ob aining a new ID (second e m in he nume a o ). Panel C o Table A8 shows ha he wel a e gains o he coun e ac ual in e en ion ($PBA 5.56) a e abo e and beyond he di ec cos o he in e en ion and yield la ge p i a e wel a e gains han he ac ual in e en ion ($PBA 3.2). Thus, he e is a pe -pe son wel a e loss o indi iduals o $PBA 2.3 ha can be a ibu ed o he usabili y issues ha p e en ed a success ul use o he online pla o m. In e es ingly, hese wel a e changes a e d i en by high-income indi iduals who alue mo e he educ ion in he ime needed o he enewal p ojec (WH> WM> WL). In addi ion, he gains achie ed by he coun e ac ual in e en ion almos ully o se he di ec and indi ec e ec s o he in e en ion on go e nmen spending (MV PFC= 0.99).34 34No e ha his is a a he conse a i e es ima e since he online enewal can also e- 36 The esul s unde sco e he impo ance o use - iendly digi al pla o ms o ansac wi h he go e nmen . Digi iza ion o go e nmen ansac ions b ings he p omise o inc easing ci izens’ wel a e by educing ansac ion cos s. This p omise can only be ul illed i he inno a ions a e b oadly accessible. They also emphasize ha he cos s o ob aining alid IDs migh be he e ogeneous, and hus uni e sal access o alid iden i ica ion equi es di e en policy ools. Re- minde s may be pa icula ly bene icial o lowe -income households ha end o be mo e p one o beha io al biases (Schilbach e al., 2016), and digi iza ion may be ela i ely mo e bene icial o weal hie indi iduals acing ela i ely highe ansac ion cos s. VII Conclusion and Policy Implica ions In a se ing o high ansac ion cos s and low compliance a es, we s udy he ex en o which ex message eminde s, and a link o an online p ocedu e, inc ease he enewal a e o na ional ID ca ds in Panama. ID enewal is no only impo an in insically, as ca ying an up- o-da e ID is equi ed by law, bu also o i s unc ional u ili y in accessing public and p i a e se ices. The u ili y o he na ional ID inc eased ma kedly du ing he COVID-19 pandemic, as i became he p ima y ehicle o dis ibu ing subsidies o ood and o he basic goods. O e all, we ind ha eminde s a e e ec i e in p omo ing ID enewal, con- sis en wi h p io esea ch on eminde s, which is abundan in he beha io al economics li e a u e. Fu he mo e, up ake in ID enewal is posi i ely associ- a ed wi h up ake o social p og ams, pa icula ly ele an du ing he pandemic. duce he go e nmen cos s o enewing an ID g, which a e none heless le cons an in his calcula ion. 37 Table 1: Summa y s a is ics Admin da a Admin da a Phone su ey sample HH su ey All ID exp. da e 1-3/2020 (1) (2) (3) (4) Age 43.382 42.131 41.147 46.739 Male 0.446 0.389 0.402 0.476 Cohabi s wi h pa ne 0.647 0.590 Ma ied 0.380 0.362 0.379 # household membe s 3.883 4.165 Has child en 0.795 0.771 0.670 0.583 # o child en 1.874 1.829 1.285 1.288 Has child en unde 6y 0.285 0.312 # o child en unde 6y 0.357 0.465 No educa ion o pa ial p ima y 0.005 0.071 P ima y o pa ial seconda y 0.137 0.201 Seconda y educa ion 0.221 0.288 Some e cia y educa ion 0.140 0.292 In e ne access 0.589 0.384 Obse a ions 11956 3459 1128 19300 No e: The able epo s means o demog aphic cha ac e is ics o di e en g oups. Column 1 epo s means using all he ini ial s udy popula ion using adminis a i e eco ds. column 2 epo s means o all s udy pa icipan s whose IDs we e se o expi e be ween Janua y 20 and Ma ch 20, 2020. Column 3 epo s means o he subsample o s udy pa icipan s who esponded o he phone su ey. Column 4 epo s means using da a om he 2018 o Panama’s Mul iple Pu pose Household su ey using su ey weigh s. The sample in column 4 is es ic ed o indi iduals o age 28 o 70, as only indi iduals in ha age ange a e eligible o ID enewals. Column 4 also uses sampling weigh s. 44 0 .1 .2 .3 .4 .5 .6 .7 .8 Cummula i e sha e o enewed ids -4 -3 -2 -1 0 1 2 3 4<-4 >4 Weeks o expi a ion da e Con ol Message Online Figu e 2: Renewals be o e and a e he Expi a ion Da e No e: The igu e epo s he cumula i e sha e o s udy indi iduals who enewed hei IDs as a unc ion o ime ela i e o he expi a ion da e, measu ed in weeks such ha he expi a ion da e co esponds o he i s day o week 0. The igu e is compu ed using he subsample o indi iduals whose IDs expi ed be ween Janua y 20 and Ma ch 20, 2020. 45 Table 2: E ec s on Renewals (Admin Da a) (1) (2) (3) (4) (5) (6) (7) (8) Renewal On ime S a ed online Comple ed online T ea men 0.105*** 0.120*** 0.125*** 0.030*** (0.019) (0.019) (0.008) (0.004) Reminde 0.123*** 0.133*** 0.001 0.002 (0.021) (0.022) (0.004) (0.002) Online 0.087*** 0.107*** 0.257*** 0.060*** (0.021) (0.022) (0.014) (0.008) Obse a ions 3,156 3,156 3,156 3,156 3,156 3,156 3,156 3,156 R-squa ed 0.241 0.242 0.149 0.149 0.155 0.283 0.103 0.129 Con ol G oup Mean 0.593 0.593 0.252 0.252 0 0 0 0 P- alue (Reminde -Online) 0.0660 0.251 0 0 ∗∗∗p < 0.01,∗ ∗ p < 0.05,∗p < 0.1 No e: The able epo s es ima es o he ea men e ec s o he in e en ion based on equa ions (1) and (2). The es ima ed sample includes all indi iduals whose IDs we e se o expi ed be o e he in e en ion was discon inued. All eg essions con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. P- alues co esponding o a es o equali y o coe icien s be ween he Message and Online ea men g oups a e epo ed in he bo om pa o he able. Robus s anda d e o s a e epo ed in pa en heses. 46 Table 3: He e ogeneous E ec s on Online Renewals by Age, Educa ion, Income and Access o In e ne Connec ion (Su ey Da a) (1) (2) (3) (4) (5) (6) Renewal S a ed online Comple ed online T ea men 0.067 0.159*** 0.043** (0.073) (0.041) (0.017) T ea men ×Age >median -0.087 -0.087** -0.035* (0.064) (0.035) (0.018) T ea men ×Uni e si y+ 0.006 0.044 0.035* (0.069) (0.041) (0.019) T ea men ×Low income 0.131* -0.050 -0.017 (0.069) (0.037) (0.016) T ea men ×In e ne access 0.085 0.044 -0.011 (0.071) (0.039) (0.017) Reminde 0.105 0.013 0.005 (0.087) (0.027) (0.014) Online 0.021 0.323*** 0.082*** (0.080) (0.061) (0.028) Reminde ×Age >median -0.092 -0.024 -0.010 (0.073) (0.023) (0.013) Online ×Age >median -0.071 -0.211*** -0.073** (0.074) (0.053) (0.030) Reminde ×Uni e si y+ -0.045 -0.016 0.000 (0.081) (0.026) (0.014) Online ×Uni e si y+ 0.058 0.100* 0.070** (0.081) (0.060) (0.034) Reminde ×Low income 0.125 0.017 0.010 (0.080) (0.025) (0.012) Online ×Low income 0.131* -0.089 -0.039 (0.079) (0.057) (0.029) Reminde ×In e ne access 0.120 0.009 -0.003 (0.086) (0.027) (0.016) Online ×In e ne access 0.050 0.116** -0.009 (0.079) (0.057) (0.029) Obse a ions 942 942 942 942 942 942 R-squa ed 0.479 0.482 0.243 0.432 0.181 0.227 Con ol G oup Mean 0.460 0.460 0 0 0 0 ∗∗∗p < 0.01,∗ ∗ p < 0.05,∗p < 0.1 No e: The able epo s es ima es o he ea men e ec s o he in e en ion on online enewals as a unc ion o demog aphic cha ac e is ics based on equa ion (3) using he sample o indi iduals in he ollow-up su ey. Each column epo s esul s om a di e en eg ession o he ou come on in e es on ea men s a us and in e ac ions o ea men s a us wi h a ec o o co a ia es ha cap u e dimensions o he e ogenei y o in e es . Columns 1, 3 and 5 epo esul s based on a speci ica ion ha pools bo h ea men a ms. Columns 2, 4 and 6 epo esul s by ea men a m. All eg essions a e es ima ed using he subsample o su ey esponden s and con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. Robus s anda d e o s a e epo ed in pa en heses. 47 Table 4: E ec s on T ansac ion Cos s, Ci izen Pe cep ions and Usage o Digi al Pla o ms (Su ey Da a) (1) (2) (3) (4) (5) (6) VARIABLES T ansac ion cos s Ci izen pe cep ions Digi al pla o ms T ea men -0.041 0.007 -0.042 (0.055) (0.050) (0.060) Reminde -0.016 -0.015 -0.031 (0.060) (0.055) (0.068) Online -0.061 0.035 -0.045 (0.065) (0.057) (0.069) Obse a ions 942 942 942 942 942 942 R-squa ed 0.253 0.253 0.239 0.240 0.347 0.347 Con ol G oup Mean 0.00 0.00 0.00 0.00 0.00 0.00 P- alue (Reminde -Online) 0.461 0.340 0.842 ∗∗∗p < 0.01,∗ ∗ p < 0.05,∗p < 0.1 No e: The able epo s es ima es o he ea men e ec s o he in e en ion based on equa ions (1) and (2). The dependen a iables a e index a iables compu ed ollowing he app oach sugges ed by Ande son (2008). All eg essions con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. P- alues co esponding o a es o equali y o coe icien s be ween he wo ea men a ms a e epo ed in he bo om pa o he able. Robus s anda d e o s a e epo ed in pa en heses. 48 Table 5: E ec s on Recep ion and Usage o Digi al Vouche s (Admin Da a) (1) (2) (3) (4) (5) (6) VARIABLES Vale Digi al (4-8/2020) Disbu sed $Spen $ T ea men 0.036** 11.339* 9.977* (0.018) (5.827) (5.701) Reminde 0.043** 12.950** 11.027* (0.021) (6.591) (6.453) Online 0.029 9.632 8.864 (0.021) (6.762) (6.612) Obse a ions 3,156 3,156 3,156 3,156 3,156 3,156 R-squa ed 0.257 0.258 0.253 0.253 0.255 0.255 Con ol G oup Mean 0.329 0.329 93.93 93.93 87.97 87.97 P- alue (di coe s) 0.510 0.610 0.734 ∗∗∗p < 0.01,∗ ∗ p < 0.05,∗p < 0.1 No e: The able epo s es ima es o he ea men e ec s o he in e en ion co esponding o equa ions (1) and (2). All eg essions con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. P- alues co esponding o a es o equali y o coe icien s be ween he wo ea men a ms a e epo ed in he bo om pa o he able. Robus s anda d e o s a e epo ed in pa en heses. 49 (a) Recep ion o Vale Digi al (4/2020-8/2020) (b) Amoun disbu sed and spen Figu e 3: T ea men E ec s by Quin iles o Pe Capi a Income No e: The igu e depic s ea men e ec s o he in e en ion on he p obabili y o ecei ing a leas one paymen o Vale Digi al be ween Ap il and Augus 2020, on he o al amoun o ans e s disbu sed o he digi al ouche and on o al ouche spending. 90% con idence in e als a e compu ed using obus s anda d e o s. 50 Table 6: E ec s on Access o Digi al Vouche s: Mechanisms (Su ey Da a) (1) (2) (3) (4) Vale Digi al (4/2020- 8/2020) Reci ed eme - gency help Expe ienced p oblems o cash ou bene- i s Index - Digi al ools Panel A: A e age e ec s (su ey esponden s) T ea men 0.068* 0.002 -0.000 0.076 (0.035) (0.037) (0.021) (0.057) Obse a ions 942 942 942 942 R-squa ed 0.328 0.262 0.180 0.238 Panel B: By locali y income ca ego y (su ey esponden s) T ea men 0.213*** 0.133 -0.133* 0.455*** (0.078) (0.102) (0.069) (0.157) T ea men X Q2 -0.229** -0.066 0.123 -0.469** (0.114) (0.126) (0.087) (0.202) T ea men X Q3 -0.117 -0.195 0.145* -0.472** (0.113) (0.128) (0.081) (0.208) T ea men X Q4 -0.118 -0.065 0.150* -0.329* (0.106) (0.131) (0.082) (0.197) T ea men X Q5 -0.187* -0.235* 0.163** -0.434** (0.107) (0.129) (0.077) (0.193) Obse a ions 942 942 942 942 R-squa ed 0.331 0.266 0.186 0.254 Con ol G oup Mean 0.312 0.649 0.0720 0.00 ∗∗∗p < 0.01,∗ ∗ p < 0.05,∗p < 0.1 No e: The able epo s es ima es o he a e age ea men e ec s o he in e en ion and he e ogeneous e ec s by income quin iles using he sample o su ey esponden s. Columns 1 epo ea men e ec s on he p obabili y o ha ing ecei ed a leas one paymen o Vale Digi al du ing Ap il o Augus 2020. Columns 2 epo s e ec s on sel - epo ed in o ma ion o ecep ion o any eme gency aid om he go e nmen (including Vale Digi al as well as o he p e-exis ing p og ams). Column 3 epo s e ec s on he p obabili y o epo ing ha ing had p oblems o cash ou o use go e nmen bene i s, Column 4 epo s e ec s on an index o usage o digi al pla o ms ela ed o Vale Digi al. All eg essions con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. Robus s anda d e o s a e epo ed in pa en heses. 51 Online Appendix A Appendix: Suppo ing E idence 52 (a) Recep ion o Vale Digi al (4/2020-8/2020) (b) Amoun disbu sed and spen Figu e A1: Robus ness: E ec s by income using sel - epo ed income ca e- go ies (su ey da a) No e: The igu e depic s ea men e ec s o he in e en ion on he p obabili y o ecei ing a leas one paymen o Vale Digi al be ween Ap il and Augus 2020, on he o al amoun o ans e s disbu sed o he digi al ouche and on o al ouche spending. The e ec s a e calcula ed based on he subsample o su ey esponden s. 90% con idence in e als a e compu ed using obus s anda d e o s. 53 Table A7: Robus ness: Mechanisms (Using sel - epo ed income) (1) (2) (3) (4) Vale Digi al (4/2020- 8/2020) Recei ed eme - gency help Expe ienced p oblems o cash ou bene- i s Index - Digi al ools T ea men 0.141** 0.099 -0.070 0.233** (0.062) (0.066) (0.043) (0.111) T ea men X Medium Income -0.120 -0.131 0.087 -0.209 (0.089) (0.089) (0.056) (0.145) T ea men X Highe income -0.112 -0.155 0.113** -0.281** (0.088) (0.097) (0.051) (0.143) Obse a ions 942 942 942 942 R-squa ed 0.362 0.273 0.189 0.275 Con ol G oup Mean 0.312 0.649 0.0720 0.00 ∗∗∗p− alue < 0.01,∗ ∗ p− alue < 0.05,∗p− alue < 0.1 No e: The able epo s es ima es o he ea men e ec s o he in e en ion by income ca ego y ob ained by es ima ing a eg ession o access o Vale Digi al on ea men s a us and i s in e ac ion wi h an indica o o sel - epo ed income ca ego ies. All eg essions con ol o s a a ixed e ec s as well as a ec o o baseline demog aphic cha ac e is ics. Robus s anda d e o s a e epo ed in pa en heses. 60 Table A8: Calcula ion o ac ual and coun e ac ual MVPFs Panel A: Common pa ame e s Ac ual in e en ion Coun e ac ual in e en ion Mon hly bene i s in $PBA (B) 100 100 Ope a ing cos s o enewing an ID in $PBA (g) 15 15 Pe -capi a cos s o sending SMSs in $PBA (5 SMS pe indi idual) 0.5 0.5 A e age ime needed o in-pe son enewals in minu es (τ) 90.73 30.85 O he cos s in $PBA ( anspo a ion, pho ocopies, ees) (p) 3.77 3.77 E ec o he in e en ion on enewals (dA/dT) 0.105 0.173 Panel B: Pa ame e s by income g oup (j) Ac ual in e en ion Coun e ac ual in e en ion Income g oups Income g oups Low Middle High Low Middle High P obabili y o ecei ing bene i s (πj) 0.32 0.36 0.24 0.32 0.36 0.24 Sha e o complie s (sj) 0.38 0.37 0.25 0.37 0.37 0.27 Mispe cep ion o bene i s (j) -0.99 -0.99 -0.92 -0.99 -0.98 -0.91 Hou ly wage in $PBA(Wj) 0.72 1.40 13.44 0.72 1.40 13.44 Renewals (p opo ion) (Aj) 0.28 0.44 0.28 0.28 0.44 0.28 Panel C: P i a e Wel a e gains and MVPFs To al p i a e wel a e change (WTP) 3.20 5.56 MPVF all (MV PF) 0.92 0.99 MPVF by income g oup (MV P Fj) 0.92 0.93 0.82 0.95 0.98 1.03 No e: Panels A and B epo calib a ed pa ame e s. See Sec ion D o de ails. Time cos s (τ) and ees (p) o enew an ID as well as p ocessing cos s (g) and SMS cos s (D) a e no malized o mon hly equi alen s o he calcula ions o MVPFs. The e ec o he coun e ac ual in e en ion on enewals is compu ed by eg essing an indica o o whe he an indi idual enews he ID o s a s he enewal p ocess online on a dummy cap u ing ac ual ea men s a us, demog aphic con ols and s a a ixed e ec s. Income g oups a e de ined based on he income ca ego ies in ou su ey da a. Low income g oup: indi iduals wi h household income be ween $PBA 0 o 500. Middle income g oup: indi iduals wi h household income be ween $PBA 500-1400. High income g oup: indi iduals wi h amily income highe han $PBA 1400. 61 B Appendix: Tex messages Remide ea men : •Reminde s sen weekly be o e he expi a ion da e: “[T ibunal Elec o al] [-NOMBRE-], u c´edula ence el DD de MES. Renu´e ala g a is. In o: h ps://sede. ibunal-elec o al.gob.pa/in o” •Reminde s sen weekly a e he expi a ion da e: “[T ibunal Elec o al] [-NOMBRE-], u c´edula enci´o el DD de MES. Renu´e ala g a is. In o: h ps://sede. ibunal-elec o al.gob.pa/in o” Online ea men : •Reminde s sen weekly be o e he expi a ion da e: “[T ibunal Elec o al] [-NOMBRE-], u c´edula ence el DD de MES. Aho a una isi a al TE, enu´e ala g a is en l´ınea: h ps://sede. ibunal-elec o al.gob.pa/ enue a/CODINVIT” •Reminde s sen weekly a e he expi a ion da e.“[T ibunal Elec o al] [- NOMBRE-], u c´edula enci´o el DD de MES. Aho a una isi a al TE, enu´e ala g a is en l´ınea: h ps://sede. ibunal-elec o al.gob.pa/ enue a/CODINVIT” •In his case, CODINVIT is an eigh -digi pe sonalized code ha is asso- cia ed wi h he ci izen’s ID and phone numbe . 62 Figu e A2: Landing page: Reminde ea men Figu e A3: Landing page: Online ea men 63 C Appendix: Online Renewal Applica ion In his sec ion we desc ibe he s eps needed o enew an ID online. S ep 1:Ve i ica ion. The use s we e asked o e i y hei iden i y by p o iding hei ID numbe , ID se ial numbe and bi h da e. Figu e A4: S ep 1: Ve i ica ion. 64 S ep 2: Regis a ion. A e e i ying hei iden i y, ci izens we e p omp ed o c ea e an accoun using an email and passwo d o hei choice. This enabled use s o sa e a d a o he ansac ion and comple e i la e . Figu e A5: S ep 2: Regis a ion. 65 S ep 3: Gene al In o ma ion. A lis o all he s eps and equi e docu- men a ion was displayed in he sc een o ou line he enewal p ocess. Figu e A6: S ep 3: Gene al In o ma ion. 66 S ep 4: Ve i ica ion o pe sonal da a. The ci izen was asked o con- i m hei pe sonal da a. I hey wan ed o modi y any hing, hey would be in o med hey had o do so in pe son a he egis a s’ o ice, and he p ocess would s op a his poin . Figu e A7: S ep 4: Ve i ica ion o pe sonal da a. 67 S ep 5: Ve i ica ion o con ac in o ma ion. He e he use s will be asked o en e he same phone numbe h ough which hey ecei ed he no i ica ion om T ibunal Elec o al. Figu e A8: S ep 5: Ve i ica ion o con ac in o ma ion. 68 S ep 6: Addi ional in o ma ion. Ci izens could upda e hei o gan dono s a us, disabili y s a us and whe he hey wan ed hei husband’s las name o be on hei ID ca d (in he case o ma ied women). Figu e A9: S ep 6: Addi ional in o ma ion. 69 G=X j=L,M,H ((πjB) + g)Aj D.1.1 E ec s o he ac ual in e en ion Le dA/dT deno e he change in he sha e o indi iduals who ob ained a new ID due o he ea men . Consis en wi h he esul s discussed in Sec ion IV.A, we assume ha he e ec o he in e en ion comes only om he eminde s—i.e., he in e en ion a ied he le el o mispe cep ions (). Thus, dA/dT =dA/d. Using he Leibniz ule one can show ha : dVj dj = [u(y0+πjB)−u(y0+πjB(1 + j)]dA dj He e we exploi he ac ha dA/dj= (dFj(c∗)/dc∗)dc∗/dj= (c∗)jdc∗/dj. Using a i s -o de Taylo se ies app oxima ion o he pe cei ed u ili y o e- newing an ID a ound he ac ual u ili y o doing so we ge : u(y0+πjB(1 + j)−u(y0+πjB) = u0(y0+πjB)jπjB) and dVj dj =−u0(y0+πjB)jπjBdA dj We hen di ide he change in wel a e due o he in e en ion by u0() o ob ain a money-me ic measu e o he a e age change in wel a e due o he in e en ion; he willingness o pay o he in e en ion: WTPj=−jπjBsj dA dT (A2) 76 He e, we assume ha he ea men e ec s o he in e en ion was cons an ac oss each ype o complie s. Thus, he change in he enewal a e among indi iduals o ype jequals he sha e o complie s o ype j(sj)mul iplied by he a e age ea men e ec o he in e en ion dA/dT. Adding (A2) ac oss ypes we ge ha he a e age change in o al wel a e due o he in e en ion is equal o he nume a o o equa ion (5). The denomina o o exp ession (5) is ob ained by conside ing he ma ginal cos o he go e nmen associa ed o he change in indi iduals enewing hei IDs due o he in e en ion (dG/dT = Pj=L,M,H (πjB+g)sjdA/dT) plus he pe -pe son cos o implemen ing he in e en ion D. Thus, equa ion (5) cap u e he a e age willingness o pay o he in e en ion among indi iduals o ype jand equa ion (5) cap u es he ma ginal change in social wel a e pe dolla spen by he go e nmen due o he in e en ion: he ma ginal alue o public unds. D.1.2 E ec s o he coun e ac ual in e en ion The coun e ac ual in e en ion a ies mispe cep ions () h ough eminde s and educes he ime needed o enew an ID (τ). We compu e he change in wel a e due o in e en ion dVj/dTC o an indi idual o income g oup jby aking he o al de i a i e o Vjwi h espec o jand −τ. Fo his we use he Leibniz ule and he same i s -o de Taylo app oxima ions desc ibed abo e. dVj dTC=∂Vj ∂ d +∂Vj ∂(−τ)d(−τ) =−jπjB∂Aj ∂j dj−jπjB∂Aj ∂−τd(−τ) + WjAjd(−τ) =−jπjB(∂Aj ∂j dj+∂Aj ∂−τd(−τ)) + WjAjd(−τ) =−jπjB(dAj dTC) + WjAjd(−τ) 77 The las e m o he exp ession comes om he ac ha he o al change in enewals due o he coun e ac ual in e en ion is dAj/dTC= (∂Aj/∂j)dj+ (∂Aj/∂(−τ))d(−τ). Fo simplici y, we assume ha he ea men e ec o he coun e ac ual in e en ion is cons an ac oss income g oups such ha dAj/dTC=sjdA/dTC. Finally, he change in wel a e o income g oup jis: dVj dTC=−jπjBsj(dA dTC) + WjAjd(−τ) (A3) Adding he exp ession in equa ion (A3) ac oss income g oups, we ob ain he nume a o o equa ion (6). D.2 Calib a ion •Income g oups (j) a e de ined based on he income ca ego ies in ou su ey da a. Low income g oup: indi iduals wi h household income be- ween $PBA 0 o 500. Middle income g oup: indi iduals wi h household income be ween $PBA 500-1400. High income g oup: indi iduals wi h amily income highe han $PBA 1400. •The mon hly go e nmen bene i s (B) a e equal o $PBA 100; he alue o he mon hly digi al ouche s. This is smalle han he pe -pe son mon hly ans e s ecei ed by bene icia ies o 120 a los 65 (Panam´a’s noncon ibu o y pension p og am) and is abo e he mon hly household ans e associa ed o Red de Opo unidades (Panam´a’s condi ional cash ans e p og ams a ge ed a amilies wi h young child en). Bo h ans- e s ep esen $PBA 120 and 50 pe mon h, espec i ely. •The cos o p ocessing a new enewal gis calib a ed using he eplace- men penal y o $PBA 15 ha indi iduals ha e o pay o ob ain a dupli- ca e o a los o damaged ID. No e ha he ID enewal due o expi a ion 78 is ee o cos in Panam´a. Fo ou calcula ions, we assume ha people enew hei ID only once a yea and di ide gby 12 so ha i is exp essed in mon hly e ms. This is an o e s a emen o he p ocessing cos s as enewals a e only due e e y 10 yea s. No e ha we keep his pa ame e cons an du ing he analysis o he wel a e impac s o he coun e ac ual policy e en hough he coun e ac ual policy could ha e also educed he p ocessing cos s aced by he go e nmen . •The pe -pe son cos o he in e en ion (D) equals $PBA 0.5. On a e - age, we sen 5 SMSs o ea ed indi iduals. The cos o each SMS was 10 cen s. •The a e age ime needed o in-pe son enewals in minu es (τ) is cali- b a ed using he o al ime spen in he enewal p ocess plus he ime spen commu ing o T ibunal Elec o al o ices. Fo his we ook he a - e age o su ey esponden s in he con ol g oup who decided o enew hei ID. •We compu e he a e age obse able ansac ion cos s pby adding up he amoun spen on anspo a ion o TE o ices plus he cos o pho ocopies o he documen a ion needed o enew an ID (when applicable) and hen a e aging ac oss su ey esponden s in he con ol g oup who enewed hei ID. •The p obabili y o ecei ing go e nmen bene i s (πj) is compu ed as he sha e o con ol-g oup indi iduals who ecei ed a leas one paymen o ale digi al, by income g oup. •The hou ly wage (Wj) is compu ed in h ee s eps. Fi s , we ook he midpoin o he o al household income ca ego y limi s based on da a 79 om he ollow up su eys. This p o ided an app oxima ion o he mon hly household income wi hin each ca ego y.We hen di ided he mon hly household income by he numbe o adul s li ing in indi idual i’s household. Finally, we di ided he pe -adul mon hly income by 160— he numbe o mon hly wo king hou s assuming 8 wo king hou s pe day and 5 wo king days. •The sha e o complie s sjis compu ed as he sha e o ea ed indi id- uals in each income g oup jwho enewed hei ID. In he case o he coun e ac ual in e en ion TC, he sha e o complie s is compu ed as he sha e o indi iduals in each income g oup jwho ei he enewed hei ID o s a ed he enewal p ocess online. •We compu ed he mispe cep ions o bene i s om enewing IDs as he alue o j o which an indi idual in g oup jwould ha e been indi e - en be ween enewing he ID. We did so by using a i s -o de Taylo app oxima ion a ound he u ili y o no enewing an ID. 80