Assessing the role of social networks on migrant labor market outcomes: evidence from a representative immigrant survey

Wo king Pape Se ies Assessing he Role o Social Ne wo ks on Mig an Labo Ma ke Ou comes: E idence om a Rep esen a i e Immig an Su ey __________________________________________ Cá ia Ba is a No a School o Business and Economics - Uni e sidade No a de Lisboa CReAM, IZA and NOVAFRICA Ana Isabel Cos a No a School o Business and Economics - Uni e sidade No a de Lisboa NOVAFRICA ISSN 2183-0843 Wo king Pape No 1601 Ap il 2016 NOVAFRICA Wo king Pape Se ies NOVAFRICA | No a School o Business and Economics|Faculdade de Economia | Uni e sidade No a de Lisboa Campus de Campolide | 1099-032 Lisboa – Po ugal | T: (+351)213 801 673 | F: (+351) 213 801 105 | www.no a ica.o g Any opinions exp essed he e a e hose o he au ho (s) and no hose o NOVAFRICA. Resea ch published in his se ies may include iews on policy, bu he cen e i sel akes no ins i u ional policy posi ions. NOVAFRICA is a knowledge cen e c ea ed by he No a School o Business and Economics o he No a Uni e si y o Lisbon. I s mission is o p oduce dis inc i e expe ise on business and economic de elopmen in A ica. A pa icula ocus is on Po uguese-speaking A ica, i.e., Angola, Cape Ve de, Guinea-Bissau, Mozambique, and Sao Tome and P incipe. The Cen e aims o p oduce knowledge and dissemina e i h ough esea ch p ojec s, publica ions, policy ad ice, semina s, con e ences and o he e en s. NOVAFRICA Wo king Pape s o en ep esen p elimina y wo k and a e ci cula ed o encou age discussion. Ci a ion o such a pape should accoun o i s p o isional cha ac e . A e ised e sion may be a ailable di ec ly om he au ho . Assessing he Role o Social Ne wo ks on Mig an Labo Ma ke Ou comes: E idence om a Rep esen a i e Immig an Su ey  Cá ia Ba is a† and Ana Isabel Cos a‡ Ap il 2016 Abs ac Wha ole do social ne wo ks play in de e mining mig an labo ma ke ou comes? We examine his ques ion using da a om a andom sample o 1500 immig an s li ing in I eland. We p opose a heo e ical model o mally p edic ing ha immig an s wi h mo e con ac s ha e addi ional access o job o e s, and a e he e o e be e able o become employed and choose highe paid jobs. Ou empi ical analysis con i ms hese indings, while ocusing mo e gene ally on he ela ionship be ween mig an s’ social ne wo ks and a a ie y o labo ma ke ou comes (namely wages, employmen , occupa ional choice and job secu i y), con a y o he li e a u e. We ind e idence ha ha ing one mo e con ac in he ne wo k is associa ed wi h an inc ease o 11pp in he p obabili y o being employed and wi h an inc ease o abou 100 eu os in he a e age sala y. Howe e , ou da a is no sugges i e o a ne wo k size e ec on occupa ional choice and job secu i y. Ou indings a e obus o sample selec ion and o he endogenei y conce ns. JEL classi ica ion: D8, F22, J3 Keywo ds: Social Ne wo ks; In e na ional Mig a ion; Wage De e mina ion; Labo Ma ke In eg a ion  The au ho s a e g a e ul o help ul sugges ions om Ana Balcão Reis, Ana Ga cia, Bien Baloca ing, F ancesco Ces a i, Julia Sei he , Ma ilde G ácio, Ped o B inca, José Ta a es, and Te esa Molina-Millán. This pape makes use o a ep esen a i e immig an su ey gene ously unded by he EU NORFACE P og amme on In e na ional Mig a ion, unde he lead o P incipal In es iga o s Ca ia Ba is a and Gaia Na ciso. † No a School o Business and Economics, CReAM, IZA and NOVAFRICA. Email: ca ia.ba is[email p o ec ed] ‡ No a School o Business and Economics, and NOVAFRICA. Email: cos a.anai[email p o ec ed]m 1 1. In oduc ion The s udy o social ne wo ks eme ged in he economics li e a u e as a way o explain in o ma ion misma ches and o he ma ke ic ions. Ea ly empi ical s udies show ha abou hal o employed indi iduals ely on amily and iends o ind jobs o o ha e access o job in o ma ion (G egg and Wadswo h, 1996; Addison and Po ugal, 2001). The main heo e ical hypo hesis linking social ne wo ks o labo ma ke ou comes p oposes he exis ence o an in o mal channel, usually o med by ela i es, iends and acquain ances, ha p o ides indi iduals in o ma ion no a ailable h ough o mal sou ces. Such channel mi iga es job sea ch ic ions in wo dis inc ways: by making a ailable in o ma ion o job seeke s abou employmen oppo uni ies, while also p o iding employe s e e ences abou he wo ke s. In he ecen decades, a g owing li e a u e has in es iga ed he ole o social ne wo ks on mig an ou comes. In one o i s mos p ominen indings, mig an s end o clus e in g oups and sha e in o ma ion among hem (see Edin e al, 2003 and Munshi, 2003), which illus a es he pa icula impo ance o social ne wo ks o mig a ion. Being newcome s o he labo ma ke , mig an s a e in need o in o ma ion abou job openings and he cha ac e is ics o he labo ma ke , pa icula ly upon a i al o he hos coun y. On he o he hand, ne wo ks may also be use ul om he pe spec i e o he employe who o en lacks in o ma ion abou he newly a i ed mig an s. Despi e he expec able impo ance o social ne wo ks on mig an labo ma ke ou comes, he e is s ill a lack o obus empi ical e idence on his esea ch ques ion. This pape a emp s o ill his gap by using a andom sample o 1500 immig an s om 110 di e en na ions esiding in I eland in 2010. We use de ailed su ey da a o mig an s and hei con ac s, which enables us o measu e he social ne wo k o each mig an , and o ind in his way empi ical e idence linking ne wo ks o he labo ma ke ou comes o immig an s. The co e hypo hesis o his s udy is ha , immig an s wi h la ge ne wo ks a e mo e exposed o in o ma ion and 2 he e o e a e mo e likely o be employed and o hold highe paid jobs. We also analyze o he labo ma ke ou comes such as occupa ional choice and job secu i y (meaning he job s abili y o e ed by he ype o job con ac ). Ou pape con ibu es o he p e ious li e a u e by s udying he ole o mig an ne wo ks a a ious dimensions o labo ma ke in eg a ion. We examine h ee main ques ions: (i) is he p obabili y o being employed highe o indi iduals wi h mo e con ac s; (ii) wha is he causal e ec o he ne wo k size on subsequen wages: do immig an s wi h la ge ne wo ks ha e highe wages; (iii) is he e a selec ion p ocess wi h he choice o wo king among immig an s. Iden i ica ion o he causal e ec s o ne wo k size on mig an ou comes is complica ed by he possibili y o se e al endogenei y p oblems. Fi s , he suspec ed causal ela ionship be ween he mig an ne wo k size and he espec i e labo ma ke ou come can be simul aneously de e mined by unobse able cha ac e is ics. Second, endogenei y can also occu h ough a selec ion p ocess ha leads some immig an s o choose o pa icipa e in he labo ma ke while o he s do no – which esul s in a sample selec ion p oblem in ha i is no possible o obse e he wages o hose who do no wo k. In o de o ackle hese possible es ima ion biases, and o ensu e he obus ness o ou esul s we es ima e se e al dis inc models: Linea P obabili y Model (LPM), Two-S age Leas Squa es (2SLS), Hecki model and IV-Hecki model. We ind e idence ha immig an s wi h a la ge ne wo k size a e mo e likely o be employed and ea n highe sala ies. Mo eo e , ou empi ical analysis con i ms ha he e is, in ac , a sel -selec ion p ocess in luencing he decision o wo k. Howe e , ou da a indica es ha ne wo k size does no seem o in luence occupa ional choice o he job secu i y o immig an s. Ou esul s a e in line wi h he p e ious indings in he li e a u e ela ed o wages 3 and employmen . The impac o he ne wo k size on he occupa ion and job secu i y a e new indings o he li e a u e. The s uc u e o he emainde o he pape is o ganized as ollows. Sec ion 2 e iews he main esul s ound in he li e a u e. Sec ion 3 shows a heo e ical amewo k o model he impac o he ne wo k size on he employmen and wages o immig an s. Sec ion 4 desc ibes he me hodology used and he iden i ica ion issues. Sec ion 5 p esen s he da a used and he desc ip i e s a is ics. Finally, Sec ion 6 includes he empi ical esul s and sec ion 7 concludes. 2. Li e a u e Re iew This sec ion p esen s a sys ema ic analysis o he p e ious indings on he opic and ela es he pas li e a u e wi h ou empi ical s a egy. We summa ize he impac o social ne wo ks bo h on mig a ion and on labo ma ke ou comes. As ini ially p oposed by Sjaas ad (1962) and Ha is and Toda o (1970), mig a ion is o en an unce ain and isky in es men . Howe e , mig a ion ne wo ks can lowe he cos s and unce ain y o ollowe mig an s, as e i ied by McKenzie and Rapopo (2010). Umblijs (2012) p esen s heo e ical and empi ical e idence ha i po en ial mig an s ha e access o a ne wo k a he des ina ion, mo e isk-a e se mig an s will mig a e han i hey had no access o hese ne wo ks 1 . Cal o-A mengol and Zenou (2005) we e one o he i s au ho s o pu o wa d a heo e ical model o he ole o ne wo ks on labo ma ke ou comes pa icula ly wages and employmen . The model accoun s o a job-ma ching p ocess, in which wo ke s ind jobs h ough hei social ne wo k, and shows ha in he s eady-s a e, labo ma ke ou comes a e posi i ely co ela ed ac oss ime and ac oss agen s wi hin a ne wo k. None heless, i assumes ha ne wo ks a e exogenous, a s ong assump ion ha has no been alida ed by empi ical esea ch (a ew examples include Munshi, 2003 and An, 2015). 1 Fo u he e idence o he impo ance o social ne wo ks o mig a ion, see: Beine e al. (2011), Ba is a and Umblijs (2014), Ba is a and Umblijs (2016), Ba is a e . al (2016). 4 La e , Wahba and Zenou (2005) ex ended he model o Cal o-A mengol and Zenou (2005) by di e en ia ing be ween low- and high-educa ed indi iduals and compa ing he e iciency o using ne wo ks o ind a job wi h o he sea ch me hods. The unde lying assump ion made by he au ho s is ha low-educa ed indi iduals only use in o mal job sea ch me hods o ind a job, while high-educa ed indi iduals use bo h o mal and in o mal me hods. Conside ing popula ion densi y as a p oxy o he size o he ne wo k in Egyp , hey p o ide empi ical e idence ha condi ional on being employed, he p obabili y o inding a job h ough iends and ela i es, compa ed o o he sea ch me hods, is highe in dense a eas han in less dense a eas. Gi en he impo ance o ne wo ks as an in o mal ins i u ion wi h consequences in he labo ma ke , one would expec i s e ec o be s onge among mig an s ( ela i e o na i es) as hey a e newcome s o he ma ke . Munshi (2003) a gues ha bo h mig an s and employe s a he des ina ion lack ull access o in o ma ion abou each o he , eason why hey a e in need o job e e als. Social ne wo ks ha e he ole o dec easing he asymme y o in o ma ion be ween bo h agen s. Using a sample o Mexican mig an s in he US, he s udy inds ha mig an s wi h la ge ne wo ks a e mo e likely o be employed and hold highe paying jobs upon a i al. Also, Chen (2009) uses he p opo ion o labo mig an s in he home illage as an indica o o he illage social ne wo k o s udy he e ec s o in e nal mig a ion in China and inds ha la ge ne wo ks a e associa ed wi h highe wages. Recen ly, Ke and Mando (2015) p o ide e idence ha immig an s in he US end o clus e in he same occupa ions as immig an s om he same na ionali y. A esul e i ied by Pa el and Vella (2007) ha ind addi ional e idence o a wage p emium o hose immig an s who choose o wo k in he mos popula occupa ions o hei ne wo ks due o possible ma ke powe among hese g oups. Al hough he co ela ion be ween ne wo ks and highe wages may seem in ui i e, some o he s udies ha e poin ed o an opposi e di ec ion. Fo ins ance, Da che Lou y (2006) and 5 Long e . al (2013) ind e idence o a wage penal y among hose indi iduals who use ne wo ks o ind jobs o e o he sea ch me hods. Thei iew es s on he assump ion ha wo ke s who do no ob ain be e job o e s h ough o mal channels use ne wo ks o ind jobs as a las eso . Fo ha eason, ne wo k use s will hold lowe paid jobs compa ed o non-use s. In ou pape , we will u he explo e he e ec s o ne wo ks in he labo ma ke using a sample o in e na ional mig an s li ing in I eland. By con olling o a a ie y o indi idual cha ac e is ics, and add essing po en ial endogenei y p oblems, we a e able o p o ide no el insigh on his issue. Con a y o he exis ing li e a u e, ou main de ini ion o ne wo ks is no only based on geog aphical p oximi y o place o o igin. Ou measu e o ne wo ks includes non-mig an s and mig an s om he same coun y, bu also om di e en coun ies. In spi e o he g ea impo ance o mig an communi ies, i is plausible ha hey also sha e and a e p o ided wi h in o ma ion abou he labo ma ke by o he esiden s. This app oach is e en mo e pe inen in he con ex o I eland as mig an s come om a a ie y o coun ies (including highly de eloped and de eloping economies), mos ly wo k in high-skilled jobs and a e, on a e age, highly educa ed 2 . 3. Theo e ical Model In o de o ame ou esea ch, we p opose an immig an ’s sea ch beha io model, adap ed om Pa el and Vella (2007), o s udy he heo e ical mechanisms linking ne wo ks o wages and employmen . Job sea ch is made using o mal and in o mal me hods, whe e he la e e e s o he indi idual ne wo k o employed iends and ela i es. The model assumes a con inuous ime amewo k in which job o e s can a i e om ei he o mal o in o mal channels ollowing a Poisson p ocess and ha immig an s, a e a i ing o a new coun y, seek o maximize he expec ed alue o discoun ed u u e income using discoun a e 𝑟. Indi iduals 2 Cen al S a is ics O ice – I eland (census epo o 2011) 6 (2) (3) ecei e job o e s h ough he o mal channel wi h a i al a e 𝑝, and om he in o mal channel wi h a i al a e 𝑝𝑁. As long as he immig an does no pa icipa e in he labo ma ke , he ecei es compensa ion 𝑏, which can be in e p e ed as leisu e o a di e en ype o compensa ion. Wages o e ed h ough he o mal channel a e ep esen ed by 𝑤 and ha e dis ibu ion 𝐹(𝑤), while wages o e ed h ough he ne wo k channel a e ep esen ed by 𝑤𝑁 and ha e dis ibu ion 𝐹𝑁(𝑤𝑁). The model assumes ha , as long as he mig an inds a job wi h wage 𝑤 o 𝑤𝑁, he will keep i o e e . The alue o wo king in a job ound h ough he o mal channel and he in o mal channel is, 𝑊(𝑤)= 𝑤 𝑟 and 𝑊(𝑤𝑁)= 𝑤𝑁 𝑟, espec i ely. Thus, he low alue o being unemployed, 𝑟𝑈, is: 𝑟𝑈=𝑏+𝑝∫𝑚𝑎𝑥[𝑊(𝑤)−𝑈,0]𝑑𝐹(𝑤)+ 𝑝𝑁∫max [𝑊(𝑤𝑁)−𝑈,0]𝑑𝐹𝑁(𝑤𝑁) (1) The i s e m on he igh -hand side o he equa ion, 𝑏, s ands o he ins an aneous payo o being unemployed, while he second and he hi d e ms ep esen he expec ed alue om being employed in a job ound ia he o mal and ne wo k channels espec i ely. The decision o en e ing he labo ma ke is made when he indi idual inds a job o e wi h wage ealiza ion 𝑤 o 𝑤𝑁 in which he alue o being employed, (𝑤) o (𝑤𝑁) is highe han he alue o no en e ing he labo ma ke , U. The la ge he ne wo k size, he mo e job o e s a gi en indi idual will ecei e h ough he in o mal channel. Assuming ha he numbe o people in he ne wo k is gi en by 𝑛 and ha he a i al job a e 𝑝 is cons an among all indi iduals in he ne wo k, 𝑝𝑁 is exp essed by: 𝑝𝑁= ∑𝑝𝑖 𝑖∈𝑛 =𝑛𝑝 The indi idual will accep he job o e i 𝑤 is a leas as la ge as he ese a ion wage, 𝑤𝑅. The ese a ion wage is sol ed by W(𝑤𝑅) = 𝑈. No ing ha 𝑟𝑈=𝑤𝑅 and 𝑊− 𝑈=𝑤−𝑤𝑅 𝑟. Adding his o equa ion (1), one ob ains he ollowing simpli ied equa ion o he ese a ion wage: 13 had o ul ill he ollowing equi emen s: 1) be 18 yea s old o olde ; 2) no I ish o B i ish bo n 10 ; 3) ha e a i ed in I eland be ween 2000 and six mon hs p io o he in e iew da e; and 4) no an I ish o B i ish ci izen. The andom sampling p ocedu e ollowed h ee s eps: i s , 100 Enume a ion A eas (EA) we e andomly selec ed ou o he 323 Elec o al Dis ic s. Second, 15 households we e chosen wi hin each EA. Finally, i he household had mo e han one membe eligible o ake pa in he su ey, he indi idual esponden was andomly selec ed based on a nex -bi hday ule. Due o missing ele an in o ma ion abou eligibili y o nine esponden s, he inal numbe o immig an s included in he sample is 1,491. 5.2. De ining Ne wo k Size In o de o es ima e he e ec o he ne wo k size on wages in he con ex o mig a ion one i s needs o de ine he ne wo k size o each immig an . We ollow (Pa el and Vella, 2007) and include in he ne wo k he indi iduals ha we e wo king a he ime he su ey was conduc ed, and he e o e we e mo e able o p o ide inside in o ma ion abou he labo ma ke 11 . We use h ee di e en ques ions in he su ey o co e all he con ac s o he immig an s in I eland. The i s one conce ns he composi ion o he household membe s o he esponden 12 : “Please indica e all he pe sons who belong o his household”. The second ques ion conce ns he con ac s who he esponden knew li ing in I eland be o e he had mo ed o he coun y: “Be o e coming o I eland, how many people did you know who we e al eady li ing in I eland (a he ime you mo ed)”? 13 ; The hi d ques ion includes he con ac s li ing in I eland whom, a he ime o he su ey, he mig an had mos con ac wi h: 10 As epo ed in Ba is a and Na ciso (2013), “B i ish ci izens we e excluded due o he close his o ical ies be ween I eland and G ea B i ain.” 11 We also es ed ne wo k de ini ions including indi iduals ha we e no wo king. Howe e , we ob ained weake esul s, which sugges s ha labo ma ke in o ma ion is sha ed by hose who a e inside s in he labo ma ke . 12 Household membe s a e de ined as hose who usually sleep and ea in he same uni . 13 People belonging o he same household a e excluded om his ques ion. We decided o include hese con ac s once hey may ha e been an impo an channel o in o ma ion o he mig an s bo h be o e and a e he mig a ion p ocess. 14 “Cu en ly, who a e he people (excluding people you li e wi h o people you knew be o e coming o I eland) ha you ha e mos con ac wi h in I eland?”. Using his in o ma ion we a e able o obse e he size o he ep esen a i e ne wo k o each mig an . 5.3. Desc ip i e S a is ics Table 1 includes he main desc ip i e s a is ics abou he mig an s’ cha ac e is ics. The i s a iable co esponds o he cu en mon hly ne wages exp essed in eu os. As we can obse e, only 957 esponden s epo ed a posi i e wage 14 . Ou da a indica es ha a ound 63% o he indi iduals epo ed o be wo king as hei main occupa ion. The size o he mig an ne wo k is he explana o y a iable o main in e es . Ne wo k size is a disc e e a iable ha assumes alues be ween 1 and 12 in ou da a. The a e age numbe o people in he ne wo k is app oxima ely h ee. The sample is made up o a highly educa ed class o mig an s who had, on a e age, 15 yea s o educa ion, which oughly co esponds o a Bachelo ’s deg ee. The a e age age is app oxima ely 33 yea s old. The a e age numbe o immig an s pe coun y is 10004. Finally, he a e age mon hly household income, excluding he esponden ’s, is 1128 eu os and a ound 46.5% o he immig an s ha e a child. 15 6. Resul s 6.1. Employmen The heo e ical model p oposed in sec ion 3 p edic s ha he size o he ne wo k is posi i ely co ela ed wi h he p obabili y o he immig an being employed. Table 2 p esen s he empi ical analysis o his p edic ion. Employed is a bina y a iable ha assumes alue 1 i 14 Those indi iduals o whom we did no obse e a posi i e wage ha e he ollowing occupa ions: unpaid housewo k, s uden , e i ed and unemployed o a e no allowed o wo k due o isa issues. 15 The supplemen a y appendix includes Table A.1. which shows he dis ibu ion o o eign-bo n indi iduals in he sample by con inen and Table A.1., a desc ip ion o he dis ibu ion o he op na ionali ies. 15 he indi idual is employed and alue 0 i o he wise 16 . We s a by using he LPM o s udy he ela ionship be ween he employmen s a us and ou main a iable o in e es , ne wo k size 17 . The esul s indica e ha ha ing one mo e pe son in he ne wo k is associa ed wi h an inc ease o 3 pe cen age poin s (pp) in he p obabili y o being employed – an es ima e signi ican a he 1% signi icance le el. Mo eo e , emales a e less likely o be employed compa ed o males, as he gende coe icien is nega i e and highly signi ican . Being mo e educa ed is also posi i ely co ela ed wi h employmen , one mo e yea o school inc eases he p obabili y o being employed by 1.3 pp. Age and age squa ed a e also s ong de e minan s o he p obabili y o being employed, al hough in e sely ela ed. As he immig an ge s olde , he is less likely o be employed. In e es ingly, he numbe o yea s since mig a ion does no seem o in luence he employmen . The same holds ue o ma i al s a us. The esul s a e obus once endogenei y conce ns a e accoun ed o . The ne wo k size e ec becomes highe in magni ude sugges ing ha ha ing one mo e con ac in he ne wo k inc eases he p obabili y o being employed by 11.3 pp and emains posi i e and signi ican a he 10% le el. Ou ins umen al a iable seems o be alid as i passes he weak iden i ica ion es : he C agg-Donald Wald F es p esen ed in he end o he able is 12.249 implying a s ong associa ion be ween he s ock o immig an s and he numbe o con ac s in he ne wo k size. 18 These empi ical esul s he e o e lend empi ical suppo o he i s es able implica ion o ou heo e ical model. 6.2. Wages 16 We conside as employed indi iduals who epo ed o be wo king as he main occupa ion. 17 Table A.4. o he supplemen a y appendix includes he P obi es ima ions whose sign, signi icance and magni ude a e simila o he LPM es ima ions. 18 S aige and S ock (1997) indica e ha , o he case o a single endogenous eg esso , he ins umen is s ong i i passes he h eshold o 10 in he i s -s age F-s a is ic. 16 The second main p edic ion o ou heo e ical model is ha indi iduals’ wages a e a posi i e unc ion o he size o he in o mal ne wo k. To empi ically es his hypo hesis, we employ h ee di e en models: OLS, Hecki and IV-Hecki . The es ima ed coe icien s and espec i e s anda d e o s a e p esen ed in Table 4. Mon hly wage appea s in he loga i hmic o m gi en he skewness p esen in he a iable (1.9318, a conside able igh -hand skewness). We begin ou analysis by in oducing he OLS es ima ion esul s (Column 1). Ne wo k size is s a is ically signi ican a he 5% le el wi h a coe icien o 0.0260. Tha is, ha ing one mo e pe son in he ne wo k size is associa ed wi h a 2.60 pe cen inc ease in wages. Indi idual’s gende does no seem o be co ela ed wi h he sala y ea ned by he mig an . The coe icien is nega i e (sugges ing ha , on a e age, emales end o ecei e less han males), al hough i is no signi ican . Mo eo e , he numbe o yea s o schooling is posi i ely co ela ed wi h he wages ea ned. One mo e yea in school is associa ed wi h a 2.14 pp inc ease in wages, a esul signi ican a he 10% le el. The numbe o yea s in I eland since mig a ion, is also an impo an posi i e de e minan o wages, indica i e o he p esence o a p ocess o acquisi ion o human capi al in he hos coun y. Age and age squa ed a e no co ela ed wi h wages in he OLS eg ession. Finally, ma ied immig an s do no ea n mo e on a e age compa ed wi h o he ma i al s a us. In addi ion o he main indi idual cha ac e is ics explaining wages, namely gende , yea s o schooling, yea s in I eland, ma i al s a us, age and con inen ixed e ec s, we include u he con ols ha may also po en ially a ec ing wages, o inc ease he compa abili y o he di e en es ima ion models we use. Al hough he OLS es ima o allows one o ha e a i s insigh on he ela ion be ween ne wo k size and wages, i may yield biased es ima es as sample selec ion and o he endogenei y conce ns migh be ac i e in ou sample. To es and co ec o he hypo hesis o sample selec ion, we employ a Heckman selec ion model, included in he second column o 17 Table 4. The in e se Mills a io ep esen ing he la en selec ion ac o is nega i e and s a is ically signi ican , indica ing ha ha ing a child and a highe household income is nega i ely co ela ed wi h he p obabili y o en e ing he labo ma ke . We can conclude ha he e is a selec ion bias on unobse ed cha ac e is ics ha u ns he sample o hose mig an s o whom we obse e wages di e en om he emaining ones. Conside ing ha unobse ed cha ac e is ics o he immig an s e lec hei unobse ed abili y, he Hecki es ima e o ne wo k e ec s e eals ha column (1) o e es ima ed he ne wo k e ec s o less able people who gain mo e om he in o mal channels. A b ie e lec ion on why less able people ake mo e ad an age om hei ne wo ks leads us o conside ha lowe ea ning-abili y mig an s may be elying mo e on hei social ne wo k as hey ind i mo e di icul o acqui e jobs h ough o mal me hods. Compa ing he OLS wi h he Hecki es ima es, we can see ha when we ake in o accoun selec ion in he decision o wo king on unobse ed abili y, he ne wo k e ec is s ill signi ican a he 10% le el and ou main coe icien o in e es is now 0.0196, lowe han he OLS es ima e. A e co ec ing o sample selec ion, age and age squa ed become signi ican in opposi e di ec ions. While age is posi i ely ela ed wi h he mig an s’ wages, age squa ed in luences wages in a nega i e way. This implies a diminishing ma ginal e ec o age, i.e., as he mig an ge s olde , he e ec o age on wages lessens. This esul is obus o he IV- Hecki model. All he o he coe icien es ima es emain simila o he OLS model. Column 3 p esen s he esul s o he Hecki model wi h he ne wo k size ins umen ed, ou IV-Hecki model. A e con olling o endogenei y, we ob ain ou bes es ima e o he causal e ec o ne wo ks on wages, which is s ill posi i e and signi ican a he 10% le el. Ha ing one mo e pe son in he ne wo k is associa ed wi h an inc ease o 6.46 pe cen in wages. This esul sugges s ha bo h he OLS and Hecki model unde es ima e he e ec s o 18 ne wo k size. Ou ins umen al a iable is signi ican and alid wi h a i s -s age F alue o 20.845. IV-Hecki esul s a e mo e obus when compa ed o he o he wo models, as i simul aneously deals wi h he wo main sou ces o po en ial es ima ion biases in ou empi ical analysis. Ne e heless, he obus ness o ou main coe icien o in e es in he h ee models p o ides clea e idence ha he size o mig an social ne wo ks has a posi i e impac on labo ma ke ou comes. Table A.3. o he supplemen a y appendix p esen s he i s -s age eg essions o he Hecki and IV-Hecki models. 6.3. Occupa ional Choice and Job Secu i y In addi ion o employmen and wages, we also examine he impac o ne wo k size on o he labo ma ke ou comes, namely occupa ional choice and job secu i y. As men ioned in Pa el and Vella (2007) and Ke and Mando (2015), immig an s end o choose he same occupa ions as immig an s om he same backg ound, and he e o e mig an s belonging o he same ne wo k may enjoy a la ge ma ke powe in some gi en sec o s o he economy. In ou empi ical analysis we es i ha ing mo e con ac s can in luence indi iduals o choose a ce ain ype o occupa ion. Tables 4 o 6 include he es ima ions o he immig an s’ main occupa ions. Ou da a ules ou he hypo hesis ha ha ing mo e con ac s can in luence mig an s o choose a ce ain occupa ion o e he o he ones, as he signi icance o he esul s is no obus o endogenei y o o he inclusion o con inen ixed e ec s in he eg ession. Howe e , he esul s sugges some e idence o a possible e ec ela ed wi h low-skilled jobs (Table 4) o which he coe icien s a e signi ican i we no accoun o con inen ixed e ec s and emains posi i e in ou equa ions. In wha conce ns job secu i y, i.e., he employmen s abili y o e ed by he ype o con ac , we examine he p obabili y o ha ing a ce ain ype o job con ac when he 19 ne wo k size is la ge . Table 7 p esen s he esul s o he di e en ypes o job con ac s: pe manen , empo a y, no con ac o e en sel -employmen . We ind weak e idence ha ne wo k size is co ela ed wi h pe manen and empo a y con ac s o o being sel -employed. E idence is s onge o indi iduals who do no ha e any employmen con ac . In e es ingly, ne wo ks seem o lead immig an s o o he ypes o job si ua ion a he han jobs wi h no con ac as he coe icien is signi ican and nega i e. This is, ha ing one mo e con ac dec eases he p obabili y o ha ing a job wi h no con ac by 17 pp. Fo his eason, ne wo ks seem o play an impo an ole in dec easing p eca ious wo k. 7. Concluding Rema ks This pape examined he impac o la ge ne wo k sizes on immig an ’s wages, p obabili y o being employed, occupa ional choice and job secu i y. We ind e idence ha ha ing one mo e con ac in he hos coun y’s labo ma ke is associa ed wi h an inc ease o 11pp in he p obabili y o being employed and wi h an inc ease o abou 100 eu os in he a e age sala y. Howe e , he e idence ela ed o he impac o ne wo ks on job secu i y and occupa ional choice is a he weak o non-exis en . Expanding he p e ious indings in he li e a u e ha ha e been mainly ocused on de eloping coun ies, his s udy concludes ha ne wo ks a e also impo an o mig an s in de eloped coun ies. In o ma ion lows can play an impo an ole o immig an s in a new labo ma ke , a cen al conclusion o policy make s who wish o c ea e be e mig a ion expe iences. Ou esul s may e lec demand and supply in luences in he labo ma ke . On he demand side, employe s wan ing o educe hei sc eening cos s will employ immig an s based on e e als o , i hey a e sa is ied wi h he pe o mance o a ce ain ype o wo ke s, hey a e mo e likely o hi e people om he same backg ound. F om he supply pe spec i e, upon a i ing in he hos coun y, mos immig an s a e s ill seeking o job oppo uni ies (as ou da a sugges s, only 13.48% o hem had al eady a job o e when hey mo ed o I eland). I 20 indi iduals a e in con ac wi h mo e people who a e employed, i is likely ha hey will ha e be e in o ma ion abou oppo uni ies in he job ma ke . Al hough we a e no able o disen angle such possible mechanisms in ou pape , hey p o ide an in e es ing opic o u u e esea ch. Re e ences Addison John T., and Ped o Po ugal. 2002. “Job Sea ch Me hods and Ou comes,” Ox o d Economic Pape s 54, 505-533. An, Weihua. 2015. “Ins umen al Va iables Es ima es o Pee E ec s in Social Ne wo ks”. Social Science Resea ch Volume 50. Pages 382–394. Ba is a, Ca ia, and Gaia Na ciso. 2013. “Mig an Remi ances and In o ma ion Flows: E idence om a Field Expe imen ”, IZA Discussion Pape s 7839. Ba is a, Ca ia, and Janis Umblijs. 2014. “Mig a ion, isk a i udes, and en ep eneu ship: E idence om a ep esen a i e immig an su ey”. IZA Jou nal o Mig a ion, 3: 1-25. Ba is a, Ca ia, and Janis Umblijs. 2016. “Do Mig an s Send Remi ances as a Way o Insu ance? E idence om a Rep esen a i e Immig an Su ey”, Ox o d Economic Pape s, 68 (1): 108-130. Ba is a, Ca ia, and Ped o Vicen e. 2011. “Do Mig an s Imp o e Go e nance a Home? E idence om a Vo ing Expe imen ”. Wo ld Bank Economic Re iew, 25(1): 77-104. Ba is a, Ca ia, Julia Sei he , and Ped o C. Vicen e, 2016. “Mig a ion, Ins i u ions and Social Ne wo ks in Mozambique”, NOVAFRICA Wo king Pape . Beine, Michel, F édé ic Docquie and Çağla Özden (2011) “Dissec ing Ne wo k Ex e nali ies in In e na ional Mig a ion”, CESi o Wo king Pape : Labou Ma ke s, No. 3333. Cal o-A mengol, An oni, and Y es Zenou. 2005. “Job Ma ching, Social Ne wo k and Wo d-o -Mou h Communica ion”, Jou nal o U ban Economics 57, 500 – 522. Came on, A. Colin, and P a in K. T i edi. 2005. Mic oeconome ics: Me hods and Applica ions. Camb idge: Camb idge Uni e si y P ess. Chen, Yunsong. 2014. “Do Ne wo ks Pay O Among In e nal Mig an s in China?” Chinese Social Re iew, Volume 45, Issue 1, pages 28-54. Co es, Pa icia. 2008. "The E ec s o Low Skilled Immig a ion on US p ices: e idence om CPI da a", Jou nal o Poli ical Economy, ol. 116, n:3. Da che Lou y. 2006. “Some Con ac s A e Mo e Equal han O he s: In o mal Ne wo ks, Job Tenu e, and Wages”, Jou nal o Labou , 24, 299-318. 21 Edin, P., P.F ed iksson, and P.Aslund. 2003. "E hnic Encla es and he Economic Success o Immig an s—E idence F om a Na u al Expe imen ," Qua e ly Jou nal o Economics, Vol. 118, No. 1, pp. 329—357. G egg, Paul, and Jona han Wadswo h. 1996. “How E ec i e a e S a e Employmen Agencies? Jobcen e Use and Job Ma ching in B i ain”, Ox o d Bulle in o Economics and S a is ics, 58, 443-457. Ha is, John, and Michael Toda o. 1970. “Mig a ion, Unemploymen and De elopmen : A wo-sec o analysis”, Ame ican Economic Re iew 60: 126–42. Ke , William, and Ma in Mando . 2015. “Social Ne wo ks, E hnici y, and En ep eneu ship”, Na ional Bu eau o Economic Resea ch, Wo king Pape n. 21597. Long, Wenjin, Simon Apple on, and Lina Song. 2013. “Job Con ac Ne wo ks and Wages o Ru al-U ban Mig an s in China”, IZA Discussion Pape s 7577, Ins i u e o he S udy o Labo . McKenzie, Da id, and Hillel Rapopo . 2010. "Sel -Selec ion Pa e ns in Mexico-U.S. Mig a ion: The Role o Mig a ion Ne wo ks," The Re iew o Economics and S a is ics, MIT P ess, ol. 92(4):811-821. Munshi, Kai an. 2003. “Ne wo ks in he Mode n Economy: Mexican Mig an s in he U.S. Labo Ma ke ,” The Qua e ly Jou nal o Economics, MIT P ess, ol. 118(2), pages 549- 5999. Pa el, K ishna, and F ancis Vella. 2007. “Immig an Ne wo ks and Thei Implica ions o Occupa ional Choice and Wages” IZA Discussion Pape s 3217, Ins i u e o he S udy o Labo (IZA). Sjaas ad, La y. 1962. “The Cos s and Re u ns o Human Mig a ion”, Jou nal o Poli ical Economy 70:80. S aige , Douglas, and James H. S ock. 1997. “Ins umen al Va iables Reg ession wi h Weak Ins umen s,” Econome ica, 65, 557–86. Umblijs, Janis. 2012. “The E ec o Ne wo ks and Risk A i udes on he Dynamics o Mig a ion”, Ox o d IMI Wo king Pape s, 54/2012. Wahba, Jackline and Y es Zenou. 2005. “Densi y, Social Ne wo ks and Job Sea ch Me hods: Theo y and Applica ion o Egyp ”, Jou nal o De elopmen Economics, Volume 78, Issue 2, Pages 443-473. Heckman, James. 1979. “Sample Selec ion Bias as a Speci ica ion E o ”, Econome ica, 47: 153-161. 22 Table 1. Selec ed Desc ip i e S a is ics Table 2: Employed as he bina y dependen a iable Woold idge, J. M. 2006. In oduc o y Econome ics. A Mode n App oach. (Thi d Edi ion). Thomson Sou h-Wes e n. 24 Umblijs, Janis. 2012. “The E ec o Ne wo ks and Risk A i udes on he Dynamics o Mig a ion”, Ox o d IMI Wo king Pape s, 54/2012. Appendix: LPM (1) IV (2) Ne wo k Size 0.0288*** (0.00615) 0.113* (0.0665) Female -0.0866*** (0.0257) -0.0828*** (0.0277) Yea s o Schooling 0.0133** (0.00475) 0.0113* (0.00556) Yea s in I eland -0.000491 (0.00486) -0.00630 (0.00669) Age 0.0417*** (0.00942) 0.0433*** (0.0101) Age^2 -0.000447*** (0.000126) -0.000458*** (0.000135) Ma ied -0.0128 (0.0284) -0.0526 (0.0425) Cons an -0.361** (0.172) -0.647** (0.262) Con inen Fixed E ec s YES YES Va iables Obs. Mean S d. De . Max. Min. Main Dependen Va iables Mon hly ne wage (eu os) 957 1665 987.9 10500 100 Employed (dummy) 1481 0.627 0.484 1 0 Main Independen Va iables Ne wo k Size 1280 3.077 2.165 12 1 Female (dummy) 1491 0.541 0.498 1 0 Yea s o schooling 1483 14.69 2.798 17 0 Yea s in I eland 1489 5.348 2.863 11 0 Ma ied (dummy) 1491 0.424 0.494 1 0 Age 1491 32.59 8.025 72 18 Ins umen al Va iable S ock o immig an s in I eland (yea 2000) 1485 10004 12919 44633 2 Exclusion Res ic ions Mon hly Household Income (eu os) 1077 1128 1746 17500 0 Ha ing child en (dummy) 1491 0.465 0.499 1 0