Essays on Labor Economics and Entrepreneurship

Transcription

1 Essays n Labr Ecnmics and Entrepreneurship Item Type text; Electrnic Dissertatin Authrs Córdva Gnzález, Karina Elizabeth Publisher The University f Arizna. Rights Cpyright is held by the authr. Digital access t this material is made pssible by the University Libraries, University f Arizna. Further transmissin, reprductin r presentatin (such as public display r perfrmance) f prtected items is prhibited except with permissin f the authr. Dwnlad date 04/07/ :19:52 Link t Item

2 Essays n Labr Ecnmics and Entrepreneurship by Karina Elizabeth Córdva Gnzález A Dissertatin Submitted t the Faculty f the Department f Ecnmics In Partial Fulfillment f the Requirements Fr the Degree f Dctr f Philsphy In the Graduate Cllege The University f Arizna

3 2 THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE As members f the Dissertatin Cmmittee, we certify that we have read the dissertatin prepared by Karina Elizabeth Córdva Gnzález entitled Essays n Labr Ecnmics and Entrepreneurship and recmmend that it be accepted as fulfilling the dissertatin requirement fr the Degree f Dctr f Philsphy Rnald L. Oaxaca Price V. Fishback Tauhid Rahman Date: August 2nd, 2013 Date: August 2nd, 2013 Date: August 2nd, 2013 Final apprval and acceptance f this dissertatin is cntingent upn the candidate s submissin f the final cpies f the dissertatin t the Graduate Cllege. I hereby certify that I have read this dissertatin prepared under my directin and recmmend that it be accepted as fulfilling the dissertatin requirement. Dissertatin Directr: Rnald L. Oaxaca Date: August 2nd, 2013

4 3 Statement by Authr This dissertatin has been submitted in partial fulfillment f requirements fr an advanced degree at The University f Arizna and is depsited in the University Library t be made available t brrwers under rules f the Library. Brief qutatins frm this dissertatin are allwable withut special permissin, prvided that accurate acknwledgment f surce is made. Requests fr permissin fr extended qutatin frm r reprductin f this manuscript in whle r in part may be granted by the head f the majr department r the Dean f the Graduate Cllege when in his r her judgment the prpsed use f the material is in the interests f schlarship. In all ther instances, hwever, permissin must be btained frm the authr. Signed: Karina Elizabeth Córdva Gnzález

5 4 Acknwledgments I am very grateful t my advisr, Dr. Rnald Oaxaca fr his guidance and many valuable lessns taught frm the very first time we met. I als thank Dr. Price Fishback and Dr. Tauhid Rahman, members f my dissertatin cmmittee, fr their great advice, feedback and supprt. I benefited tremendusly frm fruitful cnversatins with them that nt nly imprved my academic wrk but als psitively impacted me as a persn. I wuld als like t thank Dr. Manuela Angelucci, wh is a great leader by example and whse wrk ethic is admirable; Dr. Rabah Amir, fr his encuragement during my first years in graduate schl; Dr. Anna Breman, Dr. Lisa Ordóñez, Dr. Martin Dufwenberg, fr their valuable supprt and advice in different stages f my research; and Richard Kiser, fr his utstanding assistance at the Ecnmic Science Labratry. I am very grateful t Dr. Albert Díaz Cayers, Dr. René Zenten, and all the fellws, schlars and staff at the Center fr U.S.-Mexican Studies at UCSD, fr an amazing experience full f research pprtunities and stimulating discussins during the academic year I als thank the staff at the Department f Ecnmics and the Office f Internatinal Students Services at the University f Arizna fr their friendly assistance and invaluable help thrugh many demanding administrative requests and bureaucratic prcesses. I gratefully acknwledge financial supprt frm the Fulbright Cmissin in Mexic, the Ministry f Public Educatin, SEP, and the Natinal Cuncil fr Science and Technlgy, CONACYT, during different stages f my graduate studies. My surce f strength is the amazing peple that surrund me. I thank frm the bttm f my heart my beautiful family, whse prayers and psitive thughts keep me safe; and my wnderful friends, wh always manage t keep me cmpany and t keep me sane despite the distance. I culd nt be here withut the lve, encuragement and supprt f my parents, Andrés Córdva Calderón and Rsa María Gnzález Maldnad, and my dear brther Víctr Andrés Córdva Gnzález. I we everything t them, and every effrt I d is t hnr the example they gave me. T my husband, William Rbert Ingersll, I am thankful beynd wrds fr standing n my side and being always there fr me with the right wrds and his thughtful actins full f endless patience, jy and lve. Finally, I want t thank Gd s presence in my life, acknwledging that in Him everything is pssible.

6 5 Dedicatin T Rssy, Andrés, Víctr and Will, whse lve and faith inspire me and supprt me thrugh everything I d.

7 6 Table f Cntents List f Figures List f Tables Abstract CHAPTER 1. Intrductin CHAPTER 2. Cllective Remittances and the 3x1 Prgram in Mexic: Lcal Labr Market Effects Intrductin Ptential Effects f Cllective Remittances n Lcal Labr Markets Channel 1: Labr Demand Channel 2: Incentives t Migrate Data and Descriptive Statistics Data Surces Descriptive Statistics Empirical Strategy Results Cnclusins CHAPTER 3. Chice and Perfrmance Under Stress: Are Men and Wmen Different? Intrductin Treatment, Experimental Design, Steps, and Data Treatments and Experimental Design Experimental Steps Data Methdlgy High Emtinal Lad and Crtisl Stress and prductivity Effect f stress n crrect answers Des stress increase anyne s prductivity? Effect f stress n attempts and mistakes Stress and entrepreneurial chice Stress and wrng chices Decmpsing the stress-induced incme lsses

8 7 Table f Cntents - Cntinued Stress-induced lsses by negative life events, baseline stress, and gender Stress and nn-maximizing chices: sme mechanisms Stress and nn-maximizing chices: pessimistic and ptimistic beliefs Stress and nn-maximizing chices: cnfusin and mistakes Stress and nn-maximizing chices: risk preferences Stress r Cgnitive Lad? Cnclusin and Discussin APPENDIX A. Full Regressin Results frm Chapter APPENDIX B. Experimental Prtcl, Instructins and Tasks frm Chapter References

9 8 List f Figures Figure 2.1. A Psitive Shift f the Labr Demand Figure 2.2. A Psitive Shift f the Labr Supply Figure 2.3. Labr Demand and Labr Supply Shift Simultaneusly Figure 3.1. Change in crtisl frm baseline levels by gender Figure 3.2. Quiz 1 and 2 crrect answers by treatment and gender Figure 3.3. Quiz 1 and 2 attempted answers by treatment and gender Figure 3.4. Quiz 1 wrng answers by treatment and gender Figure 3.5. Entrepreneurial chice by treatment and gender Figure 3.6. Nn incme maximizing entrepreneurial chice by treatment and gender Figure 3.7. Subjective - bserved prbabilities and prductivity by treatment and gender Figure 3.8. Subjective and bserved prbabilities by treatment and gender. 95 Figure 3.9. Incnsistencies chices, beliefs; Wrng recall Figure Density f risk attitudes by treatment and gender

10 9 List f Tables Table 2.1. Classificatin f Prjects Financed with Cllective Remittances thrugh Prgram 3x1 Para Migrantes Table 2.2. States, Municipalities and Husehlds that Participate in Prgram 3x1 para Migrantes Table 2.3. Municipalities in MxFLS That Participate in Prgram 3x1 para Migrantes with Prjects financed with Cllective Remittances Table 2.4. Number f MxFLS Municipalities Participating in the 3x1 Prgram and Average Amunt Invested (US Dllars), by Type f Prject. 51 Table 2.5. Descriptive Statistics f Migratin and Remittances fr Husehlds and Individuals in MxFLS Municipalities, 2002 and Table 2.6. Individual Characteristics f Men and Wmen living in a MxFLS Municipality participating in 3x1 Prgram during , by age.. 52 Table 2.7. Changes in Emplyment and Wage Rates fr Men and Wmen in Municipalities Participating in the 3x1 Prgram, by Age Table 2.8. Effect f Cllective Remittances n Emplyment and Migratin in Municipalities that participate in the 3x1 Prgram Table 2.9. Effect f Cllective Remittances n the Weekly Number f Hurs Wrked and Hurly Wage Rate in Municipalities that participate in 3x1 Prgram Table 3.1. Experimental design and sample sizes Table 3.2. Experimental sessins by date and treatment Table 3.3. Difference in means, by treatment cnditin and gender Table 3.4. Effect f high emtinal lad n change in crtisl by gender Table 3.5. Quiz 1 and Quiz 2 crrect answers by treatment and gender Table 3.6. Quiz 1 and Quiz 2 attempted and wrng answers by treatment and gender Table 3.7. Pled upgrade likelihd by treatment and gender Table 3.8. T much upgrade because f likely mistake r cnfusin by treatment and gender Table 3.9. ATT n wrng entrepreneurial chice by treatment and gender. 105 Table Lsses and Returns n Quiz 2 (US Dlls) Table Return n quiz2 and lsses frm wrng upgrades. Hetergeneity by life events and gender Table Return n quiz2 and lsses frm wrng upgrades. Hetergeneity by baseline self-reprted stress and gender Table Incnsistencies and mistakes by treatment and gender: Optimism fr Highest Upgrade

11 10 List f Tables - Cntinued Table Incnsistencies and mistakes by treatment and gender: Optimism fr Lwest Upgrade Table Incnsistencies and mistakes by treatment and gender: Differences in subjective and bserved prductivity Table Incnsistencies and mistakes by treatment and gender: Cnfused Chices and Wrng Recllectin Table Risk attitudes by treatment and gender Table A.1. Effect f Cllective Remittances n the Prbability f Being Emplyed in Municipalities that participate in 3x1 Prgram Table A.2. Effect f Cllective Remittances n the Prbability f Wishing t Migrate in Municipalities that participate in 3x1 Prgram Table A.3. Effect f Cllective Remittances n the Prbability f Participatin in the Labr Frce in Municipalities that participate in 3x1 Prgram 118 Table A.4. Effect f Cllective Remittances n the Weekly Number f Hurs Wrked in Municipalities that participate in 3x1 Prgram Table A.5. Effect f Cllective Remittances n the Hurly Wage Rate in Municipalities that participate in 3x1 Prgram Table B.1. Order f experiment tasks, by treatment type

12 11 Abstract This dissertatin is cmpsed f tw essays that relate tpics in the fields f labr ecnmics, migratin, experimental ecnmics and entrepreneurship, taking int accunt a gender perspective. The first essay examines cllective remittances, thse sent by migrants assciatins t be invested in cmmunity prjects in their hmetwns, matched by gvernmental funds thrugh the Mexican prgram 3x1 Para Migrantes. This study evaluates the effect f cllective remittances n the prbability f wanting t migrate, being emplyed and in the labr frce, and n the amunt f hurs wrked f adult men and wmen in 2002 and 2005 in Mexic. Cllective remittances have a psitive, albeit mdest, impact n the emplyment and labr frce participatin f adults in participant municipalities, but n effect n the preferences t migrate. Imprtant differences are bserved by type f prject executed and by gender and age chrt, with yunger men and wmen benefiting the mst frm investments in schls and sprts facilities. The secnd essay cnducts a series f labratry experiments t test the hypthesis that, while stress wrsens entrepreneurial chices and utcmes fr all, it des s mre fr wmen than men. Results shw that the effects f stress n chice and perfrmance are mre negative fr wmen. Experimentally-induced stress causes mre lng-lasting prductivity lsses fr wmen, and additinal lsses fr making chices that d nt maximize incme given ne s prductivity. The negative treatment effect n wmen s prductivity, chice quality, and earnings is driven by wmen wh experienced negative life events. The mechanisms that affect chices als differ by gender. Men are mre likely t present incnsistencies during a series f entrepreneurial decisins, and wmen t have inaccurate beliefs abut their perfrmance.

13 12 CHAPTER 1 Intrductin This dissertatin is cmpsed f tw essays, ne that studies cllective remittances as a ptential tl fr lcal develpment and the generatin f jb pprtunities, and ne that analyzes stress and the mechanisms thrugh which it can affect perfrmance and entrepreneurial chices. Bth essays take int accunt a gender perspective, and they can be related by their cntributin t the literature in labr and develpment ecnmics, and the design f public plicy interventins. The first essay fcus n the ecnmics f immigratin and labr ecnmics. I study cllective remittances, the amunt f mney sent by assciatins f Mexican migrants in the United States t their hmetwns in rder t be invested in cmmunity prjects, and hw they can affect cmmunity develpment, individual labr decisins, and the verall labr utcmes in Mexican municipalities. I analyze the prgram 3x1 fr Migrants, a unique strategy by the Mexican gvernment t incentivize the scial participatin f migrants clubs in cmmunity prjects. Given sme challenging characteristics f the available data, ne f the cntributins f this study is the discussin and implementatin f a series f strategies in rder t deal with identificatin challenges in a cntext f a prgram in which participants self-select t btain the treatment. In rder t d s, I study the effects f the 3x1 Prgram at the individual level, ding an intensive margin analysis f a sample f participant municipalities, and using a rich lngitudinal dataset that allws the estimatin f a fixed effects mdel that cntrls fr nn-bservable characteristics. Migrant netwrks and the remittances they send cllectively are nt a cmmn bject f research in cmparisn with research n individual migrants and family remittances. Netwrks are usually defined by gegraphic r scial criteria and they

14 13 cannt be tracked as easily as family ties. Mst studies agree n the fact that migrants netwrks facilitate and incentivize migratin, leading t a cnsequent reductin in the lcal labr supply. Hwever, if we cnsider cllective remittances, the impact f migrants netwrks n ecnmic utcmes may be different thrugh ther channels nt yet studied. My research sheds sme light n the dynamics f netwrks and remittances in a way that is relatively new. This cntributes t the understanding f a different dimensin f the migratin prcess that can be a tl fr lcal develpment with direct participatin f civil sciety, and can als help t imprve the design f prgrams in Mexic and ther cuntries with high levels f migrant participatin and remittances flws. The secnd essay presents the analysis frm a series f labratry experiments in which I study high emtinal lad and uncertainty as surces f stress, and hw they can affect entrepreneurial chices and utcmes. Entrepreneurship is defined as the prpensity t take risks fr a prfit, which depends n the entrepreneur s perfrmance that can be measured as their prductivity. High emtinal lad (anxiety, sadness, wrry) is elicited thrugh a vide clip, and uncertainty is induced by limiting the amunt f infrmatin individuals have in rder t make decisins. I test if the effects f stress n entrepreneurship ccur thrugh changes in prductivity, risk aversin, accuracy f subjective expectatins f future prductivity, and cgnitive impairment. I analyze t what extent stress makes entrepreneurial chices nn-incme-maximizing, and hw respnses t stress vary by gender. The main hypthesis t test is that, while stress wrsens entrepreneurial chices and utcmes fr all, it des s mre fr wmen than men. Results shw that stress, thrugh high emtinal lad and uncertainty, decreases prductivity, lwers entrepreneurial chices, increases nn-maximizing chices, and reduces verall prfits, especially fr wmen. Men are mre likely t present incnsistencies during a series f entrepreneurial decisins, and wmen are mre likely t have inaccurate beliefs abut their perfrmance. Wmen shw a greater prclivity

15 14 t have experienced events that culd lead t higher chrnic stress (as indicated by the fact that they have experienced parental divrce r death f clse relative, have lw-educatin father, and higher-than-median self-reprted stress and crtisl), and when this happens they are mre susceptible t these experiment s stressrs and the treatment effects are larger. These findings suggest that there are ecnmic benefits t reducing stress and uncertainty, bth in the wrkplace and utside. Mrever, given that wmen are bth subject t mre stressrs and respnd mre negatively t stress than men, the differential incidence f, and the respnse t, stress by gender may be ne determinant f why wmen are under-represented in high-pressure activities, jbs, and prfessins.

16 15 CHAPTER 2 Cllective Remittances and the 3x1 Prgram in Mexic: Lcal Labr Market Effects 2.1 Intrductin Migratin and remittances have a very imprtant rle in the Mexican ecnmy. In 2010, the number f Mexicans wrking r lking fr a jb in the US accunted fr 16.4 percent f the ecnmically active ppulatin in Mexic. 1 These migrants send mney back hme in such a way that family remittances accunt fr abut 2.05 percent f the GDP in 2010 (6.24 percent f the ttal incme in the current accunt), 2 which makes remittances the secnd mst imprtant surce f incme, just after il. The imprtance f family remittances led the Mexican gvernment t take actins t facilitate the prcess t remit and decrease its csts. A large number f Mexican migrants living in the US create clubs and assciatins that send mney t supprt cmmunity prjects in their Mexican hmetwns. 3 Increases in these cllective remittances and the imprtant prjects carried ut in different municipalities mtivated the participatin f the Mexican gvernment. The result was 3x1 Para Migrantes, an innvative prgram in which the gvernment cllabrates with migrant netwrks in rder t encurage the use f remittances t imprve living cnditins in Mexican twns. Mst studies f cllective remittances s far fcus n the structure f the migrants netwrks, their rganizatin structure, and 1 Own calculatins with data frm the Natinal Institute f Gegraphy, Statistics and Infrmatin (INEGI), fr Mexic; and the Current Ppulatin Survey fr the United States. 2 Althugh the imprtance f family remittances remains, remittances frm Mexican immigrants living in the US significantly reduced after 2008, due t the financial crisis in the US. Family remittances decreased frm a histrical maximum f USD$26 billin in 2007, t USD$21 billin in 2010, slwing recvering since then (Banc de Mexic, 2013). 3 I refer t these as migrant clubs r netwrks, but they are als cmmnly knwn as Hmetwn Assciatins (HTAs).

17 16 sme evaluatins abut the rle f the 3x1 prgram as a tl fr prvisin f public gds and demcratic participatin in disadvantaged regins. 4 In cntrast, the effect f this prgram n the Mexican labr market is the bject f study in this paper. The term cllective migrant refers t higher level and relatively permanent migrant rganizatins which functin n a binatinal basis fstering bth scial and cultural slidarity between Mexicans and their hmetwn cnstituency with mutual scial, plitical and ecnmic benefits [Delgad and Rdríguez (2001), p.747]. Cnsequently, the idea f cllective remittances implies a jint fund that migrants assciatins use t finance cmmunity prjects [... ], leading the migrants t lbby at different gvernment levels in rder t execute cmmunity wrks that histrically have been ignred, and stimulating the evaluatin f public plicy and accuntability practices [Mctezuma and Pérez (2006), p. 125, wn translatin]. 5 In Mexic, these cllective remittances have cnstantly increased ver the last 30 years, in such a way that migrants clubs started t gain plitical influence in their hmetwns in Mexic. At an early stage these cllective remittances were mstly directed t scial and religius activities. Mre recently, they have been fcused n activities related t generatin and imprvement f health and educatinal services, urban infrastructure, access t electricity, schlarships funds and prductive prjects in rural areas. The increasing amunt f cllective remittances in the last years attracted the attentin f lcal gvernments during the decade f 1990, leading t the creatin f mutual funds designed t create bigger cmmunity prjects. The Mexican state f Zacatecas was the first f its type. A prgram called Un pr Un (One fr One) was created in 1992, in which municipal gvernments matched funds with the cllective remittances received by the cmmunity in rder t finance prjects in the 4 See fr example: Duquette-Rury (2010), Aparici and Meseguer (2011, 2012), Delgad and Rdríguez (2001), Fx and Bada (2008), Fernández, García and Vila (2006), and Mctezuma (2005). 5 Althugh these definitins are based n the Mexican migratin experiences, imprtant examples f cllective remittances in different hst and hme cuntries are mre cmmn nw (see Schüttler (2008) fr a discussin f evidence frm Mexic, Mrcc, and the Philippines. See Burgess (2012) fr a cmparisn f cllective remittances in Mexic and El Salvadr).

18 17 regin. The plitical influence f these migrants clubs grew quickly, and the state gvernment started t participate in this prgram as well, knwn by then as Ds pr Un (Tw fr One). In 2002, the Natinal Ministry f Scial Develpment, SEDESOL, incrprated this prgram as part f the Natinal Develpment Plan f President Fx s administratin. Since then, this strategy is knwn as Prgrama 3x1 para Migrantes, r Three fr One Matching Funds Prgram Fr Migrants, indicating that the amunt f cllective remittances is matched with a similar amunt f mney frm the municipal, state and federal levels f gvernment, under the cnditin that the mney has t be invested in cmmunity prjects. The rules f peratin in this prgram are straightfrward since there is n prir selectin frm the gvernment t participate in it. The interest t participate must cme frm the cmmunity and the migrants club, and they have t apply fr funds t the municipal gvernment first and then fr the state and federal funds thrugh SEDESOL. These three lcal actrs (cmmunity, migrants and municipal gvernment) decide the terms f implementatin f the prject thrugh a designated cmmittee. Since municipalities self-select t jin the prgram, it might be the case that thse that participate are mre likely t have mre resurces and initial infrastructure t rganize and start prjects, althugh the Mexican gvernment is seeking t increase participatin in mre marginalized cmmunities as a way t help develpment in pr areas. 6 In 2003, the secnd year f the prgram s peratin, migrants clubs invested USD$8.9 millin in their hmetwns. These cllective remittances almst dubled in 2006 when migrants cntributed t the 3x1 Prgram with USD$16.9 millin. The states in which migrants hmetwn assciatins cntributed the mst are the nes with a strnger traditin f migratin (Zacatecas, Jalisc, Michacan), where migrants are histrically mre rganized. Hwever, the 3x1 Prgram has incentivized 6 A detailed histry f the develpment f the prgram can be fund in Fernández (2006). See Aparici and Meseguer (2012) fr a nice shrt summary f the implementatin f this prgram.

19 18 the creatin f new migrants assciatins: accrding t the data available in the directry f rganizatins registered in the Institute f Mexicans Living Abrad, 65 percent f the 2,309 hmetwn assciatins listed were created between (IME, 2013). 7 Despite this cntinuus increase in the participatin and cllabratin f rganized migrants, cllective remittances invested thrugh the 3x1 Prgram are still less than 1 percent f family remittances sent t Mexic. 8 In terms f federal transfers t municipalities, cllective remittances represented less than 5 percent f the ttal amunt f the transfers made frm the federal gvernment t municipalities via funds fr scial infrastructure and municipality s strengthening. 9 Nevertheless, the scpe f the prjects funded thrugh cllective remittances can be larger thanks t the matching nature f the 3x1 Prgram. There are numerus studies that fcus n the effect f remittances that migrants send t their families in their cmmunities f rigin. Remittances, as a surce f higher incme, may have psitive effects n cnsumptin and investment decisins. They may relax the current budget cnstraint f the migrants family r may be the nly way t vercme credit cnstraints fr imprtant investment decisins that require a cnsiderable amunt f mney. In this sense, remittances can have an imprtant rle fr the ecnmic develpment f the hme cuntries. On the ther hand, sme studies shw that family remittances may have negative effects n the number f hurs wrked by the members f the husehld wh receive remittances. Others find that the psitive effect f remittances is nt big enugh t cunteract the influence f migratin incentives fr kids, wh reduce their schling attendance in rder t 7 It is wrth nting that the rganizatins vluntary register themselves in this list, s it is pssible that mre rganizatins exist but pt nt t be registered. 8 Cllective remittances represented percent f family remittances in 2003, and percent in In Zacatecas, the state with higher inflws f remittances, cllective remittances represented 5 percent f the ttal amunt received in 2006 frm the Fnd de Aprtacines para la Infrastructura Scial Municipal and Fnd de Aprtacines para el Frtalecimient de ls Municipis. In Veracruz and Hidalg, the states with the smaller levels f cllective remittances, they represented 0.01 and 0.16 percent f the transfers frm these funds, respectively.

20 19 migrate. 10 Regarding the effect f migratin and family remittances n labr supply at the hme cuntry, different studies shw ambiguus results. 11 Cllective remittances are pssible when migrants netwrks exist. Netwrks are usually defined by gegraphic r scial criteria (fr example, if migrants cme frm the same hmetwn, r if they are frm the same religius grup). Migrants netwrks facilitate migratin. They prvide infrmatin t ptential migrants, they are a safety net in the hst cuntry t thse wh migrate fr the first time, and they may even help t finance the trip t leave the hme cuntry. Netwrks incentivize migratin thrugh these channels, leading t a cnsequent reductin in the lcal labr supply. 12 Nevertheless, the effect f migrants netwrks n labr markets may be quite different thrugh ther channels when cnsidering the use f cllective remittances fr lcal investment. When a municipality is receiving cllective remittances that are invested in prductive prjects, this may lead t the imprvement f ecnmic cnditins in the cmmunity with the added effect that makes it mre appealing t stay in the municipality and get a jb there, as an alternative t migrate. Then we may see a reductin in the levels f migratin in the municipality and therefre sme effects in the labr market, either in the number f jbs available, r decisins n the number f hurs wrked. The impact f migrants assciatins thrugh cllective remittances n lcal labr markets, t my knwledge, has been nt analyzed yet, basically due t the lack f data. Hwever, in the Mexican experience, availability f data is pssible thanks t the recrds f the 3x1 Prgram frm a previus qualitative evaluatin. I match 10 Fr family remittances and labr supply in Mexic, see Airla (2008), Amued-Drantes and Pz (2006) and Hansn (2007). Fr remittances and their impact n the generatin f family businesses r micrenterprises, see Wdruff (2007) and Wdruff and Zenten (2001). Fr migratin incentives, remittances and children schling attendance see McKenzie and Rapprt (2006). 11 Cfr. Hansn (2007), and Amued-Drantes and Pz (2006), which als have divergences due t the methdlgy used. See Görlich, Mahmud and Trebesch (2007), fr an analysis f the effect f remittances n labr supply, husewrk and schling in husehlds with at least ne migrant, with infrmatin frm Mldva. 12 See Kla and Özden (2008) wh studied labr supply f husehlds with migrant members, taking int accunt the netwrk effect measured by religius attachment in India.

21 20 this municipality-level database with individual and husehld-level data frm the Mexican Family Life Survey (MxFLS) in rder t get infrmatin abut sciecnmic characteristics, emplyment status, family remittances and migratin experiences. I fcus this study n the labr market fr men and wmen f age 25 and lder, given that mst adults at this stage are less likely t be enrlled in schl, and mre likely t participate in the labr frce. Since the perid f study f this prgram is shrt and fr years when the prgram was starting, it is mre prbable t expect a quick respnse, if any, frm thse wh have a mre active presence in the labr market, at least in the shrt run. Sme issues arise in this analysis, thugh. Given the nature f the 3x1 Prgram, municipalities self-select in rder t participate. Labr frce participatin, even cntrlling fr individual and husehld characteristics, might imply prblems f endgeneity. Different specificatins and the use f a rich panel database in this case are useful fr vercming these cncerns. Taking this int accunt, I find that cllective remittances thrugh participatin in the 3x1 Prgram have an ambiguus impact n the labr market in the municipalities with prjects financed by these funds. The ttal per capita amunt invested in municipality prjects thrugh the 3x1 Prgram has a psitive effect n the prbability f being emplyed and participating in the labr frce, but n effect in the desire t migrate. Additinally, while men and wmen respnd similarly t the investments and prjects executed in their cmmunities by the prgram, they shw slightly different patterns that are als affected by age. Fr example, when the municipality invests cllective remittances in urbanizatin prjects, wmen and lder men are mre likely t be emplyed, while yunger men are nt, but bth yunger men and wmen are mre likely t be willing t mve ut f their twn. Meanwhile, the ppsite happens fr the lder grup. Municipalities als seem t benefit mre frm investments in schls and sprts facilities. These increase the prbability f being emplyed fr all grups, and slightly reduce the willingness t mve ut fr yunger men and wmen. While verall these results are mdest, this

22 21 is a first step taken n a better understanding n hw cllective remittances frm migrants assciatins can strengthen scial capital in a way that can be useful as a ptential tl fr lcal develpment. 2.2 Ptential Effects f Cllective Remittances n Lcal Labr Markets The 3x1 Prgram matches funds frm the municipal, state and federal gvernments t the amunt f cllective remittances prvided by the migrants assciatin in a given municipality. The prjects financed by this prgram must yield a benefit fr the cmmunity. There are 21 different types f prjects carried ut since the implementatin f the prgram. I classify these prjects in fur categries: thse related t urbanizatin f a municipality, prjects directed t health and scial infrastructure, thse fr schls and sprts facilities and prjects fcused n the imprvement f prductin and prductivity (see Table 2.1 fr the cmplete classificatin). The nature f each prject and the amunt invested in it may affect the lcal labr market thrugh tw channels: decreasing the incentives t migrate and increasing the labr demand. Each channel implies different hyptheses I test with the available data Channel 1: Labr Demand The 3x1 Prgram may increase labr demand in tw ways. First, prjects need wrkers t be carried ut, s the prgram itself generates emplyment pprtunities the mment the prgram is implemented. Secnd, nce finished, the prjects themselves may be a new surce f emplyment in the municipality, especially since an imprtant number f prjects have the bjective f imprving prductin r prductivity. Either way, cllective remittances that finance prductive prjects in municipalities may increase the demand fr labr and reduce unemplyment. If we cnsider a labr market in which there wuld nt be a clear-market equilibrium, as wuld be the case

23 22 in an ecnmy like Mexic, in the shrt run we have sticky wages abve the equilibrium wage and the effect f cllective remittances in the labr demand will nly affect the levels f emplyment, but nt the current wages. 13,14 Figure 2.1 represents this situatin, where we bserve a psitive shift f the labr demand. This is the first hypthesis t test empirically: H 1 0: Cllective remittances, thrugh the prjects financed by the 3x1 Prgram, increase the prbability f being emplyed. Anther ptential effect is that demand fr labr increases in such a way that we may see changes in the number f hurs wrked in a given frame f time. There are different causes that make this scenari t happen. One pssibility is t have an excess f labr demand. In this case previusly unemplyed peple wuld get a jb but thse already emplyed may have incentives (r may be pushed) t wrk mre hurs. Anther pssibility is that sme wrkers may be underemplyed r that there are sme cnstraints in the labr market that made nt pssible fr them t wrk their desired number f hurs. 15 If the prgram, thrugh the prjects it finances, allws these cnstraints t disappear, thse underemplyed peple may wrk mre 13 One may think that the availability f new jb pprtunities wuld attract peple wh are currently ut f the labr market s this wuld increase the labr frce participatin. Hwever, since I am assuming that wages are sticky, this shift in labr demand des nt alter wages. If peple ut f the labr frce d nt want t wrk because their reservatin wage is higher than the market wage, the availability f jbs is nt ging t alter their decisin t stay ut f the labr frce. Nevertheless, we may expect an effect n the labr frce participatin nce we cnsider hw cllective remittances invested in prductive prjects may affect the incentives t migrate (see sectin 2.2). 14 Data frm the Natinal Survey n Occupatin and Emplyment (ENOE) reprts unemplyment rates f 2.52 (men) and 3.49 (wmen) percent in 2002, and 3.43 (men) and 3.64 (wmen) percent in If we cnsider thse individuals unemplyed and thse ut f the labr frce but willing t wrk divided by the ppulatin lder than 14 years ld, these adjusted unemplyment rates are 5.48 and 7.67 percent fr men and wmen respectively in 2002, and 6.63 and 9.39 percent fr men and wmen respectively in 2005 (wn calculatins using available data frm INEGI). 15 The underemplyment rate (ttal underemplyed ppulatin divided by the ttal amunt f emplyed peple) are 6.83 (men) and 4.66 (wmen) percent in 2002, and 8.45 and 5.94 percent fr men and wmen respectively in 2005 (wn calculatins using data frm INEGI).

24 23 hurs nce the prgram is implemented in their municipality. In any f these tw ptential cases we can expect an increase in the number f hurs wrked. A first explratin f this idea wuld be develped in sectin 4. Next sub-sectin discusses an alternative channel by which cllective remittances can affect these indicatrs Channel 2: Incentives t Migrate Cllective remittances are invested in cmmunity prjects that are means t imprve scial and prductive infrastructure. As a cnsequence f these imprvements, there is an increase in the amenities available at the municipality. It culd als be the case that there are mre jb pprtunities, as discussed abve. The municipality is nw a mre attractive place, s there is a reductin in the incentives t migrate. Ptential migrants may cnsider pstpning their leaving r even feel discuraged t migrate at all. If mre peple are staying in the municipality, this implies mre peple are available t wrk there: lcal labr supply increases in the shrt run. Additinally, the new pprtunities might encurage sme f thse individuals ut f the labr frce t lk fr a jb again, s labr supply might increase, even if there is n change in migratin patterns. Since ne categry f the prjects financed by cllective remittances includes imprvement and cnstructin f schl facilities, we may think that fr sme peple the imprved access t educatin wuld lead them t leave the labr frce in rder t cntinue their studies. Hwever, given that I fcus my analysis n peple ver the age f 24, it is nt prbable that they will be interested in cming back t schl. Mrever, as lng as the individuals frm this sample are mre likely t have children and/r being the head f a husehld, it is mre attractive fr them t stay in their municipality if there are imprtant imprvements in educatin fr their children. Therefre, we may expect that cllective remittances decrease the incentives t migrate and thus increases labr frce participatin. These are the last tw hypthe-

25 24 ses I test. H 2 0: Cllective remittances, thrugh the prjects financed by the 3x1 Prgram, make it mre attractive t stay in their hmetwns and decrease the desire t migrate. H0: 3 Cllective remittances, thrugh the prjects financed by the 3x1 Prgram, incentivize mre peple t lk fr a jb, either because they decide t stay in their hmetwns r because they are encuraged by new pprtunities and amenities, increasing the prbability f participating in the labr frce. Figure 2.2 depicts the psitive shift f the labr supply as a cnsequence f the effects summarized in hyptheses 2 and 3. Overall, I expect these prjects t increase jb pprtunities (either because peple are needed t carry ut these prjects r because nce the prjects are dne they are a new surce f emplyment), and t increase the amenities in the cmmunity (since an imprtant part f these prjects are related with the imprvement f rads and schls, access t electricity, and cnstructin f parks). Therefre, at the end, these prjects increase the labr demand, mre jbs are nw available, there are fewer incentives t migrate and mre peple are willing t participate in the labr frce. The effect n the prbability f unemplyment in this sense is ambiguus: cmmunity prjects may generate mre jbs and thus increase the prbability f emplyment, but the decrease in the incentives t migrate implies that mre peple staying in the cmmunity makes cmpetitin fr a jb harder. Bth the supply and demand curves are likely t shift simultaneusly. Which effect is higher at the end is an empirical questin. This situatin is illustrated in Figure 2.3. The next sectins present the data and the strategies t identify these mechanisms.

26 Data and Descriptive Statistics Data Surces Data abut cllective remittances were btained frm the 2006 evaluatin 16 f the prgram 3x1 para Migrantes, implemented by SEDESOL, the Mexican Ministry fr Scial Develpment. This database cntains infrmatin fr all the prjects financed by cllective remittances and gvernment funds in 18 states and 735 municipalities between 2002 and 2006, including ttal amunt invested in each prject, and the nature f the prject, say, urbanizatin (bring access t electricity, redesign f streets), imprvement f schls and sprts facilities, health and scial infrastructure (supprt fr hspitals/clinics, scial assistance, cmmunity services, renvatin f histrical sites) and prjects related t imprvement f prductivity (purchase f farm machinery, tls, cnnectin f rural rads t freeways, irrigatin systems, etc). I cmplement this infrmatin with data fr states and municipalities. Grss Dmestic Prduct (GDP) at the state level was btained frm INEGI, the Natinal Institute f Statistics, Gegraphy and Infrmatin. The index f marginalizatin 17 at the municipal level was prvided by CONAPO, the Natinal Cuncil f Ppulatin. Electral data t estimate plitical preferences by municipality was btained frm IFE, the Federal Electral Institute. Data at the individual and husehld level cme frm the Mexican Family Life Survey (MxFLS), a lngitudinal database whse baseline is frm 2002 and its secnd wave infrmatin was cllected in The baseline includes 35,667 individuals in 16 The bjectives f the 2006 evaluatin were t examine the prgram s design, the cverage, peratin and basic results f the prgrams executin, but nt an analysis at the micr level the way I am prpsing here. 17 The Index f Marginalizatin cnsiders the percentage f illiterate ppulatin lder than 15 years ld, the percentage f ppulatin lder than 15 years withut elementary schl, percentage f ppulatin living in dwellings withut tilet, withut electricity, withut access t water, with sme level f vercrwding, with flr f earth, in lcalities with less than 5,000 inhabitants, and the percentage f emplyed ppulatin living n an incme lwer than the equivalent f twice the minimum wage.

27 26 8,440 husehlds frm 136 municipalities and 16 states in rural and urban Mexic. The secnd wave tracked the same individuals and husehlds three years later and included all the new-brn babies and new members added t the husehld during that perid f time. The sample f interest fr this study takes int accunt nly the riginal husehlds with their riginal members frm the 2002 baseline, present in bth 2002 and 2005 with cmplete infrmatin. This leaves us with 32,386 individuals in 7,570 husehlds. The final wrking sample is restricted t thse adults ver the age f 24 in 2002 wh were present in bth years 2002 and When I restrict the analysis t nly thse municipalities that participated in the 3x1 Prgram, the final wrking database cntains 5,620 individuals frm 2,866 husehlds. The use f the MxFLS data allws us t have a detailed picture f individual and husehld characteristics which are helpful fr the empirical strategies fllwed t test the hyptheses discussed abve. MxFLS gives us rich infrmatin abut age, educatin, ethnicity (if the individual speaks an indigenus language), emplyment histry, preferences t migrate, and migratin experience. Related t husehlds, there is enugh infrmatin t generate imprtant cntrl variables, including the number f members in the husehld (husehld size), value f assets, nn-labr incme, access t frmal credit and savings institutins (as a measure f financial cnstraints), migratin experience in the husehld, family remittances, and ecnmic and health shcks. 18 MxFLS data is merged at the municipality level with data frm the 3x1 Prgram and the rest f municipality and state-level variables Descriptive Statistics Cllective remittances had existed fr years befre the prgram 3x1 para Migrantes, and the Mexican state f Zacatecas is the paradigmatic example f the rganizatin f 18 Shcks in MxFLS are reprted at the husehld level, and they can be either the death f a husehld member, illness r serius accident that required hspitalizatin f any husehld member, unemplyment r business failure f any husehld member, hme r business lss due t any natural disaster, ttal lss f crps r lss, and rbbery r death f livestck.

28 27 migrants clubs fr the develpment f their hmetwns. Als, Zacatecas is the state with the highest value in the index f migratin intensity. 19 Althugh this state is nt included in the MxFLS sample, we still have Michacan, which is classified as the state with the secnd highest value in the index f migratin intensity, and Jalisc, the state with the secnd highest vlume f cllective remittances invested thrugh the prgram. Table 2.2 shws the distributin f Mexican states and municipalities that participate in the 3x1 Prgram: 27.5 percent f the municipalities in Mexic (in 26 ut f 32 states) had at least ne prject financed with cllective remittances during the perid With respect t the sample f interest, percent f the municipalities present in the MxFLS sample participated in the 3x1 Prgram during this perid f time. Taking int accunt the cumulative number f prjects in the prgram during the whle perid , these 52 municipalities in the MxFLS sample participated in the prgram with a ttal f 351 prjects, with ne municipality frm Nuev Len registering 111 prjects in The prjects financed by cllective remittances thrugh the prgram 3x1 para Migrantes are distributed in small rural villages f less than 2,500 inhabitants (49 percent) and urban areas (51 percent). The average amunt invested per municipality by cllective remittances thrugh the prgram was USD$120,760 in 2002 and USD$245,250 in Internatinal migratin t the US is a cmmn experience fr an imprtant share f Mexican husehlds. Table 2.5 shws this infrmatin at the husehld and the individual level in municipalities accrding t their participatin in the 3x1 Prgram. Nte that there is a clear, strnger migratin and remittances experience in thse 19 Index f Migratin Intensity includes infrmatin frm the Natinal Census, taking int accunt the percentage f husehlds wh received remittances, the percentage f husehlds with emigrants living in the US during , percentage f husehlds with circular migrants during , and the percentage f husehlds with migrants wh came back frm the US during See Table 2.3 fr the ttal amunt f prjects fr each year, and the accumulated number f prjects ver all the perid see Table 2.4 fr the average amunt invested by type f prject.

29 28 husehlds that live in participant municipalities. Of the 4,704 husehlds in the MxFLS sample in cmmunities that d nt participate in the 3x1 Prgram, 38.8 percent had at least ne relative living in the US in 2002 and Fr husehlds in participating cmmunities the percentage is 60.4 and 56.2 fr 2002 and 2005, respectively. Arund percent f the nn-participant husehlds had a current member f the husehld with a previus migratin experience, either shrt term (mre than 1 mnth, but less than 12 mnths) r lng term (stayed in the US fr mre than ne year, but returned t his/her hmetwn back in Mexic). These numbers are higher fr husehlds in municipalities that are beneficiaries f the prgram, ranging frm 0.9 t 5.2 percent. The percentage f husehlds that receive remittances is almst duble in participant municipalities in bth years, and they als receive a higher amunt f such transfers, which represent mre than a quarter f their labr incme and arund a half f their nn-labr incme. Men and wmen f age 25 and lder als shw a slightly higher prevalence f migratin experiences when they live in participant municipalities. Regardless f their rigin, mre men than wmen are willing t migrate. Men and wmen living in nn-participant municipalities are mre likely t express a desire t migrate ut f the municipality and f them, arund 8 and 12 percent (as ppsed t 16 and 21 percent in participant municipalities) want t g t anther cuntry (mstly the US), and thse wh wuld leave the cuntry will d s in rder t get a jb. We can see this fact as an indicatr f hw internatinal migratin is related with emplyment decisins. 2.4 Empirical Strategy In this sectin I prpse different specificatins in rder t test the hyptheses psited in sectin 2, and then I discuss sme issues cncerning ptential surces f endgeneity and hw t deal with them in this cntext and with the available data.

30 29 The first cncern abut these data is the nature f the 3x1 Prgram that implies, by design, a characteristic f self-selectin. The municipalities rganize with the lcal migrants netwrks t prpse a prject and an agreement f hw t execute it and then ask fr the gvernment s cllabratin. If thse municipalities participate because they have better scial capital that makes it easier t rganize and plan these prjects, r if peple in thse cmmunities are mre entrepreneurial, the effect f the cllective remittances thrugh the 3x1 Prgram wuld be verestimated. An increase in the emplyment pprtunities r a higher number f amenities due t the prgram may be driven by thse particular characteristics f the municipality and nt nly due t the investment made with the cllective remittances. The analysis by Aparici and Meseguer (2012) shws that in fact there is self selectin bias in the sample f municipalities that participate in the 3x1 Prgram. Pr municipalities are mre likely t participate, but the amunt f mney invested is negatively crrelated with the index f pverty. Similarly, the level f migratin intensity and the ppulatin size are gd predictrs f prgram participatin. There are als sme partisan effects: thse states and municipalities ruled by an authrity frm the Natinal Actin Party, PAN (the plitical party f the federal gvernment frm 2000 t 2012), have a higher prbability f participatin in the 3x1 Prgram, althugh they d nt usually get mre mney nr carry ut mre prjects. Participatin in this prgram is nt randmized. Mrever, the previus cnclusins by Aparici and Meseguer (2012) give us sme surces f selectivity bias. I have tw ptential ways t treat this prblem. First, given this surce f endgeneity, it is nt pssible t d an extensive margin analysis, that is, I cannt cnsider the whle MxFLS sample f 136 municipalities if I cannt cntrl first fr the self-selectin prblem. Therefre, I restrict my study t the 52 municipalities that participated in the 3x1 Prgram during with at least ne prject carried ut during that perid. My restricted intensive margin analysis will fcus n the effect f the cllective remittances n the labr market in thse municipalities that participate