Economics of corruption and crime: an interdisciplinary approach to behavioral ethics / by Eugen Dimant, M.Sc. in Business Sciences; M.Sc. in International Economics

ECONOMICS OF CORRUPTION AND CRIME: AN INTERDISCIPLINARY APPROACH TO BEHAVIORAL ETHICS A DISSERTATION SUBMITTED TO THE FACULTY OF BUSINESS ADMINISTRATION AND ECONOMICS UNIVERSITY OF PADERBORN IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF ECONOMIC SCIENCES - DOCTOR RERUM POLITICARUM - GRADE: SUMMA CUM LAUDE BY EUGEN DIMANT DATE OF BIR TH : 10 SEP TE MBER 1986 PLACE OF BIR T H: CHIȘINĂU, MO LD OV A M.SC. BUSINE SS SCIE N CE S : WITH HIG H DI ST IN C TION & VA L ED I CT OR I A N , 2014 M.SC. INTER NA T IO NAL EC ON O MI C S: WIT H HIG H DIS T INCTION & VA L ED I C T OR I AN , 2012 DECEMBER 2015 ii DEDICATED TO MY PARENTS GALINA AND LEONID DIMANT AND MY BELOVED ONES. iii ACKNOWLEDGMENTS This disse a ion would ha e no been possible wi hou he pe sis en and subs an ial sup- po , encou agemen , and assis ance o people om bo h he academic and social sphe e, including my doc o al supe iso and dis inguished commi ee membe s as well as my am- ily and nume ous iends. I begin by exp essing my deepes g a i ude and app ecia ion o my doc o al supe iso P o- esso Bu kha d Hehenkamp, whose encou agemen , cons uc i e eedback, and suppo made his disse a ion possible. He ully suppo ed me wi h help ul academic and pe sonal ad ice, and helped me pe se e e h ough my p ojec s. His in e es in bo h expe imen al economics and beha io al e hics was con agious, and inspi ed my line o esea ch. He en- cou aged my de elopmen by p o iding me ample eedom and a conduci e en i onmen . Fo hese, and many o he easons, I am e y indeb ed o e e y hing Bu kha d has done o me. I would also like o hank P o esso Tim K iege who has been an inspi a ion since my Bach- elo s. He spa ked my in e es in economics, has been he supe iso o bo h my Bachelo ’s and Mas e ’s hesis, and a g ea colleague and collabo a o on nume ous o my esea ch p ojec s e e since. Tim is also he eason why I decided o s ay in academia. Beyond es ab- lishing a p o essional ela ionship, Tim always had a sympa he ic ea o all so s o pe sonal and academic ques ions along he way, o which I am e y hank ul. In addi ion, I would like o hank he emaining membe s on my commi ee, namely P o es- so René Fah , P o esso Luigi Mi one, and P o esso Ul ich Schmid o ha ing igge ed my in e es in expe imen al economics, as well as being g ea colleagues and inspi a ions o my wo k. I am also e y g a e ul o he chance o de elop my skill se and lea n om he bes du ing my isi ing esea che posi ions a a ious US ins i u ions. In pa icula , I wan o exp ess my deepes app ecia ion o P o esso Daniel House , P o esso Law ence Lessig, P o esso C is ina Bicchie i, and P o esso Ga y Bol on o hos ing me a he In e disciplina y Cen e o Economic Science (ICES) a he Geo ge Mason Uni e si y, he Beha io al E hics Lab (Be- Lab) a he Uni e si y o Pennsyl ania, he Edmond J. Sa a Cen e a Ha a d Uni e si y, i and he Cen e and Labo a o y o Beha io al Ope a ions and Economics (LBOE) a he Uni- e si y o Texas a Dallas, espec i ely. Finally, ye impo an ly, I would like o hank my pa en s who ha e always been ex emely suppo i e o all o my endea o s and encou aged me o ollow my passions. Thank you o playing he u mos impo an ole in my li e and making me who I am. Since e hanks a e also due o my colleagues and belo ed iends who ga e ad ice and suppo and su e ed h ough hou s o my mu e ing abou esea ch. In pa icula , I would like o highligh Ma - ga i a S aschewski’s con ibu ion, as she had o endu e a ious esea ch- ela ed con e sa- ions and s o ies o e he las couple o yea s, as well as o epea edly p oo eading my esea ch. All o you ha e played an in eg al ole in my pe sonal and schola ly de elopmen . This disse a ion has been an equally a duous and ewa ding jou ney, and I am g a e ul o each and e e y one who was pa o his enchan ing en u e. CONTENTS Acknowledgmen s......................................................................................................................................... iii Con en s.......................................................................................................................................................... In oduc ion .................................................................................................................................................... 8 Chap e 1: Su ey ................................................................................................................................... 12 The Na u e o Co up ion: An In e disciplina y Pe spec i e .................................................. 12 1.1 In oduc ion ............................................................................................................... 13 1.2 His o y o Co up ion and Co up ion Resea ch......................................................... 15 1.3 Face s and De e minan s o Co up ion ..................................................................... 17 1.3.1 In e nal Wo ld – Ra ional Choice & Beha io al Pe spec i e...................................... 19 1.3.2 Meso Wo ld—Sociological & C iminological Fac o s ................................................. 22 1.3.3 Ex e nal Wo ld—Economic, Legal, Poli ical, His o ical and Geog aphical Fac o s 25 1.3.4 In e disciplina y Pe spec i e and Empi ical Findings ............................................... 28 1.4 Conclusion ................................................................................................................. 29 Chap e 2: Empi ics................................................................................................................................ 32 2.1 The E ec o Co up ion on Mig a ion, 1985-2000 ...................................................... 32 2.1.1 In oduc ion ............................................................................................................ 33 2.1.2 Da a and Me hodology ............................................................................................ 34 2.1.3 Empi ical esul s ...................................................................................................... 37 2.1.4 Conclusion............................................................................................................... 40 2.2 A C ook is a C ook … Bu is He S ill a C ook Ab oad? On he E ec o Immig a ion on Des ina ion-Coun y Co up ion ................................................................................. 42 2.2.1 In oduc ion ............................................................................................................ 43 2.2.2 Theo e ical Conside a ions ..................................................................................... 45 2.2.3 Empi ical Analysis ................................................................................................... 49 2.2.4 Da a ......................................................................................................................... 51 2.2.5 Empi ical Resul s ..................................................................................................... 57 2.2.6 Conclusion............................................................................................................... 71 Chap e 3: Expe imen s........................................................................................................................ 73 3.1 On Pee E ec s: Beha io al Con agion o (Un)E hical Beha io and he Role o Social Iden i y ........................................................................................................................ 73 3.1.1 In oduc ion ............................................................................................................ 74 3.1.2 Backg ound and Cou se o In es iga ion ................................................................. 80 3.1.3 The Concep ual F amewo k o Beha io al Adap a ion ............................................ 85 i 3.1.3.1 The Mechanism o Social In e ac ion and Challenges o Measu ing i s E ec s ... 85 3.1.3.2 Why Unde s anding Beha io al Adap a ion Ma e s .............................................. 90 3.1.3.3 Wha We Know: On Pee s, Beha io al Adap a ion, and Neighbo hood E ec s ... 92 3.1.4 D i e s o Beha io al Con agion: An In e disciplina y Pe spec i e .......................... 98 3.1.4.1 Concep s in Economics .............................................................................................. 99 3.1.4.2 Concep s in (Social) Psychology ............................................................................. 106 3.1.5 The Role o Social Iden i y and Obse ed (Un)E hicali y in Beha io al Con agion 111 3.1.5.1 Concep ual In oduc ion .......................................................................................... 111 3.1.5.2 Theo e ical Model ..................................................................................................... 115 3.1.6 The Expe imen ..................................................................................................... 120 3.1.6.1 Expe imen al Design and P ocedu e ...................................................................... 120 3.1.6.2 Hypo heses ................................................................................................................ 124 3.1.6.3 Resul s and Discussion ............................................................................................. 125 3.1.7 Lessons Lea ned: Policy Implica ions .................................................................... 135 3.1.8 Conclusion and Ou look ........................................................................................ 138 3.1.9 Appendix ............................................................................................................... 140 A: Da a O e iew, Robus ness Checks, and Addi ional Resul s ............................... 140 B: Social Iden i y S a emen s ....................................................................................... 148 C: Expe imen al Ins uc ions........................................................................................ 149 D: Sc eensho s o Decision Sc eens............................................................................. 152 3.2 Tax E asion Re ised: Su p ising Expe imen al E idence on he Role o P incipal Wi ness Regula ions and Di e ences in Gende A i udes ........................................ 156 3.2.1 In oduc ion .......................................................................................................... 157 3.2.2 Expe imen al Design ............................................................................................. 158 3.2.3 P edic ions and Hypo hesis ................................................................................... 160 3.2.4 Resul s and Discussion .......................................................................................... 161 3.2.5 Conclusion............................................................................................................. 166 3.2.6 Appendix ............................................................................................................... 167 Bibliog aphy ............................................................................................................................................... 168 ii Con i ma ion o co-au ho ship: “Wi h my signa u e, I con i m ha Eugen Diman has con- ibu ed a leas he equi alen e o o 1/n o he pape ha I ha e co-au ho ed (wi h n de- no ing he numbe o au ho s on he espec i e pape ).” Chap e P ojec Name To al Numbe o Au ho s Name o Co- Au ho (s) Signa u e o Co-Au ho s Published in a Pee -Re iewed Jou nal1 1 The Na u e o Co - up ion: An In e dis- ciplina y Pe spec i e 2 Tho ben Schul e  2.1 The E ec o Co up- ion on Mig a ion, 1985-2000 3 Tim K iege  Daniel Meie ieks 2.2 A C ook is a C ook … Bu is He S ill a C ook Ab oad? On he E - ec o Immig a ion on Des ina ion-Coun- y Co up ion 3 Tim K iege  Ma ga e e Redlin 3.1 On Pee E ec s: Be- ha io al Con agion o (Un)E hical Beha io and he Role o So- cial Iden i y 1 - - 3.2 Tax E asion Re ised: Su p ising Expe i- men al E idence on he Role o P incipal Wi ness Regula ions and Di e ence in Gende A i udes 3 Johannes Buckenmaie Luigi Mi one 1 A he ime o his disse a ion’s submission. 8 INTRODUCTION Due o i s ad e se na u e, he s udy o illici beha io has aken a cen e s age in cu en economic esea ch. Illici beha io can ake a ious o ms, such as co up ion, aud, and ax e asion, bu hey a e all gene ally de imen al o bo h socie y and economy, and hampe ou social li es. Unde s anding he d i e s o illici beha io and how hey in e ela e wi h ins i u ional ac o s no only d aws a clea e pic u e o how and why indi iduals beha e in illegi ima e ways, bu also how hey acili a e he c ea ion and implemen a ion o mo e e - icien and e ec i e policy measu es. These mo i a ions a e a he hea o his disse a ion. The collec ion o scien i ic wo ks in his disse a ion build upon an in e disciplina y ap- p oach. By applying su ey echniques, empi ical in es iga ions o obse a ional da a, as well as expe imen al me hodology, he esea ch p esen ed he e sheds ligh on opics e- la ed o he economics o c ime and co up ion. I employ he encompassing e m beha io al e hics o s ess he ac ha my esea ch gi es p io i y o assessing he d i e s and conse- quences o indi idual decision-making in (un)e hical se ings. I ollow Baze man and Gino (2012, p. 85) in de ining he e m beha io al e hics as “ he s udy o sys ema ic and p edic - able ways in which indi iduals make e hical decisions and judge he e hical decisions o o h- e s when hese decisions a e a odds wi h in ui ion and he bene i s o he b oade socie y.” I expand his pe spec i e by en iching he discussion on decision-making wi hin he une h- ical sphe e, and con ibu e o a b oade unde s anding o (un)e hical beha io . A o al o i e pape s a e a anged acco ding o hei opic in h ee dis inc chap e s. The s uc u e o his disse a ion ollows a mac o- o-mic o app oach, which is conduci e o lead- ing a comp ehensi e discussion on illici beha io . The o de in which my esea ch p ojec s a e discussed in his disse a ion ansi ions om a high-le el app oach o a concise beha - io al in es iga ion a he indi idual le el. To se he s age, I i s discuss he s a us-quo o exis ing heo ies and empi ical esea ch on he d i e s o illici beha io wi h a pa icula ocus, bu no limi ed o, co up ion, be o e in es iga ing he in e ela ion be ween illici beha io and ins i u ional en i onmen s employing empi ical echniques o obse a ional da a analyzing he bila e al in e dependence be ween co up ion and mig a ion. I comple e 9 he pic u e wi h a na ow mic oanalysis on he indi idual d i e s o illici beha io wi h a pa icula ocus on beha io al spillo e s. In mo e de ail, Chap e 1 deals wi h he gene al o e iew and discussion o he economics o co up ion om an in e disciplina y pe spec i e (Diman & Schul e, o hcoming). Co - up ion, as he p edominan pa o o ganized c ime, has ie ce impac s on economic and socie al de elopmen , wi h es ima es sugges ing he di ec cos om co up ion o exceed $1 illion on a global scale (T anspa ency In e na ional, 2011a). As Gi e (1999, p. 1) explains i , “co up ion is one o he mos dange ous social ills o any socie y. This is because co up- ion, like a deadly i us, a acks he i al s uc u es ha make o socie y’s p og essi e unc- ioning, hus pu ing i s e y exis ence in o se ious pe il.” By de ini ion, co up ion ep e- sen s a hidden ac ion dis ending unde he su ace o ou daily li e, ende ing eliable es i- ma es o i s magni ude and pe asi eness nea ly impossible. None heless, and o he bes o ou knowledge, co up ion has soaked h ough en i e pa s o socie y and he economy, bo h in subliminal and pe asi e o ms (c . Rose-Acke man & So eide (2011)). I also has be- come a mo e publicly discussed opic due o inc eased media co e age, ecen ly d i en by he FIFA co up ion scandals and exposed cases o pe o mance enhancing d ugs in p o es- sional spo s (Diman & Deu sche (2015)). Consequen ly, in o de o unde s and he bigge pic u e o co up ion i is impo an o b eak he unde lying mechanisms down in o an e- ceden s and e ec s o co up ion a he mic o, meso, and mac o le el. This ep esen s his chap e ’s pape , a guing ha only he conside a ion o a ional and beha io al aspec s, as well as sociological, c iminological, and ins i u ional ac o s and hei in e ac ion pain s us a comp ehensi e pic u e o why, how, and o wha ex en , indi iduals engage in illici be- ha io in gene al and co up beha io in pa icula . In Chap e 2, he empi ical analyses o obse a ional da a shed ligh on he impac o co - up ins i u ional en i onmen s on mig a ion decisions as well as on he e e se link ha is he impac o immig a ion on des ina ion coun y’s co up ion le els. In he pape by Di- man , K iege & Meie ieks (2013) discussed in Chap e 2.1, we shed ligh on he ole o co - up ion in igge ing emig a ion lows om co up ion- idden coun ies. We examine he in- luence o co up ion on mig a ion o 111 coun ies be ween 1985 and 2000. Robus e i- dence indica es ha co up ion is among he push ac o s o mig a ion, especially ueling 16 S ill, o a long ime, co up ion was mainly a esea ch opic in he ields o poli ical, sociological, his o ical and c iminal law esea ch. In he 1960s and 1970s, gene al app oaches o assessing he mechanism o co up ion c ea ed an ambiguous pic u e o i s o e all e ec s. Due o a lack o eliable da a and me hodological issues, economic esea ch emained la gely silen (My dal, 2011). A ha ime, con lic ing in e es s be ween poli icians and esea che s we e p e en ing co up ion esea ch om ad ancing. Fo example, ying o ecei e a isa o a possibly co up ion- idden coun y was almos impossible a ha ime i he ip’s pu pose—a co up ion s udy—was men ioned (Nye, 1967). On op o ha , esea ch on co up ion had su e ed om disag eemen on a o mal de ini ion and he con ex dependency o an ac , which may all unde he de ini ion o co up ion in one coun y bu no in ano he . One o he i s o - eci ed de ini ions was coined by Nye (1967, p. 419):“Co up ion is beha io which de ia es om he o mal du ies o a public ole because o p i a e- ega ding (pe sonal, close amily, p i a e clique) pecunia y o s a us gains; o iola es ules agains he exe cise o ce ain ypes o p i a e- ega ding in luence.” One d awback o his de ini ion is he inhe en ambigui y, because “all illegal ac s a e no necessa ily co up and all co up ac s a e no necessa ily illegal.” (Pe e s & Welch, 2011, p. 155) In ce ain socie ies, pa icula ac ions may al eady be conside ed a o m o co up misconduc , whe eas in o he socie ies hese ac s may well be pa o hei “ o mal du ies” and “jus poli ics.” (Pe e s & Welch, 2011) S a ing in he la e 1980s and ea ly 1990s, sound heo e ical app oaches acili a ed he schola ly e o s o s udy he mechanism o he economics o co up ion. Especially in ligh o he economic accele a ion o Asian coun ies a ha ime, esea ch was s ill unse led on whe he co up ion exhibi s only ad e se e ec s on socie ies and economics—sanding he wheels—o migh c ea e posi i e e ec s—g easing he wheels—unde ce ain ci cums ances h ough he educ ion o ine icien ed ape (D ehe & Gassebne (2013), Vial & Hano eau (2010)). Today, his a gumen was se led by sound esea ch, indica ing ha co up ion abo e all is de imen al o he socie y. These esul s a e now b oadly accep ed. Th ough he use o mo e sophis ica ed me hodological app oaches and mo e eliable da a, cu en esea ch has se led on he ac ha he gene al and long- e m de imen al e ec s o co up ion ou weigh he con ex -speci ic and sho - e med posi i e e ec s (Aid (2009), 17 Méon & Sekka (2005)). The b oade a ailabili y o huge da ase s was key o his de elopmen . Fo example, he PRS G oup in oduced he “In e na ional Coun y Risk Guide” in 1984 and T anspa ency In e na ional es ablished he Co up ion Pe cep ion Index as one o he mos acknowledged measu emen s in 1995. In he 1990s and a e he end o he Cold Wa , he i s global an i-co up ion mo emen s occu ed along wi h he democ a iza ion p ocess o many de eloping coun ies. E e since, he media has become inc easingly in ol ed in a c i ical assessmen o co up ion, d awing he public’s a en ion o i s consequences (Lamb opoulou, e al., 2005). In wha ollows, he mechanism, he an eceden s, and he e ec s o co up ion will be discussed om an in e disciplina y pe spec i e on he mic o, meso, and mac o le el. 1.3 Face s and De e minan s o Co up ion The nex sec ion cen e s on he in e disciplina y na u e o co up ion esea ch. In ou a emp o blend di e en heo ies om a ious a eas, we in oduce a s uc u al amewo k ha allows us o discuss co up ion s epwise, om wha we e e o as he inne - o-ou e - wo ld app oach. Fo his eason, we s a wi h he analysis o co up beha io in he in e nal wo ld, which comp ises a c i ical discussion o he a ional choice heo y and beha io al heo ies. Building on his, we hen add an addi ional le el o discussion a he meso le el, whe e we shed ligh on bo h sociological and c iminological ac o s. Ul ima ely, we discuss co up beha io om he pe spec i e o he ex e nal wo ld, which includes, among o he s, economic, legal, and poli ical aspec s. We belie e ha such an app oach encompasses he b ea h o scien i ic discussion on he opic o co up ion and does su icien jus ice o he di e en heo ies and app oaches ha con ibu e o a be e unde s anding o wha shapes co up beha io . Fo easons o con enience, we p o ide a g aphical illus a ion o guide he eade h ough he nex sec ion’s discussion o ac o s ha explain co up beha io . 18 Figu e 1.1: In e disciplina y Pe spec i e Ex e nal Wo ld Bu eauc a ic En i onmen Poli ical En i onmen His o y O he Ins i u ional Cha ac e is ics Legal En i onmen Economic En i onmen Geog aphy Meso Wo ld Values Social No ms S ain Theo y Educa ion Cul u e In e nal Wo ld Ra ional Choice Theo y Beha io al Pe spec i e Indi idual Co up Beha io Induces a Re oac i e E ec Decision - making - p ocess e en ually leads o Di e en ial Associa ion Theo y 19 1.3.1 In e nal Wo ld – Ra ional Choice & Beha io al Pe spec i e The in e nal wo ld ep esen s a mic o pe spec i e ha highligh s he indi idual’s in insic willingness o ac i ely engage in ac s o co up ion. This aspec comp ises pu ely a ional beha io and beha io beyond his clea -cu a ionale. He e, ligh will be shed on aspec s ha exclusi ely a ge he indi idual pe spec i e. This ep esen s a p ecise me hodological di e ence in compa ison o he agg ega e le els ha will be analyzed in subsequen chap e s. We deem i impo an o include hese di e en pe spec i es o allow o a well- ounded discussion o he an eceden s and e ec s o co up ion. Fo his pu pose, we will s a wi h a pu e ac o -based pe spec i e and hen g adually mo e owa ds an agg ega e pe spec i e. Conside ing a ional choice, his pa icula app oach in he con ex o c ime has i s oo s in he seminal con ibu ion o Ga y S. Becke , analyzing he disposi ion o de ian beha io based on cos -bene i calcula ions (Becke , 1968). Encompassing economic heo ies on c ime causa ion ha e e ol ed e e since. The a ional choice, whe he o no o succumb o co up beha io , is based on a decision p ocess in which indi iduals y o maximize hei u ili y. This is done by weighing expec ed bene i s agains expec ed cos s o de ian beha io , including oppo uni y cos s and he isk o being caugh o punished. One can use his gene al app oach o unde s and a subse o c iminal beha io , namely co up ion, by shedding ligh on he decision making p ocess o bo h he b ibe and he b ibee. Al hough oppo uni y cos s and isk calcula ion will ce ainly di e o each o he pa ies in ol ed, he basic decision p ocess is simila . (1) Oppo uni y cos s due o ime alloca ion: Whene e ime is spen on c iminal engagemen , less ime is a ailable o legal ac i i ies. The oppo uni y cos s he e o e ep esen he amoun o income, which is gi en up o a end o he al e na i e ac ion. (2) Risk calcula ion: The conside a ion o he isk o being caugh o punished. Ce ain ac ions a e less likely o be obse ed and p osecu ed and hus d i e he indi idual isk assessmen . Bo h ac o s also ep esen iable ways o de e co up beha io , o example, h ough applying mo e se e e punishmen s and inc easing he p obabili y o de ec ion. Resea ch indica es ha bo h inc easing he ce ain y and he se e i y o punishmen a e iable measu es o de e c iminal beha io , wi h he o me being backed up by mo e, consis en , 20 empi ical e idence han he la e (Nagin, 2013). Feess e al. epo ha inc easing he magni ude o punishmen — o example, up o a dea h penal y like in China—migh e en b ing abou pe e se e ec s (Feess, e al., 2014). I is easonable o assume ha unde such ci cums ances, judges would end o be mo e ca e ul in sen encing, since he condemna ion would be associa ed wi h high cos s o bo h he de endan and he judge gi en he isk o a po en ially w ong decision. Consequen ly, i espec i e o he co up ac s de ec ed, pe cen age o ac ual con ic ions migh d op, ende ing inc eased sanc ion de imen al o useless a he bes . F om a c iminal’s pe spec i e, in a si ua ion in which de ian beha io becomes mo e luc a i e due o a ce e is pa ibus dec ease in expec ed cos s, such a leeway migh induce e en mo e de ian beha io . A e all, acing bo h a d op in con ic ions and a ising es ima ed numbe o un epo ed cases may emp he go e nmen o impose e en ha de sanc ions, leading o a icious ci cle (S ein ücken, 2004). Ye , mo e o en han no , indi idual beha io goes beyond clea -cu a ional decision- making bu is bounded in e ms o o wha ex en decisions a e ho oughly elabo a ed (Gige enze & Sel en, 2002). As desc ibed be o e, he pu e a ional choice app oach lea es no oom o mo al qua els ha may in luence he calculus, al hough eal li e expe ience p o es mo als highly ele an . Ye , mo als di e no only om socie y o socie y bu also on an indi idual le el and e en om one si ua ion o ano he —especially i ac o s such as emo ions a e conside ed. Essen ially, a combina ion o all hese aspec s is needed o each a well-elabo a ed in e nal iew. Thus, in ecen yea s, he beha io al app oach, which en iches he a ional pe spec i e wi h he inclusion o psychological aspec s and biases, has been inco po a ed in o models ying o be e explain de ian beha io in gene al and co up beha io in pa icula . I has been a gued ha e en a a ional decision-make migh end up engaging in seemingly i a ional beha io ha is guided by mo e han jus a a ional calculus, bu a he is a unc ion o he unde lying en i onmen . This s eam o li e a u e has ex ended he decision space o he so-called “homo oeconomicus” by inco po a ing ac o s such as ecip oci y, emo ions, social image and he like o d aw a mo e ealis ic pic u e o human beha io (Ba be is, 2011). Clea ly, he g owing body o app oaches ep esen s an addi ion a he han subs i u ion o he a ional choice app oach. 21 A guably, pu e a ional choice conce ns a e incapable o explaining he de ac o ex en o exis ing co up ion. Lambsdo a gues ha he a ional choice heo y b ings abou wo seemingly con lic ing ou comes, one wi h and one wi hou exis ing co up ion. On he one hand, one should obse e co up ion mo e equen ly as i is he case since—a leas in he absence o no ms, alues, and he like—c iminal beha io is solely d i en by a ional calculus (Lambsdo , 2012). On he o he hand, since b ibe y is no a subgame pe ec Nash equilib ium, i s ac ual occu ence migh al eady be su p ising. In one-sho b ibe y se ings, as is usually he case, epu a ion does no play any ole, sugges ing ha he b ibee has no incen i e o ecip oca e he beha io o he b ibe . Consequen ly, he b ibe an icipa es he b ibee’s de ian beha io —e.g. pocke ing he money wi hou p o iding he espec i e se ice—and, as a esul , he should no pay any b ibes in he i s place. E en in epe i i e se ings, he exchange will e mina e e en ually, leading o wha is called an endgame e ec , sugges ing ha he b ibee will de ia e om he ecip ocal a angemen a some poin . This en ails ha by using backwa d induc ion, he b ibe will e ain om paying b ibes in he i s place as well. Accoun ing o hese seemingly con lic ing ou comes, cu en esea ch sugges s ha one’s decision-making p ocess is as ly guided by he social en i onmen and one’s pee ’s beha io (E ans e al. (1992), Glaese e al. (1996)). Among o he hings, heo e ical and expe imen al esea ch sugges s ha he e ec o beha io al con agion is media ed by he social p oximi y o he pee s (Ake lo (1997), Diman (2015)). A pe son’s ai s and beha io a e p edominan ly based on social in e ac ion (LaRossa & Rei zes, 1993); people a e no bo n wi h hem, bu a he lea ned and adap ed h ough he cou se o social in e ac ion. These pa e ns and alues can a y and de elop as ime mo es on and hey can be conside ed o be unde cons an exogenous in luence. Wha is mo e, exis ing e idence poin s a he impo ance o social no ms and alues, bu also he impac o epu a ion in epea ed game en i onmen s, in explaining co up beha io (Gäch e & Falk (2002), Milinski e al. (2002)). “Repu a ion is a powe ul o ce o s eng hening and enla ging mo al.” (Haid , 2007, p. 998) In sum, he many ac o s comp ising he in e nal wo ld can be seen as he essen ial pilla s in explaining co up beha io . Resea ch indica es, howe e , ha he decision o beha e in a co up manne is no d i en solely by in e nal ac o s. Ins ead, i is he in e play wi h he 22 social en i onmen ha impac s o o e ides he in e nal wo ld. The social na u e o humans p omo es he conside a ion o pee g oup a ilia ion and epu a ion, deeming i unlikely ha beha io in gene al and une hical one in pa icula is pu ely sel -d i en. We now u n o he discussion o meso and mac o ac o s ha add o he unde s anding o co up decision-making and build upon he in e nal wo ld. 1.3.2 Meso Wo ld—Sociological & C iminological Fac o s The meso wo ld ocuses on social in e ac ion. I is plausible o assume ha , beyond he in insic willingness, di e en componen s like ypical alues, ules, and no ms wi hin a gi en socie y ha e a s ong impac on a pe son’s decision on whe he o no o ac co up ly. The e a e many sociological ac o s and c iminological aspec s as well as heo ies ha can in luence he le el o co up beha io . Sociological Fac o s The gene al cul u e wi hin a gi en coun y can ha e a signi ican impac on indi idual decisions o engage in co up beha io . Hus ed examines he e ec o di e en cul u al aspec s and desc ibes “a cul u al p o ile o a co up coun y as one in which he e is high unce ain y a oidance, high masculini y, and high powe dis ance.” (Hus ed, 1999, p. 354) O he s udies come o a simila conclusion. Fo example, Volkema and Ge z (2001) analyzed powe dis ance and unce ain y a oidance, again showing a signi ican posi i e co ela ion be ween hese cul u al ac o s and he le el o co up ion. Recen s udies also suppo hese esul s. The wo dimensions o na ional cul u e (powe dis ance and indi idualism) mode a e he ela ionship be ween human de elopmen and co up ion (Sims, e al., 2012). This is also ue i no ms and alues a e ca ied o e om di e en cul u es h ough mig a ion. Fo example, Diman e al. ind some indica ion o such a oo p in e ec . In con inuing o conduc business as usual, he des ina ion coun ies expe ience de e io a ion in ins i u ional quali y and an inc ease in co up ion le els in he sho un. Bu hey also ind ha mig an s e en ually assimila e o he new en i onmen in he medium un (Diman , e al., 2015). Aside om he cul u al aspec s, esea ch also poin s a he ele ance o educa ion in media ing he inclina ion owa ds co up beha io . Educa ion ypically in ensi ies in he 23 p ocess o economic de elopmen wi hin a gi en coun y and con ibu es o lowe le els o co up ion (T eisman, 2000). A s udy conduc ed in Nepal indica es ha educa ion is one o he p ima y de e minan s o co up beha io . Highe educa ion is s ongly co ela ed wi h he likeliness o condemn co up beha io and he eluc ance o accep e en small b ibes (T uex, 2011). Resea ch also indica es ha he composi ion o gende in leading posi ions media es he ex en o co up ion (Sung & Chu (2003), Sung (2003), Sung (2012)). Fo example, Dolla e al. (2001) ind ha a g ea e numbe o women in ol ed in pa liamen is ypically associa ed wi h lowe le els o co up ion. Simila esul s a e common in c oss-coun y e alua ions (Swamy, e al., 2001). Typically, women end o obey socie y ules and a e less likely o ake se ious isks and he e o e less o en commi o co up ion (Esa ey & Chi illo (2013), F ank e al. (2011)). C iminological Fac o s F om a c iminological pe spec i e, co up ion is a he cen e o gene al c ime and i acili a es he pe asi eness o he c ime (Husman & Walle (2010), Shelley (2014)). The c iminological iew on de ian beha io is in e disciplina y in i sel . In pa icula , he e is a s ong in e dependence be ween he sociological ac o s and c iminology, because aspec s like cul u e and educa ion ha e an e ec on gene al c ime a es and he e o e on he le el o co up ion. The inco po a ion o a ional decision-making also ep esen s an e iden link o he in e nal wo ld (Glueck & Glueck, 2014). Su he land and C essey (2014) b ough o wa d he di e en ial associa ion heo y, concluding ha c iminal beha io is commonly lea ned and adop ed in in e ac ion wi h o he people. Aspec s such as social class, ace and uns able homes a e no only ac o s a o ing he commi men o c iminal ac i i y bu hey also inc ease he p obabili y ha people will socialize wi h pe sons o simila cha ac e . This heo y is widely suppo ed by empi ical esea ch ha ocuses on social lea ning o bo h c iminal and con o ming beha io (Ake s (2014), Cohen (2014)). A he same ime, social lea ning is no only es ic ed o small neighbo hoods o ce ain a eas, bu also does en ail an agg ega e pe spec i e on he socie al le el. The s ain heo y, i s es ablished by Me on in 1938 in a ime when he 24 mos widely accep ed hypo hesis was ha c iminal beha io can be p ima ily a ibu ed o biological disposi ion, highligh s he ele ance o social s uc u es and social p essu e in he occu ence o c iminal beha io (Me on, 1938). Whene e indi iduals eel hey a e being ea ed un ai ly by he socie y—e.g. es ic ed access o good schooling—, hey encoun e a s ess ul si ua ion, which in u n axes one’s sel -con ol (Hi schi & Go edson, 1990). This heo y sugges s ha unde hese ci cums ances, people may end o e e se he goals se by socie y and c ea e hei own goals con lic ing wi h exis ing no ms and alues. They a e likely o belie e ha he means jus i y he ends, which is conduci e o hei decision o engage in c iminal ac i i ies (Cohen (2014), Messne & Rosen eld (2014)). The basic s ain heo y, howe e , has been al e ed o e ime, e en ually leading o a mo e gene alized heo y. Indi iduals e en in a s able pe sonal en i onmen —e.g. well paid and secu e job—a e po en ially willing o pu e e y hing a isk and choose o engage in c iminal beha io . Such beha io migh s em om a biased sel -pe cep ion. Al hough well-educa ed whi e-colla indi iduals should be able o ully ake s ock o he consequences o hei co up beha io , Benson a gues ha such o ende s o en do no iew hemsel es as c iminals bu a he as good employees, jus i ying hei ac s solely on he basis o ying o en o ce he company’s success (Benson, 2014). This heo y seems o hold especially o employees in highe posi ions wi h ample esponsibili ies when hey see he chance o, o example, secu e o he people’s jobs by ac ing co up ly (Fleming & Zyglidopoulos, 2009). Such a biased sel - pe cep ion migh be he esul o bo h hypoc isy and a di e en unde s anding o wha is igh and w ong. As esea ch indica es, such an unde s anding o , o example, wha is conside ed a b ibe o a gi is con ex dependen , which a ies subs an ially ac oss coun ies (Milling on, e al. (2005), S eidlmeie (1999)). Howe e , aside om a ying pe cep ions in di e en coun ies, he a ionaliza ion p ocess is p esen in e e y socie y and i is a key de e minan o whi e-colla c ime and co up ion in pa icula . The abili y o a ionalize une hical beha io pushes ou eelings o guil and shame, ende ing co up beha io jus i iable i he e a e enough good easons (Sø eide, 2014). In line wi h he social lea ning heo y in oduced ea lie , such wo k en i onmen s can be deemed highly nega i e. I he supe iso s ac co up ly wi hou any eelings o guil , his beha io may a ec he o he 25 employees’ decision-making p ocess. Consequen ly, u he analysis is essen ial wi h espec o he ex emely high damages in ol ed in whi e-colla c imes. P osecu ion and quan i ica ion o such c imes u n ou o be ex emely ough (Lambsdo & Schulze, 2015), and e en hough nume ous cases wi h ex ensi e damage a e known, he ac ual ami ica ions emain de ious. Fu he mo e, highe le els o co up ion combined wi h weak ins i u ional s uc u es, soak h ough socie y and e en ually lead o ising gene al c ime a es, c ea ing a hos ile en i onmen and b eeding g ound o e en mo e co up ion (Cla os, 2013). This a icle now u ns o he ex e nal wo ld by shedding ligh a ac o s a he mac o le el ha in luence he ex en o co up ion. 1.3.3 EXTERNAL Wo ld—Economic, Legal, Poli ical, His o ical and Geo- g aphical Fac o s The ex e nal wo ld includes all o he elemen s ep esen ing ex insic oppo uni ies ha di ec ly o indi ec ly ha e an in luence on co up ion. Among o he s, hese a e economic, legal, poli ical, his o ical and geog aphical ac o s. Economic Fac o s Exis ing esea ch poin s a a b oad ange o economic ac o s ele an o he ex en o co up ion. Fo example, he o e all quali y o he go e nmen in a gi en coun y is a well- s udied de e minan . “Poo go e nance may a ec economic pe o mance h ough hei impac on ax e enue, public spending, and iscal de ici .” (Tanzi, 1999, p. 10) Ine icien bu eauc acy uels co up ion because i p o ides a e ile g ound o “speed money.” Such a mechanism is designed o ci cum en impeding egula o y bodies, which ep esen he majo ing edien o he g easing he wheels hypo hesis desc ibed in Sec ion 1.2. In he con ex o i m en y in highly egula ed coun ies, D ehe and Gassebne (2013) analyzed mo e han o y coun ies o se e al yea s, concluding ha he g easing he wheels hypo hesis holds e en oday. The mo e ine icien egula ions a e, he longe he delays o companies being able o s a hei business. In consequence, such ine iciency, coupled wi h he isk o losing money and business, igge hei decision o make use o speed 32 CHAPTER 2: EMPIRICS 2.1 THE EFFECT OF CORRUPTION ON MIGRATION, 1985-2000ꟸ Eugen Diman *, Tim K iege **, Daniel Meie ieks *** ABSTRACT: We examine he in luence o co up ion on mig a ion o 111 coun ies be ween 1985 and 2000. Robus e idence indica es ha co up ion is among he push ac o s o mig a ion, es- pecially ueling skilled mig a ion. We a gue ha co up ion ends o diminish he e u ns o educa ion, which is pa icula ly ele an o he be e educa ed. KEYWORDS: Co up ion; mig a ion; skilled mig a ion; push ac o s o mig a ion JEL CLASSIFICATION: D73; F22; O15 ꟸ Published in: Applied Economics Le e s (2013), 20(13), pp. 1270-1274. * Uni e si y o Pade bo n. Co esponding au ho . ** Uni e si y o F eibu g and CESi o. *** Uni e si y o F eibu g. 33 2.1.1 In oduc ion P e ious empi ical esea ch sugges s ha , besides socio-economic and demog aphic ac- o s (e.g., unde de elopmen and demog aphic p essu es), poli ico-ins i u ional condi ions (e.g., poli ical ins abili y) a e among he push ac o s o in e na ional mig a ion (c ., e.g., Ha on and Williamson (2003); Mayda (2010); D ehe e al. (2011); Docquie and Rapopo (2012)). We a gue ha a ela ed impo an push ac o , which has so a been mos ly ne- glec ed in bo h heo e ical and empi ical esea ch, is co up ion. Co up ion is associa ed wi h a numbe o un a o able ou comes. Fo one, co up ion ends o nega i ely a ec a coun y’s (sho - un) le el o economic ac i i y.1 Economies plagued by high le els o co up ion g ow mo e slowly, e.g., as co up ion con ibu es o an ine i- cien alloca ion o esou ces (Jain (2001); Campos e al. (2010)). Fo ano he , co up ion may also wo sen a coun y’s (s uc u al) socio-economic si ua ion. Fo ins ance, Gup a e al. (2002) ind ha high le els o co up ion p omo e income inequali y and he sp ead o po - e y. Fu he mo e, co up ion may also lead o subop imal pa e ns o social mobili y when i ma e s mo e s ongly o upwa d mobili y han ac ual me i s. In sum, he p e alence o co up ion is likely o wo sen indi idual wo king and li ing condi- ions o he majo i y o ci izens. I may he e o e also ma e o he calculus o a po en ial mig an . Indi idual educa ion is a pa icula ly impo an ac o in luencing mig a ion deci- sions. He e, co up ion ends o lowe he e u ns o educa ion, e.g., by con ibu ing o slow economic g ow h and unemploymen , widesp ead inequali y and he lack o social ad- ancemen . Gi en he i e e sibili y o human capi al in es men , co up ion may make i mo e a ac i e o mig a e o ecoup one’s indi idual educa ion in es men . He e, we ex- pec he highly skilled o be pa icula ly esponsi e o he p e alence o co up ion, gi en hei high le el o human capi al in es men and subsequen need o pa icula ly high (i.e., cos -e ec i e) skill p emiums.2 Based on hese lines o easoning, we hypo hesize ha co - up ion is among he push ac o s o mig a ion and pa icula ly ele an o skilled mig a ion. 1 No e, howe e , ha his assessmen does no ule ou ha co up ion may ac ually yield posi i e economic e ec s unde speci ic ci cums ances (D ehe and Gassebne , 2013). 2 In addi ion o ha , he inc ease o income inequali y and po e y caused by co up ion (Gup a, e al., 2002) may os e he poli ical demand o edis ibu ion. As he be e skilled a e ypically he ypical ne paye s o (p og essi e) income axes, his may u he uel skilled emig a ion. 34 2.1.2 Da a and Me hodology To empi ically examine his hypo hesis, we compile da a on (skilled and a e age) mig a ion, co up ion and u he con ols o 111 coun ies be ween 1985 and 2000. The summa y s a- is ics and he ope a ionaliza ion o he con ols a e epo ed in Table 1.3 Da a on ou de- penden a iables, he mig a ion a es, a e d awn om De oo (2008) who p o ides es i- ma es o he a es o skilled and a e age mig a ion o six main ecei ing coun ies (Aus alia, Canada, F ance, Ge many, he U.K. and he U.S.). He e, he skilled mig a ion a e e e s o he a io o he numbe o skilled emig an s (who exhibi a pos -seconda y ce i ica e) o he o al numbe o skilled na i es aged 25 o olde , while he a e age mig a ion a e is de ined as he a io o he o al numbe o emig an s o he o al numbe o na i es aged 25 o olde (De oo , 2008). 3 The mig a ion da a is a ailable only o h ee poin s in ime (1990, 1995 and 2000). The e o e, we use i e-yea a e ages o he explana o y a iables ( o he 1986-1990, 1991-1995 and 1996-2000 pe iods) o es ima e hei in luence on mig a ion. 35 Table 2.2.1: Summa y S a is ics and Da a Ope a ionaliza ion and Sou ces Va iable Obse a- ions Mean SD Mini- mum Maxi- mum Ope a ionaliza ion Sou ce Skilled Mig a ion 333 0.143 0.179 0.001 0.910 A e age Mig a ion 333 0.035 0.061 0.001 0.419 Co up ion 333 2.669 1.294 0 5.983 Pe Capi a Income 333 8.625 1.179 5.576 10.908 Real, PPP-adjus ed pe capi a income, logged (a) Popula ion Size 333 9.380 1.487 6.028 14.038 Popula ion size in housands, logged (a) Regime Type 333 2.169 7.072 -10 10 Re ised Combined Poli y Sco e, anging om -10 (au oc acy) o +10 (democ acy (b) Poli ical Ins abili y 333 0.300 0.758 0 3.766 Numbe o ba le dea hs in ci il wa s (de ined as a con lic wi h a leas 25 ba le dea hs pe yea ), logged+1 (c) You h Bu den 333 0.260 0.030 0.174 0.333 Numbe o people be ween he ages o 15 and 29 as sha e o o al popula ion (d) Quali y o Bu eau- c acy 333 1.890 1.200 0 4 Index o ins i u ional s eng h and quali y o a coun y’s bu eauc acy (e) T ade Openness 333 4.026 0.644 0.486 5.811 Sum o expo s and impo s as a sha e o eal GDP, logged (a) Fo me Colony 333 0.495 0.501 0 1 Time-in a ian dummy a iable (1=coun y has colonial links o one o he six majo e- cei ing coun ies) ( ) Dis ance 333 7.943 1.066 4.456 9.093 Time in a ian a iable ha measu es he dis ance be ween coun y’s capi al and he nea es capi al o one o he six majo ecei ing coun ies, logged ( ) No es : Sou ce e e s o (a) PENN Wo ld Tables (h ps://pw .sas.upenn.edu/), (b) Poli y4 Da ase (h p://www.sys emicpeace.o g/poli y/poli y4.h m), (c) PRIO Ba le Dea hs Da a (h p://www.p io.no/Da a/A med-Con lic /Ba le-Dea hs/), (d) Uni ed Na ions Popula ion Di ision (h p://esa.un.o g/unpd/wpp/Excel-Da a/popula ion.h m), (e) ICRG (2009), ( ) CEPII GeoDis Da ase (h p://www.cepii. /CEPII/en/bdd_modele/p esen a- ion.asp?id=6). 36 Co up ion da a a e d awn om he In e na ional Coun y Risk Guide (PRS G oup, 2008).4 Su eying expe s, he ICRG issues a co up ion index ha we use as ou main explana o y a iable. He e, co up ion e e s o inancial co up ion associa ed wi h conduc ing busi- ness (e.g., b ibes) as well as o he o ms o poli ical co up ion such as excessi e pa onage, nepo ism and close ies be ween poli ics and business. No e ha we escaled he ICRG co - up ion index, so ha highe alues co espond o highe co up ion le els. To a oid de ec ing only spu ious e ec s o co up ion on mig a ion, we also conside a num- be o con ounding con ols ha may simul aneously a ec co up ion and mig a ion. Fo ins ance, we con ol o he e ec o economic de elopmen , as iche coun ies a e bo h less suscep ible o co up ion and less likely o ha e high mig a ion a es. Following he em- pi ical li e a u e on he de e minan s o co up ion (Se a, 2006) and mig a ion (Docquie & Rapopo , 2012), in ou baseline speci ica ion we con ol o he e ec o pe capi a income, popula ion size, egime ype and poli ical ins abili y. As obus ness checks, we amend his model wi h addi ional con ols o demog aphic p essu e (you h bu den), ins i u ional qual- i y ( he quali y o a coun y’s bu eauc acy), ade openness and ce ain coun y-speci ic ai s ha may a ec mig a ion cos s (colonial ies and dis ance be ween sending and des i- na ion coun ies).5 Ini ial es s indica e he p esence o au oco ela ion, he e oskedas ici y and c oss-sec ional dependence, as i is common o panel da a wi h coun y-yea obse a ions. We he e o e un a se ies o pooled OLS and ixed-e ec s eg essions wi h D iscoll-K aay s anda d e o s ha a e obus o hese da a cha ac e is ics (D iscoll & K aay, 1998).6 4 We use he ICRG da a because i is a ailable since 1984, making a panel es ima ion app oach o he co up ion-mig a ion nexus possible. O he co up ion measu es a e a ailable only o sho e ime pe iods. Jain (Jain, 2001, p. 77) no es ha he a ious co up ion measu es a e usually highly co ela ed. 5 Ou indings a e also obus o he inclusion o u he con ols o eligious ac ionaliza ion, oil p oduc ion, go e nmen size, u he geog aphic and his o ic coun y cha ac e is ics (landlocked loca ion, common language) and educa ion (yea s o schooling). 6 We also expe imen ed wi h ins umen al a iable (IV) es ima ions, as e e se causa ion may be an issue. Howe e , pooled and ixed-e ec s IV-es ima ions (whe e co up ion is ins umen ed by he quali y o judicial ins i u ions and he deg ee o democ a ic pa icipa ion) do no indica e ha co up ion is endogenous o mig a ion. Also, Du bin-Wu-Hausman es s sugges s ha any endogenei y among he eg esso s does no bias ou es ima es. 37 2.1.3 Empi ical esul s The pooled OLS es ima es a e epo ed in Table 2. We ind ha co up ion has a posi i e and s a is ically signi ican e ec on bo h skilled and a e age mig a ion. Howe e , he ma ginal e ec o co up ion on skilled mig a ion ends o be app oxima ely h ee o ou imes highe han i s e ec on a e age mig a ion. This inding p o ides i s suppo o ou hy- po hesis ha co up ion as among he push ac o s o mig a ion and especially ma e s o he mig a ion decisions o he highly skilled. The ixed-e ec s es ima es—which uly conside he panel s uc u e o ou da ase —a e e- po ed in Table 3.7 We ind ha co up ion only has a posi i e, s a is ically signi ican and speci ica ion- obus e ec on skilled mig a ion, bu has no signi ican impac on a e age mig a ion. This esul u he s eng hens ou p e ious inding ha he decision o he highly skilled o emig a e is s ongly a ec ed by he disincen i e o co up ion a home. 7 No e ha all cons an in luencing ac o s (colonial ies, dis ance e c.) a e now subsumed unde he ixed e ec s. 38 Table 2.2.2: Co up ion and Mig a ion (Pooled OLS Es ima es) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Skilled Mig a ion (1) - (5) A e age Mig a ion (6) - (10) Co up ion 0.025 0.022 0.036 0.024 0.035 0.009 0.005 0.009 0.008 0.012 (0.006)** (0.004)** (0.005)** (0.006)** (0.006)** (0.001)*** (0.001)** (0.001)** (0.001)*** (0.001)*** Pe Capi a - 0.046 - 0.043 - 0.057 - 0.048 - 0.045 0.003 0.005 0.002 0.002 0.002 Income (0.002)*** (0.003)*** (0.007)** (0.003)*** (0.002)*** (0.001)** (0.001)** (0.002) (0.001) (0.001) Popula ion Size - 0.044 - 0.044 - 0.047 - 0.040 - 0.041 - 0.014 - 0.013 - 0.014 - 0.012 - 0.013 (0.001)*** (0.001)*** (0.002)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.002)** (0.001)** (0.001)*** Regime Type 0.006 0.007 0.006 0.007 0.007 0.003 0.003 0.003 0.003 0.003 (0.001)** (0.001)** (0.001)** (0.001)** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** Poli ical 0.002 0.002 0.007 0.005 0.004 - 0.003 - 0.003 - 0.003 - 0.002 - 0.003 Ins abili y (0.005) (0.005) (0.005) (0.005) (0.007) (0.003) (0.003) (0.002) (0.003) (0.004) You h Bu den 0.389 0.403 (0.361) (0.041)** Quali y o - 0.027 - 0.002 Bu eauc acy (0.010) (0.005) T ade 0.020 0.009 Openness (0.001)*** (0.001)*** Fo me Colony 0.068 0.019 (0.001)*** (0.001)*** Dis ance - 0.026 - 0.010 (0.003)** (0.002)** R 2 0.252 0.254 0.262 0.255 0.294 0.200 0.225 0.200 0.206 0.237 N*T 333 333 333 333 333 333 333 333 333 333 No es : Cons an no epo ed. D iscoll - K aay s anda d e o s in pa en heses. **p<0.05, ***p<0.01. 39 Table 2.2.3: Co up ion and Mig a ion (Fixed-E ec s Model Es ima es) (1) (2) (3) (4) (5) (6) (7) (8) Skilled Mig a ion (1) - (4) A e age Mig a ion (5) - (8) Co up ion 0.005 0.005 0.004 0.005 - 0.003 - 0.003 - 0.002 - 0.003 (0.001)*** (0.001)*** (0.001)** (0.001)** (0.001) (0.001) (0.001) (0.001) Pe Capi a - 0.048 - 0.043 - 0.045 - 0.051 0.002 - 0.001 0.001 0.001 Income (0.010)** (0.010)** (0.010)** (0.008)** (0.003) (0.003) (0.002) (0.003) Popula ion Size 0.001 - 0.001 0.002 - 0.002 0.012 0.013 0.012 0.011 (0.009) (0.009) (0.010) (0.011) (0.002)** (0.002)** (0.002)** (0.002)** Regime Type 0.002 0.002 0.002 0.002 0.001 0.001 0.001 0.001 (0.001)* (0.001)* (0.001)* (0.001)* (0.001)** (0.001)** (0.001)** (0.001)** Poli ical 0.008 0.008 0.007 0.008 - 0.002 - 0.002 - 0.002 - 0.002 Ins abili y (0.001)** (0.001)** (0.001)** (0.001)*** (0.001)* (0.001)** (0.001)* (0.001)* You h Bu den 0.190 - 0.087 (0.028)** (0.018)** Quali y o 0.003 - 0.001 Bu eauc acy (0.001) (0.005) T ade 0.006 0.003 Openness (0.006) (0.001)** Wi hin - R 2 0.174 0.182 0.176 0.176 0.145 0.145 0.132 0.130 N*T 333 333 333 333 333 333 333 333 No es : Cons an no epo ed. D iscoll - K aay s anda d e o s in pa en heses. *p<0.1, **p<0.05, ***p<0.01. 40 Finally, no e ha he esul s o he con ol a iables a e la gely in line wi h p e ious e- sea ch. Fo ins ance, bo h he pooled OLS and ixed-e ec s es ima es sugges ha skilled mig a ion is less common in iche coun ies, as p e iously epo ed in Docquie and Rapopo (2012). As ano he example, he posi i e e ec o poli ical ins abili y on skilled mi- g a ion in he ixed-e ec s eg essions is in line wi h D ehe e al. (2011). 2.1.4 Conclusion We examine he impac o co up ion on mig a ion o a panel o 111 coun ies be ween 1985 and 2000. Ou empi ical esul s indica e ha co up ion especially d i es skilled mi- g a ion, while i s e ec on a e age mig a ion is less p onounced and no s a is ically obus . Ou main inding is consis en wi h he hypo hesis ha co up ion lowe s he e u ns o ed- uca ion and consequen ly ma e s mos o he calculus o (p ospec i e) highly skilled mi- g an s. Co up ion con ol may he e o e be an impo an policy ool o ein he b ain d ain, pa icula ly when his b ain d ain is associa ed wi h p edominan ly poo de elopmen ou - comes. 41 Appendix Albania Dom. Republic I aly Papua N. Guinea Tunisia Alge ia Ecuado Jamaica Pa aguay Tu key Angola Egyp Japan Pe u Uganda A gen ina El Sal ado Jo dan Philippines U. A ab Emi a es Aus alia E hiopia Kenya Poland Uni ed Kingdom Aus ia Finland Kuwai Po ugal Uni ed S a es Bah ain F ance Libe ia Qa a U uguay Bangladesh Gabon Libya Romania Venezuela Belgium Gambia Madagasca Saudi A abia Vie nam Boli ia Ge many Malawi Senegal Zambia Bo swana Ghana Malaysia Sie a Leone Zimbabwe B azil G eece Mali Singapo e Bulga ia Gua emala Mexico Somalia Bu kina Faso Guinea Mongolia Sou h A ica Came oon Guinea-Bissau Mo occo Sou h Ko ea Canada Guyana Mozambique Spain Chile Hai i Ne he lands S i Lanka China Hondu as New Zealand Sudan Colombia Hunga y Nica agua Sweden Congo (Repub- lic) India Nige Swi ze land Cos a Rica Indonesia Nige ia Sy ia Co e d’I oi e I an No way Tanzania Cuba I aq Oman Thailand Cyp us I eland Pakis an Togo Denma k Is ael Panama T inidad 48 co up ion a e deeply en enched wi hin he people’s mindse . Bilodeau (2014) inds ha he immig an s’ ela ionship wi h hei des ina ion coun y’s poli ics is subs an ially a - ec ed by he poli ical en i onmen in hei home coun y, hus sus ainably imp in ing hei pe sonal a i udes. Along hese lines, Helliwell (2014) also ind suppo o he oo p in e - ec o us le els, which is o high ele ance in he co up ion con ex (c . Bjø nsko (2011); Ro hs ein and Eek (2009)). Thei esul s sugges ha mig an s om low- us en i onmen s ca y o e hei us -a i udes o hei des ina ion coun ies in a much mo e p onounced way han mig an s om high us en i onmen s, indica ing an asymme ic in e ela ion be- ween mig a ion and s ickiness o no ms (see also Uslane (2008)). Consequen ly, alue as- simila ion becomes unlikely in he sho un, and co up beha io emains pe sis en . Thi d, as Va ese (2011) no es, success ul c iminal beha io in a new and unknown en i on- men does no only equi e a c iminal mind, bu also an oppo uni y. I migh ake some ime a e en e ing he des ina ion coun y o comp ehensi ely adap o he new en i on- men , and o ind ways and means o success ul co up ion. I immig an s show pe sis en co up ion a i udes, he ull e ec o immig an s’ co up beha io may become isible in he a ge coun y only a e some pe iod o ime. This leads us o ou nex hypo hesis. Hypo hesis H2: The e ec on he des ina ion coun y’s le el o co up ion, ela ed o immi- g a ion om a mo e co up sending coun y o a less co up des ina ion coun y, is posi- i e. Howe e , i migh ake some ime be o e he e ec becomes no iceable. In he ollowing sec ion, we will es ou hypo heses o in es iga e which e ec s domina e. Be o ehand, a ca ea is in o de . Endogenei y is a widely acknowledged issue in empi ical co up ion esea ch.7 As is gene ally ue o empi ical panel da a esea ch, a co ela ion ex- poses a gene al cohe ence a he han ende ing a clea causal ela ionship. In ou case, co up ion could po en ially be bo h he an eceden and he e ec o o he ac o s. As al- eady indica ed abo e, Coo ay and Schneide (2014) and Diman e al. (2013) show e e se causali y be ween mig a ion and co up ion, inding ha excessi e co up ion decisi ely impac s mig a ion lows. Hence, immig a ion migh e y well lea e a co up oo p in in he 7 Fo example, he li e a u e indica es ha he ela ionship be ween co up ion and economic g ow h also holds in he e e se di ec ion (c . D ehe and Gassebne (2013); suppo ing he ‘g easing- he-wheels’ hypo hesis, and Méon and Sekka (2005), suppo ing he ‘sanding- he-wheels’ hypo hesis). 49 des ina ion coun y because o (sel -) selec ion e ec s. I , o ins ance, an hones ou side decides o lea e a co up coun y i is unlikely ha he/she will ( olun a ily) choose an equally co up des ina ion. Tha is, he le el o co up ion in he des ina ion coun y migh be ele an in shaping mig a ion lows. E iden ly, i is impo an o con ol o endogenei y as he esul s migh po en ially su e om a e e se-causali y bias. Ou app oach o how o add ess his p oblem will be p esen ed in he ollowing sec ion. 2.2.3 Empi ical Analysis The Empi ical Model Based on he p e ious heo e ical conside a ions, he discussion in his and he ollowing sec ion aims a es ing he hypo heses de eloped in Sec ion 2.2.2. Ou s a ing poin is a panel model o he o m i co up ion = α + ф i q m ig a io n  + β 1 i X   + i  + i  whe e i co up ion is he le el o co up ion in coun y i and yea , i q m ig a io n  is he o al mig a ion s ock wi h a ime lag q, 1 i X   is a condi ioning se o lagged con ol a iables and he dis u bance e m is composed o he indi idual e ec ηi and he s ochas ic dis u bance i  which is assumed o be gene a ed by a whi e noise p ocess. This speci ica ion allows es - ing he gene al e ec o mig a ion on co up ion acco ding o hypo hesis H1. Since we assume he mig a ion a iable o ha e a ime-shi e ec on co up ion, we le he independen a iable o in e es en e he model wi h a ime lag q, which may ake alues om one o i e i , o example, he maximum lag is i e yea s. This lag s uc u e allows us o di e en ia e be ween immedia e and delayed e ec s. Addi ionally, lagging he independ- en a iable o in e es dampens he p oblem o a possible endogenous ela ionship be- ween co up ion and mig a ion by elimina ing he co ela ion be ween he explana o y a iables and he e o e m. We epo he Akaike In o ma ion C i e ion (AIC) and Bayesian In o ma ion C i e ion (BIC) o allow o a compa ison o he model i o he al e na i e lag 50 selec ions.8 Assuming con ol a iables also do no ha e an immedia e e ec in he same pe iod, all o he con ols en e he model wi h a lag o -1. We p o ide esul s o ixed-e - ec s panel eg essions ha allow us o accoun o coun y-speci ic e ec s. Fu he mo e, we explo e he e ec o immig a ion om highly co up coun ies on he co - up ion le el o he a ge coun ies acco ding o hypo hesis H2. This is es ed by eg essing he o al mig a ion s ock om coun ies ha exhibi a co up ion le el which is highe han he o al a e age o e all 207 coun ies o mig a ion o igin on he co up ion le el o he des ina ion coun y, so ha we can es i a highe mig a ion s ock om mo e han p opo - ionally co up coun ies d i es he co up ion le el in he des ina ion coun y. Dealing wi h Po en ial Endogenei y We accoun o po en ial endogenei y by applying a Di e ence-GMM es ima ion in o de o exclude esul s ha migh be d i en by he unde lying econome ic app oach, and hus do no allow o s a is ically eliable in e ence. The dynamic GMM app oach de eloped by A el- lano and Bond (1991) appea s as an app op ia e app oach, as i allows calcula ing con- sis en and e icien es ima es by using lagged le els da ed in pe iod -2 and ea lie as in- s umen s. The co esponding momen condi ion can be checked using he Sa gan s a is ic ha es s he alidi y o he ins umen s. In ollowing A ellano and Bond (1991), we p o ide esul s o Di e ence-GMM es ima ions. In gene al, he esul s a e in line wi h he es ima ions p esen ed be o e. An al e na i e es i- ma ion p oposed by A ellano and Bo e (1995) and Blundell and Bond (1998) is he Sys em- GMM app oach, which pe o ms well wi h highly pe sis en da a unde mild assump ions. Howe e , he e is an impo an poin o be made abou using Sys em-GMM. Gi en ha Sys- em-GMM uses mo e ins umen s han he Di e ence-GMM, i may no be app op ia e o use Sys em-GMM wi h a da ase ha consis s o a small numbe o coun ies. In his case, his me hod is likely o exhibi a ini e sample bias as he numbe o ins umen s inc eases ex- ponen ially wi h he numbe o pe iods used. As a gued by Roodman (2009), such an o e - i ing o endogenous a iables is likely o lead o alse posi i e esul s. In addi ion, he as- sump ion o lagged con ol a iables being exogenous o he e o e m is non- i ial. Fo his 8 Plümpe e al. (2005) illus a e ha in ixed-e ec s models he lag s uc u e o he independen a iable has a la ge impac on he coe icien and he le el o signi icance. They a gue ha he e is no gene ally accep ed indica o o he de e mina ion o he leng h o he lag, howe e , he e a e se e al candida es like he -s a is ic, he R2, he AIC and he BIC ha acili a e he choice. 51 eason, we eso o he Di e ence-GMM app oach, as he a io o coun ies and ime pe i- ods used in ou panel is well balanced, hus uling ou a po en ial small sample bias (c . Alonso-Bo ego and A ellano (1999)).9 2.2.4 Da a Dependen Va iable We use he c oss-na ional co up ion a ing om he In e na ional Coun y Risk Guide (ICRG) (PRS G oup, 2008). I elies on he subjec i e assessmen o coun y expe s ypically ope a ing wi hin in e na ional non-go e nmen al o ganiza ions. As a componen o he po- li ical igh s index, i is conce ned wi h ac ual o po en ial co up ion in he o m o excessi e pa onage, nepo ism, job ese a ions, ‘ a o - o - a o s’, sec e pa y unding and suspi- ciously close ies be ween poli ics and business.10 O iginally, he alue o he index anges om 0 o 6, wi h 0 indica ing a high le el o co up ion and 6 ep esen ing a low le el. We anspose he scale o simpli y he in e p e a ion o he esul s so ha highe alues o he index indica e a highe ex en o co up ion. The main ad an age o his index is ha i is a ailable annually o a la ge sample o coun ies beginning in he ea ly 1980s, and so ena- bles us o analyze he co up ion–mig a ion nexus wi hin a panel amewo k.11 The sum- ma y s a is ics can be ound in Table S1. Main Independen Va iable o In e es Ou main independen a iable is immig a ion (mig a ion). We use he OECD In e na ional Mig a ion Da abase which p o ides annual se ies on mig a ion lows and s ocks in o OECD coun ies om 207 coun ies o o igin o he pe iod 1975–2011 (OECD, 2014a). The majo ad an age o his da ase is ha i p o ides bila e al da a and so allows dis inguishing be- ween coun ies o des ina ion and coun ies o o igin, allowing us no only o analyze he gene al e ec o mig a ion on co up ion bu also o g oup sou ce coun ies acco ding o 9 To check he obus ness o he model speci ica ion, we also un all eg essions using he Sys em-GMM. The esul s sup- po ou main indings. Howe e , he ule o humb – o keep he numbe o ins umen s less han o equal o he numbe o g oups – canno be me . E en i only he second lag is used as an ins umen o he Sys em-GMM he numbe o ins u- men s exceeds he numbe o coun ies. The esul s a e a ailable on eques . 10 h p://www.p sg oup.com/ICRG_Me hodology.aspx. 11 O he common co up ion measu es like he Co up ion Pe cep ions Index (CPI) cons uc ed by T anspa ency In e na- ional o he Con ol o Co up ion Ra ing published by he Wo ld Bank a e a ailable only om 1995 and 1996, espec i ely. S ensson (2005) and T eisman (2007) show ha all h ee measu es a e highly co ela ed. 52 hei le el o co up ion. We weigh mig a ion by he espec i e des ina ion coun y’s popu- la ion in housands in o de o accoun o he inhe en popula ion size he e ogenei y ac oss he OECD coun ies. Since di e en coun ies use di e en de ini ions o immig a- ion12 and di e en sou ces o hei mig a ion s a is ics, he OECD da abase o e s bo h da a on immig an s by na ionali y and on immig an s by coun y o bi h. Especially in he case o he mig a ion s ock a iable, he di e ences in he de ini ion play an impo an ole and mus be conside ed. The ‘coun y o bi h’ app oach akes in o ac- coun he o eign-bo n popula ion, o example, he i s gene a ion o immig an s, includ- ing immig an s ha ha e ob ained ci izenship. The ‘na ionali y’ app oach includes second and highe gene a ions o o eigne s, bu does no co e na u alized ci izens. Thus, he na- u e o he coun ies’ legisla ion on ci izenship and na u aliza ion plays a ole (Pede sen e al. (2008)). We use he immig an s s ock by ‘na ionali y’ a iable o h ee easons. Fi s , his a iable is a ailable o mo e coun y- ime obse a ions han he immig a ion s ock by ‘coun y o bi h’, hus allowing o mo e meaning ul es ima ions. Second, we ac on he assump ion ha na u alized ci izens should be pu on an equal oo ing wi h he domes ic popula ion as i is easonable o assume ha he na u alized ci izens’ magni ude o assimi- la ion is well ad anced. Thi d (and closely connec ed o he p e ious a gumen ), ou hy- po hesis H2 akes he assimila ion p ocess in o conside a ion assuming ha he ull e ec o immig a ion on des ina ion coun y co up ion occu s only a e some ime. The ‘na ion- ali y’ app oach akes up his ime dimension mo e na u ally. Con ol Va iables To a oid spu ious ela ionships be ween he dependen a iable and he independen a - iable o in e es , we employ a se o con ol a iables commonly iden i ied as po en ial de- e minan s o co up ion. In ou baseline speci ica ion we con ol, i s , o he impac o economic de elopmen measu ed by (logged) eal pe capi a income (GDP p.c.) ex ac ed om he Penn Wo ld Table (PWT) (Hes on, e al., 2012). I is a commonly used a iable o explain co up ion. The heo e ical a gumen s esses ha economic de elopmen os e s 12 Coun ies like Aus alia, Canada, he Ne he lands, New Zealand, Poland, he Slo ak Republic and he Uni ed S a es de- ine an ‘immig an ’ by coun y o o igin o coun y o bi h, while some coun ies like Aus ia, he Czech Republic, Denma k, Finland, G eece, Iceland, I aly, No way and Sweden de ine an immig an by ci izenship and inally some coun ies like Bel- gium, F ance, Hunga y, Ge many, Japan, Luxembou g, Po ugal, Spain, Swi ze land and he Uni ed Kingdom ely on sel - epo ed na ionali y (Pede sen, e al., 2008). 53 highe ins i u ional quali y, which in u n will p o ide ewe b eeding g ounds o co up ion ia implemen a ion o mo e sophis ica ed an ico up ion measu es. A highe chance o iden i ica ion and punishmen o co up ion will inc ease he expec ed cos s, and c owd ou incen i es o engage in de ian beha io (c . Se a (2006)). Along hese lines, se e al empi - ical s udies ind a obus nega i e co ela ion be ween economic de elopmen and pe - cei ed co up ion, sugges ing ha poo e coun ies exhibi highe co up ion a es (c . La Po a e al. (1999); Se a (2006); T eisman (2007)). Howe e , panel s udies based on ixed- e ec es ima ion by B aun and Di Tella (2004) ind ha an inc ease in a coun y’s weal h measu ed by GDP pe capi a also inc eases co up ion. A po en ial explana ion o a posi i e nexus be ween g ow h and co up ion is p o ided by Kindlebe ge (2000). He a gues ha mo al s anda ds anish in imes o economic booms due o a mo e p onounced mani es a- ion o g eed, e en ually unde mining he indi idual’s disposi ion o obey he law. O e all, we ollow he empi ically se led mains eam a gumen and expec ha mo e de eloped coun ies (in e ms o GDP pe capi a) expe ience lowe a es o co up ion. We also accoun o he e ec o popula ion size (PWT) on co up ion. F om he heo e ical pe spec i e, Knack and Az a (2003) sugges ha la ge poli ies may bene i om economies o scale in es ablishing poli ical and adminis a i e s uc u es, so ha a la ge coun y size migh be nega i ely co ela ed wi h co up ion. On he o he hand, small coun ies may bene i om highe manageabili y, and mo e e iciency and anspa ency in adminis a i e managemen , leading o a posi i e co ela ion be ween popula ion size and co up ion. Em- pi ical e idence shows mixed esul s. Fo one, Knack and Az a (2003) show ha he e is in- deed no clea ela ionship be ween coun y size and co up ion and ha exis ing esul s su e om selec ion bias. On he o he hand, a c oss-coun y s udy by Ta a es (2003) shows a nega i e impac o popula ion on co up ion, while Roo (1999) inds ha a la ge popula- ion is signi ican ly associa ed wi h mo e co up ion indica ing ha smalle coun ies a e less co up han la ge coun ies. We ollow he majo i y o exis ing e idence and assume ha popula ion size and ex en o co up ion go hand in hand, due o a highe numbe o po en ial b ibe s and b ibees and issues o e ec i e moni o ing, which a e likely o be mo e ex ensi e wi h a g owing popula ion size. Ali and Isse (2003) a gue ha a la ge go e nmen sec o (go e nmen size) may c ea e op- po uni ies o co up ion. The la ge he size and scope o he bu eauc acy, he mo e likely 54 i is o ind co up beha io . On he con a y, Goel and Nelson (2010) indica e ha go e n- men size migh be in e sely ela ed o a coun y’s co up ion le el. No a la ge public sec o pe se de e mines he magni ude o co up ac i i y, bu la ge go e nmen s migh in ac de o e a highe sha e o public spending o ope a i e law en o cemen aimed a de e ing de ian beha io (c . Fisman and Ga i (2002); Goel and Nelson (2010)).13 Al hough no ex- plici ly es ed o a subse o OECD coun ies, we ollow he majo i y o exis ing empi ical e idence and expec a la ge go e nmen sec o o ha e a nega i e impac on a coun y’s co up ion le el. The a iable is measu ed by go e nmen expendi u e di ided by GDP and ex ac ed o m he Wo ld De elopmen Indica o s (Wo ldBank, 2013). Fu he mo e, we con ol o democ acy measu ed by he Poli y2 index (Poli y IV P ojec , 2012), which is ound o be highly ele an in exis ing heo e ical and empi ical esea ch on co up ion. In gene al, bo h s ands o esea ch indica e ha mo e democ a ic coun ies end o be less co up (e.g., B aun and Di Tella (2004); Knack and Az a (2003); Kunico á and Rose-Acke man (2005); Shen and Williamson (2005)). F om a heo e ical pe spec i e, Shen and Williamson (2005) con end ha s a es wi h democ a ic go e nmen s a e likely o ha e mo e sophis ica ed policies and legal ins i u ions ha a e mo e likely o be independen o he eli es’ impai men . Seldadyo and de Haan (2006) a gue ha poli ical libe y imposes anspa ency and p o ides checks and balances wi hin he poli ical sys em and so ends o educe co up ion. Kunico á and Rose-Acke man (2005) sugges ha elec o al ules and po- li ical s uc u es can in luence he le el o co up ion. Poli ical pa icipa ion, poli ical com- pe i ion and cons ain s on he chie execu i e make i easie o moni o he poli ical sys em and limi poli ical co up ion.14 O e all, since bo h heo y and empi ics esona e wi h each o he , we would expec a nega i e impac o democ acy on co up ion. In addi ion, exis ing esea ch acknowledges he impo an link be ween economic eedom (ICRG) and co up ion. F om a heo e ical pe spec i e, one can a gue ha , especially in mode n economies, many es ic ions on economic eedom– in pa icula es ic ions o 13 Howe e , i is wo h no ing ha pa s o he exis ing li e a u e also poin a a di e en ela ionship be ween go e nmen sec o and co up ion. Co up go e nmen s may impose de imen al e ec s on public goods deli e y, weaken he ax mo ale and he bu eauc a ic quali y whose unc ional in e ac ion, ce e is pa ibus, likely leads o a smalle go e nmen sec o (c . F ey and To gle (2007); Hall and Jones (1999); Johnson e al. (1997); Tanzi (2013)). 14 T eisman (2007) indica es ha he ela ionship be ween democ acy and co up ion migh be mo e complex, sugges ing ha democ a iza ion inc eases co up ion in he sho un and educes i as democ acy deepens. Howe e , he composi- ion o ou da a does no allow us o examine long- un e ec s o con ols such as democ a iza ion. Thus, we eso o a sho - un examina ion o he con ol’s impac on co up ion. 55 capi al and inancial ma ke s – p o ide oppo uni ies o co up ion (c . G ae and Mehlkop (2003)). This no ion is s ongly suppo ed by he empi ical li e a u e. Goel and Nelson (2005) ind a s ong nega i e ela ionship be ween economic eedom and co up ion, whe e he ela ionship depends on a coun y’s le el o de elopmen . Paldam (2002) p esen s simila esul s sugges ing ha coun ies wi h high egula ion and li le economic eedom ha e a la ge po en ial o en seeking, esul ing in highe le els o co up ion. Suppo i e esul s o a nega i e ela ionship be ween economic eedom and co up ion a e also ound by Ali and Isse (2003), and Kunico á and Rose-Acke man (2005). We expec ha mo e economic eedom and ewe es ic ions imposed on ade a e in e sely co ela ed wi h a coun y’s co up ion le el. We measu e economic eedom by he in es men p o ile a iable o he ICRG, a guing ha a high in es men isk accompanies lowe economic eedom.15 Finally, eligion may also ma e o explaining co up ion. Religion is belie ed o play a de- cisi e ole in a ec ing co up ion le els h ough i s inhe en he e ogenei y in pu ing em- phasis on mo al alues, hones y and being in h all o au ho i y. Consequen ly, eligious s uc u es ha a e mo e hie a chical a e belie ed o be mo e conduci e o he incep ion and de elopmen o co up s uc u es (Paldam, 2001). Empi ical esea ch inds ha coun- ies wi h a p edominan ly p o es an popula ion end o ha e lowe co up ion le els, while mo e hie a chically s uc u ed eligions (such as Ca holicism, Eas e n O hodoxy and Islam) end o inc ease co up ion (La Po a, e al., 1999). We ollow Blombe g and Hess (2008) in using eligious ensions as a con ol in o de o ge an imp ession whe he a dom- inan ole o a speci ic eligious g oup, and he supp ession o eligious eedom, has an e - ec on he le el o co up ion. The a gumen is ha a dominan eligion in a coun y c ea es di e en ial access o powe , leading o a si ua ion in which less powe ul eligious g oups eso o co up ion o le eling he poli ical and economic landscape. A se o a iables does no en e ou baseline model, in pa icula economic g ow h, ade openness, in e nal and ex e nal con lic s and egime s abili y. Ra he , hey a e used o as- sess he obus ness o ou indings. The i s o hese a iables is economic g ow h (PWT) (in addi ion o he le el o de elopmen ). Ali and Isse (2003) a gue ha i coun ies wi h lowe co up ion le els g ew as e , his posi i e expe ience ough o gi e way o a s ic e igh 15 As an al e na i e, we also employed he ‘Economic F eedom’ index p o ided by he F ase Ins i u e (Gwa ney & Lawson, 2008). The esul s (based on a signi ican ly smalle da ase ) suppo ou main indings and a e a ailable on eques . 56 agains co up ion in he u u e. Tha is, economic g ow h should be nega i ely co ela ed wi h u u e co up ion. Howe e , he empi ical e idence on his a gumen is mixed. While Lei e and Weidmann (1999) ind ha GDP g ow h has a dampening e ec on coun y le el co up ion, Be die e al. (2013) ind he opposi e e ec . Howe e , o he subse o OECD coun ies (in which we a e in e es ed in) hei esul s emain insigni ican . Fu he mo e, o he s udies ind no signi ican e ec a all (c . Ali and Isse (2003); B une i e al. (1997); Mau o (1995)). Consequen ly, due o he ocus on he same subse o coun ies, we expec ou esul s o be in line wi h Be die e al. (2013) o hei subse o OECD coun ies and ex- pec no signi ican e ec in ei he di ec ion o GDP g ow h on co up ion le els. We u he mo e assess he impac o ade openness – measu ed by expo s and impo s as a sha e o GDP (PWT) – as an indica o o compe i ion.16 Lei e and Weidmann (1999) sugges ha openness o o eign ade, which is equi alen o a ela i ely s ong economic compe- i ion, is a p ima y ac o o expe iencing ela i ely low le els o co up ion. This a gumen is backed up by empi ical esea ch. Sandhol z and Koe zle (2000) ind ha economic in e- g a ion dec eases co up ion ac i i y, albei no di ec ly.17 In pa icula , he exis ing e- sea ch sheds ligh on he in e ela ion be ween openness o inancial ma ke s and co up- ion le els. Al hough no en i ely cong uen , o he mos pa he exis ing li e a u e poin s a he idea ha es ic ions b ing abou indi idual e o o bypass egula ions wi h he use o de ian beha io , such as b ibing public o icials (c . D ehe and Sieme s (2009); Edwa ds (1999)). We hus expec an in e se ela ionship be ween ade openness and co up ion, which is along he lines o he p e ious discussion on he impac o economic eedom on co up ion. We also accoun o a po en ial e ec o in e nal and ex e nal con lic s (ICRG) on co up ion. Con lic s – in e ms o domes ic and ansna ional e o ism o ci il wa – may ha e a des a- bilizing e ec on he economy which is wha we expec o show up in ou analysis. Fo in- s ance, D ehe e al. (2010) and Meie ieks and G ies (2013) show ha e o ism a ec s he 16 Al e na i ely, we p oxy ade openness by he a io o impo o GDP (He z eld & Weiss, 2003). He e, a low impo sha e implies high impo es ic ions. Consequen ly, he p esence o such es ic ions o e s an oppo uni y o b ibe (Seldadyo & De Haan, 2006). 17 Howe e , Knack and Az a (2003) a gue ha ade sha e and impo sha e o GDP a e s ongly ela ed o coun y size. Smalle coun ies end o ha e a highe ade sha e, so no con olling o popula ion he coe icien on openness is likely o e lec selec ion bias. 57 economy nega i ely and con ibu es o poli ical ins abili y. This in u n may c ea e a b eed- ing g ound and may also p o ide oppo uni ies o co up ion. Rega dless o he egime ype, egime s abili y (Poli y IV P ojec , 2012) is ano he poli ical a iable ha may a ec co up ion. As sugges ed by T eisman (2007), i may ake decades o democ a ic ins i u ions o ansla e in o low pe cei ed co up ion so ha no he cu en egime ype bu he egime s abili y a ec s he co up ion le el. This is suppo ed by an ex- eme-bounds analysis by Se a (2006), who p o ides e idence ha ac ual democ acy is weakly in e ela ed wi h co up ion, whe eas poli ical s abili y measu ed by unin e up ed democ acy esul s in educing co up ion. I is easonable o assume ha he poli ical ac- uums inhe en o uns able egimes enable auds e s o mo e easily ind means o p ecipi- a e success ul ac s o co up ion. Consequen ly, and in compliance wi h he p e iously lead discussion on he in e ela ion be ween democ acy and co up ion, we expec coun ies wi h s able egimes o be less p one o co up ion. 2.2.5 Empi ical Resul s In his sec ion, we epo ou empi ical esul s using di e en econome ic app oaches o ensu e obus ness and o accoun o possible endogenei y issues. Table 1 p esen s esul s o he baseline model wi h an al e na i e lag leng h o bo h ixed e ec s (FE) and Di e - ence-GMM es ima ions, while Tobi esul s a e gene ally p esen ed in he suppo ing in o - ma ion.18 18 I should be no ed ha he esul s o he GMM es ima es di e om hose o he FE and he Tobi es ima es in some cases. This has a leas wo easons. Fo one, he Di e ence-GMM includes he lag o he dependen a iable as an addi- ional eg esso , esul ing in a educ ion in he numbe o obse a ions o abou 10%. Fo ano he , based on he ule o humb, which decla es he numbe o ins umen s o be smalle han he numbe o c oss sec ions, only one lag is used o ins umen a ion. Howe e , in his case he GMM es ima o is no necessa ily e icien since i does no make use o all a ailable momen es ic ions. 64 Table 2.1.2: Con inued om P e ious Page VIF 1.28 1.98 1.77 1.58 1.34 2.19 1.91 1.58 Adjus ed R2 0.310 0.379 0.396 0.411 0.443 0.483 0.489 0.493 AIC 696.9321 672.1169 661.6321 648.9388 420.2725 394.3588 392.0972 387.4652 BIC 725.3623 700.7084 698.3926 677.5303 447.2379 421.3242 426.7670 414.4306 Obse a ions 429 439 439 439 348 348 348 348 No e: * p < 0.10, ** p < 0.05, *** p < 0.01; obus s anda d e o s in pa en heses; mig a ion s ock is weigh ed by popula ion. 65 Finally, he Di e ence-GMM es ima ions yield a signi ican and posi i e e ec o egime s a- bili y, indica ing ha , in he sho un, coun ies ha a e weal hie and possess a mo e s a- ble egime s uc u e a e mo e p one o co up ion. While he o e all di ec ion is he same, he e ec ’s magni ude is mo e conse a i e han he coe icien s de i ed om he FE ap- p oach and only shows up signi ican ly in he sho un. We p esen he Di e ence-GMM e- sul s in Table 3.21 A e we could no iden i y a consis en and signi ican e ec o gene al immig a ion on co - up ion suppo ing H1, we now u n o he ques ion aised in hypo hesis H2, whe he co - up ion mig a es and how long i may ake o in il a e he des ina ion coun y. Mo e speci - ically, we explo e whe he immig an s om highly co up coun ies ca y o e hei beha - io , so ha immig a ion om coun ies wi h a le el o co up ion ha is highe han he a - e age leads o an inc ease in co up ion in he des ina ion coun y. The esul s o his exe - cise a e shown in Table 4. Again, we p esen he FE and Di e ence-GMM es ima ions join ly. The esul s a e based on a calcula ion o he o al a e age o co up ion le els o e all coun- ies o each yea om which we hen de i e he mos co up coun ies a he op 50% le el.22 O e all, he esul s indica e ha immig a ion om highly co up coun ies boos s he co - up ion le el in he hos coun y, hus suppo ing ou hypo hesis H2. Acco ding o he FE es ima ion, we ind a signi ican and posi i e e ec o selec ed mig a ion on hos coun ies’ co up ion le els. The coe icien ises o a alue o 0.0099 ( o a lag o h ee pe iods), which means ha an inc ease in he mig a ion s ock o one hund ed mig an s pe one housand ci izens a ec s co up ion signi ican ly, inc easing he co up ion alue by 0.99 poin s (ou o 7). This is a aise o 14.1% o he maximum scale. As i has p e iously been he case, he esul s o he Di e ence-GMM es ima ions a e mo e conse a i e, hus ep esen ing a lowe bound esul wi h a aise o up o 4.4% o he maximum scale o he same inc ease in he mig an s- o-ci izens a io. Con e sely, he esul s o he al e na i e Tobi eg essions poin o an uppe bound esul , indica ing a aise o up o 18.1% o he maximum scale. In gene al, 21 We also calcula e a Tobi e sion o he eg essions wi h and wi hou al e na i e con ol a iables. The esul s a e in line wi h he FE es ima ion and a e p esen ed in Table S3. 22 Ou indings do no change when we conside mig a ion om e en mo e co up coun ies a he op 40% (30%, 20% and 10%) le el. 66 he Di e ence-GMM esul s a e mo e conse a i e and u n ou o be signi ican less o en compa ed o he FE and Tobi es ima ions. We ace his back o he limi ed amoun o c oss sec ions, which is a p oblem inhe en o ou ocus on OECD coun ies.23 Fu u e esea ch migh po en ially o e come his d awback by ex ending he esea ch scope beyond OECD coun ies. No iceably, while we ini ially obse e an escala ing e ec o selec i e mig a ion on co up- ion le els, he esul s a e indica i e o an assimila ion p ocess o e ime. 23 Fu he possible explana ions we e o e ed in he beginning o his sec ion. 67 Table 2.1.3: Mig a ion and Co up ion - Di e ence-GMM Reg ession wi h Al e na i e Con ols co up ion (1) (2) (3) (4) (5) (6) (7) (8) co up ion -1 0.4531*** (0.1466) 0.4341*** (0.1587) 0.4909*** (0.1740) 0.5372** (0.2237) 0.3699* (0.1919) 0.3327* (0.1975) 0.3743* (0.2000) 0.6012*** (0.0921) mig a ion -1 0.0006 (0.0038) -0.0026 (0.0061) -0.0006 (0.0047) -0.0040 (0.0056) mig a ion -5 0.0075 (0.0053) 0.0050 (0.0037) 0.0064* (0.0038) 0.0030 (0.0032) GDP p.c. -1 -0.0115 (0.5432) -1.0297*** (0.3747) -0.1208 (0.2843) -0.9525 (0.7889) popula ion -1 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000) -0.0000 (0.0000) go size -1 -0.0115 (0.0337) 0.0006 (0.0288) -0.0102 (0.0328) -0.0189 (0.0337) -0.0096 (0.0259) 0.0087 (0.0288) 0.0042 (0.0260) -0.0257 (0.0278) democ acy -1 -0.0131 (0.0251) -0.0182 (0.0237) 0.0068 (0.0313) 0.0207 (0.0950) -0.0259 (0.1338) 0.0672 (0.0821) econ eedom -1 0.0669*** (0.0254) 0.0604** (0.0242) 0.0567** (0.0222) 0.0406** (0.0162) 0.0464*** (0.0127) 0.0382*** (0.0131) 0.0423*** (0.0154) 0.0192 (0.0137) eligious ension -1 0.0418 (0.0861) 0.0520 (0.0740) 0.0792 (0.0865) 0.0707 (0.0913) 0.0674 (0.0583) 0.0735 (0.0612) 0.0555 (0.0708) 0.1102* (0.0609) GDP p.c. g ow h -0.5576 (0.8686) -0.7175 (0.7638) ade openness -1 0.0033 (0.0043) 0.0021 (0.0030) in e nal con lic -1 -0.0129 (0.0380) -0.0896* (0.0458) ex e nal con lic -1 0.0311 (0.0384) 0.0822* (0.0430) egime s abili y -1 0.0457** (0.0201) 0.0397 (0.0250) 68 Table 2.1.3: Con inued om P e ious Page Sa gan (p-Value) AR (2) (p- alue) 0.5787 0.2176 0.6018 0.2107 0.7435 0.2153 0.6894 0.1494 0.7741 0.5403 0.7274 0.5610 0.7305 0.6099 0.7374 0.4782 Ins umen s 30 31 33 31 27 27 29 27 Obse a ions 396 406 406 406 319 319 319 319 No e: * p < 0.10, ** p < 0.05, *** p < 0.01; obus s anda d e o s in pa en heses; GMM esul s based on he wo-s ep Di e ence-GMM es ima o , second lag o he dependen a iable used as GMM-s yle ins umen ; AR (2) e e s o he A elano Bond es o au o eg essi e co ela ion (o de 2); Sa gan e e s o he Sa gan es o o e iden i ica ion es ic ions; mig a ion s ock is weigh ed by popula ion. 69 Table 2.1.4: Mig a ion om Co up Coun ies and Co up ion - Fixed E ec s and GMM Reg ession co up ion Fixed E ec s GMM (1.1) (1.2) (1.3) (1.4) (1.5) (2.1) (2.2) (2.3) (2.4) (2.5) co up ion -1 0.5278*** (0.1875) 0.4694** (0.1974) 0.5085*** (0.1926) 0.3542** (0.1716) 0.3792* (0.2219) mig a ion -1 0.0087** (0.0037) -0.0016 (0.0014) mig a ion -2 0.0095*** (0.0033) 0.0026* (0.0014) mig a ion -3 0.0099*** (0.0030) 0.0031** (0.0015) mig a ion -4 0.0073*** (0.0023) -0.0004 (0.0008) mig a ion -5 0.0040* (0.0020) -0.0007 (0.0011) GDP p.c. -1 1.0742* (0.5622) 1.0156 (0.5998) 1.0565 (0.6270) 1.2442** (0.5403) 1.6372*** (0.4360) -0.3686 (0.5195) -0.3201 (0.3661) -0.4154 (0.4664) -0.1656 (0.3887) 0.1622 (0.5569) popula ion -1 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) go size -1 0.0522 (0.0388) 0.0734** (0.0338) 0.0855*** (0.0290) 0.0997*** (0.0273) 0.0989*** (0.0250) -0.0109 (0.0332) -0.0078 (0.0265) -0.0171 (0.0353) -0.0013 (0.0272) 0.0050 (0.0358) democ acy -1 -0.1738*** (0.0295) -0.1251** (0.0512) 0.0682 (0.1657) 0.1052 (0.1667) 0.1565 (0.1573) 0.0332 (0.0887) 0.0210 (0.2159) -0.0095 (0.1567) 0.0446 (0.1538) -0.0003 (0.1068) econ eedom -1 0.0741** (0.0327) 0.0721** (0.0348) 0.0672* (0.0363) 0.0616* (0.0313) 0.0425* (0.0229) 0.0505** (0.0212) 0.0382* (0.0230) 0.0410** (0.0160) 0.0393** (0.0171) 0.0433** (0.0192) eligious ension -1 0.2573* (0.1401) 0.2462* (0.1341) 0.2584* (0.1303) 0.2035 (0.1209) 0.1564 (0.1077) -0.0273 (0.0807) -0.0234 (0.0752) -0.0104 (0.0795) -0.0405 (0.0418) -0.0564 (0.0480) 70 Table 2.1.4: Con inued om P e ious Page VIF Adjus ed R2 1.24 0.382 1.26 0.381 1.27 0.423 1.26 0.459 1.27 0.494 AIC 625.1091 579.3207 501.4671 432.3486 367.1703 BIC 653.2051 607.0477 528.7663 459.2535 393.6571 Sa gan (p- alue) 0.7940 0.9690 0.8070 0.8624 0.6690 AR (2) (p- alue) 0.3989 0.3811 0.6287 0.6532 0.8664 Ins umen s 31 30 29 28 27 Obse a ions 409 388 365 345 325 379 358 338 318 298 No e: * p < 0.10, ** p < 0.05, *** p < 0.01; obus s anda d e o s in pa en heses; GMM esul s based on he wo-s ep Di e ence-GMM es ima o , second lag o he dependen a iable used as GMM-s yle ins umen ; AR (2) e e s o he A elano Bond es o au o eg essi e co ela ion (o de 2); Sa gan e e s o he Sa gan es o o e iden i ica ion es ic ions; mig a ion s ock is weigh ed by popula ion. 71 These obse a ions a e in line wi h he p e iously discussed a gumen s p esen ed by Chis- wick (1978), which a e suppo i e o he idea ha he mig an s’ assimila ion happens a di e en speeds.24 Mo eo e , he esul s o he con ol a iables a e b oadly in line wi h ou p e ious indings p esen ed in Tables 1 and 2. 2.2.6 Conclusion In his pape , we shed ligh on he impac o mig a ion on co up ion in he des ina ion coun y. Capi alizing on a comp ehensi e da ase consis ing o annual se ies on mig a ion lows and s ocks in o OECD coun ies om 207 sending coun ies o he pe iod 1984–2008, we explo ed di e en channels h ough which co up ion migh mig a e. Ini ially, he impli- ca ions migh go in o a ious di ec ions as di e en e ec s a e in place a he same ime. On one side, he exis ing li e a u e sugges s ha mig a ion could be he esul o a posi i e se- lec ion. Fo example, highly skilled people migh lea e hei home coun ies as hey expec hei indi idual li ing condi ions o imp o e. On he con a y, howe e , poo socioeconomic condi ions ypically cons i u e push ac o s o mig a ion, no only o a small posi i e selec- ion o hones people bu also o he co up ible a e age indi idual. Independen o he econome ic me hodology applied, we consis en ly ind ha (i) gene al mig a ion has an insigni ican e ec on he des ina ion coun y’s co up ion le el, and (ii) ha immig a ion om co up ion- idden coun ies boos s co up ion in he des ina ion coun y. This holds e en a e con olling o po en ial endogenei y by means o a Di e - ence-GMM es ima ion. Hence, he ea by in e na ional legisla o s (as exp essed in ecen ag eemen s by he G20 g oup) ha immig a ion may cause a p oblema ic in low o co up ion appea s jus i ied. Policy-make s will, he e o e, ha e o ake p ecau ions o a oid his p oblem. Howe e , i is no immedia ely ob ious wha he op imal esponse will be. One possibili y could be o e- s ic immig a ion by only selec ing immig an s o igina ing om non-co up coun ies. Al- e na i ely, e y ca e ul checks ad pe sonam could be conduc ed. The downside o his pol- 24 The esul s o he Tobi eg ession a e p esen ed in Table S4 in he online appendix. The esul s a e cohe en and su i e when using he al e na i e se o con ol a iables. 72 icy is ha he emaining in low o mig an s could be a he small, which migh no be op i- mal, gi en ha mos OECD coun ies ace a se e e ageing p oblem and a e in need o immi- g a ion o keep hei social secu i y sys ems sus ainable. An a guably be e s a egy could be o immunize he domes ic popula ion agains a co up a i ude b ough in o he coun y by some immig an s. This would be in line wi h Va ese’s (2011) a gumen which we may e- ph ase as ollows: success ul co up ion needs bo h a co up mind and an oppo uni y. 73 CHAPTER 3: EXPERIMENTS 3.1 ON PEER EFFECTS: BEHAVIORAL CONTAGION OF (UN)ETHICAL BEHAVIOR AND THE ROLE OF SOCIAL IDENTITY Eugen Diman * Ve sion: Decembe 2015 ABSTRACT: Social in e ac ions and he esul ing pee e ec s loom la ge in bo h economic and social con- ex s. This is pa icula ly ue o he spillo e o (un)e hical beha io in explaining how beha io and no ms sp ead ac oss indi idual people, neighbo hoods, o e en cul u es. Al hough we un- de s and and obse e he ou comes o such con agion e ec s, li le is known abou he d i e s and he unde lying mechanisms, especially wi h espec o he ole o social iden i y wi h one’s pee s and he (un)e hicali y o beha io one is exposed o. We use a a ian o a gi e-o - ake dic a o game o shed ligh on hese aspec s in a con olled labo a o y se ing. Ou expe imen con ibu es o he exis ing li e a u e in wo ways: i s , using a no el app oach o inducing social iden i ica ion wi h one’s pee s in he lab, ou design allows us o analyze he spillo e -e ec s o (un)e hical beha io unde a ied le els o social iden i ica ion. Second, we s udy whe he con- agion o e hical beha io di e s om con agion o une hical beha io . Ou esul s sugges ha a) une hical beha io is mo e con agious, and b) social iden i ica ion wi h one’s pee s and no he (un)e hicali y o obse ed beha io is he main d i e o beha io al con agion. Ou indings a e pa icula ly impo an om a policy pe spec i e bo h in o de o os e p o-social and mi i- ga e de ian beha io . KEYWORDS: Con o mi y, Con agion, Pee E ec s, Social Iden i y, Une hical Beha io JEL: D03; D73; D81 * Uni e si y o Pade bo n and Ha a d Uni e si y. This wo k has g ea ly bene i ed om con e sa ions wi h Edwa d Glaese , Daniel House , Law ence Ka z, Da id Laibson, Ul ich Schmid , and Wendelin Schnedle on he ea ly e sion o he expe imen al design. I am pa icula ly hank ul o aluable inpu om Max Baze man, C is ina Bicchie i, Ga y Bol on, Elena Ka ok, Judd Kessle , and Robe Ku zban du ing my isi ing esea ch posi ions a he Ha a d Uni e si y, he Uni e si y o Pennsyl ania, and he Uni e si y o Texas a Dallas, espec i ely. I also wan o hank René Fah , U i Gneezy, Bu kha d Hehenkamp, Rosema ie Nagel, A no Riedl, and Tim Salmon as well as he pa icipan s a nume ous wo kshops and con e ences o in e es ing discussions and help ul sugges ions. Financial suppo by he Ge man Resea ch Founda ion (DFG) h ough he SFB 901 a he Uni e si y o Pade bo n is g a e ully acknowledged. 80 chap e 3.1.6. We close wi h a concise discussion on po en ial policy ecommenda ions in chap e 3.1.7 and a conclusion and an ou look in chap e 3.1.8. 3.1.2 Backg ound and Cou se o In es iga ion A basic p inciple o classical economic heo y sugges s ha indi iduals o m a ional expec- a ions based on a ailable in o ma ion and ac on hem acco dingly. Howe e , e en he g ea John Mayna d Keynes (1936) exp essed his conce n abou he a ionali y o indi idu- als o ealize e icien in es men decisions in he long un al eady 80 yea s ago. Ins ead, Keynes expec ed indi iduals o ollow he he d, hus s essing he impo ance o pee s o many economic decisions. Since hen, a con as ing s ain o li e a u e eme ged ha ac- coun s o he ele ance o beha io al ai s on he indi idual decision-making p ocess. Un- de s anding he unde lying mechanism o pee e ec s is key o comp ehending i s impac on economic decisions and ou comes. Fo he bigge pa , exis ing esea ch on pee e ec s mainly eso s o ield expe imen s o pu ely obse a ional s udies ha a e gene ally in e- io o con olled lab expe imen s in e ms o , among o he s, a clean iden i ica ion o he ele an channels, endogenei y, and e lec ion p oblems (see Manski (2000), Falk & Fisch- bache (2002)). Only ecen ly, he e has been a push o s udy pee e ec s in he lab, allowing us o gain a deepe and o en a mo e eliable unde s anding o he unde lying mechanism Ang is (2014). Social in e ac ions in gene al and he po en ially esul ing pee e ec s in pa icula play an ins umen al ole om bo h he socie al and economic pe spec i e. Exis ing li e a u e indi- ca es ha s anda d economic o ces alone canno encompass many o he ou comes ha we obse e in eal li e. Examples a e, among o he s, he escala ion o c ime a es o he mas- si e su ge in emale labo pa icipa ion a es in Wo ld Wa II (c . Mulligan (1998), Le i (1999). See also ecen indings on pa e ni y lea e by Dahl, Løken & Mogs ad (2014)). In- s ead, social in e ac ions a e ound o o e explana ions help ul o unde s anding he causes o apid shi s in economic undamen als. Such ipple e ec s a e likely he esul o social in e ac ions, hus aising he awa eness abou he impo ance o unde s anding he unde lying mechanism o pee e ec s (Glaese & Scheinkman, 2004). Al hough explici esea ch on pee e ec s and he esul ing beha io al spillo e s (in he li - e a u e some imes e e ed o as beha io al o social con agion) has i s o igins in he la e 81 19 h cen u y, he unde lying concep has been obse ed long be o e.2 Repo edly, an ab- s use-seeming s eam o suicides happened a e eading Goe he’s The So ows o he Young We he wo hund ed yea s ago. “My iends […] hough ha hey mus ans o m poe y in o eali y, imi a e a no el like his in eal li e and, in any case, shoo hemsel es; and wha occu ed a i s among a ew ook place la e among he gene al public […]” (Goe- he, quo ed in Rose (1929, p. 29)). The widesp ead imi a ion o his beha io ga e ise o ea among he popula ion and go e nmen s, ul ima ely leading o a ban o he book in I aly, Leipzig, and Copenhagen (Phillips, 1974). The ou b eak o he Tanganyika laugh e epi- demic o 1962 in Uganda is ano he in amous example o beha io al con agion. The e, a mass hys e ia in ec ed almos 100 pupils wi h con agious laugh e , o cing se e al schools o close down o days (Rankin & Philip, 1963). Ini ially, he concep o social con agion has been in oduced in he o m o a social phe- nomenon – as opposed o a biological one – explaining why and how ce ain o ms o be- ha io soak h ough socie y ( o ea ly wo k see Baldwin (1894), Ta de (1903)). Since he 1950s, empi ical esea ch on his opic has been on he ise wi h e idence sugges ing ha he me e exposu e o and con ac wi h indi iduals o cul u e is su icien o igge beha - io al con agion. Condi ional on a su icien ly salien igge , beha io al con agion leads o beha io al adap a ion owa ds obse ed beha io . In his pape , we aim a expanding he exis ing knowledge on he d i e s o saliency, in pa icula wi h a ocus on bo h he e hicali y o he obse ed beha io and he deg ee o social dis ance o p oximi y wi h he obse ed indi idual. A long adi ion in social science highligh s he impo ance o social iden i y in unde s anding indi idual beha io wi hin he amewo k o social in e ac ions (c . Boga dus (1928)). A salien s a e o social iden i y is ound o igge a o i ism owa ds hose o s onge social kinship. The e m social iden i y is eclec ic and se e al o i s ace s ha e been s udied in exis ing economic esea ch. While he e m encompasses a b oad ange o concep ual elemen s, 2 Resea ch on beha io al con agion is agmen ed and di e en disciplines ha e in oduced own no ions and de ini ions e e ing o he same o closely ela ed concep . Exis ing esea ch in e changeably uses di e en e ms o desc ibe such si ua ions, among o he s: con o mi y, beha io al con agion, imi a ion, o beha io al adap a ion. Due o i s mo e gene ic na u e and he con ex o ou esea ch, we will mainly eso o he e m beha io al con agion o beha io al adap a ion. Fo he sake o comp ehensibili y, we will abs ain om clea ly de ining and delimi ing hose concep s o now. 82 om sha ed p e e ences and expe iences o sha ed cul u al and eligious belie s, in his pa- pe we ollow he p imal unde s anding o his e m as in oduced by Taj el and Tu ne (1979). Mo e p ecise, we e e o he exis ence o social iden i y i a pe son de i es sel -es- eem om belonging o he pee g oup and has a p e e ence o exhibi ing simila beha - io al pa e ns ( o a simila app oach, see Chen & Li (2009)). Using his de ini ion in combi- na ion wi h a simple implemen a ion o social iden i y, as applied in ou expe imen ia he obse a ion o p e e ence simila i y, is conduci e o bo h de i ing lowe bound esul s o he ole o social iden i y in acili a ing (un)e hical beha io and easie ep oducibili y o ou esul s. This line o esea ch is impo an om a policy pe spec i e in gene a ing e ec i e measu es o igge bo h mo e p o-social and less an i-social beha io , consequen ly educ- ing he o he wise esul ing economic and social ine iciencies. Wi h no able excep ions, exis ing esea ch has been s uggling o o e come a numbe o challenges o s udy clean pee e ec s in he lab, especially in con ex s whe e social iden i y plays a media ing ole. Among hese, inducing o a leas p oxying he na u al occu ing a - ia ion o social iden i y has p o en o be di icul . Following he adi ion o he minimal- g oup pa adigm (Taj el & Tu ne , 1986), psychological esea ch has in oduced a numbe o ways o p oxy social iden i y in he lab, such as ha ing pa icipan s in e ac wi h pa icipan s ha we e assigned he same colo a a a (Taj el, 1982), a simila name (Pelham, e al., 2005), same bi hday (Cialdini & DeNichols, 1989), o imagined closeness (Gunia, e al., 2009). So a , i has p o en o be challenging o s udy pee e ec s unde con olled se ings in gene al, le alone wi hin a mo e sophis ica ed social en i onmen such as a ied le els o social iden- i y. Exis ing economic esea ch has eso ed o using a dual app oach, s udying pee e ec s in he lab and in he ield, wi h bo h app oaches ha ing hei limi a ions. Applying a no el me hodological app oach o induce a ying le els o social iden i y in he lab allows us o combine he bes ea u es o bo h wo lds o s udy pee e ec s beyond wha had been pos- sible so a . 83 To ou knowledge, he i s economic con ibu ion examining his ques ion in a con olled se ing by a ying social iden i y among indi iduals is he s udy by Bohne & F ey (1999).3 They used a ough p oxy o social dis ance by a ying he deg ees o iden i ica ion and ound e idence ha social dis ance is decisi e in p edic ing he ex en o o he ega ding beha io in a dic a o game se ing.4 O he examples o economic app oaches ha b idge anonymi y and induce social iden i y, among o he s, a y he wo ding o he expe imen al ins uc ions (Ho man, e al., 1996), use ace- o- ace in e ac ion (Bohne & F ey, 1999) o show pic u es o one pa y only (Eckel & Pe ie, 2011), e eal names (Cha ness & Gneezy, 2008), e eal p e e ences such as hose o pain ings (Chen & Li, 2009), o ec ui iends and amily membe s (B and s & Solà, 2010). Such s udies ypically yield he obus inding ha s onge social iden i y igge s a o i ism. Two na u al p oblems a ise wi h he concep s used in economics so a . Fi s ly, using ace- o- ace communica ion o allowing pa icipan s o in e ac wi h iends o amily membe s in oduces se ious biases and c owds-ou he e ela ion o ue p e e ences (Ro h, 1995). Secondly, and mo e ele an o he poin o ou expe imen , he deg ee o social iden i ica ion measu ed in he lab by hese concep s can ha dly be a ied and is a he bina y. Ou app oach o using da ing-websi e ques ions as a ma ching de ice allows us o induce and exogenously a y di e en le els o social p oxim- i y in he lab in o de o s udy hei ole in he spillo e o (un)e hical beha io . I is his pape ’s aim o shed ligh on h ee ques ions s a ed abo e and o con ibu e o a be e unde s anding o he gene al mechanism o pee -e ec s. Fo his eason, we p opose a no el app oach o p oxy di e en le els o social p oximi y among pee s in a labo a o y se ing. Such an app oach allows us o exogenously a y social cha ac e is ics and s udy hei ole in beha io al con agion. We mimic social p oximi y by he use o ques ions aken om a majo Ame ican da ing websi e o cap u e indi idual p e e ences and in e es s and use he ma ching sco es o o e lapping answe s among lab pa icipan s as an exogenous 3 Al hough mo i a ed by a p e ious s udy o Ho man, McCabe & Smi h (1996), he wo k o Bohne & F ey (1999) is he i s o di ec ly a y (a p oxy o ) social dis ance among pee s. Ins ead, Ho man, McCabe & Smi h (1996) a ied he language used in he dis ibu ed ins uc ions, a guing ha “subjec s b ing hei ongoing epea ed game expe ience and epu a ions om he wo ld in o he labo a o y, and […] dic a o ins uc ions […] may imply ha he objec i e is o sha e he money wi h someone, who, hough anonymous, is socially ela i ely nea o he decision make ” (Ho man, e al., 1996, p. 655). 4 Resea ch usually e e s o his concep as social iden i y, which encompasses bo h social dis ance and i s in e se, social p oximi y. Th oughou his pape , we will mainly e e o social p oximi y. Ano he app oach o mimic social p oximi y used in economic and psychological expe imen s has been o ask pa icipan s o b ing along hei iends o ela i es and s udy hei in e ac ion in a con olled en i onmen . Among o he hings, we will discuss po en ial d awbacks o hese and ela ed app oaches in chap e 3.1.5 in mo e de ail. 84 ma ching de ice ac oss ea men s. This allows us o s udy decision-making going beyond simple ing oup - ou g oup compa isons. Ra he , ou app oach p o ides us wi h an ex ensi e a ay o possibili ies o ma ch pa icipan s acco ding o hei sha ed simila i ies. To he bes o ou knowledge, we a e he i s o use such an app oach. Thus, we no only complemen exis ing ield s udies, bu also b oaden he scope and u iliza ion o lab expe imen s in ex- plaining beha io and beha io al changes in pee se ings, especially wi hin he une hical domain. Fo his pu pose, we ex end he cu en ly exis ing app oaches by a social componen ha su icien ly conside s he ele ance o social dis ance and p oximi y o one’s pee s in a ec - ing beha io al decisions. This a emp will be a he hea o his pape , leading o a p o- posed heo e ical ex ension o Ake lo ’s (1997) and Glaese & Scheinkman’s (2004) seminal wo k on social in e ac ion, social dis ance, and con o mi y, whe e he ex en o social dis- ance is a unc ion o geog aphic loca ion. We, howe e , emphasize he ole o social dis- ance as a unc ion o an ac ual o e lap in pe sonali y based on e.g. pe sonali y ai s o in e es s. Ano he subs an ial con ibu ion o ou esea ch is he di ec compa ison o beha io al con- agion o e hical and une hical beha io . In gene al, exis ing esea ch has ocused on shed- ding ligh sepa a ely on beha io al spillo e s o ei he e hical beha io (c . Thöni & Gäch e (2015)) o une hical beha io (c . Gino e al. (2009)). Conside ing he di e ing se ings and games used in exis ing expe imen s o s udy beha io al con agion, cu en esea ch does no help o unde s and whe he and o which ex en beha io al con agion in ei he di ec- ion di e s om each o he . Ins ead, expe imen s ha add o he unde s anding o he po- en ially di e en mechanisms should place pa icipan s in a uni o m en i onmen and al- low o he spillo e o e hical and une hical beha io . Ou expe imen al se -up allows us o s udy hese ques ions. Pa icipan s play a one-sho dic a o game in which hey decide how much money o dona e o o ake away om a cha - i y, which esembles a a ia ion o a dic a o game implemen ed by Lis (2007) and Ba dsley (2008)). He e, he (un)e hicali y o ( aking away) dona ing money o he cha i y is s essed by explaining he consequences o hei beha io clea ly o he pa icipan s. Tha is, all he money ha is ( aken away) dona ed o he cha i y will (no ) be gi en o wa d o he cha i y, hus indi idual beha io will (ha m) bene i he cha i y. Hence, pa icipan s ace a iskless 85 bu in e ms o i s (un)e hicali y p ecisely de ined si ua ion in which hey ha e o decide whe he o no o pe sonally bene i a he expense o a cha i y o hei choice. A e each- ing hei ini ial decision, pa icipan s a e gi en he oppo uni y o lea n abou o he pa ici- pan s’ ini ial decisions ollowed by he op ion o e ise hei own ini ial decision. Se e al economic and psychological heo ies a e able o explain beha io al con agion e en unde ull anonymi y and wi hou obse abili y o one’s own ini ial and po en ial e ision decision, as implemen ed in ou expe imen . Among hese a e concep s ela ing o social decisions and social dis ance (Ake lo (1997), Glaese & Scheinkman (2004)), imi a ion o p e - e ences (Sliwka, 2007), social lea ning (Bandu a, 1971), no ms (Cialdini e al. (1990) and Bic- chie i (2006)), sel -expansion (A on & A on, 1986) o e en guil (Kandel & Lazea , 1992). Many o hese concep s a e no s ic ly dis inc , in bo h hei assump ions and p edic ions. Thus, in his pape we will no a emp o esol e which app oach explains beha io al con agion bes bu a he ocus on shedding ligh on he d i e s o beha io al con agion and i s in e - ela ion wi h he social iden i y dimension and he ex en o (un)e hicali y o obse ed be- ha io . 3.1.3 The Concep ual F amewo k o Beha io al Adap a ion 3.1.3.1 The Mechanism o Social In e ac ion and Challenges o Measu - ing i s E ec s A g owing body o li e a u e sugges s ha social in e ac ions a e p incipal no only o hu- mans, e.g. in making social o economic decisions, bu also o animals, e.g. in inding he igh s a egy o place o maximize one’s hun ing success (Laland, 2002). E iden ly, social in e ac ions igge di e en eac ions and ou comes, which a e e e ed o as social e ec s. Di e en s eams o li e a u e e e o such in e ac ions in di e en ways: bandwagons, con- o mi y, epidemics, he d beha io , imi a ion, neighbo hood e ec s, pee in luences, social lea ning, o social no ms (see Hyman (1942), Me on (1957), G ano e e (1979), Jones (1984), Manski (2000)). In addi ion, Pingle & Day (1996) subsume hese and o he ypes o 86 beha io , including ollowing an au ho i y, habi , hough less impulse, and hunch, as econ- omizing beha io . Some call i simply pee e ec s.5 Al hough hese no ions e e o di e en mechanisms, hei ou come is o en (bu no always) simila ; ha is, he indi idual’s adap- a ion o obse ed beha io . Howe e , some mechanisms encompass s onge in e ac ion wi h he pee s han o he s do (e.g. he pee ’s abili y o obse e my eac ion and exac change in beha io ) and in ol e mo e o less delibe a ion.6 As we will discuss in his chap e in mo e de ail, exis ing esea ch indica es ha he exis - ence and pe sis ence o social e ec s a e con ex speci ic and may e en lead o, among o h- e s, highe consump ion o alcohol and d ugs, chea ing, and smoking. The same is also ound o be ue o p o-social and coope a i e beha io . Thus, be o e u ning o he speci ic e ec s o social in e ac ion, we shall make he e o o unde s and he unde lying mecha- nisms i s . The concep o social in e ac ion lies a he hea o social psychology and sociology, o example in o de o explain he o ma ion o as es (Webe , 1978). The impo ance o social in e ac ion in explaining social phenomena has a long adi ion and is o en asc ibed o Su he land’s Di e en ial Associa ion Theo y (Su he land, 1939). Ake lo (1997) s esses he iew ha he heo y o social in e ac ion is key o unde s anding why indi iduals do no succumb o isola ed and pu ely sel -maximizing decision-making. Ins ead, his concep gi es ise o concei e an indi idual as someone who cons an ly in e ela es wi h he unde - lying social en i onmen and p oduces and deals wi h he esul ing ex e nali ies. A p inciple consequence o ex ending he a ional model based on Becke ’s ea ly wo k by such a social dimension is ha he ype o he esul ing indi idual is mo e sophis ica ed and esembles mo e closely o he in ui ion o sociologis s han he classical economis s. 5 O en, he e m “pee e ec s” is used o e e o all o hese mechanisms wi hou speci ying he exac channel h ough which beha io al adap a ion a ises. Al hough i is impo an o shed ligh on he di e en channels h ough which social in e ac ion po en ially ansi ions in o beha io al adap a ion, he ocus o his s udy lies in unde s anding he ole o exogenous ac o s such as social p oximi y ha ha e he po en ial o in luence he in ensi y o beha io al adap a ion. We shall no a emp o se le he a gumen o which app oach explains he mechanism o ou expe imen bes . Ins ead, we will discuss some o he mo e p ominen concep s and mechanisms in he ields o economics and social psychology ha esemble he mechanism o beha io al adap a ion he way i is implemen ed in ou expe imen in chap e 3.1.4. Fo ou pu pose, howe e , we will assume pee e ec s o be in play whene e an indi idual ′ beha io changes a e ha ing been exposed o beha io o indi idual , i espec i e o he di ec ion o he beha io al change. In u n, howe e , we assume beha io al con agion o be in place whene e indi idual ′ beha io changes in he di ec ion o indi idual  beha io . 6 I is wo h no ing ha i is a beyond his pape ’s scope o shed ligh on all o hese concep s. Ins ead, we will pick ou and discuss hose concep s o which we belie e a e o bigge impo ance o wha we analyze wi hin ou expe imen . 87 Manski (1993) dis inguishes be ween wo ypes o social in e ac ions.7 One is endogenous in e ac ion in he o m o , o example, in o ma ion exchange among c iminals o social no ms. The beha io o he ele an pee g oup media es he likelihood ha he indi idual will engage in he same kind o beha io . In addi ion, exogenous in e ac ion emphasizes ha he p opensi y o an indi idual o beha e in a ce ain way is also media ed by exoge- nous cha ac e is ics o he g oup such as hei a i ude owa d c ime o social and economic s a us.8 In social in e ac ions, ex e nali ies abound. The key mechanism o social in e ac ion implies ha one’s pe sonal ne bene i is a unc ion o he beha io exhibi ed by one’s ele an so- cial g oup o con ac pe son (Glaese & Scheinkman, 2004).9 Inhe en o such in e ac ions a e s a egic complemen a i ies in he o m o ci cula cascades whe e “e en i changes in undamen als c ea e only a small change in he le el o ac i i y o each indi idual, each indi idual’s small change will hen aise he bene i s o e e yone else pu suing he ac i i y” (Glaese & Scheinkman, 2004, p. 84). In p inciple, small changes in undamen als may cause la ge shi s in ou comes, which a e some imes e e ed o as he bu e ly e ec , a e m hail- ing om he chaos heo y (Lo enz, 1963). Social in e ac ions a e also highly ele an in unde s anding he sp ead o c iminal beha - io . Social in e ac ion plays a decisi e ole in he o ma ion o gangs and he ec ui men o young c iminals (see Reiss (1988) and Jankowski (1991)). This is pa icula ly ue o he c iminals’ decision o engage in illici beha io join ly (Reiss (1980)). “Social in e ac ions seem o c ea e a sense o in ulne abili y and a willingness o iola e social no ms and ake 7 Glaese & Scheinkman (2004) p o ide a mo e dis inc ca ego iza ion o he mechanisms ha gene a e social in e ac ion: physical, lea ning, s igma, and as e- ela ed in e ac ions. Fo he pu pose o his expe imen , we will ex end his ca ego iza ion in o de o be e cap u e he mechanism o beha io al changes we a e in e es ed in. We will discuss hese poin s in mo e de ail in chap e 3.1.6. 8 While hese wo aspec s ep esen in e ac ions ha a e shaped by he unde lying social en i onmen , Manski (1993) also in oduces co ela ed e ec s ha explain simila beha io as he esul o acing simila ins i u ional en i onmen s (see also Manski (2000)). This hi d aspec is no conside ed any u he since co ela ed e ec s a e no social e ec s and a e hus nei he c ea ed by social in e ac ions no c ea e social mul iplie s (see Glaese , Sace do e & Scheinkman (1996), (2003)). Using a no el app oach in ou expe imen , a clean a ia ion o bo h endogenous and exogenous in e ac ion allows us o d aw causal in e ences ha a e mo e p ecise han wha has p e iously been possible. We will e u n o his impo an poin in ou design sec ion o chap e 3.1.6. 9 The e m ‘ ele an social g oup o con ac pe son’ is ague in exis ing esea ch. A social g oup o con ac pe son is sel - epo ed and de ined om one’s indi idual poin o iew and e e s o one o mo e indi iduals whose beha io ei he has a di ec o indi ec impac on one’s well-being, e.g. h ough esou ce ex e nali ies o o he - ega ding p e e ences. While i is impo an o unde s and he di e en channels h ough which beha io al spillo e s occu , he ocus o ou s udy is o shed ligh on he d i e s ins ead. See chap e 3.1.4 o a discussion. 88 isks, as long as one is in he company o like-minded indi iduals” (Glaese , e al., 1996, p. 511). Social in e ac ion also s ongly a ec s s igma iza ion, which is impo an om an e o- lu iona y pe spec i e. As he numbe o c iminals ise, illici beha io becomes mo e com- mon and hus po en ially mo e accep ed; and as c iminali y con e ges owa ds ‘no mali y’, he (social) en s o illici beha io inc ease due o inc eased a ac i eness, social accep a- bili y and a c owding-ou o legal ac i i ies whe e ea nings om legal ac i i ies a e s olen by c iminals (see Rasmussen (1996) and Mu phy, Shlei e & Vishny (1993)). Resea che s ace subs an ial di icul ies measu ing social in e ac ions and i s esul ing e - ec s in a clean way. In a eal-wo ld ins ance, assigning changes in indi idual con o mi y o dis inc mechanisms a e subjec o iden i ica ion p oblems (Ang is , 2014). As a gued be- o e, his p oblem a ises because indi idual beha io is a ec ed by bo h endogenous (e.g. he g oup’s beha io ) and exogenous e ec s (e.g. g oup cha ac e is ics) and, in addi ion, uni o m beha io can be he esul o simila unobse ed cha ac e is ics (Manski, 1993). P e- ious empi ical esea ch in ol ed eg essing a pe son’s ac ions on he ac ion o his pee s. Howe e , Manski (1993) poin s a h ee undamen al p oblems conce ning his me hodol- ogy: i s , d awing causal in e ence is di icul when endogenei y is a p oblem, in pa icula when he indi idual’s and i s pee ’s beha io is in e ac i e and in luences each o he ci cu- la ly. Second, omi ed a iables inc ease he likelihood o spu ious co ela ions be ween ac ions. Thi d, in eali y, so ing and sel -selec ion in o pa icula neighbo hoods ende s i di icul o unde s and wha ac ually d i es beha io . A guably, empi ical esea ch in pa ic- ula aces hese challenges because one only obse es he beha io o indi iduals who sel - selec ed hemsel es in o, o example, mo ing o a be e neighbo hood, bu no o hose who decided o u n down he oppo uni y (Glaese & Scheinkman, 2004).10 O pa icula in e es o ou esea ch a e social in e ac ions leading o he sp ead o une h- ical beha io . Exis ing esea ch poin s o a s ong p esence o posi i e co a ia es ac oss in- di iduals’ decision o engage in c iminal beha io . In pa icula , Glaese , Sace do e & 10 P ominen examples and o ce ul ways o add essing hese issues include, among o he s, he wo k o Case & Ka z (1991), Ka z, Kling & Liebman (2001), Ang is & Lang (2004), Kling, Ludwig & Ka z (2005), Kling, Liebman & Ka z (2007), Ludwig & Kling (2007), Damm & Dus mann (2014), and Che y, Hend en & Ka z (2015). We will discuss hese and o he empi ical con ibu ions examining pee e ec s and beha io al adap a ion in he ield in mo e de ail in his chap e . 89 Scheinkman (1996) s a e ha only co a iance ac oss c iminal decisions o indi iduals ex- plains exis ing a iance in c ime a es, which is a beyond any heo e ical p edic ion o c ime a es. In o de o mi iga e any e m- ela ed con usion on he side o he in e es ed eade , we a e in need o a e m ha allows us o cap u e a pa icula ype o beha io ha esembles a subse o social in e ac ion and i s esul ing social e ec s. Mo e speci ically, we a e no in- e es ed in he d i e s o all kinds o beha io al changes esul ing om obse a ion, bu only in beha io al changes ha lead o a con e gence o beha io . We a e hus p oposing he impa ial e m beha io al adap a ion o cap u e such beha io .11 The e m beha io al adap a ion e e s only o a subse o social in e ac ion because social in e ac ion may lead o all kinds o social e ec s whe e he esul ing beha io may o may no con e ge owa ds wha has been obse ed. In i s basic o m, he exis ence o social e - ec s does no ell us much abou he speci ic beha io al eac ion, i any, ha ollows om his exposu e. In p inciple, he esul o social in e ac ion could lead o ei he beha io al aliena ion o adap a ion. Fo he pu pose o his s udy, we will ocus on he la e . In u n, beha io al adap a ion e e s o hose si ua ions only in which he esul ing beha io is co- he en o he beha io one has been exposed o. Ou s udy sheds ligh on he key d i e s igge ing an indi idual’s esponse o become mo e like o he s, in one way o ano he .12 Exis ing mac o- and mic o-le el da a inadequa ely ca ches he unde lying mechanisms and he causal ela ionships ele an o ou p ojec . Tha is, answe ing he ques ions o how be- ha io al adap a ion a ies wi h di e en le els o social p oximi y and whe he his in e - play is di e en o adap a ion owa ds e hical e sus une hical beha io . As has been a - gued by Ang is (2014), wi h a e excep ions like Kling e al.’s (2007) Mo ing- o-Oppo uni y esea ch, mos ield s udies on pee -e ec s su e in one way o he o he om endogenei y, 11 Beha io al adap a ion has also been s udied in he ield o e olu iona y game heo y as well as in he heo y o lea ning in games (see Sel en (1978), Ro h & E e (1995), Schlag (1998), and Apes eguia, Huck & Oechssle (2007), o only name a ew). Howe e , i is beyond he scope o his pape o ouch upon all s eams o li e a u e ha ha e shed ligh on he gene al p ocess o beha io al adap a ion. Ins ead, we ocus on he s udies mos ele an o ou expe imen and ex end ou apologies o colleagues whose esea ch emains unnamed in his pape . 12 S udying he d i e s o beha io al aliena ion is a po en ial enue o u u e esea ch, as his line o esea ch has ye o ca ch a en ion om economis s. Ake lo (1997) e e s o his beha io as he esul o s a us-seeking e o s. Beyond his, howe e , beha io al aliena ion can be he esul o e.g. one’s desi e o no be iden i ied wi h a pa icula social g oup. 96 hei indings indica e ha social in luence is ansmi ed h ough bo h channels and poin in he same di ec ion. Some s udies, howe e , ind li le e idence o neighbo hood o pee e ec s. In a highly e- ga ded s udy, E ans, Oa es & Schwab (1992) show ha a e con olling o selec ion bias, any measu able pee e ec on eenage p egnancy and school d opou a es disappea . While being ca e ul in no claiming ha no pee e ec s exis a all, hey a he poin c i ically o me hodological issues measu ing pee e ec s in a clean way. Ang is & Lang (2004) use da a om he Me opoli an Council o Oppo uni y (Me co) deseg ega ion p og am in which mos ly black s uden s a e sen o mo e a luen subu bs. Thei indings indica e ha he e a e li le, i a all, posi i e spillo e s on s uden s. Simila ly, Bu ke & Sass (2013) ind li le e idence o class oom pee e ec s on s uden achie emen o Flo ida public school s uden s. Likewise, Ludwig & Kling (2007) ind li le e idence o he con agion o c ime hy- po hesis using MTO da a. Ins ead, hei indings indica e ha c ime a es a e me ely d i en by neighbo hood acial seg ega ion. In conclusion, he exis ence o pee e ec s is up o schola ly deba e, which is mainly d i en by me hodological challenges and da a p oblems. Howe e , in ollowing Ang is (2014), he p e iously discussed MTO p og am yields he mos p omising se ing o s udy clean pee e ec s in he ield. P e ious esea ch ha u ilized MTO da a yields, among o he hings, s ong gende asymme ies in e ms o he e olu ion o c iminal beha io . The au ho s sug- ges ha hese indings can be a ibu ed o di e ences in which males and emales espond o hei en i onmen and i s in luences. Ul ima ely, his leads o di e ences in magni ude and speed a which (illici ) beha io is picked up. E idence om he Lab In wha ollows, we shall no a emp o p o ide an exhaus ing o e iew o he comp ehen- si e li e a u e dealing wi h pee e ec s in gene al. Ins ead, we will ocus on a ange o in lu- en ial s udies using lab expe imen s o shed ligh on mechanisms ela ing o beha io al spillo e s ha a e mo e in line wi h ou pape ’s ocus. La e in he pape , we will e e o hese s udies in mo e de ail whe e necessa y. O e he las decade, a comp ehensi e s eam o li e a u e s udying spillo e -e ec s and beha io al adap a ion in he lab has eme ged ha complemen he ongoing impo an wo k 97 in he ield. As discussed p e iously, he me hodological shi was s ongly d i en by chal- lenges ela ing o iden i ying hese e ec s in a clean way using obse a ional da a. Se e al esea che s claim ha al hough he mos ecen gene a ion o s udies measu ing such e - ec s wi h obse a ional da a has succeeded o make impo an s eps owa ds acking he challenges ou lined be o e, con olled lab expe imen s a e s ill he gold s anda d in educ- ing noise and po en ial con ounds (Ang is , 2014). “Howe e , e en i he se ing o e s an almos pe ec oppo uni y o iden i y pee e ec s in many o hese s udies, he impossibil- i y o con olling o all local o pe sonal con ounding ac o s and o endogenous so ing makes he iden i ica ion s a egy no ully con incing" (Falk & Ichino, 2006, p. 40). Ea ly labo a o y esea ch s udying pee e ec s and social iden i ica ion join ly has been pi- onee ed by Ho man, McCabe & Smi h (1996) and Bohne & F ey (1999). These s udies made use o a ia ion in he ins uc ion’s wo ding o enhanced ace- o- ace communica ion o s udy he ole o social iden i ica ion in gi ing decision, equi ocally inding suppo o i s ele ance (see also Cha ness & Gneezy (2008)). We will e u n o hese s udies in mo e de ail in chap e 3.1.5. O he s udies ha e looked in o pee e ec s in p oduc i i y decisions. In a highly ega ded s udy, Falk & Ichino (2006) ound obus e idence o he exis ence o pee e ec s in a p oduc i i y ask. Thei esul s indica e ha low-p oduc i i y wo ke s a e pa icula ly sus- cep ible o pee e ec s, which esul s in an o e -p opo ional aise in p oduc i i y. Follow- ing he wo k o Mas & Mo e i (2009), subsequen s udies ied o disen angle he na u ally occu ing channels o simul aneously obse ing pee s and being obse ed by pee s. Fo he mos pa , hese s udies ound he la e channel o be mo e e ec i e han he o me in boos ing p oduc i i y (c . Geo ganas, Tonin & Vlassopolous (2013); o excep ions see Veld- huizen, Oos e beek & Sonnemans (2014)). Along he lines o s udying beha io in he wo kplace, Gäch e e al. (2012) se up an expe - imen ha in es iga es ecip ocal beha io unde obse abili y o o he people’s ac ions. They ind ha he indi idual’s ex en o comply wi h no ms o ecip oci y is signi ican ly d i en by bo h pay and e o compa ison in o ma ion. Za a (2011) expe imen ally exam- ines cha i able gi ing in a social con ex . He inds ha by sys ema ically e ealing in o - ma ion, bo h he lea ning abou desc ip i e no ms ( h ough obse ing wha o he s do) and he image- ela ed conce ns ( h ough e ealing own beha io o he e e ence g oup) d i e 98 indi idual con ibu ion le els. In a mo e delinquen con ex , Falk & Fischbache (2002) in- es iga e pee e ec s in he o m o condi ional s ealing beha io . In pa icula , hey in es- iga e whe he an indi idual’s inclina ion o s eal is dependen on o he pee ’s s ealing be- ha io . Thei main indings sugges ha , on he agg ega e le el, people make s ealing deci- sions condi ional on he beha io o hei pee s. In economics, a limi ed numbe expe imen al esea ch has also poin ed a he con agion o bo h sel ish beha io and dishones y. Bicchie iy and Xiao (2009) s udy a dic a o game wi h a ying in o ma ion on o he pa icipan ’s sel ish o ai beha io , inding ha ai ness in ac ions is con agious. Mo e o he poin o ou esea ch, Innes & Mi a (2013) use a a ian o Gneezy’s (2005) decep ion game o s udy whe he dishones y b eeds dishones y. Thei ind- ings sugges ha he belie s abou o he ’s dishones y is indeed con agious, po en ially d i en by he wiggle- oom c ea ed by such social cues and hus ep esen ing a jus i ica ion de ice o one’s pe sonal dishones beha io . 3.1.4 D i e s o Beha io al Con agion: An In e disciplina y Pe spec- i e P io o del ing in o he subjec in mo e de ail, some e o will be made o disen angle many o hose exis ing concep s explaining why and unde which ci cums ances people demon- s a e a change in beha io as a unc ion o hei pee ’s beha io . In bo h economics and psychology, se e al concep s ha e been de eloped o e he pas ew decades e e ing o simila and o en he same easoning o change one’s own beha io .17 The complexi y o he sel and he dependency o one’s own beha io on wha one obse es o pee s has ecei ed a lo o schola ly a en ion (c . Baumeis e (1987), Kahneman & T e sky (2009)). I is hus impo an o shed ligh on he unde lying concep s d i ing beha io al adap a ion. In his con ex , we will di e en ia e be ween concep s ha o igina ed in he (I) economic li e a u e and in he (II) li e a u e o (social) psychology. The economic concep s include (I.1) social decisions and social dis ance (Ake lo (1997), Glaese & Scheinkman (2004)), (I.2) image e- la ed conce ns (Be nheim, 1994), (I.3) as e o con o mi y (Be nheim, 1994), and (I.4) imi a- ion (Alós-Fe e & Schlag (2009), Sliwka (2007)). In addi ion, we also discuss some (social) 17 See discussion in chap e 3.1.3. 99 psychological concep s including (II.1) social lea ning (Bandu a (1971)), (II.2) no ms (Cialdini e al. (1990), Bicchie i (2006)), and (II.3) psychological closeness and ica ious dishones y (A on & A on (1986), Golds ein & Cialdini (2007)). One poin is wo h cla i ying. In his pape , we shall no a emp o se le he a gumen which schola ly app oach explains beha io bes . Ra he , o ou pu pose, we will use he p e iously de ined uni ying e ms beha io al adap a ion o beha io al con agion h ough- ou he pape in o de o e e o one pe son’s decision o change ini ial beha io as a unc- ion o obse ed beha io om a leas one o he pe son. Using his umb ella e m will allow us o ocus on wha is ele an wi hou ge ing los in concep ual deba es. We deem i im- po an o help unde s and he ole social p oximi y plays in he change o beha io and o wha ex en p oximi y media es he spillo e o e hical and une hical beha io . I will be he nex sub-chap e s’ aim o dissec he di e en app oaches in mo e de ail and o explain he easons why his is he case. A imes, he p edic ions and he empi ical ou - comes a e iden ical, bu he unde lying o ces causing such ou comes a e di e se. In wha ollows, we will discuss he di e en heo ies ou lined abo e and hei implica ions wi h e- spec o changes in beha io . A he end o his chap e , we will ela e hese heo ies o ou expe imen and p o ide a sys ema ic b eakdown wi h espec o he i o each pa icula heo y o explain beha io in ou design.18 3.1.4.1 Concep s in Economics Social Decisions and Social Dis ance Seminal con ibu ions by Ake lo (1997) and Glaese & Scheinkman (2004) we e among he i s o p o ide an economic amewo k highligh ing he ele ance o social dis ance in a - ec ing social in e ac ion and beha io al adap a ion. Beyond a ionalizing one’s own beha - io in an isola ed en i onmen , esea ch indica es ha beha io is a unc ion o bo h pu e own-maximizing and o he - ega ding conce ns. 18 I is wo h no ing, howe e , ha ou discussion he e will be pu ely desc ip i ely. We will con as he explana o y powe o hese concep s wi hin he ame o ou expe imen in chap e 3.1.6. 100 Fo a while, adi ional economics has neglec ed such in e dependence and a he pu em- phasis on indi idualism. In eali y, howe e , decisions a e a ely b ough abou in o al iso- la ion, bu a e a he he esul o an in e play wi h one’s (social) en i onmen . A guably, indi iduals ca e abou bo h s a us conce ns (in absolu e and ela i e compa isons) and o he ’s well-being (Feh & Schmid (1999), Bol on & Ocken els (2000), Cha ness & Rabin (2002)). Because social in e ac ions ypically ende ex e nali ies wi h he po en ial o slow- ing down con e sion owa ds socially (un-)desi able equilib ia, i is impo an o unde - s and he unde lying mechanism. In his a emp o explain he connec ion be ween social in e ac ion and beha io al adap a- ion (con o mi y, in pa icula ), Ake lo (1997) made use o he New onian heo y o g a i y. Ake lo ’s app oach cen e s on explaining con o mis beha io as a unc ion o dis ance in he social space. Such di e ence in one’s social space is cha ac e ized by he di e ence in beha io be ween onesel and a pe son o g oup o ele ance. Fo his pu pose, Ake lo ap- plies he concep s o a g a i y model allowing him o a gue ha con o mi y leads o bene i s (e.g. highe indi idual u ili y o gains om ade) ha a e nega i ely co ela ed wi h dis- ance in social space.19 Ake lo pu s a educed o m concep o wa d in which one’s u ili y declines as he indi idual’s beha io de ia es om he beha io o o he s. Assuming he ex- is ence o ep esen a i e agen s, his model p edic ions yield a se o equilib ia in which he ul ima e beha io o all indi iduals is he same, hus clea ly cha ac e izing he beha io o e e y pa y.20 19 Because geog aphic loca ion is one de e minan o social in e ac ion, his app oach is a gene aliza ion o wha sociologis s ha e coined “social geog aphy”, which has been inspi ed by K ugman’s (1991) wo k on economic geog aphy. 20 Al hough ex-pos beha io is uni o m ac oss indi iduals in equilib ium, such an app oach s ill models con o mism because o he ex-an e desi e o be and beha e like o he s. Ake lo also p oposes an ex ension wi h indi idual he e ogenei y ha allows o he o ma ion o social sub-g oups wi h own no ms and alues. The ex ension models a mechanism in which ( andomly dis ibu ed) pas social loca ion o each indi idual is inhe i ed. In combina ion wi h s a ic expec a ions abou “ he posi ions o be occupied by o he s in social space […] such a model can po ay s able g oups in low le el equilib ium aps because indi idual’s incen i es o choose x o con o m wi h hose whose inhe i ed social loca ions a e close may o e whelm hei incen i es o choose x o in insic easons” (Ake lo , 1997, p. 1010). In ou expe imen , we will s udy his mechanism by holding he o he ’s beha io ixed and hus allowing o delibe a e beha io al adap a ion ee o uns able, highe -o de , o e en nonexis en belie s abou he o he pe son’s nex mo e. We will e u n o his a gumen in ou design sec ion. 101 Ul ima ely, in o de o maximize one’s in insic u ili y, he indi idual will con e ge owa ds he obse ed beha io and hus educes he exis ing social dis ance.21 Wi h educed dis- ance, social in e ac ions become mo e a o able h ough acili a ion o e.g. mu ually ben- e icial ade. Image Rela ed Conce ns In unde s anding indi idual beha io , one undamen al ques ion a ises: “How much o wha we do is he esul o ou gene ic code (na u e); and how much o i is a unc ion o ou en i- onmen , including he ac ions o hose a ound us and ou own pas ac ions (nu u e)?” (Cab al, 2005, p. 15). The h ea o epu a ion loss and being punished in he case o w ong- doing is an in eg al componen o indi idual decision making in gene al. By weighing cos s agains bene i s, he impac o possible epu a ion loss migh de e indi iduals o engage in illici beha io o any kind. Indi iduals a e s i ing o social inclusion and ecogni ion. They wan o belong o one o mo e social g oups and engage in social in e ac ion. S i ing o social iden i y is an inhe en cha ac e is ic, hus decisi ely d i ing he indi idual’s yea ning o main aining an adequa e epu a ion. In his sense, losing ace as he esul o de ian beha io may se e as an a oid- ance, as his migh esul in exclusion om he g oup (c . Be nheim (1994)). Along hese lines, Ake lo (1980) a gues ha he de ia ion om social s anda ds migh lead o loss o social epu a ion and consequen ly o os acism. Howe e , i is wo h no ing ha de ian beha io no necessa ily implies bad beha io pe se, bu has o be unde s ood as a de ia- ion om he unde lying social no m in ei he di ec ion. Repu a ion can also wo k as a means o p ese e a cohe en sel -image ha allows keeping an in e nal consis ency (c . Baumeis e (1998)). The concep o he sel has mul iple ace s and impac beha io in di - e en ways, as le el o sel - ega d and sel -es eem, he ex en and con en o sel -iden i y as well as he s uc u e o he sel -concep exhibi mo i a ional implica ions (c . Wells 21 A poin no made clea in his app oach is whe he such desi e o con o m is also dependen on he signaling alue o one’s beha io al adap a ion. The model a gues ha he desi e o con o m and educe social dis ance is en i ely d i en by in insic conce ns. Howe e , he inabili y o signal con o mi y o one’s pee s uns coun e o he ini ial concep o con o mism (in oduced by, o example, Ash (1958) and Be nheim (1994). Thus, his concep a he esembles he ideas o imi a ion o p e e ences (Sliwka, 2007) o sel -expansion (A on & A on, 1986) ha will be add essed in he nex sub- chap e s. 102 (1978)). By ha , people y o a oid a nega i e upda e o hei sel -image h ough hei ac- ions, which o he wise would esul in an in e nal con lic . Such a mechanism possesses he powe o p e en indi iduals om engaging in delinquen beha io in he i s place. This mechanism is mo e dis inc when he g oup is salien , consequen ly gi ing ise o mo e ex- ensi e alignmen wi h he g oup’s beha io . Wha is mo e, own ac ions could also be sub- jec o social signaling, which highligh s he pe cep ion o onesel by o he s and hus po en- ially igge ing con o m beha io (c . G ossman (2010)). I is impo an o clea ly dis inguish be ween hese wo mo i es and mechanisms when analyzing one’s indi idual d i e s o con o m beha io .22 Howe e , o image ela ed conce ns o be e ec i e, a se ing o epea ed in e ac ions has o be in place. In a one-sho se ing, he h ea o a epu a ion loss is negligible as he playe s won’ ace each o he o a second ime. I ecu ing in e ac ion is no aking place, any de- ian beha io can nei he be aced back o he indi idual no will be colo ed nega i ely wi h ega d o u u e collabo a ion. Exis ing esea ch suppo s his pe spec i e. Fo exam- ple, he e is comp ehensi e expe imen al li e a u e poin ing o he ac ha con ibu ions and compliance wi h unde lying no ms ise in mul iple-sho in e ac ions. In pa icula , his is ue when anspa ency (e.g. abili y o obse e o he pa icipan s’ con ibu ion decisions) o punishmen (e.g. o non-compliance wi h social no ms) is possible (c . And eoni (1995), Feh & Gäch e (2000), Came on e al. (2009), Chaudhu i (2011)). Consequen ly, in o de o epu a ion o be e ec i e (e en in absence o a punishmen -mechanism), a se ing o e- pea ed in e ac ion has o be in place and ac ions ha e o be obse able o o he people. Tas e o Con o mi y Resea ch and obse a ions o eal li e si ua ions indica e ha unde lying social ac o s de- cisi ely in luence indi idual beha io . Mo i a ion o a pa icula beha io is d i en by he inhe en desi e o be alued and o win p es ige, es eem, popula i y, and accep ance (Ellingsen & Johannesson, 2008). A guably, bo h he in o e si e cohe ence and ou wa d app ecia ion ma e s and can be achie ed h ough con o mi y in beha io . The basic idea he e is ha any in ini esimal de ia ion om e ec i e no ms migh be punished by he social 22 These aspec s will be aken in o conside a ion in he expe imen al design. We will e u n o his poin la e . 103 g oup (Be nheim (1994); see also Ake lo (1980) and Ti ole (1996) o ela ed wo k on he ole o epu a ion in indi idual decision making and beha io al alignmen ). In his seminal con ibu ion, Be nheim (1994) o malized he concep o as e con o mi y. He a gues ha he indi idual’s p e e ence o s a us is in line wi h psychological, e olu iona y, and beha io al conside a ions. In pa icula , na u al selec ion a o s conce ns o s a us as his goes hand in hand wi h g ea e oppo uni ies o ep oduc ion. F om a beha io al poin o iew, such conce ns ac as a ein o cing de ice o o m p e e ences o highe es eem be- cause es eemed indi iduals a e mo e likely o ecei e be e ea men . Al hough pa ly ep esen ing a depa u e om he adi ional o mula ion o p e e ences, his app oach does no necessa ily equi e he abandonmen o consis en , sel -in e es ed op imiza ion p ocesses. By assuming ha s a us depends on “public pe cep ions abou an indi idual’s p e e ences o e ac ions […] es eem is de e mined by expec a ions abou u- u e ac ions and ha as es and p ocli i ies a e he bes p edic o s o u u e ac ions” (Be nheim, 1994, p. 843). Unsu p isingly, esea ch indica es ha people sha e simila p e e ences and opinions wi hin he same g oup, which can a guably be bo h he an eceden and he e ec o social in e ac ion and no m alignmen .23 Aligning wi h exis ing no ms is a way o signal one’s (wish ul) belonging o a ce ain social g oup. As a gued by Hogg & Tindale (2002), enac ing in-g oup-p o o ypical beha io is a way o alida e one’s own g oup membe ship, no only o he social g oup bu also o hemsel es. By ha , aligned beha io migh unc ion as a signal o bo h he ex e nal (e.g. he pee s) and he in e nal wo ld (e.g. hemsel es) in o de o solidi y he sense o belonging. I he indi idual ca es abou s a us, one has o se he igh signals o he pee s in o de o c ea e es eem. This es eem will be p o ided in case he indi- idual beha es in a way ha is expec ed om him. In his ein, Be nheim (1994, p. 844) de i es he ollowing p oposi ion, highligh ing he de- cision p ocess leading o beha io al adap a ion: “When popula i y is su icien ly impo an ela i e o in insic u ili y (de ined as he u ili y de i ed di ec ly om consump ion), many indi iduals con o m o a single, 23 Manski (1993) e e s o his as he p e iously men ioned e lec ion p oblem, ende ing i ex emely di icul o s udy clean pee -e ec s in he ield. We will e u n o his poin sho ly in he expe imen ’s design sec ion o chap e 3.1.6. 104 homogeneous s anda d o beha io , despi e he e ogeneous unde lying p e e - ences. They a e willing o supp ess hei indi iduali y and con o m o he social no m because hey ecognize ha e en small depa u es om he no m will se i- ously impai hei popula i y.” A guably, he ex en a which people a e willing o align wi h hei pee ’s beha io is d i en by he deg ee o social iden i ica ion o he social g oup. Social iden i y heo y p o ides help ul guidance o dis inguish be ween di e en magni udes o pe sonal iden i ica ion wi h he pee ’s beha io (c . Taj el & Tu ne (1979), Taj el (1982)). Acco ding o his ap- p oach, people ca ego ize hemsel es and o he pe sons in o di e en social g oups. Mem- be ship in a g oup is de ined as he social iden i y and indi iduals s i e o enhance hei posi ion and sel -es eem h ough ac ions. These ac ions encompass he alignmen wi h ex- is ing g oup no ms. “[…] he social iden i y analysis o ca ego iza ion p ocesses sugges s ha g oup cohesion o solida i y is no only a ac ion among g oup membe s, bu also a - i udinal and beha io al consensus, e hnocen ism, in-g oup a o i ism and in e g oup di - e en ia ion, and so o h – he en i e ange o e ec s o ca ego iza ion-based depe sonali- za ion” (Hogg & Tindale, 2002, p. 65). A comp ehensi e s eam o li e a u e sugges s ha such iden i ica ion is d i en by in-g oup-ou -g oup conce ns, inding ha in-g oup a o i - ism is ound ac oss di e en con ex s and social se ings (c . Taj el (1982), Ho man, McCabe & Smi h (1996), Bohne & F ey (1999), Eckel & Pe ie (2011), Cha ness & Gneezy (2008)). F om his one can de i e he assump ion ha he s onge he desi e o be pa o he in-g oup, he s onge one’s in insic willingness o comply wi h p e alen ules and engage in con- o mi y by adap ing social no ms. Such a desi e is d i en by, o example, he inhe en ele- ance o he social image componen (c . And eoni & Be nheim (2009)). The s onge he social iden i y wi h a e e ence g oup, he s onge he e ec o social compa ison on indi- idual con o mi y. O e all, as e o con o mi y is a unc ion o exis ing social iden i y and he ex en o social compa ison o one’s e e ence g oup. De ia ing om he g oup’s beha - io (e en in ex eme cases whe e he g oup beha io is clea ly w ong, c . Asch (1958)) and hus being pe cei ed as di e en migh c ea e discom o in o m o subs an ial disu ili y. 105 Imi a ion In his subchap e , we will discuss wo s eams o li e a u e sugges ing ha beha io al ad- ap a ion is he esul o ei he pu e beha io al imi a ion (e.g. due o obse ing beha io ha leads o a supe io ou come) o he adjus men o p e e ences (e.g. con ingen on he be- ha io o he pee s). Wi h no claim o comple eness, his chap e ’s ocus will lie on he se- lec ed sample o con ibu ions discussing he opics o beha io al imi a ion and adjus men o p e e ence as a mechanism o beha io al adap a ion. Pu e Beha io al Imi a ion Imi a ion is a common beha io al ai o bo h animals and humans (Laland, 2002). While i is likely ha imi a ion is d i en by e olu iona y and cul u al ace s, i s pe asi eness can only be explained i i su icien ly o en leads o a desi able ou come. Imi a ion has a numbe o desi able ea u es because i allows one, among o he hings, o ee ide on supe io in- o ma ion (in he spi i o social lea ning, bu see also Sinclai (1990)), o sa e men al e- sou ces (in he spi i o bounded a ionali y, bu see also Conlisk (1980)), o as a means o be ega ded simila o o he s (in he spi i o he social es eem a gumen , bu see also Cho & K eps (1987)). Along hese lines, Alós-Fe e & Schlag (2009) abs ac om he s anda d Bayesian belie - based app oach and in oduce he concep o imi a ion as a belie - ee beha io al ule. The au ho s make he case ha imi a ion is igge ed by he obse a ion o a supe io ou come. In ace o di e en imi a ion concep s, he deg ee o sophis ica ion in ol ed in assessing he ex en o he obse ed agen ’s be e pe o mance clea ly domina es a simple ‘ ollow who- e e pe o ms be e han me’ ule. The eason behind his a gumen a ion is ha “ he eluc- ance o swi ch when he obse ed choices a e only sligh ly be e han he own migh no be due o swi ching cos s, bu a he o he ac ha he payo -sensi i i y o he imi a ion ule allows he popula ion o lea n he bes op ion” (Alos-Fe e & Schlag, 2009, p. 273). In essence, his app oach highligh s he idea ha beha io al adap a ion can be igge ed by beha io al imi a ion ha esul s om obse ing (su icien ly) supe io ou comes. Adjus men o P e e ences Ano he s ain o li e a u e models beha io al adap a ion in a way ha is ha d o swallow by adi ional economis s: p e e ences a e no s able pe se bu a e subjec o in e ela ions 112 may induce a ca y-o e e ec o beha io (Golds ein & Cialdini, 2007), whe eas he pe spec- i e- aking heo y assumes ha he imagined social closeness is al eady su icien o acili- a e beha io al adap a ion (Galinsky, e al., 2005). As has been discussed p e iously, psychological esea ch has in oduced a numbe o ways o p oxy social iden i y in he lab. Due o me hodological di e ences, exis ing economic e- sea ch has eso ed p ima ily o using ough p oxies o in oduce social iden i ica ion in one o m o ano he . Exempla ily, his has been achie ed by a ying he wo ding o he expe i- men al ins uc ions (Ho man, e al., 1996), using ace- o- ace in e ac ion (Bohne & F ey, 1999) o showing pic u es (Eckel & Pe ie, 2011), e ealing names (Cha ness & Gneezy, 2008), e ealing p e e ences such as hose o pain ings (Chen & Li, 2009), and ec ui ing iends and amily membe s (B and s & Solà, 2010). In line wi h mos psychological esea ch, hese s udies ind ha social iden i y igge s in-g oup a o i ism. Howe e , wo na u al p oblems a ise wi h hese app oaches o in oduce social iden i ica- ion in he lab: i s ly, he in oduc ion o po en ial biases caused by ace- o- ace communi- ca ion wi hin he lab o le ing pa icipan s in e ac wi h hei kin (Ro h, 1995). Exempla ily, i has been shown ha indi iduals disc imina e using social cues such as looks o skin colo (Eckel & Pe ie, 2011). Addi ional esea ch has in es iga ed o he - ega ding beha io as a unc ion o social dis ance. Using he dic a o game se ing, Bohne & F ey (1999) ind ha o he - ega ding beha io is mo e p onounced unde s onge iden i ica ion. They incep ed s onge iden i ica ion by he use o iden i ica ion p io o he dic a o game. Again, hei app oach is only a ough p oxy o social p oximi y and does no con ol o po en ial biases s emming om, among o he s, unobse ed a ac ion e ec s. Secondly, and mo e along he lines o ou expe imen , applying hese me hods o in oduce social iden i ica ion does no usually allow o a a ia ion o social p oximi y beyond a simple bina y ou come o no social iden i ica ion e sus some social iden i ica ion. In ou expe imen , we p opose a me hod ha allows us o na u ally a y he deg ee o social iden i ica ion in he lab wi hou ha ing o deal wi h he issues aised abo e. We will e u n o his in ou design sec ion. O e all, i may be s a ed ha di e en s eams o exis ing esea ch poin a he signi icance o social iden i y in explaining he magni ude o beha io al adap a ion hus making i wo h- while o examine his opic in mo e de ail. 113 The e hicali y o obse ed beha io migh also be decisi e o he magni ude o beha io al adap a ion. As we will a gue h oughou his pape and will be explained mo e o mally in his chap e , we expec ha obse a ions o une hical beha io ende beha io al adap a- ion mo e likely and mo e ex ensi e as compa ed o obse a ions o e hical beha io . Exis - ing empi ical li e a u e on he slippe y-slope e ec (c . Gino & Baze man (2009), Welsh e al. (2015)) as well as on he “b oken windows e ec ” (c . Beckenkamp e al. (2014), Le eb e, Pes ieau, Riedl & Ville al (2015)) suppo his poin o iew.28 A guably, he magni ude o be- ha io al adap a ion is asymme ic in one di ec ion o ano he , indica ing ha he likeli- hoods o adap ing beha io a e di e en when obse ing e hical o une hical beha io . Be- low, we will p o ide some heo e ical a gumen s o his claim be o e u ning o a mo e o - mal app oach. Indi iduals engage in e hical and une hical beha io o all so s o easons and some imes u n e il ex emely quickly unde pa icula ci cums ances (see he in amous Milg am ex- pe imen (Milg am, 1963) and Zimba do’s S an o d p ison expe imen (Zimba do, 1971)). Di e en pe spec i es a e help ul in unde s anding one’s a ionaliza ion o engage in (un)e hical beha io , such as sel - a ionaliza ion, obedience o no ms, o ac i e signaling o one’s pee s. Indi iduals end o a ionalize hei beha io wi h hemsel es while ying o a oid cogni i e s ain s emming om, among o he s, nega i e sel -image upda ing (c . Wells (1978), Baumeis e (1998)). He e, indi iduals do no wan o hink o hemsel es as a bad pe son and hus end o engage in sel -decep ion and mo al a ionaliza ion and c ea e posi i e dis- illusions ha sus ain a posi i e sel -image (c . Taylo (1989), Maza , Ami & A iely (2008)). “Sel -decep ion allows one o beha e sel -in e es edly while, a he same ime, alsely be- lie ing ha one’s mo al p inciples we e upheld. The end esul o his in e nal con game is ha he e hical aspec s o he decision “ ade” in o he backg ound, he mo al implica ions obscu ed” (Tenb unsel & Messick, 2004, p. 223). A guably, by a oiding hinking o immo al ac s as such, men al p ocesses leading o a denial o ac s acili a e a pe son’s inclina ion owa ds une hical beha io . 28 Al hough no explici ly along he lines o (un)e hical beha io , a simila a gumen can be made based on he indings wi hin he ex ensi e li e a u e on condi ional coope a ion and punishmen (see Chaudhu i (2010) o a selec i e e iew). 114 F om he pe spec i e o no m heo y, he exposi ion o une hical beha io comes wi h a el- a i ely highe isk o being condemned by and excluded om he social g oup (Bicchie i, 2006). The deg ee o which one can beha e une hically (i a all) among one’s social pee s is ypically g ounded in unce ain y. Consequen ly, a pe son’s decision o engage in ei he e h- ical o une hical beha io is no only simple cos -gain maximiza ion unde isk bu also in- ol es accoun ing o he inhe en unce ain y. The unce ain y in ol es, among o he s, he de ia ion om wha is deemed app op ia e wi hin he social g oup, epu a ion loss, and he po en ial exclusion om he g oup. F om a psychological pe spec i e, engaging in une hical beha io in ol es highe cos s, as one has o o ce ully con ince onesel ha gaining pe - sonally a he expense o ha ming o he s is accep able. P ocesses leading up o his eason- ing in ol e, among o he s, sel -decep ion, jus i ica ion, and hypoc isy. Typically, only he ac ing indi idual p o i s om une hical beha io in (non-) mone a y e ms, such as s ealing money, while ano he pe son is losing ou . In u n, howe e , no mally bo h indi iduals gain om e hical beha io such as dona ing money: he ecei ing indi idual gains in mone a y e ms and he ac ing indi idual, while losing ou in mone a y e ms, is (o e -)compensa ed om a beha io al pe spec i e by ei he he posi i e eelings o wa m glow o gi ing and al- uism (Becke (1974), And eoni (1989) & (1990)) o social p essu e (Ake lo & K an on, 2000). Plausibly, o con ince onesel o beha e une hically in ol es mo e (psychological) e o han o beha e e hically. F om a signaling pe spec i e, obse ing someone who is socially close o onesel as com- pa ed o obse ing a s ange sends a mo e salien o m o social lea ning abou app op ia e (o a leas ole a ed) beha io wi hin he social g oup. This assump ion is also along he lines o Schelling’s (1968) p ominen ly s a ed “ he mo e we know, he mo e we ca e” and a numbe o expe imen al esul s in a ious se ings (see Eckel & G ossman (1996), Bohne & F ey (1999), Cha ness & Gneezy (2008), and Gino & Gallinsky (2012)). The esul ing esolu ion o unce ain y allows he indi idual o o e ule hei own conce ns abou he inapp op ia e- ness o une hical beha io . This s eam o li e a u e sugges s ha obse ing une hical be- ha io migh igge s onge con agion as compa ed o obse ing e hical beha io . Fo hese and mo e easons, we expec beha io al adap a ion o be asymme ically biased o- wa ds une hical beha io . 115 3.1.5.2 Theo e ical Model We ex end he model in oduced by Ake lo (1997) and expand he unde lying concep s in a way ha is conduci e o unde s anding he ole o social iden i y and he (un)e hicali y o obse ed beha io in beha io al con agion.29 P e ious wo k pu emphasis on modeling so- cial in e ac ions and beha io al adap a ion as a unc ion o dis ance in he social space, e.g. esul ing om dis ance in beha io and geog aphic loca ion o one’s pee s. We, howe e , shed ligh on he ele ance o social dis ance and p oximi y esul ing om o e lapping and common in e es s and p e e ences in d i ing beha io al adap a ion. In ui i ely, one could easonably assume ha he beha io o one’s amily and iends exhibi s a mo e salien sig- nal and hus is aken mo e s ongly in o conside a ion han obse ing andom s ange s. Wha is mo e, we ollow he p e ious discussion and assume addi ionally an asymme y in beha io al con agion depending on he e hicali y o he obse ed beha io . This is a sub- s an ial ex ension o Ake lo (1997), whe e an asymme ic adap a ion in beha io canno be illus a ed. The pu pose o his sec ion is o ou line a simple educed- o m model ha allows one o d aw p edic ions abou indi idual con ibu ion decisions in he a ious con ex s ha a e e lec ed by he expe imen al design. These p edic ions can hen be es ed empi ically. The p incipal goal o his exe cise is o show how beha io al con agion is media ed by social iden i y and he (un)e hicali y o obse ed beha io and how hese ac o s a ec he adap- a ion gap, ha is he di e ence be ween one’s own and obse ed beha io , a e con agion has aken place. We will in oduce he symme ic model i s be o e u ning o he ex ension. Symme ic Con agion P io o in oducing he o mal model, we begin wi h he in ui ion o wha he model should cap u e. The unde lying idea is ha indi idual beha io is a unc ion o he pee ’s ac ions 29 In pa icula , we ocus on Ake lo ’s (1997) quad a ic u ili y e sion o he con o mi y model as his gene a es unique equilib ia om which we can de i e es able hypo heses o ou expe imen , which we also deem o be mo e in line wi h he gene al s o y o his pape . While Ake lo ’s gene al model pu s mo e emphasis on one’s own decisions, he quad a ic u ili y model has an a p io i assump ion ha pu s one’s own and pee beha io on an equal oo ing wi h espec o how beha io a ec s indi idual u ili y. Assuming ha indi iduals a e on a con inuum be ween ex emely sel ish and ex emely al uis ic, he quad a ic u ili y app oach is conclusi e. I should be no ed ha his assump ion is no c ucial o ou model’s p edic ions as we mainly ocus on he ele ance o wo ac o s, i.e. social iden i y and he (un)e hicali y o obse ed beha io , in a ec ing one’s own decisions. Thus, he equal-weigh assump ion is no decisi e in p edic ing he di ec ion in which indi idual beha io changes as a unc ion o hose wo ac o s. 116 and hus encompasses mo e han one’s sel - e e en iali y. In a si ua ion o social in e ac- ion, an indi idual is expec ed o ace a ade-o , weighing one’s own p e e ences agains he pee ’s e ealed p e e ences. To s ay in Manski’s (1993) e ms, social in e ac ion and he unde s anding ha one’s own ac ions a e e lec ed by he pee s and hus ha e an impac on o he (closely ela ed) people shapes he indi idual’s willingness o engage in beha io al adap a ion. The magni ude o which one is willing o e ise one’s own ini ial beha io and adap is i s and o emos a unc ion o he social p oximi y o he pee . “As a consequence, he impac o my choices on my in e ac ions wi h o he membe s o my social ne wo k may be he p ima y de e minan o my decision, wi h he o dina y de e minan s o choice he di ec addi ions and sub ac ions om u ili y due o he choice) o only seconda y im- po ance” (Ake lo , 1997, pp. 1006-1007). I is hus easonable o assume ha he indi id- ual’s u ili y is subjec o a ela i e e alua ion o one’s own beha io and he beha io o he pee s. Simila o Ake lo (1997), he unde lying cha ac e is ic o ou model is he ea u e ha indi idual u ili y is declining wi h inc easing dis ance be ween one’s own beha io and he pee ’s beha io . While he aim o his sec ion is o ou line a model ha explains changes in gene al beha io , we will use he esul ing p edic ions o gene a e hypo heses in he con ex o (un)e hical beha io . In ou design, we eso o a wo-s age dic a o game in which each indi idual is pai ed wi h a cha i y o his choice. Bo h, he indi idual and he cha i y s a wi h an ini ial endowmen ==∈ ℝ o equal size. A each s age ∈[1,2], each indi idual  aces he choice, , o ei he (a) dona e (pa o all o ) one’s own money o he cha i y, (b) e ain he equal spli , o (c) ake away (pa o all o ) he cha i y’s money and add i o one’s own income. We will e e o (a) and (c) as e hical o une hical beha io , espec i ely. Na u ally, he indi idual’s decision is o he o m  ∈ [−I, +I ]. The only di e ence be ween bo h s ages is he in o - ma ion se ha he indi idual possesses abou his pee ’s beha io . Tha is, a e comple ion o s age 1, he indi idual obse es a andom indi idual’s beha io om s age 1. A s age 2 ( ha is, a e he obse a ion), he indi idual is gi en he oppo uni y o e ise his ini ial decision, i desi ed. Le .∈(0,1) depic he social p oximi y o an indi idual a ime . Impo an ly, le an in- di idual’s inhe en a i ude owa ds (un)e hical beha io be desc ibed by . Tha is,  ep- esen s he indi idual ’s p e e ence o gi e o ake a pa icula mone a y amoun wi hin he 117 bounda ies o one’s income in a gi en si ua ion, hus being de ined as  ∈ [−I, +I ]. Wha is mo e, le  ep esen indi idual ′ p io (s age 1) o ac ual obse a ion (s age 2) o indi- idual ′ (un)e hical beha io . (1) =  = 1 (   )  ( ) F om his i ollows ha each playe maximizes own u ili y a S age 1 o he ollowing o m: (2)   =−1−∗−−∗− A s age 1, indi idual  has no in o ma ion abou indi idual ’s decision, as all pa icipan s ca y ou hei decisions simul aneously. Consequen ly,  eso s o o ming belie s abou he beha io o . As depic ed abo e, indi iduals ace a ade-o decision a s age 1, in which de ia ion om he indi idual inhe en cha ac e is ic  has o be weighed agains de ia ing om one’s own belie s abou he pee ’s beha io . Because no in o ma ion abou he pee s was gi en a s age 1, he esul ing decision is a simple maximiza ion p oblem o he o m: (3)  = 2∗−−2∗−1∗− . ⇔∗==−+ yielding he compa a i e s a ic: (4)  =− We can in e ha a change o  in  depends on he social p oximi y ( ha is ei he a p io in = o an upda ed belie in =) in he ollowing way: (5)  = > 0  >(   ) = 0  = (   ) < 0  < (   ) 118 In o de o s udy he adap a ion gap a s age 2, ha is he di e ence in beha io be ween indi idual  and  a e he second s age, one has o hold ’s beha io om s age 1 cons an while gi ing indi idual  he abili y o e ise his ini ial decision.30 P oposi ion: Equa ion (2) p o ides a solu ion o he maximiza ion p oblem and educes he adap a ion gap o: (6) −=−+−=∗1−+(−1) Obse e ha unlike in Ake lo ’s (1997) gene al con o mi y model, his app oach gene a es a unique equilib ium p edic ion due o es ic ions pu on he social p oximi y pa ame e , ha is 0 < .< and he linea eac ion unc ion. Essen ially, by assuming ha beha io al adap a ion is symme ic in ei he di ec ion his in- dica es ha he gap is d i en by he social p oximi y o he pee ha one is obse ing: he close indi idual a e in e ms o p oximi y he smalle is he expec ed gap and he mo e sim- ila is hei beha io . An al e na i e in e p e a ion is ha he pee e ec is s onge and leads o a mo e ex ensi e beha io al adap a ion he highe hei social p oximi y is. Asymme ic Con agion The main pu pose o he ex ended model is o allow o an asymme ic adap a ion o une h- ical and e hical beha io . I is plausible o assume ha no only obse ing bu also s a ing o ac on une hical beha io equi es a di e en mindse and igge s o he cogni i e p o- cesses han i is he case o e hical beha io . He e, he abili y o sel -jus i y beha io de- pends on he (un)e hicali y o he (obse ed) ac o he ex en ha i a ies he bounda ies o he mo al wiggle oom (Dana, e al., 2007). Empi ical suppo is p o ided by he slippe y- slope e ec (c . Gino & Baze man (2009), Welsh e al. (2015)) as well as esul s on he “b oken windows e ec ” (c . Beckenkamp e al. (2014), Le eb e, Pes ieau, Riedl & Ville al (2015)), which sugges ha une hical beha io is likely o be mo e con agious han une hical beha - io . We in oduce , which ep esen s a ac o biasing beha io al adap a ion owa ds he obse - a ion o une hical beha io . We assume ha he deg ee o he adap a ion bias depends on 30 Which is exac ly wha we will do in ou expe imen . See chap e 3.1.6. 119 he di e ence be ween one’s own and obse ed une hical beha io −. Wi h  s ill e aining i s p ope ies om (2), he indi idual’s maximiza ion is o he o m: (7)   =−1−∗−−∗∗− wi h: (8) = max (−+ 1, 1) This de ini ion illus a es ha whene e one’s own ini ial beha io is mo e e hical han wha is being obse ed om he pee beha io al con agion is s onge han obse ing mo e be- ha io ha is mo e e hical. The s eng h o he di e ence in his con agion o ce depends on he dis ance be ween own and obse ed beha io . This implies he ollowing maximiza- ion o he indi idual’s u ili y: (9)  = 2k∗−−2∗−1∗− . ⇔=−+ −+ 1 wi h he compa a i e s a ic: (10)  =∗()  Unde hese assump ions, he esul ing adap a ion gap looks as ollows: (11) −=|…|=󰈅−()∗()  󰈅 We can easily see ha he sligh ly di e en ly appea ing adap a ion gap e ains he same p ope ies as in he symme ic model. In conclusion, we expec beha io al adap a ion o be d i en by social p oximi y wi h a bias owa ds une hical beha io . Pu di e en ly, we expec he spillo e o une hical beha io o be mo e p onounced, hus leading o a s onge de- gene a ion o beha io ela i ely o he ise o good Sama i ans. 120 3.1.6 The Expe imen 3.1.6.1 Expe imen al Design and P ocedu e In o de o s udy beha io al con agion, we a e mainly in e es ed in answe ing wo ques ions wi h ou expe imen al design: i s , whe he indi iduals e ise hei ini ial beha io in ligh o obse ing pee beha io . Second, whe he and how a beha io al change depends on bo h he e hicali y o he obse ed beha io and he social iden i ica ion wi h he obse ed pee s. Ou design d aws on a unique app oach o s udy beha io al con agion in he lab ha allows us o accoun o po en ial con ounds po en ially inhe en in o pee e ec s s udies as a - gued by Manski (1993) and Ang is (2014) (see Thöni & Gäch e (2015) o a simila app oach). In o de o be ega ded as beha io al con agion, e ised beha io has o be mo e simila o obse ed beha io han one’s ini ial beha io ha one has decided upon p io o lea ning pee in o ma ion. Tha is, e ision o one’s ini ial beha io mus ollow he di ec ion o ob- se ed ini ial pee beha io . Ou basic design ollows his s aigh o wa d p ocedu e: ac ion – obse a ion o a pee – eac ion.31 Conside a a ian o a wo-playe dic a o game in which he pa icipan (dic a- o ) is ma ched wi h a cha i y ( ecipien ). The dic a o ’s ac ion space en ails aking away money om he cha i y, lea ing he ini ial si ua ion unchanged, o gi e money o a sel -cho- sen cha i y ( he basic design ollows Lis (2007) and Ba dsley (2008)). In ollowing Eckel & G ossman (1996), we use a cha i y o inc ease he saliency o he in ol ed decisions. The expe imen is played one-sho wi h a possibili y o e ise one’s ini ial beha io . Be ween he ini ial decision and po en ial e ision, indi iduals a e gi en he oppo uni y o obse e he ini ial beha io o ano he andom pa icipan . Alongside he ac ual beha io , ea men a ia ions include he al e a ion o un eiled social p oximi y in o ma ion o he obse ed pa icipan .32 Tha is, in addi ion o lea ning ac ual beha io and he amoun ha was aken away o gi en by his pa icipan , addi ional in o ma ion on he pa icipan ’s social p oxim- i y o onesel is a ied wi h he andom ea men assignmen . The ea men a ia ion lies in he in o ma ion gi en abou he social p oximi y o he obse ed pee s: no in o ma ion on p oximi y (Baseline), as well as high p oximi y (T1) and low p oximi y (T2) in o ma ion. 31 No e ha in o de o exclude any hedging conce ns h oughou he whole expe imen , in o ma ion abou he speci ics o he design we e only p o ided whe e necessa y in o de o each a delibe a e decision. Tha is, a S age 1 pa icipan s we e nei he awa e o he possibili y o obse e pee s la e on no o e ise hei ini ial decision, ensu ing unbiased ini ial beha io . 32 Hence o h, we will use he e ms social p oximi y and social iden i y in e changeably. 121 P oximi y is calcula ed based on o e lapping answe s in he lis o s a emen s used in he beginning o each session and hen p esen ed o he pa icipan s in he o m o below- o abo e-a e age p oximi y in o ma ion o he obse ed pee .33 We capi alize on a sho ened 25 i ems lis o s a emen s compiled om a majo US Ame ican da ing websi e o ensu e he alidi y o he ques ions in success ully ma ching people (see Gibbs, Ellison & Heino (2006) and Hi sch, Ho açsu & A iely (2010) o a discussion).34 Since he business concep o da ing websi es is based on achie ing high ma ching success a es, he use o such alida ed ques- ions imp o es he success o incep ing social iden i ica ion be ween pa icipan s in he lab.35 The expe imen al p ocedu e is ep esen ed by a single i e a ion o he ollowing h ee s ages: Fi s S age - The Ac ion: S a ing wi h an equal dis ibu ion o money, each indi idual de- cides whe he o (i) dona e own money o he cha i y’s accoun , (ii) no change he ini ial equal dis ibu ion, o (iii) ake money om he cha i y and add o one’s own accoun . Second S age – The Obse a ion: Each ac i e playe obse es one passi e playe o an- dom who has engaged in ei he e hical o une hical beha io . In all h ee ea men s, he exac in o ma ion en ails he mone a y amoun aken away om o gi en o he cha i y. Ex- 33 The implemen a ion o he low- and high-p oximi y in o ma ion ollowed a e y s aigh o wa d calcula ion. Fo each pa icipan o he ac i e g oup, an indi idual p oximi y sco e o bo h pa icipan s o he passi e g oup was calcula ed based on o e lapping answe s in he lis o s a emen s. F om each ac i e pa icipan ’s indi idual pe spec i e, he passi e pa icipan wi h he highe (lowe ) sco e was labeled as he high (low) p oximi y pee . In ac , his calcula ion app oach allows o he same passi e pe son o be o high (low) p oximi y o one ac i e pe son, while being o low (high) p oximi y o ano he ac i e pe son, hus uly andomizing in o ma ion. We abs ained om p o iding explici ma ching sco es o pe cen ages o e ain maximum con ol. In addi ion, his allows us o alle ia e he alse-consensus e ec , in which people sys ema ically o e es ima e he deg ee o simila i y o o he s. The p o ision o social cues o his kind allows he pa icipan s o upda e hei belie s eliably wi h espec o he ac ual deg ee o simila i y. See Ellingsen & Johannesson (2008, p. 995) o a discussion. 34 We should s ess he ac ha we epo lowe -bound esul s. We induce social p oximi y in a e y simple way by p o iding pa icipan s wi h ei he he high- o low-p oximi y signal in he social iden i ica ion ea men s 1 and 2. Al hough his app oach allows us o p o ide a comp ehensi e se o in o ma ion o induce e en mo e salien and dis inc o ms o social iden i y (i.e. by p o iding he exac ma ching sco e, he exac answe s o he ques ions, le ing pa icipan s pu di e en weigh s on ques ions acco ding o hei indi idual impo ance and so on), we eso o his easy- o-use-easy- o- ep oduce app oach. See Appendix B o he exac lis o ques ions used in ou expe imen . 35 In addi ion, we elici he s eng h o social iden i y o he obse ed pee using a a ian o he sel -e alua ion scale o one’s social iden i y ollowing Luh anen & C ocke (1992) o e i y he obus ness o ou social iden i y implemen a ion app oach. The non-pa ame ic and eg ession es ima ions yield he same o e all esul s bo h in di ec ion and in magni ude. 128 he cha i y as compa ed o hei ini ial beha io p io o ha ing obse ed a pee ’s beha io . The esul s a e signi ican a he 1% le el (p = 0.000, z = 4.365) and indica e a change in beha io almos h ee imes as la ge when une hical beha io was obse ed as compa ed o e hical beha io . The esul s a e highly sugges i e o une hical beha io being mo e con agious. Thus, hese esul s s ongly suppo ou hypo hesis H1 and con i m ha une h- ical beha io is indeed mo e con agious han e hical beha io independen o social iden i- ica ion.45 Figu e 3 illus a es ou indings. Figu e 3: Amoun Re ised (%) and Obse ed (Un)E hicali y. The igu e depic s he amoun e ised as pe cen age o one’s ini ial beha io . Any alue abo e (below) ze o indica es ha mo e (less) ECU we e gi en a e he e ision o he cha i y ela i e o one’s ini ial decision. The analysis is b oken down in o he (un)e hicali y o obse ed beha io by he ac i e pa icipan s. We a e also in e es ed in he ype o beha io al con agion ha is igge ed by social iden i- ica ion. Following hypo hesis H2, we assume ha social iden i ica ion ampli ies he con a- gion o une hical beha io in an o e -p opo ional way as compa ed o con agion o e hical 45 Unless no ed o he wise in he esul s sec ion, we ob ain simila esul s in e ms o e idence and signi icance when using absolu e ECU numbe s a he han pe cen ages. 129 beha io . We hus examine he ole o social iden i ica ion in a ec ing he magni ude and di ec ion o e ision choices. Bo h indings a e illus a ed below in igu e 4. Ou esul s obus ly indica e ha highe social iden i y indeed igge s s onge beha io al con agion, pa icula ly con agion o une hical beha io . As social iden i y inc eases, he magni ude o e ised beha io inc eases as well, peaking a -29.8% o he high p oximi y condi ion. The di e ences in beha io a e signi ican a he 1% le el (p = 0.000, z = 4.759) when compa ing beha io in he unknown p oximi y wi h he high p oximi y condi ion. Likewise, he esul s a e signi ican a he 1% le el (p = 0.000, z = -3.448) when compa ing he high p oximi y wi h he low p oximi y condi ion. He e, again, he nega i e numbe s o amoun e ised sugges ha , in e ms o magni ude o e ised beha io , une hical beha io is s ongly p onounced and hus mo e con agious han e hical beha io . O e all, we ind ample suppo o ou hypo hesis H2 and hus conclude ha he magni ude o a e ision o one’s ini ial beha io is indeed s ongly co ela ed wi h social iden i ica ion. Consis en wi h ou heo e ical model, he esul s also yield s ong suppo o he idea ha he educ ion in he adap a ion gap is d i en by social iden i ica ion. As p edic ed by ., he s onge he social iden i ica ion o one’s pee is a obus p edic o (all a he 1% le el) o one’s adap a ion gap o he obse ed pee a e obse a ion. By he numbe s, we ob ain p = 0.000 and z = 6.104 (p = 0.000 and z = -3.441) when compa ing unknown p oximi y e sus high p oximi y condi ion (high p oximi y e sus low p oximi y condi ion). We ind ha he adap a ion gap is in e sely co ela ed o he social iden i y. 130 Figu e 4: Amoun e ised (%) and adap a ion gap (%) by social p oximi y. Along simila lines, beha io al con agion as a unc ion o bo h social iden i ica ion and he (un)e hicali y o obse ed beha io is illus a ed in igu e 5. The beha io al space is de- sc ibed by h ee al e na i es and also speaks o he (non-)exis ence o obse ed pee e - ec s:  No Con agion: a e obse ing pee beha io , he pa icipan ei he did no e ise his/he ini ial beha io o e ised i in o he opposi e di ec ion o wha he/she ob- se ed he pee has done.  Con agion: a e obse ing pee ’s beha io , he pa icipan did e ise his/he ini ial decision. The e ision was di ec ed in o he di ec ion o he obse ed beha io . This beha io indica es he exis ence o beha io al con agion caused by pee e ec s. The b eakdown o beha io al changes by di e en le els o social p oximi y p o ides addi- ional e idence o hypo hesis H2. When compa ing o he condi ion whe e no social p ox- imi y o one’s obse ed pee was induced, beha io al is mo e con agious o bo h low and high p oximi y si ua ions in bo h he e hical and une hical domain. Again, since we epo 131 lowe -bound esul s, his is a s ong indica ion ha ou me hod o inducing social iden i i- ca ion wo ks and is likely o p oduce e en s onge esul s when social p oximi y would be in oduced in a mo e sophis ica ed way. Figu e 5: Beha io al change by ea men and obse ed (un)e hicali y. This igu e illus a es he ac ion o pa icipan s exhibi ing beha io al con agion b oken down in o bo h he obse ed (un)e hicali y o beha io and he social iden i y o one’s obse ed pee . Unless he ac i e pa icipan e ises his/he beha io in o he di ec ion o ob- se ed beha io he beha io is no classi ied as con agion. In e ms o explaining he d i e s o beha io al con agion, wo li e a u e s eams ha e no ye been b ough oge he in exis ing esea ch: he ole o social iden i ica ion on beha io al con agion on he one hand, and he ole o obse ed beha io al une hicali y on he o he . Ou expe imen al design allows us o do exac ly his o he i s ime and asce ain he main d i e o beha io al con agion. Following ou p e ious discussion, we assume social iden i- ica ion o be a s onge d i e o beha io al con agion han he une hicali y o obse ed beha io (H3). As a i s , we in es iga e whe he beha io al con agion is di e en unde a - ied le els o social p oximi y when di ec ly compa ing con agion in he e hical e sus he une hical domain using non-pa ame ic compa isons. Ou esul s p o ide s ong suppo ha beha io al con agion is asymme ic. In pa icula , a a ia ion in social iden i ica ion yields no signi ican a ia ion in beha io al changes in he e hical domain. Howe e , he e- sul s a e s ongly s a is ically signi ican when looking a beha io al changes as a unc ion 132 o social iden i ica ion in he une hical domain. He e, when compa ing he high p oximi y condi ion o he no p oximi y (low p oximi y) condi ion, he Mann-Whi ney-U s a is ics yield esul s ha a e highly signi ican a he 1% le el wi h p = 0.000 and z = 6.025 (p = 0.000 and z = -4.005). These esul s highligh he impo ance o ou con ibu ion in his pape : pee e ec s a e no uni o mly in place, bu a he s ongly depend on bo h he (un)e hicali y o obse ed beha io and he deg ee o social iden i ica ion o he obse ed pee . Figu e 6: Amoun e ised (%) by social p oximi y and obse ed (un)e hicali y. Wha is mo e, we an se e al eg essions, including OLS, Logi , and Tobi , whe e applicable, in o de o assess he obus ness o ou esul s. In sum, ac oss di e en speci ica ions ou esul s s ongly sugges ha he obse a ion o (un)e hical beha io does no igge any pa icula beha io , nei he e hical no une hical. Thus, he me e obse a ion o beha io alone is insu icien o he exis ence o pee e ec s in he (un)e hical sphe e, bu is a he dependen on he social iden i ica ion o one’s pee s. These indings a e in suppo o ou hypo hesis H3. Findings a e p esen ed in Appendix A. 133 All in all, we ind con incing suppo o all o ou h ee hypo heses: une hical beha io is indeed mo e con agious han e hical beha io (H1), social iden i ica ion d i es he magni- ude o beha io al con agion (H2), and social iden i ica ion is a mo e eliable p edic o o beha io al con agion han he obse ed (un)e hicali y o obse ed beha io (H3). He e, i is impo an o s ess a pa icula poin wi h espec o he in e p e a ion o he esul s. A gu- ably, he expe imen ’s amed en i onmen c ea ed by he p esence o a cha i y could po- en ially ca y a no m o p osocial beha io in he lab in and o i sel . Tha is, p io o obse - ing one’s pee s, some pa icipan s a he ma gin o beha ing une hically migh ca y he no ma i e belie ha aking om a cha i y ep esen s inapp op ia e beha io and hus ini- ially e ain o i . I so, i would come as no su p ise o obse e s onge con agion o un- e hical as compa ed o e hical beha io because hose who wan ed o beha e une hically in he i s place bu e ained om doing so migh now ind jus i ica ion in hei pee ’s be- ha io . In his espec , wo impo an ema ks should be made: i s , i any hing, such an assump ion would only explain le el e ec s bu no ea men di e ences because such be- lie s a e by expe imen al design unco ela ed wi h he ea men s. Consequen ly, i espec- i e o he exis ence o po en ial no ms, ou design ende s ou main inding alid: beha - io al con agion is highes whe e social iden i y is s onges .46 In addi ion, we elici ed incen- i ized belie s abou wha pa icipan s hough abou his pee ’s beha io p io o obse ing i . As he eg essions esul s sugges , such belie s yielded no explana o y powe and hus play no ole in nei he explaining he magni ude no he di e ences o beha io al con agion. F om a policy pe spec i e, ou esul s s ess ha social p oximi y ende s i di icul o change indi idual beha io , bu i a he ampli ies one’s ini ial (un)e hicali y. We will e u n o his poin in ou policy ecommenda ions in chap e 3.1.7. In ligh o he e y conse a i e incep ion o social iden i ica ion o p o iding e y limi ed in o ma ion on social iden i ica- ion, we deem hese esul s o ep esen a lowe bound hus s eng hening he ole o social iden i ica ion wi hin he con ex o beha io al con agion. Ou lowe -bound app oach comes om he ac ha pa icipan s we e nei he old he exac ma ching pe cen age no he ac- ual in e es s and p e e ences hey had in common. I ha ing been andomly assigned o one o he social iden i ica ion ea men s, pa icipan s only knew whe he hey we e ob- se ing a pee wi h abo e o below a e age cong uence. A mo e sophis ica ed way o induce 46 I would like o hank Ga y Bol on, René Fah , and Elena Ka ok o poin his ou and o ela ed discussions. 134 social iden i ica ion and ma ch pa icipan s acco dingly is likely o p oduce esul s ha a e mo e p onounced. The well-enginee ed mechanisms implemen ed by da ing websi es o ma ch people and achie e high success a es a e a shining example o wha is possible: excluding ma ching pa ne s based on pe sonali y ai s ha ep esen a no-go (e.g. smok- ing), pu ing emphasis on pa icula in e es s (e.g. spo s, eligion), o indi idual cha ac e - is ics (e.g. looks, educa ion). We delibe a i ely e ained om applying sophis ica ed measu es o his kind and a he eso ed o an easy- o-use-easy- o- ep oduce me hodo- logical app oach ha could be used in u u e expe imen s in which inducing salien social iden i y is key. Along hese lines, he p esence o a po en ial expe imen e demand e ec s (EDE) is wo h add essing since hei p esence has po en ially been p oblema ic o p io pee e ec s udies ( o a discussion see Thöni & Gäch e (2015)). Because we a e mainly in e es ed in ea men di e ences a he han in o e all le els, he expe imen e demand e ec is deemed less p oblema ic as long as i s exis ence and magni ude is o hogonal o he ea men a ia ion (Zizzo, 2010). None heless, we conside ed exis ing expe imen al s udies o ule ou expe i- men e demand e ec s o he ex en possible. No exclusi ely o bu p ominen ly exis ing in pee e ec s udies, o ced lea ning (i.e., o ced obse a ion o one’s pee ’s beha io ) migh po en ially induce EDE o e en lead o esen men on he side o he pa icipan s. Fo ced obse a ion migh igge hough s ela ed o being expec ed o use he in o ma ion o econside and po en ially e ise ini ial beha io . In p e ious gene al and pee e ec s udies in pa icula , his issue has no mally been o e looked, mainly o a oid sel -selec ion p oblems. Howe e , when he op ion no o lea n is wi hheld om pa icipan s, he ob- ained esul s a e po en ially con ounded. We deem his challenging o he s udy o pee e ec s and should hus be discussed. Exis ing esea ch indica es ha indi iduals some imes choose o delibe a ely emain igno- an abou he s a e o na u e (see Ca illo & Ma io i (2000), Dana, Webe & Kuang (2007), Con ads & I lenbusch (2013), Ba ling, Engl & Webe (2014), and G ossman (2014)). I p e- sen , such s a egic igno ance migh be an impo an componen o ou expe imen . A po- en ial eason no o acqui e cos less in o ma ion is ela ed o he a oidance o indulge in nega i e sel -image upda ing o guil a e sion (c . Wells (1978), Baumeis e (1998), Cha ness & Du wenbe g (2006)). One can plausibly assume ha such a e sion is e en s onge when 135 s udying pee e ec s wi hin an (un)e hical dimension. Thus, o cing pa icipan s o lea n po en ially unpleasan in o ma ion migh lead o biased beha io and migh e en inc ease EDE. In o de o add ess his challenge, ou design ollows Con ads & I lenbusch (2013) and Ba ling, Engl & Webe (2014) and d aws on a mechanism in which lea ning pee in o ma ion is olun a y (see also Eckel & Pe ie (2011)).47 Addi ionally, in o de o ule ou any epu a- ional conce ns, social lea ning, o ecip oci y, he expe imen includes an anonymously played one-sho in e ac ion wi h ano he pa icipan . Such an expe imen al design allows us o s udy beha io al con agion in he lab in an unbiased way. To ou knowledge, his ep- esen s a no el design in s udying pee e ec s in he lab in gene al and he beha io al con- agion o (un)e hical beha io in pa icula while con olling o po en ial con ounds ha pee e ec s udies su e om egula ly (see Manski (2000), Falk & Fischbache (2002), An- g is (2014)). In he ligh o ou expe imen al se -up, any ea men -speci ic in o ma ion is p o ided only a e one’s delibe a e decision o lea n pee beha io . Thus, any s ill po en- ially exis ing o m o EDE would be ully unco ela ed wi h he ea men s and hus exhibi only a gene al le el-e ec , i any. 3.1.7 Lessons Lea ned: Policy Implica ions As a gued be o e, unde s anding social in e ac ions in gene al and in pa icula he po en- ially esul ing pee e ec s is undamen al om a policy pe spec i e. I does no only help o unde s and socie al and economic ou comes beyond wha s anda d economic o ces can explain (i.e., he massi e su ge in emale labo pa icipa ion a es in Wo ld Wa II (Mulligan, 1998) o he escala ion o c ime a es (Le i , 1999)). I also allows us o implemen be e - a ge ed policy measu es o ackle a ba e y o challenges such as educing c ime a es, im- p o ing heal h condi ions, o inc easing labo ma ke pa icipa ion. “To he ex en ha he- o y and measu emen o social in e ac ions enables us o unde s and hese massi e changes, he s udy o social in e ac ions po en ially has majo policy ele ance” (Glaese & Scheinkman, 2004, p. 84). In his chap e , we will discuss some o he policy implica ions 47 Howe e , delibe a ely allowing pa icipan s o emain igno an abou pee beha io bea s he isk o sel -selec ion e ec s. I is wo h no ing ha in ou expe imen 10% o all he pa icipan s decided no o acqui e pee in o ma ion. Impo an ly, howe e , his choice is uncondi ional on he pa icipan ’s ini ial beha io , hus s ongly sugges ing he absence o any sel -selec ion mechanism leading o po en ial biases in ou analysis. 136 ha one can in e om ou esul s. We will ollow he main heme o his pape and ap- p oach his opic om wo sides: he e hical and une hical con ex ( o a b oade discussion see I lenbusch & Ville al (2015)). S a ing wi h he e hical pe spec i e, olun a y edis ibu ion o income e.g. in he o m o cha i able gi ing is an in eg al pa o humaneness, wi h up o 90% o Ame icans dona ing o cha i ies. Unde s anding he d i e s o cha i able gi ing, such as al uism and social p es- su e, has been a he hea o las decade’s esea ch, bo h in he ield and in he lab (c . Le i & Lis (2009), DellaVigna, Lis & Malmendie (2012)). S ill, we seem o ha e an impe ec un- de s anding o wha eally mo i a es gi ing (And eoni, 2006). Beyond wha we ha e since lea ned abou he exis ence o pee e ec s wi h espec o con ibu ion decisions, ou ex- pe imen yields se e al new insigh s help ul o unde s anding he ex en o which such be- ha io shows up, especially as compa ed o une hical beha io . We will e u n o his com- pa ison sho ly. In u n, une hical beha io in i s a ious o ms impai s he daily li e. Exempla ily, yea ly global ax e asion anges a an abs use $3.1 illion o 5.1% o wo ld GDP (The New Yo k Times, 2011), o e $1 illion is es ima ed in b ibes paid yea ly a ound he globe (The Wo ld Bank, 2013), and some 210 million people use illici d ugs each yea (Uni ed Na ions O ice on D ugs and C ime, 2011). He e, one migh c edibly a gue ha om a pu ely a ional sel - maximizing pe spec i e we should obse e way less illici beha io han we ac ually do. F om a game heo e ic pe spec i e, in some illici deals ha in ol e us - ela ed ac ions such as b ibe y do no ep esen a subgame pe ec Nash equilib ium, which is ue o bo h one-sho and ini ely epea ed con ex s. Thus, in many si ua ions he occu ence o illici beha io is al eady su p ising ( o a discussion, see Diman & Schul e ( o hcoming)). A so- lu ion o his con lic is he ecogni ion o , among o he s, pee e ec s. As has been ho - oughly a gued h oughou he pape , he inco po a ion o beha io al con agion allows us so explain why obse ed beha io goes seemingly beyond clea -cu sel -maximiza ion, bu is a he embedded in and he esul o a social con ex . Consequen ly, bo h aspec s aise he ollowing ques ion: how do hese indings ansla e in o a eal wo ld se ing and wha o do abou i ? While being ca e ul a d awing conc e e in e ences om a labo a o y se ing and ela ing hem di ec ly o he ou side wo ld, ou e- sul s conclusi ely indica e he spillo e o une hical beha io o be much mo e likely han 137 he spillo e o e hical beha io . Tha is, ge ing people o s a dona ing solely based on pee e ec s (e.g. h ough obse ing o he s gi ing o cha i y) has a long way o go compa ed o ha ing hem do some hing une hical. Wha we see, howe e , is he indi idual’s espon- si eness o social iden i ica ion, in pa icula on he side o emales. A po en ial solu ion o ha e people dona e mo e o a good cause is o p o ide hem in o ma ion beyond simple s a is ics on wha o he people do, i.e. amoun o money ha has been collec ed so a (like Wikipedia). Ins ead, some esea ch al eady indica es ha he incep ion o social no ms es- pecially o se ings ha mos closely ma ch he indi idual’s immedia e si ua ional ci cum- s ances ha e he s onges e ec on compliance (Golds ein, e al., 2008). Ou esul s sugges o go one s ep u he and p o ide in o ma ion ha allows us o d aw in e ences on he so- cial p oximi y o he pee s, hus inc easing he saliency o social iden i ica ion. Exempla ily, a message o he o m “People in you neighbo hood / wi h simila demog aphic cha ac e - is ics ha e dona ed an a e age o $...” would lead o pick-up a es o his beha io ha a e highe han when eso ing o a simple s a is ic. In all likelihood, a simila app oach could be applied o make people e ain om beha ing une hically. Resea ch on slippe y slope indica es ha once beha io is spoiled, e en hones people con e ge quickly o a s eady s a e wi h a ple ho a o une hical beha io s (c . Gino & Baze man (2009), Welsh e al. (2015)). E en mo e wo isome, ecidi ism a es o con ic s a e no mally e y high, leading o wha is called a ecidi ism nigh ma e. In he US ecidi ism a es a e up o 80% o e-a es s wi hin he i s 3 o 5 yea s a e hei elease om p ison (Na ional Ins i u e o Jus ice, 2014). This is pa icula ly de imen al om a wel a e pe spec- i e, as he US spends app oxima ely $75 billion on inca ce a ion and $260 billion on p e- en ion, de ec ion, and p osecu ion o c ime on a yea ly basis (Khadja i, 2015). In e ms o e ec i e coun e measu es o une hical beha io , ou expe imen al esul s indica e ha ex- posing indi iduals o un ai h ul o socially close people is likely o igge epulsion and less une hical beha io , hus con ibu ing o a posi i e ans o ma ion.48 48 Fo an app oach along simila lines, see Pennsyl ania S a e Uni e si y's Jus ice Cen e o Resea ch on desis ance om c ime (Pennsyl ania S a e Uni e si y's Jus ice Cen e o Resea ch, 2014). 144 A3: Amoun Re ised (%) as a unc ion o social iden i ica ion and obse ed une hicali y. Model 1 and 5 (Model 2 and 6) es s o he e ec o he ea men s (o obse ed une hicali y) while con olling o ini ial beha io and he obse ed adap a ion gap. Model 3 (Model 7) es s o bo h ea men e ec s and obse ed une hicali y simul aneously. Model 4 (Model 8) adds con ols o gende and some in e ac ion e ms as well as he numbe o in e es . In o de o ule ou any endogenous conce ns and s ess he e ec i eness o exogenous a ia ion o social p oximi y we also add dummies o he da ing websi e ques ions. None o he dummies u n ou signi ican in nei he model speci ica ion, hus emphasizing he obus ness o ou esul s. 145 A4: Examina ion o he adap a ion gap (%) using Logi es ima ions. Pa icipan s who did no e ise hei ini ial decision a e ea ed as adap a ion gap = 100%. 146 A5: Examina ion o he d i e s o beha io al con agion using a mul inomial logis ic eg ession. Beha io o pa icipan s who e ised hei ini ial decision bu in o he opposi e di ec ion o wha hey ha e obse ed om hei pee is ea ed as an i-con agion ( epulsion). 147 Following he eg ession esul s, ou esul s also speak o he idea ha beha io al con agion seems o acili a e he magni ude o (un)e hical beha io a he han changing indi idual beha io o he be e o wo se, espec i ely. Tha is, hose who beha ed (un)e hically in he i s place become e en mo e (un)e hical a e being exposed o hei pee s. This is ue o bo h e ision o one’s ini ial amoun and he educ ion o he adap a ion gap o one’s pee . In mo e de ail and in line wi h ou p e ious esul s, his ela ion is mo e p onounced he mo e salien social iden i y o one’s pee s is and in pa icula in he une hical domain. This is addi ional suppo o ou hypo hesis H1 ha un- e hical beha io is mo e con agious han e hical beha io . We ind simila esul s o hose who de- cided o keep he ai equal spli in he beginning, while he beha io al change o hose who dona ed ini ially seem no o be suscep ible o changes in social iden i y. The igu e below illus a es esul s. A6: The uppe (lowe ) g aph depic s he amoun e ised (%) (adap a ion gap (%)) condi ional on ini ial beha io b oken down by social p oximi y. This igu e depic s he pe cen age o ini ial beha io e ised as a unc ion o he pa icipan ’s ini ial beha io and b oken down in o di e en social iden i y ca ego ies. Essen ially, he igu e illus a es he magni- ude and di ec ion o beha io al con agion a e obse ing he passi e pee condi ional on one’s ini ial decision p io o ha ing obse ed a pee . 148 B: Social Iden i y S a emen s 1. I am a eliable pe son. 2. I am in e es ed in poli ics and/o economics. 3. Money is impo an o me. 4. I am an hones and since e pe son. 5. I am a cinephile. 6. I am in e es ed in spo s. 7. I am a eligious pe son and ai h is impo an o me. 8. I am ond o animals. 9. I am in e es ed in a and/o cul u es. 10. I am an ac i e and ad en u ous pe son. 11. I am in e es ed in ca s and/o echnology. 12. I am ond o child en and amily-o ien ed. 13. I am in e es ed in o eign languages and/o coun ies. 14. I am a wa mhea ed and help ul pe son. 15. I am a ole an pe son. 16. I like o gossip. 17. I am a ai h ul pe son. 18. I play an ins umen . 19. I like o go ou and dance. 20. I am a goal-o ien ed pe son. 21. I spend a lo o ime in on o he TV. 22. I am a sociable pe son and like o be among people. 23. I like o play ideogames. 24. I am a humo ous and en e aining pe son. 25. I am a s ong-willed pe son. A e age amoun o chosen s a emen s (ac oss all ea men s): 15.8 (63%) 149 C: Expe imen al Ins uc ions Gene al In o ma ion on he Expe imen  Fi s o all, we would like o hank you e y much o pa icipa ing in his expe imen . Please ead he ins uc ions ca e ully. The expe imen will las o abou 45-60 minu es.  Du ing he en i e expe imen , no communica ion is allowed. I he e is some hing you do no unde s and o i you ha e any ques ions, now o a some poin du ing he ex- pe imen , please aise you hand and emain sea ed. One o ou colleagues will come o you and answe you ques ion.  Du ing he expe imen , you ha e he possibili y o ea n money. The amoun you will ecei e a he end o he session depends on how many “Tale ” you ea n du ing he expe imen .  A he end o he expe imen , he amoun o “Tale “ ha you ha e ea ned will be con- e ed in o eal money a an exchange a e o 20 Tale = 1 Eu o.  All decisions you make du ing his expe imen will emain anonymous. None o he pa icipan s ge s o know he iden i y o o he pa icipan s in he expe imen and de- cisions canno be linked o a speci ic pa icipan . Mo eo e , you will be paid anony- mously a he end o he expe imen . 150 O de o E en s:  The expe imen consis s o a lis o s a emen s ha you will ecei e a he beginning and u he decisions. Explana ions and in o ma ion ela ed o hese decisions will be gi en as he expe imen p og esses. You will make hese decisions once.  Bo h you as well as a cha i able o ganiza ion o you choice (i.e. an o icially egis- e ed cha i y o ganiza ion) will be p o isionally assigned a mone a y amoun o 300 Tale each.  Du ing he expe imen you will ha e o decided on whe he you wan o…  … ake a pa o all o he money om he cha i able o ganiza ion.  … lea e he di ision o he sum o money as i is.  … gi e a pa o all o you money o he cha i able o ganiza ion.  In case you decide o ake money om he cha i able o ganiza ion, he espec i e amoun o money will be ans e ed o you indi idual cash accoun and exac ly he same amoun will be deduc ed om he cash accoun o he cha i able o ganiza ion.  Should you decide o gi e money o he cha i able o ganiza ion o you choice, he espec i e amoun o money will be deduc ed om you indi idual cash accoun and gi en o he cha i y. The expe imen e will double all ECUs emaining in he cha i y’s accoun a he end o he expe imen .  You decision emains anonymous and nei he he o he pa icipan s o he expe i- men no he expe imen e s ha e he possibili y o assign you choices o you iden- i y.  A he end o he expe imen , one pa icipan will be chosen a andom and his o he choice will be implemen ed and coun owa ds he cha i y (i.e. ha choice will be el- 151 e an o he paymen ). In pa icula , we will double he espec i e amoun and do- na e i o he cha i y a e he expe imen ends. The eceip o his dona ion will be published on he homepage o he BaER-Lab (www.bae -lab.o g) in a imely manne . All o he pa icipan s will ecei e 150 Tale (including he show-up ee) a he end o he expe imen .  The o al payo o he pa icipan s:  In case you a e he andomly chosen pa icipan 300 Tale +/- he amoun o money ha has been gi en o/ aken om he cash accoun o he cha i able o ganiza ion  In case you a e no he andomly chosen pa icipan 150 Tale  The o al payo o he cha i able o ganiza ion:  (Amoun o money in he cash accoun o he cha i able o ganiza ion o he an- domly chosen pa icipan ) × 2  A he end o he expe imen , he ele an in o ma ion on he paymen will be made isible o each pa icipan on his o he sc een.  A e he ac ual expe imen concludes, we will ask you o ill ou a ques ionnai e. Please ill ou he ques ionnai e ca e ully and u h ully. 152 D: Sc eensho s o Decision Sc eens 1. Lis o s a emen s: gene a es he p oximi y measu e in all ea men s 153 2. Fi s decision: beha ing (un)e hically (Exempla ily o he aking away decision)