Board gender diversity and dividend payout: The critical mass and the family ties effect

In e na ional Re iew o Financial Analysis 79 (2022) 101973 A ailable online 11 No embe 2021 1057-5219/© 2021 The Au ho s. Published by Else ie Inc. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). Boa d gende di e si y and di idend payou : The c i ical mass and he amily ies e ec ☆ Emma Ga cía-Meca a , F´ elix J. L´ opez-I u iaga c , d , * , Domingo Ja ie San ana-Ma ín b a Technical Uni e si y o Ca agena, School o Business, 30201 Ca agena, Mu cia, Spain b Uni e si y o Las Palmas de G an Cana ia, School o Business & Economics, 35017 Las Palmas de G an Cana ia, Spain c Uni e si y o Valladolid, School o Business & Economics, 47011 Valladolid, Spain d In angible-d i en Economy Labo a o y, Na ional Resea ch Uni e si y Highe School o Economics, Pe m, Russia ARTICLE INFO JEL codes: G35 G34 Keywo ds: Boa ds o di ec o s Di idend payou Family i ms Female di ec o s Family ies ABSTRACT We analyse he ela ionship be ween emale di ec o s and payou policy o a sample o non- inancial Spanish lis ed i ms. Based on he c i ical mass heo y, we ind an in e ed-U shaped ela ionship. Fo low le els o emale ep esen a ion in he boa d, women di ec o s inc ease di idends in o de o educe agency con lic s, and imp o e epu a ion o legi imacy. Howe e , a e an in lec ion poin , cha ac e is ics adi ionally associa ed o women, such as isk a e sion, a conse a i e and p uden inancial a i ude, and lowe o e con idence eme ge and educe di idend paymen s. Mo eo e , ou esul s sugges ha emale di ec o s play a e y di e en ole wi h he con olling sha eholde , depending on wha amily ies exis . Women di ec o s who ha e amily connec ions wi h he dominan sha eholde exhibi he same in e ed-U shaped ela ionship wi h di idends. In con as , o emale di ec o s wi h no amily ies, he ela ionship wi h di idends is U-shaped. Ou esul s show he aul lines wi hin he g oup o emale di ec o s depending on he ela ionship wi h he amily owne s, and ha he in luence o non- amily emale di ec o s only a ises when his g oup o women gain enough powe , isibili y, au ho i y, and legi imacy. 1. In oduc ion Recen esea ch has expanded ou knowledge on wha ole women play in co po a e inance (Bui, Nguyen, Pham, & Phung, 2019; Nguyen, N im, & Malagila, 2020). Some li e a u e sugges s ha boa d gende di e si y inc eases inno a ion, s akeholde empa hy, e hical beha iou and c ea i i y (Chen, Leung, & E ans, 2018; Liu, Lei, & Bu ne , 2020; Zala a, Tau ingana, & Tingbani, 2018). Howe e , he esea ch is con- lic ing, wi h some pape s a guing ha gende s e eo ypes ela ed o women’s isk a e sion and less o e con idence can lead o poo e emale inancial decisions, while o he esea che s no e he lack o any di e - ence be ween emale and male beha iou when women each high manage ial posi ions (Ga cía La a, Ga cía Osma, Mo a, & Scapin, 2017; Sila, Gonzalez, & Hagendo , 2016; Wellalage, Fe nandez, & Th ika- wala, 2020). Despi e he ele ance o di idends, he esea ch on wha consequences gende di e si y has on payou policy is qui e scan , em- b yonic, and o some ex en , con lic ing. Many s udies suppo a posi i e ela ionship be ween he p opo ion o emale di ec o s and di idend payou , bo h a he na ional le el and in wo ldwide analyses. The e is e idence o his posi i e ela ionship o he US ma ke (Benjamin & Biswas, 2019; Byoun, Chang, & Kim, 2016; Chen, Leung, & Goe gen, 2017) as well as o coun ies such as Malaysia (Tahi , Rahman, & Mas i, 2020), Nige ia (Id is, Ishak, & Hassan, 2019), Sy ia (Al-Rahahleh, 2017), China (Wellalagea, Fauzi, Wang, & Basyi h, 2014), and Spain (Puche a-Ma ínez & Bel-Oms, 2015). Recen ly, Ye, Deng, Liu, Szewc- zyk, and Chen (2019) also ound a posi i e in luence o gende di e si y on di idend paymen s in an in e na ional sample o 22 coun ies. In con as , Sanan (2019), Saeed and Samee (2017), and Mus a a, Saeed, Awais, and Aziz (2020) epo a nega i e ela ionship in India, China, Russia, and some o he Asian coun ies. O he pape s obse e li le o no di e ences in he di idend dis ibu ions o emale- and male-led ☆ We hank wo anonymous e e ees, B ian Lucey (edi o ) and Philip Jaggs o hei commen s on a p e ious e sion o he pape , and Pablo Cob e os o his help in da a collec ion. We also hank he Spanish Minis y o Economics, Indus y and Compe i i eness (ECO2017-82259-R and ECO2017-84132-R) and he Spanish Minis y o Science and Inno a ion (PID2020-114797GB-I00) o inancial suppo . All he emaining e o s a e solely he au ho s’ esponsibili y. * Co esponding au ho a : Uni e si y o Valladolid, School o Business & Economics, 47011 Valladolid, Spain. E-mail add esses: [email p o ec ed] (E. Ga cía-Meca), [email p o ec ed] (F.J. L´ opez-I u iaga), [email p o ec ed] (D.J. San ana-Ma ín). Con en s lis s a ailable a ScienceDi ec In e na ional Re iew o Financial Analysis jou nal homepage: www.else ie .com/loca e/i a h ps://doi.o g/10.1016/j.i a.2021.101973 Recei ed 17 Feb ua y 2021; Recei ed in e ised o m 14 Augus 2021; Accep ed 5 No embe 2021 In e na ional Re iew o Financial Analysis 79 (2022) 101973 2 boa d ooms (Eluyela e al., 2019; McGuinness, Lam, & Viei o, 2015). Fu he mo e, Gyapong, Ahmed, N im, and Nadeem (2019) and Benjamin and Biswas (2019) show ha his associa ion depends on owne ship concen a ion and CEO duali y. This inconclusi e e idence sugges s ha he ela ionship be ween emale di ec o s and di idend policy is mo e complex han ini ially hough and ha i may di e signi ican ly ac oss i ms. Rela ed o his poin , he c i ical mass heo y sugges s ha adding one woman o he boa d oom makes e y li le di e ence, and ha only when eaching a high enough numbe (o p opo ion) o women on he boa d can emale di ec o s c ea e a c i ical mass ha is able o exe a subs an ial in luence on boa d discussions (Joecks, Pull, & Ve e , 2013; Liu, Wei, & Xie, 2014; Shahab, N im, Ullah, Yugang, & Ye, 2020). Adding one woman o he boa d migh , as sugges ed by Adams and Fe ei a (2009), simply be conside ed a i m policy aimed a enhancing epu a ion, wi h emale di ec o s ac ing as me e okens. The e o e, he ela ionship be ween di idends and emale ep esen a ion could de ia e om a linea one, and mo e complex ela ionships migh explain he he e ogenei y in e- sul s and help o iden i y he poin a which women p o e in luen ial. The lack o conclusi e e idence ega ding boa d gende di e si y and di idend policy also a ec s amily i ms. In hese companies, he analysis is e en mo e complex since women should no be analysed as a single homogeneous subg oup. The e may be aul lines among amily and non- amily women di ec o s because emale di ec o s wi h amily ies sha e a common cul u e, common alues and pa e n o educa ion (He dhayin a, Lau, & Shen, 2021; Minichilli, Co be a, & MacMillan, 2010; Vel op, He mes, Pos ma, & de Haan, 2015). Family emale di- ec o s also ha e a s ong emo ional a achmen o he i m ha im- p o es hei le el o commi men and in ol emen in he i m, which can a ec hei a i ude o di idend payou because o hei own pa icula incen i es and in e es s. In his con ex , we add ess wo main esea ch ques ions: (1) Does he in luence o women di ec o s di e a di e en le els o emale p es- ence? (2) Is he e a di e en in luence o women di ec o s condi ional on he exis ence o amily ies wi h he dominan owne ? Using a da ase o 131 non- inancial Spanish lis ed i ms be ween 2003 and 2017 we aim o ill hese esea ch gaps. Spain p o ides a unique se ing o s udy hese ques ions as he e has been a d ama ic inc ease in he p esence o women on boa ds in ecen yea s, since he la es Spanish co po a e go e nance code ecommended a emale boa d gende quo a o a leas 30% (Cabeza-Ga cía, Fe n´ andez-Gago, & Nie o, 2018; CNMV, 2015). In addi ion, mos lis ed i ms ha e a amily dominan owne . In u n, Spanish lis ed i ms p o ide a unique ield o s udy by combining public in e es in inc easing emale ep esen a ion on boa ds wi h concen a ed co po a e owne ship and a high numbe o amily i ms. Ou esul s con i m he complexi y o he link be ween di idend policy and boa d gende di e si y: when he numbe o emale di ec o s is sca ce he e is a posi i e ela ionship. Ne e heless, his in luence u ns nega i e when he numbe o women eaches a c i ical mass. This inding challenges o he s udies ha suppo a non-di e en beha iou o men and women in high leade ship posi ions (Adams & Funk, 2012; Sila e al., 2016) and sugges s ha gende s e eo ypes o supposedly speci ic emale cha ac e is ics (e.g., mo e isk a e sion, conse a i e inancial a i ude, and na u al inclina ion o e ain excess cash) only a ise when he signi icance o he g oup is high enough. The indings also con i m ha he e ec i eness o boa d gende di e si y is condi- ional on le els o emale ep esen a ion. Thus, di idend payou and emale di e si y can be subs i u i e epu a ion p o ide s and moni o ing ools o educing agency cos s in he i m, con i ming he esul s o La Po a, L´ opez de Silanes, Shlei e , and Vishny (2000) and Smi h, Pen- na hu , and Ma ciniak (2017) ega ding he wo old ela ionship be- ween co po a e go e nance and di idend policy. Ou esul s also lend suppo o a di e en in luence be ween amily and non- amily emale di ec o s on di idend payou s, wi h an in e ed- U shaped and a U-shaped associa ion, espec i ely. We show he exis- ence o amily aul lines among women di ec o s and obse e ha hese aul lines do impac amily i m di idend policy. Thus, when he p es- ence o emale amily di ec o s is low, hese di ec o s a e less e ec i e, and di idends become mo e necessa y in o de o educe agency con- lic s be ween he dominan amily and ex e nal in es o s. Howe e , we no e ha amily women di ec o s educe di idend payou s when hei numbe on he boa d exceeds a c i ical mass h eshold. A isk a e se and conse a i e a i ude, oge he wi h he in e ac ion o inancial and socioemo ional aims, hus seems o be p e alen o amily women di- ec o s. Mo eo e , we also sugges ha a e a c i ical mass o emales wi h amily ies, he e is a “halo e ec ” ha can be compa ible wi h some deg ee o amily en enchmen . A his poin , amily emale di- ec o s can ypically pu sue amily objec i es such as long- e m su i al and amily epu a ion h ough isk a oidance and can educe di idends as a way o e ain mo e cash unde amily con ol and ensu e liquidi y in he e en o possible u u e h ea s o oppo uni ies. Howe e , non- amily emale di ec o s ha e he opposi e ela ionship wi h di idends, wi h an ini ially nega i e associa ion ha u ns posi i e a e a gi en poin . This esul indica es ha when non- amily women di ec o s do no each a c i ical mass, hey a e used by dominan amily owne s as okens ha allow hem o ob ain epu a ion and educe di - idends. Howe e , when non- amily emale di ec o s each a c i ical mass, hey gain powe , au ho i y and legi imacy, which may lead hem o make decisions ha educe agency con lic s and disgo ge i m e- sou ces om amily con ol. The esul s can be unde s ood as a con i - ma ion o he c i ical mass heo y, such ha he independen ole o hese emale di ec o s can only be de eloped once a high enough p o- po ion o non- amily ied women is achie ed. The e o e, o low le els o emale ep esen a ion he esul s sugges ha appoin ing non- amily women can be used as a oken o signal socially esponsible beha iou while educing di idends. Ou esea ch o e s wo con ibu ions o he exis ing li e a u e. Fi s , gi en ha mos p e ious s udies a e based on he agency heo y, di i- dend policy is supposed o be bo h a mechanism o alle ia ing con lic s inside i ms as well as a signal o epu a ion. We ex end he heo e ical ounda ions o he ela ionship be ween emale ep esen a ion and di idends wi h he c i ical-mass a gumen s by es ing o a non-linea ela ionship. This de ia es om exis ing s udies ha suppo a linea posi i e o a nega i e associa ion. We hus con ibu e o his li e a u e by showing ha he in luence o emale di ec o s can a y depending on he le el o boa d di e si y and on hei s eng h in he boa d oom. Second, mos p e ious li e a u e on gende di e si y is based on widely held co po a ions, excluding he companies cha ac e ized by high owne ship concen a ion and a s ong amily p esence. Ou esul s con i m a di e en pa e n o he link be ween di idends and emale di ec o s depending on he exis ence o amily ies, showing no only he di e en incen i es o each o hese wo g oups, bu also he di e en impac on di idend policies depending on he g oup’s isibili y and powe . 2. Backg ound and hypo heses de elopmen Acco ding o he agency heo y, in dispe se owne ship con ex s, di idends ac as a mechanism o mi iga e agency con lic s be ween manage s and sha eholde s. By paying di idends, manage s e u n co po a e ea nings o sha eholde s and educe ee cash lows which migh be used o ex ac p i a e bene i s. Di idends enable capi al ma ke discipline and o ce manage s o come o he inancial ma ke s in o de o aise unds om ex e nal in es o s. Simila ly, in coun ies whe e he e is concen a ed owne ship, poo in es o p o ec ion, and weak go e nance, di idends can also be a ool o mi iga e agency p oblems be ween la ge and mino i y sha eholde s. Di idends could help he dominan owne o es ablish a good epu a ion o aise ex e nal unds in he u u e, such ha di idends a e a subs i u e o legal p o ec ion. P e ious esea ch has p o ided heo e ical a ionale and empi ical suppo bo h o a posi i e and a nega i e ela ionship be ween emale E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 3 di ec o s and di idend payou . F om an agency heo y app oach, he boa d o di ec o s can a ec he i m’s payou policy h oughou he moni o ing ole (Fa inha, 2003; Hu & Kuma , 2004). In his amewo k, emale di ec o s can impac boa d pe o mance by educing agency cos s and by o e ing g ea e p o ec ion o mino i y sha eholde s h ough oughe o e sigh ( he so-called wa chdog ole). Adams and Fe ei a (2009) suppo his moni o ing ole by no ing ha emale di- ec o s ha e highe a endance eco ds and a e mo e likely o belong o moni o ing commi ees han hei male coun e pa s. In addi ion, inc easing gende di e si y in boa ds o di ec o s can imp o e boa ds’ abili y o exe cise hei con ol and s a egic oles since i educes he likelihood o ubbe -s amping managemen decisions, expands he pool o knowledge used o make g oup decisions, and accele a es inno a ion and c ea i i y (Basco & Voo decke s, 2015; Tuggle, Schna e ly, & Johnson, 2010). In line wi h he abo e a gumen s, Puche a-Ma ínez and Bel-Oms (2015), McGuinness e al. (2015), and Byoun e al. (2016) show ha he impac o boa d di e si y on di idend policy is s onge o i ms wi h po en ially g ea e agency p oblems o ee cash low, sug- ges ing ha a di e se boa d mi iga es he con lic s be ween co po a e inside s and ex e nal in es o s. The gende socializa ion heo y, based on sociological, cogni i e, and psychological pe spec i es, also highligh s he inna e di e ences be ween men and women ha can explain he di e en beha iou o emale di ec o s. Al hough hese cha ac e is ics a e no uni e sal in indi iduals, women a e o en conside ed o be mo e sympa he ic, mo e ecep i e, mo e ca ing and mo e coope a i e (Kim, Roden, & Cox, 2013). The emale s yle o leade ship is mo e in e ac i e and pa ici- pa i e han he male one, which can imp o e a boa d’s abili y o deal wi h ambigui y and unce ain y (Be inelli, Del Bosco, & Giachino, 2019). Female cha ac e is ics associa ed wi h hei bene olence, uni- e salism, inclina ion o comply wi h ules and laws, e hical beha iou o empa hy make hei leade ship mo e social and mo e s akeholde o ien ed (Malik, Nowland, & Buckby, 2021; Sun, Du a, Zhu, & Ren, 2021). Since hese cha ac e is ics migh make women di ec o s pay mo e a en ion o s akeholde demands, one would expec gende di e se boa ds o be mo e sensi i e o he di idend paymen eques s o mino i y sha eholde s. In sum, as women di ec o s a e conside ed by he abo e pe spec i es and heo ies o be oughe moni o s, p o ide s o legi imacy, signalling ools, as well as mo e sensi i e o all s akeholde s’ needs, he e a e a ionales o expec a non-negligible emale in luence on di idends, sugges ing a posi i e e ec o boa d gende di e si y on he payou a io. Ne e heless, hese same app oaches can also explain a nega i e ela ionship be ween emale ep esen a ion and di idend policy. I , ac- co ding o he agency heo y, emale di ec o s a e oughe moni o s and d i e s o co po a e go e nance e ec i eness, hen women who si on boa ds migh ac as a subs i u i e mechanism, an al e na i e o di idend payou . Thus, di idends could be used less as a moni o ing ool gi en he be e p o ec ion o all sha eholde s’ in e es s o e ed by emale di- ec o s (La Po a e al., 2000). Acco dingly, manage s and dominan owne s can incen i ize socially esponsible ac ions, in o he wo ds, appoin ing women di ec o s, and can c ea e a “halo e ec ” ha p o- mo es hei own epu a ion, inc eases hei en enchmen and de ia es a en ion away om agency con lic s wi h o he ex e nal o in e nal agen s, such as mino i y sha eholde s (Ba nea & Rubin, 2010; Bo ghesi, Hous on, & Na anjo, 2014). In u n, gende di e si y can be also a subs i u i e ool o epu a ion and di idend payou . The gende socializa ion heo y also sugges s a nega i e in luence o emale di ec o s on di idend paymen s based on he di e ences in isk a e sion be ween men and women. Women a e gene ally conside ed mo e isk a e se and conse a i e in hei inancial decisions han men (Faccio, Ma chica, & Mu a, 2016; Pal ia, V¨ ah¨ amaa, & V¨ ah¨ amaa, 2015, 2020). P e ious esea ch shows ha women end o in es less in high- isk asse s and ollow less ex eme in es men s a egies (Niessen & Ruenzi, 2007). This g ea e emale isk a e sion is con i med by s udies showing he posi i e ela ionship be ween emale in ol emen and i m cash holding (Adhika i, 2018; Zeng & Wang, 2015). Acco ding o his iew, emale-led i ms could be mo e likely o e ain cash holdings in o de o ace u u e unce ain y, and he e o e educe di idend pay- men s. Saeed and Samee (2017) show a nega i e ela ionship be ween boa d gende di e si y and di idend paymen s in eme ging coun ies, sugges ing ha women b ing new and be e in es men oppo uni ies o i ms and ha hey end o p ese e in e nal unds o in es men s. The abo e discussion sugges s ha he ela ionship be ween emale di ec o s and di idend policy is mo e complex han was a i s hough , wi h di e en a gumen s suppo ing bo h a posi i e and a nega i e in- luence. A his poin , we posi ha his associa ion migh also depend on he numbe o women di ec o s on boa ds, a possibili y no p e iously s udied. Indeed, he mixed indings in gende di e si y li e a u e and i m pe o mance could be explained by a ailu e o conside he c i ical mass heo y (Amo elli & Ga cía-S´ anchez, 2020; Joecks e al., 2013). In ligh o his heo y, he in luence o emale ep esen a ion is no linea bu condi ional on he le el o such ep esen a ion. A non-linea ela- ionship migh a oid he inconsis encies ound in p e ious esea ch and help o iden i y he poin a which emale in ol emen p o es op imal o he i m. Following his app oach, only when he e is a high enough numbe o women on he boa d can emale di ec o s c ea e a c i ical mass ha is able o signi ican ly in luence boa d discussions. Since boa ds ha e adi ionally been a homogeneous g oup o (male) di ec o s wi h simila a ibu es and iews, adding one woman would no make a di e ence because a small numbe o women in a male domina ed boa d migh be ine ec i e, wi h i being necessa y o ha e a c i ical mass o women. The c i ical mass heo y has been empi ically es ed by Joecks e al. (2013), who ound a U-shaped ela ionship be ween gende di e si y and i m pe o mance, such ha emale di ec o s a i s nega i ely a ec i m pe o mance, and a e associa ed wi h supe io i m pe o - mance only a e eaching a gi en h eshold. Consis en wi h his heo e ical amewo k, we posi ha emale di ec o s’ incen i es and beha iou may a y, depending on he le el o boa d gende di e si y, such ha he ela ionship be ween gende di e si y and di idends may be non-linea . Thus, when he le el o boa d gende di e si y is low, women may ac as okens, ha enjoy lowe s a us, p es ige and in luence compa ed o membe s o he nume ical majo i y. Recen esea ch sug- ges s ha women’s abili y o implemen hei s a egic i m ision may be limi ed by hei okenism ole wi hin a co po a e hie a chy (Glass & Cook, 2016) and ha hei abili y o d i e co po a e p ac ices is limi ed no only by hei s uc u al posi ion bu also by hei low-s a us nu- me ical mino i ies (Inge soll, Glass, Cook, & Olsen, 2019). The dis- inguishing emale cogni i e and socio-psychological cha ac e is ics ha make hei decisions di e en om men may ha e ouble su acing when he e a e ew women on he boa d. Consequen ly, below a c i ical h eshold, ha ing a mino i y o women on a boa d means ha hey ha e o deal wi h masculine no ms and male-domina ed decision making on a day- o-day basis, which can educe o e en wipe ou he impac o gende di e ences in inancial policies. Acco ding o p e ious li e a u e on c i ical mass, hese unde - ep esen ed women become “ou -g oup” boa d membe s who usually keep a low p o ile and a oid sha ing hei iews and ideas (Saggese, Sa o, & Vigan` o, 2021). The e o e, a his poin emale di ec o s can ac as me e okens and ha e no in luence o modi y accep ed beha iou o decisions because o hei sca ce isibili y, powe , au ho i y and legi i- macy. Mo eo e , women di ec o s as a mino i y g oup may no ollow expec ed gende beha iou bu , in con as , eplica e male pa e ns in o de o be accep ed in a male-domina ed cul u e. De ks, Van Laa , and Elleme s (2016) a gue ha women in i ms in which mos execu i e posi ions a e held by men may imi a e a he han change he exis ing gende hie a chy ( he queen bee phenomenon) and, a he han adding di e si y, may adjus hei sel -p esen a ion o i he masculine o ga- niza ion cul u e. In e ms o di idend policy, when he numbe o emale di ec o s is below his c i ical mass, emale di ec o s may eplica e hei masculine-coun e pa s’ iew. E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 4 Howe e , a e a gi en h eshold o gende di e si y, he in luence o women di ec o s can be mo e e ec i e. Risk a e sion, a conse a i e and p uden inancial a i ude and lowe emale o e con idence only eme ge and impac he i m’s di idend policy when emale oices can be hea d (which occu s abo e he c i ical mass poin ). A his poin , a high enough p opo ion o women on boa ds could ac as a subs i u i e mechanism o di idends o educe agency con lic s. Thus, di idends could be used less as an e ec i e moni o ing ool gi en he be e p o- ec ion o sha eholde s’ in e es s o e ed by emale di ec o s. Addi- ionally, women di ec o s could be a subs i u i e company ool o epu a ion, dec easing di idend payou as long as he numbe o emale di ec o s inc eases. This di e en ia ed ela ionship be ween women and payou depending on hei isibili y and legi imacy in he boa d can explain he he e ogenei y and equi ocal esul s ound in p e ious li e a u e, which has analysed he ole o women wi hou examining he ele ance o c i ical mass (Mus a a e al., 2020; Sanan, 2019; Wellalagea e al., 2014). The e o e, we p opose he ollowing hypo hesis: H1. The e is a non-linea ela ionship be ween emale di ec o s and i m payou . Toge he wi h all o he a o emen ioned a gumen s, ecen li e a u e sugges s ha women a e no a homogeneous g oup and ha he di e - ences among hem in e ms o amily ies mus be conside ed. This en- ails he need o explo e ce ain social-con ex ual ac o s ha migh o e a g ea e unde s anding o he mechanisms ha accoun o emale leade ship in i ms (Chadwick & Dawson, 2018; Hooble , Mas e son, Nkomo, & Michel, 2018; Pole i-Hughes & B iano-Tu en , 2019). One such con ex ual ame is co po a e owne ship and, mo e speci ically, he amily na u e o he i m. Family i ms p o ide a pa icula ly in e es ing ield o examining wha ole women play in di idend policy, since di idends in amily i ms a e conside ed a ele an mechanism o o e come agency con lic s be ween amily con olling and ou side sha eholde s (Pindado, Requejo, & To e, C. d. l., 2012). The e is also a g ea e emale p esence in amily i ms han in non- amily i ms, as shown by E ns and Young (2015), who no e ha women a e ep e- sen ed in 55% o amily i ms, and ha 70% o amily businesses a e conside ing a woman o hei nex CEO. Despi e his a ou able en i onmen o women in amily i ms and hei apid inco po a ion in o leade ship oles, p e ious li e a u e e- po s ha women wi h amily ies ha e o en played a less ac i e ole (Danes & Olson, 2003; Ma inez Jimenez, 2009). Resea ch also shows ha hese women usually ha e limi ed leade ship oles and less isibili y due o hei supposedly mo e emo ional beha iou (Chadwick & Daw- son, 2018). The ew s udies ha ha e examined he impac o emale di ec o s wi h amily ies o en igno e ha hei incen i es a e no he same as hose o o he independen di ec o s. He e, he socioemo ional app oach p o ided by he socioemo ional weal h (SEW) heo y is essen ial is-` a- is unde s anding he e ec o amily ies on emale di- ec o s and hei speci ic incen i es and beha iou . Female di ec o s wi h amily ies can be cha ac e ized as a subg oup who sha e a common cul u e, business alues, and socioemo ional goals (C uz, G´ omez-Mejia, & Bece a, 2010; Sha ma, Ch isman, & Chua, 2003). When compa ed o non- amily coun e pa s, emale di ec o s in amily i ms engage in mo e non- inancial aspec s such as main aining amily con ol and pu suing long- e m objec i es, suppo ing ans- gene a ional succession, and amily epu a ion (Be one, C uz, & G´ omez-Mejía, 2012; Chua, Ch isman, Kelle manns, & Wu, 2011). Indeed, hei nomina ion may also be ela ed o speci ic easons such as p o iding job secu i y o amily membe s, p ese ing amily con ol o ensu ing amily succession (He dhayin a e al., 2021). As mos o hese emales ha e been b ough up o wo k and pa icipa e ac i ely in he i m since childhood, hey sha e common alues and ha e s ong in- cen i es o pass on hei business o u u e gene a ions, oge he wi h a long- e m in ol emen and in e es in p o ec ing he amily epu a ion and a oiding con lic s be ween he ela i es who wo k oge he in he i m. These a i udes can a ou a posi i e in luence o amily emale di ec o s on di idend payou s and ha e been empi ically p o ed by Isako and Weisskop (2015), who show ha he highe payou gi en by amily i ms bene i s all sha eholde s, bu is d i en by he amily income hypo hesis due o he conside able amoun o amily weal h in es ed in he i m. Howe e , we also posi ha he isk a e se, conse a i e and p uden inancial a i ude o amily women can eme ge and impac he i m’s di idend policy. Thus, we sugges ha a e a c i ical mass o emales who ha e amily ies, he e is a “halo e ec ” which may p o e compa ible wi h a ce ain deg ee o amily en enchmen . A his poin , amily emale di ec o s may ypically pu sue amily objec i es such as long- e m su i al and amily epu a ion h ough isk a oidance, and may educe di idends as a way o keep mo e cash unde amily con ol, p o ec ing SEW endowmen and ensu ing liquidi y in he e en o u u e possible h ea s o oppo uni ies. Family ies may lead o di e ging iewpoin s and in e es s among amily and non- amily emale di ec o s since socioemo ional goals may no be sha ed by he non- amily emale di ec o s, who a e mo e objec- i e and independen , and possess a common eeling o non-inclusion (He dhayin a e al., 2021; Vandebeek, Voo decke s, Lamb ech s, & Huyb ech s, 2016). Women wi hou amily ies a e gene ally appoin ed h ough a compe i i e selec ion p ocess, no based on nepo ism o amily quo as, which makes hem mo e e ec i e in e ms o alle ia ing agency p oblems and no as inclined o p io i ize SEW amily aims (Gonz´ alez, Guzm´ an, Pablo, & T ujillo, 2020). In u n, a g ea e p esence o non- amily emale di ec o s may aise he ocus on inancial a he han on socioemo ional goals, he eby inc easing he conce n abou he needs o all s akeholde s, including mino i y sha eholde s by aising di idends. This iew is suppo ed by He dhayin a e al. (2021), who ind ha non- amily emale di ec o s ha e a mo e posi i e e ec on i m alue and di idend payou han amily emale di ec o s. The c i ical mass heo y also applies o his issue, such ha he in- luence o non- amily emale di ec o s on payou decisions can depend on hei isibili y and powe in he boa d. Since he di e en skills and mo i a ions ha hese emale di ec o s b ing o he boa d can only be eached a e a ce ain h eshold, we hypo hesize a non-linea associa- ion be ween non- amily women and di idends. We sugges ha when he numbe o non- amily women di ec o s is low, hey may ind i di icul o play hei ole o independen moni o s o he con olling amily because hey equi e a c i ical mass in o de o wo k as e ec i e moni o s, and e en o legi imize hei p esence on he boa d. Acco ding o Vandebeek e al. (2016), women in amily i ms ace some obs acles ela ed o hei in isibili y and glass ceiling p oblems. In u n, when hese women ha e no amily ies hese obs acles a e e en g ea e , and i is only when hey o m a signi ican subg oup ha hey a e able o be e ec i e in luence s. In he con ex o social p essu e and high egula ion o inc easing boa d gende di e si y, adding one non- amily woman o he boa d migh , as sugges ed by Adams and Fe ei a (2009), simply be conside ed a i m policy aimed a enhancing epu a ion and legi imacy, wi h hese women independen di ec o s ac ing as me e okens and inding i di icul o play a ole no only as women bu also as non- amily mem- be s. Thei mino i y posi ion p e en s hem om making a subs an ial con ibu ion o co po a e decisions because hey belong o he boa d “ou -g oup” (Saggese e al., 2021). Howe e , once he in lec ion poin is achie ed, he in luence o non- amily emale di ec o s inc eases, and hese women may exe mo e p essu e o cu ail amily exp op ia ion con lic s, which means keeping esou ces ou o amily sha eholde con ol and ins ead se ing he whole o ganiza ion. Again, and ollowing he c i ical mass heo y, we sugges ha only when he g oup o non- amily emale di ec o s is big enough in he boa d, can hey gain he equi ed powe , isibili y, au ho i y, and legi imacy needed o educe agency con lic s and keep he i m’s esou ces away om he exclusi e con ol o he amily. We hus posi ha no only amily ies bu also boa d isibili y and powe may lead o di e ging iewpoin s and incen i es be ween amily and non- amily emale di ec o s. In line E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 5 wi h his, we posi he ollowing wo hypo heses: H2a. The e is a non-linea ela ionship be ween amily emale di- ec o s and i m payou . H2b. The e is a di e en beha iou be ween amily and non- amily emale di ec o s wi h espec o i m payou . 3. Empi ical design 3.1. Sample and a iables We examine a sample o 131 non- inancial Spanish lis ed i ms be- ween 2003 and 2017. Du ing ha pe iod he a e age numbe o non- inancial i ms lis ed in he Spanish capi al ma ke s was less han 150 i ms, such ha ou sample is highly ep esen a i e o he o al popu- la ion o Spanish lis ed i ms. The combina ion o c oss-sec ion da a wi h ime se ies p o ides an unbalanced sample o 1464 i m-yea obse a- ions, wi h 84.9% o he i ms ha ing six o mo e obse a ions du ing he pe iod 2003–2017. This sample accoun ed o 98.1% o Spanish ma ke capi aliza ion in 2017. Da a on inancial s a emen s, di idends and o he i m cha ac e is ics we e aken om he Spanish S ock Ex- change Commission (CNMV, Comisi´ on Nacional del Me cado de Valo es). We hand-collec ed in o ma ion abou he p esence o women in he boa d o di ec o s using a ious sou ces. We examine boa d composi ion using he Annual Co po a e Go e nance Repo s published by he Spanish Secu i y Exchange Commission. A e iden i ying a woman as a di ec o , we analysed he amily ela ionship wi h he dominan amily owne h ough amily names, om he i ms’ websi es and media, o by con ac ing he i ms. I we iden i ied a di ec amily ela ionship, o one acqui ed h ough ma iage, we classi ied he women as amily ied, and o he wise as non- amily ela ed. In ou sample, he con olling amily main ained company con ol o 90% o he i een yea s analysed, wi h 99% o hese i ms ha ing amily membe s on he boa d. Ou dependen a iable (PAYOUT) is he di idend payou a io, measu ed as he a io o o al di idends o ne income be o e ex ao - dina y i ems. Ou esul s a e obus o al e na i e measu es o di idend payou such as di idends o e o al asse s, di idends o sales, and whe he o no he i m pays di idends. Following he ex an li e a u e, boa d gende di e si y (WOMEN) is de ined as he numbe o women di ec o s on he boa d di ided by he o al numbe o boa d membe s (Byoun e al., 2016; Chen e al., 2017; Puche a-Ma ínez & Bel-Oms, 2015; Saeed & Samee , 2017; Sila e al., 2016). Da a on amily i ms come om Bona-S´ anchez, P´ e ez-Alem´ an, and San ana-Ma ín (2019), who iden i y he dominan o ul ima e owne o Spanish lis ed companies, applying he con ol chain me hod. Acco ding o he con ol chain me hod, a i m is conside ed as a amily i m when wo condi ions a e join ly me : i s , he p incipal owne mus hold, ei he di ec ly o indi ec ly, no less han 20% o o ing igh s, and second, his dominan owne mus be an indi idual o a amily. When iden i ying a woman as a di ec o , we analyse he amily ela ionship wi h he dominan amily owne h ough amily names, om he i ms’ websi es and media, o by di ec ly con ac ing he i m. I we iden i ied a amily ela ionship ei he by blood o as an in-law, we hen classi ied he woman as a amily membe , and as non- amily o he wise. 1 We hen de ine FAM_WD and NONFAM_WD as he num- be o amily and non- amily emale di ec o s on he boa d o amily i ms di ided by he o al numbe o boa d membe s. We con ol o a numbe o i m cha ac e is ics ha migh a ec di idend policy: i m size, p o i abili y, inancial le e age, boa d size, and g ow h oppo uni ies (Bilinski & Lyssimachou, 2018; B awn & ˇ Se i´ c, 2018; Ha akeh, Lee, & Walke , 2019). Fi m size (SIZE) is he log o o al asse s, p o i abili y (ROA) is measu ed h ough he e u n on asse s, i.e., he EBITDA o o al asse s a io, he size o he boa d (BOASIZE) is he log o he numbe o di ec o s, g ow h oppo uni ies (GROWTH) is he equi y ma ke - o-book a io, inancial le e age (LEV) is he a io o o al liabili ies o o al asse s. Gi en he ele ance o owne ship s uc u e on payou policy, we use he con ol chain me hod o calcula e o ing igh s (VOTING) in he hands o he ul ima e o dominan owne . Va iables we e winso ized a he 1s and 99 h pe - cen iles in o de o educe he impac o po en ial ou lie s, excep o he dummy a iables. All he eg essions include a se o dummy a iables o con ol o indus y ( ρ i ) and ime ixed e ec s ( μ ). Ou model is as ollows: PAYOUTi = α 1+β1WOMENi +β2VOTINGi +β3SIZEi +β4ROAi +β5GROWTHi + + β6LEVi +β7BOASIZEi + ρ i+ μ + ε i (1) 3.2. Me hod Since we ha e a panel da a se , we ini ially un wo panel da a echniques (A ellano & Bo e , 1995; Blundell & Bond, 1998): he ixed e ec s me hod and he gene alized me hod o momen s (GMM). The ixed e ec s me hod in oduces i m-le el ixed e ec s o co ec o he cons an unobse able he e ogenei y ha may be p esen in he gende di e si y-di idends ela ionship. In addi ion, he p esence o women di ec o s migh be endogenous since demands o g ea e gende di- e si y on he boa d may be a ec ed by claims o di idends, which aises a conce n abou possible endogenei y (Chen e al., 2017). The GMM p ocedu e, based on he panel da a me hodology, allows us o add ess unobse ed cons an he e ogenei y, po en ial endogenei y 2 conce ns, and e e se causali y (Pindado, Requejo, & de la To e, 2014). Al hough hese wo app oaches educe possible p oblems associa ed wi h endogenei y in he gende di e si y-payou ela ionship, we also add ess his issue using wo o he es ima ion models: an ins umen al a iable app oach and he p opensi y sco e ma ching p ocedu e. The p opensi y sco e ma ching p ocedu e is an al e na i e me hod o con ol o di e ences in he speci ic cha ac e is ics o he i ms wi h emale di ec o s and o add ess endogenei y igge ed by po en ial selec ion bias (Heckman, Ichimu a, & Todd, 1998; Ka pa iˇ cius & Yu, 2018; Nekhili, Naga i, Ch ioui, & Nekhili, 2017). This echnique equi es he selec ion o wo compa able sub-samples: a se o i ms wi hou emale di ec o s and a coun e pa se o i ms wi h women di ec o s. The candida e i m o ma ching should be om he same indus y and yea as he i m wi h women di ec o s. F om among he candida e i ms we hen selec he op imal ma ch (i.e., a i m wi hou emale di ec o s) based on he nea es neighbou echnique. To es ima e he p opensi y sco e, a comp ehensi e se o i m cha ac e is ics ha cap u e he like- lihood o a gi en i m ha ing women di ec o s should be employed. 4. Resul s 4.1. Desc ip i e analysis The p opo ion o i ms wi h a leas one emale di ec o is shown in Table 1 (Panel A). The e was an inc ease in gende di e si y du ing he pe iod 2003–2017: while only 27.71% o i ms had a leas one emale di ec o in 2003, by 2017 he pe cen age had isen o 80.58%. Howe e , he pe cen age o emale di ec o s did no ise a such a as pace bu inc eased s eadily om 12.28% o all di ec o s in 2003 o 20.43% in 2017. In line wi h Sac is ´ an-Na a o and G´ omez-Ans´ on (2007), a ound hal o he i ms in he sample a e amily i ms. Ou da a show a s eady 1 La in coun ies ha e wo ad an ages ha make i easy o iden i y amily ela ionships. Fi s , he e a e wo su names, he i s being he a he ’s and he second he mo he ’s. Second, ma ied women keep hei maiden names. 2 In line wi h Woold idge (2010), we b oadly de ine endogenei y bias as any si ua ion whe e he dis u bance e m o he s uc u al equa ion is co ela ed wi h one o mo e independen a iables. E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 6 ep esen a ion o emale amily di ec o s, a ound 5% o sea s, while he pe cen age o emale non- amily di ec o s inc eases om 6% o 15% be ween 2003 and 2017. These numbe s can be compa ed wi h hose o simila p io esea ch: Saggese e al. (2021) epo a mean (median) o 11% (9.1%) emale di ec o s in a sample o p i a e medium-sized and la ge I alian high- ech i ms du ing he pe iod 2012–2015. Gonz´ alez e al. (2020) epo a igu e o 17.5% emale di ec o s in Colombia, o whom 43.2% a e amily- emale di ec o s. Gyapong, Ahmed, N im, and Nadeem (2021) epo an a e age 8.9% in Aus alia be ween 2009 and 2014. T inh, Cao, Dinh, and Nguyen (2021) epo a mean (median) o 16.8% (16.7%) among FTSE 100 i ms be ween 2006 and 2011. Ye e al. (2019) epo an 8.3% a e age o an in e na ional sample om 22 coun ies om 2000 o 2013, and Chen e al. (2017) epo 10.3% among S&P 1500 cons i uen s be ween 1997 and 2011. Taken oge he , hese numbe s unde line he ele ance o emale di ec o s in he Spanish con ex . Panel A o Table 1 also epo s some desc ip i e s a is ics o he main a iables. Mean payou in he sample is 0.29 and, on a e age, women accoun o 16.15% o all di ec o s. I should be no ed ha he a e age log o boa d size is 2.3, which means an a e age boa d size o 9.97 di ec o s. Panel B displays he Pea son co ela ion ma ix among he a iables. The e is a signi ican (al hough low) nega i e co ela ion be ween PAYOUT and WOMEN, which migh sugges an ini ially nega i e ela- ionship be ween di idend policy and emale di ec o s. The absence o high co ela ions be ween he explana o y and he con ol a iables sugges s ha mul icollinea i y is no an issue ha migh bias ou esul s, which is con i med by he low alues o he ec o in la ion ac o (VIF) in he igh -hand side column. In panel C o Table 1, we epo he es o means compa ison be- ween i ms wi h gende di e se and non-di e se boa ds. Resul s indi- ca e he e a e no s a is ically signi ican di e ences in payou policy, p o i abili y, and le e age. The da a sugges ha i ms wi h gende di e se boa ds ha e la ge boa ds, a mo e concen a ed owne ship s uc u e, a e bigge in size, and ha e ewe in es men oppo uni ies. These mul i ace ed esul s sugges ha he ela ionship be ween gende di e si y and di idend policy seems o be mo e complex han ini ially an icipa ed, and calls o u he analysis. 4.2. Baseline esul s Model 1 ( ixed-e ec s) and model 2 (GMM) in Table 2 show he e- sul s o he basic ela ionship be ween gende di e si y and di idend policy. Bo h eg essions e idence an in e ed U-shaped associa ion. These esul s a e consis en wi h H1. The es ima es show ha , a low le els o women di ec o s, he e is a posi i e ela ionship wi h payou . This is consis en wi h using di idends o educe agency con lic s, in- c ease epu a ion o legi imacy, o ake iskie co po a e decisions. Howe e , he e is an in lec ion poin 3 a e which he nega i e e ec o he pe cen age o women di ec o s p e ails. A e his poin , a highe Table 1 Desc ip i e s a is ics and co ela ion ma ix. Panel A. Desc ip i e s a is ics (%) 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 A leas one woman (%) 27.71 27.38 31.46 46.39 49.54 55.05 58.49 64.08 66.00 63.27 65.59 68.13 75.51 80.20 80.58 WOMEN 12.28 13.67 13.35 14.06 14.44 14.34 14.61 14.70 14.93 16.40 17.21 17.90 17.40 18.03 20.43 Family Fi ms 42.17 42.86 46.07 48.45 53.21 53.21 53.77 54.37 54 52.04 52.69 52.75 51.02 51.49 51.46 FAM_WD 6.12 7.25 8.01 7.08 7.52 6.34 5.85 5.66 5.56 5.91 6.16 5.70 4.57 4.86 4.95 NONFAM_WD 6.15 6.41 5.33 6.98 6.91 8.00 8.57 8.87 9.20 10.30 10.86 12.17 12.88 13.19 15.40 Mean Median S d. De 1s Q 3 d Q 4 h Q PAYOUT 0.29 0.15 2.44 0.00 0.45 4.34 WOMEN 16.15 13.33 8.74 10.00 22.22 40.00 ROA 0.05 0.05 0.10 0.02 0.09 0.48 BOARDSIZE 2.30 2.30 0.34 2.07 2.56 2.99 VOTING 31.93 25.08 21.24 14.93 49.69 91.14 GROWTH 2.80 1.72 4.32 0.99 3.03 31.44 LEV 0.64 0.64 0.23 0.49 0.77 0.99 SIZE 13.86 13.68 1.95 12.39 15.12 18.42 Panel B. Co ela ion ma ix Payou Women_Di ec o s ROA Boa dsize Vo ing G ow h Le VIF WOMEN −0.04* 1.04 ROA 0.05** −0.01 1.54 BOARDSIZE 0.02 −0.03 0.12*** 1.87 VOTING −0.02 0.12*** 0.03 −0.17*** 1.08 GROWTH 0.02 −0.05** 0.26*** −0.03 0.07*** 1.42 LEV −0.04 0.01 −0.28*** 0.03 0.05** 0.03 1.21 SIZE 0.009 0.09*** 0.10*** 0.64*** 0.01 −0.02 0.17*** 1.93 Panel C. Gende di e se boa ds e sus non di e se boa ds Gende di e se boa ds (N =850) Non-gende di e se boa ds (N =614) T-Tes Mean Median S d. De . Mean Median S d. De . PAYOUT 0.31 0.13 2.85 0.27 0.17 1.73 0.28 ROA 0.05 0.05 0.09 0.05 0.05 0.11 0.31 BOARDSIZE 2.34 2.39 0.33 2.24 2.30 0.35 5.77*** VOTING 32.99 26.24 21.21 30.45 23.93 21.21 2.26** GROWTH 2.59 1.70 3.45 3.10 1.75 5.28 −2.22** LEV 0.64 0.64 0.23 0.64 0.63 0.22 −0.16 SIZE 14.15 14.07 2.02 13.47 13.23 1.78 6.67*** 3 The in lec ion poin is be ween 12% and 13% o emale di ec o s, no a om ou median alue. E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 7 p opo ion o emale di ec o s implies a dec ease in payou policy. Thus, a non-linea unc ion eme ges as a mo e p ecise ela ionship be ween di idends and emale di ec o s and may econcile p e iously con lic ing esul s. Fo ins ance, Puche a-Ma ínez and Bel-Oms (2015) epo a posi i e ela ionship be ween emale di ec o s and di idend payou in Spanish i ms. Ne e heless, in h ee qua e s o hei obse - a ions he p opo ion o women di ec o s is below 12.5%, i.e., mos o hei obse a ions a e in he posi i e-slope pa o ou non-linea unc- ion. Ye , con a y o ou esea ch, hese au ho s ail o es he ela- ionship o high le els o emale di ec o s. Simila ly, Byoun e al. (2016) epo a posi i e ela ionship o Ame ican i ms, al hough hei a e age a io o emale di ec o s is 8.88%, some way below ou u ning poin . Fu he mo e, he nega i e ela ionship ound by Saeed and Samee (2017), which had hus a ailed o ma ch he indings epo ed in p e ious esea ch, is consis en wi h he downwa ds-slope ound in ou esul s. As ega ds he con ol a iables, he i m’s pe o mance, size, and numbe o di ec o s ha e a posi i e ela ionship wi h di idends. These esul s mean ha di idend paye s a e la ge and mo e p o i able (B a , G aham, Ha ey, & Michaely, 2005). In con as , inancial le e age and g ow h oppo uni ies ha e a nega i e ela ionship wi h di idend policy. These la es esul s may be due o he inancial needs o he as es g owing i ms and o he subs i u e disciplina y ole o inancial le e age. We es he obus ness o ou esul s wi h wo o he es ima ion models: an ins umen al a iable app oach and h ough he p opensi y sco e ma ching p ocedu e. In Table 3, we employ he ins umen al a iable app oach o ob ain he exogenous elemen om gende di- e si y (Conyon & He, 2017). In line wi h Lang and Lockha (1990), Pea ce and Zah a (1992) and Waldman, Rami ez, House, and Pu anam (2001), we use UNCERTAINTY as an ins umen o gende di e si y since i ms eac o unce ain y by designing di e se boa ds, which en- hances adap a ion o a isky en i onmen . This a iable is based on he economic policy unce ain y index (BBD index) c ea ed by Bake , Bloom, and Da is (2016) and e ined o Spain by Ghi elli, P´ e ez, and U asun (2019). This index is a measu e o unce ain y ela ed o u u e economic policy and egula o y ou comes. Speci ically, he index is buil as a weigh ed mean o h ee componen s. The i s and mos ele an componen includes media e e ences o policy unce ain y. The second ac o measu es unce ain y abou u u e changes in ax code p o isions. The hi d componen is based on he disag eemen be ween he con- sume p ice index and go e nmen spending o p oxy unce ain y abou iscal and mone a y policy. Thus, we de ine UNCERTAINTY as he qua e ly a e age o he BBD index in yea . Panel A o Table 3 shows he esul s o he i s -s age eg ession, whe e he dependen a iable is he pe cen age o women (Model 3). Fo b e i y, we only epo he es ima es o he p incipal a iable. As can be seen, unce ain y posi- i ely a ec s he p esence o women on boa ds. Panel B o Table 3 shows he second-s age eg essions. The esul s co obo a e hose epo ed in Table 2: he e is an in e ed U-shaped ela ionship be ween he p o- po ion o emale di ec o s and di idend payou , wi h an in lec ion poin being eached a a ound 11% o women di ec o s. We now un he analysis h ough he p opensi y sco e ma ching p ocedu e. To es ima e he p opensi y sco e, we use a numbe o i m cha ac e is ics ha a e supposed o cap u e he likelihood o a gi en i m ha ing emale di ec o s. We use VOTING o in oduce he powe o he dominan owne o design boa d composi ion (Campopiano, De Massis, & Chi ico, 2014; Delmas & Ge gaud, 2014; Dou, Zhang, & Su, 2014), GROWTH o con ol o he possible ela ionship be ween co po a e pe o mance and emale di ec o s (Fa ell & He sch, 2005), SIZE due o he g ea e p opensi y o la ge i ms o hi e emale di ec o s (Pe e son & Philpo , 2007; Te jesen, Sealy, & Singh, 2009), and DEBT since Gillan & S a ks (2000) a gue ha ins i u ional in es o s may d i e Table 2 Fixed e ec s and GMM es ima es. Model 1 Model 2 Fixed e ec s GMM WOMEN 0.765*** (4.30) 0.192*** (4.65) WOMEN 2 −0.031** (−5.83) −0.007*** (−6.21) ROA 0.014* (1.94) 0.725*** (3.97) GROWTH −0.001 (−0.90) −0.001** (−2.41) LEV −0.09* (−1.94) −0.512*** (−3.87) SIZE 0.039 (0.65) 0.058** (2.45) VOTING 0.003* (1.99) 0.005 (0.90) BOARDSIZE 0.297* (1.95) 0.569*** (6.66) In e cep 1.530* (1.87) 3.025*** (4.40) m 2 −0.97 Z1 16.97*** Z2 27.65*** Z3 28.46*** Hansen es 87.07 F es 10.31*** 395.58*** U es 4.30*** 4.65*** Adj. R 2 0.37 No. o obse a ions 1464 1464 This able epo s he es ima ed coe icien s ( -s a is ics based on obus s anda d e o ). The dependen a iable is PAYOUT ( o al di idends o ne income). WOMEN is he p opo ion o emale di ec o s, ROA is e u n on asse s, GROWTH is g ow h oppo uni ies, LEV is inancial le e age, SIZE is i m size, VOTING is he p opo ion o o ing igh s o he ul ima e sha eholde , and BOARDSIZE is he size o he boa d. Model 1 includes indus y and ime dummy a iables, and Model 2 includes ime dummy a iables. Hansen is a es o o e -iden i ying es ic ions unde he null hypo hesis ha all ins umen s a e unco ela ed wi h he dis u bance p ocess. m 2 is he s a is ical es o he lack o second-o de se ial co ela ion in he i s -di e ence esidual. z 1 is he Wald es o he join signi icance o he epo ed coe icien s. z 2 is he Wald es o he join signi i- cance o ime dummies. z 3 is he Wald es o he join signi icance o indus y dummies. F is he es o he join signi icance o all he coe icien s. U is he es o he nonlinea e ec o emale di ec o s. * , ** , ***: s a is ically signi ican a 1, 5 and 10%, espec i ely. Table 3 Ins umen al a iable es ima es. Panel A. Fi s -s age eg essions Model 3 Dependen a iable WOMEN UNCERTAINTY 0.163*** (8.66) F es 10.05*** Adj. R 2 0.152 No. o obse a ions 1464 Panel B. Second-s age eg essions Dependen a iable PAYOUT WOMEN 0.196*** (2.60) WOMEN 2 −0.008*** (−2.41) ROA 0.180*** (2.80) GROWTH −0.002*** (11.9) LEV −0.009** (−2.36) SIZE 0.002*** (3.66) VOTING 0.001*** (3.35) BOARDSIZE 0.013*** (3.82) In e cep −0.004 (−0.55) F es 110.20*** U es 2.12** Adj. R 2 0.37 No. o obse a ions 1464 Es ima ed coe icien s ( -s a is ics based on obus s anda d e o ). The depen- den a iable is PAYOUT ( o al di idends o ne income). WOMEN is he p o- po ion o emale di ec o s, UNCERTAINTY is an index o economic policy unce ain y, ROA is e u n on asse s, GROWTH is g ow h oppo uni ies, LEV is inancial le e age, SIZE is i m size, VOTING is he p opo ion o o ing igh s o he ul ima e sha eholde and BOARDSIZE is he size o he boa d. All he e- g essions include indus y and ime dummy a iables. F is he es o he join signi icance o all he coe icien s. U is he es o he nonlinea e ec o emale di ec o s. * , ** , ***: s a is ically signi ican a 1, 5 and 10%, espec i ely. E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 8 i ms owa d g ea e di e si y. The ac ha he CEO can simul aneously play he ole o boa d chai man migh also in luence he composi ion and e ec i eness o he boa d (Bea , Rahman, & Pos , 2010). Thus, we also include he a iable PRESI_DUAL, which equals 1 when he same pe son has bo h jobs. In Table 4, we epo he esul s o he p opensi y sco e ma ching p ocedu e. As can be seen, esul s a e ully consis en wi h hose p e- sen ed in Tables 2 and 3: he e is a non-linea ela ionship be ween he p opo ion o emale di ec o s and di idend payou , wi h an ini ially posi i e e ec ha u ns nega i e a e an in lec ion poin (a ound 12%– 13% o emale di ec o s). Company p o i abili y inc eases di idends, and le e age plays a subs i u i e ole as a mechanism o manage ial discipline. In o de o es he ex en o which ou esul s migh be a ec ed by he measu e o payou , we conside some al e na i e me ics o ou dependen a iable (Table 5). We use di idends scaled by o al asse s (Model 5), di idends scaled by sales (Model 6), and we conside he likelihood o paying di idends in Model 7 (Table 5). Since he dependen a iable in Model 7 is he likelihood o paying di idends, in his column we use he ins umen al a iables p obi me hod. O e all, ou indings a e in line wi h hose p e iously epo ed and co obo a e he non-linea ela ionship be ween women in boa ds and payou policy, wi h an ini ially posi i e e ec ha u ns nega i e a e a gi en poin . 4.3. Female di ec o s in amily i ms We now analyse he speci ic e ec o amily ies. In u n, we ocus on amily i ms and classi y each emale di ec o depending on he links wi h he dominan owne amily. Since he gene a ional s age o amily i ms migh in luence di idend policy, we include he a iable GEN- ERATION as a measu e o iden i y which amily gene a ion con ols he i m (Luba kin, Schulze, Ling, & Dino, 2005; Michiels, Voo decke s, Lybae , & S eij e s, 2015). This a iable akes he alue o one o ou depending on whe he he amily i m is he i s , second, hi d o ou h gene a ion, espec i ely. Fu he mo e, as subs i u e measu es o he gene a ional e ec , we es ima e al e na i e models ha include wo con ol a iables: CEOage, de ined as he age o he CEO, and FAM_MD, de ined as he numbe o amily male di ec o s on he boa d o amily i ms di ided by he o al numbe o boa d membe s. In o de o es how amily and non- amily emale di ec o s a ec di idend policy, we es ima e Models 8 o 19 in Table 6. The columns di e in e ms o he dependen a iable: di idends o ne income (Models 8 o 11), di idends o o al asse s (Models 12 o 15), di idends o sales (models 16 o 19). The esul s sugges in e es ing di e ences compa ed o hose men ioned ea lie . When we ocus on emale di- ec o s who ha e amily ies wi h he con olling amily (Models 8, 9, 12, 13, 16, and 17), we ind an in e ed-U shaped ela ionship be ween women on boa ds and di idends. This migh mean ha , when amily emale di ec o s ca y li le weigh on he boa d, emale ep esen a ion inc eases di idend payou because o amily- ied women’s in e es and desi e o p o ec he amily image and epu a ion (G´ omez-Mejía, Hay- nes, Nú˜ nez-Nickel, Jacobson, & Moyano-Fuen es, 2007). Addi ionally, a hese low o mode a e le els o emale p esence, women di ec o s may ha e a g ea e abili y and mo e incen i es han hei non- amily coun e pa s o limi he oppo unis ic ac ions o manage s (Te jesen e al., 2009; Villalonga & Ami , 2006). Howe e , he p esence o amily women on he boa d beyond he op imal le el migh exace ba e o he agency con lic s since he boa d would lack he necessa y independence o ac as an e ec i e con ol mechanism and would he e o e p io i ize amily in e es s and incen i es o he de imen o hose o ou side in- es o s. Acco dingly, o high enough le els o emale p esence, hese ies wi h he con olling amily migh p e ail o e emale p e e ences, such ha he “halo e ec ” c ea ed by appoin ing a high numbe o ( amily) women di ec o s could be compa ible wi h a ce ain deg ee o Table 4 P opensi y sco e ma ching es ima es. Model 4 WOMEN 0.751*** (4.22) WOMEN 2 −0.030*** (−5.77) ROA 1.340* (1.83) GROWTH 0.03 (0.41) LEV −0.072* (−1.77) SIZE 0.005 (0.10) VOTING −0.004 (−1.08) BOARDSIZE 0.204 (0.77) In e cep 1.524* (1.88) F es 43.7*** U es 2.86*** Adj. R2 0.29 No. o obse a ions 1228 Es ima ed coe icien s ( -s a is ics based on obus s anda d e o ). The dependen a iable is PAYOUT ( o al di idends o ne in- come). WOMEN is he p opo ion o emale di ec o s, ROA is e- u n on asse s, GROWTH is g ow h oppo uni ies, LEV is inancial le e age, SIZE is i m size, VOTING is he p opo ion o o ing igh s o he ul ima e sha eholde and BOARDSIZE is he size o he boa d. All he eg essions include indus y and ime dummy a iables. F is he es o he join signi icance o all he co- e icien s. U is he es o he nonlinea e ec o emale di ec o s. * , ** , ***: s a is ically signi ican a 1, 5 and 10%, espec i ely. Table 5 Al e na i e dependen a iables. Dependen a iable: Di /A (GMM) Dependen a iable: Di /Sales (GMM) Likelihood o di idends (IV P obi ) Model 5 Model 6 Model 7 WOMEN 0.040** (2.44) 0.066*** (4.54) 0.239*** (5.39) WOMEN 2 −0.001*** (−2.61) −0.001*** (−3.22) −0.006*** (−4.45) ROA 0.096*** (10.12) 0.024*** (2.73) 0.128* (1.73) GROWTH −0.001*** (−11.35) −0.003*** (4.54) −0.011* (−1.74) LEV −0.004 (−1.09) −0.009 (−0.21) −0.032 (−0.81) SIZE 0.001* (1.64) 0.008*** (8.40) 0.019 (1.05) VOTING 0.001*** (2.68) 0.004*** (10.67) 0.003*** (2.79) BOARDSIZE 0.019*** (3.57) 0.013** (2.36) 0.416*** (3.71) In e cep −0.102*** (−2.93) −0.107*** (8.27) 0.297 (1.15) m 2 −0.44 0.91 Z1 44.48*** 41.15*** Z2 8.65*** 32.76*** Z3 11.25*** 38.04*** Hansen es 76.54 80.37 F es 1266*** 249.24*** U es 2.40** 2.48*** χ 2 2915.21*** Log likelihood −4840.91 Wald es exogenei y 32.52*** No. o obse a ions 1464 1464 1464 Es ima ed coe icien s ( -s a is ics based on obus s anda d e o ). The depen- den a iable is di idends o o al asse s (Model 5), di idends o o al sales (Model 6), and he likelihood o paying di idends (Model 7). WOMEN is he p opo ion o emale di ec o s, ROA is e u n on asse s, GROWTH is g ow h oppo uni ies, LEV is inancial le e age, SIZE is i m size, VOTING is he p o- po ion o o ing igh s o he ul ima e sha eholde and BOARDSIZE is he size o he boa d. All he eg essions include indus y and ime dummy a iables. Hansen is a es o o e -iden i ying es ic ions, unde he null hypo hesis ha all ins umen s a e unco ela ed wi h he dis u bance p ocess. m 2 is he s a is ical es o lack o second-o de se ial co ela ion in he i s -di e ence esidual. z 1 is he Wald es o he join signi icance o he epo ed coe icien s. z 2 is he Wald es o he join signi icance o ime dummies. z 3 is he Wald es o he join signi icance o indus y dummies. F is he es o he join signi icance o all he coe icien s. U is he es o he nonlinea e ec o emale di ec o s * , ** , ***: s a is ically signi ican a 1, 5 and 10%, espec i ely. E. Ga cía-Meca e al. In e na ional Re iew o Financial Analysis 79 (2022) 101973 9 Table 6 Family s. non- amily emale di ec o s. Dependen a iable: Payou (GMM) Dependen a iable: Di /A (GMM) Dependen a iable: Di / Sales (GMM) Model 8 Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Model 16 Model 17 Model 18 Model 19 FAM_WD 0.428*** (2.65) 0.255*** (3.89) 0.235*** (3.29) 0.077*** (4.88) 0.139** (2.01) 0.079** (2.34) FAM_WD 2 −0.012*** (−2.60) −0.009*** (−3.33) −0.007*** (−3.14) −0.002*** (−4.76) −0.004** (−2.02) −0.003*** (−2.90) NONFAM_WD −0.286** (−2.56) −0.177** (−2.30) −0.157*** (−4.06) −0.015*** (−7.87) −0.094** (−2.31) −0.058** (−2.08) NONFAM_WD 2 0.014*** (3.97) 0.008*** (3.17) 0.004** (2.52) 0.0005*** (6.75) 0.004*** (3.10) 0.002*** (2.78) ROA 0.271* (1.75) 0.469*** (3.09) 0.346* (1.85) 0.977*** (3.95) 0.093*** (3.78) 0.043*** (6.03) 0.044*** (4.38) 0.041*** (6.11) 0.074** (2.40) 0.035*** (3.68) 0.01 (0.77) 0.035*** (4.01) GROWTH 0.002 (0.47) −0.003 (−1.16) 0.006 (0.56) −0.013** (−2.36) −0.005** (−2.32) −0.005*** (−6.03) −0.004** (−2.11) −0.002*** (−2.73) 0.004 (1.23) −0.003** (−3.58) 0.004 (1.36) −0.003*** (4.35) LEV −0.650** (−2.14) −0.010 (−0.10) −0.185*** (2.78) −0.656*** (−5.13) −0.002 (−0.21) −0.001 (−1.08) −0.001 (−0.41) −0.005** (−2.45) −0.001 (−0.90) 0.001 (−0.35) −0.020** (−2.11) −0.003 (−0.95) SIZE 0.085* (1.97) −0.005 (−0.30) 0.213*** (3.29) 0.054** (2.41) 0.003 (1.58) 0.004 (0.82) 0.002 (1.27) −0.002 (−0.43) 0.017** (2.01) 0.003*** (3.00) 0.0025 (1.22) 0.002** (2.18) VOTING 0.005* (1.85) −0.003 (−0.38) 0.001* (1.71) −0.003*** (−3.08) 0.001* (1.92) −0.003** (−2.16) 0.007 (1.32) 0.002 (1.18) 0.006 (0.40) −0.001*** (−3.81) 0.001 (1.22) −0.001*** (4.20) BOARDSIZE 0.266* (1.90) 0.177** (1.95) 0.500* (1.82) 0.455*** (4.43) 0.003 (0.41) 0.008*** (3.07) 0.006 (0.55) 0.010*** (3.60) 0.026* (1.65) 0.010** (2.19) 0.023** (2.19) 0.015*** (3.23) GENERATION 0.299*** (2.76) 0.162* (1.82) −0.002 (−0.72) 0.010 (−2.35) 0.004 (0.58) 0.008* (1.95) CEOAge −0.004* (−1.81) −0.006*** (−2.84) −0.003*** (−5.96) −0.004*** (8.38) −0.003*** (−4.27) −0.004*** (6.30) FAM_MD 0.005** (2.23) 0.005*** (2.56) 0.001*** (4.61) 0.001*** (3.45) 0.001*** (2.81) 0.001*** (2.52) In e cep −2.261*** (−2.57) 0.282 (0.61) −2.758*** (−3.01) 1.883** (2.45) −0.047* (−1.97) −0.027*** (−4.05) −0.021 (−0.81) −0.024*** (−4.90) 0.158 (0.74) −0.061*** (−5.77) −0.041* (−1.65) −0.065*** (−5.97) m 2 −0.49 −0.61 −0.73 −0.35 0.64 0.19 0.54 0.19 0.23 −0.80 −0.82 −0.82 Z1 20.60*** 4.76*** 4.08*** 11.49*** 5.20*** 53.99*** 8.86*** 57.58*** 25.6*** 23.63*** 4.60*** 21.51*** Z2 5.99*** 3.63*** 8.35*** 4.63*** 6.04*** 28.65*** 4.88*** 9.06*** 9.6*** 7.04*** 4.3*** 7.79*** Z3 9.67*** 26.39*** 16.25*** 29.36*** 7.05*** 39.51*** 6.72*** 16.46*** 5.91*** 31.68*** 6.34*** 38.62*** Hansen es 34.64 68.40 34.45 45.44 24.18 61.53 29.98 49.67 27.10 61.60 31.76 65.79 F es 637.21*** 227.90*** 201.98*** 653.31*** 81.04*** 3106.09*** 95.60*** 742.53*** 39.32*** 3487.56*** 192.4*** 1732.72*** U es 2.50*** 3.06*** 2.56*** 2.30** 2.99*** 1.31* 2.04** 3.29*** 2.78*** 3.73*** 2.20** 4.85*** No. o obse a ions 745 745 745 745 745 745 745 745 745 745 745 745 Es ima ed coe icien s ( -s a is ics based on obus s anda d e o ). The dependen a iable is di idends o ne ea nings (models 8 and 9), di idends o o al asse s (models 10 and 11), di idends o o al sales (models 12 and 13), and he likelihood o paying di idends (models 14 and 15). FAM_WD (NONFAM_WD) is he p opo ion o emale di ec o s wi h (wi hou ) amily ies wi h he con olling amily, ROA is e u n on asse s, GROWTH is g ow h oppo uni ies, LEV is inancial le e age, SIZE is i m size, VOTING is he p opo ion o o ing igh s o he ul ima e sha eholde , BOARDSIZE is he size o he boa d, and GENERATION is he gene a ional-s age o he amily i m. CEOage is he age o he CEO. FAM_MD is he numbe o amily male di ec o s on he boa d o amily i ms di ided by he o al numbe o boa d membe s. All he eg essions include indus y and ime dummy a iables. Hansen is a es o o e -iden i ying es ic ions, unde he null hypo hesis ha all ins umen s a e unco ela ed wi h he dis u bance p ocess. m 2 is he s a is ical es o he lack o second-o de se ial co ela ion in he i s -di e ence esidual. z 1 is he Wald es o he join signi icance o he epo ed coe icien s. z 2 is he Wald es o he join signi icance o ime dummies. z 3 is he Wald es o he join signi icance o indus y dummies. F is he es o he join signi icance o all he coe icien s. U is he es o he nonlinea e ec o emale di ec o s. * , ** , ***: s a is ically signi ican a 1, 5 and 10%, espec i ely. E. Ga cía-Meca e al.