Multimarket pioneers: Does multimarket contact improve their performance?

MULTIMARKET PIONEERS: DOES MULTIMARKET CONTACT IMPROVE THEIR PERFORMANCE? Bea iz Domínguez* Uni e si y o La Rioja Uni e si y o Za agoza Ronda Mise ico dia 1, 22001, Huesca (Spain) Tel:+34 974 23 93 73 (bdguez@uniza .es) Jaime Gómez Uni e si y o La Rioja Calle La Cigüeña, 60, 26006, Log oño (Spain) (jaime.gom[email p o ec ed]) Juan P. Maícas Uni e si y o Za agoza Calle A a azana 4, 44003, Te uel (Spain) (jpmaicas@uniza .es) *Co esponding au ho Keywo ds: pionee s; ecip ocal mul ima ke con ac ; non- ecip ocal mul ima ke con ac ; sphe es o in luence; mobile elecommunica ions. ACKNOWLEDGEMENTS We hank inancial suppo om he Spanish Minis y o Economy, Indus y and Compe i i eness (MINECO), he Spanish Agency o Resea ch (AEI), S a egos Resea ch G oup and Eu opean Regional De elopmen Fund (FEDER) (p ojec s ECO2017-85451-R and ECO2014-53904-R). The help ul commen s o Tobias K e schme , Gian i o Lanzolla, Lucio Fuen elsaz, Ma co Gia a ana and Zied Gued i on ea lie e sions o his pape a e also g a e ully acknowledged. 2 MULTIMARKET PIONEERS: DOES MULTIMARKET CONTACT IMPROVE THEIR PERFORMANCE? Abs ac : Fi s -mo e ad an ages (FMA) and mul ima ke con ac (MMC) ha e e ol ed independen ly in he s a egic managemen li e a u e. This is su p ising because FMA e ode as a esul o compe i ion which, in u n, is a ec ed by MMC. This pape links hese li e a u es h ough he concep o sphe es o in luence and analyses he e ec o MMC on pionee s’ p o i abili y. We use he o de o ma ke en y o iden i y sphe es o in luence and o de e mine hei dis ibu ion among mul ima ke i ms. The dis ibu ion o sphe es o in luence allows us o dis inguish be ween ecip ocal MMC and non- ecip ocal MMC and o s udy how each o hem de e mines pionee s’ p o i abili y. We es ou hypo heses in he mobile elecommunica ions indus y. Ou indings show ha ecip ocal MMC has a posi i e e ec on pionee s’ esul s, bu non- ecip ocal MMC nega i ely a ec s hem. Keywo ds: pionee s; ecip ocal mul ima ke con ac ; non- ecip ocal mul ima ke con ac ; sphe es o in luence; mobile elecommunica ions. 3 INTRODUCTION Fi s -mo e ad an ages (FMA) exis when he pionee s in a ma ke ob ain highe bene i s in e ms o p o i abili y, alue c ea ion, o su i al han he ollowe s (Liebe man and Mon gome y, 2013). Ea ly s udies ocused on e i ying he exis ence o FMA (Ke in and Va ada ajan, 1992) and on explaining hei d i e s (Liebe man and Mon gome y, 1988). These s udies concluded ha FMA gene ally exis , al hough hey may depend on he me hodological decisions aken by esea che s (Vande we and Mahon, 1997). Recen wo k on en y iming has e ol ed in o o he ea u es o he heo y and claims ha he logic o FMA is in alid wi hou a en ion o con ingencies (Zacha y e al., 2015). This implies ha FMA a e no co ec ly explo ed by simple and di ec e ec s on pe o mance bu a e bes assessed inco po a ing con ingen ac o s (Fos u i, Lanzolla, and Suá ez, 2013). The ele an ques ion o he li e a u e, nowadays, is no whe he FMA exis bu he ac o s ha de e mine hem (Dykes and Kole , 2018; Zacha y e al., 2015). Following his ein o esea ch, some s udies ha e been o ien ed o he mac o (Gomez, Lanzolla, and Maícas, 2016; Suá ez and Lanzolla, 2007) and mic o (F anco e al., 2009) de e minan s o FMA. Despi e hei con ibu ion o FMA li e a u e, he e is s ill much o lea n abou he bounda y condi ions ha a ec he magni ude o hese ad an ages (Liebe man and Mon gome y, 2013; Zacha y e al., 2015) and, consequen ly, he oppo uni y o s udies on FMA ha imp o e ou unde s anding o he phenomenon unde no el heo e ical lenses (Dykes and Kole , 2018). In unde s anding he bounda y condi ions su ounding FMA, we base on he idea ha ollowe s can e ode he p o i s o pionee s h ough hei compe i i e ac ions (Dykes and Kole , 2018; Lee e al., 2000) and, he e o e, ha compe i ion is a signi ican d i e ha de e mines pionee ing ad an ages (Ke in e al., 1992; Zacha y e al., 2015). Besides his, wi h ew excep ions (E hi aj and Zhu, 2008; Gal-o , 1985), he li e a u e on FMA has paid li le a en ion o he e ec s o compe i ion. We con ibu e o ill his gap and o be e unde s and FMA by 4 analysing how compe i ion de e mines he magni ude o hese ad an ages. In doing so, we adop he lenses p o ided by he heo y o mul ima ke compe i ion, which depends on he exis ence o mul ima ke con ac (MMC). MMC has been de ined as a si ua ion in which i ms simul aneously compe e wi h he same i als in di e en ma ke s (Be nheim and Whins on, 1990; Spagnolo, 1999), and i has been shown o ha e impo an implica ions o i ms’ compe i i e agg essi eness (Gimeno and Woo, 1999). In pa icula , he numbe o compe i i e ac ions dec ease wi h he numbe o common ma ke s be ween i als (Young e al., 2000; Yu e al., 2009), so ha MMC migh lead o mu ual o bea ance be ween compe i o s (Ha eman and Nonnemake , 2000). These ideas sugges ha pionee s ha ha e MMC wi h hei ollowe s con on highly di e en compe i i e condi ions han pionee s wi h single-ma ke ollowe s and, he e o e, he magni ude o hei ad an ages will be di e en . Wi h his in mind, we a gue ha connec ing FMA and MMC heo ies helps us o be e unde s and he phenomenon o FMA. To explain he con ingen ole o MMC on FMA, we use he concep o sphe es o in luence. In pa icula , we adop he idea ha mul ima ke i ms do no assign he same impo ance o all ma ke s bu end o place mo e impo ance on he ma ke s whe e hey ob ain highe p o i s (Be nheim and Whins on, 1990). These impo an ma ke s a e called ‘sphe es o in luence’ and signi ican ly de e mine i ms’ compe i i e beha iou (Be nheim and Whins on, 1990; Gimeno, 1999). In his s udy, we base on he e idence suppo ing FMA and expec ha mul ima ke i ms place mo e impo ance on he ma ke s whe e hey a e i s mo e s. Thus, we iden i y sphe es o in luence as he ma ke s whe e mul ima ke i ms a e pionee s. The iden i ica ion o sphe es o in luence is impo an because i allows us o obse e he way in which hey a e dis ibu ed among mul ima ke compe i o s and, he e o e, he ype o MMC ha pionee s ha e wi h hei ollowe s. We elabo a e on he dis inc ion be ween ecip ocal and non- ecip ocal MMC (Gimeno, 1999) and con end ha no only he exis ence o MMC be ween 5 pionee s and hei ollowe s bu also i s ype is impo an in explaining he magni ude o FMA. A ecip ocal MMC be ween wo i ms is de ined as a MMC in which he pionee in ma ke n is a ollowe in ma ke m, and he pionee in ma ke m is a ollowe in ma ke n. All he o he si ua ions in which he e is a MMC bu i ms do no ha e ecip ocal posi ions in he common ma ke s all in he ca ego y o non- ecip ocal MMC. We con end ha he ecip oci y o i ms’ ma ke posi ions imp o es he e ec i eness o MMC and leads o a educ ion in he le el o compe i ion be ween pionee s and ollowe s om he beginning o hei mul ima ke ela ionship. This implies ha ecip ocal MMC has a linea , and nega i e e ec on compe i ion and a posi i e e ec on pionee s’ p o i abili y. In con as , we a gue ha he ela ionship be ween non- ecip ocal MMC and pionee s’ p o i abili y is mo e complex. Al hough non- ecip ocal MMC can also ul ima ely ha e a i al y educing e ec , his consequence only a ise when he amoun o MMC be ween he pionee and i s ollowe s is high enough o hem o ecognize hei in e dependences, gi en he lowe c edibili y and e ec i eness o his ype o MMC. This means ha non- ecip ocal MMC is no e ec i e as a i al y educing mechanism o low le els o MMC. In ac , o low le els o non- ecip ocal MMC, an inc ease in compe i ion is he mos likely ou come o an inc ease in MMC. Only when non- ecip ocal MMC accumula es up o a medium le el, i s a s ha ing educed i al y and inc eased p o i abili y as a consequence. As a esul , we p opose a U-shaped in luence o non- ecip ocal MMC on pionee s’ p o i abili y. We es ou hypo heses in he wo ldwide mobile elecommunica ions indus y. Ou sample is made up o 2,094 qua e ly obse a ions, om 2001 o 2012, ha belong o 81 pionee s in 72 coun ies.1 Ou esul s con i m he exis ence o FMA and show ha hei magni ude is dependen on he ype o MMC be ween pionee s and ollowe s. 1 Fo us, each coun y is a ma ke . Hence, he e ms coun y and ma ke a e synonymous in his s udy. 6 The con ibu ions o his s udy a e wo old. Fi s , we inco po a e MMC as a con ingen ac o ha de e mines he magni ude o FMA and, he e o e, con ibu e o he iden i ica ion o he bounda y condi ions su ounding hese ad an ages. The analysis o bounda y condi ions has o long been ecognized as c i ical o heo y ad ancemen (Gonzalez-Mulé and Aguinis, 2018). Unde s anding he si ua ions in which a heo y holds and he way in which con ingen ac o s can “…change he na u e and/o di ec ion o ela ions among a iables” is c i ical i we wan o p o ide a mo e p ecise guidance o manage s (Gonzalez-Mulé and Aguinis, 2018: 2247). In he case o he li e a u e on FMA, se e al au ho s ha e highligh ed he need o pay a en ion o con ingencies (Liebe man and Mon gome y, 2013; Zacha y e al., 2015) i a alid heo y is o be buil . Fo example, he (in)exis ence o MMC could explain why FMA could be (lowe ) highe in some si ua ions han in o he s. Simila ly, MMC heo y sugges s ha he inc eased p o i s om o bea ing in a ma ke could be ob ained a he expense o he educ ion o p o i s in o he ma ke s (see, o example, Be nheim and Whins on, 1990, McG a h e al., 1998, Fe nández and Ma in, 1998). This is impo an o ou unde s anding o FMA heo y, gi en ha he highe p o i abili y ob ained in a gi en ma ke by a pionee could be o se by he lowe p o i abili y ea ned by he pionee in he ma ke s in which his i m is a ollowe . Al hough p io s udies ha e made an e o on explaining some o hese con ingences, he e is s ill much o lea n om he in eg a ion o di e en heo ies (Dykes and Kole , 2018; Zacha y e al., 2015). In pa icula , we a e no awa e o any e o s a emp ing o in eg a e he FMA and MMC li e a u es. The inclusion o he compe i i e dynamics gene a ed by MMC allows us o be e unde s and he phenomenon o FMA. Second, his s udy ex ends he analysis o he wo li e a u es o an in e na ional se ing, adding ano he con ingency o ou unde s anding o FMA heo y. Wi h a ew excep ions (Gomez and Maicas, 2011; Yu e al., 2009), mos esea ch on bo h FMA and MMC has been limi ed o one ma ke (Kim and Lee, 2011; Makadok, 1998). In he case o he s udy o FMA his is especially 7 impo an , gi en ha he empi ical esul s o he sca ce s udies de eloped in an in e na ional se ing ha e been mixed (Dykes and Kole , 2018). I is impo an o no e ha coun ies may di e in a numbe o a ibu es such as hei cul u e o hei deg ee o ins i u ional de elopmen ha may lead o di e ences in he magni ude o pionee ing ad an ages (Dykes and Kole , 2018). Addi ionally, s udies on FMA ha ha e b oadened hei analysis o di e en ma ke s (Gómez and Maícas, 2011; Song e al., 1999) ha e no conside ed he mul ima ke na u e o i ms. Howe e , inco po a ing MMC in o he s udy o FMA in an in e na ional se ing is a na u al addi ion, as i ms o en expand ab oad (Hi e al., 2016), and, hen, migh compe e simul aneously wi h he same i als in se e al in e na ional ma ke s. The e o e, he in e na ional dimension o ou s udy con ibu es o he li e a u e on FMA and esponds o a ecen call o he necessi y o b oadening he analysis o hese ad an ages o an in e na ional con ex (Dykes and Kole , 2018). In his sense, ou esea ch adds o he e idence ha con i ms ha FMA exis in an in e na ional se ing and shows ha hei magni ude signi ican ly depends on he numbe o in e na ional ma ke s whe e pionee s mee hei ollowe s and he ype o MMC be ween hem. Simila ly, he in e na ional dimension o ou s udy con ibu es o he MMC li e a u e. I is impo an o no e ha he coo dina ion be ween he ac i i ies ha i ms de elop in di e en ma ke s, which is equi ed o mu ual o bea ance o a ise, is mo e di icul in an in e na ional se ing (Yu e al., 2009). Despi e his, ou esul s e eal ha ecip ocal MMC is e ec i e in educing i al y in his con ex . Howe e , coo dina ion migh be mo e di icul in a si ua ion o non- ecip ocal MMC, in which he numbe o MMC equi ed o mu ual o bea ance o a ise is high. LITERATURE REVIEW AND THEORETICAL FRAMEWORK In his sec ion, we i s examine he FMA li e a u e and, hen, e iew he main indings in he li e a u e on MMC. This allows us o desc ibe he main a gumen s o he wo heo ies, be o e p esen ing he model and hypo heses, which is done in he nex sec ion. 8 De e minan s o FMA FMA imply ha pionee ing in a new geog aphic o p oduc ma ke leads o inc eased pe o mance (Liebe man and Mon gome y, 1988; 1998). Al hough some s udies ha e concluded ha he o de o en y does no ha e any e ec on i m pe o mance (Golde and Tellis, 1993) o ha he e ec is nega i e (Robinson and Fo nell, 1985), mos s udies ha e con i med ha FMA exis (see Vande we and Mahon, 1997). These ad an ages ha e been shown in di e en con ex s, such as manu ac u ing (Ahlb ech and Ecke , 2013), ad e ising agencies (Magnusson, Wes john, and Boggs, 2009), he oil indus y (Masca enhas, 1992) and he mobile elecommunica ions indus y (Gómez and Maicas, 2011; Jakopin and Klein, 2012), leading o he conclusion ha FMA a e widesp ead. Di e en mechanisms ha e been a gued o explain how FMA eme ge, bu he mos accep ed idea is ha hese ad an ages a e p ima ily de i ed om he p e-emp ion o esou ces, consume swi ching cos s, and echnological leade ship (Liebe man and Mon gome y, 1988). Besides he consensus abou he exis ence o FMA and hei mechanisms, p io li e a u e also ag ees ha FMA do no las o e e and end o diminish o e ime (Liebe man and Mon gome y, 2013). This ac has led esea che s o ocus on he ac o s ha a ec he magni ude o FMA, dis inguishing be ween i m (F anco e al., 2009) and en i onmen al ac o s (Suá ez and Lanzolla, 2007). Rega ding i m-le el ac o s, s udies ha e p oposed echnological capabili ies (F anco e al., 2009) and R&D in es men s (Kim and Lee, 2011) as enhance s o he ela ionship be ween pionee ing and pe o mance. Simila ly, esea che s ha e highligh ed he e olu ion o he echnology and he ma ke (Suá ez and Lanzolla, 2007), he app op iabili y egime (Kim and Lee, 2011), and he numbe o compe i o s as en i onmen al de e minan s o he magni ude o FMA (Makadok, 1998). In his ega d, some ac o s o he compe i i e dynamics, such as imi a ion, he in oduc ion o new p oduc s and, he en y o new compe i o s in o he ma ke 9 lead o an e osion o almos all compe i i e ad an ages (Wiggins and Rue li, 2002). This means ha he magni ude o FMA signi ican ly depend on he compe i ion be ween pionee s and ollowe s (E hi aj and Zhu, 2008; Fe ie e al., 1999; Lee e al., 2000). Pionee s ha do no ace compe i ion du ing a longe ime bene i mo e om monopolis ic esul s and lea ning e ec s. In con as , FMA e ode in he gale o compe i ion (Lee e al., 2000), meaning ha he ollowe s’ in oads e ode pionee s’ pe o mance (Boulding and Ch is en, 2003). In his ega d, ollowe s may o e come he ad an ages om ea ly en y h ough imi a ion (E hi aj and Zhu, 2008; Lee e al., 2000) and some o he s compe i i e mo emen s (Fe ie e al., 1999). Fo ins ance, he as e a ollowe in oduces he same p oduc as o e ed by he pionee in he ma ke , he sho e he pionee ing ad an age (Lee e al., 2000). Addi ionally, FMA may be diminished as la e en an s launch new p oduc s (E hi aj and Zhu, 2008), make p ice educ ions and, make R&D in es men s (Fe ie e al., 1999). This e osion o FMA de i ed om compe i o s’ beha iou is an accele a ing p ocess ha places a p emium on compe i o analysis (Liebe man and Mon gome y, 2013). Based on his idea, ou s udy ocuses on he ole o compe i ion in de e mining pionee s’ p o i s in an in e na ional con ex . The in e na ional dimension o his s udy makes us adop a b oad pe spec i e o compe i ion and conside ha pionee s can compe e wi h hei ollowe s in many coun ies. In his ega d, he compe i i e beha iou o i ms di e s conside ably depending on he numbe o ma ke s ha hey sha e wi h hei compe i o s (Ka nani and We ne el , 1985; Smi h and Wilson, 1995). Consequen ly, o analyse p ope ly he e ec o compe i ion on FMA, we in oduce he heo y o MMC. MMC and i al y: The ole o sphe es o in luence Resea che s ha e de ined MMC as a si ua ion in which i ms simul aneously compe e wi h he same i als in di e en geog aphic o p oduc ma ke s (Baum and Ko n, 1999; Yu and Canella, 2013). A g owing numbe o s udies in economics (E ans and Kessides, 1994) and s a egy 16 be ween pionee s and ollowe s while he second ocuses on a si ua ion whe e hese i ms ha e non- ecip ocal MMC. ---------------------------------------------------- Inse Figu e 1 he e ---------------------------------------------------- Pionee s’ p o i abili y and ecip ocal MMC The en y o a ollowe in o a ma ke in which a pionee is he monopolis signals he s a o compe i ion. In ac , ma ke commonali y, i.e., he coincidence o i ms in he same compe i i e space, is a p e- equisi e and a d i e o compe i ion (Chen, 1996). As he pionee ’s ad an age is likely o diminish wi h an escala ion in compe i ion (Dykes and Kole , 2018; Lee e al., 2000), a pionee would p e e keeping i s monopolis ic posi ion. Howe e , en y in o a new ma ke has adi ionally been seen as an agg essi e mo emen (Ha eman and Nonnema ke , 2000; Fuen elsaz and Gomez, 2006). The e o e, once a ollowe has adop ed an agg essi e s ance h ough en y in o a ma ke , he pionee is likely o espond o his en y in o de o p o ec i s ad an age, and compe i ion is likely o escala e. This escala ion o compe i ion can ake di e en o ms, such as imi a ion (Lee e al., 2000), p oduc di e en ia ion (Ma hews, 2002), R&D in es men s (Robinson and Chiang, 2002), o p ice educ ions (Schnaa s, 1994). All hese o ms hinde he main enance o FMA and, consequen ly, lead o an e osion o hese ad an ages (Dykes and Kole , 2018; Zacha y e al., 2015). New en ies in o he ma ke s occupied by compe i o s a e also a likely ou come o an escala ion o compe i ion (Baum and Ko n, 1996, 1999), inc easing he numbe o MMCs be ween pionee s and ollowe s. In some cases, once a ollowe has en e ed a ocal ma ke , he pionee in ha ma ke esponds by en e ing he ma ke in which he ollowe ac s as a pionee and whe e i has impo an e i o ial in e es s, c ea ing a ecip ocal MMC (Ka nani and We ne el , 1985). Recip ocal MMC e e s o a si ua ion in which he pionee and he ollowe ha e ecip ocal posi ions in hei common ma ke s. Si ua ion A in Figu e 1 illus a es his. Due o he exis ence 17 o FMA, i m i has a pe o mance ad an age o e i m j in ma ke n, and i m j has an ad an age o e i m i in ma ke m. Thus, i ms i and j ha e di e en in e es s in he ma ke s in which hey compe e, so hey a ibu e di e en le els o impo ance o he ma ke s in which hey a e ope a ing: each i m gi es a g ea e impo ance o he ma ke whe e i is a pionee . Hence, i m i is expec ed o p io i ise ma ke n o e ma ke m, while i m j gi es mo e impo ance o ma ke m han o ma ke n. The exis ence o ecip ocal MMC implies he exis ence o ecip ocal e i o ial in e es s in he ma ke s whe e he pionee and i s ollowe compe e simul aneously. The li e a u e on MMC has a gued ha , as each i m p io i ises a di e en ma ke , i ends o gi e up i s i al’s impo an ma ke — ha is, he ma ke whe e he i al enjoys FMA—in exchange o he same ea men in i s own impo an ma ke (Be nheim and Whins on, 1990; So enson, 2007). This means ha he pionee / ollowe will e ain om imi a ing, launching new p oduc s, o making p ice educ ions i he ollowe /pionee beha es in he same way in he ma ke ha is mo e impo an o his i al. In o he wo ds, ou con en ion is ha ecip ocal MMC educes he incen i es ha pionee s and ollowe s ha e o compe e, leading o mu ual o bea ance. P io esea ch on MMC has concluded ha mu ual o bea ance can be eached i wo compe ing i ms ha e c edible h ea s o e alia ion (Gimeno, 1999). In a si ua ion o ecip ocal MMC, each i m is p esen in i s i al’s impo an ma ke and, consequen ly, bo h ha e c edible e alia o y h ea s. I he pionee is a acked, i could espond e ec i ely in he ollowe ’s impo an ma ke . Simila ly, he ollowe can espond e ec i ely o an a ack in he impo an ma ke o i s i al. As each i m knows i s mul ima ke i al’s capabili ies o esponding e ec i ely o an a ack, bo h a e willing o espec hei i al’s impo an ma ke in exchange o he same ea men in hei own impo an ma ke . The e o e, he e alia o y capabili y o he pionee / ollowe leads he ollowe /pionee o espec i s e i o ial in e es s. Because i als’ compe i i e ac ions e ode FMA (Lee e al., 2000), he educ ion in he numbe o compe i i e 18 ac ions ha is de i ed om ecip ocal MMC helps pionee s o p o ec hei ad an ages in he ma ke . I is impo an o no e ha his educ ion in he le el o compe i ion akes places as soon as ecip ocal MMC be ween he pionee and he ollowe appea s, leading o a di ec and posi i e e ec o his ype o MMC on pionee s’ p o i abili y. Based on his easoning, ou i s hypo hesis posi s ha : Hypo hesis 1: Recip ocal MMC has a posi i e e ec on pionee s’ p o i abili y. Pionee s’ p o i abili y and non- ecip ocal MMC Non- ecip ocal MMC migh also a ec compe i i e beha iou , ul ima ely educing he escala ion o compe i ion elici ed by a ollowe ’s ini ial en y in o he ma ke o he pionee , bu i s e ec is signi ican ly di e en han ha o ecip ocal MMC. As opposed o he ci cums ance ha gene a es he i s hypo hesis, ecip oci y in i ms’ ma ke posi ions now does no exis . While he en y in o he ma ke o he pionee means ha he ollowe is now p esen in i s i al’s impo an ma ke , we now ocus on a si ua ion in which he escala ion o compe i ion does no lead he pionee o an en y in o a salien ma ke o i s ollowe . The absence o ecip oci y al e s he way in which MMC a ec s compe i ion, especially in he case o he ollowe ’s compe i i e beha iou (Gimeno, 1999). This implies ha he e ec o non- ecip ocal MMC o e he in ensi y o i al y and FMA is di e en om he si ua ion desc ibed in Hypo hesis 1. In his si ua ion, he lack o ecip oci y implies ha he pionee and he ollowe do no ha e simila capabili ies o damage each o he . In pa icula , he ollowe ha is p esen in an impo an ma ke o i s mul ima ke i al, poses a g ea e h ea o e alia ion on he pionee . The eason is ha he ollowe can espond e ec i ely o any a ack o he pionee in he ma ke whe e his i m enjoys FMA. In con as , he pionee is no p esen in any salien ma ke o he ollowe . 19 We use Si ua ion B o Figu e 1 o be e explain hese di e ences be ween he capabili ies o he pionee and he ollowe o pose an e ec i e h ea o e alia ion in a si ua ion o non- ecip ocal MMC. Due o he exis ence o FMA, i m i would a ibu e mo e impo ance o he ma ke s in which i is he pionee . This i m could ecei e a esponse o an a ack in he ma ke in which i enjoys FMA, and his possibili y makes his pionee be p one o educe i s agg essi eness agains he ollowe in e e y ma ke in which bo h i ms compe e, seeking o sa egua d he ma ke in which i enjoys FMA. In con as , i m j is always a ollowe . Thus, i will assign simila le els o impo ance o he ma ke s in which i compe es wi h i m i, in e ms o he o de o ma ke en y. As i m j ( he ollowe ) does no ha e a s ong posi ion in any o he ma ke s in which i compe es wi h he i m i ( he pionee ), his pionee does no exe such an e ec i e h ea o e alia ion on i s ollowe when ecip oci y does no exis . The e o e, in a si ua ion o non- ecip ocal MMC, he ini ial incen i es o he ollowe o ole a e he pionee a e lowe han in a si ua ion o ecip ocal MMC. As a esul , he escala ion o compe i ion elici ed by he ini ial en y o he ollowe in o he pionee ’s ma ke is likely o las longe han in he case o ecip ocal MMC. Howe e , he lowe incen i es o he ollowe o ole a e he pionee o he ma ke may change as he amoun o non- ecip ocal MMC be ween he wo i ms inc eases (Ha eman and Nonnemake , 2000; Fuen elsaz and Gomez, 2006). We a gue ha , in a si ua ion o non- ecip ocal MMC, i is he le el o MMC which de e mines whe he he ollowe has incen i es o compe e o o collude wi h he pionee . The eason is ha , an inc easing le el o non- ecip ocal MMC s eadily inc eases he h ea o e alia ion, as he numbe o ma ke s sha ed by he wo i ms inc eases and he s akes a isk a e highe . This means ha when he amoun o non- ecip ocal MMC inc eases om low o medium, he h ea o e alia ion is no high enough o mu ual o bea ance o a ise. Thus, he ollowe has incen i es o compe e agains he pionee , wi h he subsequen nega i e e ec on FMA. Ne e heless, when he olume o non- ecip ocal MMC be ween he wo i ms is om medium 20 o high, and he h ea o e alia ion is high, he s akes a isk make he ollowe o s a ha ing incen i es o mu ually o bea and, consequen ly, his bene i s FMA. The combina ion o hese coun e ailing e ec s on FMA, i s dec easing, hen inc easing, leads o a U-shaped e ec o non- ecip ocal MMC on pionee s’ p o i abili y. F om low o medium le els o non- ecip ocal MMC, he ollowe pe cei es ha he bene i s o compe ing in he ma ke a e highe han i s cos s. This ollowe is no awa e o i s in e dependencies wi h he pionee ye and, consequen ly, has incen i es o imp o e i s posi ion h ough compe i i e ac ions. Thus, his ollowe will launch new p oduc s in he ma ke , make R&D in es men s, o engage in p ice educ ions wi h he pu pose o compe ing agains he pionee o he ma ke (Fe ie e al., 1999; Lee e al., 2000). These compe i i e ac ions will inc ease he in ensi y o ma ke i al y (Baum and Ko n, 1999) and, mo e impo an ly, will limi he abili y o he pionee o p o ec i s ad an ageous posi ion (Liebe man and Asaba, 2006). Hence, a educ ion in pionee ’s p o i s is expec ed when his i m has low o medium le els o non- ecip ocal MMC wi h i s ollowe . We expec his e ec on he pionee ’s p o i s o be di e en wi h highe le els o non- ecip ocal MMC. As non- ecip ocal MMC be ween he pionee and i s ollowe inc eases, he mechanisms ha lead o a educ ion in he le el o compe i ion be ween mul ima ke i ms—namely, amilia i y and ea o e alia ion—become ope a i e. In such a si ua ion, he ollowe pe cei es han he cos s o compe ing in he ma ke may be g ea e han i s bene i s. On he one hand, a g ea e ma ke o e lap p o ides mul ima ke i ms wi h mo e in o ma ion abou he esou ces and beha iou s o hei compe i o s, inc easing hei amilia i y. This, in u n, makes mul ima ke i ms awa e o hei in e dependencies and o he bene i s o main aining low le els o compe i i e beha iou . Fu he mo e, g ea e knowledge o compe i o s enables mul ima ke ollowe s o an icipa e pionee s’ mo emen s. Due o his, he ollowe becomes awa e o he pionee s’ capabili y o cause eal damage. 21 On he o he hand, a g ea e ma ke o e lap also inc eases he h ea o e alia ion. The e o e, a pionee ha mee s i s ollowe in mo e ma ke s can espond o i s a acks in he same ma ke whe e he a ack akes place, in any o he ma ke s whe e hey compe e, o , e en, in all o hem (Ka nani and We ne el , 1985; Po e , 1980). Thus, he po en ial damage om e alia ion inc eases wi h he numbe o common ma ke s. The unce ain y abou whe e he e alia ion will ake place and he possibili y ha i akes place in many ma ke s a he same ime lead he ollowe p one o o bea unde in e media e le els o MMC. F om his olume o MMC, he ollowe ’s compe i i e beha iou so ens and, he e o e, he numbe o compe i i e ac ions in he ma ke lessens. This educ ion in ma ke i al y helps he pionee o p o ec i s ad an ageous posi ion (Liebe man and Asaba, 2006) and inc eases he magni ude o FMA. In summa y, as opposed o ecip ocal MMC ha di ec ly so ens bo h he pionee ’s and he ollowe s’ compe i i e beha iou , we expec non- ecip ocal MMC only o educe he le el o compe i ion be ween he pionee and i s mul ima ke ollowe when i anges om medium o high.2 Thus, non- ecip ocal MMC is expec ed o imp o e pionee s’ p o i s only when he amoun o MMC be ween he pionee and he ollowe is high enough o he mechanisms o amilia i y and h ea o e alia ion o be e ec i e. We expec hese mechanisms no o be e ec i e in educing compe i ion om low o medium le els o non- ecip ocal MMC and, he e o e, in hese si ua ions ma ke o e lap is likely o elici compe i ion and nega i ely a ec he pe o mance o pionee s. Acco dingly, we p opose ou second hypo hesis as ollows: Hypo hesis 2: Non- ecip ocal MMC has a U-shaped in luence on pionee s’ p o i abili y. 2 Some esea che s a gue ha high le els o MMC migh be associa ed wi h g ea e coo dina ion p oblems, and ha i migh lead o an inc ease in compe i ion (Chuang e al., 2018). Ne e heless, ou easoning is based on p io e idence, which con i ms ha a high olume o MMC is associa ed wi h low compe i i e ac ions (Fuen elsaz and Gómez, 2006; Klein e al., 2020) and, consequen ly, wi h a highe pe o mance (Domínguez e al., 2016). 22 SAMPLE, VARIABLES, AND METHODOLOGY Sample We es ou hypo heses in he mobile elecommunica ions indus y. The in o ma ion on mobile ope a o s was ob ained om GSMA In elligence. P io s udies on FMA in he mobile elecommunica ions indus y ha e shown ha i s -mo e ope a o s ou pe o m hei i als in he ma ke (Gómez and Maícas, 2011; Jakopin and Klein, 2012), con i ming he exis ence o FMA in his indus y. We also es ed whe he pionee s ob ain be e esul s han ollowe s in ou sample and con i med ha pionee s ou pe o m hei ollowe s.3 A e con i ming he exis ence o FMA, we obse e how ecip ocal and non- ecip ocal MMC a ec s he p o i abili y o pionee s. The sample consis s o qua e ly obse a ions o 81 mobile ope a o s4 ha we e i s mo e s in 72 coun ies5, om he beginning o 2001 o he end o 2012. This leads o a sample o 2,094 pionee -coun y-qua e obse a ions. To cons uc he da abase o his s udy, we i s obse ed which i m was he pionee in each coun y and, a e wa ds, included he obse a ions o hese i ms in ou sample, whe he hey we e ope a ing a he end o ou ime ame o no . Thus, ou sample does no ha e su i al biases. 3 The esul s o hese es ima ions a e a ailable om he au ho s on eques . 4 The e is usually one pionee pe coun y. Howe e , in some coun ies, wo ope a o s launched mobile se ices wi h a second gene a ion (2G) echnology a he same ime and bo h a e iewed as pionee s. P e ious s udies ha e usually assumed 2G as he s a ing poin o his indus y due o he low numbe o subsc ibe s in he p e ious gene a ion (Gómez and Maícas, 2011). While he i s gene a ion (1G) was mainly a ge ed a business consume s, 2G was expanded o e e yone and, he e o e, a ac ed much mo e use s o mobile se ices (G ajek and K e schme , 2009). 5 The coun ies ha a e included in ou sample a e: Albania, A gen ina, A menia, Aus ia, Bela us, Belgium, Bulga ia, Chad, Chile, China, Colombia, he Congo Democ a ic Republic, C oa ia, he Czech Republic, Denma k, Ecuado , Egyp , Es onia, Finland, F ance, Geo gia, Ge many, G eece, Hong Kong, Hunga y, I eland, Is ael, I aly, Japan, Jo dan, Kazakhs an, Kenya, Kuwai , La ia, Libe ia, Macao, Macedonia, Malawi, Malaysia, Maldi es, Mon eneg o, Mo occo, Mozambique, Nepal, he Ne he lands, New Zealand, Nige ia, No way, Oman, Pales ine, Pe u, Poland, Po ugal, Qa a , Romania, Se bia, Slo akia, Slo enia, Sou h A ica, Sou h Ko ea, Spain, S i Lanka, Sudan, Sy ia, Sweden, Swi ze land, Tajikis an, Tu key, Uganda, Uk aine, he Uni ed Kingdom, and Uzbekis an. The selec ion o hese coun ies can be explained by wo ac o s. Fi s ly, in he mobile elecommunica ions indus y, only ope a o s ha ha e gained a license in he ma ke can ope a e he e. Some coun ies, such as B azil o he Uni ed S a es, do no gi e hese licenses wi h na ional co e age bu wi h egional co e age. Since we do no ha e in o ma ion o measu e MMC in egional e ms, we selec only coun ies in which licenses ha e na ional co e age. Secondly, we include only coun ies wi h a ailabili y o in o ma ion o all he a iables included in ou model. 23 The mobile elecommunica ions indus y is an ideal labo a o y o es bo h FMA and MMC. Rega ding FMA, a leas wo easons make he indus y pa icula ly sui able o his s udy. Fi s , mobile ope a o s mus ei he ob ain a license (G ube , 2005) o acqui e an incumben ope a o in he ma ke o be able o p o ide se ices in a pa icula ma ke . This equi emen allows us o de e mine he o de o ma ke en y o e e y playe and, he e o e, o iden i y he pionee s o each ma ke . Second, he exis ing li e a u e has p o ed he exis ence o FMA in he mobile elecommunica ions indus y (Gómez and Maícas, 2011; Gómez e al., 2016; Jakopin and Klein, 2012). Howe e , he eme gence o new echnologies, he inc ease in p oduc lines, de egula ion, and he in e na ionaliza ion p ocess ha e inc eased he le els o compe i ion (Jakopin and Klein, 2012). This may hinde he abili y o pionee s o ob ain highe esul s. In ela ion o MMC, he mobile elecommunica ions indus y has expe ienced a signi ican p ocess o ma ke openness since he 1990s (G ube , 2005), leading o mobile i ms’ in e na ionaliza ion. The inc ease in he numbe o ma ke s in which i ms a e p esen means ha MMC has also inc eased in his indus y. Va iables Dependen a iable P io s udies in he FMA li e a u e ha e claimed ha p o i abili y is he mos sui able measu e o obse e hese ad an ages and ha e highligh ed ha ma ke sha e and su i al may lead o misleading esul s (Liebe man and Mon gome y, 2013). Following his ecommenda ion, we use a p o i abili y a iable. In pa icula , we use he EBITDA ma gin (EBITDAmin ) ha is calcula ed as he i m’s ope a ing p o i be o e in e es , ax, dep ecia ion, and amo iza ion di ided by i s o al e enue in he ma ke . This is a qua e ly and i m-in-ma ke a iable ha has been equen ly used o measu e i ms’ pe o mance in he mobile elecommunica ions indus y (Asimakopoulos and Whalley, 2017; Domínguez e al., 2016), o in e oga e he exis ence o FMA in his indus y (Gómez and Maícas, 2011; Gómez e al., 2016; Jakopin and 24 Klein, 2012), and o measu e i ms’ pe o mance in o he con ex s as well (see, as examples, Belezon e al., 2016; S e n and James, 2016; Shapi a, 2017). Independen a iables Recip ocal mul ima ke con ac (RMMCin ). We base on pas esea ch (Gimeno, 1999) and conside ha he pionee has one ecip ocal MMC wi h i s ollowe when ou condi ions a e simul aneously me : (1) ha he ocal i m i is he pionee o ma ke n (Pin); (2) ha ano he i m j is also p esen 6 in ma ke n as a ollowe (Ijn); (3) ha i m j is he pionee o ma ke m (Pjm); and (4) ha i m i is a ollowe in ma ke m (Iim). To be e illus a e hese condi ions, we ake he case o Tele onica and Voda one as an ins ance o one ecip ocal MMC in ou sample. Tele onica was he i s g oup o launch mobile se ices in Spain (Pin), whe e Voda one en e ed some yea s la e (Ijn). In con as , Voda one was he i s mo e in he Uni ed Kingdom (Pjm), and Tele onica en e ed his coun y some ime la e (Iim). We use dummy a iables o accoun o each o hese condi ions. The p oduc o he ou dummies is equal o one only when all condi ions a e me and, he e o e, one ecip ocal MMC be ween he pionee and i s ollowe exis s. To calcula e he o al numbe o ecip ocal MMCs ha he pionee has wi h he ollowe s in ma ke n, we inco po a e all o he ma ke s (di e en om m) and ollowe s (di e en om j). As each pionee can ha e a di e en numbe o ollowe s in he ocal ma ke n, we weigh he a iable o ecip ocal MMC by he numbe o ollowe s ha he pionee has in ha ma ke (Fuen elsaz and Gómez, 2006). Thus, ou a iable, RMMCin , is a i m-in-ma ke measu e ha is calcula ed as ollows: 6 In he mobile elecommunica ions indus y, mobile ope a o s a e o en pa o a g oup, which is p esen in wo o mo e ma ke s a he same ime (GSMA In elligence, 2018). G oups a e p esen in he ma ke s h ough hei pa icipa ion in he mobile ope a o s’ equi y. When measu ing MMC, we mus ake in o accoun ha mu ual o bea ance can eme ge only when coo dina ion be ween mul ima ke i ms’ subuni s— ha is, mobile ope a o s— can ake place. G oups ha ha e a low pa icipa ion in he ope a o s’ capi al a e no allowed o make impo an decisions, such as deciding whe he o compe e agains o o o bea he compe i o s in he ma ke . Thus, o ensu e ha coo dina ion be ween mobile ope a o s could ake place, we equi ed ha g oups had a pa icipa ion o a leas 50% in he ope a o s’ capi al. Below his h eshold, we did no conside ha g oups a e p esen in he ma ke s. 25 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖𝑖𝑖 =∑ ∑ 𝑃𝑃𝑖𝑖𝑖𝑖 ×𝐼𝐼 𝑗𝑗𝑖𝑖 ×𝑃𝑃 𝑗𝑗𝑗𝑗 ×𝐼𝐼𝑖𝑖𝑗𝑗𝑗𝑗𝑗𝑗 ∑𝐼𝐼 𝑗𝑗𝑖𝑖 𝑗𝑗 The a iable e lec s he a e age amoun o ecip ocal MMCs ha he pionee (i) has wi h he ollowe s o he ocal ma ke (n). Howe e , his measu e does no ake in o accoun he asymme ies in i ms’ ma ke impo ance. By de ini ion, a mul ima ke pionee ope a es in wo o mo e ma ke s a he same ime and can gi e each o hem a di e en impo ance. To cap u e he asymme ies o ecip ocal MMCs, we mul iply RMMCin by he impo ance ha he pionee gi es o he ocal ma ke n. In pa icula , o each qua e , we conside he pe cen age o he pionee ’s o al connec ions7 ha come om his ma ke . Non- ecip ocal mul ima ke con ac (NRMMCin ). This a iable cap u es he a e age amoun o MMCs o he pionee and i s ollowe s when ecip oci y does no exis . Simila o he p io measu e, ou condi ions should be me o conside ha he e is a non- ecip ocal MMC be ween he pionee and i s ollowe : (1) ha he ocal i m i is he pionee o ma ke n (Pin); (2) ha ano he i m j is p esen in ma ke n as a ollowe (Ijn); (3) ha i m j is p esen in ma ke m as a ollowe (Ijm); and (4) ha i m i is p esen in ma ke m, wha e e i is a pionee o a ollowe (Iim). The e o e, in he case o non- ecip ocal MMC, wo di e en si ua ions a e possible. Fi s , in bo h ma ke s n and m, i m i is he pionee (Pin; Pim) and i m j is a ollowe (Iin; Iim). Second, i m j is a ollowe in bo h ma ke s (Ijn; Ijm), while i m i is he pionee in ma ke n and a ollowe in ma ke m (Pin; Iim). One example o he i s si ua ion is obse ed be ween Voda one and Tele onica in Ge many and I eland. In bo h coun ies, Voda one was he pionee and Tele onica he ollowe . An example o he second si ua ion can be ound o O ange and Voda one in he Ne he lands and he Uni ed Kingdom. Voda one was he pionee in he UK, whe e O ange was 7 We ob ained he o al connec ions o each i m om GSMA In elligence. This da a sou ce measu es a i m’s connec ions as ‘ he o al unique SIM ca ds (o phone numbe s, whe e SIM ca ds a e no used) ha ha e been egis e ed on he mobile ne wo k a he end o he pe iod’ (GSMA In elligence, 2018). 32 o hei cu en p o i abili y, as is shown by he coe icien o he ou h lag o EBITDA ma gin (β = 0.00443; p < 0.05). The coe icien o Fi m size indica es ha a bigge size nega i ely de e mines i ms’ p o i s (β = -0.000224; p < 0.05). Addi ionally, we obse e ha la ge coun ies allow pionee s o ob ain be e esul s, as can be seen in he posi i e coe icien o Ma ke size (β = 0.000156; p < 0.01). Mo eo e , GDP pe capi a has a nega i e and signi ican e ec on EBITDAm (β = -0.141; p < 0.10). Robus ness checks We conduc ed some addi ional analyses o assess he obus ness o ou esul s. Fi s , he in e na ional dimension o his s udy migh lead o he exis ence o some in e dependencies be ween i m obse a ions among coun ies. Tha is, he e migh be some in e dependencies be ween he obse a ions o a i m ha is a pionee in mo e han one coun y.12 To deal wi h his, we un a new se o es ima ions whe e we include a a iable ha accoun s o he belonging o he pionee o an in e na ional mobile g oup. In pa icula , we include he a iable G oup in each one o he models o Table 2. This a iable akes he alue o 0 when he i m does no belong o any mobile g oup, meaning ha he e a e no in e dependencies be ween i m obse a ions among coun ies. O he wise, he a iable is equal o he code iden i ica ion o he mobile g oup which he pionee belongs o, ga he ing he obse a ions o pionee s ha belong o he same mobile g oup. The esul s o his es ima ion a e consis en wi h hose epo ed in Table 2 and con i m ha ecip ocal MMC has a posi i e e ec on pionee s’ p o i abili y while non- ecip ocal MMC has a nega i e e ec on hem.13 We also e an all he models o Table 2 in a sample ha is compounded by pionee s ha , indeed, ha e FMA. Al hough we con i med he exis ence o FMA in ou sample, one could a gue ha some pionee s ha a e included may be ou pe o med by hei ollowe s and, consequen ly, ail 12 We hank an anonymous e iewe o his commen . 13 The esul s o his es ima ion a e a ailable om he au ho s on eques . 33 in enjoying pe o mance ad an ages.14 In his ega d, some s udies ha e a gued ha second en an s equen ly ob ain be e esul s han pionee s, as hey bene i om he expe ience o he pionee in he ma ke and do no con on as much unce ain y as i did (Liebe man and Mon gome y, 1998). To con on his issue, we emo e om ou sample he obse a ions o hose pionee s ha a e ou pe o med by he second en an s in he ma ke . Addi ionally, we also emo e om ou sample hose coun ies whe e pionee s ha e an a e age EBITDA ma gin ha is abo e he a e age EBITDA ma gin o hei i als, ega dless o whe he hey a e second mo e s o la e en an s. The esul s o his es ima ion a e consis en wi h he main indings epo ed in Table 2 and con i m ha ecip ocal MMC has a posi i e e ec on pionee s’ p o i abili y while non- ecip ocal MMC nega i ely de e mines he pe o mance ad an ages o pionee s. Addi ional esul s: The ela ionship be ween MMC and ollowe s’ esul s The aim o his s udy was o analyse he e ec o compe i ion on FMA in an in e na ional con ex h ough he lens o MMC. We ha e al eady shown ha ecip ocal MMC imp o es pionee s’ pe o mance while non- ecip ocal MMC nega i ely a ec s hem. Ne e heless, he p io analysis obse es he e ec s o ecip ocal and non- ecip ocal MMC only in he ma ke s whe e mul ima ke i ms a e pionee s. Ou main p emise is ha mul ima ke i ms end o be pionee s in some ma ke s and ollowe s in o he s. In ac , ecip ocal MMC equi es ha he pionee o he ma ke has he ole o a ollowe in ano he ma ke . Some s udies on MMC ha e sugges ed ha mul ima ke i ms bene i in one ma ke in exchange o ob aining lowe p o i s in o he ma ke s. This means ha mul ima ke pionee s mus sac i ice ma ke s in which hey a e ollowe s o enhance hei esul s in hei sphe es o in luence (McG a h e al., 1998). We al eady know ha ecip ocal MMC imp o es he pe o mance o pionee s in he ma ke s whe e hey enjoy FMA, bu we do no know how i a ec s he pe o mance o hese i ms in 14 Once again, we hank an anonymous e iewe o his commen . 34 he ma ke s whe e hey a e ollowe s. Following p io esea ch, a possibili y is ha mul ima ke pionee s could ob ain wo se esul s in he ma ke s whe e hey a e ollowe s in exchange o be e esul s in he ma ke s whe e hey enjoy FMA. To obse e whe he mul ima ke pionee s mus sac i ice he ma ke s in hose cases in which hey a e ollowe s, we now analyse he e ec s o ecip ocal and non- ecip ocal MMC in he ma ke s whe e mul ima ke i ms a e ollowe s. Table 3 shows he desc ip i e s a is ics o he sample o ollowe s. As shown, 2,982 obse a ions, which belong o 128 mobile ope a o s,15 a e included in ou models. The mean a iable o EBITDAm is 0.2716 wi h a s anda d de ia ion o 0.22. Rega ding he independen a iables, RMMC17 has a mean alue o 0.10. The minimum and maximum alues o his a iable a e 0 and 2.06, espec i ely. The mean alue o NRMMC is 0.44. Once again, he minimum alue is 0 bu he maximum alue is 20.83. As can be seen, ollowe s ha e a g ea e numbe o MMCs han pionee s. This is explained by how we measu e ou a iables o ecip ocal and non- ecip ocal MMC. These a iables di e en ia e be ween pionee s and ollowe s and calcula e he a e age numbe o MMC ha pionee s ( ollowe s) ha e wi h he ollowe s (pionee s) o he ma ke . As he numbe o ollowe s in he ma ke is g ea e han he 15 The 128 mobile ope a o s ha a e included ope a e in 71 coun ies. These coun ies a e he same as in he sample o pionee s wi h he excep ion o Pales ine. In his coun y, wo ope a o s we e iden i ied: Jawwal (PalTel), which was he i s mo e , and Wa aniya Mobile, ha en e ed la e . We ha e in o ma ion abou he i m a iables o his s udy o he pionee bu no o he ollowe . 16 This alue is gla ingly lowe han he mean alue o EBITDAm in he sample o pionee s, which is 0.40. This is consis en wi h he esul ha con i med he exis ence o FMA in ou sample and, consequen ly, suppo s he associa ion be ween i s -mo ing and sphe es o in luence. 17 RMMC and NRMMC we e measu ed in he same way as done in he pionee s’ sample. Howe e , in his case, we obse e i ms ha a e ollowe s in he e e ence ma ke . Following he no a ion ha we ha e used in Figu e 1 and in he a iables’ desc ip ion sec ion, we now ocus on i m j ins ead o ocusing on i m i, as in he p io analysis. Besides his, he same ou condi ions mus be me o conside ha ollowe s ha e ecip ocal o non- ecip ocal MMC wi h he pionee . Tha is, ecip ocal MMC exis s when he ollowe o he ma ke n ( i m j) is he pionee in ma ke m, and he pionee o ma ke n ( i m i) is a ollowe in ma ke m. When his ecip oci y in i ms’ ma ke posi ions does no exis , he ollowe has non- ecip ocal MMC wi h he pionee . Simila o he pionee s’ a iables o MMC, in hose ma ke s whe e he e is mo e han one pionee , we use an a e age measu e. Thus, o cons uc ollowe s’ a iables o MMC, we i s ly add up he o al numbe o ecip ocal (non- ecip ocal) MMCs ha ollowe s ha e wi h he pionee (s). A e wa ds, we di ide his numbe by he o al numbe o pionee s in he ma ke . Ne e heless, since he e is usually only one pionee pe ma ke , in he ollowe s’ sample, he denomina o o he RMMC and NRMMC a iables end o be 1. In ma ke s whe e wo i ms launched mobile se ices a he same ime, bo h a e conside ed as pionee s and he denomina o o he MMC a iables is 2. 35 numbe o pionee s, ollowe s show a g ea e numbe o ecip ocal and non- ecip ocal MMCs han do pionee s. ---------------------------------------------------- Inse Table 3 he e ---------------------------------------------------- Table 4 shows he esul s o ou es ima ion in he sample o ollowe s. Models 1 o 5 p esen he esul s o es ing how ecip ocal MMC and non- ecip ocal MMC a ec hei p o i abili y.18 Model 1 includes only he con ol a iables and he lagged alues o he dependen a iable. Model 2 es s he e ec o ecip ocal MMC. Models 3 and 4 es he lineal and quad a ic e ec s o non- ecip ocal MMC, espec i ely. Finally, Model 5 e e s o he ull model and inco po a es all he a iables oge he . Once again, he esul s o he F- es s a e epo ed a he bo om o he able and show ha Model 5 is p e e ed. We ocus on his model o p esen he esul s o ou es ima es on he sample o ollowe s. ---------------------------------------------------- Inse Table 4 he e ---------------------------------------------------- The coe icien o RMMC is posi i e and s a is ically signi ican (β = 0.0495; p < 0.10). We o e an economic in e p e a ion o his esul by analysing he e ec o one addi ional ecip ocal MMC. On a e age, an inc ease o one ecip ocal MMC wi h he pionee o he ma ke inc eases he ollowe s’ p o i abili y om 0.2686 o 0.3181. This indica es ha ollowe s ha ha e ecip ocal MMC wi h he pionee s o he ma ke ob ain g ea e pe o mance han hose ha compe e agains single-ma ke pionee s. The i al y- educing e ec o ecip ocal MMC also imp o es he esul s ha ollowe s ob ain in he ma ke , e ealing ha mul ima ke i ms do no su e om a educ ion in p o i s in he ma ke s in which hey a e ollowe s. Simila ly, non- ecip ocal MMC also de e mines he p o i s o he ollowe s. Model 5 o Table 4 shows ha NRMMC is nega i e and s a is ically signi ican (β = -0.0459; p < 0.01), and 18 Table 4 eplica es he speci ica ion models o he pionee s’ sample and, he e o e, we include he same a iables as in Table 2. 36 NRMMC2 has a posi i e bu small impac on ollowe s’ p o i abili y (β = 0.000981; p < 0.01). Followe s ha ha e non- ecip ocal MMC wi h he pionee o he ma ke ob ain wo se pe o mance han ollowe s ha compe e wi h his i m only in ha ma ke . Ne e heless, he nega i e e ec s o ha ing non- ecip ocal MMC may disappea when he numbe o hese con ac s wi h he pionee inc eases, as sugges ed by he posi i e e ec o he quad a ic e m. To con i m whe he his cu ilinea e ec akes place, we depic he ela ionship be ween non- ecip ocal MMC and ollowe s’ p o i abili y in Figu e 2. As can be seen, he p o i abili y o ollowe s dec eases as he amoun o non- ecip ocal MMC wi h he pionee s inc eases. Al hough he slope o he cu e la ens a high le els o non- ecip ocal MMC be ween pionee s and ollowe s, Figu e 2 does no show a posi i e in luence o non- ecip ocal MMC on ollowe s’ p o i abili y.19 I e eals ha an inc ease o one non- ecip ocal MMC be ween he ollowe and he pionee makes he a e age ollowe pe o mance dec ease om 0.2686 o 0.2227. ---------------------------------------------------- Inse Figu e 2 he e ---------------------------------------------------- The e ec o MMC on pionee s’ and ollowe s’ esul s Ou indings show ha pionee s ha enjoy FMA and ha e ecip ocal MMC wi h hei ollowe s ob ain g ea e p o i s han hose ha compe e wi h single-ma ke ollowe s. Simila ly, ecip ocal MMC also bene i s hese i ms in he ma ke s whe e hey a e ollowe s, allowing hem o ha e highe p o i s han ollowe s ha compe e wi h single-ma ke pionee s. Non- ecip ocal MMC is also ele an o explain he magni ude o FMA, bu i s e ec on he pe o mance o mul ima ke pionee s is nega i e. In acco dance wi h his esul , ollowe s ha 19 We calcula ed he in lec ion poin o NRMMC. I is loca ed a 23.44 non- ecip ocal mul ima ke con ac s wi h he pionee s o he ma ke . As shown in Table 3, he maximum alue o NRMMC ha ollowe s ha e wi h pionee s is 20.83. Thus, we canno con i m he posi i e e ec o NRMMC on ollowe s’ pe o mance bu only i s nega i e e ec (Haans e al., 2016). 37 ha e non- ecip ocal MMC wi h he pionee s ob ain wo se pe o mance han hose ha compe e only wi h he pionee s in one ma ke . The ques ion ha a ises a his poin is whe he he e ec o ecip ocal and non- ecip ocal MMC on i ms’ p o i abili y is s onge in he ma ke s whe e i ms a e pionee s han in he ma ke s whe e hey a e ollowe s. This poin is impo an , as i will de e mine whe he pionee s ha enjoy FMA a e able o main ain he pe o mance gap be ween hei esul s and hose o he pionee s o , con a ily, ha his gap is smalle o e ime. Tables 3 and 4 show ha he ma ginal e ec s o ecip ocal MMC and non- ecip ocal MMC a e bigge o pionee s han o ollowe s. This sugges s ha ecip ocal MMC and non- ecip ocal MMC ha e a s onge e ec on he p o i abili y ha mul ima ke i ms ob ain in he ma ke s whe e hey a e i s mo e s. The e ec s o MMC a e sp ead ac oss all he ma ke s in which mul ima ke i ms compe e. Ne e heless, hese e ec s a e s onge in he ma ke s whe e mul ima ke i ms a e i s mo e s. To illus a e his, Figu e 3 depic s he e ec s o ecip ocal MMC on pionee s’ and ollowe s’ p o i abili y. The solid line e e s o pionee s and he do ed line o ollowe s. As can be seen, he solid line is abo e he do ed line. This means ha mul ima ke pionee s bene i mo e om ha ing ecip ocal MMC wi h hei ollowe s han do mul ima ke ollowe s wi h he pionee s o he ma ke . Thus, pionee s ha enjoy FMA bene i mo e om his kind o MMC in he ma ke han hei mul ima ke ollowe s and, consequen ly, he pe o mance gap be ween hei esul s is inc eased by ecip ocal MMC. Figu e 4 depic s he e ec s o non- ecip ocal MMC on pionee s’ and ollowe s’ p o i abili y. In his case, he do ed line, which e e s o ollowe s, is abo e he solid line ha e e s o pionee s. Followe s ecei e he nega i e e ec s o non- ecip ocal MMC less han do pionee s. This esul sugges s ha non- ecip ocal MMC is a ele an ac o o explain he e osion o FMA o e ime. Al hough non- ecip ocal MMC makes bo h pionee s and ollowe s ob ain lowe p o i s in he ma ke , he nega i e e ec s a e s onge o pionee s. 38 ---------------------------------------------------- Inse Figu e 3 he e ---------------------------------------------------- ---------------------------------------------------- Inse Figu e 4 he e ---------------------------------------------------- DISCUSSION AND CONCLUSIONS Despi e he impo ance o compe i ion on FMA, ew empi ical s udies ha e add essed he ela ionship be ween compe i ion and hese ad an ages. Conside ing ha economic libe alisa ion and globalisa ion ha e con e ed MMC in o a common cha ac e is ic in an a ay o indus ies, i is su p ising ha he s a egic managemen li e a u e has no included mul ima ke compe i ion in i s analysis o en y iming. In ac , o ou knowledge, his is he i s pape ha analyses FMA om a MMC pe spec i e. We link he FMA and MMC li e a u es h ough he concep o sphe es o in luence. By linking he wo heo ies, we imp o e he analysis o FMA by conside ing how he compe i ion be ween pionee s and ollowe s in di e en ma ke s de e mines he magni ude o hese ad an ages. Ou indings show ha ecip ocal MMC imp o es he p o i abili y o pionee s ha enjoy FMA, explaining why some o hem ob ain bigge ad an ages han o he s. The p esence o pionee s in he ma ke s, whe e hei ollowe s a e i s mo e s, makes ollowe s conscious o he pionee s’ capabili y o e alia e. I mul ima ke ollowe s make an agg essi e mo emen in a ma ke , such as a p ice cu o a new p oduc de elopmen , he pionee can espond wi h he same mo emen in he ma ke whe e mul ima ke ollowe s enjoy FMA. This leads mul ima ke ollowe s o espec he ma ke whe e hei i als enjoy FMA, in exchange o he same ea men in hei own impo an ma ke s. As ollowe s educe hei compe i i e ac ions in he ma ke s, he p o i abili y o pionee s ha enjoy FMA inc eases. We also show ha ecip ocal MMC bene i s mul ima ke i ms no only in he ma ke s whe e hey a e pionee s bu also in hose ma ke s whe e hey a e ollowe s. This esul is impo an because i shows ha he imp o emen in p o i abili y ha pionee s enjoy in he ma ke s in 39 which hey ha e FMA is no main ained a he expense o impo an educ ions in p o i abili y in he ma ke s in which hey ollow. In o he wo ds, he e ec s o mu ual o bea ance a e el in all he ma ke s whe e mul ima ke compe i o s mee and no only in hose ha a e mo e impo an o hem. This esul sugges s ha he h ea o e alia ion is enough o enac mu ual o bea ance in all he ma ke s in which i ms ope a e (Fe nández and Ma ín, 1998). Addi ionally, i also e eals ha ollowe s bene i om ha ing ecip ocal MMC and shows ha his ype o MMC does no p oduce a ze o-sum esul whe e one i m gains and he o he s lose. On he con a y, e e y i m in he ma ke can bene i om a educ ion in compe i ion. Simila ly, he e ec o MMC is also signi ican in all he ma ke s in which i ms pa icipa e in a non- ecip ocal mul ima ke ela ionship. Howe e , in con as o he p io si ua ion, he e ec o non- ecip ocal MMC is nega i e o bo h pionee s and ollowe s. Thus, ou esea ch shows ha each ype o MMC has a highly di e en e ec on i ms’ p o i abili y and con ibu es o p io s udies on his issue (Gimeno, 1999). Rega ding he pionee s, an explana ion o he nega i e e ec o non- ecip ocal MMC is ha he lack o ecip oci y hinde s hei capabili y o exe a c edible e alia o y h ea on hei mul ima ke ollowe s.20 Ins ead o c ea ing incen i es o collude, non- ecip ocal MMC poses a g ea e compe i i e ension be ween pionee s ha enjoy FMA and hei ollowe s, which e eals i sel as an inc ease in compe i i e ac ions in he ma ke and, he e o e, as a dec ease in he magni ude o FMA. These e ec s o non- ecip ocal MMC a e no limi ed o a speci ic ma ke bu a e sp ead ac oss all he ma ke s whe e mul ima ke compe i o s mee . As a consequence, ollowe s ha ha e non- ecip ocal MMC wi h pionee s ob ain lowe p o i s esul s han do single-ma ke ollowe s. Finally, ou s udy also sugges s ha ecip ocal and non- ecip ocal MMC does no a ec i ms’ p o i abili y in he same way wi hin all hei ma ke s. On he con a y, ou indings sugges ha he e ec s o bo h ecip ocal and non- ecip ocal MMC a e s onge in he ma ke s whe e 20 This could also imply ha he numbe o mul ima ke con ac s in ou sample is no high enough o gene a e mu ual o bea ance. 40 mul ima ke i ms a e pionee s. Mul ima ke i ms bene i mo e om ha ing ecip ocal MMC in he ma ke whe e hey enjoy FMA. Simila ly, non- ecip ocal MMC has a s onge nega i e e ec on mul ima ke i ms’ p o i abili y when hey pe o m as pionee s. An explana ion o his esul can be ound in he di e ences be ween mul ima ke i ms’ in e es s in hei common ma ke s. Ou esul sugges s ha MMC allows mul ima ke i ms o edis ibu e hei p o i s among he ma ke s whe e hey a e ope a ing acco ding o hei own in e es s (Fe nández and Ma ín, 1998). As mul ima ke pionee s p e e o imp o e FMA han ob aining highe p o i s in ma ke s whe e hey a e ollowe s, hey ac consequen ly and edis ibu e hei p o i s in his way. Ou s udy signi ican ly con ibu es o he FMA and MMC li e a u es. Fi s , we con i m he con enience o linking MMC and FMA and show ha mul ima ke compe i ion is an impo an con ingen ac o ha explains FMA. We, hus, con ibu e o he iden i ica ion o he bounda y condi ions su ounding FMA. Addi ionally, his in eg a ion is highly ad isable because he cu en compe i i e landscape is cha ac e ised by i ms ha o en expand ab oad (Hi e al., 2016), so ha pionee s migh encoun e he same i als in di e en coun ies (Gimeno and Jeong, 2001). The heo y o MMC helps us o unde s and ac ions be ween pionee s and ollowe s ha , om a single-ma ke pe spec i e, would no make sense (McG a h e al., 1998). Mo e impo an ly, wi hou he MMC pe spec i e, he analysis o compe i ion on FMA migh lead o misleading esul s by a ibu ing o he FMA isola ing mechanisms ha a e he esul o compe i ion. Second, we also con i m he impo ance o dis inguishing be ween ecip ocal and non- ecip ocal MMC. Al hough p io esea ch has al eady highligh ed his and concluded ha ecip ocal MMC has a s onge e ec on compe i ion (Gimeno, 1999), his s udy goes beyond ha and shows ha each ype o MMC has a di e en e ec on i m p o i abili y. While ecip ocal MMC leads o a di ec educ ion on compe i ion—and, consequen ly, o an 41 imp o emen in pe o mance—non- ecip ocal MMC makes mul ima ke i ms ini ially ha e incen i es o compe e ins ead o o bea . Thi d, we ex end bo h he FMA and MMC li e a u es o an in e na ional se ing. In ela ion o he FMA li e a u e, his is impo an , as pionee s ace di e en condi ions in each coun y ha signi ican ly explain di e ences in he magni ude o hei p o i s (Dyke and Kole , 2018). Conside ing 72 coun ies allow us o b oaden ou knowledge by be e comp ehending FMA in an in e na ional se ing. This is impo an , as coun ies di e in some espec s ha ha e been sugges ed o a ec p o i abili y in an in e na ional se ing, such as di e ences in economic o ins i u ional ac o s (Goldszmid e al., 2011). Mo e speci ically, he ins i u ional se ing has been p oposed as an in e es ing a enue o analysing FMA in an in e na ional se ing (Fos u i e al., 2013). In he case o MMC, esea che s ha e highligh ed ha i ms need o coo dina e hei business uni s o bene i om mu ual o bea ance (Jayachand an e al., 1999). Coo dina ion is much mo e di icul in an in e na ional con ex because i ms ha e o deal wi h di e en ma ke condi ions ha make hem be ‘adap i e o he speci ic equi emen s o di e en coun y ma ke s’ (Yu e al., 2009: 129). Ou esul s show ha , besides hese coo dina ion challenges, he mu ual o bea ance akes place in an in e na ional se ing when i ms ha e ecip ocal MMC. Ne e heless, we do no ind suppo o he mu ual o bea ance in an in e na ional con ex when i ms ha e non- ecip ocal MMC wi h hei i als. This is in e es ing and migh indica e ha he challenges o coo dina ion ac oss in e na ional ma ke s migh be emphasised when MMC is non- ecip ocal. This s udy also has some manage ial implica ions. We add new e idence on he e ec s ha he iming o en y in o he ma ke ha e on i ms’ pe o mance. As p io s udies ha e al eady shown, manage s should conside ha he iming o en y has impo an e ec s on he 48 Shamsie, J., Phelps, C., Kupe man, J., 2004. Be e la e han ne e : a s udy o la e en an s in household elec ical equipmen . S a egic Managemen Jou nal, 25(1), 69–84. Shapi a, Z., 2017. En e ing new ma ke s: The e ec o pe o mance eedback nea aspi a ion and well below and abo e i . S a egic Managemen Jou nal, 38(7), 1416-1434. Schnaa s, S.P., 1994. Managing imi a ion s a egies: How la e en an s seize ma ke ing om pionee s. New Yo k: The F ee P ess. Smi h, F.L., Wilson, R.L., 1995. The p edic i e alidi y o he Ka nani and We ne el model o mul ipoin compe i ion. S a egic Managemen Jou nal, 16(2), 143–160. Song, X.M., Di Benede o, C.A., Zhao, Y.L., 1999. Pionee ing ad an ages in manu ac u ing and se ice indus ies: empi ical e idence om nine coun ies. S a egic Managemen Jou nal, 20(9), 811–836. So enson, T.L., 2007. C edible collusion in mul ima ke oligopoly. Manage ial and Decision Economics, 28(2), 115–128. Spagnolo, G., 1999. On in e dependen supe games: mul ima ke con ac , conca i y and collusion. Jou nal o Economic Theo y, 89(1), 127–139. S e n, I., James, S.D., 2016. Whom a e you p omo ing? Posi i e olun a y public disclosu es and execu i e u no e . S a egic Managemen Jou nal, 37(7), 1413-1430. Suá ez, F.F., Lanzolla, G., 2007. The ole o en i onmen al dynamics in building a i s mo e ad an age heo y. Academy o Managemen Re iew, 32(2), 377–392. Vande we , P., Mahon, J.F., 1997. Me a-analysis o he impac o esea ch me hods on indings o i s - mo e ad an ages. Managemen Science, 43(11), 1510–1519. Vidal, E., Mi chell, W., 2013. When do i s en an s become i s su i o s? Long Range Planning, 46(4-5), 335-347. Wa e man, L., Kou oumpis, P., 2011. Benchma king elecoms egula ion – The Telecommunica ions Regula o y Go e nance Index (TRGI). Telecommunica ions Policy, 35(5), 450-468. Wiggins, R.R., Rue li, T.W., 2002. Sus ained compe i i e ad an age: Tempo al dynamics and he incidence and pe sis ence o supe io economic pe o mance, 13(1), 81-105. Young, G., Smi h K.G., G imm, C.M., Simon, D., 2000. Mul ima ke con ac and esou ce dissimila i y: A compe i i e dynamics pe spec i e. Jou nal o Managemen , 26(6), 1217-1236. 49 Yu, T., Canella, J .A.A., 2013. A comp ehensi e e iew o mul ima ke compe i ion esea ch. Jou nal o Managemen , 39(1), 76–109. Yu, T., Sub amaniam, M., Canella, J . A.A., 2009. Ri al y de e ence in in e na ional ma ke s: con ingencies go e ning he mu ual o bea ance hypo hesis. Academy o Managemen Jou nal, 52(1), 127–147. Zacha y, M.A., Gianiodis, P.T., Payne, G.T., Ma kman, G.D., 2015. En y iming: Endu ing lessons and u u e di ec ions. Jou nal o Managemen , 41(5), 1388-1415. Figu e 1. Di e en si ua ions o MMC be ween pionee s and ollowe s Ma ke n Ma ke m A. Ma ke n Ma ke m Ma ke n Ma ke m B.1 B.2. Sou ce: Own elabo a ion based on Gimeno (1999) i n j m im jn i n im jn jm i n i m jn jm 51 Table 1. Desc ip i e s a is ics Va iable N Mean Sd. Min. Max. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) EBITDAm (1) 2,094 0.40 0.110 -0.42 0.85 1 RMMC (2) 2,094 0.07 0.17 0 1.10 -0.01 1 NRMMC (3) 2,094 0.20 0.42 0 3.45 -0.23* 0.39* 1 Fi m size (4) 2,094 14.46 52.24 0.03 649.57 0.13* -0.02 -0.05* 1 Ma ke impo ance (5) 2,094 0.53 0.39 0.005 1 0.07* -0.18* -0.29* 0.18* 1 Incumben (6) 2,094 0.49 0.50 0 1 0.01 -0.02 -0.18* -0.11* 0.29* 1 To al ma ke s (7) 2,094 7.58 7.07 1 30 -0.09* 0.36* 0.41* -0.09* -0.70* -0.23* 1 Numbe o i ms (8) 2,094 3.79 1.50 1 11 -0.07* 0.02 0.02 -0.04 -0.08* -0.26* 0.13* 1 Ma ke size (9) 2,094 54.47 189.37 0.31 1,372.75 0.19* -0.05* -0.06* 0.86* 0.19* -0.17* -0.10* -0.03 1 GDP pe capi a (10) 2,094 0.14 0.13 0.0012 0.72 -0.23* 0.22* 0.07* -0.08* -0.08* 0.20* 0.07* 0.11* -0.14* 1 P oximi y (11) 2,094 0.65 0.48 0 1 -0.007 0.16* -0.16* -0.13* -0.36* -0.05* 0.18* 0.19* -0.18* 0.32* 1 MNP (12) 2,094 0.55 0.50 0 1 -0.18* 0.19* 0.04 -0.10* -0.14* 0.17* 0.05* 0.14* -0.18* 0.52* 0.32* 1 Signi icance le el: * p<0.05 52 Table 2. The ela ionship be ween MMC and pionee s’ esul s (1) (2) (3) (4) (5) EBITDAma gin EBITDAma gin EBITDAma gin EBITDAma gin EBITDAma gin L4.EBITDAma gin 0.00390* 0.00459** 0.00309 0.00310 0.00443** (1.87) (2.24) (1.48) (1.49) (2.19) Fi m size -0.000223** -0.000234*** -0.000206** -0.000207** -0.000224** (-2.45) (-2.71) (-2.15) (-2.16) (-2.47) Ma ke impo ance -0.00128 -0.00538 -0.00477 -0.00499 -0.0133 (-0.03) (-0.13) (-0.12) (-0.12) (-0.34) Incumben 0.0265 0.0273 0.0195 0.0197 0.0187 (1.25) (1.35) (0.94) (0.95) (0.99) To al ma ke s -0.000835 -0.00138 0.000172 0.000183 -0.000671 (-0.46) (-0.73) (0.10) (0.10) (-0.37) Numbe o i ms 0.00108 0.00150 0.000159 0.0000838 0.000700 (0.18) (0.26) (0.03) (0.01) (0.12) Ma ke size 0.000167*** 0.000169*** 0.000155*** 0.000155*** 0.000156*** (6.59) (7.02) (5.87) (5.87) (6.16) GDP pe capi a -0.154** -0.162** -0.133* -0.135* -0.141* (-2.03) (-2.07) (-1.75) (-1.70) (-1.78) P oximi y 0.0314 0.0299 0.0190 0.0186 0.0132 (1.14) (1.08) (0.70) (0.66) (0.48) MNP -0.0204 -0.0220 -0.0186 -0.0187 -0.0214 (-0.98) (-1.08) (-0.92) (-0.92) (-1.09) RMMC 0.0447 0.0910** (1.16) (2.30) NRMMC -0.0472** -0.0443* -0.0634** (-2.53) (-1.65) (-2.28) NRMMC2 -0.000183 0.000217 (-0.20) (0.25) Yea ly dummies Yes Yes Yes Yes Yes Qua e ly dummies Yes*** Yes*** Yes*** Yes*** Yes*** _cons 0.380*** 0.385*** 0.395*** 0.396*** 0.408*** (8.37) (8.42) (8.62) (8.55) (8.79) N 2,094 2,094 2,094 2,094 2,094 a 1 -3.292 -3.288 -3.314 -3.314 -3.314 a 1p 0.000996 0.00101 0.000919 0.000919 0.000921 a 2 0.710 0.750 0.690 0.690 0.807 a 2p 0.478 0.453 0.490 0.490 0.420 hansen 4.269 4.243 4.010 4.009 3.879 hansenp 0.511 0.515 0.548 0.548 0.567 F- es s 1 1.35 6.42** 7.42** 10.46** F- es s 2 8.95** F- es s 3 0.04 5.41* F- es s 4 5.31** s a is ics in pa en heses; Signi icance le els: * p < 0.10, ** p < 0.05, *** p < 0.01 53 Table 3. Desc ip i e s a is ics o he ollowe s’ sample Va iable N mean sd min max (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) EBITDAm (1) 2,982 0.27 0.22 -0.98 0.77 1 RMMC (2) 2,982 0.10 0.30 0 2.06 0.12* 1 NRMMC (3) 2,982 0.44 1.78 0 20.83 0.08* 0.62* 1 Fi m size (4) 2,982 7.85 16.59 0.035 199.66 0.19* 0.11* 0.10* 1 Ma ke impo ance (5) 2,982 0.49 0.42 0.007 1 0.13* -0.06* -0.07* 0.17* 1 Incumben (6) 2,982 0.19 0.40 0 1 0.20* 0.17* 0.28* 0.03 -0.01 1 To al ma ke s (7) 2,982 6.79 6.69 1 30 0.002 0.17* 0.17* -0.01 -0.70* 0.11* 1 Numbe o i ms (8) 2,982 4.02 1.48 2 11 -0.09* -0.09* -0.09* -0.02 0.14* -0.11* -0.11* 1 Ma ke size (9) 2,982 49.20 159.46 0.30 1,372.75 0.10* -0.01 -0.02 0.86* 0.18* -0.05* -0.10* -0.05* 1 GDP pe capi a (10) 2,982 0.16 0.13 0.001 0.74 -0.05* 0.05* -0.09* -0.03 0.02 -0.15* 0.07* 0.03 -0.11* 1 P oximi y (11) 2,982 0.54 0.50 0 1 0.03 0.30* 0.21* -0.07* -0.46* -0.02 0.27* -0.20* -0.15* 0.20* 1 MNP (12) 2,982 0.60 0.49 0 1 -0.04* 0.06* -0.06* -0.02 0.02 -0.13* 0.05* 0.09* -0.13* 0.56* 0.18* 1 Signi icance le el: * p<0.05 54 Table 4. The ela ionship be ween MMC and ollowe s’ esul s (1) (2) (3) (4) (5) EBITDAm EBITDAm EBITDAm EBITDAm EBITDAm L4.EBITDAma gin 0.0705*** 0.0705*** 0.0705*** 0.0700*** 0.0702*** (26.86) (26.85) (27.09) (26.52) (26.50) Fi m size 0.00446*** 0.00433*** 0.00474*** 0.00523*** 0.00507*** (4.35) (4.19) (4.56) (4.93) (4.75) Ma ke impo ance 0.0701*** 0.0675** 0.0749*** 0.0596** 0.0553* (2.62) (2.40) (2.80) (2.07) (1.85) Incumben 0.0507** 0.0494** 0.0586*** 0.0658*** 0.0663*** (2.38) (2.33) (2.65) (2.97) (2.90) To al ma ke s 0.00298** 0.00285* 0.00329** 0.00275* 0.00253 (2.02) (1.85) (2.17) (1.77) (1.61) Numbe o i ms -0.0127*** -0.0125*** -0.0135*** -0.0122*** -0.0123*** (-2.74) (-2.71) (-2.94) (-2.64) (-2.65) Ma ke size -0.000344*** -0.000333*** -0.000369*** -0.000403*** -0.000390*** (-3.46) (-3.32) (-3.68) (-3.91) (-3.78) GDP pe capi a -0.109* -0.106 -0.111* -0.0860 -0.0805 (-1.67) (-1.61) (-1.72) (-1.40) (-1.29) P oximi y 0.0222 0.0183 0.0277 0.0351* 0.0290 (1.07) (0.79) (1.34) (1.69) (1.30) MNP -0.0228** -0.0229** -0.0240** -0.0248** -0.0265*** (-2.10) (-2.11) (-2.21) (-2.51) (-2.66) RMMC 0.0202 0.0495* (0.92) (1.81) NRMMC -0.00761*** -0.0411*** -0.0459*** (-3.53) (-7.63) (-8.02) NRMMC2 0.000970*** 0.000981*** (7.95) (8.06) Yea ly dummies Yes*** Yes*** Yes*** Yes*** Yes*** Qua e ly dummies Yes*** Yes*** Yes*** Yes*** Yes*** _cons 0.225*** 0.227*** 0.223*** 0.222*** 0.226*** (6.09) (6.01) (5.98) (5.63) (5.70) N 2,982 2,982 2,982 2,982 2,982 a 1 -2.495 -2.494 -2.499 -2.541 -2.543 a 1p 0.0126 0.0126 0.0125 0.0111 0.0110 a 2 -0.896 -0.896 -0.887 -0.841 -0.838 a 2p 0.370 0.370 0.375 0.400 0.402 hansen 53.35 53.41 53.58 51.71 51.97 hansenp 0.243 0.242 0.236 0.295 0.286 F- es s 1 0.85 12.49*** 63.47*** 70.33*** F- es s 2 69.78*** F- es s 3 63.21*** 69.58*** F- es s 4 69.78*** s a is ics in pa en heses; Signi icance le els: * p < 0.10, ** p < 0.05, *** p < 0.01 Figu e 2. The e ec o non- ecip ocal MMC on ollowe s’ esul s Figu e 3. The e ec o ecip ocal MMC on pionee s’ and ollowe s’ esul s -0,6 -0,5 -0,4 -0,3 -0,2 -0,1 00246810 12 14 16 18 20 22 EBITDA ma gin NRMMC 0 0,05 0,1 0,15 0,2 0,25 00,2 0,4 0,6 0,8 11,2 1,4 1,6 1,8 22,2 EBITDA ma gin RMMC Pionee s Followe s 56 -1,6 -1,4 -1,2 -1 -0,8 -0,6 -0,4 -0,2 00246810 12 14 16 18 20 22 EBITDA ma gin NRMMC Pionee s Followe s Figu e 4. The e ec o non- ecip ocal MMC on pionee s’ and ollowe s’ esul s