Does independent directors' monitoring affect reputation? Evidence from the stock and labor markets

Du, Jun; Hou, Qingchuan; Tang, Xuesong; Yao, Yiwei A icle Does independen di ec o s' moni o ing a ec epu a ion? E idence om he s ock and labo ma ke s China Jou nal o Accoun ing Resea ch P o ided in Coope a ion wi h: Sun Ya -sen Uni e si y Sugges ed Ci a ion: Du, Jun; Hou, Qingchuan; Tang, Xuesong; Yao, Yiwei (2018) : Does independen di ec o s' moni o ing a ec epu a ion? E idence om he s ock and labo ma ke s, China Jou nal o Accoun ing Resea ch, ISSN 1755-3091, Else ie , Ams e dam, Vol. 11, Iss. 2, pp. 91-127, h ps://doi.o g/10.1016/j.cja .2018.01.002 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/241773 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ Does independen di ec o s’ moni o ing affec epu a ion? E idence om he s ock and labo ma ke s q Jun Du a , Qingchuan Hou b , Xuesong Tang c , Yiwei Yao d,⇑ a School o Accoun ing and Finance, The Hong Kong Poly echnic Uni e si y, Hong Kong b School o Accoun ancy, Shanghai Uni e si y o Finance & Economics, China c School o Accoun ing, Sou hwes e n Uni e si y o Finance and Economics, China d Depa men o Accoun ancy, Hang Seng Managemen College, Hong Kong ARTICLE INFO A icle his o y: Recei ed 10 Oc obe 2017 Accep ed 11 Janua y 2018 A ailable online 3 Ma ch 2018 ABSTRACT Using no el da a on independen di ec o s’ opinions in China, we in es iga e he s ock and labo ma ke effec s p omp ed by independen di ec o s publicly saying ‘‘no” o majo boa d decisions. We find ha he ma ke eac s nega- i ely o modified di ec o opinions, bu posi i ely o fi ms in e locked wi h he di ec o s who said ‘‘no.”We u he find subs an ial u no e and decline in boa d sea s a e independen di ec o s issue modified opinions. O e all, we iden i y a dilemma in China whe eby he labo ma ke does no ewa d igilan di ec o s o s anding up o fi m inside s, al hough in es o s add a p emium o effec i e boa d moni o ing. Ó2018 Sun Ya -sen Uni e si y. P oduc ion and hos ing by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecom- mons.o g/licenses/by-nc-nd/4.0/). 1. In oduc ion The epu a ion- ela ed consequences o independen di ec o s a e an issue o conside able public and aca- demic in e es . Independen di ec o s a e widely belie ed o play an impo an ole in co po a e go e nance. Fama and Jensen (1983) sugges ha independen di ec o s can make dis inc con ibu ions in aligning manage s wi h he in e es s o s ockholde s. The mos significan incen i e o independen di ec o s o pu sue h ps://doi.o g/10.1016/j.cja .2018.01.002 1755-3091/Ó2018 Sun Ya -sen Uni e si y. P oduc ion and hos ing by Else ie B.V. 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/). ⇑ Co esponding au ho . E-mail add esses: [email p o ec ed] (J. Du), [email protected] e.edu.cn (Q. Hou), angxs@swu e.edu.cn (X. Tang), [email p o ec ed] (Y. Yao). q We hank he 2017 Symposium o China Jou nal o Accoun ing Resea ch o i s help ul commen s and g a e ully acknowledge he financial suppo o he Na ional Na u al Science Founda ion o China (P ojec s 71,372,206 and 71,672,152) and he suppo o he P og am o New Cen u y Excellen Talen s in Uni e si y (NCET-13–0963). We a e esponsible o any e o s o omissions. China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 HOSTED BY Con en s lis s a ailable a ScienceDi ec China Jou nal o Accoun ing Resea ch jou nal homepage: www.else ie .com/loca e/cja effec i e boa d moni o ing is o de elop hei epu a ions as decision expe s. Al hough Fama and Jensen (1983) indica e ha independen di ec o s’ epu a ions hold posi i e alue, i is difficul o empi ically e alua e such epu a ions, as fi ms a ely disclose he ac i i ies o indi idual di ec o s in he boa d oom. The e o e, many s udies ha e in e ed he alue o independen di ec o s’ epu a ions by in es iga ing he s ock o di ec- o labo ma ke esponses in he wake o ex ao dina y e en s. Fo example, fi ms wi h inside -domina ed boa ds a e mo e likely o con on financial dis ess (Gilson, 1990; Ha o d, 2003), epo ea nings es a e- men s (S ini asan, 2005), ace class-ac ion lawsui s (Fich and Shi dasani, 2007) and announce di ec o u n- o e s (Fahlenb ach e al., 2010; Ag awal and Chen, 2011). Mos o hese s udies ha e ound ha on a e age in es o s eac nega i ely o hese ad e se signals o weak boa d moni o ing and ha independen di ec o s suffe epu a ional penal ies i hey do no igilan ly moni o op managemen . Al hough his line o esea ch p o ides insigh s in o he ex pos se ling-up mechanism o di ec o s’ labo ma ke , i is limi ed in i s effo s o impu e hese co po a e ailu es o poo boa d moni o ing by independen di ec o s (Richa dson, 2005). Mo e impo an ly, in he as majo i y o fi ms ha do no ace hese c ises, di ec e idence o how independen di ec o s help o o e see boa ds emains scan . We exploi a unique da ase o independen di ec o s’ moni o ing effo s o unco e he co esponding effec on independen di ec o s’ epu a ions and ca ee p ospec s. China’s ma ke egula o , he China Secu- i ies Regula o y Commission (CSRC), equi es independen di ec o s o publicly disclose hei opinions on impo an manage ial decisions. Such di ec o s a e expec ed o say ‘‘no”in boa d mee ings i hey belie e ha a boa d p oposal is no in he in e es s o sha eholde s. Fu he mo e, an independen di ec o ’s opinion no only se es as a signal o he ex e nal ma ke ega ding his/he moni o ing effo s, bu also may gi e di ec o s legal elie om po en ial lawsui s. Thus, he unique disclosu e o independen di ec o s’ moni o ing ac i i ies in he Chinese ma ke sheds ligh on he ‘‘black box”o boa d unc ion. We use his con ex o in es iga e he consequences o he boa d dispu es ha a ise when independen di ec- o s say ‘‘no.”Fi s , we s udy whe he Chinese in es o s alue independen di ec o s’ moni o ing effo s. We expec he ma ke o in e p e independen di ec o s s anding up o fi m inside s abou any de ec ed i egu- la i ies as nega i e news, as modified opinions disclose boa d dispu es ha may be unobse able o he wise. Ou ocus is he analysis o he s ock ma ke eac ion o independen di ec o s’ opinions ega ding di ec o -in e locked fi ms. Unde he epu a ion hypo hesis, he ma ke should espond posi i ely o fi ms in e locked wi h di ec o s who say ‘‘no,”gi en ha he announcemen o a modified di ec o opinion signals effec i e moni o ing by he opinion-issuing di ec o s. In con as , he endogenous hypo hesis p edic s a neg- a i e ma ke esponse o di ec o -in e locked fi ms, as fi ms ha sha e he same independen di ec o s may e eal p oblems ha a e simila o hose o he opinion- ecei ing fi ms. Second, o examine he weal h effec s o indi idual independen di ec o s, we analyze he changes in boa d sea s o opinion- ecei ing fi ms and o he di ec o -in e locked fi ms. I independen di ec o s de elop a epu a ion o independen moni o ing by saying ‘‘no” o con olling sha eholde s o op execu i es, hen acco ding o Fama and Jensen (1983) hey a e mo e likely o keep hei boa d sea s in di ec o -in e locked fi ms o gain mo e boa d sea s a e hey issue nega i e commen s on co po a e decisions. We find ha fi ms ecei ing modified di ec o opinions sus ain nega i e cumula i e abno mal e u ns (CARs), whe eas in e locked fi ms exhibi posi i e e u ns a ound he announcemen s o modified indepen- den di ec o opinions. The e is also a significan u no e o independen di ec o s who say ‘‘no”in opinion- ecei ing fi ms subsequen o he issuance o modified independen di ec o s’ opinions. Con a ily, he e idence shows ha hese di ec o s end o lose mo e di ec o ships. Ou findings a e o some ex en con- sis en wi h he di ec o epu a ion hypo hesis, which sugges s ha Chinese in es o s alue effec i e moni o - ing conduc ed by independen di ec o s. In he long un, howe e , igilan di ec o s a e no ewa ded o hei epu a ion o effec i e moni o ing, as Fama and Jensen (1983) sugges . One possible explana ion is ha Chi- nese fi ms may sel -selec lax moni o ing o easy manage ial con ol and manipula ion. Ou s udy is closely ela ed o Jiang e al. (2016), who examine he associa ion be ween independen di ec- o s’ p opensi y o issue modified opinions and hei ca ee conce ns and media co e age. Ou s udy diffe s om Jiang e al. (2016) in wo impo an ways. Fi s , ou esul s shed ligh on he dilemma ha al hough in es o s alue effec i e boa d o e sigh , independen di ec o s’ moni o ing effo s a e no ewa ded by he di ec o labo ma ke in China, whe e con olling sha eholde s o en de e mine who can si on he boa d. Sec- ond, ou findings help diffe en ia e he al e na i e hypo heses p oposed by Fich and Shi dasani (2007) by 92 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 showing ha he ma ke may eac diffe en ly in opinion- ecei ing fi ms and di ec o -in e locked fi ms when an independen di ec o says ‘‘no.” We con ibu e o he li e a u e in se e al ways. Fi s , we help disen angle he h ee hypo heses p oposed by Fich and Shi dasani (2007). In con as o mos p e ious s udies o di ec o epu a ion, we e alua e Chinese independen di ec o s’ epu a ions by ocusing on hei modified opinions o co po a e decisions, a posi i e signal o effec i e moni o ing. Using his no el da a, we abs ac he legal liabili y hypo hesis and conduc a cleane es o diffe en ia e be ween di ec o epu a ion and he endogenous hypo heses. Second, we cla i y he ole o independen di ec o s’ epu a ions in eme ging ma ke s. In a ela ed s udy, Tang e al. (2013) find ha independen di ec o s’ opinions help o mi iga e he agency cos s o lis ed Chinese fi ms. Ou analysis complemen s his finding by showing ha al hough he s ock ma ke alues effec i e boa d o e sigh , igilan di ec o s do no ecei e ewa ds om he di ec o labo ma ke in China, by which con ol- ling sha eholde s can de e mine who is in i ed o he boa d. Thi d, ou findings in he Chinese con ex may ha e implica ions o egula o s who in end o imp o e he anspa ency and effec i eness o boa d go e nance. As he majo i y o fi ms in eme ging ma ke s a e dom- ina ed by con olling sha eholde s, which may de e mine he sla e o di ec o s, he di ec o labo ma ke he ein may d i e ou independen di ec o s who da e o s and up o fi m inside s in boa d mee ings. Wi hou concu en imp o emen in ma ke ins i u ions and go e nance s uc u es, such as in he p ocedu e o nom- ina ing and selec ing di ec o s, he expec ed effec o disclosing independen di ec o s’ opinions can ha dly be sus ained in he long un. This pape con inues as ollows. Sec ion 2in oduces he ins i u ional backg ound and de elops he hypo heses. Sec ion 3desc ibes he sample and a iables. Sec ions 4 and 5 p o ide he esul s. Sec ion 6p e- sen s he obus ness checks. Sec ion 7concludes he pape . 2. Ins i u ional backg ound and hypo hesis de elopmen 2.1. Ins i u ional backg ound In esponse o he g ea expec a ions o he imp o emen o co po a e go e nance and he mi iga ion o unneling among con olling sha eholde s, he independen di ec o sys em was in oduced in China in he 1990s. In he Guidelines o he a icles o associa ion o lis ed companies (CSRC, 1997), he CSRC sugges s o he fi s ime ha lis ed fi ms appoin independen di ec o s a hei disc e ion. Following his sugges ion, qui e a ew Chinese fi ms began o es ablish boa d posi ions o independen di ec o s, pa icula ly fi ms seek- ing o lis in o e seas ma ke s. Howe e , a wa e o financial scandals in he la e 1990s in fi ms ha had appoin ed independen di ec o s igge ed public ou c y abou he effec i eness o independen moni o ing in China. 1 The independen di ec o sys em was c i icized as being a ‘‘flowe ase”in he boa d oom, me ely deco a i e and o li le use in imp o ing co po a e go e nance. To ebuild ma ke confidence in independen boa d moni o ing, he CSRC published he Guiding opinion on es ablishing an independen di ec o sys em in lis ed companies in 2001, manda ing g ea e anspa ency in he moni o ing o boa d and manage ial decisions (CSRC, 2001). Acco ding o he 2001 opinion, by mid- 2003, all lis ed Chinese fi ms we e expec ed o ha e se up an independen di ec o sys em, unde which a min- imum o one hi d o he boa d di ec o s would be independen di ec o s. No ably, he Chinese egula o imposes an addi ional disclosu e equi emen ha diffe s om he boa d egula ions in o he majo capi al ma ke s. Lis ed Chinese fi ms should disclose any e en s o ansac ions ha independen di ec o s belie e may affec he in e es s o mino i y sha eholde s, such as he nomina ion, appoin men and dismissal o di ec- o s and senio execu i es; he emune a ion o di ec o s and senio execu i es; ma e ial in e -co po a e loans o sha eholde s and o he affilia ed en i ies; and o he issues s ipula ed in company cha e p o isions. In addi- ion, he 2001 CSRC opinion classifies independen di ec o s’ opinions in o wo majo ca ego ies: he s anda d clean opinion and he modified di ec o opinion, comp ising qualified and ad e se opinions and disclaime s o 1 Fo example, a lis ed depa men s o e, Zhengzhou Baiwen (s ock ID: 600898), which had appoin ed independen di ec o s as ea ly as 1995, was ound guil y o nume ous financial scandals in 2001. M . Jiahao Lu, he o me independen di ec o o Zhengzhou Baiwen, ecei ed public censu e om he CSRC and was fined RMB100,000 o his negligence in o e seeing he audulen fi m. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 93 opinion. I independen di ec o s disag ee on a pa icula co po a e issue, hen he fi m should disclose hei iews sepa a ely. Following he 2001 CSRC opinion, he independen di ec o s o lis ed Chinese fi ms began o publicly com- men on issues ha we e impo an o he fi ms hey se ed. In he fi s couple o yea s, no independen di ec- o s issued nega i e opinions on co po a e decisions. I was no un il 2004 ha modified independen di ec o opinions eme ged, when he CSRC ga e independen di ec o s he addi ional powe o o e see con olling sha eholde s and senio execu i es. Fo example, acco ding o a CSRC ci cula (CSRC, 2005), independen di ec o s ha e he igh o employ an independen accoun ing fi m o inspec any doub ul co po a e decisions, i all o he fi m’s independen di ec o s belie e an audi is necessa y. 2.2. Li e a u e e iew Independen di ec o s’ moni o ing o manage s is widely conside ed one o he mos impo an unc ions o co po a e boa ds in p o ec ing sha eholde in e es s. Fama and Jensen (1983) p opose ha he mos di ec incen i e o independen di ec o s o make independen judgmen s on manage ial decisions is o es ablish hei epu a ions as moni o ing expe s. They p edic ha independen di ec o s who effec i ely o e see hei se ing boa ds may be ewa ded wi h addi ional boa d sea s. An ex ensi e li e a u e examines he consequences o effec i e moni o ing and he cha ac e is ics o inde- penden di ec o s. Fo example, mo e independen di ec o s on a boa d is associa ed wi h s onge CEO u n- o e and pe o mance sensi i i y (Weisbach, 1988; Conyon and He, 2011), dec eased nega i e ma ke eac ion o announcemen s o ende offe s o bidding fi ms (By d and Hickman, 1992) and inc eased posi i e ma ke eac ion o announcemen s o poison pills (B ickley e al., 1994). Se e al ecen s udies ha e documen ed mixed e idence on he moni o ing ole o independen di ec o s. Duchin e al. (2010) and Faleye e al. (2011) poin ou ha oo s ong a p esence o independen moni o ing may unde mine co po a e pe o mance. Fahlenb ach e al. (2010) p o ide e idence ha independen di ec o s wi h execu i e expe ise do no affec he appoin ing fi m’s ope a ing pe o mance, decision making o manage ial compensa ion. Almos all o he s udies in his camp ha e e alua ed he effec i eness o independen moni o ing based on he documen ed associa ions. Few s udies ha e di ec ly examined he independen moni o ing effo s o independen di ec o s, as fi ms a ely disclose dispu es ha occu inside o he boa d oom. Ano he g owing body o li e a u e in e s he alue o independen di ec o s by in es iga ing obse able co po a e e en s. S ini asan (2005) finds ha independen di ec o s, pa icula ly hose se ing on audi com- mi ees, a e mo e likely o lea e he fi ms ha es a e hei ea nings and subsequen ly lose hei boa d sea s a o he fi ms. Fich and Shi dasani (2007) examine he ma ke eac ion o he announcemen o lawsui s and di ec o u no e in fi ms in e locked wi h aud- ain ed independen di ec o s. They find a nega i e abno mal s ock e u n associa ed wi h aud announcemen s in di ec o -in e locked fi ms. Mo eo e , independen di ec- o s gene ally e ain hei boa d sea s in sued fi ms, bu lose significan di ec o ships in in e locking fi ms. Bo h s udies explain he decline o di ec o ships o aud-affilia ed independen di ec o s as epu a ional penal ies. Howe e , independen di ec o s may be innocen in he con ex o financial aud, as hey may unco e he aud and hence should no be held accoun able (Richa dson, 2005), o hey may be unawa e o he aud due o limi ed in o ma ion om managemen (Fich and Shi dasani, 2007). Mo e impo an ly, al hough he e idence epo ed in hese wo s udies is consis en wi h he epu a ion hypo hesis, Fich and Shi dasani (2007) also poin ou ha i is difficul o exclude wo o he dis inc , bu no mu ually exclusi e, explana ions o he nega i e ma ke eac ion and decline in di ec o ships. The endogenous hypo hesis p e- dic s ha fi ms wi h simila boa d s uc u es and ope a ing en i onmen s end o appoin independen di ec- o s wi h simila a ibu es. A lawsui announcemen signals ha di ec o -in e locked fi ms a e suscep ible o simila financial aud and independen di ec o s may olun a ily educe hei o he di ec o ships o ocus hei moni o ing effo s on he sued fi ms. The o he is he li iga ion hypo hesis, unde which he ma ke is conce ned ha aud- ain ed independen di ec o s may no be able o de o e enough ime and effo o mon- i o ing di ec o -in e locked fi ms. As a esul , independen di ec o s choose o hold ewe boa d sea s o educe he po en ial li iga ion isk. 94 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 2.3. Hypo hesis de elopmen The di ec o opinion da a om he Chinese s ock ma ke p o ide a good oppo uni y o conduc cleane es s on he moni o ing and epu a ional oles o independen di ec o s. Following Tang e al. (2013),we measu e independen boa d moni o ing as he issuance o modified di ec o opinions. O he h ee hypo he- ses p oposed by Fich and Shi dasani (2007), he li iga ion hypo hesis is o less conce n in ou s udy. Acco ding o A icle 113 o China’s Company Law, di ec o s who pa icipa e in adop ing a boa d esolu- ion a e held liable o any iola ion o law o egula ion esul ing om he esolu ion. ‘‘Howe e , i a di ec- o is p o en o ha e exp essed his objec ion o he o e on such esolu ion and his objec ion was eco ded in he minu es, hen he di ec o may be exemp ed om liabili y”(S anding Commi ee o Na ional People’s Cong ess, 2005). Anecdo al e idence also sugges s ha an independen di ec o may be exemp ed om he CSRC’s en o cemen ac ion i he/she has said ‘‘no” o a boa d esolu ion ha iola es law o ma ke eg- ula ion and causes losses o he fi m. 2 As li iga ion isk is low o independen di ec o s who say ‘‘no”in boa d mee ings, we only conside he es able implica ions o he o he wo compe ing hypo heses ( epu a- ion and endogenous) conce ning he ole o independen di ec o s, as p esen ed by Fich and Shi dasani (2007). To dis inguish be ween he epu a ion and endogenous hypo heses, we fi s examine he ma ke eac ion associa ed wi h di ec o s’ opinion announcemen s in di ec o -in e locked fi ms. I Chinese in es o s alue independen di ec o s’ ac i e moni o ing effo s, hen he epu a ion hypo hesis p edic s ha he ma ke will eac posi i ely o he issuance o modified di ec o opinions o fi ms in e locked wi h he independen di ec- o s who say ‘‘no”in boa d mee ings. Unde he endogenous moni o ing hypo hesis, fi ms wi h simila fi m and go e nance cha ac e is ics end o selec independen di ec o s wi h simila a ibu es. I a modified di ec- o opinion se es as a signal o boa d dispu es o he ma ke , in es o s may suspec ha he di ec o - in e locked fi m has p oblems simila o hose o he opinion ecei e s. The e o e, he endogenous hypo hesis p edic s a nega i e ma ke eac ion associa ed wi h he announcemen o modified di ec o opinions in di ec o -in e locked fi ms. Second, we in es iga e he changes in independen di ec o s’ boa d sea s in in e locking fi ms. Fama and Jensen (1983) poin ou ha he mos di ec incen i e o independen di ec o s is o es ablish good labo ma - ke epu a ions. Those who pe o m effec i e boa d moni o ing a e likely o be ewa ded wi h addi ional boa d sea s, whe eas hose who pe o m poo ly will suffe boa d sea losses. S ini asan (2005) and Fich and Shi dasani (2007) find e idence consis en wi h he p edic ion o epu a ional loss o aud- ain ed inde- penden di ec o s. We belie e an independen di ec o can es ablish a good epu a ion o independen mon- i o ing by issuing a modified di ec o opinion. Thus, we p edic ha he possibili y o losing di ec o ships a o he fi ms is nega i ely associa ed wi h independen di ec o s’ issuance o modified opinions. In con as , he endogenous hypo hesis a gues ha independen di ec o s may olun a ily cu hei boa d sea s o educe hei wo kloads o o p o ec hei epu a ions when hey an icipa e o de ec ad e se e en s (Ye mack, 2004; Fich and Shi dasani, 2007; Fahlenb ach e al., 2010). In his con ex , we p edic a decline in independen di ec o - ships ollowing announcemen s o modified di ec o opinions. We also examine whe he independen di ec o s can e ain hei boa d sea s in he fi ms hey se e i hey say ‘‘no”in boa d mee ings. Shi dasani and Ye mack (1999) a gue ha CEOs may p e e lax moni o ing and hus conside poo moni o s as a ac i e candida es o hei boa ds. Al e na i ely, di ec o s may olun a ily lea e poo ly pe o ming fi ms a a high a e o educe he damage o hei epu a ions, limi legal liabili y o a oid he highe wo kload ha boa ds usually unde ake when pe o mance dec eases (Va eas, 1999). Based on hese wo a gumen s, we p edic ha independen di ec o s a e mo e likely o lose boa d sea s in he fi ms hey se e i hey issue modified di ec o opinions. 2 In he no o ious financial aud case o Xinjiang Tunhe (s ock ID: 600073), he fi m’s wo independen di ec o s, M . Houwen Du and M . Jie Wei, esigned om hei boa d sea s in June 2004. The boa d di ec o s and op execu i es we e la e punished by he CSRC o he fi m’s alse s a emen in i s 2003 annual epo , wi h he excep ion o M . Du, who was p o en o ha e said ‘‘no”in p e ious boa d mee ings in Ap il 2004 (Li, 2004). J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 95 3. Da a and sample desc ip ion 3.1. Da a sou ces The da a o his s udy come om he China S ock Ma ke Accoun ing Resea ch (CSMAR) da abase. The ini ial sample co e s all 23,805 independen di ec o opinions announced be ween Janua y 2005 and Decem- be 2010, inclusi e. Fi s , we exclude he ollowing obse a ions: (a) opinions no coded by he da abase (CSMAR i em c05002b = .); (b) opinions coded as ‘‘o he ype”(CSMAR i em c05002b = 7), as hey canno be di ec ly classified as ei he clean o modified; (c) opinions on Chinese companies’ ‘‘sha e ading e o m” (CSMAR i em c04002b = 12), as his is no a no mal ope a ing business decision; and (d) epe i i e obse a- ions. We ob ain 18,634 independen di ec o opinions a e he fi s sc eening p ocedu e. Second, we manu- ally iden i y 70 addi ional modified independen di ec o opinions om he o iginal independen di ec o announcemen s classified by he CSMAR da abase ei he as missing (adding 55 cases) o obscu ely as ‘‘o he ype”(adding 15 cases). Thi d, we equi e each obse a ion o ha e he necessa y CSMAR da a on ma ke capi aliza ion, he independen di ec o s’ bio da a, financial s a emen s, co po a e go e nance a iables and necessa y daily p ice da a o compu e CARs. To be consis en wi h he li e a u e, we also exclude fi ms ope - a ing in he finance indus y. O e all, 10,142 independen di ec o opinions mee he selec ion c i e ia, o which 138 a e modified di ec o opinions. The sample includes 1429 dis inc fi ms ecei ing independen di ec- o opinions and 1298 dis inc di ec o -in e locked fi ms. 3.2. Sample desc ip ion Table 1 p esen s he sample’s desc ip i e s a is ics. Panel A shows he yea ly dis ibu ion o independen di ec o opinions and he fi ms ha ecei e hem. F om 2005 o 2010, 138 modified di ec o opinions we e issued, accoun ing o 1.36% o he o al di ec o opinions. The numbe o modified di ec o opinions ange om a low o 2 in 2008 and 2009 o a high o 67 in 2006. Du ing he sample pe iod, on a e age, 2.66% fi ms ecei ed modified independen di ec o opinions. Panel B o Table 1 p esen s he b eakdown o modified independen di ec o opinions by co po a e issues. We g oup he 11 ypes o business decisions coded by he CSMAR da abase in o h ee majo ypes: (1) opin- ions on pe sonnel and compensa ion issues, (2) opinions on financial epo ing and audi ing issues and (3) opinions on majo ope a ing issues o he han pe sonnel, compensa ion, financial epo ing and audi ing issues, such as ela ed-pa y ansac ions, gua an eed loans and me ge s and acquisi ions. No e ha an inde- penden di ec o ’s opinion may in ol e mul iple co po a e e en s. The mos equen single e en discussed in modified di ec o opinions is gua an eed loans. Financial epo s, ela ed-pa y ansac ions and pe sonnel issues a e o he e en s equen ly discussed by independen di ec o s. O he 145 e en s disclosed in he mod- ified di ec o opinions, 18 in ol ed pe sonnel and compensa ion issues, 28 in ol ed financial epo ing and audi ing issues and 99 in ol ed o he majo ope a ing decisions. We also examine he deg ee o se e i y o independen di ec o opinions in Panel C o Table 1. The majo i y o di ec o opinions a e s anda d and accoun o 98.62% o he o al opinions. Abs en ion, ad e se and dis- ag eemen opinions a e he op h ee modified opinions in he sample pe iod, accoun ing o 0.45%, 0.35% and 0.27% o he o al numbe o di ec o opinions, espec i ely. 4. Ma ke eac ions o modi ied independen di ec o opinions In his sec ion, we examine whe he Chinese in es o s alue independen di ec o opinions a e con olling o opinion cha ac e is ics, co po a e go e nance ac o s, financial s a emen a iables and a se o di ec o cha ac e is ics. 4.1. Va iable desc ip ions We use CAR_EVNTFIRM o measu e he 3-day CARs a ound he da es o independen di ec o opinions in opinion- ecei ing fi ms. The ma ke model is used o measu e he daily abno mal e u ns: 96 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 ARi ¼Ri ð ^ aiþ^ biRm Þ, whe e R m is p oxied by he CSMAR alue-weigh ed e u n. The model is es ima ed o e a pe iod o 256 o 7 days be o e he opinion announcemen da e, equi ing a minimum es ima ion pe iod o 120 days. Ma ke eac ion is defined as CAR ¼PT ARi . Al e na i ely, we use CAR_LOCKEDFIRM o measu e he ma ke eac ion in di ec o -in e locked fi ms o modified independen di ec o opinions. In Panel A o Table 2, we epo he esul s o he ull sample. Consis en wi h Tang e al. (2013), we find ha he ma ke eac s nega i ely when fi ms announce ecei ing modified di ec o opinions. The 3-day CARs a e sys ema ically nega i e and significan ly diffe en om ze o. The mean (median) o he CAR is 1.04% (0.97%), sugges ing ha he fi ms suffe addi ional alue decline due o he in o ma ion eleased in modified di ec o opinions. Bo h he -s a is ics o he mean diffe ences and he z-s a is ics om he Wilcoxon es s indica e ha fi ms ecei ing modified di ec o opinions exhibi significan ly nega i e s ock e u ns compa ed wi h fi ms epo ing clean di ec o opinions. Panel A also p esen s he ma ke eac ion esul s o di ec o -in e locked fi ms. As p edic ed by he epu- a ion hypo hesis, he fi ms ha a e in e locked wi h independen di ec o s who issue modified epo s expe- Table 1 Sample o independen di ec o opinions. Panel A: Yea ly dis ibu ion Yea No. o di ec o opinions No. o fi ms ecei ing di ec o opinions Clean Modified To al % o modified Clean Modified To al % o modified 2005 1956 48 2004 2.40% 854 47 901 5.22% 2006 1545 67 1612 4.16% 807 54 861 6.27% 2007 1638 13 1651 0.79% 793 13 806 1.61% 2008 1736 2 1738 0.12% 641 2 643 0.31% 2009 1750 2 1752 0.11% 648 2 650 0.31% 2010 1379 6 1385 0.43% 786 6 792 0.76% To al 10,004 138 10,142 1.36% 4529 124 4653 2.66% Panel B: E en s ela ed o modified independen di ec o opinions Ca ego y o e en s CSMAR code F equency Clean Modified To al Pe sonnel and compensa ion Pe sonnel (1) 2661 15 2992 Compensa ion (2) 313 3 Financial epo ing and audi ing Financial epo s (3) 594 24 767 Audi ing (7) 145 4 Ope a ing issues Rela ed-pa y ansac ions (4) 4907 22 7999 Gua an eed loans (5) 1290 35 Me ge and acquisi ions (6) 207 12 Changes in owne ship (8) 498 9 Financing (9) 421 0 Disposal o asse s (10) 85 4 Miscellaneous e en s (11) 492 17 To al 11,613 145 11,758 Panel C: Ca ego y o independen di ec o opinions Ca ego y o opinions CSMAR code F equency Sha e o o al Clean Unqualified opinion 1 10,002 98.62% Modified Qualified opinion 2 22 0.22% Ad e se opinion 3 35 0.35% Disclaime o opinion 4 10 0.10% Abs en ion 5 46 0.45% Disag eemen 6 27 0.27% To al 10,142 100.00% This able p esen s he summa y s a is ics o independen di ec o opinions om 2005 o 2010. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 97 ience posi i e ma ke eac ions. The mean and median o he 3-day CARs o in e locking fi ms a e 0.31% and 0.26%, espec i ely, and he mean and median CARs o fi ms ha a e in e locked wi h independen di ec o s who issue clean opinions a e significan ly nega i e. Panel A also shows ha al hough he a e age CARs a e posi i e, 46.87% o fi ms ha a e in e locked wi h di ec o s who issue modified opinions expe ience nega i e s ock e u ns. We hen condi ion he CAR analysis on possible con ounding ea nings announcemen effec s. We define a concu en ea nings announcemen e en i i is announced wi hin he same (1, +1) window as an indepen- den di ec o ’s opinion announcemen . Panel B o Table 2 shows ha 45.51% o di ec o s’ opinions a e accom- panied by a qua e ly, in e im o annual ea nings announcemen . Only 7.72% o di ec o -in e locked fi ms publish hei own ea nings wi hin he 3-day pe iod su ounding in e locking di ec o s’ opinion announce- men s. In he ollowing sec ion, we es whe he hese con ounding ea nings announcemen s affec he ma ke eac ion o independen di ec o opinions. We pe o m he empi ical analysis by pa i ioning he ull sample in o opinion- ecei ing and di ec o - in e locked fi ms. Specifically, we use he indica o a iable MDO_EVNTFIRM o diffe en ia e be ween fi ms ha ecei e modified independen di ec o opinions and hose ha ecei e clean independen di ec o opin- ions. Simila ly, in he subsample o in e locking fi ms, we use MDO_LOCKEDFIRM o iden i y fi ms ha a e in e locked wi h di ec o s who issue modified independen di ec o opinions. We con ol o he ollowing a iables in he analyses o ma ke e u ns. Fi s , ollowing Bai e al. (2004), La cke e al. (2007) and Dey (2008), who documen agency conflic s be ween con olling sha eholde s and mino i y sha eholde s, we examine ou indi idual go e nance a iables o cons uc a go e nance index and o cap u e he agency cos s o con olling sha eholde s: (1) BLOCK, se o 1 i he pe cen age o owne - ship held by he la ges sha eholde is highe han he median and 0 o he wise; (2) CONTROL DISPERSION, Table 2 Uni a ia e analysis o he cumula i e abno mal e u ns a ound di ec o s’ opinion announcemen da es. Panel A: Full sample Sample Modified opinion No. Mean C oss- sec . Pa ell Median Sign ank es % posi i e Opinion- ecei ing fi ms 0 10,004 0.14% 2.49 ** 5.16 *** 0.22% 3.86 *** 48.07% 1 138 1.04% 2.15 ** 2.67 ** 0.97% 2.38 ** 39.86% Diff (1–0) 1.18% 0.75% (2.45 ** )(2.31 ** ) Di ec o -in e locked fi ms 0 22,008 0.08% 2.41 1.28 0.33% 11.11 46.26% 1 224 0.31% 0.94 1.05 0.26% 0.94 53.13% Diff (1–0) 0.39% 0.60% (1.18) (1.47) Panel B: Ea nings announcemen a ound he independen di ec o opinion announcemen da es Di ec o opinion Opinion- ecei ing fi ms Di ec o in e -locked fi ms Any ea nings announcemen news? Any ea nings announcemen news? NO YES To al NO YES To al 0 5462 4542 10,004 20,317 1691 22,008 54.60% 45.40% 100.00% 92.32% 7.68% 100.00% 1 64 74 138 198 26 224 46.38% 53.62% 100.00% 88.39% 11.61% 100.00% To al 5526 4616 10,142 20,515 1717 22,232 54.49% 45.51% 100.00% 92.28% 7.72% Panel A p esen s he summa y s a is ics o he ma ke eac ion o independen di ec o opinions om 2005 o 2010. The CAR a iables a e defined in Appendix A. The -s a is ics and Wilcoxon-Z s a is ics o he mean and median diffe ences a e epo ed in pa en heses. *** and ** deno e significance a he 1% and 5% le els, espec i ely. Panel B p esen s he con ingency able ha displays independen di ec o opinions and ea nings announcemen s. We define ea nings announcemen news i he fi m also announces ea nings in he 3-day (1 o +1) window o he independen di ec o ’s opinion announcemen . 98 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 second yea ’s di ec o s lis ollowing he yea he/she announces his/he opinion. Gi en he need o check di ec- o s’ u no e a es o e 2 yea s, we collec a subsample o 294 indi idual independen di ec o s who issued modified di ec o opinions om 2005 o 2010. We also examine any changes in boa d sea s subsequen o he issuance o independen di ec o opinions. We define CHGSEATS as he mean changes in boa d sea s o he 2 yea s ollowing he issuance o a di ec o ’s opinion. RETCHGSEATS measu es he mean changes Table 4 (con inued) Panel B: Analysis o cumula i e abno mal e u ns o di ec o -in e locked fi ms VARIABLES (1) (2) (3) (4) CAR_LOCKEDFIRM CAR_LOCKEDFIRM CAR_LOCKEDFIRM CAR_LOCKEDFIRM IFEXEEXPT 0.001 0.000 0.001 0.000 (1.223) (0.073) (0.295) (0.620) YEAR DUMMY Yes Yes Yes Yes PROPOSAL FIXED EFFECTS Yes Yes Yes Yes Obse a ions 22,232 6668 1661 17,491 Adj. R 2 0.103 0.114 0.081 0.103 Panel A o Table 4 p esen s he eg ession esul s o he 3-day CARs a ound opinion announcemen om 2005 o 2010 o fi ms ecei ing independen di ec o opinions. The dependen a iable is CAR_EVNTFIRM. Columns (2) o (4) p esen he eg ession esul s based on he sub-samples o diffe en opinion affai s, including pe sonnel issues in Column (2), financial issues in Column (3) and ope a ing issues in Column (4). Panel B o Table 4 p esen s he eg ession esul s o he 3-day CARs a ound opinion announcemen om 2005 o 2010 o di ec o - in e locked fi ms. The dependen a iable is CAR_LOCKEDFIRM. Columns (2) o (4) p esen he eg ession esul s based on he sub- samples o diffe en opinion affai s, including pe sonnel issues in Column (2), financial issues in Column (3) and ope a ing issues in Column (4). Robus s anda d e o s clus e ed a he fi m le el a e used o compu e he -s a is ics. The -s a is ics epo ed in pa en heses a e wo ailed. ***, ** and * deno e significance a he 1%, 5% and 10% le els, espec i ely. All o he a iables a e defined in Appendix A. Table 5 Uni a ia e analysis o e aining di ec o ships. Panel A: Change in di ec o ships o indi idual independen di ec o s who issued modified opinions om 2005 o 2010 No. o di ec o ships No. o yea s ela i e o issuing modified di ec o opinion 0+1+2 1 195 145 106 2 55 43 32 3 24 25 16 4 13 12 13 5556 6210 A e age di ec o ship 1.59 1.31 1.02 % o independen di ec o s holding mo e han one boa d sea 33.67% 29.25% 22.79% Panel B: F equency o di ec o opinions s. e aining di ec o ships No. o pinion announcemen s by indi idual di ec o s Re aining di ec o ships in opinion- ecei ing fi ms Re aining di ec o ships in in e locked fi ms No Yes To al No Yes To al Clean opinions 22,175 19,035 41,210 12,355 11,617 23,972 53.81% 46.19% 100.00% 51.54% 48.46% 100.00% Modified opinions 218 134 352 96 127 223 61.93% 38.07% 100.00% 43.05% 56.95% 100.00% To al 22,393 19,169 41,562 12,451 11,744 24,195 Chi-squa e 9.265 *** 6.376 ** This able p o ides in o ma ion abou he boa d posi ions o independen di ec o s who issued modified opinions om 2005 o 2010. Yea 0 is he yea in which he di ec o ’s opinion is announced. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 105 Table 6 Desc ip i e s a is ics o di ec o ship analysis. Panel A: Summa y s a is ics o he sample o opinion- ecei ing fi ms Va iables Fi ms ecei ing clean di ec o opinions (n = 13,423) Fi ms ecei ing modified di ec o opinions (n = 269) T-s a o es s o mean diffe ences Z-s a om Wilcoxon wo-sample es Mean Median Mean Median Di ec o ship a iables RETAIN_EVNTFIRM 0.584 1.000 0.457 0.000 4.190 *** 4.188 *** CHGSEATS 0.144 0.000 0.247 0.000 2.199 ** 2.057 ** RETCHGSEATS 0.075 1.000 0.148 1.000 2.386 ** 2.147 ** GAINSEATS 0.158 0.000 0.095 0.000 2.439 ** 2.177 ** E en cha ac e is ics CAR_EVNTFIRM 0.001 0.005 0.013 0.012 3.075 *** 3.000 *** ISSUE_PERSONNEL 0.276 0.000 0.130 0.000 5.309 *** 5.304 *** ISSUE_FINANCIAL 0.133 0.000 0.186 0.000 2.542 ** 2.542 ** ISSUE_OPERATING 0.821 1.000 0.840 1.000 0.828 0.828 Fi m cha ac e is ics FIRM SIZE 21.455 21.355 21.138 21.045 4.723 *** 4.882 *** M/B 2.987 2.056 2.779 1.767 1.180 1.954 * ROA 0.032 0.032 0.024 0.011 13.266 *** 10.580 *** ST 0.026 0.000 0.115 0.000 8.942 *** 8.916 *** MAO 0.049 0.000 0.216 0.000 12.273 *** 12.206 *** FIRM AGE 8.064 8.000 7.944 8.000 0.476 0.560 CG INDEX 1.503 1.000 1.494 1.000 0.159 0.196 Boa d cha ac e is ics BOARD SIZE 2.260 2.197 2.264 2.197 0.292 0.025 %INDBOARD 0.357 0.333 0.350 0.333 2.322 ** 1.251 CEO TURNOVER 0.339 0.000 0.539 1.000 6.861 *** 6.850 *** Di ec o cha ac e is ics FEMALE 0.114 0.000 0.100 0.000 0.696 0.696 AGE 51.078 49.000 47.729 45.000 5.217 *** 5.583 *** TENURE 3.046 3.000 2.758 3.000 2.952 *** 2.493 ** MTENURE_INDIVIDUAL 3.099 3.000 2.825 3.000 3.116 *** 2.689 *** IFTERM2 0.658 1.000 0.651 1.000 0.251 0.251 ABSENCE 0.067 0.000 0.099 0.000 4.031 *** 3.088 *** PAY 10.535 10.584 10.503 10.491 0.959 1.133 MPAY_INDIVIDUAL 10.593 10.597 10.504 10.519 1.730 * 1.746 * DIRECTORSHIPS 1.707 1.000 1.669 1.000 0.556 0.394 COMPCOMM 0.208 0.000 0.208 0.000 0.010 0.038 AUDCOMM 0.195 0.000 0.186 0.000 0.355 0.355 FINEXPT 0.294 0.000 0.361 0.000 2.830 *** 2.380 ** EXEEXPT 0.208 0.000 0.286 0.000 3.113 *** 3.112 *** Panel B: Summa y s a is ics o he sample o di ec o -in e locked fi ms Va iables Fi ms in e locked wi h di ec o s issuing clean di ec o opinions (n = 10,979) Fi ms in e locked wi h di ec o s issuing modified di ec o opinions (n = 268) T-s a o es s o mean diffe ences Z-s a om Wilcoxon wo-sample es Mean Median Mean Median RETAIN_LOCKEDFIRM 0.584 1.000 0.615 1.000 0.815 0.815 E en cha ac e is ics CAR_EVNTFIRM 0.001 0.005 0.007 0.008 1.356 1.321 CAR_LOCKEDFIRM 0.000 0.004 0.005 0.007 1.304 1.917 * (con inued on nex page) 106 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 in boa d sea s di ided by he numbe o boa d sea s in he yea o opinion issuance. GAINSEATS is he ne gain o new sea s he independen di ec o secu es subsequen o saying ‘‘no.”The a iable is se o 0 i he di ec o does no gain any new sea s. Panel A o Table 5 p esen s he changes in di ec o ships held by he 294 indi idual independen di ec o s. Yea ze o e e s o he modified opinion’s announcemen yea . The a e age di ec o ships dec ease pe sis en ly, om 1.59 di ec o ships in yea ze o o 1.02 di ec o ships in yea wo. In yea ze o, 33.67% o di ec o s hold mo e han one boa d sea , and his a io dec eases by mo e han 10–22.79% in yea wo. Ou main a iables o in e es a e he indica o a iables ha diffe en ia e be ween he independen di ec- o s who say ‘‘no”in he opinion- ecei ing (MDO_DIR) and di ec o -in e locked (MDO_LOCKEDDIR) fi ms. As discussed in Sec ion 4, we con ol o se e al a iables ha he li e a u e has shown o influence di ec- o ships a bo h he fi m and di ec o le els. Fi s , we include CGINDEX in he eg ession o con ol o he en enchmen effec s o la ge sha eholde s. As independen di ec o u no e is highly ela ed o CEO u n- o e , we also include an indica o o CEO u no e in he analysis (Fich and Shi dasani, 2007). Second, he eg ession also includes a se o di ec o -specific a ibu es, including indica o a iables such as whe he he independen di ec o is emale (Adams and Fe ei a, 2009; Gul e al., 2011; S inidhi e al., 2011), Table 6 (con inued) Panel B: Summa y s a is ics o he sample o di ec o -in e locked fi ms Va iables Fi ms in e locked wi h di ec o s issuing clean di ec o opinions (n = 10,979) Fi ms in e locked wi h di ec o s issuing modified di ec o opinions (n = 268) T-s a o es s o mean diffe ences Z-s a om Wilcoxon wo-sample es Mean Median Mean Median ISSUE_PERSONNEL 0.297 0.000 0.178 0.000 3.402 *** 3.398 *** ISSUE_FINANCIAL 0.119 0.000 0.241 0.000 4.835 *** 4.825 *** ISSUE_OPERATING 0.788 1.000 0.782 1.000 0.187 0.187 Fi m cha ac e is ics FIRM SIZE 21.523 21.410 21.245 21.157 3.240 *** 2.904 *** M/B 2.960 1.988 2.481 1.654 1.958 * 3.526 *** ROA 0.033 0.031 0.012 0.023 4.114 *** 3.219 *** ST 0.026 0.000 0.040 0.000 1.172 1.172 MAO 0.048 0.000 0.080 0.000 1.961 ** 1.961 ** FIRM AGE 8.507 9.000 7.655 8.000 2.750 *** 2.711 *** CGINDEX 1.453 1.000 1.529 2.000 1.140 1.477 Boa d cha ac e is ics BOARD SIZE 2.257 2.197 2.263 2.197 0.426 0.299 %INDBOARD 0.357 0.333 0.348 0.333 2.397 ** 2.105 ** CEO TURNOVER 0.331 0.000 0.414 1.000 2.271 ** 2.270 ** Di ec o cha ac e is ics FEMALE 0.078 0.000 0.098 0.000 0.953 0.953 AGE 51.025 50.000 47.862 46.000 4.333 *** 4.522 *** TENURE 3.130 3.000 3.144 3.000 0.107 0.567 PAY 10.611 10.597 10.520 10.564 2.342 ** 2.776 *** DIRECTORSHIPS 2.826 3.000 3.351 3.000 5.988 *** 5.489 *** COMPCOMM 0.215 0.000 0.167 0.000 1.465 1.359 AUDCOMM 0.218 0.000 0.155 0.000 1.980 ** 1.980 ** COMPCOMM_EVNTFIRM 0.208 0.000 0.115 0.000 2.856 *** 2.796 *** AUDCOMM_EVNTFIRM 0.212 0.000 0.098 0.000 3.653 *** 3.649 *** FINEXPT 0.373 0.000 0.448 0.000 2.021 ** 0.020 EXEEXPT 0.145 0.000 0.207 0.000 2.263 ** 2.262 ** Panel A shows he summa y s a is ics o he a iables used in he di ec o ship analysis based on he sample o opinion- ecei ing fi ms. Panel B shows he summa y s a is ics o he sample o di ec o -in e locked fi ms. ***, ** and * deno e significance a he 1%, 5% and 10% le els, espec i ely. All o he a iables a e defined in Appendix A. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 107 Table 7 P obabili y o e aining boa d sea s. Panel A: P obabili y o e aining boa d sea s in fi ms ecei ing independen di ec o opinions VARIABLES (1) (2) RETAIN_EVNTFIRM RETAIN_EVNTFIRM CONSTANT 2.559 *** 2.287 ** (2.735) (2.429) MDO_DIR 0.649 *** 0.796 ** (3.303) (2.078) CGINDEX 0.095 *** (2.729) MDO_DIR CGINDEX 0.105 (0.486) CAR_EVNTFIRM 0.053 0.017 (0.111) (0.035) ISSUE_PERSONNEL 0.004 0.004 (0.053) (0.049) ISSUE_FINANCIAL 0.003 0.004 (0.029) (0.049) ISSUE_OPERATING 0.222 ** 0.220 ** (2.348) (2.311) FIRM SIZE 0.081 ** 0.085 ** (2.233) (2.354) M/B 0.016 0.019 (0.923) (1.102) ROA 0.957 * 1.004 * (1.755) (1.848) ST 0.109 0.126 (0.555) (0.647) MAO 0.289 ** 0.292 ** (2.026) (2.059) FIRM AGE 0.052 *** 0.054 *** (5.901) (6.092) BOARD SIZE 0.082 0.042 (0.504) (0.260) %INDBOARD 1.685 ** 1.689 ** (2.369) (2.384) CEO TURNOVER 0.289 *** 0.294 *** (4.395) (4.478) FEMALE 0.047 0.048 (0.653) (0.656) AGE 0.004 0.004 (1.535) (1.621) ABSENCE 0.643 *** 0.650 *** (3.407) (3.441) TENURE 0.726 *** 0.726 *** (26.708) (26.730) IFTERM2 0.185 *** 0.182 *** (2.636) (2.597) PAY 0.039 0.044 (0.591) (0.661) DIRECTORSHIPS 0.022 0.021 (1.007) (0.967) COMPCOMM 0.095 0.097 (1.564) (1.595) AUDCOMM 0.043 0.044 (0.625) (0.637) FINEXPT 0.114 ** 0.112 ** (2.507) (2.459) (con inued on nex page) 108 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 Table 7 (con inued) Panel A: P obabili y o e aining boa d sea s in fi ms ecei ing independen di ec o opinions VARIABLES (1) (2) RETAIN_EVNTFIRM RETAIN_EVNTFIRM EXEEXPT 0.108 * 0.106 * (1.844) (1.811) YEAR DUMMY YES YES INDUSTRY DUMMY YES YES Obse a ions 13,692 13,692 Pseudo R 2 0.181 0.182 Panel B: P obabili y o e aining boa d sea s in di ec o -in e locked fi ms (1) (2) (3) (4) (5) VARIABLES RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM CONSTANT 1.214 1.153 1.149 1.146 1.153 (1.120) (1.054) (1.051) (1.046) (1.054) MDO_LOCKEDDIR 0.244 0.875 ** 0.841 * 0.881 ** 0.956 ** (1.255) (1.989) (1.899) (1.986) (2.159) CGINDEX 0.022 0.022 0.022 0.022 (0.503) (0.499) (0.503) (0.509) MDO_LOCKEDDIR CGINDEX 0.405 * 0.414 * 0.404 * 0.419 * (1.845) (1.872) (1.844) (1.895) CAR_EVNTFIRM 0.099 0.105 0.091 0.107 0.116 (0.191) (0.203) (0.175) (0.206) (0.222) CAR_LOCKEDFIRM 0.417 0.444 0.442 0.440 0.446 (0.672) (0.713) (0.711) (0.708) (0.718) ISSUE_PERSONNEL 0.061 0.057 0.057 0.057 0.058 (0.626) (0.590) (0.587) (0.590) (0.594) ISSUE_FINANCIAL 0.036 0.036 0.036 0.035 0.035 (0.382) (0.378) (0.385) (0.373) (0.368) ISSUE_OPERATING 0.052 0.046 0.045 0.047 0.045 (0.471) (0.419) (0.404) (0.422) (0.409) SAME INDUSTRY 0.021 0.020 0.032 0.020 0.018 (0.203) (0.190) (0.301) (0.190) (0.173) MDO_LOCKEDDIR SAME INDUSTRY 0.538 (0.959) FIRM SIZE 0.035 0.035 0.035 0.035 0.035 (0.855) (0.839) (0.842) (0.837) (0.838) FIRM AGE 0.040 *** 0.041 *** 0.041 *** 0.041 *** 0.041 *** (3.748) (3.815) (3.806) (3.815) (3.816) M/B 0.012 0.012 0.012 0.012 0.012 (0.798) (0.810) (0.809) (0.817) (0.814) ROA 0.462 0.485 0.485 0.481 0.477 (0.707) (0.741) (0.741) (0.734) (0.728) ST 0.004 0.006 0.009 0.016 0.008 (0.017) (0.025) (0.036) (0.063) (0.034) MDO_LOCKEDDIR ST 0.186 (0.311) MAO 0.381 ** 0.383 ** 0.386 ** 0.383 ** 0.347 ** (2.252) (2.261) (2.273) (2.262) (1.992) MDO_LOCKEDDIR MAO 0.684 (1.204) BOARD SIZE 0.222 0.232 0.233 0.233 0.229 (1.123) (1.165) (1.171) (1.169) (1.152) %INDBOARD 2.347 *** 2.342 *** 2.342 *** 2.342 *** 2.364 *** (2.746) (2.739) (2.740) (2.740) (2.765) (con inued on nex page) J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 109 Table 7 (con inued) Panel B: P obabili y o e aining boa d sea s in di ec o -in e locked fi ms (1) (2) (3) (4) (5) VARIABLES RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM RETAIN_ LOCKEDFIRM CEO TURNOVER 0.389 *** 0.388 *** 0.388 *** 0.388 *** 0.389 *** (5.191) (5.180) (5.176) (5.173) (5.186) FEMALE 0.010 0.013 0.015 0.012 0.013 (0.082) (0.107) (0.122) (0.104) (0.104) AGE 0.003 0.003 0.003 0.003 0.003 (0.810) (0.783) (0.799) (0.782) (0.766) ABSENCE 1.013 *** 1.014 *** 1.017 *** 1.015 *** 1.014 *** (3.887) (3.901) (3.906) (3.901) (3.903) TENURE 0.783 *** 0.783 *** 0.784 *** 0.783 *** 0.783 *** (21.714) (21.758) (21.758) (21.760) (21.721) IFTERM2 0.204 ** 0.206 ** 0.205 ** 0.206 ** 0.204 ** (2.209) (2.230) (2.223) (2.231) (2.208) PAY 0.096 0.098 0.097 0.098 0.099 (1.248) (1.261) (1.250) (1.267) (1.273) DIRECTORSHIPS 0.058 * 0.059 * 0.059 * 0.059 * 0.060 * (1.756) (1.807) (1.805) (1.808) (1.837) COMPCOMM_EVNTFIRM 0.004 0.004 0.004 0.004 0.006 (0.043) (0.041) (0.040) (0.042) (0.058) COMPCOMM 0.122 0.119 0.118 0.119 0.117 (1.153) (1.124) (1.121) (1.125) (1.111) AUDCOMM_EVNTFIRM 0.008 0.010 0.010 0.010 0.009 (0.084) (0.102) (0.107) (0.104) (0.096) AUDCOMM 0.030 0.028 0.030 0.028 0.028 (0.274) (0.252) (0.268) (0.253) (0.253) FINEXPT 0.040 0.041 0.040 0.042 0.043 (0.521) (0.538) (0.522) (0.541) (0.558) EXEEXPT 0.050 0.056 0.053 0.056 0.057 (0.448) (0.499) (0.473) (0.499) (0.508) YEAR DUMMY YES YES YES YES YES INDUSTRY DUMMY YES YES YES YES YES Obse a ions 5294 5294 5294 5294 5294 Pseudo R 2 0.213 0.214 0.214 0.214 0.214 Panel C: P obabili y o e aining boa d sea s in fi ms ecei ing modified independen di ec o opinions VARIABLES (1) (2) RETAIN_EVNTFIRM RETAIN_EVNTFIRM CONSTANT 0.732 2.172 (0.231) (0.662) MDO_DIR 0.486 * 0.687 (1.771) (1.416) CGINDEX 0.206 (1.197) MDO_DIR CGINDEX 0.184 (0.700) CAR_EVNTFIRM 0.983 0.743 (0.474) (0.359) ISSUE_PERSONNEL 0.128 0.078 (0.409) (0.241) ISSUE_FINANCIAL 0.505 * 0.552 * (1.826) (1.933) ISSUE_OPERATING 0.553 0.581 * (1.601) (1.657) (con inued on nex page) Line missing 110 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 Table 7 (con inued) Panel C: P obabili y o e aining boa d sea s in fi ms ecei ing modified independen di ec o opinions VARIABLES (1) (2) RETAIN_EVNTFIRM RETAIN_EVNTFIRM FIRM SIZE 0.004 0.022 (0.036) (0.194) M/B 0.062 0.060 (1.340) (1.305) ROA 1.590 1.494 (0.906) (0.851) ST 0.655 0.751 (1.345) (1.535) MAO 0.142 0.080 (0.329) (0.186) FIRM AGE 0.044 0.044 (1.101) (1.101) BOARD SIZE 0.238 0.110 (0.375) (0.167) %INDBOARD 1.081 0.743 (0.450) (0.303) CEO TURNOVER 0.641 ** 0.681 *** (2.532) (2.585) FEMALE 0.077 0.087 (0.269) (0.299) AGE 0.005 0.003 (0.457) (0.326) ABSENCE 1.606 *** 1.671 *** (2.582) (2.703) TENURE 0.657 *** 0.668 *** (8.269) (8.158) IFTERM2 0.412 * 0.385 (1.681) (1.524) PAY 0.432 * 0.449 * (1.846) (1.877) DIRECTORSHIPS 0.015 0.002 (0.144) (0.017) COMPCOMM 0.392 0.416 (1.428) (1.543) AUDCOMM 0.014 0.000 (0.048) (0.001) FINEXPT 0.190 0.194 (0.943) (0.954) EXEEXPT 0.188 0.210 (0.834) (0.910) YEAR DUMMY YES YES INDUSTRY DUMMY YES YES Obse a ions 859 859 Pseudo R 2 0.236 0.243 Panel D: P obabili y o e aining boa d sea s in in e locked fi ms wi h di ec o s who issue modified independen di ec o opinions VARIABLES (1) (2) RETAIN_LOCKEDFIRM RETAIN_LOCKEDFIRM CONSTANT 7.095 7.487 (1.524) (1.617) MDO_LOCKEDDIR 0.696 * 1.416 ** (1.900) (2.099) (con inued on nex page) J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 111 Table 7 (con inued) Panel D: P obabili y o e aining boa d sea s in in e locked fi ms wi h di ec o s who issue modified independen di ec o opinions VARIABLES (1) (2) RETAIN_LOCKEDFIRM RETAIN_LOCKEDFIRM CGINDEX 0.017 (0.066) MDO_LOCKEDDIR CGINDEX 0.471 (1.284) CAR_EVNTFIRM 0.893 1.059 (0.465) (0.495) CAR_LOCKEDFIRM 3.449 2.946 (1.295) (1.073) ISSUE_PERSONNEL 0.687 0.661 (1.579) (1.516) ISSUE_FINANCIAL 0.085 0.085 (0.229) (0.218) ISSUE_OPERATING 0.609 0.486 (1.542) (1.198) SAME INDUSTRY 0.008 0.007 (0.021) (0.017) FIRM SIZE 0.139 0.129 (0.820) (0.767) FIRM AGE 0.053 0.068 (1.258) (1.526) M/B 0.054 0.054 (1.000) (1.026) ROA 3.354 3.463 (1.236) (1.320) ST 1.113 0.998 (1.170) (1.059) MAO 0.660 0.828 (0.890) (1.126) BOARD SIZE 0.233 0.278 (0.333) (0.363) %INDBOARD 0.296 0.243 (0.076) (0.061) CEO TURNOVER 0.050 0.016 (0.149) (0.047) FEMALE 0.561 0.506 (0.946) (0.824) AGE 0.002 0.001 (0.106) (0.060) ABSENCE 3.406 *** 3.515 *** (3.136) (3.255) TENURE 0.858 *** 0.882 *** (5.274) (5.251) IFTERM2 0.149 0.123 (0.420) (0.342) PAY 0.692 ** 0.758 ** (2.270) (2.349) DIRECTORSHIPS 0.222 * 0.227 * (1.672) (1.676) COMPCOMM_EVNTFIRM 0.256 0.232 (0.479) (0.419) COMPCOMM 0.139 0.233 (0.298) (0.488) AUDCOMM_EVNTFIRM 0.697 0.691 (1.167) (1.128) (con inued on nex page) 112 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 COMPCOMM and AUDCOMM i he di ec o si s on he compensa ion commi ee o audi commi ee (Da idson e al., 1998; S ini asan, 2005) and FINEXPT and EXEEXPT i he di ec o has any financial o execu i e expe ise (DeFond e al., 2005). O he a iables include di ec o enu e (TENURE), whe he some di ec o s a e se ing hei second e m on he boa d (IFTERM2), di ec o pay (PAY) and he numbe o boa d posi ions held as an independen di ec o (DIRECTORSHIPS). Following Fich and Shi dasani (2007), we also conside whe he he di ec o is a compensa ion o audi commi ee membe in an opinion- ecei ing fi m (COMPCOMM_EVNTFIRM and AUDCOMM_EVNTFIRM, espec i ely) in he analysis o e aining di ec o ships in in e locking fi ms. When in es iga ing he changes in he boa d sea s o indi idual di ec o s, we calcula e boa d mee ing absence (ABSENCE), mean di ec o emune a ion (MPAY_INDIVI- DUAL) and mean enu e (MTENURE_INDIVIDUAL) o each indi idual di ec o . Table 6 epo s he desc ip i e s a is ics o he con ol a iables in he di ec o ship analysis. 5.2. P obabili y o e aining di ec o ships in opinion- ecei ing i ms In his sec ion, we examine whe he independen di ec o s e ain boa d sea s a e hey issue opinions on co po a e decisions. We fi s es ima e he ollowing eg ession: P obðRETAIN EVNTFIRMi¼1Þ ¼c0þc1MDO DIRiþc2CGINDEXiþc3MDO DIR CGINDEX i þc4CAR EVNTFIRMiþc5ISSUE PERSONNELiþc6ISSUE FINANCIALi þc7ISSUE OPERATINGiþc8FIRM SIZEiþc9M=Biþc10ROAiþc11ST i þc12MAOiþc13FIRM AGEiþc14BOARDSIZEiþc15%INDBOARDi þc16CEO TURNOVERiþc17FEMALEiþc18AGEiþc19ABSENCEi þc20TENUREiþc21IFTERM2iþc22PAY iþc23DIRECTORSHIPSi þc24COMPCOMMiþc25AUDCOMMiþc26FINEXPT iþc27EXEEXPT i þX J J¼1 YEAR þX K K¼1 INDUSTRY þei whe e all o he a iables a e defined in Appendix A. Table 7 (con inued) Panel D: P obabili y o e aining boa d sea s in in e locked fi ms wi h di ec o s who issue modified independen di ec o opinions VARIABLES (1) (2) RETAIN_LOCKEDFIRM RETAIN_LOCKEDFIRM AUDCOMM 0.718 0.742 (1.189) (1.210) FINEXPT 0.452 0.424 (1.355) (1.210) EXEEXPT 0.421 0.568 (0.993) (1.180) YEAR DUMMY YES YES INDUSTRY DUMMY YES YES Obse a ions 328 328 Pseudo R 2 0.257 0.266 Panel A o Table 7 p esen s he logis ic eg essions o e aining a boa d sea in opinion- ecei ing fi ms in he 2 yea s ollowing he issuance o independen di ec o opinions om 2005 o 2010. The dependen a iable is RETAIN_EVNTFIRM. Panel B o Table 7 p esen s he logis ic eg essions o e aining a boa d sea in he sample o di ec o -in e locked fi ms. The dependen a iable is RETAIN_LOCKEDFIRM. Robus s anda d e o s clus e ed a he fi m le el a e used o compu e he z-s a is ics. The z-s a is ics epo ed in pa en heses a e wo ailed. ***, ** and * deno e significance a he 1%, 5% and 10% le els, espec i ely. All o he a iables a e defined in Appendix A. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 113 Panel A o Table 7 p esen s he logis ic eg ession esul s o e aining di ec o ships. Column (1) e eals a significan ly nega i e associa ion be ween e aining a boa d sea in an opinion- ecei ing fi m and MDO_DIR, which sugges s ha di ec o s who say ‘‘no”a e mo e likely o lea e opinion- ecei ing fi ms han he coho ha is iendly wi h he managemen on he boa d. When he CGINDEX and i s in e ac ion wi h MDO_DIR a e included in he eg ession, he coefficien o he in e ac ion e m is nega i e bu no s a is ically significan , whe eas he coefficien o MDO_DIR is s ill s a is ically significan . As expec ed, he es ima ed coefficien s o FIRM SIZE and ROA a e significan ly pos- i i e, implying ha independen di ec o s a e mo e likely o e ain hei boa d posi ions in fi ms ha a e bigge and pe o m be e . The TENURE a iable has nega i e effec s on boa d posi ions, consis en wi h he Chi- nese p ac ice ha he enu e o an independen di ec o canno exceed 6 yea s. The coefficien o IFTERM2 is s a is ically significan , which implies ha i some o he independen di ec o s a e in hei second e m, he independen di ec o who issues opinions is mo e likely o e ain his/he boa d sea s. 5.3. P obabili y o e aining di ec o ships in di ec o -in e locked i ms To es he p obabili y o e aining di ec o ships in di ec o -in e locked fi ms, we use he ollowing eg es- sion o analyze he sample o di ec o s in in e locking fi ms: P obðRETAIN LOCKEDFIRMi¼1Þ¼d0þd1MDO LOCKEDDIRiþd2CGINDEX i þd3MDO LOCKEDDIR CGINDEX iþd4CAR EVNTFIRMi þd5CAR LOCKEDFIRMiþd6ISSUE PERSONNELi þd7ISSUE FINANCIALiþd8ISSUE OPERATINGi þd9SAMEINDUSTRY iþd10MDO LOCKEDDIR SAMEINDUSTRY i þd11FIRM SIZEiþd12FIRM AGEiþd13M=Biþd14ROAi þd15ST iþd16MAOiþd17BOARD SIZEiþd18%INDBOARDi þd19CEO TURNOVERiþd20FEMALEiþd21AGEiþd22ABSENCEi þd23TENUREiþd24IFTERM2iþd25PAY iþd26DIRECTORSHIPSi þd27COMPCOMM EVNTFIRMiþd28COMPCOMMi þd29AUDCOMM EVNTFIRMiþd30AUDCOMMiþd31FINEXPT i þd32EXEEXPT iþX J J¼1 YEAR þX K K¼1 INDUSTRY þVa epsiloni Column (1) in Panel B o Table 7 shows ha he es ima ed coefficien o MDO_LOCKEDDIR is posi i e bu no s a is ically significan . When he go e nance index (CGINDEX) and i s in e ac ion wi h modified opinions (MDO_LOCKEDDIR) a e included in he eg ession, MDO_LOCKEDDIR is significan ly posi- i e, indica ing ha an independen di ec o who says ‘‘no”is mo e likely o e ain his/he di ec o ship in an in e locked fi m. In e es ingly, he in e ac ion i em in Column (2) is significan ly nega i e, sugges ing ha an independen di ec o who says ‘‘no”is mo e likely o lose his/he di ec o ship in an in e locked fi m i he fi m is hea ily influenced by he con olling sha eholde . To in es iga e he possibili y o di ec o s olun a ily educing boa d sea s in in e locked fi ms o manage li iga ion isk, when subjec ed o simila isk exposu e in he same indus y, we expand he equa ion o include he in e ac ion i em SAME INDUSTRY MDO_LOCKEDDIR. We also check whe he he main esul s hold by inco po a ing he in e ac ion be ween financial difficul y and di ec o s saying ‘‘no”in in e locked fi ms. Column (3) o Table 7 shows ha SAME INDUSTRY is nega i ely associa ed wi h he likelihood o e aining di ec o ship in di ec o -in e locked fi ms, al hough he es ima ed coefficien is no significan . The in e ac ion o SAME INDUSTRY is posi i e and insignifican , sugges ing ha di ec o s who say ‘‘no” do no in end o educe boa d sea s in in e locked fi ms ope a ing in he same indus y. Columns (4) and 114 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 hold boa d sea s. Masulis and Mobbs (2014) a gue ha fi m size is a na u al sou ce o di ec o epu a ion incen i es, gi en ha la ge fi ms affo d a di ec o g ea e isibili y, p es ige, compensa ion and likelihood o ob aining addi ional di ec o ships. Acco dingly, di ec o ships in fi ms o diffe en sizes c ea e diffe en ial incen i es o moni o senio managemen closely. We expec he epu a ion incen i es o independen di ec o s o affec hei p opensi y o say ‘‘no.”The second ins umen in he fi m-le el analysis is MPROFESSOR, he pe cen age o independen di ec o s who a e uni e si y acul y membe s. F ancis e al. (2013) find ha aca- demic di ec o s a e effec i e moni o s and play an impo an go e nance ole h ough hei ad ising and mon- i o ing unc ions. The hi d ins umen we use in he fi m-le el analysis is SAMEPLACE, an indica o se o 1 i any independen di ec o li es in he same ci y as he se ing fi m, as Alam e al. (2014) find ha geog aphic dis ance be ween di ec o s and co po a e headqua e s is ela ed o in o ma ion acquisi ion and boa d deci- sions. In he di ec o -le el analysis, he ins umen s a e he na u al loga i hm o he o al asse s o he fi ms in Table 9 (con inued) Panel C: Sensi i i y es s on he change in di ec o ships o independen di ec o s who issued opinions (1) Fi s s age (2) Second s age VARIABLES MDO_DIR CHGSEATS RET CHGSEATS GAINSEATS ISSUE_PERSONNEL 0.831 *** 0.022 0.017 0.004 (4.570) (1.298) (1.523) (0.451) ISSUE_FINANCIAL 0.490 *** 0.003 0.003 0.009 (2.983) (0.184) (0.271) (0.982) ISSUE_OPERATING 0.210 0.030 * 0.017 0.006 (1.100) (1.651) (1.334) (0.550) AGE 0.016 ** 0.005 *** 0.004 *** 0.003 *** (2.366) (7.643) (9.354) (9.008) FEMALE 0.239 0.035 * 0.021 0.036 *** (1.113) (1.673) (1.599) (2.866) ABSENCE 1.381 ** 0.049 0.063 0.048 (2.424) (0.673) (1.325) (1.112) MTENURE_INDIVIDUAL 0.181 ** 0.088 *** 0.079 *** 0.007 * (2.141) (11.725) (17.305) (1.760) IFTERM2_INDIVIDUAL 0.355 * 0.041 * 0.021 * 0.033 *** (1.897) (1.782) (1.695) (2.598) MPAY_INDIVIDUAL 0.145 0.047 *** 0.034 *** 0.033 *** (1.056) (4.005) (4.203) (4.753) DIRECTORSHIPS 0.233 *** 0.141 *** 0.015 *** 0.026 *** (2.792) (10.965) (3.746) (5.381) FINEXPT 0.129 0.112 *** 0.076 *** 0.070 *** (0.916) (6.211) (7.538) (6.475) EXEEXPT 0.287 * 0.108 *** 0.066 *** 0.063 *** (1.788) (6.906) (6.583) (7.598) YEAR DUMMY Yes Yes Yes Yes INDUSTRY DUMMY Yes Yes Yes Yes Obse a ions 15,667 15,667 15,667 15,667 Pseudo R 2 /Adj. R 2 0.134 0.104 0.078 0.056 This able p esen s he esul s o he sensi i i y es s on endogenei y. Panel A p esen s he sensi i i y es s on he 3-day CARs a ound opinion announcemen om 2005 o 2010 o di ec o -in e locked fi ms. The dependen a iables o he fi s -s age logi eg ession a e MDO_LOCKEDFIRM and MDO_LOCKEDFIRM CGINDEX and he dependen a iable o he second-s age OLS eg ession is CAR_LOCKEDFIRM. Panel B p esen s he sensi i i y es s on he p obabili y o e aining boa d sea s in di ec o -in e locked fi ms. The dependen a iables o he fi s -s age logi eg essions a e MDO_LOCKEDDIR and MDO_LOCKEDDIR CGINDEX and he dependen a iable o he second-s age eg ession is RETAIN_LOCKEDFIRM. Panel C p esen s he sensi i i y es s on he change in di ec o ships o independen di ec o s who issued opinions. The dependen a iable o he fi s -s age logi eg ession is MDO_DIR and he dependen a iables o he second-s age OLS eg ession a e CHGSEATS, RET CHGSEATS and GAINSEATS. Robus s anda d e o s clus e ed a he fi m le el (Panels A and B) and a he di ec o le el (Panel C) a e used o compu e he -s a is ics (z-s a is ics) o he OLS (logi ) eg essions. The -s a is ics (z-s a is ics) epo ed in pa en heses a e wo ailed. ***, ** and * deno e significance a he 1%, 5% and 10% le els, espec i ely. All o he a iables a e defined in Appendix A. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 121 which he independen di ec o holds boa d sea s (DIRINCENTIVE); he indica o , which is se o 1 i he independen di ec o is a acul y membe (PROFESSOR); and SAMEPLACE. In he fi s s age, we add he ins umen o he logi model o es ima e an independen di ec o ’s p opensi y o issue a modified opinion, wi h all independen a iables om he second s age se ing as con ol a iables. As he e a e h ee g oups o es s, including ma ke eac ion, e aining di ec o ships and change in di ec o ships, h ee se s o fi s -s age eg essions a e p esen ed in Table 9. Nex , we use he p edic ed p obabili y om he fi s s age o eplace he MDO indica o s in he second- s age eg essions. As MDO*CGINDEX appea s in he CAR eg ession and he es s on e aining di ec o - ships, MDO*CGINDEX is also es ima ed in he fi s s age and he p edic ed alue is included in he second-s age eg essions. As we p ima ily ocus on he epu a ion o o endogenous hypo hesis o he inde- penden di ec o , we ocus only on he es s o in e locked fi ms. Table 9 p esen s addi ional eg essions wi h ins umen a iables. In he fi s -s age eg essions in Panels A o C o Table 9, he di ec o incen i e a iable appea s o be nega i ely co ela ed wi h he di ec o ’s p open- si y o issue modified opinions, which is inconsis en wi h Masulis and Mobbs (2014). One possible explana- ion is ha independen di ec o s wi h mo e boa d sea s may end o keep silen in boa d mee ings, as hey ha e mo e o lose when s anding up o co po a e inside s. Panel A o Table 9 epo s he esul s o he ma ke eac ion o di ec o opinions in in e locked fi ms. As expec ed, he ins umen ed MDO_LOCKEDFIRM a i- able is significan ly posi i ely associa ed wi h CARs, and he ins umen ed MDO_LOCKEDFIRM*CGIN- DEX a iable is significan ly nega i ely associa ed wi h CARs. Panels B and C o Table 9 show he es s o he p obabili y o e aining boa d sea s and he change in di ec o ships. Al hough he p obabili y o e ain- ing boa d sea s in di ec o -in e locked fi ms is no significan , we find quali a i ely simila esul s ha he ins umen ed MDO_DIR is nega i ely associa ed wi h he change in di ec o ships in he 2 yea s subsequen o issuing modified di ec o opinions. The p eceding sensi i i y es s show ha ou main esul s in Sec ions 4 and 5 a e obus . 6.3. In e locking i ms ha also ecei ed modi ied independen di ec o opinions I some in e locking fi ms in ou sample also ecei ed modified independen di ec o opinions on a da e close o he e en da e, he esul s in Table 7 ha pe ain o e aining boa d sea s in di ec o -in e locked fi ms may be biased o ou findings. 4 To add ess his conce n, we conduc a sensi i i y es by excluding he in e - locking fi ms ha also ecei ed modified independen di ec o opinions wi hin a 5-yea e en window ( 2 o + 2) and e un he eg essions in Panels B and D o Table 7. In so doing, we dele e 21 dis inc in e locking fi ms and lose 132 boa d-yea obse a ions. The esul s emain quali a i ely he same. 7. Conclusions We examine he s ock and labo ma ke effec s associa ed wi h independen di ec o s’ issuance o di ec o opinions in he Chinese ma ke . We find ha he ma ke eac s nega i ely o modified di ec o opinions, bu ha di ec o -in e locked fi ms exhibi posi i e s ock e u ns a ound he opinion announcemen da es. We u - he find ha independen di ec o s a e mo e likely o lose di ec o ships a e hey issue modified opinions and less likely o gain new boa d appoin men s a e hey say ‘‘no”in boa d mee ings. Ou findings sugges ha al hough he disclosu e o independen boa d moni o ing is in o ma i e a e con olling o al e na i e expla- na ions in p e ious s udies, he epu a ion o independen moni o ing does no ewa d indi idual independen di ec o s by inc easing hei u u e di ec o ships. O e all, we en ich he di ec o epu a ion li e a u e by exam- ining he consequences o independen di ec o s’ ac i e moni o ing o he s ock ma ke and labo ma ke . Al hough ou esul s a e based on a small sample o modified opinions o independen di ec o s in China, hey may ha e impo an implica ions o he egula o s o eme ging ma ke s, whe e independen di ec o s play a c ucial ole in p o ec ing he in e es s o mino i y sha eholde s. 4 We hank he e iewe o b inging his poin o ou a en ion. 122 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 Appendix A. Va iable de ini ions Dependen a iables Abno mal e u n analysis CAR_EVNTFIRM Th ee-day (1 o +1) cumula i e abno mal e u ns (CARs) a ound he announcemen o he independen di ec o ’s opinion on he opinion- ecei ing fi m CAR_LOCKEDFIRM Th ee-day (1 o +1) CARs a ound he announcemen o he independen di ec o ’s opinion on he di ec o -in e locked fi m Di ec o u no e analysis RETAIN_EVNTFIRM Indica o a iable ha equals 1 i he independen di ec o e ains a boa d sea wi hin 2 yea s o he opinion- ecei ing fi m ollowing he issuance o he independen di ec o ’s opinion, and 0 o he wise RETAIN_LOCKEDFIRM Indica o a iable ha equals 1 i he independen di ec o e ains a boa d sea in he 2 yea s ollowing he issuance o he independen di ec o ’s opinion in his/he di ec o -in e locked fi m, and 0 o he wise CHGSEATS Mean change in boa d sea s in he 2 yea s ollowing he issuance o he di ec o ’s opinion RET CHGSEATS Mean change in boa d sea s in he 2 yea s ollowing he issuance o he di ec o ’s opinion, di ided by he numbe o boa d sea s in he yea o opinion issuance GAINSEATS Numbe o boa d sea s gained by he independen di ec o in he 2 yea s ollowing he di ec o opinion issuance. The a iable equals 0 i no new sea s a e gained CHGTASEATS Na u al loga i hm o ( he a e age o he o al asse s o all se ing fi ms/ he a e age numbe o di ec o ships in he 2 yea s ollowing he issuance o he di ec o ’s opinion) minus he na u al loga i hm o ( he sum o he o al asse s o all se ing fi ms/ he o al numbe o di ec o ships in he yea o di ec o opinion issuance) CHGMVSEATS Na u al loga i hm o ( he a e age ma ke alue o all se ing fi ms/ he a e age numbe o di ec o ships in he 2 yea s ollowing he issuance o he di ec o ’s opinion) minus he na u al loga i hm o ( he sum o he ma ke alues o all se ing fi ms/ he o al numbe o di ec o ships in he yea o di ec o opinion issuance) CHGSALESEATS Na u al loga i hm o ( he a e age o he o al sales o all se ing fi ms/ he a e age numbe o di ec o ships in he 2 yea s ollowing he issuance o he di ec o ’s opinion) minus he na u al loga i hm o ( he sum o he sales o all se ing fi ms/ he o al numbe o di ec o ships in he yea o di ec o opinion issuance) CHGPAYSEATS Na u al loga i hm o ( he a e age di ec o pay o all se ing fi ms/ he a e age numbe o di ec o ships in he 2 yea s ollowing he issuance o he di ec o ’s opinion) minus he na u al loga i hm o ( he sum o he di ec o pay o all se ing fi ms/ he o al numbe o di ec o ships in he yea o di ec o opinion issuance) T ea men a iables Abno mal e u n analysis MDO_EVNTFIRM Indica o a iable ha equals 1 i he fi m ecei es a modified independen di ec o opinion, and 0 o he wise J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 123 MDO_LOCKEDFIRM Indica o a iable ha equals 1 i he fi m is in e locked wi h an independen di ec o who issues a modified di ec o opinion, and 0 o he wise Di ec o u no e analysis MDO_DIR Indica o a iable ha equals 1 i he independen di ec o issues a modified di ec o opinion, and 0 o he wise MDO_LOCKEDDIR Indica o a iable ha equals 1 i he independen di ec o issues a modified di ec o opinion o he di ec o -in e locked fi m, and 0 o he wise Con ol a iables E en cha ac e is ics ISSUE_PERSONNEL Indica o a iable ha equals 1 i he independen di ec o ’s opinion is issued owa d a boa d esolu ion on pe sonnel issues (e.g., appoin ing o discha ging op execu i es and manage ial compensa ion), and 0 o he wise ISSUE_FINANCIAL Indica o a iable ha equals 1 i he independen di ec o ’s opinion is issued owa d a boa d esolu ion on financial epo ing and audi ing issues, and 0 o he wise ISSUE_OPERATING Indica o a iable ha equals 1 i he di ec o ’s opinion is on ope a ing issues o he han pe sonnel, financial epo ing and audi ing issues, and 0 o he wise Fi m and boa d cha ac e is ics FIRM SIZE Na u al loga i hm o o al asse s M/B Ma ke alue o equi y/book alue o equi y ROA Ne income/ o al asse s ST Indica o a iable ha equals 1 i he fi m is gi en special ea men s a us, and 0 o he wise MAO Indica o a iable ha equals 1 i he fi m ecei es a modified audi opinion, and 0 o he wise FIRM AGE Numbe o yea s he fi m’s s ock has aded on he Shanghai o Shenzhen s ock exchange EARN NEWS Indica o a iable ha equals 1 i he fi m announces ea nings in he same window (-1 o + 1) as he independen di ec o ’s opinion, and 0 o he wise EARN SURP Mos ecen ly announced ea nings minus he ea nings ou qua e s ago, di ided by he ma ke alue o equi y SAME INDUSTRY Indica o a iable ha equals 1 i he di ec o -in e locked fi m is in he same indus y as he opinion- ecei ing fi m, and 0 o he wise BLOCK Indica o a iable ha equals 1 i he pe cen age o owne ship held by he la ges sha eholde is mo e han he median, and 0 o he wise CONTROL DISPERSION Indica o a iable ha equals 1 i he ul ima e con olling sha eholde ’s con ol igh s do no equal he sha eholde ’s owne ship, and 0 o he wise INSTIHLD Pe cen age o owne ship in he fi m held by ins i u ional in es o s LESSINSTIHLD Indica o a iable ha equals 1 i he pe cen age o owne ship held by ins i u ional in es o s is less han he median, and 0 o he wise DUALITY Indica o a iable ha equals 1 i he chai man o he boa d is also he CEO, and 0 o he wise CGINDEX Composi e index calcula ed by summing up BLOCK, CONTROL DISPERSION, LESSINSTIHLD and CEO DUALITY ( anging om 0 o4) CEO TURNOVER Indica o a iable ha equals 1 i he CEO o chai man lea es he office in he 2 yea s ollowing he di ec o opinion announcemen , and 0 o he wise 124 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 BOARD SIZE Na u al loga i hm o he numbe o boa d membe s %INDBOARD Pe cen age o he boa d membe s who a e independen di ec o s Di ec o cha ac e is ics FEMALE Indica o a iable ha equals 1 i he independen di ec o is emale, and 0 o he wise IFFEMALE Indica o a iable ha equals 1 i a leas one independen di ec o o a company is emale, and 0 o he wise TENURE Yea s o se ice as an independen di ec o on he fi m’s boa d MTENURE Mean yea s o se ice as an independen di ec o o all independen di ec o s on he fi m’s boa d MTENURE_INDIVIDUAL Mean yea s o se ice as an independen di ec o ac oss all o his/he boa d posi ions as an independen di ec o IFTERM2 Indica o a iable ha equals 1 i a leas one independen di ec o has se ed he fi m o mo e han 3 yea s, and 0 o he wise IFTERM2_INDIVIDUAL Indica o a iable ha equals 1 i an independen di ec o has se ed o mo e han 3 yea s in any o his/he se ing fi ms, and 0 o he wise AGE Age o he independen di ec o MAGE Mean age o all independen di ec o s on he fi m’s boa d PAY Na u al loga i hm o annual di ec o emune a ion MPAY Na u al loga i hm o mean annual di ec o emune a ion o all independen di ec o s on he fi m’s boa d MPAY_INDIVIDUAL Na u al loga i hm o mean annual di ec o emune a ion o an independen di ec o ac oss all o his/he se ing fi m boa ds as an independen di ec o ABSENCE Pe cen age o absences o o al numbe o boa d mee ings DIRECTORSHIPS Numbe o boa d sea s held by a pe son in all o his/he se ing companies as an independen di ec o MDIRECTORSHIPS Mean numbe o boa d sea s held by all independen di ec o s on he fi m’s boa d COMPCOMM Indica o a iable ha equals 1 i he opinion-issuing di ec o is a membe o he fi m’s compensa ion o nomina ion commi ee, and 0 o he wise COMPCOMM_EVNTFIRM Indica o a iable ha equals 1 i he in e locked independen di ec o is a membe o he opinion- ecei ing fi m’s compensa ion o nomina ion commi ee, and 0 o he wise IFCOMPCOMM Indica o a iable ha equals 1 i any opinion-issuing di ec o si s on he compensa ion commi ee o he boa d, and 0 o he wise AUDCOMM Indica o a iable ha equals 1 i he opinion-issuing di ec o is a membe o he fi m’s audi commi ee, and 0 o he wise IFAUDCOMM Indica o a iable ha equals 1 i any opinion-issuing di ec o si s on he audi commi ee o he boa d, and 0 o he wise AUDCOMM_EVNTFIRM Indica o a iable ha equals 1 i he in e locked independen di ec o is a membe o he opinion- ecei ing fi m’s audi commi ee, and 0 o he wise FINEXPT Indica o a iable ha equals 1 i he opinion-issuing di ec o has financial expe ise, and 0 o he wise IFFINEXPT Indica o a iable ha equals 1 i any independen di ec o s o he fi m ha e financial expe ise, and 0 o he wise EXEEXPT Indica o a iable ha equals 1 i he opinion-issuing di ec o has execu i e expe ise, and 0 o he wise IFEXEEXPT Indica o a iable ha equals 1 i any independen di ec o s o he fi m ha e execu i e expe ise, and 0 o he wise J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 125 Ins umen a iables MDIRINCENTIVE Na u al loga i hm o he sum o he o al asse s o he fi ms in which he independen di ec o s hold boa d sea s DIRINCENTIVE Di ec o incen i e measu ed as he na u al loga i hm o he o al asse s o all o his/he se ing fi ms MPROFESSOR Pe cen age o independen di ec o s who a e uni e si y acul y membe s PROFESSOR Indica o a iable ha equals 1 i he independen di ec o is a uni e si y acul y membe , and 0 o he wise SAMEPLACE Indica o a iable ha equals 1 i any independen di ec o li es in he same ci y as he se ing fi m, and 0 o he wise Re e ences Adams, R., Fe ei a, D., 2009. Women in he boa d oom and hei impac on go e nance and pe o mance. J. Financ. Econ. 94 (2), 291– 309. Ag awal, A., Chen, M.A., 2011. Boa d oom b awls: an empi ical analysis o dispu es in ol ing di ec o s. A ailable a SSRN: h p:// ss n.com/abs ac =1362143 o h ps://doi.o g/10.2139/ss n.1362143. Alam, Z.S., Chen, M.A., Cicco ello, C.S., Ryan, H.E., 2014. Does he loca ion o di ec o s ma e ? In o ma ion acquisi ion and boa d decisions. J. Financ. Quan i a i e Anal., o hcoming Bai, C.E., Liu, Q., Lu, J., Song, F.M., Zhang, J., 2004. Co po a e go e nance and ma ke alua ion in China. J. Compa . Econ. 32, 599– 616. B ickley, J., Coles, J., Te y, R., 1994. Ou side di ec o s and he adop ion o poison pills. J. Financ. Econ. 35, 371–390. Bushee, B.J., 1998. The influence o ins i u ional in es o s on myopic R&D in es men beha io . Accoun . Re . 73 (3), 305–333. By d, J., Hickman, K., 1992. Do ou side di ec o s moni o manage s? E idence om ende offe bids. J. Financ. Econ. 32, 195–221. Chung, R., Fi h, M., Kim, J., 2002. Ins i u ional moni o ing and oppo unis ic ea nings managemen . J. Co po . Finance 8 (1), 29–48. Conyon, M.J., He, L., 2011. Execu i e compensa ion and co po a e go e nance in China. J. Co po . Finance 17, 1158–1175. Co ne , M.M., Ma cus, A.J., Saunde s, A., Teh anian, H., 2007. The impac o ins i u ional owne ship on co po a e ope a ing pe o mance. J. Bank. Finance 31 (6), 1771–1794. CSRC, 1997. Guidelines o he a icles o associa ion o lis ed companies (in Chinese). CSRC, 2001. Guiding opinion on es ablishing an independen di ec o sys em in lis ed companies (in Chinese). CSRC, 2005. Ci cula o issuing he p o isions on s eng hening he p o ec ion o he igh s and in e es s o he gene al public sha eholde s (in Chinese). Da idson III, W.N., Pilge , T., Szakma y, A., 1998. Golden pa achu es, boa d and commi ee composi ion, and sha eholde weal h. Financ. Re . 33, 17–32. DeFond, M.L., Hann, R.N., Hu, X., 2005. Does he ma ke alue financial expe ise on audi commi ees o boa ds o di ec o s? J. Accoun . Res. 43 (2), 153–193. Dey, A., 2008. Co po a e go e nance and agency conflic s. J. Accoun . Res. 46 (5), 1143–1181. Doyle, J., Ge, W., McVay, S., 2007. De e minan s o weaknesses in in e nal con ol o e financial epo ing. J. Accoun . Econ. 44, 193– 233. Duchin, R., Ma susaka, J.G., Ozbas, O., 2010. When a e independen di ec o s effec i e? J. Financ. Econ. 96, 195–214. Fahlenb ach, R., Low, A., S ulz, R.M., 2010. The da k side o independen di ec o s: do hey qui when hey a e mos needed? A ailable a SSRN: h p://www.ss n.com/abs ac =1585192. Faleye, O., Hoi ash, R., Hoi ash, U., 2011. The cos s o in ense boa d moni o ing. J. Financ. Econ. 101, 160–181. Fama, E.F., F ench, K.R., 1992. The c oss-sec ion o expec ed s ock e u ns. J. Finance 47 (2), 427–465. Fama, E.F., Jensen, M.C., 1983. Sepa a ion o owne ship and con ol. J. Law Econ. 26, 301–325. Fich, E.M., Shi dasani, A., 2007. Financial aud, di ec o epu a ion, and sha eholde weal h. J. Financ. Econ. 86, 306–336. F ancis, B., Hasan, I., Wu, Q., 2013. P o esso s in he boa d oom: is he e an added alue o academic di ec o s? Financ. Manage., o hcoming Gilson, S.C., 1990. Bank up cy, boa ds, banks, and blockholde s. J. Financ. Econ. 27, 355–387. Gul, F., S inidhi, B., Ng, A., 2011. Does boa d gende di e si y imp o e he in o ma i eness o s ock p ices? J. Accoun . Econ. 51 (3), 314–338. Ha o d, J., 2003. Takeo e bids and a ge di ec o ’s incen i es: he impac o a bid on di ec o ’s weal h and boa d sea s. J. Financ. Econ. 69, 51–83. He malin, B.E., Weisbach, M.S., 2003. Boa ds o di ec o s as an endogenously de e mined ins i u ion: a su ey o he economic li e a u e. Econ. Policy Re . 9, 7–26. 126 J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 Jiang, G., Wang, H., 2008. Should ea nings h esholds be used as delis ing c i e ia in s ock ma ke ? J. Accoun . Public Policy 27 (5), 409– 419. Jiang, W., Wan, H., Zhao, S., 2016. Repu a ion conce ns o independen di ec o s: e idence om indi idual di ec o o ing. Re . Financ. S ud. 29 (3), 655–696. La cke , D.F., Richa dson, S.A., Tuna, I., 2007. Co po a e go e nance, accoun ing ou comes, and o ganiza ional pe o mance. Accoun . Re . 82 (4), 963–1008. Li, B., 2004. Effec i e moni o ing sa es independen di ec o s om legal liabili y. Shanghai Secu i ies News, July 1. Lin, K., Pio oski, J.D., Yang, Y.G., Tan, J., 2012. Voice o exi ? Independen di ec o decisions in an eme ging economy. A ailable a SSRN: h p://ss n.com/abs ac =2166876 o h ps://doi.o g/10.2139/ss n.2166876. Masulis, R.W., Mobbs, S., 2014. Independen di ec o incen i es: whe e do alen ed di ec o s spend hei limi ed ime and ene gy? J. Financ. Econ. 111 (2), 406–429. Richa dson, S., 2005. Discussion o consequences o financial epo ing ailu e o independen di ec o s: e idence om accoun ing es a emen s and audi commi ee membe s. J. Accoun . Res. 43 (2), 335–342. Shi dasani, A., Ye mack, D., 1999. CEO in ol emen in he selec ion o new boa d membe s: an empi ical analysis. J. Finance LIV 5, 1829–1853. S inidhi, B., Gul, F.A., Tsui, J., 2011. Female di ec o s and ea nings quali y. Con emp. Accoun . Res. 28 (5), 1610–1644. S ini asan, S., 2005. Consequences o financial epo ing ailu e o independen di ec o s: e idence om accoun ing es a emen s and audi commi ee membe s. J. Accoun . Res. 43 (2), 291–334. S anding Commi ee o he Na ional People’s Cong ess, 2005. Company Law o he People’s Republic o China. Tang, X., Du, J., Hou, Q., 2013. The effec i eness o he manda o y disclosu e o independen di ec o s’ opinions: empi ical e idence om China. J. Accoun . Public Policy 32 (3), 89–125. Va eas, N., 1999. Boa d mee ing equency and fi m pe o mance. J. Financ. Econ. 53, 113–142. Weisbach, M., 1988. Independen di ec o s and CEO u no e . J. Financ. Econ. 20, 431–460. Yao, Y., Xu, L., Liu, Z., 2010. Taking away he o ing powe s om con olling sha eholde s: e idence om he Chinese secu i ies ma ke . J. In . Financ. Manage. Accoun . 21 (3), 187–219. Ye mack, D., 2004. Remune a ion, e en ion, and epu a ion incen i es o independen di ec o s. J. Finance 59 (5), 2281–2308. Yuan, R.L., Xiao, J.Z., Zou, H., 2008. Mu ual unds’ owne ship and fi m pe o mance: e idence om China. J. Bank. Finance 32 (8), 1552–1565. J. Du e al. / China Jou nal o Accoun ing Resea ch 11 (2018) 91–127 127