The NASDAQ Restructuring: Do Names Even Matter?

Transcription

1 The NASDAQ Resrucuring: Do Names Even Maer? Kevin D. Broom 1 1 Deparmen of Healh Managemen & Policy, Sain Louis Universiy, Unied Saes Correspondence: Kevin D. Broom, Assisan Professor, Deparmen of Healh Managemen & Policy, Sain Louis Universiy, Unied Saes. kbroom3@slu.edu Received: January 24, 2013 Acceped: March 15, 2013 Online Published: April 8, 2013 doi: /ifr.v4n2p1 URL: hp://dx.doi.org/ /ifr.v4n2p1 Absrac This paper examines he impac of NASDAQ s July 2006 resrucuring o creae he new Global Selec and Global Markes. The new iers changed NASDAQ from a wo-iered markeplace (Naional and Capial Markes) o a hree-iered markeplace (Global Selec, Global, and Capial Markes). I examine he asse-pricing impac on NASDAQ-lised firms affeced by he resrucuring, as well as any changes o NASDAQ s abiliy o compee wih he NYSE for new lisings. While iniial daa analysis indicaes a poenial announcemen effec for Global Selec Marke socks, furher analysis indicaes lile more han a momenum effec from overall marke movemen in he weeks prior o he creaion of he new ier. Addiionally, analysis shows ha NASDAQ became less compeiive agains NYSE in aracing new lisings afer he resrucuring. These findings cas doub on any uiliy resuling from he resrucuring. Keywords: NASDAQ, global selec, NYSE, compeiion, iered, momenum, IPO 1. Inroducion On February 15, 2006, he Nasdaq Sock Marke, Inc. (referred o hereafer as NASDAQ) announced he creaion of a new marke ier for publicly raded companies on he NASDAQ Sock Marke. The newes ier, named he NASDAQ Global Selec Marke (GSM), would have financial and liquidiy requiremens higher han any oher marke in he world. On June 26, 2006, a subsequen NASDAQ announcemen specified he approximaely 1,200 companies ha qualified for he new marke ier. Less han one week laer, on July 3, 2006, he new lising srucure ook effec. Bob Greifeld, NASDAQ Presiden and Chief Execuive Officer, promoed he new ier as a blue chip marke for blue chip companies. His announcemen implies, a leas in some manner, ha NASDAQ s inen was o creae a new, unique markeplace for blue chip companies. On he oher hand, an aricle in MarkeWach pronounced ha he new ier means lile o invesors (Jaffe, 2006). Furhermore, he aricle wen on o sae he NASDAQ's designaions are ransparen and ulimaely have more o do wih markeing han markes (Jaffe, 2006). Ulimaely, wheher or no NASDAQ s creaion of he Global Selec Marke (GSM) represens an aemp o enhance NASDAQ s repuaion is an empirical quesion. From an economic perspecive, does he exisence of a iered markeplace somehow represen a compeiive response o maximize uiliy for an exchange? Jickling (2007) provides evidence ha NASDAQ s lisings have dropped 39% during he imeframe. If a iered srucure assiss in aracing new lisings (which produce higher lising fees) or new raders (which produce more commissions on rading volume), a iered srucure would maximize NASDAQ shareholders wealh. Is he recen resrucuring a response o a decade of declining lisings, designed o maximize NASDAQ shareholders wealh? More imporanly, did i increase shareholder wealh? This sudy examines if iered srucures are an aemp by an exchange o enhance heir repuaion. I firs analyze NASDAQ-lised socks o deermine if he resrucuring signaled lower risk and greaer presige for NASDAQ hrough is impac on socks lised on he new Global Selec Marke (GSM) and Global Marke (GM) iers. Did he new ier assis GSM socks in increasing he invesor bases (i.e. visibiliy), reducing heir cos of capial, and increasing heir marke value? Any collecive impac on GSM socks would represen an indirec repuaion effec for NASDAQ. The analysis focuses on socks affeced by he resrucuring, namely he former NASDAQ Naional Marke socks ha were re-caegorized as he Global Selec Marke (GSM) and Global Marke (GM) socks. Published by Sciedu Press 1 ISSN E-ISSN

2 The second par of he sudy examines if he resrucuring enhances NASDAQ s compeiiveness in he markeplace for new lisings by focusing on any impac on is abiliy o compee for new IPO lisings wih oher U.S. exchanges. The analysis focuses on any changes in NASDAQ s abiliy o compee for new lisings in he IPO marke. A direc change in NASDAQ s compeiiveness could also represen a repuaion effec resuling from he resrucuring. If a higher proporion of new firms ha are eligible for lising on muliple exchanges chose o lis on NASDAQ afer he resrucuring, his increase provides suppor for a repuaion effec. The evidence indicaes ha NASDAQ-lised companies did no receive any announcemen effec as a resul of he resrucuring, and NASDAQ is less (no more) compeiive in he compeiion for lising wih he NYSE. My analysis indicaes NASDAQ-lised socks demonsraed a momenum effec ha sared before he resrucuring, and coninued well beyond he resrucuring, wih no noiceable change in he direcion of he sock price momenum. My analysis also indicaes he probabiliy of NASDAQ acquiring a new lising acually fell significanly afer he resrucuring, indicaing a weakened compeiive sae for he exchange. The res of he paper is organized as follows. Secion 2 reviews relaed lieraure and provides hypoheses. Secion 3 describes he daa and mehodology. Secion 4 presens he empirical findings, and Secion 5 concludes. 2. Lieraure Review and Hypoheses Meron s (1987) invesor aenion hypohesis saes ha when a firm increases is invesor base (i.e. visibiliy), i can lower heir expeced reurns, reduce heir cos of capial, and increase heir marke value. Thus, increased visibiliy can serve as a proxy for higher repuaion. This sudy seeks o deermine if NASDAQ s 2006 resrucuring resuled in any maerial changes in NASDAQ s repuaion and heir compeiiveness in he markeplace for lisings. Under he Chemmanur and Fulghieri (2006) framework, exchanges face a rade-off beween he value resuling from a higher repuaion and he value of expeced cash flows from firms lising on he exchange. The opimal lising sandards maximize he combinaion of hese wo offseing values. Consider an alernaive moivaion under his heoreical approach. The abiliy o creae a iered srucure migh allow a marke o aler his radeoff beween he repuaion value and he value of expeced cash flows. Consider a single-iered marke wih only one lising crierion, marke capializaion, and he exchange liss only firms wih a minimum marke capializaion of $100M. If he exchange decides o creae a new lower ier, wih a minimum marke capializaion of $25M for he new lower ier (ceeris paribus), he exchange doesn appear o creae maerially differen iers. If he rading srucure is he exac same, he exchange is simply allowing a new se of lower repuaion firms o lis in a rading environmen wih he same marke fricions (again, assuming he same rading echnology, he same rading rules, ec.). The exchange can hen signal ha he new lower ier is a specialized ier for smaller, emerging companies, while coninuing o promoe he higher lising sandards of he original ier. Figure 1 shows his relaionship. Figure 1. Opimal lising sandards (wih lower ier) Under he opimal sandards for an exchange having only one ier, Poin S denoes he poin where he oal value of lising fees plus repuaion value is maximized. If he exchange were o lower lising sandards, heir cash flows from lising fees would increase (from Poin A o Poin B) as hey araced new firms o lis on heir exchange (ha were ineligible for lising under he old lising sandards), bu heir repuaion value would be reduced (from Poin A o Published by Sciedu Press 2 ISSN E-ISSN

3 Poin C) as invesors view he exchange as becoming a lower repuaion exchange. The rade-off beween repuaion value and value from lising fees would resul in a shif along he Toal Value Curve (from Poin S o Poin T), bu a minimal change in he oal value of he exchange. On he oher hand, a iered srucure could have a differen effec. The creaion of a lower ier could allow he exchange o mainain heir high repuaion while increasing heir cash flows from lising fees. As long as he exchange successfully markes he lower ier as being a specialized marke for emerging companies, while minimizing similariies wih he higher ier, he exchange could allow new companies o lis on he lower ier. If successful, heir cash flow would increase (from Poin A o Poin B) while he repuaion value would remain sable (from Poin A o Poin D). The end resul would be an upward shif in oal value from Poin S o U. In he case of NASDAQ, his effec could help explain he moivaion o move away from is original single-iered srucure as firs developed in In he early 1980s, as he NASDAQ firms began o diverge ino disinc classes of larger and smaller firms, he NASDAQ divided ino he NASDAQ Naional Marke and he NASDAQ Small-Cap Marke. NASDAQ s success wih a iered srucure may have moivaed European exchanges in Belgium, Germany, France, and Holland o adop similar marke segmens in order o arac high growh companies (Mendoza, 2007). Today, he dominan European exchange using a iered srucure is he London Sock Exchange wih is Alernaive Invesmen Marke. Conversely, now consider he same single-iered marke wih only one lising crieria, marke capializaion, and he exchange liss only firms wih a minimum marke capializaion of $100M. If he exchange creaes a second higher ier, wih a minimum marke capializaion of $250M for he new higher ier (ceeris paribus), he exchange doesn appear o creae maerially differen iers. If he rading srucure is sill he exac same, he exchange is simply reclassifying an already exising subse of is lised companies as being higher repuaion firms, and his new ier sill has he same marke fricions (again, assuming he same rading echnology, he same rading rules, ec.). The exchange can hen signal ha he new higher ier is a blue chip ier for larger, esablished companies, in an aemp o compee wih oher high repuaion exchanges. The exchange would promoe he higher lising sandards ( a blue chip marke for blue chip companies ) in an effor o enhance heir repuaion value. If successful, he exchange would reap a higher repuaion value by promoing he virues of he highes ier, while coninuing o collec lising fees from he lowes ier. Figure 2 shows his relaionship. Under he opimal sandards for an exchange having only one ier, Poin S denoes he poin where he oal value of lising fees plus repuaion value is maximized. If he exchange were o increase heir lising sandards, heir repuaion value would increase (from Poin A o Poin B), bu heir cash flows from lising fees would be reduced (from Poin A o Poin C) as some firms would no longer mee he higher lising sandards. The rade-off beween repuaion value and cash flows from lising fees would resul in a shif along he Toal Value Curve (from Poin S o Poin T), bu a minimal change in he oal value of he exchange. On he oher hand, a iered srucure could have a differen effec. Figure 2. Opimal lising sandards (wih higher ier) Published by Sciedu Press 3 ISSN E-ISSN

4 The creaion of a higher ier could allow he exchange o coninue collecing cash flows from lised companies ha do no exceed he high repuaion hreshold of he upper ier. As a resul, he value from lising fees would remain sable (a shif from Poin A o Poin D). The repuaion value would increase (from Poin A o Poin B) if he exchange could successfully promoe he meris of he higher ier while minimizing awareness of he similariies wih he lower ier. The end resul would be an upward shif in oal value from Poin S o Poin U. In his example, he iers migh no exhibi marke qualiy differences beyond he marke capializaion and public floa of socks lised on each ier. The enire marke, regardless of ier, could be viewed as one big marke wih all socks exhibiing he same rading fricions, differing only by a scale effec (i.e. he iers are lile more han liquidiy iers wih he same fricions). Thus, he exchange could maximize heir oal value by implemening higher sandards hrough he creaion of an upper ier. Under his scenario, he exchange benefis by esablishing a iered-srucure and promoing he benefis of a high-repuaion, upper ier in order o enhance heir repuaion value while mainaining heir cash flows from lising fees. If so, he only maerial differences in marke qualiy beween he iers, ceeris paribus, should resul primarily from firm size and public floa. The enire markeplace, regardless of ier, could be viewed as one big marke wih all socks exhibiing he same rading fricions, bu differing only in scale (i.e. he iers are lile more han liquidiy iers wih similar rading fricions). This Repuaion Hypohesis may explain NASDAQ s moivaion for a iered marke srucure. NASDAQ s curren compeiive environmen, defined by a highly compeiive markeplace for new lisings and an ongoing wave of consolidaion in exchanges and rading plaforms (in an aemp o garner marke share), migh have creaed he need o creae he new Global Selec Marke. Having already resrucured o gain a compeiive advanage a he lower end of he marke, he GSM resrucuring appears aimed a gaining a compeiive advanage a he upper end of he marke. While he heoreical moivaion simply provides a usificaion for he reason why NASDAQ resrucured (i.e. o enhance heir repuaion value), he empirical porion of his sudy aemps o measure wheher he resrucuring acually enhanced NASDAQ s repuaion. Two echniques previously used o measure an impac on a firm s repuaion include visibiliy and compeiiveness approaches. For publicly raded firms, an asse pricing approach measures changes in a firm s visibiliy o serve as a proxy for changes in heir repuaion. Since NASDAQ began publicly rading on he NASDAQ Sock Marke in 2005, any repuaion effec can be measured direcly on NASDAQ s sock. If NASDAQ s resrucuring was designed o draw new aenion o heir markeplace, any posiive repuaion impac should resul in a posiive sock pricing effec. Addiionally, he repuaion effec could be indirec, specifically o socks lised on he newes NASDAQ ier. While researchers have ye o specifically measure any repuaion effec wih exchanges, ample evidence does exis ha repuaion (i.e. visibiliy) is priced in socks. Kadlec and McConnell (1994) find ha visibiliy changes are an imporan deerminan in explaining firm decisions o move heir lising from NASDAQ o NYSE. Thus, firms seek he repuaion effec from moving o he higher-repuaion seing (i.e. from NASDAQ o NYSE). Jain and Kim (2006) find ha firms experience posiive cumulaive abnormal reurns upon swiching heir lising from NASDAQ o he NYSE. Papaioannou, Travlos, and Viswanahan. (2008) analyze changes in operaing performance resuling from he increased visibiliy of firms moving heir lising o NYSE. They find ha increased visibiliy leads o increased operaing performance. Likewise, Baker, Powell, and Weaver (1999) argue ha visibiliy is imporan o firms. The increased visibiliy may increase informaion flow abou a firm (reduces uncerainy) and enhance he efficiency of rading in heir sock (reduces informaion asymmeries). However, hey find ha he increased visibiliy resuls from changes in marke capializaion, and no simply from he lising decision. Thus, he following hypoheses will be esed: H 0 : The NASDAQ reorganizaion had no posiive impac on heir repuaion (i.e. no indirec repuaion effec). H 1 : The NASDAQ reorganizaion had a posiive impac on heir repuaion hrough he socks lised on heir exchange (i.e. an indirec repuaion effec). Published by Sciedu Press 4 ISSN E-ISSN

5 On he oher hand, Barber and Odean (2008) propose evidence ha any asse pricing effec resuling from he resrucuring may be only emporary, resuling from he increased aenion around he iming of he announcemen. They show ha invesors are ne buyers of aenion-grabbing socks, and ha aenion-driven buying does no resul in superior reurns. Consequenly, any repuaion effec may simply be a emporary aenion effec due o he resrucuring announcemen. Therefore, he asse pricing analysis will include deermining if any repuaion effec is permanen or emporary. For exchanges in paricular, he compeiiveness approach measures changes in heir abiliy o compee for lisings, and his change in compeiiveness also serves as a proxy for repuaion. NASDAQ s enhanced abiliy o arac new lisings (from exising or new public firms) could lead invesors o expec higher fuure cash flows. Easley and O Hara (2007) sae ha exchanges collec revenues boh hrough lising fees and ransacion fees, boh of which would increase if a higher proporion of firms choose o lis on NASDAQ. Higher cash flows could lead o an expecaion of higher earnings, and hus o a posiive impac o heir sock price as invesors upwardly revise heir valuaions of he exchange s sock. Under he Chemmanur and Fulghieri (2006) framework, high-repuaion exchanges se high lising and disclosure requiremens, resuling in more precise informaion available o ousiders when evaluaing firms lised on he exchange. Exchanges can aemp o use marke segmens (iers) o enhance heir repuaion impac by implemening higher sandards and forming a new higher ier. If successful, he exchange could exploi he new ier o beer compee for lisings wih oher high-repuaion exchanges. Coffee (2002) refers o his compeiion hrough higher lising sandards as he race o he op scenario. Thus, he following hypoheses will be esed o deermine if he resrucuring had any impac on NASDAQ s repuaion hrough heir abiliy o arac new lisings: H 0 : The NASDAQ reorganizaion had no impac on heir compeiiveness in he markeplace for lisings (i.e. no repuaion effec). H 2 : The NASDAQ reorganizaion had a posiive impac on heir compeiiveness in he markeplace for lisings (i.e. a posiive repuaion effec). 3. Daa and Mehodology This sudy has wo sample ses. The firs se consiss of all NASDAQ-lised socks ha were lised on he GSM and NGM for he six monhs surrounding he resrucuring. The GSM is imporan for analyzing any poenial posiive asse pricing impac when NNM socks were elevaed o he new GSM. The remaining NNM socks ha were lef behind in he new NGM are also analyzed in order o deermine if hey had any negaive asse pricing impac for no meeing he new higher sandards of he GSM. The Cener for Research in Securiy Prices (CRSP) daabase serves as he primary daa source for idenifying all NASDAQ-lised socks, as well as o which ier hey are assigned. The sample se consiss of he 1,210 socks lised on he GSM and he 1,354 socks lised on he NGM from 1 July 31 Dec 2006 (hus eliminaing he NCM). This provides a oal of 2,564 socks for he asse pricing analysis via even sudy. A wo-sep procedure is used o calculae abnormal reurns using he Fama-French hree-facor model (1993) as a benchmark. In he firs sage, he benchmark parameers are esimaed, using a 255-day esimaion period ha ends 46 days before each even dae, using equaion 1. R ˆ ˆ R sˆ SMB hˆ HML (1) m In equaion 1, R m represens he rae of reurn of a marke index (S&P 500) for day, SMB represens he average reurn on hree small marke-capializaion porfolios minus he average reurn on hree large marke-capializaion porfolios, and HML represens he average reurn on wo high book-o-marke equiy porfolios minus he average reurn on wo low book-o-marke equiy porfolios, and is a random variable assumed o have an expeced value of zero, be homoskedasic, and be uncorrelaed wih R m or R k (for any k ), or s (for any s 1). reurns are hen esimaed in he second sage. The abnormal reurn will be calculaed using equaion 2. A R Rˆ R m ˆ ˆ R sˆ SMB hˆ HML (2) For he even sudy analysis, I use four measures o analyze abnormal reurns in order o idenify any poenial asse pricing effec resuling from NASDAQ s resrucuring. These measures are average abnormal reurn, cumulaive Published by Sciedu Press 5 ISSN E-ISSN

6 average abnormal reurn, buy-an-hold abnormal reurn, and average compounded abnormal reurn. The funcional forms of each abnormal measure are displayed in equaions 3-6. Average : Average : AAR N A 1 (3) N Buy-and-Hold : CAAR 1 N T 2 A N 1 T 1 T 2 1 ˆ 1 R 1 T 2 T 2 T1 1 BHAR ˆ, T 1, T R 2 a m (5) T1 T1 Average ed : (4) ACAR T1, T 2 1 N N 1 BHAR, T1, T 2 (6) The even sudy es saisic is he non-parameric generalized sign es (Cowan, 1992). The generalized sign es conrols for he normal asymmery beween posiive and negaive reurns during he esimaion period. The generalized sign es is a beer es for even sudies han he Paell es (1976) due o he Paell es s assumpion of cross-secional independence in he abnormal reurn. For sensiiviy analyses, I use he Fama-French four-facor model wih a momenum facor, recommended by Carhar (1997), o measure abnormal reurns. R ˆ ˆ R sˆ SMB hhml ˆ uˆ UMD (7) m In equaion 7, R m, SMB, and HML represen he same variables as he Fama-French hree-facor model. In addiion, UMD represens he average reurn on wo high prior-reurn porfolios minus he average reurn on wo low prior-reurn porfolios. Addiionally, is a random variable assumed o have an expeced value of zero, homoskedasic, and uncorrelaed wih R m, R k (for any k ), or s (for any s 1). reurns are hen esimaed in he second sage. The abnormal reurn will be calculaed using equaion 7. A R Rˆ R m ˆ ˆ R sˆ SMB hˆ HML uˆ UMD (8) The same four measures of abnormal reurn (equaions 3-6) will be used o measure any asse pricing impac due o NASDAQ s resrucuring. I es for an asse pricing impac using hree alernae even daes. The Press Release Dae is he dae of NASDAQ s original press release announcing he resrucuring (15 Feb 2005). The Idenificaion Dae is he dae NASDAQ idenified he specific socks designaed for lising on he new GSM (26 June 2006). The Effecive Dae is firs rading day of he new NASDAQ srucure (3 July 2006). As a componen of he sensiiviy analysis, hese alernae daes will consider wheher any pricing impac occurred on he iniial announcemen of he resrucuring, or on he dae ha specific socks were idenified for each ier, raher han simply he firs day of rading on he new ier. The second daa se consiss of all IPOs ha wen public in he five years surrounding he NASDAQ reorganizaion (30 monhs prior unil 30 monhs afer). The Field-Rier daase idenifies all IPOs during his ime period. I exclude all socks wihou a CRSP share class code of 11 or 12 (excludes all closed end funds, REITs, cerificaes, ADRs, Published by Sciedu Press 6 ISSN E-ISSN

7 uni russ, ec.). The CRSP and Compusa daabases provide addiional company-specific daa used in he mulivariae analysis. An iniial analysis of he IPO marke from indicaes ha a oal of 803 firms conduced IPOs in he 5-year period. A oal of 462 firms conduced IPOs in he hiry monhs before he resrucuring, compared wih 341 aferwards. During he 5-year sample period, NASDAQ araced 40% of heir IPOs afer he resrucuring, versus 60% in he same imeframe before i, implying ha he resrucuring may no have helped hem become more compeiive. This analysis does ake ino consideraion ha no all socks qualify for lising on all hree maor exchanges (NYSE, AMEX, and NASDAQ). Many firms qualify for lising on NASDAQ, bu no NYSE, due o NASDAQ s lower lising requiremens. For he analysis in he resuls secion, hese smaller firms were excluded from he sample se, hereby esablishing a condiion ha firms chose NASDAQ condiional upon heir being qualified o choose beween NASDAQ, AMEX, and NYSE. Previous research by Corwin and Harris (2001) on IPO exchange lising choice idenified ha IPOs are more likely o lis on he same exchange as heir indusry peers. Addiionally, smaller and riskier firms end o lis on NASDAQ. Consequenly, he Corwin and Harris sudy provides hree conrols for examining any poenial impac of he resrucuring on NASDAQ s compeiiveness (indusry concenraion, firm size, and firm risk). For his analysis, I ll conduc a probi model using he conrol variables idenified in he Corwin and Harris (2001) sudy. The probi model akes he form of: NASDAQ 1 Z Prob * where (-) denoes he sandard normal disribuion, denoes a vecor of coefficiens, and Z denoes a vecor of independen (i.e. explanaory) variables. In his analysis, he dependen variable will equal one if he IPO lised on NASDAQ, and will equal zero oherwise (i.e. he firm chose o lis on NYSE or AMEX). The explanaory variables comprising Z will include: -NASDAQ indusry share: indicaes he percenage of firms wihin a company s indusry, using he four-digi SIC o idenify indusry, ha are lised on NASDAQ (peer-firm lisings), - Marke value: indicaes he IPO s pos-lising marke value (shares ousanding imes share price), -Sandard deviaion: indicaes he IPO s level of risk by using he sandard deviaion, as calculaed using he five-day close-o-close reurns in he 100 rading days immediaely following is lising, -Pos_Resrucuring_IPO: indicaes a dummy variable equal o one if he IPO occurred afer 1 July 2006, and equal zero o oherwise; his variable is he variable of ineres, and will be inerpreed as suppor for H 2 (i.e. he resrucuring enhanced NASDAQ s repuaion) if posiive and significan. For sensiiviy analysis, I also conduc a logisic regression model using he same funcional form. 4. Evidence The resuls of he even sudy on he NASDAQ resrucuring, firs focusing on he Global Selec Marke socks, are repored in Table 1. Panel A shows he announcemen effec when NASDAQ firs announced he resrucuring on 15 February, A he iniial announcemen, NASDAQ did no specify which firms would be lised on which iers. While he even sudy resuls do indicae saisically significan negaive reurns for GSM socks in he days following he iniial announcemen, he negaive reurns are consisen wih he overall movemen in GSM socks in he days leading up o he press release. On average, GSM socks had a cumulaive abnormal reurn of -2.25% in he 30 rading days leading up o he announcemen. If you rese he abnormal reurn o zero afer he close of rading he day before he announcemen, he GSM socks coninued, on average, o have a -2.96% abnormal reurn over he subsequen 30 rading days. Published by Sciedu Press 7 ISSN E-ISSN

8 Table 1. Even sudy resuls, NASDAQ global selec marke (FF3FM) Day Panel A: Press Release Dae Panel B: Idenificaion Dae Panel C: Effecive Dae Posiive/ Negaive Raio Posiive/ Negaive Raio Posiive/ Negaive Raio % -0.03% 581: % 0.28% 733: *** 0.28% 0.28% 720: *** % -2.37% 501: *** -0.16% 1.75% 543: ** 0.51% 1.62% 782: *** % -2.33% 585: % 2.16% 800: *** 0.74% 2.36% 826: *** % -2.49% 522: *** 0.15% 2.31% 675: *** -0.05% 2.31% 531: *** % -2.88% 438: *** 0.57% 2.88% 797: *** -0.11% 2.20% 542: ** % -3.16% 479: *** -0.23% 2.65% 502: *** -0.22% 1.98% 532: *** % -3.21% 549: * 0.51% 3.16% 787: *** 0.09% 2.07% 646: *** % -3.41% 505: *** 0.75% 3.91% 824: *** -0.10% 1.97% 612: % -3.49% 583: % 3.86% 531: *** -0.02% 1.95% 527: *** % -3.42% 595: % 3.73% 543: ** 0.10% 2.05% 666: *** % -3.25% 690: *** -0.22% 3.51% 530: *** -0.14% 1.91% 516: *** % -3.14% 596: % 3.61% 654: *** 0.34% 2.25% 713: *** % -3.38% 491: *** -0.10% 3.51% 608: % 2.29% 647: *** % -3.23% 707: *** -0.02% 3.49% 519: *** 0.14% 2.43% 668: *** % -3.12% 656: *** 0.09% 3.58% 659: *** 0.01% 2.44% 593: % -2.25% 614: * -0.14% 3.44% 514: *** -0.56% 1.88% 417: *** % -0.01% 553: % 0.33% 709: *** -0.23% -0.23% 499: *** % -0.20% 508: *** 0.04% 0.37% 649: *** -0.04% -0.27% 579: % -0.52% 471: *** 0.13% 0.50% 670: *** -0.09% -0.36% 567: % -0.77% 499: *** 0.00% 0.50% 590: % -0.23% 694: *** % -0.67% 667: *** -0.57% -0.07% 416: *** 0.23% 0.00% 745: *** % -0.90% 484: *** -0.23% -0.30% 497: *** 0.14% 0.14% 627: ** % -0.89% 549: * -0.04% -0.34% 582: % 0.03% 565: % -0.95% 545: ** -0.09% -0.43% 570: % 0.22% 715: *** % -1.16% 553: % -0.30% 696: *** 0.16% 0.38% 709: *** % -1.22% 533: *** 0.22% -0.08% 741: *** 0.51% 0.89% 798: *** % -1.46% 476: *** 0.14% 0.06% 622: * 0.46% 1.35% 761: *** % -1.63% 492: *** -0.11% -0.05% 569: % 1.24% 588: % -1.84% 556: % 0.15% 718: *** 0.04% 1.28% 652: *** % -1.54% 734: *** 0.16% 0.31% 708: *** 0.16% 1.44% 697: *** % -1.39% 678: *** 0.51% 0.82% 804: *** 0.26% 1.70% 668: *** % -1.40% 555: % 1.27% 760: *** -0.03% 1.67% 561: % -2.96% 451: *** 0.11% 1.25% 636: *** 0.21% 2.28% 621: ** -30, % *** -3.42% *** 3.59% *** 3.29% *** 2.42% *** 2.12% 9.337*** -1,0 0.03% 1.593** 0.03% 1.362** 0.19% 3.385*** 0.18% 3.156*** -0.79% *** -0.81% *** 0, % *** -0.20% *** 0.37% 8.024*** 0.36% 7.737*** -0.27% *** -0.28% *** +1, % *** -3.14% *** 0.92% 6.592*** 0.62% 5.619*** 2.52% 9.680*** 2.25% 8.592*** n=1,210 *, **, *** denoes saisical significance a he.1,.05,.and.01 levels of significance. Similarly, Panel B shows he resuls around he dae NASDAQ idenified he socks ha would migrae o he GSM. The resuls indicae he opposie effec around he idenificaion dae. In he 30 rading days afer he announcemen, GSM socks, on average, had posiive abnormal reurns. Neverheless, his paern follows he GSM marke-wide paern in he days leading up o he announcemen. In he 30 rading days before he idenificaion dae, GSM socks had accumulaed, on average, an abnormal reurn of 3.44%. If you rese he abnormal reurn on he close of rading on he day preceding NASDAQ idenifying he fuure GSM socks, he socks only gained on average an addiional 1.25% of abnormal reurn in he nex 30 rading days. Panel C shows he resuls on he effecive dae ha rading commenced on he new GSM, July 3, These resuls are consisen wih he idenificaion dae (overall posiive movemen in he GSM sock prices in he 30 days leading up o he announcemen). Given he shor imeframe beween he idenificaion dae and he announcemen dae (one rading week), his resul is no surprising. Table 2 shows he resuls of a sensiiviy es using he Fama-French Four Facor model, which includes Carhar s momenum facor. The findings for all hree announcemen daes are similar o he hree-facor model. GSM socks show pos-announcemen abnormal reurn paerns ha are consisen wih he shor-erm momenum wihin he GSM group of socks leading up o he announcemen. Socks are falling boh before and afer he press release, and rising boh before and afer he idenificaion and effecive daes. These resuls are displayed graphically in Figures 3 and 4. Figure 3 shows he abnormal reurns, saring from he even days, whereas Figure 4 shows he abnormal reurns daing back o he beginning (-30) of he pre-even window. Published by Sciedu Press 8 ISSN E-ISSN

9 Table 2. Even sudy resuls, NASDAQ global selec marke (FF4FM) Day Panel A: Press Release Dae Panel B: Idenificaion Dae Panel C: Effecive Dae Posiive/ Negaive Raio Posiive/ Negaive Raio Posiive/ Negaive Raio % -0.03% 592: % 0.26% 712: *** 0.28% 0.28% 721: *** % -2.36% 505: *** -0.17% 1.62% 537: *** 0.46% 1.47% 776: *** % -2.32% 583: % 2.03% 791: *** 0.74% 2.21% 826: *** % -2.48% 522: *** 0.13% 2.16% 657: *** -0.01% 2.20% 568: % -2.87% 433: *** 0.55% 2.71% 786: *** -0.08% 2.12% 559: % -3.15% 479: *** -0.23% 2.48% 505: *** -0.21% 1.91% 534: *** % -3.20% 544: ** 0.48% 2.96% 781: *** 0.09% 2.00% 646: *** % -3.40% 507: *** 0.75% 3.71% 823: *** -0.10% 1.90% 612: % -3.48% 576: % 3.69% 558: % 1.92% 545: ** % -3.41% 587: % 3.58% 551: * 0.10% 2.02% 665: *** % -3.23% 679: *** -0.21% 3.37% 536: *** -0.12% 1.90% 528: *** % -3.12% 590: % 3.47% 654: *** 0.33% 2.23% 709: *** % -3.35% 487: *** -0.10% 3.37% 612: % 2.26% 644: *** % -3.19% 701: *** 0.00% 3.37% 532: ** 0.12% 2.38% 668: *** % -3.08% 649: *** 0.09% 3.46% 660: *** 0.03% 2.41% 602: % -3.04% 615: * -0.13% 3.33% 524: *** -0.56% 1.85% 416: *** % -0.01% 553: % 0.32% 707: *** -0.21% -0.21% 503: *** % -0.21% 526: *** 0.03% 0.35% 641: *** -0.06% -0.27% 566: % -0.53% 471: *** 0.12% 0.47% 667: *** -0.10% -0.37% 564: % -0.78% 504: *** 0.02% 0.49% 601: % -0.28% 650: *** % -0.68% 667: *** -0.57% -0.08% 418: *** 0.19% -0.09% 733: *** % -0.92% 491: *** -0.22% -0.30% 502: *** 0.15% 0.06% 633: *** % -0.91% 546: ** -0.06% -0.36% 571: % -0.03% 576: % -0.98% 538: ** -0.10% -0.46% 559: % 0.16% 717: *** % -1.19% 552: * 0.09% -0.37% 654: *** 0.16% 0.32% 712: *** % -1.26% 551: * 0.19% -0.18% 729: *** 0.50% 0.82% 794: *** % -1.50% 477: *** 0.15% -0.03% 626: ** 0.46% 1.28% 757: *** % -1.67% 487: *** -0.10% -0.13% 582: % 1.18% 586: % -1.87% 545: ** 0.19% 0.06% 719: *** 0.03% 1.21% 653: *** % -1.57% 740: *** 0.16% 0.22% 714: *** 0.13% 1.34% 674: *** % -1.42% 671: *** 0.50% 0.72% 803: *** 0.27% 1.61% 668: *** % -1.43% 551: * 0.45% 1.17% 763: *** -0.04% 1.57% 560: % -2.98% 458: *** 0.11% 1.14% 634: *** 0.22% 2.20% 634: *** -30, % *** -3.39% *** 3.45% *** 3.14% *** 2.40% *** 2.09% 9.009*** -1,0 0.02% 1.586** 0.03% 1.471** 0.19% 3.525*** 0.19% 3.468*** -0.77% *** -0.79% *** 0, % *** -0.21% ** 0.35% 8.164*** 0.35% 7.935*** -0.27% *** -0.28% *** +1, % *** -3.14% *** 0.82% 6.388*** 0.53% 5.358*** 2.39% *** 2.13% 8.780*** n=1,210 *, **, *** denoes saisical significance a he.1,.05, and.01 levels of significance. Global Selec Marke (Even Sudy Resuls, ) 6.00% 4.00% 2.00% Percen 0.00% -2.00% -4.00% -6.00% Even Day Press Release Dae Idenificaion Dae Effecive Dae Figure 3. GSM, pos-announcemen window Published by Sciedu Press 9 ISSN E-ISSN

10 Global Selec Marke (Even Sudy Resuls, ) 6.00% 4.00% 2.00% Percen 0.00% -2.00% -4.00% -6.00% Even Day Figure 4. GSM, full 61-day even window The socks evenually designaed for he NASDAQ Global Marke show a complee opposie paern around he same even daes. Table 3 shows he resuls of an even sudy on he 1,354 socks ha were lef behind in he middle ier as a resul of he resrucuring. While he GSM socks were clearly rending downward as a group leading up o he February 15, 2006 press release, he NGM socks were rending upward. The resuls in Panel A indicae ha NGM socks had an average abnormal reurn of 1.9% in he 30 rading days leading up o he press release, and susained ha rend for an addiional 1.25% of abnormal reurn in he 30 rading days afer he announcemen. Table 3. Even sudy resuls, NASDAQ global marke (FF3FM) Day Panel A: Press Release Dae Panel B: Idenificaion Dae Panel C: Effecive Dae Posiive/ Negaive Raio Posiive/ Negaive Raio Posiive/ Negaive Raio % -0.37% 550: *** -0.18% -0.18% 683: % 0.02% 704: ** % 1.32% 608: ** 0.14% -0.24% 717: *** -0.15% -0.04% 733: *** % 1.36% 587: *** -0.21% -0.45% 689: % -0.51% 680: % 1.32% 595: *** -0.01% -0.46% 696: % -0.64% 641: % 1.71% 629: % -0.69% 662: % -0.61% 626: ** % 1.81% 639: % -0.63% 694: % -0.62% 661: % 1.94% 662: % -0.78% 725: *** -0.18% -0.80% 695: % 2.06% 661: % -1.24% 674: % -0.70% 685: % 2.03% 675: % -1.38% 636: % -0.87% 612: ** % 2.03% 643: % -1.36% 612: *** -0.21% -1.08% 656: % 2.05% 674: % -1.38% 660: % -1.15% 627: ** % 2.08% 649: % -1.56% 696: % -1.40% 632: * % 2.08% 633: % -1.47% 688: % -1.59% 655: % 2.00% 654: % -1.64% 611: *** 0.03% -1.56% 700: * % 2.19% 688: ** -0.20% -1.84% 657: % -1.64% 650: % 2.10% 630: % -1.90% 623: ** 0.29% -1.35% 706: ** % 0.15% 651: % -2.15% 630: * -0.16% -0.16% 651: % 0.20% 641: % -2.34% 650: % 0.07% 707: ** % 0.23% 627: % -2.31% 705: ** -0.01% 0.06% 697: % 0.11% 609: ** -0.08% -2.39% 650: % -0.04% 729: *** % 0.06% 632: % -2.09% 710: ** -0.31% -0.35% 688: % 0.02% 616: % -2.26% 652: % -0.26% 685: % 0.08% 629: % -2.03% 703: * -0.02% -0.28% 657: % 0.21% 644: % -2.05% 689: % -0.66% 641: % 0.34% 693: ** -0.10% -2.15% 730: *** -0.23% -0.89% 666: % 0.41% 637: % -2.46% 693: % -1.14% 680: % 0.40% 630: % -2.37% 683: % -1.28% 700: * % 0.57% 654: % -2.39% 650: % -1.21% 689: % 0.69% 694: *** -0.37% -2.76% 645: % -1.11% 757: *** % 0.64% 700: *** -0.23% -2.99% 670: % -1.31% 680: % 0.61% 667: % -3.24% 672: % -1.56% 616: *** % 0.63% 674: % -3.38% 707: ** 0.01% -1.55% 671: % 1.24% 638: % -4.36% 687: % -2.92% 635: Press Release Dae Idenificaion Dae Effecive Dae -30, % 3.805*** 2.01% 2.444** -1.81% *** -2.37% *** -1.60% % *** -1,0 0.06% -0.44**** 0.07% ** -0.31% *** -0.31% *** 0.14% % 0.898** 0, % 0.594*** 0.21% 0.376** -0.43% *** -0.44% *** 0.07% % 0.045** +1, % 2.363*** 0.92% 0.622** -2.21% 0.296** -3.01% *** -2.77% * -3.22% *** n=1,354 *, **, *** denoes saisical significance a he.1,.05, and.01 levels of significance. Published by Sciedu Press 10 ISSN E-ISSN

11 Similarly, as shown in Panels B-C, he downward rend in NGM sock prices in he 30 rading days before he idenificaion and announcemen daes was susained over he subsequen 30 rading days. The Fama-French Four Facor models, shown in Table 4, show he same rends. Figures 5 and 6 show he resuls graphically. Wha appears in Figure 5 o be a posiive announcemen effec, followed by a susained abnormal reurn in he subsequen shor-erm, appears in Figure 6 o be lile more han shor-erm price momenum. The findings indicae ha you have wo significan porions of he NASDAQ marke moving clearly in wo differen direcions, over wo differen imeframes, as NASDAQ was iniially announcing, and hen implemening, a maor reorganizaion of heir lising environmen. Table 4. Even sudy resuls, NASDAQ global marke (FF4FM) Day Panel A: Press Release Dae Panel B: Idenificaion Dae Panel C: Effecive Dae Posiive/ Negaive Raio Posiive/ Negaive Raio Posiive/ Negaive Raio % -0.31% 564: *** -0.22% -0.22% 668: % 0.02% 704: ** % 1.25% 586: *** 0.12% -0.45% 704: ** -0.17% -0.17% 701: * % 1.29% 585: *** -0.21% -0.66% 679: % -0.63% 677: % 1.26% 601: ** -0.05% -0.71% 681: % -0.74% 655: % 1.65% 623: % -0.97% 668: % -0.70% 626: ** % 1.76% 641: % -0.91% 688: % -0.71% 669: % 1.87% 665: % -1.11% 704: ** -0.17% -0.88% 695: % 2.00% 662: % -1.57% 673: % -0.78% 689: % 1.98% 679: * -0.09% -1.66% 647: % -0.93% 624: ** % 1.96% 629: % -1.60% 616: *** -0.20% -1.13% 660: % 1.93% 664: % -1.60% 668: % -1.19% 631: * % 1.95% 638: % -1.78% 694: % -1.45% 623: ** % 1.92% 612: * 0.10% -1.68% 689: % -1.64% 652: % 1.83% 642: % -1.80% 630: * 0.02% -1.62% 698: % 2.00% 669: % -2.00% 657: % -1.69% 654: % 1.90% 623: % -2.04% 630: * 0.29% -1.40% 707: ** % 0.17% 661: % -0.26% 622: ** -0.15% -0.15% 649: % 0.27% 650: % -0.45% 651: % 0.07% 706: ** % 0.29% 626: % -0.43% 695: % 0.05% 684: % 0.18% 619: % -0.49% 657: % -0.08% 710: ** % 0.13% 624: % -0.20% 706: ** -0.34% -0.42% 676: % 0.11% 619: % -0.34% 663: % -0.33% 688: % 0.16% 629: % -0.13% 700: * -0.01% -0.34% 666: % 0.32% 654: % -0.16% 685: % -0.71% 640: % 0.42% 688: ** -0.15% -0.31% 709: ** -0.22% -0.93% 670: % 0.55% 656: % -0.67% 675: % -1.19% 680: % 0.57% 650: % -0.57% 685: % -1.34% 702: * % 0.73% 652: % -0.57% 666: % -1.28% 686: % 0.82% 681: * -0.37% -0.94% 640: % -1.18% 758: *** % 0.73% 689: ** -0.22% -1.16% 672: % -1.40% 679: % 0.68% 659: % -1.43% 670: % -1.66% 613: *** % 0.66% 657: % -1.57% 708: ** 0.00% -1.66% 664: % 1.25% 627: % -2.60% 689: % -3.06% 639: , % 1.776* 1.98% 3.355*** -2.01% ** -2.58% *** -1.67% % *** -1,0 0.07% % *** -0.30% *** -0.30% *** 0.14% % 0.51*** 0, % % 1.178** -0.45% *** -0.46% *** 0.07% % 0.297** +1, % % 2.294** -2.33% *** -3.16% *** -2.91% % *** n=1,354 *, **, *** denoes saisical significance a he.1,.05, and.01 levels of significance. Published by Sciedu Press 11 ISSN E-ISSN

12 Global Marke (Even Sudy Resuls, ) 6.00% 4.00% 2.00% Percen 0.00% -2.00% -4.00% -6.00% Even Day Press Release Dae Idenificaion Dae Effecive Dae Figure 5. NGM, pos-announcemen window Global Marke (Even Sudy Resuls, ) 6.00% 4.00% 2.00% Percen 0.00% -2.00% -4.00% -6.00% Even Day Press Release Dae Idenificaion Dae Effecive Dae Figure 6. NGM, full 61-day even window These even sudy resuls seem inconsisen wih any posiive repuaion effec NASDAQ may have inended. While a brief analysis of he 0-30 rading day window would seem o indicae ha GSM socks may have benefied from being moved ono he new ier, and ha he NGM socks may have suffered, heir abnormal reurn paerns were no differen han in he weeks leading up o he announcemen han hey were immediaely aferwards. Figure 7 shows abnormal reurns for GSM and NGM socks over he 61-day window surrounding idenificaion and effecive daes. Published by Sciedu Press 12 ISSN E-ISSN

13 Global vs. Global Selec Marke (Even Sudy Resuls, ) 6.00% 4.00% 2.00% Percen 0.00% -2.00% -4.00% -6.00% Even Day Global Selec, Idenificaion Dae Global Selec, Effecive Dae Global, Idenificaion Dae Global, Effecive Dae Figure 7. GSM vs. NGM, full 61-day even window These resuls seem consisen wih wo possible explanaions. Firs, he reorganizaion had lile o no impac on any NASDAQ-lised firms, as heir pricing appeared relaively unaffeced. Any posiive (negaive) pricing impac for GSM (NGM) socks could be explained away by echnical analysis of shor-erm price movemens. Second, he marke may have inferred which companies would fall ono which iers, and prices sared moving well before he firms were officially announced as moving o he GSM or NGM. If NASDAQ s new iered srucure benefis firms wih any enhanced repuaion effec, a beer es would be o analyze when firms cross from one ier ino a new ier. As NCM or NGM firms grow, become more profiable, and heir sock becomes more liquid, hey would mee he higher lising sandards of he NGM or GSM. Wih a subsequen move ono a higher ier wih beer visibiliy, hey could reap a posiive impac on heir sock price hrough higher levels of invesor paricipaion. Conversely, as GSM or NGM firms become less profiable, and heir sock becomes less liquid, hey would fail o mee he coninued lising sandards of he GSM or NGM, and would drop o a lower ier. Wih a subsequen move ono a lower ier wih less visibiliy, hey could face a negaive impac on heir sock price. Tables 5 and 6 explore his effec as firms cross hese boundaries. Table 5 focuses on firms moving from he NCM o he NGM, or from he NGM o he GSM. Panel A repors abnormal reurns from he dae he firm announced is inenion o move ono a higher ier, and Panel B repors reurns from he effecive dae (usually only a lag of 1-2 rading days). Mos firms announce a rise o a new ier using a formal press release, a NASDAQ announcemen, or an SEC filing. Some firms choose no o announce he move, or he announcemen could no be locaed. Thus, he number of observaions for he announcemen dae is slighly smaller han for he effecive dae. If neiher an announcemen nor an effecive dae could be esablished, he firm was hrown ou of he sample. Published by Sciedu Press 13 ISSN E-ISSN

14 Table 5. Even sudy resuls, socks moving o higher ier Day % 0.03% % 0.12% % 0.14% % 0.18% % -0.81% ** -0.97% -1.20% ** -0.47% -0.37% % -0.73% % -0.90% % -1.23% % -0.41% % -0.73% % -1.22% % -1.53% % -0.52% % -0.79% % -0.78% % -1.02% % -1.06% % -1.33% % -0.02% % -0.31% % -0.53% % -0.84% % 0.62% % 0.31% % 0.35% % -0.01% % 0.53% % 0.21% % 0.24% % -0.15% % 0.90% % 0.59% % -0.26% % -0.54% % 0.19% * -0.60% -0.01% * -0.24% -0.50% % -0.82% % -0.20% % -0.41% % -0.89% % -1.18% % -0.27% % -0.54% % -0.81% % -1.19% % -1.03% % -1.42% % -0.82% % -1.28% % -0.63% % -1.04% % -0.84% % -1.31% % -1.22% % -1.65% % -1.01% % -1.40% % -1.47% % -1.93% % -1.68% % -2.12% % -0.73% % -1.19% % -1.07% % -1.55% % -1.44% % -1.96% % -1.66% % -2.30% % -2.11% % -2.68% % -2.37% % -3.10% % -2.21% % -2.93% % -3.12% % -3.78% % -2.72% % -3.47% % -2.59% % -3.23% % -3.25% % -3.90% % -2.88% % -3.50% % -3.44% % -4.14% % -3.15% % -3.84% ** % -3.42% % -4.05% % -2.96% 2.302** 0.18% -3.66% % -3.80% % -4.52% % -3.32% % -4.04% % -3.78% % -4.52% % -3.85% % -4.62% % -3.74% % -4.44% % -3.82% % -4.52% % -3.26% % -3.86% % -4.51% % -5.10% % -3.41% % -3.94% % -4.60% % -5.30% % -4.27% % -4.93% % -4.94% % -5.73% % -4.70% % -5.48% % -5.12% % -6.01% % -4.84% % -5.64% % -5.43% % -6.37% * % -6.33% % -7.67% % -6.13% % -7.50% Panel A: Announcemen Dae n=76 Panel B: Effecive Dae n=84 FF3FM FF4FM FF3FM FF4FM -30, % % % % % % % % ,0 0.49% % % % % % % % , % % % % % % * -5.95% * -6.63% * *, **, *** denoes saisical significance a he.1,.05, and.01 levels of significance. The analysis from Table 5 indicaes ha upon announcing a move o a higher ier, and upon he beginning of rading on he new ier, firms have an immediae posiive, bu non-significan announcemen effec, followed by a shor-erm reversal over he nex 30 rading days. Boh even day windows (-1 o 0, and 0 o +1) are insignifican for boh he announcemen and effecive daes, and using boh he hree-facor and four-facor models. The longer pos-even window (+1 o +30) is significan for he effecive dae, and approaching significance for he announcemen dae. Surprisingly, NASDAQ socks appear o have a negaive pricing impac when hey move ono a higher ier. Table 6 displays resuls for when socks move o a lower ier. Even more surprisingly, NASDAQ socks appear o have a srong posiive price impac when dropping o a lower ier. While he immediae price impac doesn appear o happen, NASDAQ socks ha move ono a lower ier appreciae noiceably in he 30 rading days immediaely afer boh he announcemen and he swich. This finding is significan in 6 of he 8 specificaions (hree-facor vs. four-facor models, announcemen vs. effecive daes, and cumulaive vs. compound reurns). Published by Sciedu Press 14 ISSN E-ISSN