Mergers & Acquisitions and Innovation Performance in the Telecommunications Equipment Industry

Transcription

1 Mergers & Acqustons and Innovaton Performance n the Telecommuncatons Equpment Industry Tseveen European Unversty Vadrna Andreas Stephan Jönköpng Internatonal Busness School, DIW Berln, CESIS Stockholm ABSTRACT The telecommuncatons equpment ndustry wtnessed an enormous worldwde round of Mergers & Acqustons (M&A). Ths paper examnes the nnovaton determnants of M&A actvty and the consequences of M&A on the technologcal potental and the nnovaton performance. We extend the resource-based theory n elucdatng external technology sourcng and provde emprcal evdence on the keen relance of the equpment frms on M&A as a technology sourcng strategy for the perod Employng the matchng propensty score approach, ths study provdes evdence that mergers realze a sgnfcant growth n the nnovaton performance of frms. The post-merger nnovaton performance s, n turn, drven by both the pror success n nhouse R&D commtment and the deteroraton of nternal technologcal capabltes at acqurng frms. Keywords: Mergers & Acqustons, Innovaton Performance, Telecommuncatons Equpment Industry JEL Classfcatons: O30, G34, L63 We would lke to thank Vvek Ghosal and Mchal Grajek as well as semnar partcpants at the IIOC Conference n Savannah, the EARIE Conference n Valenca, and the Conference on the Economcs of ITC n Pars for ther helpful comments and suggestons. We are also grateful to the edtor Paul Nghtngale and three anonymous referees for ther constructve comments and suggestons on the prevous verson of the paper. The standard dsclamer Correspondng author: European Unversty Vadrna, Faculty of Economcs and Management Studes, Grosse Scharrnstrasse 59, D Frankfurt (Oder), E-mal: gantumur@euv-frankfurt-o.de

2 1. Introducton It s strkng that most Mergers and Acqustons (M&A) are assocated wth technologcal and/or regulatory shocks. Ths pattern supports the large sample results of Mtchell and Mulhern (1996) that merger actvty s related to ndustry shocks. Schoenberg and Reeves (1999) argue that deregulaton s the sngle most mportant factor determnng acquston actvty at the ndustry level. Another study by Andrade et al. (2001) shows that deregulaton becomes a domnant factor n M&A actvty after the late 1980s and accounts for nearly half of the merger actvty snce then. Le Blanc (2002) ponts out the dstncton of exogenous shocks to the ndustry structure, both technologcal (nnovaton) and nsttutonal (deregulaton), and the strategc nature of mergers to alter the ndustry structure, mprove ther compettve poston, and ncrease ther market power. At the same tme, a complementary explanaton of merger waves nvolves endogenous effects, where an ntal merger n an ndustry trggers a chan of successve mergers. The smultaneous mpact of both effects s partcularly relevant for the telecommuncatons ndustry, where a radcal change n the envronment, e.g. the 1996 Telecom Act n the US and the 1998 deadlne for telecom markets lberalzaton n Europe, pushed some poneer frms to decde and quckly embark on a technologcal merger strategy, trggerng smlar moves n the whole ndustry through some knd of subsequent race for assets and target frms (Fgure 1). Durng ths perod, R&D and nnovaton ncreasngly shfted toward the producers of telecommuncatons equpment. As the trade and regulatory lberalzaton prmarly has globalzed the demand for telecommuncatons equpment, technologcal change n the ndustry has had upstream effects on R&D (Fgure 2). Moreover, the growth n patentng has been tremendous - from 1988 to 1998 the number of communcaton equpment patents appled by the UPSTO ncreased by more than four tmes (Fgure 3). As the telecommuncatons equpment ndustry was opened to entry, some frms chose to be nche manufactures such as AVM 1 or Pandatel 2, offerng a narrow range of equpment products, whereas bg players such as Alcatel-Lucent, Noka-Semens, Motorola, Ercsson adopted a strategy to expand beyond ther former boundares. In between there are numerous frms operatng n a specfc equpment sub-sector wth a partly regonal or global focus. Through nternatonal strateges, frms were not only able to enter foregn markets, but also able to seek foregn assets (both of a tangble and an ntangble nature) and to buld R&D, and supply and producton facltes abroad. External strategc optons such as M&A provde an establshed market poston, access to exstng networks and nfrastructure, to a range of capabltes that they need n order to further develop both core actvtes and complementary actvtes. Whle the regulatory lberalzaton foremost has globalzed the demand for telecommuncatons equpment, technologcal developments have also created new opportuntes and 1 AVM operates n the area of PC connecton through dgtal communcaton technologes such as ISDN, GSM, and xdsl technology. 2 Pandatel s a leadng manufacturer of optoelectronc components and wavelength dvson multplexers.

3 threats for the equpment producers. The emergence of the New Economy and the ntroducton of new technologes such as moble phones and broadband have forced the equpment producers to reconsder ther strategy, ther technologcal base, and ther product portfolo. The technologcal process whch s progressvely blurrng the boundares between nformaton and communcatons technologes has requred technologcal dversfcaton amongst the ncumbents and opened up multple entry optons also for new players (D Mnn and Palmberg, 2006). Buldng on the future changes n the scope of dfferent product markets, the convergence of varous technology subfelds ples up arguments of scope economes, possble market share leveragng, and network effects to motvate M&A. In that context, the equpment producers had to make adequate adaptatons to the technologcal changes and quckly respond to the essental technologcal development often through technologcal acqustons. In order to enhance or to sustan ther compettve advantage M&A have, thus, amed at provdng a unque geographcal coverage or a unque range of varous products wthn and between the segments such as communcaton devces, publc and enterprse network equpment, and system and network management that can be bundled n a sngle proposal. The lterature on M&A n the telecommuncatons ndustry s extensve and mostly focuses on the telecom servce provders (e.g., Jamson, 1998; Km, 2005; Le Blanc, 2002; Rosenberg, 1998; Warf, 2003). Whle the mportance of technologcal nnovaton s wdely acknowledged wthn ths lterature, t s surprsng how lttle attenton t has receved n justfyng M&A. At the same tme, whle the ntense M&A actvty wthn the telecom servce provders passed ts rpple effects on to the telecom equpment producers, we are not aware of any study whch nvestgates the lnkage between recent rses both n M&A and nnovaton actvty n the telecommuncatons equpment ndustry. Ths paper attempts to make contrbuton on two fronts. On the theoretcal front, ths paper ncorporates the nature and type of nnovaton n whch frms engage n explanng technology mergers 3 and extends the resource-based approach n elucdatng external technology sourcng. On the emprcal front, ths paper reveals the keen relance of the telecommuncatons equpment frms on mergers as a technology sourcng strategy. It further addresses mportant exploratory ssues: frst, does the nnovaton actvty of frms depct a sgnfcant predctor of enterng the merger actvty? Second, how do frms that choose mergers and frms that stay outsde of the merger actvty dffer wth respect to ther nnovaton performance? Thrd, the follow up queston s then what are the effects of mergers on the nnovatve performance of frms f we control for the dfferences n nnovaton performance pror to M&A actvtes? The remander of the paper s organzed as follows: Secton 2 outlnes the nnovaton ratonale for mergng n the telecommuncatons equpment ndustry and gves an overvew of the correspondng merger cases. Secton 3 develops a theoretcal framework to our research questons. The emprcal methodology s presented n Secton 4. Secton 5 provdes a descrpton of the data. In 3 In the followng sectons, we use the terms mergers and acqustons nterchangeably by referrng to the combned unt.

4 Secton 6, we present and dscuss emprcal results and ther robustness. Secton 6 concludes wth a summary of our fndngs and some mplcatons for further research. 2. Innovaton Ratonale for Mergng n the Telecommuncatons Equpment Industry Telecommuncatons equpment frms seek to expand ther array of actvty, both n types of products offered and wthn geographcal areas offered. 4 The combnaton of two sets of resources and capabltes allows frms to mplement expanson strateges quckly and effcently. If the mergng frms are complementary, the acqurng frm s able to enter new, lucratve, fast-growng markets, coordnate ts product portfolo or gan new dstrbuton channels. For nstance, the acquston of the publc telecom equpment busness from Robert Bosch Telekom provded Marcon s communcaton dvson wth an excellent opportunty to enter the German market. 5 Moreover, Bosch s Publc Network s broadband wreless access products complemented Marcon s own range of access products. Another example of the transnatonal transactons was the acquston of Phllps by AT&T. 6 Through ths acquston AT&T attaned an nternatonal premer poston n transmsson networks, mcrowave transmsson access, and moble nfrastructure system solutons and provded strong strategc ft wth exstng equpment vendors n Europe. As part of ts strategc move nto next generaton IP carrer networkng, Ercsson acqured TouchWave, Inc., a Slcon Valley based provder of enterprse IP-telephony solutons. The acquston gave Ercsson fully featured Internet Protocol (IP) PBX systems that support wth voce over IP and wreless LAN termnals, enhanced ts presence n the US and targeted new dstrbuton channels. One of the sgnfcant trends n the ndustry s the separaton of component supplers that desgn and manufacture the specalzed components from systems provder that manufacture entre peces of equpment and equpment systems. 7 The component frms play a crucal role n supportng the communcatons equpment supply chan. It s at the component level that many technologcal breakthroughs have been acheved and enables the technologes for the system provders. Although a dvson of labor between systems and components manufacturng s becomng evdent, component manufacturers are provdng more ntegrated modules to ther systems customers rather than just raw components. By ncorporatng the key technologes nto ther product portfolos, systems provders can offer the techncal capabltes wthn ther equpment whch s demanded by ther servce provder customers. The mpact of greater relance by systems provders on ther supplers s that the latter 4 The merger cases we descrbe n ths secton are restrcted to the tme perod whch corresponds wth the tme frame of our emprcal analyss. 5 In ts transacton n 2000, Marcon pad USD153 mllon, whereas most of the assets to be transferred to Marcon are located n Germany. Case COMP/M.1800 (European Commsson) 6 The acquston took place n the framework of the process of restructurng of AT&T whch led to the separaton of the telecommuncatons equpment busness from other groups one year later n Case IV/M.651(European Commsson) 7 Ths trend s partcularly evdent n the optcal communcatons equpment sector (Merrll Lynch, 2000).

5 have to provde broader product lnes n order to offer a greater ntegraton role, or are mpelled to provde platform technologes 8 that can be used to ntegrate dscrete devces. In the optcal components subfeld, the most nnovaton drven sector, a wave of consoldaton occurred led by JDS Unphase, a manufacturer of fber optc products and a leader n the ntegraton of dscrete components nto mult-functonal components and hgh-performance modules. Wthn a year, JDS Unphase had acqured 11 optcal component provders. 9 These acqustons accelerated JDS Unphase s ablty to ntegrate sgnfcant technologes and assocated ntellectual propertes nto ts hgh-value optcal modules. Gven that key breakthroughs n the technology occur at the component level, systems provders seem to have adopted an ntegraton strategy, whch s to leverage the work of component supplers and then to add value and product dfferentaton mostly through system level archtecture and software. Moreover, the characterstcs of the network ndustry also gve the equpment provders an attractveness of the mergng. The frm's compettve advantage n the network ndustry comes from the capacty of ts products to network wth each other n order to form systems whch are adapted to the specalzed needs of ndvdual customers. 10 Network effects, n context of equpment manufacturng, occur when the servce provder s customers value of equpment ncreases wth the number of the customers usng that same equpment or complementary equpment. The rse n the number of servce provder customers usng the equpment ncreases the number of complements for that equpment whch n turn ncreases the value of the equpment. The complexty of the devces leads to two common costs n runnng multvendor networks: costs of learnng new devces and costs of ensurng compatblty and nteroperablty between multple devces (Forman and Chen, 2003). Network externaltes n the router equpment sector, for example, prmarly arse due to the ncompleteness of compatblty of routers and swtches (Tanaka and Murakam, 2003). Although all routers and swtches, whch fall wthn broad range of vendors classfed as networkng equpment provders 11, adopt the same nterface based on the. TCP/IP protocol, the mplementaton of the nterface depends on the vendor. Thus, compatblty wthn routers and swtchers s not perfect among vendors. Hence, the mergers n the network ndustry often produce sgnfcant cost savngs due to ts network specfcs. In some nstances wth sgnfcant economes of scale or when the costs of desgnng components to work wth dfferent systems,.e. nteroperablty, are hgh, a merger can actually be effcent for the market at least for the servce provder customers Platform technologes are technologes that can be used to facltate a broad range of applcatons based actvtes. Access to approprate platform technologes can reduce costs and avod unnecessary duplcaton of facltes, ncrease nternatonal R&D compettveness, and provde an envronment of effectve networkng and collaboraton. 9 Between June 1999 and July 2000 JDS Unphase had acqured 11 frms at a cost of US$ 60.6 bllon. 10 For more on network effects, see Katz and Shapro (1994). 11.e., data-communcatons equpment provders. 12 On the other hand, the mergers n network ndustres can also ncrease the monopoly power of the domnant frm by creatng sgnfcant barrers to entry as customers can be locked-n or ted to a partcular product by sgnfcant nvestments nto that product.

6 Csco Systems the domnant suppler n the market for networkng equpments has been attemptng to create an end-to-end servce network archtecture by pursung a vson of beng all thngs to all users. Csco s frequent acqustons were among others n the area of router technology, some of whch specfcally related to ether software or ASIC 13 development whch represents the two key technologes of router systems producers (Merrll Lynch, 2000). These ASICs are hghly propretary and a substantal proporton of a vendor s ntellectual property s contaned wthn them. Such acqustons where the merger strategy adheres to the technology value and the assocated knowledge base are extremely mportant for the establshed equpment vendors, allowng them to adapt rapdly to the dynamc market envronment. Almost all equpment sub-sectors have been prolfc acqurer of technology. Examples exhbtng Csco s 14 remarkable number of technology acqustons nclude Grante Systems, purchased for ts Ggabt Ethernet technology; Radata, Inc., whch provded Csco wth leadng semconductor technology for developng next generaton wreless networks; Komodo Technology, Inc. wth ts technology on a smooth transton path from tradtonal crcut-swtched networks to new packet-based networks. Even frms that are not large-scaled but are known for ther home-grown products have used external technology sources as a mean of value added nnovaton. Through ts acqustons of four frms wthn a year, ADVA Optcal Networkng, a German-based leadng global provder of end-toend optcal networkng solutons, attaned access to dverse knowledge bases such as software developng n ntegrated access devces, a desgnng and manufacturng of carrer class fber access equpment, and ntellgent storage area networkng gateways as well as R&D sklls and expertse. 15 Each of these transactons was done purposely to gan n-depth knowledge of each technology and to strengthen the poston as a provder of optcal network solutons. Acquston of valuable patents n order to reassert patent portfolo has been a crucal means of technologcal acquston. Acqurng Chpcom ncreased 3Com's granted and pendng U.S. patents from 97 to The patents enhanced the company's nnovaton n dverse technology areas such as ncreasng network throughput, smplfyng nternetworkng connectvty, enablng multmeda over exstng Ethernet lnes, and mprovng LAN securty. Furthermore, n moble nfrastructure equpment, Qualcomm, the world leader of CDMA dgtal wreless technology, acqured SnapTrack Inc., a leader n wreless poston locaton technology. Through ths acquston, Qualcomm obtaned SnapTrack's 13 Advances n Applcaton-Specfc Integrated Crcuts (ASIC) technology undertaken by router manufactures have been responsble for mportant mprovements n router performance. 14 Wth ts relatvely dversfed product range, Csco s unbeatable n ts record of acqustons for nstance, between 1993 and 2000 a total of 71 acqustons took place across the equpment sub-sectors. 15 For around a total of USD195 mllon, ADVA acqured Cellware Broadband (Germany), Frst Fbre (UK), Storage Area Network (UK) and R&D team from Semens n Norway n Wth the combned nnovaton of the Chpcom and 3Com development teams (1995), 3Com acheved a steady stream of new patents n the followng years.

7 patent portfolo of nearly 50 patents, ether ssued or pendng, that are crtcal to the effcent, costeffectve deployment of Wreless Asssted GPS systems. 17 Snce the telecommuncatons equpment ndustry has been shaped by hghly knowledge ntensty, the ratonale of mergng actvtes needs to take nto account technologcal changes whch have modfed the compettve and productve context of frms durng the last two decades. In other words, n order to justfy the desderatum of mergers as an external technology source, t s essental to attend to the nature and type of nnovaton n whch frms engage and ts compettve mpact upon an ndustry. 3. Theoretcal Framework Resource-based theory seeks to brdge the gap between theores of nternal organzatonal capabltes, on the one hand, and external compettve strategy theores, on the other hand (Barney and Clark, 2007). 18 Instead of vewng the frm as an organzaton to mnmze producton costs (producton based theory), the resource-based vew of the frm shfts the focus from cost- to value-consderaton and vews the frm as a creator of benefts (Hoffman and Schaper-Rnkel, 2001). Resource-based theory defnes a frm as a bundle of resources and capabltes and argues that dfferent frms resources and capabltes can be a major reason for the persstent heterogenety n a frm s performance. Ths approach focuses on the characterstcs of valuable resources and capabltes that one frm possesses and that compettor frms fnd dffcult to create on ther own. Resource-based theory, however, mplctly assumes that frms have the same capablty n appropratng the advantages and benefts generated by a technology, or that a technology s compettve advantage can be reaped by any frm no matter whether or not the frm has the capablty to assmlate and explot t. Ths assumpton has been adjusted by the absorptve capacty approach, whch suggests that whether or not a frm can dentfy and assmlate and explot external technologes vares among frms and even vares n the same frm over tme and n dfferent stuatons (Cohen and Levnthal, 1989, 1990; Lane and Lubatkn, 1998). Moreover, wth ther theory about the persstence of the large ndustral corporaton, Nelson and Wnter (1982) and Mowery, Oxley, and Slverman (1996) add a cumulatve dmenson to the theory of the frm by exposng the organzatonal capabltes whch are characterzed by tact knowledge and embeddedness n organzatonal routnes. They emphasze the role of organzatonal dfferences n the frm dynamc capabltes, especally dfferences n abltes to generate and gan from nnovaton and defne dynamc capabltes as the frm s ablty to ntegrate, buld, and reconfgure nternal and external competences to address changng envronments. Followng Joseph Schumpeter (1942), the lterature characterzes dfferent sets of ndustry condtons whch affect the compettve mpact of technologcal nnovatons. New-Schumpeteran 17 In 2000, Qualcomm pad USD 1 bllon n stock for the acquston whch should accelerate the ntroducton of powerful locaton-enabled moble devces. 18 Barney and Clark (2007) provde an n-depth survey on the resource-based theory.

8 authors (Utterback and Abernathy, 1975, 1978; Tushman and Anderson, 1986) have proposed a punctuated equlbrum model for analyzng the technologcal change process. Accordng to ths model, the evoluton of technology-based organzatons and ndustres s characterzed by long perods of stablty and marked by abrupt changes assocated wth the emergence of new technologes. Intally, the lack of a domnant desgn facltates the market entry of frms and the competton among technologcal alternatves, usually product nnovatons. Durng the consoldaton of a domnant desgn, the ndustral structure undergoes several changes due to the frms strateges towards standardzaton. After the emergence of a domnant technology, the nnovatve efforts of the frms n the market are, n general, process nnovatons, and the ndustral structure enters a phase of maturty and stable equlbrum. However, the punctuated equlbrum model does not apply to ndustral settngs that have been characterzed as contnuously chaotc, uncertan, and as hgh-velocty (Esenhardt and Bourgeos, 1988). Instead, a punctuated dsequlbrum model (Page, Wersama, and Perry, 1989) may be more approprate. Ths model assumes that n many ndustres, dscontnuty and change s the norm; that nnovaton s technologcally drven, and that there are no long perods of stable desgn convergence. Under such condtons, frms seekng technologcal know-how wll be more lkely to pursue archtectural nnovaton whch s trggered by a change n the lnkages between core components n an establshed product whle the core desgn concept behnd each component remans the same (Henderson and Clark, 1990). Although there are several other dmensons on whch t may be useful to defne the dfferent types of technologcal change, the use of the term archtectural nnovaton enables us to dentfy nnovatons that have a more sgnfcant mpact on the relatonshps between components than on the core technologes of the components themselves. The telecommuncatons equpment ndustry has experenced a great amount of technologcal nnovaton as a consequence of the convergence process between nformaton and communcaton technologes. The ntal radcal nnovaton was the applcaton of ntegrate crcut technology to allow dgtal swtchng of telephone calls whch changed the core desgn concepts of telephone systems. After the ntroducton of ths radcal nnovaton, the telecommuncatons equpment ndustry has settled on buldng archtectural knowledge that represents learnng a lttle about many dfferent possble desgns n contrast to learnng a great deal about the domnant desgn (Henderson and Clark, 1990). The frms actvely developed both knowledge about alternatve components and knowledge of how these components can be ntegrated. Much of ths nnovaton could be characterzed as archtectural snce t has led to the reconfguraton of components and changes n the way equpment addresses a varety of user needs. Wth respect to the archtectural nnovaton prevalent n the telecommuncatons equpment ndustry, establshed frms often face dffcultes n adaptng to archtectural technologcal change. These dffcultes accrue at dentfcaton and development of archtectural knowledge as well as at applcaton of new archtectural nnovaton (Henderson and Clark, 1990). In the followng, we

9 analytcally examne the merger ratonale accordng to these stages of technologcal evoluton. Certanly, the ratonales behnd mergers ascrbed to these stages are not mutually exclusve. It s feasble to assume that they are at least to some degree ntertwned between the nnovaton stages. A. Mergers for Identfcaton of Archtectural Innovaton Under condtons of uncertanty, frms face the dentfcaton problem: Frms requre sgnfcant tme and resources to dentfy a partcular knowledge as archtectural, snce archtectural nnovaton can often ntally be accommodated wthn old frameworks. Snce core desgn concept remans untouched by archtectural nnovaton, the organzaton may msunderstand the nature of the threat. Ths s dstnctve to establshed frms n the market or market leaders, mostly large frms by whch an approprate nnovaton may be screened out by the nformaton flters and communcaton channels that embody old archtectural knowledge. Moreover, economc models suggest that establshed frms have an ncentve to nvest n ncremental nnovatons that add to ther establshed knowledge base and protect or enhance ther exstng rent stream (Dos, 1988; Teece, 1996). Thus, they seek to maxmze the returns from known technology rather than devote resources for archtectural nnovatons wth an uncertan payoff. In contrast to ncremental nnovatons, archtectural nnovatons place a premum on exploraton n desgn and the assmlaton of new knowledge. Hereby, the frm's knowledge base, defned as a set of knowledge and competences, remans a prelmnary condton n the assmlaton of spllovers from R&D efforts of the envronment. Cohen & Levnthal (1989) and Rosenberg (1990) nsst on potental synerges between the frm's own knowledge base and external flows of scentfc and techncal knowledge. Internal R&D actvty does not only stmulate nnovaton, but t also enhances the frm s ablty to recognze and assmlate outsde knowledge. By contrbutng R&D to the frms absorptve capacty, however, t should be noted that technologcal performance does not necessarly depend on past or referental performance, but rather on absorptve capacty generated n the past. In other words, frms wth hgh absorptve capacty wll explot new deas regardless of ther past performance. For Rothaermel (2001), organzatonal learnng motvates exploraton allances n order to buld new competences through the exploraton of new technologcal knowledge. Wth the am of dscoverng a new technology, exploraton mergers may allow the acquston of new capabltes by sharng tact knowledge such as basc R&D related to emergng technologes. B. Mergers for Development of Archtectural Innovaton Snce success n the telecommuncatons equpment ndustry turns on the synthess of component technologes n creatve new desgns, frms nnovaton competence reles on the actve development of both knowledge about alternatve components and knowledge of how these components can be ntegrated. Thereby, an ongong nternal R&D commtment allows the frms to not only scan the envronment for component and archtectural knowledge better, but t also enables the frms to evaluate complementary technology better (Veugelers, 1997). When a frm decdes to acqure a

10 technology, ts own R&D operatons allow t to ntegrate the technology better because external knowledge sources do not automatcally fnd ther way nto the frm s nnovaton process (Cassman and Veugelers, 2002). Once a frm has recognzed the nature of an archtectural nnovaton, t encounters the next constrants to swtch to a new mode of learnng and then nvest tme and resources n learnng about the new archtecture (Henderson and Clark, 1990). Frm routnes and organzatonal approaches that are successful n one technology paradgm do not necessarly translate nto success n subsequent and other paradgms. Organzatonal nerta constrants the abltes of establshed frms because the structures and systems facltate survval n stable and predctable envronments become labltes n envronments undergong rapd change (Hannan and Freeman, 1984). Snce establshed frms focus on nurturng organzatonally embedded knowledge, ther emphass on contnuous mprovement n such knowledge can nhbt learnng new archtectural knowledge, thus becomng subject to nertal forces. Such knowledge nerta enters nto force when a frm s partcularly blocked from adjustng to the market envronment by ther pror success n developng nnovaton competences and by the fact that t must buld new archtectural knowledge n a context n whch some of ts old archtectural knowledge may be relevant (Shu-hsen Lao, 2001). In order to transfer the broad, but shallow knowledge that s partcularly mportant for archtectural nnovaton and, n turn, leverage exstng competences, frms may engage n explotaton mergers that allow them to beneft drectly from the technologcal expertse of emergent frms whch are characterzed by the absence of nternal forces of nerta and long-standng commtments to establshed value networks (Hagedoorn and Schakenraad, 1994; Rothaermel, 2001). At the same tme, the explotaton mergers enable frms to buld new upstream value chan actvtes. Rchman and Macher s (2004) fndngs from ther case study analyss contrbute to the framework that the successful development of new technologcal paradgms often requres new and dfferent routnes (Nelson and Wnter, 1982). If a frm s unable to execute a technology strategy to accommodate to an emergng archtectural nnovaton on ts own, t may elect to acqure frms that have already begun to develop or commercalze products under a new technologcal paradgm. Hence, the focal frm gans access to both new technologes and the underlyng routnes and resources that support those technologes. In turn, the creaton and the mplementaton of new routnes to develop and adjust to new technologes requre the development of mechansms that allow frms to constantly and consstently adapt to frequent archtectural nnovatons. Frm s organzatonal slack resources can encourage transfer of technology (e.g., Nohra & Gulat, 1996) and provde the resources for dscretonary nvestments such as R&D (e.g., Bourgeos, 1981). Wth an ncreasng organzatonal slack and through ts mpact on absorptve capacty, managers accept more rsk and choose mergers that provde a hgher degree of nteracton between the nvolved frms and exposure to the tact knowledge n the nvolved

11 partes and great potental for knowledge transfer compared to other external R&D sources lke lcensng and jont venture (Steensma and Corley, 2001). In ndustres lke telecommuncatons equpment manufacturng, the responsveness of organzatonal slack to explot external knowledge flows faces consderable tme constrants due to hghly compettve condtons caused by short cycles of archtectural nnovaton and to the fact that, n new product development, the strateges of telecommuncatons equpment frms are not always market drven - for the pursut of nnovaton often nvolves technologes wth applcatons that meet the future rather than present customers demand. Gven the rapd pace of nnovaton, frms may not have the tme to develop nnovatons on ther own f they seek to reman compettve andmay need to acqure such nnovatons to keep up. C. Mergers for Applcaton of Archtectural Innovaton Once any domnant desgn s establshed, the ntal set of components s refned and elaborated, and the progress takes the shape of mprovements n the components wthn the framework of a stable archtecture. The correspondng gestaton perod of technologcal nnovaton can be, however, affected by government regulatons and ndustry standards. Long gestaton perods stran captal resources of new entrants before they have successfully marketed new products. The lqudty constrants of new entrants mght enhance a barganng power of establshed frms, better enablng them to gan access to technologcal nnovatons of new entrants through mergers on favorable terms. Moreover, although a technology can be archtectural n the sense that t reles upon a new knowledge base and on a new producton processes, t does not alter the way the new products are commercalzed. Thus, f the downstream value chan actvtes of establshed frms retan ther value, establshed frms may then be n a poston to beneft from the new technology by undertakng mergers wth emergent frms (Rothaermel, 2001; Teece, 1992). At the same tme, due to a lack of strong enterprse channels, the emergent frms are less lkely to have the resources to brng an nventon to the marketplace. Ths lack of manufacturng and dstrbutng actvty can be flled by large frms whch possess a greater ablty to fnance a large amount of R&D as well. 4. Emprcal Methodology 4.1 Assessng the Merger Propensty In the frst stage of the analyss, we explore the attractveness of telecommuncatons equpment frms as merger canddates by nvestgatng the determnants of mergers. Employng a random utlty model, we consder frm s decson of whether to acqure, to be acqured, to have nvolvement n a poolng merger or to stay outsde the merger market. The utltes assocated wth each of these choces k are modeled as a functon of the frm s characterstcs X whch affect the utltes dfferently:

12 U = X β + e (1) k k k Whle the level of utlty s not observable, we can, however, nfer from the frms choces how they rank each of these alternatves. If we assume that the ej are dstrbuted Webull, the dfferences n the dsturbances are dstrbuted logstc and a multnomal logt can be used to estmate the dfferences n the βk parameters. The propensty of engagng n a merger s based on a panel that conssts of nnovaton-related and fnancal varables on both merged and non-merged frms for whch data were avalable durng the 1988 to 2004 perod. The probablty that frm chooses alternatve k s specfed as: ' ( βk ) ' ( β ) ( β β ) l exp X 1 Pr( chooses k ) = = m m ' exp l l X exp l k X where β1,..., βm are m vectors of unknown regresson parameters. An mportant property of the multnomal logt model s that relatve probabltes are ndependent from each other, whch s the so-called ndependence of rrelevant alternatves (IIA) property. In order to obtan robust standard errors of estmated coeffcents, approprate tests were conducted, whch are dscussed n the secton 6.1. In the followng, we explan the nnovaton determnants of mergers captured by the emprcal analyss and assess the plausblty of the merger choce. As noted before, the man prescrpton of the resource-based approach asserts that only resources that are valuable, rare, nmtable, and nonsubsttutable can be sources of compettve advantage. Due to the fact that such resources are descrbed as ntangble, rather than tangble, the research stream tends to be dosyncratc n that prevous studes focus on a very lmted set of resource varables or a sngle frm. Whle such studes are benefcal, they are lmted n ther generalzablty. By quanttatvely studyng resource effects across a large sample of multple frms, the results provde generalzable fndngs for the resourcebased approach. Accordng to Mchalsn et al. (1997) and Levtas and Ch (2002), ths s an mportant need because t adds broader, more robust tests of the theory. In ths paper, we employ both tangble and ntangble resources, whch are quanttatvely measurable, as the determnants of mergers. The nnovaton performance of a frm s examned wth respect to the resources for ts R&D commtment, R&D competence, absorptve capacty, and knowledge nerta. R&D commtment of the frm s an mportant resource n drvng the frm s technologcal development. Hgh R&D commtment means that the frms have suffcent capabltes to mount an effectve market challenge to hgh technology from external sourcng. To capture R&D s role n promotng technologcal nnovaton, we take the R&D ntensty as a share of R&D expendture n total assets. (2)

13 The measure of nnovaton competence s n lne wth those of Henderson and Cockburn s (1994), who provde the concept of core competence and dvde t nto two crtcal elements of competence to be measured. Whle component competence s the locally embedded knowledge and sklls, archtectural competence s the ablty to ntegrate the component competences n a new and flexble way and to develop fresh component competences as they are requred. Due to the dffculty of ganng access to the ntra-organzatonal resource-level nformaton, we restrct the measure by the component competence. Snce propretary knowledge becomes a strategcally mportant capablty n hgh-tech frms, we use the ctaton-based patent ntensty as a measure of component competence. In order to account not only for the quantty but also the qualty of the patented nventons, we measure the patent-based characterstcs of a frm usng the number of forward ctatons of patents. The number of ctatons receved by any gven patent s truncated n tme because we only know about the ctatons receved thus far. In other words, the number of forward ctatons a patent receved depends on the year of the applcaton. We, therefore, normalze the ctaton counts by ther average value calculated over all patents belongng to the same technologcal sub-class whose applcaton was fled n the same year. 19 We then weght each patent of a frm by the number of normalzed ctatons that t subsequently receved (Trajtenberg, 1990). Knowledge nerta stemmng from the use of routne problem solvng procedures, stagnant knowledge sources, and an overemphass on refnng and mprovng exstng knowledge, preventng the frm from explorng alternatve knowledge sources (Shu-hsen Lao, 2001), s measured by ts outcome, namely by R&D productvty. R&D productvty accounts for the extent to whch R&D brngs forth new knowledge and s defned as a rato of ctaton-weghted patent to R&D expendture. In ther revew of the absorptve capacty approach, Zahra and George (2002) argue that there exst substantal dfferences among the dmensons of absorptve capacty the ablty to dentfy, assmlate and explot knowledge from the envronment (Cohen and Levnthal, 1989), whch allow them to coexst and be measured and valdated ndependently. Accordng to Zahra and George (2002), the potental absorptve capacty enables the frms to be receptve to dentfcaton and assmlaton external knowledge, whle realzed absorptve capacty reflects the frm s explotaton capablty to leverage the knowledge that has been absorbed. We employ the stock of accumulated patents n order to capture the path-dependent and cumulatve nature of absorptve capacty. 20 Thereby, we are able to capture the frst two components of absorptve capacty drectly, namely the ablty to dentfy and to assmlate the new knowledge from mergers. The stock of accumulated knowledge of a frm s measured usng ctaton-based patents and calculated by applyng the perpetual nventory method by assumng a deprecaton rate of 15 percent per annum (Hall, 1990). Hence, the ndvdual patents n the 19 Ths s the fxed-effects approach proposed n Hall et al. (2001) 20 Many studed used R&D ntensty as a measurement of absorptve capacty. However, recent studes provde evdence that the two faces of R&D (Cohen and Levnthal, 1989) domnate each other n the short-run,.e. R&D expendtures are predomnantly a means of developng new knowledge and nnovaton rather than a means of buldng absorptve capacty (Schmdt, 2005).

14 frm s knowledge base provde the bass for comparng the frm s own knowledge base wth that of other frms. In order to control for the sgnfcant role of technology development as a ratonale for engagng n mergers, we nclude dummy varables whch ndcate zero (or very low) R&D ntensty and zero (ctaton-weghted) patent ntensty. 21 Our measure of organzatonal slack focuses on a frm s avalable and potental slack resources, whch represent resources avalable and not yet commtted for partcular allocatons and the future ablty to generate resources, respectvely. Geger & Cashen (2002) hghlght the role of organzatonal slack as an mportant condton that facltates nnovaton adopton and, thus, contrbutes to a frm s nnovatveness. 22 In keepng wth prevous studes (e.g., Sngh, 1986; Bromley, 1991), we rely on accountng data and measure the avalable slack usng the rato of cash flow to total assets and the potental slack usng Tobn s q. In fnancal terms, the cash flow rato represents the fnancal capabltes of frms and amounts for funds avalable to a frm for operatons, nvestments, and acqustons. We approxmate Tobn s q by calculatng the rato of the market value to the book value of a frm s assets, where the former s the sum of the book value of long-term debt and the market value of common equty. Beng a forward lookng ndcator, a hgh value of q suggests favorable growth opportuntes (e.g., Gugler et al., 2004). Besdes the large frms, the relatvely small frms n the telecommuncatons equpment ndustry have often become publc entty. 23 Gven the evdence that frms that nnovate receve a market premum over smlar frms that do not nnovate (e.g., Chaney, Devnney, and Wner, 1991), frms wth growng market value may appear to be lkely acquston targets or merger partners for mature frms lookng to absorb growth opportuntes. Frm sze and frm growth are measured by the book value of total assets and the annual growth rate of the market value, respectvely. All monetary values are deflated by a prce ndex for communcatons equpment manufacturng based on the 1999 prce Assessng the Merger Impact In the second stage of the analyss, we nvestgate the effect of mergers on the nnovaton performance of the frms. Gven the hgh degree of archtectural nnovaton affectng the telecommuncatons 21 For the frms wth mssng R&D ntensty over tme, we set the R&D ntensty equal to zero. Indeed, the manual nspecton of the observatons wth mssng values ndcates that the mssng R&D values belong entrely to that part of the relatve small frms that provde low or no R&D actvty. 22 Geger & Cashen (2002) nvestgate the dfferent dmensons of slack and fnd avalable slack to have an nverted U-shaped relatonshp wth nnovaton adopton, whle potental slack has a lnear postve relatonshp to nnovaton adopton. 23 Ths trend s reasoned by the fact that becomng an ndependent publc entty provdes smaller frms wth stock optons whch are essental for retanng hgh labor sklls and that market valuatons for ndependent component frms are much hgher than for those that closely depend on the parent frms (Merrll Lynch, 2000). 24 Moreover, all covarates n the regressons have been lagged by one year n order to avod potental endogenety problems as well as possble bases arsng from dfferent merger accountng methods and fnancal statement consoldaton.

15 equpment ndustry, we expect that successful equpment frms take advantage of technologcal mergers to enhance ther nnovaton poston and performance. Mergers may extend the technologcal base of frms nvolved, allowng them to acheve greater economes of scale and scope n R&D through more effcent deployment of knowledge resources. That s, mergers may enlarge the overall R&D budget of frms engaged, whch then enables them to tackle larger R&D projects and, thereby, ths spreads the rsk of nnovaton. Furthermore, snce n fast movng markets wth abbrevated product lfe cycles, the uncertanty and the rsk of nvestng n unproven technologes are often shouldered by outsde nvestors, the merger nsders may nvolve greater ncentves to nvest n R&D after the acqurng of an auspcous technology. Greater R&D commtment, however, does not guarantee per se that the frms wll have hgher levels of nnovaton competence wth respect to patent ntensty, especally f mergers undertaken have low knowledge and technology complementarty wth the frms nvolved. Therefore, the ntegraton of complementary knowledge and technology may ncrease nnovaton output through mergers leadng to more advanced technologes beng developed (e.g., Gerpott, 1995). If technology s playng an mportant role n the external knowledge acquston strategy, then frms undertakng mergers, whle at the same tme acqurng patents, wll renew and enhance ther path-dependent absorptve capacty n terms of accumulated patent portfolo. Gven the dffcultes faced by the development of archtectural nnovaton, we expect that technologcal mergers are used to overcome the knowledge nerta and the lack of compettve nnovaton competence as well. At the same tme, however, the organzatonal obstacles faced n mergers wth regard to the larger R&D departments as well as the mscarred and napproprate ntegraton of the technology-based frms (Duysters and Hagedoorn, 2000) may dmnsh the margnal returns from R&D nvestment n ther extent and speed. In addton, the overall performance of the merged frms wll be analyzed by means of frm sze, market value, cash-flow rato, and Tobn s q. Beng the measures of growth opportuntes and fnancal capablty of a frm, Tobn s q and cash flow gve an ndcaton to the fnancng type of a merger. The frst stage estmates may enable us to ndcate whether a merger s fnanced by stock and/or debt or nternally generated resources lke cash flow. In partcular, f frms wth hgher Tobn s q are more lkely to engage n mergers, then we predct that an average merger s fnanced through stock nstead of cash, whch may lead to a decreasng post-merger Tobn s q. The fndngs from studes on the frms post-merger performance show that, n the long-run, stock acqurers perform poorly compared to cash acqurers and matchng frms that do not merge (e.g., Loughran and Vjh, 1997) 25. Most studes nvestgate the mpact of a merger on the market value of a frm by lmtng the scope of a study to the around days of the merger announcement and, thus, gnorng the contemporaneous mpact of nnovaton as well. Gven the evdence that a contnued emphass on R&D 25 Loughran and Vjh (1997) analyze the excess return for the fve-year perod after the acquston.

16 and an ntroducton of new products (postvely) affect the market value of frms n the technologcally based ndustres (e.g., Pakes, 1985; Chaney, Devnney, and Wner, 1991) and n order to capture both coexstng effects of technologcal mergers and ther nnovaton value on the frm s market value, we consder the long-run effect smlar to the other performance determnants as descrbed n Secton 6.2. Our analyss of the effects of mergers controls for endogenety and ex-ante observable frm characterstcs usng a propensty score method (Deheja and Wahba, 2002). Controllng for the dfferences n the merged and non-merged frms performances pror to the merger, we estmate the frms post-merger nnovaton performance compared to what they would have n the absence of the merger. For each frm n the sample, let M be a merger ndcator that equals one when the frm engages n a merger and zero otherwse. We denote Y 1 as the nnovaton performance of mergng and Y 0 as the nnovaton performance of non-mergng frms and observe ( 1 ) M and, hence, Y = M Y 1 + M Y 0. Accordngly, let E Y 1 M = 1 and E Y 0 M = 0 denote average outcomes of the technologcal performances of merged and non-merged frms, respectvely. The effect we are nterested n s that of the merger on the technologcal performance of the merged frms, or the dfference between the expected nnovatve performances of the merged frms and the frms that would have experenced f they dd not merge: τ M = = = = 1 E Y1 M 1 E Y0 M 1 (3) Ths denotes the expected treatment effect on the treated. Snce we do not have the counterfactual evdence of what would have happened f a frm had not engaged n a merger, E Y 0 M = 1 s unobservable. However, t can be estmated by E Y 0 M = 0 and the effect can be then gven by the dfference n the average outcome between the merged and non-merged nnovatve performances: e τ = E Y1 M = 1 E Y0 M = 0 (4) In fact, we have observatons of the frms that dd not engage n a merger, but f the merged and the non-merged frms systematcally dffer n ther frm characterstcs, (4) wll be a based estmator of (3) (Hrano et al., 2002). Rubn (1997) and Rosenbaum and Rubn (1983, 1984) showed that a propensty score analyss of observatonal data can be used to create groups of treated and control unts that have smlar characterstcs, whereby comparsons can be made wthn these matched groups. In these groups, there are frms that have been merged and frms that have not been merged; hence, the allocaton of the merger can be consdered to be random nsde the groups of frms.

17 The merger propensty score s defned as the condtonal probablty of engagng n a merger gven a set of observed covarates ( ) ( ) X : p M = Pr M = 1 X = E M X (5) The treatment effect of a merger s then estmated as the expectaton of the condtonal effects over the dstrbuton of the propensty score n the merged sample: τ { ( ) ( ) } = = E ( ) E, =, = Y p M M 1 E Y p M M 0 M = 1 M 1 p M 1 0 The propensty score matchng reles on two key assumptons (Rosenbaum and Rubn, 1983, 1984). The frst, condtonal ndependence assumpton (CIA) requres that condtonal on the propensty score potental outcomes are ndependent of treatment assgnment. The CIA assumes that selecton nto treatment occurs only on observable characterstcs. Hence, unbased treatment effect estmates are obtaned when we have controlled for all relevant covarates. The second assumpton s the common support or overlap condton, meanng that frms must have a postve probablty of beng ether merger or non-merger rather than just havng same covarate values. In sum, the propensty score matchng reles on the strong gnorablty assumpton, whch mples that for common values of covarates, the choce of treatment s not based on the benefts of alternatve treatments. (6) 5 Data In order to examne the nteracton between merger and nnovaton actvty, a new frm-level data set s constructed whch covers the overwhelmng majorty of frms n the telecommuncatons equpment ndustry that operated n any year over the 18-yearperod, 1987 to 2004 (ncludng lagged perods). Ths data set s created by a complex matchng process of nformaton from ntally four separate databases. We defne the telecommuncatons equpment frms as those whch have prmary actvty n the communcatons equpment Standard Internatonal Codes (SIC) 3661, 3663, or The populaton of frms and ther fnancal nformaton ncludng R&D expendtures were drawn from Compustat and Global Vantage databases. After elmnatng frms wth mssng fnancal nformaton, we can dentfy a sample of 638 telecommuncatons equpment frms for those data on R&D expendtures, total assets, market value, cash flow, and long-term debt. The patent statstcs for the telecommuncatons equpment ndustry are based on the database whch s compled by the Natonal Bureau of Economc Research (NBER, Hall et al., 2001). Ths database comprses detaled nformaton on all US patents granted between 1963 and 2002 and all