1 Discussion Pper No Buyer Power nd Suppliers Incentives to Innovte Christin Köhler nd Christin Rmmer
2 Discussion Pper No Buyer Power nd Suppliers Incentives to Innovte Christin Köhler nd Christin Rmmer Downlod this ZEW Discussion Pper from our ftp server: Die Discussion Ppers dienen einer möglichst schnellen Verbreitung von neueren Forschungsrbeiten des ZEW. Die Beiträge liegen in lleiniger Verntwortung der Autoren und stellen nicht notwendigerweise die Meinung des ZEW dr. Discussion Ppers re intended to mke results of ZEW reserch promptly vilble to other economists in order to encourge discussion nd suggestions for revisions. The uthors re solely responsible for the contents which do not necessrily represent the opinion of the ZEW.
3 Ds Wichtigste in Kürze Zulieferer sehen sich immer wieder mit Kunden konfrontiert, die ufgrund von Konzentrtionsprozessen in ihrer Brnche über Käufermcht verfügen. Die Ausübung von Käufermcht wird oft ls negtiv für die Innovtionstätigkeit von Zulieferern betrchtet, d ngenommen wird, dss sie zu niedrigeren Gewinnen bei den Zulieferern und mithin uch zu niedrigeren Investitionsnreizen führt. Allerdings können Käufermcht und die dmit einhergehenden Wettbewerbsverhältnisse im Käufermrkt die Innovtionsnreize des Zulieferers durchus erhöhen. Wir betrchten sowohl die Preis- ls uch die Technologiedimension des Käufermrktwettbewerbs. Firmen, die in intensivem Preiswettbewerb stehen, können ihre Verhndlungsmcht gegenüber Zulieferern dzu nutzen, Kostensenkungen (durch Prozessinnovtionen) oder Qulitätsverbesserungen (durch Produktinnovtionen) zu fordern, um sich selbst im eigenen Mrkt von ihren Wettbewerbern bsetzen zu können. Zulieferer, deren Kunden in strkem Technologiewettbewerb stehen, müssen notwendigerweise innovtiv sein und Wissenstrnsfer vom Kunden sicherstellen. In beiden Fällen sind Käufer und Zulieferer uf Koopertion und Kollbortion ngewiesen, die erhebliche Investitionen verurschen. Dies wiederum stärkt die Verhndlungsposition des Zulieferers und erhöht dmit uch seine Innovtionsnreize. Bislng hben sich llerdings erst wenige empirische Untersuchen dem Zusmmenhng zwischen Käufermcht und Innovtionsnreizen des Zulieferers gewidmet. Die vorliegenden Arbeiten finden regelmäßig einen negtiven Effekt, verwenden jedoch keine objektiven Mße bzw. lediglich Mße uf Brnchenebene, um Käufermcht bzubilden. Drüber hinus sind sie uf Brnchen beschränkt, die ls besonders strk von Käufermcht betroffen gelten, ohne dbei die Wettbewerbsverhältnisse im Käufermrkt zu berücksichtigen. Wir überprüfen empirisch den Effekt von Käufermcht uf die Innovtionsnreize der Zulieferer sowohl in Bezug uf die Entscheidung, in Innovtionen zu investieren, ls uch uf die Höhe dieser Investitionen. Dzu verwenden wir Unternehmensdten des Mnnheimer Innovtionspnels (MIP). Unser Dtenstz umfsst Unternehmen us Deutschlnd us dem verrbeitenden Gewerbe und dem Dienstleistungssektor. Er enthält Angben zum Anteil der drei größten Kunden m Umstz des Zulieferers sowie zu den Möglichkeiten der Kunden zu konkurrierenden Zulieferern zu wechseln, sodss ds Ausmß von Käufermcht objektiv uf Unternehmensebene bgebildet werden knn. Wir zeigen, dss Käufermcht einen negtiven Effekt uf die Innovtionsentscheidung eines Zulieferers ht. Dieser negtive Effekt wird jedoch bgemildert, wenn sich ein Zulieferer mit Kunden konfrontiert sieht, die ihrerseits in strkem Preiswettbewerb stehen und über Käufermcht gegenüber dem Zulieferer verfügen. Wir finden keine Hinweise druf, dss Käufermcht einen direkten Einfluss uf die Entscheidung eines Zulieferers ht, wie intensiv in Innovtionsktivitäten investiert wird. Stttdessen finden wir, dss der Effekt von Käufermcht uf die Innovtionsintensität des Zulieferers von Grd des technologischen Wettbewerbs im Käufermrkt bhängt. Je stärker dieser Wettbewerb ist, desto niedriger sind die FuE- Investments eines Zulieferers, der sich einem Kunden mit Käufermcht gegenüber sieht.
4 Non-technicl summry With mny industries experiencing significnt concentrtion processes during the lst yers, suppliers re incresingly confronted with powerful buyers. The common belief is tht exertion of buyer power negtively ffects the innovtion decisions of suppliers. The rtionle behind this view is tht buyer power leds to decresing profits of suppliers, which t the sme time lowers their investment incentives. This explntion my be too nrrow s competition in the buyer mrket my spur suppliers' innovtion incentives. We consider both the price nd the technology dimension of buyer mrket competition. A powerful buyer confronted with strong price competition might hve the incentive to demnd lower prices or higher qulity in order to gin cost dvntge or to differentite wy from competitors. Moreover, for suppliers exposed to powerful technologiclly competing buyers it my be precondition to be innovtive nd to utilize knowledge spillovers from the buyer side. Then supplier nd buyer need coopertion nd collbortion which requires considerble investments into their reltionship. In turn this leds to stronger brgining position for the supplier nd thus increses innovtion incentives. A few empiricl studies re dedicted to the nlysis of buyer power nd suppliers' incentives to innovte nd they frequently find negtive reltionship. However these studies lck n objective mesure for buyer power or merely use industry mesures. Furthermore the focus is by now on prticulr industries which re perceived to be hevily ffected by concentrtion processes mong buyers. Besides, ll these studies tend to neglect the dimensions of competition in the buyer mrket. We nlyse the reltionship between buyer power nd suppliers' innovtion incentives empiriclly in different stges of the innovtion process. Tht includes the innovtion decision nd the decision on the intensity of innovtion ctivity. We pply firm level dt provided by the Mnnheim Innovtion Pnel (MIP). Our dtset comprises 1,129 observtions from Germn firms cross mnufcturing nd service sectors nd llows us to pply objective mesures for buyer power tking ccount of supplier s economic dependency from the lrgest three customers nd the buyers opportunities to switch to competing suppliers. We find negtive effect of buyer power on supplier s likelihood to invest in R&D. This negtive effect is mitigted by the intensity of price competition in the downstrem mrket. In contrst, we find no evidence of buyer power to ffect supplier s decision how much to invest in R&D directly. Insted, there is wek evidence tht the effect of buyer power depends on the intensity of technology competition in the downstrem mrket. The stronger the technology competition downstrem, the lower R&D investments of supplier confronted with powerful buyer.
5 Buyer Power nd Suppliers' Incentives to Innovte Christin Köhler Christin Rmmer September 2012 Abstrct Buyer power is widely considered to decrese innovtion incentives of suppliers. However, there is little empiricl evidence for this sttement. Our pper nlyses how buyer power influences innovtion incentives of upstrem firms while tking into ccount the type of competition in the downstrem mrket, nmely price nd technology. We explore this reltionship empiriclly for unique dtset contining 1,129 observtions of Germn firms from mnufcturing nd service sectors including informtion on the economic dependency of firms from their buyers. Using generlised Tobit model, we find negtive effect of buyer power on supplier s likelihood to strt R&D ctivities. This negtive effect is mitigted if the supplier fces powerful buyers operting under strong price competition. There is lso wek evidence for negtive effect of buyer power on suppliers R&D intensity if the powerful buyer opertes under strong technology competition. JEL-Clssifiction: L11, O31 Keywords: Innovtion, Buyer Power Adress: Christin Köhler Centre for Europen Economic Reserch (ZEW) Industril Economics nd Interntionl Mngement P.O. Box Mnnheim Germny Phone: E-mil: Christin Rmmer Centre for Europen Economic Reserch (ZEW) Industril Economics nd Interntionl Mngement P.O. Box Mnnheim Germny
6 1. Introduction The impct of mrket structures on innovtion ctivities hs received much ttention in innovtion reserch. The vst mjority of the literture concentrtes on the effects of horizontl competition on innovtion incentives nd neglects the incentives resulting from verticl interctions in mrkets. With mny industries experiencing concentrtion processes suppliers re often confronted with powerful buyers. Yet, reltively little is known bout how powerful buyers my ffect innovtion incentives of upstrem firms. A common belief is tht exertion of buyer power negtively ffects innovtion decision of suppliers becuse buyer power will led to decresing profits of suppliers, which t the sme time lowers their investment incentives. This is expected to reduce the vriety in suppliers' rnge of products (OECD, 1998; Europen Commission, 1999; Inderst nd Shffer, 2007). However, recent theoreticl findings suggest tht buyer power might hve positive impcts on the suppliers' innovtion incentives (Inderst nd Wey, 2007; Inderst nd Wey, 2011). So fr, the role of suppliers' innovtion incentives in the presence of powerful buyers hs been discussed lrgely from theoreticl perspective. 1 From n empiricl point of view, only few studies exist tht nlyse suppliers' incentives to innovte when fcing powerful buyer. These studies often lck n objective mesure for buyer power on firm level but rther use either ggregted industry mesures or firms' subjective ssessment whether they re confronted with powerful buyers. Furthermore, nlyses of the reltionship between buyer power nd suppliers' incentives to innovte re mostly focused on prticulr industries which re perceived to be hevily ffected by concentrtion processes mong buyers. Besides, ll these studies tend to neglect the dimensions of competition in buyer mrkets. We rgue tht it does mke difference for the upstrem firm whether it is supplying to buyer engged in intensive competition or to buyer fcing no or only low level competition. We consider both the price nd the technology dimension of competition in the downstrem mrket. A powerful buyer confronted with strong price competition might hve the incentive to lign innovtion ctivities long the vlue chin in order to gin cost dvntge or to differentite wy from competitors. Moreover, it my be precondition to be innovtive nd to utilise knowledge spillovers from the buyer side for firms supplying to powerful buyers which re exposed to intensive technologicl competition. Wht is more, fierce technologicl competition in the buyer mrket my generte new technologicl opportunities for suppliers to invest in own R&D ctivities. As consequence competition in the buyer mrket my led to incresed innovtion incentives on the supply side. We test these hypotheses empiriclly with dtset tht contins 1,129 observtions from Germn firms cross mnufcturing nd service sectors bsed on the Germn Innovtion Survey. In contrst to existing studies the dtset enbles us to pply n objective mesure for buyer power which tkes ccount of supplier s economic dependency from the lrgest three customers nd the buyers opportunities to switch to competing suppliers. We decompose the 1 For comprehensive review of the theoreticl development on buyer power see e. g. Inderst nd Mzzrotto (2008). 1
7 effects of buyer power on the suppliers' innovtion incentives into the effect on the decision to strt innovtion ctivities nd the effect on the mount of resources spent on innovtion. Our results show tht buyer power shpes supplier s innovtion incentives. In ddition, the type of competition in downstrem mrkets mtters but it ffects supplier s innovtion decision t different stges differently. We find negtive effect of buyer power on supplier s likelihood to invest in R&D. This negtive effect is mitigted by the intensity of price competition in the downstrem mrket. Our finding implies tht supplier confronted with buyer power hs higher likelihood to strt R&D ctivities, if price competition in the buyer mrket is strong. In contrst, we find no evidence of buyer power to ffect supplier s decision how much to invest in R&D directly. Insted, there is wek evidence tht the effect of buyer power depends on the intensity of technology competition in the downstrem mrket. The stronger the technology competition downstrem, the lower R&D investments of supplier confronted with powerful buyer. Our interprettion of this result is tht for firms supplying to highly competitive industries in terms of technology, it is precondition to hve R&D ctivities but it seems tht t the sme time suppliers re not ble to improve their brgining position with further investments into supplier-buyer reltionship. Hence, powerful buyers re ble to extrct lrger shre of joint profits which reduces suppliers innovtion incentives compred to independent suppliers. The reminder of this pper is orgnised s follows. Section 2 reviews the existing theoreticl nd empiricl literture on the effects of buyer power on suppliers' innovtion incentives nd explores the possible effects of the type of competition previling in the buyer mrket. Section 3 presents the dt, the vrible specifiction nd our estimtion strtegy. Descriptive sttistics s well s estimtion results re presented in section 0. Section 5 provides discussion of the results while section 6 concludes nd offers further directions of reserch. 2. Theoreticl frmework 2.1. Buyer power nd innovtion incentives Literture provides different pproches to the emergence nd impcts of buyer power. In contrct theory it is ssumed tht supplier nd buyer negotite bilterlly over prices nd quntities of the respective good or service to be trded. Given tht contrcting between the supplier nd the buyer leds to joint profit, the split of the profit then depends on the brgining position of ech contrcting prty. The strength of the brgining position nd hence brgining power is determined by the profits to be relised when the contrct is mde with n lterntive supplier or buyer. The higher such outside-option pyoffs in reltion to the counterprts outside-option pyoffs the stronger the brgining position of the respective contrctor. According to this pproch, buyer power results from the fct tht more vluble outside options re t the disposl of the buyer thereby llowing the buyer to extrct lrger shre of joint profits (Inderst nd Vlletti, 2007; Dobson nd Inderst, 2008). In this pper we dopt the view of buyer power being consequence of brgining power exerted by the downstrem 2
8 firm (buyer) on the upstrem firm (supplier) (Dobson nd Inderst, 2008; Inderst nd Mzzrotto, 2008). 2 When deciding on investment in innovtion efforts supplier will consider the discounted vlue of future rents collectble from this ctivity nd whether these rents re pproprible. Given tht buyer power results from stronger brgining position of the buyer reltive to the supplier, the effect of buyer power on suppliers' innovtion incentives seems to be cler-cut: When fcing powerful buyers, the supplier hs less incentive to innovte, s the ppropribility of innovtion rents is too low. Recent theoreticl studies show, however, tht buyer power my provide dditionl innovtion incentives for suppliers. Suppliers fcing lrge buyers hve n incentive to invest in both product nd process innovtions, given tht size is the sole source of buyer power (Inderst nd Wey, 2007). While process innovtion llows lower unit costs t high volumes compred to supplier fcing mny smller buyers, product innovtion renders higher revenues compred to the old product. Either wy, supplier innovtion leds to devlution of the buyer's outside options nd in turn strengthens the brgining position of the supplier llowing for lrger shre of joint profits. Given the life-cycle hypothesis of Utterbck nd Abernthy (1975), positive effect of buyer power on suppliers' innovtion incentives might lso occur since suppliers with few buyers my suffer less from uncertinty over innovtion demnd of buyers nd therefore hve declined risk of innovtion filure (Klepper, 1996). In ddition, lrger size of orders might induce higher incentives for suppliers to engge in R&D s there is more certinty in the sles of new products (Peters, 2000). In contrst, merger in buyer mrkets my reduce incentives for product differentition by suppliers. Product differentition is often linked to innovtion since entering new product mrkets typiclly constitutes n innovtion ctivity. In cse of buyer merger, the consolidted buyer my be better off using single sourcing strtegy, i.e. to stock only goods of one supplier. If the likelihood of buyer merger is incresing, this strtegy will led to lower degree of product differentition of suppliers (Inderst nd Shffer, 2007). Lrge buyers my hve n incentive to force their suppliers into contrcts which constitute n exclusive reltionship between supplier nd buyer. Such supply contrcts will reduce upstrem innovtion becuse suppliers will ber disdvntges of low-scle production nd hve less incentive to innovte (Stefndis, 1997). Wht is more, lrger buyers cn more credible thret to integrte bckwrds (Ktz, 1987; Inderst nd Wey, 2007) nd my intensify competition on supplier mrkets. By breking up collusion mong suppliers they lower suppliers' profits (Scherer nd Ross, 1990). This effect is incresing in the size of the buyer (Snyder, 1996; Snyder, 1998). Also, they re in position to llevite mrket entries on the supply side, e.g. by overtking fix costs of otherwise unprofitble entrnts or pre-committing some of their purchses (Dobson nd Inderst, 2008). The concern bout negtive effects of buyer power on innovtion incentives of suppliers led the UK Competition Commission (CC) to conduct mrket investigtion focusing on dverse 2 One could lso study buyer power in the frmework of monopsonistic behviour (see e. g. Ms-Colell et l., 1995). The min rgument of this pproch is tht monopsonistic firms strtegiclly reduce demnd in order to mximise profits. However, this my not pply for most supplier-buyer reltionships. 3
9 effects on competition in the supply for groceries in the UK due to the behviour of retilers. One prt of the investigtion exmined whether buyer power of retilers my "impose excessive risks nd unexpected costs on suppliers, which reduces suppliers' incentive or bility to invest nd innovte. This could led to reduced cpcity, reduced product qulity nd fewer new product offerings" (Competition Commission, 2008, p. 157). Although the CC did not find evidence tht UK grocery suppliers exhibit less innovtion efforts, they expect the innovtion performnce to be decresing in future if consequences of buyer power, e.g. retrospective price djustments or excessive trnsfer of risks, continue t the observed level (Competition Commission, 2008, p. 173). Empiricl studies frequently find negtive reltionship between buyer power nd innovtion ctivities of suppliers. These existing studies follow quite different pproches to cpture buyer power nd innovtion incentives of suppliers. Frber (1981) nlysed the effect of mrket structure in the buyer mrket on R&D efforts in supplier industries using cross-sectionl industry level dt of 50 4-digit mnufcturing SIC-industries from the US. Mrket structure in both supplier nd buyer mrkets is mesured by concentrtion rtios, reflecting the shre of industries sles generted by its four lrgest enterprises. Employing simultneous eqution model explining the shre of scientist nd engineers in the workforce, the dvertising intensity nd the seller concentrtion rte, he finds evidence tht concentrtion in the buyer mrket ffects R&D incentives of suppliers. However, the sign of this effect depends on the concentrtion in the supplier mrket. If the supplying industry is wekly concentrted, n increse in concentrtion of the buyer industry will hve negtive effect on the shre of scientists nd engineers in the workforce. Conversely, this effect is positive if the mrket concentrtion in the supplier industry is sufficiently high. This findings re in line with the results of Peters (2000) who investigtes the effect of mrket structure in the buyer mrket both on suppliers' innovtion inputs nd innovtion outputs using firm-level dt consisting of 401 Germn utomotive suppliers. Innovtion inputs re mesured by R&D expenditure divided by sles s well s by totl innovtion expenditure divided by sles. 3 Innovtion output is cptured by the introduction of product or process innovtions within two yer spn. Mrket structure in the buyer industry is represented by industry's concentrtion rtio (CR3) nd by n dditionl dummy vrible indicting whether the supplier hs 10 or more customers. Regrding innovtion intensity, the result indictes tht firms supplying highly concentrted buyer industries exhibit lower levels of innovtion intensity. The negtive correltion is mitigted, however, if suppliers re operting in concentrted industry. With respect to R&D intensity, mrket structure in the buyer industry is found to moderte the effect of mrket structure in the supplier industry. Suppliers operting in concentrted industry nd supplying highly concentrted buyer industries exhibit significntly higher R&D intensity. Conversely, suppliers operting in concentrted industry nd supplying buyer industries with low degree of concentrtion show significntly lower R&D intensities. Interestingly, there is no evidence tht the mrket structure of buyer industries hs significnt impct on the supplier's probbility to introduce new products. Also, the suppli- 3 Innovtion expenditure includes expenses not only for R&D but lso for other ctivities iming t the introduction of new products or processes, such s design, mrketing, trining nd purchse of mchinery, equipment, softwre nd intellectul property. 4
10 er's probbility to introduce process innovtions is not ffected by the concentrtion in the buyers industry but by the number of customers. Weiss nd Wittkopp (2003; 2003b) use survey dt from Germn food mnufcturers. Innovtive ctivity is mesured by the overll number of new products introduced within three yer time spn (Weiss nd Wittkopp, 2003b) nd by the number of new products with either regulr or superior qulity introduced within three yer time spn (Weiss nd Wittkopp, 2003). Mrket power of the retilers is cptured by firms' ssessment whether retilers re ble to exert pricing pressure on them on scle rnging from 1 (very low) to 5 (very high). Using smll smple of 88 nd 87 firms, respectively, they observe tht suppliers experiencing very high pricing pressure of retilers introduce significntly less new products. With respect to qulity differences mong the newly introduced products they yield mixed results. While they observe negtive reltionship between retilers' pricing pressure nd the number of new products with regulr qulity, retilers' pricing pressure does not hve significnt effect on the number of new products with premium qulity The role of downstrem competition So fr, reserchers rrely looked t the type of competition in the buyer mrket when nlysing the reltionship between buyer power nd suppliers innovtion incentives. We rgue tht it is importnt to tke the competitive environment of the buyer into ccount becuse it is likely to be trnsferred by powerful buyer to the upstrem mrket. We will distinguish between price nd technology dimension of competition. Let us first consider the price dimension of competition. If price competition in the downstrem mrket is strong, mrgins re low. Then powerful buyer my squeeze suppliers' profit mrgins to gin competitive dvntges over competitors. Supplying firms need to supply higher qulity ( higher quntity) to the sme price or conversely, the sme qulity (the sme quntity) to lower price. In result the mgnitude of expected innovtion rents my be too smll to induce R&D investments on the suppliers' side, especilly ginst the bckground of high filure risk nd the finncil burden ttched to n innovtion project. In contrst to this trditionl rgument which is brought forwrd frequently (OECD, 1998; Competition Commission, 2008), suppliers cn strengthen their brgining position reltive to the buyer by relising lower unit costs t high volumes (Inderst nd Wey, 2007). This increses process innovtion incentives for suppliers s outside options of the buyer re devluted nd suppliers cn pproprite higher shre of innovtion returns. If price competition in the downstrem mrket is strong, buyers my im to increse their mrgins by differentiting wy from ech other. This would lso decrese price competition. In this cse buyers my use their power to stimulte suppliers' product innovtion ctivities in order to lign innovtion ctivities long the vlue chin. However, investments in R&D on the supplier side my be necessry to mke use of the buyer's innovtion impulses since knowledge cquired from the buyer hs to be integrted in the supplier's knowledge stock before it cn be commercilly exploited. The effect of suppliers' in-house R&D my hence be twofold: to generte new knowledge nd to crete bsorptive cpcity which llows evlution nd exploittion of externlly vilble knowledge (Cohen nd Levinthl, 1989). This knowledge exchnge will likely result in the building of co-specilised ssets (Teece, 1986). 5
11 Such sitution obviously increses innovtion incentives on the supply side s the supplier's brgining position reltive to the buyer improves nd llows more fvourble split of the joint profit. Suppliers' innovtion incentives my be even further spurred if specific demnds of powerful buyers re nticiptory for lrger mrket segments in the future (von Hippel, 1988). Hence, suppliers exposed to buyer power nd strong price competition in the downstrem mrket hve higher innovtion incentives thn suppliers with powerful buyers tht re less exposed to price competition. We derive our first hypothesis ccordingly: Hypothesis 1: Buyer power will hve more positive effect on suppliers' innovtion incentives the stronger the price competition in the buyer mrket. Another dimension of competition in the buyer mrket is technology intensity which is likely to be trnsferred to the supplier mrket s well. Intense technology competition urges buyers to invest hevily into the development of new products nd new process technology, which reduces their profits. In the context of ptent rces, some prt of this investment my not be turned into commercil success but is sunk, hurting profits further (Fudenberg et l., 1983). Buyers with brgining power vis-à-vis their suppliers my use this power to shift significnt shre of these costs nd risks to their suppliers. In order to mintin commercil reltions with buyers, suppliers will hve to invest into own R&D efforts long the innovtion ctivities of their buyers. Consequently, suppliers' innovtion incentives re higher thn in sitution with low technology intensity in the downstrem mrket. In fct, mny studies found buyers re min source for technologicl dvnce in upstrem firms (Klevorick et l., 1995). Technology competition in the buyer mrket cn lso be seen s technologicl opportunity for suppliers. The higher R&D investments crried out in the buyer mrket, the more demnd there is lso for innovtive upstrem products (Scherer, 1982). This leds us to our second hypothesis: Hypothesis 2: Buyer power will hve more positive effect on suppliers' innovtion incentives the stronger the technologicl competition in the buyer mrket. 3. Empiricl study 3.1. Dt The empiricl prt of our study employs firm level informtion from the Mnnheim Innovtion Pnel (MIP) which consists of representtive strtified rndom smple of Germn firms. Dt collection is crried out by the Centre for Europen Economic Reserch (ZEW) on behlf of the Federl Ministry of Eduction nd Reserch. The MIP hs provided nnul informtion on innovtive behviour in the Germn mnufcturing sector since 1992 nd in the service sector since 1994 nd is t the sme time the Germn contribution to the Europen Community Innovtion Survey (CIS). Definitions of innovtion nd innovtive ctivities re 6
12 tken from the OECD s Oslo Mnul. The trget popultion of the MIP is enterprises locted in Germny with t lest five employees. 4 The 2005 survey wve of the MIP offers unique informtion on firms' mrket environment which is merged with dt from the 2006 survey wve to observe innovtion behviour of firms in the following period. Since we re interested in the interction between upstrem nd downstrem firms, we drop ll observtions of firms indicting tht privte households or public institutions re the lrgest customers. Additionlly we exclude ll firms reporting tht R&D expenditure exceeds their sles to void outlier problems. This leves us with smple of 1,137 observtions for which informtion on ll model vribles is vilble Vribles Innovtion incentives A number of uthors hve proposed different concepts for mesuring innovtion ctivities. 5 Since we re interested in innovtion incentives we choose n input mesure s proxy for the innovtion incentives of suppliers, s it represents discounted future rents ttched to innovtive efforts no mtter whether these efforts re successful. We use the R&D intensity which is defined s the expenditure on R&D ctivities divided by sles. It is widely used s mesure of innovtion input in the literture (see e. g. Cohen nd Levin, 1989; Crepon et l., 1998) Buyer power Our min explntory vrible of interest is buyer power. 6 As we define buyer power to result from reltively stronger brgining position of the buyer compred to the supplier, we hve to construct mesure which cptures whether supplier is confronted with buyer power or not. One of the fctors determining supplier s brgining position is the shre of sles generted by one buyer, s this mesure cn indicte substntil economic dependency. Once, " buyer ccounts for sufficiently lrge frction of supplier's overll business, this my led to more-thn-proportionl reduction in the vlue of the supplier's profits outside reltionship with the prticulr buyer" (Dobson nd Inderst, 2008, p. 339). This is due to the fct tht in cse the supplier loses the contrct with the buyer, the supplier's economic vibility could be undermined. Losing lrge contrct will result in free cpcity on the supplier's side nd will require the supplier to significntly lower prices in order to sell the excessive cpcity to remining buyers (Inderst nd Wey, 2007). Therefore, our mesure will include the extent to which supplier's sles depend on its three lrgest customers. The degree to which supplier is confronted with buyer power lso depends on the buyer s opportunities to switch to nother supplier. The ese of such switch is determined by the mrket structure in the supply mrket on the one hnd nd the substitutbility of the demnded product on the other. A monopoly in the supply mrket does not llow for n outside op- 4 For more detiled description of the MIP see Peters (2008). 5 For n overview see Hgedorn nd Cloodt (2003). 6 For n overview of ll pplied vrible definitions see Appendix A. 7
13 tion of the buyer, resulting in powerful brgining position of the supplier, even if the buyer is monopsonist. 7 Conversely, polypolistic supply mrket nd monopsonistic buyer mrket enble the buyer to behve opportunisticlly nd might led to hold-up. Tht is, fter the supplier crried out necessry investments to fulfil contrcted obligtions, the buyer my initite ex-post negotitions nd force the supplier to ccept conditions which reduce profit mrgins or even led to loss (Willimson, 1975). Thus, our mesure hs to include informtion bout the concentrtion in the supplier's mrket nd the substitutbility of the supplied products. Tble 1: Definitions of buyer power mesures Mesure of buyer power Shre of sles generted with the three lrgest customers Number of supplier s competitors Degree of substitutbility of supplier s product BP 1 50% BP 2 50% >5 OR High substitutbility (gree, fully gree) We consider different degrees of buyer power nd subsequently define two mesures (see Tble 1) using informtion on the shre of sles generted with the three lrgest customers, on the number of competitors in the suppliers min mrket nd whether the suppliers min product is esy to substitute by competitor products. Our first mesure for being exposed to buyer power BP 1 is dummy vrible reflecting the fct tht the three lrgest customers of supplier ccount for 50 or more percent of the sles. We interpret this s degree of dependency from buyers which could seriously undermine the economic vibility of supplying firm. 8 Hence it is included in ll mesures of supplier s exposure to buyer power. Our second mesure BP 2 equls BP 1 but tkes the vlue one only, if dditionlly the supplier hs either more thn 5 competitors or its products re esy to substitute. Compred to BP 1, this definition reflects weker brgining position of the supplier since it not only covers economic dependency in terms of sles but lso buyer s opportunities to switch to other suppliers Dimensions of competition in the buyer mrket To derive mesures for the intensity of price nd technology competition in the buyer mrket, it is desirble to hve informtion bout the identity of the most importnt buyers. Such dt is extremely difficult to obtin through voluntry surveys since most firms will refrin from 7 Such circumstnces, chrcterized by highly concentrted mrkets on both sides, hve been described s counterviling power sitution by Glbrith (1956). 8 One might object tht this mesure is not providing sufficiently ccurte degree of economic dependency, s the shre of sles generted by the lrgest single customer could be considerbly lower. However, in the merger cse Rewe/Meinl, the Europen Commission estblished tht supplier whose business with the two merging chins ccounted for more thn 22 percent hs to be considered s "economiclly dependent" on them. A survey mong grocery producers provided evidence tht this ws the most suppliers could fford to lose without serious dnger of bnkruptcy. With respect to our mesure of buyer power, the smllest possible shre of sles generted by one customer is roughly 17 percent, given tht supplier indictes he is economiclly dependent on his three lrgest customers. Hence, we consider our mesure to be sufficiently precise in order to correctly reflect serious economic dependency from buyers. 8
14 disclosing such informtion, nd sometimes confidentility greements with buyers restrict disclosure t ll. In the MIP 2005 survey, firms were sked to nme the sector of their three lrgest customers. Questionnire instructions helped firms to provide buyer sector informtion tht corresponds to 3-digit level of Nce 9, though firms did not give industry codes but short description of sectors which hve been coded to Nce 3-digits. Bsed on this sector informtion, we construct industry level mesures of competition. For the degree of price competition we use n industry's price cost mrgin (PCM) since it gives n indiction whether firms re ble to chieve mrgins high bove their mrginl costs. For the ske of interprettion, we trnsform the vrible to 1 PCM, i.e. vlues close to zero indicte low price competition in the buyer mrket nd vlues close to one refer to very intense price competition. As n indictor of technologicl competition we use sector s R&D intensity (RDint: R&D expenditure over sles) since firms will dedicte higher shre of their resources to R&D if keeping pce with technologicl chnge is crucil for competing in their mrket. We do not hve informtion on the loction of the lrgest buyers which implies tht we do not know whether they re domestic or interntionl buyers, but we do know the firms export shre in totl sles. We clculte both 1 PCM nd R&D intensity for Germny nd for OECD countries, to cpture the intensity of competition on domestic nd foreign mrkets. We weight the vlues with the respective shre of supplier s domestic nd interntionl sles. 10 In ddition, we lso introduce dummy vribles indicting the position of the buyer industry in the vlue chin. We distinguish between the production of rw mterils, intermediries, cpitl goods, consumer goods, producer services nd consumer services Competitive environment of the supplier A supplier s incentive to invest in innovtion ctivities my be shped by the competitive environment in their own mrket s well. Therefore, we control for concentrtion in the supplier's mrket since monopoly or oligopoly my llow for higher mrgins nd thus for higher investments in R&D or conversely for lower incentives to invest in R&D. 12 Concentrtion in the supplier's mrket is mesured by two dummy vribles cpturing the number of min competitors. The first dummy tkes the vlue one if the firm responded to hve no competitors nd zero otherwise. The second dummy tkes the vlue one if the firm indicted to hve t most 5 min competitors nd zero otherwise. Wht is more, there re firm chrcteristics which llevite the influence of powerful buyers nd strengthen supplier s brgining position. On one hnd, compred to single-productsupplier, high degree of product diversifiction offers more outside options to the supplier nd llows to escping from profit squeeze s described bove. Hence we include the degree of product diversifiction of supplier, mesured s the shre of sles which is not generted 9 Nce is the industril clssifiction system used in Europen Union sttistics. This study uses Nce rev For detiled description of buyer mrket competition mesures see Appendix B. 11 The definition of these industry groups cn be found in Tble C1 in the Appendix. 12 For n overview of the extensive literture deling with the effects of mrket structure on innovtion see e.g. Cohen (2010). 9
15 by supplier s min product line. 13 On the other hnd, suppliers my be very ctive in expnding their mrketing ctivities. The extension of mrketing ctivities cn ttrct new customers nd hence increse the number of the supplier's outside options. We control for this by dummy vrible indicting whether the supplier hs introduced new design or new methods for selling products in the pst three yers Further control vribles Following the literture on firms propensity to innovte (see e. g. Cohen, 1995; Crepon et l., 1998), we lso include firm size mesured by the number of employees, firm ge (in logs), firm's bility to bsorb knowledge mesured by the shre of grduted employees nd whether firm belongs to n enterprise group s explntory vribles. Moreover, we lso control for firm's sector ffilition nd whether firm is locted in the territory of the former GDR Estimtion strtegy Innovtion incentives of suppliers re shped by supplier's brgining position vis-à-vis its buyers nd by the type of competition tht chrcterises buyer mrkets. Accordingly, we model the innovtion decision of supplier to be dependent on mesure reflecting the reltion between supplier s nd buyer s brgining position, the dimensions of competition in the buyer mrket s well s further determinnts. Since we mesure innovtion incentives by R&D intensity, we hve to tke possible selection bis into ccount s this vrible is only observble for firms tht engge in reserch nd development ctivities. To control for this we pply the well-known generlised Tobit model (Heckmn, 1979). This pproch furthermore enbles us to seprte the effect of buyer power nd buyer mrket competition on the supplier's probbility to strt R&D ctivities from the effect on the decision how much to invest in R&D once the supplier decided to strt R&D. As mny uthors point out, when using generlised Tobit model one needs to mke sure to hve n exclusion restriction which explins the selection but not the structurl eqution nd is not correlted to the error term of the ltter. We use firm size to be the exclusion restriction for two resons. First, firm size will positively ffect the probbility to strt R&D ctivities s lrger firms hve n dvntge in spreding the fix costs of R&D over lrger output (Cohen nd Klepper, 1996). At the sme time firm size should not hve n effect on R&D intensity s the ltter is lredy by definition scled by size. Second, relted reserch frequently uses firm size s exclusion restriction in this context (see e. g. Griffith et l., 2006). In the first stge we estimte the probbility of supplier to spend positive mount on R&D ctivities in the next period which is followed in the second stge by the estimtion of R&D intensity given the supplier strted with R&D ctivities. 14 Anlogous to Crepon et l. (1998) we ssume tht firms tke up R&D ctivities if discounted future profits from R&D ctivities re positive. Let RD * i,t+1 be the discounted future profits from R&D of supplier i in period 13 Note tht the extent to which the lrgest three customers ccount for supplier's sles refers to the min line of product. 14 We use the R&D ctivities in t+1 to void simultneity problems, which my occur from the fct tht R&D investments of supplier cn lso hve n effect on supplier s exposure to buyer power s well s supplier s mrket environment. 10
16 t+1. These expected profits re not observble nd depend on the supplier s brgining position reltive to those of the buyer, i.e. whether the supplying firm is subject to the exertion of buyer power (BP), nd the intensity of competition in the buyer industry (BC k with k = 1, 2) reflecting the price nd technologicl dimension of competition, respectively. The cross term ccounts for possible interctions between supplier s exposure to buyer power nd the type of competition in the downstrem mrket nd llows us to test our hypotheses. In ddition we include vector of vribles reflecting supplier's competitive environment (SC) which is likely to shpe the innovtion incentives of the supplier, too. Other relevnt firm specific chrcteristics re cptured by vector of control vribles (X). The unobserved error term is represented by ε. We observe tht firms invest in R&D in t+1 if RD * i,t+1 is positive. Furthermore we ssume the true R&D intensity RDint * i,t+1 of supplier i in period t+1 to be determined by mostly identicl set of explntory vribles, i. e. the supplier s brgining position reltive to those of the buyer (BP), the intensity of competition in the buyer industry (BC k with k = 1, 2), the cross terms s well s vector of vribles reflecting supplier's competitive environment (SC). Firm specific chrcteristics re cptured by vector Y. We will estimte the following model with eqution (1) denoting the selection eqution nd eqution (2) denoting the intensity eqution. Note tht vector X is identicl to vector Y with the exception of firm size, since we need to tke the exclusion restriction into ccount. Due to the fct, tht RDint * i,t+1 is only observble when RD * i,t+1 is positive, we ssume joint normlity of both disturbnce terms ε i,t nd µ i,t. (1) (2) 4. Results 4.1. Descriptive Sttistics Tble 2 shows descriptive sttistics of the vribles of interest differentited by supplier s exposure to buyer power. The descriptive nlysis revels some interesting differences which re robust cross the two specifictions of the buyer power vrible. We find the shre of R&D performing firms to be significntly lower in the subsmples of suppliers confronted with buyer power. In fct, the shre decreses in the degree of buyer power. Contrstingly, we observe no significnt differences in the mens of R&D intensity. Considering the competition vribles in the most importnt buyer s industry, we find no significnt differences between the totl smple nd the vrious subsmples. Hence, there is no indiction for powerful buyers to hve more intensive competition my it be in terms of prices or R&D. We observe higher shre of dependent suppliers to be monopolists when we pply 11
17 the widest definition of buyer power B In ddition, suppliers being exposed to buyer power re less diversified in their product rnge, hve less employees, show less continuous R&D ctivities, re younger nd significntly higher shre of them is locted in the former Est Germny. Tble 2: Descriptive sttistics differentited by supplier s exposure to buyer power Men SD Men SD Men SD All BP1=1 BP2=1 RD t *** *** RDint t * b Buyer's 1-PCM t b Buyer's RDint t No. of competitors: None t ** *** No. of competitors: 1-5 t *** Product diversity t *** *** Mrketing t *** No. of employees (ln) t *** *** Shre of grduted employees t Continuous R&D ctivities *** *** Firm ge (ln) *** *** Prt of enterprise group t Est Germny t *** *** N 1, Results re derived from Two-smple t tests compring smple mens of dependent nd independent suppliers. Asterisks indicte the level of significnce tht the differences of smple mens re not equl 0: * p<0.10, ** p<0.05, *** p<0.01. The descriptive sttistics of the remining vribles is shown in Tble C2 in the Appendix. Dummy vrible b For detils on the clcultion see in the Appendix B Regression results Tble 3 shows the estimted coefficients using first specifiction without the interction terms. The columns 1 nd 3 show the estimted coefficients of the selection eqution while columns 2 nd 4 show the estimted coefficients of the intensity eqution. With respect to our mesures of supplier s exposure to buyer power we predominntly find effects on the decision how much to invest in R&D. Regrdless of the mesure for buyer power, the estimted coefficients show significntly negtive sign in the intensity equtions with the levels of significnce incresing the nrrower our definition of buyer power. The mgnitude of the coefficient is higher in the estimtion using BP 2, indicting tht the effect is stronger the more ccurte buyer power is mesured. For the selection eqution, we find no significnt coefficient for BP 1 while BP 2 is negtive lthough it is only wekly significnt. Regrding the dimensions of competition in the most importnt buyer s mrket no mtter whether this buyer hs stronger brgining position thn the supplier we find no evidence tht price competition is correlted to the probbility of supplier to strt R&D ctivities when pplying BP 1. For BP 2 however there is wekly significnt negtive correltion, i. e. the stronger price competition in the downstrem mrket, the lower supplier s probbility to 15 Note tht the definition of BP 2 requires the suppliers hs to hve more thn 5 competitors or high degree of substitutbility. We thus refrin from interpreting the descriptive sttistics on the number of competitors nd show them for the ske of completeness. 12
18 strt R&D ctivities. Across both model specifictions we find no evidence tht downstrem price competition ffects the R&D intensity of suppliers. Regrding R&D competition in the downstrem mrket, we find wekly significnt coefficients in the selection eqution while the coefficient estimtes re insignificnt for the intensity eqution cross both model specifictions. Hence, downstrem R&D competition intensity positively ffects the decision to strt R&D while it hs no consequences for supplier s decision how much to invest. Tble 3: Estimted coefficients of the Generlised Tobit model without interction terms RD i,t+1 RDint i,t+1 RD i,t+1 RDint i,t BP1 t ** (0.104) (0.011) BP2 t * *** (0.112) (0.012) b Buyer's 1-PCM t * (0.449) (0.045) (0.449) (0.045) b Buyer's RDint t * * (0.059) (0.004) (0.059) (0.004) No. of competitors: None t * ** ** * (0.275) (0.036) (0.276) (0.036) No. of competitors: 1-5 t (0.097) (0.009) (0.098) (0.009) Product diversity t (0.207) (0.020) (0.207) (0.020) Mrketing t *** *** (0.114) (0.011) (0.114) (0.011) No. of employees (ln) t *** *** (0.037) (0.037) Shre of grduted employees t *** *** *** *** (0.278) (0.025) (0.278) (0.025) Continuous R&D ctivities t *** *** *** *** (0.125) (0.027) (0.125) (0.026) Firm ge (ln) (0.057) (0.005) (0.057) (0.005) Prt of enterprise group t ** ** (0.099) (0.010) (0.099) (0.010) Est Germny t (0.103) (0.010) (0.103) (0.010) Mills Lmbd * (0.033) (0.032) Constnt *** *** (0.387) (0.058) (0.388) (0.058) N 1,129 1,129 LR/Wld chi P-vlue * p<0.10, ** p<0.05, *** p<0.01. Stndrd errors in prentheses. Estimtions include 20 industry dummies nd 5 dummies cpturing the buyer s position in the vlue chin. Estimtion results cn be found in Tble C 3 in the Appendix. Dummy vrible b For detils on the clcultion see Appendix B. With respect to the further explntory vribles we find highly robust results irrespective of the pplied mesure of buyer power exposure. Suppliers holding monopoly exhibit lower likelihood to invest in R&D but once they decided to strt R&D ctivities they invest more in R&D. In ddition, we find significntly positive coefficients in the selection eqution for lrger firms nd for suppliers who introduced mrketing instruments. For the shre of grduted employees we find significntly positive coefficient estimtes in both equtions of the model, implying tht higher bsorptive cpcity contributes positively to the likelihood of investing in R&D s well s to higher R&D intensity. Similrly, performing continuous R&D ctivi- 13
19 ties lso positively ffects both the likelihood of strting R&D s well s the R&D intensity. For supplier being prt of n enterprise group we find significntly negtive coefficients in the intensity eqution which indictes tht R&D is optimized within the group nd resources re llocted ccordingly mong subsidiries. Mills lmbd is significnt in our estimtions using BP 2, which is n indiction tht controlling for selection is necessry in our dtset. Let us now consider the results of the estimtions including the interction terms which re presented in Tble 4. Agin, columns with odd numbers show the estimted coefficients of the selection eqution while columns with even numbers show the estimted coefficients of the intensity eqution. Tble 4: Estimted coefficients of the Generlised Tobit model with interction terms RD i,t+1 RDint i,t+1 RD i,t+1 RDint i,t BP1 t *** (0.547) (0.060) BP2 t ** * (0.568) (0.063) b Buyer's 1-PCM t *** ** (0.554) (0.054) (0.535) (0.053) b Buyer's RDint t ** * (0.072) (0.005) (0.070) (0.004) BP b t x Buyer's 1-PCM t ** ** (0.862) (0.090) (0.891) (0.095) BP b t x Buyer's RDint t * (0.118) (0.007) (0.124) (0.009) No. of competitors: None t * ** ** * (0.278) (0.036) (0.277) (0.036) No. of competitors: 1-5 t (0.098) (0.009) (0.098) (0.009) Product diversity t (0.208) (0.019) (0.208) (0.020) Mrketing t *** *** (0.114) (0.010) (0.114) (0.011) No. of employees (ln) t *** *** (0.037) (0.037) Shre of grduted employees t *** *** *** *** (0.280) (0.025) (0.280) (0.025) Continuous R&D ctivities *** *** *** *** (0.125) (0.026) (0.125) (0.026) Firm ge (ln) (0.057) (0.005) (0.057) (0.005) Prt of enterprise group t ** ** (0.099) (0.010) (0.099) (0.010) Est Germny t (0.104) (0.010) (0.104) (0.010) Mills Lmbd * (0.032) (0.032) Constnt * (0.434) (0.056) (0.428) (0.056) Wld-Test on joint significnce of BP, BP x Buyer's 1-PCM nd BP x Buyer's RDint χ 2 (3)= 7. 94** χ 2 (3)= 9. 97** χ 2 (3)= 7.37* χ 2 (3)= 8.27** N 1,129 1,129 LR/Wld chi P-vlue * p<0.10, ** p<0.05, *** p<0.01. Stndrd errors in prentheses. Estimtions include 20 industry dummies nd 5 dummies cpturing the buyer s position in the vlue chin. Estimtion results cn be found in Tble C 4 in the Appendix. Dummy vrible b For detils on the clcultion see Appendix B. 14
20 Compred to Tble 3, we now see significnt effects of buyer power in both selection equtions. Regrdless of the mesure, we find significntly negtive coefficients of buyer power. The inclusion of the interction terms disentngles counterviling effects of buyer power nd the intensity of price competition in buyer mrkets. The estimted coefficients of the interction with price competition in the buyer mrket re significntly positive with the level of significnce being stronger for BP 1 though. Our findings imply tht hving less powerful brgining position thn the most importnt buyer reduces the likelihood of supplier to strt R&D ctivities. This negtive effect is mitigted, however, if the powerful buyer fces strong price competition. A similr effect in the selection equtions is found for the effect of price competition in buyer mrkets. In contrst to the results of the bseline model, the inclusion of interction terms yields strongly significnt coefficient estimtes in both selection equtions. Hence, we find negtive correltion of price intensity in buyer mrkets with supplier s probbility to strt R&D which suggests tht hving buyers who operte in highly competitive environment with respect to prices reduces supplier s likelihood to invest in R&D. Interestingly, if buyers hve stronger brgining position compred to suppliers, this negtive effect is mitigted. The estimted coefficients of buyer power, price competition in the buyer mrket nd its interctions re more significnt in the selection eqution which pplies BP 1. The reson for this is most likely tht the lower number of observtions in the group of BP 2 in combintion with dditionl two interction terms regrding competition intensity in buyer mrkets increses the stndrd error nd lowers the level of significnce for estimted coefficients. Considering the effect of buyer mrket R&D competition we find neither significnt effect of the min term nor the interction with buyer power. The significnt results of the intensity equtions regrding the min terms of buyer power re not robust to the inclusion of the interction terms. While BP 1 looses significnce completely, the significnce level of BP 2 drops from 1 to 10 percent. The interction effect with R&D competition intensity in the buyer mrket is significntly negtive in the specifiction pplying BP 1 lthough the level of significnce is rther low while it is insignificnt for BP 2. In ddition, the inclusion of the interction terms revels leds to significntly positive coefficient estimtes of buyer mrket R&D competition intensity which implies tht the higher the R&D intensity in the buyer mrket, the higher the R&D intensity of supplier. The results presented in column 6 suggest however, tht this positive impct is counterviled if the supplier is subject to the exertion of buyer power. We find no evidence of buyer mrket price competition intensity to ffect the R&D intensity of suppliers, neither directly nor when intercted with buyer power. Yet the results of the Wld test show tht the Null-hypothesis of the coefficient estimtes of buyer power nd its interctions being zero, cn be rejected t resonble levels of significnce. The estimted coefficients of ll other explntory vribles re robust to the inclusion of the interction terms, i.e. they keep their sign nd their significnce in the selection s well s in the intensity eqution. 15