SMES inventive performance and proï tability in the markets for technology Giovanna Padula a n, Elena Novelli b, 1, Raffaele Conti c, 2 a Universitã â€oel. Bocconiâ€, Via G. Roentgen, 1, 20136 Milano, Italy b City university London, 106 Bunhill Row, London EC1Y 8tz, UK c Catã lica Lisbon School of business and Economics, Palma de Cima, 1649-023 Lisboa, Portugal a r t i c l e i n f o Available online 11 march 2015 Keywords Markets for technologies Vertical boundaries Firm inventive performance Firm proï tability Small medium enterprises a b s t r a c t This paper studies the inventive performance and proï tability of small and medium sized ï rms (SMES) that are compared â€oetechnology specialists†to the inventive performance and proï tability of SMES that are instead vertically-integrated. In this paper perspective, â€oetechnology specialists†are ï rms that specialize upstream in generating inventions and trade those inventions in disembodied formwith other ï rms, usually through licensing agreements. Instead, vertically-integrated ï rms are those ï rms that both generate inventions and commercialize products incorporating those inventions. We argue that technology specialists achieve a higher inventive performance than vertically-integrated ï rms, since they can accumulate deeper and broader inventive experience, whilst keeping a more ï exible organizational structure. These ï rms display a lower proï tability though, due to the imperfections inherent in invention market transactions and the lower bargaining power caused by the lack of commercialization assets. The theoretical framework is tested through a cross-industry investigation on a sample of European SMES. Implications for the viability of being a technology specialist as a strategy and for the development of markets for technology are discussed & 2015 Elsevier Ltd. All rights reserved 1. Introduction Recent studies have established the increasing importance of markets for technology (e g.,, Arora et al. 2001; Krammer, 2014 Ritala and Hurmelinna-Laukkanen, 2009; Veer and Jell, 2012; Wang et al. 2012) †hereafter, MFT€ that is, markets where inventions are traded as â€oefree standing†entity, disembodied from individuals, orga -nizations and products (e g.,, Arora et al. 2001). ) In these markets ï rms can exchange their inventions for a price, usually through a licensing agreement, which is a contract where the owner of an invention allows another party the right to use or modify it in exchange of compensation (WIPO, 2014. Previous research on MFT has mainly taken a policy perspective on this phenomenon, arguing that the development of these markets allows for an efï cient division of innovative labor among small and large ï rms according to their comparative advantage†which is, respectively, doing research and generating inventions for small ï rms, and producing and marketing the ï nal products that embody new inventions for large ï rms (Arora et al. 2001; Arrow, 1983; Holmstrom, 1989. This type of conï guration is socially desirable, in principle, since every type of ï rm focuses on the activity it performs better (Firth and Narayanan, 1996; Li and Tang 2010); ) hence, a higher overall value might be generated compared to a situation where all ï rms internalize both the research and ï nal product commercialization activities (e g.,, Arora and Ceccagnoli 2006; Arora et al. 2001; Conceicao et al. 2012 However, the ï rm-level implications of MFT in terms of ï rm inventive performance (i e. the extent to which a ï rm is capable of generating valuable inventions) and proï tability have been largely neglected. It is not clear whether small ï rms are better off exploit -ing their comparative advantage in inventing by becoming â€oetech -nology specialists††that is, specializing upstream in the inventive activities and then sell their inventions in the MFT€ or whether they should vertically integrate†that is, commercialize their own inven -tions to ï nal customers. In particular, on the side of inventive per -formance, previous research on MFT has neglected largely how the interdependence between upstream invention and downstream product commercialization activities affects the ï rm's capacity to generate high quality inventions; consequently we still lack an understanding of whether the inventive performance of technology specialists overcomes that of vertically-integrated small ï rms. At the same time, on the side of proï tability, the literature on MFT has largely neglected to consider the ability of technology specialists to appropriate the economic returns of their inventions. Indeed, beco -ming a technology specialist and selling inventions to other ï rms Contents lists available at Sciencedirect journal homepage: www. elsevier. com/locate/technovation Technovation http://dx. doi. org/10.1016/j. technovation. 2015.01.002 0166-4972/& 2015 Elsevier Ltd. All rights reserved n Corresponding author. Tel.:à 39 02 58366823 E-mail addresses: giovanna. padula@unibocconi. it (G. Padula novelli@city. ac. uk (E. Novelli), raffaele. conti@ucp. pt (R. Conti 1 Tel.:à 44 20 7040 0991; fax: à 44 20 7040 8328 2 Tel.:à 351 217 214270; fax: à 351 217 270250 Technovation 41-42 (2015) 38†50 require ï rms to incur the private costs related to search and ne -gotiation in the MFT (e g. Fosfuri, 2006. In addition, being a technology specialist also implies that a ï rm lacks downstream complementary assets that have been demonstrated to be a relevant source of bargaining power (e g.,, Mcgahan and Silverman, 2006 Teece, 1986. Accordingly, we still do not know the extent to which†at the ï rm level†the economic beneï ts of being a technology specialist overcome the costs This study ï lls in these gaps by investigating the following research question: how does the choice of being a technology specialist (as opposed to being integrated a vertically ï rm) affect an SME's inventive performance and proï tability? Addressing these issues is important because it allows for an understanding as to what extent being a technology specialist is a viable strategy for an SME The rest of the paper is organized as follows. In Section 2 we present our theory and hypotheses and in Section 3 we describe the method that we used to test the hypotheses developed. In Section 4 we present the empirical results, while in Section 5 we discuss their implications to practice and theory. Finally, in Section 6, we present the conclusions from the study 2. Technology specialists vs. vertically-integrated SMES implications on inventive performance and proï tability Building on the principles of specialization and division of labor Smith, 1776 1983; Stigler, 1951; Young, 1928), literature on MFT has argued that small and large ï rms are naturally endowedwith different capabilities in inventing and commercializing; hence, they can beneï t from specializing in the activity in which they are relatively more efï cient (e g.,, Arora et al. 2001; Ceccagnoli and Jiang, 2013. In particular, we can represent the innovation value chain as the chain of activities from upstream research activities†i e.,, research and inven -tions generation†to downstream activities†i e.,, large-scale develop -ment of those inventions into products, manufacturing and marketing to the ï nal customers. Large, established ï rms, due to their highly bureaucratic structure, have a comparative advantage in performing downstream activities, which typically involve a high degree of routi -nization and standardization (Allarakhia andwalsh, 2011; Holmstrom 1989; Mangematin et al. 2011). ) Small ï rms, instead, have a compara -tive advantage in performing upstream research activities because due to the low organizational distance between managers and res -earchers (e g. Arrow, 1983; Marion et al. 2012), they are more likely to pursue risky but potentially extremely valuable technological trajec -tories (Arrow, 1983; Arora et al. 2001 These arguments suggest that, at the system level, the division of value chain activities among ï rms on the basis of their comparative advantage leads to the generation of a higher value compared to a situationwhere every ï rm performs all these activities (i e. invention development and commercialization to ï nal customers) internally Hence, based on this argument, it would appear preferable†from a social welfare perspective†if SMES specialized in upstream research activities, i e. if they became â€oetechnology specialists†(Arora et al 2001). ) However, existing research in this area provides only limited insight on whether operating as a technology specialist also brings a â€oeprivate†advantage to SMES, that is, whether technology specialists have a better performance compared to the vertically-integrated SMES, i e. those SMES that internalize all value chain activities. More precisely, existing research on MFT has provided only limited con -sideration to the interdependence between upstream invention gen -eration and downstream commercialization activities. Consequently existing research has investigated not the extent to which this interdependence affects the inventive performance of small ï rms that are technology specialists, and only focus on the generation of inventions, vs. vertically-integrated small ï rms, which internalize both activities In addition, existing research on MFT has investigated not the extent to which SMES'proï tability is affected by the choice between upstream specialization vs. vertical integration. Becoming a technology specialist implies undertaking search and negotiation activities in the MFT; hence, it might require incurring additional costs that might reduce SME's proï tability (e g. Leiblein and Madsen, 2009. The extent to which these costs overcome the beneï ts of being a technology specialist has been overlooked by extant literature. Furthermore, a technology specialist lacks downstream complementary assets that a vertically-integrated ï rm instead possesses, with possible implications on its bargaining power and consequently on its proï tability com -pared to a vertically-integrated SME (e g. Fosfuri, 2006; Leiblein and Madsen, 2009. However, these implications have been neglected by extant studies. The goal of this paper is to ï ll this gap and compare the implications for an SME of being a technology specialist vs. being vertically-integrated, in terms of both their inventive performance and proï tability. In doing so this paper contributes to improving our understanding on the performance of SMES (Hoffman et al. 1998 We argue that being a technology specialist (as opposed to being a vertically-integrated ï rm) has a positive impact on a small ï rm inventive performance for two reasons. The ï rst reason relates to the deeper and broader inventive experience that technology specialists can accumulate (Leiblein and Madsen, 2009. Technology specialists devote all their efforts and resources to their inventive activity (Arora et al. 2001). ) This makes them more likely to enjoy faster accumula -tion of inventive experience in their technological ï elds compared to vertically-integrated small ï rms†which instead spread their reso -urces and attention across upstream (i e.,, invention) and downstream i e.,, commercialization) activities. While this argument holds for any ï rm (regardless of its size), it is even more salient for small ï rms whose resource endowments are compared typically scarcer to those of larger ï rms (Teece, 1986. This implies that technology specialist SMES tend to acquire a â€oedeeper†inventive experience than vertically -integrated SMES (Dà ez-Vial, 2009; Yelle, 1979 At the same time, because technology specialists have the ultimate goal to sell or license their technologies to other ï rms Bianchi et al. 2011; Veer and Jell, 2012), they have the incentive to generate inventions that target a greater variety of business applica -tions and customer needs compared to vertically-integrated ï rms whose research activity mainly serves in-house needs (Arora et al 2001; Grant, 2002; Hicks and Hegde, 2005. This argument holds a fortiori for smaller vertically-integrated ï rms, which, due to their resource constraints, usually operate in a limited set of market niches. This implies that technology specialists tend to acquire a â€oebroader†inventive experience than vertically-integrated ï rms and this effect is even stronger in the case of SMES (Hicks and Hegde 2005). ) Both a depth and breadth of inventive experience enable lessons learned from experience to accrue more steadily, thus generating better inventions (Katila and Ahuja, 2002 The second reasonwhy being a technology specialist (as opposed to being integrated a vertically ï rm) has a positive impact on a small ï rm's inventive performance is related to the organizational structure typically characterizing technology specialists vs. vertically-integrated ï rms, which makes the former better positioned to generate valuable inventions. A vertically-integrated ï rm is likely to display tight inte -rdependences between upstream organizational units†focused on res -earch and on the generation of valuable inventions†and downstream units†commercializing those inventions embodied into products for ï nal customers (Taylor and Helfat, 2009. These interdependences are likely to inhibit the generation of path-breaking inventions and rather favor path dependence at the expense of novelty (Powell, 1992; Taylor and Helfat, 2009. A very clear illustration for this mechanism is presented by Fosfuri and Roende (2009. Vertically-integrated compa -nies are companies where an upstream R&d unit and a downstream manufacturing unit coexist. In principle the R&d unit may select the research trajectory to be pursued between multiple alternatives, which G. Padula et al.//Technovation 41-42 (2015) 38†50 39 vary in their value and novelty. For instance, the R&d unit might choose between research trajectories likely to deliver radical and extr -emely valuable inventions, and other trajectories probably resulting in incremental inventions. Choosing a research trajectory oriented to the generation of radical inventions is likely to require the creation of new sets of manufacturing routines and expertise†and so, huge adaptation costs in the production units (e g. Linton andwalsh, 2013. This implies that these more radical trajectories naturally meet a strong internal resistance (Henderson, 1993; Henderson and Clark, 1990) in vertically -integrated ï rms. To avoid a costly internal conï ict, the R&d units of vertically-integrated ï rms are likely to lean towards incremental though probably less valuable) research trajectories A large bulk of empirical evidence supports the idea that vertically -integrated ï rms present systemic resistance to generating radical inv -entions that alter the relationships among different stages of the production process (Glasmeier, 1991; Mariotti and Cainarca, 1986 O†Connor and Demartino, 2006; Tripsas, 1997. In contrast, technology specialists can take advantage of a higher degree of freedom in their decision making, which stems from the absence of the typical organ -izational and coordinative constraints that characterize vertically -integrated ï rms. Given that experimentation and risk taking are crucial in the discovery of valuable technological solutions (e g.,, Ahuja and Lampert, 2001; Gupta et al. 2006), we suggest that the greater opportunity of technology specialists compared to vertically-integ -rated ï rms to undertake radical research paths is likely to result in a greater ability to generate valuable inventions All these arguments lead us to predict Hypothesis 1. Technology specialist SMES have higher inventive performance than vertically-integrated SMES While being a technology specialist may positively affect SMES 'inventive performance, at the same time it may also hamper their proï tability for two reasons related respectively to: a) the imperfec -tions that plague the MFT functioning (e g. Cockburn, 2007; Gans et al 2008) and b) the lower bargaining power of technology specialists in negotiations due to their lack of downstream (i e. commercialization assets (Mcgahan and Silverman, 2006; Teece, 1986 Consider ï rst the imperfections that obstruct the functioning of MFT. Existing research has emphasized how the actual volume of technology transactions occurring in MFT is much lower than it could be due to several imperfections in the functioning of these markets (Gans et al. 2008; Giuri et al. 2007). ) This clearly hampers the technology specialists'possibility to make proï ts through invention trading. For instance, it is complicated usually quite for a company that has generated a new invention to identify the right buyer, since this involves scanning multiple market niches and identifying technological problems to which the invention could constitute a solution (Ceccagnoli and Jiang, 2013; Cockburn, 2007 This generates very high search costs and, consequently, a reduc -tion in the proï ts that technology specialists can generate by selling their technologies in MFT Moreover, even once a potential buyer has been found, uncertainty about the market value of an invention might obstruct the transaction Gambardella, 2013. Having generated the invention the seller, com -pared to the buyer, usually has more information regarding its true value (Gans et al. 2008). ) This information asymmetry leads to a classic adverse selection problem (Beggs, 1992; Sakakibara, 2010) because the buyers are not always capable of selecting between good and bad inventions and they make offers that take into consideration the possibility that the acquired invention might be a low quality one Cockburn, 2007. As a result, sellers of good inventions end up receiving offers that are lower than what they know would be fair which reduces the likelihood of an agreement with the buyer. A similar dynamic also occurs in the circumstance in which the invent -ing ï rm itself is uncertain about the true value of its inventions†such as in the case of very novel and path-breaking inventions. In this situation inventing ï rms tend to be overoptimistic about the quality of their own inventions (Dushnitsky, 2010; Giuri et al. 2007). ) This reduces the chance to agree with the buyer on a price, because even fair offers tend to be perceived by the inventor as too low. Overall, the difï culties in reaching an agreement about the value of the invention with any potential buyers hamper the possibility that technology specialists will generate proï ts from their inventions. Consistently with these arguments, previous research has found that divergences over the ï nancial terms of licensing agreements†which is the usual way an invention is sold in MFT€ are some of the major reasons why negotiations break down (Cockburn, 2007 Finally, the trading of inventions is plagued by possible opportu -nistic behaviors by the transaction counterparts, especially in the absence of â€oeappropriate intellectual property rules, procedures, and protection†(Gouvea et al. 2012, p. 563. Such moral hazard issues are likely to induce prospective buyers to consider with caution the option of buying an invention on the market (Dechenaux et al. 2011 Dushnitsky, 2010. This is due to the fact that the knowledge und -erlying the inventions often displays tacit components in addition to codiï ed components (Arora, 1996; Winter, 1987. The effective transfer of such knowledge, therefore, requires a certain amount of complementary effort from the inventing ï rm side to assist the buyer in the complete understanding and integration of the invention in its products (Leone and Reichstein, 2012. However, inventing ï rms might display opportunistic behaviors and try to skimp on the full effort required to transfer knowledge to the buyer (Arora, 1996. This issue is complicated further by the fact that transactions of inven -tions often require highly specialized complementary investments from the buyers, who are exposed consequently to the risk of â€oehold up†(Shane, 2002. The risk of moral hazard and hold up reduces potential buyers'propensity to acquire external inventions. From the point of view of a technology specialist, this results in a further obstacle to proï t from invention trading Besides MFT imperfections, the second reason why technology specialists tend to be less proï table than vertically-integrated small ï rms is that a vertically-integrated ï rm†by deï nition provided with downstream assets†can sell its inventions embodied into ï nal products, without having to negotiate with a counterpart; by contrast a technology specialist has to engage in a negotiation with a ï rm provided with downstream assets to sell its inventions. In this type of negotiation, the margins accruing to the technology specialist tend to be squeezed due to the stronger bargaining position of the buyer that originates from the possession of downstream (commercialization vis-a-vis upstream (research) assets (Chiu et al. 2008). ) Hence, since the possession of downstream assets represents a critical determinant of the ability to appropriate the economic returns of an invention (e g Teece, 1986), vertically-integrated ï rms are better able to proï t from their inventions compared to technology specialists Previous empirical evidence supports our line of reasoning. For instance, Arora and Nandkumar (2012), examining the software secu -rity industry, found thatmft raise the value of marketing capabilities in ensuring ï rm survival, and simultaneously decrease the value of technological capabilities. In the same vein, but using a much broader dataset on all publicly traded U s. ï rms, Mcgahan and Silverman 2006) show that the stock market value of ï rms controlling down -stream assets in a focal industry increases when outsider players generate inventions that could be commercialized fruitfully within the industry. This happens because outsiders usually do not possess the relevant downstream assets to operate in the industry. As a result insiders tend to enjoy a higher bargaining power in negotiations and eventually appropriate a greater portion of the value generated through the transaction of inventions, reducing the proï ts accruing to ï rms who do not possess downstream assets (i e. technology specialists To summarize, the imperfections of MFT and the limited bar -gaining power of technology specialists determined by their lack of G. Padula et al.//Technovation 41-42 (2015) 38†5040 downstream assets exert a negative effect on the ability of technol -ogy specialists to proï t from their inventions in MFT. Accordingly we hypothesize that Hypothesis 2. Technology specialist SMES have lower proï tability than vertically-integrated SMES 3. Method The empirical investigation of this study was accomplished on a population of European-based SMES, across all industries, within the timeframe 1996†2001. Coverage across all industries provides the advantage of permitting a systematic investigation of the study's predictions. Geographic restriction to Europe is motivated by the fact that huge institutional differences characterize markets for technology across different regions throughout the world, a circumstance that may have an impact on the performance of the ï rms under investigation in this study (Ginarte and Park, 1997. As a consequence, focusing on a speciï c and relatively homogenous geographical area may guarantee that many of these features rem -ain constant across this study sample, enabling a more robust test of the hypotheses. However, as the appropriability regime may still be expected to vary from country to country even within the European area, a control for the strength of patent protection was included in the statistical analyses (Ginarte and Park, 1997 While the 1990s were characterized by the steady increase in the volume of market transactions of inventions and by the increase in variance across ï rms in terms of their vertical boundaries and invention-commercialization choices, the greatest changes in this direction occurred†at least in Europe†in the second half of the 1990s, that is the temporal window on which this study is focused 3. 1. Sample and data We used a cross-sectional dataset of Europe-based SMES across all industries in the timeframe 1996†2001 to test our hypotheses. The choice of employing a cross-sectional dataset instead of a panel dataset is motivated by the concern for the reliability of yearly data on invention commercialization strategies, provided that our sample is composed by SMES. Indeed, forming a panel dataset would require the collection of yearly data on ï rms'invention commercialization strategies (i e. yearly data on whether each ï rm had sold or licensed its inventions to other ï rms or had embodied its inventions into products). ) Having conducted an accurate and extensive pilot search on multiple data sources (including Business & Industry, Factiva Zephyr and Securities Data Corporation databases as well as com -pany web sites and specialized websites) we discovered that collect -ing yearly data for small private companies was problematic since these ï rms do not receive systematic media coverage. Therefore it is not possible to ï nd each licensing agreement or each product launched by each of these companies in each single year reported on public sources However, our pilot search supported the idea that expanding our cross-sectional analysis to a time window of six years would lead to a reliable assessment of ï rms'strategies. In fact, we found that if a company engages in a strategy of exclusively using licensing agre -ements to commercialize its inventions, the likelihood that in a period of six years at least one of its licensing agreements will be announced on a website or in a corporate report is quite high. Sim -ilarly, if a company has launched products based on its inventions this information is likely to appear at least once on the materials we collected on the company in the six years window of reference Furthermore, the invention commercialization strategies of ï rms in our sample seem quite stable in the temporal window under inves -tigation of our study. Hence, a cross-sectional perspective seems not only a way to bypass the reliability problem that a panel approach would cause, but also a more appropriate approach from the stand -point of yearly data variability. The choice of using a cross-sectional dataset is in fact in line with other studies in the ï eld such as Arora and Gambardella (1990), Fosfuri (2006) and Gans et al. 2002 Firms that had at least one EPO-granted patent that had been applied for within the time frame 1996†2001 were included in the sample. Using the patent application date and not the grant date enables us to control for differences in delays that may occur in granting patents after the application is led ï (e g. Trajtenberg, 1990 Furthermore, the protection of patent, once granted, is retroactive and also covers the application period. Two motivations underlie the decision to include in the sample ï rms with at least one patented invention. First, patents represent an externally validated measure of inventive activity (Belenzon and Patacconi, 2013; Griliches, 1990 Second, patent protection reduces several frictions that typically characterize the trading of inventions and has a huge effect on the likelihood of selling or licensing an invention to other ï rms (e g Arora and Ceccagnoli, 2006; Gans et al. 2008). ) Hence including ï rms with at least one patented invention allows a reliable identiï cation of the ï rms â€oeat risk†of engaging in invention trading activities Company names identiï ed from the patent database have been matched with company names from the Amadeus database (Bureau Van dijk; hence both listed and non-listed companies were incl -uded in our sample. Checks for misspelling of company names were made and corrected. Subsidiaries at the parent level were then tracked on Amadeus, in order to exclude from the sample all ï rms that proved to be subsidiaries of large ï rms or joint ventures. Ama -deus was employed then to discriminate between large and small -medium ï rms. As this study is concerned with SMES, ï rms were retained in the sample only if they showed no more than 250 employees in at least one year within the timeframe 1996†2001 covered by this study. As indicated by the European commission 250 employees is the standard cut off point to identify SMES in the European context (Recommendation 2003/361/EC These sample construction rules provided the master list that was employed to collect the data that we used in this study. Data on ï rms'vertical integration and invention trade were collected and triangulated through an extensive search of press releases, including Business & Industry, Factiva, Zephyr and the Securities Data Corpora -tion (SDC) databases as well as from company web sites. In cases where this information was not available from current companies 'websites, or if the companies'websites were no longer active, the Internet Archive's Wayback Machine was used to visit the past websites (Yadav et al. 2007). ) Data on ï rms'inventive portfolios was collected using Patstat. Data on ï rms'age were obtained from company websites and Internet archives. Amadeus was employed to collect data on ï rms'proï tability and size across the whole time -frame covered by this study. Finally, to obtain data on the strength of the appropriability regime across the different European countries included in this study sample, this paper referred to publications by Park (2008) and Ginarte and Park (1997 The ï nal sample included 551 ï rms, of which 20 were technology specialists. Basic characteristics of industry afï liation and country of origins of the ï rms included in our sample are displayed in Tables 1 and 2 Table 1 reports the industry afï liation for all ï rms in the sample on the basis of US SIC codes. The table is organized to allow for an immediate comparison between the distribution across industries of the overall sample and the distribution across industries of the technology specialists. Table 1 shows that the overall sample of our European innovative SMES tends to be distributed across high and low tech industries, though a relatively larger majority of them actually belongs to high tech sectors. In fact, the most represented industries where European innovative SMES are active are SIC 35 31.4%,Industrial and Commercial Machinery and Computer Equip -ment), SIC 34 (11.98%,Fabricated Metal Products, Except Machinery G. Padula et al.//Technovation 41-42 (2015) 38†50 41 and Transportation Equipment; SIC 87 (9. 07%,Engineering, Account -ing, Research, Management and Related Services; SIC 36 (7. 80 %Electronic, Other Electrical Equipment and Components; SIC 28 7. 62%,Chemicals and Allied Products; SIC 38 (5. 26%,Measuring Analyzing and Controlling Instruments; Photographic, Medical and Optical Goods; Watches and Clocks) and SIC 30 (4. 36%Rubber and Miscellaneous Plastics Products. Overall, ï rms in these sectors constitute more than 75%of the sample, and these sectors represent predominantly†though not exclusively†high tech business activ -ities. Firms in the remaining sectors are fragmented across a high number of industries, where high tech business activities are much less represented By replicating the same analysis for technology specialists, we observe that technology specialists are concentrated relatively more in high tech industries compared to the overall sample. In particular we ï nd that the great majority of technology specialists belong to SIC 87 (65%,Engineering, Accounting, Research, Management and Related Services. Other SIC represented include SIC 28 (15%Chemi -cals and Allied Products; SIC 34 (10%,Fabricated Metal Products Except Machinery and Transportation Equipment; SIC 27 (5 %Building Construction General Contractors and Operative Builders and SIC 13 (5%,Oil and Gas Extraction. In order to understand in more detail the activity of technology specialists, which constitute the focus of our investigation, we closely investigated the inventive proï le of the companies that became technology specialists. Con -cerning the technological area of activity of technology specialists, we ï nd that among the 20 technology specialists, 55%(11 companies are in the biotech technological ï eld. Among the remaining compa -nies, 2 focus on the generation of mechanical technologies for the aeronautic and automotive sectors, 2 are in IT/electronics (generating magnetic tagging technologies and technologies for switchboards 3 companies generate chemical technologies (generating respectively thermoplastic elastomer technologies, composting technologies and chemical active ingredients), 1 company generates toys and 1 com -pany generates technologies for oil and gas offshoring Overall, technology specialists in our sample appear to be concentrated in high tech sectors characterized by strong appro -priability regimes. Moreover the majority of technology specialists are in the biotech sector. This is in line with extant studies on MFT indicating biotechnology as one of the ï elds where invention trade has developed more in the last decades (Arora et al. 2001; Bianchi et al. 2011). ) Indeed, research in this area indicates that†beginning in the 1970s, several small R&d intensive biotech companies, mostly Table 1 Industry afï liation: overall sample and technology specialists Description US SIC All ï rms in the sample Technology specialists Num%Cum%Num%Cum %Industrial and commercial machinery and computer equipment 35 173 31.40 31.40 Fabricated metal products, except machinery and transportation equipment 34 66 11.98 43.38 2 10.00 10.00 Engineering, accounting, research, management, and related services 87 50 9. 07 52.45 13 65.00 75.00 Electronic and other electrical equipment and components, except computer equipment 36 43 7. 80 60.25 Chemicals and allied products 28 42 7. 62 67.88 3 15.00 90.00 Measuring, analyzing, and controlling instruments; photographic, medical and optical goods; watches and clocks 38 29 5. 26 73.14 Rubber and miscellaneous plastics products 30 24 4. 36 77.50 Miscellaneous manufacturing industries 39 21 3. 81 81.31 Business services 73 12 2. 18 83.48 Primary metal industries 33 10 1. 81 85.30 Furniture and ï xtures 25 10 1. 81 87.11 Transportation equipment 37 9 1. 63 88.75 Paper and allied products 26 7 1. 27 90.02 Lumber and wood products, except furniture 24 7 1. 27 91.29 Stone, clay, glass, and concrete products 32 6 1. 09 92.38 Apparel and other ï nished products made from fabrics and similar materials 23 5 0. 91 93.28 Agricultural production 01 4 0. 73 94.01 Heavy construction other than building construction contractors 16 4 0. 73 94.74 Textile mill products 22 4 0. 73 95.46 Food and kindred products 20 3 0. 54 96.01 Printing, publishing, and allied industries 27 3 0. 54 96.55 1 5. 00 95.00 Building construction general contractors and operative builders 15 3 0. 54 97.10 Electric, gas, and sanitary services 49 3 0. 54 97.64 Leather and leather products 31 2 0. 36 98.00 Oil and gas extraction 13 2 0. 36 98.37 1 5. 00 100.0 Building materials, hardware, garden supply, and mobile home dealers 52 2 0. 36 98.73 Mining and quarrying of nonmetallic minerals, except fuels 14 1 0. 18 98.91 Construction special trade contractors 17 1 0. 18 99.09 Petroleum reï ning and related industries 29 1 0. 18 99.27 Transportation by air 45 1 0. 18 99.46 Transportation services 47 1 0. 18 99.64 Apparel and accessory stores 56 1 0. 18 99.82 Personal services 72 1 0. 18 100.0 Total 551 20 Table 2 Country of origin: overall sample and technology specialists Country All sample Technology specialists Number%Cum%Number%Cum %Italy 300 54.45 54.45 France 73 13.25 67.70 1 5. 00 5. 00 Finland 38 6. 90 74.60 3 15.00 20.00 Great britain 37 6. 72 81.32 8 40.00 60.00 Netherlands 29 5. 26 86.58 Spain 25 4. 54 91.12 1 5. 00 65.00 Norway 19 3. 45 94.57 3 15.00 80.00 Germany 16 2. 90 97.47 1 5. 00 85.00 Denmark 14 2. 54 100.00 3 15.00 100.00 Total 551 20 G. Padula et al.//Technovation 41-42 (2015) 38†5042 US-based, entered the industry. Through time the sector in the US consolidated towards a structure of small upstream technology specialists (Arora et al. 2001), trading their inventions to down -stream companies. The analysis of the characteristics of our sample indicates that also in Europe, small biotechnologies ï rms tend to represent a high portion of the ï rms operating in MFT In Table 2 we report the distribution of our sample across countries. We note that 54.45%of the sample is composed by Italian companies, 13.25%by French companies, 6. 90%by Finnish compa -nies, 6. 72%by British companies and 5. 26%by Dutch companies. The remaining 13.43%is composed by Spanish, Norwegian, German and Danish companies. We did not impose any geographic restriction in our sample, which included all European ï rms available in the Amadeus database having been granted at least 1 patent that had been applied at the EPO ofï ce in 1996†2001, and having no more than 250 employees in the same period Therefore the distribution of our sample is to some extent also informative of the geographical distribution of the population of these types of ï rms. The predominance of Italian ï rms in our sample is consistent with the evidence that the Italian economy is essentially based on small and medium enterprises. For instance in 1991,24. 2%of manufacturing ï rms in Italy had less than 10 employees, compared to 13.3%in the UK and 7. 8%in Germany OECD, 1997 It is also interesting to note that when we move to the sub -sample of technology specialists, the distribution indicates that 40 %of the sample is composed by British companies; another 45%is equally distributed amongst Danish, Finnish and Norwegian com -panies and ï nally, France, Spain and Germany constitute 5%of the sample each. Britain, Denmark and Norway are the countries where the ratio specialists vs. non-specialists is the highest (specialists constitute, respectively, 22%,21%and 16%of the companies from those countries in the sample) compared to the other European countries included in our sample 3. 2. Variables 3. 2. 1. Dependent variables This study employs two dependent variables corresponding to two distinct dimensions of ï rm performance: inventive performance and proï tability. As already speciï ed, in order to investigate ï rm inven -tive performance, we refer to patent data. However, patents substan -tially vary in their economic and technological value (Griliches, 1984 Sreekumaran et al. 2011; Trajtenberg, 1990. Thus, patent citations are a better indicator of the importance or value of patents than simple patent counts (Frietsch et al. 2014; Galasso and Simcoe, 2011; Hall et al.,, 2005; Hess and Rothaermel, 2011; Kelley et al. 2013; Trajtenberg 1990). ) Following extant literature in this area, we measure ï rm inventive performance using a citations-based index, i e. weighting each patent i of the ï rm by the actual number of citations (Ci) that it subsequently received (Trajtenberg, 1990. In particular, for each ï rm in the sample, the Inventive performance variable was computed as Pn i  1ã°1ã Ciã, where n is the count of the EPO-granted patents that had been applied for by the focal ï rm within the timeframe 1996†2001 and Ci is the number of citations subsequently received by each patent. Existing research suggests that the use of a citation-based measure of inventive activities effectively captures the value of the inventions developed by the ï rm (e g. Galasso and Simcoe, 2011; Hess and Rothaermel, 2011; Trajtenberg, 1990), hence the use of this indicator is consistent with the theory developed in this paper In calculating this variable two important issues were taken into account. First, citation counts are truncated inherently (Hall et al 2005; Rosenzweig and Mazursky, 2013. Patents continue to receive citations for long periods of time, while we observe only citations up to a certain point in time. Moreover, citations to patents applied for in earlier time periods (that had a longer time window to be cited cannot be aggregated and compared with citations to patents applied for more recently. In order to address this concern, for each patent of each ï rm in our sample, we counted the number of citations received in the ï rst three years after patent grant Second, inventors are likely to patent their inventions in multiple patent ofï ces. In these cases the same invention receives a different patent number, although the two patents are â€oeequivalent†from an invention standpoint. In particular extant literature suggests that unlike US patents, a large share of EPO patents are cited indirectly through their non-EPO equivalent (Hall et al. 2007). ) A proper count of forward citations should therefore also include citations received by patent equivalents (Harhoff et al. 2006). ) In order to address this issue we used the Patstat dataset to reconstruct patent families and track all citations received by each patent of each ï rm in the sample, including those to the patents non-EPO equivalent. This variable construction provided the measure of invention performance employed in this study To assess ï rm Proï tability, we calculated for each ï rm in the sample the average of the company's Return on Assets (ROA obtained within the timeframe 1996†2001 3. 2. 2. Independent variable Our independent variable indicates whether or not a ï rm had been a technology specialist within the timeframe under investiga -tion in this study. Consistently with our theory we use the expression technology specialist to indicate a ï rm that commercializes its inve -ntions exclusively as free standing entities as opposed to integrating these inventions into products. To identify ï rms that have traded their inventions we refer to ï rms who have engaged in invention licensing activities, following a large body of prior research in this area (e g. Arora et al. 2001; Fosfuri, 2006; Somaya et al. 2011; Leone and Reichstein, 2012. Accordingly, for each ï rm a dichotomous variable, Technology specialist, was constructed and valued 1 if we found evidence that, in the period 1996†2001,1) the ï rm engaged in licensing activities and (2) had embodied not its inventions into products, 0 otherwise. For each of the companies in our sample, we extensively searched different sources available (including Business & Industry, Factiva, Zephyr and the Securities Data Corporation (SDC databases as well as company websites and specialized websites) to identify announcements and reports mentioning the name of the ï rm. We used the Internet Archive's Wayback Machine to visit the version of the websites published in the period 1996†2001 (Yadav et al. 2007). ) We read the full text of all announcements. To assess whether the company had engaged in licensing activity we referred to the content of the announcement. For instance, we identiï ed as technology licensing agreements those cases in which the announce -ment: a) mentioned the transfers of inventions from the focal ï rm to other ï rms; b) included words such as â€oelicense†or â€oelicensingâ€; or c mentioned that the focal ï rm received a payment for the transfer of its invention (e g.,, referring to some speciï c licensing terms such as â€oeroyalties†or â€oefeesâ€. We also reviewed these sources to assess whether the ï rm, in the six years of interest, had embodied not the invention into products. To assess this, we referred to the same sources mentioned above. In many cases, companies explicitly spec -iï ed their strategy on their website or in their reports. When this information was not available, we searched the company website to check whether any products were advertised. We also searched news and specialized press to identify any announcements regarding product launches. Finally, in some cases we used the presence of manufacturing facilities to assess whether the company was active in the product market 3. 2. 3. Control variables There is a recognized†although controversial†relationship bet -ween a ï rm's size and its inventive performance (e g.,, Berends et al 2014; Cohen and Klepper, 1991; Feldman, 1997; Koen, 1992; Freeman G. Padula et al.//Technovation 41-42 (2015) 38†50 43 and Soete, 1997; Revilla and Fernandez, 2012; Rothwell and Zegveld 1982; Rubenstein and Ettlie, 1983; Shefer and Frenkel, 2005. The size of a ï rm may also affect its proï tability in different ways (e g. Berc -ovitz and Mitchell, 2007; Mas-Ruiz and Ruiz-Moreno, 2011. Indeed compared to large ï rms, small ï rms may be less able to exploit economies of scale and scope, or be more ï nancially constrained (e g Teece, 1986) which may cause a negative effect on the cost of capital e g. Apitado and Millington, 1992; Beedles, 1992. Although this study sample is formed by SMES, a variance across the size of the ï rms in the sample is present and may affect the result of the statistical analysis. To control for these effects, we calculated the variable Size for each ï rm as the minimum number of employees between 1996 and 2001 This study also controlled for ï rms'age. On one hand, age may affect the ability of a ï rm to build a reputation as a competent, reliable and trustworthy inventing ï rm, and consequently may have a positive impact on the chance to have accepted its inventions by the market and proï t (Danneels, 2002; Dowling and Helm, 2006; Katila, 2002 On the other hand, age may create organizational inertia and so neg -atively affect the ï rm inventive performance (Katila, 2002; Sørensen and Stuart, 2000. To control for all these effects, this study employed an Age variable that was constructed for each ï rm as a count of the years elapsed from the ï rm's foundation year to 2001 Characteristics of the ï rms'inventive portfolio may also have an impact on ï rm performance. One of the more critical characteristics in this regard is the generality of a ï rm's inventive portfolio, i e. the attitude of a ï rm to generate inventions more broadly applicable to a wide range of markets (Bresnahan and Trajtenberg, 1995; Gambardella and Giarratana, 2013; Hall et al. 2000; Valentini, 2012. By allowing access to a wider array of markets, a more general inventive portfolio may provide the ï rms with bigger market size, with positive effects on proï tability. To control for these effects, we measured the generality of the inventions according to the procedure employed by Trajtenberg et al. 1997). ) This measure accounts for the extent to which citations received by a patent are spread across different technological classes Speciï cally, for each patent i granted to the ï rms in our sample and applied in 1996†2001, the patent generality measure was calculated for each patent i, as follows: Patent generalityiâ 1ï j  1 s 2 ij where s 2 ij indicates the share of citations received by patent i from patents belonging to patent class j, out of n patent classes. To account for any forward citations truncation issues (Hall et al. 2005; Rosenzweig and Mazursky, 2013) we calculated the measure by using the citations received by each patent in the ï rst three years after patent grant. The patent generality measure was averaged then at the ï rm level to obtain a Firm invention generality variable We also control for the strength of the appropriability regime which might exert an important impact not only in determining the strategy chosen by a ï rm to commercialize its inventions (Teece's 1986) but also on the ï rm propensity to engage in invention activities in the ï rst place (Dosi et al. 2006; Laursen and Salter Table 3 Descriptive statistics and pairwise correlationa Descriptive statistics and pairwise correlation Variables Description Obs Mean SD Min Max 1 2 3 4 5 6 7 1. Inventive performance Pn i  1ã°1ã Ciã where n is the count of the EPO-granted patents that had been applied for within the timeframe 1996†2001 and Ci is the number of citations subsequently received by each patent 551 3. 492 7. 447 1. 000 106.000 1 2. Proï tability Mean of the company's Return on Assets (ROA obtained within the timeframe 1996†2001 551 5. 230 15.050 ï¿81.043 57.600 ï¿0. 133***1 3. Technology specialist Dummy variable taking value 1 is the ï rm engaged in licensing activity and did not sell products in the period 1996†2001 551 0. 036 0. 187 0. 000 1. 000 0. 067 ï¿0. 333***1 4. Size The minimum number of employees of the ï rm between 1996 and 2001 551 55.595 54.877 1. 000 248.000 0. 107**0. 110**ï¿0. 132***1 5. Age Count of the years elapsed from the ï rm's foundation year to 2001 551 37.031 34.305 1. 000 394.000 ï¿0. 056 0. 107**ï¿0. 152***0. 343***1 6. Firm invention generality Firm level mean of Patent Generalityi. Patent Generalityiâ 1ï j  1 s 2 ij where s2ij indicates the share of citations received by patent i from patents belonging to patent class j out of n patent classes 551 0. 365 0. 167 0. 000 0. 775 ï¿0. 024 0. 013 ï¿0. 033 ï¿0. 03 ï¿0. 093**1 7. Patent strength For each European country included in this study sample, average of the 1995 and 2000 patent protection indices obtained by Park's (2008 study (normalized 551 0. 898 0. 026 0. 788 0. 921 0. 025 0. 082*ï¿0. 089**ï¿0. 067 ï¿0. 042 ï¿0. 051 1 a*po0. 1;****po0. 05;*****po0. 0 G. Padula et al.//Technovation 41-42 (2015) 38†5044 2014; Teece, 1986. Hence, we controlled for the differences in the strength of patent rights across the European countries represented in this study sample by constructing a Patent strength control variable on the basis of the index of patent protection developed by Park 2008). ) This study was an update to 2005 and an extension to 122 countries of a previously developed patent protection index by Ginarte and Park (1997) covering 110 countries and referring to a time span from 1960 to 1990. In both studies, the index of patent protection was constructed, per country per quinquennium†within the timeframe 1960†1990 (Ginarte and Park, 1997) and the time -frame 1995†2005 (Park, 2008) †on the basis of ï ve categories of patent laws: 1) extent of coverage; 2) membership in international patent agreements; 3) provisions for loss of protection; 4) enforce -ment mechanisms; 5) duration. For each country and for each period they then scored each of these categories a value ranging from 0 to 1 and then summed up to constitute an overall value of patent index per country per period) ranging from 0 to 5. To construct our measure of Patent strength, per each European country included in this study sample, we took the average of the 1995 and 2000 patent protection index values. These results were normalized then so that the strongest possible level of patent protection is equal to 1 Finally, we included a set of Industry dummies in order to control for industry (deï ned at the level of one digit SIC code speciï c effects. Descriptive statistics and correlations are displayed in Table 3 4. Results We ï rst estimated the impact of being a technology specialist through an OLS regression (Table 4). Model 4. 1 estimates the inventive performance of the ï rm as a function of its choice to become a technology specialist and a set of controls, and tests Hypothesis 1 that technology specialist SMES display a higher inventive performance than vertically integrated SMES€ against the null hypothesis that the inventive performance of technology specialist SMES is not statistically signiï cantly different from the inventive performance of vertically integrated SMES. Results of Model 4. 1 show that the coefï cient of the variable technology specialist equals to 0. 386, which means technology specialists display an inventive performance about 47%greater than the inventive performance of vertically integrated ï rms (p value o0. 10. Model 4. 2 estimates instead the proï tability of the ï rm as a function of its choice to become a technology specialist and a set of controls, and tests Hypothesis 2 that technology specialist SMES dis -play a lower proï tability than vertically integrated SMES€ against the null hypothesis that the proï tability of technology specialist SMES is not statistically signiï cantly different from the proï tability of vertically integrated SMES. Results of Model 4. 2 show that the coefï cient of interest equals to ï¿0. 490, which implies technology specialists are about 39%less proï table than vertically integrated SMES (p value o0. 01 To account for the possibility that ï rms'choice to become a tech -nology specialist is endogenous to their performance, we emp -loyed a two stage least square model (2sls)( Wooldridge, 2002. In implementing this model, we have used the variable Technology speci -alist as the dependent variable of the ï rst equation, and Inventive performance and Proï tability, respectively, as the dependent variables of the second stage. We selected the average proportion of technology specialists in the same country and similar size of the focal ï rm as an instrument for the variable Technology specialist. The rationale behind this choice is related to the fact that some exogenous characteristics of the country's institutional environment (for instance, the Intellectual Property right (IPR) protection or the extent of local competition) may affect a SMES'decision to become technology specialists, and this inï uence varies according to the ï rm category size. Hence, in calc -ulating this variable we have grouped SMES in two groups: ï rms with less than 38 employees and ï rms with over 38 employees, where 38 employees is the median number of employees of ï rms in our sample In Table 5a and b we report the results from the 2sls. Model 5. 1 estimates the ï rst stage equation, which shows how the average proportion of technology specialists in the same country and similar size of the focal ï rm is correlated positively with the likelihood of the focal ï rm being a technology specialist. Model 5. 2 estimates the inventive performance of the ï rm as a function of its choice to become a technology specialist and a set of controls. Model 5. 3 estimates the ï rm proï tability as a function of the ï rm choice to become a tech -nology specialist and a set of controls. Results from bothmodel 5. 2 and Model 5. 3 largely conï rm the results of the OLS model and show that being a technology specialist has a positive impact on the inventive performance of a ï rm and a negative impact on ï rm proï tability consistent respectively with our Hypotheses 1 and 2 A possible concern regards the small number of technology specialists in our sample (20 over 551. To increase comparability among technology specialists and vertically integrated SMES (and also to further address any endogeneity issue) we replicated the analysis on a subsample which included technology specialists and a control group constituted by an equal number of similar non-technology specialists. In particular, we used a propensity score matching method to select the group of vertically integrated ï rms, similar to the technology specialist ï rms along several important dimensions which could determine the ï rm choice to become a technology specialist Dehejia and Wahba, 2002; Hasan et al. 2011; Rosenbaum and Rubin 1983), including ï rm age, size, ï rm invention generality, industry afï liation and appropriability at the country level. For each technology specialist, the closest matching company among the vertically inte -grated ï rms was chosen. We replicated the OLS regression analysis using this subsample of 40 companies. Results are reported in Table 6 Model 6. 1 estimates the inventive performance of the ï rm as a function of its choice to become a technology specialist and the set of controls, while Model 6. 2 estimates instead the proï tability of the ï rm as a function of its choice to become a technology specialist and the set of controls. The results support both Hypotheses 1 and 2 We also used a quantile regression to estimate the relationship between the choice of being a technology specialist and the ï rm's inventive performance and proï tability. In fact, the distributions of the two dependent variables (Inventive performance and Proï tability Table 4 OLS regression estimationa, b Model 4. 1 Model 4. 2 Inventive performance Log Proï tability Log Technology specialist 0. 386*ï¿0. 490 ***0. 206)( 0. 124 Size (Log) 0. 147***0. 006 0. 034)( 0. 020 Age (Log) ï¿0. 109**0. 072 ***0. 045)( 0. 027 Firm invention generality Log ï¿0. 001 ï¿0. 031 0. 057)( 0. 034 Patent strength (Log) 0. 129 ï¿0. 096 1. 197)( 0. 719 Industry dummies Included Included Constant 0. 671*4. 182 ***0. 352)( 0. 211 N. Observations 551 551 R squared 0. 073 0. 099 a*po0. 1;****po0. 05;*****po0. 01 b Since the minimum of the variable ï rm invention generality is added 0, we 0. 01 to the variable before taking the logarithm; since the minimum of the variable proï tability is ï¿81.043, we added 81.053 (min à 0. 01) to the variable before taking the logarithm G. Padula et al.//Technovation 41-42 (2015) 38†50 45 are characterized by heavy tails. Other studies, whose dependent variables were characterized by heavy tails, have employed a quantile regression (Coad and Rao, 2008; Koenker and Bassett, 1978. Results available upon request) are again consistent with our theory 5. Discussion These results have important implications for practitioners and researchers 5. 1. Implications to practice The results from this paper enhance our understanding of the viable strategies a small ï rm can choose for proï ting from its inventions. Our ï ndings suggest that, because of the imperfections that plague technology markets and of the low bargaining power of ï rms lacking downstream assets (i e.,, technology specialists), the choice of simply selling inventions disembodied from products in the MFT (as opposed to directly commercializing them to ï nal custo -mers) might not be the best option for SMES. To be sure, these results reï ect what happens on average across all industries in all European countries. However, our research might suggest that the viability of a technology specialist strategy would increase in those industries and /or countries where the strength of the IPR regime or the tendency to engage in trust-based behaviors limits the imperfections that hamper the well-functioning of MFT. In this respect, future research might better elaborate on the role of environmental and ï rm contingencies that make technology specialist SMES more proï table than vertically-integrated SMES The results from this paper also raise implications for policy makers. A key conclusion of past literature of MFT is that the diffusion of technology specialists†and the consequent development of MFT€ is socially desirable because it facilitates the division of innovative labor amongst small and large ï rms, which tend to have a comparative advantage, respectively, in generating inventions and commercializing them (e g.,, Teece, 1986. However, our study shows that while technology specialist SMES have better inventive performance com -pared to vertically-integrated SMES, they also display worse pro -ï tability, an outcome that over time might reduce the overall number of ï rms that choose this strategy. Hence, our ï ndings have relevant policy implications, because they might imply that the number of SMES deciding to become technology specialists might be lower than optimal For policy-makers, this emphasizes the importance of designing mechanisms that reduce the high transaction costs that plague MFT in order to increase technology specialists'proï tability. For instance policy makers could favor the emergence of specialized intermedi -aries†that is, ï rms providing services such as patent evaluation patent monetization and patent management, which might contri -bute to solve some of the imperfections affecting MFT (e g.,, informa -tion asymmetries between buyers and sellers. The investigation of Table 5 a Two stage least square regression estimation: ï rst stagea, b Model 5. 1 Technology specialist Instrumental variable 1. 446 ***0. 177 Size (Log) 0. 000 0. 007 Age (Log) ï¿0. 025 ***0. 009 Firm invention generality (Log) ï¿0. 010 0. 011 Patent strength (Log) 0. 240 0. 249 Industry dummies Included Constant 0. 101 0. 069 N. Observations 551 b Two stage least square regression estimation: second stagea, b Model 5. 2 Model 5. 3 Inventive performance (Log) Proï tability (Log Technology specialist 2. 765***ï¿2. 249 ***0. 694)( 0. 437 Size (Log) 0. 167***ï¿0. 008 0. 038)( 0. 024 Age (Log) ï¿0. 014 0. 003 0. 057)( 0. 036 Firm invention generality (Log) 0. 015 ï¿0. 044 0. 064)( 0. 040 Patent strength (Log) 1. 077 ï¿0. 797 1. 363)( 0. 859 Industry dummies Included Included Constant 0. 237 4. 078 ***0. 514)( 0. 324 N. Observations 551 551 a*po0. 1;****po0. 05;*****po0. 01 b Since the minimum of the variable ï rm invention generality is added 0, we 0. 01 to the variable before taking the logarithm; since the minimum of the variable proï tability is ï¿81.043, we added 81.053 (min à 0. 01) to the variable before taking the logarithm G. Padula et al.//Technovation 41-42 (2015) 38†5046 the role of intermediaries on the liquidity, transparency and efï -ciency of the MFT may constitute a promising new line of inquiry for future research 5. 2. Implications to theory This paper also has implications to theory. First of all, this study better speciï es the idea, developed by previous contributions on MFT (e g.,, Arora et al. 2001), that the division of innovative labor amongst small and large ï rms is optimal for the overall economy Indeed, our study shows that small ï rms that specialize in upstream activities, as opposed to spreading their limited resources among research and commercialization activities, generate more valuable inventions. However, since technology specialists are relatively less proï table than vertically-integrated ï rms, only a few SMES may decide to specialize upstream. Thus, the potential social beneï ts associated with the generation of better inventions by technology specialists might not be realized fully Second, this paper also has implications for research on ï rm survival. Recent studies have demonstrated the role of inventive performance on ï rms'survival (e g.,, Ceï s and Marsili, 2006. By emp -hasizing that technology specialists achieve a superior inventive perf -ormance but that these ï rms experience worse proï tability, this study calls for the need to improve our understanding of the relationship between inventive performance and ï rm survival. For instance, future research could analyze whether the positive relationship between a ï rm's inventive performance and survival only holds for those ï rms that have acquired the downstream assets needed to commercialize their inventions to the ï nal customers Finally, the results from this paper might also have implications for research on Venture capital (VC. It is established well that Venture capital (VC) is crucial for small and young ï rms perfor -mance (Bottazzi and Da Rin, 2002; Samila and Sorenson, 2011 The results of this study would suggest that the VC role could be particularly important when MFT are not well functioning, and so vertical integration is a better option for small ï rms in order to proï t from their inventions. Indeed, VC might provide ï nancially -constrained small ï rms not only with the necessary resources to invest in the acquisition of downstream assets (e g.,, Stucki, 2014 but also with the managerial expertise required to commercialize their products to the ï nal customers (e g.,, Robson and Bennett 2000). ) Hence, an interesting avenue for future research could be the exploration of the role played by VC as a potential substitute for MFT 5. 3. Limitations This study has some limitations. First, we only considered SMES inventive performance and proï tability, but other performance dim -ensions could be evaluated. For instance, future studies could inves -tigate whether the decrease in proï tability experienced by technol -ogy specialists is traded actually off with superior growth outcomes in terms of company size, or if the choice of being a technology specialist is more appropriate to foster ï rms'adaptability in the face of a changing environment. In this respect, it could be interesting to replicate this study across different time windows in order to see how ï rms using different strategies react to environmental and macroeconomic shocks Second, our sample of technology specialists is limited quite. Two issues might be considered in this respect, i e. whether (1) the data are representative of the overall populations of technology specialists in Europe in the time period considered;(2) the results obtained are reliable. With regard to the ï rst point, it should be noted that very limited data are available publicly on the population of SMES who are technology specialists. Hence, we believe that the selected sample is valuable because it allows us to provide an analysis on the behavior of a relevant type of ï rms that otherwise could not be investigated Regarding the second point, when the independent variable does not display relevant variation, there is a high risk of the results not being statistically signiï cant (e g.,, Wooldridge, 2002. Nevertheless our results are signiï cant. Hence we could argue that what we presented was a conservative estimate of the real effect and that our results could be even stronger if we had more variation in our main independent variable, that is, if we had a larger number of technol -ogy specialists Third, this study only focuses on small and medium ï rms. How -ever, the choice between selling an invention in the MFT and emb -edding it into a product to be sold to ï nal customers might in principle also regard other players, like large ï rms or users. Future research should therefore investigate to what extent and under what con -tingencies these players sell their ideas to other ï rms rather than selling their products to ï nal customers. In addition, we employ a cross sectional perspective in our analysis. The investigation of the same issue in a longitudinal perspective could potentially lead to Table 6 OLS regression estimation (after matching) a b Model 6. 1 Model 6. 2 Inventive performance (Log) Proï tability (Log Technology specialist 0. 732**ï¿0. 738 *0. 329)( 0. 366 Size (Log) ï¿0. 128 ï¿0. 103 0. 151)( 0. 168 Age (Log) ï¿0. 110 0. 319 0. 204)( 0. 227 Firm invention generality (Log) 0. 074 ï¿0. 105 0. 222)( 0. 247 Patent strength (Log) ï¿0. 382 ï¿2. 489 3. 833)( 4. 260 Industry dummies Included Included Constant 0. 720 4. 140 **1. 606)( 1. 785 N. Observations 40 40 R squared 0. 240 0. 212 a*po0. 1;****po0. 05;*****po0. 01 b Since the minimum of the variable ï rm invention generality is added 0, we 0. 01 to the variable before taking the logarithm; since the minimum of the variable proï tability is ï¿81.043, we added 81.053 (min à 0. 01) to the variable before taking the logarithm G. Padula et al.//Technovation 41-42 (2015) 38†50 47 insightful results, and also offer the possibility to control for ï rm time -invariant heterogeneity. However, we acknowledge that the panel data on technology commercialization strategies of small private ï rms represents a challenging task since these data are not available on public or commercial dataset. Future research may consider the possi -bility of using a survey approach in order to obtain some insight using a longitudinal dataset Finally, this paper refers to the use of licensing agreements as evidence of the fact that a ï rm pursued a â€oetechnology specialist†strategy, i e. it proï ted from its inventions by licensing them in disembodied form to other ï rms as opposed to incorporating them into products, and investigates the implications related to the use of this strategy on ï rm proï tability and inventive performance. It would be very interesting for future research to also investigate the antecedents of (i e. the reasons behind) the ï rm's choice to use licensing agreements. In addition to a monetary reason, there might be other strategic reasons that led ï rms to license their inventions such as entering a foreign market or setting an industry standard. In this respect, whilst the use of secondary data allows conducting such investigation only to a very limited extent, the use of surveys or in -depth case studies could provide new relevant insights in this area 6. Conclusion This study investigates the effects of being a technology specialist on ï rm inventive performance and proï tability. In particular, the results from this paper show that technology specialist SMES are better performers than vertically-integrated small ï rms in terms of inventiveness, but worse performers in terms of proï tability. Focus -ing their attention on inventive activities allows technology specia -lists to learn how to generate higher quality inventions faster than vertically-integrated ï rms, due to the deeper and broader inventive experience technology specialists have the chance to accumulate Arora et al. 2001; Katila and Ahuja, 2002; Yelle, 1979; and to their ï exible organization structure (Henderson and Clark, 1990; Fosfuri and Roende, 2009. However, the imperfections that plague the functioning of MFT (Cockburn, 2007), and the higher bargaining power of ï rms possessing commercialization assets vis-a-vis rese -arch assets (e g. Arora and Nandkumar, 2012; Teece, 1986), lead technology specialists to experience a signiï cantly lower proï tability than vertically-integrated ï rms This study has a number of practical and theoretical implications as already discussed in the previous section. One of the most inte -resting contributions is probably that it emphasizes the existence of a conundrum between what would be socially optimal†that is, the division of innovative labor amongst large ï rms, specializing in down -stream commercialization, and small ï rms, specializing in inventing†and what would be privately optimal for small ï rms, which instead have the incentive to integrate downstream to better capture the economic returns from their inventions. Hence, this study calls for the need to design and implement institutional mechanisms aimed at addressing this conundrum and reconciling private with social bene -ï ts in the context of ï rms'inventive activity Acknowledgements Wewould like to thank the editor, Jonathan Linton, and the two anonymous reviewers for their very constructive comments throughout the review process. We are also grateful to Alfonso Gambardella for his very insightful guidance in the earlier versions of this paper. The paper also beneï ted from the comments of the participants at the Academy of Management Conference (2010 and the Strategic Management Conference (2010. Giovanna Padula and Elena Novelli wish to gratefully acknowledge the PRIN funding support (Grant 2006135451) from the Italian Ministry of University and Research (MIUR. Elena Novelli also gratefully acknowledges the ï nancial support of the Economic and Social Research Council (ESRC) Future Research Leaders Scheme (Grant ES/K001388/1 References Ahuja, G.,Lampert, Morris C.,2001. Entrepreneurship in the large corporation: a longitudinal study of how established ï rms create breakthrough inventions Strateg. Manag. J. 22 (6†7), 521†543 Allarakhia, M.,Walsh, S.,2011. Managing knowledge assets under conditions of radical change: the case of the pharmaceutical industry. Technovation 31 (2/3 105†117 Apitado, V p.,Millington, J. K.,1992. Restrictive loan covenants and risk adjustment in small business lending. J. Small Bus. Manag. 30 (1), 38†48 Arora, A.,1996. Contracting for tacit knowledge: the provision of technical services in technology licensing contracts. J. Dev. Econ. 50 (2), 233†256 Arora, A.,Ceccagnoli, M.,2006. Patent protection, complementary assets and ï rms 'incentives for technology licensing. Manag. Sci. 52 (2), 292†308 Arora, A.,Fosfuri, A.,Gambardella, A.,2001. Markets for Technology: The Economics of Innovation and Corporate Strategy. MIT Press, Cambridge MA Arora, A.,Gambardella, A.,1990. Complementarity and external linkages: the strategies of the large ï rms in biotechnology. J. Ind. Econ. 38, 361†379 Arora, A.,Nandkumar, A.,2012. Insecure advantage? Markets for technology and the value of resources for entrepreneurial ventures. Strateg. Manag. J. 33 (3 221†251 Arrow, K. J.,1983. Innovation in large and small ï rm. In: Ronen, J. Ed.),Entrepre -neurship. Lexington Books, Lexington, pp. 15†28 Beedles, W. L.,1992. Small ï rm equity cost: evidence from Australia. J. Small Bus Manag. 30 (3), 57†65 Beggs, A w.,1992. The licensing of patents under asymmetric information. Int J. Ind. Organ. 10, 171†191 Belenzon, S.,Patacconi, A.,2013. Innovation and ï rm value: an investigation of the changing role of patents, 1985†2007. Res. Policy 42 (8), 1496†1510 Bercovitz, J.,Mitchell, W.,2007. When is more better? The impact of business scale and scope on long-term business survival, while controlling for probitability Strateg. Manag. J. 28, 61†79 Berends, H.,Jelinek, M.,Reymen, I.,Stultiã ns, R.,2014. Product innovation processes in small ï rms: combining entrepreneurial effectuation and managerial causa -tion. J. Prod. Innov. Manag. 31 (3), 616†635 Bianchi, M.,Cavaliere, A.,Chiaroni, D.,Frattini, F.,Chiesa, V.,2011. Organisational modes for open innovation in the biopharmaceutical industry: an exploratory analysis. Technovation 31, 22†33 Bottazzi, L.,Da Rin, M.,2002. Venture capital in Europe and the ï nancing of innovative companies. Econ. Policy 17, 229†270 Bresnahan, T. F.,Trajtenberg, M.,1995. General purpose technologies â€oeengines of Growthâ€? J. Econom. 65, 83†108 Ceccagnoli, M.,Jiang, L.,2013. The cost of integrating external technologies: supply and demand drivers of value creation in the markets for technology. Strateg Manag. J. 34, 404†425 Ceï s, E.,Marsili, O.,2006. Survivor: the role of innovation in ï rms'survival. Res Policy 35 (5), 626†641 Chiu, Y. C.,Lai, H. C.,Lee, T. Y.,Liaw, Y. C.,2008. Technological diversiï cation complementary assets, and performance. Technol. Forecast. Soc. Change 75 6), 875†892 Coad, A.,Rao, R.,2008. Innovation and ï rm growth in high-tech sectors: a quantile regression approach. Res. Policy 37 (4), 633†648 Cockburn I.,2007. Is the market for technology working well? Obstacles to licensing and ways to overcome them. In: Proceedings of the Conference on Economics of Technology Policy, Monte Veritã, Switzerland Cohen, W. M.,Klepper, S.,1991. The tradeoff between ï rm size and diversity in the pursuit of technological progress. Small Bus. Econ. 4, 1†14 Conceicao, O.,Fontes, M.,Calapez, T.,2012. The commercialization decisions of research-based spin-off: targeting the market for technologies. Technovation 32, 43†56 Danneels, E.,2002. The dynamics of product innovation and ï rm competences Strateg. Manag. J. 23 (12), 1095†1121 Dechenaux, E.,Thursby, J.,Thursby, M.,2011. Inventor moral hazard in university licensing: the role of contracts. Res. Policy 40 (1), 94†104 Dehejia, R. H.,Wahba, S.,2002. Propensity score-matching methods for nonexperi -mental causal studies. Rev. Econ. Stat. 84 (1), 151†161 Dà ez-Vial, I.,2009. Firm size effects on vertical boundaries. J. Small Bus. Manag. 47 2), 137†153 Dosi, G.,Marengo, L.,Pasquali, C.,2006. How much should society fuel the greed of innovators? On the relations between appropriability, opportunities and rates of innovation. Res. Policy 35, 1110†1121 Dowling, M.,Helm, R.,2006. Product development success through cooperation: a study of entrepreneurial ï rms. Technovation 26, 483†488 Dushnitsky, G.,2010. Entrepreneurial optimism in the market for technological inventions. Organ. Sci. 21 (1), 150†167 Feldman, M.,1997. Small Firm Innovative Success: External Resources and Barriers Maryann Feldman and Associates, Baltimore MD G. Padula et al.//Technovation 41-42 (2015) 38†5048 Firth, R. W.,Narayanan, W. K.,1996. New product strategies of large, dominant product manufacturing ï rms: an exploratory analysis. J. Prod. Innov. Manag. 13 334†347 Fosfuri, A.,Roende, T.,2009. Leveraging resistance to change and the skunk works model of innovation. J. Econ. Behav. Organ. 72 (1), 274†289 Fosfuri, A.,2006. The licensing dilemma: understanding the determinants of the rate of technology licensing. Strateg. Manag. J. 27 (12), 1141†1158 Freeman, C.,Soete, L.,1997. The Economics of Industrial Innovation. Pinter, London Frietsch, R.,Neuhã¤usler, P.,Jung, T.,Van Looy, B.,2014. Patent indicators for macroeconomic growth†the value of patents estimated by export volume Technovation 34, 546†558 Galasso, A.,Simcoe, T.,2011. CEO overconï dence and innovation. Manag. Sci. 57 1469†1484 Gambardella, A.,Giarratana, M.,2013. General technological capabilities, product market fragmentation, and markets for technology. Res. Policy 42, 315†325 Gambardella, A.,2013. The economic value of patented inventions: thoughts and some open questions. Int. J. Ind. Organ. 31 (5), 626†633 Gans, J. S.,Hsu, D. H.,Stern, S.,2002. When does start-up innovation spur the gale of creative destruction? RAND J. Econ. 33 (4), 571†586 Gans, J. S.,Hsu, D. H.,Stern, S.,2008. The impact of uncertain intellectual property rights on the market for ideas: evidence from patent grant delays. Manag. Sci 54, 982†997 Ginarte, J. C.,Park, W. G.,1997. Determinants of patent rights: a cross-national study Res. Policy 26, 283†301 Giuri, P.,Mariani, M.,Brusoni, S.,Crespi, G.,Francoz, D.,Gambardella, A.,Garcia -Fontes, W.,Geuna, A.,Gonzales, R.,Harhoff, D.,Hoisl, K.,Le Bas, C.,Luzzi, A Magazzini, L.,Nesta, L.,Nomaler, Ã.,Palomeras, N.,Patel, P.,Romanelli, M Verspagen, B.,2007. Inventors and invention processes in Europe: results from the Patval-EU survey. Res. Policy 36 (8), 1107†1127 Glasmeier, A k.,1991. Technological discontinuities and ï exible production net -works. The case of Switzerland and the world watch industry. Res. Policy 21 469†485 Gouvea, R.,Linton, J. D.,Montoya, M.,Walsh, S. T.,2012. Emerging technologies and ethics: a race-to-the-bottom or the top? J. Bus. Ethics 109 (4), 553†567 Grant, R.,2002. Contemporary Strategy Analysis. Blackwell Publishing, Malden MA Griliches, Z.,1984. R&d, Patents and Productivity. University of Chicago Press Chicago Griliches, Z.,1990. Patent statistics as economic indicators: a survey. J. Econ. Lit. 27 1661†1707 Gupta, A k.,Smith, K. G.,Shalley, C. E.,2006. The interplay between exploration and exploitation. Acad. Manag. J. 49 (4), 693†706 Hall, B.,Jaffe, A.,Trajtenberg, M.,2000. Market value and Patent Citations: A First Look. NBERWORKING Paper no. 7741. National Bureau of Economic Research, Inc Cambridge, MA Hall, B. H.,Jaffe, A.,Trajtenberg, M.,2005. Market value and patent citations. RAND J. Econ. 36 (1), 16†38 Hall, H. B.,Thoma, G.,Torrisi, S.,2007. The market value of patents and R&d evidence from European ï rms. Working 1paper 13426. National Bureau of Economic Research, Cambridge, Mass Harhoff, D.,Hoisl, K.,Webb, C.,2006. European patent citations†How to count and how to interpret them? University of Munich and CEPR (London), University of Munich, and OECD Hasan, I.,Kobeissi, N.,Wang, H.,2011. Global equity offerings, corporate valua -tion, and subsequent international diversiï cation. Strateg. Manag. J. 32 (7 787†796 Henderson, R.,1993. Underinvestment and incompetence as responses to radical innovation. Evidence from the photolithographic industry. RAND J. Econ. 24 (2 248†270 Henderson, R.,Clark, K. B.,1990. Architectural innovation: the reconï guration of existing product technologies and the failure of established ï rms. Adm. Sci. Q 35 (1), 9†30 Hess, A m.,Rothaermel, F. T.,2011. When are assets complementary? Star scientists strategic alliances, and innovation in the pharmaceutical industry. Strateg Manag. J. 32, 895†909 Hicks, D.,Hegde, D.,2005. Highly innovative small ï rms in the markets for technology. Res. Policy 34 (5), 703†716 Hoffman, K.,Parejo, M.,Bessant, J.,Perren, L.,1998. Small ï rms, R&d technology and innovation in the UK: a literature review. Technovation 18 1), 39†55 Holmstrom, B.,1989. Agency costs and innovation. J. Econ. Behav. Organ. 12 (3 305†327 Katila, R.,2002. New product search over time: past ideas in their prime? Acad Manag. J. 45 (5), 995†1010 Katila, R.,Ahuja, G.,2002. Something old, something new: a longitudinal study on search behaviour and new product introduction. Acad. Manag. J. 45 (6 1183†1194 Kelley, D. J.,Ali, A.,Zahra, S. A.,2013. Where do breakthroughs come from Characteristics of high-potential inventions. J. Prod. Innov. Manag. 30 (6 1212†1226 Koen, M. S.,1992. Business Intellectual Property Prediction. MOSCI Corporation Rolla MO Koenker, R.,Bassett, G.,1978. Regression quantiles. Econometrica 46 (1), 33†50 Krammer, M. S. S.,2014. Assessing the relative importance of multiple channels for embodied and disembodied technological spillovers. Technol. Forecast. Soc Change 81, 272†286 Laursen, K.,Salter, A.,2014. The paradox of openness: appropriability, external search and collaboration. Res. Policy 43 (5), 867†878 Leiblein, M. J.,Madsen, T. L.,2009. Unbundling competitive heterogeneity: incentive structures and capability inï uences on technological innovation. Strateg Manag. J. 30, 711†735 Leone, M. I.,Reichstein, T.,2012. Licensing-in fosters rapid invention! The effect of the grant-back clause and technological unfamiliarity. Strateg. Manag. J. 33 (8 965†985 Li, H.,Tang, M.,2010. Vertical integration and innovative performance: the effects of external knowledge sourcing modes. Technovation 30, 401†410 Linton, J.,Walsh, S.,2013. The effect of technology on learning during the acquisition and development of competencies in technology-intensive small ï rms. Int. J. Entrep. Behav. Res. 19 (2)( 3 Mangematin, V.,Errabi, K.,Gauthier, C.,2011. Large players in the nanogame dedicated nanotech subsidiaries or distributed nanotech capabilities? J. Tech -nol. Transf. 36 (6), 640†664 Marion, T. J.,Friar, J. H.,Simpson, T. W.,2012. New product development practices and early-stage ï rms: two in depth case studies. J. Prod. Innov. Manag. 29 639†654 Mariotti, S.,Cainarca, G. C.,1986. The evolution of transaction governance in the textile clothing industry. J. Econ. Behav. Organ. 7, 351†374 Mas-Ruiz, F.,Ruiz-Moreno, F.,2011. Rivalry within strategic groups and conse -quences for performance: the ï rm-size effects. Strateg. Manag. J. 32 1286†1308 Mcgahan, A m.,Silverman, B. S.,2006. Proï ting from technological innovation by others: the effect of competitor patenting on ï rm value. Res. Policy 35 (8 1222†1242 O†Connor, G. C.,Demartino, R.,2006. Organizing for radical innovation: an explora -tory study of the structural aspects of RI management systems in large established ï rms. J. Prod. Innov. Manag. 23, 475†497 OECD, 1997. Small Business, Job creation and Growth: Facts, Obstacles and Best Practices. Available at www. oecd. org/cfe/smes/2090740. pdf Park, W. G.,2008. International patent protection: 1960†2005. Res. Policy 37 761†766 Powell, T. C.,1992. Strategic planning as competitive advantage. Strateg. Manag. J. 13 7), 551†558 Revilla, A j.,Fernandez, Z.,2012. The relation between ï rm size and R&d productivity in different technological regimes. Technovation 32, 609†623 Ritala, P.,Hurmelinna-Laukkanen, P.,2009. What's in it for me? Creating and appropriating value in innovation-related coopetition. Technovation 29 819†828 Robson, P. J. A.,Bennett, R. J.,2000. SME growth: the relationship with business advice and external collaboration. Small Bus. Econ. 15, 193†208 Rosenbaum, P. R.,Rubin, D. B.,1983. The central role of the propensity score in observational studies for causal effects. Biometrika 70 (1), 41†55 Rosenzweig, S.,Mazursky, D.,2013. Constraints of internally and externally derived knowledge and the innovativeness of technological output: the case of the United states. J. Prod. Innov. Manag. 31 (2), 231†246 Rothwell, R.,Zegveld, W.,1982. Innovation and the Small and Medium-Sized Firm Pinter, London Rubenstein, A. H.,Ettlie, J. E.,1983. Radical Technology, Organization Science Structure and Context in the Innovation Process, Applied science and Technol -ogy Association, Evanston IL Sakakibara, M.,2010. An empirical analysis of pricing in patent licensing contracts Ind. Corp. Change 19 (3), 927†945 Samila, S.,Sorenson, O.,2011. Venture capital, entrepreneurship, and economic growth. Rev. Econ. Stat. 93, 338†349 Shane, S.,2002. Selling university technology: patterns from MIT. Manag. Sci. 48 1), 122†138 Shefer, D.,Frenkel, A.,2005. R&d, ï rm size and innovation: an empirical analysis Technovation 25, 25†32 Smith, A.,1776 1983. The Wealth of Nations, Penguin Papers, Harmondsworth, UK Somaya, D.,Kim, Y.,Vonortas, N s.,2011. Exclusivity in licensing alliances: using hostages to support technology commercialization. Strateg. Manag. J. 32 159†186 Sørensen, J. B.,Stuart, T. E.,2000. Aging, obsolescence, and organizational innova -tion. Adm. Sci. Q. 45 (1), 81†112 Sreekumaran, S n.,Mathew, N.,Nag, D.,2011. Dynamics between patent latent variables and patent price. Technovation 31, 648†654 Stigler, G.,1951. The division of labour is limited by the extent of the market. J. Polit Econ. 59, 185†193 Stucki, T.,2014. Success of start-up ï rms. The role of ï nancial constraints. Ind. Corp Change 23 (1), 25†64 Taylor, A.,Helfat, C. E.,2009. Organizational linkages for surviving technological change: complementary assets, middle management, and ambidexterity Organ. Sci. 20 (4), 718†739 Teece, D. J.,1986. Proï ting from technological innovation: implications for integra -tion, collaboration, licensing and public policy. Res. Policy 15, 285†305 Trajtenberg, M.,1990. A penny for your quotes: patent citations and the value of innovations. RAND J. Econ. 21 (1), 172†187 Trajtenberg, M.,Henderson, R.,Jaffe, A.,1997. University vs. corporate patents: a window on the basicness of invention. Econ. Innov. New Technol. 5 (1), 19†50 Tripsas, M.,1997. Unraveling the process of creative destruction: complementary assets and incumbent survival in the typesetter industry. Strateg. Manag. J. 18 S119†S142 G. Padula et al.//Technovation 41-42 (2015) 38†50 49 Valentini, G.,2012. Measuring the effect of M&a on patenting quantity and quality Strateg. Manag. J. 33, 336†346 Veer, T, . Jell, F.,2012. Contributing to markets for technology? A comparison of patent ï ling motives of individual inventors, small companies and universities Technovation 32, 513†522 Wang, Y.,Vanhaverbeke, W.,Roijakkers, N.,2012. Exploring the impact of open innovation on national systems of innovation†a theoretical analysis. Technol Forecast. Soc. Change 79, 419†428 Winter, S g.,1987. Knowledge and competence as strategic assets. In: Teece, D. J Ed.),The Competitive Challenge: Strategies for Industrial Innovation and Renewal. Ballinger, Cambridge, MA, pp. 159†184 WIPO, 2014. âoe http://www. wipo. int/export/sites/www/freepublications/en/licen sing/903/wipo pub 903. pdfâoe Wooldridge, J. M.,2002. Econometric Analysis of Cross-section and Panel Data. The MIT Press, Cambridge, Massachussets Yadav, M. S.,Prabhu, J. C.,Chandy, R. K.,2007. Managing the future: CEO attention and innovation outcomes. J. Mark. 71 (4), 84†101 Yelle, L. E.,1979. The learning curve. Historical review and comprehensive survey Decis. Sci. 10, 302†328 Young, A.,1928. Increasing returns and economic progress. Econ. J. 38, 527†542 G. Padula et al.//Technovation 41-42 (2015) 38†5050 SMES inventive performance and profitability in the markets for technology Introduction Technology specialists vs. vertically-integrated SMES: implications on inventive performance and profitability Method Sample and data Variables Dependent variables Independent variable Control variables Results Discussion Implications to practice Implications to theory Limitations Conclusion Acknowledgements References

< Back - Next >

Overtext Web Module V3.0 Alpha
Copyright Semantic-Knowledge, 1994-2011