Venture Capital investments and patenting activity of hightech start-ups: a micro-econometric firm-level analysis Fabio Bertoni, Annalisa Croce, Diego D’Adda** Department of Management, Economics and Industrial Engineering Politecnico di Milano Abstract The aim of this paper is to analyse empirically the impact of Venture Capital (VC) finance on the innovation output of new technology-based firms (NTBFs) as is reflected by their patenting activity. In particular, we compare the patenting rates of VC-backed and non VC-backed NTBFs. To investigate whether VC investments spur patenting activity, we consider a unique longitudinal dataset composed of 351 Italian NTBFs operating in high-tech manufacturing industries and software, 33 of which are VC-backed. We estimate different econometric models on panel data, controlling for factors that may affect firm’s patenting behaviour other than the presence of VC, like founders’ human capital and use of other sources of financing. The results show that VC investments positively affect subsequent patenting activity and that before receiving VC, VC-backed firms do not exhibit a higher patenting propensity than other firms.
JEL codes: D92, G24, L21 Keywords: Venture capital, Patents, New technology-based firms __________________________________________________ The financial support provided by the VICO FP7 project is gratefully acknowledged. We are grateful to Massimo G. Colombo, Mike Wright, Raffaele Oriani, Henri-François Boedt, Stefano Brusoni and participants in the XVII AiIG Annual Conference, the 12th ISS Conference, the XXII RENT Conference, the DRUID Winter Conference 2009 for helpful comments on this and related works. The usual disclaimer applies. ** Corresponding author: Diego D’Adda, Politecnico di Milano, Department of Management, Economics and Industrial Engineering, Piazza Leonardo da Vinci, 32, 20133, Milan, Italy. ph: +3902-2399-3974; fax: +39-02-2399-2710. E-mail address:
[email protected].
1. Introduction Venture capital (VC) is generally considered by both scholars and practitioners as the most suitable financing mode for young high-tech start-ups (new technology-based firms, NTBF). In fact, it is contended in the financial literature that this financing mode offers a fundamental contribution to the success of NTBFs (see, for instance, Sahlman 1990, Gompers and Lerner 2001, Kaplan and Strömberg 2001, Denis 2004). In this paper we focus attention on the innovation output of NTBFs, as is reflected by their propensity to patent, and the extent to which VC has a positive (or negative) impact on it. Whether access to VC finance fosters the innovative activity of portfolio companies is a matter of empirical test. Previous evidence, deriving mainly from industry level analysis (Kortum and Lerner 2000, Tykvova 2000, Ueda and Hirukawa 2008a), highlights a positive relation between VC finance and patenting activity1. Fewer studies analyze the effect at firm-level and find mixed evidence (Engel and Keilbach 2002, Baum and Silverman 2004). In this work we resort to a hand-collected 10 year long longitudinal dataset composed of 351 Italian NTBFs to analyze the effect of VC investments on firm’s patenting in the years that follow the first round of VC. Sample firms were established in 1980 or later, were owner-managed at foundation, survived as independent firms up to January 2004, and are observed from 1994 up to 2003. They operate in high-tech manufacturing sectors and in software. Most conditions which are typically found to favour the development of the VC sector in a country, like the presence of liquid stock markets, large pension funds sector, and flexible labor markets (see e.g. Jeng and Wells 2000, Da Rin et al. 2006, Black and Gilson, 1996) were lacking in Italy during the observation period of this study.2 As a consequence VC was less developed in Italy than in countries in which empirical studies are often conducted like the US, the UK or Israel. In such an underdeveloped VC market the estimate of the effect engendered by the presence of a VC investor on firm’s innovation activity presents some peculiarities
1
A related stream of the literature deals with the impact on patenting and innovation of later stage investments like Leveraged Buyouts and Private Equity (see for instance the recent work by Lerner, 2008) 2 Another factor which is often associated to a flourishing VC market is favourable taxation for equity investments. The capital gain tax rate was indeed very low or even, under some conditions, null in Italy during this period. However, there was no reduced levy for early-stage or high-tech investments and this, actually, relatively favoured late-stage investments in mature industries, crowding out early-stage high-tech investments. 2
compared to the one which have been highlighted elsewhere. On the one hand, the limited number of investors suggests that they may be in a better position to “cherry pick” the best firms because of lack of competition. On the other hand, a “self-selection out” problem (Bertoni et al. 2010) may arise: firms with the best future performance might prefer to abstain from looking for VC investments due to the high opportunity cost of this search. The presence of these two phenomena might engender an endogeneity problem, leading to a bias in the estimated effect of VC financing on NTBFs performance. The direction of this (potential) bias is not clear a priori because these sorting effects push in opposite directions. The results of our estimates support the view that VC investments have a positive effect on firm’s patenting activity over and beyond the (observable) effect attributable to sorting. Indeed, we find no evidence of positive (or negative) sorting: before receiving VC, VC-backed firms do not exhibit a different patenting propensity than other firms, in line with the findings of previous work on VC investments in Europe (Engel 2002, Bottazzi et al. 2008, Colombo and Grilli 2009). The paper is structured as follows. In the next section we survey the literature on the effects of VC investments on firm’s innovation. In Section 3 we describe the sample of firms that are considered in the empirical analysis. In section 4 we illustrate the empirical methodology. Section 5 reports the results. Robustness tests are described in Section 6.Finally, Section 6 concludes.
2. Related literature 2.1. The contribution of VC investment to the innovation output of portfolio firms The financial literature highlights several motives explaining why access to VC finance stimulates the innovation performance of portfolio NTBFs. First of all, NTBFs are the firms which are most likely to be financially constrained (see e.g. Carpenter and Petersen 2002, Colombo and Grilli 2007). Owing to their superior screening capabilities (Chan 1983, Amit et al. 1998), VC investors can identify firms with great innovative projects, the quality of which remains hidden to other investors, and provide them with the financing necessary to support R&D. Higher R&D investments, made possible by the presence of VC, in turn boost firm’s innovation output. Moreover, patents, that have been often used as proxy of firm’s innovation activity (as we do in this work), have a direct cost to the firm, that is likely to be sizeable for (small) NTBFs. They may also entail nontrivial indirect costs (Hsu and Ziedonis 2007) like public disclosure of the invention, training of technical personnel about patenting rules and procedures, and the opportunity cost of time devoted to the patenting process (e.g. the communication between inventors and attorneys/patent agents). Cohen et al. (2000) find that the 3
cost of applying for a patent and the cost of defending a patent in a court are important reasons that discourage smaller firms from patenting (see also Lerner 1994). VC finance allows portfolio firms to meet these costs, causing an increase in their patenting rate over and above the one generated by the increase of their R&D expenditures. Second, VC investors are no silent partners (Gorman and Sahlman 1989, Barry et al. 1990). On the one hand, they actively monitor the behaviour of entrepreneurs of portfolio companies (Lerner 1995, Kaplan e Strömberg 2003). On the other hand, they make use of specific financial instruments and contractual clauses (e.g. stage financing) that protect their investments from opportunistic behaviour on the side of entrepreneurs, and create high-powered incentives for them (Sahlman 1990, Gompers 1995, Hellmann 1998, Kaplan and Strömberg 2003, 2004). This tighter discipline results in greater innovation productivity and greater innovation outcome. Third, VC investors perform a key coaching function to the benefit of portfolio firms (Gorman and Sahlman 1989, MacMillan et al. 1989, Bygrave and Timmons 1992, Sapienza 1992, Barney and Busenitz 1996, Sapienza et al. 1996, Kaplan and Strömberg 2004). In fact, they provide these firms with valuable advising services in fields such as strategic planning, marketing, finance and budgeting, and human resource management, in which these firms typically lack internal capabilities. Moreover, portfolio firms can take advantage of the network of social contacts of their VC investors with potential customers, suppliers, and alliance partners (Colombo et al. 2006, Hsu 2006,Lindsey 2008). They also find it easier to get access to external resources and competencies that would be out of reach without the endorsement of the VC investor because of the quality certification effect of VC investments (Stuart et al. 1999). From the above it follows that the resource and competence endowment on which VC-backed firms can rely substantially exceeds the one of their non-VC-backed counterparts. Hence the expected returns to R&D investments are greater for the former type of firms. This again leads to greater R&D investments and innovation output. Nonetheless, it has also been argued by previous studies that the agency relation between VC investors and entrepreneurs may engender conflicts (Ueda 2004, Atanasov et al. 2006, Masulis and Nahata 2009). A negative effect on the innovation activity of portfolio firms may ensue. First, VC investors may have objectives and strategies that are different from those of entrepreneurs. Disagreements may absorb the entrepreneurs’ effort and attention to the detriment of the pursuit of innovative projects (Dushnitsky and Lenox 2006). Second, VC finance might pose appropriability hazards for portfolio firms because VC investors might poach the innovative business ideas of entrepreneurs and exploit them in their absence (Ueda 2004).The associated appropriability hazards may induce entrepreneurs to take decisions that are detrimental to firm’s innovation performance. For instance, they 4
may concentrate investments in the protection of firm’s existing technologies rather than in the development of new technologies. 2.2. The effect of VC financing on the innovation of portfolio companies: survey of empirical literature The work by Kortum and Lerner (2000) is probably the first large-scale study dealing with the impact of VC on innovation. They analyze patents data and VC finance across twenty manufacturing industries between 1965 and 1992 in the USA. They use a patent production function at industry level, controlling for the arrival of technological opportunities that could lead to a spurious relation between VC and patenting activity. 3 They document that VC does indeed spur patenting activity. Its effect is found to be even more positive than the one of traditional corporate R&D. They also conduct a firm level analysis on 122 VC-backed and 408 non VC-backed companies to check whether the effect of VC derives only from a change in firm’s patent propensity (i.e. using patents more to protect their proprietary technology instead of relying on different appropriability mechanisms). The results suggest that VC-backed companies i) have patents that are more frequently cited, and so supposedly are of better quality, and ii) engage more in trade secret litigations than non VC-backed firms (i.e. they don’t make less use of other appropriability mechanisms like trade secrets). In a similar study, Tykvova (2000) using again a patent production function, finds that VC investments have a highly significant positive effect on patenting activity in a sample of ten German industries observed between 1991 and 1997. The estimated effect of VC is lower than in Kortum and Lerner (2000). However, the extent of this effect might be underestimated due to the inclusion in the analysis of private equity investments in addition to classical VC investments. Ueda and Hirukawa (2008a) exploit the same methodological framework, extending data used by Kortum and Lerner (2000) up to 2001 to include also the NASDAQ bubble period. Using patents as the dependent variable they confirm Kortum and Lerner’s (2000) results at industry level. However they argue that these results could be explained by a different patent propensity between VC-backed and non-VC-backed firms, and that the use of total factor productivity (TFP) growth, as a measure of innovative activity alternative to patents, may rule out this problem. They do not find support for the argument that VC investments have a positive impact on TFP growth. In a related work, Ueda and Hirukawa (2008b) address the causality concern be-
3
They use i) R&D expenditures to control for the arrival of technological opportunities that could be anticipated only by economic actors at that time and ii) a policy shift, that was unlikely to be related to the arrival of opportunities, in an instrumental-variable regression. 5
tween VC investments and innovation: they investigate whether it is VC that spurs innovation or if it is the arrival of a new technology that increases demands for VC by driving creation of new firms. Using TFP as measure of innovation they find support for both of these hypotheses in computer and communication industries. On the contrary they find a negative relation between VC and TFP growth in drugs and scientific instruments. Unlike the abovementioned studies that are conducted mainly at industry level, other scholars move to the firm level for a more in-depth understanding of the relationship between VC and innovation activity and its causality direction. In particular, one should acknowledge that patents can be considered as a signal (Spence 1973) of firm’s quality, since i) the examination process provides a certification function for the underlying invention, and ii) they are costly to obtain. This may have controversial effects on the likelihood of a firm obtaining VC. On the one hand, VC investors may be attracted by firms with superior quality certified by the patents they have been granted. Moreover, patents may have a signalling function to outside investors other than VC investors, thus favouring the exit strategy of these latter through a trade sale or a IPO. On the other hand, to the extent that this signal alleviates adverse selection problems, it may destroy the very source of the competitive advantage in screening enjoyed by VC investors, with an opposite effect on the likelihood of the focal firm obtaining VC. Whatever the net effect of these opposed forces, the above reasoning argues in favour of the need for adequate controls for the endogeneity of VC investments in assessing their innovation impact at firm level. 4 Engel and Keilbach (2002) consider a sample of 142 German VC-backed companies that were established between 1995 and 1998 and rely on matched pair techniques to study the effects of VC on innovation output. Focusing on the very early stage of firms’ life, they show through probit estimation that firms’ prefoundation patenting behaviour and human capital characteristics do affect the probability of VC involvement. They then resort to the propensity score method to build their control sample. The average number of patents in the VC-backed group is (weakly) significantly higher than in the control group. Baum and Silverman
4
Hsu and Ziedonis (2007) provide evidence that patents do signal the unobserved economic value of start-up firms to external investors. In fact, while examining 370 US semiconductor start-ups receiving VC, they find that pre-money valuation is positively affected by the stock of patent applications held by the firm. This relation is particularly significant in early funding rounds and when funds are provided by prominent VC investors. Moreover, they find that patent applications stock positively affects the probability of an investor exit through IPO. Similarly, Munari and Toschi (2007) using a sample of 332 VC-backed companies in the nanotechnology sector, document that the patent portfolio in nanotech patent classes is positively associated with the amount of VC finance. 6
(2004) directly address the causality issue between VC investments and firms’ patenting activity. Their sample is composed of 204 biotechnology start-ups located in Canada. Relying on time series regression techniques, they find that the amount of pre-IPO financing is positively affected by patent applications and patents granted in the year before the receipt of VC finance. Furthermore they do not find support for the hypothesis that patenting activity, measured both by number of patent applications and patents granted, is positively related to the amount of VC financing in the previous year. Other studies have relied on innovation indicators other than patents. Hellmann and Puri (2000) analyze the impact of VC investments on product market behaviour in a sample of 173 start-up companies that are located in Silicon Valley. They find that pursuing an innovator strategy 5 positively affects the probability of obtaining VC. Moreover VC finance reduces time to market especially for innovator firms. Da Rin and Penas (2007) examine the innovation strategies of a sample of Dutch firms, out of which 91 are VC-backed. They contrast a “make” innovation strategy, based on internal R&D investments, with a “buy” strategy that relies on R&D outsourcing, and compare the effects of VC finance and public subsidies on firms’ choice of R&D strategy. Interestingly, they find that public financing merely relaxes firms’ financial constraints without shifting the balance between influencing the choice between “make” and “buy” R&D strategies. Conversely, obtaining VC shapes the subsequent innovation strategy of portfolio firms in favour of a “make” strategy, causing a permanent shift toward in-house R&D. Chemmanour et al. (2008) analyse the causality direction between VC investments and TFP growth in a large sample of US private firms. They find that VCbacked firms exhibit higher TFP growth than non-VC-backed firms prior to receiving VC. Moreover VC-backed firms’ TFP growth is greater even after receiving VC. They also try to disentangle the relative size of the “sorting” and “treatment” effects using endogenous switching regression and matching techniques. They document that VC financing boosts the TFP of sample firms, suggesting that venture capitalists provide significant value to portfolio firms. To sum up, the above mentioned studies are not unanimous as to the positive effect of VC investments on firm’s innovation output and to the causality direction between the two. In this study we empirically test whether VC investments foster patenting activity of Italian NTBFs controlling for firm’s quality (proxied by the human capital of founders and firm’s patent track record) and for the effects of
5
They measure innovator ex-ante strategy through interviews, trying to capture if firms are introducing a new product or service that i) is not a close substitute of a product/service already offered on the market, ii) is expected to outperform product/services already offered in the market, or iii) satisfies either of the two criteria above. 7
capital injections from other sources (i.e. cash flows, increase in debt, public subsidies).
3. The sample In this paper we use a unique, hand collected longitudinal dataset relating to a sample composed of 351 Italian NTBFs that are observed over the ten year period between 1994 and 2003. Most sample firms are privately held during the observation period (98%). They were established in 1980 or later, were owner-managed at founding time and have survived as independent firms up to the end of 2003. They operate in high-tech manufacturing sectors and software. 6 The sample of NTBFs was extracted from the 2004 release of the RITA (Research on Entrepreneurship in Advanced Technologies) directory, developed at Politecnico di Milano. The RITA directory presently is the most complete source of information on Italian NTBFs and it has been used in several previous studies (e.g. Colombo et al. 2004, Colombo et al. 2006, Colombo and Grilli, 2005, 2009, Bertoni et al. 2009)7. Assessing the representativeness of our sample with the population of Italian NTBFs using official statistics is not possible for several reasons. First, the notion of representativeness requires a precise definition of the unit of analysis which is, in this domain, quite slippery (see e.g. Aldrich et al. 1989). Second, national official statistics do not provide themselves a reliable description of the population of Italian NTBFs. On the one hand, in Italy, most individuals who are defined as selfemployed by official statistics (i.e. “independent employees”) actually are salaried workers with atypical employment contracts. Unfortunately, on the basis of official data such individuals cannot be distinguished from owner-managers of a new firm. This means that the official number of NTBFs in Italy is enormously inflated, especially in sectors like software where atypical employment contracts are very common. In addition, official data do not distinguish firms that were established by one or more entrepreneurs (i.e. owner-managed firms) from firms that were created as subsidiaries of other firms. This again inflates the number of NTBFs. Lastly, there are no official statistics about M&As: therefore one cannot distinguish firms that were acquired (and thus lost independence) while keeping their legal status,
6
More precisely, the following sectors are included: computers, electronic components, telecommunication equipment, optical, medical and electronic instruments, biotechnology, pharmaceuticals, advanced materials, robotics, process automation equipment, and software. 7 The RITA directory also includes firms in high-tech services rather than manufacturing. Since the dependent variable in this study is based on patents, we excluded (with the exception of software) service companies. Patents are rarely used by these firms and so cannot be regarded as a reliable indicator of their innovative output. 8
from independent NTBFs. While the RITA directory obviously does not offer a complete coverage of the population of Italian NTBFs, it is quite unlikely that it excludes potential candidates for VC investment. The RITA directory was created in 2000 and it was updated in 2002 and 2004 8. Altogether, the 2004 release of the RITA directory comprises 1,974 firms that complied with the abovementioned criteria relating to industry of operations, age and independence. Data, collected by survey, concern to the human capital characteristics of firm’s founders, the characteristics of the firms including access to VC finance, the identity of VC investors, receipt of public subsidies, and the evolution over time of firm’s employees. The term “VC” is used in this study in a quite broad tense. We include among VC investments all equity (or equity-like) financing provided by external investors to high-tech entrepreneurial firms in the early stages of their life (seed, start-up and expansion capital). The source of financing may be independent financial intermediaries, diversified financial firms, and non-financial firms (i.e. corporate venture capital). Data on VC finance was cross-checked with those available from public sources and commercial databases. In addition, financial and economic data including the evolution over time of firm’s sales from 1994 onwards, and data on patent activity during firm’s entire life were obtained from public sources (i.e. the AIDA and CERVED databases and the databases of patent offices accessed through the
[email protected] search engine, respectively). Due to the procedure that was used to create the RITA directory, our dataset has several strengths with respect to those used in previous studies, as well as a few weaknesses. First, to the best of our knowledge this is the first study that uses a long longitudinal firm-level dataset for both VC-backed and non-VC-backed privately held firms to detect the effects of VC investments on firm innovation output.
8
For its construction several sources were used. These included: i) the lists of the companies that are members of the national entrepreneurial associations of the focal industries; ii) the lists of the members of the regional sections of the Italian entrepreneurial association (Confindustria); iii) the lists of the members of the local Chambers of Commerce; iv) the lists of companies that participated in the most important industry trades and expositions; and v) the lists of companies that purchased advertising services in popular off-line (e.g. Kompass) and on-line (e.g. Infoimprese.it) directories. Moreover, the RITA directory includes: vi) the population of young firms that were granted by the Italian communication authority (AGCOM) a license to provide telecommunication services (including Internet access services), vii) the population of NTBFs that were incubated in a science park or a business innovation center (BIC) affiliated with the respective national associations, viii) all NTBFs which VC investors that adhere to the Italian financial investor association (AIFI) disclosed to have invested in. Lastly, information provided by the national financial press, specialized magazines, and other sectoral studies was also used in the compilation of the directory. 9
In addition, as a consequence of the process used to build the sample, the dataset includes NTBFs that are typical targets of VC investors. Conversely, lifestyle firms and other non-growth-oriented firms that would be very unlikely to be selected by VC investors are excluded from the data set. This yields more precise estimates of the relevant counterfactual (i.e. the innovation output VC-backed firms would experience if they were not VC-backed) than would be possible if we considered only VC-backed firms, or if low-tech firms and high-tech lifestyle companies, which clearly have different finance needs, were included in the sample. Second, our coverage of VC investments is more comprehensive than what found in most other studies on the Italian market, which only rely on public sources and commercial databases. Taking into consideration the sample of 351 RITA firms used in this study, a χ 2 tests show that there is no statistically significant difference between the distribution of these firms across industries and the corresponding distribution of the population of 1,275 RITA firms from which the sample was drawn (χ2(3)=3.38). Regarding the weaknesses of the sample used in this work, the most serious problem, which is common to survey-based studies (for exceptions see e.g. Delmar and Shane 2006, Eckhardt et al. 2006), is survivorship bias: only firms that survived up to the survey date could be included in the sample. In principle, attrition might generate a sample selection bias in our estimates. For instance, NTBF failure rates are likely to decrease with access to VC finance because VC-backed firms allegedly benefit from greater endowment of financial and other resources (Puri and Zarutskie 2008). If there were a negative correlation between patenting and exit, the impact of the VC variables on firm patenting activity might actually be higher than that highlighted by our empirical analysis. 9 In fact, we were not able to rigorously control for this selection bias. The best we can do is to implement a partial control that is described in Section 6. Table 1 shows the sectoral distribution of the sample of VC-backed firms and of firms patenting at least once during the observation period. Out of the 351 firms that are considered in this work, 33(16.8%) obtained VC finance and 30 (15.2%) were granted a patent. There are considerable differences across industries as to both the share of VC-firms out of the total number of firms and the share of patenting firms. Higher patenting in an industry is always associated with a greater share of VC-backed firms. Firms in biotechnology, pharmaceutics and advanced materials are both the most likely to patent (27.3%) and to obtain VC finance (18.2%),
9
However, the positive impact of VC finance on survival should not be taken for granted. VC-backed firms might be more risk-prone than their non-VC-backed peers and actually have a lower likelihood of survival (Manigart and Van Hyfte 1999). Moreover, VC-backed firms are more likely to be acquired, and so to exit from the sample, as trade sales as an important exit mechanism for VC investors. 10
while the opposite happens for firms in software (3.8% and 7.5%, respectively). More generally the two rankings of industries according to propensity to patent or likelihood of obtaining VC money coincide. Insert Table 1 around here As an additional preliminary evidence of the relationship between VC and patenting it is interesting to check whether there is any noticeable difference in the likelihood to patent between VC-backed and non-VC-backed firms. Table 2 shows that out of the 318 non-VC-backed firms, 292 (91.8%) did not obtain any patent in the period under analysis, while the same happens, out of the 33 VC-backed firms, only for 23 companies (69.7%). In other words 10 out of 33 (30.3%) VC-backed firms hold at least one patent against 26 out of 318 (8.2%) non-VC-backed firms. The null hypothesis that the propensity to patent does not differ between VCbacked and non-VC-backed firms is rejected at conventional confidence levels (2(1)=15.90). Insert Table 2 around here Moreover, as is apparent from Table 3,among the 10 VC-backed patenting firms only two applied for a patent before obtaining the first round of VC. Even more striking is the figure related to the granting date: all the patents of VC-backed firms were granted after the entry of the VC investor. These figures seem to suggest that, among VC-backed firms, an increase in patenting activity occurs after the entry of the VC investor. Insert Table 3 around here Note however that this statistics do not take into account firm-specific characteristics (i.e. individual heterogeneity) that are likely to affect both the likelihood to obtain VC and patenting activity. Thus a more systematic analysis is needed in order to assess the allegedly positive effect of VC finance on patenting.
4. Econometric methodology 4.1. Model specification In this work our aim is to study the relationship between innovation output (measured by patenting) and VC financing. For this purpose, we estimate two sets of econometric models for panel-data. Table 4 shows the dependent and explanatory variables included in the models. 11
Insert Table 4 around here The first two variables shown in Table 4 (Npatents(t), Dpatents(t)) are the dependent variables of the econometric models. Dpatents(t) is a dummy variable equal to 1 if a firm applies for at least one patent in year t (and the patent is subsequently granted). Npatents(t) measures the number of applications of the firm in year t for patents that are later granted by the patent office. First, we estimate the conditional distribution of Dpatents(t) using a random effects Logit model. Second, we estimate a random effects Negative Binomial model whose dependent variable is Npatents(t). To test the effect of VC on patenting activity we include a dummy variable, VCbacked(t-1), which is equal to 1 if the focal firm received VC financing in year t-1 or before (that is, the firms is or has ever been VC-backed).10 We also include a set of variables aimed at controlling for firm’s unobserved heterogeneity in the ability to patent. Following previous studies (Blundell et. al. 1995, Ahuja and Katila 2001, Dushnitsky and Lenox 2005),we use a variable that measures firm’s prior patent stock, Patent stock(t-1). Other variables measure human capital characteristics of firms’ founders; in particular we include indicators of the work experience and of the educational background of the founding team. Human capital is likely to deeply affect both the probability of obtaining VC finance and the probability to apply for a patent; therefore its inclusion is crucial in order to rule out spurious relations between VC finance and patenting activity. We also check for other sources of financing to control for a potential “capital injection” effect of VC. In fact patents are costly, especially for young small firms like the NTBFs under examination here. It may be the case that the mere injection of capital explains the patent application, thus driving the relation between VC and patenting activity. In this case the effects of other funding sources should be similar to the one of VC. For this reason we include measures of the increase in total assets (ΔTotalAssets(t-1)), increase in debt (ΔDebt(t-1)), the amount of cash flow (CashFlow(t-1)), the participation in a EU funded research project (DEUProject(t-1)), and the receipt of public subsidies (DPublicSubsidies(t-1)). Finally, control variables include firm size and age alongside firm-specific effects (capturing timeinvariant individual heterogeneity) and year dummies (capturing time-variant common effects).
10
Note that VC-backed(t-1) does not switch back to 0 when VC investors exit. The reason is that firms that obtained VC are inherently different from those that never did. For instance, VC finance is assumed to signal the quality of a firm to uninformed third parties, making it easier for the firm to obtain access to additional resources. This effect is likely to persist even after the exit of the VC investor. 12
4.2. Estimation methodology As mentioned before, we resort to random effects panel data models in our econometric analysis. In a nonlinear context, like in binary choice and count models, the direct estimation of the unobserved individual effects introduces an incidental parameters problem, with estimates of both coefficient and individual effects being inconsistent (see Lancaster, 2000). Moreover, the use of fixed effect estimators is highly inefficient on our sample because it would reduce the sample to the set of firms that “switch” (patent) at least once in the observation period, both for binary and count models, which, as was shown before, represents a minority (15.2%) of our sample. Moreover fixed effects models are more subject to measurement error which, in the case of patenting, is a serious concern. Therefore we opt for using random effects models, that hinge on the assumption of strict exogeneity of the regressors. In other word, we require that only contemporaneous regressors matter once the unobserved individual effect is also conditioned on. This assumption holds if we rule out the presence of correlation between individual effects and the regressors, and reverse causality, i.e. feedback from lagged responses to current regressors. The correlation between the individual effect and the regressors would make the estimates inconsistent. To overcome this drawback, at least partially, one possible solution is to add more structure and specify a conditional distribution for the individual effects. Some restrictions on the relationship between individual effects and regressors are thus imposed. Following Blundell et al. (1995) we use the patent application stock, comprehensive of the pre-sample period, as a proxy of unobserved heterogeneity, i.e. firm’s ability to patent.
5. Results Table 5 shows the results of the econometric estimates. In each model the estimated coefficient of VC-backed(t-1) is positive and statistically significant. When a NTBF receives VC, its patenting activity (measured alternatively by the likelihood of obtaining one or more patents, DPatents(t), or the number of granted patents, NPatents(t)) increases significantly, even when controlling for time-invariant heterogeneity and common exogenous shocks. Moreover, when other controls are added in the model specification, the coefficient of VC-backed decreases only slightly (in the Logit estimates) or increases (in the Negative Binomial estimates). Let us briefly consider the effects on patenting of the control variables. Firm size has a positive significant effect on patenting while the age coefficient is never significant. As Cohen et al. (2000) pointed out the bigger the firm the higher the probability of using patents, in accordance with the costs of application and enforcement which are sizeable for a small firm. The variable measuring the patent stock is positive and (weakly) significant only in model (3) with full specification; 13
the limited patent track record of the NTBFs in our sample may be the reason. Among human capital characteristics, the most notable effect is the one related to the technological education of the founders; its coefficient is positive and significant (at 5% or lower) in all the models. The technological work experience of founders also is positive and significant in models (3) and (6). Conversely, their commercial work experience seems to influence only the likelihood of applying for a patent, but not the number of patents granted. Finally, in models (3) and (6) we introduce accounting variables to control for other sources of capital injection. The ratio of cash flows on total assets has a positive and significant effect on both the likelihood of obtaining one or more patents and the number of patents granted. A similar, although less robust, effect is detected for the increase of debts over total assets. This suggests that debt and cash flows are two ways through which NTBFs finance patent application filing. On the contrary public subsidies, be they local, national or European, does not influence patenting activity. Insert Table 5 around here To assess the economic magnitude of the effect of VC investments on patenting activity, we used models (3) and (6) to predict the probability of obtaining one or more patents and the expected number of granted patents, contingent on whether a firms is VC-backed or not. The figures in Table 6 document that, for the “median” NTBF, being VC-backed increases the likelihood of patenting almost threefold and the expected number of patents almost fivefold. This last result is extremely in line with what Kortum and Lerner (2000) find for US in the period 1990-1994, where the average number of patent for a VC-backed firm is 12.74 and for a non VCbacked is 2.40 (therefore the average VC-backed firm is responsible of 5.3 times the number of patents with respect to the average non VC-backed). Insert Table 6 around here
6. Robustness tests We perform a number of robustness checks relating to the effect of VC on patenting activity. The results of these additional estimates are not reported in the text for the sake of concision but are available from the authors upon request. As we mentioned before, our estimates are possibly affected by a survivorship bias, therefore we implement a direct test to check its presence in the spirit of Wooldridge (2995) and Semykina and Wooldridge (2006). We initially focused attention on the RITA 2000 sample. This sample, composed of 401 firms, was selected according to the same criteria and strategy used for the sample examined in the present study (Colombo et al. 2004). Some 31 sample firms were VC-backed at the beginning of 2000. We examined the exit rate of sample firms in the period 2000–2003. Twelve 14
VC-backed firms either ceased operations or were acquired (38.7%). Some 89 nonVC-backed firms exited the sample in this period (24.1%). Then we estimated a probit model of firm exit in 2000–2003 conditional on survival up to the end of 1999; the dependent variable in this model is the (probability of) survival of sample firms in 2000–2003. The independent variables include human capital variables, receipt of VC investments before 2000, firm-specific characteristics (e.g. firm size and age in 1999), and other controls. We used the estimated coefficients for this sample selection model to compute the inverse Mill’s ratio of survival for each firm-year observation included in the sample analysed in the present work. This time-varying ratio was then inserted as a control for survivorship bias in a pooled probit model11 with the same specification as model (3) and robust standard errors clustered at the firm level. The non-significance of the correction term suggests that survivorship bias is not affecting the results presented earlier. Second, the Blundell et al. (1995) approach has some drawbacks when, like in our sample, the pre-sample period and the patenting history of companies are limited. Therefore, we also estimate, obtaining results consistent with the ones reported in section 5, models (3) and (6) including firm-specific averages of timevariant explanatory variables in Mundlak’s style (see Mundlak 1978). The underlying assumption is that unobserved individual effects will depend only upon the average of the other observable time-variant regressors. Finally, we address reverse causality by taking a second look at the statistics illustrated in Table 3. The main reason why the association between VC and patenting could be flawed by reverse causality is that firms might want to protect their knowledge from the threat of expropriation by VCs (Ueda, 2004). As a consequence, even if their innovative activity is unchanged we would observe an increase in patenting. Econometrically, patenting, as a measure of innovative activity, suffers from a measurement error which is correlated to one of the covariates (i.e. VC). However, if this is true, we should also observe a significant patenting activity right before (close to) the first investment by VC. VC investments are not exogenous and unpredictable events for entrepreneurs, hence we would expect them to protect their intangible capital (knowledge) by depositing patents before starting the negotiation with VCs. However we observe little evidence of an increase in patenting due to the attempt to limit expropriation by VC and only 2 patent applications are filed before the arrival of VC. Reverse causality seems thus not to be a major concern in this study.
11
We use a probit assuming bivariate normality of error terms, adapting Vella (1992) “two-step” test of selection bias. 15
7. Conclusions This work contributes to the limited micro-econometric literature that has analyzed the firm-level effect of VC investments on innovation. In particular, we have examined the effect of VC investments on firms’ patenting activity using firm-level longitudinal data on a sample of 351 Italian NTBFs that operate in high-tech manufacturing and software. Our results can be summarised as follows. First, we find evidence that firm’s propensity to patent is significantly higher for VC-backed firms than for their non-VC-backed counterparts. Moreover both the likelihood of obtaining one or more patents and the number of granted patents increase significantly after a firm receives the first round of VC. These effects are both statistically significant and of significant economic magnitude. Moreover, VC-backed firms do not exhibit such a high patenting propensity before receiving VC. So, our results support the view that VC is beneficial to firm’s innovation. There are two possible, non-competing arguments to explain the positive effect of VC investment on firm’s patenting activity. First, it has been shown elsewhere (Bertoni et al. 2009) that VC finance results in the relaxation of NTBFs’ financial constraints. As more financial resources are available to VC-backed firms than to their non-VC-backed counterparts, the former firms may increase their R&D investments after receiving VC, which in turn leads to greater patenting activity. Moreover, previous studies (see e.g. Cohen et al. 2000) have documented that patent application and enforcement is costly. This reduces the propensity to patent of financially-constrained non-VC-backed NTBFs, independent of R&D expenditures. Nonetheless, our estimates suggest that the positive effect of VC finance on patenting persists after controlling for changes in the cash flows of firms and in their ability to obtain debt finance. Therefore, there is more to VC investments than simply the provision of additional financial resources to financially constrained firms. Unfortunately, while our estimates support the view that VC investments add value to portfolio firms, resulting in superior innovation outcome, they do not tell us why. For instance, VC-backed firms may be forced to streamline their R&D operations and abandon “pet” research projects, because of the tighter financial discipline engendered by the presence of the VC investor in firm’s equity capital. This leads to superior R&D productivity. Alternatively, the VC investor may act as a “coach” to the advantage of portfolio firms. Sponsorship by the VC investor may also make it easier for them to find alliance partners (Colombo et al. 2006, Hsu, 2006, Lindsey 2008) . In turn, this increases the expected marginal returns to R&D expenditures, leading to greater innovation input and output. Disentangling the relative importance of these different channels through which VC investments may contribute to firms’ innovation is an interesting avenue for future research. Whatever the source of the beneficial effect engendered by VC investments, the evidence that they have a dramatic positive influence on NTBFs’ patent activity, 16
has important policy implications. In Europe the VC sector is far less developed than in the USA, the UK, or Israel, a situation that is well represented by Italy. Although an analysis of the determinants of this situation is beyond the scope of the present study, the findings presented here indicate that the development of VC financing should figure prominently in the innovation policy agenda of European governments. In spite of the interest of this study, several directions for future research seem very promising. First, we aim to enhance our dataset including data from the new release of RITA which will be released in late 2009. By doing so we will overcome a relevant limitation of the present study, which is the small number of VC-backed firms and VC-backed patenting firms. Second, the question arises as to whether the results obtained can be generalised to countries other than Italy. Third, it could be interesting to analyse whether the extent of the positive treatment effect of VC investment on firm’s patenting activity depends on the type of investor. Different typologies of VC can have different objectives and characteristics (Tykvova 2006, Tykvova and Walz 2007). These differences are likely to influence the effect of VC finance on firm innovation and may help to differentiate the relative importance of these different channels through which VC investments may contribute to firm’s innovation. Obtaining a better understanding of the role of these factors represents a key priority for making further steps in assessing the economic effects of VC. Therefore, the creation of long longitudinal data sets relating to privately held firms located in different countries and backed by different types of investors is clearly a priority in this field.
17
8. References Aldrich, H.E., Kallenberg, A., Marsden, P., Cassell, J., (1989). “In pursuit of evidence: sampling procedures for locating new businesses”. Journal of Business Venturing 4, 367-386. Amit, R., Brander, J., Zott, C. (1998). “Why do venture capital firms exist? Theory and Canadian evidence”. Journal of Business Venturing 13, 441. Audretsch, D.B., 1995. “Innovation and industry evolution”. Cambridge, Mass: MIT Press. Ahuja, G., Katila, R. (2001). “Technological acquisitions and the innovation performance of acquiring firms: a longitudinal study”. Strategic Management Journal 21: 267-294. Atanasov V.A., Ivanov V.I., Litvak K. (2006). “VCs and the expropriation of entrepreneurs”, SSRN Working paper series, n. 905923. Barney, J.B., Busenitz, L.W. (1996). “New venture teams’ assessment of learning assistance from venture capital firms.” Journal of Business Venturing, Vol. 11, pp. 257. Barry, C.B., Muscarella, C.J., Peavy, J.W., Vetsuypens, M.R. (1990). “The role of Venture Capital in the creation of public companies”. Journal of Financial Economics 27, 447–471. Baum, J.A.C., Silverman, B.S. (2004). “Picking winners or building them? Alliance, intellectual, and human capital as selection criteria in venture financing and performance of biotechnology start-ups”, Journal of Business Venturing, n. 19, pp. 411-36. Bertoni F., Colombo M.G., D'Adda D., Grilli L. (2010), “VC financing and the growth of new technology-based firms: correcting for sample self-selection” in Audretsch D.B., Dagnino G.B., Faraci R. and Hoskisson R.E., eds., New Frontiers in Entrepreneurship: Recognizing, Seizing, and Executing Opportunities, (Springer). Bertoni F., Colombo, M.G., Grilli, L., (2009).“Venture capital and the growth of new technology-based firms: longitudinal analysis of the role of investor type”. Working paper. Black B.S., Gilson R.J., (1998). “Venture capital and the structure of capital markets: banks versus stock markets”, Journal of Financial Economics, 47(3): 243-277. Blundell, R., Griffith, R., Van Reenen, J. (1995). “Dynamic count data models of technological innovation”. Economic Journal vol.105: 333-344. Bottazzi, L., Da Rin, M., Hellmann, T. (2004). “The Changing Face of the European Venture Capital Industry: Facts and Analysis” Journal of Private Equity, 7, pp. 26-53. 18
Bygrave W., Timmons J. (1992). “Venture Capital at the Crossroads”. Harvard Business School Press, Boston, MA. Carpenter R.E., Petersen, B.C.(2002). “Is the growth of small firms constrained by internal finance?”, The Review of Economics and Statistics, 84, pp. 298-309. Chan Y.S. (1983). “On the positive role of financial intermediation in allocation of venture capital in market with imperfect information”. Journal of Finance, 35(5): 1543-1568. Chemmanur, T., Krishnan, K., Nandy, D., 2008."How Does Venture Capital Financing Improve Efficiency in Private Firms? A Look Beneath the Surface," Working Papers 08-16, Center for Economic Studies, U.S. Census Bureau. Cohen W.M., Nelson R.R., Walsh J. (2000). “Protecting Their Intellectual Assets: Appropriability Conditions and Why U.S. Manufacturing Firms Patent (or Not)”. NBER Working Paper 7552 (Feb). Colombo M.G., Delmastro M., Grilli L., (2004), Entrepreneurs' human capital and the start-up size of new technology-based firms. International Journal of Industrial Organization, vol.22, pp. 1183-1211. Colombo, M.G., Grilli, L., (2005). “Founders' human capital and the growth of new technology-based firms: a competence based view”, Research Policy 34, 795-816. Colombo, M.G., Grilli, L., (2007). “Funding gap? Access to bank loans by hightech start-ups”, Small Business Economics 29, 25-46. Colombo, M.G., Grilli, L., (2009). “On growth drivers of high-tech start-ups: the role of founders’ human capital and venture capital”, Journal of Business Venturing, forthcoming. Colombo, M.G., Grilli, L., Piva, E., (2006). “In search for complementary assets: the determinants of alliance formation of high-tech start-ups”, Research Policy 35, 1166-1199. Da Rin, M., Nicodano, G., Sembenelli, A., (2006). “Public policy and the creation of active venture capital markets”, Journal of Public Economics 90, 16991723. DaRin M., Penas M.F. (2007). “The Effect of Venture Capital on Innovation Strategies”. NBER Working Paper No. W13636. Available at SSRN: http://ssrn.com/abstract=1033761 Delmar, F., Shane, S., (2006). “Does experience matter? The effect of founding team experience on the survival and sales of newly founded ventures”, Strategic Organization 4, 215-247. Denis, D.J. (2004). “Entrepreneurial finance: an overview of the issues and evidence”. Journal of Corporate Finance 10: 301-326. Dushnitsky, G., Lenox, M.J. (2005). “When do incumbents learn from entrepreneurial ventures? Corporate venture capital and investing firm innovation rates.”. Research Policy, 34, pp. 615-639. 19
Dushnitsky, G., Lenox, M.J. (2006). “When does corporate venture capital investment create firm value?”. Journal of Business Venturing 21 (6), 753-772. Eckhardt, J.T., Shane, S., Delmar, F., (2006). “Multistage selection and the financing of new ventures”, Management Science 52, 220-232. Engel D., Keilbach M. (2002). “Firm Level Implication of Early Stage Venture Capital Investment. An empirical Investigation”. Discussion Paper No. 02-82. Centre for European Economic Research (ZEW). Gompers P.A. (1995) “Optimal investment, monitoring, and the staging of venture capital”. Journal of Finance 50 (5): 1461-1489. Gompers, P., Lerner, J. (2001). “The venture capital revolution”, Journal of Economic Perspectives 15, 145–168. Gorman M., Sahlman W.A. (1989). “What do venture capitalist do?”. Journal of Business Venturing 4: 231-248. Hellmann, T., (1998). “The allocation of control rights in venture capital contracts”. Rand Journal of Economics 29,57– 76. Hellmann T., Puri M. (2000). “The interaction between product market and financing strategy: the role of venture capital”. The Review of Financial Study, 13(4): 959-984. Hsu, D.H., (2006). “Venture Capitalists and Cooperative Start-up Commercialization Strategy”. Management Science, 52: 204-219. Hsu, D.H., Ziedonis R. (2007). “Patents as Quality Signals for Entrepreneurial Ventures”. Working Paper. Jeng L.A., Wells P.C. (2000). “The determinants of venture capital funding: evidence across countries”, Journal of Corporate Finance 6, 241-289. Kaplan S.N., Strömberg P. (2001). “Venture capitalists as principals: Contracting, Screening and Monitoring”. American economic Review, 91: 426-430. Kaplan S.N., Strömberg P. (2003). “Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts”. Review of Economic Studies, 70:2, p. 281. Kaplan S.N., Strömberg P. (2004). “Characteristics, Contracts and Actions: Evidence From Venture Capitalists Analyses”. Journal of Finance 59, no. 5 : 2177-210 Kortum S., Lerner J. (1998). “Does venture capital spur innovation?”. NBER Working Paper, 6846, Cambridge, Massachussetts. Kortum S., Lerner J. (2000). “Assessing the contribution of venture capital to innovation”. Rand of Journal Economics 31 (Winter). Lancaster T. (2000), “The Incidental Parameter Problem Since 1948”, Journal of Econometrics, 95, 391-413. Lee P.M., Wahal S. (2004). “Grandstanding, certification and the underpricing of venture capital backed IPOs.” Journal of Financial Economics, 73, pp. 375407. 20
Lerner J. (1995). “Venture capitalists and the oversight of private firms”. Journal of Finance, 50: 301-318. Lerner J. (1994). “The Importance of Trade Secrecy: Evidence from Civil Litigation”. HBS Working Paper 95-043. Available at SSRN: http://ssrn.com/abstract=6089. Lerner J., SØrensen M., Strömberg P. (2008). “Private equity and long-run investment: the case of innovation”. In Globalization of Alternative Investments, Working Papers Volume 1. The Global Economic Impact of Private Equity Report 2008. World Economic Forum series. Levin R.C., Klevorick A.K., Nelson R.R., Winter, S.G. (1987).“Appropriating the Returns from Industrial Research and Development”, Brookings Paper on Economic Activities 3, 783-820. Lindsey L. (2002). “The venture capital Keiretsu effect: An empirical analysis of strategic alliances among portfolio firms”. Working paper, Stanford University. MacMillan I.C., Kulow D.M., Khoylian R. (1989). “Venture Capitalists’ Involvement in Their Investments: Extent and Performance.” Journal of Business Venturing, Vol. 4, p27. Manigart S., Hyfte M. (1999). “Post-Investment Evolution of Belgian VentureCapital Backed Companies: An Empirical Study”. Paper presented at the Babson Entrepreneurship Conference. Masulis R.W., Nahata R., (2009). “Venture capital conflicts of interest: evidence from acquisitions of venture backed firms”, ECGI Finance working paper, n. 211/2008. Maula M.V.J. (2001). “Corporate Venture Capital and the Value-Added for Technology-based New Firms”. Doctoral dissertation, Institute of Strategy and International Business, Helsinki University of Technology. Maula M.V.J., Murray G. (2001). “Complementary Value-Adding Roles of Corporate Venture Capital and Independent Venture Capital Investors”. The Journal of BioLaw & Business 5(2) Megginson W., Weiss K. (1991). “Venture Capitalist certification in initial public offerings”. Journal of Finance 46: 879-903 Munari F., Toschi L. (2007). How good are VCs at valuing technology? Analysis of patenting and VC investments in nanotechnology. Working paper. Mundlak Y. (1978). “On the Pooling of Time Series and Cross Section Data”, Econometrica, vol. 46(1), pages 69-85. Puri M., Zarutskie, R., (2008). “On the lifecycle dynamics of venture-capital- and non-venture-capital-financed firms”, CES Working paper, n. 08-13. Repullo S., Suarez J. (2004). “Venture Capital Finance: A Security Design Approach”. Review of Finance 8: 75–108, 2004. 21
Sahlman W.A., (1990). “The structure and Governance of Venture-Capital Organizations”. Journal of Financial Economics: 473-521. Sapienza H.J. (1992). “When do venture capitalists add value?”. Journal of Business Venturing, 7: 9-27. Sapienza H.J., Manigart S., Vermeir W. (1996). “Venture Capital governance and value added in four countries”. Journal of Business Venturing, 11: 439-469. Semykina, A., Wooldridge, J.M., (2006). “Estimating panel data models in the presence of endogeneity and selection: theory and application”, mimeo, Michigan State University. Siegel R., Siegel, E. MacMillan, I., (1988). “Corporate venture capitalists: autonomy, obstacles, and performance”. Journal of Business Venturing 1, 275– 293. Spence M.,(1973). “Job Market Signalling”. Quarterly Journal of Economics, 87: 355-374. Sykes H.B. (1986). “The Anatomy of a Corporate Venturing Program: Factors Influencing Success.” Journal of Business Venturing 1: 275. Sykes, H., (1990). Corporate venture capital: strategies for success. Journal of Business Venturing 5, 37–47. Stiglitz J., Weiss A. (1981).“Credit Rationing in Markets with Incomplete Information”. American Economic Review, 71: 393-409. Stuart T.E., Hoang H., Hybels R.C. (1999). “Interorganizational Endorsements and the Performance of Entrepreneurial Ventures”. Administrative Science Quarterly, 44. Tykvová T., (2000). “Venture capital in Germany and its impact on innovation”. Social Science Research Network Working Paper. Ueda M., (2004). “Banks versus venture capital: project evaluation, screening, and expropriation”, Journal of Finance 59, 601-621. Ueda M., Hirukawa M. (2008a). “Venture Capital and Industrial 'Innovation'” (September 9, 2008). Available at SSRN: http://ssrn.com/abstract=1242693 Ueda M., Hirukawa M. (2008b). “Venture Capital and Innovation: Which is First?”. (September 14, 2008). Available at SSRN: http://ssrn.com/abstract=1242698. Vella, F. (1992). “Simple Tests for Sample Selection Bias in Censored and Discrete Choice Models”. Journal of Applied Econometrics. 7: 413-421. Wooldridge, J.M., (1995). “Selection corrections for panel data models under conditional mean independence assumptions”, Journal of Econometrics 68, 115132.
22
Tables Table 1 - Distribution of sample firms, VC-backed firms and patenting firms by industry Industry
Number of sample firms
Biotechnology & Pharmaceutics
22
4
18.2%
6
27.3%
ICT manufacturing
120
13
10.8%
21
17.5%
Automation & Robotics
49
4
8.2%
3
6.1%
Software
160
12
7.5%
6
3.8%
Total
351
33
16.8%
30
15.2%
VC-backed firms N. %
Patenting firms N. %
Table 2 – Venture Capital finance and patenting activity Non-VC-backed N.
%
292 26 318
91.8 8.2 100.0
Not-patenting Patenting Total
VC-backed N. 23 10 33
Total %
N.
%
69.7 30.3 100.0
315 36 351
89.7 10.3 100.0
Table 3 – Venture Capital finance and timing of patenting activity
Patent applications Patents granted
… before VC entry … after VC entry … before VC entry … after VC entry Total
Number of patents …
Number of firms patenting …
3 54 0 57 57
2 10 0 10 10
23
Table 4 – Dependent and explanatory variables of patenting activity Variable NPatents(t) DPatents(t) Employees(t-1) Age(t) VC-backed(t-1) Patent stock(t-1) Techworkexp
Comworkexp
Otherworkexp
Techeduc Ecoeduc ΔTotalAssets(t-1) ΔDebt(t-1) CashFlow(t-1) DPublicSubsidies(t-1) DEUProject(t-1)
24
Description Number of applications of the firm in year t for patents which are later granted by the patent office Dummy variable equal to 1 when firm applies for at least one patent in year t and the patent is subsequently granted Logarithm of the size of the firm at year t-1 measured by the number of employees (including owners that have an active role in the management of the firm) Logarithm of firm’s age at year t Dummy variable equal to 1 if firm received VC financing in or before year t-1 Firm’s patent stock in year t-1 computed as NPatents(t-1) + 0.85 Patent stock(t-2) Average number of years of technical work experience of founders in the same sector of the start-up before firm’s foundation. Average number of years of commercial work experience of founders in the same sector of the start-up before firm’s foundation. Average number of years of work experience of founders in sectors other than the one of the start-up before firm’s foundation. Average number of years of scientific and/or technical education of founders at university level. Average number of years of economic and/or managerial education of founders at university level. Increase in the amount of assets between year t-2 and year t-1 (difference in logarithm) Increase in the amount of debt between year t-2 and year t1 over the amount of total assets Cash flow in the year t-1 over the amount of total assets Dummy variable equal to 1 if firm received public subsidies (national or local) in year t-1 Dummy variable equal to 1 if firm was involved in a research project financed by the European Commission in year t-1
Table 5 – Effect of Venture Capital on patenting activity: results of the estimates
VC-backed(t-1) Employees(t-1) Age(t)
(1) 1.248 (0.551) 0.652 (0.219) -0.577 (0.358)
** ***
Patent stock(t-1) Techworkexp Comworkexp Otherworkexp Techeduc Ecoeduc
DPatents(t)
NPatents(t)
Logit random effects
Negative binomial random effects
(2) 1.174 (0.609) 0.585 (0.234) -0.480 (0.386) -0.007 (0.096) 0.069 (0.044) 0.096 (0.055) 0.026 (0.031) 0.246 (0.104) 0.168 (0.344)
* **
*
**
ΔTotalAssets(t-1) ΔDebt(t-1) CashFlow(t-1) DPublicSubsidies (t-1) DEUProject(t-1) Software ManufacturingICT BioPharma Constant
Number of Observations Number of Groups
-1.630 (0.830) 0.593 (0.726) 1.032 (0.913) -5.617 (0.984)
**
***
-1.892 (0.851) 0.349 (0.749) 0.505 (0.926) -6.523 (1.239)
**
***
(3) 1.091 (0.632) 0.481 (0.245) 0.097 (0.524) 0.215 (0.123) 0.127 (0.049) 0.096 (0.056) 0.050 (0.032) 0.270 (0.106) 0.004 (0.445) 0.052 (0.702) 2.449 (1.299) 4.535 (1.878) 0.128 (0.539) -0.253 (0.763) -2.484 (0.885) -0.613 (0.742) 0.138 (0.874) -7.714 (1.563)
* **
(4) 1.310 (0.447) 0.414 (0.188) -0.473 (0.319)
*** **
* *** *
**
(5) 1.598 (0.518) 0.438 (0.199) -0.407 (0.353) -0.077 (0.051) 0.060 (0.042) 0.049 (0.050) 0.014 (0.028) 0.258 (0.092) 0.072 (0.318)
*** **
***
* **
***
***
-1.689 (0.667) 0.238 (0.554) 0.730 (0.736) -1.183 (0.905)
**
-1.973 (0.743) 0.185 (0.625) 0.631 (0.843) -2.361 (1.046)
***
**
(6) 1.602 (0.656) 0.431 (0.230) -0.212 (0.504) 0.053 (0.079) 0.099 (0.049) 0.077 (0.054) 0.039 (0.031) 0.278 (0.100) -0.096 (0.454) -0.100 (0.693) 1.361 (1.060) 3.850 (1.623) -0.338 (0.490) -0.150 (0.653) -2.455 (0.849) -0.459 (0.720) 0.196 (0.854) -3.560 (1.458)
2,266
2,151
1,371
2,266
2,151
1,371
351
333
224
351
333
224
** *
**
***
**
***
**
Note: ***, ** and * denote respectively significance level below 1%, 5% and 10%. Reporting coefficient of Logit and Negative Binomial regressions estimated using random effect models. Standard errors in brackets.
25
Table 6 – Estimated effect of VC investments on firms’ patenting activity Probability of patenting Expected number of patents a) VC-backed(t-1)=1 0.0193 0.1224 b) VC-backed(t-1)=0 0.0066 0.0247 Ratio ( a / b ) 2.9384 4.9610 Note: predictions using estimated models (3) and (6), obtained by setting the value of all explanatory variables at the median value with the exception of VC-backed(t-1) that alternatively takes value 1 and 0.
26