«John R. Graham, Co-Supervisor David T. Robinson, Co-Supervisor Manuel Adelino Alon P. Brav Manju Puri Aaron K. Chatterji Dissertation submitted in ...»
Technological Uncertainty. I ﬁrst explore heterogeneous eﬀects of innovation deteriorations on CVC initiations across uncertainty levels that ﬁrms face in their informational environment. The working hypothesis is that the impact of innovation deterioration should be stronger when the uncertainty level is higher, that is, when identifying valuable innovation opportunities becomes more diﬃcult and information is therefore more valuable. I estimate an extended model based on the OLS model (4.1) and 2SLS models (4.3) and (4.4). The sample is categorized into two subgroups 38 by the median of uncertainty levels of ﬁrms’ informational environment, indicated by Iuncertainty.
The results are reported in Table E.1, which shows that the causal relation between deterioration in innovation and the decision to engage in CVC investment is stronger when there is higher demand to acquire information on new technologies and new markets, favoring the informational rationale behind CVC.
This result cannot be explained by the interpretation that ﬁrms make CVC investments before acquiring a new technology, as a way to wait for the uncertainty to resolve. Indeed, CVC investments seldom evolve to acquisition of the portfolio company. Recent studies examine acquisition cases when CVC investors acquire portfolio companies in which they invested (Benson and Ziedonis, 2010; Dimitrova, 2013). In general, the acquisition of portfolio companies is rare—fewer than one-ﬁfth of CVC investors acquired their portfolio companies. CVC units that did conduct such acquisitions acquired fewer than 5% of their portfolio companies (that is, one out of 20 investments).
Managerial Desperation and Leapfrog Innovation. Early research shows that desperate managers, after experiencing a negative shock, might aggressively seek outside solutions to the deterioration, which typically lead to even worse outcomes (Higgins and Rodriguez, 2006). Therefore, one could reasonably worry that the result simply documents that desperate managers are more likely to conduct CVC investment for leapfrog innovations. I investigate this issue by studying the success rate of the portfolio companies invested by CVCs categorized by the severity of innovation declines at initiation. If the concern is indeed the case, we would expect CVC parents that experienced the largest hit before initiating to have lower performance because they likely made the decision out of desperation. In Table E.2, I ﬁnd that those CVCs whose parents’ performance decline the most actually score a similar, if not higher success rate compared to other CVCs.
39 Financial Returns. What is the role of ﬁnancial condition in ﬁrms’ decision to operate a CVC? On the one hand, anecdotal cases (e.g., Google, Intel) give us the impression that CVC is an investment channel for cash-rich ﬁrms to make equity investments in the startup market. In contrast, the structure and features of CVC investments could lead to the hypothesis that CVC could be poor-man’s innovation, that is, declining ﬁrms are more ﬁnancially constrained and cannot conduct internal R&D or M&As, which are on average more costly than CVC. In Table E.3, I show that the main result is robust on the subsample of ﬁrms whose KZ-index is below the industry median or whose cash-ﬂow ratio is above the industry median (less ﬁnancially constrained).
More Robustness. To conﬁrm that the results are not driven by the sampling process or speciﬁcations, I conduct an array of robustness checks. In the Appendix, I show that: the result is not sensitive to the length used to capture innovation changes (τ “ 3 in the paper); the result is robust to removing ﬁrms that are large/small, that are from speciﬁc industries (such as IT or pharmaceuticals), or that are located in speciﬁc locations (in California); and that the result also holds for deteriorations of product market performance such as ROA and growth rate in sales.
40 5 CVC Operations: Select, Acquire, and Integrate Information Section 4 presents evidence consistent with the view that the information acquisition motive drives the initiation of the CVC life cycle. To explore the information acquisition view further, this section examines how CVCs select which portfolio companies to acquire valuable information from and identiﬁes the information spillover from those startups to CVC parent ﬁrms. Empirically, I construct a comprehensive data set on innovation-related activities in both CVC parents and the entrepreneurial sector. Using this database, I can test the information acquisition view by examining whether CVCs target entrepreneurial companies that could potentially provide higher informational value to the parents, and by tracking the dynamic of incorporating new information into corporate activities within parents.
5.1 CVC Portfolio Formation I start by examining how the selection of portfolio companies reﬂects the CVC information acquisition rationale. Selecting portfolio companies involves trading oﬀ multiple factors that determine the eﬃciency of information acquisition. The 41 ﬁrst consideration is the technological proximity between the parent ﬁrm and a startup. The conceptual idea is that investing in technologically proximate companies facilitates the process of absorbing and integrating information, therefore creating greater informational beneﬁt (Cohen and Levinthal, 1990; Dushnitsky and Lenox, 2005b). The second factor is incremental informational value through investment.
Indeed, investing in companies with very similar knowledge sets adds little marginal informational beneﬁt, although it could be eﬃcient for creating synergies (Bena and Li, 2014). The third determinant is the availability of alternative information acquisition channels. The working hypothesis is that CVC investors should pursue information that would be diﬃcult to acquire without the CVC channel, that is, we should expect CVC investment to concentrate on companies with little informational communication otherwise.
To empirically analyze how CVC parent ﬁrms balance these economic forces in selecting portfolio companies, I construct a data set by pairing each CVC i with each entrepreneurial company j that was ever invested by a VC. I remove cases when the active investment years (between initiation and termination) of CVC ﬁrm i and the active ﬁnancing years of company j (between the ﬁrst and the last round of VC ﬁnancing) do not overlap. I estimate a probability model on this sample to predict the decision of CVC i investing in company j, that is,
The key variables of interest in model (5.1) are TechProximity, Overlap, and SameCZ, which capture the informational relation between a CVC parent ﬁrm i and an entrepreneurial company j, echoing the three potential portfolio determinants outlined above.1 The ﬁrst measure, Technological Proximity (TechProximity), is calculated as the Cosine-similarity between the CVC’s and the startup’s vectors of patent weights across diﬀerent technology classes (Jaﬀe, 1986; Bena and Li, 2014). A higher Technologial Proximity indicates that the pair of ﬁrms works in closer areas in the technological space.
The second measure, Knowledge Overlap (Overlap), is calculated as the ratio of—(1) numerator: the cardinality of the set of patents that receive at least one citation from CVC ﬁrm i and one citation from entrepreneurial company j; and (2) denominator: the cardinality of the set of patents that receive at least one citation from either CVC i or company j (or both). A higher Knowledge Overlap means that the pair of ﬁrms shares broader common knowledge in their innovation.
To provide a clean interpretation of the estimation, both Technological Proximity and Knowledge Overlap are measured as of the last year before CVC i and company j both enter the VC-startup community. For example, if ﬁrm i initiates the CVC in 1995 but company j obtained its ﬁrst round of ﬁnancing in 1998, the measure is constructed using the patent proﬁles in 1997. The rationale for this criterion is to mitigate the potential interactions between CVCs and startups before investment.
To construct a proxy for the availability of alternative information acquisition channels, I rely on recent studies showing that geographic proximity inﬂuences the intensity of knowledge spillover between ﬁrms (Jaﬀe et al., 1993; Peri, 2005). The 1 The Appendix describes the methodology identifying innovation activities of entrepreneurs through merging patent data sets with VentureXpert and deﬁnes those variables more formally.
43 main variable is a dummy indicating whether CVC ﬁrm i and company j are located in the same Commuting Zone (CZ). I use CZ as the geographic delineation because it has been shown that CZ is more relevant for geographic economic activities (Autor, Dorn, and Hanson, 2013; Adelino, Ma, and Robinson, 2016) and innovation spillover (Matray, 2014). Projecting the information acquisition hypothesis on this context, we should expect that CVCs invest less in companies that are in the same geographic location, from which they could learn through the more inexpensive mechanism of local knowledge spillover.
Table 5.1 presents coeﬃcients estimated from model (5.
1). In column (1), a positive and signiﬁcant coeﬃcient means that the Technological Proximity between a CVC and an entrepreneurial company increases the likelihood of CVC deal formation. This result is consistent with the interpretation that CVCs select companies from which they are more capable of absorbing knowledge for their core business.
Column (2) examines the eﬀect of Knowledge Overlap. The negative coeﬃcient means that after conditioning on the technological proximity, CVC parent ﬁrms prefer to invest in companies with diﬀerent knowledge bases. In other words, CVCs select portfolio companies through which they are exposed to more new innovation knowledge. Importantly, this result could potentially distinguish the information acquisition rationale for CVC with the alternative rationale that CVC is conducted for product market synergies and asset complementarity. Under non-informational strategic concerns, ﬁrms favor targets with both close technological proximity and high knowledge overlap in order to achieve economic synergies (Bena and Li, 2014).
In column (3), I study the eﬀect of alternative information acquisition channels, speciﬁcally knowledge spillover, on CVC portfolio selection. The literature on VC, and on investment more broadly, has documented a “home (local) bias” phenomenon— 44 Table 5.1: The Selection of CVC Portfolio Companies This table studies how CVCs strategically select portfolio companies. I construct a cross-sectional data set by pairing each CVC i with each entrepreneurial company j that was ever invested by a Venture Capital investor. I remove cases when the active investment years of CVC ﬁrm i (between initiation and termination) and active ﬁnancing years of company j (between the ﬁrst and the last round of VC ﬁnancing)
do not overlap. The analysis is performed using the following speciﬁcation:
IpCV Ci -T argetj q “ α`β1 ¨T echP roximityij `β2 ¨Overlapij `β3 ¨SameCZij `γˆX`εij, where the dependent variable, IpCV Ci -T argetj q, is equal to one if CVC i actually invests in company j, and zero otherwise. Technological Proximity is calculated as the Cosine-similarity between the CVC’s and startup’s vectors of patent weighting across diﬀerent technological classes (Jaﬀe, 1986; Bena and Li, 2014). Knowledge Overlap is calculated as the ratio of the cardinality of the set of patents that receive at least one citation from CVC ﬁrm i and one citation from the entrepreneurial company j, and the cardinality of the set of patents that receive at least one citation from either CVC i or company j (or both). Geographical distance is measured using a dummy variable if the CVC ﬁrm i and company j are located in the same Commuting Zone (CZ), IpSameCZq. The Appendix deﬁnes those variables more formally. In order to provide a clean interpretation of the estimation, both Technological Similarity and Knowledge Overlap are measured as of the last year before CVC i and company j both enter the VC-startup community, and the goal is to mitigate the potential interaction between them in the VC-startup community. Fixed eﬀects at CVC ﬁrm and entrepreneurial company level are included. T-statistics are shown in parentheses and standard errors are clustered by CVC ﬁrm. *, **, *** denote statistical signiﬁcance at the 10%, 5%, and 1% levels, respectively.
45 when investing in companies that are geographically closer, investors can better resolve the information asymmetry problem and conduct more eﬃcient monitoring (Da Rin, Hellmann, and Puri, 2011). In column (3), however, I ﬁnd that CVCs do not really invest in their “home” companies. The dummy variable indicating that the CVC and the startup are located in the same Commuting Zone negatively aﬀects the probability of investment, which is consistent with the explanation that CVC parent ﬁrms can acquire information from startups in the same CZ through local innovation spillover (Matray, 2014), which decreases the marginal beneﬁt of making a CVC investment in them.
Overall, Table 5.1 shows that CVCs strategically select information sources and invest in companies from which they could acquire beneﬁcial information. They invest in companies that work in similar technological areas and possess knowledge new to the parent ﬁrm. They are less likely to invest in companies located in the same geographic areas from which they could gain information through inexpensive local knowledge spillover.
5.2 Internalizing Acquired Information