Abstract
We investigate the influence of government grants, venture capital (VC), and private equity (PE) funding on innovation in newly public firms. We examine innovation inputs (R&D), innovation outputs (patents), and the quality thereof (patent citations). We contribute to understanding of the mechanisms between government grants and subsequent VC and PE funding and innovation. We find that grants encourage VC funding but not PE funding. Grants and VC/PE funding are generally complements regarding innovation except grants substitute for VC funding on innovation inputs. Furthermore, we observe that the firm-level heterogeneity of VC/PEs significantly influences innovation in portfolio companies.
JEL Classification:
1. Introduction
Innovation is important to firm performance and survival. New venture firms tend to pursue a strategy of innovation to enhance their competitive advantage, performance outlook, and valuation (Terziovski, 2010). Accordingly, new venture firms may seek outside resources, including funding, to enhance their knowledge base and capability to innovate (Yli-Renko et al., 2001). Three sources for such funding are government grants, venture capital (VC) firms, and private equity (PE) firms (Bruton et al., 2010; Rind, 1981). However, new venture firms often face extensive constraints on their ability to acquire funding, in part due to potentially severe information asymmetries. The literature indicates that new venture firms may use a variety of mechanisms to signal their quality in order to reduce these information asymmetries.
Scholars have argued that such signals may derive from a variety of actions, including forming an alliance with a prominent partner (Stuart et al., 1999), possessing industrial and entrepreneurial experience (Burton et al., 2002; Eisenhardt and Schoonhoven, 1990; Hsu, 2007), or exhibiting particular top management team characteristics (Higgins and Gulati, 2006; Zhang and Wiersema, 2009). Recent studies suggest that patents can also help alleviate information frictions in the market for entrepreneurial capital and find that patents increase the likelihood of VC funding (Conti et al., 2013; Farre-Mensa et al., 2016; Häussler et al., 2012; Hoenen et al., 2014; Hsu and Ziedonis, 2013; Zhou et al., 2016). A limited number of related studies have examined how government subsidies influence access to financial resources by generating a certification effect—a signal of quality (Lerner (1999) for VC and Feldman and Kelley (2006), Meuleman and De Maeseneire (2012), and Wang and Thornhill (2010) for external capital (debt and equity)). However, there is little evidence of the impact of government subsidies (such as grants) on subsequent VC/PE funding 1 and the interplay between government subsidies, VC/PE, and innovation.
The impact of both VC (early-stage financing) and PE (later-stage financing) funding on the financial performance of new venture firms has been highly studied in the literature (Bruton et al., 2010; Fitza et al., 2009). However, there has been less focus on the impact of VC and PE funding on innovation (Kortum and Lerner, 2001; Lerner et al., 2011; Mollica and Zingales, 2007; Popov and Roosenboom, 2009).
Differences between VC and PE firms may lead to differences in the innovation outcomes of portfolio companies. We view VCs and PEs as distinctly different funding sources with different non-pecuniary support approaches and resources (Park and Steensma, 2012). For example, VC firms typically raise funds with fixed life spans (7–10 years), whereas PE firms may not have such a limit. VC, in our investigation, includes seed, startup, early stage, and expansion stage funding; thus, it supports the preliminary stages of a company’s life. We view PE as typically funding later-stage activity where companies are past the initial growth phase and risks may be lower. However, such companies can still benefit from investments in innovation and from the managerial guidance typically facilitated by PE firms. Such differences between VC and PE firms may lead to differences in the innovation outcomes of portfolio companies.
Furthermore, the influence of government grants on innovation has received limited attention (Branstetter and Sakakibara (2002), Bronzini and Piselli (2016), Clausen (2009), David et al. (2000), and Jaffe and Le (2015) focus on innovation outcomes). Since firms can access grants, VC, and PE funding, at various stages of their development, and grant funding may indeed encourage subsequent VC or PE funding, this leaves us with incomplete knowledge of whether and why government grants might encourage subsequent VC or PE funding and whether and why grants might supplement or complement VC and PE funding with regard to innovation.
To address these gaps, we investigate how government grants, VC funding, and PE funding influence innovation inputs (R&D expenditure) and outputs (patents), and the quality of those outputs (patent citations). 2 We draw upon the logic of information economics (Akerlof, 1970; Riley, 2001; Spence, 1974) and the knowledge-based view of the firm (Grant, 1996a, 1996b; Kogut and Zander, 1992b) to develop our hypotheses. We analyze the impact of VC, PE, and grant funding using a sample of 436 firms that list on the Australian Stock Exchange (ASX). We examine newly public firms at the point of their initial public offering (IPO). Innovation in new venture firms is often undertaken with the intention to achieve listing on the stock market through an IPO: a highly desired outcome for both new venture firm entrepreneurs and investors (Bruton et al., 2010; Wu, 2012).
We acknowledge that in focusing on newly listed companies, we implicitly assume that the company has been profitable/successful enough to reach the IPO stage. 3 However, we impose this requirement on both the treatment sample (VC-/PE-backed companies) and the control sample (non-VC-/PE-backed companies). This allows a comparison of firms that are at a similar stage in their life cycle (Maug, 2001). Subsequently, we compare survived companies with other survived companies to draw conclusions about the impact of VC/PE backing. 4 We do not directly examine unlisted companies, 5 many of whom might not survive to reach an IPO and thus the results may not be generally applicable. However, the Australian stock market facilitates the listing of relatively small companies, with measures such as not imposing a minimum stock price (unlike the United States, which imposes a US$1 stock price) and thus newly listed firms may have some similarities to unlisted companies. Subsequently, while our results are most directly applicable to newly public companies, they may have implications for unlisted companies that receive VC/PE backing.
We analyze the level of innovation inputs (R&D expenditure) and outputs (patents), and the quality of those outputs (patent citations) and we distinguish between whether the company had a government grant, VC funding, or PE funding at the time of listing on the ASX. 6 We also take steps to mitigate analytical risks, including endogeneity and systemic differences between VC-/PE-backed firms and non-VC-/PE-backed firms, in addition to conducting numerous sensitivity tests.
We find that grants increase innovation inputs, outputs, and output quality. VC funding increases innovation inputs. PE funding increases innovation outputs and the quality thereof. We find grant recipients who receive VC or PE funding produce more innovation outputs and outputs of higher quality than do grant recipients who do not receive VC or PE funding, suggesting that VC and PE funding complements government grants regarding innovation outputs and output quality. However, we also have two unexpected findings. First, grants substitute for VC funding regarding innovation inputs while they complement PE funding. Second, the receipt of a grant encourages a VC, but not a PE, to fund a new venture firm. In our discussion, we develop deeper explanations of these two unexpected findings. In addition, our model control variables include discernible attributes of VC/PEs; thus, we provide insight on the types of VC and PE firms that encourage innovation thereby providing additional insights for practitioners, especially entrepreneurs, who might select between different funding sources.
This investigation of funding sources for new venture firms primarily contributes to the literature on innovation while it also adds to the understanding of funding source heterogeneity. We examine government grants, VC funding, and PE funding as distinct antecedents of innovation where innovation is decomposed into innovation inputs, outputs, and quality of the outputs. Importantly, the joint consideration of government grants, along with VC and PE funding, adds a novel aspect that has received limited investigation. 7 We argue that government grants provide an externally validated signal of quality, which encourages subsequent funding by VC/PE firms and the results demonstrate that grants, VC funding, and PE funding differ in their relationship with innovation in new venture firms and differ in their impact on innovation inputs and outputs. In addition, this study provides insights that directly apply to managers of new venture firms and to managers of VC and PE firms.
2. Theory and hypotheses
The theoretical basis for our analysis is information economics (Akerlof, 1970; Riley, 2001; Spence, 1974), which we supplement with perspective from the knowledge-based view (Grant, 1996a, 1996b; Kogut and Zander, 1992b). Unlike prior studies on the influence of VC and PE funding that generally consider the resultant valuation or performance of the firm, we investigate innovation. Innovation is an important strategic outcome for new venture firms, especially prior to IPO. Innovation is generally sought because it is a path to competitive advantage and higher performance (Crossan and Apaydin, 2010; McGrath et al., 1996). The innovation literature has associated the innovativeness of firms with a variety of causal factors including knowledge, resources, capabilities, networks, and the management of the firm (Ahuja, 2000; Henderson and Cockburn, 1996). We build on this prior work in our investigation of the influence of government grants and VC/PE on innovation. First, we examine the mechanisms for the direct relationships between funding sources and innovation. We follow by considering mechanisms for the influence of government grants on subsequent VC or PE funding. Finally, we consider the complementarity of VC and PE funding with government grants regarding innovation. 8
2.1. The influence of grant funding on innovation
The logic of governments offering research grants follows the traditional market failure argument (Arrow, 1962). Given that innovation is generally associated with economic growth (Abromovitz, 1956; Solow, 1957), the policy intent is to fund socially beneficial activity where returns to private investors are viewed inadequate due to incomplete appropriability and knowledge spillovers (Clausen, 2009; Mahoney and Qian, 2013). There is some evidence that government grants encourage innovation inputs and outputs in the United States (Lerner, 1999; Toole and Czarnitzki, 2007) and Europe (Aerts and Schmidt, 2008; Clausen, 2009; Colombo et al., 2011). However, the question of whether grants stimulate or substitute private R&D investment is not clear (David et al., 2000). As a possible reason for this, Clausen (2009) argues that the impact of subsidies depends on their purpose and grants for activity “far from the market” complement existing R&D, otherwise grants substitute for existing R&D.
Importantly, at the firm level, research grants are generally expected to be earmarked for research expenditure. Thus, the expectation, backed by empirical research (Clausen, 2009), is that government grants positively influence innovation inputs (R&D expenditure). Similarly, the expectation, backed by empirical research (Colombo et al., 2011), is that government grants positively influence innovation outputs (patents). Even though grants do not directly provide innovation outputs, the general logic is that grant funding enables higher quality research. A government grant enhances the firm’s ability to procure capable knowledge resources such as more knowledgeable and better educated personnel, purchase equipment, and expand testing (Clausen, 2009; Kaplan and Stromberg, 2004; Podolny, 1993).
Most importantly, innovation quality has received limited attention. Our argument is that government grants engender the firm with an externally validated signal of quality. The literature in information economics indicates that credible signals of quality have substantial influence on a firm’s strategic interactions with other firms and market participants and thus can mitigate issues of information asymmetry that might otherwise influence decisions (Akerlof, 1970; Ragozzino and Reuer, 2011; Riley, 2001; Spence, 1974).
A government grant, which is awarded in a competitive process with expert referees, provides an external quality validation (Link and Scott, 2010). This panel of expert referees is intended to be knowledgeable, objective, and dispassionate (Graffin and Ward, 2010; Tolbert et al., 2011). Thus, a government grant is a publicly visible, external validation of both some level of current success and of expected future success. As such, a government grant reduces the uncertainty associated with new venture firms by informing others of its favorable prospects (Lee et al., 2001) and provides a legitimating function in the marketplace (Tolbert et al., 2011). In other words, a government grant is imbued with “meaning” beyond the monetary value of the funds.
In sum, we contend that the signaling mechanism of a government grant enables higher innovation output quality (patent citations) as well as supports higher innovation outputs (patents) for the firm. Hence
Hypothesis 1. Government grants are positively associated with innovation inputs, outputs, and the quality of those outputs.
2.2. The influence of VC and PE firms on innovation
In the knowledge-based view of Kogut and Zander (1992) and Grant (1996a, 1996b), proprietary knowledge is the fundamental source of competitive advantage. Generating firm-specific knowledge through innovation can allow firms to obtain a competitive advantage. In general, firms seeking to grow knowledge will benefit from additional funding, which may enable investment in a larger stock of strategic assets (Dierickx and Cool, 1989), complementary resources (Barney, 1991), and/or individuals who possess specialized knowledge (Grant, 1996a, 1996b). In relation-specific knowledge (Arend et al., 2014), firms synergistically integrate knowledge with partners. VC and PE firms are recognized as providing knowledge and strategic resources to their portfolio companies (Arikan and Capron, 2010; Ragozzino and Reuer, 2011). We differentiate between VC and PE firms as their impact on innovation may differ. We define VC firms as those that invest in early-stage companies and PE firms as those that invest in later-stage companies.
VC and PE firms play a role in generating innovation by providing capital, improving governance, and establishing a credible signal to attract quality employees (Lee et al., 2001; Wallace et al., 2016). As they fulfill simultaneous roles as providers of funds and investors, VC and PE firms encourage innovation in their portfolio companies as a means to generate competitive advantage and higher valuations (Arend et al., 2014; Wu, 2012). Furthermore, as VC and PE firms invest in multiple portfolio companies, they can both acquire skills from managing those companies and can facilitate knowledge sharing between portfolio companies (De Clercq and Sapienza, 2005). This external knowledge can help portfolio companies to increase their know-how and thus innovation (Kogut and Zander, 1992).
There is some empirical evidence consistent with the notion that VCs and PEs promote innovation (Kortum and Lerner (2000, 2001), Lerner et al. (2011), and Mollica and Zingales (2007) for the United States, Guo and Jiang (2013) for China, and Popov and Roosenboom (2009) for Europe). This prior literature suggests that PEs mainly contribute to innovation outputs (patents), whereas VCs mainly focus on increasing the firm’s innovation inputs (R&D). This is consistent with the distinct objectives of VCs and PEs. PEs tend to invest in later-stage companies and fulfill a role of providing support necessary to capitalize on latent innovation. In contrast, VCs tend to invest in early-stage companies with nascent innovativeness and high potential (Metrick and Yasuda, 2010). Companies at this stage increase their knowledge base, competitive advantage, and value by further investment in innovation, that is, by increasing R&D expenditure.
However, in addition to VCs and PEs focusing on innovative growth and applying management capability to a portfolio company, they also provide signaling value. Analogous to our arguments for a government grant, both VC and PE funding engender the firm with a credible signal of quality and legitimacy (Lee et al., 2001) for its stage of development. Due to this quality signal, VC funding results in the procurement and retention of top research staff and an increase in innovation inputs, output, and output quality, extending from our previous arguments for grants.
However, we contend that PE funding will have a different outcome based on the quality signal it provides. PE firms, as later-stage investors, focus on contributing to the management of their portfolio companies (Kaplan and Stromberg, 2004) with a focus on capitalizing on the latent innovation of portfolio companies. Such management skill is essential for producing quality innovation (Bantel and Jackson, 1989). Thus, PEs are likely to focus on innovation outputs and output quality. This is because the knowledge-based resources they attract to portfolio companies tend to have development and commercialization capabilities rather than research and invention capabilities as in the case of VCs. This argument suggests that PE funding results in increased innovation outputs and output quality. Therefore, we propose
Hypothesis 2a. VC funding is positively associated with an increase in innovation inputs, outputs, and the quality thereof in portfolio companies.
Hypothesis 2b. PE funding is positively associated with an increase in innovation outputs and the quality thereof in portfolio companies.
2.3. Grants as a signal for VC and/or PE funding
Young firms often face excessively costly external finance due to frictions such as information asymmetry, asset intangibility, and incomplete contracting (Holmstrom, 1989). Rationales for government grants include the undervaluation of the social benefits of innovation by the private sector and that small firms underinvest in early-stage R&D due to financial frictions. Criticism of R&D subsidy programs is that they crowd out private investment (Wallsten, 2000) or allocate funds inefficiently (Lerner, 2009). More recent evidence suggests that public research subsidies stimulate private R&D investments (e.g. Becker, 2015; Clausen, 2009; Görg and Strobl, 2007; Hottenrott et al., 2015). The empirical evidence shows that direct subsidies are especially effective to stimulate innovation in areas with higher degrees of innovation novelty (Beck et al., 2017).
Government agencies that run US funding programs use their internal scientific and technical capabilities and also involve prestigious external review partners to better inform the process (Howell, 2017; Pahnke et al., 2015). Grant selection processes are highly competitive and based on meritorious criteria where submissions go through numerous rounds of rankings by expert teams (Hsu, 2006). The Australian grant process that we investigate generally has similar characteristics to the United States (see Supplementary Appendix A.1 for institutional detail).
The outcome of the grant process conveys a strong signal of scientific and technical merit, conveying market relevant information about grantee quality. Islam et al. (2018) argue that although the signaling mechanism from the receipt of a government research grant has attributes typical to the signaling literature, there are some differences. Grant processes are anticipated to only benefit high-quality firms and prevent a noisy signaling environment that would undermine a signal’s value. A firm applying for a grant, going through the grant process, and then having its technology validated by winning the grant introduces an external gatekeeper into the signaling mechanism. This external gatekeeper should prevent low-quality firms from being able to falsely signal a higher quality type and benefiting from this strategy.
Importantly, Islam et al. (2018) note that the grant process has two central characteristics of an efficacious signal: observability and costliness (Connelly et al., 2011). The award of a grant is a public process and is clearly observable from the public agency and media reports. The application process takes some time and resources and can be a costly process for young firms that have limited resources. Islam et al. (2018) suggest that grant success has two features that are valued in the signaling literature: a veracious signal (Busenitz et al., 2005) that is accompanied by a valuable interorganizational tie (Park and Mezias, 2005).
New venture firms may seek VC/PE funding to further increase their knowledge base, competitive advantage, and financial value. VCs and PEs investigate potential portfolio companies as part of the investment selection process. These investigations are costly and may not result in accurate quality determinations (De Treville et al., 2014). Firms that win grants can use the award as an externally validated signal of accomplishment (Islam et al., 2018). Lerner (1999) suggests that these awards play an important role in certifying firm quality to private investors.
We argue that credible signals of quality from a government grant will positively influence the investment decision of VCs and PEs. This is an important extension to the idea that VC funding can provide a signal to future potential financiers (Ragozzino and Reuer, 2011). Furthermore, to the extent that VCs, more than PEs, tend to focus on innovative early-stage firms, we expect that government grants encourage VC funding more than they encourage PE funding. This is because the signaling benefits of grants are likely to be more salient for VCs because early-stage firms typically feature higher levels of information asymmetry. Therefore, we predict
Hypothesis 3a. The presence of a government grant is (strongly) positively associated with VC funding.
Hypothesis 3b. The presence of a government grant is positively associated with PE funding.
2.4. Grants as a substitute or complement for VC and/or PE funding
VC and PEs are anticipated to encourage innovation in their portfolio companies in order to achieve more favorable exit outcomes (Rind, 1981). While government grants may provide support to build knowledge or to expedite its building, it is not clear that VCs and PEs will continue investing in innovation in their portfolio companies that previously received a grant. That is, whether government grants substitute or complement VC or PE funding regarding innovation.
Both VCs and PEs seek capital appreciation as this increases the eventual value at which they can sell/exit their portfolio companies. While capital appreciation may be achieved by growing innovation, it may also be achieved in some cases by cutting costs. Costs could be cut by limiting further investment in innovation (R&D) and focusing on commercialization. For example, Engel and Keilbach (2007) find that German VC/PEs focus on bringing a company’s existing innovations to the market. This possibility has a higher probability for firms having a strong pre-existing knowledge and patent base, such as firms in prior receipt of a government grant.
We present two arguments that VC/PE funding and grants are complementary. We first expect that the quality signal of a grant stimulates earlier VC/PE investment in a portfolio company than would normally be made. This earlier investment enables VC/PEs to contribute their skills and expertise at an earlier stage, when such skills may be especially useful, thereby complementing the receipt of a grant. A government grant provides an externally validated signal that the firm has a high level of innovativeness and potential for value. As indicated in Hypothesis 3, this signal promotes VC/PEs to fund new venture firms by mitigating issues of information asymmetry that might otherwise deter such VC/PE funding. The signaling effect of a grant is greatest for companies in which the information-asymmetry problem is greatest: relatively innovation-intensive, early-stage new venture firms. Thus, by ameliorating information-asymmetry-based barriers, the grant encourages earlier investment by both VCs and PEs. The earlier funding suggests the VC/PE will need to continue to invest in R&D. Furthermore, one of the main benefits of VC/PE involvement is the contribution of management expertise. The contribution of such expertise, especially at an earlier stage in the company’s life cycle, increases innovation outputs and quality following the arguments of Hypotheses 2a and 2b. So, while grant recipients already have received capital to invest in innovation and knowledge building, our argument is that both VCs and PEs further increase innovation inputs, outputs, and output quality.
In a similar manner to the argument for earlier investment, we contend that the pre-screening and credible signals of quality of a government grant reduce perceived investment risk, thereby encouraging both VCs and PEs to invest larger amounts in innovation in these firms. The receipt of a grant helps to mitigate information asymmetry. Reducing information asymmetry enables VC/PEs to make decisions that are more informed, thereby reducing investment risk by reducing the likelihood of miscalculating the expected returns and the riskiness thereof. VC or PEs also provide the management resources necessary to capitalize on the company’s latent innovative potential. In this way, they support the firm to convert its knowledge and resources into innovation outputs and focus on outputs of high quality (c.f. Bena and Li, 2014). Therefore, we predict
Hypothesis 4a. VC funding complements government grants regarding innovation inputs, outputs, and the quality of those outputs.
Hypothesis 4b. PE funding complements government grants regarding innovation inputs, outputs, and the quality of those outputs.
3. Data and variables
We examine the impact of government grants, VC funding, and PE funding on IPO firms at the time of listing on the stock exchange. We focus on newly public firms which allows us to examine firms at a similar stage in their life cycle, allows us to obtain data on the firm’s financials and innovation portfolio, thus comparing ‘like with like’ firms.
3.1. Sample and data
Our empirical investigation focuses on the Australian market which provides several advantages. Latent macroeconomic factors that could drive innovation (Anokhin and Wincent, 2012) are subdued. In high-innovation regions (such as the U.S. and Europe), it can be relatively more difficult to identify whether funds cause an increase in innovation, or whether innovativeness arises due to other macroeconomic factors. That is, it is relatively more difficult to eliminate macroeconomic factors as an alternative explanation for observed high innovation levels. By contrast, Australia is a country that has strong sovereign governance and well-developed principles of corporate governance (Gallagher et al., 2013), but has historically featured relatively low levels of innovation (Gans and Hayes, 2010). Thus, in Australia, a relation between VC/PE funding and innovation is likely to reflect the impact of this funding; by contrast, in high-innovation countries the relationship between VC/PE funding and innovation may reflect other latent economic-growth factors. Further, the government grants in Australia are competitive, in contrast to semi-automatic R&D subsidies, 9 assuring that the grants possess market-based characteristics.
Our sample is comprised of 436 10 IPOs that listed on the Australian Stock Exchange (ASX) between January 1995 and December 2005. The sample composition by year is shown in Table 1. Our sample drops to 411 IPOs in the models that require VC/PE intensity variables. 11 We identify the VC/PE backed firms from Thomson VentureXpert, SDC platinum and shareholder information in prospectuses. We obtain IPO prospectuses from the Connect 4 and FinAnalysis database. We hand-collect firm characteristics from each company prospectus. Hand-collected data on government grants is from AusIndustry, patents from IP Australia and patent citations from the European Patent office. We obtain VC and PE attributes from VentureXpert. There are two major spikes in IPO-activity in our data: 1999-2000 (coinciding with the tech-boom), and 2005 (coinciding with favorable financial and market conditions). The time-based variation is consistent with prior evidence on time-variation in IPO-activity (Ivanov and Lewis, 2008).
Sample composition by year.
IPO: initial public offering; PE: private equity; VC: venture capital.
3.2. Dependent variables
We examine innovation inputs, outputs, and output quality. We measure innovation inputs by R&D expenditures. R&D expenditures are logical antecedents to innovation outputs. However, not all R&D produces tangible outcomes. Nonetheless, the level of R&D has been commonly used as a proxy for innovation inputs (Hitt et al., 1991; Lavie and Rosenkopf, 2006; He and Wang, 2009). R&D spend is measured as the natural log of one plus the firm’s R&D expenditures, In(R&D +1). We measure innovation outputs by examining the number of patents (Hall et al., 2005; Kotha et al., 2011) and the quality of innovation outputs by the number of patent citations (Ahuja and Katila, 2001; Hall et al., 2005; Joshi and Nerkar, 2011; Wu, 2012) at the time of the IPO.
The models control the log of the firm’s age in order to control the influence of time on the number of patents and citations. The models that examine innovation outputs also control the firm’s R&D spend, as R&D expenditure can drive patenting.
We acknowledge that the use of patents as a measure of innovative behaviour has received some criticism for being both under-inclusive and over-inclusive (Engel and Keilbach, 2007). It is under-inclusive because not all inventions are patentable. It is over-inclusive because firms might simply patent all patentable inventions, whether or not they are commercially viable. Nevertheless, using patents is still the dominant approach to measuring innovative output (Kortum and Lerner, 2000; Kotha et al., 2011). 12
3.3. Independent variables
Our independent variables are indicator (dummy) variables. The variable Grant has a value of one if the firm received a government grant at a time when it did not have VC/PE funding. In all cases where the firm receives both a grant and VC/PE funding, the grant can be seen to pre-date the VC/PE funding. We similarly create indicators for whether the firm is VC backed or PE backed, which equal one if the firm receives VC or PE funding (respectively) and zero otherwise.
Our control variables reflect VC/PE attributes and general firm-level variables, that may influence innovation. We also control for ownership and governance structure. Appendix A1 details each control and their measurement.
3.4. Summary statistics and univariate analysis
The summary statistics are listed in Table 2. Firms that are VC or PE backed generally have higher levels of innovation outputs (patents) and grant recipients have higher levels of innovation inputs (R&D), outputs (patents), and quality thereof (patent citations). We also find that 19.6% of all VC/PE backed firms are also grant recipients (27% of all VC backed firms and 4.7% of all PE backed firms). By contrast, 7.7% of all non-VC/PE backed firms are grant recipients. VC/PE-backed and non-VC/PE backed firms are of similar age and size, which is consistent with prior Australian evidence (Da Silva Rosa et al., 2003). The results suggest that Australian VCs do not take firms public at an earlier stage than non-VC-backed firms.
Summary statistics.
This table contains the sample means for the sample and for sub-samples of companies based on whether they have VC and/or PE funding or have received a grant.
PE: private equity; VC: venture capital.
The correlation statistics are in Table 3. There is a positive (and statistically significant) correlation between R&D and patenting. There is also a significant correlation between many of the control variables. However, we check that the results are robust to collinearity concerns by ensuring that they hold if we simply omit the controls, or replace them with a set of principal components (whose eigenvalues are at least one).
Correlation statistics.
This table contains the pairwise correlation statistics for the variables. Figures in brackets are p values.
PE: private equity; VC: venture capital.
4. Analysis and results
We first examine the relationship of grant, VC, and PE funding on innovation. We next examine whether grants encourage VC or PE funding. We close by presenting robustness tests where we mitigate issues associated with endogeneity and sample selection. We summarize the hypothesis predictions and findings in Table 4.
Hypothesis predictions and findings.
PE: private equity; VC: venture capital.
4.1. The influence of government grants, VC funding, and PE funding on innovation
This section presents the primary tests of our hypotheses. In all cases where the firm receives both VC or PE funding and a grant, the grant precedes the VC/PE funding. Thus, a positive coefficient on the interaction term of grants and VC/PE funding indicates that VC/PEs are able to build upon the latent innovative capability of the firm. The models for Hypotheses 1, 2, and 4 have the following form
where Innovation represents the measures of innovation (i.e. R&D spend, Patents, or Patent Citations),
Multivariate regressions for innovation inputs, outputs, and output quality.
This table analyzes the impact of grant and VC/PE funding on innovation inputs, outputs, and output quality. The models are Tobit models. All models include year dummies, ASX industry group dummies and a constant (suppressed), and use robust standard errors clustered by industry group. Brackets contain p-values and superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively.
PE: private equity; VC: venture capital.
We find support for Hypothesis 1 in columns 4–9. Government grants are associated with significantly higher levels of innovation inputs (R&D spend), outputs (patents), and higher quality outputs (patent citations). The results on the impact of grants on innovation outcomes (patents) are consistent with international evidence (Branstetter and Sakakibara, 2002; Bronzini and Piselli, 2016; Jaffe and Le, 2015).
Hypothesis 2a is strongly supported for innovation inputs with mixed support for output and output quality. VC funding is associated with significantly higher levels of R&D (columns 4 and 7). It is important to note that the coefficients on the direct effects of interacted terms, in columns 7, 8, and 9, are conditional on the other interacted terms being at zero. The models for patents (columns 5 and 8) indicate a positive relationship until the interactions are added. The coefficient −0.394 in column 8 indicates that firms with VC funding but without a grant are not associated with higher patents (outputs). Furthermore, direct effects in column 9 (coefficient −125.536), for firms with VC funding but without a grant, indicate lower levels of patent citations. In unreported tests that exclude all grant recipients from the sample, we also find that VC funding is negatively associated with patenting whether or not we control for other fund-level characteristics. Overall, this suggests that VCs focus more on increasing innovation inputs and facilitate innovation outputs mainly if the firm has already invested capital in innovation, as manifested by the receipt of a grant. Supporting Hypothesis 2b, PE funding significantly increases patenting and patent citations (columns 5, 6, 8, and 9); however, it is associated with lower levels of R&D. This finding is consistent with arguments that PEs focus on conversion of knowledge into innovation outcomes in later-stage companies.
Hypotheses 4a and 4b are supported for innovation outputs and output quality but findings are mixed for innovation inputs. Thus, VC and PE funding are generally complements to government grants. The positive coefficients on the interaction terms “VC backed × Grant” and “PE backed × Grant” in columns 8 and 9 indicate VCs and PEs are able to build upon government grants to encourage innovation outcomes. The evidence in column 7 indicates that grants tend to reduce the amount that VC-backed firms spend on R&D. This indicates that for VCs, prior grant funding acts as a substitute for innovation inputs. A deeper investigation of these results can be accomplished by considering the direct and the interaction effects. In firms without a grant, the net effect of VC funding on R&D is 0.286 (the coefficient on the VC-backed indicator). In firms that have a government grant, the net effect of VC funding on R&D is 0.008 (= 0.286 – 0.278). Thus, while VCs invest less in R&D in firms that have a grant, VC funding is positively associated with R&D for both grant recipients and non-recipients. Interestingly, Model 9 indicates that VC support results in higher output quality for firms having a grant and less for firms without a grant, suggesting that VCs build upon the firm’s latent innovative resources. PEs support increases in R&D expenditure for both grant recipients and non-recipients. Overall, these results suggest that VCs and PEs are able to build upon the foundation provided by government grants to increase innovation, at least in regard to innovation outputs and output quality.
4.2. The influence of government grants on VC funding and PE funding
To examine Hypothesis 3 (a and b), we analyze the factors that determine VC/PE funding. We use logit models where the dependent variable is an indicator that the firm is VC backed or PE backed. The model includes year dummies, industry group dummies, and cluster standard errors by industry group (following Petersen (2009)) and has the following form
where VC/PE Funding is an indicator in three separate models that the IPO firm receives either VC or PE funding, PE funding, or VC funding, Grant is an indicator that the firm received a government grant,
The results are shown in Table 6. The receipt of a grant significantly increases the likelihood of receiving VC funding (indicated by the significant, positive coefficient on the Grant variable in column 2). The receipt of a grant does not significantly increase the likelihood of PE funding (indicated by the insignificant coefficient on the Grant variable in column 3) but does not significantly decrease it either. VC/PE sector activity (VC/PE intensity) significantly increases the likelihood of either VC or PE funding. There is some evidence of a quadratic (inverse U-shaped) between PE funding and firm size. This suggests that PEs are not only less likely to invest in very small firms (suggesting some aversion of highly speculative firms) but are also less likely to invest in very large firms (possibly suggesting some preference for innovative growth). Furthermore, the industry-average profitability (Ind Ave Return on Equity) is negatively related to VC funding. This implies that VCs prefer low-ROE industries, which are typical of start-up industries. Overall, these results provide support for Hypothesis 3a that grants influence VC funding. However, contrary to Hypothesis 3b, government grants do not significantly influence the likelihood of PE funding.
Determinants of VC/PE funding.
This table examines the factors that determine VC/PE funding. The dependent variable is an indicator that equals one if the firm receives VC or PE funding, as indicated in the column title. The model is a logit model, include year dummies and industry dummies, and use robust standard errors clustered by industry group. The model also includes a constant term. Brackets contain p-values and superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively.
PE: private equity; VC: venture capital.
We also consider the results of our controls for the attributes of VC/PEs—size, syndication, overseas location, and portfolio spread (number of companies divided by size of the VC/PE firm). First, large VC/PEs are associated with increases in innovation outputs and output quality. This result is consistent with the idea that larger VC/PEs tend to be more experienced and better resourced, placing them in a better position to contribute both financial and management resources to the portfolio company. Second, syndicated VC/PEs are associated with increases in innovation inputs and with decreases in innovation outputs and output quality. This is consistent with the arguments of Chahine et al. (2012) who highlight that VC syndication can create a “bystander” effect whereby individual VCs assume that the other VCs engage in monitoring/value creation, resulting in individual VC/PEs free-riding and causing an overall lack of monitoring. Third, VC/PEs from overseas locations are associated with increases in innovation inputs and with decreases in innovation outputs. This is consistent with the view that geographic distance impedes collaboration (between companies) and the exchange of information (Pirinsky and Wang, 2006; Ragozzino and Reuer, 2011), which would be necessary for the VC/PE firm to convert innovation inputs into quality outputs. Fourth, VC/PE portfolio spread is associated with decreases in innovation inputs, outputs, and output quality. This finding provides further support for the “limited attention” problem where the VC/PE cannot devote high levels of attention to each individual firm because they spread their resources across too many firms (Lopez-de-Silanes et al., 2015).
4.3. Additional robustness tests
We undertake several tests to mitigate endogeneity concerns and/or sample-selection effects. We run two sets of two-stage least-squares analyses (Greene, 2008; Wooldridge, 2002) with different instruments. The method and instruments are described in Supplementary Appendix A2. The results are shown in Table 7. The VC and PE funding variables have the same sign and similar levels of significance to those in the main models. This suggests that our results are robust to endogeneity concerns and lend some support to the idea that VC and PE funding leads to innovation.
Two-stage least-squares models.
This table contains regression models that control for endogeneity. The first set of 2SLS models (columns 1–3) replace the VC/PE variables with the predicted values from an estimation of the logit model in Table 6 (with the grant-dummy omitted from the model). The second set of models (columns 4–6) use a model similar to that in Samila and Sorenson (2011). Supplementary Appendix A2 describes the instruments used in the first stage regression. All models contain year dummies, industry dummies and a constant, and use robust standard errors clustered by industry group. Brackets contain p-values. Superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively.
PE: private equity; VC: venture capital.
A core endogeneity concern is that VC and/or PE firms simply cherry-pick innovative firms before they undertake the IPO, meaning that the relation between VC/PE funding and innovation is an artifact of selection rather than causation. The critical concern in our analysis is time because time is required between R&D expenditure and patents and between patents and patent citations. Thus, we mitigate this by examining sub-samples where we exclude any VC-/PE-backed firm that we can identify as having received its first round of capital within 1, 2, or 3 years of the IPO (for three different sub-samples). The results shown in Table 8 also help to mitigate concerns over the limitations of the 2SLS approach to endogeneity concerns (Reeb et al., 2012; Semadeni et al., 2014). The relationship between VC/PE funding and innovation in these sub-samples is similar to the core results. In Table 9, we examine indicators that equal one if the firm received VC or PE funding at least 1, 2, or 3 years before the IPO. We find that the longer the VC/PE firm has invested in the firm, the greater is its contribution to innovation. This is consistent with the VC-/PE-innovation relation holding more for longer-horizon investments, where the VC/PE can plausibly contribute more to the firm’s innovation.
Excluding firms with under 1, 2, or 3 years investment.
This table contains models that focus on the time for which a firm has received investment. Columns 1–3 exclude any firm that we can identify as having received under 1 year of investment from the VC/PE firm. Columns 4–6 and 7–9 do similarly but for 2 years and 3 years, respectively. The models are Tobit models with a lower bound of zero. Brackets contain p-values and superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively. All models include year dummies, industry group dummies, and a constant (suppressed) and use robust standard errors clustered by industry group.
PE: private equity; VC: venture capital.
Examining time of investment.
This table contains models that examine indicators for whether the company received funding of a venture capital (VC) or private equity (PE) firm for at least 1, 2, or 3 years (for three separate indicators). We construct these indicators by identifying, from VentureXpert, the companies that have received under 1, 2, or 3 years of investment. The indicators equal one if we can identify that the company has received under that many years of investment and equal zero otherwise. The models are Tobit models with a lower bound of zero. The brackets contain p-values and superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively. All models include industry dummies and year dummies and a constant (suppressed) and robust standard errors clustered by industry group.
Furthermore, we conduct falsification-type tests. We expect that the hypothesized benefits of VC, PE, and grants would not increase with a firm’s profitability. We interact the firm’s profitability (proxied by its return on equity), with the VC, PE, and grant in Table 10. However, as patents and citations do not directly involve expenditure, it is only plausible to examine R&D. The interactions of PE and grants with ROE are significantly negative and the VC interaction is insignificant. This indicates that the benefits of VC, PE, and grants to R&D do not increase with the firm’s profitability. That is, firms that do not need VC, PE, or grants benefit less from them which suggests a causal relationship in our results.
Falsification tests.
This table contains models that examine the relationship between R&D expenditure, ROE, and the receipt of VC/PE funding or a grant. The models are Tobit models with a lower bound of zero. The brackets contain p-values and superscripts ***, **, and * denote significance at 1%, 5%, and 10%, respectively. All models include industry dummies and year dummies and a constant (suppressed) and robust standard errors clustered by industry group.
PE: private equity; VC: venture capital.
We take steps to address sample selection issues using the propensity score approach, detailed in Supplementary Appendix A2. We also examine sub-sample regressions of firms whose size is in the top 50% of the sample. We test whether VC/PE funding increases innovation in a set of firms (large firms) that are less likely to be innovative. The results are qualitatively similar to our main findings.
We further take steps to mitigate concerns over the modeling method. First, given that the number of patents and number of citations are count data, we ensure that the results are robust to using a Poisson or negative binomial model. The results are also robust to using an ordinary least squares (OLS) model. Second, the reported models include indicator variables and cluster standard errors by industry, consistent with the approach to control for unobserved heterogeneity suggested in Petersen (2009) and Gormley and Matsa (2014). Nonetheless, we also obtain similar results if we use “industry adjusted” dependent variables (which subtract the industry average for the dependent variable from the firm’s value of that variable). We also obtain similar results to the baseline patents and patent citations models if we examine patents or patent citations scaled by the firm’s R&D spend. We also mitigate concerns that the output results (patent citations) merely reflect the firm’s self-citations to its own patents, using an “adjusted” citation measure equal to the number of citations less the firm’s number of patents.
We take steps to address issues with interpreting coefficients in non-linear and limited dependent variable models, such as Tobit (Hoetker, 2007; Wiersema and Bowen, 2009). The Tobit model is essentially a linear model in which the distribution of the dependent variable is truncated. Thus, the concerns in relation to marginal effects are less severe than with logit or probit models. Nonetheless, in Table 11, we provide marginal effects, standard errors, and p values in which we explicitly account for the distributional assumptions of the Tobit model. We do this for the VC-/PE-related variables in columns 7–9 of Table 6. The main finding is that the signs and significance of the core variables are similar.
Marginal effects for the results in Table 5.
This table contains marginal effects, standard errors, z-statistics, and p-values computed in a way that explicitly accounts for the truncation of the dependent variable in Table 5. These are computed directly after running the models that generate columns7–9 in Table 5 (as indicated in the panel title). They are computed with the command “mfx, predict(e(0,.)) varlist(X)” in STATA after running the Tobit model, where X denotes the names of the variables for which marginal effects are requested. The results are qualitatively similar if we use the STATA command “margins, dydx(X),” where X represents the list of variables.
PE: private equity; VC: venture capital.
5. Conclusion
We examine the influence of government grants, VC funding, and PE funding on innovation in new venture (pre-IPO) firms. Each of these funding sources comes bundled with an additional set of resources, which we hypothesize can contribute to the investee’s innovative success. Firms have a choice of funding sources (Hsu, 2004), sources also have decision discretion (De Treville et al., 2014), and sources of capital can provide different non-pecuniary resources (Park and Steensma, 2012). We use the logic of information economics and the knowledge-based view and hypothesize the mechanisms whereby government grant funding influences innovation and subsequent VC/PE funding prior to IPO.
We find that the presence of a government grant has implications both for the firm’s capacity to engage in innovation and its ability to attract additional funding. The presence of a grant is complementary with the impact of VC and PE funding on innovation outputs and on output quality. This highlights that the grant, by mitigating issues of information asymmetry, can help VC and PE firms to identify innovative portfolio companies, whose innovations they can help to “bring to market.” However, we find that for grant recipients, VC funding is a substitute for innovation inputs (R&D), while PE funding is a complement. The finding on VCs is unexpected. Our argument on PEs is that the signal of quality from a government grant supports the decision to invest earlier than would typically occur. This results in PEs taking on portfolio companies that will still benefit from higher R&D investment. For VCs, we offer a supposition that the signal from a government grant does more than reduce information asymmetry. It conveys prior investment in R&D inputs such that the VC can quickly capitalize on the existing knowledge advantage by supporting innovation outputs and output quality as the means by which the VC can create and capture value. This would be especially the case for VCs (as opposed to PEs) as VCs tend to invest smaller amounts of money in each company than do PEs, resulting in greater capital constraints in VC-backed companies. Thus, VCs would focus on investing capital on commercialization, which is where the probability of a large gain is higher, when the company has already invested sufficiently in R&D.
We find that while government grants encourage VC funding, they do not lead to subsequent PE funding. The differential impact of grants on VC and PE funding likely reflects the tendency of VCs to invest in early-stage firms, which typically feature greater issues of information asymmetry, and for which the signaling benefits of grants are likely to be more important. Government grants act as a credible signal of the firm’s innovativeness and quality and this signal has powerful influences.
Our approach contributes to the literature as it considers the role of government grants in attracting subsequent VC and PE funding; separately examines both VC and PE funding, whereas prior studies tend to focus on VC funding only; investigates the interactive effects of government grants with both VC and PE funding which lacks investigation; and examines innovation inputs, outputs, and output quality, whereas prior studies typically examine only one or two aspects of innovation. As a specific insight into the mechanism regarding innovation outcome quality, we use the knowledge-based view to argue for the signaling value of retention, as well as procurement, of top research staff. Our investigation of innovation complementarity between forms of funding is novel. Our work is consistent with, but extends, recent work arguing that pre-IPO funding sources provide a signal to subsequent investors after IPO (Ragozzino and Reuer, 2011).
We also contribute to the specific literature on government grants, by showing that VC/PE funding and government grants are generally complements on innovation. However, grants and VC funding are substitutes on innovation inputs (R&D expenditure). We present logic for this substitute’s relationship based on the theorized value creation objectives of VC firms in conjunction with the signal of the government grant.
We further contribute to the literature by examining the discernible attributes of VC/PEs and their relationships with innovation. The literature on VC and PE funding has acknowledged the heterogeneous nature of VC and PE firms (Bruton et al., 2010). While prior work has examined heterogeneity of VC/PEs with regard to financial performance, innovation performance is understudied. Thus, our results provide an empirical foundation for future work in this area.
Our results are of interest to managers of new venture firms and managers of VC/PE firms (as well as policy makers who have increasingly focused on avenues through which to increase corporate innovation (Neville and Sorensen, 2014)). Our empirical results indicate that the heterogeneity in VC/PEs also significantly influences innovation in portfolio companies. 13 Managers of new venture firms may gain a different view of the value and strategic benefit of government grants and managers of VC and PE firms may gain additional insight into the evaluation of potential portfolio companies.
However, this study is not without limitations. First, as previously acknowledged, we use a sample of IPO firms and patent data to indicate innovation output and patent citations to indicate innovation quality. Future researchers could investigate alternative measures of innovation output and output quality, such as trademarks and/or designs. Second, our data are cross-sectional. A longitudinal analysis could provide insight into the persistence of the effects of grants and VC/PE funding. Third, it is foreseeable that the impact of grant schemes in encouraging VC/PE funding, and innovation, will vary with the details of those grants. Thus, future research could further consider how the role of grants, VC, and PE firms varies across countries with different legal regimes, intellectual property laws, and grant schemes.
Supplemental Material
AJM-16-0288.R2_Main_Doc_with_AE-AD – Supplemental material for Innovation in newly public firms: The influence of government grants, venture capital, and private equity
Supplemental material, AJM-16-0288.R2_Main_Doc_with_AE-AD for Innovation in newly public firms: The influence of government grants, venture capital, and private equity by George A Shinkle and Jo-Ann Suchard in Australian Journal of Management
Footnotes
Acknowledgements
This paper benefited from comments from Mark Humphery-Jenner, Martin Kenney, Brian McCann, Elaine Hutson, Rik Nuenighoff, Ronan Powell, Zach Sautner, Bart Sharp, and Jason Zein. We also thank the seminar participants at UNSW, the Third Innovation Academics’ Workshop on “Evaluation and Innovation” (2013), the Workshop on Venture Capital and Private Equity in the Asia Pacific Region (2013), the Finance Research Network (FIRN) Annual Meeting (2013), the ACERE meeting (2014), and the Asia-Pacific Innovation Conference (2014). We also thank Cambridge Associates LLC for providing us with their survey of venture capital and private equity activity in Australia. All authors contributed equally; authors are listed alphabetically.
Final transcript accepted 29 August 2018 by Karen Benson (AE Finance).
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We thank the Australian Research Council (Grant DP140103039) for financial assistance.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
