Amsterdam Business School - MSc Finance Master Thesis Corporate Finance
Supervisor: Dr. T. Caskurlu
Innovation after a Private Equity Takeover:
The Impact on R&D and Knowledge Acquisition
Floor Sougé
1July 1, 2018
Abstract: Private equity is expected to outperform the market in order to meet the required risk-returns by (institutional) investors, which puts pressure on increasing the underlying portfolio companies’ value. This paper investigates the impact of private equity on the innovation strategy after a takeover by comparing materialized private equity transactions to withdrawn private equity transactions. Innovation strategy can be described as a plan made by companies to encourage advancements in technology or services and makes a distinction between internal (autonomous R&D) and external innovation (knowledge acquisition). The empirical research consists of a difference-in-difference model using U.S. private equity and patent data from 1997 to 2017. The results find that there is noticeable evidence for an increase in knowledge acquisition, but that private equity ownership does not impact internal innovation. Therefore, this research suggests that the innovation strategy after a private equity takeover shifts from an internal innovation strategy to an external innovation strategy.
Keywords: private equity, innovation strategy, R&D, knowledge acquisition
1
Statement of Originality
This document is written by Student Floor Sougé who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2 Table of content Introduction ... 3 Literature review ... 7 Hypothesis ... 14 Methodology ... 16 Data ... 18 Data collection... 18 Sample construction ... 19 Descriptive data ... 21 Results ... 26 Robustness tests... 33 Concluding remarks ... 38 Limitations ... 38 Conclusion ... 38 Recommendation ... 40 References ... 41 Appendices ... 44
Appendix A: Variable calculations ... 44
Appendix B: Descriptive data ... 45
3 Introduction
Private equity as an alternative ownership structure – compared to for example an initial public offering (hereafter: IPO) – has been growing rapidly over the last twenty years and has been outperforming the market (Flood, 2017; Bain & Company, 2018). This trend is expected to continue, however, critics are sceptical as the return on investment is decreasing over the years. Although with an expected average annual return of 9.4% for the next decade, which comes down from 11.2% in 2009, private equity remain better performing than any other asset class (Flood, 2017). Private equity investments are currently back to a ‘pre-financial crisis’ 2007 level and Apollo Global closed its last buyout fund last year with a record amount of $24.7bn (Flood, 2017). Yet, with the high influx of private equity capital, there is the risk of an overheated market resulting from the quest for good-yielding investment opportunities by private equity firms. This puts pressure on the private equity investments teams to meet the expectations of their (institutional) investors. Consequently, the pressure is forwarded onto private equity’s portfolio companies to realize results and create value. Therefore, these companies need to innovate to continuously improve their operating performance.
This paper sheds light on the ongoing debate about private equity performance on both fund and portfolio level, and more specific how this performance improvement is achieved through innovation. On fund level, some studies claim that private equity has been structurally outperforming the market (Kaplan & Schoar, 2005; Kaplan & Strömberg, 2009; Harris, Jenkinson & Kaplan, 2014; Gompers, Kaplan & Mukharlyamov, 2016). Other studies claim that these findings are not correct, as the data is incomplete and risk is not accurately measured. Adjusting for these considerations in private equity fund analysis, has led to arguments of underperformance relative to the market (Moskowitz & Vissing-Jørgensen, 2002; Kaplan & Schoar, 2005; Phalippou & Gottschalg, 2009; Braun, Jenkinson & Stoff, 2017). On portfolio level, certain studies claim that private equity increases value of the portfolio firm through operating and governance improvements and tax benefits (Kaplan & Strömberg, 2009; Boucly, Sraer & Thesmar, 2011; Bernstein, Lerner, Sorensen & Strömberg, 2017; Liu, 2017). Others argue that increase in profitability comes at a price and private equity rather destroys value (Lichtenberg & Siegel, 1990; Davis, Haltiwanger, Handley, Jamin, Lerner & Javier, 2014). The short term interest of private equity firms is linked to asset disposal and excessive job losses in order to create profitability. This research focuses on the impact of private equity firms on portfolio company’s innovation strategy.
4 The innovation literature landscape describes two innovation strategies, which are evaluated in this paper related to private equity takeovers. Companies need to innovate in order to remain competitive, remain profitable, increase market share and/or – especially from a private equity perspective – to realize investor’s return objectives. Innovation strategy, being defined as a plan to encourage advancements in technology or services, makes a distinction between internal innovation (independent and autonomous research and development; hereafter: R&D) and external innovation (knowledge acquisition). Internal innovation is advantageous for companies due to potentially a high return on investment, tailor-made application and disruptive power. However, investments in R&D are sometimes difficult to finance due to asymmetric information issues. Autonomous R&D investments often come with large initial investments, opportunity and adverse selection costs and return on investment only on the medium to long term (Hall, 2002; Li, 2011; Sears, 2018). From a private equity perspective to increase profitability on the short term, the internal innovation strategy is possibly not always appealing. The second innovation strategy, external innovation, means acquiring know-how from outside sources instead of in-house R&D and implement these acquired innovations into the existing business. This is also referred to as knowledge acquisition. The main advantage of knowledge acquisition is that, to a certain extent, it provides a ‘proven’ effectiveness, which diminishes opportunity costs and the time constraint of in-house R&D (Sears, 2018). Due to synergy potential and avoiding R&D duplication, external innovation is positively related to innovative outperformance.
The link between private equity firms’ impact on innovation is profound by change of ownership, management and/or the board. New management tend to contribute more to a new corporate strategy, supporting private equity firms’ aim to increase a portfolio company’s value in a limited time. Research indicates that also new board members have a positive influence on innovation compared to companies remain with their existing boards (Fauver, Hung, Li & Toboada, 2017). After a private equity takeover the majority of the management teams is on average replaced, unless the existing management possesses outstanding managerial skills or specialized skills. Private equity firms are found to have better managerial skills compared to traditional family-, founder- or governmental companies. Outstanding managerial skills are beneficial for company growth, business development and attracting talent (Bloom, Sadun & van Reenen, 2015).
This paper contributes to the existing literature to fill the gap by including the impact on both internal and external innovation strategy after a private equity takeover. It extends the literature on the impact of internal innovation after an acquisition (Seru, 2014) or more
5 specific after a leveraged buyout (hereafter: LBO) (Lerner, Sorensen & Strömberg, 2011). It complements Liu (2017), who investigates the determinants of private equity’s portfolio company growth, but does not specifically include innovation strategy in his paper. Moreover, this paper relates to Bernstein, who examines both innovation strategies after an IPO, allowing to make a comparison between private equity portfolio companies and listed companies. To the best of my knowledge, this particular research has not been done before.
Formulating this literature gap into a research question, the following question is proposed: do private equity firms influence the innovation strategy after a takeover? A thorough answer includes the two different perspectives on innovation strategy; internal innovation and external innovation. Based on the existing literature and the following assumptions, hypotheses are formulated as a tool to answer the research question. Private equity’s focus of maximizing (short term) return leads to cost cutting, which among others may limit the R&D expenditure. A more (short term) cost-cautious approach to R&D expenditure is expected not to have a negative impact on the result proxy of R&D, i.e. patents, due to improved efficiency. For this assumption it holds that the amount of patent applications does not change, however a higher percentage of the patents is granted and the quality and economic value of patents increases. Recalling that internal innovation often need to reckon with high initial investment costs, opportunity costs and return on investments often only on medium to long term, one may argue that private equity firms may want to shift their focus – on the assumption it is more cost-efficient – to external innovation.
This paper examines an event study about the impact of private equity ownership on the innovation strategy of the portfolio companies, for which a difference-in-difference analysis is conducted. In order to answer the aforementioned research question, the empirical research investigates three dimensions of the innovation strategy after a private equity takeover: R&D expenditure, efficiency and quality in the results of R&D (i.e. patents) and knowledge acquisition. This analysis includes a total of 588 reported deal announcements between 1997 and 2017 in the U.S. market, meaning that both the target company and the private equity firm as acquirer were based in the United States. A balanced timeframe of five years before and after the deal announcement is applied to incorporate the full impact of private equity ownership.
The experiment consists of two groups: a treatment group and a control group. The treatment group consists of successful private equity takeovers and the control group covers the transactions that did not materialize (withdrawn deals). An important requirement for the sample construction was that the private equity firm obtained a controlling share after the
6 deal, to correctly measure the impact. Therefore, it is required that the private equity firm owns at least 50% of the shares after the deal, but was not the majority shareholder before the deal. The treatment group consists of 480 materialized transactions.
This research has to overcome endogeneity issues, as withdrawing a deal influences the company’s public reputation. To discover the reason for a deal to be withdrawn, articles from newspapers were consulted. This research makes a distinction between withdrawn deals for internal criteria and external criteria. Internal criteria for inclusion in the exogeneous dataset are, amongst others, not being able to set an appropriate price for all parties concerned or unsuccessful negotiations on other matters. The external criteria to remove deals from the endogenous dataset to create an exogenous dataset are: no information found, overbid by another acquirer, regulatory issues or lack of financing. The control group consists of 108 withdrawn transactions, of which 57 are also included in the exogenous control group.
The results are checked for robustness over three different axes. First, the event study in the main research consists of a balanced timeframe of five years prior and five year after a deal announcement. To evaluate the timing of the impact, the research is also conducted in a balanced three year framework. Second, to confirm the results are robust for any timeframe and rule out the possible time trends, the dataset is split in half. The empirical model is applied for deal announcements between 1997 – 2006 and 2007 – 2017. The former timeframe relates closer to the existing literature, and the latter timeframe shows newer results. Third, the level of innovation within a company depends strongly on the industry, therefore, a distinction is made between low- and high innovative industries to test the impact of a private equity takeover. This robustness examines a potential bias of high innovative companies overruling low innovative companies in the dataset.
The rest of the paper proceeds as follows: the first section reviews the existing literature. Based on the literature review, the different hypotheses are formulated. In the second section, the research design is explained in detail. This consists of the methodology, data collection and sample construction. In the third section of this paper, the results of the empirical research are presented and discussed. This also includes the analysis to control for robustness. The fourth section gives the overall conclusion and further recommendations. References and additional explanatory tables are found in the appendices.
7 Literature review
The literature review provides an in-depth inside in the private equity and innovation strategy literature landscape. Following a brief introduction of private equity in the U.S. market, the dynamics of private equity firms vis-à-vis its (institutional) investor in relation to the (short term) return requirement are explained. Continuing, the constraints and challenges the private equity sector may be facing, in particular the impact of change of ownership (and management) on innovation are discussed. Also, the determinants of innovation are considered. The final part of the literature review explains financing innovation. This closes the circle as it returns to private equity investments.
After the private equity rise in the 1980s, corporate investments financed through private equity largely vanished in the 1990s. Private equity returned after the millennium switch and became increasingly popular. The year 2007 accounted a record amount of $369bn invested in private equity funds. Nevertheless, after the financial crisis fund volume and deals declined and it took more than a decennium to recover and to return to equal levels again. Liu (2017), Robinson and Sensoy (2016) and Kaplan and Schoar (2005) explain the fluctuations in private equity investments through the pro-cyclicality of the funds related to the market, meaning that private equity fund performance shows a similar trend compared to any other financial instrument in the market. On the contrary, Braun et al. (2017) and Hooke and Yook (2016) argue that the private equity market is a mature industry and very competitive for good-yielding investment opportunities.
The different private equity firms have a similar business structure, following the same five phases of: fundraising, acquiring portfolio companies, increase performance and value of portfolio companies, exit the portfolio companies and, finally, close the fund. First, fund raising can be targeted to different types of investors; varying from institutional investors like pension funds or insurance companies to family offices and high net-worth individuals. Also, so-called ‘fund of funds’ investors invest in individual private equity funds. The justification for these medium to long-term, often less liquid investments is the projected above average returns. Second, private equity firms make direct investments by acquiring target companies or by taking majority or (strategic) minority stakes in these target companies. As strategic buyers typically overbid financial acquirers, acquiring companies is challenging. Gorbenko and Malenko (2014) explain these higher bids through the synergy potential for the strategic
8 buyer, which the financial acquirer does not have. Third, key focus is creating value by maximizing the growth potential of a portfolio company. Kaplan and Strömberg (2009) claim that private equity investments contribute to improvements in financial, governance and operating performance. This is elaborated further on in the literature review. Fourth, private equity firms exit a portfolio company once the projected value increase has been realized. An exit occurs usually after approximately five to six years (Kaplan & Strömberg, 2009). The portfolio company is about two-third of the times sold to a strategic buyer or another investor, i.e. secondary buyout. Another option for a private equity to exit a portfolio company is through an IPO, which holds for fourteen percent of the cases. Six percent of the portfolio companies goes bankrupt during the private equity ownership (Kaplan & Strömberg, 2009). Fifth, the fund is closed. These private equity funds are closed-end, meaning investments will only be returned after all the invested portfolio companies are exited. On average, the funds are closed after ten years, as it takes this amount of time to go through all the phases (Braun et al., 2017).
Certain studies claim that private equity funds structurally outperform the market, others are more sceptical on the performance and draw a more critical view on the future implementation. This section discusses the constraints and challenges private equity firms are facing in this respect.
The rationale for investors to invest is the assumption that private equity investments outperform the public equity sector (Kaplan & Schoar, 2005; Harris et al., 2014; Gompers et al., 2016). Harris et al. (2014) find that the returns on private equity investments are 20% higher compared to the S&P 500 index, which relates to an annual outperformance of 3%.
However, there is an ongoing debate whether the outperforming returns are manipulated and too optimistic due to inflated accounting valuation and biased sample selection (Phalippou & Gottschalg, 2009). In addition, Gompers et al. (2016) argue that academic research about private equity has its limitations, as most of the portfolio and investment data is not disclosed. Also, Korteweg and Sorensen (2017) show that the difference between the top and bottom quartile return can differ up to 8 percentage points annually. Such a large differentiation in returns for different funds can be a reflection of, among others, fund size, type of fund, or risk profile of the fund. Diversification or specialization of the fund could also influence the return on investment, because, according to portfolio theory, diversification decreases risk and increases returns (Lossen, 2007). Private equity firms invest the funds on behalf of their investors for which the private equity firms
9 receive a management fee over the capital raised and often a success fee over the returns realized. Phalippou and Gottschalg (2009) claim that based on the net return to investors, including the different fees, private equity underperforms by 6% compared to the market. Continuing, Moskowitz and Vissing-Jørgensen (2002) and Harris et al. (2014) emphasize on the illiquidity and (idiosyncratic) risks of private equity, for which a premium should be paid. Taken this into account, the difference between a (reported) gross return and a (real) net return after fees and risk premium, can be substantial.
Moskowitz and Vissing-Jørgensen (2002) therefore rise a valid question: why would investors still invest in private equity funds? Braun et al. (2017) share this view and address a potential negative snowball effect. The relative poor performance of private equity firms makes investors hesitative to invest, which makes it harder to raise new capital. This creates additional pressure on the private equity investment teams, constantly aiming for better-yielding investment opportunities. However, an increased risk profile may present a sliding scale that eventually may lead – despite (projected) high returns of individual investments - to disappointing returns on fund level because of a poor risk-return ratio. As a consequence, private equity firms cannot simply assume that investors, investing in previous funds, would reinvest in new funds, which used to be a common practice (Braun et al., 2017). This puts even more pressure on private equity fund performance. These findings are in line with Kaplan and Schoar (2005), who present similar findings from the opposite point of view: firm performance and fundraising are positively related, meaning that for well performing private equity firms, fundraising is relatively easier. That said, it should be noted that the relation between firm performance and fundraising is concave.
To handle aforementioned issues and overcome the pressure of the investors, private equity firms tend to invest in poor performing companies with growth potential. This creates opportunities for performance improvements. This section discusses the target company selection by private equity firms and private equity’s impact on company value.
Private equity captures approximately 4% on average in each industry, measured by sales (Bernstein et al., 2017). Liu (2017) describes the target identification as “cherry-picking” undervalued companies. Gorbenko and Malenko (2014) find that private equity targets mature and poor performing companies. Dittmar, Li and Nain (2012) formulate similar findings differently: private equity targets companies that show high potential for value improvement. Combining the three views constructs a pool of typical private equity portfolio companies. These companies do not meet their value potential yet, but do have
10 upscale potential. Private equity firms aim to make a turnaround on these companies. Partners or executives in a private equity firm have strong incentives to invest in potential growth companies, as their pay check depends on the returns (Axelson, Strömberg & Weisbach, 2009). Acharya, Gottschalg, Hahn and Kehoe (2013) find that private equity partners with a consultancy or industry background are more persistent in increasing company operating performance.
Private equity portfolio companies tend to grow faster compared to its peers (Boucly et al., 2011; Bernstein et al., 2017). Boucly et al. (2011) find that this growth leads to more profitability which causes better access to issuing debt and increase of the capital expenditures. Private equity firms add value to the portfolio company by providing ‘an invisible stamp’, which lowers credit constraints (Boucly et al., 2011). Liu (2017) also finds that private equity adds value to the portfolio company through increase in operating earnings, operating cash flow and profitability. In addition, Guo, Hotchkiss and Song (2011) find that the increase in leverage results in a larger tax shield, leading to increase in available cash flows.
Nevertheless, private equity firms are ‘accused’ of destroying value rather than creating it. As profitability is measured as EBITDA2 over total assets, Lichtenberg and Siegel
(1990) find that short term profitability is increased by selling assets and, thus, cutting the total assets. Liu (2017) criticizes this claim, as he does not find results of asset disposal. Furthermore, the profitability increase of portfolio companies does not only increase through revenue increase, but also due to cost reduction. This includes job losses (Davis et al., 2014). The average job loss holds for 6% over five years after a private equity takeover compared to a same industry control group. Contradicting to their own results, Davis et al. (2014) also find that private equity portfolio companies have relative high job creation levels, due to job reallocation.
Addressing the change of ownership and potentially change of management in a takeover, private equity impacts the operating business of portfolio companies including the innovation strategy. This section discusses the rationale for change of management and its impact on innovation.
Change of shareholder, in management and/or in the board is reflected in the corporate strategy of the company: new people, new vision. Seru (2014) investigates the
11 change of R&D activity after a merger or acquisition in a conglomerate, leading to a conclusion that the quality of the patents decreases. The quality of patents, measured by patent citations, denotes the economic power of the innovation (Hall, Jaffe & Trajtenberg, 2005) The decrease in patent quality is in contrast with the findings Lerner et al. (2011). Lerner et al. (2011) research a different type of acquisition, namely a leveraged buyout. Even though the level of patent applications does not change over time, the quality increases. In short, less but more meaningful patents are submitted after a LBO. Bernstein (2015) also finds a decrease in patent application after an IPO, as innovation strategy is transferred to an acquisition strategy.
For a private equity firm, the management of its portfolio companies is crucial for its success. While investing in a new portfolio company, a private equity firm has two options: continue with and incentivize the existing management team or replace the existing management team. In the first scenario, the portfolio company’s management team is sometimes offered to co-invest and become shareholder to create a strong incentive to perform. Management could receive a large upscale, but are only able to sell their shares after the private equity has exited the company to create long-term commitment. Bargeron, Schlingemann, Stulz and Zutter (2017) finds that in this scenario a higher premium (10%-18%) is paid in the takeover. The higher premium is explained by the high level of human capital of the CEO. In the second scenario, the existing management is replaced in order to improve the company’s overall performance or realize a certain turnaround. Kaplan and Strömberg (2009) claim that two-third of the portfolio company’s CEOs is replaced within four years after a takeover.
Kaplan, Klebanov and Sorensen (2012) find general ability and execution skills to be more positively related with company success than interpersonal skills. Bloom et al. (2015) confirm the managerial outperformance of private equity portfolio companies compared to family-run, founder- and government owned companies, based on performance monitoring, effective targets and performance incentives. Well managed companies grow faster and are able to attract higher skilled employees (Bloom et al., 2015). As management makes the company’s decisions, amongst others about the innovation strategy, management performance is vital for private equity portfolio companies. Moreover, change in the board is found to be increasing company value (Fauver et al., 2017), underlining the turnover rate after a private equity takeover.
12 One of the key aspects to meet private equity’s requirements for performance improvement is innovation. This section focusses on the definition, determinants and financing of internal innovation.
Innovation is best described as the realization of ideas that create value for companies by means of products and services. Companies need to innovate in order to remain competitive, remain profitable, increase market share and/or – especially from private equity perspective – to realize investors’ return objectives (Kogan, Papanikolaou, Seru & Stoffman, 2017; Ortiz-Villajos & Sotoca, 2018). Innovation strategy makes a distinction between internal innovation (independent and autonomous R&D) and external innovation (knowledge acquisition) (Cassiman & Veugelers, 2006).
The advantage of internal innovation is the tailor-made approach and – if successful – yields high returns for the company. However, the often uncertain outcome of R&D investments requires a large tolerance for failure (Holmstrom, 1989). This tolerance relates to the large initial investments and opportunity cost of internal innovation. In the event of a successful innovation, companies can secure this innovation through patents. Patents are protected by intellectual property rights, which means exclusive rights of the usage of this innovation for a certain period of time. Patents are therefore a proxy for the results of R&D (Hall et al., 2005). However, patents are to a certain extent an inaccurate way of measuring innovation, as not all forms of innovation are or can be protected by patents. Companies need to make a trade-off between disclosure to the outside world resulting from a patent process and the economic value of innovation by keeping the information confidential without any patent protection (Moser, 2013).
The other innovation strategy is the external innovation, which is also referred to as knowledge acquisition. External innovation is described as the acquisition of innovation from outside sources rather than the autonomous in-house R&D and implement these acquired innovations into the existing business. Acquiring ‘proven’ effective innovation diminishes the opportunity costs and creates R&D synergy (Letina, 2016; Sears, 2018). From an innovation perspective, this results in innovative outperformance compared to competitors. However, external innovation is not completely risk free, as it deals with adverse selection cost related to acquisition. Bena and Li (2014) find that companies with a large patent portfolio and low R&D expenses acquire R&D intensive firms with a limited patent portfolio.
Moreover, the existing literature does not agree on the preferred innovation strategy for companies. On the one hand, Cassiman and Veugelers (2006) debate that both strategies are complementary to each other, depending on the industry which strategy is more profound.
13 On the other hand, Berchicci (2013) and Sears (2018) suggest a preference of external innovation over an internal innovation. Berchicci (2013) emphasizes on the innovative excellence of external innovation up to a certain threshold. The benefits of the external innovation should outweigh the cost of searching, coordinating, moderating and implementing. Sears (2018) argues that the long lead time of autonomous R&D creates a competitive disadvantage with peers. This is especially profound in high-innovative industries.
Innovation strategy is sensitive to internal (i.e. company) and external (i.e. market and government) influences. Balsmeier, Fleming and Manso (2017) argue that higher board independence is positively related with innovation; more patent applications and patent quality. However, they also argue that in high independent boards there is more focus on the known research areas rather than explore new possibilities. Hirshleifer, Low and Teoh (2012) explain that, within innovative industries, overconfident CEOs are positively related with innovation. A CEO with an entrepreneurial spirit is less risk-averse and therefore more empowered to invest in innovation. This leads to a positive effect with more patent applications and citations. Bénabou, Ticchi and Vindigni (2015) find that the opposite also holds, as religiosity has a negative impact. Related to the time constraint of internal innovation (Sears, 2018), CEO’s incentive to innovate is suppressed. A long-term compensation plan and job security can overcome the lack of motivation to invest in internal R&D. (Manso, 2011).
Moreover, the market has also an impact on innovation, as competition and imitation accelerate innovation (Bessen & Maskin, 2009). The government may stimulate the innovative environment through tax schemes and legislation. Mukherjee, Singh and Žaldokas (2017) denote that an increase in taxes has a negative effect on innovation. Bessen and Maskin (2009) and Moser (2013) find that strong intellectual property rights rather discourage than encourage innovation for the next generation. The concern of investing in innovation which might (partly) overlap with protected by intellectual property demotivates the incentive to innovate due to the aforementioned costs of innovation. Modest intellectual property rights encourages competition and innovation (Moser, 2013). Atanassov (2013) debates the effect of hostile takeovers on innovation. This is found to be beneficial for innovation and economic growth, as the implementation of antitakeover legislation has a negative effect on innovation.
Furthermore, finance and financial development are key drivers of economic growth (Brown, Fazzari & Petersen, 2009), and thus to its extent innovation. However, investing in
14 innovation is difficult for companies with financial constrains due to adverse selection cost (Hall, 2002). Internal financing, e.g. retained earnings, is preferred to overcome asymmetric information issues. Financial constraints drive companies to the use of external financing (Frank & Goyal, 2003). External financing is prioritized by debt over equity, because the tax shield on the interest payment of debt makes it cheaper than equity. This theoretical model is called the pecking order theory (Myers & Majluf, 1984). Based on information asymmetry, innovation is associated with high risk and high returns (Li, 2011) causing difficulties obtaining external financing. Fama and Fetch (2005) refute the pecking order theory, arguing that issuing equity is not the last resort. Their proposed model, i.e. trade-off model, diminishes adverse selection cost. The models are complementary.
Hypothesis
Based on the discussed existing literature, hypotheses are derived to answer the research question. This research intends to measure the influence of private equity ownership on innovation, by answering the question:
Do private equity firms influence the innovation strategy after a takeover?
Concluding from the literature, private equity firms aim to turnaround poor performing or undervalued companies, create value and exit the company to realize a projected return. The hypotheses follow the perspective of the private equity firm. To achieve projected returns, the portfolio company needs to improve operating performance and increase profitability. The profitability is increased by on the one hand expanding the revenues and on the other hand cutting cost. Relating to Boucly et al. (2011), Davis et al. (2014) and Bernstein et al. (2017), especially in case of a short term horizon, investments with returns on the long term are often cut to increase short term profitability. This leads to the first hypothesis:
Hypothesis 1: A private equity takeover decreases the R&D expenditure
Even though the target company is limited by a budget (hypothesis 1), the portfolio company is supposed to increase operating performance. Hence, results of innovation (measured by patents) is expected to become more efficient. Private equity firms are found to be excel in managerial skills (Bloom et al., 2015), which comes in useful to increase efficiency in innovation output. Moreover, Lerner et al. (2011) emphasize on the constant levels of patent
15 application after a LBO, but with higher quality. Bernstein (2015) finds decreasing patent application after an IPO and Seru (2014) finds less patent quality after a takeover by conglomerates. Taking these findings into account, the efficiency after a private equity takeover is assessed in twofold. First, the relative patent grants (#grants / #applications) increases after the takeover and second, the quality or economic power of the patents, measured through patent citations, increases after the takeover. The following hypotheses are constructed:
Hypothesis 2: A private equity takeover increases the efficiency and quality of internal innovation after a takeover, by means of:
Hypothesis 2A: A private equity takeover does not change the amount of patent applications.
Hypothesis 2B: A private equity takeover increases the amount of patent granted relative to patent applications
Hypothesis 2C: A private equity takeover increases the economic power of a patent via an increase of patent citations.
As internal innovation (R&D) is found to be a risky investment (Li et al., 2011) due to the opportunity cost and the investment in innovation does not show immediate results, private equity might be hesitative to make such investments. The average five year private equity investment in a portfolio company might not be sufficient to benefit from internal innovation (Sears, 2018). The alternative, external innovation, is found by Bernstein (2015) to be more persistent than internal innovation after an IPO. Berchicci (2013) emphasizes on the innovative excellence of external innovation compared to internal innovation. Combining these considerations, proposes the following hypothesis:
Hypothesis 3: A private equity takeover increases portfolio company’s external innovation
16 Methodology
As an integral part of this paper, an event study is carried out in order to analyse the portfolio company’s internal and external innovation after a private equity takeover. The timeframe of the deal announcements of private equity takeovers included in this event study include the five years before and the five years after the respective announcements and covers the period from 1997 to 2017. This deal announcement timeframe is to some extent arbitrary in a sense that there is no data available yet for the full five years for deal announcements made as from 2013, which will limit the observations. The target company and acquirer firms of the deals incorporated in this paper are all located in the United States.
The quantitative analysis uses the same research design as Seru (2014) and Liu (2017). Both papers construct an event study to measure the impact of an acquisition on a target company by implementing a difference-in-difference model. This model compares the situation before and after a takeover with a treatment group of materialized transactions and a control group of withdrawn transactions. The following difference-in-difference model is applied in this research:
𝑌𝑖𝑡 = 𝛼 + 𝛽1𝑃𝑜𝑠𝑡𝑖𝑡+ 𝛽2𝐷𝑒𝑎𝑙𝑖 + 𝛽3𝑃𝑜𝑠𝑡𝑖𝑡∗ 𝐷𝑒𝑎𝑙𝑖 + 𝛿𝑍𝑖𝑡 + 𝐹𝐸 + 𝜀𝑖𝑡 (1)
Formula (1) measures the impact of the private equity deal on innovation. The difference-in-difference methodology allows to detect different performance over time, i.e. before and after the takeover, and to detect differences in performance compared to a peer group, i.e. materialized deals versus withdrawn deals. Formula (1) is constructed the following. The dependent variable Y can be categorized into the three research categories related to the three hypotheses. In the first category internal innovation is measured through portfolio company’s R&D expenditure. The second category refers to the efficiency and quality of R&D results, i.e. patent applications, relative grants and citations. The third and final category relates to knowledge acquisition by the portfolio company.
Following, post is a dummy variable and refers to the occurrence of the event. Before the event, post has the value of 0, which changes to the value of 1 in the year of the acquisition. In the tables discussed in the result and robustness section, the Post variable is called After. Deal is a dummy variable and denotes whether or not the acquisition materialized. Withdrawn transactions have a value of 0 compared to materialized transactions
17 with a value of 1. The main coefficient of interest is β3 relates to the interaction term
Post*Deal. The interaction term is a dummy variable with value of 1 after a materialized acquisition, or 0 otherwise. Furthermore, α denotes the constant and ε the error term. The variables are bound to the company (i) and time (t).
In formula (1) matrix Z denotes a set of control variables. Seru (2014) has found that these variables influence the deal negotiations and growth potential. Seru’s broad approach of the variable assessment makes the control variables applicable in this research. The control variables in Z include: size, R&D expenditures, Tobin’s Q, profitability, operating cash flow, retained earnings, book leverage, asset tangibility, company’s age, company’s age squared, the Herfindahl index and the Herfindahl index squared. The squared firms age is included as a measure of business cycles. The squared Herfindahl index controls for potential non-linearity between innovation and market competition. Table A1 in appendix A presents the calculation of the variables in formula (1), as these are scaled and adjusted for outliers.
To correct for miscalculations, the standard errors are altered and the fixed effects are applied. The standard errors are adjusted for robustness. This application modifies the assumption about the independence of the standard errors. The robust standard error estimation assumes that the error terms are not identically distributed. Subsequently, the models are controlled for external influences. The fixed effects include company, industry and year. These effects eliminate irregular trends. Moreover, these fixed effects omit the constant (α) and the deal variable (β2 * Deal) in the regression.
Coefficient β3 denotes the impact of the private equity after the deal and, therefore,
confirms or rejects the aforementioned hypotheses. R&D expenditure is expected to decrease after the private equity takeover. Hence, β3 is expected to present a negative sign. The
effectiveness of relative patent grants and the patent quality are supposed to increase which leads to a positive coefficient. In short, referring to the hypotheses discussed before, hypotheses with an increasing result show a positive sign, whereas hypotheses with a decreasing result show a negative sign. If the coefficient β3 does not show any significance,
the treatment group does not perform different after a private equity takeover than before or compared to the control group.
18 Data
Data collection
For the acquisition data, the Thomson One Database (SDC Platinum) is consulted. This dataset provides all private equity transaction announcements within the 1997 – 2017 timeframe. A bug occurred in the Thomson One Database for retrieving the deal observations from 10 may 2000. These three observations were added to the dataset manually. Within the dataset, an acquisition is considered a private equity transaction when the deal is done by a financial investor (and not a strategic investor), a private equity firm on behalf of its institutional investors, or through a leveraged buyout supported by private equity capital. If the acquirer is an investor group, the first mentioned acquirer of the group is assumed to invest the largest share and is therefore considered as the main investor of the investor group. The label of a private equity acquisition is given when the acquirer’s SIC code equals 6799. This SIC code means that the acquirer is an alternative investor, which not specified in any other category. Within the alternative investor category (SIC code 6799), ‘oil royalty traders’, ‘patent traders’ and ‘real estate investors’ are removed from the dataset. This leads to a dataset in which all acquirers are private equity firms.
In line with the discussed literature review with respect to limitations on data availability (Phalippou & Gottschalg, 2009; Gompers et al., 2016), the dataset used is based on transactions with the following characteristics: after the takeover, the private equity firm owns and controls a majority share, whereas before the transaction they were no or only minority shareholders. Also, the transaction involves an acquisition of at least 25% of the shares. Limitations on data availability primarily occurs in case of takeovers of listed companies by private equity. In such a situation, the company that has been taken over is likely to delist and consequently is no longer obliged to file its accounts to the SEC (Liu, 2017). If the portfolio did not report filings in the following years after a private equity, the takeover was removed from the dataset. An exception is made for deal announcement after 2013. Despite this limitation, the dataset remains sufficiently large to provide a proper analysis.
The effectiveness and quality, i.e. the result, of company internal innovation is measured through patent data, which were directly retrieved from the United States Patent and Trademark Office (USTPO) Bulk Data Storage System (BDSS). This database contains all U.S. patent applications since early 1900s and is regularly updated. Even though this database has its limitations, it is preferred over the HBS database (Li et al., 2014), which
19 observations only date until 2010. To properly analyse the patents and their citations, the company identifiers of the patents of the USPTO BDSS database needs to be linked to the company identifiers of the Compustat database. The nonexistence of a proper linking table complicates matching the company identifiers. To match both databases, first, the Stoffman’s linking table (Kogan et al., 2017) is merged into the USPTO database, allowing patents up to 2010 to match correctly with the company identifiers based on patent grant number. Second, the company names are adjusted by removing punctuation marks and additions, such as corp, corporation or inc, amongst others. Based on name similarity the company identifiers are expanded to the years after 2010 and are matched to the patents. Despite the risk of incorrect matches or missed matches (Li et al., 2014), this is the only alternative to create a complete, up-to-date database for patent applications. The most accurate measure involves manually matching the company identifiers to the patents, which is beyond of the scope of this paper. Consequently, all companies, who did not apply for their first patent in or before 2010, were eliminated from this dataset. This has created a biased, but still sufficient large dataset. Another element to take into considerations is that the citations per patent, i.e. the patent quality or economic power, is manually derived per patent and added to Stoffman’s database. Therefore, patent quality or economic power could only be addressed for the patents registered until 2010. For citations the same holds that manually deriving and potentially updating pre-2010 citations per patent is beyond the scope of this paper.
The database Compustat annual filings in the Wharton Research Data Services (WRDS) database is consulted for financial and accounting information of the companies that were taken over through private equity. This database is used to obtain this data related to R&D expenditures, acquisition expenditure and the control variables. As the database does not provide a variable for the founding year to derive the age of the portfolio company for the control variables, the age is calculated as the amount of years the company is present in the Compustat database.
The Thomson One dataset and USPTO dataset are merged into the Compustat dataset to create the final dataset.
Sample construction
To measure (internal) innovation correctly is rather complicated, because innovation to a certain extent is intangible. R&D expenditures as a percentage of total assets is a clear indication of the importance given by the company’s management to innovation. This covers the development phase of innovation (Seru, 2014). For the productivity and success of R&D
20 patents act as a proxy. The citations per patent are a measurement of patent quality and to its extent economic power (Hall et al., 2005)3. The patent data – as measurement of innovation
productivity – has its limitations (Seru, 2014). It is fair to say that patent data is not a perfect reflection of a company’s innovation activities, because not all innovation are patented; for confidentiality reasons, financial reasons or other specific reasons (Moser, 2013).
Annual acquisition and R&D data do not indicate the impact the investments made in the years before. For both variables a similar formula is applied to also include these historic investments. These manipulated variables reflect all investments that are not yet depreciated and are explained as follows. A lagged version of acquisition data represents a more accurate value, as the implementation after an acquisition takes time (Faleye & Mkrtchyan, 2017). Faleye and Mkrtchyan (2017) propose this formula as conceptualization of Mintzberg’s realized strategy theory. Formula (2) is used to control for these lagged variables.
𝐴𝑐𝑞𝑢𝑖𝑠𝑖𝑡𝑖𝑜𝑛 𝑆𝑡𝑟𝑎𝑡𝑒𝑔𝑦𝑖𝑡 = 𝐴𝑐𝑞𝑠𝑖𝑡−1+ 0.8 ∗ 𝐴𝑐𝑞𝑠𝑖𝑡−2+ 0.6 ∗ 𝐴𝑐𝑞𝑠𝑖𝑡−3+ 0.4 ∗ 𝐴𝑐𝑞𝑠𝑖𝑡−4 + 0.2 ∗ 𝐴𝑐𝑞𝑠𝑖𝑡−5
(2)
Similarly, R&D expenditure is capitalized to include the previous year’s investments (Ladika, 2017). The development and research phase of innovation consist of multiple years. Formula (3) includes the R&D expenditures that are not yet depreciated. XRD in formula (3) refers to the R&D expenditure.
𝑅&𝐷 𝑒𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒𝑖𝑡 = 𝑋𝑅𝐷𝑖𝑡+ 0.8 ∗ 𝑋𝑅𝐷𝑖𝑡−1+ 0.6 ∗ 𝑋𝑅𝐷𝑖𝑡−2+ 0.4 ∗ 𝑋𝑅𝐷𝑖𝑡−3+
0.5 ∗ 𝑋𝑅𝐷𝑖𝑡−4 (3)
Formulas 2 and 3 are applied as dependent variables in the research.
As explained in the methodology section the analysis includes a treatment group of successful private equity transactions and a control group of withdrawn transactions. However, the research is confronted with endogeneity issues (Seru, 2014; Liu, 2017). Announcing the deal and subsequently withdrawing the deal has an impact on the target company and its reputation (Liu, 2017). Deals are withdrawn for various reasons, which are
21 reviewed for every withdrawn deal in the dataset in an attempt to solve the endogeneity issues. A deal withdrawal is evaluated in this research as exogenous if the withdrawal is motivated by disagreement between the acquirer and the target. This includes negotiation failure, acquisition price disagreement or a different vision on the future of the company, amongst others. The deal withdrawal is reviewed as endogenous when the acquirer was overruled by a higher bid from a competitor, issues with regulations occurred or when there was a lack of finance. Due to the ambiguous factor of no information, the rationale of the deal withdrawal is unknown and therefore marked as endogenous. The endogenous withdrawals are subtracted from the dataset in order to create a dataset that consists solely of exogenous deals. Table 1 presents the amount of deals that are removed for endogenous reasons. The reasons for deal withdrawal are found by using public sources, such a recognized newspapers (amongst others, Wall Street Journal and the New York Times) or company press releases.
Table 1 Reasons for deal withdrawal.
This table presents an overview of the endogenous reasons for deal withdrawal. The first row denotes the full sample, followed by the reasons to withdraw the deal announcement. The bottom row shows the amount of companies in the exogenous subsample.
Total withdrawn deals 108
No information 25
Overbid 18
Regulatory issues 4
Lack of financing 4
Exogenous withdrawn deals 57
Descriptive data
In total 588 deals are included in this research, of which 480 were successful and 108 were withdrawn. After removing the endogenous withdrawn deals, it leaves with 57 different target companies. A distinction is made for firms with at least one patent application, as not all companies are similar innovative. In total there are 129 deal announcements for target companies with at least one patent application, of which 118 were successful and 11 withdrawn. The exogenous subsample remains with five withdrawn deals. The overview of the deal sample size is presented in table 2 and plotted over the years in graph 1. The relative large size of the treatment group does not affect the regressions. Nevertheless, it should be noted that for approximately 20% of the successful deals there is at least one patent application, whereas for withdrawn deals this is only close to 10%. This is also visible in
22 graph 1 as the gap between total deal announcement and successful deals becomes smaller over time.
Table 2 Deal sample size.
This table presents an overview of the amount of investigated deals. The first row denotes the total deals, divided in the successful and withdrawn deals (column 2 and 3). The last row is the exogenous subsample of column 3. A distinction is made between deals with or without patent applications.
All deals Successful Withdrawn full sample
Withdrawn subsample
Private equity deals 588 480 108 57
With > 0 patent applications 129 118 11 5
Moreover, graph 1 presents the deal announcements, the materialized transactions and the withdrawn transactions over the timeframe. In this graph two different trends can be observed. The first trend shows a certain cyclicality of the deal announcements. For the first ten years in the timeframe, i.e. 1997 to 2007, the ratio between materialized deals and withdrawn deals is approximately constant. As aforementioned, in 2007 a record in fund value was raised by the private equity firms, which results in a peak in graph 1 in 2009. The second trend noticeable is a decrease in deal withdrawals over the years, in both absolute and relative measurement. The decrease can be explained by a more efficient approach of private equity firms, deal announcement in a later stage or less volatile macroeconomic factors. The impact of the uncertainty of the beginning of financial crisis (2007/2008) is noticeable in the peak of withdrawn deals announced in 2006. However, for the data in the last two years, i.e. 2016 and 2017, the graph potentially shows biased data. Private equity deals announcement in those years can still potentially be withdrawn in 2018 or later.
23 Graph 1
Private equity deals per year, announced, effective and withdrawn.
This graph illustrates the total amount of deal announcements (blue), the amount of effective deals (red) and the amount of withdrawn deals (green) between 1997 and 2017. The effective and withdrawn deals add up to the total deal announcement. The overall total is 588 deal announcements, with 480 effective and 108 withdrawn.
Table 3 provides the summary of the statistics of the main variables, in which column 1 and 2 show the results separated for the treatment and control group, respectively. It can be observed that the treatment group is more patent intensive than the control group. Graph 2 emphasizes on this distinction. Nevertheless, the median for in model 1 and model 2 for patent application is zero, meaning that the majority of the observations does not have at least one patent application, confirming the data presented in table 2. Moreover, a difference can be observed in relative R&D expenditure similar to patent applications. The mean in the treatment group is almost double the value of the mean in the control group, whereas the median value is approximately the same. An explanation for the difference in patents and R&D expenditure could relate to the innovative intensiveness of the industry, which is stronger present in the treatment group than the control group. For both variables, the standard deviation in the treatment group is sufficient large to expect outliers. Excluding these industries is assessed in the robustness section of the results. Looking at the relative patent grants, both the treatment and control groups behave similarly and 80% to 85% of the patent applications are granted. However, for both the treatment and the control group, the mean does not reflect change over time, which is illustrated in graph 3.
24 Table 3
Descriptive statistics.
This tables reports the summary statistics of the main variables split for materialized transactions (column 1) and withdrawn transactions (column 2). The variables of denotes the dependent variables of the regression (patent application and grants per year, R&D expenditure and acquisition appetite). In column 1 and column 2 the mean, median and standard deviation per variable are presented. Column 3 denotes the difference in the mean.
(1) (2) (3)
Treatment group Control group Mean
Difference
Mean Median SD Mean Median SD
# patent applications 68.24 0.00 684.72 3.87 0.00 24.11 64.37 # patent applications, ln 0.82 0.00 1.70 0.26 0.00 0.89 0.56 Acquisition, adjusted 4.05 4.24 2.40 3.60 3.89 2.25 0.45 Acquisition, % 0.03 0.00 0.07 0.02 0.00 0.06 0.01 Patent cites, ln 0.15 0.00 0.58 0.07 0.00 0.39 0.08 Patent grants, % 0.80 0.90 0.27 0.85 1.00 0.26 -0.05 R&D expense, % 0.15 0.05 0.29 0.07 0.04 0.09 0.08
R&D expense, adjusted 0.43 0.14 0.82 0.20 0.11 0.23 0.23
Observations 3679 889
Furthermore, a trend can be observed in graph 2 reviewing the patent applications over time around a private equity takeover (t=0). For the treatment group the patent applications show a double concave pattern: for the timeframe -5 to 0 and the timeframe 0 to 5. One could argue that a private equity wants to materialize the R&D investments after a takeover, which would explain the increase in t=1 and t=2. This is flattened out later in time, returning at t=5 to t=0 level. The control group, on the other hand, shows a different pattern over time. Similar to the treatment group, it shows a fairly constant or slightly decreasing trend in the two years leading up to a deal announcement, however, the impact of the withdrawal is clearly noticeable. At t=1 the control group shows a steep reduction. In the years following to t = 5, a recovery pattern can be observed, nevertheless, it does not return to pre-withdrawal levels.
As aforementioned, a striking time trend is noticeable in graph 3. The patent grants relative to patent applications show a decreasing trend over the years. For the treatment group this is a fairly constant line, however, the control group shows more fluctuations. Especially at t=1, the year after a private equity deal announcement, the relative patent grants drop almost 25%. Target companies of withdrawn deals recover till t=3 after which the similar trend as before the deal announcement and the treatment group is noticeable. Hence, the withdrawing a deal after a public announcement has its effect on the company.
25 Graph 2
Patent applications in the timeframe.
The graph illustrates the development of patent applications over time. The timeframe denotes the years around the private equity takeover announcement (t=0). The blue line denotes the treatment group and the red line the control group. The deal announcement (t=0) is within the timeframe 1997 – 2017 and only includes U.S. firms. The value of the line is the natural logarithm of patent applications.
Graph 3 Relative patent grants in the timeframe.
The graph illustrates the trend of relative patent grants over time related to a deal announcement (t=0). Relative patent grants is measured as patents granted over patent applications. The blue line denotes the treatment group and the red line denotes the control group. The deal announcement (t=0) is within the timeframe 1997 – 2017 and only includes U.S. firms. The value of the line is in percentages.
26 Results
This section analyses and debates the empirical results of this paper. A difference-in-difference regression is conducted to measure the impact of a private equity takeover on innovation strategy. The main coefficient of interest is the interaction term, i.e. after*treatment, as it denotes private equity’s impact. The control variables are for brevity not mentioned in the tables. The results presented examine the hypotheses in order to answer the research question. It should be noted that the treatment variable included in formula (1) is omitted from the regression, due to the conflict with firm fixed effects.
Table 4 Effect of a private equity takeover on R&D expenditure.
This table reports the coefficients of a panel data regression (firm and time) whether a private equity takeover impacts the R&D expenditures of a portfolio company. The dependent variables are R&D expense (models 1 and 2) and adjusted R&D expense (models 3 and 4). The adjusted R&D expense includes all R&D that is not yet depreciated. The independent variables are dummy variables for event time before (=0) or after (=1) the deal, whether in the treatment group (=1) or the control group (=0), and the interaction term. The time frame is 1997 – 2017 for event time = 0. The regressions are controlled for time-, industry-, and firm fixed effects and additional control variables (size, R&D expenditures, Tobin’s Q, profitability, operating cash, retained earnings, book leverage, asset tangibility, age, age squared, Herfindahl index and Herfindahl index squared). For models 1 and 3 the full sample is applied, where models 2 and 4 use the exogenous subsample. In the table are the coefficients, standard errors (in parentheses), number of observations and the R-squared presented. The standard errors are adjusted for robustness. *, ** or *** denote the significance level at, respectively, 10%, 5% or 1%.
R&D expense, % Adjusted R&D expense, ln Full sample Sub sample Full sample Sub sample
(1) (2) (3) (4) After * Treatment 0.0278 (0.0170) 0.0418*** 0.0194 0.0671** (0.0161) (0.0242) (0.0288) After -0.0154 -0.0317* 0.0346 -0.00259 (0.0149) (0.0171) (0.0283) (0.0346) Control variables Y Y Y Y Time FE Y Y Y Y Industry FE Y Y Y Y Firm FE Y Y Y Y Observations 565 508 554 497 R2 0.943 0.945 0.975 0.975
27 Table 4 presents the results of the empirical research on the effect of a private equity takeover on R&D expenditure, which relates to the first hypothesis. Based on the existing literature, the R&D expenditure is expected to decrease as private equity cuts cost to increase profitability. To measure the impact on R&D expenditures two dependent variables are applied: the R&D expenditures relative to total assets and the natural logarithm of the adjusted R&D expenditure. The adjusted R&D expenditure includes the not yet depreciated R&D expenditures from previous years (formula 3), as the returns of R&D cannot always be obtained in the same year (Ladika, 2017; Sears, 2018).
For the first dependent variable, R&D expense, the results from the full sample and subsample show a similar pattern, in respectively model 1 and model 2, however with different significances. Firms invest less in the R&D department over time, however for a private equity takeover it increases. Model 1 in table 4 presents the full sample, which does not show any statistical significance. The economic significance is debatable as R&D expenditure should increase by approximately 36 percentage points to have a one percentage point impact on company value. This seems too high to be economic significant. The high-value of R-squared suggests that the private equity does not influence R&D expenditure rather than the lack of relation due to scattered data. In the subsample in model 2 the results show a significance on the interaction term at a one percent level. This implies that after a private equity takeover the R&D expenditure increases with 4.18 percentage points relative to the exogenous withdrawn control group. From an economic significance perspective, the R&D expenditure has to rise 23.9 percentage points to obtain an additional percentage relative to total assets. This is better manageable than in model 1, however, it still seems not too realistic. Moreover, the independent variable relating to the event time is significant at ten percent level. After a deal announcement, whether or not successful, the R&D expenditure decreases by 3.17 percentage points compared to prior an announcement. The significance in this variables compared to the significance of the interaction term suggests that withdrawn transactions spend seriously less on R&D after a deal withdrawal to statistically significant influence the coefficient sign. Recalling that the ratio between successful and withdrawn private equity deals in this papers is approximately 80-20, the negative impact of withdrawn transaction on R&D expenditure is sincere.
The results of the second dependent variable, the adjusted R&D expense, are presented in model 3 and model 4 of table 4. For the full sample in model 3 no statistical significance is found on the impact of private equity on adjusted R&D expenditure. Similar to model 1, the high R-squared value confirms the lack of influence of private equity firms on
28 R&D rather than scatter data leading to indecisiveness. The exogenous sample in model 4 denotes a statistical significant coefficient for the interaction term. After a successful private equity takeover the adjusted R&D expenditure increases by 6.7 log points. This relates to a 15.6% increase in the sample mean and an 8.2% increase in the standard deviation. Model 4 confirms the findings in model 2.
Concluding from table 4, the results are not in line with the expectations and, to some extent, contradicting. In the full sample, model 1 and 3, a private equity takeover does not influence the R&D expenditure. However, according to model 2 and 4, a private equity takeover influences the R&D expenditure and has a positive relation. The results of table 4 reject the first hypothesis that private equity firms decreases the R&D expenditure. The rationale for private equity to cut cost in order to increase profitability does not apply for the R&D expenditure. An argument in favour of the presented results could be that private equity firms invest in the R&D department to increase company growth and in this manner increase profitability. The counter argument simultaneously holds that the benefits of R&D are most likely outside the timeframe of a private equity investment (Sears, 2018). This could be refuted by implying that private equity can still profit of the R&D investments by receiving goodwill during the portfolio company’s exit. Arguments in favour or against these results and the relation to the existed literature could be going on, however, further research is requested to air this issue. This is emphasized on in the recommendations.
Table 5 presents the empirical research on the impact of a private equity takeover on the efficiency and quality internal innovation, which relates to hypothesis 2. The efficiency of internal innovation is expected to increase to enhance company value. Internal innovation efficiency is measured through three channels, namely: patent applications, relative patent grants and citations. Patents are a proxy to express the results of R&D. Patent citations is defined as the amount of times referred to the patent by another patent and is a proxy of the quality and economic power of the patent.
The amount of patent applications are explored in model 1 and 2 of table 5. The natural logarithm of the amount of patent application is taken to control for outliers. In both models the main coefficient of interest, the interaction term, does not show any statistical significance. This is line with the expectation from the existing literature. The high value of R-squared confirms the lack of impact of a private equity takeover rather than scattered data leading to inconclusive results. The different sign in the coefficient of the interaction term in model 1 and 2 emphasizes the statistical insignificance, which consequently complicates the
