Universiteit van Amsterdam - Amsterdam Business School
Acquiring from a former VC investor – does it pay off?
An empirical study of acquisitions of venture-backed companies
Master Thesis
Student: Elitsa Krumova (11360712)
Program: MSc Finance: Corporate Finance
Date: August, 2017
2
Statement of Originality
This document is written by Elitsa Krumova, who takes 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 source 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.
3
Acknowledgement
I would like to express my gratitude to my supervisor Dr. Tolga Caskurlu for the suggestion of the topic of this research as well as for the valuable guidance throughout the completion of this thesis. Special thanks go to my family and friends for the support during my studies.
4 ABSTRACT
This paper examines 2 311 acquisitions of venture capital-backed U.S. private companies that occurred between 1989 and 2016. Acquirers are U.S. public companies who are either formerly venture capital-financed or not venture capital-capital-financed. My main focus are deals in which acquirer and target share at least one common venture capital (VC) investor. I explore the value impact of the common VC connection on acquirer’s short-term stock market performance.
My results confirm that the effect of the common VC connection depends on the geographical proximity between acquirer and target as well as on the reputation of the VC firm. I find that when a target is located further apart from the acquirer, sharing a common VC is positively perceived by the market. In the cases of more highly reputable common VC firm, the acquirers achieve negative abnormal returns. However, the analysis implies that when the common VC is relatively younger, bidders are able to achieve positive abnormal returns.
5
Table of Contents
Chapter 1: Introduction ... 6
Chapter 2: Literature review ... 8
2.1 Social connections ... 8
2.2 Venture capital financing and venture capital reputation as a signals ... 11
2.3 Studies on M&A transactions and stock market performance ... 12
2.4 Hypothesis Formulation ...-... 13
Chapter 3: Methodology ... 15
3.1 Event study methodology ... 15
3.2 Regression models and variables definitions ... 16
Chapter 4: Data, Sample Construction and Summary Statistics ... 19
4.1 Data sources and sample construction ... 19
4.2 Descriptive statistics ... 20
Chapter 5: Results, Discussion and Robustness Tests ... 22
Chapter 6: Limitations ... 26 6.1 Method ... 26 6.2 Data ... 26 Chapter 7: Conclusion ... 28 References ... 29 List of Tables Table 1: Number of Acquisitions of Venture Capital-Backed Companies by Year and Sample ... 33
Table 2: Number of Acquisitions of Venture Capital-Backed Companies by Industry and Sample ... 34
Table 3: Summary Statistics ... 36
Table 4: Acquirer’s Cumulative Abnormal Returns (CARs) around Deal Announcement ... 38
Table 5: Regressions for Acquirer’s Cumulative Abnormal Returns (CARs) around Deal Announcement ... 40
Table 6: Regressions for Acquirer’s CARs around Deal Announcement & VC Reputation ... 41
Table 7: Regressions for Acquirer’s Cumulative Abnormal Returns (CARs) around Deal Announcement ... 43
6
Chapter 1: Introduction
The focus of this research are deals in which formerly VC-funded U.S. public companies make acquisitions from their previous VC investors. Does the common VC firm presents a source of informational advantage that lessens the magnitude of the principal- agent problem inherent in mergers and acquisitions? Or, does the connection misleads the acquirer and allures him in value destructive deals? I am interested in measuring the value impact on the bidder’s short-term stock performance produced in such transactions.
My topic is related to academic research in the areas of mergers and acquisitions, venture capital, venture capital certification and social and inter-firm connections’ impact on firm decision making. The thesis is closely linked to two earlier studies of venture backed-acquisitions. Gompers and Xuan (2009) and Kamath and Yan (2009) both examine transactions in which acquirers and targets share at least one common VC investor. However, they do not provide unanimous evidence for the effect of this shared VC sponsor on acquirer’s short-term stock performance. Kamath and Yan (2009) report negative average announcement returns of approximately 3%. Gompers and Xuan (2009) document the opposite results - positive deal announcement returns of around 2.6% to 2.8%. Furthermore, Kamath and Yan (2009) do not distinguish among VC firms while Gompers and Xuan (2009) conclude that the positive effect of acquirer and target sharing a common VC investor is driven by the more prestigious VC firms. Given the contradictory results reported in these studies and the fact that they are both a decade old, it is interesting to reexamine the topic. I use a more extended time period that covers both the period investigated in the before mentioned papers as well as more recent data. In addition, I retest the conclusion of Gompers and Xuan (2009) that acquirers gain only if the common VC firm is among the more reputable VC investors. In brief, my thesis aims to resolve the puzzling findings of earlier studies in the same field. I also contribute to relatively limited literature on the role of venture capitalists as intermediaries between their current and former portfolio companies (Lindsey (2008), Masulis and Nahata (2011), Gompers and Xuan (2009), Kamath and Yan (2009)).
I investigate the impact of target and acquirer sharing a common VC investor by using a sample of 2 311 acquisitions of VC-backed U.S. private companies between 1989 and 2016. Acquirers are U.S. public companies who are either formerly venture capital-backed or not venture capital-backed. My study confirms that the market reaction to deal announcement varies in specific cases. The effect differs when the information asymmetry problem is more exacerbated or target and VC firm characteristics serve as additional market signals. Due to the small sample size, my results are not robustly significant across different empirical models. Nevertheless, the analysis confirms that the common venture capital connection is value creating when acquirer and target are headquarters in different states. The analysis related to VC firm reputation yields results opposite to the findings in Gompers and Xuan (2009). When the
7
shared VC investor is less experienced, acquisitions produce positive deal announcement returns for the bidder. This tendency could be explained with a potential rent-seeking behavior on the part of the prestigious VC firm as suggested in Kamath and Yan (2009). My results imply that only more experienced VCs are able to extract higher premiums that net out the positive impact of the connection.
The remaining part of the paper is organized as follows. Chapter 2 provides an overview of academic research related to my topic. Chapter 3 presents the methodology used as well as definitions for key variables. Chapter 4 describes data, sample construction and summary statistics. Chapter 5 reports the empirical results of my study – univariate and multivariate analysis of cumulative abnormal returns as well as robustness checks. Chapter 6 discusses limitations of my research. Chapter 7 concludes.
8
Chapter 2: Literature review and Hypothesis Formulation
This chapter provides a brief overview of the strands of academic research connected to my analysis. It is divided in three parts. It first covers studies on the impact of social connection in business and finance, in general, as well as in venture capital, in particular. I continue by summarizing key findings about the signaling effect of venture capital funding and venture capital reputation. The third part presents important literature about merger and acquisitions outcomes. In the last subsection, I develop the hypothesis that I will be testing.
2.1 Social connections
A. Social connections studies in general finance spheres
Academic research on the role that social networks play in financial decision making has been on the rise in recent years. Cohen, Frazzini, and Malloy (2008) document that information is transmitted through educational networks between mutual fund managers and executives at publicly traded companies. Fund managers appear to be more likely to invest in companies in which they have a connection and they achieve superior results in the latter investments. Kunhen (2009) reviews past business connections between mutual funds and advisory firms and finds out that these relations increase the likelihood of an advisory firm being hired. However, the latter fact is not associated with better or worse financial outcomes for fund shareholders. The author, hence, suggests that the positive impact of information transmission and the negative one of favoritism net each other out.
Social links impact has been more exhaustively examined in the light of CEO connections and board of directors interlocks. Schonlau and Singh (2009) stipulate that boards with better developed networks (higher number of connections to other boards, based on board interlocks) have access to superior information and hence are better able to exercise their monitoring and advising functions to the benefit of the company. The paper examines the short-term and long-term post-merger performance of companies with better connected (central) board of directors in comparison with firms with less connected ones (non-central). It demonstrates that firms with central boards enjoy higher buy-and-hold returns (BHAR) as well as larger increases in return on assets (ROA) 3 years after deal completion. Cai and Sevilir (2009) study the effect of board connections on M&A deals outcomes in terms of short-term and long-term performance, takeover premium and investment banking fees paid. They define two types of board relatedness: first degree connection when acquirer and target have at least one common board director before deal announcement and second degree – when at least one director from acquirer and one from target have been simultaneously directors in the board of a third firm. They find that mergers with both type of connections exhibit positive abnormal returns around announcement date. However, when it comes to performance over the long run, only transactions for second connection show superior results to the control. The paper suggests that second degree informational advantage is used for value creation as deals
9
are undertaken when there are opportunities for better performance of combined entity. In first degree connection deals the informational advantage could be used for the benefit of the acquirer and at the expense of target. Ishii and Xuan (2013), in the context of merger and acquisitions (M&A), examine social ties. The latter are defined in terms of common educational and/or work background of executives and board directors. Their results show a negative relation between the degree of social connectedness of acquirer and target and the 3-day cumulative abnormal returns around the announcement day for acquirer, target as well as combined entity. The accounting performance of the new company one year after merge is worse than the expected performance of the separate entities. The authors conclude that social ties in the M&A transactions have value destructive consequences. Larckera et al. (2013) research how well connected a company is based on its number of common board directors across different firms and how this connectedness influences future long-term performance. They point out that companies with better connected boards experience superior future excess returns, higher profitability in terms of ROA and are able to outperform the analysts’ consensus earnings forecasts. The authors stipulate that their evidence implies that the advantages of board connectedness are not immediately internalized in the stock prices. El-Khatib Fogel and Jandikc (2015) investigate the impact of CEO personal network on M&A deals and report higher probability of highly connected CEOs to engage in M&A transactions. Such deals also appear to be value-destroying. Moreover, these executives receive higher compensation following the deal. The authors conclude that positive effects of access to superior private information are outweighed by the negative effects of managerial entrenchment.
Hence, the quality and character of the social connection could have a varying effect on financial decision outcomes. The links do not have a uniform influence on takeover deals. That influence is often a function of other sub-factors.
B. Social connections studies in the case of venture capital (VC) firms
There is a growing number of papers focused on the impact of social ties in the domain of venture capitalist (VC) firms. Shane and Cable (2002) research the impact of social connections between entrepreneurs and seed stage investors on the financing of entrepreneurs’ businesses. They conclude that social ties are used as a source of additional information in the decision whether to sponsor the venture or not. Hochberg, Ljungqvist and Lu. (2007) examine the performance of VC funds and how the latter is impacted by VC networks and the degree of relatedness within these networks. A better co-investment network leads to better fund performance as well as higher probability of survival for a company after VC exit. Sorenson and Stuart (2001) study the interfirm relations between venture capitalist firms, relations formed via syndicated investments. They demonstrate that these ties are a powerful source of informational advantage that facilitates VC investments and even decreases the preference for investment in businesses that are located in geographically proximate areas. Du (2016) investigates the homophily effect in the case of syndicated VC investments. Based on his research VC firms are more likely to co-invest with counterparts
10
similar to them. Jääskeläinen and Maula (2014) examine how direct and indirect social ties among the VC network influence the probability of the VC firm realizing its investment exit in a foreign market. They find that the stronger the ties of the firm, the smaller the firm’s preference for local exits.
However, the main body of research has been concentrated on the connections that VC firms create among each other. More papers investigating the ties between VC firms and former portfolio companies as well as among former portfolio companies have started to emerge. Lindsey (2008) conducts a study of the role of VC funds in the formation of strategic alliances among portfolio companies. She finds evidence that companies with a common VC investor are significantly more likely to enter into alliance. The observed effect is more important for same industry alliances as well as businesses whose competitiveness largely relies on intangible assets. The formation of such alliances, in addition, improves the likelihood for a successful exit for the VC fund (when the alliance is created while the VC still has a stake in either of the entities). Masulis and Nahata (2011), in accordance with the hypothesis that VC firms “certify” the quality of a target, find that acquirers of VC-backed private companies experience higher abnormal returns around announcement date. In addition they further show that VC funds experience certain conflicts of interest during M&A deals that affect target valuation and, thus, acquirer returns. The takeover premium for targets backed by VC funds that are approaching planned liquidation is significantly lower as the VC investors are under higher pressure to exit their positions. There is also evidence for a “wealth redistribution” effect from target to acquirer when a VC firm has stakes in both acquirer and target.
Apart from Masulis and Nahata (2011), there are two more papers that research acquisitions of venture-backed companies. Gompers and Xuan (2009) and Kamath and Yan (2009) examine transactions in which acquirers and targets share the same VC investor. They document that the presence of at least one common VC sponsor increases the likelihood of an acquisition. However, the observed impact on the bidders’ short-term stock performance largely differs in the two studies. Gompers and Xuan (2009) suggest that the common financial sponsor serves as an informational link. They report positive deal announcement returns of around 2.6% to 2.8% for the purchasing entity. The returns are even higher when the common venture capitalist has more experience, the target is younger or situated at a greater geographical distance from the bidder. On the other hand, Kamath and Yan (2009) find evidence for a rent-seeking behavior on the part of the common VC firm. When the bidders purchase from their old VC sponsor, they experience negative average announcement returns of approximately 3% as well as weaker operating performance in the year following the deal. Though, in deals with a common VC firm, both Gompers and Xuan (2009) as well as Kamath and Yan (2009) document a significant tendency in the preferred method of payment. Such transactions are more often financed with equity, presumably as a consequence of the reduced information asymmetry between target and acquirer.
Therefore, given the contradictory evidence provided in the above described studies, it will be interesting to reexamine their results.
11
2.2 Venture capital financing and venture capital reputation as a signals
Venture capital firms predominately invest in companies with high information asymmetry problems, mainly early stage and technology ventures (Gompers (1995)). Apart from the access to financial resources (which is usually scarce in such entities), VC firms contribute to their new ventures with expertise and connections to prospective partners, investors, human capital (Gompers (1996), Brav and Gompers (1997)). They could also improve the venture’s likelihood of survival. Puri and Zarutskie (2012) demonstrate that VC-backed companies experience lower failure rates in comparison to their non-VC-VC-backed counterparts.
Krishnan, Chemmanur and Nandy (2011) demonstrate that the involvement of VC firms improves the total factor productivity of the financed company. This efficiency gain is even stronger when the investor is among the more reputable VC firms. Nahata (2008) confirms that companies sponsored by more reputable VC firms enjoy better asset productivity. Ventures backed by VCs, compared to non VC-backed start-ups, are more successful in finding commercial and research partners according to Hsu (2006). They also more easily attract human capital (Hellmann and Puri, 2002).
Evidence for the value added by VC firms is also provided in Brav and Gompers (1997). They report superior post-IPO stock market performance for VC-financed companies. Baker and Gompers (2003) investigate the board of directors structure of companies taken public by their VC sponsors. VC-financed entities have fewer inside directors and more independent outsiders in comparison to non-private equity supported peers. They conclude that the better quality of the board is associated with the superior post-IPO stock returns achieved by the VC-backed companies.
The presence of a VC firm in a young company is positively interpreted by other market participants. The grandstanding hypothesis of Gompers (1996) establishes the venture capitalist funding as a form of endorsement, “certification”. Lee and Wahal (2004) also provide evidence for the certification function of VC firms in the light of IPO underpricing. Megginson and Weiss (1991) find support for the certification role of VC firms during initial public offerings (IPOs). The presence of a venture capitalists is a proof of the company’s quality and, hence, diminishes the inherent information asymmetry. VC-backed companies experience lower underpricing and overall costs and attract more reputable underwriters.
However, this certification effect is not unanimous. Gompers (1996) shows that young VC firms are willing to accept a higher underpricing during an IPO as they need to build a reputation, create a track record of successful exits and attract investors in their new funds. Consequently, less established VCs also tend to liquidate their positions earlier compared to more experienced peers. Lee and Wahal (2004) also point out that the positive signal of the VC funding is reduced in the case of younger VC firms or VC firms with less IPO exits. Gompers, Kovner, Lerner and Scharfstein (2006) demonstrate that financing by a more experienced VC firm improves the likelihood of success for entrepreneurs. Sorensen (2007) provides evidence that more experienced VC firms are usually better at both selecting and monitoring their
12
investments. Furthermore, Hsu (2004) demonstrates that entrepreneurs are more likely to choose a better reputable VC firm over a less reputable one even at the cost of a less favorable valuation offer.
Ragozzino and Blevins (2016) study various dimensions of the venture capital investment and their effect on the exit characteristics. The research shows that reputation of the VC firms, number of VC firms financing the company, duration of the investment, age of the company at first round as well as the amount of monetary funds provided directly impact the exit decision.
The importance of VC syndication network for finding prospective investments, described in Lerner (1994), Du (2016), Hochberg et al. (2007) and Sorenson and Stuart (2001), suggests that high quality ventures will attract a higher number of VC sponsors. Also, the expected negative reputational effect of inviting peers to co-invest in low quality ventures further confirms that a high number of investors is a positive signal. Syndication diminishes the severity of the information asymmetry problem (Lerner, 1994). Barry, Muscarella, Peavy and Vetsuypens (1990) show a negative relation between the number of VC firms and the underpricing incurred during IPO. Guler (2007) demonstrates that VC firms remain more involved in a portfolio company when the amount of the investment is larger.
2.3 Studies on M&A transactions and stock market performance
There is a large body of academic studies on mergers and acquisitions. Earlier research provides relatively exhaustive answers to the questions how varying deal, acquirer and target characteristics influence the market perception of the transaction.
A paper by Moeller, Schlingemann and Stulz (2004) shows that, on average, smaller acquirers are able to earn higher abnormal returns in comparison to larger bidders. Faccio, Mcconnell and Stolin (2006) demonstrate, using a sample of 17 western European countries, that acquirers of unlisted targets earn higher average abnormal returns – a positive average abnormal return of 1.48% compared to -0.38% for acquirers of public targets. Chang (1998), Hansen and Lott (1996) as well as Fuller, Netter and Stegemoller (2002) demonstrate that public acquirers of private firms achieve positive abnormal returns. However, in contrast to acquisitions of public targets, the stock purchases of private firms yield higher returns than the cash purchases of private firms. In addition, Fuller et al. (2002) find evidence that, in transactions involving private targets, the higher the relative size of the target, the higher the abnormal returns achieved by the bidder. Again, in such transactions, stock deals outperform cash ones. However, Fuller et al. (2002) hypothesizes that the observed phenomenon could be due to the so called liquidity effect – private companies have less restricted access to finances and, hence, might be willing to accept a lower biding price. Though, such explanation is less likely in the case of companies backed by venture capital funds. On the other hand, Chang (1998) explains the results with the creation of a large blockholder and the monitoring benefits brought by the target shareholders.
13
Lang, Stulz and Walking (1991) test the free cash flow hypothesis of Jensen (1988) and demonstrate that acquirers with high market-to-book ratios (a proxy measure of Tobin’s Q) achieve higher abnormal returns compared to bidders with low market-to-book indicators. Similar evidence is found in Servaes (1991). In addition Maloney, McCormick and Mitchell (1993) show that bidders with higher pre-merger leverage levels outperform less indebted bidders. Another important characteristic of merger and acquisition to be taken into account is the fact that these deals are inclined to cluster by industry (Andrade, Mitchell and Stafford, 2001).
In addition, earlier studies have found an interesting link between geographical proximity and takeover occurrence and performance. Kedia, Panchapagesan and Uysal (2008) show that when the headquarters of acquirer and target are situated within 100 km of each other, these bidders outperform almost twice acquirers of more distant firms. Gompers and Xuan (2009) also demonstrate that distance significantly influences stock performance around announcement. Erel, Liao and Weisbach (2012) investigate cross-border transactions and find that the closer geographic location positively influences the probability of a deal. Distance has been observed as an important indicator in other financial domains as well. Malloy (2005) documents that analyst located in closer geographical proximity to the companies they cover are able to provide more accurate forecasts due to informational advantage. Coval and Moskowitz (1999) find that U.S. investment managers are more likely to include in their equity portfolios companies more closely situated to them. The preference for local businesses is stronger when the prospective firms are as well small, have high leverage or non-internationally tradable goods. Coval and Moskowitz (2001) provide evidence of nearby investments yielding significantly higher returns for mutual funds.
2.4 Hypothesis Formulation
The focus of this research are deals in which formerly VC-funded acquirers purchase companies from their former VC sponsors. I am interested in measuring the value impact on the bidder’s short-term stock performance produced in such transactions. Given the contradictory evidence in similar studies (Kamath and Yan, (2009) and Gompers and Xuan (2009)), I reexamine their results. Does the common VC firm presents a source of informational advantage that lessens the magnitude of the principal- agent problem inherent in M&A transactions? Or, does such connection misleads the acquirer and allures him in value destructive deals?
Based on the discussed academic research on the role of personal and inter-firm ties as well as on the certification effect of VC investors, I formulate the following hypothesis:
Hypothesis 1a: Acquirer’s stock performance at deal announcement will be positive when acquirer and
target share at least one common VC firm.
Given that the distance between bidder and target significantly influences deal performance, I account for the geographical proximity of the two companies. Acquirers in close by acquisitions are expected to have
14
additional sources of information about the target and to be able to more easily conduct due diligence. As the information asymmetry problems will be larger in takeovers of geographically distant companies, I hypothesis that:
Hypothesis 1b: When acquirer and target share at least one common VC firm, the positive acquirer stock
performance at deal announcement will be higher if acquirer and target are located further apart.
As presented in the previous subsections the effect of the VC certification is not unanimous and could vary with the number and quality of the financial sponsors. The screening and monitoring capabilities of the VC firm are function of its experience. Its reputation influences the exit decision as well. Based on the literature, I expect that:
Hypothesis 2: When acquirer and target share at least one common VC firm, the positive acquirer stock
performance at deal announcement will be stronger if the common VC sponsor is more reputable.
In the next Chapter I proceed by explaining the econometric model to be used in the testing the above hypothesis in the next Chapter. I also define independent variables based on earlier studies described in this section.
15
Chapter 3: Methodology
3.1 Event study methodology
In order to test the value impact of acquisitions in which both acquirer and target have been financed by a common venture capitalist, I use an event study methodology. The method has been widely employed in finance, accounting and management research. As per 2004 the event studies papers published in leading journals exceeded 500 (Kothari and Warner, 2004).
The methodology is based on assessing the stock market reaction to an unexpected event. As the markets are assumed to be semi-strong efficient, the bidder’s stock price is supposed to reflect all discounted future cash flows of the firm. Therefore, any abnormal price movements around the deal announcement are expected to be a consequence of the news release and represent the so called excess or abnormal returns. Assuming that the efficient market hypothesis (EMH) holds true, the markets should be able to timely incorporate all new information relevant to the company’s value (McWilliams and Siegel, 1997; MacKinlay, 1997). A benefit of this approach is that share prices could not be as easily manipulated as accounting data and, hence, they could be considered as a more unbiased measure of the company’s true value (McWilliams and Siegel, 1997). A number of studies have found evidence that the stock performance around the event of interest is a good predictor of future performance. For instance, Healy, Palepu and Ruback (1992) investigate the post-merger operating performance of large public U.S. companies and find positive connection between market reaction at announcement day and post-merger operating cash flows. In a more recent paper by Higgins and Rodriguez (2006), the authors, through a study of pharmaceutical acquisitions, demonstrate a relation between positive abnormal returns and improved ex post innovative performance. The constraints and flaws of the event studies method are discussed in the Limitations section.
The abnormal return generated from each takeover announcement is interpreted as the market adjustment to the arrival of new information. It is computed as the difference between the firm’s actual realized return and the return expected had the deal not been disclosed (referred to as “normal” return). For the calculation of the “normal” rate of return I use the market model which is noted as a standard practice in McWilliams and Siegel (1997). Under the market model the rate of return on the stock price of firm i is regressed on the return of the market portfolio of stocks (Rmt) over the time interval T. Interval T (referred to as the estimation window) precedes the event of interest. The below model is used:
where αi is the intercept, βi is the systemic risk for firm i and εiT is the error term. Longer estimation periods are preferable as they permit to capture the normal return of a stock (McWilliams and Siegel, 1997).
16
Based on the above formula, I calculate the expected return for each firm i if no acquisition related information had been made public. Then, I proceed to calculate the excess return around the deal announcement – this is the difference between the observed real return and the forecasted one:
and are the parameters derived from the ordinary least squares regression of the return of firm i
on the market index during the estimation period. As markets might start adjusting the share price a
short period prior to the acquisition disclosure or continue to do so for a short period afterwards, a standard practice in event studies is to sum the abnormal returns over a period of trading days around the event of interest (referred to as event window). The following equation is used for the computation of the cumulative abnormal returns (CAR) for each firm i:
where the event window is defined as starting m days before the event (day 0) and finishing n days after the event, abbreviated as [-m;+n]. The length of the event window is usually a discretionary choice on the part of the researcher as scholars do not unanimously prescribe a definite time frame for conducting the analysis. On the one hand, too short event windows might fail to capture the information leakages relevant to the takeover announcement or the gradual information dissemination following the deal disclosure. On the other hand, extended event windows pose the threat of importing the valuation effect of confounding announcements. However, the majority of previous studies, among which MacKinlay (1997) and McWilliams and Siegel (1997), recommend using a shorter period in order to minimize the possibility of concurring events driving the abnormal stock movements. For instance Chang (1998) as well as Song and Walkling (2000) opt for a two-day event window. Moeller et al. (2004), Andrade, Mitchell and Stafford (2001), Gompers and Xuan (2009), Ishii and Xuan (2013) are examples of papers employing a three-day windows. Yet others such as Fuller et al. (2002) use a five day window.
3.2 Regression model and variables definition
A. Regression model
Based on the above described methodology, I am using the below regression models in order to test the hypotheses from Chapter 2:
17
In the following sub-section I describe the construction of key variables.
B. Variables definitions
Explanatory variables
Common VC. As the main purpose of this research is to assess whether purchasing companies from their
former VC sponsors is value-enhancing or value-destructive for the acquirers, the primary explanatory variables is Common VC. This is a dummy variable taking the value of 1 when both target and acquirer have been financed by venture capitalists and have at least one common VC investor. Respectively the variable equals 0 when different VC firms have funded the acquirer and the target or when the acquirer has not been backed by a VC prior to deal announcement.
backed Acquirer. This is a dummy variable equal to 1 when both acquirer and target have been
VC-backed. Its purpose is to differentiate between the two control samples and to isolate the effect of the common VC sponsor.
As discussed in Chapter 2 Literature review geographical distance has been identified as an important factor in M&A deals.
Same State. This variables accounts for geographical proximity between target and acquirer. It is a binary
variable that equals 1 when the headquarters of both acquirer and target are located in the same state.
Apart from the above described independent variable, I include explanatory variables related to the quality and magnitude of the VC investment in the target. Research on VC investments and exits success (Ragozzino and Blevins (2016), Gompers et al. (2010), Lee and Wahal (2004), Hochberg et al. (2007)) has often identified these as important determinants of the how successfully the VC firms manage to close their positions in the portfolio ventures and the premium they earn.
Years from 1st VC Investment in Target. This is a measure of the duration of the VC sponsorship, the time
period during which the company was held in the portfolio of its investors. It is target is computed as the difference between deal announcement and first VC investment in the target (in years). Gompers (1996) suggests that less reputable VC firms are inclined to exit earlier as they need to build their reputation and raise capital for new funds.
18
Number of Target affiliated VCs. The variable represents the number of venture capitalist who have
invested in the target company. I take the log of this number as the informational effect of each additional firm is not equal. The use of the measure is based on Ragozzino and Blevins (2016).
Average Firm Age. A proxy for the venture capitalist reputation is for how long it has been active investor. I
define Firm Age as the difference between the date a portfolio company first received a VC investment and the date when the VC firm first started to invest ever. The definition is based on Hochberg et al. (2007) and Gompers, Kovner, Lerner and Scharfstein (2006). Some studies such as Sorenson and Stuart (2001) have used the year when a VC firm was founded instead of the one it started making investments. However, Thomson One PE has a large amount of missing data regarding the foundation period of some VCs. Hence, instead of dropping such observations, I opt for using the date when a VC started actively investing. I believe that the latter date is even a more accurate proxy for the period when the firm started building its reputation and connections in the VC industry. Afterwards, I calculate the Average Firm Age across all disclosed VC sponsors of a target.
Average Firm Investment Experience. An alternative measure of venture capital firm experience follows
the definition used in Gompers, Kovner, Lerner and Scharfstein (2006) as well as Gompers and Xuan (2009). It is based on the number of companies a VC firm has sponsored prior to financing the target company in question. By the year of first financing round in the target, each VC firm investment activity is adjusted with average prior investment activity of the VC firms in the Thomson One Private Equity module. For the computation a logarithm is employed because each additional investment contributes more to the reputation of a firm that has engaged in relatively less investments than it does to the reputation of a firm that already has many companies in its portfolio (Gompers et al. (2006)). Hence, the experience of the firm is determined as logarithm of one plus the number of companies the VC firm has invested in prior to the first investment round of the target company. From this value I deduct the logarithm of one plus the average number of companies the average VC firm has invested in prior to the same year. Again, I take the average across all disclosed VC sponsors of the targets in my sample.
For both variables, I create dummies that designate if the measure of firm experience is above or below the median experience in the distribution.
Control variables
In addition to the above described measures, I employ a number of control variables related to deal, acquirer or target characteristics. These include both continuous and binary measures for acquirer size, acquirer level of debt and cash holdings, acquirer book-to-market ratios, relative deal value, method of payment, same industry deals, target’s age at acquisition or at first round of VC disbursement. The use of the variables is motivated by the studies discussed in Chapter 2 Literature review. Control variables precise definitions are provided in table descriptions
19
Chapter 4: Data, Sample Construction and Summary Statistics
The current chapter presents the data sources and sample construction procedure as well as summary statistics.
4.1 Data sources and sample construction
In my research I use several databases. I obtain information related to VC firms and their portfolio companies from the Private Equity (PE) module of Thomson ONE (former VentureXpert). From PE Investments sub-module I download all VC investments made in U.S. companies between Jan 1st 1970 and Dec 31st 2016. Following Kamath and Yan (2009), I exclude the following type of investors: individuals, corporate venture programs, insurance subsidiaries and affiliates, incubator and development programs, government affiliated programs, university affiliated programs, endowments, foundations or pension funds. VC sponsors could be both U.S. and non-U.S. based. I require that at least one of the of the company’s investors has been disclosed.
Information regarding deals is retrieved from the Thomson ONE M&A module. I include only completed transactions announced between Jan 1st 1989 and Dec 31st 2016 where the acquirers are U.S. public companies and the targets are U.S. private companies. I restrict the observations to transactions with disclosed deal value. Transactions classified as spinoffs, tender offers, repurchases and other specific deals are excluded. Following earlier research on acquisitions of VC-backed targets (Masulis and Nahata (2011), Gompers and Xuan (2009), Kamath and Yan (2009)), I exclude deals in which the bidder acquired less than 100% of the target. Companies from the financial services and utilities sectors are not included (2-digit SIC codes 60 to 67 and 40 to 49). Disclosure of both acquirer’s and target’s names is required.
I then merge the M&A dataset with the VC investments one. I match the observations by company names – first for the targets and subsequently for the acquirers. For this purpose, I have applied a detailed name standardization routine to all companies. I also manually verify the correct names for a large number of observations that fall under certain criteria (for instance, in either dataset the name fields contain parenthesis or abbreviations such as “aka”, acquirer and target names coincide, etc.). Furthermore, I require that VC funding occurs prior to the deal announcement. In order to perform crosschecks with the sample I construct, I also download information on all VC exits via an acquisition between Jan 1st 1989 and Dec 31st 2016. However, due to the decrease in the quality and coverage of VC exits in the former VentureXpert (Kaplan and Lerner (2016)), I choose not to use this data directly, but perform the matching procedures described above.
The final sample contains only VC-backed targets with either VC-backed or non VC-backed acquirers. I proceed by identifying which of the VC-funded acquirers share at least one common VC sponsor with their target. Again, I match on firm name after a name standardization procedure has been executed.
20
In addition, in order to assess VC firm reputation and experience, I also obtain investments data by year and firm in the period Jan 1st 1960 to Dec 31st 2016. The source is again the PE module of Thomson ONE. Then, by year of first investment in the target and firm name, I add firm specific information for each VC firm financing the unique targets in the sample.
Stock returns for the bidders are retrieved from The Center for Research in Security Prices (CRSP). Following the methodology discussion in the previous section, I require the availability of bidder’s share prices for up to 250 trading days prior deal announcement and 5 trading days after. Companies with more than 30 trading days of missing information are excluded. Accounting data for the fiscal year preceding the transaction is obtained from Compustat North American Annual databases (Compustat).
4.2 Descriptive statistics
Applying the selection criteria described above leads to a sample of 2 311 acquisitions of VC-backed targets. There are three sub-samples. The group of deals where acquirer and target have at least one common VC investor is my group of interest (hereafter “Common VC Deals”). It consists of 194 (8.4%) unique observations. There are two control sub-samples – transactions where acquirer and target are both VC-financed, but do not share a common sponsor (hereafter “No Common VC Deals”) and transactions where the acquirer is not VC-funded (hereafter “Not VC-backed Acquirer Deals”). They contain 1 001 (43.3%) and 1 116 (48.3%) unique deals, respectively.
Table 1 (on page 33) presents a split of the sample and sub-samples by year of deal announcement. The number of deals is highest in the years of the so called dot-com bubble as well as the years directly preceding and following it. The transactions in the “Common VC Deals” group peak between 1995 and 2003 and decline afterwards. The observations in the two control sample “Not VC-backed Acquirer Deals” also decline after the mid-2000s, but at a slower pace. Besides that they are relatively uniformly distributed. The group “No Common VC Deals” is approximately uniformly distributed after the transactions increase in the mid-1990s.
An industry breakdown of the deals using Fama- French 48-industry categories is displayed in Table 2 (on page 34). A split by acquirer industry shows that Business Services, Computers, Electronic Equipment and
Pharmaceutical Products are the most widely presented sectors in all three sub-samples. An important number of deals are also assigned to Measuring and Control Equipment, Electrical Equipment, Medical Equipment. Panel B of Table 2 shows the transactions depending on the industry of the target. The industry composition exhibits the same pattern as in Panel A. Most popular categories remain Business Services, Computers, Electronic Equipment and Pharmaceutical Products. Finally, Panel C depicts distribution of same industry acquisitions across the sub-samples. Same industry deal are transactions in which acquirer and target belong to the same Fama-French 48- industry category. Given, the similar industry structures
21
presented in Panels A and B, it is not surprising that most of the deals are classified as same industry ones. The share of the latter varies from 58% to 66% in the three groups.
Provided the industry and year clustering observed in the above tables, I include industry and year fixed effects in the regression analysis.
Summary statistics about key acquirer, target and deal characteristics are provided in Table 3 (on page 36). Venture capital-financed acquirers are relatively smaller in terms of assets compared to non-venture capital-financed acquirers. However, the difference on this dimension is not significant across the sub-samples. The mean market capitalization of the three groups exhibits the opposite trend. It appears that the market values more positively the former venture-backed acquirers in my sample. Book to market is slightly higher for non-VC-backed bidders. The patterns in cash to assets and debt to assets ratios are more clear. Venture capital-financed acquirers have higher share of their assets in cash, cash equivalents and term investments while the share of long term debt, current portion of long-term debt and short-term debt is lower.
The group of not venture capital-financed bidders tends to acquire older targets that have been venture capital financed for a longer period. The sub-sample Common VC appears to purchase the youngest companies and the companies from which VC sponsors exit the fastest. Unsurprisingly, targets of bidders from Common VC group have been funded by a larger number VC investors.
Venture capital-financed bidders are more likely to purchase targets located in the same state. The trend is stronger for Common VC group. The probability that a transaction involves acquirer and target with the same Fama – French 48-industry category is comparable across samples. However, in accordance with earlier research there is a strong pattern in the preferred method of payment. The probability of Common VC acquisitions to be financed entirely with equity is substantially higher. Expectedly, the likelihood of transaction paid 100% in cash has the opposite trend. Percentage of deal paid in stock (cash) denotes the percentage of the transaction value paid with stock (cash). The Common VC group acquirer uses, on average, more stock and less cash to finance his deals.
22
Chapter 5: Results, Discussion and Robustness Tests
In order to assess the market reaction to acquisitions of VC-backed private targets I use the event study method described in Chapter 3. I am particularly interested in the cases when bidders are formerly VC-financed companies and they acquire companies from the portfolio of their former VC investors.
For each unique deal in my sample, I calculate the cumulative abnormal returns around the announcement period. I start by forecasting the expected normal returns had the transaction not been announced. I use the market model and choose the value-weighted CRSP index as a proxy for the market index. As a robustness check, returns using equally-weighted CRSP index as well as the Standard and Poor’s index are estimated. Forecasted normal returns remain qualitatively unchanged.
Gompers and Xuan (2009) as well as Ishii and Xuan (2013), who also study the role of inter-firm connections in M&A deals, set the estimation period to -200 to -20 days prior deal announcement. They require that no more than 30 days of daily stock prices are missing per firm. I adopt similar approach in my research. Normal returns are forecasted over an estimation window of 200 days, set to [-220;-20] trading days prior deal announcement. However, I have tested the effect of using both larger and smaller windows. Estimation windows of [-250;-20] and [-150;-20] have led to qualitatively similar predictions for the companies’ normal rates of return.
Following the literature on event studies summarized in Chapter 3, I choose a short event window of [-2;+2] trading days around announcement day (day 0). In this way I will be better able to isolate any abnormal movements in the acquirer’s market valuation that linked to the takeover announcement and minimize the threat of confounding events during the event window. The cumulative abnormal returns are winsorized at the 2% level.
I proceed with the univariate and multivariate analysis of acquirer’s cumulative abnormal returns.
Univariate analysis of cumulative abnormal returns (CARs)
Table 4 (on page 38) presents a univariate analysis of the cumulative abnormal returns (CARs) for each of the three sub-samples in my research – deals where acquirer is not VC-financed (group (1)), deals where acquirer is VC-financed, but do not share a common VC investor with his target (group (2)) and deals where acquirer is VC-financed and has a common VC sponsor with his target (group (3)).
Bidders who share a connection with their targets outperform their counterparts in the control sub-samples. They achieve an average CAR of 1.82% against mean CARs of 0.98% and 0.45% for groups (1) and (2) respectively. I also check whether these results are statistically distinguishable between my sample of
23
interest and each of the control ones. Based on a t-test that accounts for the uneven sample sizes and variances, only the difference in stock market performance between groups (2) and (3) is statistically different from zero at the 10% level. However, as hypothesized in Chapter 2, the market reaction should vary in specific cases when the information asymmetry problem is more exacerbated or specific target and VC firm characteristics serve as additional market signals. Therefore, I split each sub-sample based on such measures..
I first check the difference in market reaction depending on the geographic proximity of acquirer and target. As discussed in Chapter 2, the greater distance poses difficulties for the acquirer to assess the quality and true value of the target, to conduct due diligence as well as to integrate the target in its business. On the other hand, proximity facilitates all these processes. This suggests that the market reaction would not be unanimous. The analysis in Table 4 confirms that the common venture capital connection is more valuable in deals between companies with headquarters in different states. Common VC group achieves substantially higher abnormal returns of 2.80% compared to 0.92% and 0.34% for sub-samples (1) and (2) accordingly. The reported performance is statistically different between the sub-samples. On the other hand, when the transaction involves companies in the same state, there is no distinguishable difference between the Common VC group and the control ones. It appears that proximity diminishes the informational advantage brought by the common VC investor. The data in this dimension is in accordance with the findings of Gompers and Xuan (2009).
Another case of important information asymmetry are acquisitions of young companies. Analysis in Table 4 provides evidence for that. When target’s age at deal announcement is below the median age, acquirers from Common VC group outperform significantly their counterparts in the control groups. The bidders from the sample of interest achieve a mean CAR of 2.44% against 0.60% and 0.17% for groups (1) and (2). Though, the difference is statistically distinguishable only in comparison with the No Common VC group. This could be due to the fact that Not VC-backed Acquirer group generally purchases more mature companies as reported in Table 3. It is possible that the bidders select later-stage VC-financed companies over seed-stage for instance.
As discussed in previous sections the certification role of VC firms varies with their reputation in the market (Gompers, (1996)). I employ three separate proxies for the prestige of VC investors. Detailed definitions are provided in Chapter 3.
I start by a split that accounts for the duration of VC involvement. When the number of years between first investment and exit are below the median, the presence of a common VC firm seems to be beneficial. The mean CAR of acquirers in the sample of interest amounts to 2.45%. The market reaction is qualitatively and statistically significant. CARs for the control sub-samples are much lower – at levels of 0.40% and 0.46%, respectively.
24
Afterwards, I use a measure of how long on average the VC firms financing a target have been active investors. Average VC firm age is the difference in years between the date the average target VC sponsor first invested in it and the date this firm first started to ever invest. Acquiring from a VC investor with below median age is positively perceived by the market when bidder and target share a common VC sponsor. Mean CAR is 3.61% for the sample of interest. Not VC-backed Acquirers and No Common VC achieve CARs of 0.55% and 0.38%, respectively. The mean difference with each control group is statistically significant at 5% level.
Another proxy for the VC firm prestige follows a measure developed in Gompers, Kovner, Lerner and Scharfstein (2006). It is computed as the log of one plus the number of companies the VC firm has invested in prior to investing in the target less the log of one plus the number of companies the average VC company invested in prior that same year. Acquisitions from VC firms with below the median experience are value-creating for the bidder from the Common VC group. Mean CAR is 3.10% and statistically distinguishable from the control sub-samples. Acquisitions from more experienced VC investors are qualitatively and quantitatively similar across the three groups.
The analysis related to VC firm reputation yields results opposite to expectations outlaid in Chapter 2 and to the findings in Gompers and Xuan (2009). This tendency could be explained with a potential rent-seeking behavior on the part of the VC firm as suggested in Kamath and Yan (2009). Their study attributes a value destructive effect on acquisitions from former VC sponsor without distinguishing between VC firm reputation. It is possible that only more experienced VCs are able to extract higher premiums that net out the positive impact of the connection.
In order to further assess the results of the univariate analysis, I proceed with a regression one.
Multivariate analysis of cumulative abnormal returns (CARs)
I first conduct OLS multivariate regressions for hypothesis 1a and 1b. The results are presented in Table 5
(on page 40) . Key independent variable is a dummy variable Common VC that takes the value of one when acquirer and target have at least one common venture capital investor, and zero otherwise. I control for the difference between VC-backed and non-VC-backed acquirers by introducing a dummy variable equal to one when the acquirer is formerly VC-backed and zero otherwise.
Specifications (1) to (3) test the effect of acquirer and target having a common VC investor. Models (4) to (5) account for the geographical proximity between the two companies. The coefficients on my main explanatory variable are positive across all five specifications, but remain statistically insignificant throughout (1) to (4). When I introduce an interaction term between the Common VC variable and the dummy for same state deal, the effect of the shared VC sponsor is statistically distinguishable. It leads to 1.8% higher CARs for bidders in the sub-sample of interest. The results are in accordance with the observed
25
data in the univariate analysis. The impact of the shared VC investor on CAR is dependent on whether the target is closely located to the bidder.
I then proceed by examining how the reputation of a VC investor impacts the acquirer’s realized return. I am interested in whether the common VC sponsor connection is differently perceived by the market when this VC is more or less reputable. This analysis is presented in Table 6 (on page 41) . As previously discussed I use three different proxies for VC reputation is order to check whether the findings in Gompers and Xuan (2009) are robust to different definition of VC firm prestige. Specifications (1) to (4) use the reputational proxy of how long the target has been held in the VC portfolio; specifications (5) to (8) employ the time the average target VC firm has been an active investor prior to funding the target in question. Models (9) to (12) use the measure adopted in Gompers and Xuan (2009). The results are comparable and consistent across all three specifications. However, the coefficients on the main variable of interest are significant only in models (7) and (8). These models use the reputational proxy for average VC firm age (this is the difference in years between the date the average target VC sponsor first invested in it and the date this firm first started to ever invest). If the common VC firm has been active investor for longer than the median firm, any positive effect from the VC connection is alleviated. A net negative –0.8% CAR is estimated. Due to the small sample size the rest of the specifications remain statistically undistinguishable.
Both univariate and multivariate analysis confirm the value creating effect of having a common VC sponsor when acquirer and target are more distantly located. However, when I take into account the prestige of the VC investor, my results refute the conclusion made in Gompers and Xuan (2009). It appears that more reputable VCs are able to extract a higher premium at the expense of bidders as suggested in Kamath and Yan (2009).
Robustness checks
In order to assess the robustness of my results, I conduct the same regression analysis using a different event window. I opt for a period of [-1;+1] trading days around deal announcement. The same event window in employed by Gompers and Xuan (2009). The cumulative abnormal returns are again winsorized at the 2% level. Models in Table 7 (on page 43) and Table 8 (on page 44) are reciprocal to the ones used in Tables 5 and 6.
The results for specifications (1) to (5) in Table 7 remain qualitatively similar to the ones presented in Table 5. The coefficient on the variable Common VC is statistically not significant, but very close to the significance level of 10% (t-stat of 1.64). Results reported in Table 6 remain robust as confirmed by the analysis in Table 8.
26
Chapter 6: Limitations
This section briefly discusses the main limitations of my research. I proceed by describing the sources of concern regarding the adopted methodology and then regarding the data and sample construction.
6.1 Method
A major limitation of the employed methodology is the fact that the event studies method is heavily dependent on three key assumptions – markets are efficient, the event is not anticipated and there is not any confounding impact from other events. Thus, if any of the before-mentioned is violated, the validity of the results and interpretations made in this paper could be biased. I have chosen to use short event windows in my study in order to minimize the risk of concurrent events driving the stock performance around the announcement day. I also check in Thomson One M&A that none of the takeover announcements in my sample has started as a rumor. Therefore, the major threat to the validity of the method employed in my research remains the efficient market hypothesis assumption. A number of empirical studies (Fama, 1998; Fama and French, 2007; Shleifer, 2000) do not refute the rationality of markets, but do question their ability to remain efficient all the time.
Another drawback is the impossibility to use accounting measures in order to assess the impact of the acquisition on the joined firm or to compare the accounting and stock market performance. For the calculation of return on equity (ROE), return on assets (ROA) or another indicator of profitability of the combined firm, I would need the financial data of both acquirer and target prior to the effective date of the deal. As the targets in my sample are private, they are not obliged to disclose such information and their financials are not readily available in databases such as Compustat. An alternative approach would be to manually collect accounting information regarding the target companies. The Securities and Exchange Commission requires the acquirers to disclose the financial data of their newly purchased firms when the deal value equals or exceeds 10% of the acquirer’s own market valuation. Masulis and Nahata (2011) impose such a 10% cutoff as their study includes private targets and are, thus, able to analyze the post-merger financials of the combined entity. However, adopting such a constraint in my research poses a twofold problem. First, it will gravely diminish my sample of interest – the sample of acquirers and targets sharing at least one common venture capital sponsor. Second, manually collecting the information for all the entities in my three samples (one sample of interest and two control ones) is beyond the timeframe of my research. Therefore, such an analysis is left for a wider and more extensive study of the current topic.
6.2 Data
The main challenge for studies of venture capital firms and their portfolio companies remains the relative scarcity of the publicly available information. As noted in Kaplan and Lerner (2016), contrary to mutual
27
funds, private equity firms are generally not obliged to disclose neither their financials nor their portfolio holdings. The small number of data providers gather information from public fillings, news releases, large institutional investors and limited and general partners at private equity firms. Given that most of these platforms are U.S. based, it is not surprising that the firms and portfolio companies covered are predominately U.S. based as well.
The two databases, most prevalent in academic research are Venture Source, a unit of Dow Jones, and the private equity module of Thomson One (former VentureXpert, a unit of Thomson One’s discontinued Venture Economics division). A recent assessment of the quality of venture capital data by Kaplan and Lerner (2016) points out that even these platforms suffer from incompleteness and serious inconsistencies. They also face a decline in the quality of the information provided during the last decade. The quality issue is particularly exacerbated in the former VentureXpert due to the decrease in outcome coverage (Kaplan, Sensoy and Stromberg, 2002). Overall, Dow Jones’ Venture Source is viewed as a more extensive source for exits information while Thomson’s private equity section provides more complete data on investments.
However, further concern with both databases is that they update backwards the names of the private equity firms as well as portfolio companies. Keeping only the most current names makes tracking historical records problematic. This coupled with the lack of widely used identifiers (such as cusip, permno or gvkey) poses great difficulties for scholars. In fact, matching within the former VentureXpert is also no less of a challenge as many portfolio companies have not been assigned an identifier and no unique indicators have been created for firms nor for funds. In addition, it is very hard to introduce a common practice for filling in company names within a database and near impossible across different data providers. How long a name should be, whether references to past company names should be included, how the latter should be coded, what the correct spelling of a name is are among the decisions usually left at the discretion of the individual data collection specialist.
Thus, an important concern in my study are the sample construction procedure and sample selection. On the one hand, it is possible that sample bias has been inherent to the original data as the names of some of large private equity investors have not been disclosed in Thomson ONE. Furthermore, as discussed above, the majority of the data gathered comes from limited and general partners at private equity firms and, therefore, coverage is dependent on the connections of the data provider. On the other hand, I also cannot exclude the option of introducing bias during the sample construction. I employ a detailed standardization routine to all company and firm names, perform several crosschecks between Thomson One sub-modules for PE investments and PE exits and Thomson ONE M&A module. I also manually verify the correct names for a large number of observations that fall under certain criteria (for instance, they contain parenthesis or abbreviations such as “aka” or “fka” in the name fields). Hence, it is still possible that I have dropped unnecessarily observations or failed to rectify some of the name discrepancies present. Though, given the scrutiny with which the matching procedures were conducted, misallocation of observations are less likely.
28
Chapter 7: Conclusion
In conclusion, my analysis confirms that the market reaction to deal announcement varies in specific cases. The effect differs when the information asymmetry problem is more exacerbated or target and VC firm characteristics serve as additional market signals. Due to the small sample size, my results are not robustly significant across different empirical models. Nevertheless, the analysis confirms that the common venture capital connection is value creating when acquirer and target are headquarters in different states. The analysis related to VC firm reputation yields results opposite to the findings in Gompers and Xuan (2009). When the shared VC investor is less experienced, acquisitions produce positive deal announcement returns for the bidder. This tendency could be explained with a potential rent-seeking behavior on the part of the prestigious VC firm as suggested in Kamath and Yan (2009). My results imply that only more experienced VCs are able to extract higher premiums that net out the positive impact of the connection.
References:
1) Akerlof, G. 1970. The market for lemons: Quality uncertainty and the market mechanism. Quart. J. Econom. 84 4889–5000.
2) Andrade, G., Mitchell, M., Stafford, E. (2001). New evidence and perspectives on mergers, Journal of Economic Perspectives 15, 103-120.
3) Baker, Malcolm P., and Paul A. Gompers, 2003, The determinants of board structure at the initial public offering, Journal of Law and Economics 46, 569-598.
4) Barry, C., Muscarella, C., Peavy, J., Vetsuypens, M., 1990. The role of venture capital in the creation of public companies: evidence from the going-public process. Journal of Financial Economics 27, 447–472.
5) Brav, A. and Gompers, P.A. (1997). Myth or reality? The long-run underperformance of initial public offerings: Evidence from venture and nonventure capital-backed companies. Journal of Finance, 52, 1791–1821.
6) Brown, S., Warner, J. (1985). Using daily stock returns: The case of event studies. Journal of Financial Economics 14, 3-31.
7) Cai, Y., Sevilir, M., (2009), Board Connections and M&A Transactions, Journal of Financial Economics, 103, pp. 327 – 349
8) Chang, S., 1998. Takeovers of privately held targets, method of payment, and bidder returns. Journal of Finance 53, 773–784.
9) Cohen, L., Frazzini, A., Malloy, C., (2008), The small world of investing: board connections and mutual fund returns, Journal of Political Economy 116, 951–979
10) Coval, J., and T. Moskowitz, 2001, Geography of Investment: Informed Trading and Asset Prices, Journal of Political Economy 109, 811–841.
11) Coval, J., Moskowitz, T., 1999. Home bias at home: local equity preference in domestic portfolios. Journal of Finance 54, 2045–2074.
12) Da Rin, M., Hellmann, T., & Puri, M. L. (2011). A Survey of Venture Capital Research. (CentER Discussion Paper; Vol. 2011-111). Tilburg: Finance.
13) Du, Q. (2016). Birds of a feather or celebrating differences? The formation and impacts of venture capital syndication, Volume 39, Part A, Pages 1–14
14) El-Khatib, R. , Fogel, K. and Jandikc, T. (2015), CEO network centrality and merger performance, Journal of Financial Economics, Volume 116, Issue 2, May 2015, Pages 349–382
15) Erel, I., Liao, R. C. and Weisbach, M. S. (2012), Determinants of Cross-Border Mergers and Acquisitions. The Journal of Finance, 67
16) Faccio M, Mcconnell J, Stolin D. 2006. Returns to acquirers of listed and unlisted targets. Journal of Finance and Quantitative Analysis 41: 198-220
