The influence of VC-involvement on the post-
IPO valuation of firms
Do VCs add value within the portfolio firm?
ABSTRACT
Using a sample of 1.095 IPOs from the period 2005-2012, we research the non-financial value added by VCs to their portfolio companies. Next to that, the difference between US- and European based VCs is investigated. Due to the activities and roles VCs take on next to financing we suggest that VC-backed firms are valued higher than non-VC backed firms. Also, because of a higher level of experience we expect US-based portfolio firms to be priced higher than European firms. To prove our hypothesis we use a difference in means test and a regression analysis with a basic valuation model. Our results show that VCs add value within the portfolio firm. Our multivariate analysis shows no difference between European and US-based IPOs. However, the difference in means test shows, contrary to our hypothesis, a lower underpricing for European firms. The methodology is novel to this research domain but has proven sufficient to measure valuation effects.
JEL-classification: G24, G32
Keywords: Venture Capital, Valuation, IPO Combined master thesis for the programs: MSc. Finance
MBA specialization Small Business & Entrepreneurship Supervisor:
K. Sklavos
2 1. Introduction
In his analysis on venture capital literature, Barry (1994) suggested the following question for future research: ‘Do venture capitalists add value within the portfolio firm?’. The literature is still divided whether venture capital funds (VCs) add to the success of a start-up. Most recently, Bessler and Seim (2012) found in their empirical analysis that a sample of European VC-backed start-ups consistently and significantly outperformed a group of non VC-backed start-ups. They researched the level of underpricing and the long-run financial performance metrics of the initial public offering (IPO) of over 500 firms for the period 1996-2010. A seminal empirical paper by Brav and Gompers (1997) with a period of 20 years and a sample consisting of almost a thousand NYSE venture-backed IPOs came to a similar conclusion using equal weighted returns. However, many studies find different results. For example, Brau et al. (2004) examined the IPOs of a set of small VC-backed manufacturing firms for underpricing and a set of three-year performance metrics. Contrary to other literature, their results indicate no significant differences between VC- and non VC-backed firms. This paper extends to the existing literature by answering the following research question: ‘Do VCs add value to start-ups, and if so, are US VCs more capable than European VCs in doing this?’.
This research is an addition to the current literature for two reasons. Up till now the literature has predominately used underpricing and other return metrics to measure the performance of non VC-backed against VC-VC-backed IPOs. However, stock market returns can be a weak test if one wants to know whether VCs add value to a portfolio company. Following the methodology of Gregory and Whittaker (2012), a basic valuation regression model is used based on book value, abnormal earnings and an ‘other information’ variable. The ‘other information’ variable can explain effects on the value of the firm that are not explained by book value or residual income. This way it can be confirmed whether or not investors value VC-involvement.
3 more mature and experienced (Ooghe, Manigart, & Fassin, 1991). Therefore, it can be expected that there is a difference in the performance between US and European IPOs of VC-backed firms.
The remainder of this research is structured as follows. Next, a theoretical background will be given regarding the value-adding activities of VCs and the differences between European- and US-based VCs. This section will end with the development of the hypothesis. The third section will outline in more detail the methodology as is suggested by Gregory and Whitaker (2012). The fourth section will contain the results and a discussion of the tests. The fifth and final section will present the conclusion of this study.
2. Theoretical background
Venture capital funds (VCs) are an intermediary between investors and entrepreneurs. VCs serve as an equity capital provider for high risk high growth companies, mostly start-ups. In 2011 almost $30 billion got invested in over nearly 4.000 venture capital deals in the US alone, making it an important part of our financial system (NVCA, PWC, 2012). They channel capital from pension funds, banks, individuals, corporations, insurance companies and other sources of capital to portfolio firms.
Fried and Hisrich (1994) give an overview of the benefits that VCs give from a supply-side perspective as well as from a demand-side perspective. From a supply-side perspective VCs serve as an information-producing and decision-making agent for investors. They provide the benefits of lower information gathering costs than their capital providers could achieve individually. Besides that, it acquires economies of scale for service providers such as attorneys, accountants, management recruiters and other industry contacts. Lastly, the investors benefit from the learning curve of the VC when it comes to screening, information gathering and overseeing or guiding new enterprises.
4 ventures. They provide this capital without the confidentiality issues that go paired with going public, thus start-ups do not lose potential economic rents on proprietary information. Most importantly, next to providing capital, the VC may provide business advice, contacts and reputational capital that are not available from the public equity markets.
The VC-process
The activities of VCs are described in five steps by Tyebjee and Bruno (1984). First, a potential deal comes to the attention of a VC, this event is called deal origination and can manifest itself in several ways. Since a small firm with cash needs is unlikely to be actively identified by VCs, a lot of deals originate by referral or through other VCs looking for a co-investor. Other ways are by cold contacts or technology scans.
The second activity described by Tyebjee and Bruno (1984) is deal screening, a process in which the number of potentially originated deals gets cut down to a manageable stack of deals. This is done by preferences for technologies, markets and products that the VC is more familiar with. In the third step, deal evaluation, the business proposal undergoes a more thorough assessment. Fried and Hisrich (1994) indicate several screens a potential deal goes through within the VC. First of all, the potential deal has to fit the VC in terms of investment size, industry and geographical location. After the VC firm-specific screen a generic screen is applied in which the business plan is reviewed in terms of generic criteria. Only after passing through these first two screens additional information is gathered on a proposal for a more thorough evaluation. Variables in this assessment are not only the idea on grounds of risk, return, growth and market potential but also the entrepreneurs’ capacities are considered.
5 willingness to invest in the firm may be an important factor in the lead VCs decision to invest, basically a simple sanity check with an experienced and knowledgeable party. A second reason behind syndication is tradition; in early stages the US venture capital industry was small, therefore investments got syndicated in order to diversify investment risks (Bruton, Fried, & Manigart, 2005). Currently VCs are of a size that financial risk can be diversified within the portfolio of the VC itself, therefore, syndication is merely for nonfinancial resources like legitimacy and knowledge sharing (Sorensen & Stuart, 2001). The fourth step in the sequence described by Tyebjee & Bruno (1994) is deal structuring, which encompasses the negotiation on the price and equity share as well as the covenants the entrepreneur has to adhere to.
Non-financial value-added
The post-investment activities are the fifth and last step given by Tyebjee & Bruno (1994); after the deal is struck the role of the VC changes from financier into a consulting role. It lends non-financial resources to the firm in order to maximize its success rate. Next to that, it is common practice that the VC takes seats in the board of directors of the portfolio firm.
In a multi-theoretical approach Wijbenga et al. (2003) indicate three value-adding roles of VCs through their positions taken on the board of directors of the new venture. Based on the three functional roles of the board of directors the post-investment activities are:
i) Networking activities
VCs are considered to have a boundary spanning role, aiding the portfolio firm in getting acquainted with potential suppliers, customers, service providers, investors and other industry contacts (Timmons & Bygrave, 1986).
ii) Monitoring activities
6 evaluation of the strategy and market opportunities of the portfolio firm (Harrison & Mason, 1992).
iii) Strategy-making activities
The board of directors takes part in strategic decision-making and has the responsibility of decision-maker of last resort (Fama & Jensen, 1983). The seats held by VCs on the board thus actively partake in formulating the venture’s business strategy, and in case of short-term crises or problems they offer assistance (Harrison & Mason, 1992).
Next to these activities other value-adding inputs are personnel management, management recruitment and development and operations (MacMillan, Kulow, & Khoylian, 1989). Moreover, VCs leave their mark on the portfolio firms by making them adopt certain decision-making styles and techniques as they require reporting- and information gathering standards (Ehrlich, de Noble, Moore, & Weaver, 1994; Flynn & Forman, 2001). Other intangible resources brought to the table by VCs are credibility towards suppliers, customers and bankers (Timmons & Bygrave, 1986) and even moral support for the CEO (Fried & Hisrich, 1995). All in all, VCs have a large impact on the organizational development and structuring of the portfolio firm. In a survey containing 20 papers on non-financial value added by VCs, Large and Muegge (2008) identified all the value-adding inputs. They created a typology according to an external orientation containing two categories and an internal orientation containing six categories (table 1).
Table 1: Non-financial value-adding inputs by VCs
Internal External Recruiting Mandating Strategizing Mentoring Consulting Operating Legitimation Outreach
Source: Large and Muegge (2008)
7 limited involvement or laissez faire, moderate involvement and close tracker involvement. According to Sapienza (1992) the nature and style of VC-CEO interactions have a significant impact on the value of venture capitalist involvement. He concludes that more involvement and an open relationship would result in superior results regarding value creation.
VCs and IPOs
Venture capital funds differ from other intermediaries not only in their specialization for new ventures with high information problems. Bascha and Walz (2001) give as a main characteristic the limited period for which VCs finance their portfolio firms. This financing constraint is considered an instrument by the limited partners of the venture capital funds to ensure the distribution of investment returns. Therefore, after a limited time period the VC seeks to divest the portfolio firm and reap its returns. Bascha and Walz (2001) distinguish between five exit mechanisms, namely liquidation (in case of failure), share repurchases by the founder (buy-back) and selling shares to institutional investors (secondary purchase), trade sales (TS) and initial public offerings (IPO). The latter two are the most popular and account for approximately two-thirds of the divestment methods in Europe. However, public offerings have increased in popularity since 1997 as where trade sale (selling the company as a whole to another firm) decreased in popularity (European Investment Bank, 2001). IPOs are also preferred for the return they offer VCs, which is with 60%, four times larger than the average return associated with acquisitions (Black & Gilson, 1999).
9 Thus, VC managers ought to add non-financial value to a portfolio firm, plus the debated certification role of VCs dictates that the monitoring activity means a decreased level of information asymmetry. It can be assumed that investors in the secondary market appreciate these factors and include it in the firm value. Therefore, the following hypothesis is deduced:
Hypothesis 1: VC-backed firms are more valuable post-IPO than non VC-backed firms
Differences between US and European VC
Bessler and Seim (2012) suggest that there is a difference between the performance of US and European VCs due to the information advantage of US-based VCs because of their higher reputation and greater experience. Ooghe et al. (1991) and Manigart (1994) support the proposition that US VC-backed firms perform better than European VC-VC-backed firms because of the experience and better position in the institutional environment. The private equity and venture capital financial model was initially developed in the US and designed to fit the US institutional environment (de Lima Ribeiro & Gledson de Carvalho, 2008). There is clear evidence that a better institutional- and legal environment reduce the perceived risk of investing, therefore reducing the information asymmetry problem and the level of underpricing (Hopp & Deher, 2007; Engelen & Essen, 2010). Thus, a clear link between the institutional environment and the level of underpricing might indicate that there might be performance differences between US and European VCs.
10 of contact between the VC manager and the CEO of the firm is on average more than twice as high for the US (194.05 hrs) compared to Europe (77.16 hrs). In a survey amongst CEOs and their VC manager, Sapienza (1992) found that the perceived value of the involvement was greater when it was frequent and open. Following the arguments that there is a difference in the level of syndication, the institutional environment, experience, interaction with the portfolio firms and emphasis on NFVA-roles between European and US-based VCs the following hypothesis can be deduced:
Hypothesis 2: US VC-backed firms are more valuable post-IPO than European VC-backed firms
3. Data and Methodology
The research methodology employed in this research is borrowed from capital market-based accounting research. It is well-known in the accounting literature to test, for instance, whether markets value research and development activities by companies (Lev & Sougiannis, 1996). In their paper on the valuation effects of corporate social performance (CSP), Gregory and Whittaker (2012) argue that in certain occasions concentrating on stock market returns can be a weak test. With a simple numerical example they show that unless firms actually change their CSR policy, return tests do not have the power to detect CSP. For the case of venture capital-involvement return tests can show whether or not VC-backed firms outperform non VC-backed firms, however, it fails to include the valuation of the portfolio firm by shareholders post-IPO. Conform the key value driver formula (Koller, Goedhart, & Wessels, 2010) the continuing value of a firm is a product of the invested capital, the return on invested capital, the cost of capital and the growth rate1. The return metrics can be considered the cost of equity, and are only a small part in the valuation formula. Therefore, concluding that VCs add value by using only return metrics can be misleading. Next to that, the non-financial value added by VCs is most likely shown in the value through a higher growth rate and a larger return on invested capital. Although the cost of equity is part of the key value driver formula, measuring equity capital directly is closer to measuring the true value of the firm. Gregory and Whittaker (2012) use a methodology that can detect valuation effects of CSP in a more precise manner than tests based upon Tobin’s Q. The
1
11 valuation model employed by them is developed by Ohlson et al. (1995) and is described in more detail in the next section. The starting point is the rational market valuation of the firm, assuming that in an efficient market the value of the firm is equal to the present value of its expected future cash flows. The analysis boils down to an OLS regression model with valuation measures and an ‘other information’ parameter to distinguish VC-involvement. This method has proven accurate to test for effects of R&D-expenditure, brands and CSP on value (Barth, Clement, Foster, & Kasznik, 1998; Gregory & Whittaker, 2012; Lev & Sougiannis, 1996). In the next section the variables will be discussed in more detail.
3.1 Variables
In this section the dependent, independent and control variables used in the regression analysis will be further discussed.
3.1.1 The basic valuation model
The regression model is based on a basic valuation model which expresses future cash flows in terms of abnormal earnings, cost of equity and accounting book value (Rees, 1997; Ohlson, 1995). The basic valuation model can be expressed in the following way:
∑
Pt is the share price, bt the book value and xt is the earnings minus the required return of investors
on the opening book value (abnormal earnings). For re a market index can be used, in this case the
12
Chart 1: Return on MSCI World Index (%)
This chart shows the returns on the MSCI World Index for the years 2005-2012. In the table at the bottom of the graph you can find the returns per year.
The basic model will regress the share price on book value per share, abnormal earnings per share and an ‘other information’ variable, besides the control variables. By regressing at a per share level, problems of heteroskedasticity can be avoided that arise when using regular book value or abnormal earnings (Barth, Clement, Foster, & Kasznik, 1998; Barth & Clinch, 2009). Lastly, in the multivariate analysis the dependent variable is the log of the share price of the portfolio firm after the IPO. This deviation from the methodology described by Gregory and Whittaker (2012) is because a poor distribution of share prices across the sample, in order to come to a normal distribution the log of the share prices is used.
Since expected future earnings is mostly not publicly available information except for analyst forecasts, Gregory and Whittaker (2012) operationalize the formula with the assumption that the growth rate of the future cash flows will remain constant on the long term. As mentioned above, Rees (1997) and Ohlson (1995) extend the model to incorporate an additional parameter that can be regarded as any information that helps explain the value of the firm that is not captured by earnings or book value. With these three variables the basic form of the regression model can be formulated as (Gregory & Whittaker, 2012; Rees, 1997):
13 The test is to see whether the coefficient returns either a negatively or positively significant explanation to the effect of the ‘other information’ variable (γit) on firm value. In our first hypothesis
this regressor will be the VC-dummy variable which is 1 when the firm that goes public is VC-backed and zero otherwise. The second hypothesis will be tested by using the other information variable as a dummy-variable to distinguish between US and European IPOs on a sample of only VC-backed firms.
3.1.3 Other variables
Past literature, mainly on the effects of VC-involvement on underpricing, has shown that a few variables are important to the magnitude of underpricing. Since this research is on the effect of VC-involvement on the shareholder valuation of the firm shortly after the IPO, it can be expected that these variables have an equally moderating effect. The variables are shown with a brief description in table 2.
Table 2: Control variables for VC-research
Variable Description
Deal size Smaller issues go paired with more speculation and volatility, affecting valuation by a higher share price (Beatty & Ritter, 1986)
Flotation The VCs and entrepreneurs have an incentive to get the most out of a deal at the IPO, therefore, the larger the public offering the more effort is done to increase the value of the share price at IPO. Measured as percentage of total equity capital offered publicly (Large & Muegge, 2008)
Hot/cold market In a hot market there is an above mean number of IPOs paired with a higher level of underpricing and volatility (Ritter, 1984).
Technology firm dummy
Because technology firms’ products are more difficult to value higher levels of underpricing were found, this might be similar for our research (Loughran & Ritter, 2004)
Syndication partners The number of VCs or other firms that took part in the syndicate before bringing the portfolio firm public (Wang, Wang, & Lu, 2003; Barry, Muscarella, Peavy, & Vetsuypens, 1990)
Average experience of syndication partners
The average years since the date of incorporation of the VCs in the syndicate (Barry, Muscarella, Peavy, & Vetsuypens, 1990; Lee & Wahal, 2004)
VC-14 involvement. Underpricing is the return for investors that had shares allocated in the primary market which they would sell at the end of the first trading day on the secondary market.
Underpricing is formulated in the following way (Ritter, 1984):
Where Pi is the share price of firm i on closing (CP) and on opening (OP) of the first trading day
on the secondary market. UP is the magnitude of underpricing for firm i. In the analysis where underpricing is used as the dependent variable or in the difference of means tests, outliers are deleted.
3.2 Data
The sample for VC-backed IPOs is drawn from the Zephyr database on mergers, acquisitions and other deals. The initial sample of 1.459 VC-backed IPOs that took place between 1997 (or 2001 for the US) till 2013 was screened for the following criterion: (1) the geographical location of the target firm is either West-Europe or the United States; (2) penny stocks with an offer price of € 1,- or less are excluded (n=823); (3) Firms that are in the banking or insurance industry (SIC-codes between 6000-6999) are excluded (n=236); (4) Lastly, the sample is manually checked to confirm that the vendor is a VC and not a private equity fund (n=177).
Penny stocks and firms in the banking or insurance industry are excluded following the methodology of Bessler and Seim (2012). The underlying reason is that penny stocks show a higher volatility. Next to that, banking and insurance firms that go public are often (investment) funds or holding companies. The distinction between VCs and PEs is important since the operations of PEs in respect to VCs are different in the sense of the non-financial value added. Where VCs have a focus on fast growth and building an organization to create value, PEs have a tendency to transform an organization in order to create value.
15 the geographical location of the target firm is either Western-Europe or the United States. Banking and insurance firms are excluded from the control sample as well.
Data on annual share prices, book value and EBIT is drawn from the Orbis database. These variables were available for all the target firms in the sample from 2005 on. For the regression model this limits the sample to 88 VC-backed firms and 1.007 non VC-backed firms. Since a number of the IPOs take place after 2005 limited data is available for the consecutive years post-IPO. For instance, a portfolio firm that had its IPO in 2010 maximally has two years of abnormal earnings reported. In many cases the data for 2012 is not available yet, which thins out the sample even more.
Table 3 shows the IPOs per country of the whole sample. The complete sample consisting of 1.095 portfolio firms stems from 21 different countries. With 44% of the firms from the US, the sample is nearly equally divided in firms from the US and Western Europe. A big proportion of the sample stems from the four biggest western European economies, France, Germany, the UK and Italy; they are best represented with a combined number of 432 IPOs. The smallest country in the sample is Gibraltar with two IPOs over the period 2005-2012.
Table 3
IPOs per Country 2005-2012
Country IPOs Country IPOs
United States 487 Luxembourg 12
France 156 Denmark 11 Great-Brittain 126 Austria 10 Germany 101 Turkey 5 Italy 49 Portugal 5 Norway 25 Greece 4 Sweden 22 Cyprus 4 Belgium 21 Ireland 4 Switzerland 17 Finland 3 Spain 16 Gibraltar 2
the Netherlands 15 Total 1095
16
Chart 2: IPOs per year for the period 2005-2012
This chart shows the IPOs per year for the period 2005-2012 for the whole sample. The average amount of IPOs per year for the sample is 184. The years 2005, 2006 and 2007 are considered ‘hot markets’ because of their higher than average number of IPOs.
4. Results and analysis
In this section the results are presented and discussed for both hypotheses. The first hypothesis focuses on whether or not VCs add value to the portfolio firm. The chosen methodology, as elaborated on in the previous section consists of a univariate, bivariate and multivariate analysis. The second hypothesis focuses around the difference between European and US-based portfolio firms. A difference in means test and regression analysis is given for the VC-backed sample divided up into European and US-based firms. Lastly, we will look into the impact of syndication of VCs and their experience measured by age.
4.1 VC-backed vs. non VC-backed
The first subsection of this chapter will focus on the first hypothesis; the difference between VC-backed and non VC-VC-backed firms when going public. The next section will show the descriptives and the difference in means test, followed by the correlation analysis and the regression analysis. The results are discussed within this section.
17 4.1.1 Descriptive statistics and univariate analysis
Table 4 shows the descriptive statistics for the independent, dependent and control variables of the regression model for the VC-backed firms. The table indicates that a VC-backed firm in an average IPO sells 38,61% of its shares for € 268 mln with an average offer price of € 12,97. The average closing price on the first day of trading on the secondary market is a lot higher with € 16,41. The trend in the share price after the first month differs a lot per share as indicated by rather large standard deviations. The sample shows on average positive abnormal earnings for all the subsequent years after the IPO. Since the majority of the deals in the sample took place in 2005/2006, the second and third years show a lower minimum in abnormal earnings as well as in the mean of the share prices caused by an overall dip in the market as can be seen in chart 1. Nevertheless, the hot and cold market dummy-variable shows that with 73,8%, almost three-quarter of the deals was timed in a hot market, supporting the notion of Gompers (1994) of the added value of market timing of VCs. The syndicates that are behind the timing of the VC-backed IPOs consist on average out of 2,092 VCs with a maximum of nine parties included and on average a little over 31 years of experience. This is quite a lot compared to other research samples in previous literature where the median was between 6 and 10 years (Gompers, 1996; Wang, Wang, & Lu, 2003).
In table 5 the descriptive statistics are depicted for the same set of variables for the sample of non VC-backed firms. The total sample of 1005 IPOs went to the secondary market for an average offer price of € 10,64 and closed the day at € 14,61. The IPOs yield on average € 188,3 mln for 30,52% of the equity capital of the company. A similar trend as in the VC-backed sample can be recognized for the share price for the subsequent years with a dip in the second and third year. The same can be seen in the abnormal earnings for those years.
The first series of tests conducted is a difference in means test presented in table 6. The first column of the table, control, is the mean of the backed sample minus the mean of the non VC-backed sample.
18
Table 4
Post-IPO Variables: Summary Statistics for Venture Capital-Backed IPOs, 2005-2012
Standard
Variable n Mean Std. Dev Minimum Maximum
Share price (€)
Offer price 88 12,97 9,6 1,87 54,4
First day closing price 88 16,41 14,76 2,05 91
After 1 week 88 16,38 14,82 2,04 91 After 1 month 88 16,44 14,76 2,14 91 Year 1 79 15,40 26,45 0,28 201,71 Year 2 75 11,53 15,15 0,4 98,07 Year 3 67 12,91 19,46 0,43 145,55 Year 4 63 14,27 29,67 0,35 222,55 Year 5 52 16,14 31,27 0,06 205,19 Year 6 22 23,15 28,45 1,02 109,72
Book value per share, t=0 (€ x 1000) 86 0,062 0,257 0,001 2,323
Abnormal earnings (€ x 1000) Year 1 77 73620 208882 -204927 1529682 Year 2 74 44232 142994 -483151 692732 Year 3 66 37996 159573 -831842 626150 Year 4 64 38501 153774 -404763 598541 Year 5 57 78614 173587 -150805 669319 Year 6 36 90100 244620 -722932 700725 Deal characteristics Deal size (€ x 1000) 88 268001 353668 8906 2106420 Flotation rate 65 38,61 18,95 6,22 100 Syndication partners 87 2,092 1,308 1 9
Avg. years of experience syndicate 79 31,12 26,83 4 147,5
Hot/cold market dummy-variable 88 0,738 0,442 0 1
Firm characteristics
Technology dummy-variable 88 0,432 0,498 0 1
US/EU dummy-variable 88 0,4205 0,496 0 1
19
Table 5
Post-IPO Variables: Summary Statistics for Non Venture Capital-Backed IPOs, 2005-2012
Standard
Variable n Mean Std. Dev Minimum Maximum
Share price (€)
Offer price 1005 10,64 10,31 1 165,03
First day closing price 1005 14,61 49,03 0,065 1415,35
After 1 week 1005 14,68 49,01 0,071 1451,35 After 1 month 1005 14,64 48,88 0,048 1415,35 Year 1 865 12,12 30,76 0,01 734,55 Year 2 749 9,11 12,9 0,01 143,15 Year 3 664 10,44 14,79 0,03 198,78 Year 4 600 11,37 18,85 0,01 262,95 Year 5 418 12,34 19,41 0,03 261,02 Year 6 144 17,34 26,33 0,09 225,45
Book value per share, t=0 (€ x 1000) 964 0,017 0,109 -0,932 0,002
Abnormal earnings (€ x 1000) Year 1 853 48439 403528 -1733804 9311708 Year 2 744 -2253 998603 -25257385 8085113 Year 3 660 37416 379504 -1985617 8036136 Year 4 599 43296 329738 -2147535 5040059 Year 5 417 56895 447168 -1419599 6458215 Year 6 144 92792 642574 -2507366 6579770 Deal characteristics Deal size (€ x 1000) 1007 188319 737379 1 14873586 Flotation rate 725 30,52 20,71 0,003 100
Hot/cold market dummy-variable 1007 0,609 0,488 0 1
Firm characteristics
Technology dummy-variable 1004 0,274 0,447 0 1
US/EU dummy-variable 1007 0,447 0,497 0 1
This table shows the number of firms per variable, n, for the control group with the mean, standard deviation, minimum and maximum. The data covers the firms that went public in the period 2005-2012. The share price in euros spans the period from IPO till 6 years after. The offer price is the price set by the underwriter and the syndicate that takes the firm public, the subsequent prices are market prices. The book value per share is the net asset value at the time the firm went public divided by the amount of shares it issued. Abnormal earnings are the EBIT of the firm minus the book value of the previous year times the expected return on equity based on the MSCI World Index. Flotation rate is the percentage of shares the firm offered publicly as percentage of equity. The hot-cold dummy is 1 when the IPO was in a hot market. The technology-dummy is 1 for a technology-firm. The US/EU dummy is 1 when the firm is from the US.
20
Table 6
Test of Differences in means
Venture Capital-Backed vs. Non VC-backed
Variable VC-control t-statistic p-value
Return on share price (%)
Underpricing -0,265852 -6,880*** 0,0000 Underpricing 6 days -0,267426 -7,119*** 0,0000 After 1 month -0,050321 -6,273*** 0,0000 Year 1 -0,010758 -0,133 0,8946 Year 2 -0,044539 -0,802 0,4252 Year 3 0,017956 0,186 0,8531 Year 4 0,02335 0,189 0,8504 Year 5 -0,007298 -0,045 0,9641 Year 6 0,38545 1,249 0,2252 Abnormal earnings (€ x 1000) Year 1 25180,96 1,058 0,2935 Year 2 46485,822 2,796*** 0,0066 Year 3 580,36 0,029 0,9765 Year 4 -4794,57 -0,249 0,8038 Year 5 21719,73 0,944 0,3489 Year 6 -2692,03 -0,066 0,9477 Deal characteristics Deal size (€ x 1000) 79682 2,114** 0,0374 Flotation rate 8,09 3,439*** 0,001
Hot/cold market dummy-variable 0,129 2,758*** 0,0071
Firm characteristics Technology dummy-variable 0,158 2,955*** 0,004 US/EU dummy-variable -0,0265 -0,499 0,6189
This table shows the differences in mean between the group of VC-backed firms and the non VC-backed firms. The column VC-control is the mean of the VC-backed sample minus the mean of the non VC-backed firm for the item in the row. Underpricing is the return of the first day of trading on the secondary market; for underpricing 6 days, the closing price after 6 days of trading is used to account for less liquid markets. For the years after the IPO the closing price of the first day is the denominator, hence the formula is (share price year x – closing price 1st day) / closing price 1st day. Abnormal earnings are
21 The parametric t-test of difference in means, with the null hypothesis of equal means between groups, shows a distinct difference in the level of underpricing. It shows that the initial return of the VC-backed sample is 26,58% lower than the non VC-backed sample. With a t-statistic of -6,880 this result is significant at the 1%-level. This means that the certification hypothesis for this sample holds as proposed by Megginson and Weiss (1991), however, their levels of underpricing for the sample period 1983-1987 was far lower with 7,1% vs. 8,2% (t=-7,82). More recent studies show more similar levels of underpricing, such as Bradley and Jordan (2002) with 30,37% vs. 16,62% (t=8,96).
The returns in percentages are calculated as the buy and hold-return from the closing price of the first day, thus using the first established market price (Hamao, Packer, & Ritter, 2000). On the short term there is a significant difference in performance, where the non VC-backed sample outperforms the VC-backed sample with 5,03%. One explanation of this short-term difference in performance between the two groups is an underpricing effect that takes more time to fade away. This mainly depends on the liquidity and the size of the market the IPO took place in. An underpricing of six days has been used in previous literature, for example Espenlaub et al. (1999) made use of it in their study of IPOs in the UK. The returns for the subsequent years, do not show any significant differences between the two samples. The means are reasonably close to each other with an exception of the sixth year, which most likely is caused by the small sample for the post-IPO year (36 venture-backed vs 144 non venture-venture-backed firms). The abnormal earnings show a similar picture for the years one till six, except for the second year which shows a large and significant difference between the two sub-samples. In the second year post-IPO, which for a lot of the companies in the sample is 2007/2008, abnormal earnings take a diving. This fall is far less severe for the VC-backed sample, since they remain in the positive where the non VC-backed sample on average takes a loss of € 2.25 million. It can be assumed that this fall in earnings is due to the financial crisis that started to affect the economy from 2007 onwards.
22 (p=0,0374). Secondly, venture capital-backed IPOs have a flotation rate that is on average 8,09% (p=0,001) higher than the control group. Remarkably, the hot/cold market-dummy shows that 73,87% of the sample against 60,8% of the control group is issued during a hot market. Lastly, the technology-dummy indicates that, with 15,8%, a significantly higher portion of the VC-backed sample compared to the control group is active in the high-technology industry. Therefore, we can conclude that VC-backed firms generally go for a larger IPO with a higher equity stake. This, and the fact that they seem to be better at timing the market larger deal sizes. The fact that VCs serve a higher number of technology firms can be explained from both the VCs and entrepreneurs perspective. VCs might have a preference for technology firms because of the relatively high returns that are made in this industry; however, it is also possible that technology firms are more actively seeking VC-funding than other industries.
4.1.2 Bivariate analysis
23
Table 7:
Correlation coefficients (n=509)
Variable log(SPT0) NAVPS t0 PVAE t1 PVAE t2 PVAE t3 VC Technology Hot/cold Flotation log(DealS) UP Return3Y
log(SPT0) 1,00 NAVPS t0 0,08 1,00 PVAE t1 0,07 0,22 1,00 PVAE t2 0,08 0,09 0,93 1,00 PVAE t3 0,07 -0,04 0,89 0,95 1,00 VC 0,10 0,09 0,13 0,11 0,10 1,00 Technology 0,11 0,09 0,03 0,01 -0,01 0,13 1,00 Hot/cold 0,08 0,02 0,00 -0,01 -0,02 0,03 0,06 1,00 Flotation 0,02 -0,03 -0,02 -0,05 -0,05 0,10 -0,07 0,11 1,00 log(DealSize) 0,18 0,13 0,11 0,10 0,10 0,17 0,13 0,08 0,41 1,00 Underpricing -0,15 -0,04 -0,04 -0,07 -0,01 0,03 -0,03 0,04 0,06 0,09 1,00 Return3Y 0,19 -0,02 0,03 0,04 0,04 -0,03 -0,06 -0,09 0,05 0,21 0,18 1,00
24 The control variable deal size shows sizable significant correlations with all the dependent variables and the VC-dummy. As it is posing a threat for multicollinearity it will be left out of the regression model. The other control variables for flotation (Flotation), the technology-dummy (Technology) and the hot or cold-market dummy show small and insignificant correlation coefficients with most variables.
Lastly, the variables underpricing (Underpricing) and the buy-and- hold return for the third year (Return3Y) show mainly insignificant and small correlation coefficients except for with each other and as mentioned before with the natural log of the deal size. The buy-and-hold return is calculated from the end of the first trading day till the annual stock price of the third year after the IPO. These two measures will be used in a fourth and fifth model to establish the comparability between the sample used in this study and previous literature. This can add to the validity of the novel methodology of using a basic valuation model in the VC-literature.
4.3 Multivariate analysis
The univariate analysis shows some significant differences between the VC-backed firms and the non VC-backed firms in the trend of the share price and return, but also in the control variables. The bivariate results support a difference in the two groups based on the significant positive correlation of the VC-dummy and other independent variables. In this section five different regression models will further explore the impact of VC-involvement.
25
For the second and third model the second and/or third year of the present value of abnormal earnings per share are left out of the equation. The reason for this is to check for the effects of multicollinearity due to the high correlations between the three variables. Next to that, it allows us to run the equation with a larger sample since the sample thins out as the years progress.
The other two models test for the initial return (underpricing) and the three-year return on the stock, the most used dependent variables in regression models for VC-research (Brau, Brown, & Osteryoung, 2004; Barnes, Cahill, & McCarthy, 2003; Barry C. B., 1994; Bessler & Kurth, 2007; Bradley & Jordan, 2002; Tykvova & Walz, 2007; Megginson & Weiss, 1991). In these models the future cash flows are left out of the equation as the dependent variable is not a valuation variable anymore. However, the variable book value per share will remain in the equation since it is a commonly used control variable in IPO-research (Brau et al., 2004).
26
Table 8
OLS Models for Venture and Nonventure-Backed IPOs 2005-2012
Dependent variable = 1. Share price (t=0) 2. Share price (t=0) 3. Share price (t=0) 4. Underpricing 5. 3-year Stock return
Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat)
Intercept 1,62*** 9,338 1,834*** 14,458 1,880*** 17,577 0,552*** 5,463 0,250** 2,193
Venture capital dummy variable 0,301* 1,659 0,333** 2,016 0,379** 2,312 -0,307* -1,851 0,185 1,583
Book Value per share, t=0 0,721* 1,925 0,644** 2,081 0,442 1,539 0,348 1,128 -0,078 -0,391
PV of Abnormal Earnings per share, t=1 -6,268 -1,394 -5,426 1,283 2,153 1,61
PV of Abnormal Earnings per share, t=2 2,396 0,388 7,353* 1,815
PV of Abnormal Earnings per share, t=3 6,115 1,026
Flotation 0,001 0,44 0,001 0,334 -0,001 -0,01 -0,001 -0,556 -0,001 -0,141 Technology-dummy 0,25** 2,251 0,211** 2,047 0,178* 1,798 -0,046 -0,446 0,039 0,539 Hot-or-cold-market dummy 0,248 1,488 0,057 0,485 0,039* 1,798 -0,039 -0,409 -0,513*** -4,647 R-squared 0,039 0,035 0,027 0,008 0,013 N 514 580 651 750 516
27 t=1,925) and the second model (0,644; t=2,081). In the third model it is nearly significant but much smaller with a coefficient of 0,442 (t=1,539). Lastly, from the control variables only the technology-dummy is consistently significant in adding to the model. The variable flotation returns very small coefficients and is insignificant. The variable for timing the market is only significant in the third model where only the abnormal earnings of the first year are incorporated.
The explanatory value of all the models is very low with the highest one being the first model in which the share price gets regressed over all the variables (R2=0.039). This model also has the smallest sample size (N=514). A low R-squared is not uncommon in VC-research (Brau, Brown, & Osteryoung, 2004; Megginson & Weiss, 1991), however, the basic valuation model should yield a higher explanatory value with the share price as the dependent variable. Similar regression models but with another ‘other information’ variable result in an R-squared of around 60-70% (Gregory and Whittaker, 2012) or 44-55% (Lev and Sougiannis, 1996). Although the basic valuation model did a moderate job in explaining the dependent variable, the impact of VC-involvement turned out to be significant in all three of the models. With a positive coefficient it shows that investors take VC-involvement into account when pricing a stock.
28 the only variable with a significant negative coefficient of -0,513 (t=-4,647), which infers that issuing in a hot market would imply a lower 3-year return. However, this result is deemed as not reliable since the crisis that started in 2007/2008 has a major influence on this result.
The notion that VC-involvement has an influence on share prices is not proven significant by most literature such as Brau et al. (2004). Their use of underpricing, three year sales growth and three year stock return as dependent variables show no significant difference for VC-involvement in a sample of small manufacturing firms. Their findings contradict those in the seminal paper on IPOs and the performance of VC-backed firms by Brav and Gompers (1997). They find a significant outperformance by VC-backed firms based on equal weighted returns for a five-year period post-IPO. Bessler and Seim (2012) state ‘the empirical findings provide convincing evidence that venture-backed IPOs generate positive returns for a specific time period subsequent to the IPO’. With a post-IPO period of three years they find that on average for European post-IPOs the returns outperform those of non VC-backed firms. Our regression findings when using the same dependent variables over a period of three years were insignificant, as was shown in the fifth regression model in table 8. Next to that, the picture that our univariate analysis sketches, although not significant, is one of average underperformance by VC-backed firms over a period of five years post-IPO (table 6).
A difference in the level of underpricing has always been the main argument to prove the certification role of VCs. Our results are in line with the theory that VCs provide a reduction in the level of information asymmetry. The difference in means test (table 6) shows statistically significant result for a lower level of underpricing. Therefore, the certification role of VCs is supported in this study based on univariate analysis and the regression model with underpricing as dependent variable.
29 empirical research done on the valuation effects of VC-involvement. The ways in which VCs would add non-financial value to the portfolio firm was gathered mostly by questionnaires on the perception of events by VC-managers and entrepreneurs. Although with a slim explanatory value (R=0,039), a connection can be made that goes further than performance measures in the sense of return metrics or underpricing, dependent variables that could theoretically never measure the added value of VCs from a shareholders perspective. The small explanatory power of the model is a weak point of this study; however, it is comparable to other studies on VC-involvement with the best example being Megginson and Weiss (1991) who theorized the certification role with a regression model that had an R-square of 0,018.
The novel methodology for VC- and IPO-research used in this study shows results that are in line with the theory. The significant VC-dummy in the regression model (table 8) shows that investors in fact take into account the non-financial value added by the VCs. Since the coefficient was in all three cases positive (0.301, t=1.659), hypothesis 1, that backed firms are more valuable than non VC-backed firms post IPO, can be accepted.
4.2 Europe vs. US-based portfolio firms
This section will focus on the second hypothesis, the differences between the US and European based portfolio firms and their performance post IPO. First the descriptive statistics and the difference in means test will be discussed, followed by the multivariate analysis.
4.2.1 Univariate analysis
30
Table 9
Test of Differences in means, VC-backed sample
EU vs. US firms
Variable n Mean US n Mean EU t-statistic p-value
Return on share price (%)
Underpricing 37 0,2572 51 0,0927 -5,5612*** 0,0000 Underpricing 6days 37 0,2394 51 0,0979 -4,7989*** 0,0000 After 1 month 37 -0,0034 51 0,0147 1,9492* 0,0569 Year 1 28 -0,1761 51 -0,0414 1,2697 0,2101 Year 2 27 -0,1697 48 -0,3768 -3,1079** 0,0032 Year 3 23 0,0699 44 -0,25 -3,1005** 0,0034 Year 4 20 0,059 43 -0,197 -1,8237* 0,0753 Year 5 17 0,1648 35 -0,2102 -2,405** 0,0218 Year 6 7 0,7547 15 0,3821 -1,1247 0,2796
Abnormal earnings per share
Year 1 26 7,792 51 13,656 0,5436 0,5891 Year 2 26 -0,72 48 12,28 1,3328 0,1890 Year 3 23 1,625 43 9,356 0,9905 0,3276 Deal characteristics Offer price (€) 37 13,22 51 12,63 0,3434 0,7327 Deal size (€ x 1000) 37 254242,3 51 277982,9 0,4882 0,6275 Flotation rate 28 31,519 37 43,97 4,6556*** 0,0000 Syndication partners 36 2,5833 51 1,7451 -6,2624*** 0,0000 Avg. years of experience syndicate 33 25,85 46 34,88 1,9099* 0,0625 Hot/cold market dummy-variable 37 0,5675 51 0,8627 6,0654*** 0,0000
Firm characteristics
Technology dummy-variable 37 0,4054 51 0,4509 0,6476 0,5202
This table shows the means of the n-number of US- and EU-based firms amongst the VC-backed firms that went public in the period 2005-2012. The t-statistic is for the difference in means test, accompanied with the p-value. Underpricing is the return of the first day of trading on the secondary market; for underpricing 6 days, the closing price after 6 days of trading is used to account for less liquid markets. For the years after the IPO the closing price of the first day is the denominator, hence the formula is (share price year x – closing price 1st day) / closing price 1st day. Abnormal earnings are the EBIT of the firm
minus the book value of the previous year times the expected return on equity based on the MSCI World Index. Flotation rate is the percentage of shares the firm offered publicly as percentage of equity. Syndication partners is the number of parties in the VC-syndicate, average years of experience is the average age of the VCs. The hot-cold dummy is 1 when the IPO was in a hot market. The technology dummy is 1 for a technology-firm.
31 The level of underpricing on the first trading day is 16,45% smaller for the sample of European IPOs compared to the US-based IPOs. This difference is significant with a t-statistic of -5,5612. For six days of trading, the level of underpricing between European and US IPOs come a little closer towards each other with a difference of 14,15% (t=-4,7989). The level of underpricing for the European IPOs come reasonably close to the mean of samples in other papers such as Megginson and Weiss (1991) with 7,1% and Jelic et al. (2005) with 7,79%. The subsequent performance of the both samples differs significantly as well for the years 2-5 as can be seen in chart 3.
Chart 3: The BHR from the 1
stmonth till the 6
thyear post-IPO
This chart shows the buy-and-hold return on the portfolio firms from the 1st trading day till the the 6th year post-IPO. The sample is divided in the US-based and EU-based firms.
Based on the history of VCs as outlined in the theory section, the average years of experience of the syndicate are surprising in a sense that one would expect US-based IPOs to have a syndicate of VCs backing them with on average more experience than the European ones. The average years of experience of a US-syndicate is 25,85 years opposed to 34,88 years for European syndicates (t=1,9099). For our sample this means that the proposition of Bessler and Seim (2012), that VCs with more information advantages, higher reputation and greater experience outperform others based on the level of underpricing holds true. However, their proposition was that the US-based VCs would be the more experienced VCs. The notion that experienced VCs attain lower levels of underpricing holds true
-0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1
1MNTH YR1 YR2 YR3 YR4 YR5 YR6
32 and is a reasonable explanation why the European based IPOs have a smaller magnitude of underpricing. A possible cause is a higher survivability rate due to a milder financial model in the European countries compared to the Anglo-Saxon model in the US. Although the average years of experience do not match the expectations, the number of VCs in a syndicate does. The US-based IPOs on average have 2,5833 VCs involved whereas European IPOs have 1,7451 VCs involved (t=-6,2624). A larger syndicate suggests more non-financial value added in terms of the roles described in table 1.
33 buy-and-hold returns, this is most likely caused by the small sample size of 7 US-based and 15 EU-based portfolio firms for that year.
All in all, the univariate analysis shows very mixed results with lower levels of underpricing for European IPOs but better long-term performance of US-based IPOs. The deal characteristics are surprising in a sense that the European VCs seemed to be having better averages based on market timing and average years of experience. Based on these results it can be concluded that the age and experience of VCs have less impact on the level of underpricing at the IPO than is suggested in the literature. It is likely that the institutional environment and the country’s financial system have a larger impact. Supportive evidence for this notion comes from Engelen and van Essen (2010), who show that 10% of the variation in the level of underpricing is explained by the quality of a country’s legal framework. In the next section the multivariate analysis will be discussed.
4.2.2 Multivariate analysis
In table 10 the results of the OLS regression analysis can be found for the VC-backed sample. The first three models are the basic valuation model that is also used in the previous multivariate analysis for the VC-backed vs. non VC-backed analysis. The fourth model uses underpricing as a dependent variable and the fifth model uses the 3-year stock return as the dependent variable. The EU/US-dummy variable is 1 when it concerns a US-based IPO.
34
Table 10: EU/US-influence on Venture-backed IPO performance
OLS Models for Venture Backed IPOs 2005-2012
Dependent variable = Share price (t=0) Share price (t=0) Share price (t=0) Underpricing 3-year Stock return
Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat)
Intercept 2,682*** 3,671 2,790*** 5,448 2,828*** -0,796 0,379** 2,259 0,5 0,895 EU/US-dummy variable -0,126 -0,373 -0,212 -0,737 -0,217 -0,796 0,073 0,674 0,254 0,973
Book Value per share, t=0 2,751 0,822 0,314 0,138 0,553 0,257 -0,05 -0,299 0,057 0,168
PV of Abnormal Earnings per share, t=1 -5,769 -0,337 5,601 0,382 -0,386 -0,433
PV of Abnormal Earnings per share, t=2 -17,915 -1,325 -6,044 -0,574
PV of Abnormal Earnings per share, t=3 18,187 1,603
Flotation -0,01 -1,017 -0,004 -0,496 -0,002 -0,246 -0,003 -1,081 0,016** 2,405 Technology-dummy -0,151 -0,508 -0,089 -0,326 -0,1 -0,372 -0,157 -1,565 0,437* 1,823 Hot-or-cold-market dummy 0,143 0,232 -0,18 -0,533 -0,308 -0,999 -0,072 -0,635 -1,590*** -3,167 R-squared 0,129 0,053 0,064 0,096 0,282 N 45 52 54 63 47
35 The fifth model is the model with the largest explanatory power (R2=0,282) and has the only significant variables of all five models with the EU/US-dummy variable in this sample. However, these variables are the flotation rate, the technology-dummy and the hot- or cold-market dummy. The EU/US-dummy variable for this model yields a positive coefficient which is in line with the results depicted in chart 3.
This multivariate analysis has shown very little significant results in the difference between European- and US-based IPOs. The only differences stem from the univariate analysis (table 9), and can be seen in the magnitude of underpricing and the syndication partners and their experience. However, this did not result in a higher valuation of the firms post-IPO as shown in table 10. To gain additional insight, we will focus on the value of the characteristics of the syndicate in explaining the valuation of the portfolio firms by investors post-IPO, the level of underpricing and the three-year stock return in the next section.
4.3 Syndication and venture-backed IPO performance
36
Table 11: Syndication effects on Venture-backed IPO performance
OLS Models for Venture Backed IPOs 2005-2012
Dependent variable = Share price (t=0) Share price (t=0) Share price (t=0) Underpricing 3-year Stock return Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat) Coefficient (t-stat)
Intercept 3,568*** 4,128 3,021*** 4,971 3,163*** 5,929 0,351* 1,812 -0,027 -0,039
Syndication experience-dummy -0,866** -2,649 -0,644** -2,231 -0,689** -2,61 0,215** 2,023 0,584** 2,398
Syndication partners -0,129 -0,872 -0,059 -0,419 -0,084 -0,634 0,033 0,843 0,189 1,632
Book Value per share, t=0 2,567 0,563 -0,821 -0,286 -1,062 -0,446 -0,089 -0,521 -0,012 -0,038
PV of Abnormal Earnings per share, t=1 -15,808 -0,64 6,183 0,315 6,868 0,695
PV of Abnormal Earnings per share, t=2 -9,349 -0,657 -0,414 -0,032
PV of Abnormal Earnings per share, t=3 25,171** 2,215
Flotation -0,006 -0,601 -0,001 -0,113 -0,001 -0,979 -0,004 -1,464 0.012* 1,873 Technology-dummy -0,044 -0,143 0,039 0,132 0,048 0,166 -0,211* -1,896 0,261 1,054 Hot-or-cold-market dummy -0,32 -0,454 -0,233 -0,607 -0,327 -0,979 -0,099 -0,812 -1.363** -2,387 R-squared 0,288 0,155 0,186 0,175 0,361 N 41 47 49 56 43
This table shows the regression results of five different regression models. The first three models regress the share price after going public with the book value per share and the present value of future abnormal earnings. The first model uses the present value of abnormal earnings of the three years after going public; the second model only the two subsequent years and the third model only one year. The syndication experience-dummy variable is 1 when the syndicate of VC firms that backs the portfolio is on average older than 25 years and zero otherwise. The variable syndication partners is the amount of VCs in the syndicate. The control variables are flotation, the percentage of equity that gets offered publicly, the technology-dummy is 1 when it concerns a technology-firm and a market-timing dummy variable which is 1 when the IPO happened when the market was ‘hot’.
37 exchange, a country where VCs in general are even younger than in Europe. In our model the control variables are the flotation-variable, technology-dummy and hot-or-cold dummy.
All five models show a significant coefficient for the syndication experience dummy-variable. The coefficient ranges from -0.644 to -0.866 in the first three models, all significant on the 5%-level. Surprisingly, this means that for the experienced syndicates the share price is lower with €0,644 to €0,866 on the log scale. Next to that, the fourth model shows with a significant positive coefficient that for the experienced syndicates the level of underpricing is higher (0,215; t=2,023). Lastly, the 3-year return model indicates that the more experienced VC-syndicates have a higher long-term return (0,584; t=2,398).
38 Our results of model 4 are not in line with the results of Barry (1990) when it comes to the number of VCs or with Gompers (1996) or Wang et al. (2003) when it comes to the age of the VCs. An argument for grandstanding of younger VCs does not hold, nor does the argument that the certification role for older VCs should be stronger. When comparing our study to the aforementioned studies by Gompers (1996) and Wang et al. (2003), the main difference that can be found is the cut-off point for the dummy-variable. Although in all three cases the median is used, the median of our sample is with 25 years significantly higher. A different view on the matter is that currently, opposed as to a decade ago, investors care less about the age or the number of VCs that partake in the IPO. It is less unlikely that the change is caused by a new generation of VCs since the average age of VCs is still high and only 10% of the sample of syndicates has an average of below 12.5 years. A sensitivity analysis (unreported) in which a cut-off point of 12,5 years is used, shows insignificant but smaller coefficients. This means that investors have decreasing interest in the characteristics of the syndicate. Perhaps they feel safer and their appetite of risk has grown due to improvements in the legal framework and governance codes.
5. Conclusions
In the VC-literature, a lot of research focuses on the difference between VC-backed firms and non VC-backed firms using the initial return and other return metrics. However, the results of this research stream are still mixed. Moreover, it is often the performance of the portfolio firms that gets measured when the researchers state that they want to research the value that is added by VCs. The objective of this research is to investigate the influence of VC-involvement in the valuation of firms when they go public. Therefore, a basic valuation model will be used in this research instead of return metrics. Next to that, the differences between European and US-based IPOs of VC-backed firms are investigated. Our dataset consists of 88 VC-backed firms and 1.007 non VC-backed firms that had an IPO in the period 2005-2012. The VC-backed sample consists of 37 US-based IPOs and 51 European IPOs.
39 network that the portfolio firm can utilize but also gives the portfolio firm a certain level of legitimacy. The role of legitimation is consistent with the certification role of VCs during IPOs. They are seen as an agent that reduces information asymmetry during the event of going public. This should result in a lower level of initial return on the first day of trading on the secondary market. Our first hypothesis states that the involvement of VCs raises the value of the portfolio firm.
The VC-industry originated in the US and spread to Europe in the late 1970s. Due to the small size of the industry in an early stage VCs had the tendency to form syndicates to finance firms. In this way risk could be mitigated through diversification and at the same time the portfolio firm received a second set of screenings to make sure that it was a good investment. Based on the history of the VC-industry, our second hypothesis states that US-based VCs add more value to the portfolio firm because of their larger experience.
The first set of analysis shows that VC-backed firms are valued higher than non VC-backed firms after they go public. A difference in means test shows a significant lower level of underpricing for VC-backed firms of 26,58% compared to non VC-backed firms. This metric is the most common metric used in the literature and our results are similar to recent studies using it. Based on the lower level of underpricing it can be concluded that the certification role holds. The multivariate analysis consists of five models, of which three are using the basic valuation model with the log of the share price as the dependent variable. The other two models use underpricing and three-year buy-and-hold return as the dependent variable. The regression models return significant coefficients for the venture capital-dummy for the models with the log of the share price and underpricing as dependent variable. The basic valuation model shows that VC-backed firms are valued higher than non-VC backed firms. Next to that, they have a lower level of underpricing. Therefore, it can be concluded that VCs add non-financial value to a firm and that our first hypothesis can be accepted.
40 syndication partners is 9 years higher for the European VCs. The multivariate analysis with the basic valuation model and the EU/US-dummy variable does not return any significant results. Therefore, to gain additional insight another regression model is presented in which the experience of the VC syndicates is a dummy variable. This analysis shows us that more experienced VCs actually have a negative impact on the valuation of the portfolio firm and also results in a higher level of underpricing. Based on these results we can conclude that even though by using the more experienced VCs, which have a negative effect on the valuation and increase the level of underpricing, European VCs are doing a better job when it comes to the level of underpricing. What drives these results cannot be concluded with this dataset, but we propose that it is caused by a more stable institutional environment and less volatility. Nevertheless, the hypothesis that US-based VCs add more value to the portfolio form is rejected.
6. Limitations and future research
Financial analysis on share prices generally is done over very large sample sizes. Given the nature of the field of research of this paper, public data is very scarce. This paper is bound to a time span of eight years because of data restrictions, next to that the field of VC-research is relatively young, as is the phenomena itself. Less data points make it more difficult to find accurate and consistent prove for a hypothesis, this might also be the reason why the research in the past 15 years cannot be uniform over the phenomena of underpricing in respect to VCs. Next to that, in the light of the information age and stock market bubbles trends in the data itself must have changed radically as well. A current stream of research in IPOs with regard to VC-involvement pays attention to the impact of the legal environment on the level of underpricing (Hopp & Deher, 2007; Engelen & Essen, 2010). Including control variables that show significant in this stream of literature might enlarge the explanatory power of the general underpricing or valuation models. In our case, it might clarify the surprising results of the second hypothesis.
41 model uses past actual results to explain a variable that is theoretically based upon predictive values. Assuming that actual data can be used for look-ahead variables is not irregular in financial economic research, as pointed out by Lee & Wahal (2004).
