The role of overcommitted capital in the valuation of venture capital funds’ investments

(1)

The role of overcommitted capital in the valuation of

venture capital funds’ investments

MSc. Finance | Master Thesis

Emma den Held

Supervisor: Prof. Lopez-de-Silanes

July 2016

Abstract:

This paper studies overcommitted capital in venture capital funds and the valuation of 1,122 venture capital transactions. The results support the “money chasing deals” phenomenon that inflows of money into venture capital funds lead to higher valuation of these funds’ investments (Gompers & Lerner, 2000). Overcommitted capital is defined as the ratio of fund size and target size and is positively related to the valuation of venture capital investments. If average overcommitted capital of the funds investing in a transaction increases by 10%, the valuation of the investment increases by 9.1%. Furthermore, we show that if the investing funds have relatively more capital than other funds in the market, this has a positive effect on valuation. Our conclusions are robust to (i) adding several control variables and (ii) using the average rank of funds, based on fund size, as an instrumental variable for overcommitted capital. The results are consistent with the theory that too much money chasing a limited number of deals leads to higher prices in venture capital.

(2)

Statement of Originality

This document is written by Emma den Held who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its

references have been used in creating it.

The Faculty of Economics and Business of the University of Amsterdam is responsible for the supervision of completion of the work, not for the contents.

(3)

1. Introduction

With $2.4 trillion assets under management as of June 20151, private equity can be considered an important asset class. It seems that private equity will continue to be: 689 funds closed in 2015 and aggregate capital raised was $288 billion. Private equity refers to the asset class that includes buyout funds and venture capital funds, as well as other closely related strategies2_{. They invest into private}

companies or public companies that become private, using different investment plans. These investments need to have an attractive entry price in order to deliver superior returns to the funds’ investors. This thesis examines the role of capital in determining the valuation of private equity investments, particularly focused on venture capital. It investigates if there is a relationship between overcommitted capital, defined as the ratio of a fund’s actual size and its target size, and the valuations of venture capital investments.

The existing literature focuses on private equity performance (Harris, Jenkinson, & Kaplan, 2014; Kaplan & Schoar, 2005; Phalippou & Gottschalg, 2009; Robinson & Sensoy, 2011). The historical performance of private equity funds is uncertain, which is mainly due to the uneven disclosure of funds’ returns and questions regarding the quality of the data (Harris et al., 2014). In addition, research in private equity is directed towards the determinants of private equity performance. One of the determinants is committed capital, the money that is committed to a fund and subsequently used to make investments. It can be investigated in terms of aggregate capital flowing into the industry or capital flowing into a particular fund, which determines the fund’s size. How does it relate to performance? Do funds with more capital at their availability perform better or worse? In a frequently cited paper from Kaplan and Schoar (2005), a positive and concave relationship is found between fund size and performance of venture capital funds. Large funds perform better, but performance declines when funds become very large. Inflows of capital are also larger for funds whose past performance is better. But from the funds with the same increase in performance, top performing funds grow less than proportionally compared to lower performing funds. Since investors claim that the top funds are all oversubscribed, it seems that these funds choose to stay smaller (Kaplan & Schoar, 2005). Assuming the number of good investments in the economy is limited, funds need to make a trade-off between growing and persistence of performance.

In a more recent paper, Harris et al. (2014) regress performance on an estimate of the aggregate amount of capital flowing into private equity. They find that the inflow of capital is negatively related to performance. Hence when more capital is committed to private equity funds, performance declines. To understand what drives the decrease in performance when money flows into

1_{According to Preqin’s Global Private Equity & Venture Capital Report 2016.}

2_{Preqin makes a distinction between private equity and private capital. Private equity refers to the buyout and}

venture capital industry and other closely related strategies. Private capital is the broader term for private equity, private debt, private real estate, infrastructure and natural resources. For more information refer to Table A1.

(5)

private equity, other research examines the valuations of private equity funds’ investments. Could an increase in committed capital drive up valuations and thereby reduce performance? Gompers and Lerner (2000) present empirical evidence that the valuations of venture capital investments increase due to higher inflows of capital into venture capital funds. When funds have more capital available, they seem to pay a higher price for their portfolio companies. Another possibility is that higher valuations and increased committed capital reflect better-expected market circumstances. In that case, they would both increase even if there is no causal relationship. Gompers and Lerner (2000) address this issue and conclude that it does not appear that the positive relation between capital and prices is due to better investment prospects. The finding that more capital chasing a limited number of favorable investments increases the valuations of these investments is often called the “money chasing deals” phenomenon. It essentially represents a mismatch between the supply of and demand for capital in the private equity industry. The paper’s objective is to present additional evidence that an inflow of capital increases the valuation of investments made by venture capital funds. It is interested in the relationship between overcommitted capital, capital that funds receive in excess of their fundraising target, and the valuations of these funds’ investments. Overall, this thesis aims to shed light on the role of committed capital in determining the valuations of private equity investments.

We examine the relation between overcommitted capital and the valuations of venture capital investments based on a data set containing over 1000 venture financings between 2005 and 2015. Using a hedonic approach, valuation is regressed on characteristics of the company and environment, and average overcommitted capital of the investing funds. The results show a positive and statistically significant relationship between overcommitted capital and the valuation of venture capital investments. A 10% increase in average overcommitted capital of the investing funds increases the valuation of their investment with 9.1%. We also measure relative over-commitment, which compares overcommitted capital of the funds investing in a particular company with overcommitted capital of other funds in the market. Relative over-commitment is again positively related to valuation; if the funds investing have 10% more capital relative to funds in the market, this leads to 8.4% higher valuation of their investment.

To assess the robustness of the results, we add a variety of control variables to the original regression model. These variables address alternative hypotheses, relating to the public market valuation and differences between first and later round investors. Overcommitted capital continues to have a positive effect on valuation after adding control variables. In addition, we employ an instrumental variable approach to address omitted variable bias. The instrument used is the average rank of funds and should be related to overcommitted capital but unrelated to the success of venture capital investments. The estimates support the view that overcommitted capital leads to greater competition and higher valuations. The results extend the findings of Gompers and Lerner (2000) that an inflow of capital increases the valuations of venture capital investments.

(6)

The paper is structured as follows. Section 2 starts with an introduction into private equity and venture capital. We explain important concepts in private equity, how funds operate and the venture capital cycle. Relevant papers regarding private equity performance, committed capital and the valuation of investments are discussed. Section 3 develops the hypotheses that are tested in the remaining part of the paper and explains the methodology. Section 4 describes the data set. The empirical results are presented in section 5, which consists of three parts. First, the basic econometric analysis seeks to document a relationship between overcommitted capital in venture capital funds and these funds’ new investments. Second, robustness checks are performed. Third, the relationship is explored using an instrumental variable. Section 6 discusses the findings and some limitations. The final section summarizes the main results and provides recommendations for further research.

(7)

2. Literature review

This section starts with a brief introduction of private equity and venture capital. Important concepts that are required to understand the empirical design presented in section 3.2. are explained. Next, we discuss performance, capital flows and the valuation of private equity investments. Finally, this section covers the relevant papers to identify the gap in the literature this paper aims to fill in.

2.1. Private equity & venture capital

2.1.1. Private equity

The early history of private equity (PE) started in the United States in 1946 and developed through a series of boom and bust cycles to modern private equity. The asset class increased heavily in terms of assets under management and capital flowing into the industry. From 1980 to 2007 yearly commitment to U.S. private equity funds increased exponentially: from $0.2 billion in 1980 to more than $200 billion in 2007 (Kaplan & Strömberg, 2009). Private equity firms raise capital through one or more private equity funds. These funds are organized as limited partnerships with a fixed life, which is usually 10 years but often extended for another 3-4 years. Most funds are “closed-end” meaning investors commit a certain amount of capital that the fund can use to make investments and pay management fees. They cannot withdraw their money until the fund is terminated. Because the fund does not have to return capital to investors in an uncertain time frame, it makes it easier to invest in illiquid assets (Gompers & Lerner, 2000). In a fund, private equity firms act as general partners (GPs) and are responsible for managing the fund and its investments. They receive a management fee as well as a percentage of the fund’s profit, called carried interest. Institutional investors generally act as limited partners (LPs) and commit capital to PE funds. Funds often require a minimum amount of capital for investors. The total amount of committed capital determines the fund size. It is usually not invested immediately, but drawn down when investments are identified. Usually a private equity fund invests its committed capital in the first five years of the fund’s life. It then has another 5-8 years to return the capital to its investors (Kaplan & Strömberg, 2009). There are different types of funds that follow different strategies. For instance, venture capital (VC) invest funds in early-stage companies with high growth potential and buyout funds invest in companies from which they can extract value by holding and managing the company for some time. Rapid development and changes in the private equity industry have caused the diversity of fund strategies to grow. Investors target new areas such as private debt, private real estate, infrastructure and natural resources. We use Preqin’s terminology, which refers to private equity as “the core asset class centered on the buyout venture capital industry,

(8)

2016). Table A1 in the appendix lists the different fund types that constitute the category of private equity. The literature review will mainly focus on venture capital as the empirical work is designed around this type of funds.

2.1.2. Venture capital

The first venture capital firm was American Research and Development Corporation (ARDC). It was founded in 1946 by Harvard Business School professor Georges Doriot, MIT president Karl Compton and several others. Over the next 26 years the firm made many high-risk investments in emerging companies. One major success was the investment $70,000 in Digital Equipment Company, a pioneer in the computer industry, in 1957. Digital Equipment Company grew in value to $355 million (Gompers & Lerner, 2001). In the decade after the foundation of ARDC, the formation of venture capital limited partnerships began. Partnerships had a predetermined lifetime and needed to return the assets to investors within the specified time period. As a result, partnerships would give investors their allocation of shares in the venture capital fund’s portfolio companies instead of returning cash after a company successfully went public. In the late 1970s and early 1980s, venture capital experienced real growth when institutional investors started allocating money into the private equity industry. The number of venture capital funds multiplied. The increased competition, market conditions and other alternatives for venture capital such as foreign corporations, led to declining returns in the venture capital market. Still, venture capital funds backed many high-technology companies including Apple, Genentech and Microsoft in the 1980s and 1990s (Gompers & Lerner, 2001). A large boom began in 1995, with many initial public offerings of (computer) technology and growth companies, and busted with the Internet bubble in 2000. According to Preqin, the current venture capital market is in a boom period again. In 2015, capital committed to venture capital funds was $47 billion and included Insight Venture Partners IX, a $3.4 billion fund and the largest in history. Many funds are looking for capital, increasing the competition and making it harder for funds to identify investments at an attractive price.

Venture capital funds spend a large amount of time and resources selecting deals. The attractiveness of a company depends on several aspects such as the management team, business plan, market size and competition. VC fueled some of the most successful start-ups: Microsoft, Google, Apple and Cisco were all funded by venture capital. There is an ongoing debate on what is relatively more important for a young company to succeed: its business plan or a strong management team. To address this debate, Kaplan, Sensoy and Strömberg (2009) study the evolution of 50 VC-financed companies from early businesses to public companies. The results indicate that the initial business plan or line of business rarely changes significantly for firms that go public. An initially strong business appears to be essential for a company to succeed, however, they are able to replace their

(9)

founders and initial managers. This suggests that venture capitalists should place more weight on the business than on the management team (Kaplan et al., 2009).

The venture capital cycle explains all steps in the process of venture capital financing. To establish a fund, it needs to raise capital from investors. This capital is used to make investments. Early stage, high technology companies often face the problem of information asymmetries (Gompers & Lerner, 2001). This makes it difficult to receive financing from outside investors. Venture capital organizations, however, are specialized financial intermediaries who carefully investigate the companies that they consider investing in. This decreases the information asymmetry and allows companies to receive the funding that they could not get from outside investors. Companies can receive venture financing in multiple stages, such as the seed or early stage. They offer equity in different rounds. After funds make an investment, they start monitoring the company. Venture capitalists often take seats on the company’s board of directors, recruit new managers and assist in outside relationships. The next step for funds is to exit successful deals and use the proceeds to return capital to their investors. A commonly known exit strategy for venture capital funds is through an initial public offering (IPO), where the company issues shares to the public. The venture capital firm usually does not sell its equity stake immediately, because it would send a negative signal to the market. Banks often require a lock-up period, in which insiders (including the venture capitalists) are not allowed to sell their stakes. In a study what determines venture capital funding, Jeng and Wells (2000) find that IPOs are the main driver of venture capital. If there are many IPOs in a certain period or country, this will stimulate venture capital. Another possible exit strategy is through an acquisition. The acquirer can, for instance, be a larger player in the same industry. Finally the venture capital cycle renews itself when venture capital firms raise additional funds.

2.2. Performance, capital flows and valuation

2.2.1. Private equity performance

Performance has been the main subject in private equity research. The most cited papers regarding private equity performance are presented in this subsection. Performance can be measured in various ways. Historically, most practitioners focused on a fund’s internal rate of return (IRR) and investment multiple (Harris et al., 2014). The IRR calculates the LPs return for which all investments are realized and the money is returned to the investors. It is based on the contributions and distributions, net of fees and carried interest. For investments that are not yet realized, the IRR includes the estimated value. The investment multiple, also known as the total value to paid-in (TVPI), measures the total value of distributions and unrealized investments as a multiple of the total value of all contributions. One of the key questions in private equity is how their returns relate to

(10)

public equity returns. Kaplan and Schoar (2005) introduced the public market equivalent (PME). This measure is used to investigate the performance of private equity funds and compares an investment in a private equity fund to an investment in the S&P 500. For instance, a PME of 1.20 implies that investors ended up with 20% more than if they would have invested in the public market. Kaplan and Schoar (2005) find that private equity fund investors, on average, earn less than S&P 500 index (net of fees). They do not find the outperformance that is often given to justify investment in private equity funds. Of the funds in their sample, 78% are venture capital funds and 22% are leveraged buyout funds. Phalippou and Gottschalg (2009) estimate the performance of private equity funds both net-of-fees and gross-of-net-of-fees. They find that average net-of-net-of-fees performance is below that of the S&P 500 (by 3% per year), but gross-of-fees performance is above that of the S&P 500 (by 3% per year). The paper by Harris et al. (2014) studies the performance of 1400 U.S. buyout and venture capital funds using a new data set from Burgiss. The results are more positive than has previously been documented. Buyout fund performance has consistently exceeded that of public markets. Venture capital funds performed better than public equities in the 1990s, but less than public equities in the 2000s. In addition, Robinson and Sensoy (2011) document outperformance of approximately 15% over the fund’s life compared to the S&P 500 (net-of-fees), or about 1.5% per year.

Furthermore there is empirical evidence of persistence in performance. Kaplan and Schoar (2005) find strong persistence in performance across different funds of the same general partner. A fund that outperforms the industry is likely to outperform the market with subsequent funds. Besides, funds that have persistently performed have an advantage during fundraising. Hochberg, Ljungqvist and Vissing-Jørgensen (2014) replicate Kaplan and Schoar (2005) with their persistence test. Performance of venture capital funds increases with prior-fund performance (p < 0.001). Unfortunately, returns do not seem scalable: Lopez-de-Silanes, Phalippou and Gottschalg (2015) find evidence that there are diseconomies of scale for the number of simultaneous investments. Their results show that investments underperform when there are many simultaneous investments. To conclude, historical performance of private equity funds remains uncertain. Yet we know that there is persistence in performance and that returns are not scalable.

2.2.2. Committed capital

Figure 1 shows annual commitment to U.S. venture capital funds around the dot-com bubble. Through the bubble, annual commitments to U.S. venture capital increased rapidly until the collapse in 2000. As a result, there was a sharp decline in annual commitments. In the years that follow annual commitments remain far from the peak it reached in 2000. This pattern continues after 2005; fundraising fluctuates from year to year but moves nowhere near the amount in 2000. Kaplan, Lerner and Problem (2010) note that annual commitment does not take into account the fact that the economy and stock market have increased considerably since 1980. They show that venture capital

(11)

Figure 1 – Total annual commitment to U.S. venture capital funds

Source: Thomson One (2016)

commitments are much more stable when scaling annual commitment by the total value of the U.S. stock market at the beginning of each year. Annual commitment never represents less than 0.05% of the stock market or exceeds 0.23%, except in 1999-2001. The remaining part of this section concentrates on relevant papers regarding committed capital in private equity.

Harris et al. (2014) regress fund performance on an estimate of capital flows into the private equity industry. Performance is measured by the PME, the IRR, and the investment multiple. The results show that performance is negatively related to capital commitments for buyout funds as well as venture capital funds. When capital flows move from the bottom to the top quartile, performance measures decline: PMEs by 0.33, IRRs by 9% and investment multiples by 0.75 (Harris et al., 2014). These results suggest that an inflow of capital into the venture capital industry is associated with lower subsequent performance. This is broadly consistent with previous findings. Kaplan and Strömberg (2009) regress the “vintage year return”, the capital-weighted return to all private equity funds from a particular vintage year, on committed capital. Vintage year is the year a fund had its first takedown for investment purposes. Their results indicate a negative relation between an inflow of capital and vintage year returns. The capital committed to private equity funds in a particular vintage year can thus explain these funds’ returns in the subsequent period in which they are active. Kaplan and Strömberg (2009) also consider how past returns influence committed capital. They find that committed capital is positively related to past returns. In conclusion, the results are consistent with a boom and bust cycle: when more capital is committed to private equity funds, returns decrease. And with decreasing returns, committed capital to the private equity industry also tends to decline. Robinson and Sensoy (2011) also find a negative and statistically significant relation between capital flows and performance, as measured by the IRR. Funds that started in boom years have on average lower performance. But when they measure performance with PMEs, comparing private equity funds

(12)

to the public market, they do not find a relation. Funds that were raised during periods of high inflows of capital may have lower performance, but the same applies to the performance of the public market. A study of 200 mature European private equity funds further shows that the private equity industry is segmented within itself. For a given absolute fund inflow, an increase in the allocation of money towards a particular fund type decreases the performance of that fund type (Diller & Kaserer, 2009). Finally, the greater the amount of capital committed to private equity, the longer it takes for a fund to return a given multiple of committed capital to LPs. Because more capital leads to greater competition from other private equity funds, GPs invest their capital more slowly (Ljungqvist, Richardson, & Wolfenzon, 2009).

Besides, Harris et al. (2014) examine whether fund size affects performance. Fund size is the total amount of capital committed to a fund by its limited partners and general partners. They find no relation between fund size and performance for buyout funds but a strong positive relationship for venture capital funds3. This is in line with the results of Kaplan and Schoar (2005), who document a positive and concave relationship between fund size and performance but statistically significant for venture capital funds only. So larger funds have on average higher PMEs, but when they become very large, returns decline. Kaplan and Schoar (2005) not only investigate whether fund size affects performance, but also whether past performance affects fund size. In regressions with fund size as the dependent variable, they document a positive and statistically significant relation between past performance and fund size. They find that capital committed to the next fund of a partnership is larger when past performance is higher. The relation is also concave, meaning that the best performing funds grow less than proportionally when performance increases than do lower performing funds (Kaplan and Schoar, 2005). According to most limited partners, the best performing funds are usually oversubscribed. It appears that these funds choose to stay smaller. A summary of the literature regarding committed capital is provided in Table A2 in the appendix.

2.2.3. Valuation of investments

There are several ways to determine the value of an investment. Commonly used methods are valuation using multiples and discounted cash flow (DCF) valuation. The first method uses multiples to estimate the value of a company based on the multiples of other comparable companies. A well-known multiple is EV/EBITDA, which measures earnings before interest, taxes, depreciation and amortization (EBITDA) as a multiple of the enterprise value (EV). It is does not depend on a firm’s capital structure and therefore allows companies with different levels of debt to be compared. The DCF method determines the value of a company by discounting its future free cash flows. According to Fama (1970) a market is efficient if prices always fully reflect the available information. Positive

(13)

information about the future should increase future expected cash flows and negative information should decrease expected cash flows. Hence if investors learn that a firm’s future probability will be higher, the value of the firm should also increase. On the other hand, if the inflow of money into venture capital funds is exogenous and thus unrelated to future earnings of venture capital investments, it should be unrelated to the valuations of these investments. Venture capital funds often use the “pre-money valuation” to determine how much equity they should receive for their investment. The pre-money valuation refers to the valuation of a company prior to a financing round. It is calculated as the product of the price paid per share in the financing round and the outstanding shares prior to the financing round. Private companies usually receive multiple financing rounds with corresponding pre-money valuations.

Gompers and Lerner (2000) study 4069 venture financings and find a positive relation between the valuation of new investments and inflows of capital, which suggests that “increases in

the supply of venture capital may result in greater competition to finance companies and rising valuations” (p. 321-322). If more capital is available, venture capital funds seem to pay a higher price

for new investments. They also explore the cause of this relation, whether it is driven by better prospects for young firms or by demand pressures. To see if the first explanation holds, Gompers and Lerner (2000) look at the ultimate success of venture-backed firms and show that success rates do not differ significantly between investments made in years with high inflows of capital and other years. Their evidence is more consistent with the second explanation that demand pressures drive the positive relation between inflows of capital and valuations. Because the number of favorable investments in the industry is limited and needs to be matched with fluctuating annual commitments to venture capital funds, high inflows of capital increase the valuation of these investments and lower fund returns. This is called the “money chasing deals” hypothesis (Gompers & Lerner, 2000). A model is developed in which venture capitalists’ equity share depends on capital market characteristics. One of the empirical implications of this model is that the equity share is negatively related to capital market competition and supply (Inderst & Müller, 2004). So when competition and supply increase, rising valuations may trigger smaller equity shares of venture capitalists.

(14)

3. Hypotheses and empirical design

Section 3.1. develops the hypotheses that will be used in the empirical analysis to examine the role of overcommitted capital in the valuation of venture capital investments. Two hypotheses are formed to test if overcommitted capital increases the valuations of venture capital investments. Section 3.2. presents in detail the empirical design of this thesis, including the sample selection, a description of the main variables as well as the creation of new variables.

3.1. Hypotheses

According to finance theory, the valuation of a firm depends on the discounted value of its future cash flows. Capital flows into private equity may affect the valuation of private companies, depending on whether the inflow of capital is exogenous. If the inflow of capital is exogenous, i.e. not related to future expected returns, it should not affect the valuations of private companies because there always exist substitutes. But if the inflow of capital is not exogenous, better-expected conditions for private companies could lead to an increase in the inflow of capital and in the valuations of these companies simultaneously. In this scenario, both variables increase even if there is no causal relationship. Gompers and Lerner (2000) address this issue by looking at the ultimate success of companies funded by venture capital. They conclude that it does not appear that the positive relation between committed capital and prices is attributable to better investment prospects. There is also an alternative view, in which the inflow of capital is exogenous, but venture capital funds are restricted to invest in private companies. Then an increase in committed capital results in greater competition among private equity funds and will drive up the valuations of private companies.

The first hypothesis is based on overcommitted capital in VC funds. If a fund’s committed capital exceeds the fundraising target, the fund has more capital available than it had anticipated. The fund’s goal is to invest this capital at the best possible return and will therefore compete for a limited number of good investments. This “money chasing deals” could drive up the valuations of the fund’s investments. Hence the first hypothesis can be formulated as follows:

Hypothesis 1: Overcommitted capital in venture capital funds increases the valuation of these funds’ investments

Next, we are interested in overcommitted capital of a fund relative to other funds in the market. We compare a fund’s overcommitted capital with average overcommitted capital of funds with the same vintage year and call it relative over-commitment. It is expected that funds with more overcommitted capital than the average fund in the market pay higher prices for their investments.

(15)

This is formulated in the second hypothesis:

Hypothesis 2: Relative over-commitment in venture capital funds increases the valuation of these funds’ investments

Previous findings suggest that inflows of capital are associated with lower subsequent returns. Based on the hypotheses, we expect that valuations of venture capital investments increase as a result of overcommitted capital in these funds. If the hypotheses are true and funds with overcommitted capital pay higher prices for their investments, it becomes difficult for them to generate high returns. This approach of looking at the valuations of venture capital funds’ investments is thus closely related to research on private equity performance (Kaplan and Schoar, 2005; Phalippou and Gottschalg, 2009; Robinson and Sensoy, 2011; Harris et al., 2014).

3.2. Empirical design

This is the first paper that investigates overcommitted capital in venture capital funds. A new variable needs to be created that measure overcommitted capital for funds in the sample. We define overcommitted capital as the ratio of the fund size and the target size. Fund size is the total amount of capital committed to a fund and target size is the total amount a fund sought to raise when it began fundraising4_{. Overcommitted capital is greater than 1 if the actual size exceeds the target size and}

smaller than 1 if the actual size is below the target size. Consider the following example: two funds both have overcommitted capital of USD 10 million. Fund 1 had a target size of USD 100 million and Fund 2 had a target size of USD 25 million. Even though the amount of overcommitted capital is the same for both funds, it is relatively more for Fund 2. The overcommitted capital ratio reflects this difference. For Fund 1 the ratio is: !!"_!""= 1.1 and for Fund 2 the ratio is: !"_!"= 1.4. The higher ratio of the second fund indicates that this fund raised relatively more than the first fund.

Next, overcommitted capital is calculated for all funds in the sample. The effect of overcommitted capital can be estimated more precisely if we use the ratio than, for instance, a dummy variable that indicates overcommitted capital. We drop observations for which the ratio cannot be calculated, usually because the target size is unknown. Overcommitted capital is the variable of interest, so without this value the fund cannot contribute to the research question. Approximately 1500 funds cannot be used because of missing values for overcommitted capital. Overcommitted capital is calculated for almost 2800 venture capital funds. In addition, it is winsorized at a 1% level to reduce the effect of potential outliers5. Overcommitted capital is used to calculate the mean of overcommitted capital of funds with the same fund year (also known as “vintage year”). This gives

4_{According to the definitions of Thomson One.}

(16)

the opportunity to compare overcommitted capital of a fund to average overcommitted capital of other funds. This variable is called relative over-commitment. If relative over-commitment is greater than 1, the fund has more committed capital than the average fund in the market with the same fund year. Similarly if relative over-commitment is smaller than 1, this indicates that the fund has less committed capital than the average fund in the market. It does not necessarily indicate overcommitted capital. For instance, overcommitted capital of a fund can be 0.9 (actual fund size is below target size) and average overcommitted capital that year 0.8, leading to relative over-commitment of 1.125. Similarly, Usually multiple venture capital funds invest in one company. Therefore, average overcommitted capital as well as average relative over-commitment of the funds investing in the same company is calculated and used in the regressions.

To investigate if there is a relation between the valuations of venture capital investments and overcommitted capital, we follow the approach of Gompers and Lerner (2000). They use a hedonic regression approach to examine the pricing pattern. This method analyzes all price observations and thus includes companies that receive their first or follow-on financing. It enables research to incorporate firms that received only one financing round. Hedonic regression models, however, assume that the researcher is able to identify the factors that are necessary to determine the price. It will have little explanatory power if price determinants are not measurable or quantifiable (Gompers and Lerner, 2000). As a result, omitted variable bias may lead to coefficients being incorrectly estimated. This concern is addressed in section 5.3. We use an ordinary least squares (OLS) regression model, in which the dependent variable is the logarithm of the valuation at transaction date of venture capital investments. Valuation at transaction date is defined as the equity value of a portfolio company along with the round of financing it engaged up to that point. The independent variables are characteristics of the company and environment, and average overcommitted capital or average relative over-commitment. Characteristics of the company and environment include dummy variables for company status, industry and location and continuous variables for the number of employees, age, public market valuation and GDP deflator. A description of the variables is given in Table A3 in the appendix. We take logarithms of the continuous, non-negative variables for the reason that an increase (decrease) in these values should lead to a greater increase (decrease) in value for a larger company than a smaller company. The first specification of the model is displayed below:

!"# (!!) = !! + !!∗ !!+ !!∗ !!+ !!∗ !!+ !!∗ !"# !! + !!∗ !"# !! + !!∗ !"# !"#

+ !_!∗ !"#(!"#) + !_!∗ !"# + !_!

(1)

where Vi is the valuation at transaction date, Si is the firm’s current status, Ii is the industry in which

the firm operates, Li is the firm’s location, Ei is the number of employees, Ai is the firm’s age in

(17)

is average overcommitted capital. In another specification of the model, average overcommitted capital is replaced by average relative over-commitment:

!"# (!_!) = !_! + !_!∗ !_!+ !_!∗ !_!+ !_!∗ !_!+ !_!∗ !"# !_! + !_!∗ !"# !_! + !_!∗ !"# !"#

+ !!∗ !"#(!"#) + !!∗ !"# + !!

(2)

where ARO is average relative over-commitment and the other variables are the same as in the first equation. Some variables need additional explanation. The firm’s location is constructed according to the methodology of Gompers and Lerner (2000) and split into Eastern States, Western States, or elsewhere6. The number of employees, that is included as a measure of firm size, is available for the minority of firms. Accordingly, the regression model is estimated for a larger sample without a measure of firm size, and for a smaller sample with a measure of firm size. Valuation at transaction date is the nominal value of the venture capital investment and the sample includes valuations from 2005 to 2015. Nominal values need to be corrected for inflation by including the Gross Domestic Product (GDP) deflator as an independent variable. The GDP deflator is a measure of price inflation (or deflation) with respect to a base year7_{. Using quarterly levels of the GDP deflator, the GDP}

deflator is included for every venture capital transaction based on the quarter that corresponds to the transaction date. Next to inflation, the regressions contain a measure of the public market valuation. This is to control for the portion of the increase or decrease in venture capital prices that is attributable to better market circumstances. The level of the S&P 500 stock market index is chosen to account for the public market valuation. This index, consisting of 500 publicly listed companies, is considered a leading indicator of the U.S. equity market. Using daily historical prices of the S&P 500, its value on the transaction date is included for every venture capital transaction. Gompers & Lerner (2000) use a different approach to measure the public market valuation and construct 35 industry stock price indexes. Given the time constraints, this paper employs an overall market index.

6_{The list of Eastern and Western States is given in Table A3.}

(18)

4.

Data

This section describes the different sources used to collect the data for this paper. We motivate the choices for these sources and describe the decisions that were made regarding treatment of the data. Afterwards, we present descriptive statistics to provide a better understanding of the sample.

To collect information on venture capital investments and their valuations, the database Thomson One is consulted. This database provides access to information from different sources from Thomson Reuters. Thomson One has its own private equity module with data on private equity funds, fundraising, investments, exits and performance. Due to confidentiality agreements with surveyed funds, only aggregate performance data are available. The information about fundraising and investments, however, is detailed and therefore chosen to collect the sample for this paper.

In the private equity module two separate samples need to be generated, which are later combined. The first sample contains all U.S. venture capital transactions from January 2005 to December 2015 (the sample period). It includes relevant transaction details such as company name, transaction date, financing round, private equity fund(s) involved and the valuation at the transaction date. Several other characteristics of the company and the environment are also included such as location, industry, status, age and the number of employees. These characteristics are used as independent variables in the regressions. In total, we have information on 4,906 venture capital transactions and their valuations.

The second sample consists of U.S. private equity funds that were raised between 1990 and 2016. The funds specializing in venture capital are selected. Most funds have a fixed life of 10 years, which can be extended by a couple of years. Since the sample period for venture capital transactions is from 2005-2015, we assume that funds raised prior to 1990 do not participate in these investments anymore. Funds usually have a fixed life of 10 years (with a few years of extensions), so all funds with a fund year below 1990 are removed. To be able to calculate a fair value of overcommitted capital only funds for which fundraising has ended should be included in the sample. Two kinds of funds are deleted: funds that had their final close but were downsized and funds that are still raising. For the remaining funds the report contains information about fund size, target size, vintage year and location. As explained in section 3.2., fund size and target size are used to calculate overcommitted capital, defined as the ratio of fund size and target size. Of the 4,348 venture capital funds in the sample, we are able to calculate overcommitted capital for 2,793 funds (64%).

Two other databases are used to complement the data from Thomson One. First, the Federal Reserve Economic Data (FRED) database is consulted for the quarterly GDP deflator (base 100 in 2009). Second, a measure of the public market valuation is included in the regressions. The level of the S&P 500 is downloaded from the Center for Research in Security Prices (CRSP) database. This database has research-quality data, including daily adjusted closing values of the S&P 500 index.

(19)

Merging the samples creates the final sample. Of the 4,905 transactions 2,169 transactions (44%) can be merged with a venture capital fund. Average overcommitted capital and relative over-commitment are calculated for the funds investing in the same company. For each company we keep only one of the transactions that have the exact same date and valuation, thereby randomly selecting one of the venture capital funds that invested in that company. This does not cause any problems because average overcommitted capital and relative over-commitment are calculated for the funds in that company and are the values used in the regressions. Transactions of the same company with a different transaction date and valuation are kept in the sample. Those observations represent different financings of the same company. The final sample consists of 1,122 observations. In addition, the empirical analysis is carried out for the sample that drops all other transactions of the same company. The results are discussed in section 5.1.

Table 1 analyzes the different characteristics of the firms in the sample. The total sample has a mean of 3.56 and median of 3.61. Panel A, B and C show descriptive statistics for several firm characteristics. In each panel, t-tests for means and Wilcoxon rank-sum tests for medians are presented. This compares a group with the rest of the firms in the sample. For instance, firms that are still active have a mean and median that is significantly different from the mean and median of the rest of the firms. Most firms are still active (41%), followed by firms that have been acquired (32%) and firms that went public through an IPO (18%). Firms that went public have successfully entered the public markets and receive substantial higher valuations. Further it seems as if firms located in the Western States have on average higher valuations compared to firms in the Eastern States and elsewhere. Representing more than half of the sample (51%), firms in the Western States have a mean of 3.78. This is consistent with common knowledge that many high-tech and startup companies are located in Silicon Valley in California. There are also differences between the mean valuations in the various industries. The mean of firms in the communications, medical/health and semiconductors industry is higher than the sample mean. But according to the t-tests, they do not differ significantly from the rest. Firms in the non-high technology industry have lower valuations, significant at the 10% level. The computer related industry, varying from computer hardware, computer software and services, and internet-specific firms, represents the greatest portion of firms in the sample (33%). The medical/health industry and biotechnology industry account for 25% and 21% of firms in the sample. The communications, non-high technology and semiconductors industry each represent a smaller portion of the firms in the sample; 8%, 6% and 8% respectively. The main idea of Table 1 is to show that there are differences between the valuations of venture capital investments with different characteristics.

Table 2 presents the mean, median and standard error of the logarithm of valuation for each year over the time period. First of all, it is worth noticing that the number of observations decreases heavily over the time period. The number of observations in 2005 is 293, while the number of

(20)

Table 1 –Venture capital investments by firm characteristics

The sample consists of 1,122 venture financings of private companies between January 2005 and December 2015 from Thomson One. The valuation at transaction date is defined as the value of the portfolio company on the date of transaction (in USD millions). The table presents the mean, median, standard error, number of observations and percentage of the logarithm of the valuation at transaction date for each of the following categories: status, industry and location. T-tests and Wilcoxon rank-sum tests of a group versus the rest are performed for means and medians respectively.

Log of Valuation at Transaction Date

Mean Median Std. _Error No. of. _Obs. Percentage

Panel A: Status of firm

Firm is acquired 3.52 3.59 0.06 363 32%

Firm is still active 3.22*** 3.11*** 0.07 464 41%

Firm went public 4.31*** 4.48*** 0.08 206 18%

Other 3.75 3.94 0.12 89 8%

Panel B: Location of firm

Eastern States 3.42** 3.44** 0.07 350 31%

Western States 3.78*** 3.86*** 0.06 570 51%

Elsewhere 3.19*** 3.11*** 0.10 202 18%

Panel C: Industry of firm

Biotechnology industry 3.54 3.69 0.09 235 21%

Communications industry 3.66 3.64 0.15 94 8%

Computer related industry 3.48 3.25*** 0.09 367 33%

Medical/health industry 3.66 3.93*** 0.07 278 25%

Non-high technology industry 3.25* 3.33* 0.20 63 6%

Semiconductors industry 3.79 3.83 0.11 85 8% Total 3.56 3.61 0.04 1122 100% *** Significant at 1% level ** Significant at 5% level * Significant at 10% level.

observations in 2015 is only 14. This is due to the fact that databases can sometimes collect information about earlier financing rounds at a firm’s refinancing. A firm, for example, that refinanced in 2009 could have delivered data for two earlier financing rounds in 2005 and 2007. Because firms in later years haven’t sought refinancing yet, recent data may be less complete. This form of selection bias could affect the completeness of the sample. Table 2 also shows the significance of the t-tests and Wilcoxon rank-sum tests to be able to compare valuations in a particular year with valuations in other years. In the first couple of years, the yearly means and medians fluctuate around the sample mean and median. Valuations decrease in the period following the 2008 financial crisis. In later years, from 2012 to 2015, means and medians increase rapidly (except for 2013) and are larger than those of the total sample. The standard error of the mean is also larger in later years, indicating that the true population mean can differ substantially from the sample mean in these years.

(21)

If we look at the column with average overcommitted capital in each year, we see that the mean varies between 0.76 (in 2012) and 1.06 (in 2015). For most years, the mean is close to 1. In 2005 the funds investing in private companies had on average overcommitted capital of 1.02, meaning that they received on average some additional capital over their target size, which they could spend on investments. Over the 10-year period, average committed capital exceeds 1 in three years. It is more common that average committed capital is less than 1 and funds do not reach their fundraising target. The column next to average overcommitted capital shows the public market valuation, as measured by the level of the S&P 500 stock index. It generally follows the cyclicality of the economy. The financial crisis in 2008 was followed by a period of recession and is reflected in the level of the S&P 500 among other things. The public market valuation corresponding to the venture capital investments is at its lowest point in 2009, when the mean is 935. In 2010, 2011 and 2012 it is still below the pre-crisis level mean of 2007. In the more recent years the public market valuation increases quickly and even surpasses 2000 in 2015. Figure 2 graphically depicts the logarithm of valuation at transaction date and the public market valuation over the sample period.

Table 2 – Venture capital investments by year

The sample consists of 1,122 venture financings of private companies between January 2005 and December 2015 from Thomson One. The valuation at transaction date is defined as the value of the portfolio company on the date of transaction (in USD millions). The table presents the mean, median and standard error of the logarithm of valuation at transaction date, the mean of average overcommitted capital, the mean value of the S&P 500 used to control for the public market valuation and the number of observations for each year. T-tests and Wilcoxon rank-sum tests of a group versus the rest are performed for means and medians respectively.

Year Log of Valuation at Transaction Date

Average Public market No. of Obs.

Mean Median Std. Error overcommitted capital valuation 2005 3.42** 3.53* 0.07 1.02 1205 293 2006 3.73** 4.01*** 0.08 0.99 1312 228 2007 3.52 3.71 0.10 0.97 1476 153 2008 3.24*** 3.48* 0.13 0.96 1230 121 2009 3.52 3.44 0.10 0.99 935 124 2010 3.29* 3.19** 0.14 0.97 1141 80 2011 3.15** 2.89*** 0.16 0.92 1263 68 2012 5.02*** 4.66*** 0.36 0.76 1371 8 2013 4.27** 4.17 0.43 0.99 1605 18 2014 5.97*** 6.39*** 0.40 1.05 1938 15 2015 6.63*** 7.11*** 0.65 1.06 2087 14 Total 3.56 3.61 0.04 0.99 1264 1122 *** Significant at 1% level ** Significant at 5% level * Significant at 10% level.

(22)

Figure 2 –Venture capital investments and the level of the S&P 500

Pearson correlations are shown in Table 3. The dummy variables for the firm’s status, industry and location are not included in the correlation matrix. For descriptive statistics of these variables please refer to Table 1. From Table 3 we see that the logarithm of valuation at transaction date is positively correlated with all variables in the correlation matrix. The correlation coefficient between average overcommitted capital and the logarithm of valuation at transaction date is 0.21 and highly significant. Between average relative over-commitment and valuation the correlation coefficient is 0.20. These positive correlations are in line with the suggestion that an increase in the supply of venture capital results in greater competition and rising valuations. Not surprisingly, there is a strong correlation between average overcommitted capital and average relative over-commitment. If funds investing in a particular company have high average overcommitted capital, it is also more likely that they have relatively more capital than the market. In addition, older firms and firms with more employees have higher valuations. The correlation coefficient between the logarithm of age and the logarithm of number of employees is 0.19. Often older firms have more employees, because they have had more time to hire new employees.

0 500 1000 1500 2000 2500 0 1 2 3 4 5 6 7 8 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Pu bl ic m ar ke t v al ua9 on Lo g of V al ua9 on at Tr an sac 9o n D ate

(23)

Table 3 – Pearson correlations

The table contains the Pearson correlation matrix for seven variables including the logarithm of valuation at transaction date, average overcommitted capital, average relative over-commitment and the logarithms of age (in months), public market valuation, GDP deflator and number of employees. Correlation coefficients and their significance are presented, with p-values reported between brackets.

Variables 1. 2. 3. 4. 5. 6. 7.

1. Log of valuation at transaction date 1

2. Average overcommitted capital 0.21*** 1

(0.00)

3. Average relative over-commitment 0.20*** 0.93*** 1

(0.00) (0.00)

4. Log of age 0.37*** 0.06* -0.02 1

(0.00) (0.05) (0.62)

5. Log of public market valuation 0.19*** -0.01 0.03 0.01 1

(0.00) (0.69) (0.32) (0.85)

6. Log of GDP deflator 0.14*** -0.07** 0.08*** 0.09*** 0.13*** 1

(0.00) (0.02) (0.01) (0.00) (0.00)

7. Log of number of employees 0.38*** 0.09* 0.04 0.19*** -0.02 -0.08 1

(0.00) (0.10) (0.48) (0.00) (0.78) (0.17)

*** Significant at 1% level ** Significant at 5% level * Significant at 10% level.

(24)

5. Empirical results

This section begins with the results of the basic econometric analysis conform the empirical design in section 3.2. We interpret the results and, in the second part of this section, discuss them in terms of robustness. Finally we develop an instrumental variable (IV) approach to address omitted variable bias, for which we review the results at the end of this section.

5.1. Basic regression analysis

The ordinary least squares (OLS) regression model is presented in Table 4. The empirical results are divided into five columns, representing different specifications of the model. All regressions are performed using heteroscedasticity robust standard errors. The first column of Table 4 is a regression of valuation at transaction date on characteristics of the firm and environment only. Average overcommitted capital and average relative over-commitment are left out. Column 2 to 5 show the different specifications described in section 3.2., where one of the independent variables is average overcommitted capital or average relative over-commitment. Adding these variables increases the R2 from the base regression in all cases. In column 2, the coefficient of average overcommitted capital is 0.91 and statistically significant at the 1% level. This suggests that if average overcommitted capital increases by 10%, valuation increases by 9.1%. Since overcommitted capital is the ratio of fund size and target size and assuming target size is fixed, 10% additional capital for funds leads to 9.1% higher valuations. This result is consistent with the first hypothesis that overcommitted capital in venture capital funds increases the valuations of these funds’ investments. In column 3, average overcommitted capital is replaced by average relative over-commitment, indicating whether or not the investing funds have more overcommitted capital than the average fund in the market. The coefficient is 0.84 and significant at the 1% level. If funds investing in the same company have relatively more overcommitted capital than the average fund in the market, the valuation of that company is positively affected. More specifically, the valuation of venture capital investments increases with 8.4% if the investing funds have 10% more committed capital relative to funds in the market. Column 4 and 5 show the results of similar regressions but include the logarithm of the number of employees to account for firm size. The number of employees is available for the minority of companies and reduces the sample size. But the newly added variable increases the explanatory power of the regression model and has a positive and significant effect on valuation: an increase in the number of employees of 10%, increases valuation by 2.1%. It suggests that larger companies have higher valuations. The coefficients of average overcommitted capital and average relative over-commitment remain positive and significant but are somewhat smaller. The coefficient is 0.80 for average overcommitted capital (compared to 0.91 in the full sample) and 0.78 for average relative over-commitment (compared to 0.84 in the full sample).

(25)

Next, we look at the coefficients of the other independent variables. In different specifications of the model, the coefficient of an independent variable is usually similar. It can, however, vary sometimes in magnitude and significance level. Having gone public has a great impact on the valuation of firms. The coefficient of the IPO dummy is positive and statistically significant. In the full sample, firms that went public have 84-86% higher valuations. The magnitude decreases in the smaller sample that includes a measure of firm size: the effect on the valuation for firms that went public is approximately 40%. Venture capitalists often prefer an IPO as a way of cashing out on an investment. Giot and Schwienbacher (2007) study exit options for U.S. venture capital funds and suggest that an exit order exists; first an IPO and then possibly a trade sale. Since not all private companies can go public, the ones that do are likely the more successful ones. The location of firms also affects valuation. Firms in the Eastern and Western States have higher valuations than firms elsewhere. If the firm is located in a Western State, valuation increases with 55%. For Eastern States the effect is smaller and valuation increases with approximately 21%. The geographical patterns found are consistent with the results of Gompers and Lerner (2000), who find positive and statistically significant coefficients for Eastern and Western States. This is in line with the expectations since firms located in startup ecosystems, such as Silicon Valley, New York and Los Angeles, enjoy a variety of benefits leading to higher valuations. These benefits are, for instance, the presence of specialized intermediaries, highly skilled employees and technological spillovers (Krugman, 1991). In the smaller sample the geographical patterns do not hold and the coefficients on Eastern and Western states are not significant. The results suggest that firms in the communications, medical/health and semiconductors industry have higher valuations. The coefficients of these industries are positive in all of the regressions, but statistically significant in regressions using the full sample only. Gompers and Lerner (2000) also find that firms in these industries are associated with higher valuations. In addition they find that firms in the computer hardware industry have higher valuations, but this paper does not find empirical evidence for that. In this paper we employ a computer related industry that consists of companies specializing in computer hardware, software and the internet. No distinction between the different types of companies in the computer related industry could explain the difference with Gompers and Lerner (2000).

Furthermore, the results suggest that age, measured in months, is a proxy of future earnings and hence valuation. If a firm’s age increases by 10%, its valuation increases by 5%. The coefficient on age remains positive and highly significant in each of the regressions. Usually older venture-backed firms have revised their business plan many times and dealt with some sort of survivorship bias, which results in higher valuations. The public market valuation and GDP deflator also positively affect valuations. A 1% increase in the public market valuation leads to approximately 1-1.5% higher valuations. Similarly, if prices in general increase, which is reflected in a higher GDP deflator, valuations of new investments increase.

(26)

Table 4 – Ordinary least squares regression analysis of valuation

The sample consists of 1,122 venture financings of private companies between January 2005 and December 2015 from Thomson One. The dependent variable is the logarithm of the valuation at transaction date, defined as the value of the portfolio company on the date of transaction (in USD millions). Independent variables include dummy variables for the firm’s status (equal to 1 if the firm went public), industry and location and variables for the number of employees, firm’s age (in months), public market valuation (the level of the S&P 500 at the transaction date), GDP deflator and overcommitted capital (the ratio of fund size and target size) or relative over-commitment (compared to average overcommitted capital in the market). For each company, the mean of overcommitted capital and relative over-commitment is calculated. Robust standard errors are reported in brackets. ***, ***, and * denote significance at the 1%, 5%, and 10% level respectively.

Dependent variable: Log of Valuation at Transaction Date

Full Sample Full Sample Full Sample Incl. Firm Size Incl. Firm Size

(1) (2) (3) (4) (5)

Average overcommitted capital 0.91***

(0.16)

0.80*** (0.27)

Average relative over-commitment 0.84*** _(0.15) 0.78*** _(0.23)

IPO dummy 0.86*** 0.84*** 0.84*** 0.40*** 0.39*** (0.09) (0.09) (0.09) (0.11) (0.11) Location of firm: Eastern States 0.21* 0.21* 0.22* -0.02 -0.02 (0.11) (0.12) (0.12) (0.17) (0.17) Western States 0.55*** 0.55*** 0.55*** 0.15 0.15 (0.10) (0.10) (0.10) (0.17) (0.17) Industry of firm: Biotechnology 0.03 -0.07 -0.08 0.05 0.05 (0.19) (0.18) (0.18) (0.33) (0.33) Communications 0.50** 0.37* 0.40** 0.15 0.19 (0.22) (0.20) (0.20) (0.34) (0.35) Computer related 0.14 0.06 0.05 -0.25 -0.23 (0.19) (0.17) (0.18) (0.33) (0.33) Medical/health 0.43* 0.33* 0.34* 0.19 0.20 (0.19) (0.17) (0.17) (0.32) (0.32) Semiconductors 0.46* 0.33* 0.35* 0.16 0.20 (0.21) (0.19) (0.19) (0.38) (0.39)

Log of age (in months) 0.50*** 0.48*** 0.51*** 0.28*** 0.31***

(0.04) (0.04) (0.04) (0.08) (0.08)

Log of public market valuation 1.46*** 1.45*** 1.42*** 0.95* 0.96*

(0.28) (0.27) (0.27) (0.50) (0.49)

Log of GDP deflator 4.02*** 4.40*** 3.60*** 5.70*** 4.97***

(1.05) (1.03) (0.98) (2.03) (1.89)

Log of number of employees 0.21*** 0.21***

(0.05) (0.05)

Constant -27.84*** -30.30*** -26.56*** -31.69*** -28.71***

(5.74) (5.55) (5.28) (10.90) (10.15)

R-squared 0.27 0.30 0.30 0.32 0.33

(27)

The above analysis is also carried out for the sample that drops all other transactions of the same company. Only one observation of companies that appear in the sample multiple times, indicating different financing rounds, is kept. This sample consists of 786 observations and the results are similar. Funds’ overcommitted capital and relative over-commitment continue to have a positive effect on valuation.

5.2. Robustness checks

We document a positive relation between overcommitted capital in venture capital funds and the valuation of these funds’ investments. However, errors made in the specifications of the regression model could lead to false estimation of coefficients. This subsection aims to assess the robustness of the results by adding several control variables to the basic analysis. The variables address two alternative hypotheses. First, the public market valuation should be measured with a small-cap stock index instead of the S&P 500. Second, there are differences between first and later round investors. The main idea of Table 5 is to show that the controls have little impact on the coefficients or significance of average overcommitted capital and average relative over-commitment.

The first concern is that the measure of the public market valuation used in the regressions, namely the S&P 500 stock index, is not representative for firms in the sample. The firms that are included in the S&P 500 are mostly larger than the private companies in the final sample. Besides, they have already successfully entered the public capital markets. To address this concern, a small-cap stock index is added to the regression model. This is the Russell 2000; an index that measures the performance of 2000 small-cap companies and serves as an important benchmark for small-cap stocks in the U.S. Column 1 and 2 of Table 5 show the relevant coefficients when the Russell 2000 replaces the S&P 500 as a measure of the public market valuation. These are similar regressions as in column 2 and 3 from Table 4 except that the public market valuation is a different index. The coefficients of average overcommitted capital and average relative over-commitment in Table 5 are the same as in Table 4. The Russell 2000 has a positive and significant effect on valuation, but its coefficient is a bit smaller than the coefficient of the S&P 500. Valuations react stronger on an increase of the S&P 500 than on an increase of the Russell 2000.

The second concern is that there may be differences between first and later round investors. Therefore we control for the financing round by adding a dummy variable. This variable is equal to 1 if the venture capital investment is a second or later round financing and 0 otherwise. According to Lerner (1994), established venture capitalists tend to syndicate with each other in the first round while less established venture organizations become involved in later rounds. The experience that established venture capitalists offer is very valuable for the early-stage company. Later-round investors often pay a premium, because the firm is doing well and they are rarely asked to provide value-added services. Column 3 and 4 of Table 5 report the coefficients when adding the dummy

The role of overcommitted capital in the valuation of venture capital funds’ investments