Grandstanding, certification and the underpricing of venture capital backed IPOs

https://doi.org/10.1016/j.jfineco.2003.09.003Get rights and content

Abstract

We examine the role of venture capital backing in the underpricing of IPOs. Controlling for endogeneity in the receipt of venture funding, we find that venture capital backed IPOs experience larger first-day returns than comparable non-venture backed IPOs. Between 1980 and 2000, the average return difference ranges from 5.01 percentage points to 10.32 percentage points. This return difference is particularly pronounced in the “bubble” period of 1999–2000. Consistent with the grandstanding hypothesis proposed by Gompers (J. Financial Econ. 42 (1996) 133), we find that higher underpricing leads to larger future flows of capital into venture capital funds, particularly after 1996. Cross-sectionally, the effect of underpricing is attenuated for younger venture capital firms and those that have previously conducted fewer IPOs.

Introduction

In this paper, we examine the role of venture capital backing in the underpricing of initial public offerings (IPOs). We are not the first to conduct such an investigation. The pioneering efforts in this literature are due to Megginson and Weiss (1991) and Barry et al. (1990). Megginson and Weiss compare venture capital (VC) backed IPOs to non-VC backed IPOs matched by industry and offering size between January 1983 and September 1987. They find that the first-day returns of VC backed IPOs are significantly lower than those of non-VC backed IPOs.1 They argue that their results are consistent with the intuitively plausible idea that venture capitalists certify the true value of the firm and therefore reduce underpricing. Barry et al. focus on the monitoring role of venture capitalists in IPOs between 1978 and 1987 and find that the ownership, the length of board service, and the number of venture capitalists invested in the pre-IPO firm are negatively related to IPO underpricing. Based on this correlation, they conclude that venture capitalists are “recognized by capital markets through lower underpricing for IPOs with better monitors” (p. 447).

Since the publication of these two studies, a stream of research seeking to understand the provision of venture financing has emerged.2 Sahlman (1990) reports that “although a typical large venture capital firm receives up to 1,000 proposals each year, it invests in only a dozen or so new companies” (p. 475). Kaplan and Stromberg (2002) use unique data to describe the due diligence and analyses conducted by venture capitalists prior to the provision of financing. They show how the characteristics of and risks inherent in entrepreneurial ventures translate into specific contractual provisions with entrepreneurs. Other studies characterize the contracting environment between entrepreneurs and venture capitalists as one in which the provision of venture financing is linked to the allocation of cash flow and control rights (Sahlman, 1990; Gompers and Lerner, 1996; Black and Gilson, 1998; Hellman, 1998; Kaplan and Stromberg 2001, 2002). Such allocations take place through covenants on securities exchanged for cash, in the distribution of ownership and voting rights, and in the assignment of board seats. Both the ex ante analyses conducted by venture capitalists, and the contracts designed to mitigate information and agency problems, suggest that venture financing represents an endogenous choice made by entrepreneurs and venture capitalists – the receipt of venture funding is the outcome of protracted negotiations between venture capitalists and entrepreneurs.

This choice is reflected in the nonrandom distribution and characteristics of VC backed IPOs. We examine a sample of over 6,413 IPOs between 1980 and 2000, of which over 37% (2,383) are VC backed. VC backed IPOs show significant clustering across both industry and geographical dimensions. We find that venture financing is disproportionately provided to firms in technology-intensive industries, particularly software and commercial biological research. Consistent with the results of Lerner (1995) and Gompers and Lerner (1998), over 50% of all venture backed IPO firms in our sample are headquartered in California, Massachusetts, and Texas, states with high concentrations of venture capitalists. We also find differences in the characteristics of firms taken public by venture capitalists. Venture capitalists generally take smaller and younger firms public. These firms have lower revenues than non-VC backed firms and are less likely to be profitable. On average, VC backed IPOs employ higher quality underwriters but raise less cash than non-VC backed IPOs.

Our interest is in the underpricing of VC backed IPOs. In an ideal world, we would want to observe underpricing for a VC backed IPO and the underpricing that the same IPO would experience had it not received venture financing. This would allow us to make causal inferences about the effect of venture backing on IPO underpricing. Unfortunately, given the nonexperimental nature of the data, what we actually observe is the underpricing for a VC backed IPO and the underpricing for a non-VC backed IPO. If the provision of venture financing were random, one could simply compute differences between the first-day returns of VC backed and non-VC backed IPOs. The traditional approach would be to match VC backed IPOs to non-VC backed IPOs based on a set of characteristics and attribute the difference in underpricing to venture backing, or to estimate OLS regressions with a dummy variable for VC backing. Using such regressions, Bradley and Jordan (2002) conclude that once they control for industry effects and underwriter quality, there is no difference in underpricing between VC backed and non-VC backed IPOs. Gompers and Lerner (1997) question the certification hypothesis and show that the underpricing differential between VC backed and non-VC backed IPOs is sensitive to both estimation periods and methodologies. Although controls are undeniably important, the crux of our problem is that venture backing is not randomly distributed, but represents an endogenous choice. This introduces a selectivity bias, one that can easily reverse inferences.3

To account for this bias, we use matching methods that endogenize the receipt of venture financing and do not impose linearity or function form restrictions (see, for example, Rubin (1974), Rubin (1977); Rosenbaum and Rubin, 1983; Dehejia and Wahba, 1999). The basic intuition behind these procedures is to use a first-stage regression to predict the receipt of venture funding. Estimates from this first-stage regression are then fed into various methods of matching non-VC backed IPOs to VC backed IPOs. Addressing the endogeneity issue directly produces results that are strikingly different from earlier empirical studies. Over the entire sample period, the average first-day return of VC backed IPOs is larger than non-venture backed IPOs. The return difference ranges from a minimum of 5.0 percentage points to a maximum of 10.3 percentage points, with standard errors of 1.8 and 2.1 percentage points respectively. These return differentials are statistically significant and economically important. The average underpricing of all IPOs over the sample period is about 18.0%. Using the smallest estimate (5.0 percentage points), the underpricing differential as a proportion of total underpricing is 28.0%.

Because there is pronounced non-stationarity in the volume and underpricing of IPOs (Ritter and Welch, 2002; Loughran and Ritter, 2002a), we examine the underpricing differential over various subperiods. In particular, it is possible that our results are entirely due to the internet bubble period of 1999–2000. We find that the average underpricing differential between VC backed and non-VC backed IPOs is 25.0 percentage points in 1999–2000 and 2.0 percentage points in 1980–1998, with standard errors of approximately 9.9 and 1.0 percentage points respectively. Again, the underpricing differential is statistically significant and economically important in both subperiods. The average underpricing of IPOs in 1980–1998 is 11.0%. Therefore, the 2.0 percentage point differential represents almost 20% of total underpricing. Another way to assess the economic magnitude of these return differentials is to calculate the incremental amount of money left on the table in VC backed IPOs. Loughran and Ritter (2002b) calculate this amount as the number of shares issued multiplied by the difference between the closing price on the first day of trading and the offer price. Using their metric, the average amount of money left on the table for non-VC backed IPOs in our sample is $27 million. The average first-day return for the non-VC backed IPOs is 13.4%. Again, using our smallest estimate of the return differential (5.0 percentage points), this implies an additional $10 million left on the table by VC backed IPOs. Even excluding the bubble period, similar calculations suggest an additional $4 million left on the table.

Venture capitalists typically retain a large fraction of their equity holdings subsequent to an IPO; Megginson and Weiss (1991) report that on average, venture capitalists own 36.6% of the firm prior to the IPO and 26.3% immediately thereafter. Greater underpricing represents a real cost to the venture capital fund because there is a transfer of wealth to new shareholders. Why would venture capitalists bear this cost? It is possible that the first-day return differentials are due to our inability to perfectly account for endogeneity and for differences between VC backed and non-VC backed firms. While no endogeneity correction or system of controls is perfect, our results are robust to a variety of specifications.

One explanation is that venture capitalists receive some quid pro quo for leaving more money on the table. It is possible that underwriters preferentially allocate shares of other underpriced IPOs to venture capital firms in exchange for greater underpricing in the VC's own portfolio firm. Loughran and Ritter (2002a) provide examples in which VC firms were allocated hot IPOs that were subsequently flipped for immediate profits. Recent legal investigations into IPO practices also support this scenario (see, for example, “Something ventured, something gained?,” Wall Street Journal, 10/17/2002). However, such activity occurred largely in 1999 and 2000. The underpricing differential appears in earlier periods as well. Therefore, we doubt that such quid pro quo arrangements can explain our results.

A more likely explanation lies in the grandstanding hypothesis proposed by Gompers (1996). Venture capital firms create limited partnerships (“venture capital funds”) to raise and invest capital. These funds have finite lives, typically ten years, after which they must liquidate their investments and return money to the original providers (institutional investors such as pension funds and endowments). Although VC firms can realize returns through acquisitions as well as IPOs of their portfolio firms, the majority of returns are created by taking companies public. Because of this, establishing a reputation as a VC firm that is capable of taking portfolio companies public is critical to future fundraising. Gompers (1996) describes examples in which VC firms that are unable to take portfolio companies public are unable to raise future capital. He also describes converse examples of VC firms that, once they take a portfolio company public, are quickly able to raise additional capital. Since establishing reputation is critical to future fundraising, VC firms are willing to bear the cost of underpricing because taking a company public signals quality. Gompers (1996) argues that cross-sectionally, fundraising is less of a problem for older VC firms because their reputations are already established. In contrast, less-established VC firms need to signal quality by taking portfolio companies public. As a result, they are more willing to bear the cost of higher underpricing. Consistent with this argument, he finds that younger VC firms grandstand by taking younger companies public and allowing greater underpricing. This enables young VC firms to raise more capital in the future than they would otherwise. If grandstanding is responsible for our results, then underpricing should have a larger effect on the fundraising ability of low-reputation VC firms. Also, such VC firms should be more likely to take smaller and younger companies public.

To determine if these effects are present in our data, we first create two subsamples of VC-backed IPOs with different definitions for the “lead” VC firm in the IPO. Specifically, we identify the lead venture capital firm as the VC firm with the earliest investment, or the VC firm with the largest investment. We then estimate fundraising regressions similar to Gompers (1996) for both subsamples. The dependent variable in these regressions is the size of the next fund raised by the lead VC firm after the IPO. The primary independent variables of interest are proxies for VC reputation (VC age and the number of previous IPOs done by the VC), IPO underpricing, and interaction effects of reputation proxies with underpricing.

The regressions show that the flow of capital into the lead VC firm is positively related to VC age and the number of previous IPOs done by the VC firm. This indicates that more-reputable venture capital firms are able to raise more money. The future flow of capital is also positively related to underpricing, implying that there is a benefit to bearing the cost of underpricing. Most important, the interaction effects between VC age and underpricing, and between the number of prior IPOs and underpricing, are negative. Consistent with grandstanding, this suggests that underpricing has a larger effect on the ability of young VC firms, or those that have done fewer IPOs, to raise future capital. Since the grandstanding explanation also predicts that low-reputation VC firms are more likely to take smaller and younger companies public, we also estimate regressions of both measures of reputation on IPO characteristics. We find that younger VC firms, and those that have previously done fewer IPOs, tend to take smaller and younger companies public.

There is a marked increase in the number of new firms entering the venture capital industry in the late 1990s, roughly corresponding to the increase in IPO activity. Aggregate data show that on average, 33 new firms entered the industry annually between 1980 and 1996. Between 1997 and 2000, this more than tripled to 106 new firms entering the industry every year. If the entry of new firms increases competition, there may be a change in the willingness of incumbent firms to bear the cost of underpricing to raise further capital. We estimate fundraising regressions for the 1980–1996 and 1997–2000 subperiods separately and find that the sensitivity of future capital flows to underpricing is higher in the later subperiod. Interestingly, the interaction effects between reputation and underpricing are significant in both periods, suggesting that the cross-sectional effects have not changed.

These results favor the grandstanding explanation. However, we also consider other explanations. Gompers (1996) describes the “recycling hypothesis” as one in which investors in venture capital firms plough their profits from earlier funds back into new funds raised by the same venture capital firms. However, the recycling hypothesis does not explain why some VC firms would receive more capital, or why underpricing would have larger effects for low-reputation VC firms. Another possibility is a prospect theory explanation advanced by Loughran and Ritter (2002b). Under their explanation, issuers ignore the wealth loss from underpricing because they sum this wealth loss with the wealth gain from the IPO itself. This implies that the positive relation between underpricing and future capital flows could simply be an IPO effect, rather than an underpricing effect. To separate the two effects, we compile a sample of all venture capital firms that are in existence in a given year for the 21-year period. This sample includes both VC firms that took a company public and those that did not. We then estimate two-stage Heckman selection regressions that account for the fact that underpricing is only observed when a company is taken public. The first-stage regression predicts whether a VC firm does an IPO in that year and feeds into a second-stage regression that models capital inflows as a function of IPO underpricing. The coefficient on IPO underpricing is positive and significant in the second-stage regression, implying that underpricing has a positive effect on future capital inflows, even after correcting for the observability bias.

Taken together, the evidence suggests that the grandstanding behavior originally documented by Gompers (1996) plays an important role in describing the costs that venture capitalists are willing to bear in taking their portfolio companies public. The remainder of the paper proceeds as follows. Section 2 describes our data and sample construction. Section 3 describes the methodological approach and provides the basic underpricing results. Section 4 describes the capital flow regressions. Section 5 describes various robustness checks and examines alternative explanations. Section 6 concludes.

Section snippets

Data

Our sample of IPOs comes from data provided by Jay Ritter. This dataset includes accumulated corrections made by him to IPO data from a variety of sources. The data include offering dates, offering prices, file price ranges, closing prices, SIC codes, and underwriter rankings. Underwriter rankings are based on an amended version of the Carter and Manaster (1990) and Carter et al. (1998) rankings and are described in Loughran and Ritter (2002a). The rankings range in value from 0 to 9, with

Unadjusted first-day returns

Table 1 shows the number of IPOs and average first-day returns. First-day returns are calculated as the percentage price movement from the offering price to the closing price on the first day of trading. For VC backed IPOs, we present raw first-day returns, and first-day returns of matched non-VC backed IPOs using a methodology similar to Megginson and Weiss (1991). (The returns for non-matched, non-VC backed IPOs are not shown in the table.) Each VC backed IPO is matched with a (single) non-VC

Explaining the underpricing differential

Venture capitalists are major shareholders in firms prior to an IPO, and typically retain some portion of their stakes following the IPO. Barry et al. (1990) report that the average VC holding in pre-IPO firms is 34.3%, and 24.6% immediately following the IPO. Comparable statistics reported by Megginson and Weiss (1991) are 36.6% and 26.3%, respectively. Therefore, higher first-day returns represent a real, incremental cost to venture capitalists because of the wealth transfer to new

Robustness issues and alternative explanations

The evidence thus far is consistent with the idea that underpricing is related to grandstanding by venture capital firms. This appears to be reflected in the costs they are willing to bear as well as the cross-sectional distribution of the types of firms that engage in this activity. In this section, we consider issues that could affect our inferences and examine other potential explanations for our results.

Conclusion

We investigate the role of venture capitalists in the underpricing of IPOs between 1980 and 2000. We argue that the provision and receipt of venture financing represents an endogenous choice on the part of the venture capitalist and the entrepreneur. We use instruments correlated with this endogenous choice to control for selection bias and compare the underpricing of VC backed and non-VC backed IPOs. Our results show that VC backed IPOs exhibit greater underpricing than non-VC backed IPOs. The

References (42)

  • R. Carter et al.

    Underwriter reputation, initial returns and the long run performance of IPO stocks

    Journal of Finance

    (1998)
  • H. Chen et al.

    The seven percent solution

    Journal of Finance

    (2000)
  • R. Dehejia et al.

    Causal effects in non-experimental studiesevaluating the evaluation of training programs

    Journal of the American Statistical Association

    (1999)
  • P. Gompers

    Optimal investment, monitoring and the staging of venture capital

    Journal of Finance

    (1995)
  • P. Gompers et al.

    The use of covenantsan empirical analysis of venture partnership agreements

    Journal of Law and Economics

    (1996)
  • Gompers, P., Lerner, J., 1997. Venture capital and the creation of public companies: do venture capitalists really...
  • P. Gompers et al.

    What drives venture capital fundraising

    Brookings Papers on Economic Activity – Microeconomics

    (1998)
  • P. Gompers et al.

    The venture capital revolution

    Journal of Economic Perspectives

    (2001)
  • J. Heckman

    Sample selection bias as a specification error

    Econometrica

    (1979)
  • J. Heckman et al.

    Matching as an econometric evaluation estimatorevidence from evaluating a job training program

    Review of Economic Studies

    (1997)
  • J. Heckman et al.

    Matching as an econometric evaluation estimator

    Review of Economic Studies

    (1998)
  • Cited by (0)

    We thank Jay Ritter for providing IPO data and Alexander Ljungqvist for providing supplemental information on firm founding dates. Daniel Silver provided able research assistance and Ron Harris provided excellent computational assistance. We thank an anonymous referee, Dave Denis, Diane Denis, Yael Hochberg, Bill Megginson, Jay Ritter, and participants at the EVI conference at Yale University, the Western Finance Association meetings in Cabo San Lucas, and at the SMS conference in Paris for helpful comments and suggestions.

    View full text