Do Venture Capitalists Certify New Issues in the IPO Market? Yan Gao

Transcription

1 Do Venture Capitalists Certify New Issues in the IPO Market? Yan Gao Northwestern University Baruch College, City University of New York, New York, NY Current version: 6 Novermber 2002 Abstract In addition to corporate executives, venture capitalists (VCs) have been receiving hot IPO shares from investment banks. The cozy relationship between VCs and investment banks calls in question of the well-documented certification role of VCs in IPO market. Using IPO data from 1988 through 1999, we document strong empirical evidence against the certification role of VCs. Instead, we find that VCs are acting in the best interests of the underwriter rather than the issuing firms. As a result, presence of venture capital helps enhance the bargaining power of the underwriters against the issuers in setting the IPO offer price, aggravates the agency problem of underwriters, and thereby leading to a higher level of underpricing.

2 1. Introduction Venture Capitalists fund young, high-growth privately held companies by taking, usually, concentrated equity positions. Venture capitalists not only are large stockholders, but also take active roles in monitoring and assisting business operating and strategic planning of the invested companies. They are believed to play important roles in the process of bringing companies public (Lerner, 1994). As pointed by Levin, Ginsburg, and Rocap (2002), When VC makes its investment, typically VC obtains contractual rights to control the exit strategies, including the timing of a future IPO, selection of underwriters, priority over other shareholders in the public resale of its stock The role of venture capital investment in the creation of public companies has been examined in several empirical studies using data in 1970s and 1980s. These studies suggest a certification role of venture capital investment. According to the information asymmetry model, new issues are underpriced so as to reward the truth-telling informed investors by giving them priority in share allocation (Beatty and Ritter, 1986; Rock, 1986; Benveniste and Spindt, 1989). Due to their reputational capital at stake, venture capitalists can certify that the offering price of the issue reflects all available and relevant insider information truthfully, just like auditors and underwriters. Therefore the presence of venture capital will alleviate information asymmetry problem in new issue market, and lead to lower underpricing. Megginson and Weiss (1991) compare VC backed IPOs with a control sample of non-vc backed IPOs matched by industry and offering size from 1983 through They find that VC backed IPOs have significantly lower underpricing. In addition, the presence of venture capital serves to lower the costs of going public. Barry, Muscarella, Peavy, and Vetsuypens (1990) examine the control and monitoring behavior of VCs in a set of IPOs from 1978 to They find that IPOs under more intensive monitoring activities from VCs have lower underpricing. The evidence of the certification role played by VCs is convincing, however, it is confined in data of 1980s or earlier. In this paper, we are interested in examining the role of venture capital investment in IPOs in 1990s. Do venture capitalists certify new issues in 1990s too? This is an interesting and important issue, because it has been noticed that there are many significant changes in both the activities of venture capital investments and the IPO market in 1990s compared with earlier periods. According to Venture Economics Survey in 2002 (summarized in table 1), the number of companies invested by venture capitals, average 2

3 investment in each company, and the total amount invested in the venture capital industry all grow exponentially during 1990s. Since a major exit strategy in venture capital investment is to bring invested companies public, the increasing activity of venture capital is parallel to the accelerated pace of firms going public. The last two columns in table 1 is an excerption from Ritter (2001). It shows that the number of IPOs and the size of the aggregate IPO market have increased significantly in the 1990s. In addition, it is observed that underwriting syndicates have shrunk in size and lead managers now have more control on the selling activity (Chen and Ritter, 1999). Also observed in the 1990s are closer ties between institutional investors and high prestige investment banking firms (Dunbar, 2000). Parallel with these evidence, the magnitude of underpricing (or initial return, defined as the percent of change of the first day close price relative to the offer price) increases dramatically as well. According to Loughran and Ritter (2002b), average IPO underpricing went up from 7% in 1980s to almost 15% during , and skyrocketed to 65% in Furthermore, some of the established empirical relationships break down or reverse. Beatty and Welch (1996) and Cooney, Singh, Carter, and Dark (2000) find that the negative relation between underwriter prestige and underpricing that existed in the 1980s becomes positive in the 1990s. Loughran and Ritter (2002a) propose an agency model combining with prospect theory to explain underpricing in the IPO market. Underwriters have an incentive to choose a lower offer price not only to minimize distribution risk, but also to obtain indirect compensation. For example, using their discretion in allocating IPOs among investors, underwriters could over charge commissions and share the money left on the table through quid pro quos with rentseeking investors. 1 Under the context of prospect theory, entrepreneurs care about the change in their wealth rather than the level of wealth. Therefore, if there is an unexpected increase in wealth, e.g., when the offer price is above the maximum of the original filing price range, entrepreneurs do not mind leaving money on the table very much, and underwriters take advantage of this. The agency model combining has successfully explained the reversal of the negative relationship between initial return and underwriter ranking, as well as the significant increase in the level of underpricing over time (Loughran and Ritter, 2002b). 1 In a Wall Street Journal article on an SEC probe of mutual fund overpaying on commissions, Lucchetti (1999) states, Other fund executives point out that higher commissions can be justified by the access they can provide to initial public stock offerings. 3

4 Recent disclosures by congressional investigators about allocation of hot IPOs generally have focused on corporate executives, who could return the favor with lucrative investmentbanking assignments. However, a recent Wall Street Journal article by Smith, Grimes, Zuckerman, and Scannell (2002) points out that who else are receiving generous allotments of IPO shares were venture capitalists, who often help determine which investment banks fledging companies select when they decide to go public themselves. Indeed, the practice of allocating hot IPOs to venture capitalists has been even more widespread than those to corporate executives... Critics say such mutual back-scratching was part of a cozy relationship between Wall Street and its investment-banking clients. Given this cozy relationship between venture capitalists and underwriters, would venture capitalists act in the best interests of the issuing firm? Could venture capitalists continue to certify new issues? These are the questions we aim to address in the paper. Using a sample of IPOs during 1988 through 1999, we document empirical evidence in direct contrast to those in Barry, Muscarella, Peavy, and Vetsuypens (1990) and Megginson and Weiss (1991). We observe a significantly higher initial return in VC backed IPOs relative to non- VC backed issues, and this result is not concentrated in later years in or close to the internet bubble period. Even after controlling for industry and offer size like Megginson and Weiss (1991), VC backed IPOs are underpriced 9.6% more than non-vc backed ones. Multivariate analyses suggest that a large part of higher underpricing is due to the fact that VC backed IPOs are younger in age, associated with more prestige underwriters, and more likely to revise its offer price above the maximum of the original filing price range. After controlling for these factors, VC backed IPOs are underpriced 3.5% more than non-vc backed issues. Consistent with Beatty and Welch (1996), we observe a positive relationship between underpricing and underwriter ranking. However, this positive relationship only holds in VC backed IPOs. In non-vc backed issues, underpricing is negatively related to underwriter ranking, thought the coefficient is not statistically significant. Furthermore, the positive relationship between IPO underpricing and the offer price upgrade (or revision) dummy is significantly larger in VC backed issues. These results cannot be explained by information asymmetry model of underpricing and certification role played by venture capital investment. Instead, they suggest that 1990s, venture capitalists may collude with prestige underwriters, and aggravate the agency problem described in Loughran and Ritter (2002a). Furthermore, we examine the relationship 4

5 between underpricing and VC s quality and monitoring ability. If VCs can certify new issues, we would expect that higher VC quality and more intensive monitoring and control would lead to less underpricing. We find little evidence of that. In contrast, we observe that underpricing is positively and significantly related to lead VC age and the percent of board seats owned by VCs, and negatively but insignificantly related to VC ownership before offering and lead VC s asset under management. Why would venture capitalists collude with underwriters, or act in the best interests of underwriters rather than the issuing firm? Venture capitalists are some of the owners of the issuing firm, however, their interests are not perfectly aligned with the original owners (entrepreneurs). Even though the venture capitalists have a significant stake in the issuing company and benefit from high offer price, unlike the entrepreneurs, there are several reasons that they are more willing to underprice their issues. First, venture capitalists invest in many companies, and they may have a special relationship with underwriters in expecting to bring more of their portfolio companies public in the future. Second, a successful offering with large initial return would attract media attention. It will benefit venture capitalists to bring the next company public, and may lead to more capital flows into their managed portfolios. Lee and Wahal (2002) find that capital flows to VCs are indeed higher subsequent to offerings with high IPO initial returns. Finally, some venture capitalists are more willing to accept underpricing because of the successful co-opting of venture capital partners by some prestige underwriters. Back in 1997, Siconolfi (1997) called attention to the spinning practice that hot IPO shares are purposely placed to general partners of VC firms who are in position to direct future issuers to the underwriter. Investment banker Frank Quattrone was alleged to allocate IPO shares to the so called Friends of Frank accounts for those in a position of influence, many of them are VCs. Smith, Grimes, Zuckerman, and Scannell (2002) also pointed out that the practice of allocating hot IPOs to venture capital partners has been even more widespread than that to corporate executives. 2 Therefore, VCs may take a softer stand when negotiating with the underwriters on setting the IPO offer price, which leads to higher underpricing. 2 The article pointed out that Robert C. Kagle, a Benchmark Capital (a venture capital company) partner, along with several family members, received coveted allocations of IPOs of stocks in 12 different companies in seven different personal brokage accounts they had at Goldman Sachs Group Inc Goldman apparently fared better. Benchmark helped select Goldman to lead or jointly lead 15 of 23 IPOS by companies Benchmark invested in Goldman received an estimated $49 million in fees for managing the IPOs 5

6 The remainder of the paper is organized as follows. Section 2 describes our data and methodology. Section 3 presents and interprets our empirical results while section 4 concludes the paper. 2. Data and Method Our IPO sample is from the Securities Data Company (SDC) database, and the sample period spans from January 1988 to December We exclude closed end funds, unit issues, REITs, and those issues with offer price less than $5 per share or total proceeds less than 10 million dollars. 3 To study the long-term performance of the IPOs, they must also have CRSP price data for at least two years after the issuance. This selection criterion results in 1560 IPOs backed by venture capital and 1925 IPOs not backed by venture capital. We corrected the errors in SDC database found by Ritter. 4 SDC has very limited data on the age of the offering firms, defined as the year from the date of incorporation to the offering date. To supplement the age data from SDC database, we try to find age information from various sources. Age of firms going public after May 1996 are collected from online IPO prospectus in EDGAR. For those issued before May 1996, we try to find as many age information as possible from Lexis-Nexis, Dow Jones Interactive, firm10-ks and their amendments, firm s wedpage, and by direct contacting the firm. To assess the participation and ownership of venture capitalists, we examine the Management and Shares Beneficially Owned sections in each IPO prospectus. These sections identify board members and list equity ownership before and after the IPO for all insiders and 5% shareholders. We cross-reference the stockholders listed in the prospectus against the Pratt s Guide to Venture Capital Sources (1994 and 2001 editions) to identify those venture capital investors, and we record their percentage equity ownership and note any board membership. We also collect several characteristics of the lead VC firm from the Pratt s Guide, including lead VC age (years from founding date of the lead VC and offering date of the IPO) and asset under management. Lead VC firm is identified as the one with the most number of board seats, or the highest percent of ownership before offering if the number of board seats held by several VCs is the same. 3 Megginson and Weiss (1991) select IPOs with offer price over $5 per share and total proceeds over $3 million. We use $10 million proceeds as a threshold to take into account the increasing size of IPOs during the 1990s. 4 See Jay Ritter s webpage 6

7 3. Empirical results 3.1. Univariate analysis of IPOs with and without venture capital Table 2 presents descriptive statistics of our samples of VC backed and non-vc backed IPOs. As shown in Panel A, during our sample period from 1988 through 1999, the percentage of VC backed IPOs each year varies from 27.0% to 60.2%, and averages at 44.4%, which is much higher than 28% over the period of documented in Barry, Muscarella, Peavy, and Vetsuypens (1990). It suggests that there is much more IPOs involved with venture capital investment in the 1990s. We also report the mean initial return for these two groups of IPOs. Every year, VC backed IPOs have higher initial return than non-vc backed IPOs, and the difference is statistically significant in six out of twelve years during the sample period. It suggests that VC-backed IPOs are underpriced more, and this result is not clustered around a particular time period. In Panel B, we report that the mean initial return of the entire sample of VC backed IPOs is 29.1%, which is significantly higher than 14.8%, the mean initial return of the entire sample of non-vc backed IPOs. These results appear different from previous evidence documented in the literature. Barry, Muscarella, Peavy, and Vetsuypens (1990) find no significant differences in underpricing for VC backed and non-vc backed issues offered from 1978 through Table 3 reports industry distribution of firms going public based on their two digit SIC codes. The majority of the VC-backed IPOs (76.7%) falls within eight separate industries with a large concentration in the high technology areas. Computer equipment, electrical and electronic components, instrumentation, communication, and business services account for 61.1% of VC backed IPOs. In contrast, these industries only account for 31.3% of non-vc backed IPOs. These summary statistics are similar to those reported in Barry, Muscarella, Peavy, and Vetsuypens (1990) for the period of In addition, VC backed IPOs have a higher average initial return in six out of eight industries, and four of them are statistically significant. Given that venture capital activity and IPO underpricing tends to be clustered by industry, we follow Megginson and Weiss (1991) to construct two groups of VC backed and non-vc backed IPOs by matching with industry and offer size. For each of the 1560 VC backed IPOs in our sample, we match the company as closely as possible by offering amount to non-vc backed companies in the same three-digit SIC code industry. We are able to find a match for 1520 VC 7

8 backed IPOs. Since most of VC-backed IPOs concentrate in a few technology industries, while non-vc backed IPOs are less concentrated, several VC backed IPOs are matched to the same non-vc backed IPO, therefore we have only 636 distinct non-vc backed issues. Table 4 reports the differences in firm and offering characteristics for VC and non-vc backed IPOs. A standard t-test is used to examine the difference in mean, while a Wilcoxon nonparametric test is used to examine the difference in median. Even though we have matched the two samples as closely as possible on the offering amount, VC backed IPOs, on average, have lower offering amount ($43.13 million versus $48.11 million). Nevertheless, these two groups are not significantly different in offer price, last-12-month revenue, and book value of assets. Similar to the evidence documented in Megginson and Weiss (1991), the average proportion of book value of debt as a percentage of total book assets is significantly lower in VC backed IPOs than that in non-vc backed issues (17.6% versus 23.5%). The average initial return of VC backed IPOs is 28.72%, which is significantly higher than the average initial return of the matched sample of non-vc backed IPOs that is 19.11%. The median initial return is also significantly higher in VC backed IPOs than in their non-vc backed counterparts. This result is in contrast to that documented in Megginson and Weiss (1991). Their study shows that after matching with industry and size, the initial return of VC backed IPOs over the period of is significantly lower than that of their non-vc backed counterparts, which supports a certification role of venture capital investment in IPO pricing. Before we argue against the certification role of venture capital during our sample period, we consider the possibility that the difference in initial return may be a manifestation of other factors that affect underpricing. In addition to industry and offer size, previous literature has documented several other variables affecting initial return. Ritter (1984) find a significant negative relationship between the age of the firm going public and the corresponding initial return. Underwriter ranking has been negatively related to initial return in 1980s (Carter and Manaster, 1990), while it is positively related to initial return in 1990s (Beatty and Welch, 1996; Cooney, Singh, Carter and Dark, 2000; Loughran and Ritter, 2002b). Daily return volatility during the first 20 trading days is used to proxy for ex ante uncertainty about the issuing firms value, and Ritter (1987) documents a positive relationship between return volatility and initial return. In addition, a strong positive relationship between the offer price upgrade dummy (equals one if the offer price is above the maximum of the original file price range, and zero otherwise) 8

9 and the initial return has been documented in Hanley (1993) and Loughran and Ritter (2002a, 2002b). This is consistent with the agency model of Loughran and Ritter (2002a). The offer price upgrade dummy is a proxy for the entrepreneurs receipt of good news about their expected wealth. The larger it is, the more money the entrepreneurs are willing to leave on the table. As shown in table 4, VC backed IPOs have many different characteristics from those not backed by VC. VC backed firms are much younger in age than their non-vc backed counterparts (8.98 years versus years). Both the mean and median of firm age are significantly different. VC backed IPOs are underwritten by significantly higher quality underwriters than their matched non-vc backed IPOs. Average daily return volatility during the first 20 days aftermarket is a little higher in VC backed IPOs, though the difference is not statistically significant. In addition, VC backed firms tend to have a significantly higher offer price upgrade dummy than non-vc backed firms, indicating that VC backed IPOs are much more likely to have offer prices above the maximum of the original file price range Multivariate analysis of IPOs with and without venture capital The above results suggest that, in order to identify the role of venture capital on underpricing, we must control for the impacts of other offering characteristics in a multivariate regression setting. Table 5 reports the correlation coefficients among the firm and offering characteristic variables. VC is a dummy variable that equals one for VC backed IPOs and zero for non-vc backed IPOs. The VC dummy variable is positively correlated with underwriter ranking and the offer price upgrade dummy, while it is negatively correlated with offer size and firm age. 5 These results are consistent with those presented in table 4. Interestingly, the correlation coefficient between volatility and offer size is 13%, which is highly significant. It suggests that IPOs with larger offer size tend to have higher return volatility during the first 20 trading days. To assess the relative impact of venture capital backing in IPO pricing during 1988 through 1999, we regress initial return against the VC dummy variable for whether or not the IPO is VC backed, and several control variables that have been identified in explaining initial 5 Most of the firm age variables are missing in the SDC database. We have manually collected firms incorporation date from the IPO prospectus (after May 1996) and from 10-K annual reports and various other sources (prior to May 1996). However, we are unable to identify this information for many firms, therefore, we have only 1943 observations for the firm age variable. 9

10 returns. The results are shown in table 6. Without controlling for other variables in regression 1, VC backed IPOs are underpriced 9.6% more than non-vc backed IPOs, and the coefficient on the VC dummy is positive and highly significant. In regression 2, we include the same control variables (offer size, IPO firm age, and underwriter rank) as Megginson and Weiss (1991). The coefficients on offer size and underwriter rank dummy are strongly positive, while the coefficient on firm age is significantly negative. These results are consistent with the findings in Loughran and Ritter (2002b). They observe that, in the 1990s, larger offers have been underpriced more than smaller offers, and IPOs with higher quality underwriters have been underpriced more than those with lower quality underwriters. After controlling for these variables, VC backed IPOs are underpriced 6.27% more than non-vc backed IPOs, and the coefficient is statistically significant. It suggests that the higher level of initial returns in VC backed IPOs are partly due to the fact that they are much younger companies, and associated with higher rank underwriters. Nevertheless, this result is in contrast to the evidence in Megginson and Weiss (1991) that covers sample period from 1983 through They find that, after controlling for the same variables, VC backed IPOs have significantly lower initial returns than do their matched non-vc backed counterparts. One possibility for the discrepancy between our results and those in Megginson and Weiss (1991) is that, compared to 1980s, the underpricing mechanism may be different in 1990s. It has been shown that in 1990s, underwriting syndicates have shrunk in size and lead managers now have more control on the selling activity (Chen and Ritter, 1999). Also observed in the 1990s are closer ties between institutional investors and high prestige investment banking firms (Dunbar, 2000). These changes of market activity and corporate practice offer the right environment for incubating agency problem between issuing firms and underwriters. As pointed out by Loughran and Ritter (2002b), this agency problem that were largely latent in the 1980s increased in importance in the 1990s. One piece of evidence is that the relationship between underwriter rank and initial return was negative in 1980s, but it reversed to positive in 1990s. Therefore, VC backed IPOs may be underpriced more in our sample because of other offering characteristics that are not controlled in the above analysis. So we include two more variables in the regression, daily return volatility of the first 20 trading days and the offer price upgrade dummy. Consistent with previous literature, the coefficient on offer price upgrade dummy is positive and highly significant. Furthermore, its inclusion drastically boosts the R 2 of the 10

11 regression. In contrast, the coefficient on return volatility is not significant. After including these two additional variables in regression 3, the coefficient on the VC dummy variables is reduced to 3.5%, and becomes statistically insignificant. So far the results suggest that the group of IPOs backed by VC in 1990s is significantly underpriced more than its matched group of IPOs not backed by VC, and the higher level of underpricing is partly explained by the difference of offering characteristics between these two groups. Nevertheless, we find little evidence that the presence of venture capital can certify the new issue and lead to less underpricing. In order to account for different impacts of these variables on IPO underpricing between VC backed IPOs and non-vc backed ones, we add slope dummy variables by interacting the five control variables with the VC dummy. By doing so, we are able to examine the potential interaction between venture capitalists and other factors. When the slope dummy of logsize is included in regression 4, we observe several interesting findings. The coefficient on the VC intercept dummy switches from positive to significantly negative, the coefficient on logsize becomes negative (though not statistically significant), and the logsize slope dummy is positive and highly significant. These results suggest that the impact of offer size on initial return is significantly different between the two groups of IPOs. In VC backed IPOs, larger IPOs are underpriced more than smaller IPOs. However, in non-vc backed IPO sample, this relation is reversed, larger IPOs tend to be underpriced less. Offer size is often used to proxy for information asymmetry. Larger issues would have a lower degree of information asymmetry, thereby lower underpricing. On the other hand, large issues are harder to place. Underwriters have a strong incentive to underprice the large issues more to minimize their issuing risks. Therefore the sign of coefficient on logsize will dependent on the two countervailing driving forces of information asymmetry and underwriter risk minimization. For non-vc backed IPOs, information asymmetry seems more dominant, thereby leading to a negative coefficient on logsize. In contrast, for VC backed IPOs, underwriter risk minimization appears to be more significant, leading to a positive relationship. Like underwriters, venture capitalists are more sensitive to issuing risk as well. This is because they invest in many companies, and eventually expect all of their invested companies to go public. A failure on the current issue may jeopardize their chance of bringing more companies public in the future, and bringing in more capitals in their investment portfolios. Therefore issuing firms with VC backing are more inclined to 11

12 minimizing issuing risk, especially with large issues. It suggests that during the 1990s, VCs have been pursuing interests more in line with the underwriters, rather than playing the certification role in reducing information asymmetry. This result is consistent with the cozy relationship (or potential collusion) between venture capitalists and underwriters observed in corporate practice. More direct evidence of aligned interests between venture capitalists and underwriters is found in regression 5. When we add slope dummy of underwrite rank, the coefficient on VC dummy becomes negative too, though it is not statistically significant. The slope dummy of underwrite rank is significantly positive, suggesting that the positive impact of underwriter ranking on initial return is significantly different in VC backed IPOs than in non-vc backed IPOs. Specifically, the coefficient on underwriter rank is negative (though not statistically significant) in non-vc backed IPOs, while it is significantly positive in VC backed IPOs. 6 These results indicate that the IPOs that are VC backed and underwritten by top bankers are underpriced the most. The evidence is not consistent with a certification role played by VCs. If VCs can certify new issues by reducing information asymmetry, we would expect that the slope dummy of underwriter rank is negative. Put in another words, while higher rank underwriters lead to higher level of underpricing due to their agency problem, presence of venture capital will reduce underpricing since VC investments certify the quality of the issue. This is not what we observe. On the contrary, our evidence points to a possible collusion between VC s and prestige underwriters. As a result, the presence of venture capital seems aggravate the agency problem of underwriters. A close relationship between VCs and many large investment banks in Wall Street has been reported (Siconolfi, 1997; Smith, Grimes, Zuckerman, and Scannell, 2002). Many venture capital partners received large amount of hot IPO shares from large investment banks, and they have incentives to return the favors by not only selecting those banks to underwrite their portfolio companies going public, but also taking a softer standard in negotiating offer price. According to Levin, Ginsburg, and Rocap (2002), upon their decision of initial investment, venture capitalists obtain contractual rights in controlling the exit strategies, including the timing of an IPO, selection of underwriters, and other. It indicates that VCs could significantly increase the bargain power of underwriter in negotiating offer price with the entrepreneurs, if their interests are in line with underwriters, and thereby leading to higher level of underpricing. 6 Cooney, Singh, Carter, and Dark (2000) show that in 1990s, the relationship between initial return and underwriter rank is positive for IPOs priced above file range, but negative for those offerings priced with file range. For IPOs priced below file range, there is no significant relationship between initial return and underwriter rank. 12

13 Another interesting point, in regression 7, is that when we add slope dummy of volatility, the coefficient on the VC dummy becomes significantly negative. In addition, the slope dummy of volatility is positive and significant, suggesting that the positive impact of volatility on initial return appears much stronger for VC backed IPOs. The volatility slope dummy indicates that VC presence greatly increases, rather than alleviates, the sensitivity of initial return to pricing uncertainty. Given the same degree of valuation uncertainty increase, VC backing induces much larger increase in underpricing than non-vc backing. It suggests that asymmetric information in VC backed IPOs is taken more seriously. This evidence seems inconsistent with the certification role of VC, however it is consistent with collusion between VCs and underwriters under the agency model. In regression 8, the slope dummy of offer price upgrade dummy is significantly positive. It indicates that the coefficient on the offer price upgrade dummy is larger for VC backed IPOs than for non-vc backed IPOs. This result suggests that VC backed firms are willing to leave more money on the table when they receive good news than non-vc backed firms. Again the venture capitalists, who act in the interests of underwriters, make the entrepreneurs willing to accept more underpricing. 7 In summary, we find little evidence of a certification role of venture capital investment. On the contrary, VC backed issues have a more severe agency problem between the issuer and the underwriter, which makes the issuer willing to leave more money on the table. Rather than acting in the best interests of the issuer, VCs seem to collude with underwriters to seek their own personal rents in IPO markets Robustness tests One concern of the above results is whether they are driven by our industry matching scheme. Most VC backed IPOs are concentrated on a few technology industries, while non-vc backed IPOs are more diverse. Thus several VC-backed issues are matched to the same non-vc backed issues since we require them to be in the same three-digit SIC industry. To examine whether our results are sensitive to the sample we choose, we run the same regression using our entire non-matched sample of VC backed and non-vc backed IPOs, and the results are presented 7 Since these independent variables are correlated with each other, when we include all the five slope dummies regression 9, slope dummies of underwriter ranking and offer price upgrade become insignificant. 13

14 in table 7. As shown in regression 1, the magnitude and significance of all the coefficients with all IPOs are similar to those with matched sample of IPOs in regression 2. Our entire sample consists of 3459 IPOs, however, due to lack of firm age variable in many observations, our sample is reduced to Thus the " firm age" variable significantly limits the number of observations included in the regression. To circumvent this problem, in regression 3, we use the same sample as in regression 1 but drop the firm age variable, and find that it has little impact on the coefficients of the other variables. In regression 4 without the firm age variable, we are able to include the entire sample of 3459 IPOs. The results are qualitatively similar to those in regressions with matched sample. In general our results are quite robust regardless which subsample is used Long-term performance of IPOs with and without venture capital Our argument about collusion between VCs and underwriters are based on the assumption that initial return of IPOs is resulted from deliberate underpricing. However, one can argue that the high initial return for VC backed IPOs could result from investors over-optimism about venture capitalists ability to deliver high return rather than deliberate underpricing. In order to test this alternative explanation, we analyze the long-term performance of the two groups of IPOs with and without venture capital investment. If the alternative explanation is true, we expect that VC backed IPOs would under-perform non-vc backed issues. For each IPO in our matched sample, we calculate buy-and-hold return cumulating for one-year, two-year, and three-year holding period after issuance. Excess return is then computed by subtracting it with the NASDAQ cumulative return during the same period. 8 As shown in table 8, IPOs generally under-perform the market index in the long run. However, for all three holding periods, excess returns of VC backed IPOs are significantly higher than that of non-vc backed issues. 9 This result is consistent with Brav and Gompers (1997). It suggests that the higher initial return of VC backed IPOs is unlikely to be attributed to investors irrational overoptimism. 8 We also used CRSP equally-weighted and value-weighted indexes as benchmarks, and the results are similar to that using NASDAQ index. 9 We also examine the long-term performance for the entire unmatched sample of VC backed and non-vc backed IPOs, and the result is similar to what is reported in table 8. 14

15 3.5. Underpricing and the quality of venture capitalists To further dissect the impacts of venture capital investment on IPO underpricing, we analyze various characteristics of VC holdings prior to and after the offerings. We manually collect information on VC holdings from IPO prospectus. Since IPO prospectus is not electronically available prior to May 1996, our sample only covers time period of May 1996 through December 1999, and there are total 519 VC backed IPOs. As a hedge against risk, the majority of VC backed firms have a venture capitalist syndicate with more than one venture capitalist as a shareholder of the firm prior to going public. As shown in table 9, the average holdings by all venture capitalists prior to going public is 32.01%, and it drops to 24.09% after IPOs. Twenty two percent of the offers (or 113 firms) have venture capitalists owning 50% or more of the firm equity prior to going public, and it drops to 7.5% after IPOs. Additional statistics suggest that the decrease of VC holdings after going public is mainly resulted from a dilution of the offerings rather than VCs selling activities. After going public, there are only 31.21% of VC backed IPOs having venture capitalists selling any of their shares in the offering, and the average percent of holdings sold by VCs is only 0.04%. These summary statistics are similar to those documented in Barry, Muscarella, Peavy, and Vetsuypens (1990). While a large number of venture capitalists give up voting control of the firm, the majority retains a significant portion of their holdings in the issuing firm. In addition, average percent of board seats held by VCs is about 38%, suggesting venture capitalists have a significant control of the issuing firms. As shown above, we find little evidence supporting the certification role of venture capital investment. Further investigation of various VC characteristics may help clarify the role of VCs in IPO underpricing. Using IPOs from 1978 to 1987, Barry, Muscarella, Peavy, and Vetsuypens (1990) show that the initial return of VC-backed IPOs are inversely related to many proxy variables for VC quality and control/monitoring abilities, such as fraction of issuer s shares held by VC before the IPO, the age and the total capital under management of the leading VC. Lerner (1995) finds that venture capitalist representation on the boards of the private firms in their portfolios is greater when the need for oversight is larger. To clarify the role of venture capitalists in our study, we examine the relationship between underpricing and several variables, including VC ownership before the IPO (percent of shares owned by all VCs before the IPO), fraction of board seats owned by VC, lead VC age, and lead VC asset (asset under management by the lead VC). These variables are regressed 15

16 against initial return after controlling for other five variables in the above tables, and results are presented in table 10. When each of these quality and monitoring variables is regressed against initial return, none of the variables is significant. In regression 5 where we include all four VC quality variables, the coefficients on VC ownership before offering and VC asset are negative but not significant. In contrast, the coefficients on percent of board seats owned by VC and VC age are positive and highly significant. These results offer little evidence of the certification role of VC in IPOs. On the contrary, it suggests that IPOs under more control by VCs or controlled by better-established VC firms tend to underprice more. This is consistent with our proposition that VCs collude with underwriters, and aggravate agency problem. The greater the control of the firm by VCs, the more likely VCs can collude with underwriter and convince the entrepreneurs to leave more money on the table. 4. Summary and conclusion Discretional allocation of hot IPOs to corporate executives in exchange for future investment banking assignments has been subjected to congressional investigation. However, it was not just corporate executives, venture capitalists have received hot IPO shares too, and the practice of allocating hot IPOs to venture capitalists has been even more widespread than those to corporate executives (Smith, Grimes, Zuckerman, and Scannell, 2002). In return, venture capitalists would select these investment banks as the underwriters of their portfolio companies going public. Such mutual back-scratching relationship between venture capitalist and investment banks is subjected to criticism from the public. In this paper, we are interested in studying the role of venture capital investment in IPO underpricing. Empirical studies using data in 1980s and earlier have documented that VC presence can certify new issues by reducing information asymmetry, thereby leading to less underpricing (Barry, Muscarella, Peavy, and Vetsuypens, 1990; Megginson and Weiss, 1991). Given that there are significant changes in the IPO markets and venture capital industry in 1990s, and the emerging of a cozy relationship between venture capitalists and investment banks, we aim to address a few interesting questions. Can venture capitalist certify new issues in 1990s? Do venture capitalists act in the best interest of issuing firms? Is there any interaction between VCs and underwriters? How does it impact IPO underpricing? 16

17 Using IPO data from 1988 through 1999, we document strong empirical evidence against the certification role of VCs during this period. Instead, we find that VCs are acting in the best interest of underwriters rather than issuing firms. As a result, presence of venture capital helps enhance the bargaining power of the underwriters against the entrepreneurs in setting the IPO offer price, and thereby leading to a higher level of underpricing. The absence of a certification role of VCs in 1990s is analogous to lack of certification role of underwriters in the same time period. Underwriter s reputation ranking has been used as a proxy for their certifying ability in the literature, which is shown to be negatively related to IPO underpricing in periods before 1990s. However, this relationship is reversed in 1990s where, under the agency model, the underwriter reputation ranking is more likely to proxy underwriter s market power to extract benefits from the issuer. Likewise, VC characteristics such as the number of board seats held by VCs and the age or size of capital of lead VCs are used as proxies for VC quality and control/monitoring abilities. We find that underpricing is positively related to lead VC age and percent of board seats held by VCs. The greater the control of the firm by VC, the more likely VC can collude with underwriter and convince the entrepreneurs to leave more money on the table. 17

18 Reference Barry, C., Muscarella, C., Peavy III, J., Vetsuypens, M., The Role of Venture Capital in the Creation of Public Companies. Journal of Financial Economics 27, Beatty, R., Ritter, J., Investment Banking, Reputation, and the Underpricing of Initial Public Offerings. Journal of Financial Economics 15, Beatty, R., Welch I., Issue Expenses and Legal Liability in Initial Public Offerings. Journal of Law and Economics 39, Benveniste, L., Spindt, P., How Investment Bankers Determine the Offer Price and Allocation of New Issues? Journal of Financial Economics 24, Brav, A., Gompers P., Myth or Reality? The Long-Run Underperformance of Initial Public Offerings: Evidence from Venture and Nonventure Capital-Backed Companies. Journal of Finance 52, Carter, R., Manaster S., Initial Public Offerings and Underwriter Reputation. Journal of Finance 45, Carter, R., Dark F., Singh A., Underwriter Reputation, Initial Returns, and the Long-term Performance of IPO Stocks. Journal of Finance 53, Chen, H., Ritter, J., The Seven Percent Solution. Journal of Finance 55, Cooney, J., Singh, A., Carter, R., Dark, F The IPO Partial-Adjustment Phenomenon and Underwriter Reputation, Working Paper. Dunbar, C., Factors Affecting Investment Bank Initial Public Offering Market Share. Journal of Financial Economics, 55, Hanley, K., The Underpricing of Initial Public Offerings and the Partial Adjustment Phenomenon. Journal of Finance 34, Lee, P., Wahal S., Venture Capital, Certification and IPOs. Working paper. Lerner, J., Venture Capitalists and the Decision to Go Public. Journal of Financial Economics 35, Lerner, J., Venture Capitalists and the Oversight of Private Firms. Journal of Finance 50, Levin, J. S., Ginsburg, M. D., Rocap, D. E., Structuring Venture Capital, Private Equity and Entrepreneurial Transactions, Aspen Publishers, Inc. 18

19 Loughran, T., Ritter J., 2002a. Why Don t Issuers Get Upset About Leaving Money on the Table in IPOs? Review of Financial Studies 15, Loughran, T., Ritter J., 2002b. Why Has IPO Underpricing Increased Over Time? Working Paper. Lucchetti, A., SEC Probes Rates Funds Firms Pay for Commissions. Wall Street Journal, Sept. 16, C1. Megginson, W., Weiss K., Venture Capitalist Certification in Initial Public Offerings. Journal of Finance 46, Ritter, J., The Hot Issue Market of Journal of Business 57, Ritter, J., The Long-run Performance of Initial Public Offerings. Journal of Finance 46, Ritter, J., The Cost of Going Public. Journal of Financial Economics 19, Ritter, J., Summary Statistics on Initial Public Offerings with An Offer Price of $5.00 or More. University of Florida. Rock, K., Why New Issues Are Underpriced, Journal of Financial Economics, 15, Siconolfi, M., Underwriters Set aside IPO Stock for Officials of Potential Customers. Wall Street Journal, Nov. 12, A1. Smith, R., Grimes, A., Zuckerman, G., Scannell, K., Something Ventured and Something Gained? Wall Street Journal, Oct.17, C1. 19

20 Table 1. Venture capital investments and IPO market activity from 1990 to 2000 ($ Million). Venture capital investment in the first three columns are from: PricewaterhouseCoopers/VentureEconomics/National Venture Capital Association MoneyTree Survey, information current as of 2/20/2002. IPO market activity in the last two columns is from Ritter (2001). The number of IPOs excludes those with an offer price of less than $5.00, ADRs, best efforts offers, unit offers, Regulation A offerings, real estate investment trust (REITs), partnerships, and closed-end funds. Gross proceeds data are from SDC, and exclude overallotment options but include the international tranches, if any. From National Venture Capital Association MoneyTree Survey From Ritter (2001) Year Number of Average investment in Total amount companies invested each company ($ Million) invested ($ Million) Number of IPOs Gross proceeds ($ Millions) $2.55 $2, , $2.41 $2, , $3.63 $3, , $4.83 $4, , $3.98 $3, , $4.50 $5, , $6.29 $11, , $6.22 $14, , $7.03 $19, , $12.97 $54, , $18.24 $102, ,100 20

21 Table 2. Summary statistics of our sample of VC backed and non-vc backed IPOs over the period Panel A reports the number of VC backed and non-vc backed IPOs in our sample each year, and their mean initial returns. Our sample is from the Securities Data Company (SDC) database from January 1988 to December We exclude closed end funds, unit issues, REITs, and those issues with offer price less than 5 dollars per share or total proceeds less than 10 million dollars. Initial return is defined as percentage price change from offer price to the first-day closing price. Standard t-test is used to examine the difference of mean initial return between VC backed and non-vc backed IPOs. Panel B reports the summary statistics of mean initial return of the entire samples of VC backed and non-vc backed IPOs. * and ** denote significance at the 5% and 1% level. VC backed IPOs Year NOBS Mean initial return Non-VC backed IPOs NOBS Mean initial return T-stat (test difference in mean initial return) VC-backed IPOs as percentage of total IPOs Panel A. Analysis based on sample each year % % 2.45* 43.2% % % 2.08* 41.7% % % % % % % % % % % % % % % 3.68** 38.1% % % 4.44** 47.9% % % % % % 2.23* 30.1% % % % % % 5.76** 60.2% Average 44.4% Panel B. Analysis based on the entire sample % % 9.69** 21

22 Table 3. Industry distribution of VC backed and non-vc backed issues from 1988 through VC backed IPOs Non-VC backed IPOs Percentage of all non- VC backed T-stat of difference in mean initial return SIC code Industry description NOBS Percentage of all VC backed IPOs Mean initial return NOBS IPOs Mean initial return 28 Chemicals And Allied Products % % Industrial And Commercial Machinery And Computer Equipment % % Electronic And Other Electrical Equipment And Components, Except Computer Equipment % % Measuring, Analyzing, And Controlling Instruments; Photographic, Medical And Optical Goods; Watches And Clocks % % Communications % % Miscellaneous Retail % % Business Services % % Health Services % % All others 23.3% 59.3% 22