Sinners or Saints? Top Underwriters, Venture Capitalists, and IPO Underpricing

Transcription

1 Sinners or Saints? Top Underwriters, Venture Capitalists, and IPO Underpricing Kose John Anzhela Knyazeva Diana Knyazeva Preliminary: Do not cite or quote This version: September 6, 2018 Abstract This paper examines strategic interactions between venture capitalists (VCs) and top underwriters in the IPO process. We test two contrasting hypotheses: certification and rent extraction. On the one hand, the joint involvement of VCs and top underwriters can amplify their certification effect, reducing IPO underpricing. On the other hand, the rent extraction incentives of VCs and top underwriters, particularly in a repeated dealing context, can increase underpricing. We test these contrasting effects on IPO underpricing in the cross-section and around two regulatory shocks. First, VC-backed IPOs, VC-backed deals with top underwriters, and deals with stronger VC-underwriter ties exhibit greater underpricing. The cross-sectional results continue to hold in two-stage least squares estimation with geographical instruments. Second, the 2012 JOBS Act decreased disclosures by issuers and increased informational asymmetry. We find that IPO underpricing increased among VC-backed IPOs involving top underwriters after the JOBS Act. Third, a 2003 NASD rule limited explicit rents that may be extracted from preferential allocations of new IPO issues. We find that VC-backed IPO underpricing declined after the shock, and the effect was concentrated among VC-backed deals involving top underwriters and VCs with ties to underwriters. Importantly, we find no effect of top underwriters alone unless a VC is also involved. The observed decline in IPO underpricing cannot be explained by analyst involvement or the passage of Regulation FD. Moreover, repeat dealing between large institutions and underwriters similarly increases underpricing, but the effect does not explain the role of VC-underwriter ties and does not decline after the 2003 shock. Overall, our results support rent extraction, shed new light on the effects of VC backing, underwriters, and institutions on IPO underpricing, and emphasize the role of strategic interactions between them in the post-dot-com period. Keywords: IPOs, venture capital, underwriters, institutional buyers, repeat dealing, conflicts of interest, JOBS Act, NASD Rule 2790 JEL classification: G30, G32, G38 Kose John, New York University, Stern School of Business, kjohn@stern.nyu.edu; Anzhela Knyazeva, SEC, knyazevaa@sec.gov; Diana Knyazeva, SEC, knyazevad@sec.gov. The paper has benefited from helpful comments and suggestions of Michelle Lowry and SEC workshop participants. All errors and omissions are our own. The Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This article expresses the authors views and does not necessarily reflect those of the Commission, the Commissioners, or other members of the staff. This paper is part of the Division of Economic and Risk Analysis Working Paper Series. Papers in this series are the work of the authors and not the work of the Division or the Commission. Inclusion of a paper in this series does not indicate a Division or Commission determination to take any particular action or position. References to this paper should indicate that the paper is a DERA Working Paper.

2 1. Introduction Initial public offerings (IPOs) have long been an important mechanism for raising external financing, and IPO underpricing is an essential component of an issuer s cost of capital raised in the offering. Recently, trends in the IPO market have attracted significant attention from academics, practitioners, and regulators. For example, several papers document significant shifts in IPO activity after the 2012 passage of the JOBS Act. 1 The broader IPO underpricing literature has explored the role of top underwriters and the role of venture capitalists (VCs) in IPO underpricing. 2 However, there is a relative dearth of evidence on how these crucial market participants interact in the IPO process. This paper fills the gap and tests the effects of strategic interactions between VCs, top underwriters, and institutional investors on IPO underpricing. We focus on the post-dot-com period and test the effects in the cross-section and around two quasi-natural experiments. First, the JOBS Act reduced burdens on IPO firms by scaling back process and disclosure requirements for the vast majority of IPOs. This increased the amount of information available about the typical issuer, magnifying information asymmetries about issuer firms. Second, a 2003 NASD rule restricted preferential allocations of IPO shares. This limited private benefits that may be extracted in the IPO process through underwriter s preferential IPO allocations to VC fund managers and institutional investors, among others. We formulate two competing hypotheses about the roles of VCs and underwriters in the context of IPO pricing. First, the certification hypothesis suggests that underwriters and VCs, 1 For example, Dambra, Field, and Gustafson (2015); Dambra, Field, Gustafson, and Pisciotta (2018); Chaplinsky, Hanley, and Moon (2017); Barth, Landsman, and Taylor (2017). 2 For example, Loughran and Ritter (2004); Lowry and Schwert (2004); Lowry and Murphy (2007). Early studies find that VCs decrease underpricing by reducing uncertainty about issuers through certification and monitoring, whereas studies of IPOs during the nineties find that VCs increased underpricing (e.g., Gompers (1996); Lee and Wahal (2004); Loughran and Ritter (2004); Liu and Ritter (2011); Cliff and Denis (2004)). 1

3 through their combined reputation with market participants, credibly signal their private information about issuer quality and reduce investor information asymmetry. Empirically, the certification hypothesis would predict that IPO underpricing decreases with top underwriter and VC involvement, and the effect is stronger when both are involved in a deal. In addition, under the certification hypothesis, repeat dealing between VCs and top underwriters should reinforce incentives to carefully screen issuers and increase the strength of the signal to the market from their involvement, reducing underpricing. Moreover, when information asymmetry increases, such as when issuers are required to disclose less information publicly, VCs and top underwriters should play a greater certification role. In 2012, the JOBS Act allowed most IPO issuers to adopt scaled disclosure, which has been linked to increased information asymmetry and underpricing (e.g., Chaplinsky et al. (2017); Barth et al. (2017); Gullapalli and Knyazeva (2017)). Under the certification hypothesis, the presence of VCs and top underwriters should result in a larger reduction in underpricing after the 2012 shock. In turn, when information asymmetry decreases, the certification effect should become smaller. Second, the rent extraction hypothesis suggests that underwriters and VCs have strategic incentives to underprice the IPO at a cost to the issuer (borne by the pre-ipo investors of the issuer). For example, underwriters can benefit from IPO underpricing directly if they hold underwritten securities. In addition, underpriced deals are more likely to experience a jump in price after the IPO, which is favored by IPO investors. As a result, underwriters enjoy indirect reputational benefits of underpricing, making it easier to build the book for future IPOs. Moreover, underpricing decreases the odds of the price declining below the offer price, reducing litigation risk. Similarly, VCs may benefit from IPO underpricing. For a portion of our sample period, prior to the 2003 shock, managers of VC funds were able to receive preferential 2

4 allocations from top underwriters in other hot underpriced IPOs. During the entire period, VCs receive reputational benefits from having backed a successful IPO, making it easier to raise future VC funds (e.g., Lee and Wahal (2004); Gompers (1996)). In addition, top underwriters may be more willing to facilitate future IPO exits of this VC from other portfolio firms if previous deals were underpriced and, therefore, benefited underwriters. The rent extraction hypothesis would generally predict that the individual involvement of VCs and top underwriters results in greater underpricing, and the effect is stronger when VCs and top underwriters are jointly involved. VCs can exert significant influence on the decisions to go public as well as the choice of underwriters and deal terms with respect to other portfolio companies. In addition, when VCs transact with the same underwriters in the course of multiple deals, repeat dealing incentives are stronger and underpricing can increase the willingness of a top underwriter to facilitate future IPO exits by the VC. As a result, the hypothesis would predict that the joint presence of VCs and top underwriters and VCs with ties to underwriters would lead to greater underpricing. Further, when the information asymmetry about issuers increases, the potential for such rent extraction through underpricing can increase as well. When there is more uncertainty overall, VCs and underwriters could be in a better position to use their informational advantage to coordinate on greater underpricing. Thus, this hypothesis would predict that the joint presence of VCs and underwriters should increase underpricing after the 2012 enactment of the JOBS Act, which allowed most IPO issuers to provide less disclosure, increasing the information asymmetry. Finally, a 2003 regulatory shock allows us to further test the rent extraction hypothesis. In 2003 NASD issued a rule restricting preferential allocations of new issues by underwriters to money managers. Prior to the rule, underwriters could preferentially allocate shares in hot 3

5 IPOs, which gave rise to explicit conflicts of interest with respect to underwriting. For example, VC fund managers could agree to underprice the IPOs of firms they control and steer future investment banking business to connected underwriters in exchange for preferential allocations in other underpriced IPOs (see, e.g., Loughran and Ritter (2004) examining cases of preferential allocations in the late nineties). The rent extraction hypothesis would predict that the 2003 shock, which restricted such practices, reduced incentives for strategic underpricing in VC-backed IPOs involving top underwriters. Our main findings are threefold. First, VC-backed IPOs, on average, remained more underpriced in the two decades after the dot-com bubble. The underpricing is most pronounced for VC-backed IPOs involving top underwriters and VC-backed IPOs involving VCs with close repeat dealing ties to underwriters. Second, the 2012 shock, which increased the informational advantage of VCs and underwriters, led to an increase in underpricing of VC-backed IPOs involving top underwriters and VC-underwriter ties. Third, the 2003 NASD rule, which limited preferential IPO allocations, decreased the underpricing of VC-backed IPOs, and particularly, VC-backed IPOs involving top underwriters and VCs connected to underwriters. Overall, our results suggest that the rent extraction effect dominates the certification effect. These findings highlight strategic incentives of VCs connected to underwriters with respect to IPO pricing in general, and in conjunction with regulatory shocks that impacted their rent extraction ability related to IPO pricing. Our results shed new light on the implications of VC backing for IPO pricing and information flows of IPO issuers and uncover an important dimension of implicit incentives stemming from ties between VCs and top underwriters. We also examine other dimensions of incentive conflicts that could affect underpricing. First, underwriter-specific conflicts alone do not appear to explain our findings. While 4

6 underwriters may have their own incentives to underprice, we observe a significant interaction effect of VCs and top underwriters and a significant effect of VC-underwriter repeat dealing. Second, we consider whether ties between top underwriters and VCs are merely capturing parallel ties between top underwriters and large institutions, whereby underwriters underprice deals to build a relationship with key institutions that would buy stakes in this underwriter s future IPOs. We find evidence of higher underpricing when underwriters engage in repeat dealing with large institutions in the context of IPOs (underwriters underprice IPOs and institutions invest in IPOs backed by the underwriter). However, we do not find that such implicit incentives to underprice changed as a result of the 2003 or 2012 shocks and thus do not appear to be behind the effects of VC/underwriter ties around the shocks. We also evaluate the possibility that VCs preferences for analyst coverage, which underwriters might facilitate, can explain greater underpricing in VC-top underwriter deals. We check if the post-2003 decline in the underpricing of VC-top underwriter IPOs was due to the analyst effect (e.g., due to reduced ability to secure favorable coverage from the underwriter after the Global Settlement). We also examine the underpricing of VC-top underwriter IPOs around the 2000 adoption of Regulation FD. Overall, while important for IPO bargaining dynamics in earlier periods, the analyst channel does not appear to account for our main results on VCunderwriter strategic interaction during our sample period (2000s and 2010s). The remainder of the paper is organized as follows. Section 2 describes the institutional background and related literature. Section 3 describes data and variables. Section 4 presents the main results and robustness checks. Section 5 concludes. 5

7 2. Background and related work Institutional background: shocks Our empirical design utilizes two regulatory shocks. First, the JOBS Act enacted in April 2012 allowed companies that qualify for emerging growth company (EGC) status to provide reduced disclosure during an IPO. Close to 90% of IPO issuers meet this definition. Prior studies showed that underpricing increased after 2012 due to increased information asymmetry (e.g., Chaplinsky et al. (2017), Barth et al. (2017)). In addition, about a decade prior to the enactment of the JOBS Act, media and academic studies examined the possible presence of underwriter conflicts of interest in IPOs, such as spinning and tie-in arrangements. 3 At issue was the practice of underwriters allocating underpriced IPO shares to reward preferred investors, VC managers and other money managers, and corporate insiders motivated by quid pro quo efforts to secure future business. Investors receiving hot issue allocations are expected to use the bank for future trading business while VCs and insiders receiving hot allocations in other IPOs are expected to steer investment banking business to the bank. While this practice benefited underwriters and in a repeated game setting, the preferred investors, VCs, and insiders it disadvantaged other investors. From the standpoint of the issuer, the higher underpricing raised the cost of the IPO to the issuer. Thus, our second regulatory shock is a 2003 change that curtailed such IPO practices. On October 24, 2003, the National Association of Securities Dealers (NASD) adopted Rule 2790, Restrictions on the Purchase and Sale of Initial Equity Public Offerings, to strengthen protections for investors in IPOs by ensuring that: 3 For example, Weinberg, Ari, 2002, A spin down Wall Street's ladders, Forbes, November 6, 2002, Maynard (2002); Griffith (2003); Hurt (2005); Fanto (2008); Barondes (2005); Levy (2004); Townsend (2004). 6

8 (i) underwriters make a bona fide public offering at the offering price; (ii) do not withhold securities for their own benefit, or use allocations of IPO securities to reward persons that can direct future business to the investment bank (including money managers, such as venture capitalists and hedge fund managers, who are in a position to direct business to a broker-dealer); and (iii) ensuring that industry insiders, including underwriters and their associated persons, do not take advantage of their position to purchase new issues. 4 Rule 2790 codified elements of the prior interpretative guidance on hot IPOs (ones that trade at a premium when secondary trading begins), as well as expanded its scope to all new issuers and modified certain restrictions (for example, eliminating the ability to make a factsand-circumstances decision to allocate IPO shares to restricted persons in some cases). 5 Compliance was required by March 23, Separately from this rule change, other events affecting analysts and governance occurred in (see Appendix), including Regulation Fair Disclosure (FD), Regulation Analyst Certification (AC), the Global Settlement, and Sarbanes Oxley Act. In supplemental tests, we distinguish our findings from those effects by varying sample criteria and shock timing. Related work This paper builds on three branches of prior literature. First, most recently, several studies document an increase in information uncertainty and IPO underpricing after the 2012 JOBS Act shock (e.g., Chaplinsky et al. (2017), Barth et al. (2017)). Our tests of the 2012 shock build on 4 NYSE/NASD IPO Advisory Committee, Report and Recommendations, May 2003, pp , 5 Id. In addition, a Voluntary Initiative announced in April 2003 and effective in October 2003 involved major underwriters limiting hot IPO allocations to executives and directors. In robustness tests, we test whether dating the shock to April 2003 affects our results. 7

9 this work, using the shock as a quasi-natural experiment. Specifically, the setting allows us to examine the role of strategic interactions between VCs and top underwriters when information asymmetry becomes severe. Second, this paper relates to a large literature on IPOs and IPO underpricing (see, e.g., Ljungqvist (2007) and Lowry, Michaely, and Volkova (2017) for a survey). Some studies find a negative effect of a prestigious underwriter on underpricing while other studies find an insignificant or positive effect (e.g., Carter and Manaster (1990); Carter, Dark, and Singh (1998); Loughran and Ritter (2004); Lowry and Schwert (2004); Liu and Ritter (2010, 2011); Lowry and Murphy (2007); Lowry, Officer, and Schwert (2010); Loughran and McDonald (2013)). Evidence on the effect of VCs on underpricing has also been mixed and sensitive to the time period, with some studies finding a negative effect (consistent with certification), and other studies finding a positive or insignificant effect (consistent with rent extraction), particularly in late nineties and early 2000s (e.g., Barry et al. (1990); Megginson and Weiss (1991); Loughran and Ritter (2004); Lowry and Schwert (2004); Liu and Ritter (2010, 2011), Lowry and Murphy (2007), Lowry et al. (2010), Loughran and McDonald (2013)). Proposed explanations for a positive effect of VC backing have included grandstanding (Gompers(1996); Lee and Wahal (2004)), hot issue allocations (e.g., Loughran and Ritter (2004, 2010)), and pursuit of analyst coverage (e.g., Cliff and Denis (2004) and Loughran and Ritter (2011)). Lowry et al. (2010) note, however, that underpricing tests are sensitive to the inclusion of the dot-com bubble. A recent study by Chemmanur, Krishnan, and Yu (2018) shows that VC backing draws greater investor attention to IPOs, proxied by media attention, which enables underwriters to disseminate information and gather information from investors more effectively, leading to higher IPO and secondary market valuations. 8

10 Building on this research, we control for the direct effects of top underwriters and VCs on IPO underpricing. Crucially, we focus on the effects of strategic interaction and repeat dealing ties between top underwriters and VCs, identify the effects using quasi-natural experiments, and show that the effects of interest continue to be highly relevant in recent decades, after various regulatory interventions. We account for the analyst coverage effect and run a horse race between certification and rent extraction hypotheses. Third, several studies have considered preferential IPO allocations and analyst lust as a motivation for high underpricing during the eighties and nineties (e.g., model in Biais, Bossaerts, and Rochet (2002) and evidence in Hanley and Wilhelm (1995) and Aggarwal, Prabhala, and Puri (2002)). Loughran and Ritter (2004) and Liu and Ritter (2010) examine evidence from preferential allocation cases during the dot-com period. Similarly, some work tests whether analyst lust is a motivating factor for the underpricing of VC-backed IPOs. For example, Cliff and Denis (2004) and Loughran and Ritter (2011) linked star analyst coverage to underpricing. However, Bradley, Kim, and Krigman (2015) found that star analyst coverage stopped to affect underpricing during , including for top VC deals, due to changes such as the effects of Regulation FD and Global Settlement. Tests considering Regulation FD relate to studies showing that it reduced selective access of analysts to private information (e.g., Charoenrook and Lewis (2009); Mohanram and Sunder (2006); Findlay and Mathew (2006)). In a related paper, Hoberg and Seyhun (2009) propose and test hypotheses about collaboration between underwriters and venture capitalists, centering on IPO allocations, analyst coverage, and insider selling in IPOs during They show that VCs with more underpriced deals received more long-term marketing support and favorable analyst coverage during that period, while underwriters received repeat business and profit from allocating underpriced shares. In this paper, we account 9

11 for the potential analyst channel of VC underwriter interactions and show that the 2003 shock to preferential allocations helped reduce underpricing in VC-backed top underwriter deals, but such underpricing continues to persist in our sample ending We also examine whether reputation and repeat dealing with institutional buyers helps moderate the effect and examine changes to the role of VC-underwriter ties for underpricing after the 2012 shock. Overall, our paper contributes to existing research in several important ways. We show that IPO underpricing of VC top underwriter deals is not limited to the tech bubble and persists into late 2016, despite close to two decades of learning by market participants and regulatory interventions to address conflicts of interest in analyst coverage, governance, IPO allocations, and efforts to stimulate IPO activity. We test for causal impacts by relying on two quasi-natural experiments that affected the information environment of issuers and the repeat dealing incentives of VCs and underwriters, as well as by using geographical instruments. In addition, we examine repeat dealing between institutional investors and top underwriters. We show that such incentives contribute to IPO underpricing, remain unaffected by the 2003 shock to preferential allocations or the JOBS Act, and do not explain the VC top underwriter interactions. Finally, we do not find that analyst coverage is a significant channel for the VC top underwriter effect on underpricing during the post-dot-com period. Our paper provides significant new insights. First, we show that VC-backed deals with top underwriters exhibit significantly higher underpricing, particularly when a given VC underwriter pair has been involved in multiple IPO deals in previous years. Second, VC-backed deals ran by top underwriters experienced significantly more underpricing after the 2012 JOBS Act. This is consistent with VCs and top underwriters using their information advantage to coordinate on underpricing rather than to play a certification role. Third, we examine the effects 10

12 of VCs and top underwriters on IPO underpricing using a new quasi-natural experiment around NASD Rule 2790 shock to IPO allocations. The effect of the shock on underpricing was concentrated among VC-backed IPOs with top underwriters. Crucially, the effect was pronounced for non-star analyst covered deals, suggesting that the elimination of conflicts in research analyst coverage only partly explained the reduction in VC-backed IPO underpricing. Fourth, we examine the effects of the shock to hot allocations and analyst coverage on the underpricing of IPOs with top underwriters. We do not observe a significant effect of the 2003 shock on non-vc IPOs, suggesting that the effects of these changes involved an interaction between conflicts of interest of VCs and underwriters. Finally, we explore the role of strategic interactions between institutional investors and top underwriters. While such ties increase underpricing on average, they are not affected by the 2003 shock to preferential allocations or the 2012 shock to issuers information environment, and thus do not seem to explain the direction of the VC/underwriter effects. Overall, our evidence strongly supports the hypothesis about rent extraction by VCs and top underwriters in a repeat dealing context and does not support the certification hypothesis. 3. Data The IPO sample is obtained from Dealogic Equity Capital Markets database. Consistent with prior IPO studies, the sample excludes blank checks, special purpose acquisition companies (SPACs), deals that do not list on a major exchange (NYSE/Amex or NASDAQ), small issuers (proxied by deals with offer prices below $5 or offer value below $10 million), American Depository Receipts (ADRs) offerings, and IPOs of foreign private issuers and other issuers incorporated abroad, closed-end funds, real estate investment trusts (REITs), energy and commodity trusts, and unit offerings (e.g., partnerships). Deals for which over one year has 11

13 elapsed between announcement and offer date are also excluded from the main sample because a long time to offer may indicate obstacles arising in deal marketing or other significant changes on the issuer s end. Deals missing information on first-day returns or deal values are excluded. Our main dependent variable, underpricing, is defined as the first-day IPO return, which is a standard measure of IPO underpricing. Cross-sectional tests examine the effects of VC and top underwriter presence (see Loughran and Ritter (2004); the top underwriter indicator is based on the lead underwriter (where multiple managers are present, the name of the first manager in the underwriter table is used)). Given the hypothesis about VC/underwriter interaction and incentives, we also interact the VC backing indicator and the top underwriter indicator. In addition, we introduce measures of VC/underwriter ties and institution/underwriter ties. Those measures consider the historical overlaps between the same underwriter and VC and between underwriters and institutions involved in the deal on prior deals to capture the potential strength of incentives arising from repeated interaction. To avoid major confounding market events, the tests begin in 2001, after the dot-com period (e.g., Lowry et al. (2010) show that the dot-com period dramatically affected underpricing), and exclude financial crisis years (2008 and 2009), which were characterized by extremely unfavorable market conditions and a lack of IPO activity. Robustness tests add financial crisis years back into the sample and start the sample period earlier. Tests of the shock effects interact the post-shock period indicator with VC backing, VC/underwriter ties, or institution/underwriter ties. The post-2003 shock indicator equals 1 in the period after the adoption of NASD Rule 2790 (October 2003). Tests examining the 2003 shock use the sample period. Robustness tests end the sample period in 2005 or 2006 or 12

14 start the sample period earlier, in The post-2012 shock period indicator equals 1 in the period after the enactment of the JOBS Act (April 2012) that introduced the EGC status and associated disclosure relief, which became available to most IPO issuers after that time. The tests examining the 2012 shock use the period. Due to collinearity, tests of post-shock interaction effects consider either interaction with the VC effect or interaction with the VC/underwriter joint or ties. Robustness tests account for the effects of the 2000 Regulation FD. To account for potential industry differences in risk or investment opportunities, tests include Fama-French 49 industry fixed effects. Time period (year-quarter) fixed effects are included to account for granular time variation in IPO market trends that could affect underpricing (e.g., Derrien (2005); Ivanov and Lewis (2008); Lowry et al. (2010)). Time fixed effects override direct post-shock period effects but help demean underpricing tests of potential market-wide confounding effects due to unobservable changes to the aggregate pool of issuers. Explanatory variables include standard controls from prior studies of underpricing, as well as industry and time fixed effects. To account for partial adjustment, we incorporate a control for upward offer price revision an increase in the offer price between the initial filing date and the IPO pricing date which has been found to increase underpricing (see, e.g., Benveniste and Spindt (1989); Hanley (1993); Lowry and Schwert (2004); Cliff and Denis (2004); Habib and Ljungqvist (2001); Ljungqvist and Wilhelm (2003); Ritter and Welch (2002); Aruǧaslan, Cook, and Kieschnick (2004); Loughran and McDonald (2013)). Since issuers selling a smaller stake in the firm at IPO time (having a greater share overhang) would incur a lower opportunity cost of underpricing, we expect the stake to be negatively related to underpricing (e.g., Loughran and Ritter (2004); Loughran and McDonald (2013)). We also account for deal size; length of time from announcement to offer; market returns in the month prior to the offer 13

15 (e.g., Lowry and Murphy (2007)). Robustness tests control for additional characteristics that may affect underpricing, including issuer age (Field and Karpoff (2002); Loughran and Ritter (2004); Ljungqvist and Wilhelm (2003)); VC reputation (Krishnan, Ivanov, Masulis, and Singh (2011)); star analyst coverage from the underwriter (Liu and Ritter (2011); Cliff and Denis (2004)); private equity backing; and direct VC indicator effects in tests of VC/underwriter ties. Robust standard errors are used to account for potential heteroscedasticity. We do not cluster errors by industry and time because the main specifications already include industry and time fixed effects (Petersen(2009)). In unreported robustness tests, clustering standard errors by lead underwriter, industry, issuer location, and a combination of lead underwriter and time period does not affect the main results. Sample and variable definition details are presented in the Appendix. Summary statistics of the main variables are reported in Table 1. [Table 1] 4. Results Initial univariate tests are presented in Table 2. Panel A reports t-tests of the effects of VC backing, top underwriter presence, and their interaction on underpricing. We find that underpricing is significantly higher for VC-backed deals, deals with a top underwriter, and deals with a combination of VC backing and a top underwriter. Panel B reports t-tests of the effect of the 2003 and 2012 shocks on underpricing in subsamples based on VC and top underwriter indicators and their interaction. We find that the 2003 shock reduced underpricing in the VCbacked subsample and in the subsample of VC-backed IPOs with a top lead underwriter. However, the shock did not have a significant effect on the on underpricing of non-vc-backed IPOs or on underpricing in the full sample. The 2012 shock increased underpricing in the 14

16 subsample with a combination of VC-backing and a top underwriter (significant at 10%) but not for other deals or the overall sample of firms. Top underwriter presence was not associated with differences in the effects of either shock. [Table 2] Univariate tests do not account for potential differences in issuer characteristics that could be correlated with the variables of interest. Table 3 reports multivariate tests. Panel A reports the effects of VCs, underwriters, and their ties on underpricing during the full sample period. In line with the univariate tests, VC-backed IPOs were more underpriced on average; IPOs with VC backing and a top underwriter were more underpriced on average (significant at 10%); and the presence of VCs with stronger underwriter ties was also associated with greater underpricing. The results support the strategic rent extraction behavior of VCs, particularly when VCs have close ties to underwriters. The effect of top underwriters by themselves on underpricing is insignificant. It is possible that the certification role of top underwriters offsets their strategic incentives to underprice. [Table 3] Panel B examines the effects of the 2012 shock on underpricing of deals with VC backing and top underwriters. The EGC provisions of the 2012 JOBS Act have been linked to increased information asymmetry between issuers and investors (e.g., Chaplinsky et al. (2017) and Barth et al. (2017)). After controlling for changes in market and industry conditions during this period, we find that VC-backed deals, VC-backed deals with top underwriters, and deals with close VC/underwriter ties experienced increases in underpricing after the 2012 shock. This is consistent with VCs, in particular VCs in the presence of top underwriters and VCs with close 15

17 ties to underwriters, extracting rents rather than providing certification after information asymmetry increases. Panel C examines the effects of VCs, underwriters, and their ties around the 2003 shock, which reduced certain explicit channels for rent extraction and strategic interaction with respect to underpricing. Consistent with the rent extraction hypothesis and the findings of univariate tests, VC-backed IPOs, IPOs involving VC backing and a top underwriter, and IPOs involving VCs with close underwriter ties were more underpriced on average but became less underpriced after the 2003 shock. The results suggest that the 2003 shock weakened the strategic interaction incentives of VCs, particularly VCs with close ties to underwriters, in the context of IPO underpricing. However, the presence of a top underwriter by itself was not significant and the effect did not change significantly around the 2003 shock, suggesting that strategic interaction between top underwriters and VCs was more relevant for underpricing than the presence of a top underwriter alone. Panel D repeats the tests of the 2003 and 2012 shock effects in the full sample, yielding consistent results. Endogeneity due to nonrandom nature of VC-backing and VC/underwriter ties is a potential source of concern. Endogeneity may be due to unobservable issuer characteristics, such as issuer risk or uncertainty about issuer quality, that predict VC presence (e.g., Gompers (1995)). Deals involving riskier issuers and more uncertainty about issuer quality may also be associated with greater underpricing. In addition, endogeneity may be due to VCs and underwriters electing to repeatedly partner on deals with high underpricing if such deals are mutually beneficial. In either instance, the direction of causality would affect interpretation. We 16

18 utilize industry and time fixed effects but cannot employ firm fixed effects to address unobserved variation in issuer characteristics because IPO issuers only appear in the sample once. The use of a shock setting as part of the main tests helps to alleviate some of the endogeneity concerns. The 2003 shock affected the benefits of strategic interaction with underwriters in setting the IPO price without directly affecting issuer information risk. We perform various robustness tests to account for the effects of changes in the analyst environment and for SOX, which may have affected information risk of issuers in early 2000s and rule out those explanations for the effects of the 2003 shock on underpricing of VC-backed deals with top underwriters. Conversely, it has been argued that the 2012 shock has increased the uncertainty about IPO issuers by enabling most issuers to disclose less information. It could have either created greater demand for certification or alternatively, greater uncertainty could make it easier to VC and underwriters to coordinate on underpricing because issuers recognize that greater underpricing is expected in the presence of greater uncertainty. Other than through the effect on information uncertainty, the 2012 shock did not affect strategic dealing incentives between VCs and underwriters. Further, to attempt to further address endogeneity concerns, we use two-stage least squares, reported in Table 4. We use the density of VC firms and the amount of VC funding invested by local VC firms in any firm in the issuer s local area to predict the likelihood of VC backing (motivated by Lerner (1995); Chen, Gompers, Kovner, and Lerner (2010)). Extending the logic of other studies using distance-based instruments, we also add the density of securities firms that are near the VC firm, either for all VC firms in the area, or for VC firms involved in the deal, to predict VC/underwriter ties. First-stage F-statistics support the relevance of the instruments. It is not clear that either variable would directly increase underpricing. If greater 17

19 density of financial institutions relates to economic vibrancy and better average quality of issuers in the area, it might reduce underpricing. In the two-stage least squares setting, we continue to find larger underpricing for VC-backed deals and deals in which VCs have strong underwriter ties. [Table 4] For robustness, Table 5 repeats the analysis of VC effects in subsamples based on top underwriter presence. Consistent with the prior results, the positive relation between VC backing and underpricing, the decline in the underpricing of VC-backed IPOs after the 2003 shock, and the increase in underpricing of VC-backed deals after the 2012 shock, were most significant in the subset of deals involving a top underwriter. [Table 5] The positive effect of VC backing on underpricing, particularly in the nineties, has also been attributed to analyst lust of VCs, which may cause VCs to accept greater underpricing in exchange for star analyst coverage supplied by major underwriters that back the IPO. To the extent that the 2003 NASD rule coincided with other regulatory developments related to analyst research, it is important to consider whether our findings are due to the changing dynamics of between underwriting and analyst research. By reducing conflicts of interest between investment banking business and research coverage, Regulation AC, adopted as part of implementing SOX, and the Global Settlement, likely limited the ability of top underwriters to provide favorable analyst coverage to their IPO issuers, potentially reducing the incentive of VCs to accept greater underpricing. In Panel A, we confirm that the effect of the 2003 shock remained in the subset of deals without star analyst coverage from the underwriter. Later, in Table 8 (Panel B), we also interact VC backing and 2003 shock effects with the presence of star analyst coverage from 18

20 underwriter and do not find that it subsumes the effects of the 2003 shock on the underpricing of VC-backed deals and VC-backed deals with top underwriters. Separately, in Table 8 (Panel C), we also examine the effects of Regulation FD, which limited selective disclosure by issuers and has been shown to reduce private communications with analysts. Research coverage became subject to fewer conflicts of interest after and the ability of underwriters to provide biased coverage for their issuers was limited in the 2000s compared to the 1990s. Overall, these shifts do not appear to be driving the observed interaction effects of VCs and underwriters on IPO pricing during our sample period. The main test of the 2012 shock used the post-2012 shock indicator, consistent with the vast majority of IPOs in the post-2012 period being eligible for EGC status. In Panel A of Table 6, for robustness we replace the post-2012 shock indicator with the indicator for EGC status based on the issuer s filings. The results continue to hold and have even greater significance, suggesting that the 2012 shock effect was driven by deals affected by the 2012 regulatory reform. [Table 6] The prior analysis defined the time of the shock as October 2003, which corresponds to the adoption of NASD Rule For robustness we alter the definition of the 2003 shock to allow for early or late adjustment. Panel B redefines the shock based on the time when compliance with Rule 2790 became mandatory (March 2004). The results continue to hold. Panel C examines April 2003, when the Voluntary Initiative was announced, which may have led top underwriters to begin adjusting their IPO allocation practices ahead of the adoption of Rule The effects remain consistent with prior results. deals backed by VCs, VC-backed IPOs with a top underwriter, and VC-backed IPOs with close underwriter ties become relatively less 19

21 underpriced after the shock. However, across all IPOs, the effects of top underwriters on underpricing do not vary significantly with the shock. Repeated interaction of underwriters with institutions can create incentives to underprice deals in order to encourage institutional buyers to subscribe to the underwriter s future IPOs. In Table 7, we examine this effect. When we consider underwriter ties to institutional investors, we find that they have a significant positive effect on underpricing, consistent with an incentive to underprice deals in the presence of repeat dealing with institutional buyers. However, the effect of institution/underwriter ties does not explain the effect of VC/underwriter ties, both of which are separately associated with higher underpricing. In addition, unlike the prior results of VC tests showing that the 2003 shock moderated and the 2012 shock increased the effect of VC/underwriter ties on underpricing, respectively, the effect of institution/underwriter ties does not exhibit a similar pattern around these shocks. The effect of the 2003 shock on the role of institution/underwriter ties on underpricing is either not significant or positive, depending on the specification. Even though explicit incentives stemming from preferential IPO allocations to institutional money managers were restricted by the 2003 rule, implicit incentives from repeat dealing whereby the same institutions continue to invest in the underwriter s underpriced IPO deals, which this measure captures remained strong in the 2000s and 2010s. Thus, this result highlights the importance of implicit incentives to strategically underprice stemming from repeat dealing with institutions in the context of IPO price setting. [Table 7] Panel A of Table 8 enhances the specifications by including additional control variables for VC and PE backing, firm age, VC reputation, and star analyst rating. The main results continue to hold. In Panel B we further test whether the effect of the shock on IPO underpricing 20

22 was due to the analyst lust channel. Specifically, we consider whether the relation between star analyst coverage and underpricing changed after the shock, for all and for VC-backed IPOs. The interaction between star analyst coverage and 2003 shock indicator is not significant whereas the interaction between the VC indicator and shock indicator remains significant. The triple interaction between VC, star analyst, and shock indicator is not significant. It appears that the reduction in VC-backed IPO underpricing around 2003 cannot be solely explained by VC appetite for star analyst coverage. 6 [Table 8] In Panel C, we repeat the tests comparing the effects of the 2000 Regulation FD and the 2003 NASD rule on the underpricing of VC-backed IPOs. Regulation FD restricted selective disclosure of information by issuers, including private information disclosure to some investors or analysts (e.g., Charoenrook and Lewis (2009); Mohanram and Sunder (2006); Findlay and Mathew (2006)). By reducing the amount of information an issuer may selectively convey to analysts, this change could reduce the advantage of the underwriter s analysts in providing coverage and impair the issuer s ability to cultivate an analyst relationship with the underwriter. If the VC-underwriter effect on underpricing were driven by VC preference for analyst coverage, the effect should weaken after Regulation FD. We do not find that to be the case. When Regulation FD is examined separately, it does not have a disproportionate effect on VC-backed deals or VC-backed deals with top underwriters. The effect of Regulation FD on VC-backed IPOs does not appear to go in the direction of the main finding from the 2003 shock. To compare 6 In unreported tests, we confirm a strong positive relation from prior work through After 2000, but particularly, after the Global Settlement in 2003, the nature of assignment of star analysts to IPOs likely changed. Indeed, the incidence of star analyst coverage from the underwriter declines during the post-shock of the sample period. When assigned to an IPO stock, star analysts might cover the stock differently after the shock than before the shock. The Global Settlement likely had an aggregate, industrywide effect on the reduction of analyst bias in research reports, which may be captured by time period fixed effects. The lack of sensitivity of underpricing to star analyst coverage in the 2000s is consistent with Bradley et al. (2015) 21

23 the 2000 and 2003 shocks we consider their joint effects on the underpricing of VC-backed and top underwriter IPOs. Deals with top underwriters experience a marginal decline in underpricing after Regulation FD (depending on the specification). Overall, it does not appear that Regulation FD is driving the observed effect of VC-underwriter ties on IPO underpricing. In Table 9 we perform several additional tests to validate our shocks. First, it is possible that some confounding market or regulatory event, such as the dot-com bust or subsequent aggregate trends in the IPO market drove a systematic shift in underpricing of VC-backed and top underwriter IPOs. In Panel A we reassign shock timing to alternative time periods unrelated to our shocks. We do not find the interaction of the VC indicator with the indicators based on placebo shock definitions to be significant. Second, one of the identifying assumptions of a differences-in-differences estimation is the parallel trends assumption: IPOs with and without VCs and top underwriters need to exhibit parallel trends in underpricing prior to the shock. The presence of different trends can lead to spurious estimates in differences-in-differences tests around the shock. Panel B reports regressions of underpricing in the pre-shock periods on controls, VC indicator, time trend, and interactions of the effects of interest with the time trend. A significant interaction with the time trend would be of concern. We find that the interactions with the time trend are insignificant, consistent with parallel trends. [Table 9] Third, some other characteristics, such as issuer risk, may be driving both VC backing and underpricing, and this characteristic s effect on underpricing may have changed around the time of the shock.. To address such potential confounding, in Panel C, we repeat the tests around the 2003 and 2012 shocks substituting the VC indicator with a placebo indicator based on the issuer being in the high-tech industry. While high-tech deals had relatively higher underpricing 22

24 in some specifications (not surprising), the effects of the shocks on the underpricing of high-tech deals are not significantly different from the effects of the shock on non-high-tech firms. These tests alleviate the concern that our findings were merely due to a confounding shift in the underpricing of issuers with specific characteristics that draw VCs. Fourth, in Panel D we randomly reassign the VC indicator while preserving the same proportion of VC firms as in the actual sample and repeat the underpricing tests using the placebo VC indicator and its interaction with the shocks 10,000 times. We plot and tabulate the percentiles of the distribution of coefficient estimates on the placebo effects and controls (the actual specification using the true VC assignment is also reported for reference). Unsurprisingly, between 5% and 6% of placebo coefficient estimates are significant at the 5% level, consistent with Type I error in a random assignment scenario. Actual estimates of the VC-shock interaction of interest fall in the 99.5% tail of the distribution of placebo estimates for the 2003 shock and between 95% and 97.5% tail of the distribution of placebo estimates for the 2012 shock, suggesting that actual estimates are very unlikely to arise in a random assignment scenario. Finally, Table 10 reports additional robustness tests for the 2003 shock tests. Panel A narrows the window to end either in 2005 or 2006, rather than Panel B considers the potential effects of Section 404 of Sarbanes-Oxley, which has imposed requirements of managerial reporting and auditor attestation of internal controls over financial reporting on larger issuers. To the extent that underpricing reflects information risk and to the extent that the anticipation of these requirements regarding internal controls over financial reporting could have reduced information risk and agency concerns, one might expect underpricing to decline post- SOX for issuers affected by this requirement. Although it is not clear that SOX would affect VCbacked IPOs differently, for robustness we control for the effects of SOX or exclude affected 23

25 firms. The requirement was phased in for larger filers in November 2004 but not phased in for smaller (non-accelerated) filers during the period around the 2003 shock (it was subsequently repealed for non-accelerated filers in 2010). Similar to Chaplinsky et al. (2017), we use deal size above $75 million to proxy filer size. Repeating tests in the sample without larger deals and in the sample without larger issuers in the post-sox period does not affect the results. [Table 10] Further robustness tests are reported in the Appendix. To address potential extreme observations, Table A1 winsorizes continuous variables. Another potential concern is a shift in the characteristics of issuers classified in the two groups around the shock. Although controls for key observable characteristics are included in all specifications, dramatic differences in observable characteristics around the shock can signal a shift in unobservable characteristics of the issuer pool around the shock. Table A2 examines this issue. Panel A regresses explanatory variables on the VC indicator and VC-shock interactions, controlling for industry and time fixed effects. While many of the controls do not respond differently to the shocks, there are some differences. It could be because VC-backed deals tend to be smaller. For robustness, to increase the comparability between the characteristics of VC and non-vc IPOs around the shock, we repeat the analysis in a sample that excludes deals exceeding the 95 th percentile of size and stake sold of VC-backed deals (close to $250 million and 46%, respectively). The differences in explanatory variables become less significant. Underpricing tests continue to hold in this restricted sample, as shown in Panel B of Table A2. Finally, Table A3 in the Appendix adds financial crisis years back to the sample. The results continue to hold. 24

26 5. Conclusion This paper has examined the role of VCs and top underwriters in the IPO process, and the effects of their strategic interactions on IPO underpricing. Building on existing work, we empirically tested contrasting hypotheses about their impact using quasi-natural experiments from two regulatory shocks in the post-dot-com period. On the one hand, underwriters and VCs are information intermediaries that build reputations with issuers and investors and certify issuer quality, which can reduce underpricing. On the other hand, underwriters and VCs have strategic incentives to underprice IPOs, accumulating rents from repeat dealing at a cost to issuers. Our empirical results support the rent extraction hypothesis and are not consistent with the certification hypothesis. In the cross-section, we find that VC-backed IPOs are more underpriced, and the effect is most pronounced for VC-backed IPOs involving top underwriters. Further, we find that IPO underpricing increases in the strength of repeat dealing incentives measured by previous VC/underwriter ties. This result continues to hold in a two-stage least squares setting using geographic instruments to predict VC backing and VC-underwriter ties. Next, we tested the hypotheses around two quasi-natural experiments that impacted the ability of VCs and top underwriters to extract rents in the IPO process through underpricing. First, we found that, after the JOBS Act, increased informational asymmetries did not strengthen the certification role of VCs and top underwriters. On the contrary, underpricing of VC-backed IPOs involving top underwriters and stronger VC-underwriter connections increased after the JOBS Act. The effect was concentrated among emerging growth companies most affected by the newly reduced disclosure requirements. Second, we examined the effects of the 2003 NASD rule prohibiting preferential IPO allocations on IPO underpricing of VC-backed, top underwriters deals. We find that the shock to preferential allocations decreased the underpricing of VC-backed 25

27 IPOs, and particularly, VC-backed IPOs involving top underwriters and VCs connected to underwriters. We also examined other potential incentive conflicts and obtained three important results. First, most of our results are driven by strategic interactions between VCs and underwriters, and underwriter conflicts alone do not explain our findings. Underwriters have strong incentives to underprice, but repeat dealing with VCs that bring in a pipeline of bankable projects is critical to the observed IPO underpricing in deals with top underwriters. Second, we examined the role of repeat dealing between top underwriters and large institutions. Institutional investors buying at the offer price benefit from underpricing, and underwriters can underprice deals to build a network of institutional buyers that will invest in future IPO deals. We found that when underwriters engage in repeat IPO dealing with large institutions, IPO underpricing increases, supporting the incentive conflict effect: underwriters strategically underprice IPOs and institutions invest in IPOs backed by the underwriter. However, we did not find that such repeated dealing between institutional investors and top underwriters was affected by the 2003 shock to preferential IPO allocations or the 2012 shock to information asymmetries about issuers. Thus, the effects of repeat dealing between top underwriters and institutional buyers stem from implicit incentives for repeat business that were not affected by the elimination of explicit incentives due to preferential IPO allocations. We did not find that analyst coverage is a channel for VC-top underwriter interactions during our sample period. The effects of the 2003 shock on underpricing of VC-backed deals with top underwriters continue to hold even in the absence of coverage by the underwriter s star analysts. In addition, the passage of Regulation FD (a shock to analyst coverage) did not attenuate the role of VC-underwriter interactions for IPO underpricing. While we recognize that 26

28 analysts may be important for IPO bargaining dynamics, the analyst channel does not appear to be a significant mechanism behind the VC-underwriter effects in the post-dot-com period. Our findings continued to hold after a battery of robustness and sensitivity tests, including alternative selection criteria, additional measures, and placebo tests. Overall, our paper sheds new light on a critical component of issuers cost of capital IPO underpricing. Our results suggest that VCs and top underwriters jointly extract rents by underpricing IPO issues, particularly when repeat dealing is significant. The effects are stronger after regulatory shocks that impacted their ability to extract rents related to IPO pricing. Moreover, repeated dealing between underwriters and institutional investors increases underpricing, but the effect is likely driven by implicit business incentives and not explicit preferential allocations. This evidence sheds new light on VC-backed deals, the role of information asymmetries about issuers in the IPO process, and incentives stemming from repeat interactions between VCs and top underwriters. 27

29 References Aggarwal, Reena, N.K. Prabhala, and Manju Puri, 2002, Institutional allocation in initial public offerings: empirical evidence, Journal of Finance 57(3), Aruǧaslan, Onur, Douglas Cook, and Robert Kieschnick, 2004, Monitoring as a motivation for IPO underpricing, Journal of Finance 59(5), Barondes, Royce, 2005, NASD regulation of IPO conflicts of interest - does gatekeeping work? Tulane Law Review 79, Barth, Mary, Wayne Landsman, and Daniel Taylor, 2017, The JOBS Act and information uncertainty in IPO firms, Accounting Review 92(6), Barry, Christopher, Chris Muscarella, John Peavy, and Michael Vetsuypens, 1990, The role of venture capital in the creation of public companies: evidence from the going-public process, Journal of Financial Economics 27(2), Beatty, Randolph, and Jay Ritter, 1986, Investment banking, reputation, and the underpricing of initial public offerings, Journal of Financial Economics 15(1 2), Benveniste, Lawrence, and Paul Spindt, 1989, How investment bankers determine the offer price and allocation of new issues, Journal of Financial Economics 24, Biais, Bruno, Peter Bossaerts, and Jean-Charles Rochet, 2002, An optimal IPO mechanism, Review of Economic Studies 69(1), Bradley, Daniel, Bradford Jordan, and Jay Ritter, 2008, Analyst behavior following the IPO: the bubble period evidence, Review of Financial Studies 21, Bradley, Daniel, Incheol Kim, and Laurie Krigman, 2015, Top VC IPO underpricing, Journal of Corporate Finance 31, Carter, Richard, and Steven Manaster, 1990, Initial public offerings and underwriter reputation, Journal of Finance 45(4), Carter, Richard, Frederick Dark, and Ajai Singh, 1998, Underwriter reputation, initial returns, and the long-run performance of IPO stocks, Journal of Finance 53(1), Chaplinsky, Susan, Kathleen W. Hanley, and S. Katie Moon, 2017, The JOBS Act and the costs of going public, Journal of Accounting Research 55(4), Charoenrook, Anchada, and Craig M. Lewis, 2009, Information, selective disclosure, and analyst behavior, Financial Management 38(1), Chemmanur, Thomas, Karthik Krishnan, and Qianqian Yu, 2018, Venture capital backing, investor attention, and initial public offerings, Working paper. Chen, Shuping, DeFond, Mark, Park, Chul, Voluntary disclosure of balance sheet information in quarterly earnings announcements, Journal of Accounting and Economics 33(2), Chen, Henry, Paul Gompers, Anna Kovner, and Josh Lerner, 2010, Buy local? The geography of venture capital, Journal of Urban Economics 67(1), Cliff, Michael, and David J. Denis, 2004, Do initial public offering firms purchase analyst coverage with underpricing? Journal of Finance 59(6), Dambra, Michael, Laura C. Field, and Matthew T. Gustafson, 2015, The JOBS Act and IPO volume: evidence that disclosure costs affect the IPO decision, Journal of Financial Economics 116(1),

30 Dambra, Michael, Laura C. Field, Matthew Gustafson, and Kevin Pisciotta, 2018, The consequences to analyst involvement in the IPO process: evidence surrounding the JOBS Act, Journal of Accounting and Economics 65(2), Derrien, Francois, 2005, IPO pricing in hot market conditions: who leaves money on the table? Journal of Finance 60(1), Drake, Philip D., and Michael R. Vetsuypens, 1993, IPO underpricing and insurance against legal liability, Financial Management 22(1), Fama, Eugene, and Kenneth French, 1997, Industry costs of equity, Journal of Financial Economics 43(2), Fang, Lily, and Ayako Yasuda, 2009, The effectiveness of reputation as a disciplinary mechanism in sell-side research, Review of Financial Studies 22, Fanto, James, 2008, The continuing need for broker-dealer professionalism in IPOs, Entrepreneurial Business Law Journal 2(2), Field, Laura, and Jonathan Karpoff, 2002, Takeover defenses of IPO firms, Journal of Finance 57(5), Findlay, Scott, and Prem Mathew, 2006, An examination of the differential impact of Regulation FD on analysts' forecast accuracy, Financial Review 41, Gompers, Paul, 1995, Optimal investment, monitoring, and the staging of venture capital, Journal of Finance 50(5), Gompers, Paul, 1996, Grandstanding in the venture capital industry, Journal of Financial Economics 42(1), Griffith, Sean J., 2003, Spinning and underpricing: a legal and economic analysis of the preferential allocation of shares in initial public offerings, Brooklyn Law Review 69(2), Gullapalli, Rachita, and Anzhela Knyazeva, 2017, Foreign issuer puzzle in emerging growth company IPOs, Working paper. Habib, Michel, and Alexander Ljungqvist, 2001, Underpricing and entrepreneurial wealth in IPOs: Theory and evidence, Review of Financial Studies 14(2), Hanley, Kathleen, 1993, The underpricing of initial public offerings and the partial adjustment phenomenon, Journal of Financial Economics 34(2), Hanley, Kathleen W., and William Wilhelm, 1995, Evidence on the strategic allocation of initial public offerings, Journal of Financial Economics 37, Hoberg, Gerard, and Nejat Seyhun, 2009, Do underwriters collaborate with venture capitalists in IPOs? Implications and evidence, Working paper. Hurt, Christine, 2005, Moral hazard and the initial public offering, Cardozo Law Review 26(2), Ibbotson, Roger G., Jody L. Sindelar, and Jay R Ritter, 1994, The market's problems with the pricing of initial public offerings, Journal of Applied Corporate Finance 7(1), Ivanov, Vladimir, and Craig Lewis, 2008, The determinants of market-wide issue cycles for initial public offerings, Journal of Corporate Finance 14(5), Krishnan, C. N. V., Vladimir Ivanov, Ronald Masulis, and Ajai Singh, 2011, Venture capital reputation, post-ipo performance, and corporate governance. Journal of Financial and Quantitative Analysis 46(5),

31 Lee, Peggy, and Sunil Wahal, 2004, Grandstanding, certification and the underpricing of venture capital backed IPOs, Journal of Financial Economics 73(2), Lerner, Josh, 1995, Venture capitalists and the oversight of private firms, Journal of Finance 50(1), Levy, Eric, 2004, The law and economics of IPO favoritism and regulatory spin, Southwestern University Law Review 33, Liu, Xiaoding, and Jay Ritter, 2010, The economic consequences of IPO spinning, Review of Financial Studies 23, Liu, Xiaoding, and Jay Ritter, 2011, Local underwriter oligopolies and IPO underpricing, Journal of Financial Economics 102(3), Ljungqvist, Alexander, and William Wilhelm, 2003, IPO pricing in the dot-com bubble, Journal of Finance 58(2), Ljungqvist, Alexander, 2007, IPO underpricing, In: B.E. Eckbo (ed.), Handbook of Corporate Finance, North-Holland. Lowry, Michelle, Roni Michaely, and Ekaterina Volkova, 2017, Initial public offerings: a synthesis of the literature and directions for future research, Foundations and Trends in Finance (forthcoming). Lowry, Michelle, and Kevin Murphy, 2007, Executive stock options and IPO underpricing, Journal of Financial Economics 85, Lowry, Michelle, Micah Officer, and William Schwert, 2010, The variability of IPO initial returns, Journal of Finance 65(2), Lowry, Michelle, and William Schwert, 2004, Is the IPO pricing process efficient? Journal of Financial Economics 71(1), Loughran, Tim, and Bill McDonald, 2013, IPO first-day returns, offer price revisions, volatility, and form S-1 language, Journal of Financial Economics 109(2), Loughran, Tim, and Jay Ritter, 2002, Why don't issuers get upset about leaving money on the table in IPOs? Review of Financial Studies 15(2), Loughran, Tim, and Jay Ritter, 2004, Why has IPO underpricing changed over time? Financial Management 33, Maynard, Therese H., 2002, Spinning in a hot IPO - breach of fiduciary duty or business as usual? William & Mary Law Review 43(5), Megginson, William, and Kathleen Weiss, 1991, Venture capitalist certification in initial public offerings, Journal of Finance 46, Mohanram, Partha S., and Shyam V. Sunder, 2006, How has Regulation FD affected the operations of financial analysts? Contemporary Accounting Research 23(2), Reuter, Jonathan, 2006, Are IPO allocations for sale? Evidence from mutual funds, Journal of Finance 61(5), Ritter, Jay, and Ivo Welch, 2002, A review of IPO activity, pricing, and allocations, Journal of Finance 57(4), Townsend, Lucas, 2004, Can Wall Street's "global resolution" prevent spinning? A critical evaluation of current alternatives, Seton Hall Law Review 34,

32 Table 1. Summary statistics Sample and variable definitions are presented in the Appendix. Full sample ( , excluding the financial crisis) Obs. Mean Median SD Underpricing VC-backed Top underwriter VC-backed x Top underwriter VC/underwriter ties VC/underwriter ties (no log) Institution/underwriter ties Institution/underwriter ties (no log) D(Revision>0) Revision Stake Size Size (no log) Time to offer Time to offer (no log) Industry return Internet-related shock sample ( ) Obs. Mean Median SD Underpricing VC-backed Top underwriter VC-backed x Top underwriter VC/underwriter ties VC/underwriter ties (no log) Institution/underwriter ties Institution/underwriter ties (no log) D(Revision>0) Revision Stake Size Size (no log) Time to offer Time to offer (no log) Industry return Internet-related shock sample ( ) Obs. Mean Median SD Underpricing VC-backed Top underwriter VC-backed x Top underwriter VC/underwriter ties VC/underwriter ties (no log) Institution/underwriter ties

33 Institution/underwriter ties (no log) D(Revision>0) Revision Stake Size Size (no log) Time to offer Time to offer (no log) Industry return Internet-related

34 Table 2. Univariate tests Sample and variable definitions are presented in the Appendix. Panel A reports univariate t-tests of differences in underpricing between pre- and post-shock periods for subsamples based on the VC backing indicator, top underwriter indicator, and interaction of the two. Panel B reports univariate t-tests of differences in underpricing in the full sample comparing issuers with and without VC backing, with and without a top underwriter, and with and without a combination of VC backing and top underwriter. Panel A: Effects of VC backing and top underwriter presence on underpricing Avg. Avg. Full sample underpricing underpricing Diff. Group: yes Group: no VC-backed Top underwriter VC-backed with top underwriter Panel B: Effects of the 2003 and 2012 shocks on underpricing in the full sample and in subsamples based on VC backing and top underwriter presence Subsample Post-2003 shock avg. underpricing Pre-2003 shock avg. underpricing Diff. Post-2012 shock avg. underpricing Pre-2012 shock avg. underpricing Diff. All VC-backed Not VC-backed Top underwriter No top underwriter VC-backed with top underwriter Not VC-backed / no top underwriter

35 Table 3. Multivariate tests This table reports regressions of underpricing. The intercept, Fama-French 49 industry fixed effects, time period fixed effects, and the post-shock indicator are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t-statistics based on robust standard errors are italicized. Panels A and D use the period (excluding the financial crisis). Panel B uses the period. Panel C uses the period. Panel A: Full sample Dep. var.: Underpricing I II III VC-backed VC-backed x Top underwriter VC/underwriter ties Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R

36 Panel B: 2012 shock Dep. var.: Underpricing I II III VC-backed VC-backed x Post-2012 shock VC-backed x Top underwriter VC-backed x Top underwriter x Post-2012 shock VC/underwriter ties VC/underwriter ties x Post-2012 shock Top underwriter x Post-2012 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

37 Panel C: 2003 shock Dep. var.: Underpricing I II III VC-backed VC-backed x Post-2003 shock VC-backed x Top underwriter VC-backed x Top underwriter x Post VC/underwriter ties VC/underwriter ties x Post-2003 shock Top underwriter Top underwriter x Post-2003 shock Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R

38 Panel D: 2003 and 2012 shocks Dep. var.: Underpricing I II III VC-backed VC-backed x Post-2003 shock VC-backed x Post-2012 shock VC-backed x Top underwriter VC-backed x Top underwriter x Post VC-backed x Top underwriter x Post VC/underwriter ties VC/underwriter ties x Post-2003 shock VC/underwriter ties x Post-2012 shock Top underwriter x Post-2003 shock Top underwriter x Post-2012 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

39 Table 4. Two-stage least squares This table reports regressions of underpricing using two-stage least squares during the period (excluding the financial crisis). The intercept, Fama-French 49 industry fixed effects, time period fixed effects, and the post-shock indicator are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t-statistics based on robust standard errors are italicized. Dep. var.: Underpricing I II III IV VC-backed VC/underwriter ties Revision Stake Size Time to offer Industry return Internet-related First-stage: Revision Stake -4.9E E Size Industry return Time to offer Internet-related Instruments: # local VCs # deals local VCs deal VCs' proximity to UW local VCs' proximity to UW Obs R Adj. R First stage F-statistic

40 Table 5. Analysis within subsamples This table reports regressions of underpricing within subsamples based on the presence of a top underwriter and star analyst coverage from underwriter. The intercept, Fama- French 49 industry fixed effects, time period fixed effects, and the post-shock indicator are included but not reported in the table. Panel A uses the period. Panel B uses the period. Panel C uses the period (excluding the financial crisis). Sample and variable definitions are presented in the Appendix. t-statistics based on robust standard errors are italicized. Panel A: 2003 shock Subsample: Top Top Not a top Not a top No star analyst No star analyst Star analyst underwriter underwriter underwriter underwriter from underwriter from underwriter from underwriter Dep. var.: Underpricing I II III IV V VI VII VIII VC-backed VC-backed x Post-2003 shock VC/underwriter ties VC/underwriter ties x Post-2003 shock Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R Star analyst from underwriter 39

41 Panel B: 2012 shock Subsample: Top underwriter Top underwriter Not a top underwriter Not a top underwriter Dep. var.: Underpricing I II III IV VC-backed VC-backed x Post-2012 shock VC/underwriter ties VC/underwriter ties x Post-2012 shock Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R Panel C: Full sample Subsample: Top underwriter Top underwriter Not a top underwriter Not a top underwriter Dep. var.: Underpricing I II III IV VC-backed VC/underwriter ties Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R

42 Table 6. Alternative shock definitions This table reports regressions of underpricing using alternative shocks. Panel A uses the period and replaces the 2012 shock indicator with the EGC status indicator. Panel B uses the period and defines the shock using the mandatory compliance date of NASD Rule 2790 (March 2004). Panel C uses the period and defines the shock using the Voluntary Initiative announcement (April 2003). The intercept, Fama-French 49 industry fixed effects, time fixed effects, and the post-shock indicators are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t- statistics based on robust standard errors are italicized. Panel A: EGC indicator Dep. var.: Underpricing I II III VC-backed VC-backed x EGC VC-backed x Top underwriter VC-backed x Top underwriter x EGC VC/underwriter ties VC/underwriter ties x EGC EGC Top underwriter x EGC Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R

43 Panel B: Effective date of NASD Rule 2790 Dep. var.: Underpricing I II III VC-backed VC-backed x Post-2003 shock (effective) VC-backed x Top underwriter VC-backed x Top underwriter x Post-2003 (effective) VC/underwriter ties VC/underwriter ties x Post-2003 shock (effective) Top underwriter x Post-2003 shock (effective) Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

44 Panel C: 2003 shock (April) Dep. var.: Underpricing I II III VC-backed VC-backed x Post-2003 shock (April) VC-backed x Top underwriter VC-backed x Top underwriter x Post-2003 shock (April) VC/underwriter ties VC/underwriter ties x Post-2003 shock (April) Top underwriter x Post-2003 shock (April) Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

45 Table 7. Institution/underwriter ties This table reports regressions of underpricing on institution/underwriter ties. Columns I, II, and V use the period (excluding the financial crisis). Column III uses the period. Columns IV and VI use the period. The intercept, Fama-French 49 industry fixed effects, time period fixed effects, and the post-shock indicators are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t-statistics based on robust standard errors are italicized. Dep. var.: Underpricing I II III IV V VI Institution/underwriter ties Institution/underwriter ties x Post Institution/underwriter ties x Post-2012 shock Institution/underwriter ties x EGC EGC VC/underwriter ties Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R

46 Table 8. Additional controls This table reports regressions of underpricing with additional controls. Panel A uses the period (excluding the financial crisis). Panel B uses the period. Panel C considers the effects of Regulation FD using the and periods, as specified. The intercept, Fama-French 49 industry fixed effects, time period fixed effects, and the postshock indicator are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t- statistics based on robust standard errors are italicized. Panel A: Additional controls Dep. var.: Underpricing I II III IV VC-backed VC-backed x Top underwriter VC/underwriter ties Top underwriter Revision Stake Size Time to offer Industry return Internet-related Age VC reputation PE-backed Obs R Adj. R

47 Panel B: Controlling for star analyst coverage Dep. var.: Underpricing I II VC-backed VC-backed x Top underwriter VC-backed x Top underwriter x Post Star analyst VC-backed x Star analyst Star analyst x Post-2003 shock VC-backed x Star analyst x Post-2003 shock VC/underwriter ties VC/underwriter ties x Post-2003 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

48 Panel C: Controlling for Regulation FD Dep. var.: Underpricing I II III IV VC-backed VC-backed x Post-FD VC-backed x Post-2003 shock VC-backed x Top underwriter VC-backed x Top underwriter x Post-FD VC-backed x Top underwriter x Post-2003 shock Top underwriter x Post-FD Top underwriter x Post-2003 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

49 Table 9. Placebo tests Panel A uses placebo shocks defined at alternative points in time. Columns I-VIII use the period (excluding the financial crisis) and Columns IX-XII use the period. Panel B considers interactions with the time trend in the preshock period. Columns I-III use the pre-2003 shock portion of the period and Columns IV-VI consider the pre-2012 shock portion of the period (excluding the financial crisis). Panel C uses the high-tech indicator instead of VC backing. Columns I-II use the period and Columns III-IV use the period. Panel D reports the distribution of coefficient estimates from 10,000 regressions using placebo VC assignments determined randomly (with the incidence of the placebo assignment based on sample means during and , respectively). Estimates using the actual VC assignment are reported for comparison. The intercept, Fama-French 49 industry fixed effects, time period fixed effects, and the post-shock indicator (in Panels A, C-D) are included but not reported in the table. Sample and variable definitions are presented in the Appendix. t-statistics based on robust standard errors are italicized. Panel A: Shifting shock timing Placebo shock: January 2002 January 2001 September 2000 Dep. var.: Underpricing I II III IV V VI VC-backed VC-backed x Placebo shock VC-backed x Top underwriter VC-backed x Top underwriter x Placebo shock Top underwriter x Placebo shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R (continued): January 2000 January 2014 January 2015 VII VIII IX X XI XII VC-backed VC-backed x Placebo shock VC-backed x Top underwriter VC-backed x Top underwriter x Placebo shock

50 Top underwriter x Placebo shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R Panel B: Interaction with the time trend pre-shock (evaluating pre-shock trends in underpricing) Sample: Pre-2003 Pre-2003 Pre-2003 Pre-2012 Pre-2012 Pre-2012 shock shock shock shock shock shock Dep. var.: Underpricing I II III IV V VI VC-backed VC-backed x Time trend VC-backed x Top underwriter VC-backed x Top underwriter x Time trend VC/underwriter ties VC/underwriter ties x Time trend Top underwriter x Time trend Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

51 Panel C: Placebo assignment (high-tech indicator) Dep. var.: Underpricing I II III IV High-tech High-tech x Post 2003 shock High-tech x Post 2012 shock High-tech x Top underwriter High-tech x Top underwriter x Post-2003 shock High-tech x Top underwriter x Post-2012 shock Top underwriter x Post-2003 shock Top underwriter x Post-2012 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj.R

52 Panel D: Placebo assignment (randomized VC assignment) Actual assignment: Dep. var.: Underpricing I II VC-backed VC-backed x Post-2003 shock VC-backed x Post-2012 shock Top underwriter Revision Stake Size Time to offer Industry return Internet-related Obs R Adj. R Distribution of coefficients of interest with randomized VC assignment Sample Percentile Placebo VC-backed Placebo VC-backed x Placebo VC-backed Placebo VC-backed x Post-2003 shock Post-2012 shock % p-value 5%

53 52