Are low demand IPOs underpriced evidence from biotech IPOs

Are low-demand IPOs underpriced Evidence from biotech IPOs

ARE LOW-DEMAND IPOs UNDERPRICED? EVIDENCE

FROM BIOTECH IPOs

by

SANGKYOO KANG

A dissertation submitted to the Faculty of the Graduate School of the State University of New York at Buffalo

in partial fulfillment of the requirements for the

degree of

Doctor of Philosophy (PhD)

Department of Finance and Managerial Economics

UMI Number: 3213625

3213625 2006

Copyright 2006 by Kang, Sangkyoo

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346 All rights reserved.

by ProQuest Information and Learning Company

Copyright by

SANGKYOO KANG

May 5, 2006

To my parents, my wife, and my Lord God

9 Relation between initial return and price revision for low-demand IPOs 37

12 Determinations of the likelihood of underpricing of low-demand IPOs 52

Appendix 2: Errors in the SDC’s initial filing price and shares data 63

List of Tables

Table 3 Cross-sectional statistics for biotech IPOs, 1990-2004 24Table 4 Variations in the IPO volume and initial returns over time for

Table 5 Statistics excluding hot issue periods for biotech IPOs 27

Table 6 Cumulative average returns over the 12 months after offer for

Table 9 Comparison between zero and positive price revisions for biotech

IPOs 45

Table 10 OLS regressions by an alternative demand measure for biotech

Table 14 Logistic regressions of the likelihood of underpricing on

List of Figures

Figure 4 Graphical relation between underpricing and price revision for

Abstract

We address the unresolved issue of how negative demand information affects the pricing of initial public offerings (IPOs) Using a sample of 557 biotech IPOs over the 1990-2004 period, we show that low-demand IPOs tend to be underpriced and the extent

of this underpricing is significant The post-IPO performance of low-demand IPOs supports the claim that low-demand IPOs are mildly underpriced in the long run and suggests potential negative overreaction in the market for IPOs with low demand We use

a new partial adjustment model to show that the traditional model generally fails to demonstrate little relation between initial return and price revision for low-demand IPOs Our findings are robust to controlling for the measure-specific and industry-specific factors In logistic regressions, we find evidence consistent with the view that underwriters may have difficulty in marketing a larger IPO where investor interest is low and thus are more likely to underprice The underpricing likelihood is negatively related

to pre-IPO market returns, which is consistent with the ‘a bird in the hand’ hypothesis

We also present evidence consistent with the hostile takeover avoidance hypothesis and with the prospect theory, respectively We take the demand-dependent approach to explanation for underpricing and our approach has certain advantages This study complements the information solicitation and price adjustment model in the literature

1 Introduction

Theoretical studies have predicted demand-dependent patterns in the pricing of initial public offerings (IPOs): IPOs for which there is low investor interest will not be underpriced, while those with high demand will be substantially underpriced Yet, the present observations cannot confirm that this is the case Defined as IPOs that are priced below the initial filing price range, low-demand IPOs have an average initial return of 3.4%; Furthermore, the proportion of these low-demand IPOs are greater than that of IPOs with high demand (25 percent and 22 percent, respectively) However, there appears to be little research on the effects of low demand information on the IPO pricing decision In this paper, we attempt to address the unresolved issue of how low-demand IPOs are priced We define a low demand IPO as an IPO that has negative price revision between the filing of initial price range and the offer date

1

2

There are several reasons why research on the effects of negative information (i.e., low demand) on the pricing of IPOs is of interest First, from an investor’s viewpoint, the existence of typical pricing patterns may provide good opportunities for active investing Second, the finding of positive initial returns calls into question the informational efficiency of the IPO pricing process.3 This would provide evidence consistent with underwriters giving reward to investors who provide low interest (i.e., negative information) It is difficult to conjecture why offering that reward is consistent with the efficient IPO market Third, the extent of initial returns depends largely upon the degree

3

There are other findings that may suggest that the IPO pricing process is not efficient See Lowry and

of price adjustment to demand information learned in the premarket.4 If positive initial returns are associated with the price adjustment to negative information, this would indicate that underwriters are overadjusting the offer price to negative information Fourth, IPOs underperformed in the long run.5 If the poor long-run performance is associated with fads or overoptimism, this would indicate that the long-run underperformance of IPOs documented in the literature may be pertaining to the worst underperformance of IPOs with high demand Fifth, given the higher proportion of low-demand IPOs than high-demand IPOs, the research on the pricing of low-demand IPOs would complement the information solicitation and partial adjustment model in the IPO literature

Using a sample of 557 biotech IPOs over the 1990-2004 period, we show that more of IPOs are low-demand IPOs rather than high-demand IPOs The higher proportion of low-demand IPOs found in the sample is not driven by industry-specific factors The volume of low-demand IPOs displays variations over time Low-demand IPOs tend to be underpriced and the extent of this underpricing is significant We find that the positive average initial return on low-demand IPOs is not subject to possible aftermarket price support and periods of hot issue market

We next examine the long-run performance to see if low-demand IPOs are overpriced in the long run We compute the 12-month market-adjusted buy-and-hold returns and find little evidence for long-run underperformance for low-demand IPOs Our

4

A few of many studies of the information solicitation and price adjustment include Benveniste and Spindt (1989), Benveniste and Wilhelm (1990), Hanley (1993), Benveniste, Ljungqvist, Wilhelm, and Yu (2003), Ljungqvist and Wilhelm (2003), and Lowry and Schwert (2002, 2004)

5

The long-run underperformance of IPOs has been widely documented in the literature See Ritter (1991), Hanley (1993), and Loughran and Ritter (1995)

results support the claim that low-demand IPOs are mildly underpriced in the long run and suggest potential negative overreaction in the market for IPOs with low demand

To explore the effects of negative demand information on initial return, we analyze the relation between price revision and initial return We use a new partial adjustment model to regress initial returns on price revision and find that initial return is unrelated to negative price revision This finding is not subject to hot issue periods We reexamine the traditional partial adjustment model commonly used in the literature and show that it generally fails to demonstrate little relation (i.e., an absence of relation) between initial return and price revision for low-demand IPOs.6 To test the robustness of our results, we use proceeds change as an alternative demand measure to re-estimate the regression models We also use a different data set to investigate whether our results are general for all IPOs We collect a broader sample of 3,546 IPOs from various industries over the 1990-1998 period and re-estimate the regression models These tests show that our findings are robust to controlling for measure-specific and industry-specific factors

The positive average initial returns on low-demand IPOs demand an explanation

We take the demand-dependent approach to explanation for underpricing Possible explanations for this underpricing include: (1) put option to renege on the orders (Schultz and Zaman, 1994); (2) negative cascade (Welch, 1992); (3) winner’s curse (Rock, 1986); (4) ‘a bird in the hand’; (5) litigation risk (Tinic, 1988); (6) price stabilization (Schultz and Zaman, 1994); (7) secondary-market demand (Booth and Chua, 1996); and (8) hostile takeover (Brennan and Franks, 1997) Our approach has certain advantages over tha traditional approach First, it explicitly takes into account the level of demand Second,

6

Empirical studies using the traditional partial adjustment model include Lowry and Schwert (2002, 2004),

it provides a clearer analysis of motives for underpricing, depending on the demand level Third, it establishes new testable implications regarding the pricing of IPOs, allowing us

to connect ex ante motives with ex post outcomes

We examine the characteristics of low-demand IPOs that are underpriced We divide low-demand IPOs into two groups, based on whether initial return is positive Low-demand IPOs that are underpriced have, on average, a larger offer size, older age profile, and larger firm size They have a higher level of underwriting allocation to book managers but lower market returns for 2 months before the filing of the registration statement

In other analyses, we attempt to explore the determinants of likelihood of demand IPOs being underpriced We use multivariate logistic regressions to evaluate the effects of the selected characteristics on the likelihood of underpricing We find that price revision is unrelated to the likelihood of underpricing This result is inconsistent with the view that the lower the demand for IPO shares, the greater risks involved in making a low-demand IPO, and the stronger the motivation to underprice This has two possible explanations First, as Busaba, Benveniste, and Guo (2001) suggest, an issuer’s willingness to withdraw an IPO in the face of weak investor demand may reduce the likelihood of underpricing Second, Loughran and Ritter (2002) argue that an issuer would bargain harder over the offer price if the offer price were to be revised down Then, more aggressive bargaining tactics could actually offset the likelihood of underpricing

low-We find that the offer size is positively related to the likelihood of underpricing, which is consistent with the view that underwriters may have more difficulty in marketing a larger IPO where investor interest is low and thus are more likely to make

pricing concessions Further analysis shows that low-demand IPOs with small offer size are not only less likely to be underpriced than those with large offer size, but are also less underpriced We also find that book managers’ underwriting allocation contributes significantly to the likelihood of underpricing This finding is consistent with the view that the lower the demand for IPO shares, the larger the underwriting allocation, and the higher the likelihood of low-demand IPOs being underpriced This is consistent with Corwin and Schultz (2005), who found that price revision is negatively related to a Herfindahl index of underwriting allocations On the contrary, we find that overhang is unrelated to the likelihood of underpricing Neither book-to-market ratios nor leverage appears to be significantly related to the likelihood of low-demand IPOs being underpriced We find no significant effect of book managers’ rank on the likelihood of underpricing

The underpricing likelihood is negatively related to pre-IPO market returns This

is consistent with the ‘a bird in the hand’ hypothesis that the continuing stock market decline will lower the expectation of profits from participating in future underpriced issues, reducing the present value of future expected profits Then, more investors would prefer underpricing in the hand to uncertain profits in the future To induce investors into the offering, more of low-demand IPOs will be underpriced

We attempt to test Brennan and Franks’ (1997) hypothesis that IPO firms use underpricing to avoid hostile takeover Following Field and Karpoff (2002) and Baker and Gompers (2003), we use the fraction of venture capitalists (VCs) on the board of directors as a proxy for an IPO firm’s vulnerability to takeover and CEO turnover and find that the fraction of VCs positively affects the likelihood of low-demand IPOs being

underpriced This result suggests that low-demand IPO firms with larger fraction of VCs

as director on the board are more vulnerable to takeover and CEO turnover and therefore are more likely to underprice to decrease the probability of takeover and CEO turnover This finding provides evidence consistent with the hostile takeover avoidance hypothesis

Using the historical offer price as a reference point, rather than the current initial filing price, that issuing firms anchor on, we present evidence consistent with the prospect theory.7 We find that low-demand IPO firms are more resistant to underpricing when they anchor on the high historical offer price relative to the current offer price We find little evidence for the capital needs hypothesis that young IPO firms are less likely to underprice the shares when the offer price is revised down

The remainder of this paper is organized as follows In section 2, we discuss the pricing of low-demand IPOs, and in section 3 we provide explanations for underpricing low-demand IPOs, followed by empirical implications in section 4 Section 5 describes the sample and data collections Section 6 presents the descriptive statistics, and section 7 examines long-run performance In section 8, we critically review traditional research methodology, and in section 9 we examine the relation between initial returns and price revision Section 10 includes robustness test In section 11, we examine the characteristics of low-demand IPOs that are underpriced, and in section 12 we explore the determinants of the likelihood of underpricing for low-demand IPOs Section 13 concludes

7

See Loughran and Ritter (2002)

2 Pricing of low-demand IPOs

2.1 IPO pricing process

Figure 1 illustrates the time sequence in IPO pricing process IPO pricing begins when a firm files a preliminary prospectus with the SEC that includes an anticipated offer price The anticipated offer price is provided in the form of a price range, and the

midpoint of the initial filing price range (P ) is considered the initially expected offer

price

m

After collecting indications of interest from investors during IPO roadshows, an underwriter revises the offer price to reflect premarket investor interest The dollar value

of price revision is then measured as the dollar difference between the midpoint of initial

filing price range and the offer price (P0) and denoted as $PR:

The total change in IPO price is a sum of the price revision in the premarket and the initial price change on the first trade date We measure the total change in IPO price

Figure 1

Time sequence in IPO pricing process

The figure illustrates the time sequence in which IPO prices are set P m , P0, and P1 represent the midpoint

of initial filing price range, offer price, and first-day closing price, respectively

Initial filing date Offer date First trade date

2.2 Pricing of low-demand IPOs

Benveniste and Spindt (1989) argue that the pricing of IPOs depends largely on information learned during the filing period about demand for the shares In practice, if demand for the issue is low (high), the offer price will be revised down (up) from the midpoint of initial filing price range In keeping with common research practice, we define an IPO with low (high) demand for the shares, or a “low-demand” (“high-demand”) IPO, as an IPO that has negative (positive) price revision between the filing of initial price range and the offer date We also define an IPO with zero price revision as an

“expected-demand” IPO

Alternatively, Hanley (1993) classifies demand for IPO shares, based on the relation of the offer price to the initial filing price range: issues priced below the range are low-demand IPOs with bad information revealed in the premarket; issues priced above the range, high-demand IPOs with good information; and issues priced within the

range, expected-demand IPOs with little or no information In general, however, IPOs make adjustments in both the offer price and size to reflect demand information gathered during the filing period Hanley (1993) finds that changes in the offer price are often accompanied by changes in the number of shares offered Benveniste, Ljungqvist, Wilhelm, and Yu (2003), therefore, view an IPO with a decrease (an increase) in proceeds as a low-demand (high-demand) IPO; An IPO with no change in proceeds can

be then viewed as an expected-demand IPO

Regarding the extent to which the offer price is adjusted to low demand information learned in the premarket, Benveniste and Spindt (1989) suggest that underwriters fully adjust the offer price because no compensation should be given for negative information Loughran and Ritter (2002) also suggest that issuers bargain harder over the offer price (that is, insist on full adjustment) when premarket demand for the shares is unexpectedly weak The full adjustment results in no underpricing for low-demand IPOs.8 Similarly, Benveniste and Wilhelm (1990) predict that when bad information is revealed during the filing period, an IPO will not be underpriced

However, Benveniste, Busaba, and Wilhelm (1996) argue that underwriters have

an incentive to overstate premarket interest so as to persuade investors to purchase shares

at a higher price This incentive seems to be particularly strong when demand for IPO shares revealed in the premarket is lower than initially expected; given higher demand, underwriters would have less of incentive to exaggerate it Krigman, Shaw, and Womack (1999) argue that underwriters anchor on their own initial expectation in setting the offer price The overstatement incentive and anchoring behavior suggest that underwriters do

8

In contrast, Benveniste and Spindt (1989) suggest that underwriters partially adjust to high demand information gathered in the premarket to give reward for good information provided The partial adjustment

Figure 2

Hypothetical patterns of pricing of low-demand IPOs

The figure describes the three alternative patterns of price adjustment and IPO pricing for low-demand IPOs Three patterns of price adjustment are described below according to whether premarket demand information is fully reflected in the offer price Three patterns of IPO pricing are described below according

to the relation of the first-day closing price to the offer price P m , P0, and P1 represent the midpoint of initial filing price range, offer price, and first-day closing price, respectively The dollar value of price revision

($PR) is measured by the dollar difference between the midpoint of initial filing price range and the offer price The initial price change on the first trade date ($IR) is measured by the dollar difference between the

offer price and the first-day closing price The total change in IPO price ($ΔP) is measured by the dollar difference between the midpoint of initial filing price range and the first-day closing price The +/- signs before the denotation represent a direction of price change

-$PR

$IR -$ ΔP -$IR -$PR

In contrast, Kang (2004) suggests that underwriters wish to lock in the nonbinding bids in firm-commitment IPOs Benveniste and Spindt (1989) also note that under a firm-commitment contract, underwriters have an incentive to presell the whole sales The lock-

in and preselling incentive suggests that underwriters overadjust the offer price to low demand information revealed in the premarket Underpricing of low-demand IPOs is then

a manifestation of this overadjustment effect Figure 2 describes three hypothetical patterns of price adjustment and IPO pricing for low-demand IPOs

However, the evidence on the pricing of low-demand IPOs is at least mixed In Table 1, Hanley (1993) found that the average initial returns on IPOs priced below the initial filing price range or low-demand IPOs are not significantly different from zero in the mid-1980s Ritter and Welch (2002) and Loughran and Ritter (2004) also reported similar results for IPO samples during 1980-1989 These findings are consistent with the full adjustment hypothesis In contrast, Hanley and Wilhelm (1995) documented that in a small sample of IPOs during the mid-1980s, low-demand IPOs are overpriced, on average This result is consistent with the underadjustment hypothesis However, Kang (2004) and others found that the average initial returns on low-demand IPOs are significantly positive in the 1990s and early 2000s In addition, Cornelli and Goldreich (2003) reported that the minimum oversubscription at the offer price (defined as total demand at the offer price divided by total supply) is greater than one during 1995-1999 (p 1419) These results provide support for the overadjustment hypothesis

3 Explanations for low-demand IPO underpricing

Our focus here is on the findings of positive initial returns on recent IPOs for which demand is low The evidence of persistent positive initial returns on low-demand IPOs leaves us with a conundrum: Are these findings a just sampling phenomenon, are they a result of incentive compatibility constraint, or are they driven by economically

Table 1

Statistics from previous studies

The table collects the average initial returns and percentage of IPOs from previous empirical studies Initial returns are measured as the percentage difference between the offer price and the first-day closing (or opening) price The sample is divided into Below, Within, and Above, according to a relation of the offer price to the initial filing price range * indicates the volume-weighted average initial returns for the sample period

Sample No of Average initial returns Percentage of IPOs Empirical studies period IPOs Below Within Above Below Within Above

While Hanley (1993) found an absence of underpricing for IPOs that are priced below the initial filing price range in her sample of IPOs between 1983 and 1987, other

researchers have not been able to confirm that this is always the case (See Table 1) As shown in Figure 2, underpricing of low-demand IPOs is not a result of partial adjustment

in the offer price Benveniste and Spindt (1989) argue that investors should not be rewarded for indicating low interest Underwriters expect investors who have negative information to reveal truthfully because by keeping such information to themselves until after offer, investors cannot expect to benefit Since investors have no incentive not to reveal their negative information before offer, low-demand IPOs should not be underpriced to provide a form of reward to those investors who reveal their private information However, in fact, by overadjusting the offer price to negative information,

an underwriter indeed rewards all investors who participate in the offering It is difficult

to conjecture why this over-incorporation of negative information into the offer price is consistent with the informational efficiency of the IPO market This, in turn, suggests that

if low-demand IPOs are to be underpriced, this underpricing should be driven by other reasons than compensation, so the analysis of IPO underpricing needs a more considered approach, depending on the level of investor demand We call this a demand-dependent approach

In this section, we review alternative explanations presented in the literature for underpricing, paying attention to low investor interest We believe that our demand-dependent approach to explanation for underpricing has certain advantages over tha traditional approach First, it explicitly takes into account the level of demand Second, it provides a clearer analysis of motives for underpricing, depending on the demand level Third, it establishes new testable implications regarding the pricing of IPOs, allowing us

to connect ex ante motives with ex post outcomes

Schultz and Zaman (1994) suggest that the put option for IPO investors to renege

on their orders leads to underpricing If the aftermarket price is below the offer price, investors will renege on their purchase in IPO and will instead buy the shares in the aftermarket.9 Investors’ reneging would be of more concern to low-demand IPOs than to high-demand IPOs, because low demand for IPO shares likely discourages subsequent investing and as a consequence, stocks are more likely to trade below the offer price in the aftermarket Therefore, low-demand IPOs are more likely to be underpriced to ensure that the aftermarket price is above the offer price

Welch (1992) argues that the potential for negative cascade leads to underpricing Benveniste, Busaba, and Wilhelm (2002) and Benveniste, Ljungqvist, Wilhelm, and Yu (2003) provide evidence for such information externality in the equity market The negative cascade should be of particularly concern to low-demand IPOs because IPOs with high demand for the shares are less subject to negative cascade So, low-demand IPOs are more likely to be underpriced to prevent a negative cascade from developing in the first place

Rock (1986) argues that the winner’s curse problem leads to underpricing Uninformed investors are likely to receive a larger allocation in a low-demand IPO because informed investors withdraw from the offering, while they are likely to be crowded out by informed investors from a high-demand IPO Therefore, a low-demand IPO should be underpriced to compensate uninformed investors for the bias in the allocation Otherwise, they do not participate in a low-demand IPO Benveniste and Spindt (1989) suggest that informed investors, by contrast, can be effectively induced to

9

IPO investors can renege without legal penalty for several days after the offer date See Schultz and Zaman (1994, p 202) for details

‘take a badly received IPO off the underwriter’s hands’ (p 354) without underpricing By using its leverage to give priority in the allocation of future underpriced issues, the underwriter can expect informed investors indicating low interest to purchase shares in a low-demand IPO even if by doing so they take a loss We argue, however, that a sure profit from underpricing on the current issue is ‘a bird in the hand worth two in the bush,’ because neither the investors nor the underwriter knows for sure which issues are underpriced in the immediate future In this sense, informed investors are also likely to demand a discount in the hand rather than more in the bush in the future if they are to participate in a low-demand IPO

Litigation risk may also lead to underpricing (e.g., Ibbotson, 1975; Tinic, 1988; Hughes and Thakor, 1992; Lowry and Shu, 2002) Litigation risk is likely higher for low-demand IPOs because these IPOs are more likely to be subject to overpricing Hence, low-demand IPOs are more likely to be underpriced on possible future litigation In addition, costly price stabilization efforts in the aftermarket can lead to underpricing (e.g., Schultz and Zaman, 1994; Benveniste, Busaba, and Wilhelm, 1996; Chowdhry and Nanda, 1996; Aggarwal, 2000) Underwriters are more likely to engage in price support activities for low-demand IPOs because of the higher probability of the aftermarket price being below the offer price Therefore, underpricing is more likely to be used for low-demand IPOs to avoid the costs associated with aftermarket price support

Underpricing may be useful to stimulate secondary-market demand for stocks (e.g., Booth and Chua, 1996; Rajan and Servaes, 1997; Reese, 1998; Ellis, Michaely, and O’Hara, 2000; Aggarwal, Krigman, and Womack, 2002; Pham, Kalev, and Steen,

2003).10 It is likely that an increased interest in buying IPO shares in the aftermarket is much demanded especially when premarket investor interest is low Therefore, low-demand IPOs are more likely to be underpriced for the purpose of stimulating secondary-market interest

Brennan and Franks (1997) suggest that underpricing is also useful to achieve the greater dispersion of outside holdings and thereby reduce the probability of hostile takeover Similarly, other researchers (e.g., Burkart, Gromb, and Panunzi, 1997; Pagano and Roell, 1998; Bolton and von Thadden, 1998) suggest that an entrepreneur generally chooses a more dispersed post-IPO ownership structure to avoid excess monitoring and threat of takeover by other shareholders.11 Assuming that the lack of demand for IPO shares is largely attributable to the poor performance of an IPO firm, low-demand IPO firms would be generally more vulnerable to hostile takeover Hence, low-demand IPOs are more likely to be underpriced to secure a more diversified ownership structure.12

(I.1) A low-demand IPO with more negative price revision is more likely to be underpriced

(I.2) A low-demand IPO with larger offer size is more likely to be underpriced (I.3) A low-demand IPO is more likely to be underpriced in the depressed market

(I.4) A low-demand IPO firm that is more vulnerable to hostile takeover is more likely to underprice the shares

(I.5) A low-demand IPO firm with large capital needs is less likely to underprice the shares.

The first implication is that the lower the demand for IPO shares, the greater the risks involved in making a low-demand IPO, and the stronger the motivation to underprice the issue With lower investor interest, low-demand IPO firms will have added incentive to underprice for a favorable aftermarket ownership structure This implication

is contradicted, however, by previous research (e.g., Hanley, 1993; Ljungqvist and Wilhelm, 2003; Lowry and Schwert, 2002, 2004) that found a positive relation between the extent of underpricing and price revision

The second implication is based on an assumption that underwriters may have more difficulty in marketing a larger IPO where investor interest is low and thus are more likely to make pricing concessions Also, the larger the IPO size, the more the number of reneged shares, given the higher probability of investors’ reneging in low-demand IPOs This implication is contradicted, however, by previous studies (e.g., Beatty and Ritter, 1986; Carter, Dark, and Singh, 1998; Bradley and Jordan, 2002; Aggarwal, Krigman, and

Womack, 2002; Lowry and Schwert, 2004; Loughran and Ritter, 2004) that found a negative relation between the extent of underpricing and offer size.13

The third implication is derived from the view that as the market declines, an IPO

is more difficult to make and investors are more likely to demand compensation for participation in the new issue market The cold market can also increase the probability

of investors reneging or negative cascade developing in a low-demand IPO The continuing stock market decline will lower the expectation of profits from participating in future underpriced issues, reducing the present value of future expected profits Then, more informed investors would prefer underpricing in the hand to uncertain profits in the future This implication is contradicted, however, by previous research (e.g., Hanley, 1993; Loughran and Ritter, 2002; Bradley and Jordan, 2002; Lowry and Shu, 2002; Lowry and Schwert, 2004) that showed that the extent of underpricing is positively related to pre-IPO market returns

The fourth implication is to test Brennan and Franks’ (1997) hostile takeover avoidance hypothesis This implication has not yet been tested The final implication is that the larger the issuer’s financing need, the harder it bargains over the offer price, and the less likely the issue is to be underpriced.14 This implication has not been tested directly Instead, assuming that the detail of intended use of IPO proceeds provided in the prospectus proxies for the degree of the issuer’s financing need, the fifth implication is

13

In contrast, Megginson and Weiss (1991) and Michaely and Shaw (1994) find a positive relation between underpricing and IPO size However, Megginson and Weiss (1991) interpret the positive relation as a manifestation of partial adjustment effect, while Michaely and Shaw (1994) find the positive size coefficient statistically insignificant

14

Chemmanur and Fulghieri (1999) and Lowry (2003) highlight the issuer’s capital needs as a motive for going public

consistent with the finding of Leone, Rock, and Willenborg (2003) that underpricing is negatively related to specificity regarding the intended use of IPO proceeds.15

5 Sample and data

Our sample consists of the firm-commitment IPOs of biotech companies between

1990 and 2004.16 We limit our study sample to biotech IPOs due to data collection constraints It is conceivable that this sample restriction may limit the strength of our conclusions However, a number of biotech IPOs that went public mitigate to some extent the limitation and allow a general view regarding the pricing of low-demand IPOs Biotech companies include biotechnology and pharmaceutical companies and medical device manufactures, excluding healthcare services providers; Biotechnology and pharmaceutical companies are those with SIC codes of 2833, 2834, 2835, 2836 (Drugs),

2879 (Agricultural Chemicals), 8071, 8072 (Medical and Dental Laboratories), and 8731,

8734 (Biological Research and Testing Laboratories); and Medical device manufacturers are those with SIC codes of 3826 (Laboratory Analytical Instruments), and 3841, 3842,

3843, 3844, 3845 (Surgical, Medical, and Dental Instruments).17

We collect the initial sample from SDC Global New Issues (henceforth, “SDC”) Biotechnology category We collect additional 17 IPOs from two other sources: Recombinant Capital (“Recap”) U.S Biotechnology IPOs category and SDC VentureXpert (“VentureXpert”) Medical/Health/Life Sciences category We also use the

17

Guo, Lev, and Zhou (2004) do not include those companies of generic drugs, gene therapy, medical

biotech SIC codes (defined above) as a supplementary filter to further collect 37 IPOs

We exclude units/rights, ADRs/ADSs, IPOs with an offer price below $5.00 per share, and IPOs with no CRSP data available We also exclude IPOs made in the OpenIPO®process because of their pricing process We delete 22 firms for misclassifications of units/rights (17) and SEOs (5).18 A series of sample selections leads to a preliminary sample of 689 biotech IPOs

Next, we gather SIC codes of the preliminary sample firms from the forepart of prospectuses Most foreparts of prospectuses after May 1996 are available in the EDGAR, and those not available in the EDGAR are purchased from CCH Washington Service Bureau (henceforth, “WSB”) We then segregate the sample based on whether a firm has the biotech SIC codes Based on a reading of prospectuses and other sources, we reclassify the sample into biotech and non-biotech companies and finally obtain 557 biotech IPOs Appendix 1 describes our classification of biotech companies, and Table 2 describes our sample selection procedure

Recent empirical studies (e.g., Ljungqvist and Wilhelm, 2003; Loughran and Ritter, 2004) have questioned the quality of the SDC data for some key variables, such as founding date and number of shares outstanding Kang (2004) has also found significant errors in the SDC’s initial filing price range Therefore, we try to hand-collect all data directly from prospectuses, if possible.19 We have final prospectuses for all 557 sample firms Most prospectuses after May 1996 are available in the EDGAR, and those not available in the EDGAR are purchased from Recap (www.recap.com) and WSB (www.wsb.com)

18

Ritter (http://bear.cba.ufl.edu/ritter/SDCCOR.PDF) and Benveniste, Ljungqvist, Wilhelm, and Yu

(2003) uncover similar SDC’s misclassifications

19

Ljungqvist and Wilhelm (2003) also hand-collected all data for IPOs between 1996 and 2000

Table 2

Biotech IPO sample selection, 1990-2004

The table describes the procedure for selecting the biotech IPO sample Our sample consists of the commitment IPOs of biotech companies between 1990 and 2004 Biotech companies include biotechnology and pharmaceutical companies and medical device manufacturers; Biotechnology and pharmaceutical companies are those with SIC codes of 2833, 2834, 2835, 2836 (Drugs), 2879 (Agricultural Chemicals),

firm-8071, 8072 (Medical and Dental Laboratories), and 8731, 8734 (Biological Research and Testing Laboratories); and Medical device manufacturers are those with SIC codes of 3826 (Laboratory Analytical Instruments), and 3841, 3842, 3843, 3844, 3845 (Surgical, Medical, and Dental Instruments) We collect biotech IPOs from three sources: SDC Global New Issues (SDC) Biotechnology category, Recombinant Capital (Recap) U.S Biotechnology IPOs category, and SDC VentureXpert (VentureXpert) Medical/Health/Life Sciences category We also use the biotech SIC codes (defined above) as a supplementary filter to further collect 37 IPOs We exclude units/rights, ADRs/ADSs, IPOs with an offer price below $5.00 per share, and IPOs with no CRSP data available We also exclude IPOs made in the OpenIPO® process because of their pricing process We delete 22 misclassified firms Based on the biotech SIC codes and a further reading of prospectuses and other sources, we filter out non-biotech companies See Appendix 1 for details on the classification of biotech companies

VentureXpert Medical/Health/Life Sciences (1990-2004) 2

shocking, given the fact that these variables are widely used in empirical research Appendix 2 discusses the quality problem of the SDC data on these variables

First-day closing prices are collected from the CRSP When the first-day price is not available in the CRSP, we look at the TAQ First-day opening prices are collected from the TAQ and BigCharts.com, if possible; first-day opening prices are not available for 74 firms Data on daily stock returns after IPO are collected from the CRSP, and data

on daily market returns are collected from the CRSP value-weighted NASDAQ index file

Founding dates are obtained from prospectuses, if possible When founding date

is not available in prospectuses, we depend on other EDGAR filing documents (e.g., K) and Factiva® news sources Basically, we follow Ljungqvist and Wilhelm (2003) and Loughran and Ritter (2004) definitions of founding date When dates of incorporation and business inception are different, we choose the earlier For acquisition-related IPOs, such as “roll-ups”, we follow Ljungqvist and Wilhelm (2003) definition

10-The numbers of shares outstanding after offer are obtained from final prospectuses For firms with dual-class shares outstanding, we add up the number of shares of all classes.20 The numbers of shares retained by existing stockholders and purchased by new investors in IPO are collected from the Dilution section in prospectuses Average pre-IPO stock offer prices are collected from the Dilution section

as well The numbers of shares outstanding before offer are calculated as the number of shares outstanding after offer less the number of primary shares offered

Book values (net tangible book values) per share after offer are gathered from the Dilution section in prospectuses Values of dilution per share to investors in IPO are also gathered from the Dilution section The dollar value of dilution per share is measured as

20

See Loughran and Ritter (2004) for more about dual-class shares

the dollar difference between the offer price and the book value per share after offer Data

on total assets, stockholders’ equity, and debt are collected from prospectuses Data on other variables, such as revenue, EPS, and net operating cash flows, are collected from prospectuses as well The per share value before offer is calculated by dividing the aggregate value by the number of shares outstanding before offer

The numbers of shares allocated to book managers are collected from the Underwriting section in prospectuses Underwriter reputation rankings are obtained from

Ritter’s web page (http://bear.cba.ufl.edu/ritter/rank.pdf).21 Information on whether the IPO firm is backed by VCs are gathered from the Management, Certain Transactions, and Principal Stockholders sections in prospectuses The numbers of VCs as director on the board are gathered from the Management section in prospectuses

6 Descriptive statistics for low-demand IPOs

Table 3 provides the cross-sectional statistics of the distribution of initial returns and the average price revisions and initial returns by premarket demand level To classify the demand for IPO shares, we divide the sample into Negative, Zero, and Positive, according to a sign of price revision Alternatively, we also divide the sample into Below, Within, and Above, based on a relation of the offer price to the initial filing price range,

as well as into Decrease, Unch, and Increase, based on a direction of changes in proceeds

Price revision (PR) is measured by the percentage difference between the midpoint of the initial filing price range and the offer price Initial return (IR) is measured by the

percentage difference between the offer price and the first-day closing price

Table 3

Cross-sectional statistics for biotech IPOs, 1990-2004

The table reports the cross-sectional statistics of the distribution of initial returns and the average price revisions and initial returns by premarket demand level for

557 biotech IPOs during 1990-2004 To classify the demand for IPO shares, we divide the sample into Negative, Zero, and Positive, according to a sign of price revision Alternatively, we also divide the sample into Below, Within, and Above, based on a relation of the offer price to the initial filing price range, as well as

into Decrease, Unch, and Increase, based on a direction of changes in proceeds Price revision (PR) is measured by the percentage difference between the midpoint of the initial filing price range and the offer price Initial return (IR) is measured by the percentage difference between the offer price and the first-day

of IPOs ** and *** indicate significance at the 5% and 1% level, respectively

Table 4

Variations in the IPO volume and initial returns over time for biotech IPOs

The table shows the variations over time in the IPO volume and average initial returns by premarket demand level for 557 biotech IPOs during 1990-2004 See Table 3 for the classification of demand * , ** , and *** indicate significance at the 10%, 5%, and 1% level, respectively

Table 4 (continued)

Table 5

Statistics excluding hot issue periods for biotech IPOs

The table reports the statistics of the distribution of initial returns and the average price revisions and initial returns by premarket demand level for biotech IPOs over the 1990-2004 period, excluding 2000-2001 when low-demand IPOs had the highest average initial return See Table 3 for the classification of demand, as well as definitions of variables * and *** indicate significance at the 10% and 1% level, respectively

Approximately 52 percent of the sample has a negative price revision, while 32 percent has a positive price revision This result suggests that more of IPOs are low-demand IPOs rather than high-demand IPOs The alternative classifications of demand also show a higher proportion of IPOs with low demand in the sample This finding is consistent with the findings of previous studies using all IPOs, as shown in Table 1 These results suggest that the higher proportion of low-demand IPOs than high-demand IPOs is not driven by industry-specific factors.

22

We find that low-demand IPOs are at least as likely to be underpriced as they are

to be overpriced or fully priced This result indicates that low-demand IPOs tend to be underpriced The average initial return on low-demand IPOs is 3.7%, which is significantly positive This finding is consistent with the findings of previous research using all IPO samples (See Table 1) By alternative classifications of demand, the average initial returns on low-demand IPOs are between 3.4% and 3.5%, all of which are significantly positive

Ruud (1993) argues that underwriter price support may result in very few observations in negative initial returns So we examine whether the positive average initial return on low-demand IPOs results from possible price stabilization in the aftermarket To avoid the possibility of underwriter actively engaging in price support,

we calculate the offer-to-open returns (IRo) instead, which is measured by the percentage difference between the offer price and the first-day opening price Table 3 shows that the average offer-to-open returns on low-demand IPOs are between 3.7% and 3.9%, depending on the classification of demand, and that they are all significantly positive

22

In a sample of 3,546 IPOs in various industries during 1990-1998, Kang (2004) documented that 43 percent of the sample has a negative price revision, while 41 percent has a positive price revision

These results suggest that the phenomenon of positive average initial return on demand IPOs is robust even after we control for aftermarket price support

low-Table 4 shows the variations over time in the IPO volume and average initial returns There are more low-demand IPOs in each year throughout the sample period, except in 1999 and 2000 when only few low-demand IPOs went public In 2003 none of the IPOs had a positive price revision, while in 1994 and during 2001-2003, none of the IPOs had the offer price above the initial filing price range These results indicate that the high proportion of low-demand IPOs in the sample is not limited to any particular year

The year 1996 recorded the largest number of IPOs made (N=97), and the year 2003, the smallest number (N=7) Overall IPOs had the highest average initial return during 1999-

2000, while low-demand IPOs had the highest average initial return during 2000-2001

Next, we examine whether the positive average initial return on low-demand IPOs

is driven by hot issue periods To avoid the impact of relatively high initial returns in the hot issue periods, we exclude IPOs during 2000-2001, when low-demand IPOs had the highest average initial returns Table 5 shows the results Despite a slight decrease in initial returns, we find that even during non-hot issue periods, low-demand IPOs have a significantly positive initial return, on average Our result is robust to offer-to-open returns These results suggest that the phenomenon of positive average initial return on low-demand IPOs is not limited to hot issue periods

7 Long-run performance of low-demand IPOs

Ritter (1991) shows that IPO firms significantly underperform their matching firms for the three years after offer, suggesting that ‘in the long run, initial public offerings appear to be overpriced’ (p 3) In this section, we examine the long-run performance of low-demand IPOs to see if low-demand IPOs are overpriced in the long run In doing so, we divide the sample into Negative, Zero, and Positive, according to a sign of price revision Alternatively, we also divide the sample into Below, Within, and Above, based on a relation of the offer price to the initial filing price range Table 6 reports the cumulative average raw returns and cumulative average market-adjusted returns over the 12 months after offer, excluding the initial return, for each group We compute the 7-month and 12-month buy-and-hold returns on a calendar-day basis (212-calendar-day and 365-calendar-day, respectively), as well as on a trading-day basis (147-

trading-day and 252-trading-day, respectively) The cumulative raw return for stock i (R ) is defined as:

23

i

raw

1)1(2

, −+

R

where R i,t is the raw return on stock i on event day t and T is either the number of days for

each period (7-month or 12-month) or the number of days before being delisted, whichever is earlier.24 The cumulative market-adjusted return for stock i (R i adj) is defined as:

23

Loughran and Ritter (1995) show that IPO firms underperform their matching firms for a longer period

of time (up to 5 years) However, Brav and Gompers (1997), Brav, Geczy, and Gompers (2000), and Ritter and Welch (2002) find that IPO firms do not significantly underperform a set of nonissung firms matched

by size and book-to-market

24

Three IPOs in our sample were delisted within the 12 months after offer