Research Policy xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Research Policy
journal homepage: www.elsevier.com/locate/respol

Initial public offering and financing of biotechnology start-ups: Evidence

from Japan☆
⁎
Yuji Honjoa, , Sadao Nagaokab
a
Faculty of Commerce, Chuo University, Japan
b
Faculty of Economics, Tokyo Keizai University, Japan

A R T I C L E I N F O A B S T R A C T

JEL classification: This study explores the initial public offering (IPO) and financing of biotechnology start-ups in Japan. Using a
G32 unique data set, we find that biotechnology start-ups initially backed by venture capital (VC) firms and those
M13 originating from universities are more likely to go public within a shorter period. Moreover, we find that neither
L65 staged financing nor syndication by VC firms is associated with higher IPO value relative to investment.
Keywords: Furthermore, we provide evidence that the timing of IPOs does not depend on equity market conditions in the
Biotechnology start-up biotechnology industry, whereas IPO value tends to depend on equity market conditions. We discuss the factors
Initial public offering that explain these findings, which contradict findings in previous studies of VC investments.
Market value
Staged financing
Syndication
Venture capital

1. Introduction
Many believe that the emergence of new technologies spurs future
economic growth. Among the new technologies, biotechnology has
created a paradigm shift in the pharmaceutical and agricultural fields.
For instance, the primary drug discovery approach seems to have
shifted from chemical-based to biotechnology-based ones following the
recent developments in biotechnology.1 This paradigm shift may have a
considerable impact on industrial organization in the pharmaceutical
industry. While large pharmaceutical companies were central to drug
discoveries in the past, the division of labor among biotechnology start-
ups, universities, and pharmaceutical companies has recently become
more important. This is because the drug discovery process is highly
uncertain and includes multiple stages. Small, young firms and uni-
versities often undertake new discovery projects and are likely to play
an important role in providing new drug candidates and drug discovery
technologies. Large pharmaceutical companies may prefer alliances
with universities and biotechnology start-ups for drug discovery. At the
same time, the division of labor enables large pharmaceutical compa-
nies to reduce the high risk in the early stages of drug development and
to increase the supply of new drug candidates by aligning with bio-
technology start-ups, although the risk is ultimately borne by private
equity markets.
As biotechnology start-ups often require large research and develop-
ment (R & D) investment, they need external finance. However, bio-
technology start-ups typically cannot rely on bank loans because of the
uncertainty in R & D and information asymmetries between entrepreneurs
and creditors. To secure external financing for R & D, biotechnology start-
ups rely on equity financing. Many biotechnology start-ups, although not
all do this, expand their access to equity financing by going public; that is,
they issue an initial public offering (IPO). In particular, biotechnology
start-ups that require significant funding for R & D such as clinical research
are compelled to go public just to continue their R & D activities. In this
case, we have the saying “public or perish.”

☆
This study is part of an extended project of a research program entitled “Research on Scientific Sources of Innovations and Economic Impacts of Science” supported by the Japan
Science and Technology Agency. The first author is financially supported by a Grant-in-Aid for Scientific Research (B) (No. 26285060) from the Japan Society for the Promotion of
Science. We are grateful to the participants of the workshop held at Hitotsubashi University and Nagoya University. In particular, we thank Takanori Adachi, Carsten Bienz, Makoto
Hanazono, Yoshiaki Ishii, Daisuke Miyakawa, and the members of our research project for their valuable comments and suggestions. We also thank two anonymous reviewers for their
insightful comments and suggestions. Moreover, we would like to express our appreciation to the directors of biotechnology start-ups and venture capitalists who agree to participate in
our interviews. Needless to say, any remaining errors are the authors’ own.
⁎
Corresponding author.
E-mail address: yhonjo@tamacc.chuo-u.ac.jp (Y. Honjo).
1
Among the best-selling drugs of 2013, for example, seven drugs derived from biotechnology were ranked in the top ten for worldwide sales. For more details, see the Free Daily
Pharma Industry Newsletter website. http://www.fiercepharma.com/special-reports/top-10-pharma-companies-2013-revenue [accessed on February 17, 2015].

http://dx.doi.org/10.1016/j.respol.2017.10.009
Received 6 October 2015; Received in revised form 25 July 2017; Accepted 15 October 2017
0048-7333/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Honjo, Y., Research Policy (2017), http://dx.doi.org/10.1016/j.respol.2017.10.009
Y. Honjo, S. Nagaoka
This study explores the IPO and financing of biotechnology start-
ups, focusing on the role of venture capital (VC) firms. Recent R & D
activities in Japan have largely focused on life sciences, information
science and technology, and energy and environment.2 Undoubtedly,
biotechnology is a core field of life sciences, and many expect it to play
an important role in other fields, such as energy and environment.
Japanese organizations are important contributors to global R & D in
biotechnology. For instance, the US accounted for 41.5% of bio-
technology patent applications, based on the Patent Co-operation
Treaty, and Japan followed with a share of 12% in 2006 (OECD, 2009).
Nevertheless, biotechnology start-ups appear to be relatively inactive in
the Japanese economy, especially compared to those in the US. One
potential source for this gap is the underdevelopment of private equity
markets in Japan. Private equity financing, including VC firms, plays a
critical role in providing funds to high-tech start-ups—especially those
with large R & D investment—to grow until they exit from private
equity markets, either through an IPO or merger and acquisition
(M & A). However, private equity financing is very thin in Japan. In
particular, the scale of VC investments is relatively small, and many VC
firms are subsidiaries of banks and securities companies. These features
may adversely affect the contribution of VC firms to the development of
the biotechnology industry. Investigating the contribution of VC firms
to the IPO and market values of biotechnology start-ups in Japan can
greatly clarify the mechanism of how private equity financing con-
tributes to the development of the biotechnology industry. In this study,
we construct a unique data set of 213 Japanese biotechnology start-ups
founded in the most recent 20-year period, 1995–2014.
This study examines the time to IPO and the market value of equity
(market capitalization) at IPO (hereafter, “IPO value”) by focusing on
VC financing of biotechnology start-ups. Using a survival analysis ap-
proach, we find that biotechnology start-ups initially backed by VC
firms and those originating from universities are more likely to go
public within a shorter period.3 The results also reveal that two in-
vestment practices, often considered value enhancing—staged finan-
cing and syndication, which are very common in VC investments—are
not associated with higher IPO value relative to investment. Further-
more, we provide evidence that the timing of IPOs does not depend on
equity market conditions in the biotechnology industry, whereas IPO
value tends to depend on equity market conditions.
The remainder of this paper is organized as follows. The following
section introduces the research background, including a review of the
relevant literature on the financing of high-tech start-ups and invest-
ment practices of VC firms. Section 3 explains the method used in this
study, with the data described in Section 4. Section 5 presents the es-
timation results for the time to IPO and IPO value. Finally, we discuss
the factors that explain these findings, which contradict findings in
previous studies of VC investments, in light of prevailing theory and
evidence on staged financing and syndication.
2. Research background
2.1. Financing of high-tech start-ups
It is widely recognized that R & D and innovative activities are difficult
to finance in the market (e.g., Himmelberg and Petersen, 1994; Hall, 2002;
Hall and Lerner, 2010). This is due to the features of R & D, which make it
different from ordinary investments. First, the returns to R & D investment
are highly uncertain; therefore, external suppliers of capital, especially
2
According to the Center for Research and Development Strategy (CRDS) of Japan
Science and Technology Agency, R & D expenditures are the highest in life sciences and
clinical research, at more than three trillion yen in fiscal year (FY) 2014. http://www.jst.
go.jp/crds/report/report02/CRDS-FY2015-FR-07.html [accessed on May 23, 2016].
3
In this study, the term “start-ups initially backed by VC firms” or “initially VC-backed
start-ups” refers to start-ups that raised equity from VC firms at the foundation of the
start-ups, and the term “initial VC financing” refers to the investment of VC firms.
2
Research Policy xxx (xxxx) xxx–xxx
banks, are reluctant to provide capital for such investment. As Kamien and
Schwartz (1978) argued, external financing is difficult to obtain without
substantial tangible collateral, and failed R & D projects generally leave
behind few tangible assets of value. Specifically, as Hall (2002) and Hall
and Lerner (2010) emphasize, most R & D spending is the wages and sal-
aries of highly educated scientists and engineers. Second, it is hard for
external suppliers of capital to evaluate the actual quality of R & D activ-
ities. That is, information asymmetries between firms and external sup-
pliers of capital result in adverse selection and moral hazard that constrain
external suppliers’ decision to provide capital.
In addition, some scholars have argued that younger firms are more
likely to have trouble financing R & D in the market (e.g., Müller and
Zimmermann, 2009; Honjo et al., 2014a). This is because such firms
have limited operating histories and lack complementary assets, in-
cluding the know-how and relationships that take many years to de-
velop, although banks often use relationship lending to alleviate pro-
blems associated with information asymmetries. As Müller and
Zimmermann argued, younger firms have less collateral available to
pledge to banks, and the higher default risk of these firms is a further
age-specific impediment to bank loans.
2.2. Role of VCs and IPO performance
Because of the inherent constraints of debt financing, private equity
is an important source of finance for high-tech start-ups. As Gompers
(1995) emphasizes, venture capitalists (VCs) concentrate investments
in younger firms and high-tech industries where uncertainty and in-
formation asymmetries are significant and monitoring is valuable. At
the same time, VCs seek exit strategies for their investments, and an IPO
is undoubtedly one of the most typical exit strategies.
Many studies have emphasized the role of VCs in the IPO process.
Megginson and Weiss (1991), for example, found that VC-backed firms are
significantly younger, have greater median book values of assets, and have
a larger percentage of equity in the capital structure than their non-VC-
backed counterparts. Helwege and Packer (2009) argued that firms con-
trolled by VCs and private equity specialists are less likely to stay private
and that these firms value the IPO as an important exit strategy for outside
equity. Regarding the IPOs of biotechnology firms, Lerner (1994b) ex-
amined the timing of IPOs and private financings by VCs, using a sample
of 350 privately held VC-backed biotechnology firms in the US between
1978 and 1992. The results showed that these firms go public when equity
values are high, and that they employ private financings when equity
values are lower. The findings indicated that the likelihood of an IPO for
biotechnology start-ups increases with favorable market conditions.
In addition, many studies have examined IPO performance. For
instance, Pagano et al. (1998) estimated the effects of an IPO on firm
performance based on accounting measures, such as the return on assets
and capital expenditures, using data on Italian firms. In particular, some
studies have focused on IPO value, which means the market value of
equity at IPO, as well as the determinants of IPO value. For instance,
Deeds et al. (1997) addressed the amount of capital raised from an IPO.
If access to capital is the major goal of going public, then the success of
an IPO is measured by the amount of capital raised by the firm. Deeds
et al. defined IPO value as the total value of capital raised from the IPO
minus the underwriter's fees, and estimated the effects of location, the
number of products, and citations on IPO value, using data on 92
publicly-held firms in the US biotechnology industry.
While these studies provide interesting results, it is important to
note that there is no guarantee that a firm's value will generate suffi-
cient returns for investors, even if this value is high at IPO. From the
investors’ viewpoint, the return on investment matters.4 Several studies
4
Some scholars used the cumulative average market-adjusted return and average buy-
and-hold return post-IPO (e.g., Yung et al., 2008). However, we highlight performance at
IPO, rather than after IPO, and do not employ these measures.
Y. Honjo, S. Nagaoka
have estimated returns to VC funds, using the internal rate of return
(IRR) (e.g., Kaplan and Schoar, 2005; Harris et al., 2014). Given that
the data on the amount and timing of invested capital are not available
for our research, we address how the IPO yields a significant return on
capital invested prior to IPO. In this regard, some biotechnology start-
ups write down their capital prior to IPO to eliminate any accumulated
deficits. Indeed, it is inappropriate to use the capital after such write off
as the amount of total investment to assess the return on investment at
IPO. Additionally, firms usually issue new equity when they go public;
that is, they increase their equity at IPO. In these respects, we should
take into account the capital reductions prior to IPO and the capital
increases at IPO when defining the total equity invested, which is the
denominator of IPO value as firm performance.
2.3. Investment practices of VC firms
VCs must address the uncertainty associated with investment deci-
sions and information asymmetries between entrepreneurs and them-
selves, which often generate agency and monitoring costs. To cope with
uncertainty and information asymmetries, VCs often use staged capital
infusion; namely, “staged financing.” As Gompers (1995) argued,
staged capital infusion enables VCs to reevaluate the firm's project and
to reduce potential losses from bad decisions. Gompers also provided
evidence that the most important factor influencing total venture fi-
nancing is the number of financing rounds the firm has received.
In addition to staged financing, syndication is a widely used in-
vestment practice of VC firms. VC firms use syndication when they
jointly invest in a firm. As Lerner (1994a) argued, syndication is a
mechanism through which VCs resolve informational uncertainty about
potential investments. Lerner emphasizes that syndication leads to a
superior selection of investments, which is referred to as the “selection
hypothesis,” and provided support for this hypothesis using data on VC
investments in the biotechnology industry. Lerner examined the choice
of syndication partners in the first, second, and later rounds, and found
that in the first round, established VCs tend to syndicate with one an-
other; however, later rounds involve less established venture organi-
zations. This indicates that first-round syndication is a useful way to
cope with uncertainty and information asymmetries that are more se-
vere in the first round of financing. In this respect, the effects of syn-
dication differ between the first and later rounds of financing.
Brander et al. (2002) emphasize that returns to syndicated invest-
ments should be lower than those on standalone investments under the
selection hypothesis. This is because, if a VC's assessment yields a high
expected value, the VC accepts the project, as there is little need for a
second opinion. In this case, syndication is not chosen because it is
costly for a VC (i.e., profit dilution). Thus, if improved selection is the
3
Research Policy xxx (xxxx) xxx–xxx
Fig. 1. Percentages of venture capital investments in seed, early, expan-
sion and later stages: Japan and the US. Sources: VEC Yearbook 2015
(Japan), NVCA Yearbook 2015 (US). Notes: The figures in Japan and the
US are in FY 2014 and 2015, respectively.
central motivation for syndication, it follows that the expected values of
syndicated projects should have lower returns than those of standalone
projects. Furthermore, Brander et al. proposed an alternative role for
syndication, referred to as the “value-added hypothesis,” implying that
VCs can add value to the firms in which they invest. Their empirical
analysis using Canadian data showed that syndicated investments have
higher returns, favoring the value-added interpretation.
In addition to the above two hypotheses, Lerner (1994a) examined
later-round syndication based on the incentive for “window dressing.”
That is, VC firms have a strong incentive to participate in syndicated in-
vestment in a firm with a high probability of a successful IPO because the
investment enables VC firms to enhance their reputation in the market,
which is referred to as the “window dressing hypothesis.” Consequently,
VC firms are willing to participate in a syndicated IPO investment in the
later rounds of financing, even if the expected value from participating in
the investment is small. Finally, Lerner (1994a) and Brander et al. (2002)
discussed the risk-sharing hypothesis for syndication because this allows
VCs to support high-risk but high-return investments by pooling risks.
2.4. Biotechnology start-ups and VC firms in Japan
Motohashi (2012) argued that the business model in the bio-
technology industry in Japan differs from that in the US and such dif-
ferences can be explained by the differences in VC markets between
Japan and the US. To better understand VC investments in the bio-
technology industry in Japan, we discuss the features of VC investments
and compare them to those in the US below.
First, the scale of VC investments in Japan is small. According to the
VEC Yearbook 2015 (edited by the Venture Enterprise Center), VC in-
vestments in biotechnology in Japan amounted to 12770 million yen
(106 million USD, 1 USD = 120 yen) and accounted for 18.1% of total
VC investments in fiscal year (FY) 2014. In contrast, according to NVCA
Yearbook 2015 (edited by the National Venture Capital Association),
VC investments in biotechnology in the US amounted to 5970 million
USD—over 50 times greater than those in Japan—and accounted for
12.1% of total VC investments in 2014. Thus, the scale of VC invest-
ments is much smaller in Japan than in the US.
Second, VC firms tend to invest in the early stages of start-up firms.
Fig. 1 shows the distribution of VC investments in each stage in Japan
and the US. In Fig. 1, the amount of VC investments in seed and early
stages accounts for 57% in Japan, which is considerably higher than
that in the US, 34%.5 OECD (2015) also provides evidence that pre-seed
5
Dibner et al. (2003) argued that because of the severe market conditions in 2002,
several VCs in the biotechnology industry eschewed riskier early-stage financing and
expressed more interest in late-stage start-up firms.
Y. Honjo, S. Nagaoka
and seed stage investments represent a considerable portion of total VC
in Japan as well as in Denmark, Israel, Portugal, and Slovenia. In fact,
we show later that more than 20% of IPO firms in our data set were
initially backed by VC firms, that is, VC investments for their founda-
tions. VC firms in Japan tend to invest in the early stages because pri-
vate equity financing from individual investors is limited. For instance,
according to a survey conducted by the Japan Bioindustry Association
(JBA) in 2012, a biotechnology start-up's capital raised from individual
investors at start-up was on average 0.9 million yen, while the capital
raised from VC firms was on average 20.7 million yen (Honjo et al.,
2014b).6 These features indicate that biotechnology start-ups often rely
on VC firms as early-stage investors rather than individual investors.
From a different perspective, VC firms in Japan deliver risk money that
would have been delivered by individual investors (“business angels”)
in the US.
Third, M & As are much less common as a strategic exit in Japan
than in the US. According to the VEC Yearbook 2015, there were 36
M & As and 116 IPOs among the exits in FY 2014 in Japan, showing that
M & A exits are about one third of IPO exits.7 In contrast, according to
the NVCA Yearbook 2015, there were 459 M & As and 115 IPOs among
the exits in 2014 in the US; therefore, M & A exits are approximately
four times more frequent than IPO exits. Thus, an IPO is relatively
important for investors in Japan, including VC firms that seek a suc-
cessful exit. This implies that focusing on IPOs enables us to capture a
large majority of successful exits in Japan.
As discussed, previous studies based on US and Canadian data found
a positive effect of equity market conditions on the likelihood of an IPO
and a positive association between investment practices, such as staged
financing and syndication, and market value (e.g., Lerner, 1994a,b;
Brander et al., 2002). However, the environmental features for bio-
technology start-ups and VC investments in Japan seem to differ from
those in the US; therefore, they may have different effects on the be-
havior and performance of both biotechnology start-ups and VC firms.
In Japan, the scale of VC investments is small, but VC firms tend to
invest in the early stages. Specifically, in Japan, where private equity
markets are underdeveloped, so that bridge financing is not easily
available, biotechnology start-ups may go public, regardless of equity
market conditions, to secure funding from public equity markets for
continuing their R & D activities. Thus, biotechnology start-ups may not
be able to wait for a market recovery, even if they expect equity market
conditions to improve.
The first research focus of this study is to examine the time to IPO,
especially, whether the timing of IPOs is associated with equity market
conditions. Following this, our second research focus is to examine how
these investment practices predict IPO value, which would help clarify
the role of VCs in high-tech start-ups in Japan and in general.
3. Method
3.1. Time to IPO
First, we use a survival analysis approach to investigate the types of
firms go public in a short period, as well as the association between
equity market conditions and the timing of IPOs. Consider the prob-
ability of firm i's IPO at t, which determines the duration from founding
to IPO (hereafter, “time to IPO”). Let Ti* denote firm i's time to IPO.
6
Using data provided by the Global Entrepreneurship Monitor, Honjo (2015) also
found that among 32 countries, Japan has the lowest level of new business investment by
individuals, as measured by the ratio of individuals who have personally provided funds
for a new business started by someone else in the last three years.
7
The Venture Enterprise Center classifies exits into (i) IPO, (ii) M & A (with transfer of
management rights), (iii) trade sales (to secondary funds), (iv) wire-off/liquidation, (v)
share buyback (to managers), and (vi) others. Among these exits, IPO, M & A, trade sales,
wire-off/liquidation, and share buyback accounted for 17%, 5%, 19%, 10%, and 40% in
FY 2014, respectively.
4
Research Policy xxx (xxxx) xxx–xxx
However, we cannot observe Ti* for all firms because many firms have
not yet gone public during the observation period. Furthermore, while
some firms go public during the observation period, others are forced
out of business through bankruptcy.8 More precisely, Ti* is observable if
a firm goes public up to the censoring time, denoted by Ci, and the firm
is not forced out of business; otherwise, Ti* is unobservable. We can
∼ ∼
define observable time, Ti, such that Ti = min{Ti*, Ti , Ci} where Ti is a
competing event (event other than an IPO).
The probability of firm i's IPO is captured by a hazard, h(t;Zit), which
is a function of a vector of covariates (variables), Zit. Based on this
hazard, we can use Cox's (1972) proportional hazards model to identify
the determinants of an event. However, as discussed, some firms are
forced out of business through bankruptcy before going public. Without
considering these firms, we would produce results with a selection bias.
Therefore, we regard them as competing risks when estimating the
determinants of IPOs.
Instead of the standard proportional hazards model, we employ a
competing-risks regression and define the sub-hazard, h (t; Zit ) .9 The
specification, which is similar to the proportional hazards model, is as
follows:10
h (t; Zit ) = h 0 (t )exp(αZit ), (1)
where h 0 (t ) is the baseline sub-hazard and α is the parameter to be
estimated. By maximizing the log-pseudo-likelihood function proposed
by Fine and Gray (1999), we obtain the parameter estimates, α̂ , which
indicate whether the covariate shortens the duration of an IPO. Note
that the time to IPO indicates firm age at IPO. Even for a given age,
firms face different market conditions, according to their founding
years; therefore, we control for cohort effects in the estimation.
3.2. IPO value
Next, we investigate how investment practices predict firm perfor-
mance measured by IPO value using a cross-sectional analysis. Since we
cannot observe market values for privately held firms, we estimate the
determinants of IPO value for firms that went public during the ob-
servation period. However, many firms remain private, while some
conduct an IPO. In particular, a firm initially backed by VC firms may
be more likely to adopt various investment practices and is simulta-
neously more likely to choose IPO timing for higher market capitali-
zation. Therefore, we employ a type 2 Tobit model (or sample selection
model) while accounting for sample selection bias.
Let Yi denote the market value of equity (or the market value re-
lative to total equity invested) for firm i at IPO. We assume that this
market value is a linear function of a vector of variables, Xi. Hence, we
write this as follows:
Yi = β0 + β1 Xi + ui , (2)
where β0 and β1 are the parameters to be estimated, ui is an error term,
and ui ∼ N(0, σ2). For Eq. (2), however, Yi can be observed if
γ0 + γ1 Wi + vi > 0, (3)
where Wi is a vector of variables, γ0 and γ1 are parameters to be esti-
mated, vi is an error term, vi ∼ N (0, 12 ) , ui and vi have correlation ρ, and
λ = ρσ. Otherwise, Yi is unobservable. Among the parameter estimates,
β̂1 indicates whether the variable is associated with the market value of
8
In practice, several firms may successfully exit through acquisition by public firms.
Unfortunately, we cannot identify such firms in the data set; thus, we treat all exit cases
other than IPOs as a competing event. However, M & As are a less common exit strategy in
Japan.
9
Giot and Schwienbacher (2007) examined time-to-exit through IPO, trade sale, and
liquidation for VC-backed firms, using a sample of investment rounds. They used a
competing-risks regression in their estimation to allow for a joint analysis of exit type and
exit timing.
10
For more details on the competing-risks regression, see, for example, Fine and Gray
(1999).
Y. Honjo, S. Nagaoka
equity for biotechnology start-ups.
To estimate parameters, we can use Heckman's (1979) two-step
procedure or maximum likelihood estimation (MLE). While MLE is
more efficient than the two-step procedure under joint normality of ui
and vi , it is less robust than the two-step procedure and sometimes it is
difficult to get the problem to converge. In contrast, the two-step pro-
cedure has advantages in that the assumption of joint normality can be
relaxed and the distribution assumptions can be weakened (e.g.,
Cameron and Trivedi, 2005; Wooldridge, 2010). For these reasons, we
estimate the parameters using the two-step procedure.11
For the variables Xi in Eq. (2), we focus on the investment practices
of VC firms—specifically, staged financing and syndication, which
will be explained in more detail later—because these investment
practices not only affect the availability of equity for VC firms, but
also contribute to IPO value through the screening or value-added
activities of VC firms, as suggested by the above literature review. In
addition to investment practices, we control for equity market con-
ditions at the time of IPO for each firm. However, this specification
leaves out some important variables affecting IPO value, including the
quality of a firm, such as seeds for the firm and management cap-
ability, which could also affect the investment practices of VC firms.
For instance, a firm with high management capability may be asso-
ciated with less syndication because the firm requires less manage-
ment support from VC firms. While instruments are a standard solu-
tion to such an endogeneity issue, there are no adequate instruments
available in our data set. Thus, the estimation results do not allow us
to test the causal relationship between investment practices and IPO
value, but to show how these investment practices are associated with
IPO value.
4. Data
4.1. Data source
To construct the data set of biotechnology start-ups, we obtained a
list of firms from the JBA, which is a general incorporated foundation
that promotes development in the biotechnology industry and that
surveys biotechnology firms in Japan. However, since the JBA does not
usually provide data on the financing of biotechnology firms, especially
from VCs, we cannot identify how these firms are financed and which
ones are backed by VC firms from the data source. To collect data on the
financing of biotechnology firms, we use a database provided by Japan
Venture Research (JVR).12
We construct a data set of biotechnology start-ups by matching
firms listed in the JBA and JVR databases, and obtain data mainly from
the JVR database because it contains the required financial data for our
analysis.13 In addition to these data sources, we use the listing pro-
spectus, annual securities reports, and websites of each firm to identify
the firm's profile. Moreover, we use Kabushiki Kokai Hakusho (White
Paper on Initial Public Offerings) edited by Disclosure Jitsumu Ken-
kyukai (Pronexus Inc.) to collect data on the timing of IPOs and the
accompanying capital increases. Furthermore, we obtain data on stock
prices, equity values, and stock market indices from the Nikkei Eco-
nomic Electronic Databank System–Financial Quest.
To determine the observation period, Fig. 2 describes the trend in
11
For precise identification, we exclude at least one regressor in the selection equation
from the outcome equation when estimating the type 2 Tobit model (e.g., Cameron and
Trivedi, 2005). In our model, regressors in the selection equation (Eq. (3)) are not in-
cluded in the outcome equation (Eq. (2)).
12
JVR (Corporation) is a research and consulting company that surveys not only VCs
but also start-ups backed by VC firms or other corporations. We accessed a website
provided by JVR from December 2014 to April 2015. It is important to note that in-
formation on the website is regularly updated. For more details, see the following web-
site. http://entrepedia.jp/.
13
However, the JBA database regards a few firms listed in the JVR database as no
longer engaged in biotechnology, and we excluded these firms from the data set.
5
Research Policy xxx (xxxx) xxx–xxx
the number of firms in the Japanese biotechnology industry.14 We
found that the number of firms increased after the mid-1990s, reached
its peak in the 2000s, and decreased in recent years. The JVR database
includes several firms founded quite some time ago. Therefore, we set
20 years for biotechnology “start-up” and excluded firms older than 20
years from the data set. Although generally 20 years seem a rather long
period for the start-up stage, we adopt this window because it often
takes a long time to develop new drugs. As a result, the data set com-
prises biotechnology firms founded from 1995 to 2014. As shown in
Fig. 2, this period covers the boom times for new entrants in the Ja-
panese biotechnology industry.
As an additional data-screening step, we exclude firms that are
contract research organizations (CROs), site management organizations
(SMOs), and foreign firms with head offices located outside of Japan
from the data set. As a result, we define a biotechnology start-up as a
firm founded during the period of 1995–2014 in the biotechnology
industry in Japan, except for CROs and SMOs.
4.2. Descriptive statistics
The data set includes 213 biotechnology start-ups in Japan, of
which 33 firms (33/213 ≃ 15%) have experienced an IPO and 180 firms
have not (to December 2014).15
Fig. 3 displays the trend in the number of IPOs in each market for
the 33 IPO firms. As shown in Fig. 3, none of the IPO firms were listed
in the established stock exchanges, such as the Tokyo Stock Exchange
(TSE), but in “emerging (stock) markets”—JASDAQ, MOTHERS, Cen-
trex, and Tokyo Pro Market (formerly, Tokyo AIM)—which provide
equity financing to young and emerging small- and medium-sized firms.
This is mainly because the standard accounting criteria for financial
performance, such as profits, do not give appropriate measures in these
emerging markets. Indeed, without the emerging markets, the majority
of IPO firms would not have gone public because they had negative
profits. In the data set, 22 firms had negative operating profits in their
post-IPO accounting year, although we could not obtain the financial
statements of one firm (Ribomic) that went public in September 2014.
In this respect, the emerging markets in Japan are helpful in promoting
IPOs of biotechnology start-ups.
Surprisingly, as shown in Fig. 3, there is a steady flow of IPOs in the
Japanese biotechnology industry. The emerging markets, such as MO-
THERS and Centrex, were created in the late 1990s, and two university-
based start-ups (Anges MG and Trans Genic) went public in 2002. Since
then, biotechnology start-ups have continued to go public in the
emerging markets. More importantly, the economic recession triggered
by the Lehman Brothers collapse in 2008 resulted in depressed stock
markets in Japan. According to Kabushiki Kokai Hakusho, the number of
IPOs in all of Japan's stock markets considerably decreased after the
Lehman Brothers collapse, and this was down to fewer than 20 in 2009
for all industries. Nevertheless, as Fig. 3 demonstrates, there was no
dramatic decrease in the number of IPOs in the biotechnology industry
during this period. Although biotechnology start-ups faced unfavorable
market conditions, some might be forced to go public, partly because of
the lack of alternative private equity financing. It is possible that IPOs
of biotechnology start-ups are quite independent of the market condi-
tions in Japan.
Table 1 presents the trend in the number of biotechnology start-ups
founded and the ratio of IPO firms in each founding year. As we also see
14
Fig. 2 indicates the number of firms, based on the JBA's surveys, regardless of age.
Note that the definition of biotechnology start-ups in the JBA survey differs from our
definition, since the JBA survey excludes young but large firms. Therefore, Fig. 2 has
more firms than our data set. For more details on the JBA's surveys, see, for example,
Honjo et al. (2014b).
15
Although three of the 33 firms (LTT Bio-Pharma, ECI, and Mebiopharma) were
privately held in December 2014 by going private after the IPO, we regarded these three
firms as IPOs in the analysis.
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Fig. 2. Number of firms in the biotechnology industry of Japan. Source:

Japan Biotechnology Association 2014-nen Bio Venture Tokei Doko Chosa
Hokokusho (2014 Survey on Bio-Ventures).

Fig. 3. Number of IPOs in the biotechnology industry of Japan. Source:

Kabusihi Kokai Hakusho (White Paper on Initial Public Offerings). Notes:
JASDAQ includes NEO and JASDAQ Growth. Tokyo Pro indicates “Tokyo
Pro Market” and includes the former Tokyo AIM.

Table 1 in Fig. 2, the number of biotechnology start-ups (A + B) increased from

Distribution of biotechnology start-ups: founding year. the late 1990s to the early 2000s. In this respect, policies promoting
start-ups from universities may have induced IPOs of biotechnology
Founding year IPO (A) Non-IPO (B) Total (A + B) Ratio (A/(A + B))
start-ups.16 However, the number of biotechnology start-ups decreased
1995 1 2 3 0.333 after the late 2000s, and the low level of new entrants in recent years is
1996 1 0 1 1.000 of increasing policy concern in developing the biotechnology industry
1997 1 1 2 0.500 in Japan. Meanwhile, biotechnology-startups founded in the late 1990s
1998 1 2 3 0.333
1999 5 8 13 0.385
have a higher probability of going public (A/(A + B)). Since the
2000 2 17 19 0.105 emerging markets in Japan, such as MOTHERS and Centrex, were in-
2001 5 16 21 0.238 troduced in the late 1990s, the early biotechnology start-ups could go
2002 3 17 20 0.150 public in these emerging markets.
2003 5 20 25 0.200
Table 2 presents the summary statistics for time to IPO, defined as
2004 4 22 26 0.154
2005 3 12 15 0.200 the number of months from founding to IPO plus one, indicating firm
2006 1 19 20 0.050 age at IPO. The time to IPO for the 33 IPO firms is on average 86
2007 0 10 10 0.000 months (about 7 years), although this value is restricted to bio-
2008 1 2 3 0.333 technology start-ups that have already gone public. Table 2 also pre-
2009 0 4 4 0.000
2010 0 8 8 0.000
sents the summary statistics for time to IPO for each five-year period,
2011 0 8 8 0.000 according to firms’ founding year. While the ratio of IPO firms founded
2012 0 4 4 0.000
2013 0 6 6 0.000
16
2014 0 2 2 0.000 With regards to access to university technologies, a technological license organi-
zation (TLO), which plays an intermediary role between universities and industry, was
Total (over time) 33 180 213 0.155
introduced in 1998 to promote technology transfer from universities. In addition, the
Ministry of Economy, Trade and Industry in Japan enacted a policy entitled, “The 1000
Notes: “IPO” and “non-IPO” represent biotechnology start-ups that have experienced or
University-launched Ventures Plan” in 2001 to promote industry–academia–government
not experienced an IPO by December 2014, respectively. collaboration.

6
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Table 2 we could not obtain data for the capital increase and the number of
Summary statistics for time to IPO. financing rounds for this firm from the data sources. The median IPO
value is approximately 14 billion yen, and the mean is approximately
Founding year Time to IPO (in months)
29 billion yen. In Table 4, we present equity financing prior to IPO (i.e.,
Mean SD 25% Median 75% N private equity financing), including the total paid-in capital (paid-in
capital plus additional paid-in capital) but not subtracting capital re-
1995–1999 89.1 35.9 60.0 97.0 107.0 9
duction. Although capital reduction decreases equity, it should not be
2000–2004 86.9 41.4 44.0 105.0 125.0 19
2005–2009 75.6 19.1 80.0 83.0 84.0 5 subtracted from private equity financing. In other words, Table 4 pro-
All 85.8 36.8 58.0 89.0 117.0 33 vides gross financing by returning capital reduction to private equity
financing. Additionally, because firms usually issue new equity when
Notes: “Time to IPO” is defined as the number of months from founding to IPO plus one. they go public, we also present IPO capital increase, which indicates the
SD indicates standard deviation. N indicates the number of firms.
capital increase at IPO. Table 4 shows that the median private equity
financing is approximately 1.8 billion yen (the mean is approximately
Table 3 3.6 billion yen) and the median IPO capital increase is approximately
Distribution of biotechnology start-ups: IPO/Non-IPO and initially VC-backed/others.
2.1 billion yen (the mean is approximately 2.4 billion yen). Moreover,
Initially VC-backed (A) Others (B) Total (A + B) Table 4 presents the ratio of IPO value divided by private equity fi-
nancing prior to IPO (A/B), in addition to the ratio of IPO value minus
IPO (C) 7 26 33 IPO capital increase, divided by private equity financing ((A − C)/B).
Non-IPO (D) 10 170 180
The latter ratio proposed (hereafter, “relative IPO value”) is a useful
Total (C + D) 17 196 213
proxy to capture return on investment at IPO. Using this indicator, we
χ2 = 9.308 (p < 0.01)
find that the median return on investment in a biotechnology start-up
Notes: “Initially VC-backed” represents biotechnology start-ups that raised equity from VC
that goes public is more than five times (the mean return is over fifty
firms when they were founded. “IPO” and “non-IPO” represent biotechnology start-ups times).17
that have experienced or not experienced an IPO by December 2014, respectively. Additionally, Table 5 presents the time to IPO by dividing the data
set into initially VC-backed start-ups and others. Using data on VC-
in the late 1990s is higher, as shown in Table 1, the time to IPO does not backed and non-VC-backed firms, Lerner (2009) showed that VC-
seem to differ according to the firm's founding year. backed firms tend to be considerably younger when they go public.
It is important to note that all 33 IPO firms raised equity from VC Lerner found that the time to IPO was on average 86 months for VC-
firms before going public. This indicates that biotechnology start-ups backed pharmaceuticals and 178 months for non-VC-backed firms. As
backed by VC firms are more likely to go public, although, as Lerner shown in Table 5, the time to IPO is on average 84 months in Japan,
(1994a) suggested, it is possible that some VC firms have invested in the although we cannot calculate the time to IPO for non-VC-backed firms
later rounds of financing for firms with a high probability of going because all 33 IPO firms raised equity from VC firms before going
public, so as to enhance their reputation. To discriminate such late public. Table 5 also shows IPO values for initially VC-backed start-ups
participation, Table 3 describes the distribution of biotechnology start- and others. The median IPO value for initially VC-backed start-ups is
ups by dividing the data set into IPO firms (public firms) and non-IPO approximately 15 billion yen. While the mean for initially VC-backed
firms (private firms), and shows the number of initially VC-backed start-ups is much larger than that for others, the median is not so dif-
start-ups. We find that initially VC-backed start-ups account for more ferent. Moreover, the median relative IPO value for initially VC-backed
than 20% (≃7/33) of IPO firms in the data set. According to Table 3, start-ups is 12.9.
while the share of initially VC-backed start-ups in the entire sample is Table 6 describes the number of financing rounds for IPO firms.18
low (17/213 ≃ 8%), the IPO propensity of initially VC-backed start-ups We find that biotechnology start-ups in Japan, in both the mean and the
(7/17 ≃ 41%) is much higher than that of others (26/196 ≃ 13%). median, have approximately five financing rounds. Table 6 also pre-
There is an influential view that VC firms rarely invest in high-tech- sents the number of VC firms from the founding through the first round
nology firms in Japan (e.g., Black and Gilson, 1998). However, the of financing (number of first-round VC firms), in addition to the total
above evidence shows that VC firms have a role in providing funds to number of VC firms prior to IPO (number of VC firms) because the
biotechnology start-ups, especially as early-stage investors, partly be- effects of syndication on IPO value may differ between the first and
cause equity financing from individual investors is quite thin in Japan. later rounds of financing. The mean (median) number of VC firms prior
We measure IPO performance, using not only market value but also to IPO is approximately 14 (13), and that of first-round VC firms is 3.3
a relative measure for IPO value from the perspective of return on in- (2.0) in Japan. Lerner (1994a) found that the mean number of venture
vestment. Table 4 presents IPO values in the data set. It is important to investors is 2.2 in the US, which suggests that syndicated investments in
note that we exclude one firm (Mebiopharma), which went public in the Japan are much larger than in the US.
Tokyo Pro Market, and employ 32 firms to measure IPO values because Furthermore, Table 7 presents the market value and financing of
equity for IPO firms, according to the number of financing rounds and
first-round VC firms, respectively. Surprisingly, the mean and median of
Table 4
Market value and private equity financing for IPO firms.
IPO value are higher for biotechnology start-ups with fewer financing
rounds. Likewise, they are higher for those with fewer first-round VC
Variable Mean SD 25% Median 75% N firms. These findings indicate that IPO value may actually decrease

IPO value (A) 28621 38029 8032 13978 26660 32

Private equity financing (B) 3601 5003 837 1751 4687 32
IPO capital increase (C) 2359 1873 819 2100 2942 32 17
Table 4 indicates that the variances of relative IPO value, in addition to IPO value,
A/B 57.8 182.7 1.8 6.2 23.0 32 are high because only a few firms are highly evaluated in the markets. As shown later,
(A − C)/B 55.3 176.9 1.4 5.1 20.6 32 therefore, we use the logarithm form of IPO and relative IPO values in the estimation.
18
In the JVR database, a capital increase through a third-party allocation is defined as
Notes: “IPO value” represents the market value of equity (million yen), based on the first- a financing round. Note that a capital increase through the allocation of equity to VC
day closing price. “Private equity financing” represents invested equity defined as the sum firms and other corporations is counted, whereas a capital increase only through the
of paid-in capital, additional paid-in capital, and capital reduction (million yen) prior to allocation of equity to individuals and sole proprietorships is not counted as a financing
IPO. “IPO capital increase” represents the capital increase (million yen) at IPO. Among round. In addition, multiple capital increases within three months are counted as one
the 33 IPO firms, we employ 32 firms to measure the above variables. financing round.

7
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Table 5 involvement of VC firms at the foundation of biotechnology start-ups is

Time to IPO and IPO values: initially VC-backed/others. related to the time to IPO. In this study, we hypothesize that bio-
technology start-ups initially backed by VC firms are more likely to go
Type Mean SD 25% Median 75% N
public earlier than other start-ups. We use a dummy variable for initial
Time to IPO (in months) VC financing (VC_ST) to represent the initial involvement of VC firms at
Initially VC-backed 84.0 38.9 34.0 83.0 125.0 7 founding.19
Others 86.3 37.0 58.0 89.5 117.0 26
We also include a dummy variable for university base (UNIV) in the
IPO value (in million yen)
Initially VC-backed 51622 64597 8609 14679 142728 7 regression. While scientific development spreads rapidly at the inter-
Others 22180 25046 7455 13312 22102 25 national level, as argued by Goto (2000), universities still play a vital
(IPO value − IPO capital increase)/private equity financing role in their respective countries as a local source of scientific knowl-
Initially VC-backed 64.6 107.1 1.1 12.9 128.7 7 edge, since important know-how for commercialization of scientific
Others 52.7 193.7 1.5 5.1 12.9 25
knowledge is often embodied in scientists. As is well-known, scientific
Notes: “Time to IPO” is defined as the number of months from founding to IPO plus one. knowledge is the most important asset of biotechnology start-ups. In
“IPO value” represents the market value of equity (million yen), based on the first-day fact, start-ups that originated from technologies developed in uni-
closing price. “Private equity financing” represents invested equity defined as the sum of versities or persons associated with universities (hereafter, “university-
paid-in capital, additional paid-in capital, and capital reduction (million yen) prior to based start-ups”) account for approximately 39% of biotechnology
IPO. “IPO capital increase” represents the capital increase (million yen) at IPO. “Initially start-ups in our data set. In this study, we examine whether bio-
VC-backed” represents biotechnology start-ups that raised equity from VC firms when
technology start-ups originating from universities are more likely to go
they were founded. SD indicates standard deviation. N indicates the number of firms.
Among the 33 IPO firms, we employ 32 firms to measure the above variables.
public earlier than other start-ups.
Additionally, we include a dummy variable for spin-offs (SPINOFF)
to indicate whether the firm is a spin-off and to examine whether spin-
Table 6
Number of financing rounds and VC firms. offs are more likely to go public earlier than others.20 We obtained data
for these variables from the JVR database.
Variable Mean SD 25% Median 75% N Moreover, the likelihood of an IPO may depend on equity market
conditions, such as “hot issue” markets, in which shareholders, in-
Number of financing rounds 4.9 2.6 3.0 5.0 6.5 32
Number of VC firms 14.3 8.4 9.0 12.5 19.0 32 cluding VC firms, expect higher capital gains. The literature on finance
Number of first-round VC firms 3.3 3.6 1.0 2.0 4.5 32 has argued that the IPO market exhibits dramatic swings in issuance
that are often referred to as hot and cold markets (e.g., Ibbotson and
Notes: SD indicates standard deviation. N indicates the number of firms. Among the 33 Jaffe, 1975; Ritter, 1984; Helwege and Packer, 2004). To measure
IPO firms, we employ 32 firms to measure the above variables.
market conditions, we use a variable for the JASDAQ index (JINDEX),
which is a market capitalization-weighted index provided by the TSE
Table 7
and is based on all domestic common stocks listed on JASDAQ. The
Market value and private equity financing for IPO firms by the number of financing
rounds and first-round VC firms.
JASDAQ index was calculated as 100 on December 28, 1991. This index
was 55.25 at the end of January 1995 and 104.95 at the end of De-
Variable/type Mean SD 25% Median 75% N cember 2014.21 JASDAQ is one of the major emerging markets in Japan
and indeed, as shown in Fig. 3, some biotechnology start-ups are listed
IPO value
Number of financing rounds
on JASDAQ. The values of this variable differ across firms, according to
1–3 rounds 48280 44485 14644 32804 78877 11 the month and year of the firm's founding. While this variable is time
4–7 rounds 20038 33608 7455 10467 17408 17 variant, the others are time invariant in the estimation of the sub-ha-
8+ rounds 11036 8765 5817 8794 16255 4 zard of IPO.
Number of first-round VC firms
Furthermore, as shown in Table 1, the likelihood of an IPO differs
0–3 VC firms 35667 45569 7140 16026 41152 20
4–7 VC firms 10296 4661 7349 10412 12965 8 according to the firm's founding year, which may suggest the existence
8+ VC firms 30039 22651 13545 26220 46532 4 of cohort effects on the time to IPO. Therefore, the two cohort dummies
(IPO value − IPO capital increase)/private equity financing for the founding year (C95_99 and C00_04) are included in the re-
Number of financing rounds gression.22 Finally, the industry dummies (I_DRUG and I_HEALTH) are
1–3 rounds 137.6 290.0 2.0 21.7 120.7 11 included in the regression.
4–7 rounds 14.1 30.7 1.9 3.0 11.7 17 Next, we discuss the variables used in the estimation of IPO value.
8+ rounds 3.9 6.0 0.8 1.1 7.1 4
For the selection equation, as specified in Eq. (3), we use the above
Number of first-round VC firms
0–3 VC firms 84.4 220.6 1.4 7.1 45.0 20
4–7 VC firms 6.8 6.9 2.2 2.9 12.3 8
8+ VC firms 6.5 10.1 1.3 1.6 11.8 4 19
We could not obtain the timing data from the JVR database for individual financing
of VC firms for IPO firms and non-IPO firms, aside from initial VC financing.
Notes: “IPO value” represents the market value of equity (million yen), based on the first- 20
In the JVR database, Takara Bio was not regarded as a spin-off. However, since this
day closing price. “Private equity financing” represents the total equity invested, defined firm was founded when Takara Shuzo became Takara Holdings, which is the major
as the sum of paid-in capital, additional paid-in capital, and capital reduction (million shareholder of Takara Bio, we regarded this firm as a spin-off. Additionally, spin-offs are
yen) prior to IPO. “IPO capital increase” represents the capital increase (million yen) at often discriminated from spin-outs, which are not capitalized by their parent companies.
IPO. Among the 33 IPO firms, we employ 32 firms to measure the above variables. Since we did not completely distinguish between the two and the sample size was not
sufficiently large, we included spin-outs in the variable for spin-offs.
21
with the use of these investment practices, in contrast to the findings of, Although the TSE provides other indices for emerging markets, such as the J-stock
index, it was only possible to obtain the JASDAQ index for the entire observation period
and Brander et al. (2002). We will investigate whether these investment
from January 1995 to December 2014. Meanwhile, Lerner (1994b) constructed a bio-
practices are associated with lower IPO value after controls, using re- technology equity index using publicly traded biotechnology firms. Among publicly
gression analysis. traded biotechnology firms in Japan, there were only three firms founded before January
1995.
22
Although we may be able to define the cohort dummy for firms founded during the
4.3. Variables
period of 2005–2009 (the reference category is the period of 2010–2014), as shown in
Table 1, none of the firms founded during the period of 2010–2014 went public.
First, we discuss the variables (covariates) used in the estimation of Therefore, we treated the cohort dummy for firms founded during the period of
the time to IPO (sub-hazard of IPO). We examine whether the initial 2005–2014 as the reference category.

8
Y. Honjo, S. Nagaoka
variables (VC_ST, UNIV, SPINOFF, C95_99, and C00_04), except that the
JASDAQ index and industry dummies are excluded.
For the outcome equation, as specified in Eq. (2), we employ a
parsimonious specification focusing on investment practices, given our
small sample size. Specifically, we examine whether investment prac-
tices are significantly associated with the IPO value of biotechnology
start-ups. As Gompers (1995) stated, the number of financing rounds, in
addition to the size of each investment and total financing provided, is
an important measure of the staged investment structure of VC firms. In
this study, we use the number of financing rounds (N_ROUND) as a
proxy for staged financing and obtain the data from the JVR database.
In addition to staged financing, syndication is another major char-
acteristic of investment practices. Although VCs often invest in a firm
through multiple funds and financing rounds, it is not easy to measure
the degree of syndication. Since the impact of first-round syndication
may differ from that of later-round syndication, we measure the first
round of financing separate from the later rounds. We use the variable
for first-round syndication by the number of first-round VC firms
(N_VCF).23 The use of this variable mitigates causality issues due to the
fact that VC firms have an incentive to participate in the later rounds of
financing prior to IPO just to enhance their reputation. Additionally, we
define the variable for later-round syndication by the number of later-
round VC firms (N_VCL) to investigate the impact of later-round syn-
dication, compared to that of first-round syndication. We obtain data on
the number of VC firms from the JVR database.
Furthermore, we use the variable for the JASDAQ index (JINDEX) to
control for equity market conditions at IPO. Note that this variable is
measured at the time of the firm's IPO, so that it is time variant.
Table 8 summarizes the definitions of the variables and Table 9
shows the summary statistics for the variables used in our analysis.
5. Estimation results
5.1. Results for the time to IPO
In Table 10, we present the estimation results for the time to IPO
using the competing-risks regression. While 33 firms of 213 firms in the
data set had an IPO during the observation period, 36 firms disappeared
before going public, mainly through liquidation, and these cases are
regarded as competing risks. Table 10 provides both the estimated
coefficients and sub-hazard ratios. While we use initial VC financing
(VC_ST), university base (UNIV), spin-off (SPINOFF), market conditions
(JINDEX), cohort dummies (C95_99 and C00_04), and industry dum-
mies (I_DRUG and I_HEALTH) in column (i), we omit the industry
dummies in column (ii) because they are insignificant.
As shown in Table 10, the coefficients of VC_ST are positive at the
1% significance level and the sub-hazard ratio for initially VC-backed
start-ups is approximately three times higher than that for others. The
results reveal that biotechnology start-ups initially backed by VC firms
are more likely to go public earlier than other start-ups, which is con-
sistent with Helwege and Packer (2009). Although information asym-
metries are severe when biotechnology start-ups start their businesses,
some VC firms in Japan commit to start-up financing. VCs are generally
required not only to assess and screen viable businesses but also to
provide risk money in a timely manner. The findings suggest that VC
firms that committed to start-up financing are more likely to select and/
or induce early IPOs of their funded firms.
The coefficients of UNIV are also positive at the 1% significance
level and the sub-hazard ratio for university-based start-ups is ap-
proximately six times higher than that for others. We find that the
likelihood of an IPO is higher for university-based start-ups in the
23
More precisely, we defined this variable as the logarithm of the number of VC firms
plus one because only a few firms were not backed by VC firms through the first round of
financing.
9
Research Policy xxx (xxxx) xxx–xxx
biotechnology industry in Japan. The findings suggest that university-
based start-ups have a strong incentive to access public equity markets
quickly, presumably because they require significant infusion of risk
money to sustain their R & D activities. This may also imply that uni-
versities tend to select and license out such technologies.
In addition, the coefficients of SPINOFF are positive at the 5% sig-
nificance level. We find that the likelihood of an IPO is higher for spin-
off start-ups in the biotechnology industry in Japan. The results reveal
that spin-offs are more likely to go public earlier than others.
More importantly, the coefficients of JINDEX are insignificant, in-
dicating that equity market conditions are not significantly associated with
the sub-hazard of IPO. In other words, the timing of IPOs does not depend
on equity market conditions. Our findings are inconsistent with Lerner
(1994b), who found that a higher level of the equity index increases the
probability of an IPO. For biotechnology start-ups, external equity finan-
cing is often critical for R & D activities. At the same time, many bio-
technology start-ups indeed have negative profits; therefore, these firms
cannot rely on internal financing for R & D activities. Even if a firm as-
sesses that the current market value of equity is low, it would go public to
secure funding for R & D activities when bridge financing is not available.
Additionally, VC firms that have a short time horizon may have an in-
centive to recoup their investments in a timely manner. The findings
suggest that biotechnology start-ups often cannot afford to wait for a
higher market value because of thin private equity financing in Japan.
Furthermore, C95_99 and C00_04 have a positive effect on the sub-
hazard of IPO, and the coefficients of C95_99 are positive at the 1%
significance level. The results indicate that biotechnology start-ups
founded in the late 1990s are more likely to go public within a shorter
period.
5.2. Results for IPO value
In Table 11, we present the estimation results for IPO value, defined as
the logarithm of the market value of equity (MV), using the type 2 Tobit
model. In Table 12, we also present the estimation results for relative IPO
value, which, as shown in Table 8, is defined as the ratio of IPO value to
the total equity invested (R_MV). It is important to note that the dependent
variables are measured by using the logarithm transformation, mainly
because, as shown in Table 4, the variances of these variables are high. We
use initial VC financing (VC_ST), university base (UNIV), spin-off
(SPINOFF), and cohort dummies (C95_99 and C00_04) in the selection
equation. Considering our small sample size and multicollinearity issues in
the type 2 Tobit model, we include VC_ST, UNIV, SPINOFF, C95_99, and
C00_04 only in the selection equation, and obtain the parameter estimates
using the two-step procedure. In addition to the type 2 Tobit model, we
can use an ordinary least squares (OLS) method to estimate the parameters
in the outcome equation. While we give up a separate identification of the
sample selection bias in this case, we may be able to avoid the multi-
collinearity issues arising in the type 2 Tobit model. The estimation results
are similar to those in Tables 11 and 12, which appear in Tables A1 and A2
of Appendix A.24
To control for equity market conditions, we include the variable for
the JASDAQ index (JINDEX) in the model. The estimated coefficients of
investment practices for IPO value and relative IPO value are presented
from columns (i) to (iv). Staged financing (N_ROUND) and first-round
syndication (N_VCF) are used in the outcome equation to investigate
whether investment practices predict IPO value. For comparison, the
variable for later-round syndication (N_VCL) is also used. Since
N_ROUND, N_VCF, and N_VCL are correlated, we use each of these
variables in columns (i) to (iii), separately.25 Moreover, these variables
24
We also estimated parameters using MLE. As a result, we obtained similar results to
those shown in Tables 11 and 12 .
25
For instance, the correlation coefficients between N_ROUND and N_VCF, between
N_VCF and N_VCL, and between N_VCL and N_ROUND were −0.307, −0.333, and 0.684,
respectively.
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Table 8
Variable definitions.

Variable Symbol Definition

All firms
Initial VC financing VC_ST Dummy variable for the firm raising equity from a VC firm when the firm was founded.
University base UNIV Dummy variable for the firm related to a university when the firm started a business fulfilling any of the following conditions: (1) the firm was
founded, based on technologies originating from a university, (2) the firm conducted joint research with a university within one year, (3) the
firm was founded with support of a university, (4) the firm or a trusted organization itself proclaimed university origin.
Spin-off SPINOFF Dummy variables for the subsidiary firm.
Market conditions JINDEX JASDAQ index based on the month-end closing price, normalized by its value in January 1991 (=100).
Cohort dummies C95_99 Dummy variable for the firm founded during the period 1995–1999.
C00_04 Dummy variable for the firm founded during the period 2000–2004.
Industry dummies I_DRUG Dummy variable for the firm in drug discovery.
I_HEALTH Dummy variable for the firm in medical and health care, except for drug discovery.
Only for IPO firms
IPO value MV Logarithm of the market value of equity (million yen), based on the first-day closing price.
Relative IPO value R_MV Logarithm of the ratio of the market value of equity, based on the first-day closing price, minus capital increase at IPO, divided by the sum of
paid-in capital, additional paid-in capital, and capital reduction prior to IPO.
Staged financing N_ROUND Logarithm of the number of financing rounds by VC firms and corporations prior to IPO.
Syndication N_VCF Logarithm of the number of VC firms investing in the firm from the founding date through the first round of financing plus one.
N_VCL Logarithm of the number of VC firms investing in the firm after the first round of financing prior to IPO plus one.

Notes: All the dummy variables take the value of one if the stated condition holds, and zero otherwise.

Table 9 Table 10
Summary statistics of variables. Estimation results for time to IPO.

Variable Mean SD 25% Median 75% N Variable (i) (ii)

All firms
VC_ST 0.080 – – – – 213 Coef. SHR Coef. SHR
UNIV 0.390 – – – – 213
SPINOFF 0.066 – – – – 213 VC_STi 1.165*** 3.206*** 1.117*** 3.055***
JINDEX 72.211 23.824 50.490 66.040 95.580 25138 (0.418) (1.338) (0.390) (1.191)
C95_99 0.103 – – – – 213 UNIVi 1.765*** 5.842*** 1.799*** 6.042***
C00_04 0.521 – – – – 213 (0.448) (2.617) (0.432) (2.611)
I_DRUG 0.385 – – – – 213 SPINOFFi 1.945** 6.992** 1.937** 6.940**
I_HEALTH 0.183 – – – – 213 (0.757) (5.295) (0.756) (5.249)
IPO value JINDEXit −0.002 0.998 −0.003 0.997
MV 9.655 1.065 8.989 9.544 10.184 32 (0.007) (0.006) (0.007) (0.007)
R_MV 1.935 1.873 0.317 1.636 3.022 32 C95_99i 1.847*** 6.341*** 1.725*** 5.611***
N_ROUND 1.428 0.604 1.099 1.609 1.869 32 (0.613) (3.886) (0.556) (3.121)
N_VCF 1.178 0.757 0.693 1.099 1.701 32 C00_04i 0.507 1.660 0.510 1.665
N_VCL 2.102 1.016 1.386 2.350 2.917 32 (0.527) (0.874) (0.519) (0.864)
JINDEX 68.713 20.298 50.825 66.775 86.830 32 I_DRUGi 0.298 1.347
(0.407) (0.548)
Notes: SD indicates standard deviation. N indicates the number of firms. For all firms, I_HEALTHi −0.119 0.888
JINDEX is time variant and the others are time invariant. Among the 33 IPO firms, we (0.557) (0.495)
employ 32 firms to measure the variables for IPO value. Number of firms 213 213
Number of IPOs 33 33
Number of competing events 36 36
Log pseudo-likelihood −149 −150
Wald χ2 54.2*** 45.3***
are jointly included to investigate the association between investment
practices and IPO value in column (iv). Notes: The competing risks regression is used. Figures in parentheses are standard errors
With respect to staged financing, the coefficients of N_ROUND are adjusted for 213 firms. “Coef.” represents estimated coefficients and “SHR” represents the
sub-hazard ratio. *** and ** indicate the 1% and 5% significance levels, respectively.
negative and significant in column (i) of Tables 11 and 12 .26 They are
significantly negative or insignificant in column (iv) of these tables, VC firms.27 In the former case, a project with staged financing would
which control for the effects of syndication. Thus, we find little evi- achieve high performance through a new round of financing. In the
dence that the returns to more staged financing are higher than those to latter case, however, the project may perform poorly due to a frag-
less staged financing, which is inconsistent with Gompers (1995). The mented capital infusion and the capital shortage of early-stage invest-
findings suggest that staged capital infusion is not associated with ment by a lead VC firm. The capital constraints, low capability, and
higher IPO value. inexperience of VC firms yield small investment and/or a significant
Staged capital infusion may occur not only when a project pro- error in cost estimation; therefore, a project with staged capital infusion
gresses successfully, justifying the additional investment, but also when is more likely to be associated with poor performance.
capital invested in the prior stage is too small because of the capital For syndication, the coefficient of N_VCF is positive in column (ii) of
constraints of VC firms, or project costs (or risks) are underestimated by Table 11, and negative in Table 12, although insignificant. However, its
coefficient becomes significantly negative in column (iv) of Table 12,
which controls for the effects of staged financing and later-round syn-
26
As shown in Table 4, the variance of relative IPO values was high, and the relative dication. When the size of syndication is measured by the number of VC
IPO value of one firm was over 1000 (R_MV was over 6). For robustness, we estimated the
same models as those in Tables 11 and 12, based on the sample excluding this particular
firm. Consequently, we obtained similar results to those in Tables 11 and 12, although the 27
Lerner (2002) argued that VCs underestimate the cost of change, and pointed out
impact of stages financing became weak in columns (i) of Tables 11 and columns (i) and VC's failure to consider the costly adjustments associated with the growth of their in-
(iv) of Table 12. vestment.

10
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Table 11 Table 12
Estimation results for IPO value (MV). Estimation results for relative IPO value (R_MV).

Variable (i) (ii) (iii) (iv) Variable (i) (ii) (iii) (iv)
Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef.

N_ROUNDi −0.742*** −0.489 N_ROUNDi −1.365*** −0.915*

(0.252) (0.338) (0.459) (0.539)
N_VCFi 0.107 −0.186 N_VCFi −0.323 −0.997***
(0.225) (0.210) (0.410) (0.335)
N_VCLi −0.435*** −0.291 N_VCLi −0.879*** −0.788**
(0.147) (0.197) (0.261) (0.314)
JINDEXi 0.023*** 0.022*** 0.020*** 0.021*** JINDEXi 0.023* 0.021 0.017 0.019
(0.008) (0.009) (0.008) (0.007) (0.014) (0.016) (0.014) (0.012)
Constant term 9.124*** 8.293*** 9.347*** 9.672*** Constant term 2.917** 1.820 3.501** 5.044***
(0.762) (0.834) (0.791) (0.821) (1.401) (1.529) (1.425) (1.310)
Selection Selection
VC_STi 0.926** 0.926** 0.926** 0.926** VC_STi 0.926** 0.926** 0.926** 0.926**
(0.362) (0.362) (0.362) (0.362) (0.362) (0.362) (0.362) (0.362)
UNIVi 1.011*** 1.011*** 1.011*** 1.011*** UNIVi 1.011*** 1.011*** 1.011*** 1.011***
(0.259) (0.259) (0.259) (0.259) (0.259) (0.259) (0.259) (0.259)
SPINOFFi 1.119** 1.119** 1.119** 1.119** SPINOFFi 1.119** 1.119** 1.119** 1.119**
(0.457) (0.457) (0.457) (0.457) (0.457) (0.457) (0.457) (0.457)
C95_99i 1.544*** 1.544*** 1.544*** 1.544*** C95_99i 1.544*** 1.544*** 1.544*** 1.544***
(0.408) (0.408) (0.408) (0.408) (0.408) (0.408) (0.408) (0.408)
C00_04i 0.697** 0.697** 0.697** 0.697** C00_04i 0.697** 0.697** 0.697** 0.697**
(0.300) (0.300) (0.300) (0.300) (0.300) (0.300) (0.300) (0.300)
Constant term −2.344*** −2.344*** −2.344*** −2.344*** Constant term −2.344*** −2.344*** −2.344*** −2.344***
(0.330) (0.330) (0.330) (0.330) (0.330) (0.330) (0.330) (0.330)

ρ −0.011 −0.258 −0.160 0.037 ρ −0.337 −0.449 −0.469 −0.171

σ 0.820 0.944 0.826 0.789 σ 1.553 1.800 1.570 1.269
λ −0.009 −0.244 −0.132 0.029 λ −0.523 −0.808 −0.737 −0.217

Number of firms 212 212 212 212 Number of firms 212 212 212 212
Number of IPOs 32 32 32 32 Number of IPOs 32 32 32 32
Wald χ2 17.2*** 6.92** 17.3*** 21.1*** Wald χ2 11.3*** 2.45 13.9*** 28.9***

Notes: Heckman's two-step procedure is used. Figures in parentheses are standard errors. Notes: Heckman's two-step procedure is used. Figures in parentheses are standard errors.
“Coef.” represents estimated coefficients. *** and ** indicate the 1% and 5% significance “Coef.” represents estimated coefficients. ***, **, and * indicate the 1%, 5%, and 10%
levels, respectively. Among the 33 IPO firms, we employ 32 firms to estimate the outcome significance levels, respectively. Among the 33 IPO firms, we employ 32 firms to estimate
equation. the outcome equation.

firms in the first round of financing, we do not find that syndication is

positively associated with IPO value or relative IPO value. This in-
dicates that first-round syndication is not accompanied with high
market value of equity, even though it brings in a large amount of ca-
pital. We find little evidence that the returns to more syndicated in-
vestments are higher than those of less syndicated investments, which is
inconsistent with the value-added hypothesis proposed by Brander et al.
(2002). Moreover, the coefficients of N_VCL are negative and significant
in column (iii) of Tables 11 and 12, as well as in column (iv) of
Table 12. Thus, when the size of syndication is measured by the number
of VC firms in the later rounds of financing, we find that syndication is
negatively associated with the two IPO values, which is inconsistent
with the window dressing hypothesis discussed by Lerner (1994a).
We can offer the following interpretations of the source of the dif-
ference between our results and previous findings with respect to syn-
dication. The negative or insignificant coefficient of first-round syndi-
cation suggests that syndication is not used for adding managerial value
to a project in Japan but perhaps for selection. The value-added role of
VC firms may be relatively small due to the underdevelopment of the
VC industry in Japan. At the same time, VC firms in Japan tend to face
high variations in the level of quality of projects because there is less
screening by private equity investors such as business angels before VC
investments, which makes the screening motivation for syndication
more important.28
On the other hand, the negative coefficient of later-round syndica-
tion suggests that the window dressing hypothesis does not hold in
Japan. The same reasons as suggested for explaining the negative
coefficient of staged capital infusion for IPO value can explain the ne-
gative coefficient of later-round syndication. One such explanation is
that invested capital is too small in the early rounds of financing, and
another is that later-round syndication is a response for underestimated
project costs (or risks) by a lead VC firm. Lead VC firms that are fi-
nancially constrained invest a small amount of capital in the early
rounds of financing, so that other VC firms participate in the later
rounds. The lead VC firm's underestimation of project costs (or risks)
also results in later-round syndication. In both cases, relative IPO value
is negatively associated with later-round syndication.
Furthermore, the coefficients of JINDEX are positive and significant
in Table 11, although JINDEX does not have a sufficiently significant
coefficient for relative IPO value in Table 12. We find that the market
value of equity tends to depend on equity market conditions at IPO. The
results indicate that a market boom leads to higher IPO value in the
biotechnology industry in Japan. As a typical case, the stock prices of
biotechnology firms rose sharply after the announcement of the 2012
Nobel Prize in Physiology or Medicine by John B. Gurdon and Shinya
Yamanaka. Meanwhile, as shown in Table 10, the timing of IPOs for
biotechnology start-ups does not depend on equity market conditions,
although IPO value does tend to depend on the market conditions.

28
Another potential explanation is herd behavior shown by participating in financing 6. Conclusions
rounds. Some studies have found imitative and herd behavior in Japanese firms and banks
(e.g., Lieberman and Asaba, 2006; Uchida and Nakagawa, 2007). In financing rounds, VC
firms may have an incentive to participate by just following the behavior of the lead VC
6.1. Summary of findings
firm—that is, driven by information cascade. Such syndication does not improve firm
values, and may actually reduce IPO values by encouraging excessive investments. This study explored the IPO and financing of biotechnology start-

11
Y. Honjo, S. Nagaoka
ups in Japan. Using a unique data set, we found that biotechnology
start-ups initially backed by VC firms and those originating from uni-
versities are more likely to go public within a shorter period. We also
examined how staged financing and syndication by VC firms are asso-
ciated with relative IPO value. The results revealed that first-round
syndication is not associated with high IPO value relative to investment
and that staged financing and later-round syndication tend to be ne-
gatively associated with IPO value relative to investment. Although VCs
in Japan can increase capital availability through staged financing and
syndication, they may not be able to add value. Furthermore, we pro-
vided evidence that the timing of IPOs does not depend on equity
market conditions in the biotechnology industry, whereas IPO value
tends to depend on equity market conditions.
The environment for VC investments in Japan—more specifically,
the underdeveloped private equity financial system—would explain the
performance of these investment practices, which differs from the
prevailing views in the literature (e.g., Lerner, 1994a,b; Gompers, 1995;
Brander et al., 2002). According to the prevailing view based on in-
formation asymmetries and agency costs, staged financing is positively
associated with firm value because a firm with a high-performance
project can continue to receive new rounds of financing. From the
value-added hypothesis, early-round syndication can enhance firm
value if each VC firm in the syndication provides unique management
support. Because participation in financing a successful firm enhances a
VC firm's reputation, later-round syndication is also positively asso-
ciated with firm value. However, if staged financing and syndication
are the responses to the funding gap resulting from capital constraints
or underestimation of project costs by a lead VC firm in the early rounds
of financing, they can be negatively associated with firm value. In fact,
the size of VC investments is small in Japan, compared with that in the
US; thus, capital invested by VC firms may tend to be insufficient.
Furthermore, such underestimations are more likely to occur in
Japan because of the limited experience and expertise of VC firms as
well as high uncertainty VC firms face. VC firms in Japan frequently
invest in the early stages, and prescreening by other private equity
investors before VC investments is limited. Therefore, syndication in
Japan may more often indicate that VC firms do not have sufficient
information to justify a standalone investment. In this case, early-round
syndication is negatively associated with firm value under the selection
hypothesis. Finally, the fact that IPOs occur quite independent of equity
market conditions suggests that VC firms in Japan do not have the
ability to provide bridge financing to realize the optimal IPO timing,
unlike those in the US.
6.2. Limitations and implications
This study has several limitations. First, this study focuses on the
market value of equity only at IPO because it is difficult to identify the
timing of all investments, including the alternatives to VC investments.
Second, we could not obtain a large sample of IPOs; therefore, our es-
timations of IPO value are based on parsimonious specifications, as in
Brander et al. (2002). Third, we could not control for the endogeneity of
investment practices investigated in this study, that is, staged financing
and syndication. Thus, our findings do not allow for causal inter-
pretations. Finally, our investigation ignores private equity financing
Appendix A
For robustness, Tables A1 and A2 present the estimation results using the OLS method, corresponding to Tables 11 and 12 . We include the
variables used in the selection equation in columns (vi) of Tables A1 and A2 . Overall, we obtain similar results to those in Tables 11 and 12 .
12
Research Policy xxx (xxxx) xxx–xxx
other than from VC firms, and our findings apply only to biotechnology
start-ups in Japan.
Despite these limitations, this study makes several important con-
tributions to the literature. First, we showed that staged financing and
syndication could have quite different roles, depending on the level of
development of private equity financing. Staged financing is positively
associated with firm value if its contribution is mainly to enable bio-
technology start-ups to undertake high-risk and high-return projects by
allowing VC firms to increase their investments as the uncertainty is
resolved. However, staged financing can be negatively associated with
firm value if it is used to fill the funding gap due to underfunding by a
constrained lead VC firm or underestimation of project costs (or risks).
We also found a possibility that VC firms do not realize the optimal IPO
timing, in contrast with previous literature. While VC firms are often
expected to provide hands-on management support and syndication is
expected to enhance such role of VC firms, syndication may also serve
primarily to screen projects when VC firms lack expertise and face
highly uncertain projects. Second, our findings based on relative IPO
value indicated that the median return on investment for biotechnology
start-ups from their IPOs in Japan is over five times greater than the
total private equity financing, while the variance is considerably high.
To the best of our knowledge, empirical research on this issue is scarce
outside of North America and Europe. This is the first study on bio-
technology start-ups in Japan using return on investment, constructed
from round information, as the performance measure in Japan.
This study also has several implications. Our findings suggest sig-
nificant room for enhancing the contribution of VC firms in Japan,
which will in turn improve the performance of high-tech start-ups. First,
it would be important to strengthen the professional and financial
capabilities of VC firms. Many VC firms in Japan were subsidiaries of
banks and securities companies, and they were not active in governing
high-tech start-ups nor providing managerial support. Strengthening
these capabilities is of critical importance. The entry and growth of
independent and specialized VC firms, including corporate and foreign
ones, would also be important in this regard. Higher capabilities of VC
firms to provide management support for high-tech start-ups would
make syndication a useful way to pool the management resources of
various VC firms for improving their performance. Their higher fi-
nancial capabilities also enable VC firms to use staged financing and
syndication productively, in order to expand the scope of the projects
funded. Second, it would be important to develop private equity fi-
nancing activities, which complement those of VC firms. More active
discoveries and screening of projects by business angels and other in-
vestors in the early stages could be encouraged, enabling VC firms to
focus on higher quality projects. At the same time, institutional in-
vestors such as pension funds could be encouraged to play a larger role
in financing high-tech start-ups in the late stages, including bridge fi-
nancing. This would help prevent premature and untimely IPOs of high-
tech start-ups.
Access to public equity markets through IPO is imperative not only
for many biotechnology start-ups to secure funding for their R & D ac-
tivities in order to turn their inventions into innovations, but also for VC
firms to recoup their investments. We hope that private equity finan-
cing will improve its ability to nurture the growth of biotechnology
start-ups in Japan.
Y. Honjo, S. Nagaoka Research Policy xxx (xxxx) xxx–xxx

Table A1
Estimation results for IPO value (MV): OLS.

Variable (i) (ii) (iii) (iv) (v)

Coef. Coef. Coef. Coef. Coef.

N_ROUNDi −0.743*** −0.479 −0.660

(0.218) (0.305) (0.440)
N_VCFi 0.073 −0.180 −0.368
(0.216) (0.212) (0.279)
N_VCLi −0.442*** −0.292 −0.371
(0.141) (0.218) (0.312)
JINDEXi 0.023*** 0.024*** 0.021*** 0.021*** 0.027**
(0.006) (0.007) (0.007) (0.007) (0.011)
VC_STi 0.660
(0.510)
UNIVi −0.761*
(0.441)
SPINOFFi −0.598
(0.610)
C95_99i 0.002
(0.567)
C00_04i 0.077
(0.631)
Constant term 9.112*** 7.900*** 9.127*** 9.703*** 10.341***
(0.573) (0.545) (0.685) (0.968) (1.080)

Number of IPOs 32 32 32 32 32
F 15.3*** 6.44*** 13.1*** 8.35*** 3.91***

Notes: Figures in parentheses are robust standard errors. “Coef.” represents estimated coefficients. ***, **, and * indicate the 1%, 5%, and 10% significance levels, respectively. Among the
33 IPO firms, we employ 32 firms to estimate the equation.

Table A2
Estimation results for relative IPO value (R_MV): OLS.

Variable (i) (ii) (iii) (iv) (v)

Coef. Coef. Coef. Coef. Coef.

N_ROUNDi −1.478*** −0.986 −1.113

(0.533) (0.709) (0.860)
N_VCFi −0.438 −1.043*** −0.952**
(0.385) (0.298) (0.365)
N_VCLi −0.917*** −0.782** −0.899**
(0.313) (0.361) (0.422)
JINDEXi 0.028** 0.028** 0.023** 0.021** 0.021*
(0.010) (0.010) (0.010) (0.010) (0.012)
VC_STi 0.788
(0.738)
UNIVi −0.370
(0.710)
SPINOFFi −1.977**
(0.928)
C95_99i 0.079
(0.916)
C00_04i 0.281
(0.888)
Constant term 2.143* 0.519 2.272* 4.806*** 5.203***
(1.149) (0.837) (1.297) (1.322) (1.575)

Number of IPOs 32 32 32 32 32
F 8.16*** 3.82** 13.2*** 12.7*** 7.44***

Notes: Figures in parentheses are robust standard errors. “Coef.” represents estimated coefficients. ***, **, and * indicate the 1%, 5%, and 10% significance levels, respectively. Among the
33 IPO firms, we employ 32 firms to estimate the equation.

References Biotechnol. 21, 613–617.

Black, B.S., Gilson, R.J., 1998. Venture capital and the structure of capital markets: banks
versus stock markets. J. Finan. Econ. 47, 243–277.
Brander, J.A., Amit, R., Antweiler, W., 2002. Venture-capital syndication: improved
venture selection vs. the value-added hypothesis. J. Econ. Manag. Strat. 11, 423–452.
Cameron, A.C., Trivedi, P.K., 2005. Microeconometrics: Methods and Applications. New
York, Cambridge University Press.
Cox, D.R., 1972. Regression models and life-tables (with discussion). J. R. Stat. Soc. Ser. B
34, 187–220.
Deeds, D.L., Decarolis, D., Coombs, J.E., 1997. The impact of firm-specific capabilities on
the amount of capital raised in an initial public offering: evidence from the bio-
technology industry. J. Business Ventur. 12, 31–46.
Dibner, M.D., Trull, M., Howell, M., 2003. US venture capital for biotechnology. Nat.
Fine, J.P., Gray, R.J., 1999. A proportional hazards model for the subdistribution of a
competing risk. J. Am. Stat. Assoc. 94, 496–509.
Giot, P., Schwienbacher, A., 2007. IPOs, trade sales and liquidations: modelling venture
capital exits using survival analysis. J. Banking Finance 31, 679–702.
Gompers, P.A., 1995. Optimal investment, monitoring, and the staging of venture capital.
J. Finance 50, 1461–1489.
Goto, A., 2000. Japan's national innovation system: current status and problems. Oxf.
Rev. Econ. Policy 16, 103–113.
Hall, B.H., 2002. The financing of research and development. Oxf. Rev. Econ. Policy 18,
35–51.
Hall, B.H., Lerner, J., 2010. The financing of R & D and innovation. In: Hall, B.H.,
Rosenberg, N. (Eds.), Economics of Innovation. Elsevier, Oxford, pp. 609–639.
Harris, R.S., Jenkinson, T., Kaplan, S.N., 2014. Private equity performance: what do we
know? J. Finance 69, 1851–1882.

13
Y. Honjo, S. Nagaoka
Heckman, J.J., 1979. Sample selection bias as a specification error. Econometrica 47,
153–161.
Helwege, J., Packer, F., 2004. Initial public offerings in hot and cold markets. J. Finan.
Quant. Anal. 39, 541–569.
Helwege, J., Packer, F., 2009. Private matters. J. Finan. Intermed. 18, 362–383.
Himmelberg, C., Petersen, B., 1994. R & D and internal finance: a panel study of small
firms in high-tech industries. Rev. Econ. Stat. 76, 38–51.
Honjo, Y., 2015. Why are entrepreneurship levels so low in Japan? Jpn. World Econ. 36,
88–101.
Honjo, Y., Kato, M., Okamuro, H., 2014a. R & D investment of start-up firms: does
founders’ human capital matter? Small Business Econ. 42, 207–270.
Honjo, Y., Nagaoka, S., Nakamura, K., Shimizu, Y., 2014b. Bio-venture no kagakuteki
gensen to seicho ni muketeno kadai: “2012 nen Bio-venture Tokei Chosa” ni
Motozuite (Scientific Sources of Bio-Ventures and Issues Toward Their Growth: A
Report From “the 2012 Survey on Bio-ventures), IIR Working Paper, WP#13-22.
Hitotsubashi University (in Japanese).
Ibbotson, R.G., Jaffe, J.F., 1975. “Hot issue” markets. J. Finance 30, 1027–1042.
Kamien, M.I., Schwartz, N.L., 1978. Self-financing of an R & D project. Am. Econ. Rev. 68,
252–261.
Kaplan, S.N., Schoar, A., 2005. Private equity returns: persistence and capital flows. J.
Finance 60, 1791–1823.
Lerner, J., 1994a. The syndication of venture capital investments. Finan. Manag. 23,
16–27.
Lerner, J., 1994b. Venture capitalists and the decision to go public. J. Finan. Econ. 35,
14
Research Policy xxx (xxxx) xxx–xxx
293–316.
Lerner, J., 2002. Boom and bust in the venture capital industry and the impact on in-
novation. Fed. Reserv. Bank Atlanta Econ. Rev., Forth Quarter 2002, 25–39.
Lerner, J., 2009. Boulevard of Broken Dreams. Princeton University Press, Princeton, NJ.
Lieberman, M.B., Asaba, S., 2006. Why do firms imitate each other? Acad. Manag. Rev.
31, 366–385.
Megginson, W., Weiss, K., 1991. Venture capitalist certification in initial public offerings.
J. Finance 46, 879–903.
Motohashi, K., 2012. A comparative analysis of biotechnology startups between Japan
and the US. Soc. Sci. Jpn. J. 15, 219–237.
Müller, E., Zimmermann, V., 2009. The importance of equity finance for R & D activity.
Small Business Econ. 33, 303–318.
OECD, 2009. Biotechnology Statistics 2009. OECD, Paris.
OECD, 2015. OECD Science, Technology and Industry Scoreboard 2015: Innovation for
Growth and Society. OECD, Paris.
Pagano, M., Panetta, F., Zingales, L., 1998. Why do companies go public? An empirical
analysis. J. Finance 53, 27–64.
Ritter, J.R., 1984. The “hot issue” market of 1980. J. Business 57, 215–240.
Uchida, H., Nakagawa, R., 2007. Herd behavior in the Japanese loan market: evidence
from bank panel data. J. Finan. Intermed. 16, 555–583.
Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data, Second
Edition. MIT Press, Cambridge, MA.
Yung, C., Colak, G., Wang, W., 2008. Cycles in the IPO market. J. Finan. Econ. 89,
192–208.