Disclosure of Downside Risk and Investors' Use of Qualitative Information Evidence From The IPO Prospectus's Risk Factor Section (2016)

bs_bs_banner
International Review of Finance, 16:1, 2016: pp. 73–126

DOI: 10.1111/irfi.12066
Disclosure of Downside Risk and

Investors’ Use of Qualitative
Information: Evidence from
the IPO Prospectus’s Risk
Factor Section*
RUI DING
The UNSW Business School, The University of New South Wales, Sydney, NSW,
Australia
ABSTRACT
I use the context of a company’s initial public offering (IPO) of equity

securities as a capital-market setting to empirically study the economic con-
sequences of risk factor disclosures. Using data from Australian IPOs, I
examine the relation of textual risk disclosures in the prospectus to initial
underpricing. I find that the quantity of disclosures in the risk factor section
itself has no significant impact on initial underpricing. However, an increase
in the informativeness of risk factor disclosures is associated with lower IPO
underpricing. My results suggest that IPOs that provide informative risk
factor disclosures have less ex ante uncertainty, in the sense that the disclo-
sures help investors estimate the dispersion of secondary market value. The
effect of informative risk factor disclosures on IPO underpricing is more
pronounced for IPOs with less prestigious lead underwriters and is mainly
driven by younger firms, smaller firms, and firms with poorer operating
performance prior to their IPOs. Collectively, my findings suggest that infor-
mative disclosures of downside risk are useful for investors to evaluate IPOs.
JEL Classification: D8, G14, G24, M4.
I. INTRODUCTION
Security regulators in many countries require initial public offering (IPO) issuers
to disclose investment risk and uncertainties in the IPO prospectus. The risk
factor section provides forward-looking information about future possible eco-
nomic events that could have a significant adverse effect on an issuer’s financial
* I thank Philip Brown, Feng Chen, Julie Cotter, Jeff Coulton, Gerard Hoberg, Feng Li, Ronald
Masulis, Gary Monroe, Yifang Sun, Wei Wang, Mark Wilson, Norman Wong, participants at the
AFAANZ 2012 annual conference, and, in particular, Sudipto Dasgupta (Editor), anonymous review-
ers and Associate Editor, for helpful comments and suggestions. An earlier version of this article was
circulated under the title “Do Stock Market Investors Use the IPO Prospectus’s Risk Factors Section?
Australian Evidence”.
© 2015 International Review of Finance Ltd. 2015

International Review of Finance
performance. Clear and meaningful risk disclosures can help investors assess the
potential risk and returns of an offer and make informed decisions.1 The issue is
whether risk disclosures are as informative as intended by the regulators. This
paper empirically examines the informativeness of textual risk disclosures in the
IPO prospectus.
Risk disclosures in the IPO prospectus may be informative to investors. The
prospectus is the most important source of information for investors because
the amount of publicly available information for IPOs is often limited due to
their short reporting history. The risk factor section provides forward-looking
and valuation-relevant information. The disclosures are likely to be credible as
managers of the IPO firms face potential litigation risk and reputation costs
(Skinner 1994, 1997; Lowry and Shu 2002).
On the other hand, risk disclosures may not be informative for several
reasons. First, prior studies on corporate disclosure suggest that managers have
a self-serving bias to disclose favorable information and tend to withhold bad
news (Hermalin and Weisbach 2007; Kothari et al. 2009b). Managers may have
concerns that disclosures of negative information might result in the issue being
undersubscribed as investors react to potentially unfavorable outcomes. Second,
issuers may withhold specific risk information for fear of releasing proprietary
information (Verrecchia 1983). Third, it is often difficult for managers to esti-
mate the economic effect of an identified risk on their firm’s financial perfor-
mance and to quantify the likelihood the risk will impact their firm. Therefore,
issuers may try to avoid providing detailed risk disclosures and instead catalogue
all possible risks, making the disclosures vague and boilerplate in nature.2
I examine the informativeness of risk factor disclosures in the IPO prospectus
by testing the relation of the disclosures to initial underpricing using data from
Australian IPOs. I find that the quantity of disclosures in the risk factor section
itself has no significant impact on initial underpricing. However, I find that the
informativeness of risk disclosures (as measured by the unique content) is
inversely associated with underpricing, controlling for a variety of determinants
of IPO underpricing and issuer characteristics. The finding is consistent with the
view that informative risk disclosures reduce ex ante uncertainty regarding true
share value (Ritter 1984; Beatty and Ritter 1986; Rock 1986). Ritter (1984),
extending Rock’s (1986) underpricing model, states that the concept of ‘risk’ for
IPOs relates to informational differences and that ‘riskier firms should have
higher average initial returns than firms that are easier to evaluate’ (Ritter 1984,
p. 221). My results suggest that informative risk disclosures help investors
evaluate IPOs, in the sense that they reduce information asymmetry and help
investors make a more precise estimate of the distribution of firm value. Con-
1 See Consultation Paper 155, the Australian Securities and Investments Commission (ASIC
2011b).
2 ASIC expressed its concern that ‘often the risk disclosure in prospectuses is too general and
does not explain to investors how the risk is relevant’ (Consultation Paper 155, ASIC 2011a).
It called for increased focus and specificity in risk factor disclosures (Regulatory Guide 228,
ASIC 2011b).
74 © 2015 International Review of Finance Ltd. 2015

Disclosure of Downside Risk in the IPO Prospectus
sistent with reducing investor ex ante uncertainty, IPOs disclosing more infor-
mative content about investment risk and uncertainties have less initial
underpricing (Beatty and Ritter 1986). I find that the effect of informative risk
disclosures on IPO underpricing is more pronounced for IPOs with less presti-
gious lead underwriters.
An alternative explanation drawn from the behavioral finance literature for
underpricing suggests that IPOs are not underpriced per se, but that first-day
investor optimism drives positive initial returns (i.e., the heterogeneous beliefs
explanation) (Miller 1977; Harrison and Kreps 1978; Ritter 1991; Ritter and
Welch 2002). If the divergence of opinion between optimistic and pessimistic
investors subsequently declines, IPOs will underperform in the long run (Miller
1977). This explanation implies that IPOs with more (less) informative risk
disclosures should have less (more) post-IPO analysts’ forecast dispersion (a
proxy for heterogeneity of investor beliefs) and more (less) favorable long-run
returns. For IPOs garnering sufficient coverage, I find no evidence of a negative
relation between the disclosures of unique risk information and the dispersion
in analysts’ long-term earnings growth forecasts or annual earnings per share
(EPS) forecasts. In addition, IPOs with less unique risk disclosures do not have
less favorable abnormal long-run returns. Collectively, the results do not
support the alternative explanation for the inverse relation between informa-
tive risk factor disclosures and underpricing.
I also examine the relation between risk factor disclosures and IPO valuation
both in the IPO market and in the immediate aftermarket. I use the ratio of offer
price to benchmark value to measure IPO valuation at the offer price and the
ratio of first-day secondary market closing price to benchmark value to measure
IPO valuation in the aftermarket (Purnanandam and Swaminathan 2004;
Chemmanur and Yan 2009). For a subsample of IPO firms with matching firms
available, I find a positive and significant relation between the unique content of
risk factor disclosures and IPO valuation in the IPO market when the price-to-
value ratios are estimated based on sales or book value. This suggests that firms
are valued higher in the IPO market when they disclose more unique content of
investment risk in the prospectus.3 While the relation between the unique
content of risk disclosures and IPO underpricing for the subsample of IPOs with
positive sales prior to IPO is insignificant, I find that the most undervalued IPOs
have the largest first-day return, consistent with the asymmetric information
theory. Further analysis shows that the relation between the informativeness of
risk factor disclosures and IPO underpricing is mainly driven by younger firms,
smaller firms, and firms with negative EBITDA (earnings before interest, tax,
depreciation, and amortization) in the year prior to going public.
3 The positive relation between the unique content and IPO valuation in the IPO market is
insignificant when the price-to-value ratio is estimated based on EBITDA. A large number of
IPO firms in my sample do not have positive EBITDA in the year prior to listing. The
insignificant relation between the unique content of risk disclosures and IPO underpricing
when the price-to-value ratio is estimated based on EBITDA is likely due to the use of the
truncated sample. See the ‘Additional tests’ section.
© 2015 International Review of Finance Ltd. 2015 75

I perform additional tests to explore factors influencing an issuer’s disclosure

decision. The results show that firms of an older age disclose more unique and
standard content, consistent with firms with a longer operating history having
more knowledge about the key risks in their business and operating environ-
ment. The unique risk disclosures are inversely associated with market volatility
over the 30 trading-day period prior to the IPO and the indicator variable of
accumulated losses in the year prior to the IPO. This suggests that issuers tend
to disclose less unique or specific information on investment risk when facing
a greater extent of uncertainty. Venture-capital-backed IPOs disclose less unique
and standard content, consistent with IPOs with venture-capital backing having
stronger corporate governance and fewer incentives to provide risk disclosures.
The results also support that the pre-IPO news stories substitute for firm disclo-
sure (Schrand and Verrecchia 2005).
To address the concern that the risk disclosure variables may be endogenous,
I revisit the association between risk factor disclosures and underpricing for the
sample of IPOs with negative EBITDA prior to listing using a two-stage regres-
sion model. In the first-stage regression, I endogenize risk disclosure variables by
regressing the unique content and standard content on factors that may influ-
ence firms’ voluntary disclosure decision, respectively. The fitted values of the
unique content and standard content are then used to explain underpricing in
the second stage. I find that the inverse relation between the unique content
and underpricing is robust to accommodating endogeneity. However, the posi-
tive association between the standard content and underpricing disappears.
Finally, to test whether risk factor disclosures contain risk-relevant informa-
tion, I examine the association between the disclosures and ex post measures of
firm risk. I find that the standard content is positively associated with both the
daily stock return volatility and the negative stock return volatility in the first
year of trading. No significant results are found for the unique content, con-
servative tones, and a self-developed measure of content capturing future cash
flow risk.
My study contributes to the literature on the effect of disclosure on firms’ cost
of equity. The role of disclosure in reducing information asymmetry and the
cost of capital is one of the central debates in accounting and finance. Most
prior theoretical and empirical research suggests that a higher level of disclosure
reduces the cost of equity.4 Kothari et al. (2009a) distinguish between positive
and negative information and provide initial evidence that disclosures of posi-
tive information decrease firms’ cost of capital, whereas disclosures of negative
information increase firms’ cost of capital. The finding, however, relies on their
textual analysis to classify information as positive or negative. In addition, their
finding was based on examining a combined set of information sources (includ-
ing corporations, analysts, and business press). They cannot find the relation
when only corporate disclosure is analyzed.
4 See Diamond and Verrecchia (1991), Botosan and Plumlee (2002), Easley and O’Hara (2004),
Barry and Brown (1984), and Lambert et al. (2007).

The information provided in the risk factor section of the IPO prospectus
provides a unique setting to study the effect of disclosure on the cost of capital.
The risk factor section contains only negative information, thus the direction is
unambiguous. The IPO prospectus is a primary source of information for inves-
tors as the amount of publicly available information for IPOs is often limited.
Firms undergoing an IPO are making their first large-scale public disclosure via
the prospectus. Unlike other studies of the effect of disclosure on the cost of
capital which need to control for prior disclosure, IPO provides a natural
experiment to examine the effect of differential disclosure on underpricing. The
results of my study suggest that, in the Australian IPO setting, the overall level
of risk disclosures does not affect investors’ evaluation of share value, but
informative content helps reduce investor ex ante uncertainty regarding the
value of the shares and thus underpricing.
My study contributes to the emerging literature on market participants’ use
of qualitative information by examining the informativeness of textual risk
disclosures in the IPO prospectus and investors’ use of such information. Extant
empirical work on IPO pricing has focused on financial or numerical informa-
tion in the prospectus, which only accounts for a small portion of the prospec-
tus. Textual disclosures can convey potentially valuable information that is
absent from traditional quantitative measures. Risk factor disclosures identify
important downside risks and caveats, thereby complementing quantitative
projections and measures.
My study extends and complements a small extant literature that examines
textual disclosures in the IPO prospectus in the US setting. Beatty and Welch
(1996) count the number of captioned risks in the risk factor section to measure
management caution. Arnold et al. (2010) use word counts of the risk factor
section to measure the degree of information ambiguity. These studies focus on
the level of disclosure but do not analyze the content. Hanley and Hoberg
(2010) document that greater word content in the entire prospectus (a proxy for
premarket due diligence) results in more accurate offer prices and less under-
pricing for a sample of US IPOs. Their interpretation of the result is that more
investment in premarket due diligence by issuers and underwriters decreases the
issuing firm’s reliance on book building to price the issue and therefore reduces
the need to compensate investors for revealing information. This information
gathering perspective does not apply to the Australian IPO setting as most IPOs
in Australia follow fixed-price issuance, with the inability to change the issue
price and the quantity predetermined in the prospectus (Lee et al. 1996; How
and Yeo 2000). My study applies the asymmetric information theory (Beatty
and Ritter 1986; Rock 1986) and thus presents results in a different light.5
Further, my study focuses only on the disclosures of downside risk (negative
5 Ljungqvist (2004) notes the growing interest in the IPO experience of countries other than the
US and contends that differences in institutional frameworks allow for sharper tests of
theoretical predictions.

information) and provides evidence on how such disclosures affect investor ex

ante uncertainty. My study also examines the relation between risk factor
disclosures and long-run stock returns and the relation between risk factor
disclosures and IPO valuation, which have not been examined in prior research.
This study has important practical implications for issuers, investors, and
regulators. The Australian Securities and Investments Commission (ASIC)
expressed its concern that risk disclosures in the IPO prospectus are too general
and do not explain to investors how the risk is relevant.6 In November 2011,
ASIC issued Regulatory Guide 228 Prospectuses: Effective disclosure for retail inves-
tors (RG228). RG228 encourages issuers to highlight the key risks, explain what
these risks mean to investors, and explain how each risk might affect the issuer’s
business or the offer. My study provides evidence supporting ASIC’s recommen-
dation. It is also important for firms to understand that risk disclosures are a
critical way to communicate their strategy to investors and often have a direct
effect on their share value.
The remainder of the paper is organized as follows. Section II discusses related
literature and develops the hypotheses. Section III details the sample and
empirical models. Section IV discusses the results of the empirical tests. Section
V concludes.
II. LITERATURE AND HYPOTHESES
A. Australian IPO market and disclosure requirements

The Australian IPO market has several different features compared with the US
IPO market. First, IPOs in Australia are concentrated in the small end of the
market, whereas US IPOs are concentrated in the large end (Lee et al. 1996).
Second, most IPOs in Australia follow fixed-price issuance, according to which
price and quantity of shares are predetermined in the prospectus. The book
building method remains restricted to a few large IPOs.7 The inability to change
the issue price and the quantity is an important difference from the prevailing
US environment, where a final price is not specified until information is gath-
ered from investors during book building. With greater uncertainty regarding
pre-issue demand, the fixed-price method can result in more mispricing (Lee
et al. 1996; How and Yeo 2000; Alavi et al. 2008). The Australian method is
6 See Consultation Paper 155, the Australian Securities and Investments Commission (ASIC
2011a).
7 With book building, a preliminary offer price range is set, and then underwriters and issuers
go on a road show to market the company to prospective investors. The road show helps
underwriters to estimate demand. If there is strong demand, the underwriter will set a higher
offer price (Hanley 1993; Ritter and Welch 2002). With the fixed-price method, the prospectus
details the number of shares to be issued and the issue price, neither of which can be changed
during the course of the issue. All shares must be sold or taken up by the underwriter prior to
trading commencing on the stock exchange. The issuer is committed to a price and quantity
decided on well before listing occurs (Lee et al. 1996).

expected to increase heterogeneity in information availability between

informed and uninformed investors (Lee et al. 1996). Third, share allocation is
not explicitly regulated in Australia. The nonpublic allocation procedure
increases the possibility that favored (informed) clients crowd out the unin-
formed, thereby exacerbating the winners’ curse (Lee et al. 1996).
Third, Australian IPOs have long lockup periods, during which the pre-IPO
owners’ shares cannot be sold. The Australian Securities Exchange (ASX) may
restrict the transfer of shares issued before listing so that they cannot be sold for
a period of up to 2 years after listing (PricewaterhouseCoopers 2011).8 This is
much longer than the typical 6 months underwriter-enforced lockup period for
a US IPO.
Lastly, Australia has a much less litigious environment than the US (How and
Yeo 2000). The ability of a class action to challenge the issuer and underwriter
is much more restrictive in Australia. Litigants have to launch and fund their
own actions, which further deter shareholders from initiating lawsuits against
the issuer and underwriter (How and Yeo 2000).
In Australia, disclosure obligations of an issuer are specified under the Cor-
porations Act 2001(Cth) (the Act). The Act requires the issuer to include all the
information that investors and professional investment advisers would reason-
ably require to make an informed assessment of the issuer and the security
being offered (i.e., the general disclosure requirement) (s710). The Act further
stipulates specific disclosure requirements in the prospectus (s711). The issuer is
required to word and present the prospectus in a clear, concise, and effective
manner (s715A) and to ensure the document is not misleading or deceptive
(s728 (1)).
ASIC further issued regulatory guidance on disclosures of specific content in
the prospectus. Regulatory Guide 228 (RG 228) Prospectuses: Effective disclosure
for retail investors (ASIC 2011b) advises the issuer on how to word and present
the prospectus in a clear, concise, and effective manner. It also sets out guidance
on how to satisfy the content requirement for a prospectus. RG228 requires
issuers to explain the risks associated with their business model, the security,
and the offer. It states that risk disclosures should be specific rather than general
and that issuers should help retail investors understand the key risks. While the
regulation mandates disclosure of risk factors in the IPO prospectus, issuers can
choose the extent and content of the disclosure.
B. Asymmetric information theory and underpricing

It is widely documented that, on average, IPOs of operating companies experi-
ence large positive first-day returns. Underpricing represents the cost of initial
equity capital, assuming the closing price of a stock on its first trading day
reasonably reflects the value of the issue. Prior studies offer a number of
8 This is more common where the company gained admission to the ASX official list under the
‘asset test,’ or where transactions with related party are involved.

explanations for IPO underpricing.9 Asymmetric information theory is relevant

to the tests in this paper given its focus on risk factor disclosures in the IPO
prospectus.
Rock (1986) provides a theoretical framework on the relation between dis-
closure and IPO underpricing. In his model, issuing firms and their investment
bankers are uncertain about the true value of a share of the firm’s stock (v).
Investors are also uncertain about v, but some investors incur costs to become
better informed, and these better informed investors only bid for underpriced
IPOs. As the IPO market is characterized by quantity rationing as a market-
clearing mechanism, the bids by informed investors subject uninformed inves-
tors to a winner’s curse.10 Thus, IPOs are on average underpriced to keep
uninformed investors in the market and compensate others for becoming
informed. Rock’s model implies an equilibrium relation between expected
underpricing and the uncertainty that uninformed investors have with respect
to the true value of the IPO shares.
Ritter (1984) and Beatty and Ritter (1986) formalize this implication by
showing that, for a wide class of density functions for v, the derivative of the
IPO offer price (OP) with respect to the range of possible aftermarket prices (b −
a) is negative [i.e., ∂OP/∂(b − a) < 0]. They provide a proof that if v is uniformly
distributed on the closed interval from a to b, then expected underpricing [i.e.,
(E(v) − OP)/OP] is a positive function of b minus a. Ljungqvist (2004, p. 15) states
that a key empirical implication of Rock (1986), as formalized in Ritter (1984)
and Beatty and Ritter (1986), is that ‘underpricing should increase in the ex ante
uncertainty about the value of the IPO firm.’11
Risks and uncertainties regarding an IPO firm’s future performance are likely
to revise downward both the upper and lower ends of the interval. However, the
length of the revised interval determines whether investor ex ante uncertainty
regarding the true value of the shares is reduced by risk disclosures. If the revised
interval is larger than the original interval, then risk disclosures in the prospec-
tus should be positively related to underpricing. On the other hand, if the
revised interval is smaller than the original interval (i.e., risk disclosures reduce
the uncertainty that uninformed investors have regarding the true share value),
then risk disclosures should be negatively related to underpricing. The effect of
9 The literature offers a number of explanations for IPO underpricing, including asymmetric
information, signaling, irrationality, agency, and litigation (Ritter and Welch 2002). Agency
and irrationality explanations are primarily applied to explain the large underpricing of
Internet IPOs. Tests of the signaling explanation are mixed and the litigation explanation is
suggested to be unconvincing.
10 Uninformed investors will receive a partial allocation of underpriced IPOs but a full alloca-
tion of overpriced IPOs.
11 Beatty and Ritter (1986) contend that investors who decide to engage in information
production implicitly invest in a call option on the IPO, which will be exercised if the true
price exceeds the exercise price. The value of this option increases in the extent of valuation
uncertainty. Thus, more investors will become informed as the valuation uncertainty
increases. This results in greater underpricing, as an increase in the number of informed
investors aggravates the winner’s curse problem.

risk disclosures on IPO underpricing is not obvious and thus is an empirical

question.
C. Hypotheses
The first issue I examine is whether risk factor disclosures in the IPO prospectus
affect initial underpricing. The asymmetric information theory predicts that
underpricing increases in investor ex ante uncertainty regarding the true value
of the shares (Ritter 1984; Beatty and Ritter 1986). There are three competing
arguments regarding how risk factor disclosures may affect underpricing. The
first argument supports an inverse relation between risk factor disclosures and
underpricing. Risk factor disclosures in the prospectus may lead to a reduction
in information asymmetry by increasing the overall quantity of disclosure that
is available to the public. Once privately held information becomes publicly
available, the difference in the information set among informed and unin-
formed investors should decrease. Under this argument, risk factor disclosures
in the IPO prospectus are negatively associated with underpricing.
The second argument predicts a positive relation between risk factor disclo-
sures and underpricing. The risk factor section in the prospectus contains only
unfavorable information on possible future economic events that could
adversely affect an issuer’s financial performance. The disclosures of unfavor-
able information may increase investor uncertainty with respect to the issuer’s
future cash flows, thereby making it more difficult for uninformed investors to
value the security (Kothari et al. 2009a). Further, information asymmetry
between informed and uninformed investors may increase as a result of diver-
gent interpretation by uninformed investors about the potential impact of the
disclosed risks on the issuer’s performance. Therefore, risk disclosures may result
in an increase in the range of possible aftermarket prices estimated by investors.
It follows that risk factor disclosures should be positively associated with
underpricing.
The third argument predicts no relation between risk factor disclosures and
underpricing assuming the disclosures are boilerplate and, therefore, not infor-
mative. Several factors may contribute to meaningless risk disclosures. First,
managers of an issuing firm may not be fully aware of the key risks, the
probability the risks might impact the firm and the economic effects of the risks.
Second, the risk factor section contains negative information, which managers
may tend to withhold (Hermalin and Weisbach 2007; Kothari et al. 2009b). In
addition, managers may not disclose risk information for fear of releasing
proprietary information (Verrecchia 1983). Boilerplate risk disclosures should
have no effect on investors’ assessment of the value of the shares and informa-
tion asymmetry between informed and uninformed investors. It follows that
risk disclosures have no effect on underpricing. Hypothesis 1 is stated in null
form due to the competing arguments.
Hypothesis 1: Risk factor disclosures in the IPO prospectus are not associated
with initial underpricing.

Next, I examine the association between risk factor disclosures in the prospectus
and ex post measures of IPO risk. If risk factor disclosures report useful informa-
tion on investment risk, the disclosures are expected to be positively associated
with ex post measures of IPO risk. A positive association between risk disclosures
in the prospectus and ex post measures of firm risk would indicate that firms that
turn out to be relatively risky ex post generally disclose more risk ex ante. On the
other hand, if risk factor disclosures do not contain risk-relevant information,
risk factor disclosures will have no significant association with ex post measures
of IPO risk. Prior studies suggest that managers seem keen to create a positive
perception of their company (Kothari et al. 2009b). Information on investment
risk demands explicit attention to the downside. A bias toward the positive may
hinder managers from providing effective warning signals. Further, managers
may be unwilling to provide specific discussion of investment risk because they
feel less confident in discussing an uncertain future over which they only have
imperfect control (Malmendier and Tate 2005a, 2005b).
I use stock return volatility in the first year of trading to proxy for IPO risk.
Stock return volatility reflects the expected risk associated with an issuing firm’s
future cash flows. In addition, I also examine the volatility of negative stock
returns. Prior studies suggest that people are more sensitive to economic losses
relative to economic gains (Kraus and Litzenberger 1976; Libby and Fishburn
1977; Kahneman and Tversky 1979). Stock return volatility is a symmetric risk
measure and may lack information about negative outcomes. Downside
semivariance of returns may overcome this issue (Hogan and Warren 1974;
Jahankhani 1976). Hypothesis 2 is also stated in null form given the competing
arguments.
Hypothesis 2: Risk factor disclosures in the IPO prospectus are not associated
with market-based measures of firm risks subsequent to the commencement of
trading.
III. DATA AND METHOD
A. IPO sample construction

I obtain my initial list of all Australian IPOs issued between January 1, 1996 and
December 31, 2007 from Thomson Financial’s SDC Platinum New Issues data-
base. A total of 1661 IPOs are initially identified. I eliminate unit offerings
(bundles of shares and warrants) (254), IPOs by foreign issuers (26), closed-end
funds (16), investment companies (203), real estate investment trusts (68), and
privatized government-owned companies (4). For the remaining IPOs identi-
fied, I download prospectuses from the Connect 4 database to analyze the
content of risk factor disclosures. I eliminate 94 observations due to the absence
of filings in the Connect 4 database, 97 duplicated observations, and 41 obser-
vations due to lack of pre-IPO data to calculate required variables. This reduces
the sample to 858 IPOs.

From the SDC database, I extract equity issues and their major characteristics,
including issue amount, offer price, underwriter information, fee information,
venture-capital backing, and high-technology industry firms. I hand-collect
shares outstanding pre- and post-IPO, aggregate insider equity holdings pre-
and post-IPO, and pre-IPO financial data from the prospectuses. I hand-fill gaps
in SDC’s coverage of company incorporation dates, auditor and solicitor infor-
mation, and manually check all firms. Security prices are extracted from SIRCA’s
Core Research Database.
B. Textual risk disclosure data

For each sample IPO, I extract the risk factor section from its prospectus. The
title of section varies slightly across prospectuses, but generally is named as ‘Risk
Factors,’ ‘Investment Risk,’ ‘Investment Consideration,’ or ‘Investment Uncer-
tainty.’ The extracted section is saved as a text file for each IPO. Headers and
exhibits are purged so that I can focus on the words in the section. As the
measures used to capture the content of the risk factor section incorporate
information from prior IPOs (as discussed below), the sample is further
restricted to IPOs that are issued after December 31, 1996 in order to have
sufficient data.
C. Measures to capture the content of risk factor disclosures
i. Measure 1: Standard versus unique content

The first measure aims to distinguish between unique information and standard
information contained in the risk factor section. I apply the method developed
by Hanley and Hoberg (2010) with some modifications to accommodate the
characteristics of my sample of Australian IPOs.
For each IPO, Java codes are developed to convert the text into a numeric
word vector (which is defined as WordsK) summarizing the counts of its English-
language raw words.12 A database of the unique raw English-language words
that appear in the universe of the risk factor section of all IPO prospectuses is
built. To conserve computing space, I follow Hanley and Hoberg (2010) and
exclude articles, conjunctions, pronouns, abbreviations, certain compound
words, and any words that appear fewer than a total of five times, because they
are not informative regarding content. This leaves a vector of 5186 possible
words. Word vectors for all risk factor sections in the sample have the same
length (5186). Each element of the vector is populated by the count of the
12 Hanley and Hoberg (2010) used word roots which they identified from Webster.com. At the
time of doing this study, word roots are no longer identifiable from the website. Stemming
is an option to obtain word roots, but is inherently imprecise (Loughran and McDonald
2011). Hoberg and Phillips (2010) applied a similar text-based analysis to analyze product
market synergies and competition in mergers and acquisitions and used raw words instead
of word roots. They suggest that the result generated using raw words is similar to that
generated using word roots.

number of times the word is used in the section. Thus, the vectors measure the
total amount of information in the risk factor section. I normalize the raw word
vector (WordsK) by the total number of unique raw words used in the risk factor
section and define it as normk. As discussed in Hanley and Hoberg (2010),
normalized word vectors have elements that sum to 1 and do not sum differ-
ently when a document is larger. Scaled word counts eliminate any bias due to
size.
Next, I decompose a word vector (normtotal,k) into its standard and informative
components. Consistent with Hanley and Hoberg (2010), I define content that
is driven by information in prior IPOs as standard and content that is not
explained by prior IPOs as unique. Standard content has two primary sources:
content from recent or concurrent IPOs (content related to recent IPO condi-
tions) and content from IPOs in the same industry (industry-specific content).
I define recent IPOs as those that were filed in the 90-day period preceding
the current IPO’s initial filing date. Industry-specific content is computed from
prospectuses that were filed by firms in the same 4-digit GICS (Global Industry
Classification Standard) industry group at least 91 days prior to but no later
than 1 year before the current IPO’s initial filing date.13 I choose the 4-digit
GICS industry group because this level of industry classification is expected to
allow sample firms in the same industry group to have substantial similarity in
their operating activities, thereby capturing similarity in the risk factor section
for IPOs with a reasonable level of precision.14
For IPO i that has K IPOs filed in the 90-day period preceding its initial filings,
its exposure to the content of recent IPOs is measured by the average of the
normalized vectors of the K IPOs (normrecent,t). The normalized vector of each of
the K IPOs (normk) is computed by first obtaining the word vector of the risk
factor section (WordsK) (as discussed above) and then normalizing this word
vector by dividing the sum of its elements. The average of the normalized
vectors of recent IPOs is defined as
1 K
normrecent ,i = ∑ normk
K k =1
(1)
13 I also estimated the standard and informative content component for my sample of IPOs by
defining recent IPOs as those that were filed in 180-day period preceding the current IPO’s
initial filing date. Industry-specific content is computed from prospectuses that were filed by
firms in the same 4-digit GICS industry group at least 181 days prior to but not later than 1
year before the current IPO’s initial filing date. Similar test results were obtained.
14 Hanley and Hoberg (2010) define recent IPOs as those that were filed in the 90-day period
preceding the current IPO’s initial filing date and past industry IPOs as IPOs in the same
Fama-French 48 industry code as the current IPOs that were filed at least 91 days prior to but
no later than 1 year before the current IPO’s initial filing date. This is to ensure that the
content does not overlap with the content related to recent IPO conditions. While I believe
that estimating the unique and standard components based on the 4-digit GICS industry
group is superb than that based on the 2-digit GICS industry sector, I also estimate the
unique and standard components using the 2-digit GICS industry sector and rerun the
regressions of equations (8) and (9). The results are qualitatively similar.

Similarly, for IPO i that has P IPOs in the same 4-digit GICS industry group filed
at least 91 days prior to but no later than 1 year, its exposure to the content of
past industry IPOs is measured by the average of the normalized vectors of the
P IPOs (normindustry,i). The normalized vector of each of the P IPOs (normp) is
computed by first obtaining the word vector of the risk factor section Wordsp,
and then normalizing this word vector by dividing the sum of its elements.
1 P
normindustry ,i = ∑ normp
P p =1
(2)
In order to estimate informative and standard content, I include only IPOs that
have at least one other IPO that was filed in the 90-day window prior to the
current IPO’s initial filing date and at least one other IPO in the same 4-digit
GICS industry group as the current IPO that was filed at least 91 days prior to
but no later than 1 year before the current IPO’s filing date. Imposing this
criterion further reduces the sample size to 742.15
I then run the following regression (without an intercept) for each IPO:
normtotal ,i = α recent ,i normrecent ,i + α industry ,i normindustry ,i + ε (3)
The standard content variable is defined as follows (Hanley and Hoberg 2010):
α standard ,i = α recent ,i + α industry ,i (4)
where αstandard,i measures the relative loading of standard content, interpreted as

the proportion of standard words in the risk factor section of IPO i’s prospectus.
The content not explained by these two sources, the vector of the absolute value
of the residual, is defined as unique content. Because each regression for each
risk factor section has 5186 observations, the first-stage regression in equation
(3) has ample power to fit these coefficients.
ii. Measure 2: Conservative tone

The second measure of qualitative information captures conservatism in the use
of language in the risk factor section of the prospectus. Prior studies suggest that
less optimism or greater caution is more likely when individuals feel that they
do not have perfect control over events (Malmendier and Tate 2005a, 2005b).
Tetlock et al. (2008) document that negative qualitative information in firm-
specific news stories captures otherwise hard-to-quantify aspects of firms’ fun-
damentals and it predicts low stock returns.16 They show that negative words
have a similar effect to ‘risk’ and ‘uncertainty’ words in predicting future
performance. I define risk disclosure conservatism as the number of occurrences
of words in the Harvard IV-4 negative words list, scaled by the number of total
15 This number also excludes IPOs issued in 1996 which do not meet the criterion.
16 In order to affect stock returns, negative words should convey incremental information
about either firms’ cash flows or investors’ discount rates (Campbell and Shiller 1987, Tetlock
et al. 2008, Kothari et al. 2009a).

words in the risk factor section, minus its previous year’s mean, divided by
previous year’s standard error (Tetlock 2007).
Nit − μt −1
NegativeWords _ Harvardi ,t = (5)
σ t −1
where NegativeWords_Harvardi,t is the measure of conservatism in the risk factor
section of the IPO prospectus of firm i in year t, Nit is the count of the number
of occurrences of words in the Harvard IV-4 negative words list in the risk factor
section of the IPO prospectus of firm i in year t scaled by the total number of
words in the risk factor section, and μt−1 and σt−1 are the mean and standard error
of the distribution of N in year t − 1, respectively.
An alternative measure is based on the negative words dictionary developed
by Loughran and McDonald (2011) that takes into consideration terms that are
most likely used in financial documents (i.e., 10 Ks). I count the number of
occurrences of words in the negative words list compiled by Loughran and
McDonald (2011), and scale and standardize it.
FNit − μt −1
NegativeWords _ L & M i ,t = (6)
σ t −1
where NegativeWords_L&Mi,t is the alternative measure of conservatism of risk
factor disclosures in the IPO prospectus of firm i in year t, FNit is the count of the
number of occurrences of words in the negative words dictionary developed by
Loughran and McDonald (2011) in the risk factor section of the IPO prospectus
of firm i in year t scaled by the total number of words in the risk factor section,
and μt−1 and σt−1 are the mean and standard error of the distribution of FN in
year t − 1 respectively.
iii. Measure 3: Future cash flow risk

The third measure follows the dictionary approach and is designed to capture
future cash flow risk by focusing on the content of the risk disclosures. I identify
a portion of text in the risk factor section that provides information on possible
adverse outcomes that could arise from financial risk, business risk, macroeco-
nomic risk, litigation risk, liquidity risk, and environmental risk in the risk
factor section (RG228, ASIC 2011b) and construct a cash-flow-risk dictionary.
Holding the length of the risk factor section constant, the greater the discussion
about possible adverse impact on fundamentals, the greater the future cash flow
risk. Words related to the risk categories as outlined in ASIC’s regulatory guide
(RG228) and that frequently appear in the risk factor section are identified by
reading the section of randomly selected prospectuses representing 10% of my
IPO sample. Next, for each of sample IPOs, I count the number of times these
words are mentioned in the risk factor section of the prospectus and scale this
number by the total number of words in the section. Along the lines of previous
studies by Tetlock (2007), Tetlock et al. (2008), and Balakrishnan and Bartov
(2012), I subtract from this fraction its annual cross-sectional mean in the
previous year based on all IPOs in that previous year, and then divide by its

previous year’s standard error to obtain the standardized future cash-flow-risk

measure. Using the standardized measure is to eliminate the time series com-
ponent inherent in the risk factor section (Tetlock 2007; Balakrishnan and
Bartov 2012). The future cash-flow-risk measure is as follows:
Ci ,t − μt −1
CFRiskWordsi ,t = (7)
σ t −1
D. Empirical models
i. Risk factor disclosures and IPO underpricing

I begin by examining the relation between risk factor disclosures and initial
underpricing. The following equation is used to test the relation (equation (8)).
I control for a number of variables that have been identified in the literature to
proxy for issuer/offer risk. These include firm size (Schultz 1993), age (Beatty
1989), book-to-market, high technology (Loughran and Ritter 2004), ownership
retention (Beatty and Welch 1996), underwriter reputation (Beatty and Ritter
1986; Cater et al. 1998), venture-capital backing (Barry et al. 1990; Meggionson
and Weiss 1991), auditor size (Michaely and Shaw 1995), insider selling (Hanley
1993), total time between prospectus registration and exchange listing (Lee
et al. 1996), and the return of ASX All Ordinaries over the 30 trading days
preceding the filing date as past market returns can predict future underpricing
(Loughran and Ritter 1995; Lowry and Schwert 2004).17 All the continuous
variables in equation (8) are winsorized at 1% and 99% levels.
Ln (Underpricing )i ,t = β 0 + β1Ln ( Assets )i ,t + β2 Ln ( Age )i ,t + β3 BTMi ,t
+ β 4 HighTechi ,t + β5 % Retainedi ,t + β6Underwriteri ,t + β7VCi ,t
+ β8 Big i ,t + β9 Ln ( ASXReturn )i ,t − 30 to t + β10 % Insideri ,t (8)
+ β11Ln ( Delay )i ,t + β12 RiskDisclosurei ,t + ε it
where:
Ln(Underpricing) = ln (1 + first-day initial return)

Ln(Assets) = ln (pre-IPO total assets in millions of Australian dollars)
Ln(Age) = ln (1 + years from incorporation to IPO)
BTM = pre-IPO stockholders’ equity ÷ market capitalization immediately following
the IPO18
HighTech = 1 if a high-technology company and 0 otherwise
%Retained = % of voting common stock retained by pre-IPO shareholders
17 Lee et al. (1996) suggest that the total time from prospectus registration to the commence-
ment of exchange trading is an important proxy for fluctuations in the level of aggregate
premarket demand among investors (principally informed investors) under the Australian
issuance procedures. Issues experiencing long delays indicate that they may have had
difficulty attracting informed investors, reflecting the winners’ curse faced by the
uninformed.
18 The market capitalization of the IPO issuer is obtained by multiplying the offer price by the
total number of post-IPO shares (including those retained by owner).

Underwriter = a continuous measure of underwriter prestige using their percentage

market share for the entire sample period19
VC = 1 if issue is backed by venture capitalists and 0 otherwise
Big = 1 if a big 4/6 audit firm and 0 otherwise20
Ln(ASXReturn) = ln (1 + the return of ASX All Ordinaries measured over the 30
trading days preceding the filing date)
%Insider = shares sold by selling shareholders ÷ total shares sold in the IPO
Ln(Delay) = ln (the number of days between prospectus registration and exchange
listing)
RiskDisclosure = the aforementioned measures of risk factor disclosures in the IPO
prospectus
ii. Risk factor disclosures and future stock return volatility

Next, I examine the association between risk disclosures in the prospectus and
market-based measures of risk in the first year post IPO. To avoid inducing
survivorship bias, I compute stock return volatility for IPO firms for the
12-month post-IPO period or until delisting if this comes sooner. I also control
for other issuer characteristics. The standard deviation of ASX All Ordinaries
daily returns over the 30 trading days preceding the filing date is included to
control for the effect of market volatility prior to the filing date on future market
volatility (French et al. 1987). All the continuous variables are winsorized at the
1% and 99% levels.
Volatilityi ,t +1 = δ 0 + δ 1Ln ( Assets)i ,t + δ 2 Ln ( Age )i ,t + δ 3 BTMi ,t + δ 4 MSDReturnsi ,t − 30 to t
+ δ 5 Big i ,t + δ 6 HighTechi ,t + δ 7Underwriteri ,t + δ 8VCi ,t + δ 9 Leveragei ,t
+ δ 10 AccumLossDi ,t + δ 11RiskDisclosurei ,t + ε i ,t (9)
where:
Volatility = standard deviation of stock returns in the first year of trading, exclud-
ing initial return (SD(Returns)), or the standard deviation of negative stock return
excluding initial return (SD(Down)).
1 I
SD ( Down ) = ∑ min ( Reti , 0)2 , where Reti is the IPO firm’s daily return in day
I i =1
i, and min(Reti, 0)2 equals Ret i2 if Reti is negative and 0 otherwise.21
19 The Carter and Manaster (1990) ranking measure cannot be applied as tombstone announce-
ments are not common in Australia (Alavi et al. 2008). For robustness purposes, I also
construct an indicator variable of underwriter prestige. For each IPO, the ranking is equal to
1 if the lead underwriter is in the top quintile of the market-share-based ranking (the most
prestigious), and 0 otherwise. The market-share measure is computed by aggregating
inflation-adjusted gross proceeds of all the IPOs in the sample of individual underwriters.
The alternative variable produces similar results.
20 PriceWaterhouse and Coopers & Lybrand merged in 1998. Arthur Andersen surrendered its
licenses to practice as Certified Public Accountants after being found guilty of criminal
charges related to handling the auditing of Enron in 2002. These reduced the number of
major accountancy firms to four.
21 This downside risk measure is calculated relative to a zero return as prior studies suggest that
investors regard their original investment to be the most important benchmark in assessing
risk (Veld and Veld-Merkoulova 2008).

MSDReurns = standard deviation of ASX All Ordinaries daily returns measured over
the 30 trading days preceding the filing date.
AccumLossD = 1 if the firm has negative retained earnings during the year prior to
IPO and 0 otherwise.
All the other variables are the same as previously defined.
E. Sample descriptive statistics

Table 1 presents sample composition by issue year and GICS 4-digit industry
group. The year 2007 has the largest number of IPOs across all sample years. A
total of 168 IPOs are issued in 2007, which accounts for nearly 20% of sample
observations. This is followed by the number of IPOs issued in 2006 (130), 2005
(106), and 2004 (98). The years 1999 and 2000 also have a relatively larger
number of IPOs (76 and 99, respectively). This is primarily driven by the IPOs
in the Software & Services industry group. Across all sample observations, the
Materials industry group has the largest number of IPOs (320), followed by
Energy IPOs (106) and IPOs in the Software & Services industry group (78).
Table 1 also presents the distribution of IPOs for the subsample of 742 IPOs with
the first measure of risk disclosures and for the subsample of 831 IPOs with the
second and third measures, respectively.22
Panels A and B of Table 2 present summary statistics for the measures of risk
factor disclosures for my sample of Australian IPOs. As shown in the table, there
is substantial variation in the length of the risk factor section of the sample IPO
prospectus. For the subsample of 831 IPOs filed in the period from January 1,
1997 to December 31, 2007, the section has an average of 1633 words, ranging
from as few as 206 words to as many as 6826 words. For the subsample of 742
IPOs that are filed in the same period and have the measure of unique and
standard content, the risk factor section has an average of 1640 words with a
minimum of 206 words and a maximum of 5698 words. The words in the risk
factor section account for 4.9% of the total words in the prospectus on average.
The first measure of risk disclosures is to capture the unique information
content in the risk factor section. There are 742 observations with this measure.
The risk factor section in the sample has a standard content coefficient near 1
on average. This occurs because the average risk factor section’s standard
content is measured using the average of both recent and past industry IPOs.
The statistics suggest that the average risk factor section is similar to the average
past risk factor section in the prospectus. However, there is substantial variation
around this loading, which is the subject of my analysis. The source of standard
content is tilted toward recent IPOs.
There are 831 IPOs meeting the requirements when measures 2 and 3 are
used. The second measure captures the extent to which the issuer uses
conservative language in the risk factor section using both the words in the
22 As previously discussed, a number of IPOs are excluded because they do not meet the criteria
to calculate the measures of risk disclosures.

Table 1 Sample composition

Panel A: IPO frequencies by year of issue
Year of issue Whole Subsample Subsample

sample measure 1 measures 2 and 3
No. of IPOs Percent No. of IPOs Percent No. of IPOs Percent
1996 27 3.15
1997 29 3.38 18 2.43 29 3.49
1998 21 2.45 18 2.43 21 2.53
1999 76 8.86 57 7.68 76 9.15
2000 99 11.54 93 12.53 99 11.91
2001 23 2.68 21 2.83 23 2.77
2002 31 3.61 25 3.37 31 3.73
2003 50 5.83 32 4.31 50 6.02
2004 98 11.42 89 11.99 98 11.79
2005 106 12.35 95 12.80 106 12.76
2006 130 15.15 121 16.31 130 15.64
2007 168 19.58 173 23.32 168 20.22
Total 858 100.00 742 100.00 831 100.00
Panel B: IPO frequencies by GICS 4-digit industry group
GICS4 industry group Whole Subsample Subsample

sample measure 1 measures 2 and 3
No. of IPOs Percent No. of IPOs Percent No. of IPOs Percent
1010 Energy 106 12.35 101 13.61 104 12.52

1510 Materials 320 37.30 303 40.84 308 37.06
2010 Capital goods 53 6.18 50 6.74 51 6.14
2020 Commercial and professional 41 4.78 32 4.31 40 4.81
services
2030 Transportation 7 0.82 3 0.40 7 0.84
2510 Automobiles and components 1 0.12 0 0.00 1 0.12
2520 Consumer durables and apparel 12 1.40 6 0.81 12 1.44
2530 Consumer services 27 3.15 17 2.29 25 3.01
2540 Media 26 3.03 19 2.56 24 2.89
2550 Retailing 34 3.96 25 3.37 31 3.73
3010 Food and staples retailing 1 0.12 0 0.00 1 0.12
3020 Food beverage and tobacco 22 2.56 15 2.02 22 2.65
3030 Household and personal 3 0.35 0 0.00 3 0.36
products
3510 Healthcare equipment and 35 4.08 29 3.91 34 4.09
services
3520 Pharmaceutical, biotechnology, 38 4.43 28 3.77 37 4.45
and life sciences
4510 Software and services 78 9.09 76 10.24 77 9.27
4520 Technology hardware and 22 2.56 18 2.43 22 2.65
equipment
4530 Semiconductors and 2 0.23 0 0.00 2 0.24
semiconductor equipment
5010 Telecommunication services 20 2.33 13 1.75 20 2.41
5510 Utilities 10 1.17 7 0.94 10 1.20
Total 858 100.00 742 100.00 831 100.00
This table reports sample composition. Panel A reports the IPO frequencies by year of issuer. The first two columns
report the frequencies of sample IPOs issued in the period from 1996 to 2007. Columns 3 and 4 report the frequencies
of the subsample of IPOs with the first measure of risk disclosures. Columns 5 and 6 report the frequencies of the
subsample of IPOs with the second and third measures of risk disclosures. Panel B presents the distribution of IPOs by
GICS4 industry group.
IPO, initial public offering; GICS, Global Industry Classification Standard.

Table 2 Summary statistics

Mean Standard Minimum Median Maximum
deviation
Panel A: Risk factor text for the subsample of 742 IPO firms
Total words in 34934.540 11926.080 10960 32672 96444
document
Total words in Risk 1639.883 776.734 206 1529 5698
Factor section
Risk factor words/total 0.049 0.022 0.006 0.046 0.190
words
StandardContent 1.014 0.131 0.067 1.020 1.515
UniqueContent 1.062 0.174 0.000 1.053 1.737
Recent content 0.628 0.460 −0.627 0.648 3.748
Past industry content 0.385 0.451 −2.695 0.340 1.525
Panel B: Risk factor text for the subsample of 831 IPO firms
Total words in 34620.24 12105.84 10960 31989 96444
document
Total words in Risk 1632.941 795.094 206 1524 6826
Factor section
Risk factor words/total 0.049 0.022 0.006 0.047 0.190
words
NegativeWords_Harvard −0.007 0.9500 −2.320 −0.036 4.747
NegativeWords_L&M 0.012 0.9772 −2.986 −0.036 5.018
CFRiskWords 0.027 0.9752 −2.709 0.005 4.431
Panel C: Initial return, IPO characteristics, and other variables for the subsample of
742 IPO firms
Underpricing 0.280 0.545 −0.89 0.125 5.04
Assets 66.910 1301.153 0.000 1.264 35409.000
Age 4.789 8.406 0.008 1.574 79.181
BTM 0.327 6.383 −1.396 0.043 173.911
HighTech 0.226 0.419 0 0 1
% Retained 0.529 0.212 0 0.548 0.994
Underwriter 0.016 0.030 0 0.001 0.122
VC 0.032 0.177 0 0 1
Big 0.361 0.481 0 0 1
Lawyer 0.027 0.029 0 0.017 0.113
%Insider 0.054 0.181 0 0 1
Leverage 2.244 53.846 0 0 1462.615
AccumLossD 0.689 0.463 0 1 1
Delay 54.891 26.784 14 48 235
ASXReturn 0.020 0.036 −0.089 0.020 0.155
MSD(Returns) 0.007 0.003 0.003 0.007 0.019
PreIPONews 16.186 20.348 0 12 291
SD(Returns) 0.064 0.030 0.001 0.062 0.208
SD(Down) 0.041 0.020 0.008 0.039 0.235

Table 2 (continued )
Panel D: Initial return, IPO characteristics, and other variables for the subsample of 831
IPO firms
Underpricing 0.269 0.524 −0.890 0.121 5.04

Assets 72.097 1248.034 0.000 1.771 35409
Age 5.218 9.157 0.008 1.767 81.345
BTM 0.309 6.032 −1.396 0.044 173.911
HighTech 0.245 0.431 0 0 1
% Retained 0.539 0.212 0 0.558 0.994
Underwriter 0.304 0.460 0 0 1
VC 0.032 0.177 0 0 1
Big 0.372 0.484 0 0 1
Lawyer 0.028 0.029 0 0.017 0.113
%Insider 0.007 0.003 0.003 0.007 0.020
Leverage 2.032 50.882 0.000 0 1462.615
AccumLossD 0.668 0.471 0 1 1
Delay 54.844 26.137 14 48 235
ASXReturn 0.019 0.037 −0.107 0.0194 0.155
MSD(Returns) 0.007 0.003 0.003 0.007 0.020
PreIPONews 17.332 37.997 0 12 937
SD(Returns) 0.063 0.030 0.001 0.061 0.208
SD(Down) 0.041 0.020 0.008 0.038 0.235
Panels A and B of this table present descriptive statistics on the risk factor text for the subsample of 742
IPOs with the first measure of risk disclosures and for the subsample of 831 IPOs with the second and
third measures of risk disclosures, respectively. StandardContent and UniqueContent are the coefficients
from the first-stage regression on risk factor content for the 742 IPOs from January 1, 1997 to December
31, 2007, excluding foreign issuers, unit offerings, closed-end fund, financial firms, real estate invest-
ment trust, and privatized government-owned firms. normtotal,i = αrecent,inormrecent,i + αindustry,inormindustry,i + ε,
where StandardContent is the sum of the coefficients αrecent,i and αindustry,i, and UniqueContent is the absolute
value of the residuals. NegativeWords_Harvard is the standardized number of occurrences of words in the
Harvard IV-4 negative words dictionary. NegativeWords_L&M is the standardized number of occurrences
of words in Loughran and McDonald’s (2011) negative words dictionary. CFRiskWords is the standard-
ized number of occurrences of words in the self-developed list of words indicating cash flow risk. Panels
C and D present descriptive statistics on other related variables for the subsamples of 742 IPOs and the
subsample of 831 IPOs, respectively. Underpricing is the closing price of the first trading day less the issue
price, divided by the issue price. Assets is the assets in millions of Australian dollars at the end of the year
prior to IPO. Age is the number of years from incorporation to IPO. BTM is the pre-IPO stockholders’
equity divided by market capitalization immediately following the IPO. HighTech is an indicator
variable, which equals 1 if the issuing company is a high-technology company and 0 otherwise.
%Retained is the percentage of voting common shares retained by pre-IPO shareholders. Underwriter is
a continuous measure of underwriter prestige using underwriter market share over the entire sample
period. VC is an indicator variable, which equals 1 if the issue is backed by venture capitalists and 0
otherwise. Big is an indicator variable, which equals 1 if the IPO firm is audited by a big 4/6 audit firm
and 0 otherwise. Lawyer is the market share of the issuer’s legal counsel’s dollar market share over the
sample period. SD(Returns) is the standard deviation of daily stock returns of the first year of trading,
excluding the initial return. %Insider is shares sold by selling shareholders on total shares sold in the
IPO. Leverage is pre-IPO long-term liabilities on pre-IPO total assets. MSD(Returns) is the standard
deviation of daily stock returns of ASX All Ordinaries over the 30 trading days prior to the issue date.
ASXReturn is the return of ASX All Ordinaries measured over the 30 trading days preceding the filing
date. AccumLossD is an indicator variable, which equals 1 if the issuing firm has negative retained
earnings in the year prior to the IPO and 0 otherwise. Delay is the number of days between prospectus
registration and exchange listing. PreIPONews is the number of news stories about the issuing firm in the
year prior to going public. SD(Returns) is the standard deviation of daily stock returns of the IPO firm in
the first year of trading. SD(Down) is the standard deviation of negative stock returns of the IPO firm in
the first year of trading.

Harvard IV-4 negative words dictionary and the words in the negative words
dictionary compiled by Loughran and McDonald (2011). NegativeWords_
HarvardIVi,t ranges from −2.320 to 4.747, with an average of −0.007.
NegativeWords_L&Mi,t ranges from −2.986 to 5.018, with an average of 0.012.
The third measure CFRiskWordsi,t is the self-constructed measure of future cash
flow risk. This measure intends to capture potential adverse outcomes by count-
ing the number of words indicating financial risk, business risk, macroeconomic
risk, litigation risk, liquidity risk, and environmental risk in the risk factor
section. The average of this variable is 0.027, ranging from −2.709 to 4.431.
Panels C and D of Table 2 present descriptive statistics for initial return, IPO
characteristics, and riskiness variables for the two subsamples, respectively. As
shown in Table 2, more than 20% of the sample firms are high-technology
firms, around 3% of the firms are venture capital backed, and approximately a
third of the firms are audited by a big N auditor.23 The average initial return is
27% for the sample of 831 firms (28% for the sample of 742 firms). This is
consistent with the amount of underpricing documented in prior studies. Lee
et al. (1996) document an average of 16.41% underpricing for Australian indus-
trial IPOs issued between 1976 and 1989. Dimovski et al. (2011) document an
average of 29.6% underpricing for Australian industrial IPOs issued between
1994 and 2004. Consistent with prior studies (Lee et al. 1996), I also note a
number of firms (approximately 22% of the sample) have a negative initial
return (i.e., overpriced). The average gross proceeds for my sample of IPOs are
$25 million. The average IPO offer price is $0.53. Both the offer size and the
offer price are much lower than US IPOs. The average value of pre-IPO total
assets of the sample firms is around $70 million. At the time of listing, the
sample firms have approximately 5 years operating history on average. The
average number of days from prospectus registration to the commencement of
exchange trading is 55.
IV. RESULTS
A. Risk factor disclosures and IPO underpricing

In this section, I examine the association between risk factor disclosures in the
IPO prospectus and underpricing. As discussed in section II, the effect of risk
factor disclosures on underpricing is unclear. To the extent that risk disclosures
resolve investors’ uncertainty regarding the value of the shares, an inverse
relation between the disclosures and underpricing is expected. If, however, risk
factor disclosures increase investor ex ante uncertainty, there will be a positive
association between the disclosures and underpricing.
For comparison purpose, I first test whether a higher level of risk disclosures
in the IPO prospectus impacts underpricing. I measure the quantity of risk
23 High-technology firms are identified by the SDC new issue database and contain firms in the
IT and biotechnology industry sectors.

disclosures by the number of words in the risk factor section scaled by the total
number of words in the prospectus (RiskFactorWords). Column 1 of Table 3
presents the regression results of equation (8a). The results show that the
number of words in the risk factor section is not significantly associated with
underpricing, consistent with the quantity of risk disclosures itself having no
effect on investor ex ante uncertainty. The coefficients of other control variables
are generally consistent with the literature. For example, IPOs of high-tech
companies are more underpriced. IPOs issued by firms of a larger size, IPOs
backed by venture capitalists, and IPOs experiencing long delays between pro-
spectus registration and exchange listing are less underpriced.
I then decompose the risk factor section into the standard content
(StandardContent) (content from recent IPOs and IPOs from the same industry)
and the unique content (UniqueContent) (residual content). The standard
content component and unique content component of the risk factor disclo-
sures may have differential effect on underpricing. The standard content repeats
the risks and uncertainties that have been contained in the recent IPOs and past
industry IPOs. Such disclosures are general and unlikely to resolve uncertainty
regarding the true share value. Listing generic risk factors could increase inves-
tor ex ante uncertainty as a result of divergent interpretations by investors about
the potential impact of these risk factors on the issuing firm’s performance. The
unique content provides additional information that explains the potential
effects of a disclosed risk on the issuing firm. Compared with a general state-
ment of common risk factors, a detailed discussion may help investors improve
their knowledge about the issuing firm’s expected financial performance and
possible variance of the expected performance, thereby reducing investor ex
ante uncertainty.
Column 2 of Table 3 shows that the unique content component of the risk
factor section (UniqueContent) is negatively associated with underpricing, con-
sistent with the unique content resolving uncertainty regarding the value of the
shares. The standard content component (StandardContent), on the other hand,
is positively associated with underpricing, consistent with the standard content
increasing ex ante uncertainties. The associations of underpricing with the
unique content component and the standard content component are both
economically significant.24 A one standard deviation change in unique content
in the risk factor section is associated with an approximately 1.12% reduction in
initial returns, whereas a one standard deviation change in standard content is
associated with an approximately 2.74% increase in initial returns.25
To ensure that the results are robust to the exclusion of high-tech firms, I
rerun the regression excluding technology firms.26 The results remain relatively
similar. My results show that coefficient on the unique content component is
24 Economic significance is measured by the coefficient times the standard deviation.

25 The economic significance of 1.12% is the product of −0.147 (the coefficient) and 0.076
(standard deviation). Similarly, the economic significance of 2.74% is the product of 0.282
(the coefficient) and 0.097(standard deviation).
26 Out of the 742 IPOs with the first measure of risk disclosures, there are 168 technology IPOs.

Table 3 Risk factor disclosures and IPO underpricing

Exp. Equation Equation Equation Equation Equation
sign (8a) (8b) (8c) (8d) (8e)
Constant 1.117 0.910 1.084 1.096 1.094

(9.30)*** (5.97)*** (9.35)*** (9.41)*** (9.43)***
Ln(Assets) − −0.009 −0.009 −0.009 −0.009 −0.009
(−2.38)** (−2.08)** (−2.32)** (−2.37)** (−2.36)**
Ln(Age) − −0.009 −0.007 −0.008 −0.008 −0.008
(−0.89) (−0.63) (−0.77) (−0.77) (−0.77)
BTM − 0.025 −0.001 0.036 0.027 0.028
(0.37) (−0.01) (0.53) (0.40) (0.41)
HighTech + 0.086 0.094 0.085 0.082 0.082
(3.23)*** (2.98)*** (3.15)*** (3.05)*** (3.03)***
% Retained + 0.044 0.101 0.045 0.042 0.040
(0.82) (1.68)* (0.85) (0.78) (0.76)
Underwriter − −0.220 −0.143 −0.226 −0.220 −0.227
(−0.57) (−0.33) (−0.59) (−0.57) (−0.58)
VC − −0.162 −0.096 −0.168 −0.166 −0.165
(−4.58)*** (−1.99)** (−4.78)*** (−4.68)*** (−4.69)***
Big − −0.042 −0.043 −0.041 −0.040 −0.040
(−1.86)* (−1.77)* (−1.84)* (−1.76)* (−1.80)*
Ln(ASXReturn) + 0.155 0.053 0.124 0.140 0.141
(0.55) (0.17) (0.44) (0.50) (0.50)
%Insider + 0.027 0.026 0.032 0.028 0.027
(0.54) (0.44) (0.63) (0.57) (0.54)
Ln(Delay) − −0.237 −0.229 −0.234 −0.236 −0.235
(−8.31)*** (−7.94)*** (−8.24)*** (−8.27)*** (−8.29)***
RiskFactorWords −0.385
(−0.78)
UniqueContent −0.147
(−1.92)*
StandardContent 0.282
(2.91)***
NegativeWords_Harvard −0.021
(−2.10)**
NegativeWords_L&M −0.005
(−0.51)
CFRiskWords 0.000
(−0.01)
Number of observations 831 742 831 831 831
R2 9.64% 10.37% 9.89% 9.60% 9.57%
This table presents coefficient estimates for equation (8) and in parentheses the associated t-statistics
using heteroskedasticity robust standard errors.
Ln (Underpricing )i ,t = β 0 + β1Ln ( Assets )i ,t + β2 Ln ( Age )i ,t + β3 BTMi ,t + β 4 HighTechi ,t

+ β5 % Retainedi ,t + β6Underwriteri ,t + β7VCi ,t + β8 Big i ,t
+ β9 Ln ( ASXReturn )i ,t − 30 to t + β10 % Insideri ,t + β11Ln ( Delay )i ,t
+ β12 RiskDisclosurei ,t + ε it
Ln(Asset) is ln (pre-IPO total assets in millions), Ln(Age) is ln(1 + the number of years from incor-
poration), Ln(ASXReturn) is ln (1 + the return of ASX All Ordinaries over the 30 trading days
preceding the filing date. Ln(Delay) is ln (the number of days between prospectus registration and
exchange listing). All the other variables are as previously discussed. *, **, and *** indicate signifi-
cance at the 10%, 5%, and 1% levels, respectively.
BTM, book-to-market; VC, venture-capital backing.

negative (−0.149) and statistically significant. Coefficient on the standard

content component is positive (0.219) and significant (untabulated).
Columns 3–5 of Table 3 present the regression results on the relation between
underpricing and other measures of the risk factor disclosures. Column 3 shows
that underpricing is inversely associated with NegativeWords_Harvard. A one
standard deviation change in conservative tones is associated with approxi-
mately 0.02% reduction in initial returns. No significant relation between
underpricing and risk disclosure conservatism is found when the conservative
tones are measured by the number of occurrences of words in Loughran and
McDonald’s (2011) negative words dictionary (NegativeWords_L&M). Since the
issuing firm’s future prospects are uncertain and management has only imper-
fect control over future events, conservative risk disclosures may be perceived
by investors as more credible.27 Underpricing is not significantly related to
CFRiskWords. Overall, the results are less pronounced for the measures of risk
disclosures that capture the conservative tones and cash flow risk.
B. Risk factor disclosures and IPO underpricing –

by lead underwriter prestige
In equation (8), Underwriter is included as a covariate and as a proxy for an
agency rationale for risk factor disclosures. Underwriters deserve special atten-
tion because they are important information intermediaries in the listing
process. More reputable underwriters may help reduce information asymmetry
via their certification role (Beatty and Ritter 1986). To examine the impact of
underwriter prestige on the relation between risk disclosures and underpricing,
I interact the disclosure variables with indicator variables that partition under-
writer market shares at its median (0.1%).
Compared with those with LowUnderwriter indicator variable (Underwriter of
the median or lower), the IPOs with HighUnderwriter indicator variable (Under-
writer of higher than the median) are larger deals by older and larger sized
firms.28 Mean (median) underpricing for the LowUnderwriter and HighUnderwriter
IPOs is 0.297 (0.150) and 0.263 (0.100), respectively. Mean (median)
UniqueContent for the LowUnderwriter and HighUnderwriter IPOs is 1.021 (0.989)
and 1.103 (1.096) and Standard Content is 1.025 (1.029) and 1.004 (1.007),
respectively. The LowUnderwriter IPOs disclose less unique content and more
27 Prior studies document a negative association between negative tones and future earnings
performance (e.g., Balakrishnan and Bartov 2012). Investors revise downward their expec-
tations about the IPO firm’s future performance as a result of the risk disclosures. However,
the range (or dispersion) of possible aftermarket prices affects underpricing (Beatty and Ritter
1986). The results suggest that a more conservative tone in the risk factor section results in
a lower divergence in opinion among investors, leading to less underpricing.
28 For the subsample of 742 IPOs with the second measure of risk disclosures, mean (median)
age for the LowUnderwriter and HighUnderwriter IPOs is 3.151 years (0.943 years) and 6.437
years (3.095 years) and asset is 11.145 million (0.634 million) and 122.976 million (3.840
million), respectively.

standard content in the risk factor section of the prospectus than the
HighUnderwriter IPOs.
Table 4 presents the regression results of equations (8f) and (8g) (and
equations (8b) and (8c)). In equation (8f), I substitute UniqueContent
and StandardContent by UniqueContent*LowUnderwriter and UniqueContent*
HighUnderwriter, and StandardContent* LowUnderwriter and StandardContent*
HighUnderwriter, respectively. As column 2 shows, the relation between under-
pricing and the unique content component of risk factor disclosures is more
negative for IPOs with less reputable underwriters. In equation (8g), I substi-
tute NegativeWords_Harvard by NegativeWords_Harvard*LowUnderwriter and
NegativeWords_Harvard*HighUnderwriter. As column 4 shows, the relation
between underpricing and conservative tones (as measured by the number of
occurrences of words in the Harvard IV-4 negative words dictionary) is also
more negative for IPOs with less prestigious underwriters. Overall, these find-
ings suggest that companies taken public by less reputable underwriters have
more need to provide more unique and conservative risk disclosures in their IPO
prospectuses.
C. Risk factor disclosures and future stock return volatility

Table 5 presents the results of testing the effect of risk factor disclosure on daily
stock return volatility (equations (9a)–(9d)) and negative stock return volatility
in the first year of trading (equations (9e)–(9h)). For equation (9a), the coeffi-
cient on StandardContent is 0.017 and is significant at the 5% level, and the
coefficient on UniqueContent is −0.003 but is statistically insignificant. The
results suggest that the firms that turn out to be relatively risky ex post tend to
disclose more standard content in the risk factor section of their prospectuses.
With respect to control variables, the stock return volatility is inversely related
to firm size (Ln(Assets)), underwriter prestige (Underwriter), and venture-capital
backing (VC), and is positively related to the indicator variable of high-tech
firms (HighTech), the standard deviation of the ASX All Ordinaries daily returns
measured over the 30 trading days preceding the filing date (MSDReturns), and
the indicator variable of negative retained earnings in the year prior to listing
(AccumLossD). The results are consistent with prior studies.
For equation (9e), the coefficient on StandardContent is 0.011 and significant
at the 5% level. The standard content component of risk factor disclosures is
positively associated with the standard deviation of negative stock returns in
the first year of trading (SD(Down)). The coefficient on UniqueContent is negative
but insignificant (−0.002). The positive association between the standard
content and the standard deviation of negative stock returns suggests that firms
with greater downside risk ex post tend to disclose more standard content in the
risk factor section of their IPO prospectuses. With respect to control variables,
SD(Down) is inversely associated with firm size (Ln(Assets)), underwriter reputa-
tion (Underwriter), venture-capital backing (VC), and is positively associated
with the indicator variable of high-tech firms (HighTech), market volatility prior

Table 4 Risk factor disclosures and IPO underpricing with underwriter prestige
interaction
Variable Exp. Equation Equation Equation Equation
sign (8b) (8f) (8c) (8g)
Constant 0.910 0.890 1.084 1.069

(5.97)*** (5.84)*** (9.35)*** (9.21)***
Ln(Assets) − −0.009 −0.008 −0.009 −0.009
(−2.08)** (−1.95)* (−2.32)** (−2.31)**
Ln(Age) − −0.007 −0.006 −0.008 −0.008
(−0.63) (−0.52) (−0.77) (−0.76)
BTM − −0.001 −0.011 0.036 0.040
(−0.01) (−0.13) (0.53) (0.60)
Hightech + 0.094 0.098 0.085 0.089
(2.98)*** (3.07)*** (3.15)*** (3.32)***
% Retained + 0.101 0.104 0.045 0.049
(1.68)* (1.73)* (0.85) (0.93)
Underwriter − −0.143 0.090 −0.226 −0.220
(−0.33) (0.19) (−0.59) (−0.57)
VC − −0.096 −0.095 −0.168 −0.166
(−1.99)** (−2.06)** (−4.78)*** (−4.63)***
Big − −0.043 −0.038 −0.041 −0.041
(−1.77)* (−1.54) (−1.84)* (−1.83)*
Ln(ASXReturn) + 0.053 0.029 0.124 0.127
(0.17) (0.09) (0.44) (0.45)
%Insider + 0.026 0.028 0.032 0.027
(0.44) (0.49) (0.63) (0.54)
Ln(Delay) − −0.229 −0.234 −0.234 −0.231
(−7.94)*** (−7.98)*** (−8.24)*** (−8.17)***
UniqueContent −0.147
(−1.92)*
UniqueContent*LowUnderwriter −0.171
(−1.71)*
UniqueContent*HighUnderwriter −0.070
(−0.70)
StandardContent 0.282
(2.91)***
StandardContent*LowUnderwriter 0.356
(2.76)***
StandardContent*HighUnderwriter 0.206
(1.89)*
NegativeWords_Harvard −0.021
(−2.10)**
NegativeWords_Harvard*LowUnderwriter −0.040
(−2.64)***
NegativeWords_Harvard*HighUnderwriter −0.004
(−0.32)
Number of observations 742 742 831 831
R2 10.37% 10.83% 9.98% 10.26%
This table presents coefficient estimates for equations (8f) and (8g) as well as equations (8b) and (8c)
and in parentheses the associated t-statistics using heteroskedasticity robust standard errors.
LowUnderwriter is an indicator variable, which equals 1 if the lead underwriter(s)’ market share is at
its median or lower and 0 otherwise. HighUnderwriter is an indicator variable, which equals 1 if the
lead underwriter(s)’ market share is higher than the median. *, **, and *** indicate significance at the
10%, 5%, and 1% levels, respectively.

Table 5 Risk factor disclosure and future stock return volatility
Variable SD(Returns) SD(Down)
Equation (9a) Equation (9b) Equation (9c) Equation (9d) Equation (9e) Equation (9f) Equation (9g) Equation (9h)
Constant 0.026 0.041 0.041 0.041 0.016 0.025 0.025 0.025

(2.76)*** (12.47)*** (12.58)*** (12.41)*** (2.52)** (11.03)*** (11.09)*** (11.01)***
Ln(Assets) −0.002 −0.002 −0.002 −0.002 −0.001 −0.001 −0.001 −0.001
(−6.12)*** (−6.36)*** (−6.32)*** (−6.56)*** (−4.79)*** (−5.15)*** (−5.11)*** (−5.32)***
Ln(Age) −0.001 −0.001 −0.001 −0.001 −0.001 −0.001 −0.001 −0.001
(−1.10) (−1.30) −1.29 (−1.32) (−1.20) (−1.39) (−1.37) (−1.41)
BTM 0.009 0.005 0.004 0.005 0.008 0.004 0.004 0.004
(1.13) (0.66) 0.54 (0.69) (1.34) (0.77) (0.70) (0.83)
MSDReturns 2.323 2.209 2.190 2.215 1.673 1.596 1.587 1.600
(6.90)*** (6.55)*** (6.51)*** (6.58)*** (7.23)*** (6.84)*** (6.82)*** (6.87)***
Big −0.003 −0.005 −0.004 −0.005 −0.002 −0.003 −0.002 −0.003
(−1.23) (−2.26)** (−2.10)** (−2.27)** (−1.16) (−1.97)** (−1.86)* (−1.99)**
HighTech 0.006 0.007 0.007 0.006 0.005 0.005 0.005 0.005
(2.27)** (2.91)** (2.95)** (2.81)** (3.06)*** (3.40)** (3.46)*** (3.31)***
Underwriter −0.147 −0.134 −0.132 −0.135 −0.094 −0.086 −0.085 −0.087

(−4.88)*** (−4.60)*** (−4.50)*** (−4.60)*** (−4.40)*** (−4.22)*** (−4.17)*** (−4.25)***
VC −0.012 −0.012 −0.012 −0.012 −0.007 −0.007 −0.007 −0.007
(−5.20)*** (−5.64)*** (−5.51)*** (−5.59)*** (−4.30)*** (−4.28)*** (−4.20)*** (−4.32)***
Leverage −0.002 −0.005 −0.005 −0.005 −0.002 −0.003 −0.004 −0.003
(−0.50) (−1.15) −1.2 (−1.14) (−0.92) (−1.77)* (−1.85)* (−1.76)*
AccumLossD 0.016 0.017 0.017 0.017 0.009 0.010 0.010 0.010
(7.83)*** (8.89)*** (8.72)*** (8.84)*** (6.49)*** (7.40)*** (7.25)*** (7.48)***
UniqueContent −0.003 −0.002
(−0.50) (−0.49)
StandardContent 0.017 0.011
(2.01)** (1.99)**
NegativeWords_Harvard 0.000 0.000
(−0.51) (0.06)
NegativeWords_L&M −0.002 −0.001
(−2.42)** (−1.58)
CFRiskWords 0.001 0.001
(0.96) (1.37)
Number of observations 742 831 831 831 742 831 831 831
R2 28.41% 29.25% 29.79% 29.31% 25.48% 26.16% 26.43% 26.35%
This table presents coefficient estimates for equation (9) and in parentheses the associated t-statistics using heteroskedasticity robust standard errors. *, **, and *** indicate significance at the
10%, 5%, and 1% levels, respectively.
Volatilityi ,t +1 = δ 0 + δ 1 Ln ( Assets)i ,t + δ 2 Ln ( Age )i ,t + δ 3 BTMi ,t + δ 4 MSDReturnsi ,t −30 to t + δ 5 Big i ,t + δ 6 HighTechi ,t + δ 7Underwriteri ,t

+ δ 8VCi ,t + δ 9 Leveragei ,t + δ 10 AccumLossDi ,t + δ 11 RiskDisclosurei ,t + ε i ,t
BTM, book-to-market; VC, venture-capital backing; SD, standard deviation.
99
to listing (MSDReturns), and the indicator variable of negative retained earnings

in the year prior to listing (AccumLossD).
Next, I examine the association between risk disclosure conservatism and
daily stock return volatility in the first year of trading. Risk disclosure conser-
vatism is measured by the number of occurrences of words in the Harvard-IV-4
negative words dictionary (equation (9b)) and the number of occurrences of
words in Loughran and McDonald (2011) negative words dictionary (equation
(9c)). Columns 2 and 3 of Table 5 present regression results. NegativeWords_L&M
is inversely associated with the standard deviation of daily stock return volatil-
ity (SD(Returns)). The coefficient on NegativeWords_L&M is −0.002 and signifi-
cant at the 5% level. However, the economic significance is minimal. No
significant relation is found between NegativeWords_Harvard and daily stock
return volatility (SD(Returns)). I rerun the regressions for non-tech IPOs in the
sample and obtain similar results.29 The coefficient on NegativeWords_L&M is
−0.003 and significant at the 5% level (untabulated). The economic significance
is also minimal. For the subsample of technology IPOs, there is a positive but
statistically insignificant relation of SD(Returns) to NegativeWords_Harvard and
NegativeWords_L&M, respectively, reflecting the difficulty in valuing technology
firms due to the greater uncertainty regarding product development and sales in
high-tech industries.
The regression results of equation (9f) and equation (9g) show that both
NegativeWord_Harvard and NegativeWords_L&M are not significantly related to
the volatility of negative stock returns (SD(Down)). I rerun the regressions for
the non-tech IPOs. I find NegativeWords_L&M has a marginal inverse association
with SD(Down) (coefficient = 0.002, t = −1.91). The economic significance is
minimal. NegativeWord_Harvard is not significantly associated with SD(Down)
either. Finally, I do not find any significant relation between CFRiskWords and
the stock return volatility.
D. Determinants of risk factor disclosures

In this section, I study the determinants of risk factor disclosures. The literature
on voluntary disclosures suggests that the disclosure decision is determined by
the issuer weighing benefits against costs. I use whether the issuer is a high-tech
company as a proxy for the influence of proprietary costs on disclosure. High-
tech companies may provide less informative risk to avoid revealing proprietary
information to rivals (Verrecchia 1983). However, there may be of little benefit
in withholding negative information as the proprietary value of negative infor-
mation to rivals is potentially low and it is unlikely for negative information to
be concealed for long (Hanley and Hoberg 2012). Thus, it is ex ante unclear as
to whether high-tech companies provide less information about risks in their
IPO prospectus. I proxy the influence of agency costs and corporate governance
29 Out of the 831 IPOs with the second and the third measures of risk disclosures, there are 204
technology IPOs.

on risk disclosures with the extent of retained ownership, the monitoring role
played by professional advisers (underwriters, auditors, and legal counsel), and
whether the IPO is backed by venture capitalists (Barry et al. 1990; Lerner 1995;
Baker and Gompers 2003; Hochberg 2012). I use firm age to proxy the effects of
uncertainty about whether the manager is informed. Firms of an older age are
expected to provide more informative disclosure as managers have more knowl-
edge about the firm. To proxy the extent to which investors know of a company
before it goes public, I control for the number of press stories in the year prior
to the IPO (PreIPONews). To the extent PreIPONews substitutes for disclosure in
the prospectus (Schrand and Verrecchia 2005), it is expected to be inversely
related to the amount of risk disclosures in the prospectus. I also control for firm
size, leverage, whether the firm has negative retained earnings in the year prior
to the IPO, and market volatility over the 30 trading days prior to the filing date.
The estimation equation is stated below:
RiskDisclosurei ,t = β 0 + β1Ln ( Assets )i ,t + β2 Ln ( Age )i ,t + β3 BTM i ,t + β 4 HighTechi ,t
+ β5 % Retainedi ,t + β6 % Insideri ,t + β7 Leverage + β8Underwriteri ,t
+ β9VCi ,t + β10 Big i ,t + β11Lawyeri ,t + β12 MSD ( Returns )i ,t −1
+ β13 AccumLossDi ,t + β14 Ln ( PreIPONews ) + ε i ,t (10)
where:
Leverage = Pre-IPO long-term debt ÷ Pre-IPO total assets

Lawyer = the issuing firm’s legal counsel’s dollar market share over the sample
period
Ln(PreIPONews) = ln (1 + number of news stories in the year prior to going public)
All the other variables are the same as previously defined.

Columns 1 and 2 of Table 6 present the results of the equation (10) regres-
sion, where the dependent variables are the unique content and standard
content of the risk factor section, respectively. UniqueContent is positively asso-
ciated with firm age (Ln(Age)), book-to-market (BTM) and leverage (Leverage),
and is inversely associated with the number of news stories appearing in the
year prior to the IPO (Ln(PreIPONews)), venture-capital backing (VC), the indi-
cator variable of accumulated losses prior to listing (AccumLossD), and the
standard deviation of the ASX All Ordinaries daily returns over the 30 trading
days preceding the filing date (MSDReturns). The positive relation between firm
age and UniqueContent is consistent with the prediction that firms with a longer
operating history are more experienced and better informed, and therefore are
able to provide more specific disclosures about investment risk and uncertain-
ties. The negative coefficient for VC is consistent with the corporate governance
role of venture capitalists. The negative coefficients for MSDReturns and
AccumLossD suggest that issuers tend to provide less specific risk disclosures
when there is a greater amount of risk or uncertainty regarding the market and
the business. The results also support that the pre-IPO news stories may substi-
tute for firm disclosure and that venture-capital-backed IPOs have fewer incen-
tives to provide specific risk disclosures.

Table 6 Determinants of risk factor disclosures
102
UniqueContent StandardContent NegativeWords_Harvard NegativeWords_L&M CFRiskWords
Equation (10a) Equation (10b) Equation (10c) Equation (10d) Equation (10e)
Constant 1.104 1.090 −0.101 −0.094 0.141

(40.63)*** (48.74)*** (−0.62) (−0.55) (0.89)
Ln(Assets) 0.001 0.003 0.013 0.002 0.017
(0.42) (1.64) (1.12) (0.16) (1.34)
Ln(Age) 0.0349 0.0137 0.0114 0.0003 0.0120
(5.44)*** (2.46)** (0.29) (0.01) (0.30)
BTM 0.075 0.013 0.263 −0.326 −0.273
(1.97)** (0.38) (1.22) (−1.33) (−1.08)
HighTech 0.019 −0.037 0.148 0.038 0.190
(1.42) (−2.97)** (1.74)* (0.42) (2.2)**
%Retained 0.050 −0.087 0.202 0.285 −0.141
(1.60) (−3.44)** (1.10) (1.50) (−0.79)
%Insider −0.002 −0.020 0.070 0.216 0.303
(−0.06) (−0.67) (0.31) (0.94) (1.34)
Leverage 0.050 −0.028 −0.133 −0.131 −0.016
(2.08)** (−1.54) (−1.30) (−1.08) (−0.11)
Underwriter 0.160 −0.170 −0.324 1.204 0.062
(0.78) (−0.90) (−0.25) (0.92) (0.05)
VC −0.079 −0.057 −0.144 −0.019 0.067
(−3.21)*** (−2.09)** (−0.84) (−0.10) (0.44)
Big 0.004 −0.028 −0.022 0.147 −0.023
(0.30) (−2.89)** (−0.30) (1.89)* (−0.30)
Lawyer 0.101 −0.027 −0.958 −0.411 2.797
(0.53) (−0.16) (−0.81) (−0.36) (2.47)**
MSD(Returns) −5.019 2.138 −7.678 −11.805 −4.409
(−2.38)** (1.22) (−0.63) (−1.05) (−0.38)
AccumLossD −0.061 0.000 −0.149 −0.104 −0.169
(−4.71)*** (0.02) (−1.85)* (−1.26) (−2.08)**
Ln(PreIPONews) −0.024 −0.015 0.048 0.029 −0.013
(−4.70)*** (−4.03)*** (1.74)* (0.97) (−0.44)

Observations 742 742 831 831 831
R2 20.75% 8.77% 3.12% 2.52% 4.43%
This table presents coefficient estimates for equation (10) and in parentheses the associated t-statistics using heteroskedasticity robust standard errors. The dependent variables in equations
(10a)–(10e) are UniqueContent, StandardContent, NegativeWords_Harvard, NegativeWords_L&M, and CFRiskWords, respectively.
RiskDisclosurei ,t = β 0 + β1 Ln ( Assets )i ,t + β2 Ln ( Age )i ,t + β3 BTM + β 4 HighTechi ,t + β5 % Retainedi ,t + β6 % Insideri ,t + β7 Leverage

+ β8Underwriteri ,t + β9VCi ,t + β10 Big i ,t + β11 Lawyeri ,t + β12 MSD ( Returns )i ,t −1 + β13 AccumLossDi ,t
+ β14 Ln ( PreIPONews )i ,t + ε i ,t
Ln(PreIPONews) is ln (1 + the number of news stories about the issuing firm in the year prior to going public). All the other variables are as previously discussed. *, **, and *** indicate significance
at the 10%, 5%, and 1% levels, respectively.

The standard content is positively related to firm age and inversely related to
the number of news stories appearing in the year prior to the IPO (preIPONews),
the indicator variable of high-tech firms (HighTech), venture-capital backing
(VC), auditor size (Big), and ownership retention (%Retained). Most explanatory
variables are not significantly related to conservative tones and the words
indicating cash flow risk (equations (10c)–(10e)). In equation (10e), the issuer’s
legal counsel’s market share (Lawyer) is positively associated with CFRiskWords.
In equation (10d), Big N auditors are positively associated with conservative
tones (NegativeWords_L&M). These results are consistent with the monitoring
roles of the professional advisors.
E. Additional tests
i. Test of heterogenous beliefs explanation

In this section, I examine whether the inverse relation between risk disclosures
and underpricing is also consistent with an alternative explanation that stems
from the behavioral finance literature. This explanation suggests that IPOs are
not underpriced per se, but that first-day investor optimism drives positive
initial returns (Ritter 1991). This explanation argues that because of the hetero-
geneity of post-IPO investor beliefs and limits to arbitrage, the most opportu-
nistic investors drive first-day prices, resulting in IPOs having high initial
returns and low long-run returns (Miller 1977). I perform additional tests to
examine whether the inverse relation between informative risk disclosures and
underpricing is due to the disclosures that reduce pre-IPO uncertainty or
because the disclosures reduce heterogeneity of post-IPO investor beliefs. If the
first aftermarket prices are too high for IPOs with uninformative risk disclosures
but not for informative risk disclosures, then there would still be a negative
association between the informativeness of risk disclosures and underpricing
even if the information asymmetry effect does not exist (Ritter and Welch
2002). The alternative explanation implies that IPOs with more informative risk
disclosures should have less dispersion in analysts’ earnings growth forecasts
and annual EPS forecasts (a proxy for heterogeneity of investor beliefs) and
more favorable long-run returns.
ii. Risk factor disclosures and dispersion in analysts’ forecasts

In this section, I examine whether IPOs with more informative risk disclosures
have less dispersion in analysts’ forecasts. Because forecasts are available only
for IPOs that attract analyst following, I employ Heckman’s (1979) two-stage
approach (Rajan and Servaes 1997; Wooldrige 2009; Tucker 2010). In the first
stage, I include proxies for the potential selection bias that arises from analysts’
choice to follow a particular firm (Rajan and Servaes 1997). These include firm
size and the number of seasoned firms in the same industry at the time of the
IPO. I estimate equation (11) to produce inverse Mills ratios.

AnalystFollowing = β 0 + β1Ln ( MVE )i ,t + β2 Ln ( IndustrySize )i ,t + β3 Ln ( Asset )i ,t

+ β 4 Ln ( Age )i ,t + β5 Leveragei ,t + β6 HighTechi ,t (11)
+ β7 RiskDisclosurei ,t + ε i ,t
where AnalaystFollowing is an indicator variable which equals 1 if two or more

analysts provide either long-term earnings growth or annual EPS forecasts on
the Institutional Brokers Estimate System (IBES) within a year of a company’s
IPO date and 0 otherwise, Ln(MVE) is the natural logarithm of market value of
equity in million dollars at the time of the IPO, and Ln(IndustrySize) is the
natural logarithm of the number of firms in the same 2-digit GICS industry
sector as the IPO firm. A seasoned firm is one that is listed on ASX for more than
3 years at the time the IPO firm goes public. All the other variables are as
previously defined. These variables [i.e., Ln(Assets), Ln(Age), Leverage, HighTech,
and RiskDisclosure] are the explanatory variables in stage 2 and form a strict
subset of explanatory variables in stage 1 (Wooldrige 2009; Lennox et al. 2012).
Equation (11) is estimated using probit on all observations, assuming a nor-
mally distributed error term.
I then regress the dispersion in analysts’ long-term earnings growth forecasts
or dispersion in analysts’ annual EPS forecasts on my risk disclosure variables of
interest and controls for company characteristics. I control for selection bias by
adding the inverse Mills ratio.
GrowthDispersioni ,t +1 = β 0 + β1 Ln ( Assets )it + β2 Ln ( Age )it + β3 Leverageit

(12a)
+ β 4 HighTechit + β5 RiskDisclosureit + β6 Lambdait + ε it
EPSDispersioni ,t +1 = β 0 + β1Ln ( Assets )it + β2 Ln ( Age )it + β3 Leverageit + β 4 HighTechit

+ β5 RiskDisclosureit + β6 Lambdait + ε it (12b)
where GrowthDispersion is the dispersion in IBES analysts’ long-term earnings

growth forecasts, EPSDispersion is the dispersion in IBES analysts’ annual EPS
forecasts, Lambda is the inverse Mills ratio from equation (11), and all the other
variables are as previously defined. I calculate forecast dispersion by dividing the
standard deviation of IBES forecasts by the absolute value of the mean forecast.
Equations (12a) and (12b) are estimated using ordinary least squares on the
subsample of companies that have two or more analysts providing long-term
earnings growth or annual EPS forecasts on IBES within a year of a company’s
IPO date.
Table 7 presents the results. Only 6% and 16% of the sample IPOs have at
least two analysts providing long-term earnings growth and annual EPS fore-
casts on IBES within a year of the firm’s IPO date, respectively. As the large
majority of the standard deviations in analysts’ forecasts pertain to only two
analyst forecasts, measures of dispersion for my sample are likely to be noisy
relative to measures for more established companies. As shown in Panel A, the
coefficients for UniqueContent and StandardContent are negative, but not signifi-
cantly different from 0, when the dispersion in analysts’ long-term growth

Table 7 Dispersion in analyst forecasts
Panel A
Variable Dispersion in Dispersion in Dispersion Dispersion
Analyst earnings growth earnings growth Analyst in annual in annual
following forecasts forecasts following EPS forecasts EPS forecast
Equation (11) Equation (12a) OLS regression Equation (11) Equation (12b) OLS regression
Constant −8.334 1.078 1.348 −5.825 0.737 0.545
(−4.17)*** (1.12) (1.88)* (−4.67)*** (2.68)*** (1.98)**
Ln(MVE) 1.180 0.954
(6.64)*** (8.47)***
Ln(IndSize) −0.141 −0.471
(−0.68) (−3.52)***

Ln(Assets) 0.168 0.006 −0.018 0.101 −0.055 −0.038
(2.31)** (0.10) (−0.33) (2.17)** (−2.66)*** (−2.89)***
Ln(Age) 0.083 −0.054 −0.038 0.142 −0.056 −0.053
(0.77) (−0.66) (−0.70) (1.68)* (−3.03)*** (−2.90)***
Leverage −0.153 −0.066 −0.119 −0.280 −0.081 −0.088
(−0.37) (−0.22) (−0.40) (−0.91) (−0.86) (−1.03)
HighTech 0.620 0.207 0.153 −0.069 0.048 0.069
(2.08)** (0.82) (0.97) (−0.35) (1.01) (1.41)
UniqueContent −0.725 −0.328 −0.285 0.807 −0.100 −0.060
(−0.67) (−0.58) (−0.57) (1.18) (−0.63) (−0.37)
StandardContent 2.499 −0.279 −0.389 2.122 −0.115 −0.108
(2.21)** (−0.33) (−0.56) (2.80)*** (−0.45) −0.42
Lambda 0.118 −0.094

(0.37) (−1.56)
Observations 742 46 46 742 109 109
Pseudo R2 56.89% n/a n/a 50.94% n/a n/a
R2 n/a 5.97% 4.74% n/a 23.11% 20.37%
105
106
Panel B
Variable Dispersion in Dispersion in Dispersion Dispersion
Analyst earnings growth earnings growth Analyst in annual in annual
following forecasts forecasts following EPS forecasts EPS forecast
Equation (11)’ Equation (12a)’ OLS regression Equation (11)’ Equation (12b)’ OLS regression
Constant −5.682 0.343 0.454 −2.613 0.567 0.370
(−5.51)*** (0.78) (1.36) (−4.34)*** (3.28)*** (4.3)***
Ln(MVE) 1.012 0.736
(6.86)*** (7.62)***
Ln(IndSize) −0.121 −0.402
(−0.70) (−3.56)***
Ln(Assets) 0.111 0.040 0.025 0.240 −0.062 −0.034
(1.68)* (0.72) (0.42) (4.49)*** (−2.29)** (−2.24)**
Ln(Age) 0.059 −0.037 −0.032 0.167 −0.067 −0.061
(0.61) (−0.50) (−0.54) (2.2)** (−3.47)*** (−3.22)***
Leverage −0.472 −0.054 −0.058 −0.274 −0.072 −0.069
(−1.24) (−0.20) (−0.22) (−0.96) (−1.18) (−1.23)
HighTech 0.368 0.222 0.203 0.020 0.068 0.096
(1.48) (1.05) (1.32) (0.11) (1.62) (2.5)**
NegativeWords_Harvard 0.054 −0.019 −0.014 −0.225 0.019 0.008
(0.47) (−0.29) (−0.19) (−2.75)*** (0.80) (0.30)
Lambda 0.054 −0.103

(0.20) (−1.49)
Observations 831 53 53 831 133 133
Pseudo R2 53.92% n/a n/a 51.16% n/a n/a
R2 n/a 4.20% 4.01% n/a 19.77% 17.56%
This table reports the results from regression equations (12a) and (12b) as well as OLS regressions estimating the relation between risk disclosure
informativeness and analysts’ forecast dispersion. Panel A presents the results with UniqueContent and StandardContent as the variables of interest.
Panel B presents the results with NegativeWords_Harvard as the variable of interest. t-statistics are provided in parentheses. *, **, and *** indicate
significance at the 10%, 5%, and 1% levels, respectively.
OLS, ordinary least squares.

forecasts is the dependent variable. Similar results are obtained when the dis-
persion in analysts’ annual EPS forecasts is the dependent variable. The coeffi-
cients for NegativeWords_Harvard are statistically insignificant when the
dispersion in analysts’ long-term growth forecasts and the dispersion in ana-
lysts’ annual EPS forecasts are the dependent variables, respectively (Panel B).
Taken together, these results provide no support for an inverse relation between
risk factor disclosures in the IPO prospectus and analysts’ forecast dispersion.30
iii. Risk factor disclosures and long-run returns

Next, I examine whether IPOs with more informative disclosure of risk factors
have better abnormal long-run returns. I compute buy-and-hold return for IPO
firms as the return on a buy-and-hold investment in the firm less the return on
a buy-and-hold investment in its benchmark stock. I use three benchmarks for
the benchmark return. The first benchmark is the ASX All Ordinaries index. The
second benchmark is the matching firm based on the industry and firm size,
similar to that used in Loughran and Ritter (1995). The third benchmark is the
matching firm based on the industry, sales, and EBITDA margin.31 I compute the
buy-and-hold return for the 3-year holding period. Matching firms are included
in the computation for the full holding period or until the IPO firm is delisted.
To address survivorship issue, the total return is truncated as of the delisting
date if an IPO is delisted before the 3-year anniversary date (Loughran and Ritter
1995). If a matching firm is delisted before the end of the holding period or
before the IPO firm’s delisting day, whichever is earlier, a new matching firm is
drawn from the original list of matching candidates.
Panels A, B, and C of Table 8 present the 3-year buy-and-hold abnormal
return (BHAR) computed relative to the return of: (i) the ASX All Ordinary
Index; (ii) matching firms based on industry and first-day closing market capi-
talization; (iii) matching firms based on industry, sales, and EBITDA margin,
respectively. In all the three panels, I present the means and medians of BHARs
for the subsample of IPOs with less unique content of risk disclosures and the
subsample of IPOs with more unique content, and the difference in BHARs
between subsamples of less and more unique content in the risk factor section
of the IPO prospectus. The results show that IPO firms disclosing more unique
content of risk factors do not experience significantly better post-IPO long-run
stock returns than IPO firms disclosing less unique content, contrary to the
direction predicted under the heterogeneous beliefs hypothesis (i.e., IPO firms
disclosing more unique content about risk factors have better long-run abnor-
mal return). The median BHAR computed relative to the ASX All Ordinaries
index for the subsample of less unique content (−0.378) is significantly larger
than the median for the subsample of more unique content (−0.529).
30 I also use the IPO firm’s trading volume in shares adjusted by market average volume in
shares on the first trading date as a measure of dispersion in investors’ beliefs and find no
significant relation between the unique content and the adjusted trading volume.
31 See ‘Risk factor disclosures and IPO valuation’ section for detailed discussions on the match-
ing procedure.

Table 8 Three-year buy-and-hold abnormal returns grouped by the uniqueness of

risk factor disclosures
Panel A: ASX All Ordinaries index as the benchmark
Less unique More unique Less unique – more unique
N = 371 N = 371
Mean Median Mean Median t-test Wilcoxon z-test
BHAR 0.042 −0.378 −0.042 −0.529 0.772 3.153***
(0.440) (−0.002)
Panel B: Industry and size matched firm as the benchmark
N = 371 N = 371
BHAR −0.233 −0.029 −0.080 −0.087 −0.951 0.255
(0.342) (0.799)
Panel C: Industry, sales, and EBITDA margin matched firm as the benchmark
N = 109 N = 109
BHAR 0.088 0.015 −0.156 −0.056 1.437 1.223
(0.152) (0.221)
This table reports mean and median 3-year buy-and-hold abnormal returns (BHAR) earned by IPO
firms in portfolios formed on the uniqueness of risk factor disclosures in the IPO prospectus. IPO
firms with less or more unique risk factor disclosures are those firms with unique content below
or above the median, respectively. The BHARs are computed relative to ASX All Ordinaries index,
matching firms based on industry and first-day closing market capitalization, and matching firms
based on industry, sales, and EBITDA margin, respectively. BHAR is computed as the return on a
T T
buy-and-hold investment in its benchmark stock. BHARi ,T = ∏ (1 + Ri ,t ) − ∏ [1 + E ( Ri ,t )]. Rit is
t =1 t =1
return on an IPO firm i in month t. E(Rit) is the benchmark return. T equals the 3-year anniversary
date or the delisting date of the IPO firm i, whichever occurs sooner. *** indicate significance at
the 1% level.
I further run the following regression on the relation between unique

content of risk disclosures and 3-year abnormal returns, controlling for other
variables that may affect long-run returns.
BHARi ,t = ρ0 + ρ1Ln ( Assets)i ,t + ρ2 Ln ( Age )i ,t + ρ3 BTMi ,t + ρ4 HighTechi ,t
+ ρ5 % Retainedi ,t + ρ6 Underwriteri ,t + ρ7VCi ,t + ρ8 Big i ,t (13)
+ ρ9 Ln (Underpricing )i ,t + ρ10UniqueContent i ,t + ε i ,t
I present the results from the above regression in Table 9. In column 1, I

measure BHAR relative to the return of ASX All Ordinaries index. In columns 2

Table 9 Multivariate tests – buy-and-hold abnormal return and unique content of

risk factor disclosures
Variable Exp. BHAR (ASX BHAR (industry BHAR (industry,
sign All Ordinaries and size matched sales, EBITDA margin
as the firm as the matched firm as the
benchmark) benchmark) benchmark)
Constant 0.436 −0.271 −0.156
(1.22) (−0.50) (−0.21)
Ln(Assets) + −0.037 0.020 −0.009
(−2.39)** (0.56) (−0.14)
Ln(Age) + 0.013 0.052 −0.026
(0.21) (0.60) (−0.31)
BTM + 0.711 0.384 0.595
(2.22)** (0.62) (1.20)
HighTech − −0.401 −0.037 −0.109
(−3.08)*** (−0.20) (−0.59)
% Retained + 0.102 −0.049 0.853
(0.38) (−0.12) (1.81)*
Underwriter + 2.962 0.373 4.123
(1.39) (0.15) (1.75)*
VC + 0.190 0.353 −0.434
(0.56) (1.02) (−1.04)
Big + 0.165 0.435 −0.207
(1.25) (2.54)** (−1.15)
Ln(Underpricing) − −0.247 −0.278 −0.143
(−1.79)* (−1.19) (−0.57)
UniqueContent −0.511 −0.071 −0.310
(−1.58) (−0.15) (−0.54)
Number of observations 742 742 218
R2 2.88% 1.85% 5.13%
The dependent variable is 3-year buy-and-hold abnormal return computed relative to ASX All
Ordinary Index, matching firms based on industry and first-day closing market capitalization,
and matching firms based on industry, sales, and EBITDA margin, respectively. The independent
variable is the measure of unique content in the risk factor section of the IPO prospectus
(UniqueContent). Control variables consist of Ln(Assets), Ln(Age), BTM, HighTech, %Retained, Under-
writer, VC, Big, and Ln(Underpricing), all as previously defined. t-statistics are provided in paren-
theses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
and 3, I measure BHAR relative to the return of industry and size matched firms
and the return of industry-sales-EBITDA margin matched firms, respectively. In
columns 1–3, the coefficients on the independent variable (UniqueContent) are
all statistically insignificant. The results do not show that firms disclosing more
unique content about risk factors perform better in the 3-year holding period
post IPO, and thus do not support the alternative explanation for the inverse
relation between informative risk factor disclosures and underpricing. In terms
of control variables, column 1 shows that BTM is positively related to post-
listing performance and firms in the high-tech industry have lower BHAR.

Column 1 shows a negative coefficient on Ln(Assets), but in columns 2 and 3,

firm size does not have significant explanatory power. In column 3, retained
ownership (%Retained) has a significant positive relationship with the
3-year posting listing returns. In column 1, initial returns (Ln(Underpricing)) are
significantly inversely associated with 3-year post-listing returns, consistent
with Ritter (1991).
iv. Risk factor disclosures and IPO valuation

In this section, I examine the relation between risk factor disclosures and IPO
valuation both in the IPO market and in the immediate aftermarket. Following
prior studies on IPO valuation (Purnanandam and Swaminathan 2004;
Chemmanur and Yan 2009), I match each IPO firm with a non-IPO industry
peer with comparable sales and EBITDA profit margin that did not go public
within the past 3 years. To select an appropriate matching firm, I first consider
all ASX listed firms for the fiscal year prior to the IPO year. From these firms, I
select those that are incorporated in Australia and did not go public during the
past 3 years. For the firms selected in the previous step, I group all these firms
into 24 industry groups using the 4-digit GICS codes. I group firms in each
industry group into three portfolios based on sales revenue in the fiscal year
prior to the IPO year and then each sales portfolio into three portfolios based on
EBITDA margin (EBITDA/Sales) in the prior year. When there are not enough
firms in an industry group, I construct only two sales portfolios or two EBITDA
margin portfolios. Each IPO is then matched to the appropriate industry-sales
EBITDA margin portfolio. From this portfolio, I find a matching firm that is
closest in sales to the IPO firm. In this way, I assign each IPO firm a unique
matching firm in a given cohort year. I also use the median firm in the
industry-sales EBITDA margin portfolio (based on the medians on certain price
multiples discussed later) to compute the benchmark value for each IPO
firm.
For each IPO firm, I compute a price-to-value (P0/V) ratio, where P0 is the
offer price and V is the benchmark value computed from comparable firms’
market multiples and IPO firm’s sales, EBITDA, or book value. IPO firms
with negative sales, EBITDA, or book values prior to listing are excluded
(Purnanandam and Swaminathan 2004; Chemmanur and Yan 2009). Since
many IPO firms in my sample have negative EBITDA, the sample size is reduced
to 218 firms when the price-to-value ratio is estimated based on EBITDA. This is
smaller than the number of firms included in the tests when the ratio is
estimated based on sales (461) or book value (399). The use of positive EBITDA
as a sample-selection criterion excludes many small and young firms. My
calculation of P0/V ratio follows Purnanandam and Swaminathan (2004) and
Chemmanur and Yan (2009). The P0/V ratio for the IPO is computed by
dividing the IPO offer price multiple by the comparable firm’s market
multiple.
The offer price multiples for IPOs are computed as follows:

⎛ P 0 ⎞ = Offer price × Shares outstanding

⎝ S ⎠ IPO (14a)
Prior fiscal year sales

⎝ EBITDA ⎠ IPO (14b)
Prior fiscal year EBITDA

⎝ B ⎠ IPO (14c)
Proformabook value
where Shares outstanding refers to the number of shares outstanding at the close
of the offer date (i.e., the first day the issue is traded publicly).
The price multiples for the matching firms are computed as follows:
⎛ P⎞ Market price × Shares outstanding

= (14d)
⎝ S ⎠ Match Prior fiscal year sales
⎛ P ⎞ Market price × Shares outstanding

= (14e)
⎝ EBITDA ⎠ Match Prior fiscal year EBITDA
⎛ P⎞ Market price × Shares outstanding

= (14f)
⎝ B ⎠ Match Prior fiscal year book value
where Market price is the share price and Shares outstanding is the number of
shares outstanding for the matching firm, both are at the close of the day
immediately prior to the IPO offer date.
The P0/V ratios of the IPO firm based on various price multiples are com-
puted as follows:
⎛ P0 ⎞
⎛ P 0 ⎞ ⎝ S ⎠ IPO
= (14g)
⎝ V ⎠ sales ⎛ P ⎞
⎝ S ⎠ Match
⎛ P0 ⎞
⎛ P 0 ⎞ ⎝ EBITDA ⎠ IPO
= (14h)
⎝ V ⎠ EBITDA ⎛ P ⎞
⎝ EBITDA ⎠ Match
⎛ P0 ⎞
⎛ P 0 ⎞ = ⎝ B ⎠ IPO (14i)
⎝ V ⎠B ⎛ P⎞
⎝ B ⎠ Match
I also measure the level of valuation of each IPO stock in the immediate
aftermarket by the following ratios:

⎛ P1⎞
⎛ P1⎞ ⎝ S ⎠ IPO
= (14j)
⎝ V ⎠ sales ⎛ P ⎞
⎝ S ⎠ Match
⎛ P1 ⎞
⎛ P1⎞ ⎝ EBITDA ⎠ IPO
= (14k)
⎝ V ⎠ EBITDA ⎛ P ⎞
⎝ EBITDA ⎠ Match
⎛ P1⎞
⎛ P1⎞ = ⎝ B ⎠ IPO (14l)
⎝ V ⎠B ⎛ P⎞
⎝ B ⎠ Match
⎛ P1⎞ ⎛ P1 ⎞ ⎛ P1⎞
where , , and are similar to the ratios as defined in
⎝ S ⎠ IPO ⎝ EBITDA ⎠ IPO ⎝ B ⎠ IPO
equations (14d)–(14f), except that the Offer price (P0) is replaced by the first-day
closing price of the IPO firms (P1).
Table 10 presents the 25th, 50th, and 75th percentiles of the cross-sectional
distributions of P/V ratios based on P/S, P/EBITDA, and P/B multiples, respec-
tively. Panel A of Table 10 shows that the median P0/V ratios based on P/S,
⎛ P0 ⎞ ⎛ P0 ⎞ ⎛ P0 ⎞
P/EBITDA, and P/B multiples (i.e., , , and ) based on
⎝ V ⎠ sales ⎝ V ⎠ EBITDA ⎝ V ⎠B
matching firms are 1.615, 1.272, and 0.895, respectively. Panel B of Table 7
shows that the median P1/V ratios based on P/S, P/EBITDA, and P/B multiples
⎛ P1⎞ ⎛ P1⎞ ⎛ P1⎞
(i.e., , , and ) based on matching firms are 1.927, 1.478,
and 1.066, respectively. The median values of P0/V and P1/V ratios based on the
medians of matching portfolios are similar.32
I run a series of regressions to examine the relation between risk factor
disclosures and the P/V ratios for the subsample of IPOs with matching firms
available. I first study the relation between risk factor disclosures in the pro-
spectus and the P0/V ratios. Table 11 presents the regression results. For the
⎛ P0 ⎞
regressions with Ln as the dependent variable where the benchmark
⎝ V ⎠ sales
value V for each IPO firm is computed based on its matching firms’ Price/Sales,
the coefficient on the unique content is positive and significant at the 5% level
and the coefficient on the standard content is negative and significant at the 1%
⎛ P0 ⎞
level. When Ln is computed based on the median firm of the matching
⎝ V ⎠ sales
32 Similar to Chemmanur and Loutskina (2006), Kim and Ritter (1999), Zheng (2007), and
Johnson et al. (2010), I consider the benchmarking technique a way to produce a relative
measure of IPO firm valuation without making any judgment about whether the firm is
overvalued or undervalued.

Table 10 IPO valuation based on comparable firm multiples
Panel A
P0/V ratio based on P/S multiple (based on matching firms) P0/V ratio based on P/S multiple (based on matching portfolio)
No. of IPO firms 25% Median P0/V 75% No. of IPO firms 25% Median P0/V 75%
461 0.487 1.615 5.591 461 0.824 2.369 6.848
P0/V ratio based on P/EBITDA multiple P0/V ratio based on P/EBITDA multiple
(based on matching firms) (based on matching portfolio)
218 0.522 1.272 2.754 218 0.550 1.169 2.805
P0/V ratio based on P/B multiple (based on matching firms) P0/V ratio based on P/B multiple (based on matching portfolio)

399 0.370 0.895 2.261 399 0.384 0.891 2.290
Panel B
P1/V ratio based on P/S multiple (based on matching firms) P1/V ratio based on P/S multiple (based on matching portfolio)
461 0.589 1.927 6.692 461 0.919 2.692 8.495
P1/V ratio based on P/EBITDA multiple P1/V ratio based on P/EBITDA multiple
(based on matching firms) (based on matching portfolio)
218 0.674 1.478 3.395 218 0.691 1.465 3.279
P1/V ratio based on P/B multiple (based on matching firms) P1/V ratio based on P/B multiple (based on matching portfolio)
399 0.476 1.066 2.814 399 0.459 1.035 3.249
This table presents the 25th, 50th, and the 75th percentiles of the cross-sectional distributions of P/V ratios based on P/S, P/EBITDA, and P/B
multiples, respectively.
IPO, initial public offering; EBITDA, earnings before interest, tax, depreciation, and amortization.
113
Table 11 Risk factor disclosures and IPO valuation in IPO offer price
114
Matching algorithm P0 ⎞ P0 ⎞ P0 ⎞
Ln ⎛⎜ Ln ⎛⎜ Ln ⎛⎜
⎝ V ⎟⎠ sales as the dependent variable ⎝ V ⎟⎠ EBITDA as the dependent variable ⎝ V ⎟⎠ B as the dependent variable
Matching Median of Matching Median of Matching Median of

firm matching portfolio firm matching portfolio firm matching portfolio
Constant 0.541 0.050 −2.372 −1.393 −3.391 −2.736

(0.37) (0.04) (−1.26) (−0.74) (−2.90)*** (−2.44)**
Ln(Assets) 0.009 −0.174 0.057 0.000 0.048 0.062
(0.14) (−3.48)*** (0.80) (0.01) (0.94) (1.19)
Ln(Age) −0.128 −0.163 0.066 0.102 0.047 −0.023
(−1.26) (−1.81)* (0.83) (1.27) (0.59) (−0.30)
BTM −1.958 −0.271 −1.175 −0.352 −1.529 −2.022
(−3.35)*** (−0.66) (−2.09)** (−0.64) (−3.35)*** (−4.47)***
HighTech −0.222 −0.202 0.018 −0.009 −0.316 −0.224
(−1.05) (−1.00) (0.08) (−0.04) (−2.08)** −1.37
% Retained 0.615 0.726 −0.137 0.085 1.199 1.349
(1.19) (1.57) (−0.29) (0.18) (2.66)** (3.04)***
Underwriter 1.576 4.259 −3.534 −3.772 0.988 0.497
(0.54) (1.86)* (−1.59) (−1.66)* (0.37) (0.19)
VC 0.208 −0.311 −0.124 −0.429 0.193 −0.029
(0.62) (−0.94) (−0.44) (−1.38) (1.01) (−0.10)
Big 0.262 0.072 0.086 0.160 −0.204 0.239
(1.39) (0.40) (0.41) (0.79) (−1.34) (1.52)
Ln(ASXReturn) −3.570 −2.058 3.052 1.078 −0.724 1.220
(−1.4) (−0.90) (1.13) (0.41) (−0.36) (0.64)
%Insider −0.562 0.043 0.217 −0.325 1.045 1.805
(−1.39) (0.13) (0.74) (−1.09) (2.97)** (5.61)***
Ln(Delay) 0.390 0.178 0.334 0.312 0.446 0.145

(1.19) (0.61) (0.75) (0.71) (1.93)* (0.69)
UniqueContent 1.225 0.969 0.813 0.111 1.798 0.987
(2.08)** (1.84)* (1.13) (0.16) (3.82)*** (2.28)**
StandardContent −2.79 −0.705 0.397 0.258 −1.015 0.244
(−3.65)*** (−1.09) (0.47) (0.30) (−1.70)* (0.42)
Number of observations 461 461 218 218 399 399
R2 7.17.% 5.86% 5.68% 4.69% 12.51% 18.19%
P0 ⎞ P0 ⎞ P0 ⎞
The dependent variable is Ln ⎛⎜ , Ln ⎛⎜ , or Ln ⎛⎜ calculated as the natural logarithm of IPO offer price P0 to the benchmark value of the IPO firm.
⎝ V ⎟⎠ sales ⎝ V ⎟⎠ EBITDA ⎝ V ⎟⎠ B
The independent variables are UniqueContent and StandardContent. Control variables consist of Ln(Assets), Ln(Age), BTM, HighTech, % Retained, Underwriter, VC, Big,
Ln(ASXReturn), %Insider, and Ln(Delay). t-statistics are provided in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

IPO, initial public offering; BTM, book-to-market; VC, venture-capital backing; EBITDA, earnings before interest, tax, depreciation, and amortization.
portfolio, the coefficient on the standard content is insignificant. For the regres-
⎛ P0 ⎞
sions with Ln as the dependent variable, the coefficients on both the
⎝ V ⎠ EBITDA
unique and standard content are insignificant. For the regressions with
Ln ⎛
P0 ⎞
as the dependent variable, the coefficient on the unique content is
⎝ V ⎠B
positive and significant. The regression results when the P0/V is estimated based
on sales or book value provides evidence suggesting that IPOs are valued higher
in the IPO market when firms provide more disclosures of the unique content
of investment risk.
Next, I examine the risk disclosures in the prospectus and P1/V ratios. I use
Ln ⎛ ⎞ ⎛ P1⎞ ⎛ P1⎞
P1
, Ln , and Ln to measure the level of IPO valuation at
the first-day closing price in the IPO aftermarket. To exclude any effect from the
relation between risk disclosure and IPO underpricing, I also run regressions
with an additional control for IPO first-day return (Ln(Underpricing)). The results
from the above regressions are presented in Table 12. For the regressions with
Ln ⎛ ⎞
P1
as the dependent variable where the benchmark value V for each
⎝ V ⎠ sales
IPO firm is computed based on its matching firms’ Price/Sales, the coefficient on
the unique content is positive and significant at the 10% level and the coeffi-
cient on the standard content is negative and significant at the 1% level. When
Ln ⎛ ⎞
P1
is computed based on the median firm of the matching portfolio,
⎝ V ⎠ sales
the coefficient on the standard content is insignificant. For the regressions with
Ln ⎛
P0 ⎞
as the dependent variable, the coefficients on both the unique and
⎝ V ⎠ EBITDA
⎛ P0 ⎞
standard content are insignificant. For the regressions with Ln as the
⎝ V ⎠B
dependent variable, the coefficient on the unique content is positive and sig-
nificant. The results suggest that IPO firms that provide more disclosures about
the unique content of investment risk are valued higher in the immediate
aftermarket.
Finally, I study whether risk factor disclosures have a larger impact on the IPO
offer price or on the IPO aftermarket price. I run a regression with the depen-
⎛ P1⎞ − Ln ⎛
P0 ⎞
dent variable Ln , the difference between the natural loga-
⎝ V ⎠ sales ⎝ V ⎠ sales
rithm of IPO aftermarket price ratio and the natural logarithm IPO offer price
⎛ P0 ⎞
ratio. I include Ln as an additional control variable. I report the results
⎝ V ⎠ sales
from this regression in columns 1 and 2 of Table 13, grouped by whether I
⎛ P0 ⎞ ⎛ P0 ⎞
calculate Ln and Ln based on the matching firm or the median
of the matching portfolio. The coefficients on both the standard and unique
content have the expected signs but are insignificant. The results suggest that

Table 12 Risk factor disclosures and IPO valuation in IPO first-day closing price
116
Matching algorithm P1 P1 P1
Ln ⎛⎜ ⎞⎟ as the dependent variable Ln ⎛⎜ ⎞⎟ as the dependent variable Ln ⎛⎜ ⎞⎟ as the dependent variable
Matching Median of Matching Median of Matching Median of

firm matching portfolio firm matching portfolio firm matching portfolio
(1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (1) (2)
Constant 2.361 2.411 1.522 0.810 −1.720 −2.598 −0.555 −1.258 −1.795 −2.486 −1.111 −2.456
(1.67)* (1.64) (1.19) (0.61) (−0.91) (−1.41) (−0.29) (−0.68) (−1.61) (−2.13)** (−1.02) (−2.28)**
Ln(Assets) −0.004 −0.005 −0.0178 −0.173 0.047 0.061 −0.009 0.003 0.046 0.054 0.058 0.072
(−0.07) (−0.08) (−3.53)*** (−3.44)*** (0.67) (0.87) (−0.12) (0.04) (0.91) (1.06) (1.08) (1.39)
Ln(Age) −0.106 −0.107 −0.157 −0.145 0.052 0.072 0.081 0.098 0.063 0.076 0.001 0.026
(−1.09) (−1.10) (−1.77)* (−1.64) (0.65) (0.92) (1.01) (1.21) (0.81) (0.98) (0.02) (0.36)
BTM −1.922 −1.922 −0.285 −0.291 −1.323 −1.274 −0.510 −0.471 −1.607 −1.560 −2.106 −2.014
(−3.42)*** (−3.42)*** (−0.69) (−0.70) (−2.14)** (−2.2)** (−0.86) (−0.80) (−3.31)*** (−3.26)*** (−4.34)*** (4.17)***
HighTech −0.133 −0.128 −0.098 −0.174 0.094 0.029 0.055 0.003 −0.224 −0.301 −0.140 −0.289
(−0.64) (−0.60) (−0.49) (−0.85) (0.41) (0.13) (0.23) (0.01) (−1.46) (−1.94)* (−0.82) (−1.74)*
% Retained 0.722 0.730 0.812 0.685 0.054 −0.124 0.291 0.149 1.296 1.120 1.414 1.071
(1.50) (1.52) (1.79)* (1.50) (0.11) (−0.26) (0.61) (0.31) (2.94)*** (2.55)** (3.36)*** (2.66)***
Underwriter 1.260 1.255 3.565 3.627 −3.861 −3.728 −3.828 −3.721 −0.066 −0.117 −0.472 −0.570
(0.47) (0.47) (1.54) (1.60) (−1.74)* (−1.71)* (−1.66)* (−1.63) (−0.02) (−0.04) (−0.18) (−0.23)
VC 0.072 0.067 −0.416 −0.354 −0.236 −0.078 −0.550 −0.424 0.065 0.137 −0.166 −0.027
(0.22) (0.21) (−1.29) (−1.08) (−0.91) (−0.27) (−1.93)* (−1.34) (0.35) (0.72) (−0.56) (−0.09)
Big 0.261 0.261 0.056 0.061 0.041 0.077 0.113 0.142 −0.193 −0.200 0.256 0.242
(1.41) (1.41) (0.32) (0.35) (0.20) (0.37) (0.55) (0.70) (−1.26) (−1.32) (1.58) (1.54)
Ln(ASXReturn) −4.054 −4.058 −2.24 −2.172 2.988 2.977 1.220 1.211 −1.168 −1.044 0.801 1.043
(−1.61) (−1.61) (−0.97) (−0.96) (1.06) (1.10) (0.44) (0.45) (−0.55) (−0.51) (0.40) (0.55)
%Insider −0.421 −0.420 0.168 0.158 0.221 0.238 −0.348 −0.334 1.258 1.252 1.963 1.951
(−1.10) (−1.10) (0.50) (0.48) (0.74) (0.82) (−1.13) (−1.10) (3.51)*** (3.53)*** (6.28)*** (6.45)***
Ln(Delay) 0.017 0.003 −0.151 0.039 0.142 0.398 0.060 0.264 0.085 0.282 −0.221 0.162
(0.05) (0.01) (−0.51) (0.13) (0.32) (0.92) (0.13) (0.61) (0.39) (1.17) (−1.09) (0.75)
Ln(Underpricing) −0.047 0.675 0.943 0.755 0.690 1.344
(−0.16) (2.49)** (2.66)** (2.11)** (2.67)*** (6.39)***
UniqueContent 1.092 1.089 0.822 0.890 0.906 0.790 0.274 0.181 1.623 1.696 0.833 0.976
(1.90)* (1.89)* (1.55) (1.69)* (1.23) (1.12) (0.37) (0.25) (3.28)*** (3.52)*** (1.81)* (2.30)**
StandardContent −2.921 −2.912 −0.615 −0.768 0.553 0.411 0.370 0.256 −0.894 −1.092 0.351 −0.035
(−4.04)*** (−4.00)*** (−0.95) (−1.20) (0.65) (0.50) (0.42) (0.30) (−1.51) (−1.83)* (0.59) (−0.06)
Number of observations 461 461 461 461 218 218 218 218 399 399 399 399
R2 7.39% 7.40% 5.52% 6.81% 6.61% 9.99% 5.41% 7.64% 12.74% 14.49% 19.52% 26.35%
P1 P1 P1
The dependent variable is Ln ⎛⎜ ⎞⎟ , Ln ⎛⎜ ⎞⎟ , or Ln ⎛⎜ ⎞⎟ calculated as the natural logarithm of IPO first-day closing price P1 to the benchmark value of the IPO firm.
The independent variables are UniqueContent and StandardContent. Control variables consist of Ln(Assets), Ln(Age), BTM, HighTech, % Retained, Underwriter, VC, Big, Ln(ASXReturn),

%Insider, Ln(Delay), and Ln(Underpricing). t-statistics are provided in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
IPO, initial public offering; BTM, book-to-market; VC, venture-capital backing; EBITDA, earnings before interest, tax, depreciation, and amortization.
Table 13 Risk factor disclosures on the offer price versus the aftermarket price
Ln ⎛⎜ ⎞⎟ − Ln ⎛⎜
P0 ⎞
Ln ⎛⎜ ⎞⎟ − Ln ⎛⎜
Dependent P1 P1 P0 ⎞
variable ⎝ V ⎠ sales ⎝ V ⎟⎠ sales ⎝ V ⎠ sales ⎝ V ⎟⎠ sales Ln(Underpricing)
(Matching firm) (Matching portfolio)
Constant 1.386 1.322 1.067
(5.51)*** (5.30)** (5.08)***
Ln(Assets) −0.008 −0.009 −0.009
(−0.89) (−1.01) (−1.08)
Ln(Age) −0.008 −0.007 −0.021
(−0.50) (−0.43) (−1.53)
BTM −0.106 −0.051 −0.039
(−1.09) (−0.52) (−0.37)
HighTech 0.090 0.092 0.107
(2.45)** (2.48)*** (2.99)***
% Retained 0.210 0.188 0.202
(2.25)** (2.04)** (2.65)***
Underwriter −0.193 −0.304 −0.056
(−0.34) (−0.56) (−0.12)
VC −0.090 −0.097 −0.087
(−1.82)* (−1.85)* (−1.74)*
Big −0.002 −0.014 −0.001
(−0.05) (−0.43) (−0.04)
Ln(ASXReturn) −0.276 −0.170 −0.189
(−0.68) (−0.43) (−0.49)
%Insider 0.044 0.066 0.001
(0.62) (0.90) (0.02)
Ln(Delay) −0.304 −0.313 −0.271
(−6.32)*** (−6.65)*** (−6.23)***
Ln ⎛⎜
P0 ⎞
⎝ V ⎟⎠ sales −0.030 −0.014 −0.024
(−3.86)*** (−1.61) (−3.43)***
UniqueContent −0.113 −0.131 −0.068
(−1.16) (−1.34) (−0.74)
StandardContent 0.016 0.142 0.159
(0.12) (1.09) (1.34)
Number of 461 461 461
observations
R2 15.8% 13.59% 15.05%
The dependent variable in columns 1 and 2 is Ln ⎛ ⎞ − Ln ⎛

P1 P0 ⎞
, where the benchmark
value of the IPO firm is calculated based on P/S of either the IPO firm’s matching firm or the
median of its matching portfolio. The dependent variable in column 3 is Ln(Underpricing). The
independent variables are UniqueContent and StandardContent. Control variables consist of
Ln(Assets), Ln(Age), BTM, HighTech, % Retained, Underwriter, VC, Big, Ln(ASXReturn), %Insider,
Ln(Delay), and Ln ⎛
P0 ⎞
. t-statistics are provided in parentheses. *, **, and *** indicate signifi-
⎝ V ⎠ sales
cance at the 10%, 5%, and 1% levels, respectively.
IPO, initial public offering; BTM, book-to-market; VC, venture-capital backing.

for the sample of 461 IPO firms that have positive sales prior to their IPOs, the
risk factor disclosures do not seem to affect the offer price and the first-day price
⎛ P0 ⎞
significantly differently. The coefficient on Ln is significantly negative,
⎝ V ⎠ sales
suggesting that IPOs that are most undervalued have the largest first-day return,
consistent with the asymmetric information theory. I further run the regression
of Ln(Underpricing) on the measures of risk factor disclosures for the 461 IPO
firms with positive sales in the year prior to listing. Column 3 of Table 13
presents the results and the coefficients on both the unique and standard
content are insignificant. Similar results are obtained when I run the regressions
for the subsample of IPO firms with positive sales and EBITDA and the
subsample of positive sales and pre-issue book value.
The insignificant results suggest that firms with negative EBITDA may drive
the inverse (positive) relation between unique (standard) risk disclosures and
underpricing previously found for the whole sample. As my original sample is
significantly reduced due to imposing the positive EBITDA criterion, I compare
the characteristics between the remaining 218 IPO firms and the 524 firms
which are eliminated due to negative EBITDA. Table 14 shows that IPO firms
with positive EBITDA in the year prior to going public are significantly larger
and older than IPO firms with negative EBITDA. The IPOs issued by firms with
positive EBITDA are underwritten by more prestigious underwriters and expe-
rience significantly less delay from initial filing to listing on the exchange. IPO
firms with positive EBITDA also disclose more unique content in the risk factor
section.
I run the regression of Ln(Underpricing) on the measures of risk disclosures
separately for the IPO firms with negative EBITDA and IPO firms with positive
EBITDA. Table 15 shows that, for firms with negative EBITDA or book value,
unique content of risk factor disclosures is significantly inversely related to
underpricing and standard content is significantly positively related to under-
pricing. In light of the differences in the characteristics between the IPO firms
with negative EBITDA and the IPO firms with positive EBITDA, the results
suggest that risk disclosure informativeness plays a more significant role in
helping investors assess the risks of smaller firms, younger firms, and firms with
poorer financial performance prior to going public.33
33 The finding of a more pronounced inverse relation between informative risk factor disclo-
sures and underpricing for small firm, young firms, firms with poorer performance prior to
listing, and firms with less prestigious underwriters may also be consistent with the view that
these firms suffer more from lack of investor participation. If the issuer or underwriter
anticipates that more informative risk factor disclosures help attract retail investor partici-
pation and incorporate it into the offer price, the issuer or underwriter will set a higher offer
price, resulting in lower underpricing. This investor participation explanation is consistent
with the information asymmetry explanation because more informative risk factor disclo-
sures may attract retail investors’ participation by helping these investors gain a better
understanding of the IPO firm.

Table 14 Issue and firm characteristics comparison

IPO firms with IPO firms with Test of difference
positive EBITDA in negative EBITDA in
the year prior to IPO the year prior to IPO
N = 218: A N = 524: B A–B
Mean Median Mean Median t-test Wilcoxon

z-test
Panel A: Initial return, IPO characteristics, and other variables
Ln(Underpricing) 0.218 0.138 0.1695 0.0953 1.7745 2.424

(0.076)* (0.015)**
Ln(Assets) 2.798 2.908 −0.911 −0.438 16.175 17.308
(0.000)*** (0.000)***
Ln(Age) 1.838 2.027 0.950 0.707 11.003 9.945
(0.000)*** (0.000)***
BTM 0.959 0.114 0.065 0.032 1.121 8.259
(0.264) (0.000)***
HighTech 0.358 0.000 0.172 0.000 5.010 5.512
(0.000)*** (0.000)***
% Retained 0.601 0.642 0.499 0.512 6.144 6.726
(0.000)*** (0.000)***
Underwriter 0.034 0.014 0.008 0.000 9.211 11.565
(0.000)*** (0.000)***
VC 0.046 0.000 0.027 0.000 1.208 1.342
(0.228) (0.180)
Big 0.541 1.000 0.286 0.000 6.510 6.583
(0.000)*** (0.000)***
%Insider 0.180 0.000 0.001 0.000 8.739 13.357
(0.000)*** (0.000)***
Leverage 0.188 0.093 3.100 0.000 −1.0405 16.206
(0.299) (0.000)***
Ln(ASXReturn) 0.014 0.016 0.021 0.022 −2.219 −2.257
(0.027)** (0.024)**
Ln(Delay) 3.765 3.761 3.989 3.932 −8.913 −7.406
(0.000)*** (0.000)***
Ln(Pre−IPO News) 2.253 2.303 2.260 2.674 −4.243 −4.476
(0.000)*** (0.000)***
Panel B: Prospectus text variables
Total words in 32260.790 28878.500 36046.900 34419.000 −3.850 −5.419

document (0.000)*** (0.000)***
Total words in Risk 1559.858 1412.500 1673.176 1582.500 −1.724 −2.725
Factor section (0.086)* (0.006)***
RiskFactorsWords 0.049 0.046 0.049 0.047 −0.4519 −0.079
(0.652) (0.937)
StandardContent 1.013 1.013 1.014 1.024 −0.1506 −0.588
(0.880) (0.556)
UniqueContent 1.149 1.14239 1.025 1.011379 9.194 9.870
(0.000)*** (0.000)***
Recent 0.811 0.792 0.553 0.577 7.853 7.283
(0.000)*** (0.000)***
Past industry 0.202 0.185 0.462 0.444 −8.220 −7.727
(0.000)*** (0.000)***
This table compares the issue and firm characteristics of IPO firms with positive EBITDA prior to listing with IPO
firms with negative EBITDA. p-values are provided in parentheses. *, **, and *** indicate significance at the 10%, 5%,
and 1% levels, respectively.
IPO, initial public offering; BTM, book-to-market; VC, venture-capital backing; EBITDA, earnings before interest,
tax, depreciation, and amortization.

Table 15 Risk factor disclosures and IPO underpricing – positive EBITDA firms
versus negative EBITDA firms
Panel A: This table presents results on the regressions of Ln(Underpricing) on
UniqueContent and StandardContent, controlling for Ln(Assets), Ln(Age), BTM, HighTech,
%Retained, Underwriter, VC, Big, Ln(ASXReturn), %Insider, and Ln(Delay), separately for
IPO firms with negative EBITDA or book value 1 year prior to the IPO year and IPO
firms with positive EBITDA and book value. All the variables are as previously defined.
t-statistics are provided in parentheses. *, **, and *** indicate significance at the 10%, 5%, and
1% levels, respectively.
Whole Negative Positive

sample EBITDA firms EBITDA firms
Constant 0.910 0.946 0.932
(5.97)*** (5.15)*** (2.81)***
Ln(Assets) −0.009 −0.011 −0.015
(−2.08)** (−2.2)** (−1.42)
Ln(Age) −0.007 −0.006 −0.022
(−0.63) (−0.33) (−1.34)
BTM −0.001 0.068 −0.052
(−0.01) (0.56) (−0.39)
HighTech 0.094 0.100 0.069
(2.98)*** (2.25)** (1.48)
% Retained 0.101 0.028 0.188
(1.68)* (0.37) (1.77)*
Underwriter −0.143 −0.192 −0.141
(−0.33) (−0.21) (−0.28)
VC −0.096 −0.041 −0.167
(−1.99)** (−0.59) (−2.31)**
Big −0.043 −0.045 −0.039
(−1.77)* (−1.41) (−0.98)
Ln(ASXReturn) 0.053 0.084 0.012
(0.17) (0.21) (0.02)
%Insider 0.026 2.834 −0.018
(0.44) (3.29)*** (−0.28)
Ln(Delay) −0.229 −0.202 −0.271
(−7.94)*** (−6.11)*** (−3.5)***
UniqueContent −0.147 −0.293 0.123
(−1.92)* (−3.27)*** (0.82)
StandardContent 0.282 0.300 0.151
(2.91)*** (2.52)** (1.04)
R2 10.37% 12.66% 13.01%
My previous analysis of the determinants of risk factor disclosures finds that

younger firms and firms with poorer performance prior to listing disclose less
about investment risk. To address the concern that the relation between risk
factor disclosures and IPO underpricing may be biased due to the endogeneity

Panel B: The first two columns of this table present the results on the first-stage
regression of risk disclosure variables on all the exogenous variables in equations (8)
and (10). The third column presents the second stage regression of Ln(Underpricing) on
UniqueContentIV and StandardContentIV, where UniqueContentIV and StandardContentIV
are the unique content and standard content estimated from the first-stage
regressions. All the other variables are as previously defined. t-statistics are provided in
parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
First stage: First stage: Second stage:
UniqueContent StandardContent Ln(Underpricing)
Constant 0.909 1.069 1.398
(12.46)*** (18.35)*** (2.00)**
Ln(Assets) −0.000 0.001 −0.012
(−0.18) (0.26) (−2.23)
Ln(Age) 0.022 −0.001 0.0138
(2.29)** (−0.11) (0.65)
BTM 0.041 0.063 0.086
(0.71) (1.22) (0.73)
HighTech 0.046 −0.059 0.138
(2.77)** (−3.40)*** (1.92)*
% Retained 0.036 −0.100 0.046
(0.90) (−3.39)*** (0.45)
Underwriter 0.293 −0.530 0.052
(0.97) (−1.75)* (0.05)
VC −0.117 −0.062 −0.102
(−4.54)*** (−1.57) (−1.42)
Big 0.004 −0.017 −0.040
(0.30) (−1.38) (−1.03)
Ln(ASXReturn) −0.432 −0.054 −0.130
(−2.07)** (−0.34) (−0.31)
%Insider 0.373 0.315 −0.179
(1.21) (0.41) (4.68)***
Ln(Delay) 0.047 −0.001 −0.179
(2.63)*** (−0.10) (4.68)***
Leverage 0.060 0.000
(2.28)** (0.03)
Lawyer −0.039 0.016
(−0.17) (0.08)
MSD(Returns) −8.429 −1.36
(−3.62)*** (−0.66)
AccumLossD −0.012 0.058
(−0.63) (3.02)***
Ln(PreIPONews) −0.021 −0.010
(−3.41)*** (−2.15)**
UniqueContentIV −0.879
(−2.29)**
StandardContentIV 0.325
(0.45)
R2 16.36% 13.98% 11.5%

of the risk disclosure variables, I revisit the association between risk factor
disclosures and underpricing for the sample of IPOs with negative EBITDA prior
to listing using a two-stage regression model. In the first-stage regression, I
endogenize the risk disclosure variables by regressing the unique content
and the standard content of risk factor disclosures on all the exogenous
variables in equations (8) and (10), respectively. The fitted values of the unique
content (UniqueContentIV) and the standard content (StandardContentIV) are
then used to explain underpricing in the second stage. I find that the
inverse relation between the unique content of risk factors and underpricing is
robust to considering the endogeneity of disclosure. However, the positive
association between the standard content and underpricing disappears (Panel B
of Table 15).
V. CONCLUSION
I use the context of a company’s IPO of equity securities as a capital-market

setting to examine the economic consequences of the disclosures of downside
risks of future performance. I examine the relation between measures of the
information content in the risk factor section of the IPO prospectus and under-
pricing. I find a significant inverse association between the unique content of
the risk disclosures and underpricing, which is particularly strong for IPOs with
less prestigious lead underwriters. Further analysis shows that the results are
mainly driven by younger firms, smaller firms, and firms with poorer perfor-
mance prior to listing.
The findings are consistent with the information asymmetry theory, which
suggests that informative risk factor disclosures reduce ex ante uncertainty and
that IPOs disclosing more informative content about investment risk have less
initial underpricing (Ritter 1984; Beatty and Ritter 1986). Additional tests do not
support the heterogeneous beliefs argument for the inverse relation between
risk factor disclosures and underpricing.
I also document that the informativeness of risk disclosures in the IPO
prospectus relates to proxies for agency costs and managerial uncertainty. In
addition, IPOs that turn out to be relatively risky tend to disclose more standard
content in their risk factor section, but not more unique content. This supports
the regulators’ view that issuers can improve their risk disclosures by providing
more specific information regarding firm-level risk.
Rui Ding
The University of New South Wales
The UNSW Business School
Level 3, Quadrangle Building, UNSW, Kensington
Sydney, New South Wales 2052
Australia
[email protected]

REFERENCES
Alavi, A., P. K. Pham, and T. M. Pham (2008), ‘Pre-IPO Ownership Structure and Its
Impact on the IPO Process’, Journal of Banking and Finance, 32, 2361–75.
Arnold, T., R. P. Fishe, and D. North (2010), ‘The Effects of Ambiguous Information on
Initial and Subsequent IPO Returns’, Financial Management, 39 (4), 1497–519.
ASIC (2011a), ‘Consultation Paper 155 Prospectus Disclosure: Improving Disclosure for
Retail Investors’, Available at http://download.asic.gov.au/media/1332926/CP155
-Published-12-April-2011.pdf.
ASIC (2011b), ‘Regulatory Guide 228 Prospectuses: Effective Disclosure for Retail
Investors’, Available at http://download.asic.gov.au/media/1241450/rg228
-published-10-november-2011-1.pdf.
Baker, M., and P. Gompers (2003), ‘The Determinants of Board Structure at the Initial
Public Offering’, Journal of Law and Economics, 46, 569–98.
Balakrishnan, K., and E. Bartov (2012), ‘Cognitive Bias, Conflict of Interests, and
Analysts’ Use of Qualitative Earnings Information: Evidence from IPO Prospectus’s
Risk Factors Section’, Working Paper, New York University.
Barry, C., and S. Brown (1984), “Differential Information and the Small Firm Effect”,
Journal of Financial Economics, 13, 283–94.
Barry, C., C. Muscarella, J. Peavy III, and M. Vetsuypens (1990), ‘The Role of Venture
Capital in the Creation of Public Companies’, Journal of Financial Economics, 27,
447–71.
Beatty, R., and I. Welch (1996), ‘Issuer Expenses and Legal Liability in Initial Public
Offerings’, Journal of Law and Economics, 39, 545–602.
Beatty, R. P. (1989), ‘Auditor Reputation and the Pricing of Initial Public Offering’, The
Accounting Review, 64 (5), 693–709.
Beatty, R. P., and J. R. Ritter (1986), ‘Investment Banking, Reputation, and the
Underpricing of Initial Public Offerings’, Journal of Financial Economics, 15, 213–
32.
Botosan, C., and M. Plumlee (2002), ‘A Re-Examination of Disclosure Level and the
Expected Cost of Equity Capital’, Journal of Accounting Research, 40, 21–40.
Campbell, J. Y., and R. J. Shiller (1987), “Cointegration and Tests of Present Value
Models”, Journal of Political Economy, 95 (5), 1062–88.
Cater, R., F. Dark, and A. Singh (1998), ‘Underwriter Reputation, Initial Returns, and
the Long-Run Performance of IPO Stocks’, Journal of Finance, 53, 285–311.
Carter, R., and S. Manaster (1990), “Initial Public Offerings and Underwriter Reputa-
tion”, Journal of Finance, 45 (4), 1045–67.
Chemmanur, T. J., and E. Loutskina (2006), ‘The Role of Venture Capital Backing in
Initial Public Offerings: Certification, Screening, or Market Power’, Boston College,
Working Paper.
Chemmanur, T. J., and A. Yan (2009) ‘Product Market Advertising, IPO Valuation, and
Long-run Stock Returns’, Working Paper, Boston College.
Corporations Act 2001 (Commonwealth of Australia) Available at https://
www.comlaw.gov.au/Details/C2015C00336.
Diamond, D., and R. Verrecchia (1991), ‘Disclosure, Liquidity, and the Cost of Capital’,
Journal of Finance, 46 (4), 1325–59.
Dimovski, W., S. Philavanh, and R. Brooks (2011), ‘Underwriter Reputation and Under-
pricing: Evidence from the Australian IPO Market’, Review of Quantitative Finance and
Accounting, 37 (4), 409–26.
Easley, D., and M. O’Hara (2004), ‘Information and the Cost of Capital’, Journal of
Finance, 59 (4), 1553–83.
French, K., G. Schwert, and R. Stambaugh (1987), ‘Expected Stock Returns and Vola-
tility’, Journal of Financial Economics, 19, 3–29.

Hanley, K. W. (1993), ‘The Underpricing of Initial Public Offerings and the Partial
Adjustment Phenomenon’, Journal of Financial Economics, 34, 231–50.
Hanley, K. W., and G. Hoberg (2010), ‘The Information Content of IPO Prospectuses’,
The Review of Financial Studies, 23, 2821–64.
Hanley, K. W., and G. Hoberg (2012), ‘Litigation Risk, Strategic Disclosure and the
Underpricing of Initial Public Offerings’, Journal of Financial Economics, 103 (2),
235–54.
Harrison, M., and D. Kreps (1978), ‘Speculative Investor Behavior in a Stock
Market with Heterogeneous Expectations’, Quarterly Journal of Economics, 92, 323–
36.
Heckman, J. (1979), ‘Sample Selection Bias as a Specification Error’, Econometrica:
Journal of the Econometric Society, 47, 153–62.
Hermalin, B. E., and M. S. Weisbach (2007), ‘Transparency and Corporate Governance’,
Working Paper, National Bureau of Economic Research.
Hoberg, G., and G. Phillips (2010), ‘Product Market Synergies and Competition in
Mergers and Acquisitions: A Text-Based Analysis’, Review of Financial Studies, 23 (10),
3773–811.
Hochberg, Y. (2012), ‘Venture Capital and Corporate Governance in the Newly-Public
Firm’, Review of Finance, 16 (2), 429–80.
Hogan, W. W., and J. M. Warren (1974), ‘Toward the Development of an Equilibrium
Capital-Market Model Based on Semivariance’, Journal of Financial and Quantitative
Analysis, 9, 1–12.
How, J., and J. Yeo (2000), ‘The Pricing of Underwriting Services in the Australian
Capital Market’, Pacific-Basin Finance Journal, 8, 347–73.
Jahankhani, A. (1976), ‘E-V and E-S Capital Asset Pricing Models: Some Empirical
Tests’, Journal of Financial and Quantitative Analysis, 11, 513–28.
Johnson, W., J. Kang, and S. Yi (2010), ‘The Certification Role of Large Customers in
the New Issues Market’, Financial Management, 39 (4), 1425–74.
Kahneman, D., and A. Tversky (1979), ‘Prospect Theory: An Analysis of Decision
Under Risk’, Econometrica: Journal of the Econometric Society, 47, 263–91.
Kim, M., and J. Ritter (1999), “Valuing IPOs”, Journal of Financial Economics, 53 (3),
409–37.
Kothari, S., X. Li, and J. Short (2009a), ‘The Effect of Disclosure by Management,
Analysts, and Financial Press on the Equity Cost of Capital: A Study Using Content
Analysis’, The Accounting Review, 84 (5), 1639–70.
Kothari, S., S. Shu, and P. Wysocki (2009b), ‘Do Managers Withhold Bad News?’,
Journal of Accounting Research, 47, 241–76.
Kraus, A., and R. Litzenberger (1976), ‘Skewness Preference and the Valuation of Risk
Assets’, Journal of Finance, 31, 1085–100.
Lambert, R., C. Leuz, and R. Verrecchia (2007), ‘Accounting Information, Disclosure,
and the Cost of Capital’, Journal of Accounting Research, 45 (2), 385–420.
Lee, L., S. Taylor, and T. Walter (1996), ‘Australian IPO Pricing in the Short and Long
Run’, Journal of Banking and Finance, 20, 1189–210.
Lennox, C., J. Francis, and Z. Wang (2012), ‘Selection Models in Accounting Research’,
The Accounting Review, 87 (2), 589–616.
Lerner, J. (1995), ‘Venture Capitalists and the Oversight of Private Firms’, Journal of
Finance, 50, 301–18.
Libby, R., and P. C. Fishburn (1977), ‘Behavioural Models of Risk Taking in
Business Decisions: A Survey and Evaluation’, Journal of Accounting Research, 15,
272–92.
Ljungqvist, A. (2004), ‘IPO Underpricing: A Survey’, in B. E. Eckbo (ed.), Handbook in
Corporate Finance: Empirical Corporate Finance, North Holland Publishing Co,
Amsterdam, pp. 375–422.

Loughran, T., and B. McDonald (2011), ‘When Is a Liability Not a Liability? Textual
Analysis, Dictionaries, and 10-Ks’, Journal of Finance, 66, 35–65.
Loughran, T., and J. Ritter (1995), ‘The New Issues Puzzle’, Journal of Finance, 50 (1),
23–51.
Loughran, T., and J. Ritter (2004), ‘Why Has IPO underpricing Changed Over Time?’,
Financial Management, 33 (3), 5–37.
Lowry, M., and G. Schwert (2004), ‘Is the IPO Pricing Process Efficient?’, Journal of
Financial Economics, 71, 3–26.
Lowry, M., and S. Shu (2002), ‘Litigation Risk and IPO Underpricing’, Journal of
Financial Economics, 65, 305–65.
Malmendier, U., and G. Tate (2005a), ‘Does Overconfidence Affect Corporate Invest-
ment? CEO Overconfidence Measures Revisited’, European Financial Management, 11,
649–59.
Malmendier, U., and G. Tate (2005b), ‘CEO Overconfidence and Corporate Invest-
ment’, Journal of Finance, 60 (6), 2661–700.
Meggionson, W., and K. Weiss (1991), ‘Venture Capitalist Certification in Initial Public
Offerings’, Journal of Finance, 46, 879–903.
Michaely, R., and W. Shaw (1995), ‘Does the Choice of Auditor Convey Quality in an
Initial Public Offering?’, Financial Management, 24 (4), 15–30.
Miller, E. (1977), ‘Risk, Uncertainty and Divergence of Opinion’, Journal of Finance, 32,
1151–68.
PricewaterhouseCoopers (2011), ‘Listing a Company on the Australian Securities
Exchange’, Available at http://www.pwc.com.au/legal/assets/Listing-Company
-ASX-jul10.pdf.
Purnanandam, A., and B. Swaminathan (2004), ‘Are IPO Underpriced?’, The Review of
Financial Studies, 17 (3), 811–48.
Rajan, R., and H. Servaes (1997), ‘Analyst Following of Initial Public Offerings’, Journal
of Finance, 52 (2), 507–29.
Ritter, J. (1984), ‘The ‘Hot Issue’ Market of 1980s’, Journal of Business, 57, 215–
40.
Ritter, J. (1991), ‘The Long-run Performance of Initial Public Offerings’, Journal of
Finance, 46 (1), 3–27.
Ritter, J., and I. Welch (2002), ‘A Review of IPO Activity, Pricing and Allocations’,
Journal of Finance, 57, 1795–828.
Rock, K. (1986), ‘Why New Issues are Underpriced’, Journal of Financial Economics, 19,
269–82.
Schrand, C., and R. Verrecchia (2005), ‘Information Disclosure and Adverse Selection
Explanations of IPO Underpricing’, Working Paper, University of Pennsylvania.
Schultz, P. (1993), ‘Unit Initial Public Offerings’, Journal of Financial Economics, 34,
199–229.
Skinner, D. (1994), ‘Why Firms Voluntarily Disclose Bad News?’, Journal of Accounting
Research, 32 (1), 38–60.
Skinner, D. (1997), ‘Earnings Disclosures and Stockholder Lawsuits’, Journal of Account-
ing and Economics, 23 (3), 249–82.
Tetlock, P. (2007), ‘Giving Content to Investor Sentiment: The Role of Media in the
Stock Market’, Journal of Finance, 62, 1139–68.
Tetlock, P., M. Saar-Tsechanksy, and S. Macaskassy (2008), ‘More Than Words:
Quantifying Language to Measure Firms’ fundamentals’, Journal of Finance, 63,
1437–67.
Tucker, J. (2010), ‘Selection Bias and Econometric Remedies in Accounting and
Finance Research’, Journal of Accounting Literature, 29, 31–57.
Veld, C., and Y. Veld-Merkoulova (2008), ‘The Risk Perceptions of Individual Investors’,
Journal of Economic Psychology, 29 (2), 226–52.

Verrecchia, R. (1983), ‘Discretionary Disclosure’, Journal of Accounting and Economics, 5,

365–80.
Wooldrige, J. (2009), Introductory Econometrics, Fourth Edn. South-Western Cengage
Learning, Mason, Ohio.
Zheng, S. C. (2007), ‘Are IPOs Really Overpriced?’, Journal of Empirical Finance, 14,
287–309.

Disclosure of Downside Risk and Investors' Use of Qualitative Information Evidence From The IPO Prospectus's Risk Factor Section (2016)

Uploaded by

Copyright:

Available Formats

Disclosure of Downside Risk and Investors' Use of Qualitative Information Evidence From The IPO Prospectus's Risk Factor Section (2016)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Disclosure of Downside Risk and Investors' Use of Qualitative Information Evidence From The IPO Prospectus's Risk Factor Section (2016)

Uploaded by

Copyright:

Available Formats

bs_bs_banner

International Review of Finance, 16:1, 2016: pp. 73–126

Disclosure of Downside Risk and

I use the context of a company’s initial public offering (IPO) of equity

© 2015 International Review of Finance Ltd. 2015

74 © 2015 International Review of Finance Ltd. 2015

© 2015 International Review of Finance Ltd. 2015 75

I perform additional tests to explore factors inﬂuencing an issuer’s disclosure

76 © 2015 International Review of Finance Ltd. 2015

© 2015 International Review of Finance Ltd. 2015 77

information) and provides evidence on how such disclosures affect investor ex

II. LITERATURE AND HYPOTHESES

A. Australian IPO market and disclosure requirements

78 © 2015 International Review of Finance Ltd. 2015

expected to increase heterogeneity in information availability between

B. Asymmetric information theory and underpricing

© 2015 International Review of Finance Ltd. 2015 79

explanations for IPO underpricing.9 Asymmetric information theory is relevant

80 © 2015 International Review of Finance Ltd. 2015

risk disclosures on IPO underpricing is not obvious and thus is an empirical

© 2015 International Review of Finance Ltd. 2015 81

III. DATA AND METHOD

A. IPO sample construction

82 © 2015 International Review of Finance Ltd. 2015

B. Textual risk disclosure data

C. Measures to capture the content of risk factor disclosures

i. Measure 1: Standard versus unique content

© 2015 International Review of Finance Ltd. 2015 83

84 © 2015 International Review of Finance Ltd. 2015

where αstandard,i measures the relative loading of standard content, interpreted as

ii. Measure 2: Conservative tone

© 2015 International Review of Finance Ltd. 2015 85

iii. Measure 3: Future cash flow risk

86 © 2015 International Review of Finance Ltd. 2015

previous year’s standard error to obtain the standardized future cash-ﬂow-risk

i. Risk factor disclosures and IPO underpricing

Ln(Underpricing) = ln (1 + ﬁrst-day initial return)

© 2015 International Review of Finance Ltd. 2015 87

Underwriter = a continuous measure of underwriter prestige using their percentage

ii. Risk factor disclosures and future stock return volatility

88 © 2015 International Review of Finance Ltd. 2015

All the other variables are the same as previously deﬁned.

E. Sample descriptive statistics

© 2015 International Review of Finance Ltd. 2015 89

Table 1 Sample composition

Year of issue Whole Subsample Subsample

No. of IPOs Percent No. of IPOs Percent No. of IPOs Percent

Panel B: IPO frequencies by GICS 4-digit industry group

GICS4 industry group Whole Subsample Subsample

No. of IPOs Percent No. of IPOs Percent No. of IPOs Percent

1010 Energy 106 12.35 101 13.61 104 12.52

90 © 2015 International Review of Finance Ltd. 2015

Table 2 Summary statistics

© 2015 International Review of Finance Ltd. 2015 91

Underpricing 0.269 0.524 −0.890 0.121 5.04

92 © 2015 International Review of Finance Ltd. 2015

A. Risk factor disclosures and IPO underpricing

© 2015 International Review of Finance Ltd. 2015 93

(−4.70)* (−4.03)* (1.74)* (0.97) (−0.44)