Governance Role of Media Information in IPO Market- [629320]

Governance Role of Media Information in IPO Market-
Oriented Pricing
Journal: IEEE Access
Manuscript ID Access-2018-25388
Manuscript Type: Original Manuscript
Date Submitted by the
Author:19-Dec-2018
Complete List of Authors: Zhang, Fei ; Chongqing University, Management
Zhou, Xiao-Hua ; Chongqing University, Management
Su, Jiafu ; Chongqing Technology and Business University
Tsai, Sang-Bing; Univ Elect Sci ,
Zhai, Yu-Ming ; Shanghai Institute of Technology
Keywords: Decision making, Decision theory, Economics
Subject Category<br>Please
select at least two subject
categories that best reflect
the scope of your manuscript:Industry applications, Education
Additional Manuscript
Keywords:SMEs, uncertainty

For Review OnlyIEEE Access

Governance Role of Media Information in IPO Market-Oriented Pricing
Fei Zhang1, Xiao-Hua Zhou1 and Jiafu Su2,*, Sang-Bing Tsai3,4,*, Yu-Ming Zhai5
1 School of Economics and Business Administration, Chongqing University, Chongqing 400044,
China; [anonimizat](F. Z.); [anonimizat] (X. Z )
2 Chongqing Key Laboratory of Electronic Commerce & Supply Chain System, Chongqing
Technology and Business University, Chongqing 400067, China ; [anonimizat] (J.S.)
3 Zhongshan Institute, University of Electronic Science and Technology, Zhongshan 528400,
China; [anonimizat] (S.T.)
4 College of Business Administration, Capital University of Economics and Business, Beijing
100070, China
5 School of Economics and Management, Shanghai Institute of Technology, Shanghai 201418,
China; [anonimizat] (Y.Z.)
* Correspondence: [anonimizat] (J.S.); [anonimizat] (S.T.)
Abstract: The market-oriented reform of IPO can alleviate the financing constraints of SMEs in China and promote the
optimal allocation of social resources and sustainable economic development. Media information plays a key governance
role in IPO market pricing. We construct a theoretical model of the influence of media signals on IPO pricing, which
describes the micro process in which uncertain signals in media influence retail investors' decisions and IPO underpricing.
Then, we take 516 SMEs listed in a-share from July 2009 to December 2012 as samples for empirical test, and establish
an in-depth learning model for text analysis with Java programming to measure Chinese media tone. Finally, the results
of the model analysis are verified by empirical results. Our results show that, authoritative media with high credibility
can reduce the uncertainty of information source, attract more investors' attention, and improve the valuation and demand
of retail investors. The higher the media credibility is, the higher the IPO underpricing rate is. The uncertain tone of the
media will increase the decision-making cost of investors, reduce the valuation expectation and demand of the secondary
market, and lead to a lower IPO underpricing rate.
Keywords: media information; uncertainty; IPO underpricing. text analysis; SMEs; decision analysis

1. Introduction
The difficulty of financing in small and medium
enterprises (SMEs) is a common economic problem in all
countries of the world. Especially for developing countries
with short economic marketization, SMEs financing
constraints have become an important bottleneck
restricting social innovation and sustainable economic
development. In 2005, the United Nations put forward the
concept of inclusive finance, emphasizing the three
principles of SMEs financing: equal opportunity,
sustainability and affordable costs. The main reasons for
the financing difficulty of SMEs in China are single
financing channel and serious information asymmetry. In
recent years, the Chinese government has paid more and
more attention to building a diversified financing system
and promoting the development of direct financing
channels, such as encouraging equity financing for SMEs,
promoting the normalization of new share issuance, setting
up a science and technology board and piloting an IPO
registration system.
As an important direct financing channel, IPO can not
only alleviate the financing constraints of SMEs and realize the optimal allocation of social funds, but also
promote technological innovation and promote sustainable
economic development [1,2]. However, there is still
information asymmetry in the IPO process. In order to
attract investors to participate in the IPO, the issuer usually
issues shares at a price lower than the intrinsic value of the
company, resulting in IPO underpricing [3-5]. Ritter [6]
pointed out that IPO underpricing is an indirect cost of
enterprises in the financing process, and sufficient
information disclosure before listing can reduce the
financing cost of enterprises. Li et al. [7] pointed out that
information cost determines the degree of financing
constraints faced by enterprises, and reducing information
asymmetry can improve the financing ability of enterprises.
Ang and Brau [8] found that there was a negative
correlation between enterprise information transparency
and IPO cost, which significantly affected the sustainable
development of enterprises. Just like information
disclosure of the corporate and due diligence by financial
intermediaries, the media as an independent third-party
information channel also plays an important role in
mitigating IPO information asymmetry. The main content
of this paper is to study the information governance roles Page 1 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

2
of media in the IPO process by using Chinese SMEs' IPOs
as samples.
Media information has the characteristics of easy
observation and low cost [9], and the media can dig out
information in-depth by tracking financial hotspot issues
and play a role in information production [10]. In an IPO
environment with highly asymmetric information, media
reports can prompt investors to quickly understand listed
companies and effectively reduce information asymmetry
between investors and issuers [11]. However, in recent
years, with the development of behavioral finance theory,
scholars have found that the information that the media
conveys to investors is often accompanied by sentimental
color [12]. Tetlock [13] studied the contents of the Wall
Street Journal and found that the negative tone of the media
has a significant inhibitory effect on the stock price. Media
sentiment can influence investors' decision-making
behavior by guiding investors' psychological expectations,
which in turn affects the market price of IPOs. Therefore,
media information is not only a carrier for transmitting
company quality signals, but also an important factor for
changing investor's beliefs and expectations [10].
Unlike the accuracy of data information, textual
information in media reports has natural uncertainty [14].
The uncertainty of media information is mainly reflected
in two aspects: information source and information content.
On the one hand, the differences of the credibility of media
sources will lead to different certainty in the signal
transmission process. Janney and Folta [15] pointed out
that not all media information has the same value, and the
signals sent by authoritative media are often more effective
in reducing information asymmetry. On the other hand, the
content of media information contains many tones that
express uncertain meanings. Uncertain tone will make
investors' valuation and decision-making process more
complicated, which in turn affects investors' demands for
IPO. In addition, mainstream media in developed countries
in Europe and America are characterized by independence
and marketization, while mainstream media in China are
often subject to administrative intervention, so Chinese
media reports are often cautious about the use of negative
words. Chinese SMEs often have higher risks and
uncertainties in the process of growth. In order to balance
the administrative control and maintain independence, the
media may use more uncertain words for reporting SMEs.
In the Chinese stock market, the proportion of retail
investors is more than 70%, and they usually only get
company information through media channels. Compared
with developed countries, China's institutional background
and investor structure are different, causing the uncertainty
of media information exert different impact on the IPO
market. Although there are many research literatures on
media information and IPO underpricing, existing
literature usually assumes that signals from different media
are equally credible, without considering the impact of
media credibility on investors. European and American
scholars have done a lot of researches on media tone
analysis in English context, but Chinese scholars' research
on Chinese media tone is still in its infancy, and few
scholars have carried out in-depth analysis on uncertain
tone, negative tone and positive tone in Chinese media
reports. In addition, the existing literatures usually adopt
empirical analysis methods, and do not theoretically
analyze the micro-mechanism of media credibility or
media tone affecting investor decision-making. In order to
make up for these shortcomings, we have carried out
innovative research on the relationship between media
information and IPO underpricing from two aspects:
theoretical model and empirical test. Our research finds
that the uncertainty of media information can inhibit the
sentiment and demand of retail investors. When retail
investors make investment decisions, they rely more on
high-credibility media reports. The higher the media
credibility is, the higher the IPO underpricing rate will be.
The more uncertain the media is, the lower the IPO
underpricing rate will be.
The main contribution of this paper is to construct a
theoretical model of media signal transmission in the
context of China's IPO book-building system and media
control. Our model portrays the micro-process of the
uncertainty signal transmitted by the media affecting the
decision-making of retail investors. For the first time, we
take the uncertainty of media information as the research
object, and study the influence of the uncertainty of media
source and media content on the degree of IPO
underpricing of SMEs. In addition, we establish a Java
deep learning model for text analysis in the JRE
environment, and apply Text Mining to the process of
analyzing Chinese media tone. This is a useful supplement
to the Chinese media tone research system that is still in
the exploration stage. Finally, we selected 516 SMEs listed
on the A-share market from July 2009 to December 2012
as samples, using empirical methods to verify the
conclusions and related mechanisms of the theoretical
model.
The rest of the paper is structured as follows: Part 2
is literature review. Part 3 establishes a theoretical model
and proposes hypotheses. Part 4 designs the methods of
empirical research, including sample selection, media
report collection, text analysis and variable description.
Part 5 shows the results of the empirical test and verifies
the three hypotheses of theoretical analysis. Part 6 is the
conclusion of this paper. Page 2 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

3
2. Literature Review
As for the causes of IPO underpricing, information
asymmetry theory is the mainstream theory generally
accepted by scholars [3,5,16]. On the basis of information
asymmetry theory, principal-agent theory [17], signal
theory [18] and "winner's curse" hypothesis [5] are derived.
With the media playing an increasingly important role in
the information transmission of financial markets, more
and more scholars have begun to study the impact of media
reports on IPO pricing and market performance [19-20].
Traditional research literatures mainly study the
impact of media on IPO from two aspects: information
asymmetry and behavioral finance. Literatures based on
information asymmetry perspective hold that media can
improve the efficiency of information dissemination and
reduce the degree of information asymmetry. Pollock and
Rindova [21] pointed out that the media can reduce
information friction in the IPO process and help investors
quickly form a basic understanding of new listed
companies. Frankel and Li [22] believe that media reports
before IPO can weaken the information advantage of
informed traders and play the information governance
function. Dyck [23] found that media reports can reduce
the private cost of external investors in information
processing, ensure the diversity of market participants and
increase the effectiveness of the market. Xiong Yan et al.
[24] found that media reports can reduce the degree of
information asymmetry in the pricing of primary market,
make the quotations of institutional investors more close to
the intrinsic value of IPO, and improve the pricing
efficiency of new issue. Wang Changyun et al. [25] studied
the influence of the company's media management on IPO
pricing, and found that companies use the publicity
function of media to release value information in a timely
manner and improve the pricing efficiency of the primary
market.
Scholars from the perspective of behavioral finance
believe that media reports will affect investors' emotional
fluctuations and lead to IPO mispricing. Tetlock [13]
found that the greater the fluctuation range of investor
sentiment, the greater the degree of asset mispricing would
be. Many scholars have found that media reports can guide
and strengthen the sentiment of investors in the IPO market
[26]. Cook et al. [27] believe that underwriters may collude
with the media and make profits from it, and underwriters
guide investors to generate irrational emotions with the
help of positive reports and make the IPO price deviate
significantly from the intrinsic value of the IPO. Liu et al.
[28] found that media reports before IPO can attract
investors' attention and stimulate more demands for new
shares, thus leading to a positive correlation between the
amount of media reports and IPO discount. Shao xinjian et
al. [29] found that positive media news can significantly increase investors' attention and optimism, thus leading to
a large upward deviation of IPO price from the intrinsic
value. Chen Pengcheng and Zhou Xiaohua [30] found that
media emotions can lead to cognitive bias of retail
investors and stimulate IPO market demand, leading to
higher first-day returns and poorer long-term performance.
Traditional literatures mostly adopt simple methods
to measure media information, for example counting the
total number of mentions of IPO companies by media and
simply judging the tendency of media reports [31-33]. In
recent years, with the wide application of text mining
technology in finance and the rapid development of
communication studies, scholars began to deeply study the
impact of media tone and media sources on IPO. Janney
and Folta [15] found that although media reports can
alleviate the information asymmetry between IPO
companies and investors, information from authoritative
media is more valuable to investors. Pollock et al. [34]
believed that media coverage of IPO not only includes
positive information and negative information, but also
includes neutral information or uncertain information.
Orhun et al. [11] used signaling theory to investigate the
credibility of the media and uncertain tone of the media
coverage about an IPO performance.
3. The Model
3.1. Problem Description
We learn from Chen Pengcheng and Zhou Xiaohua
[30] and create a mathematical model that influences the
decision-making of investors. The main issues involved in
the model are described as follows:
We assume that the total number of new shares issued
is 1 unit in the context of Chinese bo ok-building. The
issuer knows the intrinsic value of the IPO and is rational
enough. His knowledge of the company's intrinsic value is
𝑉~𝑁(𝑉̅,1/𝜌𝑣), which is an unbiased prior estimate. Since
the issuer and underwriter need to give a preliminary
valuation range before the inquiry, the investors can
observe the issuer's prior estimate. We assume that all
investors are uncertainty -averse and risk -averse, and their
utility function is CARA function, and the risk aversion
coefficient is γ. The investo rs of primary market are the
institutional investors who participate in the inquiry, and
they are rational enough. The investors of secondary
market are represented by retail investors, and they are in
bounded rationality. The underwriter determines the to tal
off-line allotment and the offering price according to the
results of the inquiry. The proportion of shares alloted to
the institutional investors off-line is 𝑎, and the remaining
shares 1−𝑎 is issued online to the retail investors.
Institutional investors have rich professional
knowledge and valuation experience. They can obtain Page 3 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

4
more accurate value information by reading prospectus,
participating in roadshows and field research. Therefore, it
is assumed that institutional investors' valuation of IPO
will not be influenced by the media coverage. In the
inquiry stage, the issuer transmits value signal to the
institutional investors, and we assume that the signal
structure is 𝑆𝑖=𝑉+𝜀, 𝜀~𝑁(0,1/𝜌𝜀), where 𝜀 and 𝑉
are independent of each other, 𝑖=1,2…𝑁. Institutional
investors have heterogeneous beliefs, so the
implementation values of the signal are different, but the
signal structure is common knowledge. Let the
implementation value of the signal is 𝑠𝑖, which indicates
the valuation level of the institutional investor 𝑖, which is
expressed by the way of accumulated bidding enquiry.
Due to the lack of professional knowledge and
valuation ability, retail investors cannot form private
information on company value. They obtain valuation
information through institutional investors' biddings and
media reports. It is assumed that media signals enable retail
investors to generate new insights into IPO's valuation and
drive them to trade. The media signals received by retail
investors are represented by 𝑆𝑗, 𝑗=1,2…𝑀, where 𝑀 is
the total number of retail investors affected by the media
signals. The implementation value of 𝑆𝑗 is 𝑠𝑗, indicating
the extent how the retail investor 𝑗 is affected by the
media signals. The amount of change in th e IPO valuation
of retail investors caused by the media signal is 𝛥𝑉, then
the signal structure can be expressed as 𝑆𝑗=𝛥𝑉+𝜑,
𝛥𝑉~𝑁(𝛥𝑉̅̅̅̅,1/𝜌𝛥), 𝜑~𝑁(0,1/𝜌𝜑). 𝛥𝑉 is closely related
to the quality of the media signal, and 𝛥𝑉 can b e
decomposed into 𝛥𝑉=𝑐(𝛥𝑉𝑜−𝛥𝑉𝑢), 0<𝑐<1, where
𝑐 is the media credibility, and 𝛥𝑉𝑜 is the change amount
of IPO valuation caused by the certain media signal. 𝛥𝑉𝑢
is the uncertainty's compensation. Because retail investors
are unc ertainty aversion and risk aversion, the lower
(higher) the media credibility is and the more (less)
uncertain the tone is, the more pessimistic (optimistic) the
retail investor’s judgment on the IPO's value will be. And
they will require revenue compensat ion for uncertain
signals, resulting in a smaller (larger) valuation variation
𝛥𝑉. Therefore, the mean value of 𝛥𝑉 (𝛥𝑉̅̅̅̅) can be used as
an indicator to measure the quality of media signals.
We divide the IPO pricing process into four periods.
As shown in Figure 1, the underwriter invites institutional
investors to participate in the inquiry during the 𝑡0 period,
and the institutional investors will feedback quotation and
demand information to the underwriter. During the 𝑡1
period, the underwriter and the issuer jointly determine the
final issue price 𝑝1 based on the feedback from the
institutional investors, and then place the placement to the
institutional investors. During the 𝑡2 period, new shares
are publicly traded in the secondary market and reach
equilibrium trading price 𝑝2. During the 𝑡3 period, the new shares return to its intrinsic value and are liquidated.
In the period of 𝑡3, because the rationality degree and the
influence degree caused by the media of two types of
investors are different, they have different opinions on the
judgment of t he value of new shares. Among them,
institutional investors believe that new shares are
liquidated at intrinsic value 𝑝3=𝑉, while retail investors
believe that new shares are liquidated at 𝑝3=𝑉′=𝑉+
𝛥𝑉.

Figure 1. Schematic diagram of the time
division of the IPO pricing process.
3.2. Equilibrium Analysis
1. Determination of the IPO issue price of the primary
market
During the 𝑡0 period, institutiona l investors judge
the value of the IPO based on the private signals obtained,
and determine t he bid price and the optimal purchase
amount to maximize its expected utility during the 𝑡3
period. The optimal purchase amount is the optimal
demand of institutional investors, so there is:
max
𝑞0𝑖𝐸[−𝑒−𝛾[𝑊0𝑖+𝑞0𝑖(𝑉−𝑝1)]], (1)
where 𝑞0𝑖 is the demand for new shares of institutional
investor 𝑖 during 𝑡0, 𝑊0𝑖 is his initial wealth, and 𝑝1 is
the issue price. Basing on the properties of the CARA
utility function and the normal distribution assumption,
equat ion (1) can be converted to:
max
𝑞0𝑖𝛾𝑊0𝑖+𝛾𝑞0𝑖[𝐸(𝑉 | 𝑆𝑖=𝑠𝑖)−𝑝1]−
(1/2)𝛾2𝑞0𝑖2var(𝑉 | 𝑆𝑖=𝑠𝑖), (2)
where 𝐸(𝑉 | 𝑆𝑖=𝑠𝑖) is a posteriori estimate of
institutional investors, which is generated by evaluating
the new shares based on their private signal, and
var(𝑉 | 𝑆𝑖=𝑠𝑖) represents the variance of the estimate.
Combined with Kyle lemma, there is:
𝐸(𝑉 | 𝑆𝑖=𝑠𝑖)=𝜌𝑉𝑉+𝜌𝜀𝑠𝑉−𝜌𝜀𝑠𝑃
𝜌𝑉+𝜌𝜀 ，
var(𝑉 | 𝑆𝑖=𝑠𝑖)=1
𝜌𝑉+𝜌𝜀. (3)
We take the derivative of eq. (2) with respect to 𝑞0𝐼,
and let the derivative be 0 . Then we can calculate the
optimal demand of institutional investors is:
𝑄0𝑖=1
𝛾(𝜌𝑉𝑉+𝜌𝜀𝑠𝑖−𝜌𝑉𝑝1−𝜌𝜀𝑝1). (4)
equilibrium price
Issue pricing
inquiry
t0
t1
t2
t3
Liquidatio n Page 4 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

5
When the primary market go into equilibrium, the
sum of the optimal demand of institutional investors is
equal to the total quantity of the placement to institutional
investors, ie ∑ 𝑄𝑁
𝑖=1 0𝑖=𝑎. Combined with eq. ( 4), the
issue price at the 𝑡1 period can be found as:
𝑝1=𝑁(𝜌𝑉𝑉+𝜌𝜀𝑠𝑁)−𝑎𝛾
𝑁(𝜌𝑉+𝜌𝜀), (5)
where 𝑠𝑁=1
𝑁∑ 𝑠𝑖𝑀
𝑖=1, 𝐸(𝑠𝑁)=𝑉.
2. Determination of the equilibrium transaction price of
the secon dary market
During the 𝑡2 period, because institutional investors
are not affected by media signals, and they believe that the
fluctuation of stock price caused by retail transactions is
temporary, so they do not change their posteriori estimate
about IPO value. In addition, in reality, the off -line
allotment of shares has a three -month lock -up period,
resulting in institutional holdings still unable to trade
during the 𝑡2 period, so the demand of institutional
investors during the 𝑡2 period is still 𝑄0𝑖.
In addition to obtaining public disclosure information
and institutional quotation information, retail investors can
also receive media information during the 𝑡2 period. On
this basis, retail investors determine their own demand to
maxi mize their utility during the 𝑡3 period, so there is:
max
𝑞2𝑗𝐸[−𝑒−𝛾[𝑊2𝑗+𝑞2𝑗(𝑉′−𝑝2)]], (6)
where 𝑊2𝑗 is the initial wealth of retail investors in 𝑡2
period, 𝑉′ is the liquidation value expected by retail
investors in 𝑡3 period. And eq. (6) can be written as:
max
𝑞2𝑗𝛾𝑊2𝑗+𝛾𝑞2𝑗[𝐸(𝑉| 𝑆𝑁=𝑠𝑁,𝑆𝑗=
𝑠𝑗)−𝑝2]−(1/2)𝛾2𝑞2𝑗2var(𝑉|𝑆𝑁=𝑠𝑁,𝑆𝑗=
𝑠𝑗). (7)
We take the derivative of eq. (7) with respect to 𝑞2𝑗,
calculate the posterior estimates and variances of the retail
investors ( the same with calculation method of eq. (3)),
and find out their optimal new shares demand:
𝑄2𝑗=1
𝛾(𝜌𝑉𝑉+𝑁𝜌𝜀𝑠𝑁
𝜌𝑉+𝑁𝜌𝜀+𝜌𝛥𝛥𝑉̅̅̅̅+𝜌𝜑𝑠𝑗
𝜌𝛥+𝜌𝜑−
𝑝2)(𝜌𝑉+𝑁𝜌𝜀)(𝜌𝛥+𝜌𝜑)
𝜌𝑉+𝜌𝜀+𝜌𝛥+𝜌𝜑. (8)
Under the equilibrium condition of the secondary
market, the sum of the total institutional investors' demand
and the total retail investors' demand is 1 unit, that is,
∑ 𝑄𝑁
𝑖=1 0𝑖+∑ 𝑄2𝑗𝑀
𝑗=1 =1. Then the equilibrium
transaction price of the secondary market is determined as
follows: 𝑝2=𝜌𝑉𝑉+𝑁𝜌𝜀𝑠𝑁
𝜌𝑉+𝑁𝜌𝜀+𝜌𝛥𝛥𝑉̅̅̅̅+𝜌𝜑𝑠𝑀
𝜌𝛥+𝜌𝜑−
𝛾(1−𝑎)
𝑀𝜌𝑉+𝜌𝜀+𝜌𝛥+𝜌𝜑
(𝜌𝑉+𝑁𝜌𝜀)(𝜌𝛥+𝜌𝜑) , (9)
where 𝑠𝑀=1
𝑀∑ 𝑠𝑗𝑀
𝑗=1, and 𝐸(𝑠𝑀)=𝛥𝑉̅̅̅̅.
3.3. Hypotheses
Combining eq. (5) and eq. (9), we get the IPO
underpricing rate:
𝐼𝑈=𝐸(𝑝2)−𝐸(𝑝1)
𝐸(𝑝1)=𝑁𝛥𝑉̅̅̅̅(𝜌𝑉+𝜌𝜀)+𝛾𝑎
𝑁𝑉(𝜌𝑉+𝜌𝜀)−𝛾𝑎−
𝛾(1−𝑎)(𝜌𝑉+𝜌𝜀+𝜌𝛥+𝜌𝜑)
𝑀(𝜌𝑉+𝑁𝜌𝜀)(𝜌𝛥+𝜌𝜑)𝑁(𝜌𝑉+𝜌𝜀)
𝑁𝑉(𝜌𝑉+𝜌𝜀)−𝛾𝑎 , (10)
It can be clearly seen from eq. (10) that the IPO
underpricing rate 𝐼𝑈 is positively correlated with 𝑀. It
can be inferred that if the media reports about IPO are more,
the number of retail investors affected by media signals
will be more, resulting the att entions and the demands of
the secondary market towards IPO become higher, and the
degree of IPO underpricing will be higher. On this basis
we propose the first hypothesis:
Hypothesis 1. The number of media reports is positively
correlated with the IPO und erpricing rate. The
more(less) the media reports are, the higher (lower) the
IPO underpricing rate will be.
Learning from the research of Orhun et al. [11], we
define media credibility 𝑐 as the degree to which investors
believe that the source of media signals is trustworthy, and
𝑐 is expressed as the proportion of authoritative media
reports to the total number of reports about the compony.
Therefore, in the case of controlling the total number of
media reports, the greater the number of authoritative
media reports is, the higher the credibility of the media will
be. Authoritative media has more audiences, wider
coverage and stronger credibility than ordinary media. On
the one hand, it means that authoritative media can
influence a larger number of retail investors (ie 𝑀 is
larger), so that the secondary market shows more attentions
and demands for IPO. On the other hand, authoritative
media can reduce the uncertainty of signal sources and
strengthen the existing beliefs and perceptions of retail
investors, making them believe that the media information
contains lower uncertainty and the investment value is
higher (ie 𝛥𝑉̅̅̅̅ is larger), thereby further increasing the
attentions and demands of optimistic investors on IPO. (Eq.
(8) shows that the valuation change 𝛥𝑉̅̅̅̅ is positively
correlated with the optimal new shares demand 𝑄2𝑗.)
Based on the analysis of the above two aspects, more
authoritative media reports will stimulate higher secondary
market demand and valuation levels, and further increase
the initial price of new shares under the influence of Page 5 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

6
optimism mood and herd effect. (Eq. (9) shows that the
equilibrium price of secondary market 𝑝2 is positively
correlated with 𝑀 and 𝛥𝑉̅̅̅̅), and finally shows a higher
IPO underpricing rate. Thus, the second hypot hesis can be
obtained:
Hypothesis 2. Media credibility is positively correlated
with IPO underpricing rate. The higher (lower) the media
credibility is , the higher (lower) the IPO underpricing
rate will be.
It can also be seen from eq. (10) that the IPO
underpricing rate 𝐼𝑈 is positively correlated with 𝛥𝑉̅̅̅̅.
That is, the more uncertain the media tone is, the smaller
the ( 𝛥𝑉̅̅̅̅) will be, and the lower the IPO underpricing rate
will be. Next, we analyze this in conjunction with the
actual IPO process. Uncertain media t one makes potential
investors think that the company contains more uncertainty
or unknown distribution risk, and investors will demand
higher risk compensation to hold the asset, resulting in a
lower valuation level of IPO (ie 𝛥𝑉̅̅̅̅ is smaller). In
additi on, the uncertain tone will increase the difficulty of
secondary market investors' IPO valuation, making their
decision -making process more complicated. The Ellsberg
paradox points out that due to the aversion to uncertainty,
investors will allocate less w eight to assets with uncertain
information [35]. Therefore, when more uncertainty
signals are received, investors may look for alternative
assets resulting in the decrease of the new shares demand.
(𝑄2𝑗 is negatively correlated with uncertain media tone),
which reduces the price of new stocks in the secondary
market ( 𝑝2 is negatively correlated with uncertain media
tone) and ultimately shows a lower IPO underpricing rate.
In summary, we get th e third hypothesis of this paper:
Hypothesis 3. Uncertain media tone is negatively
correlated with IPO underpricing rate. The higher(lower)
the media's uncertainty is, the lower (higher)the IPO
underpricing rate will be.
Figure 2. Schematic of the impact of the media
information's uncertainty on IPO underpricing.
Based on the above analysis, the uncertainty of media
information exists in both signal source and signal content.
We take media credibility as the reverse proxy variable of
the uncertainty of signal source, and lower credibility of
media tone will increase uncertainty in the process of
signal transmission. While the uncertainty of media tone
increases the uncertainty in the signal content. Either the
uncertainty in the signal transmission process or in the
signal content will reduce the IPO demand and valuation
level of the secondary market, which will eventually lead
to a lower IPO underpricing rate. In order to understand the
impact mechanism of media information's uncertainty on
IPO underpricing more intuitively, Figure 2 summarizes
the theoretical analysis framework of the above analysis.
Subsequently, we will design an empirical analysis method
to empirically test the above three hypothesises and their
mechanisms of action.
4. Empirical Design
4.1. Sample and Data
In July 2009, the China Securities Regulatory
Commission (CSRC) implemented the reform of IPO
book-building and liberalized the pricing limit based on the
PE ratio. In 2013, IPO is suspended. After the IPO was
restarted in January 2014, the PE ratio window guidance
for IPO pricing was again implemented, and the daily limit
of 44% on the first day of listing was set. Therefore, in
order to obtain market-based pricing data, the sample
selected in this paper is from July 2009 to December 2012.
The initial sample contains 548 SMEs listed on the A-share
market during this period. SMEs are identified according
to the "Notice on the Issuance of Standard Provisions on
Page 6 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

7
the Classification of SMEs" (MIIT Joint Enterprise [2011]
no. 300). Then, based on the research needs, we conduct
the following screening of the initial samples: (1) exclude
companies that did not have media reports before the IPO,
(2) exclude listed companies in the financial industry, and
(3) exclude companies with missing data. The resulting
empirical sample totaled 516 companies. The listed
company data used in this article are from the China Stock
Market & Accounting Research (CSMAR) Database and
the Wind Financial Information Database.
4.2. Measurement of Media Credibility
Referring to the previous literature on IPO media
research [36], the media reports in this paper are from the
China Core Newspapers Full-text (CCNF) Database. The
statistical methods are as follows: (1) The time span for
limiting the search is from the prospectus announcement
day to the day before the date of listing. Taking into
account the timeliness of media reports, the media has
begun to vigorously promote IPO after the announcement
day. At this time, the retail investors will start to pay
attention to IPO. Besides, because the issue pricing has
been completed when the IPO announcement was released,
the time span we choose can rule out the possible impact
of media reports on institutional inquiry. (2) We use the
short name, full name and stock code of the listed company
for topic search, then eliminate the irrelevant reports by
manual screening. (3) There is no limit to the types of the
media in summing up the total amount of media reports to
avoid omissions. Therefore, our media sources includes
not only official and market-oriented financial newspapers
but also industry newspapers and local newspapers.
Through the above method, the total number of reports of
each IPO company (indicated by Med) is calculated, and
each report content is downloaded. Finally we obtain a
total of 4122 reports.
Media credibility (represented by CM) is the
proportion of authoritative media reports to the total
number of reports. In order to count the number of reports
of authoritative media, we draw on the classification
criteria of authoritative media by previous scholars [36-37].
According to the authority, influence and popularity, we
identify the following eight newspapers as authoritative
media sources: China Securities Journal, Securities Times,
Securities Daily, Shanghai Securities News, China
Business News, Economic Observer, 21st Century
Business Herald and First Financial Daily. Then,
according to the above search method, the number of
authoritative media reports of each company (indicated by
CMed ) is counted. Finally, the media credibility of each
sample company can be calculated by:
𝐶𝑀 =𝐶𝑀𝑒𝑑 /𝑀𝑒𝑑 ×100% . (11) 4.3. Measurement of Media Tone
We use text analysis methods to measure media tone
by calculating the proportion of tone's feature word to the
full text. To measure the positive and negative media tone,
we draw on Wang Changyun and Wu Jiawei [10] and
construct an optimized tone's lexicon suitable for Chinese
financial context. Their lexicon only includes positive
word list and negative word list. The difference is that our
research focuses on the uncertainty of media content, so
our lexicon contains not only positive and negative word
list, but also uncertain word list.
All of the calculation programs of text analysis are
run in the Java Runtime Environment. The specific
analysis process includes five steps. The first step is
preprocessing of media text. We use Java programming as
a pre-processing tool to remove irrelevant Html tags, charts,
symbols, English words and numbers. Then we set stop
words. The second step is spliting sentences and automatic
word segmentation. The sentence is used as the smallest
information analysis unit, and the preprocessed text is
decomposed. Then the sentence is processed into word
segmentation by the JiebaTokenFilter which is a tokenizer
for Chinese. The third step is constructing a financial tone's
lexicon. We translate the uncertain, negative and positive
word lists included in the Loughran and Mcdonal [38]
word lists into Chinese. Then, according to the Modern
Chinese Dictionary and the Chinese context, the Chinese
translation of the three word lists are sorted out. And we
obtain preliminary tone's lexicon suitable for Chinese.
Combining with the Chinese sentiment dictionary
published by CNKI, we introduce Java deep learning
model to match, supplement and de-duplicate the
preliminary lexicon. Finally, the financial tone's lexicon
suitable for Chinese context was obtained. The three word
lists include 271 uncertain words (eg. approximate ,
possible , and maybe ), 4370 negative words (eg.
destruction , rejection , bankruptcy , and difficulties ), and
2256 positive words (eg. beneficial , successful , and
powerful ). The forth step is identification and judgment of
media tone. We load the disassembled word vector and
tone's lexicon into the calculation model. Then the
calculation model uses the tone's lexicon to judge the word
vectors in the sentence one by one and detect whether there
is negative logic in the sentence. The basic principle of the
model algorithm is shown in Figure 3. The fifty step is
calculating the media tone index. We assume that the
measure of text tone follows the principle of linear
superposition, and thus we can calculate the media tone as
follows: Page 7 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

8
𝑈𝑛=
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑢𝑛𝑐𝑒𝑟𝑡𝑎𝑖𝑛 𝑤𝑜𝑟𝑑𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑤𝑜𝑟𝑑𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠×
100% ,
𝑁𝑒𝑔 =
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑤𝑜𝑟𝑑𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑤𝑜𝑟𝑑𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠×
100% ,
𝑃𝑜𝑠 =
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑤𝑜𝑟𝑑𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑤𝑜𝑟𝑑 𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑑𝑖𝑎 𝑟𝑒𝑝𝑜𝑟𝑡𝑠×
100% , (12
)
where Un, Neg and Pos respectively express
uncertain, negative and positive media tone.

Figure 3. Schematic diagram of the basic
algorithm of the Java model.
4.4. Measurement of Media Tone
The dependent variables are defined as follows: The
IPO underpricing rate ( IR) is the change ratio of the closing
price of the first day to the issue price. This paper mainly
studies the pricing of IPO in the secondary market on the
first day of listing. Therefore, the post-issued fully diluted
P/E ratio ( PE) based on the first-day closing price is
selected as the proxy variable of the secondary market
valuation level. In order to increase the robustness of the
research conclusions, the ratio ( P) of the first-day P/E to
the industry average P/E is used to measure the relative
valuation of individual stocks in the secondary market. In
view of the differences in the number of shares issued, the
comparability of different companies on the first day of
trading is relatively poor. So the first day turnover rate ( TO)
is selected as the proxy variable for the demand for new
shares in the secondary market.
Our explanatory variables includes the total number
of media reports ( Med), media credibility ( CM), uncertain
media tone ( Un), negative media tone ( Neg) and positive
media tone ( Pos). In terms of controlling variables, we have controlled the fundamental variables, distribution and
market performance factors of IPO. In addition, we also
control the impact of the total number of media reports in
the empirical test of how media credibility affects IPO
underpricing ratio. The specific definition of the control
variables is shown in Table 1.
Table 1. Control Variable Definition Table.
Variable
Symbol Variable
Definition Measurement Me thod
Age company age ln(number of years +1)
Indus Industry class 11 industry dummy variables
Asset asset size ln(total assets of the year before
IPO) Debt asset -liability
ratio the ratio of liabilities to total assets
Sales Sales ln(sales of the year before IPO)
ROE return on equity the ratio of net profit to net assets
VC VC background venture capital supports 1,
otherwise 0 UW underwriter
reputation lead underwriter in the top ten is 1,
otherwise 0 Size Issuance ln(IPO financing scale)
MR market returns
market index yield on IPO day
Rate online success
rate online ticket divided by valid
subscription
4.5. Descriptive Statistics
The descriptive statistics of the data are reported in
Table 2. The average and median values of IR are 24.183%
and 17.054% respectively, indicating that the IPO samples
have a higher degree of underpricing on average. The
average PE is 51.822, and the mean and median of P are
1.970 and 1.708 respectively, which means that the
valuation of IPOs after the close of the first day is usually
much higher than the industry average. The average
turnover rate on the first day was as high as 67.70%,
indicating that the IPO generally had a high trading volume
on the first day. The median total number of media reports
is 7.988, which means that there are about 8 reports per
IPO company. The average CM is 68.848, indicating that
the authoritative media coverage accounted for an average
of 68.848% of all reports. In the three media tone, the
content of the uncertain media tone is the least, and the
mean and median are only 0.125% and 0.109%,
respectively. The average negative media tone is 0.553%,
while the positive media tone is 4.39 times the negative
media tone, indicating that the positive words used in
media reports are far more than negative words, which is
consistent with other scholars' research: the media is tend
to report IPO positively [10]. In addition, we also
performed a correlation test and a multicollinearity
diagnosis on the data. The variance spread factor (VIF) of
the variables were less than 5, indicating that the model did
not have significant multicollinearity problems.

Page 8 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

9
Table 2. Descriptive statistics for each
variable.
Variables Mean
Mean S.D.
Std.Dev. Min
Min Median
Median Max
Max IR /% 24.183 32.085 –
26.333 17.054 221.000
PE 51.822 28.035 7.230 45.390 189.650
P 1.970 1.166 0.362 1.708 9.483
TO 0.677 0.229 0.120 0.770 0.960
Med 7.988 5.175 1.000 5.000 68.000
CM /% 68.848 37.668 0.000 66.667 100.000
Un /% 0.125 0.131 0.000 0.109 1.218
Neg /% 0.553 0.621 0.000 0.388 5.690
Pos /% 2.427 0.863 0.579 2.101 11.288
5. Results
5.1. Regression Analysis and Results
We use the OLS regression analysis to test the three
hypotheses of theoretical analysis. Table 3 shows the test
of hypothesis 1 and hypothesis 2 with the IPO underpricing
rate as the explanatory variable. Column (1) shows
regression results with only 11 traditional IPO control
variables, with an adjusted R2 of 36.38%. Column (2)
increases the total number of media reports ( Med) as an
explanatory variable, and the adjustment R2 increases to
38.87%, indicating that the overall explanatory ability of
the regression model has increased after adding Med .
Med is positively correlated with the IPO underpricing
rate at the 1% level. On average, for every 1% increase in
media coverage, the IPO underpricing rate will increase by
0.0275. The results are consistent with the model analysis,
and also in line with other research [26]. The more the
number of media coverage is, the higher the investor
sentiment will be, and the higher the investor's attention
will also be , finally resulting in a higher underpricing rate.
So the hypothesis 1 is verified.
Column (3) and column (4) add media credibility
(CM) as explanatory variable. The results show that CM is
significantly positively correlated with IPO underpricing
rate. Among them, column (4) controls the total number of
media reports, the overall explanatory ability of the
regression model was further increased. The adjusted R2
increases by 40.97%. From column (4), the higher the
media credibility, the higher the IPO underpricing. On
average, for every 1% increase in media credibility, the
IPO underpricing rate will increase significantly by 0.1543.
Orhun et al. [11] found that the credibility of media reports
before IPO was significantly positively correlated with the
IPO stock price changes in the first week, indicating that
media credibility has an important impact on investors'
attention and sentiment. Therefore, Hypothesis 2 is
verified. In the previous section, we have investigated the
mechanism of media credibility affecting IPO
underpricing. That is, authoritative media will increase the
level of secondary market valuation ΔV̅̅̅̅ and new stock
demand Q2j. In order to verify the mechanism, Table 4
takes the first day P/E ratio ( PE) and the relative P/E ratio
(P) as proxy variables for the secondary market valuation
level. Columns (1) to (4) report the impact of media
information on the secondary market valuation. Column (1)
and column (3) show that the regression coefficient of
Med is significantly positive, and t values are 1.98 and
1.85, respectively. It indicates that the more the media
reports are, the higher the level of investor attention will
be, leading to a higher secondary market valuation level.
Column (2) and column (4) add CM as an explanatory
variable. It can be found that the coefficients of CM are
significantly positive after controlling the number of media
reports. It can be seen that if the credibility of media
reports is higher, the secondary market investors will give
a certain premium to the new shares, thereby increasing the
valuation level of the secondary market. Column (5) and
column (6) show a significant positive correlation between
CM and first day turnover. Based on the deterministic
effect, investors will allocate more funds to IPO with
higher credibility in media reports, increasing the demand
for new shares in the secondary market.Page 9 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Table 3. Media credibility and IPO underpricing rate.
Variables (1) (2) (3) (4)
Med 0.0275 ***(5.18)
CM /% 0.1320*** (4.86) 0.1543*** (5.80)
Med (control ) 0.0283*** (6.08)
Age 0.0171 (0.42) 0. 0189 (0.44) 0. 0155 (0.46) 0. 0152 (0.41)
Indus -0.0058 (-0.40) -0. 0054 (-0.55) -0. 0057 (-0.56) -0. 0058 (-0.77)
Asset 0.1358*** (5.34) 0.1183*** (4.28) 0.1940*** (4.83) 0.1380*** (3.53)
Debt -0.3566*** (-2.73) -0.3711*** (-3.09) -0.3036*** (-2.63) -0.3366*** (-3.05)
Sales -0.0493* (-1.95) -0.0443 (-0.92) -0.0408 (-1.51) -0.0402 (-0.24)
ROE 0.4139*** (2.91) 0.4050*** (2.85) 0.4797*** (2.60) 0.4885*** (2.47)
VC -0.0321 (-0.98) -0.0353 (-1.09) -0.0254 (-0.94) -0.0280 (-0.97)
UW -0.0181 (-0.72) -0. 0130 (-0.61) -0. 0133 (-0.75) -0. 0114 (-0.63)
Size -0.2021*** (-8.26) -0.2732*** (-8.82) -0.2913*** (-7.84) -0.2383*** (-8.46)
MR 2.5265*** (7.39) 2.1464*** (7.31) 2.6280*** (6.91) 2.6081*** (6.75)
Rate -0.2606*** (-10.29 ) -0.3939*** (-10.86 ) -0.3345*** (-9.83) -0.2859*** (-
10.46 ) C 0.9879*** (9.34) 0.3489*** (9.25) 0.3624*** (7.76) 0.2288*** (7.42)
Adj R2 0.3638 0.3887 0.3846 0.4097
The t statistic is in parentheses. ***, **, and * indicate significant at the 1%, 5%, and 10% levels, respectively.
Table 4. Media credibility and first-day P/E ratio, P/E ratio and turnover rate.
PE P TO
(1) (2) (3) (4) (5) (6)
CM /% 6.2735**(2.0
8) 0.2417*(1.69) 0.1442***(6
.00) Med 1.2732**(1.9
8) 1.4109**(2.1
8) 0.0497*(1.85) 0.0550**(2.0
4) 0.0131***(3.
02) 0.0163***(3
.88) Age 1.7855(0.57) 2.0565(0.66) 0.0455(0.35) 0.0559(0.43) -0.0465**( –
2.20) -0.0403**( –
1.99) Indus 0.5719 (0.82) 0.5980 (0.86) 0.0070 (0.24) 0.0080 (0.28) 0.0017 (0.3 7) 0.0023
(0.52) Asset 2.4544(0.66) 1.4968(0.40) 0.1752*(1.73) 0.1383(1.29) 0.1392***(5.
57) 0.1172***(4
.83) Debt -2.3183**( –
2.37) -2.2813**( –
2.34) -0.7123*( –
1.75) -0.6981*( –
1.72) -0.1917***( –
2.91) -0.1832***( –
2.89) Sales -1.3022( –
0.41) -1.9585*( –
0.62) -0.1117( –
0.85) -0.0864( –
0.65) -0.0512**( –
2.39) -0.0361*( –
1.75) ROE 3.3313**(2.3
9) 2.3491**(2.2
8) 0.0290**(2.0
8) 0.0669**(2.1
9) 0.0244(0.42) 0.0018(0.03)
VC -0.9982( –
0.37) -1.0171( –
0.37) -0.0448( –
0.39) -0.0455( –
0.40) -0.0036( –
0.20) -0.0032( –
0.18) UW -0.6124(-
0.22) -0.5974( –
0.22) -0.0602( –
0.53) -0.0596( –
0.52) -0.0285( –
1.54) -0.0289( –
1.63) Size -9.3275***( –
2.67) -8.6212**( –
2.46) -0.3575**( –
2.45) -0.3303**( –
2.26) -0.1530***( –
6.49) -0.1368***( –
6.01) MR 8.6905***(2.
80) 8.0243**(2.5
7) 3.7955***(2.
93) 3.5389*** (2.
72) 1.0290***(4.
91) 0.8758***(4
.33) Rate -1.1533***( –
3.54) -1.0822***( –
3.30) -0.4904***( –
3.61) -0.4630***( –
3.38) -0.2297***( –
9.44) -0.2133***( –
9.03) c 5.6567***(4.
83) 5.0436***(4.
12) 2.3387***(4.
79) 2.1025***(4.
12) 0.9118***(11
.53) 0.7708***(9
.71) Adj R2 0.2947 0.3321 0.3094 0.3139 0. 3424 0. 3634
Table 5 tests the relationship between the three media
tone and the IPO underpricing rate. Un is negatively
correlated with IPO underpricing ratio, and the correlation
coefficient (-0.0954) is significant at the 1% level. On
average, for every 1% increase in uncertian media tone,
IPO underpricing rate significantly reduced by 0.0954. If there is more uncertain tone content in media reports, the
uncertainty-averse investors think that the risk of holding
new shares will be greater, resulting in the initial price
performance is weaker, which leads to a lower
underpricing rate. Neg is also negatively correlated with
the IPO underpricing rate, and the at the correlation (-
0.0443) is significant at the 1% level. On average, for every Page 10 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

2
1% increase in negative tone, the IPO underpricing rate is
significantly reduced by 0.0443. More negative media tone
will increase the pessimism of secondary market investors,
which can significantly restrain investors' demand for new
shares, resulting in lower IPO underpricing rates. The
coefficient of Un is obviously greater than Neg. That is,
the influence of uncertainty tone on IPO underpricing is
stronger than negative tone, which indicates that investors
are more averse to future uncertain events than negative
events that have already occurred. Pos is positively
correlated with IPO underpricing ratio, and the correlation
coefficient (0.0147) is significant at the 5% level. Positive
media tone will induce optimism among investors in the
secondary market. Investors affected by optimism and
herding will over-purchase IPO, leading to strong
performance of new stocks and higher underpricing rates.
The correlation coefficient of negative tone is greater than
the correlation coefficient of positive tone, which indicates
that the same degree of positive tone has less influence on
IPO underpricing than negative tone. The result is in line
with prospect theory, that is, investors are more sensitive
to negative or loss information. Another literature founds
that the impact of negative and positive information on the
securities market is asymmetric, and the impact of negative
information on short-term securities prices is usually
stronger than the same level of positive information [27].
The above results verify Hypothesis 3, that is, the higher
the uncertainty tone content of the media report before the
IPO is, the lower the IPO underpricing rate will be.Page 11 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

In Table 6 and Table7, we examine the mechanism
by which the uncertainty of media tone affects IPO
underpricing. That is, uncertain media tone will reduce
secondary market valuation ΔV̅̅̅̅ and new stock demand
Q2j. The three columns in Table 6 use Un, Neg and Pos
as explanatory variables, and use PE and P as the proxy
variables for ΔV̅̅̅̅. It can be seen from column (1) that both
PE and P are negatively correlated with uncertain media
tone, and both were significant at the 1% level. If media
reports contain more uncertain tone, the uncertainty-averse
investors will demand more compensation for holding the
new shares, thus reducing the valuation of IPOs in the
secondary market. In addition, column (2) shows that PE
and P are negatively correlated with negative media tone,
and their coefficients are significant at the 1% level. In
addition, PE and P are positively correlated with positive
media tone, but the regression coefficients for Pos and P
are not significant. The three columns in Table 7 further
validates the impact of different media tone on the new
stock demand Q2j of the secondary market. Among them,
Un and Neg are negatively correlated with the first-day
turnover rate, which is significant at the 5% and the 1%
level respectively. Pos and first-day turnover rate are
significantly positively correlated at the 10% level. Un
and Neg may reduce investors' risk appetite, stimulate
investors' pessimism and restrain the investor's new stock
demand, which leads to a decline in the volume and
activity of new stocks. Conversely, positive media tone
may spur investor optimism and new stock demand,
resulting in higher activity and trading volume. Page 12 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Table 5. Media tone and IPO underpricing rate
Variables (1) (2) (3)
Un /% -0.0954***( -3.50)
Neg /% -0.0443***( -4.12)
Pos /% 0.0147**(2.22)
Age 0.0157(0.16) 0.0190(0.15) 0.0165(0.12)
Indus -0.0067( -0.61) -0.0060( -0.59) -0.0063( -0.41)
Asset 0.1049***(5.53) 0.1073***(4.80) 0.1863***(5.45)
Debt -0.2886***( -2.85) -0.2965**( -2.54) -0.2609***( -2.74)
Sales -0.0443*( -1.96) -0.0472( -1.63) -0.0429*( -1.92)
ROE 0.4134***(3.01) 0.4659***(2.97) 0.4075***(2.83)
VC -0.0392( -0.02) 0.0356(0.23) 0.0390(0.09)
UW -0.0154( -0.79) -0.0144( -0.84) -0.012 3(-0.64)
Size -0.2227***( -8.58) -0.2370***( -8.09) -0.2509***( -8.38)
MR 2.8144***(7.09) 2.3565***(6.77) 2.6273***(7.44)
Rate -0.3723***( -9.18) -0.3907***( -8.07) -0.2755***( -10.37)
c 0.3406***(9.50) 0.2582***(9.56) 0.4069***(6.70)
Adj R2 0.3895 0.3959 0. 3694

Table 6. Media tone and IPO first-day P/E and P/E ratio
（1） （2） （3）
PE P PE P PE P
Un
/% -0.3711***( –
4.03) -0.2529***( –
3.53)
Neg
/% -0.0806***( –
2.86) -0.0781***( –
2.77)
Pos
/% 0.1329**(2.0
5) 0.0101(1.40)
Age 1.5123(0.48) 0.0388(0.30) 1.7188(0.55) 0.0425(0.32) 1.5188(0.48) 0.0350(0.27)
Indu
s 0.5292 (0.76) 0.0063 (0.22) 0.5588 (0.80) 0.0065
(0.22) 0.5306
(0.76) 0.0054
(0.19) Asse
t 3.9758(1.09) 0.2376(1.57) 3.5309(0.96) 0.2179(1.42) 3.9866(1.09) 0.2348(1.55)
Debt -2.2278**( –
2.27) -0.6825*( –
1.67) -2.1727**( –
2.21) -0.6563( –
1.60) -2.2286**( –
2.27) -0.6773*( –
1.66) Sales -0.0577( -0.02) -0.1600( -1.23) -0.2935( -0.09) -0.1515( –
1.16) -0.0604( –
0.02) -0.1603( –
1.23) ROE 3.7529**(2.43
) 0.0643**(2.18
) 3.7370**(2.44
) 0.0133**(2.0
4) 3.6667**( 2.4
3) 0.0154**(2.0
4) VC -0.9899( -0.36) -0.0454( -0.40) -1.1472( -0.42) -0.0504( –
0.44) -1.0007( –
0.36) -0.0445( –
0.39) UW -0.4489( -0.16) -0.0525( -0.46) -0.3514( -0.13) -0.0501( –
0.44) -0.4519( –
0.16) -0.0537( –
0.47) Size -8.8555**( –
2.53) -0.3442**( –
2.36) -8.6436 **(-
2.47) -0.3311**( –
2.27) -8.8693**( –
2.50) -0.3393**( –
2.32) MR 9.0872***(2.9
1) 3.8754***(2.9
7) 8.5758***(2.7
1) 3.7582***(2.
85) 9.0782***(2.
92) 3.9463***(3.
04) Rate -1.1225***( –
3.32) -0.4581***( –
3.25) -0.9987***( –
2.81) -0.4318***( –
2.91) -1.1199***( –
3.43) -0.4772***( –
3.50) c 5.8236***(4.9
6) 2.4060***(4.9
2) 5.8296***(4.9
7) 2.4061***(4.
92) 5.7894***(4.
16) 2.4043***(4.
14) Adj
R2 0.3059 0.2969 0.2776 0.2976 0. 2859 0. 2663
Table 7. Media tone and IPO first day turnover rate
Variables (1) (2) (3)
Un /% -0.1215**( -2.76)
Neg /% -0.0510***( -3.13)
Pos /% 0.0147*(1.75)
Age -0.0475**( -2.23) -0.0444**( -2.10) -0.0492**( -2.31) Page 13 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

2
Indus 0.0017 (0.36) 0.0020 (0.42) 0.0013 (0.28)
Asset 0.1563***(6.33) 0.1440***(5.83) 0.1557***(6.29)
Debt -0.1849***( -2.78) -0.1687**( -2.56) -0.1825***( -2.74)
Sales -0.0639***( -3.03) -0.0583***( -2.77) -0.0638***( -3.01)
ROE 0.0045 (0.07) 0.0294(0.51) 0.0264 (0.46)
VC -0.0033( -0.18) -0.0002( -0.01) -0.0026( -0.14)
UW -0.0308*( -1.65) -0.0326*( -1.76) -0.0297( -1.59)
Size -0.1506***( -6.34) -0.1429***( -6.06) -0.1490***( -6.26)
MR 1.0348***(4.89) 0.9461***(4.45) 1.0704***(5.07)
Rate -0.2170***( -9.47) -0.1966***( -8.22) -0.2265***( -9.20)
c 0.9300***(11.70) 0.9304***(11.81) 0.8908***(9.43)
Adj R2 0.3227 0.3314 0. 3193
5.2. Robustness Test
We used the following three methods for robustness
test. Firstly, We add some new control variables to the
regression model, including the total number of words in
the media text, the dummy variable of the state-owned
holding company, the dummy variable of the high-tech
enterprise, the total number of IPO in the A-share market
within the three months prior to listing. After the above
treatment and testing again, it was found that the
coefficient values and significance levels of the main
variables did not change much. Secondly, we use relative
calculation methods to measure the explained variables.
After using P to replace PE as a proxy variable for the
IPO's relative valuation in the secondary market, the
relevant conclusions did not produce substantial changes.
The regression results can be seen in Table 4 and Table 6.
Thirdly, we replace the newspaper media data of CNKI
with online media data. In the era of media diversification
and network facilitation, retail investors' attention to online
traditional media has increased significantly. Therefore,
this paper uses Baidu News Search Database to re-measure
the explanatory variables of media information according
to the statistical way in Section 4.2. Finally, the empirical
results of the retest did not materially change.
6. Conclusions
This paper examines the information governance role
of media in the IPO financing process of SMEs in China
from both theoretical and empirical aspects. We construct
an IPO pricing model to describe the micro mechanism of
media signals influencing investors' decisions. Model
analysis finds that IPO underpricing rate is significantly
negatively correlated with the uncertainty of media
information. Then, we take the SMEs listed in Chinese A-
share market as empirical samples, use text analysis to
measure media credibility and media tone, and verify the conclusion of the theoretical model by empirical analysis.
The theoretical and practical implications of this paper are
as follows:
Firstly, our model and empirical results show that the
credibility of media sources influences the valuation and
demand for IPO by retail investors. Investors rely more on
credible media reports when making investment decisions.
Authoritative media can reduce the uncertainty of media
information sources, attract more attention from retail
investors, and strengthen the existing belief of optimistic
investors, so as to increase retail investors' expectation of
IPO valuation and demand enthusiasm. As a result, the
higher the media credibility, the higher the I PO
underpricing rate. This conclusion shows that highly
credible authoritative media reports play an important role
in reducing the uncertainty in the IPO process. Investors
have a natural dependence on authoritative media in the
environment of information overload, and they will give
certainty premium to the companies reported by
authoritative media. Therefore, authoritative financial
media should take the lead in establishing industry self-
discipline mechanism and purifying financial media
environment. We can also establish a credibility ranking
mechanism for authoritative media to reduce the cost for
investors to screen reliable information.
Secondly, we establish a deep learning model for text
analysis using Java programming language, and apply text
mining to analyze and measure Chinese media tone for the
first time. Combined with empirical test and theoretical
analysis, we find that the uncertain tone of media will
increase the decision-making cost of retail investors and
reduce their valuation and demand for IPO. The more
uncertain the media tone, the lower the IPO underpricing
rate. Due to the higher uncertainty of the development
prospects of SMEs, investors usually regard the uncertain
tone in media reports as a risk warning message. The Page 14 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

2
uncertain tone in media reports plays an important role in
suppressing the high degree of underpricing.
Thirdly, our empirical results also prove that the
positive (negative) media tone has a significant impact on
retail investors' sentiment. Positive (negative) media tone
can stimulate (restrain) investor optimism and demand for
new shares, leading to higher (lower) IPO underpricing
rate. The information conveyed by media reports to
investors is usually emotional, and media emotions may be
used to induce investors' mispricing. In reality, some media
report falsely or exaggerate in order to attract investors'
attention or accept the interests of listed companies.
Therefore, supervision departments need to improve the
incentive and punishment mechanism to promote the
objective neutrality of financial media, and take serious
legal sanctions against the "collusion" of media and
companies.
This paper focuses on the analysis of the impact of
media reports on retail investors in the secondary market,
but whether media reports significantly affect the inquiry
behavior of institutional investors in the primary market
still needs to be further studied. The media data used in this
paper are limited to news reports in the database of cnki.
However, in recent years, a large number of online media,
we-media and social media have emerged, which have
profoundly changed the way people obtain information.
Therefore, future studies can further investigate the
influence of other non-traditional media on investors.
References
1. Daily, C.M.; Certo, S.T.; Dalton, D.R.; Roengpitya,
R. IPO underpricing: A meta-analysis and research
synthesis. Entrep. Theory Pract. 2003 , 27, 271 –295.
2. Reilly, F.K.; Hatfield, K. Investor experience with
new stock issues. J. Financ. Anal. 1969 , 25, 73–80.
3. Ritter, J.R.;Welch, I. A review of IPO activity,
pricing, and allocation. J. Financ. 2002 , 57, 1795 –
1828.
4. Easley, D.; O’hara, M. Information and the cost of
capital. J. Financ. 2004 , 59, 1553 –1583.
5. Rock, K. Why new issues are underpriced. J. Financ.
Econ. 1986 , 15, 187 –212.
6. Ritter, J.R. The costs of going public. J. Financ. Econ.
1987 , 19, 269 –281.
7. Li, R.; Liu, W.; Liu, Y.; Tsai, S.-B. IPO Underpricing
after the 2008 Financial Crisis: A Study of the
Chinese Stock Markets. Sustainability. 2018, 10,
2844.
8. Ang, J.S.; Brau, J.C. Firm transparency and the costs
of going public. J. Financ. Res. 2002 , 25, 1–17.
9. Connelly, B.L.; Certo, S.T.; Ireland, R.D.; Reutzel,
C.R. Signaling theory: A review and assessment. J.
Manag. 2011, 37, 39–67. 10. Wang, C.Y.; Wu, J.W. Media tone, investor
sentiment and IPO pricing. J. Financ. Res. 2015 , 9,
174–189.
11. Guldiken, O.; Tupper, C.; Nair, A.; Yu, H. The
impact of media coverage on IPO stock performance.
J. Bus. Res. 2017 , 72, 24–32.
12. You, J.X.; Wu, J. Spiral of silence: Media sentiment
and the asset mispricing. Econ. Res. J. 2012, 7 , 141 –
152.
13. Tetlock, P.C. Giving content to investor sentiment:
The role of media in the stock market. J. Financ. 2007 ,
62, 1139 –1168.
14. Arnold, T.; Fishe, R.P.H.; North, D. The effects of
ambiguous information on initial and subsequent IPO
returns. Financ. Manag. 2010 , 39, 1497 –1519.
15. Janney, J.J.; Folta, T.B. Signaling through private
equity placements and its impact on the valuation of
biotechnology firms. J. Bus. Venturing. 2003 , 18,
361–380.
16. Loughran, T.; Ritter, J. Why has IPO underpricing
changed over time? Financ. Manag. 2004 , 33, 5–37.
17. Akerlof, G.A. The market for “lemons”: Quality
uncertainty and the market mechanism. Uncertain.
Econ. 1978 , 235, 237 –251.
18. Allen, F.; Faulhaber, G.R. Signalling by underpricing
in the IPO market. J. Financ. Econ. 1989 , 23, 303 –
323.
19. Bhattacharya, U.; Galpin, N.; Ray, R.; Yu, X. The
role of the media in the internet IPO bubble. J. Financ.
Quant. Anal. 2009 , 44, 657 –682.
20. Reuer, J.J.; Tong, T.W. Discovering valuable growth
opportunities: An analysis of equity alliances with
IPO firms. Organ. Sci. 2010 , 21, 202 –215.
21. Pollock, T.G.; Rindova, V.P. Media legitimation
effects in themarket for initial public offerings. Acad.
Manag. J. 2003, 46, 631– 642.
22. Frankel, R.; Li, X. Characteristics of a firm's
information environment and the information
asymmetry between insiders and outsiders. J.
Account. Econ. 2004 , 37, 0–259.
23. Dyck, A.; Volchkova, N.; Zingales, L. The corporate
governance role of the media: Evidence from Russia.
J. Financ. 2008, 63, 1093 –1135.
24. Xiong, Y.; Li, C.Q.; Wei, Z.H. Media coverage and
IPO pricing efficiency: Based on information
asymmetry and behavioral finance perspective. J.
World Econ. 2014 , 5, 135 –160.
25. Wang, C.Y.; Wu, J.W.; Sun, Y.M.; Gan, S.L. Media
information management behavior and IPO pricing
efficiency of the company. Manag. World. 2015 , 1,
118–128.
26. Quan, X.F.; Yin, H.Y.; Wu, H.J. Analysis of
Asymmetric Effects of Media Coverage on IPO Page 15 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

3
Stock Price: Evidence from Chinese Growth
Enterprise Market. Account. Res. 2015 , 6, 56–63.
27. Cook, D.O.; Kieschnick, R.; Ness, R.A.V. On the
marketing of IPOs. J. Financ. Econ. 2006, 82, 35–61.
28. Liu, L.X.; Sherman, A.E.; Zhang, Y. The long-run
role of the media: Evidence from initial public
offerings. Manag. Sci. 2014, 60, 1945 –1964.
29. Shao, X.J.; He, M.Y.; Jiang, P.; Xue, Y.; Liao, J.C.
The management of media public relation, investor
sentiment and security offerings. J. Financ. Res. 2015 ,
9, 190 –206.
30. Chen, P.C.; Zhou, X.H. Institutional investor private
information, retail investor sentiment and IPO first-
day return. Chinese. J. Manag. Sci. 2016, 24, 37–44.
31. Reuer, J.J.; Tong, T.W. Discovering valuable growth
opportunities: An analysis of equity alliances with
IPO firms. Organ. Sci. 2010 , 21, 202 –215.
32. Bednar, M.K.; Boivie, S.; Prince, N.R. Burr under the
saddle: How media coverage influences strategic
change. Organ. Sci. 2013, 24, 910 –925.
33. Tan, D. Making the news: Heterogeneous media
coverage and corporate litigation. Strat. Manag. J.
2016 , 37, 1341 –1353.
34. Pollock, T.G.; Rindova, V.P.; Maggitti, P.G. Market
watch: Information and availability cascades among
the media and investors in the US IPO market. Acad.
Manag. J. 2008, 51, 335 –358.
35. Park, H.D.; Patel, P.C. How does ambiguity influence
IPO underpricing? The role of the signalling
environment. J. Manag. Stud. 2015 , 52, 796 –818.
36. Huang, J.; Guo, Z.R. Efficiency of news media report
and capital market pricing: An analysis based on
synchronism of stock price. Manag. World. 2014 , 5,
121–130.
37. Wang, M.Z.; Li, D. Media public relations in capital
market: Empirical evidence from IPO of Chinese
enterprises. Manag. World. 2016, 7, 121 –136.
38. Loughran, T.; Mcdonald, B. IPO first-day returns,
offer price revisions, volatility, and form s-1
language. J. Financ. Econ. 2013, 109, 307 –326.

Page 16 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Author Bios
Fei Zhang is a Prof. at School of Economics and Business Administration, Chongqing
University. He got the PHD at Chongqing University in China.
Xiao-Hua Zhou is a Prof. at School of Economics and Business Administration,
Chongqing University. He got the PHD at Chongqing University in China.
Jiafu Su is a Prof. at Chongqing Key Laboratory of Electronic Commerce & Supply
Chain System, Chongqing Technology and Business University. He got the PHD at
Chongqing University in China.
Sang-Bing Tsai is a Prof. at Zhongshan Institute, University of Electronic Science and
Technology of China. He is Business Management PhD. He got the PHD at Nankai
University in China.
Yu-Ming Zhai is a Prof. at School of Economics and Management, Shanghai Institute
of Technology. He got the PHD at Chongqing University in China.Page 17 of 17
For Review OnlyIEEE Access
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60