An Empirical Study of the Bidding Behave under the Bookbuilding
Mechanism
Jian Zhang
1
and Li Houjian
2
1
College of Economics of Sichuan Agricultural University, Chengdu, China
2
College of Management SICAU, Chengdu, China
Keywords: Inquiry Object, Winner’s Curse, Institutional Reform.
Abstract: In this paper, we set up the difference of average biddings of inquiry object with information superiority and
inquiry object with disadvantage by analysing 45630 detailed biddings of 443 companies from 2010 to 2012
in A share of our country and test the Winner Curse in three-factor model of A share market in the inquiry
system.
1 INTRODUCTION
It is wildly accepted that the participation of
institutional investor in securities market can improve
market information efficiency and thus improve asset
allocative efficiency and enhance the stability of
financial market. Since 2006, the pricing mechanism
of new share has been carried out that IPO of China’s
A share enquires institutional investor to determine
price range and issuer and underwriter get initial
bidding distribution from Road show. Obviously, the
biddings of institutional investors participating
inquiry of new stock issue have significant impact on
new stock pricing.
The most influencing literature on inquiry
mechanism is Benveniste and Spindt (Benveniste and
Spindt, 1989), Benveniste and Wilhelm (Benveniste
and Wilhelm, 1990). They believe underwriter can
get access to the evaluation of new stock and demand
information from informed investor by inquiry
mechanism in financial markets with asymmetric
information. Cornelli and Goldreich (Cornelli and
Goldreich, 2001) and Ljungqvist and Wilhelm
(Ljungqvist and Wilhelm, 2002) show that when
underwriter has oversubscribed shares distribution
right, free rider phenomenon during inquiry will drop
and issuance pricing efficiency will improve if more
equity is allocated to native IPO investors and
institutional investors frequently involving IPO
inquiry.
2 HYPOTHESIS
This paper assumes that issuer can’t perfectly forecast
the market price of new stock and investor has
information on new stock price (Winner Curse of
Rock, 1986). When informed that one new stock is of
investment value, investors with information
superiority will improve declaration value and
number to squeeze the one with information
disadvantage out the issuing market. The specificity
of Chinese stock market result the simultaneous
processing of online and offline subscription and
institutional investor can only choose one method.
Moreover, offline subscription will undergo inquiry
and the inquiry and subscription amount impact
issuing price and final allocated number a lot. Online
retail investors and the rest institutional investors
subscribe new stock on the offline inquiry, which
won’t have real impact on issue price. This differs
with the hypothesis of asymmetric information of
institutional investor and retail investor (Hanley and
Wilhelm, 1995). In this paper, we think it a
reasonable assumption that information divide also
exists among institutional investors. Firstly, different
institutional investors hold different message on new
stock. Secondly, investment experience and
background can also influence information capacity.
Thirdly, the relationship with principal underwriter
determines. In this allocation system, institutional
investors with information superiority improve
offline subscription price to squeeze institutional
investors with information disadvantage out of
431
Zhang J. and Houjian L.
An Empirical Study of the Bidding Behave under the Bookbuilding Mechanism.
DOI: 10.5220/0006027904310435
In Proceedings of the Information Science and Management Engineering III (ISME 2015), pages 431-435
ISBN: 978-989-758-163-2
Copyright
c
 2015 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
431

effective price range so as to improve lot winning
rate, while institutional investors with information
disadvantage judge the investment value and price at
the average expectation. To sum up, we bring out the
research hypothesis.
H: the difference of average biddings of
institutional investors with information superiority
and information disadvantage is in direction
proportion to offline over-subscription ratio.
The common way to divide information among
investors in empirical are (1) Institutional investors
are with information superiority while retail investors
with information disadvantage. (2) Domestic
investors are with information superiority while
foreign investors with information disadvantage.
These ways can’t test winner curse when allotment of
shares differs among stocks with different
underpricing rate. What’s more, Chinese inquiry
objects are highly concentrated in territory and
proportion of foreign investors QFII is super low.
Thus the investor information can’t be distinguished
according to the way in classical documents.
3 DATA AND MODLE DESIGN
The data are listed companies with IPO of A share
from November 2010 to October 2012 with deletion
of individual major financial insurance companies
and few individual investors’ biddings. The final
research sample contains 463 listed companies and
45630 biddings and subscriptions of institutional
investors.
We use total amount of subscription as
measurement index of institutional investor
participating subscription. In the sample, total
subscriptions of fund company, security company,
insurance company, safe company, finance company,
recommended institutional investors and QFII
account for separately44.75%, 25.44%, 12.34%,
9.11%, 4.62%, 3.55% and 0.19%. Recommended
institutional investors account for 47.44% of total
institutional investors in number but only 3.55% in
total subscription total amount.
When divided by territory, there are 85
institutional investors in Beijing, about 19.41% of the
total number, 106 in shanghai accounting for 23.93%,
92 in Guangdong accounting for 20.77% and 160 in
the rest areas of the country, about30.62%of the total
number. The paid-in subscription of institutional
investors in Beijing, Shanghai, Guangzhou and the
rest areas separately accounts for 21.75%, 32.49%,
21.49% and 24.26%. Offline institutional investors
are concentrated in both category and territory.
We differ institutional investors by involvement
level of inquiry object and whether is underwriter. We
rank inquiry object by subscription amount and
selection the top 9 as information superiority
institutional investors and the last 30% as information
disadvantage institutional investors. There are total
398 latter inquiry objects, accounting for 30% of the
total subscription amount. Moreover, we regard
recommended institutional investors as information
superiority investors and others as information
disadvantage investors. The two explaining variables
diffp1 and diffp2 are constructed by computing the
difference of average biddings of information
superiority and disadvantage institutional investors.
2. We select ln offline subscription multiple as
explaining variable of offline subscription popular
degree.
3. company characteristics in the research sample is
controlled by net margin per share, issuing scale,
total assets one year before issuing, asset-liability
ratio one year before issuing, company age. The
impact of intermediary to IPO pricing is
controlled by introducing underwriter fame, the
dummy variable.
The research hypothesis H describes the relation of
bidding differences of good and bad institutional
investors with IPO underpricing rate. We set up the
following model with Underpricing and
Dumunderpricing as the explained variables and
Inolmeanp and recommendmeanp as explaining
variables.
dif
f
p1=ß
0
+ß
1
lnoffline+ß
2
Size+ß
3
Plev+ß
4
Age+ß
5
neps+
ß
6
underwriter+ß
7
sentiment+ß
8
Pprice+ß
9
MSM
(1)
diffp2=ß
0
+ß
1
lnoffline+ß
2
Size+ß
3
Plev+ß
4
Age+ß
5
neps+
ß
6
underwriter+ß
7
sentiment+ß
8
Pprice+ß
9
MSM
(2)
If ß
1
is significantly positive in model (1) and (2), the
research hypothesis H is stated, or it is not stated.
4 EMPIRICAL RESULTS
Firstly, rank the offline over-subscription ratio
according to categories with highest 30% company as
Group high and lowest 30% as Group low. And then
conduct mean test to the high subscription group and
low subscription group: diffp1and diffp2. From the
mean test result shown in Table 1, we can see that the
average bidding difference of inquiry objects diffp1,
when distinguishing information according the
proportion of single institution subscription in total
ISME 2015 - Information Science and Management Engineering III
432
ISME 2015 - International Conference on Information System and Management Engineering
432

Table 1: Inter-block Mean Value Analysis when Grouping as Under-pricing Rate.
GEM diffp1 GEM diffp2
Obs mean std. Obs mean std.
High 63 0.513 2.19 High 60 -0.56 2.92
Low 63 -0.58 2.32 Low 60 0.47 2.78
diff 63 1.097 3.12 diff 60 -1.03 4.07
mean(diff) = mean(high - low) mean(diff) = mean(high - low)
Ha: mean(diff) > 0 Pr(T > t) = 0.004 Ha: mean(diff) > 0 Pr(T > t) = 0.004
SME diffp1 SME diffp2
Obs mean std. Obs mean std.
High 57 0.737 2.43 High 56 -0.07 2.07
Low 57 -0.63 3.22 Low 56 0.55 2.7
diff 57 1.37 3.92 diff 56 -0.62 3.17
mean(diff) = mean(high - low) mean(diff)=mean(high-low)
Ha:mean(diff) > 0 Pr(T > t) = 0.004 Ha:mean(diff) > 0 Pr(T > t) = 0.004
subscription, locates at GEM and SME, and the
average of high subscription group is significantly
higher than the low one.
Table2 shows the regression result of
Hypothesis 1. As for GEM, estimated coefficients
of lnoffline in the models of diffp1 with and without
control variables are separately 0.341and 0.358 and
both are significant at the level 10% and 5%. This
explains that diffp1 is in positive correlation with
lnoffline. However, estimated coefficients of
lnoffline in the models of diffp2 with and without
control variables are separately -0.209 and -0.2569
and are significant at the level 10%. Meanwhile, as
for SME, estimated coefficients of lnoffline in the
models of diffp1 with and without control variables
are separately0.744 and0.529 and are significant at
the level 1% and 5%. This explains that diffp1 is in
positive correlation with lnoffline. However,
estimated coefficients of lnoffline in the models of
diffp2 with and without control variables are
separately-0.446 and -0.526 and are non-significant
at the level 10%.
The above regression analysis shows that diffp1
is positive correlated with the offline over-
subscription ratio of new shares in the research
sample. This means that recommended inquiry
institution can’t be investors with information
superiority. The above analysis support the
hypothesis H1: average bidding difference of
information superiority and information
disadvantage institutional investors is in direct
proportion to offline over-subscription ratios.
Therefore, we test the winner curse hypothesis by
bidding data of inquiry object in Chinese A share.
5 CONCLUSIONS
We select 463 A share IPO companies form
November 2011 to October 2012 as the sample and
analyze offline bidding characteristics of inquiry
object. The result shows that when dividing
information quality by new share inquiry
involvement level, the difference of average
biddings of institutional investors with information
superiority and information disadvantage is in
direction proportion to offline over-subscription
ratio. Therefore the Winner Curse of Rock is
supported in inquiry object bidding of Chinese A
share. Thus, further reasonable adjustment on new
stock issuing mechanism is needed. Autonomous
placing right of underwriter in offline issuing and
detailed release of placing situation should be
focused. Enhancing direct financing supply and
punishment on false disclosure, improving
An Empirical Study of the Bidding Behave under the Bookbuilding Mechanism
433
An Empirical Study of the Bidding Behave under the Bookbuilding Mechanism
433

responsibility of related financial intermediaries
and market pricing efficiency of new stock are
needed.
REFERENCES
Benveniste, L M., Paul A. Spindt,How Investment Banks
Determine the Offer Price and Allocation of New
Issues. Journal of Financial Economics,
24(5):pp.343-362, 1989.
Benveniste, L. M., Wilhelm, W. J., A Comparative Ana-
lysis of IPO Proceeds Under alternative Regulatory
Environments. Journal al of Financial Economics,
28(1-2): pp.173-207, 1990.
Cornelli Francesca, David Goldreich, Bookbuilding and
strategic allocation. Journal of Finance,
56(6):pp.2337-2369, 2001
Ljungqvist, A., Wilhelm, W.J., Does prospect theory
explain IPO market behavior? Journal of Finance.
60(4): pp.1759–1790, 2005
Rock,K, Why new issues are underpriced. Journal of
Financial Economics, 15(1):pp187-212, 1986.
Hanley K, W and W. J. Wilhelm, Evidence on the
strategic allocation of initial public offerings. Journal
of Financial Economics, 37(2):pp.239-257, 1995.
Table 2: Regression Result of Hypothesis.
Start-up Board
Uncontrolled
variable
Controlled variable Uncontrolled variable Controlled variable
Explained
variable
diffp1 diffp1 diffp2 diffp2
Parameter Coef. t value Coef. t value Coef. t value Coef. t value
Constant term 0.968
*
-1.83 -3.82 -0.82 0.634 0.89 4.098 0.69
lnoffline 0.341
*
1.93 0.358
**
1.97 -0.209 -0.84 -0.256 -1.15
neps -0.0165 -0.74 0.02 0.8
sentiment 137.13
**
2.04 -128.19
*
-1.69
size 0.11 0.24 0.005 0.01
plev -0.014 -1.37 -0.007 -0.55
age 0.113 0.53 -0.169 -0.57
underwriter 0.794
***
2.49 -1.059
**
-2.57
Pprice 0.0485 1.43 -0.069
*
-1.71
Sample value 212 212 202 202
Adjusted R2 value 0.213 0.3692 0.1012 0.3674
F（Wald） value
3.87 3.58 2.55 2.84
VIF 1.21 1.23
ISME 2015 - Information Science and Management Engineering III
434
ISME 2015 - International Conference on Information System and Management Engineering
434

Table 2: Regression Result of Hypothesis (cont.).
SME Board
Uncontrolled
variable
Controlled variable Uncontrolled variable Controlled variable
Explained
variable
diffp1 diffp1 diffp2 diffp2
Parameter Coef. t value Coef. t value Coef. t value Coef. t value
Constant term -1.85
***
-3.25 -4.23 -0.78 1.521
***
3.26 4.089 0.72
lnoffline 0.74
***
3.41 0.529
**
2.33 -0.446 -1.75 -0.526 -1.38
neps -1.375 -1.12 -0.343 -0.47
sentiment 0.05 1.64 -0.046
*
-1.89
size 0.214 0.41 -0.179 -0.33
plev -0.004 -0.24 -0.034 -0.14
age 0.564
**
2.16 -0.945
**
-2.44
underwriter 0.273 0.63 -0.945
**
-2.44
Pprice -0.039 -0.99 0.008 0.17
Sample value 190 190 188 188
Adjusted R2 value 0.113 0.3485 0.1306 0.2875
F（Wald） value
11.63 20.68 2.55 2.84
VIF 1.47 1.46
Note: t value is got by the standard deviation of Whiterobust. *** is significant at the level of 1%;** is significant at the level of 5%;* is
significant at the level of 10%. VIF is variance inflation factor. Coef. means coefficients
An Empirical Study of the Bidding Behave under the Bookbuilding Mechanism
435
An Empirical Study of the Bidding Behave under the Bookbuilding Mechanism
435