An Empirical Test on Initial Public Offering Stock Underpricing in China A-shares Market

(1)

An Empirical Test on Initial Public Offering

Stock Underpricing in China A-shares Market

Qin Meng

Master Thesis University of Groningen Faculty of Economics and Business

Master of Science, Business Administration of Finance

Student Number: S2079488

Email: mengqina@hotmail.com

Supervisor:

Dr. Auke. Plantinga

(2)

2

Abstract

The purpose of this thesis is to examine the “Multiple Hysteresis Effect” (this terminology is proposed in Hu and Guo, 2009) on initial public offering (IPO) underpricing in China’s A-shares market. I study initial returns of 505 IPOs issued between 10th July 2009 and February 25th 2011 in A-shares markets in China, about three years after the Chinese Equity Division reform. I select IPOs issued within one certain week as a portfolio and finally I get 151 portfolios as my samples.

I find that the average IPO underpricing for A-shares is 44.16%, but it is higher in Shenzhen Security Exchange (SZSE) than in Shanghai Security Exchange (SHSE), which are 45.69% and 30.35%, respectively.

I certify the existence of the hypothesis of “Multiple Hysteresis Effect” on IPO underpricing both by theoretical and empirical research. With a multiple hysteresis model, I find that the present IPO initial return is highly correlated with the latest two periods’ lagged IPO returns. In addition, with a multi-factor regression analysis, I find that IPO underpricing is related to the IPO size, last year’s EPS, IPO P/E ratio and turnover ratio on the first trading day. This finding shows that the primary factors in the framework of speculation are “Hysteresis” factors.

Through theoretical and empirical research, the paper gets the conclusion that the market highly speculation is the main reason of IPO underpricing.

JEL Classification: G11; G14; G15.

(3)

3

1. Introduction

An initial public offerings (IPOs) is when a company (called the issuer) issues common stocks to the public for the first time. IPO underpricing is a widespread phenomenon in the world. Almost all the stock markets on the world have different levels of IPO underpricing. A modest level of underpricing in the IPO market is acceptable as a means of attracting sufficient investors to take part in buying the newly issued stocks. But if one IPO is underpriced beyond the normal level, this issuing company will suffer an unnecessary loss. Therefore, it is important for the issuer to find a balanced level of underpricing.

A lot of researches on the IPO underpricing varies from one country to another. As Su and Fleisher (1997) inferred, underpricing in Japan is 12%, 15% in U.K., 16% in U.S., 60% in Korea, 78% in Brazil and 948.59% in China between December 1986 and January 1996. These figures show that IPOs in developed countries are generally much lower underpriced than in developing countries. Compared with other countries and areas, the magnitude of underpricing in China’s IPO stock market is even more serious so that it has attracted quite some attention.

(5)

5

In order to keep the stock market stable and protect the rights of public investors, from June 5, 2006, the China Securities Regulatory Commission (CSRC) promotes Full Circulation. It is a reform that all A-shares1 of the public companies should be fully circulated in the market to be traded without any constraints, including the non-circulation share. At first, most researchers expected that the underpricing of IPOs would be diminished (see Liu and Xiong, 2005). As of July 1, 2011, the reform has been mainly completed. However, estimation results show that, after the reform, the IPO underpricing does not change radically. Why is this fact still severe in China’s stock market after the reform of Chinese Equity Division? Is there any other factors leading to the irregular evidence? In attempting to explain the puzzle, several academic researchers have formulated different models. However, even within one market, no clear explanation can suffice to account for the apparent underpricing of new issues in China’s A-shares markets. Therefore, an understanding of the characteristics and performance of the Chinese IPO market would be of great value to investors and scholars at home and abroad.

In this paper, firstly I review previous researches with regard to IPO underpricing. Then I examine the IPO underpricing using more recent sub-period data when China’s A-shares have been fully circulated. Specifically, I focus on both SHSE and SZSE from October 7, 2009 to February 25, 2011. After some tests, I find that the average level of underpricing in A-Share market is 44.16%. Then I attempt to explain why the level of IPO underpricing in China is still high.

Based on the “Conformity behavior theory” as the main theoretical basis (see Welch, 1992), I consider that individuals tend to make their behavior conform to the behavior of similar previous individuals, so IPOs’ former performances will affect the newly issued IPOs’ performances because of the speculative factors of the potential investors in China’s stock market. Hu and Guo (2009) also mentioned that this kind of conformity behavior has multiple lagged effects, which indicates that the present IPO

1

(6)

6

return is affected by multiple lagged IPO returns previous instead of a single lagged IPO return. They called this kind of effect as “Multiple Hysteresis Effect”.

The dominant contribution of this paper is that I test the existence of the “Multiple Hysteresis Effect” hypothesis on IPO underpricing via examining and analyzing the mentality and behavior characteristics of investors in China’s IPO markets. Then I test this hypothesis during the period when A-shares have been in full circulation. Moreover, as the thesis studies the IPO underpricing phenomena in the framework of speculation, I add five general factors that might explain the IPO underpricing. These factors are the size of IPO, previous year’s earnings-per-share, price-to-earnings ratio of the first-trading day, turnover rate of the first-trading day and the rate of return on common stockholder’s equity by the fiscal year end before the IPO.

(7)

7

2. Literature Review

2.1 Phenomena of IPO underpricing in China’s stock market

The severe underpricing problem is a unique feature of China’s primary market. Mok and Hui (1998) found an underpricing of 289% for a sample of 101 Shanghai IPOs listed and in A-share market from May 1992 to December 1993. Su and Fleisher (1999) selected 308 domestic A-shares that were newly issued between 1987 and 1995. They found an underpricing level as high as 948.6% before January 1, 1996. A more updated research by Tian (2003) reported an average initial returns is 267% with the median as 131% for the IPOs from 1991 to 2000 in China’s market, which, as we can see in the below graph 1, is much higher than other countries around the world. These reported underpricing levels in the Chinese market are much higher than the average level of 60% in the emerging markets (Jenkinson and Ljungqvist, 2001). However, despite many studies on the IPO underpricing, few studies have been done to investigate the reasons of IPO underpricing in China.

Graph 1

(8)

8

2.2 Different Studies on Reasons of China’s IPO Underpricing

Classical explanations for IPO underpricing are asymmetric information (Rock, 1986), institutional explanations (Jenkinson and Ljungqvist, 2001) and ownership and control (Jenkinson and Ljungqvist, 2001). In these theories, there are some classical models and hypothesis like the winner’s curse model (Rock, 1986), ex ante uncertainty hypothesis (Ritter, 1984; Beatty and Ritter, 1986), signaling model (Allen and Faulhaber, 1989; Grinblatt and Hwang, 1989; Welch, 1989, 1996) and manager’s strategic underpricing explanation (Aggarwal, Krigman and Womack, 2001).

2.2.1 Asymmetric Information

Rock (1986) suggests that the existence of asymmetric information among different investors is able to explain IPO underpricing. He proposed a “Winner’s Curse” hypothesis. This hypothesis showed that, if a group of investors has private information on the IPO, the informed investors would put money into such shares and crowd out the other investors; if bad issues were offered, these investors would withdraw money from the market. Because uninformed investors know that they are in a disadvantageous situation of information, they require higher returns as compensation. As a result, the offering firm must price the shares at a discount in order to make sure that they can attract more uninformed investors to purchase the issue. Later, Wu (2001) reports that the subscription rate in China is positively related to IPO underpricing in support of the winner's curse model.

Mok and Hui (1998) tested the pricing of IPOs in the early years of China’s stock market before 1993, while Su & Fleisher (1999) and Allen & Faulhaber (1989) focused on the signaling model to explain the IPO underpricing in China. They all considered that asymmetric information is the main reason of underpricing. Yu and Tse (2006) showed that the winner’s curse hypothesis is the main reason for the large IPO underpricing in China instead of the signaling hypothesis and manager’s strategic hypothesis.

(9)

9

limit the underpricing. As a result, they concluded that the information asymmetry theory is doubtful to explain the high level of underpricing.

2.2.2 Ex ante Uncertainty Hypothesis

Among those tested classical models, Mok and Hui (1998) tested the ex ante uncertainty hypothesis, but they tested only one proxy of ex ante uncertainty, i.e. the inverse of new funds raised. Later, Yu (2003) used three proxies: the standard deviation of aftermarket returns, the offer size and the age of firms, to examine the ex

ante uncertainty hypothesis. Their result all shows that investors’ high ex ante

uncertainty about the firm's value is one of the main reasons for the high IPO underpricing in China.

2.2.3 Signaling Model

Yu (2003) also examined the signaling model. They tested eight key empirical implications (IPO size, ownership, age, total net asset after the issue, P/E ratio, days elapsed between IPO date and the first trading date, beforemarket return and aftermarket return ) of the signaling model, some of which have been examined in Su and Fleisher (1999), but the methodology adopted and the conclusion made are different. Using data from November 1995 to December 1998, their results showed that the signaling hypothesis does not hold on the Chinese market during the sample period.

2.2.4 Informational Cascade Theory

Welch (1992) proposes an “informational cascades” theory to explain why IPOs are on average severely underpriced. The values of the new securities issued are highly uncertain to individual market participants who cannot estimate the value of the firm with high precision. Therefore, when IPO shares are sold sequentially, later potential investors may ignore their private information and conform previous investors' behavior. If earlier IPO had a high initial return, a prospective investor can interpret a successful initial sales effort to imply that earlier investors had favorable information about the offering, in turn giving him an additional incentive to invest. Hence, we can see that investors’ approaches after some time can infer information from investors’ previous approaches.

(10)

10

(11)

11

3. Research Data & Methodology Description

3.1 The Basic Theory Support of the Multiple Hysteresis Effect

Hypothesis 1: “The performance of a newly issued IPO is affected by previous IPOs; returns that are more recent have a stronger impact on the present return; the returns of previous times of IPO are positively related to IPO underpricing.” (Welch, 1992)

The main aspect of this thesis is analyzing the speculative behavior of the individual investors on the IPO underpricing, which is similar to the “Conformity behavior theory” in Welch (1992). As mentioned in such article, investors’ decisions after some time can be affected by the investors’ decisions earlier. When IPO stocks are sold continuously, subsequent potential investors can learn from the buying behavior of investors prior. It is obvious that the percentage of individual investors has become larger in China’s stock market. Unlike the large financial institutions, most of these investors are not able to get enough and accurate information of the newly issued firms, so valuating the new securities is highly uncertain to individual market participants. As a result, a potential individual investor can copy successful purchasing behavior from previous investors, implying that earlier investors had favorable information about the offering which giving later investors an additional incentive to invest. This incentive can lead them to ignore their private information and mimic earlier investors’ behavior.

Through the analysis of “informational cascades” in Welch (1992), I think that the “Multiple Hysteresis Effect” proposed by Hu and Guo (2009) should exist in China’s IPO market and this effect is caused by the “Conformity behavior theory”. The underpricing of stocks issued in a certain time T will be affected by the underpricing of firms issued two previous periods. In the article, I will explain the theoretical support of the hypotheses in detail.

3.2 The Theoretical Explanation of the Hysteresis Effect

Some of the following theoretical explanation of “Multiple Hysteresis Effect” is cited from Hu and Guo (2009).

Firstly, we can assume that there is a company A issuing stock at time T0. The

(12)

12

which affect the value of the IPO firm A and G(x) is a function of firm value evaluated by investors. Theoretically, the true price of the stock P should be a positive function f (*) of the intrinsic value of the company, so we can indicate that the true stock price P of the company should be:

P = f (θ 0) = f [G (R0)]. (3.1)

Because different investors have different estimations with respect to the same firm, we assume a set below so that all the different values (θ *0) are belonging to it:

θ *0 [Gmin (R0), Gmax (R0)] (3.2)

Where Gmin (R0) refers to the minimum value estimated by the investors in the

market and Gmax (R0) refers to the maximum value. Correspondingly, different

investors estimate the issued price P* should also belong to a set like this:

P* {[ f [ Gmin (R0) ], f [ Gmax (R0) ] } (3.3)

Because that the underwriter wants to attract less informed investors to buy the IPO stock, they are more likely to make the IPO issuing price closer to the lower price evaluated by investors. We now assume that the reality IPO issuing price P** is set to the minimum price the investors estimated, that is:

P** = f [G min (R0) ].

As a result, the return of the first-trading day r0 can be represented as:

r0 = (P - P**) / P** = f [G (R0)] / f [ Gmin (R0) ] – 1 (3.4)

Since f [ G (R0) ] > f [ Gmin (R0) ], r0 > 0.

Secondly, we assume that at time T1, there is also a company B issuing stocks on

the market. Just before the period that company B is issued, the potential investors will notice that, in the last period T0, the return r0 of the initial day is higher than zero.

Then they will predict that in this period T1, the newly issued stock may also have a

potential excess return ζe1 on the first-trading day, and then they will consider ζe1 as a

new factor to estimate the value of firm B. Now the set of different values from all investors can be written as:

θ *

1 [ Gmin (R1 , ζe1), Gmax (R1 , ζe1) ] (3.5)

Similarly, the return on the first-trading day of T1 can be represented as:

r1 = f [G (R1)] / f [Gmin (R1, ζe1) ] – 1 (3.6)

Where the potential excess return ζe1 is a function only relative to the return of

the initial day of last period T0, which is:

ζe

1 = h (r0) (3.7)

(13)

13

company to issue, and its initial return rn can be expressed as:

rn = f [G(Rn)] / f [Gmin (Rn , ζen)] – 1, and ζen = h (r0, r1, r2 … rn-1). (3.8)

Thus it can be seen that the IPO price is not only affected by the own factors like intrinsic value of such company, but also affected by the IPO prices of the previous companies. As a conclusion, I think the hypotheses that the “Multiple Hysteresis Effect” should exist in China’s IPO market and this effect is caused by the “Conformity behavior theory” are both reasonable.

3.3 Empirical tests of the Hysteresis Effect 3.3.1 Description of data

The data are retrieved from the Datastream, the IPO companies’ financial statements published on East Money (www.eastmoney.com), Yahoo Finance (finance.yahoo.com) as well as the website of Shanghai Security Exchange (www.sse.com.cn) and Shenzhen Security Exchange (www.szse.cn). I selected those firms issued in both Shanghai Security Exchange (SHSE) and Shenzhen Security Exchange (SZSE) in A-shares markets in China between July 2009 and February 2011. There are 462 A-shares in SZSE and 43 A-shares in SHSE, and the total amount is 505 companies.

As mentioned before, A-shares refer to the ordinary domestic share designated only for private Chinese citizens and they are traded on SHSE and SZSE. In China, this kind of shares is CNY (Chinese Yuan Renminbi) dominated and only can be bought and sold by individual and legal persons within the mainland China.

There is a little difference between SHSE and SZSE. SHSE is fully committed to the goal of large and medium sized state-owned enterprises, while SZSE is committed to its mission to develop China’s multi-lever capital market system and it gives full support to development in small and medium businesses. Otherwise, these two exchanges have the same conditions and requirements of listed companies and investors, in terms of commission, administration fee, opening and closing time, trade cost, EST.

(14)

14

than one firm being issued on the markets at the same time, so I choose IPOs data within one certain week as a portfolio, then I calculate the average return of each portfolio. Finally, my sample data is 151 portfolios.

3.3.2 Methodology

The initial return on the first public day is:

IR = (P1-P0) / P0 (3.9)

Where: IR is the initial return on the first-trading day, P0 is the IPO offering price on the issuing day,

P1 is the closing price on the initial day.

I use a multiple regression model to study the lagged initial returns on IPO underpricing, in order to test the existence of the hypothesis that is called “multiple hysteresis effect” on the underpricing. Here based on this hypothesis, I think there would be a strong impact on the present IPO initial return from recent three-lagged variable, so the model could be set as follows:

IR𝑖 = c + α𝑖−1IR𝑖−1+ α𝑖−2IR𝑖−2+ α𝑖−3IR𝑖−3 + 𝜇𝑖 (3.10)

In addition, from the hypothesis of hysteresis effect, the return would have an autoregression phenomenon. At the same time, if the variances of errors also change over time, rather than systematically with one of explanatory variable, this refers to another effect, known as “(G) ARCH effect”. ARCH (Autoregressive Conditional Heteroscedasticity) model and the GARCH (General Autoregressive Conditional Heteroscedasticity) model both allow variances depending on its own lags and these approaches are now widely used in researching the effects based on high frequency time series data. As we know, conditional heteroscedasticity will bias the estimate results of OLS model, so it is required to check the existence of (G) ARCH effect in order to eliminate this kind of errors. Hu and Guo (2009) mentioned that IPO returns show a (G) ARCH effect, so I will test if there is (G) ARCH effect. Then based on this result, I will decide which model (either OLS or (G) ARCH) would be adopted.

3.4 Multi-factor Analysis of IPO Underpricing

(15)

15

Although the influence from the “Multiple Hysteresis Effect” is substantial, investors also consider other factors. This paper also analyzes these various factors comprehensively.

Su and Fleisher (1998) mentioned that any attempt to explain the positive IPO return should be framed in three aspects: the determination of IPO price, the determination of the size of initial sale and other factors affecting investors' demand for shares. In addition, as this thesis specially focuses on China’s stock market, variables characteristic of China are also taken into account. Therefore, I selected following factors to explain cross-sectional data in IPO initial returns. They are the logarithm of the average IPO size (LSIZE), the average previous year’s EPS (PEPS), the average IPO P/E ratio of the first-trading day (PER), the logarithm of the average turnover rate of the first-trading day (LTR) and the logarithm of the average return on common stockholders’ equity by the fiscal year end before the IPO (LPROE). Finally, in order to know the possible influence of markets on IPO initial underpricing, I add two A-share market returns to my model. They are market returns of Shenzhen Stock Exchange A-shares Index (SZMKTR) and market returns of Shanghai Stock Exchange A-share Index (SHMKTR)

Therefore, the multi-factor model equation is like this: IR_𝑖 = C + 𝛼_𝑖−1IR_𝑖−1 + 𝛼_𝑖−2IR_𝑖−2+ 𝛼_𝑖−3IR_𝑖−3 +𝛽₁LSIZE + 𝛽₂PERS + 𝛽₃PER + 𝛽₄LTR

+𝛽5LPROE + 𝛽6SZMKTR + 𝛽7SHMKTR + 𝜇𝑖 (3.11)

 LSIZE

Hypothesis 2: The offering size of the firm is negatively related to IPO underpricing

LSIZE refers to the natural logarithm of the average IPO issued size measured in CNY (Chinese Yuan Renminbi). The IPO issued size is closely related to the IPO underpricing and it is an important factor in IPO pricing efficiency measured either Money Left on the Table2 or short-term performance.

In 1999, Su and Fleisher found that the size of initial offering has a negative

2

“Money Left on the Table” is reflective of first day performance of the IPO. It is defined as the IPO initial return multiples issued size. In other words, it is the first-day profit received by investors who were allocated shares at the offering price. It represents a wealth transfer from the shareholders of the issuing firm to these investors.

(16)

16

correlation with the IPO initial return. Ranjan and Madhusoodanan (2004) also concluded that underpricing is more severe in the case of smaller issue sizes. Until now, many researchers have published similar results worldwide. As IPO size has been regarded as a proxy of the firm’s risk, and then the size could be measured through the amount of funds corresponding to the issue, so I think that a relatively smaller issue size will be more underpriced than a larger one. In other words, a negative relationship exists here. In order to decrease the range of the variable, here I run a regression with the natural logarithm of the average IPO size of all portfolios.

 PEPS

Hypothesis 3: The EPS for the previous year of IPOs is negatively related to IPO underpricing.

PEPS is the average EPS (earnings-per-share) of the previous year of all portfolios, and here EPS is calculated as a company's annual total earnings (usually after interest and taxes) divided by the total number of ordinary shares outstanding (after deducting the preference share).

As early as 1980s, Krinsky & Rotenberg (1989) and Ritter (1984) pointed out that the accounting data prior to issuance are related to the firm’s IPO value. So being one kind of accounting data, EPS of the previous year could be regarded as a fundamental indicator to determining an “intrinsic value” of a certain stock. Since EPS of the previous year is one of the easiest metrics that can be got from most financial statements without costs, it is an important indicator of a firm’s previous performance before this firm is listed. In addition, EPS also indicates the potential return on individual investments, so most retail perspective investors and potential investors often compare different companies’ EPS and then estimate IPO’s strength.

(17)

17

 PER

Hypothesis 4: The IPO P/E ratio is negatively related to IPO underpricing

PER mentioned in the equation refers to the average IPO P/E ratio of stocks of all portfolios. IPO P/E ratio is calculated by offering price to the earning-per-share of the company on the issued day. It is a multiple to measure how much the investors are willing to pay per dollar of earnings. For example, if a company were currently trading at a multiple (P/E ratio) of 20, the interpretation is that an investor is willing to pay $20 for $1 of current earnings.

The IPO P/E ratio is also an important financial indicator for potential investors to calculate speculative opportunities on an IPO stock. Generally, stocks with higher forecast earnings growth will usually have higher P/E so as higher return, while stocks expected to have lower earnings growth will, in most cases, have a lower P/E. By comparing prices and earnings-per-share for a company, one can analyze the market's stock valuation of the company, and then his shares relative to the income of the company is actually generating.

Cheung et al. (2009) found out that IPO underpricing is significantly related to P/E ratio (negative). Liu (2003) argues that P/E ratio is close to the demand of the shares of IPO. Liu mentioned that P/E ratio is a measure of the price level of the IPO. If the IPO is priced too high, the demand will decrease accordingly. Therefore, the initial return will be lower due to less demand. So here I agree that IPO P/E ratio is negatively related to IPO.

 LTR

Hypothesis 5: The turnover ratio of the firm is positively related to IPO underpricing.

LTR is the logarithm of the average turnover ratio of all portfolios. Turnover ratio is calculated as the first-day trading volume divide by the number of shares outstanding. This ratio represents the frequency of trading in approaching the high closing price used in calculating the first day returns. The more the frequency of trading is, the more investors and their funds are attracted into the market, and the more investors pay attention to such stocks for speculating instead of holding on it to get a long-term return.

(18)

18

higher first day turnover rates of A-share IPOs imply that domestic investors are more interested in the short-term investment and the market is highly speculative. Obviously, the higher the first day turnover rate, the higher the level of underpricing. Therefore, the turnover ratios in the first trading day are expected to be positively related to the first day returns. Here I run a regression with the natural logarithm of the average turnover ratio.

 LPROE

Hypothesis 6: The ROE on common stockholders’ equity of the firm is positively related to IPO underpricing.

LPROE is the logarithm of the average return on common stockholders’ equity (ROE) by the fiscal year end before the IPO, the formula is:

Return on Equity = Net Income after Tax / Common Shareholder's Equity, Where, Shareholder’s Equity = Total Assets - Total Liabilities.

Return on equity is used as a measurement for firm performance that investors should consider. A business that has a higher return on equity is more likely to be one that is capable of generating cash internally. The benefit comes from the earnings reinvested in the company at a high ROE rate, which in turn gives the company a high growth rate

Cheung et al. (2009) found a result that ROE is positively related to IPO underpricing in two periods (from January 1, 1992 to July 27, 1999, and from January 1, 2005 to December 31, 2005). But in the whole period from January 1, 1992 to December 31, 2005, the P-value is higher than 10%. According to this, I consider that there should be a positively relationship between ROE and IPO underpricing. Here I run a regression with the natural logarithm of the value.

 SZMKTR & SZMKTR

SZMKTR is the daily returns of SZSE A-Share Index. Constituents for SZSE A-Share Index are all listed A-shares at Shenzhen Stock Exchange. SHMKTR is the daily return of SHSE A-Share Index. Constituents for SHSE A-Share Index are all listed A-shares at Shanghai Stock Exchange.

(19)

19

conclusion, I add these two A-share market returns to the hysteresis model as well as the multi-factor model.

Descriptive statistics for the above variables are shown in Table 1.

We can obviously see that the IPO market is highly speculative since that, on the first day, the average turnover ratio is more than 71.33%. Moreover, both the market returns are smaller than zero on these IPO days, which indicates that, on these first trading days, the market shows a negative return, so that the most investors should lose money in the secondary market, especially in the small and medium stock market.

The average IPO P/E ratio of companies listed in China is 55.60, more than two times than that of China’s A-share market3, indicating that investors are expecting higher potential earnings growth on the first trading day. This number also implicates that the IPO market is euphoria with a high level of speculation.

Table 1 Details of Original Data

3

See Yahoo Finance, the average P/E ratio of A-share market in China in 2010 is 22.65.

(20)

20

Table 2 Overview of the Hypotheses, Variables and Prediction about the Relationship between Each

Variable and the IPO Initial Return on the First-Trading Day Note:

IR1 is the first-order lagged of the initial return. IR2 is the second-order lagged of the initial return. LSIZE is the logarithm of the average IPO size. LTR is the logarithm of the average turnover rate of the first-trading day. PEPS is the average previous year’s EPS. PER is the average IPO P/E ratio of the first-trading day. LPROE is the logarithm of the average ROE by the fiscal year end before the IPO.

Category Hypotheses Variables Exp. Sign

Lag Variables H1 IR1 +

IR2 + Other Variables H2 LSIZE - H3 PEPS - H4 PER - H5 LTR + H6 LPROE +

(21)

21

4. Empirical Results and Analysis

4.1 Multiple Hysteresis Effect

4.1.1 Multiple Hysteresis Model

Table 3 shows the fundamental characteristics of the IPO initial return.

Table 3 Fundamental Features of the Initial Return.

Features Mean Median Std. Dev. Maximum Minimum Results 44.16% 35.10% 0.3653 202.78% -11.05% Features Skewness Kurtosis Jarque-Bera P-Value

Results 1.3379 5.5980 86.9359 0.0000 Market SHSE SZSE SHSE & SZSE IPO initial returns 30.35% 45.69% 44.16%

According to table 3, we can find that the maximum IPO return is 202.78% and the minimum is -11.05%. We also can see that the average underpricing for IPO is 44.16% after the reform of Equity Division, which is much lower than those research findings before (e.g. Mok and Hui (1998): 956%). This may be because that firstly, with the development of China’s stock market, the severe underpricing phenomenon in the primary market has been gradually reduced. Secondly, it has been mentioned above that in this period, from July 2009 to February 2011, both the stock markets show a negative market return. In entire stock market in China, the IPO initial returns cannot be highly beyond rationality. However, compared with the mature security markets in developed countries like USA, UK and Japan (where the returns of IPO underpricing are range from 10% to 20%), China’s securities market still shows a much higher degree of “Money Left on the Table”

Moreover, the IPO initial return’s kurtosis is higher than the normal value of 3 and the skewness is not equal to 0, which means the distribution of the average IPO return tends to exhibit leptokurtosis distribution that has a fat tail and excess peakedness. Meanwhile the P-value of IPO initial return equals to zero, which also certified the same conclusion.

(22)

22

IPO stocks sold is not time-continuous. It is meaningless to research the hysteresis effect in different exchange separately because the whole A-shares IPO market is time-continuous. Furthermore, although both exchanges focus on different types of firms, they have the same conditions and requirements to investors and companies. In other words, investors can choose either exchange to invest. Consequently, I think it is logical and reasonable to combine all the firms issued in both exchanges and make them as portfolios.

Table 4 shows the autoregression of the IPO return in an OLS model.

Table 4 Autoregression Model Result 1 of the IPO Return

Notes:

1) IR1, IR2 and IR3 are the first-order, second-order and third-order lagged returns of IPO initial return, respectively. C is the constant.

2) The equation of Regression 1 is: IRi= C + αi−1IRi−1+ αi−2IRi−2+ αi−3IRi−3+ 𝜇𝑖

Variable Coefficient Std. Error t-Statistic Prob.

IR1 0.3363 0.0781 4.3045 0.0000

IR2 0.2510 0.0799 3.1419 0.0020

IR3 0.0042 0.0775 0.0548 0.9564

C 0.1646 0.0420 3.9171 0.0001

R-squared 0.3042 Durbin-Watson stat 1.8609 Adj.R-squ. 0.2897 Prob (F-statistic) 0.0000

We can see evidently that the first-order and second-order lagged returns are positively significant to IPO initial return, and their P-values are both lower than 1%. Moreover, their coefficients are high, which means the previous initial return highly impact on the current initial return. Nevertheless, the third-order lagged return is not statistically significant and its coefficient is low. This result can confirm hypotheses 1 that the “Multiple Hysteresis Effect” exists on IPO returns only for two orders, so as to IPO underpricing.

In order to obtain a more accurate model, I remove the third-order lagged variable, and then I put the first-order and second-order lagged IPO returns into Eviews and test again. I get regression 2 in table 5 below:

(23)

23

Hysteresis effect is shown as follows:

IRi = 0.1531 + 0.3551*IRi-1 + 0.2821*IRi-2 + 𝜇𝑖 (4.1)

Table 5 Result of Autoregression Model 2 of the IPO Return

Notes:

1) IR1 and IR2 are the first-order and second-order lagged returns of IPO initial return, respectively. C is the constant.

2) The equation of Regression 2 is: IR_i= C + α_i−1IR_i−1+ α_i−2IR_i−2+𝜇_𝑖

Variable Coefficient Std. Error t-Statistic Prob.

IR1 0.3551 0.0792 4.4811 0.0000

IR2 0.2821 0.0787 3.5849 0.0005

C 0.1531 0.0425 3.6039 0.0004

R-squared 0.3131 Durbin-Watson stat 1.9199 Adj.R-squ. 0.3037 Prob (F-statistic) 0.0000

Table 6 Autoregression Model Result with Market Variable

Notes:

1) The significance levels reported are: * Significant at 10%, ** Significant at 5% and *** Significant at 1%.

2) Standard errors are shown in parentheses.

3) IR1 and IR2 are the first-order and second-order lagged returns of IPO initial return, respectively. SZMKTR is the daily returns of SZSE A-Share Index. Constituents for SZSE A-Share Index are all listed A-shares at Shenzhen Stock Exchange. SHMKTR is the daily return of SHSE A Share Index. Constituents for SHSE A-Share Index are all listed A-shares at Shanghai Stock Exchange. 4) The equation of Regression 3 is:

IRi= C + αi−1IRi−1+ αi−2IRi−2+ β₁SZMKTR + β₂SHMKTR + 𝜇𝑖

Variable Regression3 Regression4 Regression5

IR1 0.3665*** (0.0806) 0.3609*** (0.0802) 0.3658*** (0.0804) IR2 0.2824*** (0.7990) 0.2872*** (0.0795) 0.2862*** (0.0790) SHMKTR 1.3223 (3.5902) -0.9083 (1.6802) SZMKTR -2.1619 (3.0735) -1.1620 (1.4366) C 0.1440*** (0.0443) 0.1470*** (0.0440) 0.1436*** (0.0442) Observations 149 149 149 R-sq. 0.3168 0.3145 0.3162 Adj. R-sq. 0.2979 0.3000 0.3020 F-statistics 16.6957*** 22.1732*** 22.3489***

(24)

24

think about the performance of market return to make the IPO initial return more accurate. So I added two A-shares market returns to the model, then I got regression 3 in table 6. However, I find that the market returns have no significance on IPO initial return, whatever the SHSE or SZSE is. Therefore, the model (4.1) is reasonable.

4.1.2 Robustness Tests on Multiple Hysteresis Model

Firstly, I use Breusch-Godfrey Serial Correlation LM Test to examine whether there is serial correlation up to lagged order P in the IPO initial return, where P is equal to 4 in this case.

The test result is shown in the table 7. The first part of the output presents the test statistics and associates probability values. The P-values are both 0% in the top part, which represents that the IPO initial return has serial correlation. The tested regression is depicted in the second part of the table. The low probabilities of RESID (-1) and RESID (-2) indicate that the model should include two lagged explanatory variables. This result is consistent with the model (4.1) and supports the hypothesis as well.

Table 7 Result of Breusch-Godfrey Serial Correlation LM Test

F-statistic 16.7004 Prob. F(4,146) 0.0000 Obs*R-squared 47.4011 Prob. Chi-Square(4) 0.0000

Variable Coefficient Std. Error t-Statistic Prob. C -0.0010 0.0250 -0.0404 0.9678 RESID(-1) 0.3667 0.0790 4.6401 0.0000 RESID(-2) 0.2761 0.07925 3.4863 0.0006

Secondly, I use Chow-Test to examine the stability of the “Multiple Hysteresis Effect” model according to season. I separate the whole period in five seasons. The results are as follows in table 8:

(25)

25

Table 8 Chow Breakpoint Test of Multiple Hysteresis Model

Note: P-values are shown in parentheses.

Time Point 01/01/2010 01/04/2010 01/07/2010 01/10/2010 01/01/201 F-statistic 1.3145 (0.2720) 0.8150 (0.4876) 0.3541 (0.7862) 0.5067 (0.6783) 0.8809 (0.4526) Log likelihood ratio 4.0534 (0.2558) 2.5261 (0.4706) 1.1028 (0.7764) 1.5756 (0.6649) 2.7286 (0.4354) Wald Statistic 3.9436 (0.2676) 2.4450 (0.4853) 1.0624 (0.7862) 1.5202 (0.6776) 2.6428 (0.4500)

Thirdly, in Hu and Guo (2009), they mentioned that the IPO returns show a (G)ARCH effect in two lags, so I use the ARCH-LM test to examine whether the coefficients in the regression are zero with two lags (the result is in table 9). Their p-values are higher than 5% and the null hypothesis is accepted, so we can say that the series have no ARCH effects. That is to say, variance of the series does not fluctuate over time. Moreover, the GARCH (1, 1) model in table 9 also gives the same outcome. As a result, we can make a conclusion that the model is unconditionally heteroscedasticity and volatility does not cluster. This outcome does not support the conclusion from Hu and Guo (2009) that the IPO initial return has (G) ARCH effect.

Table 9 Heteroscedasticity Test: ARCH-LM Test and GARCH (1, 1) Test

ARCH-LM Test

F-statistic 1.3368 Prob. F(2,146) 0.2659 Obs*R-squared 2.6795 Prob. Chi-Square(2) 0.2619 Variable Coefficient Std. Error t-Statistic Prob.

C 0.1073 0.0276 3.8831 0.0002 RESID^2(-1) 0.0479 0.0821 0.5826 0.5611 RESID^2(-2) 0.1224 0.0820 1.4925 0.1377 R-squared 0.0180 F-statistic 1.3368 Adjusted R-squared 0.0046 Prob(F-statistic) 0.2659

GARCH(1,1) Test

GARCH = C(2) + C(3)*RESID(-1)^2 + C(4)*GARCH(-1)

Variable Coefficient Std. Error z-Statistic Prob. C 0.38012 0.0321 11.8386 0.0000

Variance Equation

(26)

26 4.2 Other Factors of IPO Underpricing 4.2.1 Multi-factor Regression Model

Table 10 Result of Multi-factor Regression Model

Notes:

1) The significance levels are: * Significant at 10%; ** Significant at 5% and *** Significant at 1%. 2) Standard errors are shown in parentheses.

3) The table shows the OLS parameter estimates for the Multiple Hysteresis Effect model includes a constant (C), the first-order and second-order lagged of the initial return (IR1 and IR2), the logarithm of the average IPO size (LSIZE), the logarithm of the average turnover rate of the first-trading day (TR), the average previous year’s EPS (PEPS), the average IPO P/E ratio of the first-trading day (PER), the logarithm of the average ROE by the fiscal year end before the IPO (LPROE), the average market return of Shenzhen Stock Exchange A-shares Index (SZMKTR) and the average market return of Shanghai Stock Exchange A-share Index (SHMKTR)

4) Regression 6 equation: IRi= C + αi−1IRi−1+ αi−2IRi−2+ β₁LSIZE

+ β₂PERS + β₃PER + β₄LTR + β₅LPROE + β₆SZMKTR + β₇SHMKTR + 𝜇𝑖

Regression 9 equation: IRi= C + αi−1IRi−1+ αi−2IRi−2+ β1LSIZE + β2PERS

+β₃PER + β₄LTR + β₅LPROE + 𝜇𝑖

Regression 10 equation: IR_i = C + α_i−1IR_i−1+ α_i−2IR_i−2+ β₁LSIZE + β₂PERS +β₃PER + β₄LTR + 𝜇𝑖

Variable Reg. 6 Reg. 7 Reg. 8 Reg. 9 Reg. 10

(27)

27

Continue Table 10:

Variable Reg. 11 Reg. 12 Reg. 13 Reg. 14 Reg. 15 Reg. 16

IR1 0.2513*** (0.0745) 0.3965*** (0.0803) 0.3340*** (0.0803) 0.3382*** (0.0787) 0.2337*** (0.0729) 0.3131*** (0.0766) IR2 0.2253*** (0.0670) 0.2594*** (0.0783) 0.2984*** (0.0792) 0.2698*** (0.0780) 0.2104*** (0.0709) 0.26209*** (0.0746) LSIZE -0.0516** (0.0217) -0.0440** (0.0196) PEPS -0.2930*** (0.1056) -0.1421 (0.0993) PER -0.0034** (0.0016) LTR 0.5481*** (0.0977) 0.6123*** (0.0957) LPROE -0.1136 (0.0756) C 0.8299*** (0.1398) 0.3211*** (0.0858) 0.2517*** (0.0809) 0.3610*** (0.1061) 0.4615*** (0.0613) 0.0242 (0.1018) R-sq. 0.4957 0.3362 0.3227 0.3340 0.4650 0.3217 Adj. R-sq. 0.4779 0.3224 0.3086 0.3202 0.4539 0.3074 F-statistics 27.9162*** 24.4764*** 23.0241*** 24.2395*** 41.7234*** 22.4522***

Table 10 above shows the results of multi-factor regression models.

Similar to the tests previous, regressions 6-8 indicate that market returns do not play a role in IPO initial return, no matter what the market is and how the market performs. As a result, I remove away both market returns from the model.

It is obvious that in all regressions in table 10, the first and second-order lagged IPO returns are highly statistically significant whose probabilities are lower than 1%. These outcomes fully support the hypotheses 1 that the performances of previous two lagged returns are one of the main factors of IPO underpricing.

(28)

28

result is consistent with the hypothesis 2 and other precious studies like Su & Fleisher (1999).

Moreover, I think this is also a reason to explain the difference between the average IPO return in SHSE and SZSE. As mentioned above, the average IPO return in SHSE is 30.35% while in SZSE it is 45.69%. The SHSE issues large and medium business while the SZSE focuses on the medium and small companies. From the following table 11, we can see that the IPO size in SHSE is larger than that in SZSE. According to the hypotheses 2 and the result of empirical test, the IPO underpricing level in SHSE is lower than that in SZSE.

Table 11 Features of IPO Size in Different Exchanges

Features Mean Median Std. Dev. Maximum Minimum SZSE 31.4690 27.0000 23.3710 399.8000 9.0000 SHSE 1581.6690 220.0000 4303.1060 25570.5900 54.9000

From regression 6-9 in table 10, the probabilities of PEPS are higher than 5%, but after removing all the insignificant variables from the model, the PEPS is much stronger significant (see regression 10 and 11). So here, we can get the conclusion that the average earnings-per-share of the previous year is negatively related to IPO underpricing. The outcome confirms the hypothesis 3.

In regression 6 in table 10, the coefficient of the IPO P/E ratio is -0.0034 and its significance is at 1% level. Although the result shows that the coefficient is close to zero, the level is significant. Consequently, we can conclude that the higher P/E ratio stocks are corresponding to a less underpricing of IPOs. The finding supports the hypothesis 4 as we expected.

The turnover ratio is a signal of speculation that how the market reacts to the new shares. In table 10, we see that the P-values of the LTR in all the regressions are positively significant at a 1% level and the coefficients of the LTR in those regressions in table 10 are around 0.5. These results indicate that the turnover ratio is highly positively related to the IPO underpricing, which verifies the hypotheses 5. It is obvious that when IPOs are more underpriced, investors are more likely to sell their shares on the first trading day. Moreover, from table 6 above, we can also see that the market is highly speculative because the average first day turnover ratio is more than 71.33%.

(29)

29

are significantly related with IPO initials return except LPROE. The logarithm of the average ROE of the fiscal year end before the IPO (LPROE) do not have any impacts on the IPO return because all the significant levels are much greater than 5% and the correlations are very close to zero. Moreover, the signs are negative. So the hypothesis 6 cannot be supported here. This may be because that the ROE of previous years is a kind of financial proxy that is used to compare the performance of a company with other firms in the same industry. This value measures the long-term performance of the company in the management and operation, but it does not impact on the IPO issuing price too much as well as the short-term speculative behavior of the investors on the first-trading day.

In order to find a better regression to match the model, I also remove the variable LPROE and market returns from the factors that affect IPO underpricing. Then we get Regression 10 and we can see that it is much better fitted to the model as here all the parameters are strongly significant. Finally, we get the most advanced and suitable multi-factor regression model:

Regression 10: IR_i = C + α_i−1IR_i−1+ α_i−2IR_i−2+ β₁LSIZE + β₂PERS

+β₃PER + β₄LTR + 𝜇_𝑖 (4.2)

(30)

30

4.2.2 Robustness Tests on Multiple Hysteresis Model

Table 12 Chow Breakpoint Test of Multi-factor Regression Model

Note: P-values are shown in parentheses.

Time Point 01/01/2010 01/04/2010 01/07/2010 01/10/2010 01/01/2011 F-statistic 1.0099 (0.4319) 0.7708 (0.6291) 0.8034 (0.6007) 1.3111 (0.2436) 1.7074 (0.1025)

Log likelihood ratio 8.8080 (0.3588) 6.7708 (0.5616) 7.0500 (0.5312) 11.3348 (0.1834) 14.5939 (0.0675) Wald Statistic 8.0789 (0.4258) 6.1665 (0.6286) 6.4271 (0.5995) 10.4887 (0.2324) 13.6594 (0.0911)

(31)

31

5. Conclusion and Further Research

This paper focuses on the underpricing of IPOs in China. I studied the initial returns of 505 IPOs issued between July 2009 and February 2011 after the “Chinese Equity Division Reform” in A-share markets in China. I select IPOs issued within one certain week as a portfolio and finally get 151 samples.

I find that the average IPOs underpricing for A-share is 44.16% three years after the reform, which is much lower than the IPO returns before the reform. This indicates that equity division problem is the main reason of the severe high level of IPO underpricing before the Equity Division Reform. After the reform, the IPO market has underpricing performance, but not as high as previous study, so the invest behavior of IPO market is becoming more rational. Moreover, IPO initial return is higher in SZSE than SHSE, which are 45.69% and 30.35%, respectively.

I examine the hypothesis proposed by Hu and Guo (2009) of “Multiple Hysteresis Effect” on IPO underpricing, and then I use theoretical and empirical research to support the existence of “Multiple Hysteresis effect”. I find that the newly IPOs’ performances are highly related to the former two lagged IPOs’ performances because of the speculative factors of the potential investors in China’s stock market. This result supports the findings from Hu and Guo (2009). However, unlike the results reported in Hu and Guo (2009), I don’t find any (G) ARCH effect on IPO initial returns.

(32)

32

no relationship between the ROE of common stockholders’ equity of the firm and the market returns.

Based on all the theoretical and empirical research, this thesis gets a conclusion that the high speculation of the market is the main reason for the IPO underpricing.

(33)

33

References

Aggarwal, R., Krigman, L., & Womack, K. (2002). Strategic IPO Underpricing, Information Momentum, and Lockup Expiration Selling. Journal of Financial Economics, 66, 105-137.

Allen, F., & Faulhaber, G. R. (1989). Signaling by Underpricing in the IPO Market. Journal of Financial Economics, 23, 303-324.

Arosio, R., Guidici, G., & Paleari, S. (2000). What Drives the Initial Market Performance of Italian IPOs? An Empirical Investigation on Underpricing and Price Support (Unpublished Working Paper). Vienna Institute of Technology.

Beatty, R., & Ritter, J. (1986). Investment Banking, Reputation and Underpricing of Initial Public Offerings. Journal of Financial Economics, 15, 213-232.

Bradley, D., & Jordan, B. D. (2002). Partial Adjustment to Public Information and IPO Underpricing. Journal of Financial and Quantitative Analysis, 37, 595-616.

Carter, R., & Manaster, S. (1990). Initial Public Offerings and Underwriter Reputation. Journal of Finance, 45, 1045-1067.

Chan, K., John, K. C. W., & Junbo, W. (2004). Underpricing and Long-term Performance of IPOs in China. Journal of Corporate Finance, 10, 409-130.

Chang, E., Chen, C., Chi, J., & Young, M. (2008). IPO Underpricing in China: New Evidence from the Primary and Secondary Markets. Emerging Market Review, 9, 1-16.

Chen, G., & Gao, N. (2000). IPO Underpricing Degree and Reasons in China. Financial Research Journal (Chinese), 8, 1-15.

Chen, A., Hung, C., & Wu, C. (2002). The Underpricing and Excess Returns of Initial Public Offerings in Taiwan Based on Noisy Trading: a Stochastic Frontier Model. Review of Quantitative Finance and Accounting, 18, 139–59.

Cheung, Y. L., Ouyang, Z.W., & Tan, W.Q. (2009). How Regulatory Changes Affect IPO Underpricing in China. China Economic Review, 20, 692-702.

Derrien, F. (2005). IPO Pricing in “Hot” Market Conditions: Who Leaves Money on the Table? Journal of Finance, 60, 487-521.

(34)

34

Grinblatt, M., & Hwang, C.Y. (1989). Signaling and the Pricing of Unseasoned New Issues. Journal of Finance, 44, 393-420.

Guo, F. (2009). The Empirical Research on IPO Under-Pricing in Chinese Market. Working Paper. Shandong University.

Hanley, K. (1993). The Underpricing of Initial Public Offerings and the Partial Adjustment Phenomenon. Journal of Finance, 34, 231-250.

Hu, J., & Guo, F. (2009). The Multiple Hysteresis Effect on IPO Underpricing of China’s Listed Company: From Theoretical Demonstration to Empirical Demonstration. Shangdong Social Science, 11.

Jagannathan, R., & Gao, Y. R. (2005). Are IPOs underpriced? A closer examination. Working Paper, Kellogg Graduate School of Management, Northwestern University.

Jenkinson, T., & Ljungqvist, A. (2001). Going Public: The Theory and Evidence on How Companies Raise Equity Finance. Oxford University Press, Oxford and New York.

Kim, M., & Ritter, J.R. (1999). Valuing IPOs. Journal of Financial Economics, 53, 409-437

Li, P. (2008). An Empirical Study on IPO Underpricing under Full Circulation in China. Asian Social Science, 4, 1.

Retrieved from: http://www.ccsenet.org/journal/index.php/ass/article/view/2094/1972

Liu, Y., & Xiong, P. (2005). Equity Separation, Government Regulations, and Chinese IPO Underpricing Puzzle. Economic Research Journal, 5, 85-95.

DOI: cnki: ISSN: 0577-9154.0.2005-05-008

Loughran, T., & Ritter, J. (2004). Why has IPO Underpricing Changed over Time? Financial Management, 33, 5-37.

Loughran, T., & Ritter, J.R. (1995). The New Issues Puzzle. Journal of Finance, 50, 23-51.

Ma, S., & Faff, R. (2007). Market Conditions and the Optimal IPO Allocation Mechanism in China. Pacific-Basin Finance Journal, 15, 121-139.

Mok, H. K., & Hui, Y. V. (1998). Underpricing and Aftermarket Performance of IPOs in Shanghai, China. Pacific-Basin Finance Journal, 6, 453-474.

(35)

35

Ranjan, N., & Madhusoodanan, T. P. (2004). IPO Underpricing Issue Mechanisms and Size. Retrieved from SSRN: http://ssrncom/abstract=520744

Ritter, J., & Welch, I. (2002). A Review of IPO Activity, Pricing, and Allocations. The Journal of Finance, 57, 828-1795.

Rock, K. (1986). Why New Issues are Underpriced. Journal of Financial Economics, 15, 187-212.

Su, D., & Fleisher, B.M. (1998). What Explains the High IPO Returns in China? Emerging Market Quarterly, 2, 5-20.

Su, D., & Fleisher, B.M. (1997). An Empirical Investigation of Underpricing in Chinese IPOs. Pacific-Basin Finance Journal, 7, 173-202.

Tian, L. H. (2003). Financial Regulations, Investment Risks and Determinants of the Excess Chinese IPO Underpricing. Working Paper, Peking University Management School and London.

Yu, T., & Tse, Y. K. (2005). An Empirical Examination of IPO Underpricing in the Chinese A-Share Market. China Economic Review, 17, 363-382.

Wang, Y., & Zhang, H. (2008). Short Sales Constraints, Divergence of Opinion and IPO Overvaluation. Management Science and Engineering, 15, 1163 – 1168. DOI: 10.1109/ICMSE.2008.4669057

Welch, I. (1992). Sequential Sales, Learning, and Cascades. Journal of Finance, 47, 695-732.

Welch, I. (1989). Seasoned Offerings, Imitation Costs, and the Underpricing of Initial Public Offerings. Journal of Finance, 44, 421-449.

Willenborg, M. (1999). Empirical Analysis of the Economic Demand for Auditing in the Initial Public Offerings Market. Journal of Accounting Research, 37, 37-225.

Zhang, H., & Wang, Z. (2009). The Study of Equity Division Reform Effect on Chinese Listed Companies Financing Behaviors. Asian Social Science, 4, 1.

Retrieved from: http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=05303567&tag=1

Zheng, S. (2007). Are IPOs Really Overpriced? Journal of Empirical Finance, 14, 287-309.

(36)

36

Appendix

Graph I

The Residual of IPO Initial Return

Table I

Time Series of IPO Issuing

(37)

37

Table II

The Correlation Matrix

IR1 IR2 LSIZE PEPS PER LTR LPROE C IR1 1 IR2 -0.4409 1 LSIZE -0.2061 0.0540 1 PEPS -0.0457 -0.0742 0.5636 1 PER -0.0924 0.0833 0.3637 0.2114 1 LTR -0.3000 -0.1222 0.3090 0.1176 0.1359 1 LPROE 0.1528 -0.0070 -0.0539 -0.4540 -0.3582 0.0267 1 C 0.0787 -0.1010 -0.6253 -0.7571 -0.6747 -0.0066 0.6986 1 Note: IR1 is the first-order leg of the initial return. IR2 is the second-order leg of the initial return.

An Empirical Test on Initial Public Offering Stock Underpricing in China A-shares Market