SEARCH

SEARCH BY CITATION

Keywords:

  • Initial public offerings;
  • Long-run performance;
  • China stock market;
  • Signaling hypothesis;
  • Divergence of opinion hypothesis
  • G15;
  • G32

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

This study examines the long-run performance of 936 Chinese initial public offerings (IPOs) over the period 1996 to 2005 (the post-issue return evidence ends in June 2008). Using a number of empirical methods, including event-time and calendar-time approaches based on a size and industry matching firm benchmark, we find a significant long-run overperformance using the equal-weighted buy-and-hold abnormal returns, although not for the value-weighted returns, suggesting that the performance of small-size IPO firms is superior to that of large-size IPO firms. The significant overperformance disappears, however, when using cumulative or calendar-time abnormal returns. The preset study provides out-of-sample evidence in an emerging market context in support of Fama's (1998) argument that reported long-run performance is sensitive to the method of analysis. Finally, based on a rich set of explanatory factors as proxies for both signaling and ex-ante uncertainty characteristics, our results are supportive of the signaling hypothesis, but inconsistent with the divergence of opinion hypothesis.


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

Previous studies document two important aspects of firms that engage in initial public offerings (IPOs): underpricing of the stocks at issue and subsequent long-run underperformance as measured by abnormal returns (ARs) (see e.g. Loughran et al., 1994; Ritter and Welch, 2002; Yong, 2007; Chambers and Dimson, 2009). The evidence on initial underpricing is consistent in the published literature; empirical evidence on the long-run performance of IPOs is much less so. Using a sample of 1526 US IPOs from 1975 to 1984, Ritter (1991) finds that IPOs substantially underperform a comparison group of matching firms over 3 years after the first day of trading. Loughran and Ritter (1995) investigate 4753 US IPOs from 1970 to 1990 and document that average 3-year and 5-year buy-and-hold returns for IPOs are 8.4 and 15.7%, respectively, which are significantly lower than the 35.3 and 66.4% found for their matching firms. Fama (1998, p. 283), however, critically reviews the literature on this topic and argues that: “apparent anomalies can be due to methodology and most long-term return anomalies tend to disappear with reasonable changes in technique.” In examining 3661 US IPOs over the period 1935–1972, Gompers and Lerner (2003) confirm that the long-run performance of IPOs depends on the method used to measure returns. They find a long-run underperformance of IPOs over 5 years after listing by using value-weighted buy-and-hold AR (BHAR); however, such underperformance disappears when using cumulative AR (CAR) or equal-weighted BHAR. In addition, the calendar-time analysis shows that IPO returns are quite close to the market return, indicating no abnormal performance in the long run.

Empirical investigations of the long-run performance of IPOs in other developed markets also provide mixed results. For example, Hwang and Jayaraman (1995) report a significantly negative CAR of −14.98% for 182 Japanese IPOs relative to an equal-weighted market index return. This result, however, is reversed when a value-weighted index return is used, with a significantly positive CAR of 16.44%. Espenlaub et al. (2000) examine 588 UK IPOs over the period 1985–1992 based on several alternative methods under the event-time framework. They find substantially negative AR over 3 years after listing, irrespective of the benchmarks applied, but the significance of long-run underperformance is less marked when ARs are measured under the calendar-time framework. Kooli and Suret (2004) investigate the long-run performance of 445 Canadian IPOs from 1991 to 1998, finding that IPOs significantly underperform their matching firms over 5 years after listing using the value-weighted CAR, whereas the observed long-run underperformance becomes less statistically significant with the use of the BHAR. Further results from their calendar-time analysis show that IPOs exhibit long-run underperformance under the equal-weighting scheme, but not under the value-weighting scheme.

According to Fama (1998), the debate over long-run performance cannot be put to rest without more out-of-sample assessments. To that end, the present study investigates the long-run performance of IPOs in China’s stock market, one of the most important emerging markets in the world. Our particular focus on China’s stock market is motivated by several considerations. First, China’s impact on world affairs has increased substantially in recent 10 years and, from a financial market perspective, has become increasingly important to global investors. In particular, with the introduction of the Qualified Foreign Institutional Investors (QFII) Program into the China capital market and foreign investment banks being allowed to underwrite A-share IPOs, more foreign institutional investors are now able to access this potential. Therefore, fully understanding the mechanism and behavior of stock returns in China’s stock market is now of great interest to global investors. However, China’s stock market is less than 20 years old and its institutional setting, as well as trading practices, are relatively new and somewhat different from and independent of those in developed markets. For example, it is hard to short sell in China’s stock market and, hence, risk-free arbitrage is likely to be difficult. In addition, China’s stock market is dominated by individual investors as the mutual fund industry is in the infant stage of development.1 China’s stock market thus provides a unique environment for investigating stock price behavior, and out-of-sample evidence from this emerging market will help to shed some light on the question of whether IPO long-run performance is due to sample and market-specific characteristics.

Second, although a variety of studies have confirmed severe initial underpricing of Chinese IPOs (see e.g. Mok and Hui, 1998; Su and Fleisher, 1999; Chen et al., 2004; Chi and Padgett, 2005a; Gannon and Zhou, 2008; Guo and Brooks, 2008), very little research has been conducted on long-run performance in China’s stock market. Extant studies are limited either from the short sample periods used or by the incomplete datasets examined, and, hence, the existing results are by no means conclusive. For example, Mok and Hui (1998) investigate 101 A-share IPOs listed on the Shanghai Stock Exchange (SHSE) from 1990 to 1993 and find a significantly positive cumulative average daily return for 350 days after listing. Chen et al. (2000) examine 277 A-share IPOs from 1992 to 1995 and report a significant 3-year underperformance. Using a dataset of 570 A-share IPOs from 1993 to 1998, Chan et al. (2004) report insignificant long-run underperformance using size, book-to-market ratio (B/M), and size and B/M matching firm benchmarks. Chi and Padgett (2005b) report an average market-adjusted CAR of 10.3% and BHAR of 10.7% over 3 years after listing for 409 A-share IPOs issued in 1996 and 1997. Cai et al. (2008) examine 335 IPOs listed only on the SHSE, and document that IPOs underperform the market by up to 30% in the long run over the period 1997–2001.

Our purpose is to extend previous studies through a systematic and comprehensive examination of the long-run performance of Chinese IPOs over the period 1996–2005, which will enable a comparison of long-run performance of IPOs over a long-term horizon and across distinct firm-specific and institutional characteristics. In particular, the present study supplements existing financial literature in the following major aspects. First, we examine the most up-to-date and comprehensive dataset of 936 Chinese IPOs, which incorporates post-issue return evidence up to June 2008. Unlike previous Chinese studies that generally employ the equal-weighted event-time approaches to measure the market-adjusted long-run performance (see e.g. Mok and Hui, 1998; Chen et al., 2000; Chi and Padgett, 2005b; Cai et al., 2008), we are the first study, to our knowledge, to introduce a novel size and industry matching firm benchmark as well as to compare the results between the event-time and calendar-time approaches, between the equal-weighting and value-weighting schemes, and between the market and matching firm benchmarks. We find a long-run overperformance when using the equal-weighted BHAR calculation based on the size and industry matching firm benchmark. The overperformance relative to the matching firm benchmark is 4.6% (t-statistic = 2.09) over 24 months and 8.6% (t-statistic = 3.57) over 36 months after listing. The long-run overperformance becomes much smaller and insignificant, however, with the use of the value-weighted returns, suggesting the superior performance of small-size IPO firms in the long run. In addition, IPOs do not appear to underperform or outperform their matching firms under other event-time and calendar-time approaches. Therefore, we provide out-of-sample evidence in an emerging market context in support of Fama’s (1998) view that the sensitivity of long-run performance depends on the choice of empirical methods. We also demonstrate that the controversy over mixed results concerning IPO long-run performance is not simply sample and market specific.

Finally, we examine a number of proxies for signaling and ex-ante uncertainty characteristics in explaining the long-run abnormal performance of Chinese IPOs. We find a significantly positive relationship between IPO long-run performance and quality of IPO firms, consistent with the prediction of the signaling hypothesis (Allen and Faulhaber, 1989; Welch, 1989). For example, within the first 3 years after listing, IPO firms making seasoned equity offerings (SEOs) perform much better than those making no SEO; there is a significantly positive relationship between earnings per share (EPS) prior to issue and the long-run performance of IPOs; and IPO firms with low standard deviation of post-issue returns outperform those with high standard deviation of post-issue returns. Moreover, we find that IPO firms exhibiting higher ex-ante uncertainty are more likely to generate higher returns in the long run, inconsistent with the prediction of the divergence of opinion hypothesis (Miller, 1977). For example, small-size IPO firms significantly outperform big-size IPO firms; and IPO firms classified in the information technology (IT) industry category significantly outperform those classified under non-IT industry categories. We attribute the observed long-run overperformance to the compensation for ex-ante uncertainty.

The remainder of this paper is organized as follows. The following section describes the data and sample selection, as well as highlights both event-time and calendar-time approaches to calculating long-run AR. Section 3 discusses the empirical results. Section 4 contains the cross-sectional analyses performed in exploring possible sources of IPO long-run abnormal performance. The final section concludes this study.

2. Data and Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

2.1. Data and Sample Selection

The sample consists of 936 A-share IPOs in China’s stock market over the period 1996 to 2005, excluding IPOs with an offering price of less than RMB2.00 per share to avoid the influence of extreme price movements in low-priced stocks. The final sample is comprised of 578 IPOs listed on the SHSE and 358 IPOs listed on the Shenzhen Stock Exchange (SZSE), covering 86.5% of the A-share IPOs that went public during the sample period. Data on the listing dates, offering dates, offering prices, closing prices on the first day of trading, gross proceeds, EPS prior to issue, and government and institutional ownership are collected directly from the GTA database.2 We use the official websites of the SHSE (http://www.sse.com.cn) and the SZSE (http://www.szse.cn) to double-check the accuracy of these data. Other data, such as the total return index for each IPO and non-IPO firm as well as the market index return, are collected from DataStream. The total return index is adjusted by DataStream for stock splits, stock dividends, and rights offerings.

Table 1 presents the distribution of IPOs by the year of listing from 1996 to 2005 by number of IPOs and gross proceeds. The gross proceeds are adjusted using the consumer price index (CPI; 2005 = 100).3 The annual number of IPOs in our sample ranges from a low of 14 in 2005 to a high of 175 in 1997. The number of IPOs listed on the SHSE far exceeds the number of IPOs listed on the SZSE, mainly due to the suspension of new issues on the SZSE from 2001 to 2003.4

Table 1.   Distribution of 936 Chinese initial public offerings (IPOs) by year of listing, 1996–2005 The sample consists of 936 Chinese IPOs listed on the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) over the period 1996 to 2005. This table presents the distribution of IPOs by the year of listing, in terms of the number of IPOs and aggregate gross proceeds. The aggregate gross proceeds are presented in millions of RMB, adjusted using the consumer price index (2005 = 100). US$1 was approximately RMB8.07 on 31 December 2005.
 SHSESZSEWhole market
NGross proceedsNGross proceedsNGross proceeds
19966210 073.48708682.6613218 756.14
19977222 966.0910336 389.5517559 355.64
19984823 737.954016 994.198840 732.14
19994526 685.045024 595.609551 280.64
20008459 858.544626 552.1313086 410.67
20017055 141.4107055 141.41
20026651 151.8406651 154.84
20036749 968.3506749 968.35
20046124 924.58389272.119934 196.69
200532854.54112610.03145464.57
Full sample578327 361.82358125 096.27936452 458.09

Table 2 presents the distribution of IPOs by gross proceeds, showing that small-size IPOs with gross proceeds of less than RMB100m represent 6.52% (61 IPOs) of the sample and 1.01% (RMB4592.20m of the RMB452 458.09m total) of aggregate gross proceeds, while large-size IPOs with gross proceeds of more than RMB1000m represent 6.84% (64 IPOs) of the sample and 34.85% (RMB157 668.71m of the RMB452 458.09m total) of aggregate gross proceeds. The remaining 811 IPOs (84.64% of the sample) have recorded gross proceeds of RMB290 197.17m. Compared with average gross proceeds (RMB349.43m) raised by the 358 IPOs listed on the SZSE, average aggregate gross proceeds (RMB566.37m) raised by the 578 IPOs listed on the SHSE are significantly higher (t-statistic = 4.40).

Table 2.   Distribution of 936 Chinese initial public offerings (IPOs) by gross proceeds, 1996–2005 The sample consists of 936 Chinese IPOs listed on the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) over the period 1996 to 2005. This table presents the distribution of IPOs by gross proceeds, in terms of the number of IPOs and aggregate gross proceeds. The aggregate gross proceeds are presented in millions of RMB, adjusted using consumer price index (2005 = 100). US$1 was approximately RMB8.07 on 31 December 2005.
 SHSESZSEWhole market
NGross proceedsNGross proceedsNGross proceeds
Gross proceeds ≤ RMB100201488.99413103.21614592.20
RMB100 < Gross proceeds ≤ RMB30021445 198.2216331 689.1937776 887.41
RMB300 < Gross proceeds ≤ RMB1000293142 179.2814171 130.49434213 309.77
Gross Proceeds > RMB100051138 495.331319 173.3864157 668.71
Full sample578327 361.82358125 096.27936452 458.09

2.2. Methodology

Given problems with the event-time approaches and the specification of t-statistics for non-zero AR, we will evaluate the long-run performance of Chinese IPOs by computing the event-time and calendar-time AR,5 based on two alternative benchmarks: (i) a size and industry matching firm benchmark for the event-time approaches; and (ii) the intercepts derived from the capital asset pricing model (CAPM) and the Fama and French (1993) three-factor model for the calendar-time approaches.

2.2.1. Event-time Approaches

The average annual number of trading days over the period 1996–2005 in China’s stock market is 242, with an average of 20 trading days per month. Hence, we define an event month as a 20-trading-day interval following the first day of trading. The initial return period is defined to be month 0, and the aftermarket period includes the following 36 months, where months are defined as successive 20-trading-day periods relative to the first day of trading. Therefore, month 1 consists of event days 2–21; month 2 consists of event days 22–41, and so forth. The equal-weighted and value-weighted BHAR and CAR are computed over 3 years after listing, excluding initial returns because it is frequently impossible to satisfy all investor demand for shares at the offering price due to an inadequate supply and high demand in China’s new issue market (Chi and Padgett, 2005a).

An obvious difficulty in measuring long-run performance is choice of an appropriate benchmark (see e.g. Ritter, 1991; Fama and French, 1996; Barber and Lyon, 1997).6 In the present study, each IPO firm in our sample is matched to a non-IPO firm that is selected based on size and industry characteristics. Following Ritter’s (1991) methodology, on 31 December of each year from 1996 to 2005, all firms listed on the SHSE and SZSE that have been traded for at least 3 years are ranked by their market values. The non-IPO firm classified under the same industry category and with the closest market value to an IPO firm is chosen as the matching firm. The market value of the matching firm should not be greater than 120% or less than 80% of the market value of the IPO firm, which is calculated as the product of shares outstanding and the closing price on the first day of trading. If it is impossible to find a matching firm with a close market value in the same industry category, an alternative firm with the closest market value in another similar industry category is chosen. As a result, 846 IPOs in the sample are matched with firms in the same industry category and 90 IPOs are matched with firms in a similar industry category.7

2.2.2. Calendar-time Approaches

To control for event clustering and potential cross-sectional correlations among individual firms, the Fama and French (1993) three-factor model is often used in assessing long-run performance of IPOs (see e.g. Loughran and Ritter, 1995; Espenlaub et al., 2000; Gompers and Lerner, 2003). We estimate both the CAPM and the Fama and French (1993) three-factor regressions to test the time-series significance in the pattern of long-run returns. The difference between the monthly return on a portfolio of IPO firms and the monthly risk-free rate is used as a dependent variable in the regressions. Like Wang (2004) and Chang et al. (2010), we use monthly return on the 3-month household deposit interest rate as the risk-free rate, because the data on the 3-month Chinese T-Bills (started in 2001) are not appropriate for the present study. The monthly returns on the IPO portfolios are regressed either on market premium, which is calculated as the difference between the monthly value-weighted return on the SHSE and SZSE A-Share Indexes and the monthly risk-free rate for the CAPM, and on market premium, size premium, and value premium for the Fama and French (1993) three-factor model. The intercepts from the regressions are used as indicators of the risk-adjusted long-run performance of the calendar-time IPO portfolios. Significantly positive or negative intercepts are evidence of the existence of IPO long-run overperformance or underperformance after controlling for market, size, and value factors.

Following the methodology proposed by Fama and French (1993), we construct portfolios that mimic size and value factors in China’s stock market. On 30 June of each year t from 1995 to 2008, all stocks listed on the SHSE and SZSE are divided into two size groups, small (S) or big (B), according to whether their market values are below or above the median market value of all stocks. All stocks are also divided into three value groups, high B/M (H), medium B/M (M), or low B/M (L), according to whether the values of their B/M at the end of year t − 1 are included in the top 30, middle 40, or bottom 30 percentile, respectively. Finally, we construct six portfolios, SL, SM, SH, BL, BM, and BH, from the intersections of two size groups and three value groups in June of year t and calculate monthly value-weighted returns on the six portfolios from July of year t to June of year t + 1. For example, SL portfolio contains stocks in both the small-size group and the low-B/M group, and BH portfolio contains stocks in both the big-size group and the high-B/M group. SMB represents the return on a zero-investment portfolio formed by subtracting the mean return on three big-size firm portfolios (BL, BM, and BH) from the mean return on three small-size firm portfolios (SL, SM, and SH), and HML represents the return on a zero-investment portfolio formed by subtracting the mean return on two low-B/M firm portfolios (SL and BL) from the return on two high-B/M firm portfolios (SH and BH).

3. Empirical Results and Analyses

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

3.1. Event-time Results

Table 3 reports the equal-weighted and value-weighted BHAR of 936 IPOs for 1–36 months after listing. The results show significantly negative equal-weighted and value-weighted BHAR in the first 2 months, implying a short-run underperformance of Chinese IPOs.8 However, long-run performance exhibits different characteristics. The results presented on the left-hand side of Table 3 show that the equal-weighted returns on IPO firms are significantly greater than those of their matching firms in the long run. For example, the overperformance relative to the matching firm benchmark is 4.6% (t-statistic = 2.09) after 24 months and 8.6% (t-statistic = 3.57) after 36 months. On the right-hand side of Table 3, no significant overperformance or underperformance is found in the long run based on the value-weighted returns. For example, the BHAR drop to 1.3% (t-statistic = 0.84) after 24 months and to 3.7% (t-statistic = 1.22) after 36 months.9

Table 3.   Post-issue buy-and-hold abnormal returns (BHAR) on 936 Chinese IPOs, 1996–2005 The sample consists of 936 Chinese initial public offerings (IPOs) listed on the Shanghai Stock Exchange and the Shenzhen Stock Exchange the period 1996 to 2005. This table presents the equal-weighted and value-weighted BHAR for subsequent 1–36 months after listing, based on the size and industry matching firm benchmark. Following the calculation introduced by Loughran and Ritter (1995), we define the size and industry matching-firm-adjusted BHARi,T on the IPO i as: inline image where BHRi,t and BHRb,t are the buy-and-hold return on IPO i and its matching firm, respectively. Ri,t and Rb,t are the monthly return on IPO i and its matching firm, respectively, in month t following listing. The equal- or value-weighted buy-and-hold abnormal return inline imageon a portfolio of N IPOs for a holding period T months is computed as: inline image where ωi is the weight. ωi = 1/Nis employed to calculate the equal-weighted abnormal returns and inline image is employed to calculate the value-weighted abnormal returns, where MVi is the market value of the IPO firm i, calculated as the product of shares outstanding and the closing price on the first day of trading, adjusted using CPI (2005 = 100). Lyon et al. (1999) suggests using bootstrapped skewness-adjusted t-statistics to evaluate whether the BHAR are significantly different from zero: inline image where inline image where inline image is the estimate of the coefficient of skewness. The t-statistics are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
MonthEqual-weightedValue-weighted
BHRi,tBHRb,tBHARTt(BHART)BHRi,tBHRb,tBHARTt(BHART)
 1–0.0100.010–0.020(–2.49)**–0.0170.000–0.017(–2.13)**
 2–0.0100.011–0.021(–1.88)*–0.024–0.005–0.019(–1.89)*
 30.0110.027–0.016(–1.32)–0.0140.006–0.020(–1.90)*
 40.0340.047–0.013(–1.05)–0.0030.017–0.020(–1.57)
 50.0530.056–0.003(–0.04)0.0120.020–0.008(–0.46)
 60.0650.0630.002(0.26)0.0220.032–0.010(–0.57)
 70.0720.0610.011(0.99)0.0150.022–0.007(–0.34)
 80.0660.0650.001(0.37)0.0080.0070.001(0.14)
 90.0750.078–0.003(–0.11)0.0110.0100.001(0.16)
100.0860.087–0.001(–0.04)0.0110.0100.001(0.19)
110.0910.0880.003(0.43)0.0170.0150.002(0.27)
120.1020.0940.008(0.56)0.0210.0150.006(0.34)
130.1030.0920.011(0.81)0.0200.0130.007(0.62)
140.1080.0860.022(1.34)0.0190.0120.007(0.77)
150.1080.0850.023(1.37)0.0160.0080.008(0.48)
160.1120.0790.033(1.98)**0.0120.0020.010(0.58)
170.1300.0880.042(2.23)**0.0140.0040.010(0.83)
180.1330.0840.049(2.52)**0.0110.0010.010(0.81)
190.1410.0840.057(2.89)***0.0100.0000.010(0.67)
200.1590.0930.066(3.37)***0.0210.0100.011(0.69)
210.1770.1120.065(3.19)***0.0270.0160.011(0.92)
220.1870.1300.057(2.88)***0.0320.0200.012(1.02)
230.1940.1450.049(2.25)**0.0340.0220.012(0.93)
240.2170.1710.046(2.09)**0.0460.0330.013(0.84)
250.2260.1760.050(2.13)**0.0550.0390.016(1.14)
260.2370.1810.056(2.35)**0.0660.0500.016(0.84)
270.2470.1910.056(2.36)**0.0720.0550.017(0.94)
280.2380.1950.063(2.54)**0.0670.0460.021(1.24)
290.2560.2230.063(2.60)***0.0750.0540.021(1.32)
300.2800.2220.068(2.94)***0.0880.0610.027(1.46)
310.3060.2350.071(3.19)***0.1090.0800.029(1.28)
320.3480.2670.081(3.26)***0.1230.0870.036(1.39)
330.3910.3080.083(3.34)***0.1450.1130.032(1.19)
340.4200.3410.079(3.28)***0.1620.1260.036(1.27)
350.4520.3670.085(3.49)***0.1850.1470.038(1.26)
360.4800.3940.086(3.57)***0.2000.1630.037(1.22)

Table 4 reports the results of the equal-weighted and value-weighted CAR of 936 IPOs by summing monthly AR for the 1–36 months after listing. Consistent with the short-run underperformance presented in Table 3, the significantly negative short-run CAR are found in the first 4 months after listing for both equal-weighted and value-weighting schemes. However, both equal- and value-weighted CAR are not significantly different from zero in the long run. For example, the equal-weighted and value-weighted CAR are 2.4% (t-statistic = 0.87) and −1.5% (t-statistic = –0.54), respectively, over 24 months after listing, while the equal-weighted and value-weighted CAR are −0.9% (t-statistic = −0.23) and −4.4% (t-statistic = −1.23), respectively, over 36 months after listing. Based on the empirical evidence presented in Tables 3 and 4, we conclude that the long-run performance of Chinese IPOs looks distinctly different depending on the method and weighting scheme employed, which supports Fama’s (1998) argument that long-run performance is sensitive to choice of method.

Table 4.   Post-issue cumulative abnormal returns (CAR) on Chinese IPOs, 1996–2005 The sample consists of 936 Chinese initial public offerings (IPOs) listed on the Shanghai Stock Exchange and the Shenzhen Stock Exchange over the period 1996 to 2005. This table presents the equal-weighted and value-weighted CAR for the 1–36 months after listing, based on the size and industry matching firm benchmark. The equal-weighted or value-weighted average size and industry matching-firm-adjusted return ARt on a portfolio of N IPOs in event month t is computed as: inline image where ωi is the weight. ωi = 1/N is used to calculate the equal-weighted abnormal returns and inline image is used to calculate the value-weighted abnormal returns, where MVi is the market value of the IPO firm i, calculated as the product of shares outstanding and the closing price on the first day of trading, adjusted using the consumer price index (2005 = 100). Ri,t and Rb,t are monthly return on IPO i and on its matching firm, respectively, in event month t following listing. Therefore, the matching-firm-adjusted cumulative returns, CART, from event month 1 to event month T are the summation of ARt: inline image The t-statistics for the average abnormal returns (ARt) are computed for each month t as: inline image, where Nt is the number of firms trading in month t and sdt is the cross-sectional standard deviation of the adjusted returns in month t. The t-statistics of the CARt are determined and computed using the methodology used by Ritter (1991): inline image, where csdt is computed as: csdt = [t × var + 2 × (t - 1) × cov]1/2. var is the mean cross-sectional variance over the 36 months and cov is the first-order autocovariance of the ARt series. The t-statistics are reported in parentheses. *** and ** denote statistical significance at the 1 and 5% level, respectively.
MonthEqual-weightedValue-weighted
ARtt(ARt)CARTt(CART)ARtt(ARt)CARTt(CART)
 1−0.020(−3.50)***−0.020(−3.49)***−0.018(−3.09)***−0.018(−3.09)***
 2−0.002(−0.39)−0.022(−2.85)***−0.003(−0.63)−0.021(−2.73)***
 30.006(1.19)−0.016(−2.55)**−0.002(−0.34)−0.023(−2.22)**
 4−0.001(−0.14)−0.017(−2.40)**−0.001(−0.26)−0.024(−2.05)**
 50.003(0.75)−0.004(−0.28)0.012(2.58)***−0.012(−0.91)
 60.003(0.72)−0.001(−0.03)0.005(1.15)−0.007(−0.46)
 70.004(0.76)0.003(0.20)0.003(0.61)−0.004(−0.23)
 8−0.002(−0.75)0.001(0.01)−0.003(−0.79)−0.007(−0.45)
 9−0.003(−0.59)−0.002(−0.14)−0.006(−1.47)−0.013(−0.79)
10−0.001(−0.09)−0.003(−0.16)−0.006(−1.57)−0.019(−1.16)
110.002(0.58)0.000(−0.02)0.004(1.02)−0.015(−0.83)
120.008(1.80)0.007(0.38)0.004(0.97)−0.011(−0.57)
130.002(0.43)0.009(0.45)0.007(1.53)−0.004(−0.19)
140.007(1.62)0.016(0.78)0.003(0.83)−0.001(−0.03)
15−0.001(−0.34)0.015(0.70)−0.004(−1.17)−0.005(−0.26)
160.004(0.95)0.019(0.86)−0.001(−0.09)−0.006(−0.27)
170.002(0.40)0.021(0.88)−0.006(−1.50)−0.012(−0.53)
180.004(1.18)0.025(1.09)0.005(1.25)−0.007(−0.32)
190.006(1.29)0.031(1.37)−0.003(−0.71)−0.010(−0.44)
200.005(1.23)0.036(1.45)0.005(1.23)−0.005(−0.19)
21−0.002(−0.55)0.034(1.30)0.006(1.31)0.001(0.04)
22−0.007(−1.38)0.027(0.98)−0.006(−1.30)−0.005(−0.18)
23−0.004(−1.01)0.023(0.88)−0.004(−0.91)−0.009(−0.34)
240.001(0.25)0.024(0.87)−0.006(−1.37)−0.015(−0.54)
25−0.001(−0.22)0.023(0.80)−0.002(−0.34)−0.017(−0.57)
26−0.003(−0.67)0.020(0.67)−0.004(−1.01)−0.021(−0.71)
27−0.003(−0.68)0.017(0.59)−0.002(−0.38)−0.023(−0.79)
28−0.004(−0.94)0.013(0.46)0.001(0.06)−0.022(−0.78)
29−0.002(−0.65)0.011(0.34)0.000(0.03)−0.022(−0.73)
300.003(0.78)0.014(0.45)0.002(0.46)−0.020(−0.67)
310.005(1.17)0.019(0.59)0.007(1.54)−0.013(−0.41)
32−0.005(−0.99)0.014(0.41)−0.006(−1.18)−0.019(−0.55)
33−0.005(−1.03)0.009(0.27)−0.007(−1.45)−0.026(−0.75)
34−0.006(−1.28)0.003(0.08)−0.006(−1.10)−0.032(−0.87)
35−0.005(−0.96)−0.002(−0.05)−0.005(−1.12)−0.037(−1.06)
36−0.007(−1.42)−0.009(−0.23)−0.007(−1.51)−0.044(−1.23)

3.2. Analyses of Event-time Results

This subsection contains a detailed discussion of the long-run performance of Chinese IPOs. First, the equal-weighted BHAR are significantly positive over 36 months after listing, clearly implying long-run overperformance in China’s stock market Our results are in stark contrast to the long-run underperformance typically found in developed markets (see e.g. Ritter, 1991; Levis, 1993; Loughran and Ritter, 1995). One possible explanation for this disparity could be survivorship bias. However, an in-depth inspection of all IPOs in our sample reveals that only two IPO firms were delisted or suspended within 3 years after listing.10 Therefore, the results in this study are likely free from any survivorship bias and the fact that almost all IPO firms did survive for more than 3 years after listing also supports the finding of long-run overperformance of Chinese IPOs. In contrast, approximately 29% of 3361 US IPOs included in the sample examined by Gompers and Lerner (2003) were delisted prior to their third anniversary. Moreover, Bhabra and Pettway (2003) report that US IPO firms delisted due to financial distress show significant underperformance compared with surviving IPO firms.

Tables 3 and 4 highlight another important feature of both BHAR and CAR that is related to the weighting schemes. For example, the equal-weighted BHAR are significantly positive in the long run, whereas the value-weighted returns are much lower and insignificant. Although both equal-weighted and value-weighted CAR are insignificant in the long run, the equal-weighted CAR are much higher than the value-weighted CAR. Therefore, the equal-weighted returns are generally greater than the value-weighted returns, suggesting that the performance of small-size IPO firms is superior to that of big-size IPO firms. Our results provide out-of-sample evidence supporting Fama’s (1998, p. 296) view that apparent anomalies typically “shrink a lot and often disappear when event firms are value weighted rather than equal weighted.”

To confirm the influence of the size effect on long-run performance, we categorize all 936 IPOs in our sample into size quintiles based on gross proceeds, adjusted using the CPI (2005 = 100). The equal-weighted BHAR over 3 years after listing presented in Table 5 clearly show that IPOs in each of the five size groups exhibit long-run overperformance, but that the BHAR decrease with increasing gross proceeds. In particular, IPOs in the smallest size group with gross proceeds of less than RMB176m outperform their size and industry matching firms by 22.5% (t-statistic = 4.95), whereas IPOs in the second smallest size group with gross proceeds between RMB176m and RMB266m outperform their matching firms by 16.7% (t-statistic = 3.99). Although IPOs in the other three groups with gross proceeds of more than RMB266m also slightly outperform their matching firms, the results are statistically insignificant. Table 5 also reports the equal-weighted CAR in each of the five size groups. The result consistently shows that IPOs are strongly tilted toward small-size firms; that is, only IPO firms in the smallest size group generate a significantly positive CAR (9.6%; t-statistic = 2.33).

Table 5.   Post-issue buy-and-hold returns (BHAR) and cumulative abnormal returns (CAR) on Chinese IPOs, 1996–2005, sorted by size quintiles The sample consists of 936 Chinese initial public offerings (IPOs) over the period 1996 to 2005. All IPOs are sorted into size quintiles by gross proceeds. This table presents the equal-weighted BHAR and CAR over 36 months after listing, based on the size and industry matching firm benchmark. The gross proceeds are presented in millions of RMB, adjusted using the consumer price index (2005 = 100). US$1 was approximately RMB8.07 on 31 December 2005. The bootstrapped skewness-adjusted t-statistics are used to test for the significant difference in buy-and-hold returns between IPOs and their matching firms. The t-statistics are presented in parentheses. *** and ** denote statistical significance at the 1 and 5% level, respectively.
Gross proceeds quintilesNBHARt(BHAR)CARt(CAR)
Small (RMB34 < Gross proceeds < RMB176)1870.225(4.95)***0.096(2.33)**
2 (RMB176 < Gross proceeds < RMB266)1870.167(3.99)***–0.043(–0.57)
3 (RMB266 < Gross proceeds < RMB366)1880.010(0.41)–0.029(–0.45)
4 (RMB366 < Gross proceeds < RMB546)1870.048(1.05)–0.027(–0.47)
Big (RMB546 < Gross proceeds < RMB12 546)1870.002(0.01)–0.040(–0.87)
Full sample9360.086(3.57)***–0.009(–0.23)

Compared to big-size firms in China’s stock market, small-size firms, which publish relatively less information to investors before going public, normally represent less established and/or more risky firms. For example, a majority of IPO firms classified in the IT industry category, which covers high-technology and Internet-related firms, are young and generally small. These firms are expected to enjoy higher growth in the near future, but also to face greater risk compared to traditional industry firms. Specifically, 57 IPO firms classified in the IT industry category generate an equal-weighted BHAR of 59.5% (t-statistic = 2.91) over 3 years after listing, based on their matching firm benchmark (see Panel B of Table 10). However, the small-size firm stocks are more vulnerable to the price impact associated with institutional trading (Lakonishok et al., 1992); that is, the small-size IPO firms are more likely to be subject to price manipulation because it is easier for institutional investors to control and possibly push their share price upward, thus leading to greater speculation and uncertainty as to their future price performance (Chi and Padgett, 2005b; Aggarwal and Wu, 2006). In addition, Pastor and Veronesi (2003) argue that these small-size firms face substantial uncertainty regarding future profitability, which contributes to their high value because it raises expected future payoffs without affecting the discount rate.

3.3. Robustness Tests of Event-time Results

Cai et al. (2008) examine the long-run performance of 335 IPOs listed on the SHSE over the period 1997–2001. They report significantly negative equal-weighted BHAR and CAR relative to the market benchmark over 3 years after listing, which is different from our results based on the matching firm benchmark. They show that the 24-month equal-weighted market-adjusted BHAR and CAR are −19.60% (t-statistic = −7.45) and −10.88% (t-statistic = −5.72), respectively, while the 36-month equal-weighted BHAR and CAR are −29.57% (t-statistic = −15.27) and −24.97% (t-statistic = −10.89), respectively.

We conjecture that this disparity in results is due to the different benchmarks applied. To rule out the influence of time period and sample differences, we re-examine the long-run performance of IPOs relative to the size and industry matching firm benchmark, using the same time period and sample used by Cai et al. (2008). Table 6 shows a significant long-run overperformance for equal-weighted matching-firm-adjusted BHAR over the period 1997–2001, although not for other event-time approaches, consistent with our previous findings over the entire sample period. Therefore, it is highly likely that the benchmark employed plays an important role in the results.

Table 6.   Post-issue buy-and-hold abnormal returns (BHAR) and cumulative abnormal returns (CAR) on Chinese IPOs, 1997–2001 The sample consists of 319 Chinese initial public offerings (IPOs) listed on the Shanghai Stock Exchange (SHSE) over the period 1997 to 2001. This table presents the equal-weighted and value-weighted BHAR and CAR for the 3–36 months after listing based on the size and industry matching firm benchmark. The bootstrapped skewness-adjusted t-statistics are used to test for the significant difference in buy-and-hold returns between IPOs and their matching firms. The equal-weighted BHAR and CAR for the 3–36 months after listing, relative to the return on the SHSE A-Share Index, are extracted from tables 3 and 4 of Cai et al. (2008). The t-statistics are presented in parentheses. *** and ** indicate statistical significance at the 1 and 5% level, respectively.
MonthSize and industry matching firm benchmarkMarket benchmark (Cai et al., 2008)
Equal-weightedValue-weightedEqual-weighted
Returnst-statisticReturnst-statisticReturnst-statistic
Panel A: BHARt
 3–0.003(–0.08)–0.005(–0.36)–0.120(–4.57)***
 60.001(0.34)–0.003(–0.08)–0.104(–4.01)***
 90.017(0.78)0.014(0.91)–0.129(–4.95)***
120.025(1.09)0.021(1.13)–0.125(–4.69)***
150.043(1.59)0.041(1.55)–0.153(–5.47)***
180.044(2.15)**0.041(1.47)–0.177(–6.54)***
210.050(2.39)**0.032(1.26)–0.176(–6.41)***
240.060(2.46)**0.041(1.32)–0.196(–7.45)***
270.072(2.80)***0.050(1.48)–0.225(–9.06)***
300.073(3.02)***0.051(1.49)–0.265(–11.28)***
330.077(3.09)***0.051(1.46)–0.279(–13.46)***
360.087(3.54)***0.048(1.56)–0.296(–15.27)***
Panel B: CARt
 30.003(0.20)–0.004(–0.26)–0.001(–0.09)
 60.021(1.10)0.018(1.01)0.012(1.21)
 90.019(0.90)0.017(0.89)–0.002(–0.18)
120.020(0.81)0.013(0.57)–0.011(–0.85)
150.029(1.15)0.022(1.03)–0.023(–1.52)
180.024(0.85)0.015(0.59)–0.040(–2.52)**
210.042(1.26)0.033(1.10)–0.057(–3.26)***
240.020(0.62)0.009(0.31)–0.109(–5.72)***
270.008(0.21)–0.008(–0.24)–0.145(–7.25)***
30–0.007(–0.18)–0.013(–0.39)–0.175(–8.06)***
33–0.007(–0.16)–0.020(–0.54)–0.211(–9.57)***
36–0.008(–0.21)–0.015(–0.44)–0.250(–10.89)***

To further confirm the sensitivity of the choice of benchmarks, we compare the long-run performance of another subsample that includes 246 IPOs listed on the SHSE or SZSE over the period 2002 to 2005 on the basis of both the market and matching firm benchmarks. Like Cai et al. (2008), we use the appropriate index to benchmark each individual IPO firm’s performance depending on the stock exchange of listing. The SHSE or the SZSE A-Share Index provides the benchmark for IPOs listed on the SHSE or SZSE, respectively. Consistent with the results found by Cai et al. (2008), a significant market-adjusted long-run underperformance is found, as shown in Table 7. Specifically, the 24-month equal-weighted market-adjusted BHAR and CAR are −17.2% (t-statistic = −5.88) and −14.4% (t-statistic = −9.39), respectively, while the 36-month equal-weighted BHAR and CAR are −24.2% (t-statistic = −10.02) and −22.0% (t-statistic = −12.04), respectively. However, when the matching firm benchmark is used, only the equal-weighted BHAR are significantly positive and no long-run overperformance or underperformance is found using other event-time approaches, confirming our previous result that the long-run performance of Chinese IPOs is sensitive to the choice of methods and benchmarks.

Table 7.   Post-issue buy-and-hold abnormal returns (BHAR) and cumulative abnormal returns (CAR) on Chinese IPOs, 2002–2005 The sample consists of 246 Chinese initial public offerings (IPOs) over the period 2002 to 2005. This table presents the equal-weighted and value-weighted BHAR and CAR for the 3–36 months after listing based on the size and industry matching firm benchmark. The bootstrapped skewness-adjusted t-statistics are used to test for the significant difference in buy-and-hold returns between IPOs and their matching firms. This table also presents the equal-weighted BHAR and CAR for the 3–36 months after listing based on the market benchmark. The Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) A-Share Indexes provide benchmarks for 197 IPOs listed on the SHSE and 49 IPOs listed on the SZSE, respectively. The t-statistics are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
MonthSize and industry matching firm benchmarkMarket benchmark
Equal-weightedValue-weightedEqual-weighted
Returnst-statisticReturnst-statisticReturnst-statistic
Panel A: BHARt
 3–0.044(–2.15)**–0.047(–2.23)**–0.028(–2.98)***
 6–0.045(–2.46)**–0.048(–2.29)**–0.045(–3.36)***
 9–0.052(–2.58)***–0.056(–2.40)**–0.073(–3.63)***
12–0.039(–1.11)–0.043(–1.74)*–0.075(–3.85)***
15–0.011(–0.13)–0.018(–0.21)–0.076(–3.89)***
180.015(0.68)0.005(0.18)–0.098(–4.70)***
210.025(1.07)0.011(0.43)–0.127(–5.18)***
240.041(2.04)**0.016(0.92)–0.172(–5.88)***
270.054(2.46)**0.021(1.24)–0.198(–6.92)***
300.064(2.73)***0.024(1.36)–0.214(–8.18)***
330.073(3.27)***0.028(1.48)–0.241(–9.49)***
360.095(3.75)***0.033(1.59)–0.242(–10.02)***
Panel B: CARt
 3–0.045(–2.14)**–0.045(–2.18)**–0.032(–2.51)**
 6–0.055(–2.37)**–0.058(–2.83)***–0.049(–3.22)***
 9–0.061(–2.48)**–0.069(–2.96)***–0.079(–4.22)***
12–0.041(–1.50)–0.049(–1.84)*–0.085(–5.15)***
15–0.031(–0.71)–0.034(–1.52)–0.087(–5.77)***
18–0.027(–0.62)–0.038(–1.48)–0.104(–6.89)***
21–0.014(–0.26)–0.025(–0.90)–0.126(–7.84)***
24–0.022(–0.36)–0.033(–0.59)–0.144(–9.39)***
27–0.017(–0.27)–0.020(–0.35)–0.161(–10.56)***
30–0.009(–0.14)–0.012(–0.16)–0.200(–11.64)***
33–0.027(–0.34)–0.032(–0.30)–0.211(–11.91)***
36–0.024(–0.27)–0.029(–0.21)–0.220(–12.04)***

3.4. Calendar-time Results

Similar to Gompers and Lerner (2003), we first examine the return of IPO portfolios for each calendar month from January 1998 to December 2006 (108 months), during which time the performance is calculated for monthly returns on the portfolio containing firms that went public during the (t − 36, t − 13) period, adjusted by the size and industry matching firm benchmark. Table 8 reports the equal-weighted and value-weighted mean monthly AR in each calendar year from 1998 to 2006. However, the equal-weighted and value-weighted mean and median returns for all 108 months are statistically insignificant. For example, the equal-weighted and value-weighted mean returns are −0.02% (t-statistic = −0.09) and −0.08% (t-statistic = −0.36), respectively.

Table 8.   Monthly calendar-time abnormal returns on initial public offering (IPO) portfolios, 1998–2006 The sample consists of 936 Chinese IPOs over the period 1998 to 2006. The monthly abnormal returns are compared with the size and industry matching firm benchmark. Each month t from January 1998 to December 2006, the returns on the portfolio including IPOs that went public during the (t − 36, t − 13) period are calculated. The returns are either equal-weighted or value-weighted by the market value using the closing price on the first day of trading multiplied by shares outstanding, adjusted using the consumer price index (2005 = 100). This table presents the average monthly number of IPOs and the average monthly returns on the IPO portfolio in each year from 1998 to 2006. This table also presents the mean and median monthly returns over the entire 108 months. The t-statistics are presented in parentheses.
YearNEqual-weightedValue-weighted
1998220.90.00730.0024
1999304.1–0.00170.0006
2000212.2–0.0044–0.0046
2001189.50.00090.0007
2002221.7–0.0091–0.0086
2003179.0–0.0045–0.0027
2004122.30.00440.0011
2005155.10.00270.0025
2006144.40.00240.0018
Mean –0.0002–0.0008
t-statistic (–0.09)(–0.36)
Median –0.0031–0.0076
t-statistic (–0.19)(–0.47)

Gompers and Lerner (2003) argue that if IPOs outperform on a risk-adjusted basis, the time-series portfolios of IPOs should consistently outperform relative to an explicit asset pricing model. Therefore, we use the intercepts derived from the CAPM and the Fama and French (1993) three-factor model as indicators of risk-adjusted performance. The intercepts derived from both one-factor and three-factor OLS regressions, presented in Panel A of Table 9, are insignificantly different from zero, suggesting that IPOs do not appear to exhibit significant over- or underperformance on a calendar-time basis.

Table 9.   Capital asset pricing model (CAPM) and Fama–French (1993) three-factor regressions on IPO portfolios, 1998–2006 The sample consists of 936 Chinese initial public offerings (IPOs) over the period 1998 to 2006. The portfolio in each month t from January 1998 to December 2006 includes IPOs that went public during the (t – 36, t – 13) period. Our regression uses the difference between the monthly returns on a portfolio of IPOs and the risk-free rate (Rp,t – Rf,t) as the dependent variable. The monthly returns on the IPO portfolios are regressed on (Rm,t – Rf,t) for the CAPM and on (Rm,t – Rf,t), SMBt, and HMLt for the Fama and French (1993) three-factor model, where Rp,t represents the monthly return on the IPO portfolio; Rf,t represents the monthly return on the 3-month household deposit interest rate; Rm,t represents the monthly return on the value-weighted Shanghai Stock Exchange and the Shenzhen Stock Exchange A-Share Index; SMBt represents the monthly return on small-size firms minus the monthly return on big-size firms; and HMLt represents the monthly return on high- book-to-market ratio (B/M) firms minus the monthly return on low-B/M firms. Panel A of this table estimates the OLS regression; Panel B uses the weighted least squares regression, where observations are weighted by the square root of the number of IPOs in the portfolio in a given month. The t-statistics are presented in parentheses. *** denotes significance at the 1% level.
 CAPMFama and French (1993) three-factor model
Equal-weightedValue-weightedEqual-weightedValue-weighted
Panel A: OLS regressions
Intercept–0.0015 (–0.73)0.0000 (0.00)–0.0012 (–0.78)0.0004 (0.21)
Rm,t – Rf,t0.9402 (17.70)***0.8839 (17.46)***0.9442 (21.75)***0.8925 (17.01)***
SMBt  0.3345 (5.30)***0.0673 (0.90)
HMLt  –0.1672 (–3.12)***0.0142 (0.21)
Adjusted R20.9110.9050.9420.905
Panel B: Weighted least squares regressions (weights equal to square root of the number of IPOs in portfolio)
Intercept–0.0004 (–0.24)0.0003 (0.20)–0.0007 (–0.45)0.0008 (0.43)
Rm,t – Rf,t0.9627 (29.27)***0.9254 (25.66)***0.9714 (28.20)***0.9366 (23.56)***
SMBt  0.2999 (5.73)***0.0464 (0.74)
HMLt  –0.1488 (–2.89)***0.0318 (0.51)
Adjusted R20.9220.9350.9560.935

However, given that the CAPM and the Fama and French (1993) three-factor OLS regressions weigh each month equally, any over- or underperformance may be reduced if it is correlated with the number of IPOs in the portfolio. We thus also estimate the weighted least squares (WLS) regressions using the square root of the number of IPOs in the portfolio as weights. The results from both WLS regressions, presented in Panel B of Table 9, are very similar to those obtained using the OLS regressions: the intercepts are not significantly different from zero. Therefore, we conclude that there is no overperformance or underperformance of Chinese IPOs on a calendar-time basis.

4. Cross-sectional Results and Analyses

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

To explore potential sources, and trace the influence, of firm-specific and institutional characteristics on the long-run abnormal performance of Chinese IPOs, we examine a series of proxies for the testable signaling and ex-ante uncertainty hypotheses. Table 10 presents the cross-sectional patterns of long-run performance of Chinese IPOs over the period 1996–2005. We report the results of the cross-sectional analyses based on the 36-month equal-weighted matching-firm-adjusted BHAR because other approaches produce neither long-run overperformance nor underperformance (see Tables 3 and 4). In Table 10, column 1 explains the standard classifications of each category; column 2 provides the number of IPOs in each category; columns 3 and 4 report the buy-and-hold returns on IPO firms and their matching firms, respectively; column 5 shows the BHAR, which are calculated as the difference in buy-and-hold returns between IPO firms and their size and industry matching firms; and column 6 presents the bootstrapped skewness-adjusted t-statistics to test for the statistical significance of BHAR.

Table 10.   Long-run performance of 936 Chinese IPOs sorted by signaling and ex-ante uncertainty characteristics The sample consists of 936 Chinese initial public offerings (IPOs) over the period 1996 to 2005. In Panel A, all IPOs are sorted by proxies for the signaling characteristics: (i) the initial return, which is defined as the percentage difference between the offering price and the closing price on the first day of trading, relative to the contemporaneous return on the market index; (ii) EPS prior to issue, which is measured by the average earnings per share for the last 3 years before listing; (iii) the standard deviation of the first 242-trading-day returns; and (iv) the number of seasoned equity offerings (SEOs) within the first 3 years after listing. In Panel B, all IPOs are sorted by proxies for the ex-ante uncertainty characteristics: (i) offering size, which is calculated as the offering price multiplied by the number of shares sold to the public, adjusted using the consumer price index (2005 = 100); (ii) association with the information technology (IT) industry category; (iii) time lag between the offering and listing dates, which is defined as the number of days between the offering date and the first day of listing; and (iv) government and institutional ownership, which is defined as the proportion of state-owned and institutional shares in the total ordinary shares at issue. This table presents the equal-weighted BHAR as well as buy-and-hold returns (BHR) for IPOs and their matching firms over 36 months after listing, based on the size and industry matching firm benchmark, in each category. The bootstrapped skewness-adjusted t-statistics presented in parentheses are used to test for the significant difference in BHR between IPOs and their matching firms. *** and * denote statistical significance at the 1 and 10% level, respectively.
 NRaw BHRMatching firm BHRBHARt(BHAR)
Panel A: Proxies for signaling characteristics
Initial return
 Low 30%2810.8720.7350.137(1.54)
 Mid 40%3740.4630.3690.094(1.33)
 High 30%2810.1380.1140.024(0.65)
EPS prior to issue
 Low 30%2810.0810.120–0.039(–0.45)
 Mid 40%3740.5210.4410.080(1.94)*
 High 30%2810.8240.6050.219(3.43)***
Standard deviation of the first 242-trading-day returns
 Low4680.4760.3180.158(2.85)***
 High4680.4840.4700.014(0.43)
Number of SEOs within the first 3 years after listing
 Non-SEO6400.2070.224–0.017(–0.09)
 SEO2961.0700.7610.309(4.12)***
Panel B: Proxies for ex-ante uncertainty characteristics
Offering size
 Small 30%2811.1200.9240.196(3.83)***
 Mid 40%3740.3410.3070.034(1.16)
 Big 30%2810.025–0.0210.046(1.29)
Association with IT industry category
 IT570.9180.3240.595(2.91)***
 Non-IT8790.4520.3980.053(1.38)
Time lag between the offering and listing dates
 0–14 days1100.5820.5400.042(0.38)
 15–21 days4080.5350.4280.107(1.47)
 22–28 days1480.3640.2340.130(1.63)
 29 days and more2700.4190.3710.048(0.87)
Government and institutional ownership
 Low 30%2810.6470.4860.161(1.64)
 Mid 40%3740.4370.4030.034(0.65)
 High 30%2810.3700.2900.080(1.56)

4.1. Long-run Performance of Initial Public Offerings Categorized by Signaling Characteristics

The signaling hypothesis developed by Allen and Faulhaber (1989) and Welch (1989) assumes that good-quality firms attempt to distinguish themselves from bad-quality firms by underpricing new issues, which is a cost that bad-quality firms cannot absorb. Models of the signaling hypothesis imply that although good-quality firms are initially underpriced to a greater degree, they will be rewarded by SEOs in the near future. The prediction from this hypothesis is of a positive relationship between the long-run performance of IPOs and the quality of IPO firms. In the present study, we use the following proxies for signaling characteristics: (i) the initial return, which is defined as the percentage difference between the offering price and the closing price on the first day of trading, relative to the contemporaneous return on the market index; the SHSE and SZSE A-Share Indexes provide the benchmark for IPOs listed on the SHSE and SZSE, respectively; (ii) the EPS prior to issue, which is measured by the average EPS for the 3 years before listing; (iii) the standard deviation of the first 242-trading-day returns; and (iv) the number of SEOs within the first 3 years after listing.

We first sort the 936 IPO firms into three groups, low, medium, or high, according to whether their initial returns are included in the lowest 30, middle 40, or highest 30 percentile, respectively. Panel A of Table 10 shows that all three groups generate some statistically insignificant long-run overperformance. An analysis of variance (anova) test further shows that none of the three groups has a significantly different BHAR from the others (F-statistic = 0.41), implying that the initial return is not an effective signal in predicting long-run performance of Chinese IPOs.

All IPO firms are also sorted into three groups based on the value of EPS prior to issue, low, medium, or high, according to whether the value of EPS is included in the lowest 30, middle 40, or highest 30 percentile, respectively. Panel A of Table 10 shows that the 281 IPO firms in the low group with EPS of less than RMB0.280 slightly underperform their matching firms by −3.9% (t-statistic = –0.45), that the medium group generates a weakly significant overperformance of 8.0% (t-statistic = 1.94), and that the 281 IPO firms in the high group with EPS in excess of RMB0.434 significantly outperform their matching firms by 21.9% (t-statistic = 3.43). Furthermore, an anova test shows that at least one of the three EPS groups has a significantly different BHAR from the others (F-statistic = 3.82).

Following Su (2004), we define a high-quality (low-quality) firm as one with a low (high) standard deviation of the first 242-trading-day returns, and then divide all IPOs into two equal groups. Panel A of Table 10 shows that the high-quality group generates a significantly positive BHAR of 15.8% (t-statistic = 2.85), whereas the low-quality group generates an insignificant BHAR of 1.4% (t-statistic = 0.43). The difference in BHAR between the two groups is statistically significant (t-statistic = 2.01).

Finally, we sort all IPO firms into two groups according to the number of SEOs within the first 3 years after listing: the non-SEO group contains 640 IPO firms making no SEO and the SEO group contains 296 IPO firms making at least one SEO. The results presented in Panel A of Table 10 show that the non-SEO group generates an insignificantly negative BHAR of −1.7% (t-statistic = −0.09), whereas the SEO group generates a significantly positive BHAR of 30.9% (t-statistic = 4.12). The difference in BHAR between the two groups is statistically significant (t-statistic = 3.26).

In sum, the empirical evidence suggests that IPO long-run performance is positively related to the quality of the IPO firms, which supports the signaling hypothesis that the market receives efficient signals distinguishing good-quality IPO firms from bad-quality IPO firms in the long run, because good-quality IPO firms normally have higher EPS prior to issue, experience less price fluctuation in the long run, and have more opportunities to issue SEOs in the near future after listing.

4.2. Long-run Performance of Initial Public Offerings Categorized by Ex-ante Uncertainty Characteristics

The divergence of opinion hypothesis originally proposed by Miller (1977) assumes that differences in opinions between optimistic and pessimistic investors can lead to asset overvaluation and subsequent underperformance. Given ex-ante uncertainty about the value of IPO firms, optimistic investors are more likely to reveal their opinion by actively purchasing shares, whereas pessimistic investors mostly stay out of the market. If true, this means that share prices will be biased upward as long as there are divergent opinions and sufficient demand in the market. The divergence of expectations will narrow and prices will consequently adjust downward with the spread of information. This divergence of opinion hypothesis predicts that there will be a negative relationship between the long-run performance of IPOs and ex-ante uncertainty. To test for this, we use a number of proxies for ex-ante uncertainty surrounding IPO events: (i) the offering size, which is calculated as the offering price multiplied by the number of shares sold to the public, adjusted using the CPI (2005 = 100); (ii) the association with IT industry category; (iii) the time lag between the offering and listing dates, which is defined as the number of days between the offering date and the first day of listing; and (iv) the government and institutional ownership, which is defined as the proportion of state-owned and institutional shares in total ordinary shares at issue.

All 936 IPOs are first sorted into three size groups, small, medium, or big, according to whether the offering size is included in the smallest 30, middle 40, or biggest 30 percentile, respectively. The small-size group consists of 281 IPO firms with an offering size of less than RMB222.34m; the big-size group consists of 281 IPO firms with an offering size in excess of RMB443.37m. Panel B of Table 10 shows the long-run overperformance of all three groups: 19.6% (t-statistic = 3.83) for the small-size group, 3.4% (t-statistic = 1.16) for the medium-size group, and 4.6% (t-statistic = 1.29) for the big-size group. The result that small-size IPO firms generate relatively high long-run performance is consistent with our earlier results (see Table 5). Further analysis based on an anova test shows that at least one group has a significantly different BHAR from the others (F-statistic = 2.19).

Following Ritter (1991), we also investigate industry effect on long-run performance. We examine the difference in BHAR between 57 IPO firms classified in the IT category and 879 IPO firms classified in other industry categories. Panel B of Table 10 shows that IPO firms classified in the IT industry category produce a significantly positive BHAR of 59.5% (t-statistic = 2.91), whereas IPO firms in non-IT industry categories produce an insignificantly positive BHAR of 5.3% (t-statistic = 1.38). The difference in BHAR between the two groups is statistically significant (t-statistic = 2.12).

An important characteristic of China’s IPO market is the longer time lag between the offering and listing dates of A-share IPOs compared with a relatively short time lag in developed markets, implying more ex-ante uncertainty for successful subscribers (Chen et al., 2000). We sort all IPOs into four groups based on the number of days between the offering date and the first day of trading: Group 1 (0–14 days), Group 2 (15–21 days), Group 3 (22–28 days), and Group 4 (more than 28 days). In Panel B of Table 10 we find some long-run overperformance in all four groups, but none of it is statistically significant. An anova test shows no significant difference in BHAR among the four groups (F-statistic = 0.88).

A unique characteristic of China’s stock market is the existence of state-owned and institutional shares that cannot be publicly traded (see e.g. Chen et al., 2000; Chang et al., 2010).11 To examine the impact of ownership structure on IPO long-run performance, we sort all IPO firms into three groups based on the proportion of state-owned and institutional shares at issue: 281 IPO firms with this proportion higher (lower) than 68.75% (55.72%) comprise the high (low) ownership group, while the middle 374 IPO firms form the medium ownership group. Panel B of Table 10 shows some insignificant long-run overperformance in all three groups: 8.0% (t-statistic = 1.56), 3.4% (t-statistic = 0.65), and 16.1% (t-statistic = 1.64) for the high, medium, and low ownership groups, respectively. An anova test shows insignificant differences in BHAR among the three groups (F-statistic = 1.08).

In sum, empirical evidence is inconsistent with the divergence of opinion hypothesis, which expects a negative relationship between IPO long-run performance and ex-ante uncertainty. We find that small-size IPO firms or IPO firms classified in the IT industry category, which are usually less established and/or more risky, are more likely to generate higher long-run returns. In addition, no significant evidence suggests a positive or negative relationship between IPO long-run performance and the time lag between offering and listing dates or the degree of government and institutional ownership.

4.3. Results and Analyses of Cross-sectional OLS Regressions

Finally, in order to identify possible explanations for the long-run performance of Chinese IPOs discussed in Table 10, we carry out two cross-sectional OLS regressions using the 36-month equal-weighted BHAR and CAR, based on the size and industry matching firm benchmark, as dependent variables:

  • image(1)

where IR represents IPO initial returns; EPS represents IPO firms’ EPS prior to issue; SD represents the standard deviation of the first 242-trading-day returns, excluding the initial returns; SEO is a dummy variable set to 1 if the IPO firm makes at least one SEO within the first 3 years after listing, and 0 otherwise; LnSize represents the natural log of the offering size; IT is a dummy variable set to 1 if the IPO firm is classified in the IT category, and 0 otherwise; LnLag represents the natural log of days between the offering and listing dates; State represents the proportion of state-owned and institutional shares at issue; and Exg is a dummy variable set to 1 if the IPO firm is listed on the SHSE, and 0 otherwise. Panel A of Table 11 presents the descriptive statistics of major explanatory variables used in the cross-sectional regressions and Panel B reports the Pearson correlations among all explanatory variables. In general, the correlation coefficients are not high enough to give rise to a multicollinearity problem in the cross-sectional regressions.12

Table 11.   Descriptive statistics and Pearson correlations among major explanatory variables The sample consists of 936 Chinese initial public offerings (IPOs) listed on the Shanghai Stock Exchange and the Shenzhen Stock Exchange A-Share Index over the period 1996–2005. Panel A of this table presents descriptive statistics of the explanatory variables used in the OLS regression analysis of the long-run performance. IR represents the initial returns; EPS represents EPS prior to issue; SD presents the standard deviation of the first 242-trading-day returns; (SEO (seasoned equity offering) represents the number of SEOs within the first 3 years after listing; LnSize represents the natural log of the offering size, adjusted using the consumer price index (2005 = 100); IT represents that the IPO firm is classified in the information technology industry category; LnLag represents the natural log of the time lag between the offering and listing dates; State represents the proportion of state-owned and institutional shares in the whole ordinary shares at issue; and Exg represents the exchange of listing. Panel B presents Pearson correlation coefficients of major explanatory variables.
 Proxies for signaling characteristicsProxies for ex-ante uncertainty characteristics
IREPSSDSEOLnSizeITLnLagState
Panel A: Descriptive statistics
Mean1.2290.4100.0270.32110.3480.0603.1630.168
Median1.0990.3780.0260.00010.3650.0002.9960.000
Maximum8.3021.5440.0522.00013.2921.0005.9450.850
Minimum–0.0900.0740.0110.0008.1510.0002.1970.000
Standard deviation0.8290.1890.0070.4790.7460.2380.5660.253
Panel B: Pearson correlations
EPS0.026       
SD0.020–0.048      
SEO–0.2520.1370.073     
LnSize–0.3560.081–0.082–0.257    
IT0.0770.114–0.019–0.009–0.048   
LnLag0.054–0.1000.0390.0780.145–0.008  
State0.017–0.0990.0330.0530.020–0.0730.119 
Exg–0.036–0.075–0.072–0.2020.1960.033–0.210–0.095

The results presented in Table 12 generally confirm the results shown in Table 10.13Table 12 first shows a significantly positive coefficient of EPS (t-statistic = 2.56), which is inconsistent with Cai et al. (2008), who report a negative relationship between EPS prior to issue and the long-run performance of Chinese IPOs and attribute the long-run underperformance they find to manipulation of earnings. The discrepancy might be due to the different time periods as well as sample composition and/or the different benchmarks used. Table 12 also reports a significantly negative coefficient of SD (t-statistic = −2.10) and a significantly positive coefficient of SEO (t-statistic = 1.81), consistent with Chen et al. (2000). The result is robust to the alternative definitions of SEO proposed by Yu and Tse (2006). The above results generally support the signaling hypothesis that good-quality IPO firms with higher EPS prior to issue tend to experience less price fluctuation after listing and to have more SEO opportunities in the near future after listing. However, we find an insignificantly negative coefficient of IR (t-statistic = −1.05), which is not strongly supportive of the previous findings of a significantly negative relationship between IPO underpricing and long-run performance (see e.g. Chi and Padgett, 2005b; Cai et al., 2008). Su and Bangassa (2011) argue that insufficient supply of and high demand for new issues and/or the previous record of large initial returns attract a great many investors in pursuit of the risk-free profits of Chinese IPOs without concern for the quality of the IPO firms. Therefore, initial returns of IPOs are not an effective signal to predict the long-run performance of Chinese IPOs.

Table 12.   Results of cross-sectional OLS regressions The sample consists of 936 Chinese initial public offerings (IPOs) listed on the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange over the period 1996 to 2005. This table presents the results of two cross-sectional OLS regressions as follows: BHARsiorCARsi = α0 + β1IRi + β2EPSi + β3SDi + β4SEOi + β5LnSizei + β6ITi + β7LnLagi + β8Statei + β9Exgi + εi, where BHAR and CAR represent the 36-month equal-weighted BHAR and CAR, respectively, based on the size and industry matching firm benchmark; IR represents the initial returns; EPS represents EPS prior to issue; SD presents the standard deviation of the first 242-trading-day returns; SEO (seasoned equity offerings) is a dummy variable set to 1 if the IPO firm makes at least one SEO within the first 3 years after listing, and 0 otherwise; LnSize represents the natural log of the offering size, adjusted using the consumer price index (CPI) (2005 = 100); IT is a dummy variable set to 1 if the IPO firm is classified in the information technology industry category, and 0 otherwise; LnLag represents the natural log of the time lag between the offering and listing dates; State represents the proportion of state-owned and institutional shares in the whole ordinary shares at issue; and Exg is a dummy variable set to 1 if the IPO firm is listed on the SHSE, and 0 otherwise. The t-statistics are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
VariableEqual-weighted BHAREqual-weighted CAR
Coefficientt-statisticCoefficientt-statistic
Intercept0.0099(0.14)0.0113(0.31)
IR−0.1003(−1.05)−0.0917(−0.81)
EPS0.1564(2.56)***0.1165(2.55)**
SD−0.1244(−2.10)**−0.1050(−1.89)*
SEO0.1961(1.81)*0.1389(2.21)**
LnSize−0.1699(−2.81)***−0.0896(−2.04)**
IT0.3663(2.80)***0.2105(2.39)**
LnLag−0.0498(−0.72)−0.0284(−0.54)
State−0.1064(−0.65)0.0310(0.17)
Exg−0.1042(−2.24)**−0.0692(−1.68)*
Adjusted R20.3663 0.2578 

In contrast to the prediction of the divergence of opinion hypothesis that IPO firms with wider ex-ante uncertainty will have greater long-run underperformance, we find a significantly negative coefficient of LnSize (t-statistic = −2.81), and a significantly positive coefficient of IT (t-statistic = 2.80). These results are in accordance with those reported by Chi and Padgett (2005b) and Chang et al. (2010). We attribute the observed long-run overperformance of small-size IPO firms or those in the IT industry sector to the compensation for higher ex-ante uncertainty. In addition, the estimation result shows that the coefficient of Exg is significantly negative (t-statistic = −2.24), implying that the IPOs of firms listed on the SZSE have relatively higher long-run returns than those listed on the SHSE, which is consistent with the earlier results that the size of IPO firms listed on the SZSE is significantly smaller than that of IPO firms listed on the SHSE (see Subsection 2.1.).

The coefficient of LnLag is negative but insignificant (t-statistic = −0.72), inconsistent with Chen et al. (2000). This conflicting result could be due to different the sample periods used. During the sample period 1992 to 1995 examined by Chen et al. (2000), the average lag between the offering and listing date is 264 days, while the average lag falls to 27 days over the period 1996 to 2005 (Guo and Brooks, 2008). Decline in the average lag between offering and listing dates weakens the impact of the time lag on the long-run performance of IPOs.

We find an insignificantly negative coefficient of State (t-statistic = −0.65), consistent with Chang et al. (2010). However, Chi and Padgett (2005b) and Cai et al. (2008) report a significantly negative and positive relationship, respectively, between government ownership and long-run performance. A number of previous studies examine the influence of ownership structure on stock return in China’s stock market, which is very different from that in developed markets, and the results are mixed. A general consensus regarding China’s stock market behavior is that more inefficient management is inferred for listed firms with a higher proportion of state-owned and institutional shares. For example, Gao (1996) attributes the poor performance of state-owned enterprises to the high degree of government and institutional ownership. Chan et al. (2004) also consider government and institutional ownership to be an indicator of bureaucratic control and operating inefficiency. However, Qi et al. (2000) argue that legal entity and institutional shareholders help monitor, discipline, and motivate managers so that better long-run performance will ensue, which is confirmed by Sun et al. (2002), who provide evidence showing that government ownership is positively related to firm performance in China’s stock market.

5. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix

This study extends previous work on the long-run performance of Chinese IPOs by examining a more recent and comprehensive sample over the period 1996 to 2005. We conclude that evidence concerning the relative long-run performance of IPOs depends on the methods and benchmarks used to measure performance, which is confirmed by the return calculations derived from both equal-weighted and value-weighted BHAR and CAR over 36 months after listing (up to June 2008). A significant overperformance is found using the equal-weighted BHAR based on the size and industry matching firm benchmark. This result is free from survivorship bias and can be attributed to the superior long-run performance of small-size IPOs. The long-run overperformance does not prevail when employing CAR or value-weighted BHAR calculations. Additionally, our evidence of insignificant intercepts for the CAPM and the Fama and French (1993) three-factor regressions indicate no abnormal performance in the long run. The out-of-sample evidence in an emerging market context supports Fama’s (1998) view that the sensitivity of long-run performance depends on the choice of empirical methods and demonstrates that the controversy over mixed results concerning IPO long-run performance is not simply sample and market specific.

Finally, in segmenting the sample by a series of firm-specific and institutional characteristics, we find that IPO performance varies significantly when looking at the number of SEOs within the first 3 years after listing, EPS prior to issue, the standard deviation of post-issue 240-day daily returns, offering size, and association with the IT industry category. Our results are supportive of the signaling hypothesis that higher-quality IPO firms will exhibit higher performance in the long run. However, the results do not support the divergence of opinion hypothesis; we attribute the long-run overperformance to the compensation for ex-ante uncertainty.

Footnotes
  • 1

    Additional institutional features of the China stock market, such as the unique ownership structure and the process of share issue revitalisation, are discussed in Sun and Tong (2000, 2003).

  • 2

    See Guo Tai An (GTA) Information Technology.

  • 3

    The CPI reflects changes in the cost to the average consumer of acquiring a basket of goods and services that may be stable or change in a 1-year interval. The historical data on annual CPI (2005 = 100) in China over the period 1996 to 2005 are collected from the official database of the National Bureau of Statistics of China (NBSC). http://www.stats.gov.cn/english/statisticaldata/yearlydata/.

  • 4

    In 2000, the China Securities Regulatory Commission prepared to set up a Small and Medium Enterprises (SMEs) Board on the SZSE, resulting in the 3-year suspension of IPOs on the SZSE.

  • 5

    The two event-time approaches of BHAR and CAR are widely used by academic researchers to evaluate long-run performance of IPOs. For example, Ritter (1991) uses CAR with monthly portfolio rebalancing to calculate long-run returns, whereas Conrad and Kaul (1993) and Barber and Lyon (1997) argue that a downward bias in CAR may be introduced with the use of independent monthly rebalancing. Loughran and Ritter (1995) calculate BHAR to avoid problems caused by frequent transactions and to reduce the statistical bias in CAR. Fama (1998) and Mitchell and Stafford (2000) further discuss the potential problems of BHAR and CAR. They document that BHAR can magnify the long-run abnormal returns, even if this occurs only in a single period, and that CAR might be a better, less biased method because they eliminate the compounding effect of poor performance in a single year. Gompers and Lerner (2003) argue that BHAR and CAR suffer from a cross-sectional correlation with t-statistics in the event-time tests due to common shocks in the returns of IPO firms.

  • 6

    Barber and Lyon (1997, p. 370) argue that matching firms are the best benchmark for measuring long-run performance, because “by matching sample firms to control firms on specified firm characteristics, we are able to alleviate the new listing bias (since both sample and control firms must be listed in the identified event month), the rebalancing bias (since the returns of the sample and control firms are compounded in an analogous fashion), and the skewness bias (since abnormal returns calculated using this control firm approach are reasonably symmetric).”

  • 7

    The Appendix presents the distribution of all IPOs in our sample based on the Industrial Classification of Listed Companies issued by the China Securities Regulatory Commission (CSRC) as of 2004.

  • 8

    Guo and Brooks (2008) report a mean initial return of 378.41% for 1393 IPOs over the period 1985 to 2005, while only less than 1% of IPOs generate a negative return on the first trading day. Such a significant money effect attracts a large number of investors interested in risk-free profits from the China IPO market. For example, we calculate the winner lottery ratio of 590 IPOs over the period 2001 to 2008, using the number of successful subscribers divided by total valid subscribers. The extremely low winner lottery ratio of 0.26% indicates the greater demand for an IPO compared with the offering size. In addition, the high turnover ratio of 63.95% for all IPOs on the first trading after listing also reflects the fact that a majority of successful subscribers sell their shares in an attempt to obtain a high initial return, resulting in short-run underperformance.

  • 9

    To control for the potential skewness of the distribution of the BHAR, we report the bootstrapping skewness-adjusted t-statistics suggested by Lyon et al. (1999). The conventional t-statistics are not reported for the sake of brevity, but are available upon request.

  • 10

    Huafu Top Dyed Melange Yarn (002040), listed on the SZSE, was suspended for around 8 months within 3 years after listing; Baotou Aluminium (600472), listed on the SHSE, was delisted before its third anniversary.

  • 11

    The ownership structure of Chinese listed firms has some unique features not found in developed markets. New issues in China reflect only a small proportion of outstanding shares, with the majority of shares owned by the government and other legal entities, because most listed firms are state-owned enterprises. In China, ordinary shares of a typical listed firm can generally be classified into two categories: tradable shares and non-tradable shares. Tradable shares include A-shares and B-shares. Non-tradable shares include state-owned shares, which are held by the State Asset Management Bureau; institutional shares, or legal entity shares, which are held by other state-owned enterprises; and employee shares, which are held by employees and cannot be traded until they become tradable A-shares.

  • 12

    To rule out concern over the potential influence of multicollinearity, we calculate the variance inflation factors (VIF). VIF is defined as 1/(1 − R2), where R2 is obtained from the regression of the variable on all other regressors specified in the model. All the multiple regressions yield a value of VIF close to 1.0 or 2.0, far smaller than the commonly accepted threshold of 10 (Neter et al., 1985), indicating no evidence of multicollinearity in this study.

  • 13

    Our results are also robust to the use of the 2-year BHAR or CAR as the dependent variable. Not all results are reported for the sake of brevity, but are available upon request.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix
  • Aggarwal, R., and G. Wu, 2006, Stock market manipulations, Journal of Business 78, pp. 19151953.
  • Allen, F., and G. Faulhaber, 1989, Signaling by underpricing in the IPO market, Journal of Financial Economics 23, pp. 303323.
  • Barber, B., and J. D. Lyon, 1997, Detecting long-run abnormal stock returns: The empirical power and specification of test statistics, Journal of Financial Economics 43, pp. 341372.
  • Bhabra, H., and R. Pettway, 2003, IPO prospectus information and subsequent performance, Financial Review 38, pp. 369397.
  • Cai, X., G. S. Liu, and B. Mase, 2008, The long-run performance of initial public offerings and its determinants: The case of China, Review of Quantitative Finance and Accounting 30, pp. 419432.
  • Chambers, D., and E. Dimson, 2009, IPO underpricing over the very long-run, Journal of Finance 64, pp. 14071443.
  • Chan, K., J. B. Wang, and K. C. Wei, 2004, Underpricing and long-term performance of IPOs in China, Journal of Corporate Finance 10, pp. 409430.
  • Chang, X., S. H. Lin, L. H. K. Tam, and G. Wong, 2010, Cross-sectional determinants of post-IPO stock performance: Evidence from China, Accounting and Finance 50, pp. 581603.
  • Chen, G., M. Firth, and J. B. Kim, 2000, The post-issue market performance of initial public offerings in China’s new stock markets, Review of Quantitative Finance and Accounting 14, pp. 319339.
  • Chen, G., M. Firth, and J. B. Kim, 2004, IPO underpricing in China’s new stock markets, Journal of Multinational Financial Management 14, pp. 283302.
  • Chi, J., and C. Padgett, 2005a, Short-run underpricing and its characteristics in Chinese initial public offering (IPO) markets, Research in International Business and Finance 19, pp. 7193.
  • Chi, J., and C. Padgett, 2005b, The performance and long-run characteristics of the Chinese IPO markets, Pacific Economic Review 10, pp. 451469.
  • Conrad, J., and G. Kaul, 1993, Long-term market overreaction or biases in computed returns? Journal of Finance 48, pp. 3963.
  • Espenlaub, S., A. Gregory, and I. Tonks, 2000, Re-assess the long-term underperformance of UK initial public offerings, European Financial Management 6, pp. 319342.
    Direct Link:
  • Fama, E. F., 1998, Market efficiency, long-term returns, and behavioural finance, Journal of Financial Economics 49, pp. 283306.
  • Fama, E. F., and K. French, 1993, Common risk factors in the returns on bonds and stocks, Journal of Financial Economics 33, pp. 356.
  • Fama, E. F., and K. French, 1996, Multifactor explanation of asset pricing anomalies, Journal of Finance 51, pp. 5584.
  • Gannon, G., and Y. Zhou, 2008, Conflicts of interest and China’s A-share underpricing, International Review of Financial Analysis 17, pp. 491506.
  • Gao, S., 1996, China’s economic reform (St. Martin’s Press, New York; Macmillan Press London).
  • Gompers, P. A., and J. Lerner, 2003, The really long run performance of initial public offerings: The pre-NASDAQ evidence, Journal of Finance 58, pp. 13551392.
  • Guo, H., and R. Brooks, 2008, Underpricing of Chinese A-share IPOs and short-run underperformance under the approval system from 2001 to 2005, International Review of Financial Analysis 17, pp. 984997.
  • Hwang, C. Y., and N. Jayaraman, 1995, Long-term performance of IPOs and non-IPOs: Japanese evidence, Advances in Pacific Basin Financial Markets 1, pp. 317337.
  • Kooli, M., and J. M. Suret, 2004, The aftermarket performance of Canadian IPOs, Journal of Multinational Financial Management 14, pp. 4766.
  • Lakonishok, J., A. Shleifer, and R. W. Vishny, 1992, The impact of institutional trading on stock prices, Journal of Financial Economics 32, pp. 2344.
  • Levis, M., 1993, The long run performance of initial public offerings: The U.K. experience 1980–1988, Financial Management 22, pp. 2841.
  • Loughran, T., and J. R. Ritter, 1995, The new issues puzzle, Journal of Finance 50, pp. 2351.
  • Loughran, T., J. R. Ritter, and K. Rydqvist, 1994, Initial public offerings: International insights, Pacific-Basin Finance Journal 2, pp. 165199.
  • Lyon, J. D., B. Barber, and C. L. Tsai, 1999, Improved methods for tests of long run abnormal stock returns, Journal of Finance 54, pp. 165201.
  • Miller, E. M., 1977, Risk, uncertainty and divergence of opinion, Journal of Finance 32, pp. 11511168.
  • Mitchell, M. L., and E. Stafford, 2000, Managerial decisions and long-term stock price performance, Journal of Business 73, pp. 287329.
  • Mok, H. M. K., and Y. N. Hui, 1998, Underpricing and aftermarket performance of IPOs in Shanghai, China, Pacific-Basin Finance Journal 6, pp. 453474.
  • Neter, J., W. Wasserman, and M. H. Kutner, 1985, Applied linear statistical models (2nd edn), (Richard D. Irwin, Homewood, IL).
  • Pastor, L., and P. Veronesi, 2003, Stock value and learning about profitability, Journal of Finance 58, 17491789.
  • Qi, D., W. Wu, and H. Zhang, 2000, Shareholding structure and corporate performance of partially privatized firms: Evidence from listed Chinese companies, Pacific-Basin Finance Journal 8, pp. 587610.
  • Ritter, J. R., 1991, The long run performance of initial public offerings, Journal of Finance 46, pp. 327.
  • Ritter, J. R., and I. Welch, 2002, A review of IPO activity, pricing and allocations, Journal of Finance 57, pp. 17951828.
  • Su, D., 2004, Leverage, insider ownership, and the underpricing of IPOs in China, International Financial Markets, Institutions and Money 14, pp. 3754.
  • Su, C., and K. Bangassa, 2011, Underpricing and long run performance of Chinese IPOs: The role of underwriter reputation, Financial Markets and Portfolio Management 25, pp. 5374.
  • Su, D., and B. M. Fleisher, 1999, An empirical investigation of underpricing in Chinese IPOs, Pacific-Basin Finance Journal 7, pp. 173202.
  • Sun, Q., and W. H. S. Tong, 2000, The effect of market segmentation on stock prices: The China syndrome, Journal of Banking and Finance 24, pp. 18751902.
  • Sun, Q., and W. H. S. Tong, 2003, China share issue privatization: The extent of its success, Journal of Financial Economics 70, pp. 183222.
  • Sun, Q., W. H. S. Tong, and J. Tong, 2002, How does government ownership affect firm performance? Evidence from China’s privatization experience, Journal of Business Finance & Accounting 29, pp. 127.
    Direct Link:
  • Wang, C., 2004, Relative strength strategies in the China stock market: 1994–2000, Pacific-Basin Finance Journal 12, pp. 159177.
  • Welch, I., 1989, Seasoned offerings, imitation costs, and the underpricing of initial public offerings, Journal of Finance 44, pp. 421449.
  • Yong, O., 2007, A review of IPO research in Asia: What’s next? Pacific-Basin Finance Journal 15, pp. 253275.
  • Yu, T., and Y. K. Tse, 2006, An empirical examination of IPO underpricing in the Chinese A-share market, China Economic Review 17, pp. 363382.

Appendix

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Empirical Results and Analyses
  6. 4. Cross-sectional Results and Analyses
  7. 5. Conclusions
  8. References
  9. Appendix
Table 13.  Distribution of 936 Chinese IPOs by industry category, 1996–2005 The sample consists of 936 Chinese IPOs. This appendix presents the distribution of 936 IPOs listed on the the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) over the period 1996 to 2005 by industry category and the distribution of 576 initial public offerings (IPOs) classified in the manufacturing industry by industry subcategory, according to the “Industry Classification of Listed Companies” issued by the China Securities Regulatory Commission (CSRC) as of 2004, in terms of the number of IPOs and aggregate gross proceeds. The aggregate gross proceeds are presented in millions of RMB, adjusted using the consumer price index (2005 = 100). US$1 was approximately RMB8.07 on 31 December 2005.
Category (Subcategory)SHSESZSEWhole market
NGross proceedsNGross proceedsNGross proceeds
A: Agriculture, forestry, livestock, farming, and fishery239211.09103518.543312 729.63
B: Mining1219 775.3264631.881824 407.20
C: Manufacturing341151 437.1923586 702.67576238 139.86
 C0: Food and beverage3012 880.83166248.994619 129.82
 C1: Textile, clothes, and fur218257.87164428.063712 685.93
 C2: Timber and furniture21015.481209.1631224.64
 C3: Paper making and printing134914.1951441.27186355.46
 C4: Petroleum, chemistry, rubber, and plastics6022 811.245820 361.4011843 172.64
 C5: Electronic 228689.48113998.323312 687.80
 C6: Metal and non-metal5538 474.123819 188.819357 662.93
 C7: Machinery, equipment, and instruments8432 955.456121 094.7814554 050.23
 C8: Medicine and biological products4316 726.23247741.286724 467.51
 C9: Other manufacturing114712.3051990.60166702.90
D: Electric power, gas, and water production and supply2822 882.20134335.704127 217.90
E: Construction189783.2861902.852411 686.13
F: Transport and storage3329 688.06136777.054636 465.11
G: Information technology3926 928.21185192.105732 120.31
H: Wholesale and retail trade288215.84173502.964511 718.80
I: Finance and insurance628 663.872683.88829 347.75
J: Real estate207104.9961558.07268663.06
K: Social services128416.98163446.282811 863.26
L: Communication and cultural industry21734.061492.7632226.82
M: Comprehensive163520.73152351.53315872.26
Full sample578327 361.82358125 096.27936452 458.09