Empirical study on initial public offering (IPO) underpricing and long-run performance: Evidence from China s A-share market

Transcription

1 Afrcan Journal of Busness Management Vol. 7(11), pp , 21 March, 2013 Avalable onlne at DOI: /AJBM ISSN Academc Journals Full Length Research Paper Emprcal study on ntal publc offerng (IPO) underprcng and long-run performance: Evdence from Chna s A-share market Nu Xa*, Song Junyan and Guo Pe College of Economcs and Management, Chna Agrcultural Unversty, Bejng100083, Chna. Accepted 18 June, 2012 hs paper studes the underprcng and long-term performance of A-share ntal publc offerngs (IPOs) n Chna from June 19, 2006 to June 30, he average underprcng of A-share IPOs s 107%, and the average abnormal returns over one and three years after lstng are both sgnfcantly negatve at the 1% level. Consderng both market rsks and stock rsks, abnormal returns over one year cover the nterval [-1.71, -0.54], and abnormal returns over three years cover [-1.54,-0.27]. he underprcng of A-share IPOs s sgnfcantly affected by market volatlty, ex-ante uncertanty, lucky number rato, and frst day tradng rato. In the long run, aftermarket abnormal returns over one year are sgnfcantly determned by market volatlty, ex-ante uncertanty ando a lesser extent, ntal excess returns. Over three years, abnormal returns are drven by market volatlty, offerng prce, ex-ante uncertanty and frst day tradng rato. It s ndcated that the ntal excess returns does not show any nfluences on abnormal returns. Key words: Intal publc offerng (IPO), underprcng, long-run performance, Chna s A-share market. INRODUCION Snce stock exchanges establshed n December 1990, Chna s A-share market (A-share market: radng market of shares lsted n Chna for domestcs wth RMB transactons. B-share market: radng market of shares lsted n Chna for foregners and resdents from Hong Kong, Macau or awan, denomnated n RMB but wth foregn currency transactons) has rased 49.4 trllon Yuan n total. he number of IPO stocks and the amount rased has grown rapdly after full crculaton. Data released by Chna Securtes Regulatory Commsson (CSRC) shows that there were 70 IPOs rasng bllon Yuan n 2006 and 349 IPOs wth bllon Yuan, accompaned by and 211% growth of the IPOs number of and the amount of rased money respectvely. Publc offerng n Chna has become one of the most mportant ways of fnancng for large enterprses, also one of the basc alternatves for nvestors. Most emprcal studes on IPOs n developed and *Correspondng author. E-mal: nuxcau@126.com. el: emergng economes focus on fnancal and operatng performance of lsted frms, relyng on specfc accountng performance measures. Howeverhe analyss of ntal and long-run performance of IPOs n developed countres and certan emergng economes has attracted researchers n the past few years. Yet, lttle s known about the behavor of IPOs n Chna. In ths paper, we frstly examne the performance of the IPOs utlzng more recent sub-perod data from 1993 to 1998 and B-share IPOs from June 19, 2006 to June 30, Another contrbuton mght be that we examne what market factors determned by the characterstcs of Chnese stock market can help explan IPOs underprcng n Chna. In lne wth prevous studes, we fnd an average underprcng of A-share IPOs s 107%. he underprcng s manly drven by market volatlty (MV), ex-ante uncertanty (EX) and turnover rate (R) on the frst tradng day, and frst day tradng rato (R). Furthermorehe underprcng can be also explaned by the characterstcs of Chnese stock market, especally the characterstcs of nvestors. We also nvestgate the long-term performance of IPOs after the lstng. Despte the postve ntal IPO returns, we

2 Xa et al. 853 fnd that the average abnormal returns over one and three years after lstng are both sgnfcantly negatve at the 1% level, consderng both market and stock rsks, abnormal returns over one year cover the nterval [-1.71, -0.54], and abnormal returns over three years cover [-1.54,-0.27]. Aftermarket abnormal returns over one year are sgnfcantly determned by MV, EX and Intal excess returns. Over three years, abnormal returns are drven by market volatlty (MV), offerng prce (OP), ex ante uncertanty (EX) and frst day tradng rato (R). Intal excess returns doesn t show any nfluences on abnormal returns. he purpose of ths study s not lmted to addng another pece of evdence to the vast lterature on IPO underprcng. Rather, we would lke to nvestgate how the underprcng of IPOs s affected by some market factors determned by the characterstcs of Chna s stock market. he rest of ths paper s organzed as follows: formulaton of hypotheses that s to be examned and the methodology adopted; descrpton of data and report of statstcal results; emprcal results; summary and concluson. LIERAURE REVIEW Intal publc offerng (IPO) underprcng means IPOs offer the nvestors wth postve ntal returns on the frst tradng day. Chen et al. (2004) found an average ntal return of 267% for the IPOs from 1991 to 2000 and 145% for IPOs from 1992 to 1997, respectvely. Xu and Lang (2010) reported an underprcng of 91.6% for a sample of IPOs lsted from 2006 to Some other lteratures also show that the large underprcng magntude n the Chnese IPOs market perssts after full crculaton (Yn and Wang, 2008; Xe, 2010). Prevous studes ndcated that the mxed results are found n IPOs long-run performance. Ch and Padgett (2005) proved that postve long-term returns have been ganed by IPOs between 1996 and 1997 n Chna. Wang (2009) found IPOs n Chna between 1999 and 2002 got an average of 37-month excess return of -23.4%. here are many explanatons behnd underprcng IPOs such as asymmetrc nformaton, nsttutonal explanatons, and ownershp and nsder control. Accordng to the market volatlty hypothess (Jog and Wang, 2002)he magntude of underprcng may depend on market volatlty; n other words, ssuers try to mnmze the probablty of unsuccessful ssues by lowerng prces as long as market volatlty s hgh. Wang (1997) and Chen et al. (2008) evdenced ts adaptablty to Chna. Allen and Faulhaber (1989) argued that frms sometmes offer IPOs prced below ther ntrnsc value to sgnal ther qualty to nvestorshus expectng to have a better chance at offerng subsequent seasoned ssues at hgher prces. o the nvestors, lower offer prce means lower rsk and more opportuntes to get speculatve gans, and then they wll trade actvely on the market, further pushes prce up, whch n return enlarge expand the magntude of underprcng. In Chnahe IPOs are often prced accordng to Prce/earnngs (PE) ratos. Han and Chen (2001) favor ths theory n ther research. Accordng to asymmetrc nformaton theoryhe uncertanty about the value of recent IPOs, s hgher than that about long-hstory frms. Investors worry about the future performance of IPOs, whch s referred to as ex ante uncertanty. Beatty and Rtter (1986) ndcate that underprcng should ncrease wth ex ante. Prevous studes Mok and Hu (1998), Chau et al. (1999), Chen et al. (2004) and Xe (2010) explored how the above mentoned theores explan the hgh IPO underprcng n Chna wth vared methods. Zou (2003) takes the opnon that the aftermarket performance could be explaned by the prevously mentoned determnants of ntal performance, Levs (1993) and Paudyal et al. (1998) argued that ntal excess returns mght be due to ntal over-optmsm n the market. Accordngly, such ssues should underperform the market n the long run. In contrast, f IPOs attan ther equlbrum value at ntal returnsher long-run performance should not be sgnfcantly dfferent from that of the market. hs paper contrbutes to the lterature by examnng emprcally IPO underprcng and long-run performance n Chna n lght of the classcal models. We consder three hypotheses that may explan the IPOs underprcng n Chna. hey are the wnner s curse hypothess (Rock, 1986)he ex-ante uncertanty hypothess (Rtter, 1984; Beatty and Rtter, 1986), and the sgnalng hypothess (Allen and Faulhaber, 1989). DAA AND RESEARCH DESIGN Data he data are selected from 116 A-share stocks that were offered between June 19, 2006 and June 30, 2007 (wth full crculaton) n Chna and have traded for three years. 6 stocks (merger, dvson or ssued at a premum) were excluded, and the remanng 109 stocks become the study sample. he data come from several sources, ncludng the snghua Fnancal Research Databaseradng database from GA (Guo a An Informaton echnology Company) and Dongxng Securtes onlne tradng software real-tme quotes system. We processed the basc data wth the EXCEL software, and then use SPSS 12.0 software for emprcal analyss. Calculaton of IPO underprcng and long-term Performance IPO underprcng s often n terms of ntal return, namely the magntude of prce changes from the ssue day to the frst tradng day. Consderng the transacton costs, IPO ntal return s calculated as follows: P P C, t,0 r (1) P, o P,0

3 854 Afr. J. Bus. Manage. where r s the ntal return of IPO, at the frst tradng day,, 0 P s the closng prce of IPO C s P s the offer prce of IPO, and the transacton cost per share. akng the effect on new shares prces nto account, whch market changes (or ndustry dfferences) may have durng the perod from ssue day to the frst tradng dayhe adjusted return should be calculated, namely the ntal excess return. he adjusted return s then calculated as the ntal return for IPO mnus the benchmark return on a correspondng reference portfolo. It s mportant to specfy an approprate benchmark. Ether theoretcally or practcally, t s not obvous wth what ndex or portfolo the IPO returns should be compared. We use more than one ndex: Shangha and Shenzhen A-share ndex n Chna (CMI) and SFC ndustry ndex (IND), where the latter ndexes serve as reference portfolos for IPOs accordng to ther ndustry classfcaton. In contrast to ntal excess returns, long-run performance seems to be more complcated. Accordng to Barber and Lyon (1997)here s no consensus on the approprate way of calculatng long-run abnormal returns. In ths paper, we consder dfferent forms and models. We use the market adjusted model and to calculate the cumulatve abnormal return (CAR) and buy hold abnormal return (BHAR), respectvely. Frst, we compute cumulatve abnormal returns (CARs) and buy-and-hold abnormal returns (BHARs) usng the market-adjusted model over one- and three- year ntervals (12 and 36 tradng months, respectvely) after the IPO lstng, exclusve of the ntal returns. CAR, s, e r rcrp e ts BHAR, ( 1 r ) 1 (1 rcrp ) 1 t1 t1 {12,36} (3) CAR, where, s e s the cumulatve abnormal return (CAR) for IPO from the startng month s after tradng to the annversary month e (12 and 36 months); r and r crp are the monthly returns for IPO and on a correspondng reference portfolo n perod t, respectvely; BHAR, s the buy-and-hold abnormal return for IPO n perod (12 and 36 months), and t s the aftermarket tradng day. Secondhe returns usng the market-adjusted model are not adjusted for rsk other than that prevalng n the market as a whole, so we use the Sharp-Lntner captal asset prcng model () to calculate the abnormal return to take nto consderaton the rsk of ndvdual IPOs. AR r rf rcrp rf where t (2) (4) AR, s the abnormal return usng for frm n month t; r s the rsk-free rate proxed as a short-term one-month rate for bank deposts, and s the rsk of IPO and s taken from the regresson model (that she slope obtaned from regressng [ r - r f ] on [ rcrp, t - r f ] for the estmaton perod). Wth AR calculated, I apply the same two forms, CARs and BHARs, mentoned earler. Followng Rtter (1991), we calculate the wealth relatve (WR) to compare the average buy-and-hold return (BHR) on a portfolo of IPOs relatve to the average BHR on a correspondng reference portfolo to nterpret the performance of IPOs. A WR greater than 1.00 means that IPOs outperform ther correspondng reference portfolo, and vce versa: WR 1 AvgBHR, IPOs ={12,36} (5) 1 AvgBHR, CRP where WR s the wealth relatve over perods, and AvgBHR, IPOs and AvgBHR, CRP are the average BHR on a portfolo of IPOs and on a correspondng reference portfolo over perods, respectvely. Determnants of IPO underprcng and long-term performance o better understand the magntude of observed ntal and long-run performance of IPOs, we conduct several cross-sectonal regressons of ntal excess returns and long-term abnormal returns to dentfy the sgnfcance of selected exogenous varables mentoned n the hypothess. As far as ntal excess returns are concerned, we estmate the followng model: Ar 1EX 2OP 3MV 4PER 5MR 6R (6) where Ar s the ntal excess return of IPO that refers to the level of underprcng; EX refers to the ex ante uncertanty measured by the standard devaton of daly returns of IPO one year after offcal lstng; OP s the offer prce of IPO ; MV s market volatlty, whch s calculated, followng Paudyal et al. (1998), as the standard devaton of daly market returns over the two months before the closng date of subscrpton to buy shares of frm ; PER s the prce/earnngs rato of IPO ; MR s the lucky number rato calculated as total nvestment amount dvded by ssuable poton. Wth regard to the determnants of aftermarket performance, we rely on the same ndependent varables that explan ntal excess returns, accordng to Zou (2003). And based on the Levs s (1993) argument, we add ntal abnormal returns nto the model. We estmate the followng model to explore the explanatory power of the model: AFMARKAR Where,, 1Ar 2EX 3OP 4MV 5PER 6MR 7 AFMARKAR, IPO over perods. DESCRIPIVE SAISICS R (7) s the aftermarket abnormal return for Descrptve statstcs analyss of ntal excess returns In ths area, we test whether IPOs are underprced, n other words,

4 Xa et al. 855 able 1. Descrptve statstcs of IPO ntal returns. Varable IPO CMI IND Mean Medan Maxmum Mnmum Standard devaton Note: he table shows basc descrptve statstcs for IPO ntal returns of Chna A-share market of sample perod. CMI = the captal market ndex; IND = the ndustry ndex. he table ncludes the mean, medan, maxmum, mnmum, and standard devaton values for IPO ntal returns and ther correspondng reference portfolos. able 2. Descrptve statstcs of IPO ntal excess returns. Frms wth >0 abnormal returns Frms wth <0 abnormal returns Mean t-statstc Medan z-statstc Mean wealth relatve Medan wealth relatve CMI *** *** IND *** *** ar r r crp Note: he table shows ntal excess returns for 119 Intal publc offerng (IPOs), whch s calculated as,where r s the ntal excess return of frm from subscrpton tme to the closng of the frst tradng day, s the ntal raw return of frm from r subscrpton tme to the closng of the frst tradng day, and crp s the raw return on a correspondng reference portfolohat s general market ndex or ndustry ndex over the same perod. he t-statstc for the average excess returns s computed ar t( ar) ( ( ar) n) ar as, where s the sample average of ntal excess returns, and ( ar ) s the cross-sectonal sample standard devaton of ntal excess returns. he z-statstc s based on the Wlcoxon sgned-rank test. We provded the number of frms that experence postve or negatve ntal excess returnshe mean and medan values of ntal excess returns, and the t- and z-statstc values wth ther sgnfcance levels. Mean (medan) wealth relatve s calculated as the rato of 1 plus the mean (medan) ntal raw returns of IPOs dvded by 1 plus the mean (medan) ntal raw return on a correspondng reference portfolo. ***Sgnfcant at the 1% level. ar whether nvestors, on average, outperform the market through buyng IPOs at subscrpton prces and sellng them on the frst tradng day. As shown n ables 1 and 2, on average, IPOs yelded 107.7% on the frst tradng day, whch are far above the average return on correspondng reference portfolos. Some nvestors obtaned superor ntal returns as hgh as 496%, and others acheved ntal returns as low as 4.5%. he mean (medan) ntal excess return s around 107.4% (85.6%), t would be obtaned by an nvestor who bought IPOs at the offer prce and sold them at the end of the frst tradng day. Alsohe results showed that 64.2% of IPOs (70 of 109) provde nvestors wth postve ntal excess returns. Usng two benchmarks, CMI and INDhe results from the parametrc t-statstc (the nonparametrc z-statstc) reveal that the mean (medan) ntal excess return s sgnfcantly postve at the 1% level. hese fndngs ndcated that IPOs n Chna are grossly underprced, n lne wth the fndngs reported n the relevant lterature (Ch and Padgett, 2005; Xu and Lang, 2010). What s morehe level of underprcng s hgher than that observed n most countres (Loughran et al., 2003). Descrptve statstcs analyss of long-term abnormal returns In ths area, we consder whether IPOs sustan ther ntal excess returns and provde nvestors wth postve abnormal returns over a long perod. ables 3 and 4 showed summarzed statstcs for the long-term abnormal returns of IPOs usng two benchmarks, CMI and IND, respectvely, over one- and three-year perods. After full crculaton, IPOs n Chna A-share market both yelded negatve returns, over one- and three-year perods. Long-run abnormal returns over one-year perod cover the nterval [-1.71, -0.54], and over three-year perod cover the nterval [-1.54, -0.27]. In generalhe buy-and-hold strategy produces, on average, lower returns for nvestors than does the cumulatve abnormal return (CAR) strategy. he parametrc test statstcs are sgnfcant negatve at the 1% level for all models, whch means nvestors acheve abnormal negatve returns, and ther IPOs nvestments underperform market over one- and three-year perods. he null hypothess that the medan abnormal returns of IPOs are not dfferent from zero cannot be rejected, because the nonparametrc Wlcoxon sgned-rank test statstcs are not sgnfcant. he mean and medan wealth relatves for all models are all less than 1, whch means f an nvestor bought each IPO for an equal amount of money at the closng prce of ts frst tradng day and held t untl the frst or thrd annversary, he or she would have acheved a return less than that acheved by nvestng n the whole market. he mean wealth relatve of 0.59 for IPOs, n a best-case scenaro, mples that an nvestor would have to nvest 69% more to get the same performance as the market. Moreover, n a worst-case scenaro, an nvestor would need to nvest 300% more to catch up to

5 856 Afr. J. Bus. Manage. able 3. Abnormal returns for IPOs VS. Alternatve ndexes based on CAR. IPO VS. CMI MAM 1 year 3 year 1 year 3 year Number Frms wth > 0 abnormal return Frms wth 0 abnormal return Frms wth 0 abnormal return Mean abnormal returns *** -3.49*** *** -8.99*** t-statstc Medan abnormal returns IPO VS. IND Number Frms wth > 0 abnormal return Frms wth 0 abnormal return Frms wth 0 abnormal return Mean abnormal returns *** *** *** *** t-statstc Medan abnormal returns Note: he table provdes abnormal returns of ntal publc offerng (IPOs) over one and three years. he aftermarket abnormal returns are calculated based on cumulatve return usng the market-adjusted model and the captal asset prcng model (). he market-adjusted return ( MAM ) r, t rcrp where r and r crp are the monthly returns for IPO and on a correspondng reference portfolo n month t, respectvely. r r [ rcrp rf ], where s the abnormal return usng, r f s the rsk-free rate proxed as short-term one-month rate for bank deposts, and s the rsk of IPO, and s gven from the regresson, whch s the slope obtaned from regressng [ r rf ] on [ rcrp rf ] for the estmaton perod. CAR, s, e r rcrp e ts where CAR,, s e s the cumulatve abnormal return (CAR) for IPO from the startng month s after tradng to the annversary month e (12 and 36 months); r and r crp are the monthly returns for IPO and on a correspondng reference portfolo n perod t, respectvely. We provde the number of frms that experence postve or negatve aftermarket abnormal returnshe mean and medan values of aftermarket abnormal returns, and the t- and z-statstc values wth ther sgnfcance level.cmi= the captal market ndex; IND=the ndustry ndex. ***Sgnfcant at the 1% level. the market performance (the mean wealth relatve s only 0.25 over a three-year perod usng the IND as a benchmark). hese results support the dea that, n the long run, IPOs cannot sustan ther ntal excess returns, and provde nvestors wth negatve abnormal returns, n lne wth the fndngs reported n the lterature for IPOs n some economcs (Rtter, 1991; Gregory et al., 2010; Chang et al., 2010). EMPIRICAL RESULS AND ANALYSIS Intal excess returns In ths area of the analyss, we examne the determnants of ntal excess returns or the level of underprcng. Regresson results for Equaton 6 are shown n able 5. As Jog and Wang (2002) suggested, we found a sgnfcantly (1%) postve relatonshp between market volatlty and ntal excess return. It proved that behavors of nvestors n Chna A-share market are easly nfluenced by the stock market volatlty. hey tend to nvest prudently when stock market s n severe volatlty, and nvest actvely when stock market s smooth. Such characterstcs of nvestors behavor lead lsted companes to ssue IPOs wth a prce below ts ntrnsc value when stock market s n severe volatlty. he estmaton results showed that the coeffcent on ex-ante uncertanty s sgnfcantly postve at the 5% level, supportng Beatty and Rtter s (1986) argument that nvestors seek hgher returns to compensate for ther angst about future performance of IPOs. o nvest n a

6 Xa et al. 857 able 4. Abnormal returns for IPOs VS. Alternatve ndexes based on BHAR. IPO VS. CMI MAM 1 year 3 year 1 year 3 year Number Frms wth > 0 abnormal return Frms wth 0 abnormal return Mean abnormal returns t-statstc *** *** *** *** Medan abnormal returns z-statstc Mean wealth relatve Medan wealth relatve IPO VS. IND Number Frms wth > 0 abnormal return Frms wth 0 abnormal return Mean abnormal returns t-statstc *** *** *** *** Medan abnormal returns z-statstc Mean wealth relatve Medan wealth relatve Note: he table provdes abnormal returns of ntal publc offerng (IPOs) over one and three years. he aftermarket abnormal returns are calculated based on buy-and-hold return usng the market-adjusted model (MAM) and the captal asset prcng model (). he market-adjusted return ( MAM ) r, t rcrp where r and r crp are the monthly returns for IPO and on a correspondng reference portfolo n month t, respectvely. r r [ rcrp rf ], where s the abnormal return usng, r f s the rsk-free rate proxed as short-term one-month rate for bank deposts, and s the rsk of IPO, and s gven from the regresson, whch s the slope obtaned from regressng [ r rf ] on [ rcrp rf ] for the estmaton ( 1 r ) 1 (1 rcrp ), {12,36}, BHAR, t1 perod. BHAR, 1 s the buy-and-hold t1 abnormal return for IPO n perod (12 and 36 months), and t s the aftermarket tradng day. We provde the number of frms that experence postve or negatve aftermarket abnormal returnshe mean and medan values of aftermarket abnormal returns, and the t- and z-statstc values wth ther sgnfcance level. Mean (medan) wealth relatve s calculated as the rato of 1 plus the buy-and-hold aftermarket raw returns of IPOs dvded by 1 plus the mean (medan) buy-and-hold aftermarket raw return on a correspondng reference portfolo. CMI= the captal market ndex; IND=the ndustry ndex. ***Sgnfcant at the 1% level. lsted company, an nvestor must take more energy to collect relevant nformaton; otherwse, he or she wll face a wnner curse. Most nvestors n Chna s stock market are ndvdual nvestors, who are eager to get rch quckly. hey nvest n stocks to get speculatve returns, and ther nvestment decsons manly rely on back alley news, stock comments and newspapers artcles, etc. herefore, n the subscrptonhey would rather be unnformed nvestors than collect nformaton, so that ther adverse selecton seems more serous. Such characterstcs determne that nvestors n Chna face hgher rsk when they subscrbe. Furthermore, nvestors requre hgher returns to compensate for the rsk. So IPOs cannot but be underprced. he coeffcent of lucky number rato s postve and sgnfcant (5%), whch suggests that IPO underprcng s related to the absorpton capacty of the market, n lne wth Paudyal et al. (1998). When the demand for IPOs s hgh, nvestors receve fewer shares than they ordered. In turnhey try to buy from the market at a hgher prce on the frst day and a perod after to get quck fortunes. Investors overreacton pushes prces up, further, enlarge the magntude of IPO underprcng. hs phenomenaoo, s determned by Chna stock market characterstcs,

7 858 Afr. J. Bus. Manage. able 5. Determnants of ntal excess returns of IPOs. Independent varable Dependent varable: Intal excess returns (AR) CMI IND Intercept ** ** EX ** ** OP MV *** *** PER MR ** ** R 0.012* 0.012* R Adj. R F-value 7.319*** 7.165*** DW-stat Note: he table shows the results from multvarate cross-sectonal regresson analyss based on the followng model: Ar EX OP MV PER MR R , where Ar s the ntal excess return of IPO that refers to the level of underprcng; EX refers to the ex ante uncertanty measured by the standard devaton of daly returns of IPO one year after offcal lstng; OP s the offer prce of IPO ; MV s market volatlty, whch s calculated, followng Paudyal et al. (1998), as the standard devaton of daly market returns over the two months before the closng date of subscrpton to buy shares of frm ; PER s the prce/earnngs rato of IPO ; MR s the lucky number rato calculated as total nvestment amount dvded by ssuable poton. CMI= the captal market ndex; IND= the ndustry ndex; DW-stat. = Durbn-Watson statstc. ***Sgnfcant at the 1% level. **Sgnfcant at the 5% level.*sgnfcant at the 10% level. namelyhe majorty of nvestors ndvduals lack n nvestment knowledge and sklls wth a mmature nvestment mentalty. he postve and less sgnfcant (10%) coeffcent for frst day tradng rato showed that IPOs are more underprced n a hot market. Belevng the exstence of IPO myth, nvestors expectaton for hgher return from IPOs leads to grumous speculatve atmosphere, whch make the prce too hgh n the secondary market. hen a hgh IPO underprcng emerged. However, other factors such as frm-specfc characterstcs (prce/earnngs rato) and prcng mechansm (offer prce) are nsgnfcant n explanng IPO underprcng, ndcatng that nvestors n Chna s A-share market pad lttle attenton to the actual value of the lsted companes. he ntal IPO excess returns are sgnfcantly affected by market factors (market volatlty, ex-ante uncertanty, lucky number rato and frst day tradng rato) and nsgnfcantly nfluenced by stock factors. hs fndng showed that nvestors are more concerned about short-term returns that IPOs can brng about when makng the decson. Furtherhs reflects, n Chnese A-share market, nvestors buy IPOs for speculaton rather than nvestment. he model only explans 26.2 and 25.7% (usng CMI and IND respectvely) of the varablty n ntal excess returns of the sample of Chnese A-share IPOs, ntal returns, whch showed the general explanatory power of the three hypotheses earler mentoned. F-value s sgnfcant at the 1% level, ndcatng that the equaton has some lnear relatonshp, and moreoverhe model s set up correctly. DW-statstc s close to 2, ndcatng that the regresson does not nclude seral autocorrelaton. he lower adjusted R 2 s mean the model explaned only a part of the hgh IPO underprcng n Chna s A-share market, and they are lower than that of smlar studes about some other economes (Guo et al., 2006; Ljungqvst et al., 2006). hese results show there s stll gap between Chna marketzaton level and that of mature market economy. What s more, Chna has ts unque prcng mechansm and tradng system, so the reasons for IPO underprcng n Chna are more complcated, and exstng theores cannot fully explan t. Long-term abnormal returns In ths area of the analyss, we examne the determnants of long-run performance or long-run abnormal returns. Regresson results for Equaton 7 are shown n ables 6 and 7.

8 Xa et al. 859 able 6. Multvarate cross-sectonal regresson analyss of the determnants of long-term abnormal returns of IPOs (CMI). Independent varable Market-adjusted model Dependent varable: Long-term abnormal returns 1 Year 3 Year Market-adjusted model CR BHR CR BHR CR BHR CR BHR Intercept *** *** *** *** *** ** *** ** Ar *** * *** * *** ** EX *** *** *** *** *** ** *** ** OP ** ** ** MV *** *** *** *** *** *** *** *** PER MR 0.275*** 0.385** 0.275*** 0.382** 0.365* 0.489*** 0.333*** (3.203)*** R ** ** R Adj. R F-value *** 9.698*** *** 9.769*** 5.159*** 7.801*** 7.575*** 7.937*** DW-stat Note: he table shows the results from multvarate cross-sectonal regresson analyses of the determnants of aftermarket abnormal returns over one and three years wth the whole A-share market as the correspondng reference portfolos. he followng model s employed: AFMARKAR, 1Ar 2EX 3OP 4MV 5PER 6MR 7R, where AFMARKAR, s the aftermarket abnormal return for IPO over perods, whch takes the form of cumulatve returns (CR) or buy-and-hold returns (BHR) usng market-adjusted model or the captal asset prcng model (); Ar s the ntal excess return of IPO that refers to the level of underprcng; EX refers to the ex ante uncertanty measured by the standard devaton of daly returns of IPO one year after offcal lstng; OP s the offer prce of IPO ; MV s market volatlty, whch s calculated, followng Paudyal et al. (1998), as the standard devaton of daly market returns over the two months before the closng date of subscrpton to buy shares of frm ; PER s the prce/earnngs rato of IPO ; MR s the lucky number rato calculated as total nvestment amount dvded by ssuable poton. ***Sgnfcant at the 1% level. **Sgnfcant at the 5% level. *Sgnfcant at the 10% level. As seen n ables 6 and 7he long-term performance of IPOs n Chna s A-share market over a one-year perod s sgnfcantly (1%) affected by market volatlty, ex-ante uncertanty, ando a lesser extent, ntal excess returns. he negatve coeffcent estmate of ntal excess returns s consstent wth Levs (1993) and Paudyal et al. (1998). hs fndng suggested that nvestors are overoptmstc about the performance of IPOs n the short run, but grow more pessmstc over tme, whch s determned by Chnese unque market characterstcs: heavy speculatve atmosphere and hgh turnover of new shares. here s no prce lmtng lmts on the frst tradng day n Chna, and n addtonhere s frequent ssue of new shares. herefore, most of the prmary market nvestors sell ther shares to recoup funds to subscrbe for another IPO stock on the frst tradng day and sell n a short perod after. hs pushes the prce down, and leads to negatve long-term abnormal returns. ables 6 and 7 showed, over three-year perod, abnormal returns are sgnfcantly affected by market volatlty, offer prce, and to a lesser extent, ex-ante uncertanty and frst day tradng rato, consstent wth the hypothess. Intal excess return s nsgnfcant n explanng t. he postve and sgnfcant coeffcent for the market volatlty showed that the larger the market volatlty she hgher the long-term abnormal returns bears. he negatve and sgnfcant coeffcent for the offer prce suggests IPOs wth hgher offer prce provde nvestors wth negatve abnormal returns n long run. he postve coeffcent for the ex-ante uncertanty ndcated the uncertanty of IPO companes actual value dmnsh over the tme whle nvestment rsk decreasng, so the IPOs become hot, whch provde nvestors wth postve abnormal returns n long run, n lne wth Omran (2005) fndng. he negatve coeffcent for the frst day tradng rato reflected that nvestors grow more pessmstc over tme, gradually pushng the prce down. he ntal IPO excess returns sgnfcantly affect abnormal returns over one-year, but have nsgnfcant nfluence on abnormal returns over three-year, ndcates that the nformaton asymmetry resultng n the IPO underprcng was made up n three years or less but over one year. hs s determned by the A-share market s characterstcs n Chna that nformaton asymmetry can

9 860 Afr. J. Bus. Manage. able 7. Multvarate cross-sectonal regresson analyss of the determnants of long-term abnormal returns of IPOs (IND). Dependent varable: Long-term abnormal returns Independent varable 1 Year 3 Year Market-adjusted model Market-adjusted model CR BHR CR BHR CR BHR CR BHR Intercept *** *** *** *** * Ar *** -.254** *** EX *** ** *** OP *** *** *** ** MV *** *** 68.91*** *** *** *** *** *** PER *** 0.022*** ** 0.019*** MR *** R ** *** * ** ** R 2 % Adj. R 2 % F-value 8.503*** 6.170*** *** *** 5.524*** 5.513*** 7.154*** 7.110*** DW-stat Note: he table shows the results from multvarate cross-sectonal regresson analyses of the determnants of aftermarket abnormal returns over one and three years wth ndustry as the correspondng reference portfolos. he followng model s employed: AFMARKAR, 1Ar 2EX 3OP 4MV 5PER 6MR 7R, where AFMARKAR, s the aftermarket abnormal return for IPO over perods, whch takes the form of cumulatve returns (CR) or Ar s the ntal excess return of IPO EX refers to the ex ante uncertanty measured by the standard devaton of daly returns of IPO buy-and-hold returns (BHR) usng market-adjusted model or the captal asset prcng model (); that refers to the level of underprcng; one year after offcal lstng; OP s the offer prce of IPO ; MV s market volatlty, whch s calculated, followng Paudyal et al. (1998), as the standard devaton of daly market returns over the two months before the closng date of subscrpton to buy shares of frm ; PER s the prce/earnngs rato of IPO ; MR s the lucky number rato calculated as total nvestment amount dvded by ssuable poton. ***Sgnfcant at the 1% level. **Sgnfcant at the 5% level. * Sgnfcant at the 10% level. be made up n a short tme. As the spotlght of the stock market, IPOs are well studed by underwrters, research nsttutes, and etc. here s lots of publc evaluaton nformaton about them, whch greatly shorten the tme that prces use to come to a reasonable level. It s worth notng that factors such as the global fnancal crss had ts effects on the overall trend n the market. Recallng that most IPOs were traded n the market over one year but less than three years before the economc slump, and that they were among the bg stock wnners n the bull market, they experenced a huge sell-off n the subsequent perod of bear market. herefore, nvestors seem to descend a worrment, and grow more pessmstc, whch speed up the dsappearance of IPO underprcng. he R 2 of the ftted models provdes a better explanaton for the behavor of IPOs over a one-year perod compared wth a three-year perod. he hghest adjusted R 2 over a one-year perod s 46.6% compared wth 31.2% over a three-year perod, whereas the lowest R 2 s 25.3% over a three-year perod compared wth just 21.4% over a three-year perod. F-value s sgnfcant at the 1% level, ndcatng that the model s set up correctly. DW-statstc s close to 2, ndcatng that the regresson does not nclude seral autocorrelaton. he R 2 s over one- and three-year perods are lower than smlar study of other economes (Omran, 2005), agan provng that exstng theores cannot fully explan IPOs aftermarket performance n Chna. Concluson hs paper studes the underprcng and the long-term performance of 109 A-share IPOs ssued n Chna between June 19, 2006 and June 30, 2007, and explore ther reasons. Consstent wth results from prevous studes, we found there s a huge underprcng of A-share IPOs, as the average return on the frst tradng day s 107%. akng both market rsk and stock rsk nto consderaton, abnormal returns over one year cover the nterval [-1.71, -0.54], and abnormal returns over a three-year perod cover the nterval [-1.54, -0.27]. he ntal IPO excess returns are sgnfcantly affected

10 Xa et al. 861 by market volatlty, ex-ante uncertanty, lucky number rato and frst day tradng rato, are nsgnfcantly nfluenced by stock factors such as prce/earnngs rato and offer prce. hs fndng suggests nvestors behavor plays a major role n IPO underprcng and nvestors buy IPOs amed at speculaton rather than nvestment. In the long run, aftermarket abnormal returns over a one-year perod are sgnfcantly affected by market volatlty, ex-ante uncertanty, ando a lesser extent, ntal excess returns. Over three years, abnormal returns are drven by market volatlty, offerng prce, ex-ante uncertanty and frst day tradng rato. Intal excess returns sgnfcant nfluence on abnormal returns over one year ndcates Chna A-share market s neffcent wth one year after IPO. However, ntal excess returns do not show any nfluences on abnormal returns over three years, whch s determned by the Chnese A-share market s characterstc that nformaton asymmetry can be made up n a short tme. hs study on the IPO underprcng, long-run performance and ther determnants wth full crculaton n Chna contrbute to lterature for provdng new evdence. Snce t s not long after full crculatonhe sample demandng for IPOs tradng for full three years s lmted. Further study can track panel data of Chna A-share market to examne the aforementoned results. REFERENCES Allen F, Faulhaber G (1989). Sgnallng by Underprcng n the IPO Market. J. Fn. Econ. 23: Barber B, Lyon J (1997). Detectng long-run abnormal stock returns: the emprcal power and pecfcaton of test statstcs, J. Fnanc. Econ. 43: Beatty RP, Rtter JR (1986). Investment bankng reputaton, and the underprcng of ntal publc offerngs, J. Fnanc. Econ. 15: Chau C, Cccotello CS, Grant C (1999). Role of ownershp n Chnese prvatzaton: Emprcal evdence from returns n IPOs of Chnese A-Shares, Adv. Fnanc. Econ. 4: Chen YL, Xong H, Luo QL (2008). Emprcal Study of IPO underprcng n Chnese securty market under full crculaton, reform and strategy, p. 9. Chen G, Frth M, Km JB (2004). IPO underprcng n Chna s new stock market, J. Multnatonal Fnancal Manag. 14: Gregory A, Guermat C, Al-Shawawreh F (2010). UKIPO: Long Run Returns, Behavoural mng and Pseudo mng, J. Bus. Fnanc. Account. 37: Guo RJ, Lev B, Sh C, (2006). Explanng the short- and long-term IPO anomales n the US by Rand D, J. Bus. Fnanc. Account. 33: Han DZ, Chen J (2001). Emprcal Study of IPO underprcng n Chna, Statst. Res. p.4. Ch J, Padgett C (2005). he Performance and Long-run characterstcs of the Chnese IPO Market. Pac. Econ. Rev. 10: Levs M (1993). he long-run performance of ntal publc offerngs: he UK experence Fnanc. Manag. 22: Ljungqvst A, Nanda V, Sngh R (2006). Hot markets, nvestor sentment, and IPO prcng, J. Bus. 4: Loughran, Rtter JR, Rydqvst K (2003). Intal publc offerngs: Internatonal nsghts. Mok MK, Hu YV (1998). Underprcng and after-market performance of IPOs n Shangha, Chna, Pacfc-Basn Fnanc. J. 6: Omran M (2005). Underprcng and Long-run performance of Share Issue Prvatzatons n the Egyptan Market. J. Fnanc. Res. 2: Paudyal K, Saadoun B, Brston RJ (1998). Prvatzaton ntal publc offerngs n Malaysa: Intal premum and long-term performance, Pacfc-Basn Fnanc. J. 6: Rtter JR (1991). he long-run performance of ntal publc offerngs. J. Fnanc. 46:3-27. Rtter JR (1984). he hot ssue market of 1980, J. Bus. 57: Rock K (1986), Why new ssues are underprced?, J. Fnanc. Econ. 15: Wang JB (1997). Measure of new shares expected excess returns and the possble explanaton. Econ. Res. p.12. Wang Z (2009). Emprcal Study of IPO long-term returns based on tme seres, Busness mes. Chang X, Ln S, Lews HK, George W (2010). Cross-sectonal determnants of post-ipo stock performance: evdence from Chna, Account. Fnanc. 50: Xe JL (2010). Study of A-share IPO underprcng under full crculaton, Fnanc. Econ. p.2. Xu SX, Lang SX (2010). Study of Chna A-share IPO underprcng under full crculaton. Stock Mark. 10: Yn XB, Wang R (2008). Study of IPO underprcng under full crculaton, Econ. Issues p.2. Zou J (2003). Study of determnants of IPO underprcng and long-term abnormal returns Under Verfcaton Mechansm, Report of Shenzhen Stock Exchange Research Insttute, Shenzhen Res. Inst. p