IPO initial returns in China:Underpricing or overvaluation?

Shunlin Song,JinSong Tan,Yang Yi

aSchool of Accountancy,Central University of Finance and Econom ics,China

bSchool of Business,Sun Yat-sen University,China

cCenter for Accounting,Finance and Institutions,Sun Yat-sen University,China

IPO initial returns in China:Underpricing or overvaluation?

Shunlin Songa,*,JinSong Tanb,c,Yang Yib,c

aSchool of Accountancy,Central University of Finance and Econom ics,China

bSchool of Business,Sun Yat-sen University,China

cCenter for Accounting,Finance and Institutions,Sun Yat-sen University,China

A R T IC L E IN F O

Article history:

Availab le online 28 January 2014

JEL classif cation:

G 14

G 30

G 39

IPO initial retu rn

IPO underp ricing

IPO overvaluation

Post-IPO long-run stock performance

Determ inants

This paper separates the amount o f IPO underp ricing(p rimary market underpricing)and overvaluation(secondarymarketovervaluation)from the valueof an IPO’s initial return to evaluate the relative im portance of these two factors and theirmain determ inants.Using data on the IPOs o f 948 Chinese f rm s,we fnd thataverage initial returnsare66%and thatunderpricing and overvaluation arebetween 14–22%and 44–53%,respectively,depending on themethod used to assess f rm s’intrinsic values.In addition,whileboth thevalueof the initial return and the extent o f overvaluation are signif cantly negatively related to post-IPO long-run stock performance,overvaluation can predict post-IPO performance better than the valueof the initial return.Valueuncertainty in IPOs ispositively related to both underpricing and overvaluation,and both theunderw riter’s reputation and theexistenceo fpricing regu lation arepositively related to underp ricing.Investorsentimenthasa positiveef ectonovervaluationbuthasno ef ectora negative ef ect on underp ricing.Overall,our results suggest that in China overvaluation accounts fora larger proportion o f the initialreturn than underpricing, and that underpricing and overvaluation have diferent determ inants.

?2014 Production and hosting by Elsevier B.V.on behalf o f China Journalof Accounting Research.Founded by Sun Yat-sen University and City University o f Hong Kong.

1.Introduction

On Novem ber 5,2007,PetroChina(601857),them ost p rof tab le com pany in Asia,retu rned to the A-share m arket at the peak of a bu llmarket.The initial return(also called the f rst-day return)o f PetroChina peakedat 163%,m ak ing it theworld’sm ost valuable com pany bym arket cap italization at thatm om ent.However,the stock p riceof PetroChina gradually dropped after its IPO,depreciating by 76%over thenext three years.Consequently,countlessholders of A-shares in PetroChina lost a great deal o fmoney.However,in China’s IPO market,stories like that o f PetroChina are common.While IPO initial returns are extremely high,the post-IPO stock performance ofmany companies is very poor.In theory,there are on ly two cases in which IPOs would experience an abnormally high initial return:the o fer p rice of the IPO is too low,indicating that there is underp ricing in the p rim arymarket,or the f rst-day closing price is too high,indicating that there is overvaluation in the secondarymarket(Han and W u,2007).

Therefore,a crucialquestion iswhich of these two cases is relativelym ore important in explaining the phenom ena of extrem ely high initial retu rns in China:underpricing in the prim arym arket or overvaluation in the secondary m arket?W e believe that the answer to this question is no t only vital to stud ies of IPO pricing but also has im portant im p lications fo r the regu latorswho m ust decidewhether and how to reform IPO pricing in China.Un fortunately,when discussing initial returns in China,themedia seldom distinguish between IPO initial returns and underpricing.M oreover,the literature usually measures IPO underp ricing w ith IPO initial returns and does not diferentiate between the extent o f underpricing and the extent of overvaluation(e.g., Wang et al.,2009;Zhang and Liao,2011).

However in the Chinese stockmarket,f rst-day closing p ricesareo ften signif cantly overpriced and thus the initial return and the underpricing are two totally dif erent concep ts.If the extent o f underp ricing and overvaluation isnot separated from the value of the initial return,we not on ly cannot evaluate the relative importance o f these two parts,but also cannot precisely exam ine the main determ inants of underpricing and overvaluation.

M o tivated by this p rob lem and gap in the literature,this paper clarif es the two concep ts of IPO initial return and IPO underpricing,separates IPO underp ricing and IPO overvaluation from the value of the initial return,investigates the relative importance o f these two parts,and then exam ines their main determ inants. Estimating IPO f rms’intrinsic values is the key to measuring the extent of underp ricing and overvaluation. Twomethods for assessing intrinsic value are used and compared in this paper:analyst forecastsand comparisons to sim ilar f rms(we explain these two methods in detail in Section 4).

Using data from the IPOs o f 948 Chinese f rms between 2006 and 2011,our results show that,during the sample period,when a f rm’s intrinsic value ismeasured by analyst forecasts,underpricing isabout22.2%and overvaluation is about 44.1%(tw ice asmuch as underpricing).However,when intrinsic value is assessed th rough com parable f rm s,underpricing is about 13.6%and overp ricing is abou t 52.7%(th ree tim es asm uch asunderpricing).These results consistently show that overvaluation accoun ts for the largest propo rtion of the value of the IPO initial return,suggesting that the f rst-day closing price being overvalued by investo rs ism o re important in explaining the high initial return of Chinese IPOs than the of er p rice being underpriced by the issuer.

In addition,both the value of the initial return and the extent of overvaluation are found to be signif cantly negatively associated w ith post-IPO long-run stock performance.According to the regression coef cients o f the initial return and the overvaluation,the extent o f overvaluation,asmeasured by analyst forecasts,predicts post-IPO long-run stock performance better than the value of the initial return,suggesting that extracting the overvaluation from the initial return has value in terms of forecasting post-IPO stock perform ance.Additionally,the overvaluation measured by analyst forecasts predicts post-IPO long-run stock perform ance better than the overvaluation m easured by com parison to sim ilar f rm s,suggesting that analyst fo recasts are a m ore accu rate way to m easu re the exten t o f overvaluation than using com parable f rm s.

Wealso fnd that thevalueuncertainty of IPOs ispositively related to both underpricing and overvaluation. Both underw riter reputation and pricing regu lation are positively related to underpricing.Investor sentiment hasa positiveefecton overvaluation buthasno ef ector a negative efecton underp ricing.Overall,our results suggest thatovervaluation accounts for the largest p roportion o f the valueof initial returns in China,and that underpricing and overvaluation have dif erent determ inants.

These f ndings contribute to the literature in two ways.First,while p rior studies imp ly that IPO initial returns consist of two parts,underp ricing and overvaluation(e.g.,Cao and Dong,2006;Han andW u,2007),there hasbeen little em p iricalwo rk that attem p ts to quan tify the relative im po rtance of these two parts.This paper shows that overvaluation accounts for themajority of an IPO’s initial return.This fnding contributes to studiesof IPO pricing by pointing outwhich partso f the initial return deservem oreattention.It also has implications for regulatorswho need to consider how to reform IPO p ricing structures,by show ing that it ismore im portant to control f rst-day overvaluation than to reduce primary market underp ricing. M oreover,decomposing the initial return can also help to evaluatemore comp rehensively the potentialef ects o f IPO p ricing reform.

A second way that thispaper contributes to the literature is thatwesp lit the initial return into underp ricing and overvaluation,and investigate the main determ inants of these aspects separately,which can help us fu rther understand the determ inants of the value o f IPO initial retu rns and p rovides a reference fo r further studies that exam ine the determ inan ts of these two aspects.

This paper is d ivided in to f ve sections.In Section 2,we introduce the institu tional background of China’s IPO m arket.Section 3 provides a theoretical analysis.Section 4 presentsourmethodo logy.Section 5 reports our em pirical results and Section 6 concludes the paper.

2.Institutional background

During the processo f an IPO,the behavior of listed companies,underw riters and investors are in fuenced by both the institutional environment and institutional arrangements.Therefore,beforewe analyze the composition of IPO initial returnsand their determ inants in China,it isnecessary to brief y introduce the institutional environm ent and institutional arrangements in China’s IPO market.

2.1.IPO institutional environment

IPO initial retu rns can be exp lained by both rationaland non-rational theories.G iven that them arket efciency in China is relative low and the p rim arym arket isno t a com petitive one,non-rationalexp lanationsm ay bemore appropriate(Han and W u,2007).First,the extent of themarketef ciency of China’s stockmarket is lower than that of a developed market.W hile the ef ciency o f the U.S.stockm arket hasbeen recognized by both academ ics and p ractitioners,Chinese scholars generally believe that China’s stock market has not yet reached a level o f sem i-strong form market ef ciency.For exam ple,many studies,such as Lu and Zhou (2007)and Xu et al.(2011)have found evidence against China having sem i-strong form market ef ciency. China’sstockmarket isan emergingmarketw ith a very shorthistory,characterized by inexperienced investors and intensespecu lation.AsSu(2008)and Tian(2010)pointout,most participants in China’sstockmarketare individual investors(retail investors),who prefer specu lation rather than value investment.Thismeans that stock p rices in China’s stock market are easily in fuenced by investor sentiment.

Second,the prim arym arket in China isnot yet com petitive.In a com petitive p rim arym arket,underw riters have an incentive to underprice of er p rices in tentionally to avoid the risk o f IPO failure.However,there are very few exam p leso f IPO failures in China because investo rs in China havea very strong in terest in subscribing to new shares(becauseo f thehigh initial returns in China).1In June 2011,Baling Technology(002592)was forced to cease its IPO because the num ber of book building participants(m ostly institutional investors)did not reach the quota o f 20.It became the f rst failed IPO in China(Baling Technology IPO later succeeded in November 2011).The second failed IPO(Longmaster Information&Technology,300288)did not appear until January 2012.The com pany succeeded in holding an IPO in February 2012).The fact that thereare very few failed IPO cases suggests that there is a low risk o f IPO failure in Ch ina.Due to the low risk of IPO failure,f rmsand underw ritershave lower incentives to undervalue ofer p ricescompared to those in a developed market.Thismeans that rational theoriesmay be relatively less important in exp laining thevalueof initial returns in Chinese IPOs.

2.2.IPO institutional arrangements

In China,the government regulatesmany diferent aspectso f IPOs and thus the evolution of IPO institutional arrangementsgoeshand in hand w ith changes in lawsand regu lations.The regu lator contro ls the IPO p rocess in two main ways:IPO qualif cations and IPO pricing.In relation to IPO qualif cations,the CRSC(China Securities Regulatory Comm ission)implemented an app roval system2Theapprovalsystem means thatwhen new stocksare issued,the issuer shou ld notonly com pletely disclose f nancialand non-fnancial inform ation but also m eet the regulator’s listing criteria.The Issuance Exam ination Comm ittee of the CSRC determ ines whether the app licantm eets the listing criteria.The d iferencebetween an app rovalsystem and a registration system iswhether the regulator judges the valueof IPO f rms.The sponsor system means that theunderw riterand representative of thesponsor is responsible for recommending and guiding the issuer;conducting due diligence;verifying the truth,accuracy and completeness of the issuer’smaterials;and assisting the issuer in bu ild ing a strict in formation d isclosure system.in 2001,under which the Issuance Exam ination Comm ittee o f the CSRC exam ineswhether the applicantmeets the regu lator’s listing criteria.Com p lem enting the app roval system,a sponso r system was adopted in 2004,under which an underw riter conducts due d iligence and verifes the tru th,accu racy and com p leteness of the issuer’sm aterials.

Table 1 The evolution of China’s IPO system since 2006.

In relation to the second aspect of IPO pricing,China in troduced the book build ing app roach in 2005, which is the dom inant method of issuing in most countries.3The book bu ild ing approach can be divided into two stages:the prelim inary inquiry and the accum ulated bidding inquiry.The issuer and the lead underw riter f rst determine the range of o fering p rices through a prelim inary inquiry and then conf rm the o fering p rice through the accumulated bidding inquiry.W hether an accumulated bidding inquiry is required dependson the f rm’s size and the time of the IPO.Before 2012,small f rms could choose whether to conduct an accum ulated bidding inquiry and this has been voluntary fo r all f rm s since 2012.Under the book building approach,theunderw riter seeks dem and inform ation from institu tional investors and determ ines the o fer price w ith the issuer according to the demand information received.IPO p ricing reform in June 2009 further improved the book building approach.However,China’s regu lator has not yet thoroughly adop ted themarket-based IPO pricing app roach used in developedmarketsand still controls IPO p ricing through“W indow Guidance.”4“W indow Guidance”is a kind of regulation w ith Chinese characteristics.According to“The M easures for the Adm inistrations of Securities Issuanceand Underw riting(Septem ber 19,2006),”the PEm ultip le used to determ ine the ofering p rice,generally,cannotexceed 30.In China,the regu lator’s“W indow Gu idance”can substantially reduce the o fering price.A lthough“W indow Guidance”wasgiven up in the IPO reform o f June 2009,itwas restarted under the IPO reforms o f M ay 2012.The history of IPO reform in China shows that the CSRC is still hesitant in deciding whether it isnecessary to regu late IPO p ricing.Thus research that discusses the prosand conso f IPO p ricing regulation is still highly relevant in a Chinese context.

Table 1 summarizes the history of China’s IPO system since the split share reform in 2006.

3.Theoretical analysis and predictions

3.1.IPO initial returns:underpricing or overvaluation

M any theorieshave been proposed to exp lain the puzzle o f IPO initial returns.These theories can be categorized by whether they assum e investorsare rational or not(Han and W u,2007).A lthough current studies m ainly explain IPO initial returns in theUnited Statesusing theoriesbased on asymm etric in formation,Ritter and W elch(2002)argue that information asymmetry cannot fully explain initial returns in the United States, which can beashigh as18%,and call formoreexplanationsbased on non-rational(or sem i-rational)investors or issuers.Purnanandam and Swam inathan(2004)fnd that the median IPO o fer p rice is overvalued by 14–50%relative to industry peers,depending on the peer-m atching criteria,which suggests that initial returns m ay be a resu lt of optim istic investo r sentim en t.O ther studies such as those of Ljungqvist et al.(2006),Derrien(2005)and Dorn(2009)show that an overvalued f rst day closing p rice is a result of irrational investor sentiment.

In China,some studies use information asymmetry theory to explain initial returns(e.g.,Guo and Zhao, 2006;W ang et al.,2009;Zhang and Liao,2011),while othersuse investor sentiment theory to exp lain initial returns(e.g.,Cao and Dong,2006;Jiang,2007;Han and W u,2007).W e believe that extremely high initial returns in China(up to 66.3%in the period from 2006 to 2011)are unlikely to be ab le to be explained under the assum ption o f rational investors.In China,the real question is perhaps notwhether the f rst-day closing p rice of IPOs isovervalued or not,but the extent o f thisovervaluation.The extent o f the overvaluation isan unanswered question thatm ust be answered w ith em pirical evidence.

Stud ies have shown that the f rst-day closing p rice of IPOs usually reverses after the IPO,which raises the question o f post-IPO long-run underperfo rm ance(R itter and W elch,2002;Jiang,2007).This question is related towhether thehigh initial returnso f IPOsaredue to underpricing or overvaluation.If the initial return m ainly resu lts from underp ricing,then itwou ld be expected that initial returnswould either not be related or positively related to post-IPO long-run performance.Intuitively,IPO underp ricing shou ld not be related to the long-run aftermarket performance.However,signaling theory in fnance suggests that IPO underp ricing is positively related to post-IPO long-run performance.According to the signaling model,high-quality IPO f rms aremore likely to set a lower o fer p rice in an IPO,which deters lower quality f rms from im itating the f rm and recoups their up-front sacrif ce post-IPO through future issuing activity(W elch,1989;G rinblatt and Hwang,1989).The signaling model has received some support from empirical evidence(e.g.,Su and F leisher,1999).However,if initial retu rns are m ain ly a result of the overvaluation o f the f rst-day closing p rice,we would expect the initial returns to be negatively related to post-IPO long-run perform ance because the overpriced f rst-day closing price w ill be co rrected gradually by the secondary m arket.

In sum,according to this argument,if we split the initial return into underpricing and overvaluation,we p redict that underp ricingw illnotbe related to orw ill be positively related to post-IPO long-run performance, and that overvaluation w ill be negatively related to post-IPO long-run performance.

3.2.Determ inants of underpricing and overvaluation

In a developed m arket,themain participants in an IPO are the issuer,the underw riter and the investors. What is special in China is that the regulator isalso an important participant in IPOs.Under the book building approach,the issuer and theunderw riter negotiateover the resu ltof the prelim inary inquiry and determ ine theof er p rice,but the regulator can cap themaximum PEmultip le of theo fer p rice.W eexpect that,in China, a f rm’s IPO of er pricew illbemainly determ ined by the characteristicso f the issuer and the underwriter,and whether there is IPO pricing regu lation.Additionally,the f rst-day closing pricew ill bemainly inf uenced by investor sentiment,and the characteristics of IPO f rm smay in fuence the efectso f investor sentiment on the f rst-day closing price.Thus,we expect a f rm’s f rst-day closing price to bem ain ly af ected by investor sentim ent and the issuing f rm’s characteristics.Due to space lim itations,we on ly consider the m ost im portan t characteristics of the issuer and the underw riter,nam ely,the value uncertainty of IPOs and the underw riter’s reputation.

3.2.1.The value uncertainty of IPOs

The value uncertainty of IPOs contributes to underpricing because there is asymmetric information between the issuer and the investors in that the true value of the IPO f rm is known by the issuer but not by the investors.Therefore,investors require a lower of er price to compensate for their information uncertainty risk and the issuer needs to set a lower o fer p rice to attract these unin formed investors(Beatty and Ritter,1986).According to thisargument,wepredict thata higher value uncertainty o f IPOsw illbeassociated w ith greater IPO underp ricing.

The value uncertain ty o f IPOs can also afect overvaluation.Existing studies suggest that the stock p rices of f rms w ith a higher value uncertainty are more likely to be afected by investor sentim ent and specu lative behavior(e.g.,Baker and Wurgler,2007).According to M iller(1977),under the assump tion of a short-sale constraint and heterogeneous expectations,stock prices only refect themost optim istic investors’expectations.Given that a higher value uncertainty is often related to higher heterogeneous expectations,the stock pricesof f rmsw ith a higher valueuncertainty aremore susceptible to op tim istic investor sentiment.Additionally,higher value uncertainty isalso related to speculative behavior,and thuscan lead to the overvaluation of stock prices.Given that investor sentiment and specu lative behavior are common in China’sstockmarket,we predict that a higher value uncertainty for IPO f rm s is positively related to the overp ricing o f the f rst-day closing p rice.

3.2.2.The underw riter’s reputation

To reduce the risk o f value uncertainty facing outside investors,the issuer can signal its fundamental value inmany ways,including by increasing the retained p roportion of outstanding shares(Brealey et al.,1977)and employing a reputed underw riter(Beatty and Ritter,1986).Researchers in China have found evidence that the reputation of the auditor and venture capital companies have an efect on the extent of IPO underp ricing (W ang et al.,2009;Zhang and Liao,2011),but these studies fail to f nd evidence that the underw riter’s reputation af ects underp ricing(Guo and Zhao,2006;Song et al.,2011).

Theoretically,the reputation of the underw riter cou ld either increaseor decrease the extent of underp ricing. On the one hand,according to the signaling hypothesis,underw ritersw ith a higher reputation could send a positive signal to outside investors and m itigate the value uncertain ty o f IPOs,thus reducing underp ricing (Beatty and R itter,1986).On the other hand,acco rding to the conf ict of interest hypo thesis,there isa con f ict of in terest between the underw riter and the issuer.Underw riters have an incen tive to set a lower ofer p rice to reduce the risk o f IPO failureand to cater to their customers(institutional investors)(Beatty and Welch,1996; Guo and Zhao,2006).Compared w ith underw ritersw ith a low reputation,thosew ith high reputationshave stronger bargaining power in setting the of er price and are thusmore capab le of increasing IPO underp ricing. Furtherm ore,compared w ith large f rms,small f rms haveweaker bargaining power in setting the o fer price and thus their shares aremore likely to be underpriced by their underw riters.According to this analysis,we therefore do not have a clear prediction of the direction o f the ef ect o f an underw riter’s reputation on the extent o f underp ricing.

3.2.3.IPO pricing regulation

Asmentioned in the institutional background section,during some periods in our samp le,the regulator stipu lated that the ofer prices of IPO f rmswere not allowed to exceed a certain PE mu ltiple(for examp le, 30).In this situation,IPO pricing regulation wou ld have decreased the o fer price and thus increased IPO underp ricing.Liu etal.(2011)haveshown that deregu lating IPO p ricing indeed reduces the levelof underpricing(asmeasured by IPO initial returns).W e therefore predict that IPO p ricing regu lation is positively related to underp ricing.

3.2.4.Investor sentiment

Investo r sentim ent afects underp ricing.The puzzle o f the ho tm arket iswell docum en ted in academ ia.The initial retu rns and issue vo lum e o f IPOs fuctuate periodically.The initial returnso f IPOs du ring som e periods arem uch higher than the average level and there arem ore f rm s going public in som e periods than in others (e.g.,Ritter,1984).Thepuzzle of thehotmarket suggests that IPO f rms takeadvantage of investors’op tim istic sentim ent.Under the chargemode of underw riting in China,5It is reported thatunderw ritersgenerally chargeunderw riting by segment.W ithin the rangeo f the issuer’sexpected fnancing amount, theunderw riter can usually obtain an underw riting feeof about 3%(which is rather low)but for the partexceeding theexpected amountof f nancing the underw riting can charge asmuch as 10%.we believe that underw riters have an incentive to exploit the optim istic sentiment o f investors to increase an IPO f rm’s ofer price,thus increasing their underw riting fee.Therefore,we p redict that therew ill be a negative relationship between investor sentiment and underp ricing.

Investor sentiment also afects overvaluation.According to the hypothesis of heterogeneous expectations p roposed by M iller(1977),in a marketw ithout short-selling,when investors have divergent opinions about a f rm’s fundamentalvalue,thestock priceo f thisstock w illon ly refect theexpectationso f themostop tim istic investo rs,leading to an overvalued stock price.Past studies,such as those by R itter and W elch(2002)and L jungqvist et al.(2006),have shown that investor sen tim en t can exp lain the high initial returns o f IPOs in theUnited States.In China,short-selling isnotavailab le,and there iso ften a very high divergence of opinions among investors(asshown by thehigh turnover in the f rst-day trading of IPOs).So the closing p ricesof Chinese IPO f rmsarevery likely to be sub ject to op tim istic investor sentim ent.Studies from China,such as Jiang (2007)and Han and Wu(2007),have also found evidence that investor sentiment is related to the overvaluation o f f rst-day closing p rices.W e thus p redict that investor sentim entw illbe positively related to overvaluation.Additionally,according to the p revious analysis,we p redict that the higher the value uncertainty o f IPOs,the larger the ef ect of investor sentiment on the overvaluation.

Overall,we p redict that the value uncertainty o f IPOsw ill be positively related to both underpricing and overvaluation,the underw riter’s reputation w ill be either be positively o r negatively related to underp ricing, IPO p ricing regulation w ill have a positive inf uence on underp ricing,investo r sen tim en t w ill be positively related to overvaluation,and the value uncertainty of IPOs and investor sentim en t w ill have an interactive efect on overvaluation.Table 2 summarizes these predictions.

4.Sample andmethodology

4.1.Sample and data sources

To testour predictions,we initially collected data from the IPOsof 994 f rms for theperiod from September 19,2006 to December 31,2011.We selected September 19,2006 as the start time because this was the date when the CSRC enacted the“M easures for the Adm inistrations o f Secu rities Issuance and Underw riting.”It was also at this tim e that the sp lit share refo rm ended.By selecting IPOs that occurred after the new regu lationswere enacted in Septem ber 2006,we avoid poten tial d iscrepancies in the data caused by changes to regulations and institutions.Our samp le period ends on December 31,2011 becausewewant to ensure that therewas at least one year of post-IPO stock perform ance data for each exam ined f rm.Based on the initial sample,we exclude f rmsw ithout suf cient data on analyst forecasts,stock p ricesand other f rm characteristics,which left a sam ple of 948 IPO f rms.The p rocess o f samp le selection is reported in Tab le 3.

Table 2 Predictions of the determ inants o f IPO underp ricing and overvaluation.

Table 3 Sam ple selection p rocess.

Analyst forecast data was collected from theW IND and CSMAR(China SecuritiesM arket and Accounting Research)databases,and all o ther data,includ ing IPO initial retu rns,post-IPO stock perfo rm ance and f rm characteristics,were ob tained from the CSM AR database.It is im po rtan t to note that W IND collects analyst fo recasts from m any secu rities com panies but not all of them,so we co llected add itional data on analyst forecasts from CSM AR to comp lement the om ission o f some o f these forecasts in theW IND database.

4.2.M odel construction and variable defnitions

4.2.1.M odel construction

W e use the follow ing m odels to exam ine the ef ects of IPO initial returns,IPO underpricing and IPO overvaluation on post-IPO long-run stock performance:

In M odel1,thedependent variable is BHAR(Buy and Ho ld AbnormalReturn),which represents thepost-IPO long-run stock performance.This can be exp ressed as BHAR240,BHAR480 and BHAR720,rep resenting the one-year,two-year,and three-year post-IPO BHAR,respectively.Themain independent variable is IR, which is a p roxy for IPO initial returns.In addition,draw ing on the literature(e.g.,Zhang and Liao,2011; Song et al.,2011),we include Underw riter(underw riter repu tation),Topone(shareho ld ing p roportion o f the largest shareho lder),EPS(earnings per share),Age(f rm age),Size(f rm size),I.Board(dumm y variables for listing board ef ects),I.Year(dumm y variab les for year efects)and I.Indu(dummy variables for industry efects)as contro l variab les in ourmodel.Based on our theoretical analysis,we predict that IR isnegatively related to BHAR.The variab les used in M odel 1 are defned in detail in Table 4.

In M odel2,themain independent variablesare IRUP(IPO underp ricing)and IROP(IPO overvaluation), instead of IR.The control variab les are the same as those in M odel 1.According to our theoretical analysis, we p redict that IROP w ill have a negative relationship w ith BHAR,and that IRUP w ill not be related to or will be positively related to BHAR.

We use the follow ing model to investigate the determ inants o f IPO initial returns,IPO underp ricing and IPO overvaluation:

In M odel3,the dependen t variable can either be IR(IPO initial return),IRUP(IPO underpricing)o r IROP (IPO overvaluation).Based on our theoretical analysis,we include the follow ing four variab les as themain independent variables:Uncer(value uncertainty of the IPO),Underwriter(underw riter reputation),PEcontrol (whether IPO pricing is regu lated)and Sent(investor sentiment).Referring to the literature,such as Zhang and Liao(2011)and Song et al.(2011),we control for the follow ing variab les in M odel3:Topone(shareho lding proportion o f the largest shareholder),EPS(earnings per share),Age(f rm age),Size(f rm size),I.Board (dumm y variab les for listing board ef ects),I.Year(dummy variab les for year ef ects)and I.Indu(dumm y variables for industry efects).W e p redict that Uncer,Sent and PEcontrol are positively related to IR;that Uncer and PEcontrol are positively related to IRUP;that Uncer and Sent are positively related to IROP;and that Underw riter is either positively o r negatively related to IRUP.

4.2.2.Variable defnitions

4.2.2.1.IPO underpricing and IPO overvaluation.Estimating an IPO f rm’s intrinsic value(or fundamental value)is the key to measuring IPO underpricing and overvaluation.Two methods for estim ating intrinsic value are used and com pared in ou r study:the m ethod of analyst fo recasts and the m ethod of com parison to sim ilar f rm s.

4.2.2.1.1.Themethod of analyst forecasts.Taking advantageof uniquedata from analyst forecasts,wemeasu re the intrinsic value of IPO f rm s based on analyst fo recast prices.6By reading the abstracts of analyst reports,we fnd that analysts forecast the intrinsic value(reasonable price)o f an IPO f rm rather than the f rst-day closing p rice.Therefore,itmakes sense to use analyst forecast prices tomeasure intrinsic value.W e f rst com pu te them ean o f each analyst’s fo recast p rices(analystsusually fo recast a range o f in trinsic values for each IPO f rm),and then calcu late themean o f all analyst forecast p rices thatwere released before the IPO.7Fo r instance,an IPO f rm receives a p rice fo recast from three analysts,and their forecast prices are 11–13,12–14 and 13–15, respectively.Them ean forecastprices for each analystare12((11+13)/2),13 and 14 respectively,and theoverallm ean forecastprice is13 ((12+13+14/3)).Wewould then use thismean forecast price(13)to measure the intrinsic value of the f rm.We exclude observations in which forecast p ricesare provided by analysts af liated to the underw riters and observations that are not the analyst’s last forecast price released before the IPO.8During theentireprocessofsam pleselection,weobtained 11,471 analyst forecasts.A fter deleting 136 forecasts from analystswhowere af liated w ith underw riters and 532 repeated forecasts,we had a f nal totalo f 10,957 analyst fo recasts for all of the sam p le f rms.In our dataset,92%of IPO f rmshave at least three analysts p roviding forecast pricesand 74%of IPO f rmshave at least fveanalystsp roviding forecast p rices.Given this large number o f analyst forecasts,we can avoid the subjectivity and randomness thatwould result if forecast p rices from a single analystwere used.

4.2.2.1.2.Themethod of comparable f rms.Fo llow ing Purnanandam and Swam inathan(2004),we use the p roducto f the IPO f rm’s industry peers’PEsand the IPO f rm’sEPS tomeasure the f rm’s intrinsic value.For each IPO in our sample,we fnd a non-IPO industry peerw ith comparab le sales and net income that did not go public in the previous three years.Specif cally,we select comparab le f rmsusing three steps:(1)to obtain app rop riate comparab le f rms,we f rst consider all f rms in the CSM AR database for the fscal year before the IPO and then exclude f rm sw ith negative PE ratios orw ith PE ratios exceeding 1009Generally,f rm s with a PE ofmo re than 100 are rare and are thus not suitab le asmatching f rms.and f rm s thatwent public in the previous th ree years;(2)we group f rm s in each industry in to 4(2*2)portfo lios based on sales and net incom e,and also group the IPO f rm s(ou r sam p le f rm s)into 4(2*2)portfolios in the sam eway;(3)each IPO f rm is then matched to the approp riate industry-sales-incom e bracket.Using this portfo lio,we select a comparab le f rm that is closest in sales size to the IPO f rm.

Webelieve that compared w ith thismethod of selecting comparable f rms,using analyst forecastsmay have some advantages in predicting an IPO f rm’s intrinsic value.First,given that analysts are experts in their industry,theymay beable to choosemore appropriate comparab le f rms(mostof the analystsuse comparable f rms’PEs to estimatean IPO f rm’sPE).Additionally,analystsgenerally adjust theestimated PE o f IPO f rms according to f rm-specif c information,such as the extent of industry competition and grow th potential,so their estimation o f an IPO f rm’s PE ratio may more accurate.

4.2.2.2.Post-IPO long-run stock performance

Table 4 Def nition of variables.

Post-IPO long-run stock performance ism easured by BHAR.In theabove formu la,BHAR(2,n)is thebuy and ho ld abnormal return of IPO f rms from thesecond day to the n th day after the IPO(thisdoesnot include the f rst day of the IPO);Rmtis themarket return for day t,where t belongs to the range from 2 to n.According to the requirementso f diferent stock performance periods,we setn to be equal to 240 days,480 daysand 720 days,which results in m easurements for BHAR240,BHAR480 and BHAR720,as reported in Table 4.10Because of the lim itations o f the dataset,the num ber of observations for BHAR720 and BHAR480 is sm aller than for BHAR240 and thus when we use BHAR720 and BHAR480 as dependent variab les the sam p le is sm aller than when the full sam p le is used.

4.2.2.3.Investor sentiment.There isno standard way in academ ia to measure investor sentiment,and scho lars usually useprincip le componentanalysis to integrateseveralvariables into a comprehensive factor for investor sentiment.In the spirit of Baker and W urgler(2007)and Wu etal.(2012),we use the follow ing four variables to create an overall investor sentiment index:(1)market turnover(monthly data);(2)discountsof closed-end funds(m onthly data);(3)the number of shareho lders that open new stock accounts(monthly data);(4)and the three-month stock return of the ShanghaiComposite Index.

4.2.2.4.Value uncertainty and underwriter reputation.In previous studies,the divergence o f analyst earnings forecasts is o ften used to measure a f rm’s information uncertainty(e.g.,Barron et al.,1998;Zhang,2006).Thus,we use the d ivergence of analyst fo recast prices to m easure an IPO f rm’s value uncertain ty.Intuitively, the repu tation of the underw riter w illalso be associated w ith their underw riting incom e(e.g.,M egginson and W eiss,1991;Liu et al.,2011),thuswe use the rank o f underw riting incom e to m easure underw riter reputation and create a variable that equals 1 if the underw riter is in the top 10,and 0 otherw ise.11U sing the top 10 as the cut-of point is comm on in the literature.Our emp irical results are not af ected if the top eight is used as the cut-of instead o f the top 10.

Table 5 D escriptive statistics.

4.3.Descriptive statistics and correlation analysis

Table 5 p resents the descrip tivestatistics for themain variables.The resu lts in Table5 show that(1)during our sampleperiod(2006–2011),them ean IPO initial returns in Chinawere66.3%;(2)when intrinsic valuewas m easured using analyst forecast p rices,themean IPO underpricing and overvaluation were 22.2%and 44.1%, respectively,butwhen the intrinsic valuewasmeasured using a comparable f rm’sPE and the IPO f rm’sEPS, themean underp ricing and overvaluation were 13.6%and 52.7%,respectively;(3)compared w ith those estim ated using analyst forecasts,the underp ricing and overvaluation estimated using com parable f rmshave largerm inim um s,m axim um s and standard errors,suggesting that using analyst forecasts is a m o re p reciseway to pred ict in trinsic value(because it is less likely to p roduce extrem e values);(4)them ean BHAR 240 days, 480 days and 720 days after the IPO were-14.4%,-7.3%and 4.1%,respectively,and the m edians were -14.2%,-17.7%and-7.9%,respectively,suggesting thatmore than half of IPO f rmsunderperform themarket and conf rm ing that there is a phenom enon of post-IPO long-run underperformance in China.

Table 6 reports the correlation analysis of the main variab les.The resu lts in Tab le 6 show that(1)the correlation coef cient between IR(IPO initial returns)and IROP(IPO overvaluation)is larger than the coef cientbetween IR and IRUP(IPO underp ricing),which indicates that the sizeo f the initial return ismore related to overvaluation than overp ricing(we thus conclude that overvaluation accounts for a larger part o f the initial return);(2)IRUP1 and IRUP2,and IROP1 and IROP2 are positively and signif cantly related, suggesting that the two methods we use to estimate intrinsic value are at least som ewhat consistent;(3)IR, IROP1 and IRUP2 are negatively and signif cantly related to BHAR240(the one year post-IPO abno rm al retu rn),suggesting that higher initial returns,higher overvaluation(as estim ated by analyst fo recasts)and higher underp ricing(asestim ated using com parable f rm s)are associated w ith poo rer post-IPO long-run stock performance;(4)IR,IRUP1,IRUP2,IROP1 and IROP2 arepositively and signif cantly correlated w ith Uncer (thevalueuncertainty of the IPO),Sent(investor sentiment)and PEcontrol(whether there is regulation of IPO p ricing);(5)IRUP1 is positively and signif cantly related to Underwriter(underw riter reputation);(6)andPEcontrol and Sent are positively and signif cantly related becausea period of bu llish investor sentiment coincided w ith a period of IPO p ricing regulation.However,it is important to note that the resu lts from the correlation coef cients on ly provide prelim inary conclusions and further regression analysis based on theory is needed.

Table 6 Correlation analysis.

5.Empirical results

5.1.IPO initial returns:underpricing or overvaluation

Fig.1 p lots IPO underp ricing and overvaluation asm easured by analyst forecast p rices fo r the sam p le f rm s from 2006 to 2011.It shows that,from 2006 to 2011,although the value o f initial returns f rst increases and then decreases substantially,overvaluation consistently accounts for a larger proportion o f the initial return than underp ricing.We believe that the huge fuctuation in the value of initial returns is the resu lt o f variation in the f rm characteristics,changes in the IPO p ricing regulation and thestockmarket environm ent during the sampleperiod.W e thuscategorizeour samp le into groupsby listing board,m arketenvironmentand IPO pricing regu lation,and then compare the initial return,underp ricing and overvaluation w ithin these diferent groups.We also compare the dif erence between the underp ricing and the overvaluation in each group.

Table 7 p resents the results based on this group comparison.The resu lts in the f rst row o f Table 7 show that the average initial return was 66%during the sam p le period.W hen in trinsic value is m easured by analyst forecasts,underp ricing is abou t 22%and overvaluation is about 44%(tw ice as large as underpricing).W hen in trinsic value ism easured using com parab le f rm s,underpricing is abou t 14%and overvaluation is about 53%(m ore than three times as large as underp ricing).The results for the grouping by listing board show that,when intrinsic value ismeasured by analyst forecasts,the initial return,underp ricing and overvaluation in themain board are lower than those in the SM E board and GEM board.When intrinsic value ismeasured using comparab le f rms,underpricing in them ain board is lower than in the other boards, but overvaluation in them ain board is almost the same as in the other boards.Third,the results for the grouping by investor sentiment show that,in periods of high investor investment,the initial return, underpricing and overvaluation are signif cantly higher than during periods w ith weak investor sentiment. The dif erence between underp ricing and overvaluation is also larger in periods w ith strong investor sentim ent.Fou rth,the resu lts for the group ing by IPO pricing regu lation show that,com pared w ith the period in which the CSRC regulated IPO pricing,the initial retu rn,underp ricing and overvaluation are signif cantly lower in the period when the CSRC deregulated IPO p ricing.In summ ary,the results from both the fu ll samp le and the sub-samp les show that overvaluation accounts for a signif cantly larger parto f the initial return than underp ricing and is thusmore important than underpricing in explaining the value o f initial returns in China.

Figure 1.IPO underpricing and IPO overvaluation from 2006 to 2011.

It is worth mentioning that the IPO f rms’intrinsic values estimated both by analyst forecasts and using comparab le f rms are likely to be afected by investor sentiment.However,we fnd that overvaluation is signif cant larger than underpricing both in periodsof high and low investor sentiment,suggesting thatmeasurem ent bias resulting from investor sentiment does not af ect our conclusion.Furthermore,because China’s stock m arket lacks a sho rt-selling m echanism,stock prices arem ore o ften overvalued rather than undervalued,which leads to the possibility that ou rm ethodsm ay overestim ate the in trinsic value of IPO f rm s.Therefo re,if it were possib le to rem ove this k ind o f positive deviation in estim ating in trinsic value,overvaluation would be likely to account for an even larger part o f the initial return,supporting the conclusion that IPO overvaluation accounts for a larger part of the IPO initial return than underp ricing.Therefore,we can conf dently conclude that an overvalued f rst closing price isamuch more important issue than the underp ricing o f of ers in China’s IPO market.

5.2.IPO initial return and post-IPO long-run stock performance

In this subsection,we further investigate the ef ects of IPO initial returns,underpricing and overvaluation on the post-IPO long-run stock perform ance.This subsection has two m ain purposes:to exam ine the relationship between these variab les and to evaluate and compare the twom ethodswe used tomeasure underp ricing and overvaluation.

The results in Table 8 show that(1)contro lling for other variab les,IR(IPO initial return)is signif cantly negatively related to BHAR(post-IPO long-run stock performance),suggesting that the initial return contains at least some component of overvaluation;(2)IROP(IPO overvaluation)is signif cantly negatively related to BHAR,which isconsistentw ith our prediction that theovervalued f rst-day closing p rice of IPO f rm s reverts to its intrinsic value;(3)when intrinsic value ismeasured by analyst forecasts,underp ricing(IRUP1)is positively related to post-IPO long-run performance,whereaswhen intrinsic value ismeasured using comparable f rm s,underp ricing(IRUP2)is negatively related to post-IPO long-run perform ance.Ou r theo retical analysis p redicts that underpricing wou ld either no t be related to or would be positively related to post-IPO long-run perform ance.W e thus believe that them ethod of analyst fo recastsm easu res underp ricing and overvaluation better than themethod o f using comparab le f rms.

We can also compare thedif erence in the coef cientsof IR and IROP1,IR and IROP2,IROP1 and IROP2 using the SUEST test.The resu lts in Tab le 8 show that,no matter whether BHAR240,BHAR480 or BHAR720 isused as thedependent variab le,the coef cientsof IROP1 aresignif cantly larger in their abso lute value than those of IR and IROP2,and the coef cientsof IR and IROP2 are not signif cantly diferent.These results imp ly that overvaluation estimated by using analyst forecasts isbetter at predicting post-IPO long-run performance than overvaluation estimated using comparable f rms.This resu lt further supports the conclusion that themethod of using analyst forecasts is better atmeasuring underpricing and overvaluation.

In add ition,the resu lts in Table 8 show that the regression coef cien tsof IR,IROP1 and IROP2 increase as the tim escale o f the dependent variables increases from BHAR240 to BHAR480 and BHAR720.This suggests that the f rst-day closing p rice of IPO f rm s reverts gradually to its intrinsic value.Lastly,the resu lts in Table 8 show that f rm performance(EPS)ispositively related to post-IPO long-run stock performance,and f rm size(Size)is negatively related to post-IPO long-run stock perform ance.This suggests thatm o re pro f tab le f rm s and sm aller f rm s have better post-IPO long-run perform ance.

Table 7 IPO underp ricing and overvaluation:group comparison.

5.3.Determinants of IPO initial returns,underpricing and overvaluation

The results for the determ inants o f IPO initial returns,overp ricing and overvaluation are p resented in Table9.In the foregoing subsection,we show that themethod of using analyst forecasts isbetter atmeasuring underpricing and overvaluation than that o f using comparab le f rms.In this subsection,we thus only report the results for the determ inants o f underp ricing and overvaluation asmeasured by analyst forecasts.

The results in Column 1 of Tab le 9 show that after contro lling for other factors,the value uncertainty of IPOs(Uncer),investor sentiment(Sent)and whether IPO pricing is regulated(PEcontrol)aresignif cantly positively related to IPO initial returns(IR).This suggests that f rmsw ith high value uncertainty and f rmsw ith IPOs in periods o f high investor sentiment have higher initial returns.IPO pricing regu lation by the CSRC signif cantly increases initial retu rns.The resu lts in Co lum n 2 show s that the value uncertainty of IPOs, whether IPO pricing is regu lated,and the underw riter’s repu tation(Underw riter)are all signif can tly positively related to IPO underpricing(IRUP1).The resu lts in Colum n 3 show that the value uncertain ty of IPOs (Uncer),investor sentiment(Sent)and whether IPO pricing is regu lated(PEcontrol)aresignif cantly positively related to IPO overvaluation(IROP1).Additionally,the resu lts in Columns1 to 3 show that f rm performance (EPS)is positively related to underpricing but negatively related to overvaluation,and that f rm age(Age)is signif cantly negatively related to overvaluation but insignif cantly related to underpricing.This suggests that f rm characteristicsmay have dif erent impactson underp ricing and overvaluation.Given thatmost o f these results are consistentw ith our predictions,we do not further elaborate on them in this section.

These resu lts show that Uncer and PEcontrol af ect both IPUP1 and IPOP1.We thus further exam ine the diference in theef ectso f these two variab leson IPUP1 and IPOP1.The results in Tab le9 show that the coeff cients of Uncer do no t dif er between Co lum n 2 and Co lum n 3(SUEST test:chi2(1)=0.28).However,the coef cien ts o f PEcontrol d if er signif cantly between Co lum n 2 and Colum n 3(chi2(1)=51.25),suggesting that whether the CRSC regulates IPO pricing has d if erent im pacts on underpricing and overvaluation.W e use the samp le period adjacent to the IPO pricing regu lation(2008 and 2009)to further exam ine the ef ectsof IPO pricing regu lations on underpricing and overvaluation.Because the m arket environm ent changes quick ly in China,lim iting our samp le to a narrower w indow w ill help us to evaluatem ore precisely the consequence of the deregulation o f IPO p ricing.The resu lts in Columns 4 and 5 show that,during the sample period o f 2008–2009(there are 73 observations before and 95 observations after the regulation),PEcontrol is signif cantly positively related to IRUP1 but not signif cantly related to IROP1.This indicates that deregulating IPO p ricing reduces underpricing but not overvaluation,which is consistentw ith our predictions.

The results in Column 2 show that the underw riter’s reputation is positively related to underpricing,supporting the conf ict of interesthypothesisbut not the signaling hypothesis.To further provide evidence for the con f ict of interest hypothesis,we classify the full samp le into two groupsby f rm size(issuance size)and f nd that a positive relationship between the underw riter’s repu tation and underpricing on ly exists in the IPOs of sm aller f rm s(these resu lts are reported in Co lum ns 6 and 7).A ccording to our theoretical analysis,these results are consistent w ith the conf ict o f in terest hypothesis.The resu lts in Co lum ns 6 and 7 also show that investor sentiment(Sent)is signif cantly negatively related to underpricing in small f rms,suggesting that underw riters take advantage o f investor sentiment toward small f rms to benef t themselves.

Our theoretical analysis suggests that value uncertainty in IPOsand investor sentimentmay have an interaction efect in overvaluation,but our regressions in Co lumns 1 to 3 only exam ine themain ef ect.In view of this consideration,we further add an interaction term Uncer*Sent in Columns 8 and 9.The results in Col-um ns 8 and 9 show that the coef cient of Uncer*Sent in Co lum ns 9 is signif can tly positive,suggesting that the higher the value uncertainty o f IPOs,the larger the ef ect of investor sentiment on overvaluation.

Table 9 Determinants of IPO initial return,IPO underpricing and IPO overvaluation.

Overall,the resu lts in this subsection are consistent w ith our theoretical analysis,indicating that distinguishing between underp ricing and overvaluation is vitalwhen investigating their determ inants.

5.4.Robustness tests

To check the robustness of our resu lts,we conducted a series o f sensitivity analyses related to aspects o f sample selection,model construction and variablemeasurement.Due to space lim itations,the resu ltsof these tests are no t repo rted but are availab le on request.

5.4.1.Sensitivity analysis on sample selection

(1)During the samp le period,therewas an important reform in IPO pricing.The CSRC deregu lated IPO p ricing in June 2009.Asa robustnesscheck,we divided the fullsamp le into two sub-samplesbefore and after this reform and then repeat our regressions.The resu lts show that themain conclusions from the fu ll samp le app ly to these sub-samp les.

(2)According to the def nitions of IPO underpricing and overvaluation,underp ricing and overvaluation shou ld,in general,be positive.Therefore,we exclude observationsw ith negative underp ricing or negative overvaluation(346 observations)and repeat our analysis.The results from this reduced sample do not change our m ain conclusions.

(3)Firm characteristicsm ay d ifer between listing boards,so we divide the sam p le in to two sub-sam p les: main boards and other boards(SM Esboard and GEM board)and repeat our resu lts.Again,themain conclusions from the full sample app ly to these sub-samples.

5.4.2.Sensitivity analysis onmodel construction and variablemeasurement

(1)W emeasure IPO f rms’intrinsic values by using analyst forecasts and com parison to sim ilar f rms.To reduce theoptim istic biaso f analyst forecasts,we used themedian andm inim um instead of themean o f analyst forecast prices to m easure in trinsic value and repeat our analysis.W e also use all of the f rm’s industry peers as com parab le f rm s rather than selecting a com parab le f rm to m easu re in trinsic value and repeat ou r analysis.W e also use the rank o f underw riter size(by total asset)instead o f underw riter income to measure underw riter reputation and repeat our analysis.Overall,the results show that these changes in variab lemeasurement do not af ect ourmain resu lts.

(2)In the analysis in Table 8,we include the variab les IRUP1 and IROP1,and IRUP2 and IROP2 into the regressionmodelat the same time.Asa sensitivity test,we add IRUP1,IROP1,IRUP2 and IROP2,one by one into the model.The results show that,although the size and signif cance o f the coef cient o f IRUP2 decrease somewhat,themain conclusions of our analysis remain the same.

6.Conclusions and implications

Ourm ain resu lts can be summ arized as follow s.First,average IPO initial returns are 66%,w ith underpricing and overvaluation between 14–22%and 44–53%,respectively,depending on themeasure used.Second, while both the initial return and overvaluation are signif cantly and negatively related to post-IPO long-run stock performance,overvaluation predicts post-IPO performance better than the initial return.Third,the value uncertainty o f IPOs is positively related to both underp ricing and overvaluation.Both underw riter reputation and p ricing regu lation arepositively related to underpricing.Investor sentimenthasa positiveef ecton overvaluation but hasno ef ect or a negative ef ect on underpricing.Overall,our resu lts suggest that themain reason for extremely high IPO initial returns in China is that f rst-day closing p ricesare overvalued,and that underp ricing and overvaluation have dif erent determ inants.

Our f nd ings have im portan t im p lications fo r bo th academ ics and regu lators.First,aswe have shown that overvaluation ismore important than underp ricing,future studies should paymore attention to the causesof overvaluation and tomeasures to controlovervaluation in IPOs.W hen deciding the path of IPO reform,regulators shou ld consider not only how to reduce underp ricing and increase p ricing ef ciency in the primary market,but also how to reduce overvaluation in the secondary market.Second,aswe show that separating underpricing and overvaluation from the initial return is necessary,future studies could use the methods described here tomeasure underp ricing and overvaluation,and then investigate their determ inantsmore p recisely and exp lore thepotentialef ectsof IPO reform more comprehensively.Finally,our resultsalso serveasa rem inder for scholars that using IPO initial returns to m easure IPO underpricing in China is likely to be inaccu rate.

This study is subject to som e lim itations.Them ost im portant is that the intrinsic value of IPO f rm s ishard to m easu re.A lthough them ethods used here have som e advantages,bias rem ains.W e expect that im proved measures could be put forward to evaluatemore precisely the relative importance o f IPO underp ricing and overvaluation,investigate their determ inantsmore accurately,and evaluate the consequences of IPO p ricing reform more ef ciently.

Acknowledgements

Wew ish to thank several anonymous reviewers,DonghuiWu(the editor),Zhaoyang Gu(the discussant) and other workshop participants at the 2013 CJAR Symposium for their insightful and helpful comments. However,we are com p letely responsible for the content o f this paper.W e gratefu lly acknow ledge suppo rt received from the National Natural Science Foundation o f China(G ran t No.71272196)and the Hum anities and Social Science Foundation o f Guangdong Province(G rant No.2012JDXM-0002).Thisstudy isalso supported by grants from the“Pro ject 211”Fund of the CentralUniversity o f Finance and Econom ics(CUFE), the“2011 Synergetic Innovation”Key Pro ject on“Development of Pub lic Accounting Profession”o f the CUFE,the Beijing M unicipal Comm ission o f Education“Joint Construction Project”and the Beijing M unicipal Comm ission o f Education“Pilot Reform of Accounting D iscipline Clustering”.

Baker,M.,W urgler,J.,2007.Investor sen timent in the stock m arket.J.Finance 21(2),129–151.

Barron,O.E.,K im,O.,Lim,S.C.,Stevens,D.E.,1998.Using analysts’forecasts to measure properties o f analysts’in formation environment.Account.Rev.4(73),421–433.

Beatty,R.P.,Ritter,J.R.,1986.Investmentbanking,reputation,and the underpricing of initialpublic of erings.J.Financ.Econ.15(1), 213–232.

Beatty,R.P.,W elch,I.,1996.Issuer expenses and legal liability in initial pub lic of erings.J.Law Econ.39,545–602.

Brealey,R.,Leland,H.E.,Pyle,D.H.,1977.Informational asymmetries,fnancialstructure,and f nancial intermediation.J.Finance 32 (2),371–387.

Cao,F.Q.,Dong,X.L.,2006.An empirical study on the rationality of IPO pricing.J.Finance Econ.32(6),3–14(in Chinese).

Derrien,F.,2005.IPO pricing in“HOT”m arket conditions:who leavesmoney on the table.J.Finance 60(1),487–521.

Dorn,D.,2009.Does sentiment drive the retail demand for IPOs?J.Financ.Quant.Anal.44(1),85–108.

Grinb latt,M.,Hwang,C.Y.,1989.Signalling and the p ricing of new issues.J.Finance 44(2),393–420.

Guo,H.,Zhao,Z.Y.,2006.An em pirical analysis of the efect o f underw riter reputation on IPO pricing,initial return and long-run perform ance.M anage.W orld 3,122–128(in Chinese).

Han,L.Y.,W u,Y.R.,2007.Investor sentimentand the puzzleof IPOs:underpricing or overpricing.M anage.W orld 3,51–61(in Chinese). Jiang,H.B.,2007.The pricing behavior of A-share IPO in inef cientmarkets.J.Financ.Res.8,90–102(in Chinese).

Liu,Z.Y.,Zheng,K.,H e,Y.N.,2011.Is the f rst-stage book building refo rm ef ective.J.Financ.Res.4,158–173(in Chinese).

L jungqvist,A.,Nanda,V.,Singh,R.,2006.H otm arkets,investor sentiment,and IPO pricing.J.Bus.79(4),1667–1702.

Lu,Z.,Zhou,H.F.,2007.M omentum and reversal in China’s stock market.Econ.Res.J.9,145–155(in Chinese).

M egginson,W.L.,Weiss,K.A.,1991.Venture capitalist certif cation in initial public o ferings.J.Finance 46(3),879–903.

M iller,E.M.,1977.R isk,uncertainty,and divergence of opinion.J.Finance 32(4),1151–1168.

Purnanandam,A.K.,Swam inathan,B.,2004.A re IPOs really underpriced?Rev.Financ.Stud.17(3),811–848.

Ritter,J.R.,Welch,I.,2002.A review of IPO activity,p ricing,and allocations.J.Finance 57(4),1795–1828.

Ritter,J.R.,1984.The“hot issue”marketo f 1980.J.Bus.57(2),215–240.

Song,S.J.,Cao,H.,Yang,K.,2011.Investor attention and IPO anomalies–evidence from Google Trend volum e.Econ.Res.J.S1,145–155(in Ch inese).

Su,D.,Fleisher,B.M.,1999.An emp irical investigation of underpricing in Chinese IPOs.Pac.-Basin Finance J.7(2),173–202.

Su,D.W.,2008.N oise trading and m arket quality in Chinese stock m arkets.Econ.Res.J.9,82–95(in Chinese).

Tian,L.H.,2010.On the fnancial regulation,investment risks and IPO underp ricing.J.Financ.Res.4,85–100(in Chinese).

Wang,B.,Xin,Q.Q.,Yang,D.M.,2009.Whether auditor reputation infuences share pricing—evidence from IPO market in China.Acc. Res.11,73–96(in Chinese).

W elch,I.,1989.Seasoned o ferings,im itation costs,and the underpricing of in itial public oferings.J.Finance 44(2),421–449.

Wu,Y.R.,Pan,K.,Hu,S.M.,Jiang,J.,2012.A new explanation of industry analyst earnings forecastbias.Econ.Res.J.4,149–160(in Chinese).

Xu,N.H.,Hong,T.,W u,S.L.,Xu,X.Z.,2011.Information fow m odel,investor psychologicalb ias and stock price comovem ent.Econ. Res.J.4,135–146(in Ch inese).

Zhang,X.Y.,Liao,L.,2011.VCs’backgrounds,IPO underpricing and post-IPO performance.Econ.Res.J.6,118–132(in Chinese).

Zhang,X.,2006.In formation uncertainty and stock returns.J.Finance 61(1),105–137.

*Corresponding author.

E-mail address:dalin507@163.com(S.Song).

Production and hos ting by Elsevier

1755-3091/$-see frontmatter?2014 Production and hosting by Elsevier B.V.on behalf of China Journal of Accounting Research. Founded by Sun Yat-sen U niversity and City U niversity of H ong Kong.

h ttp://dx.doi.org/10.1016/j.cjar.2013.12.001