Market Segmentation and Information Values of Earnings Announcements: Some Empirical Evidence from an Event Study on the Chinese Stock Market

Transcription

1 SMU ECONOMICS & STATISTICS TISTICS WORKING PAPER SERIES Market Segmentaton and Informaton Values of Earnngs Announcements: Some Emprcal Evdence from an Event Study on the Chnese Stock Market Y Gao Y K Tse October 2002 Paper No ANY OPINIONS EXPRESSED ARE THOSE OF THE AUTHOR(S) AND NOT NECESSARILY THOSE OF THE SCHOOL OF ECONOMICS & SOCIAL SCIENCES, SMU

2 Market segmentaton and nformaton values of earnngs announcements: Some emprcal evdence from an event study on the Chnese stock market Y. Gao Graduate School of Busnss Unversy of Chcago Chcago, IL and Y.K. Tse School of Economcs and Socal Scences Sngapore Management Unversy Sngapore Revsed: October 2002 Abstract: Ths paper nvestgates the tradng actves of two dstnct classes of shareholders, namely, the Chnese domestc nvestors and the foregn nvestors n the segmented Chnese A-share and B-share markets, respectvely. We conduct an event study on the annual earnngs announcements based on two dfferent accountng standards: IAS and PRC GAAP. The earnngs announcements based on IAS and PRC GAAP are value relevant. The nvestors n the B-share market react to both the IAS and PRC GAAP earnngs announcements, whle the nvestors n the A-share market pay more attenton to the PRC GAAP earnngs reports. In the B-share market, posve abnormal returns are assocated wh posve earnngs surprse and negatve abnormal returns go wh negatve earnngs surprse. We fnd pre-event abnormal tradng volumes whout sgnfcant prce changes for the A shares, whch may be due to exstng nformaton n the A-share market pror to earnngs announcements. The post-event abnormal tradng volumes last for a longer perod n the A- share market than n the B-share market. JEL classfcaton: G14; G15 Keywords: Earnngs announcement; Event study; Market segmentaton Acknowledgement: The authors are ndebted to the three anonymous referees for ther helpful comments and suggestons. The frst draft of ths paper, wh a dfferent tle, was wrten when both authors were at the Natonal Unversy of Sngapore. Correspondng author: Y.K. Tse, School of Economcs and Socal Scences, Sngapore Management Unversy, 469 Buk Tmah Road, Sngapore ; Tel: ; Fax: ; emal: yktse@smu.edu.sg

3 1. Introducton An mportant move n Chna s (People s Republc of Chna (PRC)) economc reform s to reactvate the stock exchanges after four decades of abandonment. Accompaned by regulatory reforms, a move to partally prvatze state-run enterprses appeared. The government selected some state-owned plants and restructured them to form new frms as parts of the orgnal enterprses. These frms were then lsted on one of the two exchanges, namely, the Shangha Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE), where two types of shares are lsted: A shares, whch are ssued strctly to the PRC czens, and B shares, whch are ssued only to foregners. Cross lstng between the two exchanges s not allowed. A frm may ssue state shares, legal-enty shares, A shares and B shares. However, only A shares and B shares are ssued to the publc and traded n the market. Xu (2000) provded an account of the mcrostructure of the Chnese stock market. Fung and Leung (2001) dscussed the legal aspects and the corporate-governance ssues of the Chnese fnancal markets. The A- and B-share markets are rgdly segmented, causng hgh prce premums of the A shares relatve to the correspondng B shares, although the owners of the A- and B-shares have the same rghts. Internatonal accountng standards (IAS) are used to prepare accountng reports for the B-share holders, whle Chna s accountng regulatons (PRC generally accepted accountng prncples (PRC GAAP)) are used to prepare accountng reports for the A-share holders. Most emprcal works on the segmentaton of the Chnese stock market compare the behavor of the A shares and B shares usng regresson and correlaton analyss. For example, Chu and Kwok (1998) examned the cross correlaton of the returns n the A- and B-share markets. Ther correlaton analyss showed that the returns n the B-share market lead the returns n the A-share market. Fung, Lee and Leung (2000) used a latent varable model to examne the prcng of the A and B shares. They argued that the prcng of the two markets reflects dfferent fundamental forces. Xu (2000) analyzed the tme seres return and volatly patterns of the Shangha market. Usng contegraton analyss, Chan, Cheng and Fung (2001) concluded that the A- and B-share prces do not follow the same dynamcs. Overall, these results suggest that there s a dfference n the prce dynamcs of the two markets, but there s also relevant nformaton flow between the two markets. It wll be nterestng to go deeper at the mcro level to examne f there s any dfference n the reacton of each market to news and nformaton. In ths paper we adopt the event study methodology and focus on the dfferental reactons of the A- and B-share nvestors to earnngs announcements. 1 We compare the behavor of the prce reacton of the A shares and B shares n response to the companes earnngs announcements under dfferent accountng standards. Whle tradng volume has generally been gnored n prevous research as an ndcator of dfferental market behavor, n ths paper we nvestgate the tradng volumes of the A shares and B shares n reacton to the earnngs announcements. As shown by He and Wang (1995), tradng volumes around earnngs-announcement dates have mportant mplcatons for the nformaton values of earnngs announcements. Our event study on the effects of annual earnngs announcements based on PRC GAAP and IAS on the A- and B-share markets shows that there are promnent abnormal returns and tradng volumes durng the wndows of earnngs announcements, whch s consstent wh 1 To the best of our knowledge, Poon, Frth and Fung (1998) s the only paper n the lerature that uses an event study to examne the segmentaton of the Chnese stock market. The event n Poon et al. s the nal lstng of the B shares. As forcefully argued by Fama (1991), an event study allows a researcher to examne the market n a mcroscopc manner and s nsensve to model msspecfcaton that may affect a regresson model. 1

4 prevous results that earnngs nformaton s value relevant n the Chnese stock market. The abnormal tradng volumes n the A-share market persst for a longer perod than those n the B-share market, whch supports the fndngs of Su and Flesher (1999), who reported that news s more persstent for the A-shares than for the B-shares. The pre-event abnormal tradng volumes whout sgnfcant prce changes for the A shares may be due to some preevent exstng nformaton n the A-share market. Usng the earnngs numbers based on PRC GAAP, we fnd sgnfcantly posve abnormal returns n the good-news event wndow and sgnfcantly negatve abnormal returns n the bad-news event wndow for both the A- and B- shares. Wh the earnngs compled under IAS, however, we fnd only sgnfcantly posve abnormal returns n the good-news event wndow and sgnfcantly negatve abnormal returns n the bad-news event wndow n the B-share market. In the A-share market sgnfcantly negatve abnormal returns are observed n the bad-news wndow but we fnd no sgnfcantly posve abnormal returns on the good-news event day. Ths result suggests that the Chnese nvestors are not much concerned about the accountng nformaton based on IAS. Ths agrees wh Chen, Frth and Km (2002), who reported that IAS earnngs are value relevant to B shares, whle nvestors n the A-share market put most weght on PRC GAAP. The rest of ths paper s organzed as follows. Secton 2 provdes a survey of the Chnese stock market. Secton 3 revews the relevant lerature on tradng and nformaton, and outlnes the hypotheses to be examned n the event study. Secton 4 presents the methodology and descrbes the data. Secton 5 reports the results and the nterpretaton. Secton 6 summarzes and concludes. 2. The Chnese Stock Market and Accountng Practces 2.1 The Shangha Stock Exchange and Shenzhen Stock Exchange There are two offcal natonal exchanges n Chna, namely, the SHSE establshed on December 19, 1990, and the SZSE founded on July 3, In the past ten years, the sze of the two exchanges expanded rapdly. At the end of 2000, there were 572 and 514 frms lsted on the SHSE and SZSE, respectvely. From 1992, the Chnese government allowed a selected lst of frms to ssue tradable shares to foregn nvestors, whch are called B shares so as to dstngush them from those ssued only to the PRC czens. Although the B shares are also lsted on the two domestc exchanges, they are denomnated n US dollars on the SHSE and n Hong Kong dollars on the SZSE. At the end of 2000, 559 A shares and 55 B shares were lsted on the SHSE, and 499 A shares and 59 B shares were lsted on the SZSE. 2.2 Market Segmentaton Lke many developng countres, Chna set up legal restrctons on the foregn ownershp of domestc equy n order to mantan the control over local frms, especally those companes that are of strategc and natonal mportance. A major reason for ths arrangement s to attract foregn funds whout worryng about the loss of ownershp control. Thus, n Chna a local frm may ssue two dfferent types of shares, namely, A shares and B shares. Foregn nvestors are only allowed to hold the B shares but not the A shares. On the other hand, Chnese czens cannot buy foregn currences freely. Thus, the local people have ltle chance to nvest n the foregn stock markets. The lack of nvestment alternatves s a possble reason for the prce premum of the A shares relatve to the B shares. 2

5 A unque feature of the Chnese stock market s that the markets for the A shares and B shares are completely segmented durng our perod of study, 2 whle the segmentaton n most other markets are only partal. In the latter case, foregn nvestors are allowed to own only the foregn class of shares, whle domestc nvestors can own both local and foregn shares. Thus, the stock market n Chna was completely segmented, although the owners of the A shares and B shares have equal rghts. The contrast between the scale and transacton n the A-share and B-share markets s very clear: the number of lsted frms, stocks, and ssued shares, the market capalzaton, the tradng volume, the deal number, and the turnover of the A shares are much larger than those of the B shares. At the end of 2000, there were mllon A-share accounts and mllon B-share accounts on the SHSE. On the SZSE, the fgures were mllon and mllon for the A-share and B-share accounts, respectvely. In the A-share market, ndvdual nvestors domnate, whle n the B-share market the percentage of nstutonal nvestors s much hgher than that n the A-share market Accountng Practces The companes that ssued B shares need to prepare two sets of fnancal statements, one s based on IAS for the B-share holders, and the other s based on PRC GAAP for the A-share holders. The PRC GAAP and IAS numbers are released to the share holders on the same day. The level of dsclosure s hgher for the B shares than for the A shares, and the IAS are more conservatve than the PRC GAAP. 4 Thus, the prof numbers, asset numbers, and book values dffer between the IAS and PRC GAAP statements. The dfferences between the two sets of numbers are dsclosed n the PRC GAAP statements, and the B-share annual reports nclude the A-share accounts as supplementary nformaton. Thus, there are no barrers to the A-share holders to access the data from the IAS fnancal reports. Lkewse, the B-share holders can access the data from the PRC GAAP reports. Hence, s possble to use both the PRC GAAP and IAS data to make nvestment decsons. 3. Lerature Revew and Formulaton of Hypotheses In ths secton we frst revew the lerature on the effects of nformaton on stock tradng, followed by a bref survey of related emprcal research on the Chnese stock market. 3.1 The Effects of Informaton on Beaver (1968) nvestgated the reacton of stock prce and volume to earnngs announcements. He argued that abnormal tradng volumes reflect the degree to whch ndvdual nvestors n the market revse ther expectatons n reacton to earnngs announcements, and abnormal returns reflect the aggregate or average revson n expectatons. Bamber (1986) demonstrated that earnngs announcements are nformatve about frm prospects. Km and Verreccha (1991a) developed an analytcal framework to show that when traders have dfferent belefs, the level of dfferental pre-dsclosure precson 2 On February 19, 2001, the Chna Secures Regulatory Commsson announced that Chnese czens are allowed to hold and trade B shares. However, our study perod does not extend to ths date. 3 At the end of 2000 the SHSE nstutonal nvestors held 0.41% of the market value of the A shares, whle n the B-share market the nstutonal nvestors owned 5.52% of the market value. On the SZSE, the fgures were 0.49% and 3.02% n the A- and B-share markets, respectvely. 4 Major dfferences between PRC GAAP and IAS are summarzed by Bao and Chow (1999). 3

6 of nformaton s related to the change n tradng volume n response to publc dsclosure of nformaton. Zebart (1990) and Ajnkya, Atase and Gft (1991) explaned the varatons n tradng volumes n terms of pre-dsclosure asymmetry. Bamber (1986, 1987), Holthausen and Verreccha (1990), Zebart (1990), Terpstra and Fan (1993), Kross, Ha and Hefln (1994), Atase and Bamber (1994) explaned the varatons n the tradng volume n terms of unexpected earnngs, frm sze, rsk changes, market-wde volume nfluences and the magnude of assocated prce reacton. Kandel and Pearson (1995) suggested a model based on dfferental nterpretaton around publc announcements. They demonstrated that abnormal tradng volume may occur even when the announcements do not produce value effects. Recently, He and Wang (1995) constructed a dynamc model of dfferental nformaton and behavor of stock tradng to ncorporate the emprcal fndngs n the lerature. The man results are: () Prvate nformaton does not only cause tradng n the current perod, but also generates possble tradng afterwards; () Publc nformaton leads to tradng n the current perod and hgh volume appears around the announcement date; and () New nformaton, prvate or publc, brngs about hgh volume accompaned by large prce changes whle exstng nformaton only generates hgh volume whout large prce changes. There has been some emprcal research on specfc markets. Terpstra and Fan (1993) analyzed forecasts of Hong Kong frms earnngs and gave addonal support for the tradngvolume theory that nvestor dsagreement over the nterpretaton of nformaton leads to ncreased tradng. Chung and Lee (1998) studed the Japanese stock market and nvestgated the dfferental tradng actves exhbed by dfferent types of nvestors n response to earnngs announcements. Cho and Choe (1998) provded emprcal evdence for the effect of annual earnngs announcements on nvestors tradng volume n the Korean stock market. In an event study, Bhattacharya, Daouk, Jorgenson, and Kehr (2000) nvestgated a sample of Mexcan corporate news announcements and found that there are no unusual reactons n returns, volatly of returns, volumes of trade and bd-ask spreads around news announcements. They attrbuted such a curous case to the unrestrcted nsder tradng and argued that rankng emergng stock markets n terms of ther market ntegry s a methodology that can be used wh the lmed data avalable n these markets. Ths argument shows that extendng the research to other emergng stock markets such as the Chnese stock market s of partcular mportance. 3.2 Selected Emprcal Lerature on the Chnese Stock Market Earler emprcal research on the Chnese stock market has focused on the correlaton structure of the A shares versus the B shares, and the puzzle that the A shares are traded at a hgh premum versus the B shares (Baley (1994); Ma (1996); Sun and Tong (1998); Baley, Chung and Kang (1999); Poon, Frth and Fung (1998); Su and Flesher (1999)). Recently, the relevance of accountng nformaton based on dfferent accountng standards (IAS and PRC GAAP) for the stock prces n the Chnese stock market has been extensvely studed. One lne of research s to examne the role of earnngs n the valuaton of lsted companes (Chen, Chen and Su, 2001; Haw, Q and Wu, 1999). The emprcal results suggest that the earnngs reported under PRC GAAP are value relevant to the Chnese nvestors. Another drecton s to compare the degrees of value relevance of earnngs compled under PRC GAAP and IAS (Bao and Chow, 1999; Chen, Frth and Km, 2002). Bao and Chow (1999) found that the earnngs based on PRC GAAP and IAS are both sgnfcantly assocated wh the B-share prces, but accountng nformaton based on IAS has greater nformaton contents for B shares than that based on PRC GAAP. Chen, Frth and Km (2002) reported that IAS earnngs are value relevant to B-share prces and returns, whle 4

7 nvestors n the A-share market put more weght on PRC GAAP. It s only untl recently that the A-share nvestors started to pay attenton to the nformaton based on IAS. 3.3 Annual Earnngs, Concurrently Announced Cash Dvdends and Stock Dvdends Annual earnngs, cash dvdends, and stock dvdends are usually announced concurrently n Chna. To conduct an event study on annual earnngs announcements, we need to dsentangle other nosy effects. Haw, Q and Wu (2000) studed the Chnese A-share market and found that good-news frms tend to release ther reports earler than the bad-news frms. Chen, Frth and Gao (2002) reported that unexpected earnngs announcements have an mpact on stock prces, that the earnngs sgnal s stronger (or weaker) when the earnngs surprse s corroborated (or dluted) by a stock dvdend surprse of the same (or oppose) drecton, and that unexpected cash dvdends have ltle mpact on the earnngs sgnal. In our study of earnngs announcements, we check the unexpected stock dvdends and exclude the announcements from the sample when the earnngs and stock dvdends change n oppose drectons. Thus, n event, f the earnngs surprse s posve (or negatve), but the correspondng stock dvdend per-share s less (or more) than the dvdend per-share announced n the prevous year s fnancal report, the event s deleted from the sample. 3.4 Some Hypotheses In ths secton we formulate some hypotheses to be examned n the event study. (1) Are the tradng volumes n the announcement perod accompaned by large prce changes? He and Wang (1995) showed that volume may lag behnd the nformaton flow when the nformaton s prvate. They demonstrated that exogenous nformaton, whch ncludes new prvate sgnals and publc announcements, generates tradng together wh large prce changes, whle volume generated by exstng prvate nformaton s not accompaned by sgnfcant prce changes. As defned by He and Wang, earnngs announcement s exogenous. Thus, we would expect to see hgh volume surroundng the news release accompaned by sgnfcant abnormal returns n both the A- and B-share markets. (2) Wll the nvestors react to both the IAS and PRC GAAP earnngs announcements? As ndcated n Chen, Frth and Km (2002), IAS earnngs nformaton s closely related to the prces and returns of B shares and PRC GAAP earnngs nformaton does not brng ncremental nformaton to the B-share nvestors. Lev (1988) argued that accountng nformaton has dfferent values to dfferent classes of nvestors. Km and Verreccha (1991a, 1991b) provded a theoretcal model to support Lev s argument. Cready (1988) reported that large (nstutonal) nvestors are more responsve to earnngs announcements than small (ndvdual) nvestors. Snce most B-share nvestors are large nternatonal fnancal nstutons and foregn nvestors fnd more dffcult to acqure local nformaton, they tend to focus on and utlze the accountng reports. These arguments constute the ntuon for the followng hypothess: H 1 : Foregn nvestors react to both the IAS and PRC GAAP earnngs announcements. 5

8 Chen, Frth and Km (2002) found that Chnese nvestors put more weght on PRC GAAP earnngs and started to use the IAS earnngs nformaton only recently. Thus, we set up the second hypothess as follows: H 2 : Chnese nvestors react to the PRC GAAP earnngs announcements, and, to a less extent, the IAS earnngs nformaton. (3) Does publcly released news based on dfferent accountng standards have any predctable effects on the A- and B-share markets? Publcly released news may generate market reacton. How the markets respond to dfferent news s an nterestng queston. We examne the effects of good and bad news on the segmented markets and establsh the followng hypotheses: H 3 : Good news generates posve abnormal returns on announcement dates n both the A- and B-share markets. H 4 : Bad news generates negatve abnormal returns on announcement dates n both the A- and B-share markets. 4. Data and Methodology 4.1 The Data The perod under study s from June 1995 through May The Appendx provdes the lst of selected companes from SHSE and SZSE. Several crera were used to screen the data. The frst step s to select the stocks. We pck the frms that have lstngs n both A shares and B shares. In the SHSE we obtan 41 frms, whle n the SZSE the number s 42. Among these, one frm on the SHSE and 21 frms on the SZSE are followed by fewer than three analysts from the Internatonal Brokers Estmate System (IBES). These frms are screened out. The last requrement for the remanng 40 frms on the SHSE and 21 frms on the SZSE s that ther stocks should be traded n the market for at least one year. Fnally, 112 annual earnngs announcements from 38 frms on the SHSE and 44 annual earnngs announcements from 21 frms on the SZSE survve the screenng. We conduct our study on two dfferent reportng standards, namely, IAS and PRC GAAP. The sample szes under dfferent crera of earnngs surprse based on the two accountng standards are summarzed n Table 1. Daly stock prces, tradng volumes, market ndces, the forecast of earnngs and number of analysts from IBES were collected from the Datastream. 5 Actual earnngs and dvdend nformaton were obtaned from the annual reports of the companes. Dates of announcements were gathered from the SHSE and SZSE. 6 5 In Chna there are no offcal earnngs forecast reports. We use the earnngs forecasts by IBES, an nstuton well-known for the US frms. The frms n our sample are studed by at least 3 analysts durng the examnaton perod. Ang and Ma (1999) measured the transparency of the Chnese capal market by means of the ndvdual analysts forecasts. They argued that although the errors of analysts forecasts for the Chnese stocks are much hgher than those for Hong Kong and other Asa Pacfc stocks, the forecasts are stll of value. 6 The dates of the announcements were collected from the offcal webses of the SHSE ( and SZSE ( Both webses are authorzed by the Chnese Secures Regulaton Commsson (CSRC). We also cross-checked the report dates aganst the Chna Secures, Shangha Secures and Secures Tmes, whch are authorzed fnancal newspapers n Chna. 6

9 4.2 The Methodology Classc event-study methodology s appled to examne the nformaton contents of earnngs-per-share announcements. The announcement date s defned as day 0, and the estmaton perod s from day 150 to day 21. The total estmaton perod covers 130 tradng days. The event wndow of nterest begns from day 20 and ends on day +10. Let R denote the return of a secury represented as the h event on day t. Usng the contnuously compoundng method, R s calculated as (dvdend s ncluded when s dstrbuted): R (1) = ln P ln P, t 1 where P s the secury prce n the h event on day t. We follow Brown and Warner (1985) and calculate the abnormal daly return usng three dfferent measures: mean adjusted, market adjusted and market-model adjusted. Let A denote the excess (abnormal) return of the secury n the h event on day t. The daly excess return n the event wndow s calculated usng the followng three methods: (1) mean-adjusted return, A = R R, where R s the arhmetc average of the secury s daly returns n the ( 150, 21) estmaton perod of the h event, (2) market-adjusted return, A = R R mt, where R mt s the return of the ndex on day t, and (3) market-model-adjusted return, A = R ˆα ˆβ R mt, where ˆα and ˆβ are obtaned usng ordnary least squares. The ndex return n each segmented stock market s used to proxy the market return. We adopt the returns of the Shangha A-share stock ndex, the Shangha B-share stock ndex, the Shenzhen A-share stock ndex, and the Shenzhen B-share stock ndex to represent the market returns of the Shangha A-share, the Shangha B-share, the Shenzhen A-share and the Shenzhen B-share markets, respectvely. These ndces are all value-weghted. The ndex return R s also computed usng the contnuously compoundng measure,.e., mt R ln P ln P (2) mt = mt m, t 1 where P mt s the market ndex on day t. We consder the abnormal returns and calculate both the parametrc t-statstc and the non-parametrc Corrado (1989) rank statstc. However, as the results for the two approaches are qualatvely smlar, we only report the results for the parametrc t-test. The daly crosssectonal average excess return of all the secures n the N events, namely A t, s calculated as: A = 1 N t A N = 1 (3) and the t-statstc for A t s computed as: T = A S( A ) (4) t t 7

10 where S( A ) s the standard devaton of the abnormal return n the estmaton wndow defned as: 1 S ( A) = (5) A t t= 150 To dfferentate the markets reacton to dfferent nformaton, we classfy the annual earnngs announcements nto good news and bad news. Usng the annual earnngs numbers based on IAS and the earnngs forecasts data collected from IBES, we defne and calculate the earnngs surprse accordng to the followng formula: ES = ( AE EE ) EE (6) where AE s the actual earnngs based on IAS n the h event and EE s the estmated earnngs reported by IBES. Snce the IBES earnngs forecasts are for the B shares and there are no offcal earnngs forecasts for the A shares, we use another formula to calculate the earnngs surprse when the annual earnngs numbers are based on PRC GAAP,.e., ES (7) = ( AE AE, t 1 ) AE, t 1 where AE, t s the actual annual earnngs based on PRC GAAP reported n the h event and AE, t 1 s the prevous year s annual earnngs based on PRC GAAP for that company. To screen out the good news, we try two dfferent thresholds. Good earnngs surprse s defned as ES above 0% and 20%. Smlarly, for the bad news, the two standards are set as earnngs surprse below 0% and 20%. However, the results for the two thresholds are que smlar, and we shall only report the results of good news defned by ES > 20% and bad news defned by ES < 20%. To examne the changes n the tradng volume upon earnngs announcements, abnormal daly tradng volume s calculated as the dfference between the tradng volume and the mean daly volume of that stock over the event perod normalzed by the standard devaton. Each stock s daly tradng volume s the turnover as measured by the number of shares. Followng Brown and Warner (1985) and Corrado (1989), a t-test s appled to examne the sgnfcance of the normalzed abnormal tradng volume n the event wndow. Specfcally, we defne EV as the excess tradng volume of the secury n the h event on day t,.e., EV = V V (8) where V s the volume of the h event on day t and V s the smple average of the secury s tradng volume of the h event durng the ( 150, 21) estmaton perod. The standarddevaton-normalzed abnormal volume, E, s calculated as: E = EV S( EV ) (9) where SEV ( ) s s estmated standard devaton gven by: 8

11 1 S ( EV ) = 130 t= EV (10) 2 For each day t, the cross-sectonal average abnormal (excess) volume of all the events, E, s computed as: t Et = 1 N E N = 1 (11) and the t-statstc for day t s obtaned by T = 1 N t E = 1 N (12) Ths test statstc depends on the cross-secton ndependence of the secures excess volume for correct specfcaton. Brown and Warner (1985) showed that the test has good power when the ndependence assumpton holds. 5. Emprcal Results 5.1 s and s Tables 2 and 3 present the abnormal returns and abnormal tradng volumes n the event wndows of good and bad news, respectvely, based on IAS earnngs announcements. Lkewse, Tables 4 and 5 report the results for the PRC GAAP earnngs announcements. As the results for the three methods of calculatng the abnormal returns descrbed n Secton 4.2 are very smlar, to save space we only report the results for the market-model-adjusted abnormal returns. Usng a one-tal test at the 5% level of sgnfcance, we hghlght the sgnfcant t-statstcs n bold face. It can be seen that there are more cases of sgnfcant abnormal returns n the A-share market compared to the B-share market. The latter, however, has a very clear-cut pattern n that the B-share prces react posvely to good news and negatvely to bad news, for both the IAS and PRC GAAP reports. Furthermore, the reacton occurred mostly on the event day. On the other hand, the reactons n the A-share market appear to be rather unsystematc. We shall return to ths pont later. For the tradng volume, we fnd sgnfcant patterns around the earnngs-announcement dates. There are sgnfcant posve abnormal tradng volumes around the announcement dates, reflectng the nature of earnngs announcements as exogenous nformaton descrbed by He and Wang (1995). Ths s true for both the A- and B-share markets. We can see that n the A-share market, large abnormal tradng volumes whout relevant prce changes appear n the event wndows of both good and bad earnngs announcements. The abnormal tradng volume may become promnent even two or three weeks before the announcement. Accordng to the fndngs of He and Wang (1995), ths may mply some exstng nformaton n the A-share market. There are consderable post-event reactons up to day +5 for the B- share market. In comparson, the post-event reactons n the A-share market extend up to day +10. The abnormal tradng volumes persst for a longer perod for the A shares than for the B shares. Also, the abnormal tradng volume measure s generally larger for the A-share market than the B-share market n the announcement wndow. For the A-share market, abnormal 9

12 volume s more promnent when surprses are based on PRC GAAP, whether they are good or bad surprses. 5.2 Good News versus Bad News Good News From Table 2 we can see that abnormal returns are sgnfcant on days 8, 1, +2 and +4 for the A shares, and on day 0 for the B shares, when earnngs surprses are based on IAS. When we further consder the abnormal return values, we fnd that the abnormal returns are posve n the B-share market but negatve n the A-share market on the event day as well as the subsequent two days. An explanaton for ths phenomenon may be that the good news defned here actually s not good to the Chnese nvestors. As we use the IBES forecast data that are offered by the foregn nstutonal brokers and the foregn nvestors estmate may not represent the local nvestors forecast. We try the naïve forecast measure as an alternatve: we replace the IBES forecast value wh last year s IAS earnngs. However, we stll get smlar reacton patterns n both the A- and B-share markets. In Table 4 abnormal returns are sgnfcant on days 0, +2 and +5 for the A shares, and on days 0 and +6 for the B shares, when earnngs surprses are based on PRC GAAP. The eventday abnormal returns are posve for both the A- and B-share markets. For the A-share market, the abnormal returns are posve for days 0 and +1, wh reversal appearng n the subsequent three days. Thus, when PRC GAAP earnngs are used, the anomaly of the A- share returns reactng negatvely to good-news surprses s elmnated. Overall, the evdence seems to support hypothess H 3 for the B shares. In partcular, the good-earnngs nformaton based on IAS and PRC GAAP can both be used as sgnals n the B-share market. For the A-share market, however, stock returns do not move n lne wh the posve IAS earnngs nformaton. Instead, Chnese nvestors are more lkely to react only to posve PRC GAAP annual earnngs nformaton Bad News From Table 3 we can see that abnormal returns are sgnfcant on days 4, 0, +4 and +6 for the A shares, and on day 0, +1 and +2 for the B shares, when surprses are based on IAS. The abnormal returns are negatve n both markets, whch s consstent wh the bad-news effects found n other stock markets. In Table 5 sgnfcantly negatve abnormal returns are found on days 4 and 0 n the A-share stock market, and on days 0 and +3 n the B-share market, when surprses are based on PRC GAAP. Thus, hypothess H 4 s supported for both the A shares and B shares. The negatve earnngs nformaton based on IAS and PRC GAAP have nformaton values for the A- and B-share markets. Indeed, n the A-share market the prces react drastcally on bad-news announcements, as demonstrated by the magnude of the drop on the event day. In comparson, the B-share market reacts rather mldly to bad-news announcements. 5.3 IAS versus PRC GAAP The analyss of the effects of good and bad news based on IAS and PRC GAAP shows that the nvestors n the B-share market react to both the IAS earnngs and PRC GAAP earnngs nformaton, whch s consstent wh the fndngs by Bao and Chow (1999), who found that earnngs based on PRC GAAP and IAS are both sgnfcantly assocated wh the B share prces. Our event-study results thus support hypothess H 1. 10

13 The A shares react sgnfcantly to the earnngs announcements based on PRC GAAP. Although sgnfcantly negatve abnormal returns of the A shares appear on the event day of negatve earnngs surprse based on IAS, the abnormal returns n the A-share market are not sgnfcant on the event day of posve earnngs surprse based on the IAS fnancal reports. We conclude that the Chnese nvestors react only partally to the IAS earnngs reports, and pay more attenton to the PRC GAAP earnngs nformaton. Thus, hypothess H 2 s supported. Ths result agrees wh Chen, Frth and Km (2002), who reported that nvestors n the A-share market put more weght on PRC GAAP and started only recently to pay attenton to the nformaton based on IAS. There s an nterestng result here that the A shares react to the IAS-defned bad news but not to the IAS-defned good news. 6. Summary We have conducted an event study on the annual earnngs announcements based on two dfferent accountng standards, namely, IAS and PRC GAAP, n the Chnese segmented stock market. The annual earnngs announcements based on IAS and PRC GAAP are value relevant. Investors n the B-share market react to both the IAS and PRC GAAP annual earnngs announcements, whle nvestors n A-share market pay more attenton to the PRC GAAP earnngs reports. In the B-share market, posve abnormal returns are assocated wh good-earnngs surprses and negatve abnormal returns go wh bad-earnngs surprses. In the A-share market, however, ths pattern s not so clear-cut. Negatve abnormal returns (though not statstcally sgnfcant) are observed on the event day when good-earnngs surprses are based on IAS. The pre-event abnormal tradng volumes whout sgnfcant prce changes for the A-share market suggest some exstng nformaton n the A-share market. The post-event abnormal tradng volumes persst a longer perod n the A-share market than n the B-share market. Snce February 2001, the Chna Secures Regulatory Commsson allowed Chnese czens to hold and trade both A and B shares. Ths change elmnates the rgd segmentaton between the A- and B-share markets. As our results support the fndng that A-share nvestors have started to pay attenton to ISA reports, whether the exstng pattern of nformaton values of earnngs announcements wll change s an nterestng queston. Ths wll be a challengng topc for future research 11

14 References 1. Ajnkya, B.B., Atase, R.W., and Gft, M.J., of tradng and dsperson n fnancal analysts earnngs forecasts. Accountng Revew 66, Ang, J.S., and Ma, Y., Transparency n Chnese stocks: A study of earnngs forecasts by professonal analysts. Pacfc-Basn Fnance Journal 7, Atase, R.K., and Bamber, L.S., volume reacton to annual accountng earnngs announcements. Journal of Accountng and Economcs 17, Baley, W., Rsk and return on Chna s new stock markets: Some prelmnary evdence. Pacfc-Basn Fnance Journal 2, Baley, W., Chung, Y. P., and Kang, J.K., Foregn ownershp restrcton and equy prce premums: What drves the demand for cross-border nvestments? Journal of Fnancal and Quantatve Analyss 34, Bamber, L.S., The nformaton content of annual earnngs releases: A tradng volume approach. Journal of Accountng Research 24, Bamber, L.S., Unexpected earnngs, frm sze and tradng volume around quarterly earnngs announcements. Accountng Revew 62, Bao, B.H., and Chow, L., The usefulness of earnngs and book value for equy valuaton n emergng capal markets: evdence from lsted companes n the People s Republc of Chna. Journal of Internatonal Fnancal Management and Accountng 10, Beaver, W.H., The nformaton content of annual earnngs announcements, emprcal research n accountng: selected studes. Journal of Accountng Research (Supplement) 6, Bhattacharya, U., Daouk, H., Jorgenson, B., and Kehr, C. H., When an event s not event: the curous case of an emergng market. Journal of Fnancal Economcs 55, Brown, S.J., and Warner, J.B., Usng daly stock returns: the case of event studes. Journal of Fnancal Economcs 14, Chan, K.C., Cheng, L.T.W., and Fung, J.K.W., Ownershp restrctons and stockprce behavor n Chna. The Chnese Economy 34, Chen, C., Chen, S., and Su X., Is accountng nformaton value relevant n the emergng Chnese stock markets? Journal of Internatonal Accountng, Audng and Taxaton 10,

15 14. Chen, G., Frth, M., and Gao, N., The nformaton content of concurrently announced earnngs, cash dvdends, and stock dvdends: An nvestgaton of the Chnese Stock Market. Journal of Internatonal Fnancal Management and Accountng 13, Chen, G., Frth, M., and Km, J.B., The use of accountng nformaton for the valuaton of dual-class shares lsted on Chna s capal markets. Accountng and Busness Research 32, Cho, J. S., and Choe, C., Explanatory factors for tradng volume response to annual earnngs announcements: Evdence from the Korean stock market. Pacfc-Basn Fnance Journal 6, Chu, A.C.W., and Kwok, C.C.Y., Cross-autocorrelaton between A shares and B shares n the Chnese stock market. Journal of Fnancal Research 21, Chung, D. Y., and Lee, J., Ownershp structure and tradng volume reacton to earnngs announcements: Evdence from Japan. Pacfc-Basn Fnance Journal 6, Corrado, C. J., A nonparametrc test for abnormal secury-prce performance n event studes. Journal of Fnancal Economcs 23, Cready, W. M., Informaton value and nvestor wealth: The case of earnngs announcements. Journal of Accountng Research (Sprng), Fama, E., Effcent capal markets II. Journal of Fnance 26, Fung, H., Lee, W., and Leung, W.K., Segmentaton of the A- and B-share Chnese equy markets. The Journal of Fnancal Research 23, Fung, H., and Leung, W.K., Chnese fnancal lberalzaton: Implcatons for corporate governance. The Chnese Economy 34, Haw, I., Q, D., and Wu, W., Value relevance of earnngs n an emergng capal market: the case of A shares n Chna. Pacfc Economc Revew 4, Haw, I., Q, D., and Wu, W., Tmelness of annual report releases and market reacton to earnngs announcements n an emergng capal market: The case of Chna. Journal of Internatonal Fnancal Management and Accountng 11, He, H., and Wang, J., Dfferental nformaton and dynamc behavour of stock tradng volume. Revew of Fnancal Studes 8, Holthausen, R.W., and Verreccha, R. E., The effect of nformedness and consensus on prce and volume behavour. Accountng Revew 65, Kandel, E., and Pearson, N. D., Dfferental Interpretaton of Publc Sgnals and Trades n Speculatve Markets. The Journal of Polcal Economy 103,

16 29. Km, O., and Verreccha, R. E., 1991a. volume and prce reactons to publc announcements. Journal of Accountng Research 29, Km, O., and Verreccha, R. E., 1991b. Market reacton to antcpated announcements. Journal of Fnancal Economcs 30, Kross, W., Ha, G., and Hefln, F., A test of rsk clentele effects va an examnaton of tradng volume response to earnngs announcements. Journal of Accountng and Economcs 18, Lev, B., Toward a theory of equable and effcent accountng polcy. The Accountng Revew (January), Ma, X., Capal controls, market segmentaton and stock prces: Evdence from the Chnese stock market. Pacfc-Basn Fnance Journal 4, Poon, W. P. H., Frth, M., and Fung, H. G., Asset prcng n segmented capal markets: Prelmnary evdence from Chna-domcled companes. Pacfc-Basn Fnance Journal 6, Su, D.W., and Flesher, B. M., Why does return volatly dffer n Chnese stock markets? Pacfc-Basn Fnance Journal 7, Sun, Q., and Tong, W. H. S., The effect of market segmentaton on stock prces: the Chna Syndrome. Journal of Bankng and Fnance 24, Terpstra, R. H., and Fan, D. K. K., Dsperson of fnancal analysts forecast of earnngs per share and tradng volume: the Hong Kong experence. Pacfc-Basn Fnance Journal 1, Xu, C.K., The mcrostructure of the Chnese stock market. Chna Economc Revew 11, Zebart, D.A., The assocaton between consensus of belefs and tradng actvy surroundng earnngs announcements. Accountng Revew 65,

17 Appendx: Selected frms lsted on the SHSE and SZSE ncluded n the event study The frms n the sample are those that (1) ssue both A and B shares, (2) are followed by three or more analysts of IBES, and (3) are publcly traded n the market for more than one year. There are 38 frms n the SHSE and 21 frms n the SZSE SHSE SZSE 1. Chna Frst Pencl 1. Anhu Gujng Dstllery 2. Chna Textle Machnery 2. Bengang Steel Plates 3. Shangha Dazhong 3. Changcha 4. Eastern Communcatons 4. Chna Internatonal Marne Contaners 5. Helongjang Electrc Power 5. Chna Merchants Shekou 6. Hero (Group) 6. Chna Southern Glass Holdng 7. Huangshan Toursm 7. Chna Vanke 8. Huaxn Cement 8. Chongqng Changan Automoble 9. Jnzhou Harbor (Group) 9. Foshan Electrcal and Lghtng 10. Phoenx 10. Guangdong Electrc Power Development 11. Shanggong 11. Guangdong Provncal Expressway Development 12. Shangha Automaton Instrumentaton 12. Hube Sanonda 13. Shangha Chlor-Alkal Chemcal 13. Konka Group 14. Shangha Dajang (Group) 14. Shenzhen Chna Bcycle 15. Shangha Desel Engne 15. Shenzhen Chwan Wharf 16. Shangha Erfangj 16. Chna Fangda Group 17. Shangha Forever 17. Shenzhen Nanshan Power Staton 18. Shangha Frendshp & Overseas Chnese 18. Shenzhen Petrochemcal Industry (Group) 19. Shangha Haxn (Group) 19. Shenzhen Specal Economc Zone (Group) 20. Shangha Jnjang Tower 20. Wefu Fuel Injecton 21. Shangha Jnqao Export Processng Zone Development 21. Wux Ltle Swan 22. Shangha Lanhua Fber 23. Shangha Lujazu Fnance & Trade Zone Development 24. Shangha Materal Trade Center 25. Shangha Narcssus Electrc Applance 26. Shangha New Asa (Group) 27. Shangha Posts & Telecommuncatons 28. Shangha Refrgerator 29. Shangha Rubber Belt 30. Shangha Sanmao Textle 31. Shangha Shanglng Electrc Applance 32. Shangha Steel Tube 33. Shangha Tre & Rubber 34. Shangha Vacuum Electron Devces 35. Shangha Wng Sung Statonery 36. Shangha Worldbest 37. Shangha Yaohua Plknton Glass 38. Tanjn Marne Shppng 15

18 Table 1: Sample sze under dfferent levels of earnngs surprse based on IAS earnngs announcements and PRC GAAP earnngs announcements. For the IAS earnngs announcements, the earnngs surprse s defned by ES = ( AE EE ) EE, where AE s the actual earnngs based on IAS reported n the h event and EE s the estmated earnngs reported by I/B/E/S. For the PRC GAAP earnngs announcements, the earnngs surprse s defned by ES = ( AE AE, t 1 ) AE, t 1, where AE, t s the actual annual earnngs based on PRC GAAP reported n the h event and AE, t 1 s the prevous year s annual earnngs based on PRC GAAP for that company. Earnngs Surprse (ES) IAS PRC GAAP ES > 20% ES > ES < ES < -20% 58 69

19 Table 2: return and abnormal tradng volume for IAS earnngs (market-model-adjusted abnormal return, earnngs surprse exceedng +20%) Event Day A Shares B Shares

20 Table 3: return and abnormal tradng volume for IAS earnngs (market-model-adjusted abnormal return, earnngs surprse below 20%) Event Day A Shares B Shares

21 Table 4: return and abnormal tradng volume for PRC GAAP earnngs (market-model-adjusted abnormal return, earnngs surprse exceedng +20%) Event Day A Shares B Shares