2016 Inernaional Conference on Advanced Educaion and Managemen Engineering (AEME 2016) ISBN: 978-1-60595-398-4 The Effecs of Rumors on Sock Prices: A Tes in an Emerging Marke Yan ZHANG 1,2 and Hao-jia CHEN 1 1 Inernaional Business Faculy, Beijing Normal Universiy, Zhuhai, 519087 China 2 School of Business, Macau Universiy of Science and Technology, Macau, China Keywords: Rumors, Sock marke, Abnormal reurns. Absrac. The purpose of his sudy is o invesigae he effec of rumors (released on he web) in he sock marke on common sock reurns raded in he Shenzhen and Shanghai Sock Exchange. The sample consiss of 188 rumors menioned in he easmoney.com and 10jqka.com.cn. The resuls show ha he significan average cumulaive abnormal reurns (CAR) are observed in he beginning 5 days prior o he publicaion dae and he significan CARs are also deeced in 10 days pos o he publicaion dae. Furhermore, he difference of ype of rumors may be an imporan explanaion of he price movemen. Inroducion According o Merriam-Webser dicionary, a rumor is 1. a alk or an opinion widely disseminaed wih no discernible source. 2. a saemen or repor curren wihou known auhoriy for is ruh. Wih he remendous developmen of informaion echnology and elecommunicaion, where lierally counless websies as well as cha rooms and forums provide informaion on endless opics, we can found more and more opinions, esimaions and predicions of invesors and privae analys concerning publicly raded socks a hese locaions, which can be described as rumors. The number of rumors has increased significanly during he las years. Wysocki (1999), and Tumarkin and whielaw (2001)iniiae he ineres regarding he effec of inerne rumors on sock process and abnormal reurns(ars). Since hen, he quesion of wheher rading based on rumors would benefi invesors has been invesigaed exensively. Sudies invesigaing he effec of rumors are mainly relaed o he even sudy, which empirically es around he publicaion dae of he rumors, wheher he AAR (Average Abnormal Reurn) and CAR (Cumulaive Abnormal Reurn ) are saisically significan or no. Our sudy focus on an emerging counry, China, for he reason ha he Chinese sock marke is full of rumors and many invesors in China share a belief ha here is informaion leakage and insider rading. We aemp o examine he effec of rumor clarificaion on sock reurns in China. Lieraure Review The empirical sudies in his area repor mixed resuls. Diefenback (1972) and Logue and Tule (1973) iniiae he sudies and repor ha rumors have no value for invesors. Laer sudies, on he oher hand, repor ha informaion no maer provided by heard on he sree or analyss conain valuable informaion o invesor. Lloyd-Davies and Canes (1978) sudy he financial analyss recommendaions as discussed in he HOTS column of WSJ. They repor ha buy recommendaions and sell recommendaions provide significan opposie abnormal reurns. The former is associaed wih significan posiive abnormal reurns, he laer, wih significan negaive abnormal reurns on he day of publicaion. They conclude ha analyss and invesmens advisors are imporan o invesors by providing valuable service. Liu e al. (1990) exends he Lloyd-Davies and Canes (1978) sudy wih more recen sample based on furher analyses of he effecs of he single-company versus muli-company recommendaions, and he rading volume around he publicaion day. Their findings are in he line wih hese of Lloyd-Davies and Canes (1978) sudy. 143
More recenly, Mahur and Waheed (1995) invesigae he sock price behavior of firms ha are favorably menioned in he `Inside Wall Sree column of Business Week. The resuls deec he exisence of posiive significan abnormal reurns on he day before he publicaion dae, and wo days afer he publicaion dae. Uriel Spiegel e al. (2010) sheds ligh on he rajecory ha informaion spreads wih he approach of he even day. I is deeced ha as he day of he even draws closer, he inner circle of informed invesors widens and more invesors become aware of he impending even. On he day of he published rumor, a furher increase in AAR occurs, afer he even day he AAR sabilizes and mainains is level. In he period afer he rumor is published, which acually akes place, he CAR coninue o rise, while for hose rumors ha acually did no maerialize, he CAR declines afer publicaion. Halil KiymalI (2002) invesigaed he effecs of sock marke rumors(gossips) on he prices of socks raded a he Isanbul Sock Exchange. The empirical findings sugges ha here are saisically significan abnormal reurns around he publicaion dae. (-value is 3.57 during he 5 days pre-publish day o pos-publish day i.e.(-5,+5)). Furhermore, i is found ha here are differences in sock price reacion wih aspec o he size of he firm. The small firm sample appears o be more speculaive and more sensiive o price reversal. Xiaolan Yang e al. (2014) analyze he effecs of official rumor clarificaion on Chinese sock reurns under differen marke condiions. The resuls show ha he CAR afer he clarificaion even is significanly posiive in a bull marke, and significanly negaive in a bear marke. Moreover, in boh bull and bear markes, invesors are unable o disinguish beween rumors ha prove rue and hose ha prove false, or beween srong and weak rumor denial. In general, he sudies on sock marke rumors or analyss recommendaions suppor he view ha informaion provided o invesors is valuable. The objecive of his sudy is o invesigae he quesion of wheher sock marke rumors have any effec on he price of common socks raded a he Shanghai and Shenzhen Sock Exchange by examining he easmoney.com, 10jqka.com.cn. Two main issues are explored. One is wheher he informaion provided in he inerne creaes a price flucuaion on sock prices hrough publiciy impac. The second issue invesigaed is wheher he classificaion of rumors may explain he price movemens around he publicaion of rumors. Daa and Mehodology The sudy uses he sock marke rumors/gossips published in he easmoney.com, and 10jqka.com.cn during he period of 21 July 2015 and 2 May 2016. Table 1 repors he sample selecion. We observed 324 rumors in oal, in which we eliminaed he rumors published on he same opic and hose abou he same firm in subsequen weeks and hose firms wih missing sock price daa. The ne sample consiss of 188 rumors. Moreover, we caegorize he ypes of rumors ino hree major groups. Caegory 1 conains rumors abou mergers and acquisiions; Caegory 2 conains rumors abou asse resrucuring; Caegory 3 conains rumors abou posiive changes in a firm s operaions. Table 1. The sample selecion. Number of Rumors All rumors published 324 Less: Subsequenly published rumors 84 Less: Missing Daa 52 Ne Sample 188 Ca.1(rumors abu mergers and acquisiion) 44 Ca 2(rumors abou asse resrucuring) 58 Ca.3(Rumors abou posiive changes in a firm's operaions) 86 144
Even sudy mehodology is employed o analyze he effecs of rumors/gossips on sock prices as surveyed by Brown and Warner (1985). The analysis periods exend from even day -20 o +20.The abnormal performance in he analysis period is esed for. For each even i, abnormal reurns (ARi,), are calculaed for each day in he analysis period. AR = R α + β R ) (1) i, i, ( i i m, Where, ARi, is abnormal reurn for securiy i on he day ; Ri, is observed reurn for securiy i on he day ; α + β R ) is expeced reurn for securiy i on he day. Average abnormal reurns ( i i m, (AARi, )across all firms (n) in he sample are calculaed for each day in he analysis period. N, ε i, i= 1 AAR i = (1 n) (2) I means ha he AARs are he average differences beween all he observed reurns and he expeced reurns, where he expeced reurns in our sudy are approximaed by he SSE (Shanghai Sock Exchange) Composie Index and Shenzhen Componen Index. The null hypoheses o be esed are he average abnormal reurns in he even period are equal o zero. The following es saisic is used. n 1 ε i, = (3) n i= 1 σ i Where ε i, is error erm and σi is he sandard deviaion in he esimaion period. Empirical Resul We calculaed he daily average abnormal reurns (AARs) for all rumors over -20 and +20 period relaive o he even day 0. Only AARs for -10 and +10 period are repored on Table 2. The resul indicaes he AARs are 0.97%, 0.0337%, 1.151%, 0.409% and 0.349% respecively, when he AAR reaches he maximum a he dae 2 before he publicaion dae. These resuls are saisically significan a 1% level. On he dae afer he publicaion, he AARs are eiher posiive or negaive for he reason ha he invesors have differen esimaions o he sample firm. The AARs are 0.046%, -0.985%, -1.955% and 0.874% respecively. These resuls are saisically significan a 1% level. Prior o he publicaion dae Table 2. Average AARs on differen day prior o or afer publish dae. AARs (%) value Afer publicaion dae AARs (%) value -10-0.617-4.669 1 0.046 0.663-9 0.073 3.301 2-0.985-4.167-8 1.294 1.286 3-1.955-5.67-7 0.665 1.366 4 0.874 2.698-6 0.77 3.131 5-0.276-0.853-5 0.97 4.04 6-1.79-1.476-4 0.337 1.366 7 0.906 1.043-3 1.151 5.018 8-2.408-1.548-2 0.409 4.859 9-0.017-0.06-1 0.349 3.291 10-1.708-1.107 Table 3 repors he average cumulaive abnormal reurns (CARs) relaive o he marke reurns for various even windows. I shows ha firms experience abnormal reurns jus before and afer publicaion dae. For example, during (-1,+1), (-2,+2), and (-3,+3) periods, CARs are 0.74%, 1.24%, 145
1.34% respecively. All of hem are saisically significan a 5% level. So i is concluded ha he rumors in he sock marke provide valuable informaion o invesors. We furher analyze he behavior of sock price prior or afer he publicaion of he rumors. CARs in he pre-publicaion periods are negaive and saisically significan. For example during (-3,-1), (-5,-1), (-10,-1), and (-20,-1) periods, CARs are -1.72%, -1.42%, -1.29% and 1.36% respecively. The resuls of (-3,-1), (-5,-1) are saisically significan a 1% level.ohers are saisically significan a 10% level. CARs afer publicaion periods are posiive and saisically significan. For example during (+1, +3), (+1,+5), and (+1,+10) periods, CARs are 2.78%,1.49%, -1.29% and 1.69% respecively. The resuls are saisically significan a 1% level. I indicaes ha he invesors gain he obvious abnormal reurns afer he periods of he publicaion dae. Table 3. Average CARs for differen even windows (percen). Even Windows CAR(%) -value Combined period (-1,+1) 0.74 3.34 (-2,+2) 1.24 2.29 (-3,+3) 1.34 2.44 (-5,+5) 0.25 1.35 (-10,+10) 0.63 3.28 (-15,+15) 0.21 1.04 (-20,+20) 0.29 0.68 Prior o publish daa (-3,-1) -1.72-4.77 (-5,-1) -1.42-2.71 (-10,-1) -1.29-1.58 (-20,-1) 1.36 1.62 Afer publish daa (+1,+3) 2.78 5.56 (+1,+5) 1.49 2.69 (+1,+10) 1.69 3.61 (+1,+20) 1.04 0.22 Table 4 repors he resul of CARs for differen caegories rumors, say Ca.1, Ca.2, and Ca.3 menioned above. The resul indicaes when he rumors are abou mergers and acquisiions (i.e. Ca.1), he CARs are -2.63%, -3.57% and -4.32% respecively during he periods of (-5, -1), (-10,-1) and (-20,-1), which are all saisically significan. The CARs are 4.41%, 4.73% and 4.23% respecively during he periods of (+1, +5), (+1, +10) and (+1, +20), which are all saisically significan. When he rumors are abou asse resrucuring (i.e. Ca.2), he effecs are no clear. When he rumors are abou posiive changes in a firm s operaions (i.e. Ca.2), he CARs are 2.07%, 3.01% and 3.24% respecively during he periods of (+1, +5), (+1, +10) and (+1, +20), which are all saisically significan. 146
Table 4. The CARs under differen caegories rumors. Ca. 1 Ca. 2 Ca. 3 CARs(%) -value CARs(%) -value CARs(%) -value (-1,+1) 0.22 0.34 2.38 1.57 2.44 1.56 (-2,+2) 1.07 1.24 1.99 1.09 3.06 1.84 (-3,+3) 2.48 1.94 2.78 1.21 5.09 2.03 (-5,+5) 5.52 2.75-3.54-1.37 4.51 1,38 (-10,+10) -1.12-1.09-2.21-1.32 0.87 0.61 (-15,+15) 2.34 2.83-1.12-1.03 0.81 0.56 (-20,+20) -0.81-1.17 0.11 1.53-0.04-0.07 (-5,-1) -2.63-1.66 2.31 2.67 2.48 1.58 (-10,-1) -3.57-2.52 1.16 1.19 0.92 1.62 (-20,-1) -4.32-2.79-0.57-0.25 1.87 0.92 (+1,+5) 4.41 3.17-2.43-1.79 2.07 2.39 (+1,+10) 4.73 2.87-4.56-2.63 3.01 3.04 (+1,+20) 4.23 3.21 1.97 0.48 3.24 2.58 Summary and Conclusion We examine he effec of rumors on sock reurns in he Chinese sock marke by conducing even sudy over 2015-2016. The empirical findings sugges ha here are significan abnormal reurns around he publicaion dae. Furhermore, here are differences in sock price reacion wih respec o he caegory of he rumors, which is he evidence regarding irraional invesors behavior in China. Reference [1] Wysocki P. Cheap alk on he web: he deerminans of posiing on sock message boards, working Paper, Universiy of Michigan School, November, 1999. [2] Tumarkin and Whielaw, R.F. News or noise? Inerne message board aciviy and sock prices, Financial Analysis journal, 57, 41-51, 2001. [3] Diefenback, R. How good is insiuional brokerage research, Financial Analys Journal, 28, 54-60, 1972. [4] Logue, D. and Tule, D. Brokerage houses invesmen advices, The Financial Review, 8, 38-54, 1973. [5] Lloyd-Davies, P. and Canes, M. Sock prices and publicaion of second-hand informaion, Journal of Business, 51, 43-56, 1978. [6] Liu, P., Smih, D. and Syed, A. Sock reacion o he Wall Sree Journal s securiies recommendaions, Journal of Financial and Quaniaive Analysis, 25, 399-410, 1990. [7] Mahur, I. and Waheed, A. Sock price reacions o securiies recommended in Business Week s inside Wall Sree, The Financial Review, 30, 583-604, 1995. [8] Uriel Spiegel, Tchai Tavor and Joseph Templeman Effec of Rumors on Financial Marke Efficiency, Applied Economics Leers, 2010, 17, 1461-1464. [9] Halil Kiymaz, The Sock Marke Rumors and Sock Prices: A Tes of Price Pressure and Size Effec in An Emerging Marke, Applied Financial Economics, 2002, 12, pp 69-474. [10] XiaoLan Yang Rumor Clarificaion and Sock Reurns: Do Bull Markes Behave Differenly from Bear Markes? Emerging Markes Finance & Trade/ January- February 2014, Vol. 50, no. 1, pp 197-209. [11] Brown, S.J. and Warner, J.B. Using daily sock reurns: The Case of even sudies, Journal of Financial Economics, 14, 3-31, 1985. 147