The issue about market efficiency has proceeded for more than 30 years,but market efficiency and its assumptions have got the widespread approbation.Chinese stock market efficiency research has the certain scholarship andpractice value to the analysis and the developments of Chinese stock market.Particularly the event studies to aim at different events about semi-strong formtests is becoming the one of hot researches about market efficiency tests oncurrent Chinese stock market.As for the market efficiency of the three levels, the scholars of manycountries put forward various test methods. Among them, the test methodabout semi-strong form efficiency is the event study mainly, including theresearches about the year announcement, stock splits, SEOs and merges, andso on.Current, the research result that there has been about semi-strong formtests in China think consistently that current stock market of China hasn'tattained semi-strong form efficiency. In this paper, we make SEOs of Chinesestock market as research object, and apply the event studies, and test ifChinese stock market has been semi-strong form efficiency.During testing, we use the market model to calculate normal return in thispaper:First, we make use of the market model, by the sample data in theestimate window, to estimate square distance coefficient. Then we apply theequation that has been estimated to calculate the normal return in the eventwindow, so we can calculate the abnormal return in the event window. Then,by the method of equal weights and value weights, we compute the averagevalue of cumulative abnormal return (CAR). The test statistic that we use totest the average value of CAR is:We proceed short-term events studying firstly. The samples we choice arethe companies which had have the events of seasoned equity offerings (SEOs)during the year 2000 to 2002 in Shanghai Stock Exchange and ShenzhenStock Exchange, this kind of company sums up to 54.And we eliminate fourcompanies which are special, then there are 50 companies that we use to testfinally. In these 50 companies there are 23 companies of Shanghai StockExchange and 27 companies of Shenzhen Stock Exchange. During testing, weuse the data of the close price of stock everyday and Composite Index ofShanghai Stock Exchange and Composite Index of Shenzhen Stock Exchange,and we also use the total subscribed capital of these companies before theystart the events of SEOs. This investigative estimate window is the time of(-100,-21) day before the event happens, totally 80 days. And the eventswindow is the time of (-20,20) day, totally 41 days.From the calculation results we can find some phenomena, that theCAR s of the three group samples of total samples, Shanghai samples andShenzhen samples haven't obvious fluctuating when the time hasn't get to just2 day before the event day, and isn't statistically significant,showing themarket doesn't react to the information before SEOs. This has expressed theinefficiency of market. The CAR of the day just before the event day and theevent day isn't statistically significant,but the CAR has obvious fluctuating,showing that the market starts to react to the information. During the 20 daysafter the event day, the absolute value of the CAR of total samples is calmfirstly, then increases continually, then is calm, and the test statistic J isstatistically very significant. On the one hand, this shows that the marketdoesn't react enough to the information before SEOs and there areunderreaction. On the other hand, because the fluctuating of the CAR iscalm during the 20 days after the event day, so showing that the market hasabsorbed much of the information .So we can conclude that although themarket doesn't achieve the semi-strong form efficiency, there has beenimproving on the efficiency degree.By calculation results, we can also observe that the CAR of the bigcompany samples has have some changes in tow or three days in the front andback of the event day, but it is calm in the twenty days after the event day.Moreover, the test statistic J of the CAR only in number –20,2,3,11,12 dayof the event window is statistically significant at 10% confidence level, it isn'tstatistically significant in other days. However, the absolute value of theCAR of the small company samples has been obviously enlarging in tow orthree days in the front and back of the event day, and it continues enlarging inthe twenty days after the event day. Moreover, in number 1 to number 20 dayafter the event day, the absolute value of the CAR all is statisticallysignificant at 5% confidence level. This shows that the test about the smallcompany samples proves that the market is underreaction to the information.So we can conclude that the result of the test has grasped the phenomenon ofsmall stocks of Chinese stock market obviously, at the same time, it showsthat the small company samples have a big effect on the total samples. So ifwe only concerned the big companies without considering the smallcompanies, the market would have neared to semi-strong form efficiency. We then proceed the long-term events studying, still using the marketmodel to compute the normal return. The samples we choice are thecompanies which have have the events of seasoned equity offerings (SEOs)during the year 2000 to 2003 in Shanghai Stock Exchange and ShenzhenStock Exchange, this kind of company sums up to 75. But in this paper whenwe proceed the long-term events studying, the data we use are month data andthe estimate window is the time of (-48,-13) month before the event happens,totally 36 months, and the events window is the time of (-12,12) month,totally 25 months. Therefore, passing by the standard of data choice, thesuitable sample companies are 34, 19 in Shanghai Stock Exchange and 15 inShenzhen Stock Exchange. During calculating, we use the month return of thesingle stock and that of the market and the month total market value of the lastmonth of SEOs. The month returns of the single stock and that of the markethave considered the reinvestment of the cash bonus. (The calculation of thereturn is by the method of value-total equity weighted average.)From the results we can see that the CAR calculated by equal weightsis almost negative in the whole 25 months of the event window. And theabsolute value of the CAR has continued enlarging in the front months ofthe event day, then it has been calm for several months after the event day, andafter that it continues enlarging. These show that the sample companies havehave the reaction for the events in the time before the events, and have still thereaction after the events. And there is the underreaction of the market to SEOs,so the market hasn't achieved the semi-strong form efficiency. And we canfind that the absolute value of the CAR of the big company samples isbigger than that of the small company samples, this shows that there is nophenomenon of small stocks. Furthermore, whether the big company samplesor the small company samples, the test statistic J of which is hardlystatistically significant,we think that the reason of these results is that thenumber of the samples is too few. Furthermore we can find that the absolutevalue of the CAR calculated with value weights at one month before SEO isbigger than that calculated with equal weights average of total samples,because the contribution of the big companies to the abnormal return is biggerin this test. However, we find that significance of the test statistic J* of theCAR* calculated with value weights is smaller than that of the test statistic Jof the CAR calculated with equal weights, except that in the months ofnumber 0, 10, 11, 12 the test statistic J* is statistically significant at 5%confidence level, in other months after the event day it is statisticallysignificant at 10% confidence level. From this we can know that using valueweights can lower statistically significance of the abnormal return. |