This paper reviews some previous evidence which obviously indicate that the relation between abnormal returns on stock price drift round earning announcement trends to be positive in a short period, because of slow and incomplete market response to information can be explained by behavioural finance which strongly contradicts semi-strong form market efficiency that the stock price reflect all publicly available information and only information which is not available to the public can generate abnormal returns for investors on their investments. This paper aims to use previous evidence to prove the existence of abnormal returns around earnings announcement released and give a comprehensive analysis about the reason why this phenomenon occurred and this influence in different size firm will be also discussed. This paper reviews some previous evidence in order to analyze whether the efficient market hypothesis is still valid under the abnormal returns of positive earnings announcements released and give a comprehensive analysis about the reason why this phenomenon occurred. In the first part, the relation between the abnormal returns on stock price drift and semi-strong efficient hypothesis will be discussed. In the second part, it mainly demonstrates the drift on stock price around the announcement day and gives a comprehensive analysis about how quickly the stock price of react to the positive surprise earnings announcements. In the third part, the firm size effect on stock price drift will be also analyzed. Finally, this paper points out the limitation in pervious research. (1) For different national institution systems is rare. (2) The comparison of different reactions to earnings announcement of different stock markets in developed and developing countries is rare. (3) The research of Different attitudes to the earnings announcement between individual investors and institutional investors is rare.
Introduction When firms release their earnings announcement and related accounting information to the public, the institutions and individual analysts will compare this information with their previous estimation according to the inside and outside issues which can affect the company's future performance carefully, in order to pursue excess or abnormal returns. These data in reports are generally higher or lower than analysts' estimates because of the surprised issues in earnings announcement, which usually affect the stock price of these companies significantly. Positive surprised earnings announcements typically cause a positive signal to their current and potential investors who concern the future performance of these firms (Needham, 2007). Conversely, a negative earnings announcement usually provides a negative signal to their investors which can act on the stock price. According to Eugene Fama (1970), market efficiency can be explained by three forms: weak form, semi-strong form and strong form efficiency. This paper aims to demonstrate whether the investor can gain abnormal returns from positive earnings announcement of a firm, because based on semi-strong form efficiency hypothesis which provided by Fama in 1970, the stock price is completely reflected by all public information, which means the future performance cannot be predicted and estimated by any other technical nor fundamental analysis which aims to gain abnormal returns. Nevertheless, some empirical studies which focus on the evaluation of speed of adjustment of stock prices to earning announcement show that market efficiency hypothesis is invalid. DeBondt and Thaler (1987) analyzed the relation between the stock prices and earnings announcement, which shows a positive estimated abnormal stock returns around the announcement day. Furthermore, some evidence indicated that the average initial response to earning announcements is an overreaction to earnings. Firstly, the purpose of this study is to analyze whether the efficient market hypothesis is still valid under the abnormal returns of positive earnings announcements released. Secondly, it will discuss the drift of stock price around the announcement day and give an analysis about the time of how quickly the stock price of react to the positive surprise earnings announcements. Thirdly, investigate the firm size effect in stock price change around earnings announcement released. Finally, give a conclusion and an implication from this study.
Analyzing in semi-strong form EMH Semi-strong form efficiency was defined by Fama (1970) that stock price completely reflects all available information which released to the public. A test of semi-strong form efficiency was performed by Fama, Fisher, Jensen, and Roll (1969) indicated that investors cannot gain any abnormal returns under public information, which means under efficient capital markets, stock prices fully reflect all the information in a rapid and unbiased period and provide unbiased estimates of the underlying values (Basu, 1977). In conclusion, if semi-strong form efficiency is valid, abnormal returns by acting on public announcements is possible. However, many literatures indicated that the market is inefficient. William (1991) says the earnings announcements contain some information which is not available to the public. Foster et al (1984) constructed 10 portfolios on the size of earnings surprise and analyzed stock returns. He found there was a large abnormal return on announcement day and investors can obtain this excess return when the positive earning information was published. Moreover, he also recognized that there was significant stock price adrift around the announcement day which can be explained by delayed adjustment of stock market to new information (under-reaction). Ball and Kothari (1991) supported this conclusion by their empirical test that earnings announcement usually include information which not available to the market and significant excess return will be generated on the announcement day. Hereafter, Jegadeesh and Livnat (2006) say that abnormal stock returns, especially the post announcement abnormal returns are result from earning announcement. All of this evidence clearly and strongly demonstrates that price announcements contain information not available to the market and the stock price cannot fully reflect all the information released to the public, which against semi-strong form EMH.
2. Analyzing in stock price drift around announcement day Ball and Brown (1968) find there was an abnormal return around earnings announcement day and the internal and external factors which cannot be released to the public will potentially affect the intrinsic value of stock price (Charles et al, 1970). Joy and Jones (1979) concluded their studies and said the phenomenon of generating abnormal returns from earnings announcements can prove the market is inefficient.
2.1 Pre- Earning Announcement Drift Pre-announcement means private information used in anticipation of public disclosure in order to pursue abnormal returns when related information published to the market. The first conclusive evidence of pre-announcement price drift was provided by Keown and Pinkerton (1981), which is based on the underlying theory stock price behavior. In this paper, they assumed that any observed increasing in stock price would be directly prove the existence of insider trading and concluded that the leakage was significantly pervasive insider trading. Foster et al (1981) concluded that the early earnings announcement may contain some information relevant to the later and currently affect it in stock price. This appearance can be defined as pre-earnings announcement drift, which is similar to post-earnings announcement drift because of the under-reaction of investors to the changing in earnings announcement. Ball and Brown (1968) used their empirical evidence to demonstrate that around 85% to 90% of stocks price changed before the announcement released, because these information will directly influence the confidence of their current or potential investors to the company's future performance they invested. The result from Jarrell and Poulsen (1989) also support Ball and Brown (1968). Later, Paul (1994) used his empirical test to find share prices would increase dramatically in the 30 days prior to the god earnings announcement released. However, Sanders and Zdanowicz (1992) used their empirical test to examine pre-announcement activities and argued that: "The previous literature did not isolated the interval over which insider trading could take place, namely around the unpublicized initiation date." Furthermore, they also found there was no evidence to prove the price drift which caused by the pre-bid activities before the announcement date. Finally, they conclude abnormal returns only occurred after the financial information published and those speculation activities before the announcement date almost certainly based on the leaked information they got. The relation between stock price and pre-announcement trading volume activities is also discussed by Jarrell and Poulsen (1989). They used historical data to analyze the relation between the trading volume and abnormal returns, which is based on stock price drift around announcement day. According to their studies, they found abnormal trading volume activities were significant over the whole pre-announcement period of time. Based on their research, they found there was a significantly positive abnormal return may be accrued before the announcement published, but increased trading volume activity is just possible. However, Sanders and Zdanowicz (1992) criticized that there might be some noises in the methodology which was used by Jarrell and Poulsen in 1989. They argued that "as a result of the failure to take into account both day of the week effects and serial correlation, the test is misspecified".
2.2 Post- Earning Announcement Drift The abnormal returns before earnings announcement are relatively easy to understand because the leakage of information into the market, but not the same as abnormal returns after the earnings announcement (PEAD). Ball and Brown identified post-earnings announcement drift (PEAD) in 1968 and subsequent research always supported their view comprehensively. In order to confirm the existence of PEAD, Forster et al (1984) used their empirical test and said there was positive correlation between abnormal stock returns and the unexpected issues of the earning post-earnings announcement released. The post-earnings announcement drift phenomenon directly challenges the semi-strong form efficient markets hypothesis (EMH), which indicates that the stock price fully reflect all publicly available information instantaneously. Meanwhile, Abarbanell and Bernard (1992) compared market and analyst reactions to earnings announcement released. According to their studies they found that analysts are more efficient than the market. However, Fama (1998) said it is just a market anomaly in the stock market, which can be solved by three-factor module. Sometimes, previous studies also implied that abnormal returns could be generated by using investment methodology and strategies based on price-change persistency subsequent to earnings announcements (Lev and Ohlson, 1982). Kyle (1985) used his result to support (Lev and Ohlson, 1982), there was a substantial return of PEAD can be treated as compensation for the unpredictable information released in earnings announcement for informed traders to noise traders. However, one argument supported by the EMH says that: "If any systematic method of obtaining abnormal excess returns exists and if that method becomes known to the public, then the mechanics of an efficient market will negate the realization of any further benefits derived from the use of that method" (Bidwell and Riddle, 1981). The evidence shows a strong belief in market efficiency was the accumulation of results which indicate that there is an existence of persistent price adjustments after earnings announcements, but it is just an instantaneous adjustment, the stock market is still efficient. However, Bernard and Thomas (1990) say that the post-earnings announcement drift means the anomaly in market efficiency is a failure to reflect all the information into the stock price completely, because of the relatively slow reflection to the announcement published from people. It is difficult for people to understand why stock prices can not respond the information in earning announcement published completely and immediately. In order to explain this question, Bernard and Thomas (1989) and Freeman and Tse (1989) use their empirical evidence to provide one indication that could help people to understand preliminary. Specifically, they found that there was a huge disproportionately of the post-announcement drift is delayed until the subsequent earnings announcement released. Later, Booth, Kalunki and Martikainen (1996) summarized Bernard and Thomas (1989) that it is so absolute, but at least, a part of the responses to changes indicated around earnings announcements has been delayed. In other words, why this situation happened is just because not all the investors can get available information from earnings announcement and assimilate it and the existence of transaction cost in this processing. Bhushan (1994) tried to find out the relation between the post-earnings announcement drift and proxies for transaction costs. According to his empirical test, he found even in an efficient security market, the transaction will still exist because sophisticated investors will not trade until the transaction costs will be exceed by their expected profit return.
2.3 How quickly security prices assimilate new information Fama (1998) said that "most long-term return anomalies tend to disappear with reasonable changes in technique and alternative approaches are used to measure them." Grossman (1976) shows that price can reveal all private information to passive investors. However, Grossman and Stiglitz (1980) argue that only if transaction costs are zero, the prices should assimilate new information immediately and completely. As a matter of fact, the transaction cost cannot be zero. Watts (1978) also supported this view, he said only those people who can avoid the direct or indirect transaction costs can generate abnormal returns after earnings announcement released. According to Forster et al (1984), although there was positive correlation between abnormal stock returns and the unexpected issues in period of earning announcement released, he found the speed of adjustment to any information contained in announcement is gradual but not instantaneous. Bernard and Moreover, Thomas (1990) also pointed out that the existence of drift stock price will not exceed 1 hour after the earnings announcement released. In order to explain and solve this problem, Dann et al. (1977), the first research group in this domain, used high frequency data to analyze the speed of price adjustment. They found that all of the potential trading profits dissipated and disappeared within 5 minutes after the announcement released. Moreover, Patell and Wolfson (1984) showed that abnormal returns would disappear within the first 30 minutes of earnings announcement released and most of the stock price returned to normal price within 10 minutes, which is similar with previous studies of Dann et al. (1977). In addition to this, Brooks (2003) found an immediate price adjustment phenomenon to overnight unpredicted events. However, De Bondt and Thaler (1985) argued that the price decreases and rebounds in subsequent periods due to the overreaction. If these points of view hold up, it means the strong form of market efficiency hypothesis is not valid. As can be seen, all the evidence above support that there is quick adjustment of prices drift after the earnings announcement released. However, there is a problem that all of these studies do not pay more attention on the impact of the different market structures, which can influence the process of price adjustment strongly. Francis et al. (1992) compared Paris Bourse market response to earnings announcements released during trading and non-trading hours. As a result, they found there was no evidence to prove that an inner reaction to overnight announcements at the opening of security market which is supported by Grossman (1976) that the trading is necessary to assimilate new information into the stock price, but the adjustment of stock price is not immediate to overnight announcements. Nevertheless, Greene and Watts (1996) found a different result from Francis et al. (1992). In order to explain their opinion, they analyzed NYSE and NASDAQ. As a result, they found there was an immediate price adjustment in these two markets which can be explained by the weakness of the information content of annual earnings announcements in two markets. Cao et al. (2000) used their result to support this point of view, stock prices will adjust to overnight announcements immediately. Thereby, as can be seen from the previous studies, the speed of adjustment of stock prices is related to market structure, not hold unique evaluation criteria.
Firm size effect Linda (1987) said relative to the announcements of larger firms, smaller firms can generate a greater stock increasing in trading which persists for a longer time. Cheng Fan-fah (2008) used their empirical evidence to prove the impact of firm size on the stock prices of earnings announcement released are usually gained by the earnings response coefficient (ERC) and the ERC of smaller firms will be usually less than the larger firms, which shows a significantly negative correlation between firm sizes and abnormal returns. In the next parts, some issues which can affect the stock price in different size firms will be discussed.
3.1 Information asymmetry in small size firm announcement Ball and Kothari (1991) said the earnings announcement and other financial statements in large companies are more transparent than small companies and the abnormal stock return of small size firm is relatively higher than large size firms around the day of earnings announcement released because of the information asymmetry. Due to the less information published by small firms, investors should know the earnings announcement of small size firms includes lots of information which not available to the market, in other words, if investors early access the positive asymmetric information of earnings and revenue announcement from the inside of small size firms and invest on them, the more opportunities will be generated for them to obtain excess and abnormal returns than the people who invest on large size firms before the earnings announcement released.
3.2 Abnormal trading volume reaction in small size firms Beaver (1968) concluded that the trading volume indicated a consensus expectance among investors with complete information released. Freeman (1981) debated on whether there will be a relation between firm size effect and the existence of trading volume. According to his studies, the result showed an analogous inverse relation between firm size and trading volume around the date of earnings announcements released. Atiase (1985) proposed his point of view that investors prefer to discovery and pursue more related information and unexpected news of earnings announcements of small firms, which leading to a significant and sustained trading volume reaction. Linda (1987) used her empirical test to support Atiase (1985) that there was a positive correlation between the unexpected earnings announcement and the abnormal trading volume reactions. Relative to the earnings announcements of large firms, trading volume reaction of small firms is relatively larger than big firms after the earnings announcement released in a long period of time which can generate a significant stock price drift.
Conclusion This paper reviews some previous evidence in order to analyze whether the efficient market hypothesis is still valid under the abnormal returns of positive earnings announcements released. Meanwhile, the issues which can cause the drift of stock price around the announcement day and how quickly the stock price of react to the positive surprise earnings announcements are also discussed. According to this paper, I found there is a significant price drift around announcement day, because of slow and incomplete market response to new information published, which can be explained by behavioural finance. Furthermore, the length of time of assimilating new information is very fast, but it also depends on the market structure we discussed. Moreover, it also investigates the firm size effect in stock price change around earnings announcement released, a negative relation between the firm size and price drift was confirmed. All of these studies obviously indicate that there will be an abnormal return on stock price drift round earning announcement day in a short period which against semi-strong form market efficiency hypothesis that the stock price fully reflects all publicly available information and any abnormal returns cannot be generated by any other fundamental nor technical analysis. However, there are also some limitations in pervious research. (1) For different national institution systems is rare. (2) The comparison of different reactions to earnings announcement of different stock markets in developed and developing countries is rare. (3) The research of different attitudes to the earnings announcement between individual investors and institutional investors is rare.