2. LITERATURE REVIEW
2.4 Post-Earnings Announcement Drift
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
13
2.4 Post-Earnings Announcement Drift
Ball and Brown (1968) first documented the phenomenon of post-earnings announcement drift that stock prices continue to drift after earnings announcements.
Since then, researchers have investigated the phenomenon and attempted to provide explanations. Competing explanations for post-earnings announcement drift generally fall into two categories. One is the model used to calculate abnormal returns, which leads to incomplete risk adjustment in the estimation of abnormal returns. Kim and Kim (2003) argue that most of prior studies related to post-earnings announcement drift may use the mis-specified models and fail to adjust raw returns fully for risk. The other category of explanations suggests that, the stock prices fail to fully reflect the current earnings surprise. Kormendi and Lipe (1987) and Freeman and Tse (1989) suggest that responses to current earnings reflect at least some of the implications for future earnings, but that doesn’t mean the immediate response is complete. Bernard and Thomas (1989) show that the evidence is inconsistent with the explanations based on incomplete risk adjustment but due to delayed price response. Why does the market fail to response to earnings information instantaneously?
One possibility is that transactions costs impede a complete and instant response to earnings information. Bhushan (1994) uses the informational efficiency framework, in this perspective, trading and investment by professionals help bring prices consistent with fundamentals. However, since that transaction costs can prevent professionals from trading in its shares, firms can be mispriced. Bhushan (1994) divides transactions costs into two parts: direct costs (include percent bid-ask spreads and commissions) and
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
14
indirect costs (include the adverse price effect and the delay in processing the transaction). Due to the evidence revealing that direct costs of trading are inversely related to share price and larger trades of stocks can be accomplished without delay or adverse price impact, the proxy for the inverse of direct and indirect costs of trading are share price and annual dollar trading volume, respectively. The paper shows that the post-earnings announcement drift is positively related to transactions costs and suggests that transactions costs are an important determinant of the efficiency of capital markets.
Ng et al. (2008) use standard market microstructure features to examine the effect of transaction cost in the post-earnings announcement drift. They suggest weaker abnormal returns at earnings announcement and higher returns at the subsequent period, for firms with higher transaction costs. Thus, transaction costs restrain profitable trades by informed investors that are required to drive the market price in line with the fundamental value at the time of earnings announcement. Transaction costs can provide an explanation for the persistence and existence of post-earnings announcement drift (Ng et al. 2008; Chung and Hrazdil 2011).
It is also possible that that the market is incapable of fully interpreting the implications of earnings information due to information processing capabilities. Liang (2003) examines the relation between information processing biases and post-earnings announcement drift. The empirical evidence shows that drift positively relate to heterogeneous information and negatively relate to the change in uncertainty around earnings announcements. It seems that two important factors which explain drift are investors’ overconfidence about their private information and under confidence of more
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
15
reliable information. Ayers et al. (2011) also state that large traders’ under reaction and the related drift may be referred to a longer price discovery process when earnings are more difficult to interpret. Asthana (2003) argues that as a result of information revolution, the cost of accessing and shipping information of companies is greatly reduced. Such information revolution indeed affect the informational efficiency of the capital market and reduce the post-earnings announcement drift, after controlling for several factors, such as time, size, investor sophistication, and sign of analysts’ forecast errors, etc. That is, the advance in information technology may reduce trading friction and promote informational efficiency. Engelberg (2008) examines the relation between information processing cost and post-earnings drift. The paper suggests that information is heterogeneous in type and classify then into hard (such as an income statement of a firm) and soft (such as the transcript of firms’ conference call) information. Hard (soft) information has higher (lower) processing costs, which lead to under reaction phenomenon after earnings announcement (i.e., when information processing is costly, information may not be incorporated into stock prices instantly and completely).
Besides, readability of quarterly reports can also affect information processing capabilities. For example, Lee (2012) investigates how readability of quarterly reports affects the speed at which earnings news is impounded into stock prices. The results reveal that less (more) of the earning-related information is reflected in stock prices during 10-Q filing (post-filing drift) window for firms with poorer readability disclosure in quarterly reports. That is, only providing more disclosures of quarterly report do not
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
16
facilitate market efficiency, quarterly report readability should further be considered.
In addition to the speed of investors’ response to new information that may lead to post-earnings announcement drift, studies also attribute the market reaction phenomenon to analysts’ slowness in revising their earnings forecasts. Such as Zhang (2008), examines the responsiveness of sell-side security analysts’ forecast revisions after quarterly earnings announcements and shows that the earnings response coefficient in the event window is significantly higher and the corresponding post-earnings announcement drift is significantly lower for firm-quarters with responsive analysts.
Thus, responsiveness of analysts will reduce the drift and contribute to market efficiency.
Several studies examine whether individual investors or institution investors influence post-earnings announcement drift. Bartov et al. (2000) assume that there are two types of investors in the market. Market participants who are experts in gathering and processing public information called sophisticated investors, and others called unsophisticated investors. Institutional investor holdings of a stock are used as a proxy for investor sophistication. They illustrate that, because sophisticated investors are expected to characterize the process underlying earnings correctly and unsophisticated investors perceive the process to be a seasonal random walk, the degree of abnormal returns after earnings announcement is inversely related to the proportion of firm’s stock held by institutional investors. However, the results of tests evaluating the validity of institutional holdings as a proxy for their variable are mixed, which may affect the persuasiveness of their findings. Hirshleifer et al. (2008) examine the relation between
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
17
actual daily signed trades made by individual investors after earnings surprises and subsequence returns, which is distinct from the indirect method used by prior studies (i.e., examine the fraction of shares held by institutions). They suggest that individual investors do not cause post-earnings announcement drift. Different from past studies, Ayers et al. (2011) argue that after earnings announcements, small (large) traders trade in the direction of seasonal random-walk-based (analyst-based) earnings surprises. Small traders’ fail to digest the time-series property of earnings, which lead to delayed small trades and larger traders have a longer price discovery process that is reflected in the delayed large trades. They also find that the more these traders react to the earnings news during the announcement period, the lower magnitude of the post-earnings announcement drift will occur.
Based on prior studies, this study investigates whether VACs are associated with post-earnings announcement drift.
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
18