CORPORATE ANNOUNCEMENTS OF EARNINGS AND STOCK PRICE BEHAVIOR: EMPIRICAL EVIDENCE

Transcription

1 CORPORATE ANNOUNCEMENTS OF EARNINGS AND STOCK PRICE BEHAVIOR: EMPIRICAL EVIDENCE By Ms Swati Goyal & Dr. Harpreet kaur ABSTRACT: This paper empirically examines whether earnings reports possess informational value that leads to changes in the equilibrium market price of the stocks traded on the Bombay Stock Exchange (hereafter BSE). The main objective of the study is to investigate the impact of quarterly earnings announcements on the stocks constituting the Sensex. The study covers 119 quarterly earnings information releases of various firms of the sensex of the total number of disclosures. The study adopted event study methodology of MacKinlay to examine the impact of quarterly earnings announcements on stock returns and applied Markowitz model to calculate the variance of abnormal returns of securities included in the portfolio. The AARs and CAARs of the overall sample of 119 announcements are insignificant at 5% level of significance. The overall sample was divided into 2 sub samples of good announcements and bad announcements constituting 74 and 45 announcements respectively. No evidence of significant abnormal returns could be found in both these sub-samples. This implies that the quarterly earnings announcements do not have any pre-return or post-return effects on the firms included in the sensex. It may also be inferred that these announcements carry little information value for investors. INTRODUCTION Economists are frequently asked to measure the effects of an economic event on the value of firms and index. Using financial market data, an event study measures the impact of a specific event on the value of a firm and index. The usefulness of such a study comes from the fact that, given rationality in the marketplace, the effects of an event will be reflected immediately in security prices. Thus a measure of the event's economic impact can be constructed using security prices observed over a relatively short time period. The information content of earnings is an issue of obvious importance for investors. Company earnings announcements are closely-watched events, being the main source of new information about company performance. It has been well-documented that prices respond quickly to the information in earnings announcements with investors reacting positively to good news and negatively to bad news.

2 However, when companies meet expectations, the price response is typically close to zero. The effect of the earnings announcement drift has been found to persist for a significant time, creating post earnings announcement drift. However, the magnitude of the effect and the time taken to incorporate the information content of the magnitude of the effect and the time taken to market and depend on characteristics of the firm under study. Thus, the effect of the earning announcements is an important empirical matter in the capital market, influencing the movement of the share prices. REVIEW OF RELATED LITERATURE In the international scenario, there have been a number of empirical studies on earning announcements. Ball and Brown (1968) identified that about 10%-15% of the information contained in announced earnings had been anticipated by the month of the preliminary announcement. In a leading paper Beaver (1968) examined the extent to which common stock investors perceive earnings to possess informational value. The study investigated the reaction of investors against the annual earnings releases as reflected in the volume and price movements of common stocks. Beaver hypothesized that if earnings reports convey information in the sense of leading to changes in the equilibrium value of the current market price, the magnitude of the price change (without respect to sign) should be larger in the week of the announcement than during the non-report period and reports evidence of it. As defined by Fama (1970), a stock market is efficient if prices always fully reflect available information. Information is divided into three subsets, distinguishing between weak, semi-strong and strong form efficiency with respect to historical prices, publicly available information, and private information, respectively. One interesting insight from this literature is the revelation in Grossman and Stiglitz (1980) that prices can only fully reflect costless information, since there must be a return to acquiring information at a cost, otherwise there will be no information acquisition. In two papers, Brown and Warner (1980; 1985) examine the behavior of the stock returns around the announcements using monthly and daily return data respectively. In these extensive studies, they examine different abnormal return generating methods and conduct various sensitivity analyses. Foster et al, (1984), found abnormal stock returns after announcement of Quarterly earnings. Certain studies further identified some firm-specific factors, viz, market capitalization, functioning of concerned stock market, etc. which elucidates the dynamics of stock returns around the earnings announcement date in a systematic manner. Atiase (1985) focuses on firm size (capitalization) as one such characteristic. He argues that the amount of unexpected information conveyed to the market by actual earnings reports should be inversely related to firm capitalization. Grant (1990) observed that the market in which a firm s securities are traded often determined the behaviour of the stocks return around the earning announcements. This insight indeed led to a revised definition of efficiency in Fama (1991), where two versions of the hypothesis that security prices fully reflect all available information are given. The strong version stipulates that information and trading costs are always zero, while the weaker version states that prices should reflect information to the point where the marginal benefits of acting on information do not exceed the cost. Jegadeesh and Lakonishok (1996) assumed that the delay of investors reaction can reach as much as three quarters of a year. However, this delay is not always symmetrical. Overall, postannouncement results suggest that

3 investors over-react when the earnings surprise from the prior quarter is not repeated in the next quarter: investors are disappointed by firms meeting expectations when they were expecting another positive surprise, and they are pleasantly surprised by firms meeting expectations when they were expecting another negative surprise. Some of the above-mentioned studies show that companies that unexpectedly reported inauspicious results often yield slightly higher abnormal rates of return in post-announcement period (in absolute values) than the companies that reported surprisingly good results. Since Ball and Brown first documented the phenomenon of post-earnings announcement drift, the price response to earnings announcements has been analyzed in many different ways. Risk-based or transactions costs arguments for the price drift have not had much success, whereas arguments based on behavioral arguments appear to explain this effect. A related strand of literature, reviewed in Verrecchia (2001), has dealt with the theoretical modeling of how the disclosure of information affects investors as reflected in stock prices and trading volume. Liang (2003) explains the drift by investor under-reaction to public information. The price response for earnings announcements follows an S-shape, with the most sensitive price response being to small positive or negative earnings surprises, whereas the response for larger surprises is proportionally lessened. Thus, hitting the consensus earnings forecast or just missing it has the most effect on price. INDIAN CONTEXT Several studies focus on certain firmspecific characteristics for a better explanation of stock returns around the earnings announcements. In the Indian scenario, there has been a limited number of research studies on earning announcements. Obaidullah (1990) investigated the market reaction to halfyearly earnings announcements and found that the Indian stock market is semi-strong efficient. Prior research on stock return reaction to earnings announcements on the BSE is limited, both in terms of the number of studies and the samples used in these studies (Eser, 1994). This study, first, examines if security prices change in response to earnings announcements. For this purpose, the paper investigates whether the magnitude of the price change (without respect to sign) is larger on the announcement day than during the nonevent period. Second, the earnings announcements are assigned to good news and bad news subsamples by using a simple expectation model. Then, for each group, the changes in stock prices around the announcements are examined to verify whether statistically significant price changes occur in the predicted direction (i.e. a positive price activity for good news and conversely for bad news). Chaturvedi (2000) studied the stock price reaction to semi-annual earnings announcement and found abnormal returns both during the pre and post announcement dates. Mallickarjunappa (2004) found that the Indian stock market is slow in incorporating the quarterly earning information. A similar study by Gupta (2006) investigated the information content of earnings announcement made in the year ending of companies constituting CNX Nifty Index and found significant abnormal returns around the announcement date. Das et al, (2007) studied the effect of quarterly earnings announcements made by the large companies do not have substantial impact on stock return. However, this study was limited to a single quarter earnings releases comprising only 30 announcements, and moreover to effect of clustering of multiple events in the overall return behavior of the firms was not incorporated. In the present study, an effort is made to overcome this limitation and investigate the information content of quarterly earnings

4 announcements considering multiple event periods covering 119 announcements. RESEARCH METHODOLOGY This study empirically investigates the impact of quarterly earnings announcements made by 30 companies constituting the BSE-Sensex on stock price movement. The firms comprising Sensex have relatively large market capitalization in comparison to other firms listed in different indices of Bombay Stock Exchange (BSE), hence they have been termed as large firms. The announcements made by these large firms correspond to the quarterly earnings from the first quarter to the last quarter of the financial year The requisite data were retrieved from the Capitaline Data Base. The importance of clustering of events has been considered in the design of the significance test of abnormal returns for understanding the announcement effect. Clustering of events takes place when a number of quarterly earnings announcements are made on a single date. This necessitates the incorporation of the variance as well as covariance terms of the aggregated sample abnormal returns across securities. The main objective of the study is to investigate the impact of quarterly earnings announcements on the stocks constituting the Sensex. If quarterly earnings announcements have informational significance for investor, then such announcement should induce abnormal stock return during the period surrounding the announcement. The study also investigates whether stocks manifest price drift over a period of time corresponding to good and bad announcements. Usually, statistically significant price changes occur in the predicted direction, i.e. upward drift in the case of good announcements and downward drift in the case of bad announcements. Event Study methodology using daily returns along with market model is used for the purpose of analyzing the quarterly earnings announcements effect. An event study examines the behavior of a sample of firms experiencing a common type of event, example, earning announcement, stock split, issue of new debt or equity, merger and acquisition and so on. This method was originally introduced by Ball and Brown (1968) and Fama et al, (1969) and subsequently modified by Campbell et al. (1997), and MacKinlay model (1997) to determine the abnormal return, AAR and CAAR, hitherto been referred to as MacKinlay model (1997). The dates on which earnings announcements are released are defined as the event dates (t=0) provided that the stocks have been traded on that date. In the eventuality of announcement of quarterly earnings during non-trading days, the trading day immediately following the announcement date has been assigned as event date. This has been done in order to investigate the announcement effect immediately after the announcement. The effects of quarterly earnings announcements on equity share prices have been examined by taking daily-adjusted market price data. The 15 days surrounding the announcement of earning (i.e., t = -15 to t = +15) is assigned as the event period. The market model of Sharpe (1964) has been used to estimate the expected returns on a stock: Ri = α i + β i R m + e i where, R i is the return on security i on day t, α i and β i are the ordinary least squares estimate for i s market model parameters, R m is the return on the market index on day t, e i is an abnormal return, which is an independently and identically distributed residual error term.

5 Ordinary least squares regression is performed to estimate the coefficients of the market model using the estimation window. The estimated coefficients, αi and βi are used to form prediction of R i during the event period. Thus, the abnormal return (ARi) for security i on event day t is calculated as: e i = AR i = R i - α i - β i R m where, E(e i ) =0 and var (e i ) = σ²e i The abnormal return observations need to be aggregated in order to draw overall inferences about the impact of the event on the price of the securities. The aggregation is done along two dimensions- through time for an individual security and then aggregation both across securities and through time. The average abnormal return, AAR i for the event day t for a sample of N securities is calculated as shown under in order to eliminate the effect of any one or group of securities on the abnormal return: AR t = (1 / N) ARit for t = -15,..., +15 The Cumulative Average Abnormal Return (CAAR) was subsequently calculated for event days t 1 through t 2 in order to determine the cumulative effect of AARs on days surrounding the event as: CAAR (t 1,t 2 ) = AARt When AAR is calculated on a given event day, the positive and negative price activities in reaction to unexpected earnings performance would cancel out each other and as such the overall sample was subdivided on the basis of good announcements and bad announcements. In this case of individual securities that are not subjected to clustering phenomenon, if there is an increase in Earning Per Share (EPS) for the quarter under study compared to the previous quarter, the announcement will be designated as good announcement. On the other hand, if there is a decline in earnings for the quarter under study relative to the previous quarter, then it will be inferred as bad announcement. Next, the null hypothesis H o, that the event has no impact on the distribution of returns, is to be tested. Now, if quarterly earnings announcements have some informational content for the investors, then AAR for the days surrounding the announcement will be greater than AAR during non-event days for good announcement and vice versa for bad announcement. The significance of abnormal return was tested by using models specified by MacKinlay (1997). Using models specified by Mackinlay (1997) for event day, z-statistic for the AAR on N securities is: Z = AAR t [Var(AAR t )]½ where, Var(AAR t ) is the variance of AAR for N securities on event day t Var(AAR t ) = Var(AR i ) and N² Var(AR it ) = (AR it )² (t 2 -t 1-1) In order to test CAAR for N securities over T days (event days t 1 through t 2 ) the z-statistic is computed as: Z t = CAAR(t 1, t 2 ) [Var{CAAR(t 1, t 2 )}]½ where, Var{CAAR(t 1, t 2 )} = Var(AAR t )

6 EMPIRICAL RESULTS AND FINDINGS Table 1 highlights the AAR, CAAR and the corresponding z-values for each of the 30 days of the event window for all the 119 quarterly earnings announcements. On the day of earning announcement, the AAR is %, with z-value of % and CAAR is % with z-value of %, both of which are insignificant at 5% level of significance. Further analysis indicate that the AARs and CAARs for each day prior to and post announcement days are equally insignificant at 5% level of significance. TABLE 1: Significance Test for AAR and CAAR for full sample Days AAR z-statistics CAAR z-statistics -15 NA NA NA NA

7 Figure 1 and Figure 2 shows the AARs and CAARs respectively. We find that on the day of announcement, the AAR attains the maximum value of %. However, the AAR fluctuates randomly prior to and post announcement event period, and no inferences can be made that the abnormal returns are having increasing or decreasing trend due to quarter earnings announcements. The CAARs of the firms attained negative values for all the event days. Moreover, the cumulative effect of abnormal returns showed downward drift continuously. This might reflect underperformance of the index with respect to the expected model of market return. But statistically significant abnormal return could not be found during this period. The effect of announcement is not discernable. 1 0,5 0-0,5-1 -1, Figure 1: Average Abnormal Return (AAR)

8 0-0,2-0,4-0,6-0,8-1 -1, Figure 2: Cumulative Average Abnormal Return (CAAR) Table 2 presents the AARs, CAARs and corresponding z-values using MacKinlay model for the sub-sample of good announcement during the event period. The total number of announcements in this sub-sample is 74. On the day of earnings announcement, the AAR is % and CAAR is %, both of which are insignificant at 5% level of significance. The AAR has been statistically significant at 5% level of significance and attains the maximum positive value of % a day after the announcement date. Since the abnormal return provides a measure of expected change in security holder s wealth, the value of AAR reflects that the good announcements have positive effect on the stock prices of the securities. However, the effect subsidies in the following days. Statistically significant AARs have been noticed in the event days. The movement of AAR across the event days was random and, therefore, no definite conclusionn could be arrived at about the effect of good announcement on the sensex. Also the CAAR fluctuates random with a downward trend. Table 2: Significance Test for AAR and CAAR Sub-Sample: Good Announcements Days AAR z-statistics CAAR z-statistics -15 NA NA NA NA

9 Table 3 shows that the result involving bad announcement sub-sample. There are 45 bad announcements in this sub-sample. On the day of earnings announcement, the AAR was % and CAAR was %, both of which are insignificant at 5% level of significance. Here also the AARs of this set of firms are statistically insignificant prior to and post announcement day, except the prior to the announcements. So, it is inferred here that no association can be establish between this abnormal return and the announcements. Since large firms are subjected to more scrutiny by the investors, it is, therefore, expected that the share price reaction to such announcements should be statistically insignificant.

10 TABLE 3: Significance Test for AAR and CAAR Sub-Sample: Bad Announcements Days AAR z-statistics CAAR z-statistics -15 NA NA NA NA

11 CONCLUSION This study examined the impact of quarterly earnings announcement on the security returns of the security returns of the firms constituting the sensex. To investigate the announcements effects, quarterly earnings disclosures made by the 30 companies BSE-Sensex were considered. The study covers 119 quarterly earnings information releases of various firms of the sensex of the total number of disclosures. The study adopted event study methodology of MacKinlay to examine the impact of quarterly earnings announcements on stock returns and applied Markowitz model to calculate the variance of abnormal returns of securities included in the portfolio. The AARs and CAARs of the overall sample of 119 announcements are insignificant at 5% level of significance. The overall sample was divided into 2 sub samples of good announcements and bad announcements constituting 74 and 45 announcements respectively. No evidence of significant abnormal returns could be found in both these sub-samples. Further, it could not be established that the share price drift positively in the case of good announcements and drift negatively in the case of bad announcements. The reason for such an observation could be that most of the prior studies were based on samples constituting both large as well as small firms; whereas the present study comprised large firms only. Large firms are subjected to greater attention by the market participants and, therefore, fundamental information is quickly incorporated into prices, leaving no scope for systematically earning superior returns. This implies that the quarterly earnings announcements do not have any pre-return or post-return effects on the firms included in the sensex. It may also be inferred that these announcements carry little information value for investors. The index also reflects higher level of market efficiency where the influence of non-fundamental factors, which are largely behavioral, is the least.

12 About Authors Ms Swati Goyal, Assistant Professor, S.D. College for Women,Moga, Punjab, Pin Dr. Harpreet Kaur Kohli, Assistant Professor, Dept of Distance Education (Commerce), Punjabi University, Patiala,