University of Amsterdam
BSc in Economics and Business Economics Specialization in Finance
Dividend Announcement Effect under Asymmetrical Information
An event study of dividend announcement effect under asymmetrical information in 2009-2019
Che An 11797428
Supervisor: Drs. P.V.Trietsch, M.Phil June, 30th, 2021
This study aims to revise dividend announcement effect and the enhancement under asymmetrical information by using recent data from 2009-2019. Past literatures suggested that dividend announcement will lead to significant high abnormal return for firm, and is especially high for firms with higher asymmetrical information. The methodology includes event study on dividend announcement effect using CAAR and cross-sectional analysis on the effect of asymmetrical information in forms of three proxies selected, namely, number of analysts, institutional ownership, and corporate governance index. Partly consistent with past literatures, this study found significant high abnormal return following dividend initiation announcement of firms, but rejects the hypothesis that dividend announcement is enhanced under higher asymmetrical information in the sample period of 2009-2019. The failing in rejecting the hypothesis of asymmetrical information enhance dividend announcement effect rests on the fast growing technology to information gathering process and increased legislation of the general economic and markets, which sparks potential future researches from this aspect.
Statement of Originality
This document is written by Student Che An who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
UvA Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
1 Introduction ... 4
2 Literature Review ... 5
2.1 Dividend Payout Reasoning ... 5
2.2 Dividend Announcement Effect ... 5
2.2.1 Signaling Theory ... 6
2.2.2 Free Cash Flow Theory ... 6
2.2.3 Agency Cost Theory ... 7
2.3 Asymmetrical Information ... 7
2.3.1 Pecking Order Theory ... 7
2.3.2 Signaling Theory (with Asymmetrical Information) ... 8
2.4 Asymmetrical Information Measurement ... 8
2.4.1 Number of Analysts ... 9
2.4.2 Corporate Governance Index ... 10
2.4.3 Institutional Ownership ... 10
2.5 Empirical Evidence ... 11
2.5.1 Dividend Announcement Effect ... 11
2.5.2 Dividend Announcement Effect with Asymmetrical Information ... 13
3 Data ... 14
3.1 Variable Summary and Data Collection ... 14
3.2 Data Analysis ... 16
3.3 Hypothesis Summary ... 17
4 Methodology ... 18
4.1 Event Study Specifications ... 18
4.2 Construction of the Dividend Announcement Model ... 19
4.3 Cross-Sectional analysis for asymmetrical information ... 20
5 Results ... 22
5.1 Event Study Results ... 22
5.2 Cross-sectional Analysis Results ... 23
6 Conclusion ... 26
7 References ... 27
The studies of dividend announcement effect are normally set to an event study methodology that examine the relation between stock return and the dividend announcement (Abdullah et al. 2002).
The disparity of stock return lies within the effect of the prior announcement. There are various prevalent theories set to explain the dividend announcement effect.
Miller and Modigliani (1961) make a large contribution by introducing explicitly, the issuance of cash-dividend is a signal to investors sent out by managers. The studies of John and Williams (1985), Bhattacharya (1979, 1980) and Miller and Rock (1985), showed further that changes in dividend announcement are conducted by managements, at a cost, to purposely send out signals to outside investors about future earnings prospects. Furthermore, the study by Jensen (1986) of free cash flow theory, suggested that an increased amount of dividend is also a decrease in free cash flows on hand of a firm. Combining these prior researches, the dividend announcement effect can be concluded such that the increase in dividend announcement, intentionally by management, are aimed to send out positive signals about a firm’s current operations and future prospects that will increase investors’ confidence.
Hence, there are abnormal return for the firm.
The link between asymmetrical information and dividend announcement effect is a less popular study among past economists. This paper aims to integrate the idea of asymmetric information into the study of dividend announcement effect. By establishing a model for abnormal earnings using dividend announcement under and without the effect of asymmetrical information. Even though past papers have introduced several proxies on asymmetric information, this paper researches the power of these proxies to introduce the more dominant proxy. It is of great importance to provide a better proxy for the quantifying process of asymmetric information for better analytical power of forecasting models regarding dividend announcements. A general goal of this paper is to revise the study and results of past literature on similar topics . By implementing a more recent sample period of 2009-2019, the consistency of the dividend announcement effect in a more recent time horizon is tested, as well as the measurements of asymmetrical information and their effect on dividend announcement effect.
Hence, the central research question for this paper is: To what extend is the dividend announcement effect enhanced under higher asymmetrical information?
This paper is organized as follows. Firstly, the past literature on the topic will be reviewed.
Then, the data needed will be explained, collected, and analyzed on basic statistic level. The methodology will be performed and subsequently, the result of the tests will be presented. Lastly, the conclusion will be drawn and interpreted economically.
2 Literature Review
In this section, past literatures are reviewed to provide insights of many frameworks associated with dividend announcement effect and how it differs after including the effect of asymmetrical information. Firstly, dividend payout reasoning will be explained. Secondly, past literatures on three theories, the Signaling Theory, the Free Cash Flow Theory, and the Agency Cost Theory, without asymmetrical information effects will be discussed. Thirdly, past literatures on two frameworks, the Pecking Order Theory and the Signaling Theory, with effects of asymmetrical information will be discussed. Fourthly, the measurement of asymmetrical information will be discussed and three potential proxies, Number of Analysts, Institutional Ownership, and Corporate Governance Index, will be rationalized. Lastly, past empirical evidence of dividend announcement effect under asymmetrical information will be presented.
2.1 Dividend Payout Reasoning
Firms payout dividends under various conditions. Most generally, when a firm has plentiful free cash flows on hand, they may dispense them as dividends payouts to reward their investors. Even though a firm’s free cash flows are those left after making all necessary investment for future growth, questions can still be asked as in why would firms pay their investors additionally by issuing dividends, when they can hold the cash on hand for other uses? This question is justified as firms indeed have many other ways to distribute their free cash flows. The study by Richardson (2006) gave a few examples. Firms may reserve these free cash flows for future unexpected opportunities such as mergers and acquisitions. Furthermore, firms may hold excess cash as reserves, in order to have a safe liquidity cushion to fall back on when future economic prospects are uncertain and have recession probabilities.
However, dividends are just as important as the abovementioned rationale for free cash flows.
Besides the attraction of dividends payout as a reliable source of return for investors, they also serve as a crucial intermediary of communication between firms and its investors. The core to this communication intermediary, is the signal sent out by firms, regardless of it being a good or bad signal, to investors as information to react upon. Ideally, firms capitalized the information content of a dividend payout to send out good signals, such that to accumulate more earnings, through stock and investments.
Alternatively, investors can receive information, such as the firm's current operations and future growing prospects, through dividend payouts, to adjust their investments accordingly.
2.2 Dividend Announcement Effect
Most of the essential studies on the subject of information content of dividend payouts originate from the fundamental paper by Miller and Modigliani (M&M) in 1961. M&M (1961) advanced the Dividend Irrelevance Theory, which linked dividend payouts to stock prices. The theory incorporates
the assumption that, in a perfectly frictionless world, markets perform as efficiently as possible. They reasoned that while dividend payout increases, it is done so at a loss of capital value of the shareholders, which will compensate each other. Hence, the share price of a firm, along with its value, is not affected by a distribution of dividend. Nevertheless, the assumption of a perfectly frictionless world is lacking validity justification empirically. Various theories built upon the Dividend Irrelevance Theory provided insights to this real world imperfection.
2.2.1 Signaling Theory
Among the many imperfections overlooked in the Dividend Irrelevance theory, asymmetrical information is linked closely to dividend payouts, which then provides significance to the Signaling Theory. The cash flow signaling theory was developed years after the work of M&M in 1961 by Bhattacharya (1979), Miller and Rock (1985), and John and Williams (1985). The models state that, in a frictional world where asymmetric information persists, better informed management use dividend payouts as a signal to deliver the future prospects of their firms to less informed outsiders, in a costly manner (Vieira, 2007).
In a study about the impact of dividend payment, Miller and Rock made the assumption that the magnitude of asymmetric information can be measured with respect to the current internal cash flow of a company, and that the magnitude of symmetric information can be assessed with the planned investment and value of assets of the company. On the authority of Miller and Rock, assuming a constant amount of external financing and investment, cash dividend payouts are generally coupled with the operating cash flow of the company. If the dividend payout announced is higher than the expectations from the market, it discloses an potential increase in the future cash flow of the company, which will in turn raise their stock prices due to reliable future prospects. Hence, the theory theorized that when dividend increases, stock price also increases, which holds conversely as well. In addition, the magnitude of cash dividend payouts is correlated to the magnitude of permanent earnings, which the stock value is then affected.
2.2.2 Free Cash Flow Theory
Another form of information conveyed is the free cash flow usage of the firm. Jensen proposed the free cash flow theory in his study in 1986. As reported by Jensen (1986), free cash flow is the cash leftover after all necessary investments for future growth. The excess cash on hand of the firm can cause agency conflicts between the shareholders and managers. Shareholders would favor excess cash usage to maximize firm value which then increases stock value and dividend payouts which is a form of return to their investments. In contrast, managers have incentives to pursue overinvestments on negative net present value projects to maximize short term company performance but harm the future earnings prospects of the firm. Therefore, announcing dividend payouts, will send out signals to outside investors
that the company is not making overinvestments on inadequate projects that will harm the future earnings prospects of the firm.
The Free Cash Flow Theory explains the stock reaction of investors due to the dividend announcement, based on signals that the investment policy of the firm has changed. The theory thus hypothesizes that if there are unexpected dividend payouts, stock prices will increase accordingly.
2.2.3 Agency Cost Theory
The study of Easterbrook (1984) showed similar predictions with the study of Jensen (1986) regarding dividend announcement as a signal regarding free cash flow usage. Easterbrook proposed a different reasoning where agency cost theory acts as fundamental rationale. As stated in Easterbrook’s study, the separation of control in a firm between the manager and the shareholders will incentivize managers to misuse free cash flow of the firm for personal gains. Managers may pursue personal leisure in the form of inadequate asset acquisition, such as purchasing a private jet for an unreasonable large number of business trips. A cash dividend payout on a regular basis will signal outside investors that the managers are aware of their actions. Whereas, a decrease in dividend payout will carry a bad signal that management can potentially allocate more of this excess cash and resources into perquisites.
Hence, from the viewpoint of agency cost theory, a reduction in dividend payout will reduce the equity value of a company. Therefore, a negative price effect is produced due to the information conveyed by the dividend announcement.
2.3 Asymmetrical Information
The above mentioned studies of Easterbrook (1984), Miller and Rock (1985), and Jensen (1986), merely treated asymmetrical information as a friction of the market. It is treated as a characteristic of the market instead of a variable where it can have quantitative effects on the dividend announcement effect. The studies ignored the effects of different sizes and levels of asymmetrical information on the market reaction of dividend announcement. Past literatures have concluded two predominant theory of quantitative asymmetrical information effects, that is the Pecking Order Theory and The Signaling Theory.
2.3.1 Pecking Order Theory
Myers and Majluf (1984) provided a modification of Pecking Order Theory which included a quantitative measure for asymmetrical information and how it can affect dividend announcement effect.
According to them, in a world where asymmetrical information persists, a firm may underinvest in particular situations. The possibility of underinvesting translates into an ex-ante possible loss of firm value, which arises out of the lemons problem considering the new capital issuance. They stated that the possibility of underinvesting can be reduced by making use of retention to accumulate slack.
Consequently, the excess retention to accumulate slack comes from a decrease of dividend payouts.
Therefore, dividend payout policies can be seen as a control of underinvesting problems caused by asymmetrical information.
Hence, with other variables being unchanged, this theory hypothesized that the higher the magnitude of asymmetrical information, the lower the dividends. Hence, according to pecking order theory, dividend announcement effect is diminished under higher asymmetrical information.
2.3.2 Signaling Theory (with Asymmetrical Information)
Miller and Rock (1985) also proposed an advanced model of Signaling Theory which included asymmetrical information as a scalar variable instead of a characteristic of markets. This advanced model built upon the classical model of signaling theory by incorporating the size effect of asymmetrical information on dividend announcement effects. It further states that, in equilibrium, a firm signals their higher levels of current earnings and future earnings prospects by issuing higher dividends to stand out among those with lower levels of current earnings and less optimistic prospects. Moreover, when under asymmetrical information in a frictional market, dividend payouts are higher compared to that under perfectly frictionless market. These arguments by Miller and Rock (1985) suggest that, with all other things equal, firms with higher magnitude of asymmetrical information will have to pay higher levels of dividends in order to send out the same signal regarding the identical amount of current earnings than those with lower magnitude of asymmetrical information.
Therefore, the advanced model of Signaling Theory predicts that, the higher level of asymmetrical information, the higher the dividends payout. Hence, according to signaling theory, dividend announcement effect is enhanced under higher asymmetrical information.
2.4 Asymmetrical Information Measurement
Measuring asymmetrical information can be done by evaluating and measuring proxies, as it cannot be directly measured or observed. The explanatory power and forecasting power of the empirical model lies greatly within variables selection (Mitra and Owers, 1995). Even though several studies (Mitra and Owers, 1995; Bhushan, 1989; Agbetonyo et al., 2018; etc.) have provided proxies for asymmetrical information, there is not a dominant proxy determined for asymmetrical information both theoretically and empirically among these studies. Furthermore, many studies (Easley et al., 1996;
Llorente et al., 2002; etc.) provide proxies of asymmetrical information between informed and underinformed investors, rather than between firms and outside investors (Bharath et al., 2006).
However, asymmetrical information between firms and outside investors is of greater importance when applying to dividend announcement effect. Therefore, this paper proposes three potential proxies of
asymmetrical information between firms and outside investors, namely number of analysts, corporate governance index, and institutional ownership. These proxies will be reviewed in the following section.
In summary, the relation between asymmetrical information and the three proxies selected is inversely correlated. The higher number of analysts, institutional ownership, and corporate governance index will indicate a lower asymmetrical information.
2.4.1 Number of Analysts
Number of analysts is used as a proxy to determine the asymmetrical information between firms and their investors (Bhushan, 1989; Brennan and Hughes, 1991). It is assumed that as the number of analysts increase, there will be lower asymmetrical information between firms and investors. It is supported by various literature that the number of analysts could be a superior proxy that can better capture the magnitude of asymmetrical information in comparison to other proxies.
Bhushan (1989) suggests that the quantity of resources devoted to studying firms and obtaining private information to contemplate for the existing information gap will decrease alongside an increase in the number of analysts following that firm. Hence, the asymmetrical information between managers and outside investors will gradually decrease when the number of analysts following a firm increases.
Brennan and Hughes (1991) as well as D’Mello and Ferris (2000) provided similar arguments. They start off their argument stating that the number of analysts following a firm is a proxy for information flow, as the number increases, the flow experiences less friction and can act more efficiently and timely.
Furthermore, they stated that analysts act as an intermediary of information between investors and firms.
Hence, the more number of analysts, the more continuous the flow rather than discrete and gapped, and thus lower asymmetrical information. Lang and Lundholm (1996) provided more intuitive rationale stating that more analysts choose to follow firms with greater degree of disclosure about their information. This then means that the higher number of analysts following a firm, the natural state of the firm is of less asymmetrical information.
This paper adapts the number of analysts as one of the three proxies tested for asymmetrical information measurement. Even though the limitation exists due to difficulties in finding the accurate number of analysts at the exact time of dividend announcement, the validity of this proxy is still justified. The reason is that information and disclosure is built up gradually through following analysts.
With the argument from Lang and Lundholm (1996), the number of analysts acquired in present time is still sufficient to explain the magnitude of asymmetrical information at the time of the dividend announcement.
2.4.2 Corporate Governance Index
Corporate governance index serves as an important source of information to outside investors about the governance side of a company. Governance of a company can include various important aspects of a company, such as board structure, shareholder option, compensation and investment decisions. It is assumed that firms with higher levels of corporate governance are voluntarily disclosing more information to outside investors. As more private information is disclosed to the public, investors learn more about the company, which led to lower levels of asymmetrical information.
The effect of corporate governance on asymmetrical information has been studied in various papers. However, they are mostly assessing limited aspects using a particular mechanism of corporate governance. Wruck (1993) assessed the role of executive compensation structure on reducing agency cost problems and asymmetrical information. Cai et al. (2006) and Holm and Scholer (2010) suggest that the more independent the board is, the lower the asymmetrical information. They argued that an independent board has the ability and incentive to reject bad investment decisions to save cash to adopt better projects. The study of Chen and Nowland (2010) shared similar results and argued that board independence will contribute to increasing the interests of minority investors through reducing the underinvestment problem.
Empirically, Van der Bauwhede et al. (2008) argued, by focusing on the European firms, that disclosure of corporate governance information will reduce asymmetrical information through mitigating agency cost problems. By disclosing and improving corporate governance, investors will gain more confidence in the other reports disclosed by the firm.
2.4.3 Institutional Ownership
Institutional ownership is considered a proxy for asymmetrical information as institutional ownership has effects on firms information environment. It is assumed that the higher percentage of institutional ownership, the lower asymmetrical information exists between the firm and its outside investors. In contradict, the study of Heflin and Shaw (2000) suggested that higher percentage of institutional ownership can lead to a more heavy adverse selection problem, which means the higher institutional ownership, the higher asymmetrical information. There exists several counter arguments.
The counterarguments were explained in more details in the paper of Boone (2014).
The paper suggested that managers have incentives to attract and keep long-term investors, and institutional investors usually accounts for large proportions of total investors in bulk. Institutional investors of a firm are usually small in numbers but high in investment per institution. Therefore, managers have incentives to meet the information criterion of institutional investors. The result of Boone (2014) were in support of this argument, where the firms with higher institutional investors are
often more timely and willingly to disclose company information via forecast reports. Moreover, Boone (2014) also agreed with Healy and Palepu (2001) that institutional investors act as important information intermediaries. This is due to the information acquired from the firm and released to the public by the institutions.
Therefore, it is thus hypothesized that lower institutional ownership causes higher asymmetrical information. Hence, using institutional ownership as a proxy for asymmetrical information, where under pecking order theory, the higher asymmetrical information will lead to diminished dividend announcement effect.
2.5 Empirical Evidence
The empirical evidence of past studies on dividend announcement effect without asymmetrical information will be discussed. In addition, a separate discussion will be carried out for past studies that included asymmetrical information as a variable in the research of dividend announcement effect. This is due to the scarcity of including asymmetrical information as a variable in empirical studies.
2.5.1 Dividend Announcement Effect
Summary of Notable Empirical Findings of Dividend Announcement Effect
Authors Sample Set Results
Aharony and Swary (1980)
Three sets of sample:
increase/decrease/stable dividend announcement in mature market (NYSE)
Increased dividend announcement led to abnormal return, decreased dividend announcement led to negative return, and stable dividend announcement led to normal return.
Asquith and Mullins (1983) and;
Healy and Palepu (2018)
Dividend initiation and termination of mature market (US Major Stock Exchanges)
Dividend initiation led to abnormal return, and dividend termination led to negative return.
Abdullah et al. (2002) Three sets of sample:
increase/decrease/stable dividend announcement in emerging market (KLSE)
Decreased dividend announcement led to negative return, increased and stable dividend announcement have no significant effects.
Various empirical studies have been established to find the relationship between dividend announcement effect and earning situations of firms due to stock market reaction (Asquith and Mullins, 1983; Healy and Palepu, 2018; Abdullah et al., 2002; etc.). The results mainly supported the hypothesis that dividend contains information contents. These studies generally recorded a positive relation between stock returns and dividend announcement significantly. This finding supports the various theories mentioned in above sections of dividend announcement effect. Namely, the theories that state
an increase in dividend announced will increase stock price, and inversely, a decrease in dividend announced will decrease stock price. The empirical research on the price reaction of the market to dividend announcements is performed in settings with rich variations. The variations include separation of increase, decrease and stable dividend announcements, initiation and termination of dividend payouts, and a mixed combination of dividend announcements.
In the study of Aharony and Swary (1980), in an attempt to examine the reaction from market to quarterly dividend announcements, they used 3,399 quarterly dividend announcements as sample, covering 149 industrial companies from 1963 to 1976 on the New York Stock Exchange (NYSE). The firms all meet the data availability requirement: 1) quarterly earnings per share and quarterly cash dividends per share from Compustat, 2) daily return rates from CRSP, 3) dividend announcement date available in annual issues of Moody’s Dividend Record, and 4) announcement date of quarterly earnings per share available through Wall Street Journal. Aharony and Swary divided the sample into increase, decrease, and stable dividends in order to find out dividend announcement effect with sign changes.
They used market models to compute the abnormal return with an event window of [-20; 20]
surrounding the dividend announcement date. Their results indicated a positive abnormal return associating with firms announcing increased dividends, a negative abnormal return associating with firms announcing decreased dividends, and only a normal return for firms with stable dividend announcement, where one day surrounding the dividend announcement date are found most statistically significant. This supports the information content of dividends and dividend announcement effect theories.
Asquith and Mullins (1983) introduced the study of determining the relation between initiation of dividend announcement and stock prices. They believe the best simulation for an unexpected dividend event is dividend initiation. In quest of their research to find the above mentioned relation, Asquith and Mullins adopted a sample size of 168 companies that either, for the first time ever, initiated dividends or have not been paying dividend for the past 10 years and resumed paying. The result of the study suggests that firms experience significant abnormal return after the dividend announcement initiation. Thus, it is hypothesized that dividend announcement initiation is capable of releasing positive information about the firm, which is in accordance with the signaling theory of dividend announcement effect. Healy and Palepu (2018) continued and advanced the work of Asquith and Mullins (1983) by researching the changes in earnings before and after a dividend initiation or termination. They extended the sample selection process by including two sets of samples. The first sample consists of 131 firms that either paid an initial dividend or started paying again after 10 years. The second sample consists of 172 firms that either terminated dividends for the first time or terminated dividend payout after paying consistently for 10 years. The results of the study show an increase in earnings for dividend initiation and a decrease in earnings for dividend termination. Furthermore, the study concluded a positive
correlation between abnormal stock price and dividend initiation, and a negative correlation from dividend termination after the announcement year.
Unlike the study of mature markets that have produced concrete results, studies on emerging markets have shown less dominant results and experienced theoretical variations. Abdullah et al. (2002) studied the dividend announcement effect on Kuala Lumpur stock exchange (KLSE) in order to verify theory validity in emerging markets. The study followed the methodology of Aharony and Swary (1980) and included 187 observations listed on KLSE between the years of 1996 to 1999. The findings diverge from past studies of mature markets, where only the positive relation between an increased dividend announcement and abnormal return were significant. This finding differs slightly from past studies performed in the US and UK. Therefore, this paper will conduct empirical analysis in the mature markets of the US stock exchange with most stability, as emerging markets can post various uncertainties. This is to better carry out the purpose of this paper to identify asymmetrical information in context of dividend announcement effect.
2.5.2 Dividend Announcement Effect with Asymmetrical Information
The past empirical findings on dividend announcement effect incorporating asymmetrical information has fewer studies and less predominant results. The study of directly relation between asymmetrical information and dividend announcement effect is scarce. Even though there are studies on dividend announcement effect, which links dividend policy with abnormal return, and studies on asymmetrical information and dividend policy, there is less studies on the direct relation between abnormal return caused by dividend announcement effects and asymmetrical information.
One of the few studies that carried out researches of how dividend announcement causes abnormal return under asymmetrical information is the study of Mitra and Owers (1995). The study adopted different characteristic on a firm level to account for degree of asymmetrical information. The characteristics covered are firm size, insider ownership of equity in numbers and percentages separately, and number of following analysts to act as proxies for asymmetrical information. Mitra and Owers utilized a sample set of 80 dividend initiations between 1976 and 1987 obtained from CRSP daily database, Moody’s Annual Dividend Record, and Wall Street Journal Index. Established around the proxies, dividend initiations in the sample are further divided into two categories based on the magnitude of asymmetrical information, namely a low and a high level of information asymmetry. The market model was then applied to construct the abnormal earnings surrounding the dividend announcements. The results showed that announcements of dividend initiation leads to high abnormal returns significantly. Furthermore, the results indicated a stronger association between dividend announcement and abnormal return for firms with high asymmetrical information. This supports the
hypothesis made stemming from Signaling Theory, where higher asymmetrical information will lead to enhanced effect of dividend announcements.
3.1 Variable Summary and Data Collection
Summary of Variables
Variable Identifier Description Source
Stock Return Rit Daily stock return of individual stock i at time t
CRSP Master Market Return Rmt Daily return of S&P 500 at time t CRSP Master Dividend
DIVI The dividend initiation date of individual companies
CRSP Stock Event
NUMANAL Number of analysts following the company
IBES Historical Summary Statistic Institutional
INST The institutional ownership percentage of the company
Thomson Reuters Institutional (13f) Corporate
CGI The Corporate governance index of the company
Sustainalytics Historical Weighted Scores
The sample set is split into two sample sets due to data availability. They were:
Sample set 1 with 126 observations including all variables except CGI.
Sample set 2 with 56 observations including all variables.
Each observation in the two sample set correspond to one firm that initiated dividend during the period of 2009 to 2019 actively trading on major US stock exchange platforms (AMEX, NASDAQ, and NYSE). The initial sample included 175 observation each corresponding to a firm that initiated dividends in the period 2009 to 2019. The time period of 11 years through 2009 to 2019 is due to major event correction. This paper aims to test the empirical validity of dividend announcement effect with more recent information under general economic environment. Therefore, due to the major stock market and global economic downturn of the great recession of 2008 and the ongoing COVID-19 pandemic that started in 2020, the sample period is set in between the two major events. Important initial data, such as dividend initiation, number of analysts following the firm, Corporate Governance Index, Institutional Ownership, and daily return of the stock, and daily return of S&P 500, are collected through various data base and will be described. The final sample set will also be presented in forms of a statistic summary table, followed by correlation test results.
Firm dividend initiation (DIVI), defined as the first dividend payout since the firm went public, is collected through the CRSP Stock Event Database. Through the time period of 2009 to 2019, the database reported 175 dividend initiations, each corresponding to one firm and the dividend initiation date. The 175 firms and initiation dates reported are thus, the initial sample firms and time period in question.
From the 175 dividend initiating firms, the number of analysts following each firm (NUMANAL) is collected through the IBES Historical Summary Statistic Database. The database returned data for 147 firms out of the original 175 firms. The number of analysts following the firm is selected base on the closest recorded date prior to event date(dividend initiation date). Due to data availability, the initial sample size is narrowed down to 147 dividend initiation.
By researching through Thomson Reuters Institutional Managers (13f) Holdings Database, the database returned the institutional ownership(INST) percentages for all 147 firms remaining in the sample set. The institutional ownership is selected based on the closest recorded date prior to the event date.
The daily returns of stocks from each firm in the sample set are collected through the CRSP Master database. The daily returns of each firms must contain data at least for 15 days before and after the event date (dividend initiation date). This is to construct the Cumulative Average Abnormal Return (CAAR) for the event period. After accounting for data availability of the 147 stocks in the sample set, only 126 remains with full data availability. Together with S&P 500 daily return acquired through CRSP Master Database, the daily abnormal return(AR), average abnormal return(AAR), and cumulative average abnormal return(CAAR) can be computed.
Lastly, this paper followed the idea of Elbadry et al. (2015) to adopt a composite index of corporate governance to proxy for asymmetrical information. The ISS ESG Governance Quality Score (GQS) is used as a representative of the corporate governance index. This is due to the vast aspect of corporate governance included. However, due to limited availability of data, only 56 out of the 126 observations in the sample set are returned. The data availability for Corporate Governance Index(CGI) is collected through the Sustainalytics Historical Weighted Scores Database.
Therefore, in order to not limit the scale of the empirical study carried out, the sample set is split into two samples. The first sample includes the 126 observations with all necessary data available for all variables except corporate governance index. The second sample includes the 56 observations including all data for all variables.
3.2 Data Analysis
Table 1: Descriptive Statistics of independent Variables
Descriptive statistics of all available variables under the two sample sets respectively. The mean, standard deviation (Std Dev.), minimum (Min), and maximum (Max) value of the variables are presented in the table respectively. Unavailable data are denoted as a dash in the column in the respective unavailable sample set.
N = 126
Variable Mean Std Dev. Min Max.
CAR[-1;1] 0.0206 0.1421 -0.094 0.1655
NUMANAL 5.2857 5.8005 1 33
INST 0.4869 0.3464 0 1.2232
CGI - - - -
CAR[-1;1] 0.017 0.0484 -0.0618 0.1655
NUMANAL 7.7857 7.4314 1 33
INST 0.6664 0.3168 0 1.2232
CGI 54.6814 9.7116 41 86
All data presented in both sample sets showed a positive mean as presented in table 1, which was reasonable due to the data characteristics. NUMANAL takes the form of an integer and cannot be lower than 0 (no analysts following the firm). INST takes the form of a percentage with in the range of 0% to 100%. CGI is a score in range of 0 to 100 base on definition. The positive mean of CAR of event window [-1;1] showed positive support towards the empirical research of this paper indicating a positive abnormal return around the event date.
The standard deviations of most CAR and INST in both sample sets as presented in table 1 were between 0 and 0.5 which indicated a fairly small spread of the data. The NUMANAL and CGI have a high standard deviation in both sample, due to their discontinuous value measurement. In a relatively compact sample size, variables with discontinuous measurements could be subjected to a higher standard deviation.
The minimum and maximum value of the variables as presented in table 1 provided fewer insights. However, the maximum value of CAR is exceptionally higher than 0 of zero cumulative abnormal return at 0.1655 for both sample size, and the minimum value of -0.094 and -0.0618 for the two sample size respectively, are significantly closer to 0. This is another support dividend announcement effect exists. For, INST, even though it is described as a percentage between 0 and 1, the maximum value recorded is 1.2232 for both samples. It is economic unrealistic to have an ownership percentage over 1 (100%). However, this can be explained through a short selling situation where in a point in time, there is more than one party of investors are claiming ownership of the shares of a company. Although, the data is technically impossible, it can still provide an economic interpretation
of institutional ownership. This is because for an institutional ownership percentage to be above 1 in short selling situations, the original percentage should already be high. Therefore, this paper accept the institutional ownership value that is larger than 1, as it still provides information as intended for asymmetrical information.
Table 2: Correlation of dependent variables
The correlation between each independent variable is presented for both sample sets. Unavailable data are denoted as a dash in the column in the respective unavailable sample set.
NUMANAL INST NUMANAL INST CGI
NUMANAL 1 1
INST 0.3822 1 0.1581 1
CGI - - 0.4221 -0.2950 1
The correlation of all dependent variables, namely number of analysis (NUMANAL), institutional ownership (INST), and corporate governance index (CGI) is presented in table 2. The correlation between NUMANAL and CGI appears to be high at 0.4221 in the second sample set. This is due to the fact that number of analysts is one of the considerations included in composition of the CGI. Nevertheless, number of analysts is considered a strong proxy for asymmetrical information as suggested by several literatures (Bhushan, 1989; Brennan and Hughes,1991). CGI also captures information beyond number of analysts due to it well-rounded composition. Furthermore, the correlation between CGI and INST is negative at -0.2950 as presented in table 4, this is also reasonable as the grading composition of CGI believe higher institutional ownership has bad influence for corporate governance, which will lead to a lower CGI rating.
3.3 Hypothesis Summary
The summary of hypothesis is presented. Hypothesis 1 is to test for dividend announcement effect by using dividend initiation date and daily stock return, to determine the level of abnormal return caused by dividend initiations. Hypothesis 2 is a general hypothesis to investigate the effects of asymmetrical information to dividend announcement effect. Hypothesis 2 is divided into three sub hypothesis, namely hypothesis 2.1, 2.2, and 2.3, each for one information asymmetry proxy, to test for their effect on abnormal return. Furthermore, the relation between asymmetrical information and the three proxies selected is inversely correlated as suggested by the past literatures in section 2.4. The higher number of analysts, institutional ownership, and corporate governance index will indicate a lower asymmetrical information.
Summary of Hypothesis
Hypothesis Expected Result Test
1 There is Cumulative Abnormal Return (CAR) by dividend initiation
CAR exists due to dividend initiation t-test
2 The dividend announcement effect is enhance under asymmetrical
Abnormal return increases under higher asymmetrical information
2.1 Number of analysts have no effect on abnormal return
Abnormal return decrease when number of analysts increases
2.2 Institutional ownership has no effect on abnormal return
Abnormal return decreases when institutional ownership increases
2.3 Corporate Governance Index (CGI) has no effect on abnormal return
Abnormal return decreases when CGI increase t-test
4.1 Event Study Specifications
In a classical fashion, market reaction is measured by implementing event study methodology.
The fundamental data are closing prices of stocks. As suggested by Mackinlay (1997), the event study methodology requires to set an event window and an estimation window under a specification. As presented in Graph 1, the even window [t1;t2] has the following specifications where t1 needs to be low enough and t2 needs to be high enough relative to the event date (t0). More specifically, the time period [t1;t2] includes the event date t0 and have a spam long enough to capture the effect of the event. The estimation window [T1;T2] has the following specifications where T2 should be low enough to avoid influence of the event at t0 and T1 should be low enough to capture an significant amount of information for estimation.
In this paper, an estimation window of 100 days was selected from the suggestion from Cox and Peterson (1994), where the study argued the sufficiency of a 100-days estimation window which
can avoid including other unexpected events in the estimation window. Furthermore the last date of the estimation window T2 was set to be 5 days before the start of the event window, where T2=t1-5 to avoid the influence of the event.
4.2 Construction of the Dividend Announcement Model
The event window selected for the dividend announcement effect study was [-5;5], which is 5 days prior and post of the dividend announcement date. The event window is selected based on the argument made by Agbetonyo et al. (2018), to study not only the immediate reactions of the dividend announcement effect of 1 day prior and post of the initiation, but to also examine how investors react to the dividend initiation after a longer period of time, namely 5 days prior and post of the dividend initiation. Therefore, following the event study methodology construction mentioned in section 4.1, the estimation period was [-110;-10]. This insured to exclude the event effect on this estimation period and provided a long enough time frame for estimation for return parameters.
The Market Model (CAPM) was used to construct for the expected daily return E(Rit) for the stock i on day t was constructed as follows:
𝐸(𝑅𝑖𝑡) = 𝛼𝑖+ 𝛽𝑖𝑅𝑚𝑡+ 𝜀𝑖𝑡 (1)
Where 𝛼𝑖 and 𝛽𝑖 were ordinary least square (OLS) estimation from the estimation period, and 𝑅𝑚𝑡 is the daily market return on day t. The daily market return was proxied by the S&P 500 historical return. The next step was to construct for the abnormal return for each stock i on day t, it was calculated as the difference between the actual return of stock i on day t and the expected return E(Rit) estimated from the market model:
𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− E(𝑅𝑖𝑡) (2)
However, since the paper aims to research the dividend announcement effect, following the event study methodology, the abnormal return for the entire event window needs to be computed. This is to detect whether a significant difference is present, when comparing abnormal return within the window and in other periods. In times, even when abnormal returns are detected, it must be checked and proved that the significant increase in returns are not caused by a biased time series or an unexpected event other than the event being studied. Following Suwanna (2012), identically and independently distributed is a basic assumption for daily abnormal returns. Furthermore, it is also assumed that, over a long period, stock prices will gradually move towards the expectation value of the mean value.
Therefore, the cumulative abnormal return (CAR) is computed as follows:
𝐶𝐴𝑅𝑖𝑡 = ∑ 𝐴𝑅𝑖𝑡
The cumulative abnormal return (equation 3) can be tested for significance with a simple t-test for the sample observation N for each day t in the event window [-5;5]. This study test against the null hypothesis of:
𝜇𝐶𝐴𝑅= 0 or there is no cumulative abnormal return Hypothesis(1) The abnormal return and the cumulative abnormal return of the N observation (firms) in the sample were then averaged for each time t across the event period [-5;5]. This is to test for the significance of each average abnormal return (AAR) and cumulative average abnormal return (CAAR).
4.3 Cross-Sectional Analysis for Asymmetrical Information
The market friction caused by the asymmetrical information gives incentive to study the second goal of this paper, the effect of asymmetrical information on dividend announcement effect. The model for testing asymmetrical information was a multi-factor linear model as shown in equation 4 and equation 5. The event window is selected to be [-1;1] that captures the most effect of dividend announcement. An OLS regression was than performed to determine the asymmetrical information effect on the significance of abnormal returns. Building upon the market model and the cumulative abnormal return equation, the multi-factor linear model took the following form:
𝐶𝐴𝑅[−1;1]= 𝛽0+ 𝛽1NUMANAL + 𝛽2𝐼𝑁𝑆𝑇 + 𝜀𝑖𝑡 (4)
𝐶𝐴𝑅[−1;1] = 𝛽0+ 𝛽1NUMANAL + 𝛽2𝐼𝑁𝑆𝑇 + 𝛽3𝐶𝐺𝐼 + 𝜀𝑖𝑡 (5) where equation 4 denotes the regression for the first sample of 126 observations excluding corporate governance index effect, and equation 5 denotes the regression for the second sample of 56 observations including all variables but with less observations. 𝐶𝐴𝑅[−1;1] was the 3-days CAR dependent variable, 𝛽0 was the regression constant term (intercept), 𝛽1,2,3 were the coefficient of each individual independent variables, and 𝜀𝑖𝑡 was the error term of the regression.
A t-test was then performed on both of the regression models to test against the following three null sub-hypothesis:
𝑛𝑜 𝑒𝑓𝑓𝑒𝑐𝑡 𝑓𝑟𝑜𝑚 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑎𝑛𝑎𝑙𝑦𝑠𝑖𝑠 𝑜𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 ℎ𝑦𝑝𝑜𝑡ℎ𝑒𝑠𝑖𝑠(2.1) 𝛽2= 0
𝑛𝑜 𝑒𝑓𝑓𝑒𝑐𝑡 𝑓𝑟𝑜𝑚 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑎𝑙 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 𝑜𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 ℎ𝑦𝑝𝑜𝑡ℎ𝑒𝑠𝑖𝑠(2.2) 𝛽3= 0
𝑛𝑜 𝑒𝑓𝑓𝑒𝑐𝑡 𝑓𝑟𝑜𝑚 𝑐𝑜𝑟𝑝𝑜𝑟𝑎𝑡𝑒 𝑔𝑜𝑣𝑒𝑟𝑛𝑎𝑛𝑐𝑒 𝑖𝑛𝑑𝑒𝑥 𝑜𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 ℎ𝑦𝑝𝑜𝑡ℎ𝑒𝑠𝑖𝑠(2.3)
Sub-hypothesis 2.1, 2.2, and 2.3 tests individually the significance of impact of each asymmetrical information proxies on abnormal returns of dividend initiation announcements. The combined results and analysis of the three sub-hypothesis can thus be answering the second main hypothesis of this paper:
𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑 𝑎𝑛𝑛𝑜𝑢𝑛𝑐𝑒𝑚𝑒𝑛𝑡 𝑒𝑓𝑓𝑒𝑐𝑡 𝑖𝑠 𝑒𝑛ℎ𝑎𝑛𝑐𝑒𝑑
𝑢𝑛𝑑𝑒𝑟 ℎ𝑖𝑔ℎ𝑒𝑟 𝑎𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐𝑎𝑙 𝑖𝑛𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛 ℎ𝑦𝑝𝑜𝑡ℎ𝑠𝑖𝑠(2)
5.1 Event Study Results
Event study results of dividend announcement effect with average abnormal return (AAR) and cumulative average abnormal return (CAAR) at daily frequency and the respective t-value, from 5 days prior and post to the dividend initiation announcement.
Date in Event Time AAR t-test CAAR t-test
-5 -0.0612% -0.1439 -0.0612% -0.0925
-4 -0.7233% -1.7001* -0.7845% -1.1847
-3 0.0904% 0.2125 -0.6329% -0.9557
-2 -0.0249% -0.0585 0.0655% 0.0989
-1 0.0017% 0.0039 -0.0232% -0.0351
0 0.3204% 0.7532 0.3221% 0.4864
1 1.0314% 2.4243** 1.3518% 2.0413**
2 0.1716% 0.4043 1.2030% 1.8166*
3 0.3193% 0.7504 0.4909% 0.7413
4 0.1282% 0.3014 0.4475% 0.6757
5 -0.2336% -0.5491 -0.1054% -0.1592
***, **, and * indicate statistical significance at the 1%, 5%, and 10% respectively
The average abnormal return (AAR) and the cumulative average abnormal return (CAAR) at a daily frequency for 5 days prior and post the event date is reported in table 3. The t-test results are also presented in the table for each AAR and CAAR respectively. The t-test for both AAR and CAAR at event date t=1 are statistically significant at 0.05 level. The null hypothesis 1 of there is no cumulative abnormal return can thus be rejected at event date t=1 at 0.05 significance level. Furthermore, the null hypothesis can also be rejected at 0.1 significance level at event date t=2 due to the statistical significance of CAAR at event date t=2. Moreover, a negative abnormal return at event date t= -4 was also found statistically significant at the 0.1 significance level.
The first hypothesis of this paper was therefore proven, such that dividend initiations have an significantly high positive effect on abnormal returns, which means dividend announcement effect exists. The statistical significance for the CAAR at event date t=1is important to the economic interpretation of the result. Even though the AAR showed significance statistically at event date t=1, it is not sufficient to conclude that dividend initiation caused the effect. As mentioned in the methodology section, the construct of CAR is to avoid an unexpected random increase in abnormal return. By constructing and interpreting the CAAR at event date t=1, it can be concluded that the significance of CAAR proves the event (dividend initiation) caused an abnormal return across the event window. Thus, hypothesis 1 is proved, such that dividend announcement effect exist. Moreover, the significance found
at 0.1 significance level for CAAR at event date t=2 shows that there is effect spillover from the announcement date t=0 and one day after the announcement date t=1. This is consistent with the view of information flow such that information takes time to be collected and interpreted by investors, and thus investors have different reaction time.
In conclusion, the event study results of dividend announcement effect is consistent with the results of prior studies (Mitra and Owers, 1995; Miller and Rock, 1985), that dividend initiation announcements will result in significantly high abnormal returns. This result is also consistent with the various theory of dividend announcement effect, such as the Signaling theory by Miller and Rock (1985) and the Free Cash Flow Theory by Jensen (1986).
5.2 Cross-sectional Analysis Results
Cross-sectional analysis for CAR regressions for asymmetrical information effects under daily sampling.
This table presents the cross-sectional regression of CAR[-1;1] using asymmetrical information proxies:
The regression used associated with results presented in column 1 is:
𝐶𝐴𝑅[−1;1]= 𝛽0+ 𝛽1NUMANAL + 𝛽2𝐼𝑁𝑆𝑇 + 𝜀𝑖𝑡 The regression used associated with results presented in column 2 is:
𝐶𝐴𝑅[−1;1]= 𝛽0+ 𝛽1NUMANAL + 𝛽2𝐼𝑁𝑆𝑇 + 𝛽3𝐶𝐺𝐼 + 𝜀𝑖𝑡
Where CAR[-1;1] is the cumulative abnormal return for each firm in event window [-1;1]. The results presented consists of two sample sets with column 1 represents the sample set of 126 observation excluding variable CGI, and column 2 represents the sample set of 56 observations with all variables included. The slope 𝛽 is presented for each variable under two separate sample set, followed by the standard deviation for each variables in brackets. The R2 is also presented for each model to show the proportion of variance explained by the model inputs.
Variables (1) (2)
NUMANAL 0.0007 0.0019**
INST 0.0045 -0.0114
CGI - -0.0009
Constant 0.0044 0.0614
Observations 126 56
R2 0.010 0.067
***, **, and * indicate statistical significance at the 1%, 5%, and 10% respectively, based on standard robust error robust for the presence of heteroscedasticity and serial correlation. Results of unavailable data of the forecasting variable in the respective sample period are presented with a dash.
Table 4 presents the cross-sectional analysis results to determine the effect of asymmetrical information on abnormal return, and hence investigate the second hypothesis of this paper, that dividend announcement effect is enhance under higher asymmetrical information. The model specification includes the dependent variable CAR at event window [-1;1] which the selection is based on the result of the event study on dividend announcement effect carried out in the above sections where CAR[-1;1]
covers the event date and had highest significant abnormal return. The regression also included three dependent variables which is the three asymmetrical information proxies selected in this paper, namely, number of analysts following the firm (NUMANAL), institutional ownership of the firm (INST), and the corporate governance index of the firm (CGI). The estimated slope 𝛽̂ of each independent variable can be interpreted as the impact per unit of the variables on the CAR. Illustrated as an increase per number of analysts following the firm will result in a 0.0007 increase in CAR of the firm in the event window [-1;1].
The most results in both sample sets are not able to reject the null hypothesis of dividend announcement effect is enhance under higher asymmetrical information. The results showed merely no significant results except in the second sample set with 56 observations that included all variables, which returned a significant result for NUMANAL with 0.0019 at 0.05 significant level. This means that an additional analysts following the firm will result in a 0.0019 increase in CAR of the firm in the event window [-1;1]. Therefore, the null sub-hypothesis 2.1 of number of analysts have no effect on abnormal earning is rejected at a 0.05 significant level in the second sample set. Thus, proved the sub- hypothesis 2.1, such that number of analysts have an effect on firm’s abnormal return.
The statistically insignificant result can still be interpreted for a general relation between the dependent variable and the independent variable. The results of the cross-sectional analysis for both sample indicated a positive relation between number of analysis and the CAR of the firm. As the number of analyst is a proxy for asymmetrical information, this showed a possible relation such that firms with higher asymmetrical information receives less abnormal return through dividend announcement effect.
Thus if taking number of analysts as the asymmetrical information proxy, the results contradicted our second hypothesis, such that dividend announcement effect is enhanced under higher asymmetrical information. This is consistent with past literatures of Mitra and Owers (1995).
However, the results of the second sample set with 56 observations as presented in column 2 in table 4 showed a general evidence for hypothesis 2. The coefficient of INST and CGI is negative, which suggested a negative relation of CAR with institutional ownership and corporate governance index. As institutional ownership and corporate governance index is reversely correlated to asymmetrical information, the results showed that CAR is higher under higher asymmetrical information. This can
then be translated to that dividend announcement effect is enhanced under higher asymmetrical information.
The results of this cross-sectional study on the effect of asymmetrical information on dividend announcement effect cannot provide statistically significant proof to hypothesis 2, such that dividend announcement effect is enhanced under higher asymmetrical information. When comparing past literatures of Mitra and Owers (1995) that significantly provided proof for this hypothesis, the limitations of the cross-sectional study lies in the limited observations due to data availability. Even though the general methodology of this paper resembles the methodology in the paper by Mitra and Owers (1995), the scale of the empirical study is relatively smaller. This may contribute to the insignificant results returned in this paper and therefore compromise the internal validity of this study.
However, disregarding the effect of a smaller sample, the asymmetrical information effect on dividend announcement effect may be subjected to changes in different times. The past literatures (Bhattacharya, 1979; Miller and Rock, 1995; Mitra and Owers, 1995) on this topic that showed significant results supporting that dividend announcement effect is enhanced under higher asymmetrical information is performed on older sample periods, where the most recent sample period being 1976- 1987 from Mitra and Owers (1995). The information environment has changed and improved since 2002 (Abraham, 2008). The general information context of the 1970s and 1980s is drastically different from the sample period of this study, from 2009 to 2019.
The information flow of the 21st century is largely improved such that classical proxies of asymmetrical information are not efficient anymore. Moreover, the general information environment between firms and investors are improved to be more symmetrical under more and more specific and well-rounded market legislation, such as the continuous update of the International Financial Reporting Standards (IFRS). Therefore the effect of asymmetrical information on dividend announcement effect is diminished by the general economic and market environment.
This paper aims to examine the effect of dividend announcement on the return of firms and investigate the impact of asymmetrical information on the dividend announcement effect. The past studies suggested that dividend announcement will result in a significantly high abnormal return for firms, and that such effect is enhanced under higher asymmetrical information. This paper revised this topic by implementing a more recent sample set from 2009 to 2019 on US listed firms to test the consistency of dividend announcement effect against different time periods. Moreover, this paper revised the measurement of asymmetrical information by selecting various proxies of information asymmetry across various past studies, in order to find the effect of these proxied on dividend announcement effect.
The results of this paper showed that dividend announcement effect exists, such that dividend initiation announcements will result in highly significant abnormal return. Furthermore, the abnormal return is most significant on the event date and on the day after. However, when considering the effect of asymmetrical information, such that higher asymmetrical information will enhance the dividend announcement effect, is not significant. There is no significant empirical evidence that suggested dividend announcement of firms with higher asymmetrical information will experience higher abnormal returns. This may come directly from the sample size limitation, such that dividend initiation selected in the period 2009-2019 is of a smaller sample size due to data availability. However, the insignificance of asymmetrical information on dividend announcement effect can also be explained through the drastically improvements in general economic and market information environment. The degree and effect of asymmetrical information is diminishing along technology improvements on information access and increasing market legislations, such as the update on financial reporting standards.
For further research purposes, improvements can be made on sample size and variable selection, such as enlarging the sample set and variables, by reducing limitations of data availability.
Moreover, the information asymmetry measurements can be studied from the aspect of technology availability between regions and investors over classical financial variables, in order to study the effect of technological advancements on asymmetrical information.
In consistency with past studies, this paper found that dividend announcement effect exists such that dividend initiation will result in an significantly high abnormal return for firms. In contrast with past study, the effect of asymmetrical information is not detected in this paper from a recent sample period of 2009-2019.
Abdullah, N.-A.H, Abdul Rashid, R., & Ibrahim, Y. (2002). The Effect of Dividend Announcements on Stock Returns for Companies Listed on the Main Board of the Kuala Lumpur Stock Exchange.
Malaysian Management Journal. Pp. 85-98.
Abraham, R. 2008. Mobile phones and economic development: Evidence from the fishing industry in India. In MIT Press 2008, Volume 4, Number 1, Fall 2007, 5-17
Agbetonyo, Y.S., Fromont, E., & Viviani, J. (2018). Asymmetric Responses to Dividend Announcements: A Case for Ambiguity. Revue De L'ofce, 77-104.
Aharony, J., & Swary, I. (1980). Quarterly Dividend and Earnings Announcements and Stockholders' Returns: An Empirical Analysis. Journal of Finance, 35, 1-12.
Asquith, P., & Mullins, D. (1983). The Impact of Initiating Dividend Payments on Shareholders' Wealth. The Journal of Business, 56, 77-96.
Bharath, S.T., & Dittmar, A.K. (2006). To Be or Not to Be (Public). Capital Markets: Market Efficiency.
Bhattacharya, S. (1979). Imperfect Information, Dividend Policy, and "The Bird in the Hand" Fallacy. The Bell Journal of Economics, 10(1), 259-270.
Bhattacharya, S. (1980). Nondissipative Signaling Structures and Dividend Policy. The Quarterly Journal of Economics, 95(1), 1-24.
Bhushan, R. (1989). Firm characteristics and analyst following. Journal of Accounting and Economics, 11, 255-274.
Boone, A.L. (2014). The effect of institutional ownership on firm transparency and information production.
Cai, C.X., Keasey, K., & Short, H. (2006). Corporate Governance and Information Efficiency in Security Markets. European Finance.
Chen, E., & Nowland, J. (2010). Optimal Board Monitoring in Family-Owned Companies: Evidence from Asia. Corporate Finance: Governance.
Cox, D., & Peterson, D.R. (1994). Stock Returns following Large One‐Day Declines: Evidence on Short‐Term Reversals and Longer‐Term Performance. Journal of Finance, 49, 255-267.
D'Mello, R., & Ferris, S. (2000). The Information Effects of Analyst Activity at the Announcement of New Equity Issues. Financial Management, 29(1), 78-95.
Easley, D., Kiefer, N., O'Hara, M., & Paperman, J. (1996). Liquidity, Information, and Infrequently Traded Stocks. The Journal of Finance, 51(4), 1405-1436.
Easterbrook, F. (1984). Two Agency-Cost Explanations of Dividends. The American Economic Review, 74(4), 650-659. Retrieved June 29, 2021.
Elbadry, A., Gounopoulos, D., & Skinner, F.S. (2015). Governance Quality and Information Asymmetry. Emerging Markets Economics: Firm Behavior & Microeconomic Issues eJournal.
Healy, P., & Palepu, K. (2001). Information Asymmetry, Corporate Disclosure and the Capital Markets: A Review of the Empirical Disclosure Literature. Journal of Accounting and Economics, 31, 405-440.
Healy, P., & Palepu, K. (2018). Earnings Information Conveyed by Dividend Initiations and Omissions.
Heflin, F., & Shaw, K. (2000). Blockholder Ownership and Market Liquidity. The Journal of Financial and Quantitative Analysis, 35(4), 621-633.
Holm, C., & Schøler, F. (2010). Reduction of Asymmetric Information Through Corporate Governance Mechanisms – The Importance of Ownership Dispersion and Exposure toward the International Capital Market. Corporate Governance: An International Review, 18, 32-47.
Jensen, M. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. The American Economic Review, 76(2), 323-329.
John, K., & Williams, J. (1985). Dividends, Dilution, and Taxes: A Signalling Equilibrium. The Journal of Finance, 40(4), 1053-1070.
Lang, M., & Lundholm, R. (1996). Corporate Disclosure Policy and Analyst Behavior. The Accounting Review, 71(4), 467-492.
Llorente, G., Michaely, R., Saar, G., & Wang, J. (2002). Dynamic Volume-Return Relation of Individual Stocks. The Review of Financial Studies, 15(4), 1005-1047.
Michael J. Brennan, & Patricia J. Hughes. (1991). Stock Prices and the Supply of Information. The Journal of Finance, 46(5), 1665-1691.