• No results found

Stock price reactions to dividend changes in the Chinese and the US utility industries

N/A
N/A
Protected

Academic year: 2021

Share "Stock price reactions to dividend changes in the Chinese and the US utility industries"

Copied!
35
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

This paper examines stock price reactions to cash dividend announcements in the period of 2010-2014 across the Chinese and the US utility industries. As the level of asymmetric information in China is higher than in the US, this paper explores whether price reactions to dividends are larger in China than in the US. By conducting a general event study, this paper finds a weakly positive relationship between dividend changes and stock price reactions for all utility firms. However, the results do not provide any significant difference regarding the price reactions between the Chinese and the US utility industries.

JEL classification: G15; G32; G35

Key words: dividends; stock prices; payout policy; information asymmetry

Student name: Ranxin Duan Student number: S2014262 Program: IFM

Supervisor: Dr. J.H. von Eije Accessor: Dr. W. Westerman

Word count: 11,636 Date: 19-06-2015

Stock price reactions to dividend changes in the Chinese and the US utility industries

Abstract

(2)

2 / 35

1. Introduction

A major goal of a company is to maximize its shareholder’s wealth. In this paper I want to explore whether dividend payments do generate shareholder value. For around 60 years, since Lintner (1956), dividend change announcements as well as their stock price reactions have been investigated. Miller and Modigliani (1961) indicate that in a perfect market, dividend payments cannot add value to the shareholders. This means that if one assumes that their

“irrelevance theory” is correct, dividends can only contribute to shareholders’ value if the market is not perfect. In this paper, the focus will lie with the imperfection of asymmetric information, which in this case implies that managers know more about the prospects of the firm than the investors (Myers and Majluf, 1984). In particular, I am interested to know if changes in dividend policy made by the managers will influence the share prices and that may be the case if the dividend decision of mangers comes as a surprise to investors. China and the United States (US) are leading representatives of developing and developed countries and it is likely that the level of asymmetric information is higher in China than in the US (La Porta et al., 1997). Therefore, I compare the effect of dividends on stock prices between China and the US. More specifically, as the utility industry has been seldom investigated due to its regulated characteristics, I aim to test the effect of (cash) dividend payments on stock prices in the utility industry and to compare the reactions of the Chinese and the US utility industries.

Based on the dividend signaling effect (Bhattacharya, 1979; John and Williams, 1985;

Miller and Rock, 1985) and free cash flow hypothesis (Jensen, 1986), prior researches

concluded a positive effect of dividend increases announcements and a negative effect of

dividend decreases announcements (Petit, 1972; Lang and Litzenberger, 1989; Grullon et al.,

2002, 2005). Recent studies (Balachandran, Faff and Nguyen, 2004; Kim and Gu, 2009; Ali

and Chowdhury, 2010; Aamir and Shah, 2011; Sare, Pearl-Kumah and Salakpi, 2014)

investigated the stock price reactions to dividends across industries. However, only a few of

them focused on the utility industry where firms are regulated and generally promulgate high

dividends (Miller, 1986; Chatfield & Sisneros, 1989; Hansen, Kumar and Shome, 1994). As a

regulated industry, the Chinese and the US utility industries do have different characteristics

and also suffer from different levels of asymmetric information. Therefore it is interesting to

figure out whether the stock price reactions to dividend changes in the utility industry exist (in

(3)

3 / 35

line with findings in other industries), and if so, whether the stock price reactions differ between the two countries.

Using a sample of 571 cash dividend announcements across the Chinese and the US utility industries during the period 2010-2014, univariate tests and cross-sectional regressions are conducted to examine the effect of dividends changes on stock prices in this paper. The results of univariate tests indicate a positive effect of dividend (increase) announcements, but the negative influence of dividend decrease announcements can hardly be found. Regression results further confirm a weakly positive relationship between dividend changes and stock price changes using a relatively longer event window period (in the utility industry), if one assumes the size effect of dividend changes on stock prices in China is similar with the effect in the US. However, the positive relationship is not supported if one distinguishes the dividend effect in China from the effect in the US. In terms of the effect of cash dividend announcement between the Chinese and the US utility industries, univariate test results show significant and larger cumulative abnormal returns in China than in the US across all announcements. In contrast, the regression results which control for the size of dividend changes, slope dummy, firm size, leverage, industry and earning changes do not show a significant difference of dividend effect between the Chinese utility industry and the US utility industry. Therefore the larger dividend effect in the Chinese utility industry is not supported if controlled for these variables.

The contributions of this paper can be found in at least four areas. Firstly, this paper extends the existing literature by addressing the effect of dividend announcements on stock prices in the utility industry. Secondly, it compares the effect of the Chinese and the US utility industries with recent data. Thirdly, it provides insights for international investors who want to invest in the US and Chinese utility industries. Fourthly, this paper brings additional knowledge on changes in dividend policy to managers who work for the Chinese or the US utility industry, when they intend to change their dividend policies.

The rest of the paper is organized as follows. Section 2 presents the literature review of

dividend policy and the difference of asymmetric information between China and the US,

while section 3 shows information on data collection, data, variables measurements and

(4)

4 / 35

research methods. Section 4 analyzes and discusses the main results of this research. In the end, section 6 concludes this paper.

2. Literature review

2.1. Dividend policy in China and in US

A different background between China and the US such as stock market and government control may contribute to different dividend policies (Mei and Yang, 2013). In order to better understand the dividend policy made by managers in the two countries, dividend policy in China and the US will be discussed separately in this sub-section.

2.1.1 Dividend policy in China

Different from some other developed stock markets such as the US, the Chinese stock market (excluding Hong Kong and Maucao) has a short history of less than 25 years.

Establishments of Shanghai Stock Exchange in December 1990 and Shenzhen Stock Exchange in July 1991 generated the Chinese stock markets. This enabled Chinese companies to publicly list on the exchange and it gives the investors the opportunities to trade in the stocks (Chen, Firth and Gao, 2002). However, not all shares are tradable in China. According to the previous literature (Gao and Kling, 2008; Chen, Jian and Xu, 2009), non-tradable shares count for approximately 60% of the total numbers of shares in China. With the high proportion of non-tradable shares, the stock market liquidity in China is low, which may increase the possibility of paying dividend policy to enhance the market activeness and investors’ enthusiasm (Gao and Kling, 2008; Chen et al., 2009).

Chinese dividend payments exist in two forms: stock dividends and cash dividends. A company may choose to pay stock dividend only, pay cash dividend only, pay both cash dividend and stock dividend, and do not pay any dividend at a certain time (He et al., 2009).

In China, the dividend policy in listed companies gives the impression of randomness and blindness and it lacks continuity and stability (Mei and Yang, 2013). For one time period, companies may choose not to pay dividends and in another time period companies may sudden issue both a stock dividend and a cash dividend (Mei and Yang, 2013).

Moreover, it is not a big secret that Chinese listed firms are heavily controlled by the State.

(5)

5 / 35

Xu, Wang and Xin (2010) point out that two-thirds of Chinese listed companies are state-controlled, either directly or indirectly. Under such circumstances of state-control, government with different views than the individual investors, are likely to focus more on state-owned enterprises’ value-accumulation instead of paying cash dividend immediately (Mei and Yang, 2013). Hence, the operators of such enterprises will take the views of the government rather than the views of normal individual investors in decision making. Besides, the growth of firms also influences the dividend policy. In China, a large number of firms are in their growth period, therefore their resources are limited and they need to accumulate cash for taking on good investments in the future (Mei and Yang, 2013).

In terms of dividend ratios, the average dividend yield for Chinese listed firms from 1990

1

to 2010 demonstrated a value of only 0.55%, which is significantly lower than the average rate of 2.19% in the US (Mei and Yang, 2013). However, the average dividend yield for Chinese listed firms shows a substantial increase after 1990s, with an average rate reaching to 1.5% approximately during the period 1998 to 2010 (Jiang and Kim,2014). The dividend payout ratio for Chinese listed firms decreases during the period 1998-2010. With an average payout ratio of 69% in 1998 and a low ratio of 37% in 2010, the average rate during the period is 51% (Jiang and Kim, 2014). In the utility industry, dividend yield ratio fluctuates from 1% to roughly 4% with an average rate of 1.97% during the period 1996-2008 (Fang and Xu, 2008). Compared with the average dividend yield of 1.27% for other industries, the average dividend yield in the utility industry is higher and has also a high volatility.

2.1.2. Dividend policy in the US

Two major stock exchanges in the US are the New York Stock Exchange (NYSE) and the Nasdaq Stock Exchange (NASDAQ). Established in 1792, NYSE is the oldest and largest stock exchange in the US market with a history of almost 200 years and it contains 1867 listed firms up to 2014. The NASDAQ, operating since 1971, now is the second largest stock exchange in the US and is characterized by its high number of technological firms.

There are several characteristics of dividend policy of the US firms. First, dividend payout is well regulated in the US (Rozeff, 1982). Three general ways of dividend payments are cash

1 Chinese stock market starts in 1990s

(6)

6 / 35

dividend, stock repurchase and stock dividend. The dividend payment for the majority of listed firms are rather frequent, namely on a quarterly basis (Mei and Yang, 2013).

Second, shareholder ownership in the US provides more freedom. According to Mei and Yang (2013), the proportion of shares of listed firms for individuals, corporations and governments in the US are 53.5%, 34% and 12.5% respectively. Therefore dividend policies of firms are generally not set by the government and the primary goal of the management is to satisfy individual benefits. Consequently, managements are more likely to use cash dividend policy to satisfy shareholders and to increase stock prices.

Dividend ratios can bring in the most upfront views of dividend policy of a country. Mei and Yang (2013) stated that the average dividend yield ratio for the US from 1970 to 2010 is 1.97%. This ratio is slightly lower than the average rate in UK (2.19%) at the same period, but it is significantly higher than that in other countries such as Japan and Germany with an average dividend yield ratio of 1.74% and 1.11% respectively. In the US, the average dividend payout ratio was kept at a relatively high level in 1960s (56%), however, the ratio fell heavily afterward with an average rate of 34% from 1998 to 2012 (Damodaran Online, 2015).

In terms of dividend ratio in the utility industry, Morgan (2011) performed a research of S&P 500 firms including 34 utility firms up to 2010. According to the research, the average dividend payout ratio and dividend yield ratio in 2010 for the utility industry are 56% and 4.5 %, which are much higher percentages than for the overall S&P 500 with 23% and 1.2%

respectively. Another recent research (Simshauser and Catt, 2012) which compared the dividend yield ratio of utility industry between US, EU and Australia, figured out the average dividend yield ratio of the US utility industry (3.2%) is lower than average dividend yield of EU and Australia with 5% and 4.3% respectively. However, compared with the Chinese utility industry with an average dividend yield of 1.97%, the dividend yield in the US is still higher.

2.2. Stock price reactions to dividends in the utility industry

Recent studies, examining the impact of dividend announcements on stock prices, take a

focus on specific industries (Balachandran et al., 2004; Kim and Gu, 2009; Ali and

Chowdhury, 2010; Aamir and shah, 2011; Sare et al., 2014). Balachandran et al. (2004)

(7)

7 / 35

studies at the industry effect and compared effects of dividends announcements on three industries, namely financial firms, non-financial firms and resources firms. With a large sample from the financial and the non-financial industry, their results show significant and positive abnormal returns in the financial and non-financial industries. However, the resources industry (that mostly consisted of mining firms) did not indicate any significant reactions of stock prices. Later, Ali and Chowdhury (2010) investigate the commercial banking industry in Bangladesh, where they find insignificant results as well. Most recent literature (Sare et al., 2014) exploring the Ghana Stock Exchange compared stock prices reactions to dividend announcements in three different industries as well. Based on their findings, manufacturing industries show significant positive cumulative abnormal returns five days before and after the dividend initiation date while no significant results are found for the financial services industry and in the “other” industry. Thus, industry effects may exist and may influence the relation between stock prices and dividend changes.

The utility industry, mainly consisting of electricity, gas & water supply and other related sectors, is the one generally being excluded from investigation when researchers want to observe stock price reactions to dividend changes in a region or a particular country (e.g., Von Eije and Megginson, 2008). This is mostly due to the special dividend policy in the regulated utility industry.

It has been argued that a high dividend payout is likely to be adopted in the utility industry (Miller, 1986; Smith 1986; Smith and Watts 1992). Moyer, Chatfield and Sisneros (1989) argued that monitoring activities performed by security analysts are lower in utility industry compared with other industries. For utility firms, company control delegated to management is less and the majority of control is taken under regulatory supervision (Hansen et al., 1994).

Therefore, regulator supervision on one hand can reduce the management expropriations and

manager-shareholder conflicts (Hanson et al., 1994). One the other hand, as the regulators are

not allowed to own company stocks in the utility firms (Hanson et al., 1994), there is potential

shareholder-regulator conflict which calls for active monitoring. Smith (1986) and Hansen et

al. (1994) argue that, by promoting a high dividend payout strategy, company investors can

effectively monitor both managers and regulators. High dividend payments can reduce the

agency costs caused by manager-shareholder conflicts, as it can reduce free cash flows and

(8)

8 / 35

thus limit mangers overinvestment behaviours (Jensen, 1986). In term of aligning interests between shareholders and regulators, a high dividend payment policy can attract more external investors and force regulators to set the fair value of return on capital (Hansen et al., 1994). When a regulator sets a rate of return below fair value, capital market monitoring will reveal the difficulties in raising capital which will damage the regulators’ reputation. Utility firms provide basic utility services (electricity, gas, water) to all families, therefore their customer base is quite large. Utility firms and regulators in the utility industry are therefore also (indirectly) monitored by their customers as any damage will ruin not only investors but their basic customers. Moreover, Hansen et al. (1994) argued that the costs of monitoring (i.e flotation costs and indirect cost of mis-reactions) through dividends are lower in the utility industry. Thus, a high dividend payment policy is more likely to appear in the utility industry.

The distinctive characteristic of dividend policy in utility industry increases the relevance of investigating the relationship between dividend changes and stock prices at industry level.

However, to my knowledge, only a very few studies explore the impact of dividends changes on stock prices in the utility industry. Shelor and Officer (1994) investigated the electricity firms in the US using data between 1976 and 1989, which confirmed the positive and significant influence of dividend increases announcements and the negative and significant impact of dividend decrease/omissions on stock prices. Aamir and Shah (2011) conducted a similar research on the Pakistan oil & gas sector and their results shows a highly significant positive effect of dividend announcements on stock prices one day before and after the event date.

As suggested by dividend signaling theory and free cash flow hypothesis, positive dividend

(increase) announcements should have a positive signal on stock prices as it can reveal the

positive perspective of future cash flows and the reduce the overinvest problems caused by

managers (Bhattacharya, 1979; John and Williams, 1985; Miller and Rock, 1985; Jensen,

1986). Applying the same theories, dividend decreases which reduce the confidence of

shareholders about the future perspective of a firm and provide more opportunities for

managers’ overinvestment behaviors, will cause a substantial decrease on stock prices and

firm value. Even though the utility industry tends to have high dividend payout policy, the

industry specificity is not necessarily a factor that will change the sign of interaction between

(9)

9 / 35

dividend changes in announcements and stock prices. In line with the shareholders’ reactions to dividend in the general literature and the little literature which investigated the effect of dividend announcements in utility industry (Shelor and Officer, 1994; D’Souza, Jacob, and Willis, 2015), I hypothesized:

H1: There is a positive relationship between dividend changes and stock price changes in the utility industries of China and the US.

2.3. Differences in stock price reactions to dividends

As both China and the US generally adopt different dividend payout policies, the relation between dividend changes and stock prices may also varies across the two counties. However, which country has a stronger effect?

Dividend signaling theory argues that dividend policy changes can convey information about the future cash flows and the financial perspectives of the firm (Miller and Rock, 1985;

John and Williams, 1985). The theory implies a positive relation between information asymmetry and dividend announcements (Miller and Rock, 1985). When there exists asymmetric information between managers and shareholders, dividends signaling effects arise as mangers know more than investors and shareholders can receive the information about the future perspective of the firm from the dividends announced by managers (Miller and Rock, 1985; Bhattacharya, 1979; Jensen, 1986). Under symmetric information, both shareholders and managers obtain equal information about dividends; information revealed can then hardly influence the stock prices via the signaling effect. Thus, the higher the level of asymmetric information, the higher the sensitivity of the dividend signaling effect (Dionne, and Ouederni, 2011). From the agency perspective, dividends are used as a tool to reduce the asymmetric information and corresponding agency costs (Jensen and Meckling, 1976; Easterbrook, 1984;

Jensen, 1986) to align the interests of managers and shareholders. As discussed before,

dividend policy can also be used to reduce the agency costs between regulators and

shareholders (Hansen et al., 1994). When the information asymmetry is high, the increased

possibility of agency problems makes the dividend increase announcements to become a more

efficient tool to reduce agency problems and convince the shareholders. Thus, the

corresponding reaction of the stock prices should then be higher as well.

(10)

10 / 35

In the utility industry firms generally have high dividend payouts, but the level of information asymmetry is lower as most information has already been exchanged within the public (Shelor and Officer, 1994; McLaughlin and Safieddine, 2008). However, the level of information asymmetry across countries varies and thus may result in the differences in marginal effects between dividend changes and stock price changes across countries. In China, firms are considered to be opaque and with strong government control, either directly or indirectly (Xu, et al., 2010). This enhances the secrecy of firm information. Su, Xu and Phan (2008) argued that emerging markets such as China are restricted by the capability of legal enforcement, incomplete legal regulations and weak corporate governance. Thus investors will face more information asymmetry problems in the capital market (Chen, 2004; Peng, 2004; Xu and Wang, 1999). In the US, the proportion of listed international firms and the proportion of individual investors are higher compared with the Chinese stock market (Mei and Yang, 2013), therefore the level of asymmetric information is considered to be lower in the US due to the better monitoring in the market. As a common law country, the US has also better corporate governance together with well-modified rules and regulations (La Porta et al., 1997) than civil law countries (La Porta et al., 1997). This may reduce the information asymmetry due to enhanced regulatory reinforcement, especially after the introduction of Sarbanes–Oxley Act in 2002 (Holmstrom and Kaplan, 2003). By comparing developing countries and developed countries, Gao and Zhu (2012) also conclude that the level of information asymmetry in China is obviously higher than the level in the US.

As the asymmetric information in China is higher than in the US, and asymmetric information can influence the strength of the relation between dividend changes and stock prices, I hypothesize that:

H2: Dividend announcements have larger stock price reactions in the Chinese than in the US utility industry.

3. Methodology

3.1. Data and sample

Using the most recent data within a 5-year period from 2010 to 2014, I collected 203 and

368 cash dividend announcements (increase or decrease) in the Chinese and the US utility

(11)

11 / 35

industry respectively

2

.

Sample firms in the Chinese market consist of 63 out of 78 utility firms from Shanghai Stock Exchange and Shenzhen Stock Exchange

3

. In the US, announcements of dividend changes are collected from 78 firms out of 100 utility firms from NYSE and NASDAQ stock exchange during the same time period

4

. To keep it at comparable level, the definition of utility industry for both countries contains two common sub-sectors, namely gas & water and electricity based on the DataStream classification. In the US, the major utility firms are listed in the NYSE (71 out of 78 firms available), while in China the number of utility firms listed at the Shanghai Stock Exchange is two thirds of total utility firms available. Even though utility firms in the US regularly publish dividend announcements on a quarterly basis, the 78 firms only announced 368 dividend changes within a 5-year period which implies that firms adjust dividend payments on average once per year. Chinese utility firms which issue dividend announcements on an annual basis, release 203 dividend changes announcements during the same period. Compared with the US utility firms, firms in the Chinese utility industry announce dividend changes less frequently.

Table 1 illustrates the information about dividend changes involved in the announcements for the Chinese and the U.S. utility firms respectively. The dividend announcements collected are categorized into two groups based on the dividend changes (ΔDIV): positive dividend announcements and negative dividend announcements. Consistent with Chen et al. (2002) and Bozos, Nikolopoulos and Ramgandhi (2011), positive dividend announcements refer to those announcements with dividend increases (ΔDIV>0) and negative dividend announcements refer to those announcements with dividend decreases (ΔDIV<0). During the period 2010-2014, the majority of collected announcements convey dividend increases. This applies to both markets. However, the number of dividend decrease announcements released by the US utility firms is extremely low, namely 27 announcements. The remaining 341 US announcements are all positive dividend announcements and thus the US utility firms seem

2 The research only includes the announcements that show dividend changes. Dividend announcements which show similar dividend payments with the previous period are omitted.

3 Only 63 out of the 78 Chinese utility firms announced dividend changes during the period 2010-2014. The remaining 15 utility firms in China do not provide any dividend change information.

4 Only 78 out of the 100 US utility firms announced dividend changes during the period 2010-2014. The remaining 22 utility firms in the US do not provide any dividend change information.

(12)

12 / 35

more likely to increase dividends to convince investors instead of decreasing dividends.

Compared with the US, the announcements in the Chinese utility industry consist of 133 dividend increase announcements and 70 dividend decrease announcements. The relatively high number of dividend decrease announcements during the same period already shows the high flexibility and the randomness of dividend policy in the Chinese utility industry.

Table 1

Dividend changes in the announcements from 2010 to 2014

This table reports the Chinese and the US observations of dividend change categories in the utility industry during the period 2010-2014. The dividend changes (ΔDIV) are calculated as dividend changes scaled by the stock prices one day before the event.

U.S. China Total

ΔDIV<0 27 70 97

ΔDIV>0 341 133 474

Total 368 203 571

Dividend announcements including date, value of dividends, and corresponding stock prices of Chinese utility firms are collected from Sina Finance, while the dividend announcement information of the US utility firms are collected from NASDAQ official websites where contains dividend history of all the US firms. Financial data needed to control for firm size, leverage, intra-industry classification and earnings per share are generated from DataStream for both the Chinese firms and the US firms

5

.

3.2. Event study

In order to test whether there is a positive relationship between dividend changes and stock price changes and whether the dividend effect is stronger in Chinese than in the US utility industry, an event study including univariate tests and regression analysis has been performed.

3.2.1. Univariate tests

Following the traditional event study methodology, abnormal return (AR) and cumulative abnormal return (CAR) around the dividend announcement day has been estimated (MacKinlay, 1997). The dividend announcement day refers to the day of dividend declaration (event day). The event day refers to day 0 and the event window is known as the period

5 Some financial data in 2014 currently are not available in DataStream. In total, only 518 observations instead of 571 observations with full data can be included in the regression.

(13)

13 / 35

around event day. The cumulative abnormal return during the event window (i.e. 3 days and 5 days) is used to the test the influence of dividend changes on stock prices, as days prior to the event day may convey information early in the market and sometimes same information is reflected by stock prices after the event day (MacKinlay, 1997).

To generate the AR, actual return and expected return for each event i has been collected separately. The expected return is estimated based on a market model using an estimation window of 100 trading days before the event window. According to the market model equation, coefficients used to calculate the expected return has been estimated. The market model equation is indicated as follows:

R

it

i +

β

i

R

mt

it

(1) R

it

refers to the return for firm at time t, where t refers to the days in the 100-day estimation window for event і. R

mt

is the market return at time t. In the equation, α

i

is the intercept, β

i

is the systematic risks of security and Ɛ

it

refers to the error term.

Estimated parameters of regression coefficients α

i and

β

i

are utilized to predict the expected returns during the event window. The abnormal return in the each event window day is calculated as the difference between the actual return of one firm and the expected return by applying the following equation:

AR

it

= R

it

- (α

i +

β

i

R

mt

) (2) An event window of τ days, which contains the event day and a same number of days before and after the event day, has been used to examine abnormal returns for event і. In this research, an event window of 3 trading days (τ=3) are used to generate the corresponding CAR

for event і. The Chinese stock market is less sufficient compared with the US stock

market and the dividend information convey are thus likely to be slower in China, therefore I applied an alternative event window of 5 days for calculating CAR

to robust the results using a shorter event window period. CAR

is the calculated as the accumulation of abnormal returns during the event window period. The mean of CAR

for all related events has been calculated and named as Cumulative Average Abnormal Return during the period τ (CAAR

τ

).

By testing whether the CAARs of Chinese announcements and of the US announcements are

different from zero via mean test (t-test), one can examine the effect of dividend changes on

stock prices for the both countries. Furthermore, by applying the same method to compare the

(14)

14 / 35

mean difference of abnormal returns between China and US (CAAR difference), one can also figure out whether the dividend announcement effect is stronger in one country than another.

MacKinlay (1997) stated that the mean test is applicable when the sample size is large enough and when the data are normally distributed. However, median tests (non-parametric tests) should be used instead if sample data are not normally distributed. In order to prevent such problems, I also conducted a Wilcoxon signed rank test

6

and a Mann–Whitney U

7

test to examine the median of CARs during the event window as the robustness check.

3.2.2. Regression analysis

In order to further investigate the relationship between dividend changes and stock price changes and the country effects on the relationship, regression analyses have been utilized as the second method by applying the following model:

CAR

= a1 + a2* ΔDIV

i

+ a3* China

i

+ a4*ΔDIV

i*

China

i

+ a5*Firm size

i

+ a6*Leverage

i

+

a7*Electricity

i

+ a8 *ΔEPS

i

+ Ɛ

i

(3) In order to conduct the regression analysis, the dependent variable CAR

using event

window of 3 and 5 trading days respectively, has been collected respectively for event i. The first independent variable ΔDIV

i

refers to the dividend change per share for event i scaled by the stock price one day before the event (Chen et al., 2002). By integrating the ΔDIV

i

in Equation 3, one can examine the hypothesis 1 (which argues that a positive relationship between stock prices and dividend changes exists) via coefficient a2. The second independent variable China is a country dummy used to compare the difference of CARs between the Chinese and the US utility firms, where the variable equals to 1 when it is applicable to Chinese utility firms and otherwise equals to 0. By including the country dummy in the regression model, one can test which country has a larger dividend announcements effect on stock prices. However, the CAR reactions to one unit of dividend changes in China may not be the same as in the US. Thus, a slope dummy ΔDIV

i

* China is included as the third independent variable, in case there is a different effect of the size of dividend changes between the two countries.

Firm size and leverage, electricity and ΔEPS are the firm-level controls for regression

6 The Wilcoxon signed rank test is used to test whether the median is different from zero.

7 The Mann–Whitney U test is used to compare the median of two samples.

(15)

15 / 35

model, when investigating the relationship between CAR and ΔDIV and the country differences between the relationships. Firm size

i

is the size of the firm which announces the event i and the nature log value of total assets is used as proxy for firm size. In the model, Leverage

i

is measured by the debt ratio, calculated as the fraction of total debt in total assets of firm i in the year/quarter

8

of the announced dividend changes. In the sample, the utility industry composes two sub-sectors: electricity and water & gas. Electricity

i

is a dummy variable to control the intra-industry difference between two sub-sectors; a variable equals to 1 if the event i is announced by an electricity firm, otherwise 0. Chen et al. (2002) and He et al. (2009) argues that Chinese firms announce annual dividend and earnings at the same time while in the US, the dividends and earnings are announced at different date. The variable ΔEPS

i

is therefore included as a control to bridle the effect of earnings announcements, and it is estimated as function of Earning Per Share (EPS) changes scaled over the stock prices one day before the event.

3.3. Descriptive statistics9

The CAAR changes during the event window of 5 days for the Chinese and the US utility firms are shown in Figure 1. It is clearly that the CAARs of positive announcements in the Chinese and the US utilities are both positive and keep increasing during the event window days. Besides, regarding the negative announcements, the CAARs decrease below zero one day and two day after the event for US and China utility firms respectively, even though CAARs of both countries experience an increase on the event day.

In Figure 1, CAAR regarding all dividend announcements containing both positive announcements and negative announcements is increasing during the event window days for both countries. However, based on the following line chart, the CAAR of Chinese positive announcements increase heavily than the US CAAR during the event window days, while the CAAR of the US negative announcements suffers a larger decrease after the event day.

8 For Chinese utility firms which announce dividend annually, I used annual data to generate the corresponding leverage and firm size. For the US utility firms which announce dividends quarterly, I used quarterly data to generate the corresponding leverage and firm size.

9 To avoid the effect of outliers, data collected from the utility industry are winsorized at 5% for the whole sample (2.5% from the top and 2.5% from the bottom).

(16)

16 / 35 Fig. 1. Cumulative average abnormal return during the event window days. This graph illustrates the CAAR of announcements collected from two days before an announcement to two days after the announcement. The CAAR of Figure 1 is calculated based on the ARR of each event window days, see Appendix A.

Table 2 illustrates the summary statistics of dependent variables, independent variables and control variables applicable for regression regarding all announcements in Chinese and the US utility industry during 2010-2014. In total, 518 observations have been investigated in the regression analysis.

Table 2

Descriptive statistics for variables

This table reports the descriptive statistics for all variables in Equation 3, for all announcements collected in the Chinese and the US utility industry during the year 2010-2014. CAR (-1, 1) and CAR (-2,2) are dependent variables using event windows of 3 and 5 days respectively. For other variables, they are defined in line with the meanings in Equation 3. Due to missing values in 2014, only 518 observations are used in the regression analysis.

Mean Median Maximum Minimum Std. Dev. Observations

CAR(-1,1) 0.003 0.002 0.063 -0.054 0.024 518

CAR(-2,2) 0.004 0.003 0.083 -0.062 0.030 518

ΔDIV 0.001 0.001 0.021 -0.013 0.006 518

China 0.317 0.000 1.000 0.000 0.466 518

DIVCH10 0.001 0.000 0.021 -0.013 0.005 518

Firm size 15.813 15.825 18.811 12.130 1.480 518

Leverage 0.372 0.355 0.721 0.074 0.127 518

Electricity 0.595 1.000 1.000 0.000 0.491 518

ΔEPS 0.002 0.001 0.059 -0.031 0.015 518

10 DIVCH refers to the item “ΔDIV *China”

-0.50%

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

-2 -1 0 1 2

Chinese positive Chinese negative Chinese all US positive US negative US all CAAR during the event window of 5 trading days

Day

(17)

17 / 35

Among the 518 observations, the mean value and the median value of CARs using an event windows of 3 days are both positive. This implies positive stock price reactions caused by dividend announcements. In line with the result using an event window of 3 days, CAR (-2,2) which has lager event window period also presents a positive mean and a positive median.

Based on the descriptive statistics, the average mean and median of ΔDIV across the whole sample are positive. Therefore, the average dividend change across all dividend announcements collected from the Chinese and the US utility industry conveys dividend increase information. The country dummy China with the mean value of 0.372 and a median value of 0 reflects that the majority of the announcements are collected from the US utility industry and the median is a US firm. The slope dummy DIVCH in Table 2 shows similar results with variable ΔDIV, except for the median with a value of zero.

In terms of control variables, firm size has the lager mean and median compared with the other variables. The mean value and median value of leverage across utility firms are very close, which indicates that the total debt is roughly 36% of total assets, if averaged over the utility firms in the two countries. The intra-industry dummy electricity with a mean value of 0.595 and a median value of 1, shows that more observations are collected from the electricity firms than from the non-electricity firms. Finally, descriptive statistics for ΔEPS shows positive earning changes on average during the same year/quarter as dividend change announcements.

Table 3

Correlation Matrix

The table reports the correlations between regression variables across whole sample. The dependent variable CAR, the independent variables (ΔDIV, China and DIVCH) and the control variables (Firm size, Leverage, Electricity and ΔEPS) are defined in same manner as Table 2.

CAR(-1,1) CAR(-2,2) ΔDIV China DIVCH Firm size Leverage Electricity ΔEPS

CAR(-1,1) 1

CAR(-2,2) 0.790 1

ΔDIV 0.074 0.095 1

China 0.064 0.057 0.183 1

DIVCH 0.082 0.108 0.926 0.223 1

Firm size -0.015 -0.011 0.096 0.152 0.089 1

Leverage 0.070 0.007 0.047 0.261 0.044 0.397 1

Electricity 0.057 -0.016 0.068 0.182 0.078 0.101 0.191 1 ΔEPS 0.067 0.041 0.403 0.116 0.423 0.127 -0.004 0.024 1

(18)

18 / 35

Table 3 illustrates the correlations across dependent variables, independent variables and dependent variables. It is obvious that CAR (-1,1) and CAR (-2,2) are highly correlated, as they present for the abnormal returns for the same announcement in the event windows of 3 days and 5 days respectively. Besides, there is no doubt that DIVCH is highly correlated with ΔDIV, since DIVCH stands for the product of ΔDIV and country dummy China. Except for CAR, ΔDIV and DIVCH, all other variables are not highly correlated with each other.

Therefore, there is no multicollinerity problem in the regression.

4. Result

4.1. Univariate results

Univariate tests which examine the significance of the mean and median of CAR across all announcements, positive announcements, and negative announcements are illustrated in Table 4. The mean and the median of CARs are tested through t-tests, Wilcoxon signed rank tests and Mann–Whitney U tests. For each type of announcements, test result for China, the US and the two countries together are presented separately.

Panel A of Table 4 shows the test statistics for mean and median of CAR (-1,1) and CAR (-2,2) across all announcements. Based on the result, there is a significant and positive mean and median of CAR for utility firms in China, US and both countries together when announcing dividend changes. Therefore, dividend announcements, no matter positive or negative, overall have a positive impact on stock prices.

Panel B of Table 4 illustrates the univariate test results regarding the positive dividend

announcements. The results of t-tests and Wilcoxon signed rank tests provide highly

significant, positive mean and median of CARs among all three observation groups (i.e. All,

China and US). This indicates a positive effect of positive dividend announcements on stock

prices. This result is more in line with the assumptions in the first hypothesis which argued a

positive effect of dividend increases news. Among the negative announcements, Panel C does

not show any significant results for mean tests. The result of median tests is in line with the

mean test for negative announcements, except for the weakly significant and negative median

of CAR (-1,1) across the US negative announcements. This may indicate a negative effect of

negative dividend announcements in the US utility industry. Regarding the Chinese negative

(19)

19 / 35 Table 4

Univariate tests results for mean and median of CAR (-1,1) and CAR (-2,2)

The table reports results of mean tests and median tests for CARs using 3 days and 5 days respectively. The mean test applied in this paper is t-test while median tests used refer to Wilcoxon signed rank tests. Panel A, B, C represent the results for three different announcements categories respectively in China and the US utility industry during the year 2010-2014. Results of each panel are presented into three groups as well. P-values of test statistic are presented in parentheses. *, ** and ***

indicate the significance level of 10%, 5% and 1%, respectively.

Panel A:All announcements

All China US

CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2)

Mean 0.003*** 0.004*** 0.006*** 0.008*** 0.001* 0.002**

(0.001) (0.000) (0.002) (0.002) (0.054) (0.029)

Median 0.003*** 0.004*** 0.006*** 0.007*** 0.002** 0.002**

(0.001) (0.000) (0.003) (0.002) (0.028) (0.012)

Panel B: Positive announcements

All China US

CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2)

Mean 0.005*** 0.003*** 0.009*** 0.011*** 0.002** 0.003***

(0.000) (0.000) (0.001) (0.001) (0.030) (0.006)

Median 0.004*** 0.003*** 0.006*** 0.008*** 0.002*** 0.002***

(0.000) (0.000) (0.003) (0.001) (0.007) (0.003)

Panel C: Negative announcements

All China US

CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2)

Mean -4.72E-04 -2.78E-04 0.002 0.001 -0.003 -0.002

(0.446) (0.473) (0.326) (0.459) (0.360) (0.436)

Median 0.002 4.45E-04 0.005 0.001 -0.005* -0.001

(0.446) (0.497) (0.166) (0.243) (0.087) (0.143)

announcements in the utility industry, the insignificant, positive mean and median of CAR (p >

0.1) implies even a positive effect of negative dividend announcements in China.

Table 5 shows the significance of mean difference and median difference between Chinese

CAR and the US CAR in the utility industry. According to the test results, across all

announcements, the mean difference and median difference between the Chinese CAR and

the US CAR are significantly positive (p<0.1). Therefore Chinese dividend announcements

overall have larger CAR than the US announcements, which is consistent with the

propositions in the second hypothesis. Similar results can also be found for positive

announcements, which show a larger CAR in China as well. Among the negative

announcements, the mean difference between China and US are insignificant. However, the

(20)

20 / 35 Table 5

Univariate test results for mean difference and median difference between China and the US

The table reports the unpaired t-test results and Mann–Whitney U te

sts

for mean difference and median difference among three announcements groups: all announcements, positive announcements, and negative announcements. Mean difference is estimated as the difference between Chinese CAAR and US CAAR and it is positive if the Chinese CAAR is larger than US CAAR. Median difference is difference between the median of Chinese CARs and median of US CARs and it has a positive value if Chinese median is larger than the US median. P-values of t-statistic are presented in parentheses. *, **

and *** indicate the significance level of 10%, 5% and 1%, respectively.

All Positive Negative

CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2) CAR(-1,1) CAR(-2,2) Mean difference 0.005*** 0.006** 0.007 *** 0.008 *** 0.005 0.003

(0.009) (0.015) (0.002) (0.003) (0.274) (0.415)

Median difference 0.004** 0.004* 0.004** 0.006* 0.01* 0.002

(0.032) (0.054) (0.042) (0.052) (0.052) (0.110)

Mann–Whitney U test results indicate a weakly significant and positive median difference between China and the US when using an event window of 3 days. This may imply a weak support for larger (positive) CAR in China among the negative news as well.

4.2. Regression result

Table 6 and Table 7 illustrate the cross-section regression results across all dividend announcements in the Chinese and the US utility industries using event windows of 3 days and 5 days respectively. In order to prevent the heteroskedasticity problems, all regression results in this section are adjusted with White corrections for the standard errors.

Table 6 shows the regression results across all dividend announcements in the Chinese and

the US utility industry using an event window of 3 days. Among the 518 observations, model

1 which shows a positive sign of ΔDIV implies a positive relationship between dividend

changes and cumulative abnormal returns. However, the positive coefficient of ΔDIV is not

significant (p>0.1), therefore the first hypothesis which argues a positive relationship between

dividend changes and stock prices is not supported. Model (2) shows a positive coefficient of

variable China, which may imply a larger CAR caused by dividend announcements in China

than in the US. However, the result is insignificant (p>0.1), either. Therefore, the effect of

dividend announcements in China is not significantly different from the effect in the US. As a

result, the second hypothesis which proposes a larger effect of dividend announcements in

China due to a higher level of asymmetric information is not supported as well. In line with

(21)

21 / 35

the results in the precious two models, model (3) and model (4) which further includes the independent variables China and the slope dummy do not show any significant results either.

Table 6

Regression results of CAR (-1,1) with White corrections

This table reports the relationship between dividend changes and stock prices and the relation difference between Chinese and the US utility industries by using an event window of 3 days. The dependent variable is CAR (-1,1). ΔDIV which represents the dividend changes is measured as dividend changes scaled by the stock price one day before the event. China is a dummy variable to control for country and it equals to 1 if for dividend announcements in China and 0 otherwise. The slope dummy is the product of ΔDIV and China. Firm size is the first control variable which is calculated by the log value of total assets. Leverage which used to control firm leverage level is measured by the debt ratio when firms announce dividend changes. Electricity is an intra-industry control dummy to control and it has a value of 1 if the announcement belongs to the electricity industry and otherwise 0.

ΔEPS is the control to bridle the effect of earnings announcements, and it is measured as EPS changes scaled over the stock prices one day before the event.

Constant indicates the constant term of the regression equation. *, ** and *** indicate the significance level of 10%, 5% and 1%, respectively. P-values are shown in parentheses.

(1) (2) (3) (4)

ΔDIV 0.217 0.199 -0.024

(0.287) (0.344) (0.965)

China 0.002 0.002 0.001

(0.469) (0.554) (0.604)

ΔDIV* China 0.271

(0.649)

Firm size -0.001 -0.001 -0.001 -0.001

(0.136) (0.136) (0.129) (0.133)

Leverage 0.016 0.015 0.015 0.015

(0.111) (0.154) (0.154) (0.152)

Electricity 0.002 0.002 0.002 0.002

(0.328) (0.362) (0.381) (0.390)

ΔEPS 0.087 0.114 0.084 0.078

(0.306) (0.182) (0.326) (0.361)

Constant 0.012 0.012 0.012 0.012

(0.260) (0.258) (0.245) (0.250 )

Observations 518 518 518 518

R squared 0.017 0.016 0.018 0.018

Table 7 illustrates the regression results using a relatively longer event window period,

as the less efficient market in China may have a lasting effect caused by the dividend

announcements. Different from results using an event window of 3 days, model (1) of Table 7

(22)

22 / 35 Table 7

Regression results of CAR (-2,2) with White corrections

This table reports the relationship between dividend changes and stock prices and the relation difference between Chinese and the US utility industries by using an event window of 5 days. The dependent variable is CAR (-2,2). The independent variables (ΔDIV, China and ΔDIV* China) and the control variables (Firm size, Leverage, Electricity and ΔEPS) are defined in same manner as Table 6.

Constant indicates the constant term of the regression equation. *, ** and *** indicate the significance level of 10%, 5% and 1%, respectively. P-values are shown in parentheses.

(1) (2) (3) (4)

ΔDIV 0.491* 0.457* -0.149

(0.068) (0.093) (0.825)

China 0.004 0.003 0.003

(0.276) (0.380) (0.463)

ΔDIV* China 0.738

(0.324)

Firm size -0.001 -4.93E-04 -0.001 -0.001

(0.599) (0.587) (0.553) (0.572 )

Leverage 0.004 0.001 0.001 0.002

(0.734) (0.920) (0.917) (0.900)

Electricity -0.001 -0.002 -0.002 -0.002

(0.599) (0.542) (0.499) (0.484)

ΔEPS 0.013 0.076 0.008 -0.008

(0.895) (0.458) (0.942) (0.942)

Constant 0.011 0.011 0.011 0.011

(0.456) (0.426) (0.394) (0.407)

Observations 518 518 518 518

R squared 0.010 0.006 0.012 0.014

shows a positive and weakly significant (p<0.1) relationship between ΔDIV and CAR. This indicates a weakly positive relationship between dividend changes and stock price changes, which may support the assumptions in the first hypothesis. Consistent with the results in Table 6, model (2) of Table 7 does not show any significant coefficient for the country dummy China. However, the coefficient is shown with lower p-values using a relatively longer event window period. This implies that any effect of dividend announcements in China may require an even longer event window for investigation. Model (3) of Table 7 still shows a positive and significant relationship between dividend changes and stock price changes, if one assumes that stock price reactions in China and in the US are similar. A larger effect of dividend announcements in Chinese utility firms cannot be found in Model (3), either.

Regarding the slope dummy, results in model (4) show an insignificant coefficient of ΔDIV*

(23)

23 / 35

China. Thus the size effect of dividend changes in the Chinese utility industry is not significantly different from the effect in the US utility industry. As a result, hypothesis 2 which postulates a lager effect of dividend announcements in the Chinese utility industry cannot be supported. Besides, the relationship between dividend changes and cumulative abnormal returns in model (4) is insignificant. Therefore, the first hypothesis that argues a positive relationship between dividend changes and stock price changes cannot be supported if one assumes that the size effect of dividend change in China differs from the effect the US.

The insignificant result in model 4 implies an insignificant result in the US utility industry (see Appendix B which illustrates a blurred relationship between dividend changes and cumulative abnormal returns for the US utility industry).

4.3. Robustness test

In the event study, I use the CAR to measure the influence on stock returns and then test the relationship between dividend changes and stock price changes. However, the regression results may be threatened by the normality problems for CARs. In order to bring the values of CARs more closely, I robust our results following the method in Von Eije and Wiegerinck (2010). As the CAR is always larger than -1 (-100%), I transform the dependent variable CAR by adding the value 1 and then take the natural logarithm of the transformed CAR as the new dependent variable.

Table 8 illustrates the robustness check results using the natural logarithm of the transformed CAR as the dependent variable. For the CARs using a shorter event window period (3 days), the robust test results do not show any significant relationship between dividend changes and stock price changes. Besides, the dividend announcements effect in the Chinese utility industry is not significantly different from the effect in the US utility industry.

This result is consistent with the findings in Table 6.

In terms of the robustness check results applying a relatively longer event window period

(5 days), the results in Table 8 are not materially different from the findings in Table 7, which

concluded a weakly significant and positive relationship between dividend changes and stock

price changes in model 3, but no significant difference of stock price reactions between the

Chinese utility industry and the US utility industry in model (3) and model (4).

(24)

24 / 35 Table 8

Robustness test on cross-sectional regression using logarithmically transformed CARs

The table reports the regression result of robustness test. The dependent variable of the regression is

natural logarithm of the transformed CAR, and the transformed CAR is calculated as the sum of

original CAR and value 1. LnCAR3 shows the results for dependent variable using an event window of 3 days, while LnCAR5 shows the results for the dependent variable using an event window of 5 days.

The independent variables and control variables are defined in same manner as Table 6. The results in Table 8 are also adjusted with White corrections. *, ** and *** indicate the significant level of 10%, 5%

and 1%, respectively. P-values are shown in parentheses.

LnCAR3 LnCAR5

(1) (2) (3) (4) (1) (2) (3) (4)

ΔDIV 0.223 0.207 -0.014 0.488* 0.458* -0.132

(0.274) (0.324) (0.979) (0.068) (0.089) (0.844)

China 0.002 0.001 0.001 0.003 0.003 0.002

(0.516) (0.607) (0.658) (0.324) (0.440) (0.527)

ΔDIV* China 0.269 0.718

(0.649) (0.333)

Firm size -0.001 -0.001 -0.001 0.001 -4.60E-04 -4.42E-04 -4.87E-04 -4.62E-04 (0.144) (0.145) (0.138) (0.142) (0.611) (0.624) (0.589) (0.608)

Leverage 0.016 0.015 0.015 0.015 0.003 0.001 0.001 0.001

(0.118) (0.159) (0.159) (0.156) (0.804) (0.936) (0.933) (0.916)

Electricity 0.002 0.002 0.002 0.002 -0.002 -0.002 -0.002 -0.002

(0.338) (0.365) (0.386) (0.394) (0.567) (0.535) (0.491) (0.477)

ΔEPS 0.085 0.113 0.082 0.077 0.013 0.077 0.008 -0.007

(0.315) (0.182) (0.333) (0.368) (0.901) (0.451) (0.937) (0.949)

Constant 0.011 0.011 0.012 0.011 0.010 0.010 0.010 0.010

(0.279) (0.279) (0.266) (0.271) (0.462) (0.468) (0.434) (0.448)

Observations 518 518 518 518 518 518 518 518

R squared 0.016 0.015 0.017 0.018 0.010 0.005 0.012 0.014

4.4. Discussion

The results of the univariate tests indicate a positive effect of dividend (increase)

announcements on stock prices in the Chinese and the US utility industries. Regression results

using a longer event window period of 5 days together with the robustness results taking the

logarithm value of CAR to bring data more closely together, confirm the findings in

univariate results and indicate a positive relationship between dividend changes and stock

price reactions for the whole sample. This result suggests the existence of the dividend

signaling effect and the free cash flow hypothesis (Bhattacharya, 1979; John and Williams,

1985; Miller and Rock, 1985; Jensen, 1986) and supports a positive effect of dividend

(25)

25 / 35

increase announcements and/or the positive effect of overall dividend announcements in other industries (Lang and Litzenberger, 1989; Benartzi, Michaely, and Thaler, 1997; Ryan, Besley, and Lee, 2000; Grullon et al., 2002; Bozos, Nikolopoulos and Ramgandhi, 2011). However, if one assumes the stock price reactions to dividend changes in China to differ from the reactions in US, the positive relationship can hardly be confirmed. Moreover, negative effect of dividend decrease announcements observed in this research is not obvious, as the univariate test results only show a weakly significant but negative median of CAR for negative announcements in the US utility industry. In the Chinese utility industry, the mean and the median of cumulative abnormal returns are even positive but insignificant. This contradicts previous findings based on dividend signaling and free cash flow theories, which conclude on a negative effect of dividend decrease announcements (Lang and Litzenberger, 1989; Shelor and Officer, 1994; Ryan et al., 2000; Grullon et al., 2002; Bozos et al., 2011). A possible explanation is the immature stock market in China and random dividend policy in China. As explained in the literature review section, the Chinese dividend policy is relatively at random which allows the managers randomly initiate/omit dividends or even issue the stock dividends and cash dividends together (Mei and Yang, 2013). In order to attract and satisfy investors, managers might then even actively decrease cash dividends and distribute new stock dividends for investors when there is a financial problem within the firm.

According to Li and Liu (1997), stock price reactions to stock dividend announcements are much stronger than cash dividends. Therefore, when firms decrease/omit cash dividends but increase/ provide stock dividends, there may lead to positive abnormal returns ultimately.

Whether such effects occur would require further future analysis.

Based on the univariate test results, both the US and the Chinese utility industries show a significant and positive effect of dividend (increase) announcements on stock prices. Different from previous US stock market results which found a dividend signal effect (Petit, 1972;

Lang and Litzenberger, 1989; Shelor and Officer, 1994; DeAngelo,DeAngelo, and Stulz,

2006), early studies hardly provided any significant dividend effect on stock price in the

Chinese stock market using early sample periods (Li and Liu, 1997; Chen et al., 2002; He et

al., 2009). According to the later literature, the insignificant results are probably caused by the

inefficient Chinese stock market which can hardly convey the correct dividend information

(26)

26 / 35

through announcements. Zhou (2011) which examined the relationship between dividend announcements and stock prices using different sample periods found that stock prices reactions to dividend announcements can be influenced by the sample period and the maturity of the stock market. Our results which uses the most recent data (2010-2014), confirms the positive effect of dividend (increase) announcements in China. Therefore, the significant effect of Chinese dividend announcements and positive announcements on stock prices may indicate an improvement in the efficiency of the Chinese stock market. This may provide international investors with an insight to focus on Chinese stock market, as it is improving.

For managers working in China, it is important for them to criticize their dividend policies as dividend announcements can be used as a tool to convince investors and increase stock prices.

In terms of the difference of dividend effects between China and the US, univariate tests

show significantly larger cumulative abnormal returns in China than in the US among the

positive announcements group and all announcements group. In contrast, the regression

analysis and robustness tests suggest that the stock price reactions to (size effect of) dividend

changes in the Chinese utility industry are not significantly different from the reactions in the

US utility industry. These conflicting results request for caution, as the univariate test results

do not take into account the influence of other variables on stock prices. Some variables such

as the size of the dividend changes can also influence the size of abnormal returns (Michaely,

Thaler and Womack, 1995). The regression results which control for the size of dividend

changes and other control variables (i.e. firm size, leverage, industry and ΔEPS) are more

reliable than univariate results. Thus, the larger dividend announcements effect in China

proposed in the second hypothesis cannot be supported. This contradicting result for the

second hypothesis may be caused by the level of asymmetric information in the utility

industry. Even though La Porta et al. (1997) found that the level of asymmetric information is

higher in China than in the US, the difference of those two countries in the level of

asymmetric information in the utility industry could not be proven here. Prior literature

(Filbeck and Hatfield, 1999; McLaughlin and Safieddine, 2008; Fremeth and Holbur, 2012)

argues that level of information asymmetry in the regulated industries like utilities is lower

compared with other industries. As the utility industry is better monitored than many other

industries (Smith 1986; Hansen et al., 1994), the level of asymmetric information in Chinese

Referenties

GERELATEERDE DOCUMENTEN

The perceptions of residents regarding the potential impacts of tourism development in the Soshanguve community are presented in the form of effects on one’s personal life

Based on this thesis it seems that airlines in general are able to generate a positive abnormal shareholder return during the 5 years after a merger, as a significant positive

The subtraction of the government expenditure renders remains of 0.6 which are allotted to increasing private investment (Felderer,Homburg, 2005,p.171). In the works by

The converted colours of the 76 sources were plotted in relation to standard MS, giant and super giant stars on the colour-colour diagram in Fig 4.7 and in the colour-magnitude

Dittmar claims that the use of stock options may increase the number of firms that choose share repurchases over the dividend distributions, as the later lead to the

The w lines following 1c, 2c, and 3c in the listing show the minimum column widths specified by the ‘w’ keys in the format; 35 pt is 7 times TABLE’s default column width unit of 0.5

The LaTeX package decision-table provides a command \dmntable, which allows for an easy way to generate decision tables in the Decision Model and Notation (DMN) format. 1 ) This

Looking at the five institutions financial system, internal structure, industrial relations, education and training and inter-firm relations, India seems to be more of a LME