Investor Preference and Market Reaction to Cash Dividends ------Evidence from China’s A Share Stock Marke Hong Xiao

Ming Fang Xiao

Xiao Hui Qu

Email: [email protected] Department of Accounting ,Xiamen University,China Tel:13600902515

Investor Preference and Market Reaction to Cash Dividends —!Evidence from China’s A Share Stock Marke① Hong Xiao

Ming Fang Xiao

Xiao Hui Qu

Abstract: This paper examines whether investors have different preferences for relative higher cash dividends and relative higher future capital gains with respect to market phase. We get two groups in each year and find that investors almost always prefer ‘relative higher capital gain groups’ rather than ‘relative higher cash dividend groups’ except for the high cash dividend group and high capital gain group in the outer group in spite of different market phase. We are the first one to research investors’ preference in cash dividends and capital gains. In an emerging and transition economy, these results have some implications for government regulation of financial market. Keywords:

cash dividend;

investor preference;

capital gains

1. Introduction Since its founding in 1990②, China’s stock market has grew more than 20 years and presented many different characteristics compared to western stock markets. An interested phenomenon is that china’s listed companies like to issue stock dividend but are reluctant to pay cash dividend before 2009. What’s more, companies which issue stock dividend or mixed dividend usually have significant positive market abnormal return around dividend announcement date. Another phenomenon in Chinese A share stock market is that more and more listed companies begin to pay cash dividends compared to stock dividends since 2009. And most studies in china show negative market abnormal return for companies paying cash dividends around dividend announcement date. According to Gordon’s(1962) dividend growth model, the value for investors buying stocks is decided by cash dividends and capital gains, ①

This research is supported by National Natural Science Foundation of China (Grant 70972112)

and National soft science Foundation of China（2010GXS5D221）and The major project of Ministry of education, humanities and social science key bas(2010JJD630004). ②

Shanghai stock exchange was founded in 1990 and Shenzhen stock exchange was founded in 1991.

that’s to say paying cash dividend and capital gains can both increase investor’s value. If above theory is right, rational investors should not respond negatively to those companies paying cash dividends. Why? The purpose of this study is to examine whether investors in Chinese stock market have preference for cash dividends and capital gains with respect to market phase. We hypothesize that investors would prefer stocks which expect to gain high capital gains in bull markets, and investors would prefer stocks which expect to pay a higher ratio of cash dividends in bear markets. It means investors will reverse their preferences in different market phase. Classical financial theory which based upon efficient market and rational economic man hypothesis believes that the form of payment of dividend will not affect shareholder wealth and company value; therefore, investor has no preference between cash dividend and non-cash dividend. Miller and Modigliani(1961) suggested that in a perfect capital market, company’s market value had nothing to do with the dividend policy. Obviously, this theory can’t explain above phenomenon in china. Gordon(1962) finished his book and put forward the dividend growth model which cash dividends can increase company’s market value. However, Gordon didn’t tell us whether investors have any preference between cash dividends and capital gains. Fama(1974), Li hongzhi(1985) and Graham(1985) studied the relationship between cash dividends and company’s investing and financing policy, and they found cash dividends had nothing to do with company’s investing and financing policy. This study result was in accordance with MM theory. Contrary to Fama and Graham, Higgins(1972)believed that cash dividend is a function of profit and investment, thus difference in cash dividends was due to different profit and investment demand and cash dividend was related to company value. Bhattacharya(1979), Miller and Rock(1985), John and Williams(1985) put forward a basic model about cash dividend policy though they defined different information content and signaling cost in their respective models. Asquith and Mullins(1983) chose 168 companies from 1964 to 1980 which paid cash dividend for the first time or paid the next cash dividend at least ten years later, and chose companies that paid increased and three consecutive years cash dividends as a comparison sample, founding that the larger the scale to pay cash dividend, the greater the stock price rise. In china, Chen Xiao, Chen Xiaoyue(1998) who studied the initial dividend policy of the companies listed since 1996, concluded that compared to the pure stock dividends and mixed dividends, cash dividends were not welcomed by investors. Furth more, He Tao and Chen Xiao(2002) used regression method to study cash dividend’s market response both pre and after cash dividend announcement date from 1997 to 1999, showing that investors had no preference in companies which paid cash dividends and paying cash dividends can’t significantly enhance company’s market value after controlling variables such as year, company size, earning, audit results... Wei Gang(1998) chose dividend announcement information of Chinese A share stocks in 1997 and used event study method to study market price response to stock dividend and cash dividend, revealing that cash dividend was not welcomed by investors

compared to stock dividend. Xiong Wei and Hu Jundi(2003) used Shenzhen stock market’s data in 2000 to study market response to cash dividend announcement in Shenzhen stock market and found cash dividend was not an effective signaling tool in china’s stock market, can not boost company’s stock price and bring abnormal return to investors. Almost all cash dividend policy studies in china mainly focus on market response to cash dividend compared to stock dividend and mixed dividend, or its difference between initial cash dividends and consecutive cash dividends, or investor’s reaction to cash dividend increase or decrease, or cash dividend under different shareholder structures and its consequences. But, no one study whether investors have any different preference among companies that all pay cash dividends under different market phase. Do investors more prefer companies paying a high ratio of cash dividend or companies expecting to gain a higher capital gain among companies which all pay cash dividends. Previous studies about cash dividend policy in china have some drawbacks. Firstly, sample used by some scholars is too small which just covers one or two years. For example, Xiong Wei and Hu Jundi(2003) just took data in 2000 as their study sample and make conclusion. We believe the sample is too small so that the conclusion may not be reliable. Secondly, all scholars studying cash dividend in china have neglected a very important variable that is market phase. We know actually there is no perfect market and not all investors are rational men, moreover, due to cognitive biases, emotions and other psychological factors, people may behave differently from normal situation. As bull market, investors may have high expectations to gain high capital gains; by contrast, in bear market, investors may reverse their preference to pursuing for stocks that pay a high ratio of cash dividend because it sounds impossible to gain high capital gains in a bear market. In this paper, we get two sets of comparisons: firstly, we compare sample which all have paid cash dividends to examine whether investors have any preference between cash dividends and capital gains, and we call them within-group comparison; secondly, we compare sample which has paid cash dividends and sample that has not paid cash dividends to find investors preference concerning to cash dividends and capital gains, then we call them outer-group comparison. 2. Data The sample used in this study was drawn from the Chinese CSMAR and Wind database from 2005 to 2012. To qualify for inclusion in the sample, all companies must satisfy the following requirements:① all companies must have trading price information in the yearly final trading day.② all companies must have at least 260 days’ trading information before their dividend announcement date.③, all companies must not now or ever be special treated companies (that’s to delete all the ST, SST companies in the sample).The sample data used in the comparison as table1 and table 2. Table 1 The sample data used in the within-group comparison

year Total numbers③ sample percentage

year listed companies cash dividends no cash dividends sample percentage

2005 602

2006 697

2007 790

2008 829

2009 985

2010 1300

2011 -

2012 -

149 25%

169 25%

208 27%

338 41%

460 47%

571 44%

-

-

Table 2 The sample data for the outer-group comparison 2005 2006 2007 2008 2009 2010 2011 1377 1421 1530 1604 1700 2063 -

2012 -

570

595

641

731

825

898

-

-

418

417

448

471

567

600

-

-

988 72%

1012 73%

1089 71%

1202 75%

1392 82%

1498 73%

-

-

3. Methodology and hypotheses We divided the sample into 4 groups based on two indicators:

D1 P0

P − P0 D1 and 1 P0 P0

D1: expected cash dividend in next year P0: a stock’s closing price at the end of the current year P1: a expected stock closing price in the end of the next year Table3 The sub-sample based on two indicators HIGH④ MEDIUM⑤ P −P 1

LOW⑥

0

P0

HIGH

(high, high)

(high, medium)

(high, low)

MEDIUM

(medium, high)

(medium, medium)

(medium, low)

LOW

(low, high)

(low, medium)

(low, low)

As for the within-group, we rank companies which pay cash dividends into two matching groups according to

③ ④ ⑤ ⑥

P − P0 D1 ratio and 1 ratio each year. For sample P0 P0

Total numbers are the annual number of listed companies that paid cash dividends. HIGH refers to the former 25% MEDIUM refers to the middle 50% LOW refers to the later 25%

belonging to (high, medium) and (high, low), we call them ‘high cash dividend group’, and call their matching group (medium, high) and (low, high) as ‘high capital gain group’. By the same token, we call (medium, low) as ‘medium cash dividend group’, and call (low, medium) as ‘medium capital gain group’. So, we have two matching groups for each year. In terms of the outer-group, we rank companies which pay cash dividends based on

P − P0 D1 ratio, and companies which do not pay cash dividends based on 1 ratio, P0 P0

thus, we get three counterpart groups: high

medium

P1 − P0 D , and low 1 P0 P0

Table4 D1 P0

P1 − P0 P0

⑨

vs low

D1 P0

⑦

vs high

P1 − P0 D , medium 1 P0 P0

⑧

vs

P1 − P0 . P0

Counterpart groups by based on two indicators HIGH MEDIUM LOW

HIGH

MEDIUM

LOW

HIGH vs HIGH

MEDIUM vs MEDIUM

LOW vs LOW

Two bull market periods and two bear market periods have been defined for the purposes of study. The market phase periods used in this study and their respective durations appear below: Table5 The market phase periods and their respective durations Period Date A share Index Duration Bull1 2005/6-2007/10 998-6124 28 months Bull2 2008/11-2009/7 1664-3478 9 months Bear1 2007/11-2008/10 6124-1664 12 months Bear2 2009/8-2012/11 3478-1949 39 months We use event study method to calculate market responses to cash dividend announcements and set four event periods respectively with an event-period window of ±3days, ±10days, ±20days,±30days on either side of the cash dividend announcement date. The estimate period window is -260 day to -60 day with the announcement date as 0 day. Calculating stock’s normal rate of return: Rit = α i + β i Rmt + μ it ⑦ ⑧ ⑨

The high D1/P0 means the top25% The medium D1/P0 means the middle50% The low D1/P0 means the lower-ranked 25%

Where: Rit : the t day return on security i

α i : the intercept term β i : the parameter estimate Rmt : the t day return on the market

μ it : an error term The daily abnormal return for security i on the t day is defined as: ARit = Rit − E ( Rit ) Where: Rit : the observed t day return on security i E ( Rit ) = α i + β i Rmt The cumulative abnormal return for security i on the t day is defined as:

CARit =

t

∑ AR

t =− n

t=-1 … 1; -10 … 10; -20 … 20; -30 … 30

it

CARt =

1 n ∑ CARit n i

TCAR =

CARt S (CARt )

2

n

and

S 2 (CARt ) =

1 n ∑ (CAR jt − CARt ) n − 1 j =1

Where: n; the total numbers of observations We put forward below hypothesis: For the within-group: H1: In the period of bull market, the ‘high capital gain group’ will have significantly greater CAR than the ‘high cash dividend group’, and it will reverse in the period of bear market. H2: Both in the period of bull and bear markets, the ‘medium capital gain group’ will have significantly greater CAR than the ‘medium cash dividend group’. For the outer group: H3: In the period of bull market, the ‘high capital gain group’ will have significantly greater CAR than the ‘high cash dividend group’, and it will reverse in the period of bear market. H4: Both in the period of bull and bear markets, the ‘medium capital gain group’ will have significantly greater CAR than the ‘medium cash dividend group’.

H5: Both in the period of bull and bear markets, there is no significant difference on CAR value between the ’low capital gain group’ and the ‘low cash dividend group’. 4. Results

The market reaction to the ‘high capital gain group’⑩ and ‘high cash dividend group’ ⑪around cash dividend announcement date in the periods of bull market and bear market. Group1 vs group2, group3⑫ vs group4⑬ are the compared results of within-group; group a⑭ vs group b⑮, group c⑯ vs group d⑰, group e⑱ vs group f⑲ are the compared results of outer group. Table6 The CAR value, Dif (1-2)⑳ value and t test for group1 and group2 in the period of bull1 market Bull1 Event 2005 2006 window Group1 Group2 Dif(1-2) T(CAR) Group1 Group2 Dif(1-2) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.0388 0.0567 -0.0179 -0.23 0.0396 -0.0206 0.0602 1.41 [-20,20] 0.0436 0.0844 -0.0408 -0.58 0.0142 -0.0144 0.0286 0.78 [-10,10] 0.029 0.0923 -0.0633 -1.04 0.00338 -0.0338 0.0372 1.38 [-3, 3] 0.00973 0.0479 -0.0382 -0.9 -0.0117 -0.0256 0.0139 0.72 Table7

Event window [-30,30] [-20,20] [-10,10] [-3, 3] Table8

⑩ ⑪ ⑫ ⑬ ⑭ ⑮ ⑯ ⑰ ⑱ ⑲ ⑳

21

The CAR value, Dif (a-b)21 value and t test for group a and group b in the period of bull 1 market Bull1 2005 2006 Group a Group b Dif(a-b) T(CAR) Group a Group b Dif(a-b) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) 0.0844 0.0168 0.0676 0.78 -0.1237 0.0204 -0.1441 -3.39*** 0.1040 0.0180 0.0860 1.08 -0.1157 0.0148 -0.1305 -3.92*** 0.1454 0.0234 0.1220 1.65 -0.0553 0.00362 -0.0589 -2.60*** 0.1095 0.0187 0.0908 1.34 -0.0429 -0.00970 -0.0332 -2.16** The CAR value, Dif (1-2) value and t test for group1 and group2 in the period of bull2 market

We use group1 to represent ‘high capital gain group’ of within-group We use group2 to represent ’high cash dividend group’ of within-group Group3 refers to ‘medium capital gain group’ of within-group Group4 refers to ‘medium cash dividend group’ of within-group Group a refers to ‘high capital gain group’ of outer group Group b refers to ’high cash dividend group’ of outer group Group c refers to ‘medium capital gain group’ of outer group Group d refers to ’medium cash dividend group’ of outer group Group e refers to ‘low capital gain group’ of outer group Group f refers to ’low cash dividend group’ of outer group Dif(1-2)= CAR of group1- CAR of group2 Dif (a-b) = CAR of group a – CAR of group b

Event window [-30,30] [-20,20] [-10,10] [-3, 3] Table9 Event window [-30,30] [-20,20] [-10,10] [-3, 3] Table10 Event window [-30,30] [-20,20] [-10,10] [-3, 3] Table11 Event window [-30,30] [-20,20] [-10,10] [-3, 3] Table12

Bull2 2008 Group1 CAR(%) Group2 CAR(%) Dif(1-2) CAR(%) T(CAR) 0.0794 0.0232 0.0562 2.01** 0.0713 0.0335 0.0378 1.65* 0.0492 0.0188 0.0304 1.73* 0.0121 0.0085 0.00355 0.35 The CAR value, Dif (a-b) value and t test for group a and group b in the period of bull2 market Bull2 2008 Group a CAR(%) Group b CAR(%) Dif(a-b) CAR(%) T(CAR) -0.0770 0.0114 -0.0884 -1.27 -0.0503 0.0102 -0.0605 -1.05 -0.0296 0.0103 -0.0399 -1.37 -0.0197 0.00354 -0.0233 -1.95* The CAR value, Dif (1-2) value and t test for group1 and group2 in the period of bear1 market Bear1 2007 Group1 CAR(%) Group2 CAR(%) Dif(1-2) CAR(%) T(CAR) 0.0378 -0.0678 0.1056 2.82*** 0.0314 -0.0172 0.0487 1.75* 0.0378 -0.0175 0.0553 2.9*** 0.0148 -0.00573 0.0205 1.35 The CAR value, Dif (a-b) value and t test for group a and group b in the period of bear1 market Bear1 2007 Group a CAR(%) Group b CAR(%) Dif(a-b) CAR(%) T(CAR) -0.1487 -0.0532 -0.0955 -2.91*** -0.1014 -0.0178 -0.0837 -3.32*** -0.0587 -0.0154 -0.0433 -2.80*** -0.0134 -0.00797 -0.00547 -0.61

The CAR value, Dif (1-2) value and t test for group1 and group2 in the period of bear2 market Bear2 Event 2009 2010 window Group1 Group2 Dif(1-2) T(CAR) Group1 Group2 Dif(1-2) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%)

[-30,30] -0.0605 -0.0331 -0.0275 [-20,20] -0.0411 -0.0316 -0.00951 [-10,10] -0.0217 -0.0163 -0.00538 [-3, 3] -0.00852 -0.00536 -0.00316

-1.12 -0.47 -0.39 -0.34

0.0148 0.00782 0.00355 -0.0032

0.0307 0.0172 0.00995 0.00133

-0.016 -0.0094 -0.0064 -0.00453

-0.95 -0.64 -0.57 -0.59

Table13

Event window [-30,30] [-20,20] [-10,10] [-3, 3]

The CAR value, Dif (a-b) value and t test for group a and group b in the period of bear2 market Bear2 2009 2010 Group a Group b Dif(a-b) T(CAR) Group a Group b Dif(a-b) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) -0.1805 -0.0392 -0.1413 -5.77*** -0.0632 0.0318 -0.0951 -3.06*** -0.1253 -0.0323 -0.0930 -4.76*** -0.0609 0.0262 -0.0871 -3.86*** -0.0642 -0.0186 -0.0456 -3.62*** -0.0257 0.0140 -0.0397 -2.96*** -0.0264 -0.00491 -0.0215 -2.78*** -0.00628 0.00401 -0.0103 -1.17

The market reaction to the ‘medium cash dividend group’ and ‘medium capital gain group’ around cash dividend announcement date in the periods of bull market and bear market. Table14 The CAR value, Dif (3-4)22 value and t test for the period of bull1 market Bull1 Event 2005 window Group 3 Group 4 Dif(3-4) T(CAR) Group 3 CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.0317 0.0141 0.0176 0.17 0.0406 [-20,20] 0.0663 0.00379 0.0625 0.81 0.0279 [-10,10] 0.1113 0.00515 0.1061 1.74* 0.00554 [-3, 3] 0.1004 0.00486 0.0955 2.37** -0.0017

group 3 and group 4 in

Table15 The CAR value, Dif (c-d)23 value and t test for the period of bull1 market Bull1 Event 2005 window Group c Group d Dif(c-d) T(CAR) Group c CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.1760 0.0420 0.1340 4.48*** 0.0715 [-20,20] 0.1668 0.0401 0.1267 5.14*** 0.00883

group c and group d in

22 23

Dif (3-4) = CAR of group 3 – CAR of group 4 Dif(c-d) = CAR of group c – CAR of group d

2006 Group 4 Dif(3-4) T(CAR) CAR(%) CAR(%) 0.0569 -0.0163 -0.19 -0.00695 0.0348 0.48 -0.0146 0.0201 0.4 -0.00449 0.00279 0.09

2006 Group d Dif(c-d) T(CAR) CAR(%) CAR(%) 0.0372 0.0343 1.43 0.00113 0.00769 0.39

[-10,10] [-3, 3]

0.1189 0.0295

0.0385 0.0150

0.0804 0.0145

4.02*** -0.00698 1.53 -0.0237

-0.0114 -0.0215

0.00444 -0.00218

0.31 -0.23

Table16 the CAR value, Dif (3-4) value and t test for group 3 and group 4 in the period of bull2 market Event Bull2 window 2008 Group 3 CAR(%) Group 4 CAR(%) Dif(3-4) CAR(%) T(CAR) [-30,30] 0.0513 0.0299 0.0214 0.6 [-20,20] 0.039 0.0268 0.0122 0.41 [-10,10] 0.0343 0.000395 0.0339 1.58 [-3, 3] 0.0236 -0.013 0.0366 2.78*** Table17 The CAR value, Dif (c-d) value and t test for group c and group d in the period of bull2 market Event Bull2 window 2008 Group c CAR(%) Group d CAR(%) Dif(c-d) CAR(%) T(CAR) [-30,30] 0.0109 0.0311 -0.0203 -1.37 [-20,20] 0.00241 0.0311 -0.0287 -2.40** [-10,10] 0.0169 0.0224 -0.00553 -0.65 [-3, 3] 0.00426 0.00106 0.00320 0.58 Table18 The CAR value, Dif(3-4) value and t test for group 3 and group 4 in the period of bear1 market Event Bear1 window 2007 Group 3 CAR(%) Group 4 CAR(%) Dif(3-4) CAR(%) T(CAR) [-30,30] -0.0642 -0.109 0.0448 0.96 [-20,20] -0.0161 -0.057 0.0409 0.98 [-10,10] 0.0236 -0.03 0.0536 1.85* [-3, 3] 0.0106 -0.0265 0.0371 2.1** Table19 The CAR value, Dif (c-d) value and t test for group c and group d in the period of bear1 market Event Bear1 window 2007 Group c CAR(%) Group d CAR(%) Dif(c-d) CAR(%) T(CAR) [-30,30] 0.0294 -0.0378 0.0672 3.86*** [-20,20] 0.0278 -0.0182 0.0461 3.29*** [-10,10] 0.0250 -0.00730 0.0323 2.98*** [-3, 3] 0.00590 -0.00674 0.0126 1.85*

Table20 The CAR value, Dif(3-4) value and t test for group 3 and group 4 in the period of bear2 market Bear2 Event 2009 2010 window Group 3 Group 4 Dif(3-4) T(CAR) Group 3 Group 4 Dif(3-4) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] -0.0306 -0.0493 0.0187 0.49 0.0917 0.0442 0.0476 2.17** [-20,20] -0.0392 -0.0309 -0.00826 -0.3 0.0474 0.0268 0.0206 1.19 [-10,10] -0.0118 -0.0245 0.0127 0.6 0.0202 -0.00045 0.0206 1.35 [-3, 3] -0.00681 -0.00894 0.00213 0.21 0.00116 -0.00756 0.00872 0.85 Table21 The CAR value, Dif (c-d) value and t test for group c and group d in the period of bear2 market Bear2 Event 2009 2010 window Group c Group d Dif(c-d) T(CAR) Group c Group d Dif(c-d) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.00723 -0.0518 0.0590 3.36*** 0.0733 0.0475 0.0259 2.27** [-20,20] 0.00149 -0.0403 0.0418 3.04*** 0.0464 0.0274 0.0189 1.99** [-10,10] 0.0128 -0.0200 0.0328 3.29*** 0.0215 0.00785 0.0137 1.95* [-3, 3] -0.00082 -0.00904 -0.00822 1.36 -0.00013 0.000387 -0.00052 -0.10 Table22 The CAR value, Dif (e-f)24 value and t test for group e and group f in the period of bull 1 market Bull1 Event 2005 2006 window Group e Group f Dif(e-f) T(CAR) Group e Group f Dif(e-f) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.2915 0.0851 0.2064 1.15 -0.2969 0.0368 -0.3337 -1.21 [-20,20] 0.2615 0.0640 0.1975 1.18 -0.1909 -0.00243 -0.1885 -1.23 [-10,10] 0.2361 0.0508 0.1853 1.15 -0.0872 -0.0110 -0.0762 -1.06 [-3, 3] 0.1876 0.0228 0.1647 1.00 -0.0564 -0.0170 -0.0393 -0.72 Table23 The CAR value, Dif (e-f) value and t test for group e and group f in the period of bull 2 market Event Bull2 window 2008 Group e CAR(%) Group f CAR(%) Dif(e-f) CAR(%) T(CAR) [-30,30] 0.0446 0.0636 -0.0190 -0.90 [-20,20] 0.0245 0.0421 -0.0176 -1.04 [-10,10] 0.0285 0.0337 -0.00522 -0.40 [-3, 3] 0.00891 0.0152 -0.00624 -0.82 24

Dif (e-f) = CAR of group e – CAR of group f

Table24 the CAR value, Dif (e-f) value and t test for group e and group f in the period of bear 1 market Event Bear1 window 2007 Group e CAR(%) Group f CAR(%) Dif(e-f) CAR(%) T(CAR) [-30,30] 0.1033 0.00573 0.0976 4.05*** [-20,20] 0.0708 0.0163 0.0546 2.88*** [-10,10] 0.0341 0.00323 0.0309 2.09** [-3, 3] 0.0204 0.00369 0.0241 2.41** Table25 The CAR value, Dif (e-f) value and t test for group e and group f in the period of bear 2 market Bear2 Event 2009 2010 window Group e Group f Dif(e-f) T(CAR) Group e Group f Dif(e-f) T(CAR) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) CAR(%) [-30,30] 0.1256 -0.0368 0.1624 6.54*** 0.1164 0.0532 0.0632 3.84*** [-20,20] 0.0905 -0.0322 0.1227 6.54*** 0.0658 0.0324 0.0334 2.44** [-10,10] 0.0492 -0.00776 0.0569 4.44*** 0.0437 0.0167 0.0270 2.52** [-3, 3] 0.0231 -0.00785 0.0309 3.40*** 0.0207 -0.00248 0.0232 2.90*** ***, **and* significant at the 1%, 5% and 10% level respectively. Based on table 6, table 8, table10 as well as table12, we find that dif(1-2) values are significantly positive in year 2007(bear market period) and 2008(bull market period). It means that investors have preference for ‘high capital gain group’ rather than ‘high cash dividend group’, though, the dif(1-2) value in 2005(bull market period) , 2009(bear market period) and 2010(bear market period) are negative ,and dif(1-2) value in 2006(bull market period) is positive, but all of them are not significant. So, we have evidence to support the former part of H1, but the latter part of H1 is rejected. From table 14, table 16, table18 and table20, we find that dif(3-4) values are significantly positive in year 2005(bull market period), 2007(bear market period), 2008(bull market period) and 2010(bear market period). It shows that investors prefer for ‘medium capital gain group’ rather than ‘medium cash dividend group’ though the dif(3-4) values in 2006(bull market period) and 2009(bear market period) are also positive, but both of them are not significant. Then, H2 is accepted. In the outer-group, table7, table9, table11 and table 13 reveal that the dif(c-d) values are all significant negative whether it is in bull or bear market except for the year 2005(bull market period) with not significant positive values. It provides us investors always prefer for the ‘high cash dividend group’ rather than the ‘high capital gain group’ which is contrast to the result of the within-group. Obviously, the latter part of H3 is rejected because we do not find investors reverse their preferences. Table15, table17, table19 and table21 show us no matter in the period of bull

market or bear market, investors always have preference for the ‘medium capital gain group’ rather than the ‘medium cash dividend group’ ,and this is in accordance with the result in the within-group. Undoubtedly, we have evidence to support H4. In terms of the ‘LOW vs LOW’ group, it concludes that investors always prefer for the ‘low capital gain group’ rather than the ‘low cash dividend group’ in the period of bear market, however we do not find any significant difference between ‘low cash dividend group’ and ‘low capital gain group’ for investors based on table 22 to table 25. The H5 is half rejected. 5. Conclusions

In this paper, we carry out an in-depth empirical study on whether investors have different preference between cash dividends and capital gains based on two groups (the within-group and the outer group). We find investors always prefer for capital gains rather than cash dividends for listed companies that all pay cash dividends both in the period of bull market and bear market. That’s to say when all listed companies pay cash dividends, investors are much keener at capital gains and buy more stocks that is assumed to gain a relative higher capital gain in the future. Furthermore, when investors face choice between get a relative higher capital gain but no cash dividend and get a relative higher cash dividend but lower capital gain, the results is a little complicated. Investors prefer for a relative higher capital gain but no cash dividend group when

D1 ratio is not high enough, and they reverse P0

their preferences when have a high probability to get high cash dividends regardless of the market phase. We think the possibilities may be as follows: firstly, the average D1 ratio in the high cash dividend group is about 2.8% which is equivalent to the P0

one-year interest rate of fixed deposit (2%-3.5%). Actually, the 2.8% is investors’ opportunity cost to choose a high capital gain group other than a high cash dividend group, and they believe it is not worth to do so when the

D1 ratio is not lower than P0

the one-year interest rate of fixed deposit. However, they will not bear a high opportunity cost when choose a relative higher capital gain but no cash dividend group if the

D1 ratio is lower than the one-year interest rate of fixed deposit. P0

Secondly, from the perspective of behavioral finance theory, investors feel one bird in hand is better than two birds in wood. When it has a high probability to obtain a high certain income, people tend to avoid risks, thus to get realized gains rather than high future gains. The results of this study generally support our hypothesis that investors prefer for

‘relative higher capital gain groups’ rather than ‘relative higher cash dividend groups’ both in the period of bull and bear markets. Our conclusions are generally consistent with many Chinese scholars’ conclusions such as Wei Gang(1998) and Chen Xiao, Chen Xiaoyue(1998) because the purposes of investors to choose stock dividend rather than cash dividend is also intending to gain a higher future capital gain. It is a very interesting phenomenon in china that investors almost always choose to get a relative higher future capital gain rather than a relative cash dividend and in most cases, Investors do not respond positively to cash dividends. This is quite different from many western scholars’ findings. Based on our classical financial theory, it should have no difference to obtain cash dividends or capital gains. We think the reasons may be as follows: 1, china’s security market is far beyond a perfect market due to a limited history and a regulated market economy, moreover, value investment philosophy in china has not been widely accepted and many investors buy shares in order to obtain short-term price spread. Finally, there is not a complete efficient market in the world and investors are always with limited rationality due to cognitive biases, emotions and other psychological factors. 2, the cash dividends paid by many companies are too small which sounds trivial for investors compared to obtaining capital gains. In our sample, from 2005 to 2010, the majority (about 65%) of the

D1 ratio is less than 2% that is even lower than the P0

one-year interest rate of fixed deposit (2%-3.5%) except for the high cash dividend group in the outer group. By contrast, the majority of the

P1 − P0 ratio is more than P0

120 percent in the bull market. Undoubtedly, huge differences in yield rate make investors prefer ‘relative higher capital gain groups’ rather than ‘relative higher cash dividend groups’. References

[1] Miller, Merton H. and F. Modigliani, 1961,”Dividend Policy, Growth and the Valuation of Shares,” Journal of Business 34, P411 —433. [2] Gorden, M., 1962, ’The Savings, Investment and Valuation of a Corporation,” Review of Economics and Statistics, Feb. [3] Fama E F. Efficient capital markets: A review of theory and empirical work [J]. Journal of Financial, 1970, 33(25): 383—423. [4] Pettit R R. Dividend announcements, security performance and capital market efficiency [J]. Journal of Finance, 1972, 35(12): 993—1 007. [5] Bhattacharya, Sudipto, 1979, “Imperfect Information, Dividend Policy and ’The Bird in the Hand’ Fallacy”, Bell Journal of Economics 10, pp. 259 —270 [6] Asquith and Mullins, 1983, “The impact of initiating dividend payments on shareholders’ wealth,” Journal of Business 56, 77- 96

[7] Brown, S.J. and J.B Warner, “Using Daily Stock Returns, the Case of Event Studies,” Journal of Financial Economics, 1985, pp.3-31 [8] John, Kose and Joseph Williams, 1985, “Dividends, Retention and Taxes: a Signaling Equilibrium,” Journal of Finance 40 , pp. 1053 —1070. [9] Scott D. Below and Keith H. Johnson, 1996, “An Analysis of Shareholder Reaction to Dividend Announcements in Bull and Bear Markets,” Journal of Financial and Strategic Decisions, 1996, 9. [10] Wei Gang, 1998, “An Empirical Study of Dividend Policy for Chinese Listed Firms,” Economic Research Journal, 1998, 6. [11] Yu Qiao, Cheng Ying, 2001, “The Firm’s Dividend Policy and Stock Market Fluctuation in China,” Economic Research Journal, 2001, 4 [12] Chen Xiao, Chen Xiaoyue, Ni Fan, 2001, “An Empirical Study About Signaling Effect of Initial Dividend for Chinese Listed Firms,” Economic Research Journal, 1998, 5.