Signaling theory

Published:

Dividend policy of government owned company in China ------study from signaling theory

Chapter 1

This dissertation examines the possible association between government control level and cash dividend, other emphasized elements and dividend policy of the listed companies in Chinese security market, using sample of 456 stated owned listed companies which pay dividends in 2006-2008. This paper concludes that the higher level of government control, the more dividends would be paid. There is no great significant of investment opportunities and profitability for dividend in government owned firms. But profitability does go the same direction with dividend while investment is negative related to dividend. However, in zero-government share owned firms, two variables are all negative to dividend. They do not have regression relation with dividends, except profitability in government owned companies.

1.1 dividend policy

The Modigliani-Miller (1961) dividend policy theory implies that the perfectly efficiency market have full information and the dividend policy is no relative to the future earnings or the firm value or cash flows in the firms. This theory has been criticized by most practitioners and financial managers, who argue that the MM theory is based on perfect capital market, which is not true in the real world. The differentiation in tax rate, interests and information between managers and shareholders leads to different policy in dividend. The controversial in dividend policy have been listed as ‘puzzle' in the corporate finance. There is no single explanation can solve it, and now, the theory of dividend has been developed in five main streams: the signaling theory and the information problem, the agency theory, the transaction cost and residual theory, bird in hand, the taxation effect hypothesis and clientele preference (Baker et al., 2002).

Lady using a tablet
Lady using a tablet

Professional

Essay Writers

Lady Using Tablet

Get your grade
or your money back

using our Essay Writing Service!

Essay Writing Service

The asymmetric information between managers and shareholders is generated in signaling models. The proponents of this theory believe that the dividend is used as a means to signal future earnings, especially, the higher value firms (see the studies of Harada and Nguyen, (2005); Manakyan and Carroll, 1990).; and the controversial findings indicate indicates that changes in payout policies are not motivated by firm's desire to signal their true worth to the market (Benartzi, 1997; Dhillon), while others find no evidence such as DeAngelo et al. (1996).

The agency hypothesis indicates that the high payment of cash dividends results in more frequently capital rising in the capital market which causes the higher levels of market monitoring and lower agency costs as a result of increased scrutiny the capital market places on the firm (Easterbrook, 1984). According to Troung (2007), largest shareholder can influence the dividend decision of a listed company to benefit his or her own interest.

1.2 Research Objective

This dissertation will employ dividend policy in perspective of ownership structure to explain the effectiveness of government shareholders of shares on dividend policy of Chinese listed firms. This dissertation studies the signaling content of dividend policy of Chinese listed firms under the background of the reform of share structure and dividend policy. China is a civil law country with weak investor protection in the perspective of corporate governance. Consistent with La Porta et al.(2000),who document that companies in civil law countries have a lower dividend payout ratio, among the lowest in the world and around 57% of Chinese listed companies do not pay out dividends at all in 2006 to 2008. According to the report of Kui and Mako (2006), there was almost zero dividend payment for government and other shareholders before 2005. All the dividends were used as re-investment or accumulated as equity in firms. However, there exists some high dividend payout across Chinese listed companies, with some as high as 170% (calculation based on the Wind Information_2009).This phenomenon induces me to explore the different dividend policies of Chinese listed companies, as the researches of dividend policy in Chinese security market is still staying in large, especially, the dividend policy in government owned companies. Thus, research on the dividend policy in Chinese market can benefit from several aspects.

1.3 structure of this thesis

This thesis organized in 5 sections. The first section reviews the relative theories on signaling model, as well as agency theory, ownership structure, taxation effect hypothesis and clientele preference. The second part will review the empirical studies on signaling theory of dividend as well as some empirical studies in Chinese market. And the forth section provides some background about dividend policy in government owned firms in Chinese market. And the last section, follows empirical study in examines all the variables in signaling model for government controlling firms. The last section is conclusion.

Chapter 2: Literature Review

2.1 introductions

Lady using a tablet
Lady using a tablet

Comprehensive

Writing Services

Lady Using Tablet

Plagiarism-free
Always on Time

Marked to Standard

Order Now

This chapter discusses previous studies related to dividend policy; focus on signaling model and agency theory. Other theories and explanations also presented in this chapter, namely: ownership structures, taxation effect hypothesis and clientele preference.

2.2 Signaling model of dividend

Miller and Modigliani (1961) demonstrate that given the firm's operation policy, the dividend policy would not effect on the market value of the firm which measure the return on the cash flow of the firm. Now, signaling theory concerns the use of dividends to act as a signal to transfer information to the outside market. The assumption is that managers know more about the true value of the firm than the outside investors. And the managers, especially the undervalued firms, prone to transfer private information; hence there would be fewer gaps in information holdings between the investor and managers.

Bhattacharya (1979, 1980) developed inter-temporal signaling model from non-dissipative labor market signaling model which based on Spence's term ‘costless signal' that relative to the full information equilibrium, with the assumptions that the tax rate in dividend is higher than capital gains, the investors know less about profitability and the managers maximize the shareholders interests. The most important assumption in Bhattacharya's model is that if the payoff is insufficient to cover the dividends, the firms have to collect outside funds and pay the cost for it. Thus the equilibrium for this is to pursuit for a high return from the new project which will offset the disadvantage of dividend payoffs (e.g. the tax of dividend). In Bhattacharya model, the insiders of firms are assumed to solve the one-period earning to maximize the shareholders' objective function, and after the dividend payment, they sell to the new group of shareholders who receives the payoff generated by the project at the second period.

It was criticized that why do firms use dividends to signal their prospect? Some proposed that the repurchases would be better in signaling. Although it ignored the incorporation of other sources of information (e.g. moral hazard), the model reveals the positive relationship between future dividends and future cash flow, and showed the positive relationship between dividend changes and the price reaction which consistent to other model, John and Williams (1985), Miller and Rock (1985). The reasons for the firms use the dividend as signal was explained in John and Williams (1985) model, for the shareholders tend to dilute their ownership for rising cash in the second period, and the firms intent to retain the original shareholders simultaneously to rise their stock price, which complement to the short of Bhattacharya's model. John and Williams' (1985) simple equilibrium model begins under the condition that there exists a signaling equilibrium with dissipative dividends; taxes are paid only on dividend, no transaction cost on issuing, retiring and trade shares, sources and uses of funds are fully observed. In this model, the tax is the critical. It narrowed the Bhattacharya's (1980) model; give the dividends distribution only when the cash demand for the firm and stockholders exceeds its internal supply of cash. The contribution of the John and Williams's model provided great extension on signaling equilibrium research, such as applied it into the securities dividends rather than on common stocks, extend it into the repeated game theory.

To extend the previous models of the signaling, Miller and Rock (1985) developed a new model which links the earnings and time with 2 periods. They affirmed John and Williams' simple model of that the outsiders compute the investment, which is also the only signal, of corporate from dividends, current cash flows and the amount of external financing if all the parameters are announced. Meanwhile, they combine the risk-adjustment discount rate for the firm's expected earning from the Fisherian optimum and a persistence parameter for the effect of the dividend announcement together. The assumption of this model is the uncertainty of the firm's dividend, investment and financing decision and future earnings, which are the main conditions for firms to violate the Fisher rule, especially the poor performance company who is going to fool the market, pays a higher dividend currently by cutting investment below the optimal level of the second period.

It is obviously that this model is not tax basis which is totally different from Bhattacharya (1980), John and Williams (1985), according to Miller and Rock (1985), the feature of their model abandon one or more of the troublesome assumption. The strengths are the timing consistency, preannouncement value and post-announcement value, which also count the discount rate in. In the financing perspective, it filled relationship between the expected internal cash flow and the financing announcement in detail. Miller and Rock (1985) also illustrate the consistent condition under asymmetric information, and weight the proportional value of each holders based on MM theory, equally, imply the dividend make sense as signals for good news, not for bad-news firm. Their model more tends to support MM theory while Bhattacharya's tends to amend the original model, also, without tax, dividend can substitute for the share repurchase.

Lady using a tablet
Lady using a tablet

This Essay is

a Student's Work

Lady Using Tablet

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Examples of our work

Based on the above literatures, one hypothesis have been confirmed and come to agree that the dividend changes are indeed convey future earnings, but if the direction of the dividend changes leads to the changes of earnings has not. Bhattacharya (1980) focus on the transaction cost for outside financing , John and Williams(1985) ascribe the taxes are the core costs of singling, then Miller and Rock (1985) explain the dissipative cost as the manager's manipulation. They all suggest other alternative model which carries the same signal but with lower cost can be used. Thus the share-repurchase was mentioned, but John and Williams (1985) hold that the corporate insiders should covey information more efficiency and less costly, not only trough share-repurchase. Allen, Bernardo, and Welch (2000) involve the different roles for dividends and repurchases and evidenced they are not substitutes. They indicate that firms with more asymmetric information and more severe agency problem will use dividend rather than repurchases.

Ambarish, John and Williams (1987) explained it is the signal of dividends and investment. They give the efficiency equilibrium model when the dividends and new issues transfer private information at a lower cost than only the new issue. Investment is the core element in their model, but all the factors are subject to the non-mimicry constraint in efficient signaling equilibrium when analysis the efficient signaling equilibrium of the valuable firms. This model extended the model of John and Williams (1985) and deeper the discussion of Miller and Rock (1985) in the conflict between the firm and investors. The two different types of firms will be distinguished by selecting different fractions of firms with attributes of above elements, for the valuable firms prefer to report its true attribution, the less valuable firm would not follow it. The limitation is how to compare a total different projects invested by two firms, it is not realistic for investors to learn the private attribute, symmetric information exception. In the resulting signaling equilibrium, almost every firm invests less than its optimum under symmetric information, distributes a larger residual dividend, and thereby reveals its current cash. Also, Myers and Majluf (1984) show that insiders optimally forego projects with positive present value if their firms must sell risky securities to raise the required capital.

Other model about dividend signaling as Bernheim (1991) who examined the relationship of dividend and share repurchase with the condition that taxes on dividend is more than that on repurchases. In his model, the firm can control the proportion of the total payout to affect the amount of taxation payment, and the payout form is dividend rather than repurchases. And a high quality firm can choose the optimal amount of taxes to transfer the positive information.

Allen, Bernardo and Welch (2000) present the explanation on the inclination of dividend payment rather than repurchase, and the inclination on dividend smoothing. There are different groups of investors who are taxed differently, public and corporate pension funds, labor unions and other untaxed institutions are the primary holders who have greater incentive to be informed about the firm. They state their assumption of dividend is one way of attracting institution, for the restrictions in institutional charter and prudent man rules that make it more difficult for many institutions to purchase investments with low dividend payouts. Then they develop the model with asymmetric information, and attract institutions via clientele effects. According to the competitive equilibrium model, the high quality firms must prefer to pay dividend while the low quality firms must do not have the incentive to mimic high quality types and pay zero dividends, since once they mimic it will lead to a higher share price if institutions do not detect their true quality, if they do no be detected, the share price will be lower than the price before dividend payment because of dissipative tax in dividend. Thus, taxable dividends are desirable because they allow firm's management to signal the good quality of their firms. Paying dividends increases the chance that institutions will detect the firm's quality. In the perspective of dividend smoothing, firms that pay dividends are unlikely to reduce the amount of the dividend, because the clientele precisely a kind of investors that will punish them for it. Thus the managers would keep dividends relatively smooth after weight the positive share price response to the announcement of dividends against the consequences of the shareholders force to cut the dividends in response to poorer performance later.

Above all, we reviewed the literatures of dividend on cash flow, and know the proportion has reach an agreement, and equally important, cost of dividend and its substitute for share repurchase is still in controversial, one comment from Amihud and Murgia (1997) is that dividend news would not be informative if not for the higher tax that they impose on shareholders. And other theories on dividend include reputation (Gillet et al., 2008), maturity (Grullon et al., 2002) and risk (Kumar, 1988) is arising currently.

The signaling problem can be analyzed by transforming it into an equivalent problem, labeled the non -mimicry problem which had been developed by Bar- Yosef and Huffman (1988), who derived the equilibrium dividend signaling model form the Black- Scholes option pricing formula with the assumption that the expected value for managerial compensation is the optional dividend decision, then the equilibrium dividend contains penalty part of the manager's compensation, which will deters the manager from signaling false information. It treats the cash dividend payment as a function of the cash flow risk. The theoretical results explain the empirical results reported on the relationship between dividend and risk while the empirical results is negative between these variables

Ambarish, John and Williams (1987) constructed a signals period model with dividends, investment, and stock repurchases. However, Ambarish, John and Williams (1987) quoted the sources and uses of funds to explain the demand of stockholders for liquidity; and deducted that dividends have not been shown to be an efficient signal competitive with other technologies for transfer information to the market. But it indeed maximizes the current wealth of shareholders while the outsiders ignore other information or other information would not increase the cost, such as personal tax information. Williams (1988) developed a multi-period model with these elements and showed that in the efficient signaling equilibrium, firms typically pay dividends, choose their investment in risky assets to maximize net present value, and issue new stock. Constantinides and Grundy (1989) focused on the interaction between investment decisions and repurchases and financing decisions in a signaling equilibrium. They indict that a straight bond issue cannot act as a signal when investment fixed, but a convertible bond issue can. When investment is not fixed but chosen optimally, the conclusion would no longer true, a straight bond issue can act as s signal.

2.3 Agency theory

Agency costs arise from the conflicts between the firm's stakeholders. This is because higher dividend payouts make it more likely that firms will have to resort to external finance more frequently, which in turn facilitates the monitoring of firms by external capital markets (Easterbook, 1984). Given the emphasis placed on dividends as a signal of future prospects by institutions, it would appear that institutions do not regard themselves as substitute signaling mechanisms (zhang, 2002). Although Miller and Modigliani (1961) assume that managers and shareholders have the same objectives, this is not true in the real world. Managers may have different objectives to shareholders, allowing them to operate in a way which is not optimal for the shareholders, and this leads to a conflict between the two parties. Managers tend to take the entire benefit at the expense of the shareholders. This explanation has been developed as part of the agency cost theory of dividend policy (Al-Malkawi, 2005).

The payment of dividends has been suggested as a possible solution to reduce this agency problem. For example, Easterbrook (1984) explained that paying a large dividend would drive a firm to finance from capital markets in the long term. As a result, professional investors would be able to monitor the behaviors of managers, which in turn would allow shareholders to monitor managers at lower cost. Therefore, dividend payments increase management scrutiny by outsiders and hence motivate mangers to disclose new information and reduce agency costs in order to secure needed funds. Therefore, increasing dividends will cause managers to act more in line with the interest of shareholders rather than their own interest (Al-Malkawi, 2005).

Jensen (1986) makes a similar argument based on the shareholder-manage agency relationship, where managers pay dividends to reduce the firm's discretionary free cash flow that they could use to fund sub-optimal investments and gain privates from them, but diminish shareholder' wealth. That is, excessive cash balances give mangers more investment opportunity and flexibility, which may be harmful to shareholders. Agency explanations of corporate dividend policy are hence based on the implicit assumption that dividend payout policy can be used as a corporate governance mechanism, acting as a monitoring and/or disciplining device (George, 2004).

Equally important, the firm's dividend policy is effective in reducing the expected agency costs may also depend on its ownership structure (which will be discussed in next section). The conflicting between controlling shareholders and other shareholders has raising in recent literatures, such as Shleifer and Vishny (1997), Horndness (2002). According to the ownership separation and control literature, government ownership is likely to control incentive problems between managers and fixed claim-holders by reducing discretion over firm's projects (Gul, 1999). Additionally, the agency relationship between shareholders and bondholders may be reflected in the firm's dividend policy (Jensen and Meckling, 1976; Myers, 1977). According to this view, shareholders can expropriate wealth from bondholders by means of excessive dividend payments, financed by reducing investments (investment-financed dividends) or issuing new debt (debt-financed dividends). This would in turn imply a negative relationship between dividend payouts and agency proxies for those firms that are more susceptible to this type of agency problem (Al-Malkawi, 2005). For example, shareholders of firms with greater growth options can be argued to have

2.4 Ownership structure

Walker and Petty (1978) report that private or closely held companies in Anglo- Saxon countries like USA and UK, rarely pay any dividends at all, while publicly traded companies pay out sizeable cash dividends. As regards other less developed countries, Naser et al. (2004) and Al-Malkawi (2005) report that the main corporate feature in Arab countries is that organizations are run and controlled by controlling shareholders. They highlight the feature of families as controlling shareholders who executive a degrees of management activities or at least influence on management activities. These activities are based on recommendations from influential family members, which may cause problems between the interests of minority and majority shareholders. The large shareholders may take the role of effectively monitoring the activities of the firm's managers and insider shareholders (Redding ,1997) and they also have the power to expropriate the small outside shareholders (Gugler and Yurtoglu, 2001).

Add government owned firms on dividend policy

2.5 Taxation effect hypothesis and Clientele preference

Another explanation for the relevance of dividend payout policy is the taxation effect hypothesis, which suggests that dividends are subject to a higher tax cut than capital gains. It is further argued that dividends are taxed directly while capital gains tax is not realized until the stock is sold. Therefore investors prefer to retain profit in firms rather than distributing cash dividends because of the higher tax. The advantage of capital gain treatment may lead investors to favor a low dividend payout rather a high payout. This theory suggest that firm should keep dividend payments low if they want to maximize stock prices (e.g. Brennan, 1970; Bhattacharya, 1979; Miller and Rock, 1985; John and Williams, 1995; Baker et al., 1999).

However, one of the important studies that discussed the tax effect hypothesis is Brenan's paper (Brennan, 1970), which show that there is a positive linear relationship between pre-tax returns of stocks and dividend yield based on the after- tax capital asset pricing model, suggesting that an increase in dividend yield should be associated with an increase in the pre-tax returns in order to compensate investors for the tax disadvantages (Al-Malkawi, 2005).

However, several studies related the dividend tax association to the Clientele effect. Miller and Modigliani (1961) coined the term the ‘dividend clientele effect' for the tendency of certain investors to be attracted to a specific dividend paying stock. Miller and Modigliani agreed that the clientele theory does affect the firm's dividend policy, but due to their assumption of perfect capital market, no single client can have an affect on the market, thus leaving the firm value unaffected by dividend or capital gain. In the real world, taxes (clientele effects) exist and this allows the creation of specific types of client. Therefore, each firm will look to attract their clientele, represented by the investors that favor the firm's payout ratio. Thus, tax-included clientele preference may affect the firm's dividend policy.

The discussion regarding clientele theory is elaborated here by stating the tax clientele effect. Due to different rate of tax among different investors in the market, the investors tend to be attracted to dividend policies that are most suitable to their tax requirements/cuts, thus giving them the most revenue. This notion is known as the tax clientele effect, as highlighted in the above paragraph. Elton and Gruber (1970) published one of the earlier studies which support empirically the idea that the dividend-tax clientele effect is a strong factor affecting investors' decisions. Furthermore, several recent studies also came to the same conclusion, such as Long (1978), DeAngelo and Masulis (1980).

Chapter 3 Empirical Studies in Dividend Policy

3.1 Introduction

Various empirical studies on dividend policy found in the literature are discussed in this chapter with consideration to the signaling model, agency theory, ownership structures.

3.2 signaling model

Empirical test involves the signaling explanation yield mixed results, even in conflicting.

Much evidence e.g., Asquith & Mullins (1983), Kalay & Lowenstein (1986), shows that dividend changes are positively associated with stock returns in the days surrounding the dividend change announcement. Bernheim and Wantz (1995), Brooks, Charlton and Hendershott (1998), and Nissim and Ziv (2001) support the signaling hypothesis by finding an association between dividend increases and future profitability. While other studies, e.g. Watts (1973), Benartzi, Michaely, and Thaler (1997) and DeAngelo, DeAngelo and Skinner (1996), do not support the hypothesized relation between dividend changes and future earnings. Most of the empirical examined the signaling theories from varies hypotheses:

the dividend change is in the positive relationship to the earnings change; the dividend change implies the change of stock price, and they are in the same direction; the tax on the dividend constraint the behavior of a firm, and they are in different direction.

3.2.1 The relation between earnings, profit, stock price and dividend

Watts (1973) tested directly the relationship between future changes of profitability and current and past dividend policy with the sample from 1947 to 1966 for a total of 310 firms with complete information for the period and run firms by time series estimations. The average and median coefficients, although positive, are small and insignificantly different from 0 in most of his specifications. He thus argues that there must be other reasons besides signaling for the positive reaction of stock prices to dividends. DeAngelo et al. (1996) also find no evidence after examined the relationship of them. While Benartzi et al. (1997) show that earnings growth rate does not increase as dividend changes, which has been argued that it supposed be attributed to inadequate controls for expected changes in profitability (Nissim and Ziv 2000). Grullon et al.(2002) find that increase dividends experience increases in earnings during the same year, but no increase thereafter, while firms that decrease dividends experience decrease in earnings during the same year and increases thereafter. Their result was supported by Dhillon et al.(2001), who also shows that earnings increase relative to analyst forecast after dividend increases that exceed forecasted dividend increases, but unexpected future earnings are unrelated to dividend decreases. Erik Lie (2003) exam quarterly data from 1980 to 1998 as a sample and find that 661 dividend decreases and 484 dividend omissions. The results show that operating performance declines before both dividend decrease and omission announcements and increases thereafter. Other evidence from Brickley (1983) and Haely and Papelu (1988) are also support the signaling hypothesis.

Haely and Papelu (1988) find that dividends are more related to present and past changes in profitability than to future changes in profitability. So is Lintner (1956), who provided the empirical observation that dividend depends in part of the current earnings and part of the dividend of the previous year, thus challenging the dividend signaling hypothesis. Later it was confirmed by Fama and Babiak (1968). They found that the frequency of a dividend increase depended on both the number of occasions on which earnings had risen and on how recently these had risen, after studying of 392 companies in the United States between 1946 and 1964. According to Lintner(1956), firms increase dividends only when they are sure they can sustain the higher dividends, this means, the firms determine payout ratio when they have long-run target in mind (Grullon et al., 2002).

Harada and Nguyen (2005) examine the relationship between dividend adjustments, and subsequent operating performance and long-term stock returns for a large sample of Japanese firms. Compare to the previous conflict results, the condition under which the dividend adjustment take place can substantially improve the relationship between dividend changes, earnings changes, and stock returns. They generate a logic model based on five fundamental variables, the most significant of which are the level and the change of operating profits, and the level of dividend payments. Firms are expected to increase their dividends when operating profits are high and increasing, and when their dividend payments are low relative to other firms. The assumption is that expected dividend increases (decreases) are more informative than unexpected increases (decreases). In particular, unexpected dividend increases appear to derive from overly optimistic managers whose signaling behavior adds more noise than information. Removing these cases produces a more significant association between dividend changes and subsequent earnings. Risk-adjusted stock returns are also shown to be consistent with the predicted change in earnings.

Manakyan and Carroll (1990) use the Granger tests of causality and nonparametric test to exam the relationship of dividend and earning. Results are consistent with the hypothesis that dividend signals are followed by unanticipated changes in earnings in the subsequent two quarters. The necessary condition for the existence of this functions is that dividend signaled changes in performance be realized. The evidence supports the existence of a signaling value function that dividend signals are associated with unanticipated changes in short-term earnings. Results of the Wilcoxon signed-rank tests support the assertion that unexpected dividend changes precede changes in short-run earnings in a direction consistent with the signal given. The causality tests indicate that in general, earnings cause dividends, while there is evidence of a simultaneous relationship between unexpected dividends and earnings in the near tem. These conclusions are robust to different measures of dividend signals and different measures of resulting unexpected performance.

The timing of observation is a core factor in testing the relationship between stock price change, earnings, and dividend. The study of Brickley (1982) analyzes the behavior of stock prices, earnings and dividends around dividend announcements and compares the relationship between these variables for special dividends versus regular dividends. The specially designated dividends means that dividend is labeled as extra or special compare to the unlabeled dividend increases by the manager. The basic sample used consists of 165 specially designated dividend declared from 1969 through 1979 on common stock listed on the New York or American Stock Exchange. Common stock returns around the announcements of specially designated dividends and regular (unlabeled) dividend increases support the notion that management uses the labeling of dividend increases to convey information to the market about future dividends and earnings. Both specially designated dividends and regular dividend increases appear to convey positive information and the regular dividend increase providing a more positive message.

From the above studies, we can draw simple conclusion that most of them exam the relative figures from the relationship of dividend and past earnings, current earnings and future earnings, particularly, indicate the effect of short term earnings; and the balance of the current dividend and future dividend. Most of them indicate that the dividend indeed signals the stock price change and earnings.

3.2.2 The relationship between capital cost and dividend

Tax-based signaling models propose that the higher tax on dividends is a necessary condition to make them informative about companies' values, lots of the empirical evidenced this theory, some market support while others not. Signaling models predict that firm value is more sensitive to a more costly signal, as know that the marginal cost of dividend signal be increased as relative dividend taxation increase. The result of Bernheim and Wantz (1995) suggest that the return is sensitive to the cost margin, which is favorable to the signaling hypothesis. They tested the regression relation for both the 1962-1988 and 1978 -1988 sample periods present that the coefficients of A DIV * THETA are negative and estimated quite precisely and the coefficients of ADlV* HIRATED are all negative while the ADIV*CAP are all positive and highly statistically significant. It reveals that the dividend announcement will drive up the share price, and share price rise in the same direction of the tax rate change and also the raise in capacity utilization. These findings are favorable to the signaling hypothesis. The problem of this analysis is one possible criticism of this analysis is the variables may not be an appropriate measure of the tax burden on dividends, relative to undistributed profits.

However, the result of Bernhardt et al. (2005), who test the monotonicity prediction for the stock return, price and dividend with the sample of 1962 - 1996, reject the original signaling theory, they find that the information content in dividends is not positively related to the marginal cost. As depicted that the cost of the signal increase means higher tax rate, the higher quality firms prefer to differ themselves with a smaller amount of the signal from others, which will lead to a greater revision of the market estimation of the firm. Thus, the higher taxed in dividend income signal a greater increase in firm value, only if the increased value exceeds the increased tax. On the contrary, a given decreased cost will be associated with a greater negative excess return ad high dividend tax (Bernhardt et al., 2005). But finally, their finding only evidenced the strong relationship between tax and excess return, not the dividend relate to the tax. Dong Ming & Chris Robinson(2004), also insist that the taxes on dividends is more heavily than capital gains is the main reason for the individual investors prefer to be paid in stock dividends but not in cash. The dividend tax drives to the dividend behavior which complement to the Bernhardt's result.

Amihud and Murgia(1997), test the taxation signaling theory in Germany with the data from1988 to 1992 by 200 companies, where dividends are not tax-disadvantaged and in fact are taxed lower for most investor classes, some even pose none, and their model should predict that dividends give no signal. But their result shows that stock price reaction is relative to dividend news as it is in the United States. This suggests that beyond taxation, dividends have information content that can be explained by other factors. They give the reasons to explain the difference that the companies do not pay a higher tax when they allocate income to dividends instead of retaining it, but to the investors, the tax burden of dividends is lower which lead to a conflict conclusion from the theories in the United States.

3.2.3 Risks and dividend policy

Other signaling tests extend to the national economy, debt, cash flow, risks, Firm maturity and governance, such as the studies of Kalay (1980), Kale and Noe (1990), Brook et al. (1998) and Modigliani (1989). It is likely that the national economy will affect the ability of a firm to generate earnings and to pay dividends, and the impact varies over time (Brickley, 1982). In a negative earnings environment, dividend cutting firms perform better than firms that resist a dividend decrease (Harada and Nguyen, 2005; Healy and Palepu, 1988). Harada and Nguyen (2005) analysis the sample from the first and second sections of the Tokyo Stock Exchange covers from 1992 to 2002 with 13,708 samples, find that the relationship of earnings changes and dividend decreases is less significant while dividend increase is positively related to the operation performance changes. They also contrast risk adjusted stock returns with operating profitability to exam the impact of dividend changes with different time horizon, and suggest that investors do not react consistently to the signals conveyed by dividend increases and the firms prefer to improve their profitability. According to the research of Harada and Nguyen (2005), the performance of the firms tend to be health, increase in dividends in a favorable condition experience higher growth in earnings than those firms who increase dividend in unfavorable condition.

Grullon et al. (2002) import 7642 samples from 1967 to 1993 from US security market to examine the relation between dividend changes and changes in systematic risk of firms, changes in profitability and the risks. The result firmly rejects their original hypothesis, the profit in fact decrease, but not increase after a dividend increase. And dividend increase is associated with subsequent decreases in risk and the initial market reaction to the increased of dividend is strongly associated with the risk decline. They are in negative relation. They link the dividend to the life cycle theory of a firm with the hypothesis that the larger amount of dividend payment is the sign of longer maturity process. The result incidental reject the implications of the cash flow signaling models, while they show the risk decline is a sort of decline in cost of capital which accounts for the decline in the firm's growth prospect as the result of the positive price reaction to the dividend increase announcement. Briefly, an increase in dividend signals the firm is in maturity phase, shrinking investment opportunity, declining in risk, declining in return and profit.

Johnson et al. (2006) introduce the dividend signaling model into closed end equity funds that adopt explicit policies committing them to pay minimum dividend yields. These policies represent deliberate attempts to reduce share price undervaluation relative to. Funds that adopt minimum dividend policies experience reductions in their share price discounts, trade at smaller discounts than other funds, earn greater excess returns following policy adoption, and their managers survive longer than other managers do (Johnson et al., 2006). The results are broadly consistent with the predictions of dividend signaling models although the signaling costs are different, and suggest that high quality closed-end funds can reduce undervaluation via dividend policy. It consists to the theory in clientele effect, but it still cannot prove the signaling is the only one reason for funds pay dividend and the relationship between discount rate and the transaction cost.

3.3 agency theory

The theory has largely been tested in empirical studies, especially for the large companies which engage large equity ownership of outside blockholders, although the variables used are different. Such as Gul (1999) pooled cross-sectional time series data and employs investment opportunity set (IOS) to explain the relationship between government ownership of shares and corporate policy in Chinese stock market. The result shows that government ownership is positively associated with debt financing and dividend policy while IOS is found to be negative associated with debt financing and dividend payments. Thus, this paper only employees the government owned companies to exam the possible relationship between the preferences of larger shareholders and dividend policy.

Mitton (2004) also evidences that the positive relationship between corporate governance and dividend payout in the perspective of agency models of dividends from a sample of 365 firms from 19 countries. He also shows that stronger opportunities firms have, lower dividend payouts, and profitability makes little effect on the dividend payout ratios. And the research of Cheung et al.(2005) just compliment to this result. They analyze a sample of 412 publicly listed Hong Kong firms during 1995-1998, and find little relationship between family ownership and dividend policy. Only for small firms, there is a significant negative relationship between dividend payouts and family ownership up to 10% of the company's stock and positive relationship for family ownership between 10% and 35% of total company shares outstanding and dividend yield for small market capitalization firms.

3.4 ownership structure

Several studies have suggested that dividend payouts can play a useful role in reducing conflict between insider and outside owners. When insider owners pay cash dividends, they return corporate earnings to investors and are no longer using these earnings to benefit themselves (La Porta et al., 2000). Nevertheless, the percentage of earnings that can be used as dividend depends on the ownership structure. Glen et al.(1995), Gul (1999) and Naser et al.(2004) specify that in emerging markets, government ownership is a major determinant of the dividend decision-making process. Gul (1999) suggests a positive association between government ownership and dividend, arguing that firms with high government ownership is comparatively less difficult to finance investment projects and hence can afford to distribute more dividends. On the other hand, firms with lower (zero) government ownership face difficulties in raising finance, and consequently rely instead on retained earning for investments, thereby, paying less dividend. Glen et al.,(1995) contend that investors need to be protected in countries with poor legal systems, and because governments are powerful investors, they should act as a safeguard for the minority shareholders by monitoring the insider shareholders and forcing them to pay cash dividend as return of investment.

Naser et al.,(2004) suggest that in the GCC countries, where legal protection is limited, governments have a strong desire to build the reputations of firms and to avoid the exploitation of minority shareholders by paying them high dividends. They further assert that the need for such a reputation has significant effects in young stock exchanges where there is no history of the good treatment of minority outside shareholders. In addition, this need is greater when there is high uncertainty about the future cash flows of firms (Al-Malkawi, 2005).

Al- Malkawi (2005) found that among large shareholders, the government is one of the most influential in affecting the dividend policy of firms listed on the Amman Stock Exchange. He explains that the government acts on behalf of the citizens who do not control the firm directly. Therefore, in such firms ‘ a double principal-agent' conflict will exist because an agency conflict may occur between citizens and government representatives, as they might not act in the citizens' best interests, and on the other hand between representatives and other managers. The solution to this problem is higher payment of dividends, which may reduce the cash flow available to managers, thus reducing the agency problems in the firm. This explanation consists with the findings of Gugler (2003) who examined dividend policies of Australian firms.

3.5 taxation effect and clientele effect

Several empirical studies have examined Brenan's model. Of those studies, Black and Scholes' paper (1974) examines Brennan's findings but the results showed that investors can take dividends to be irrelevant while constructing their portfolios. The study of Litzenberger and Ramaswamy (1979) did not agree with the argument of Black and Scholes about the tax effect. Litzenberger and Ramaswamy (1979) found that there is a positive relationship with a high significance level between dividend yield and the stock return. Therefore they supported the dividend tax effect. However, Miller and Scholes (1982) pointed out a number of drawbacks, showing that Litzenberger and Ramaswamy ignored information effects and announcements of dividend omissions, thus causing a bias in the results. Litazenberger and Ramaswamy (1982) address this issue by using expects short-term dividend yields in order to remove the bias and their results showed significant positive coefficients of dividend yield. Keim (1985) also found a statistically significant positive dividend yield coefficient.

On the other hand, the results of studies by Lewellen et al.(1978), show only weak evidence to support the clientele effect hypothesis. Nevertheless, it should be noted that a number of studies argue that taxation theory in explaining dividend policy may not depend on tax alone: amongst those are Brennan and Thakor (1990), and DeAngelo (1991).

3. 6 Empirical Researches in Chinese Market

3.6.1 Introduction

Various empirical studies on dividend policy found in Chinese literature are reviewed in this chapter with consideration to the signaling model, agency theory, ownership structures.

3.6.2 Signaling model

Kong Xiaowen and Yu Xiaokuan(2003) researched market reaction to the dividend announcement and the content of dividend information based on the trading date in 2001 to analysis the signal transferring effect. They use the cumulative abnormal return rate to exam the market reaction to different dividend delivery polices, and test the relationship between current return and the future return. The result shows that the dividend announcement will increase the CAR, no matter about the forms of the dividend policy. And the CAR in none distribution companies are highest, and the distributed companies is better than that of none distributed ones in the future profit making. Therefore, it can be drawn that the dividend distribution policy was in a great casualness, for it failed in transferring the information of the future return of the companies through different forms of the dividend policy. However, Jia Yan (2003) and Xiong Wei (2003) point that cash dividend can not become effective signal transmit tool, the investors cannot get abnormal return as it cannot cause the stock price go up.

Liu Changkui and Wanghong (2007) used the multi-regression analysis model based on the listed companies which with high cash dividend distribution from Shanghai security trading in 2006, and examined the main element that impact the dividend is: profitability and the net value of assets. Song Xianzhong (2006), employed CAR and multi-regression analysis model to analysis financial data from 190 companies from 1999 to 2002, and indicated that listed company do not adopt dividend policy when they suffering financial problems; the market reaction to the cash dividend is relatively short, and the persistence and the reaction is not strong; most investors do not care about the financial strategies of a company, the signaling effect is not significant in China.

Xiao Lingfei (2007), selects 411 listed companies and analysis their financial data for 5 years, with assessment method for signal indicator also multi-regression analysis for relation of the change rate of cash dividend in current year and the financial indicators of next year to analysis the signal transferring effect. The result was drawn that:

There is a strong substitute relationship between the cash dividend and bonus shares; the current cash dividend rate changes and the relative value of changes of earnings per share for the next years is significantly in the negative relation, while the price earning ratio for the second year is in positive significant, and it also in positive significant relation to the net assets per share for the next year. Meanwhile, Luo Zhengying (2007) explained the failure of signaling effect of dividend in Chinese market attribute to the speculate activities and irrational expectation which cause the time lag in adjusting their portfolio, even ignore the signaling of cash dividend. And he suggests that it works only under the marco policy changed in china.

Li Zhuo (2007, 2008) tested the relationships of types of dividend, dividend payment rate and future profitability, furthermore, investigated the effects of large shareholders' cash dividend on the earnings persistence. And he give the conclusion that the earnings persistence for companies which distribute cash dividend is better than that without dividend distributions, and also the net profit is much stronger in the future; and among the distributed companies, the relationships between dividend payment rate and the earnings persistence is not in simple learner regression; the preference of larger shareholders would not effect the earning persistence. Tang Wanhong (2007) studies the impact of personal tax and the reform of shareholders structure on the cash dividend, and shows that signaling transmission effect in deed strengthening after her test the data from Shanghai Stock Market.

3.6.3 Ownership structure

Zhang(2001) concludes that concentration equity as well as exists several large shareholders is the most effective as the role of corporate governance after analyzing the equity ownership structure of thirty-six agricultural listed companies as sample. Other empirical studies in ownership structure show a positive relationship between large shareholders and cash dividend payment amounts. Such as Li (2009) compared dividend policy and ownership structure based on the year of 2006 which year completed in share split reforming in Shanghai and Shenzheng security market. The result shows that less activities of misappropriate benefits of small shareholders by inside shareholders after increasing share dispersing; and the dividend payout tend to be higher, for the large shareholder prefer to get high return on shares by cash dividend payment.

Lu and Wang(1999) employed 38 variables to explain the dividend policy of listed firms. They get 11 main factors which are all significant with cash dividend, and draw the conclusions from the study are that the higher of the large shareholders' ownership, the higher of the dividend payment; further, the higher of the government shareholders of shares, the higher of the inside control and the lower of the dividend payment; meanwhile, the lower of the government shareholders of shares, the higher level of firm growth and the higher of stock dividend payment and lower cash dividend payment.

3.6.4 Agency cost

Zhou and Xiao (2009) evidence the relationship of manager and dividend payment based on the empirical study of agency cost. They find the wages for managers positively relate to the cash dividend, although it presents as stage correlation. Wu et al,. (2005) explain it that the more of large shareholders in boards, the less tendency of cash dividend payment for agency cost control. And the boards are preferred to benefit from large amount of cash, even from further investment.

Liao and Fang (2005) distinguish the over invested firms by analysis the growth opportunity and cash flow to evidence the conflicts between shareholders and managers based on agency theory. They selected 315 firms as sample which increased dividend of 10% from 2000 to 2002, and found that the existence of government shares increased the possibility of over investment, and impact the cash dividends.

Kong(2006) describes about how agency problems among controlling shareholders and minority shareholders impact on cash dividend and employees financial data from 2003 to 2006 to build stepwise regression models in the framework of different control degree. The result shows that the stronger controlling status of large shareholders, especially the government shareholders, the more cash dividend payment without considering of future investments.

3.6.5 Summary

Moreover, several studies have investigated whether investors may favor firms paying high dividend for reasons other than tax, since high dividends reduce information asymmetry and agency problems.

CAR is widely employed into researching the dividend announcement and the market reaction to the event to calculate the excess return of stocks to explain the dividend effect. According to Xiao (2007), it is not accurate to attribute the price changes of stocks to the dividend changes, because there are lots of different information reach to the market at the same time. CSRC (China Securities Regulatory Commission) prescribed listed companies must proclaim profit distribution projection at the same time when come out with annual report, that"s to say, information about dividend and profit as well as auditing report are available to all investors in market, which leads failure of maintain the clearness of event window by shortening it (Hu Zhiqiang, 2005).

Meanwhile, Luo Zhengying (2007) explained the failure of signaling effect of dividend in Chinese market attribute to the speculate activities and irrational expectation which cause the time lag in adjusting their portfolio, even ignore the signaling of cash dividend. And he suggests that it works only under the marco policy changed in china. Jia yan (2006)However the security market of our country has some unique characters, such as have stage characteristic, not continuous, dividend payment rate is low and so on.

Otherwise, the cross-section regression methods, time serials regression methods and sensitivity tests have not employed to analysis the signaling effect on dividend in Chinese market. In 2005- 2006, the marco policy on dividend and stock market has changed which require the listed company to distribute. Moreover, it is suggested that the CSRC (China Securities Regulatory Commission) establishes rigorous standards of issuing new shares, strengthens financing management to standardize the stock market and increases the proportion of current shares, encourages organization investors to boost up the stabilities and maturities of the stock market ( Qian Zhengchong, 2005)

Chapter 4 Background of government business firms

4.1 Introduction

This chapter provides a simple introduction of the government owned firms in Chinese and some basic rules on dividend distributions.

4.2 government owned firms in China

According to the argument of Krueger (1999), the government owned firms has the pressure to hire politically connected people rather than those best qualified to perform desired profits. Such firms also focus on social and political objectives, rather than profit maximum. There are 4 types of government owned firms in China. In this paper, we considering those government firms which are bound up by the corporate law and do not have the responsibility of social service, they pay dividends to government. In considering an appropriate annual dividend, boards should have regard to the government's preference for profits of government owned firms to be distributed as cash, rather than retained as equity; given this is the prime means by which the Government can realize a return on its equity investment. It is therefore expected that government owned firms will adopt a dividend distribution target in excess of the standard benchmark when appropriate, for example, when after tax profit is not required for capital adequacy or when after tax profit will not be used for investing in value adding opportunities.

On the other hand, Government business is not subject to share market and capital market scrutiny and thus do not have the incentives provided by share price movements and credit ratings to reflect the preference of providers of equity and debt capital. The dividend policy guidelines for government business enterprises, therefore, convey the government's dividend preferences and expectations to provide dividend preferences of government owned firms and expectations to provide government owned firms with the framework and discipline for making dividend payments. Dividends represent a return on the government's equity investment and are paid by GBE's from profits.

Shareholder preferences for dividends have been emphasized in different theories, such as clientele theory (Miller & Modigliani, 1961), ‘bird-in-hand' theory (Gordon, 1963) (Wei and Xiao, 2009). Most of the empirical studies review shareholder preferences for different levels of cash dividends, instead of their preferences for cash dividends over stock dividends. It strongly indicate that Chinese companies prefer to pay stock dividends as the data stated by Wei and Xiao (2009), there are about 34 percent of listed Chinese companies paid stock dividends while some 47 percent paid cash dividends from 1993 to 2006. This feature has been changed, as the innovation begins in 2005 for stock market. According to the data collected from Wind Information 2009, the firm paid stock dividend is only 13.18 percent of total listed Chinese companies from 2006 to 2008. The unique feature in Chinese listed company is the large proportion of non-publicly tradable shares (NPTS), about 1.8830 percent of total shares 2006 to 2008. It commonly holds that the alternative to tradable shares signals good news in long term but the price of the stock will decrease. This paper will only consider the cash dividend, and ignore of stock dividend.

Followings will employ samples from the same market, and divide them into different panels based on the firm characteristic, such as government owned or zero government shares of firms to exam the differences of above variables in government owned firms and zero government shares of firms in signaling model.

Chapter 5 Methods and Results

5.1 introductions

The previous chapters presented the objectives of this thesis and reviewed relative researches on dividend policy as well as some background of Chinese security market. This chapter will specify the research hypotheses and methodology to exam the signaling dividend policy in Chinese market.

Section 5.2 will describes the hypotheses in the context of signaling and agency theories based on previous studies. Section 5.3 discusses the research model and section 5.4 describes data source and collection. Section 5.5 presents the exam results.

5.2 Hypothesis

In signaling model, firms with growth opportunities have higher dividend payment in order to signal to the market that they have better earnings prospects or growth opportunities. Hypotheses 1 is: Dividend is positively associated with current firm profitability and does not signal future profitability.

For the reasons of probability of growth in the firm, ultimate managers prefer to retain a greater proportion of cash flows if they anticipate new project investment and to avoid high cost of external financing. Therefore, this paper predicts a negative relationship between investment and dividend payments.

Hypotheses 2: dividend is negative to investment indicators.

According to the argument of Krueger (1999), the government owned firms has the pressure to hire politically connected people rather than those best qualified to perform desired profits. Such firms also focus on social and political objectives, rather than profit maximum. For the faith that government-owned firms must be less efficient and less profit making in China.

Hypothesis 3 is: the dividend payout is negatively associated with government shares.

Dividend policy of firms is determined by the preferences of the controlling stockholders. Large investors choose to invest in large corporations because it lowers their transaction costs. Since these institutions prefer dividends, the large corporations choose to pay dividends, while the small corporations, owned by individuals, do not. The results show that firm size and liquidity explain the decision of whether to pay dividends well, whereas existing informational explanations (such as monitoring and signaling) explain the level of dividends well.

Hypotheses 4 is: Dividend payout is positively associated with firm size.

some anecdotal evidence suggest a positive association between government ownership and dividends since firms with government ownership have relatively less difficulty raising funds to finance investments and can therefore afford to pay dividends. On the other hand, firms with low government ownership are more likely to have difficulty raising funds and therefore are likely to rely on retained earnings for investments.

Hypotheses 5 is: dividend payout ratio is positively associated with debt in government owned firms and negatively relative in non government owned firms.

Table 1 basic variables employed and hypothesis signs corresponding regression coefficients

Table 1 basic variables employed and hypothesis signs corresponding regression coefficients

variables

Description

hypothesis

DIV

dividends per share/ stock price, where stock price is the year end price.

GOV

percentage of shares owned by government

(-)

SIZ

natural log of total assets

(+)

PRO

operating incomes/ market value of equity

(+)

INV

return on assets

(-)

DT

total debt/ total equity

(+)

5.3 Models

To test the above hypotheses and investigate the impact of signaling model on dividend payment in Chinese listed firms. This paper organizes empirical testing of the following variables in the equation as:

Div t-1= α +β1Govt-1+β2Invt + β3Sizt-1 + β 4Prot + β 5Dtt-1 (1)

Based on the hypotheses 1 and 2, the dividend signals profitability and the investment opportunities. This model will use dividend variable in year 2006 to indicate the profitability and investment in next two years. While based on the hypotheses 3, 4 and 5 which draw from the agency theory and ownership structure, we use the current year to exam their relationship with dividend payment.

Here, Div represents for dividend rate, is the rate of net cash dividend paid in accounting period. Div= dividend per share/ price per share, most of the previous studies employed dividend payout ratio in proxy of dividend per share and dividend yield, such as Rozeff (1982) and Jensen et al.,(1992).

Gov= total government owned shares/ total shares, in order to find out whether the dividend payout ratio is affected by ownership structure, here use the percentage of shares owned by government as it has been used by Gul (1999). For the non-government owned shareholders, the indicator is used as missing value for the above equation.

Siz= natural log of total assets. It is expected to positive to dividend policy, according to the assumption that large firms will pay high dividends to reduce agency costs (Gul,1999).

Dt= total debt/ total equity, which is used by Gul (1999)

Pro=operating incomes/ market value of equity, which is used by Gul (1999)

Inv= return on assets, which is used by Gul (1999)

βi= the correlation coefficient for each variable

5.4 Data Description

5.4.1 Data source:

Dividend, government shareholders of shares, assets, equity, profit and revenue from investment and debt are collected for listed