Behavioral Motives of the Payout Policy Choice: Literature Review

The question of the significance of the payout policy in terms of value creation has been in the works for over 50 years now. These endeavors have led to the establishment of some classic theories that explain the different patterns in a company’s payout policy choices such as the signaling theory, the agency costs theory, the clientele theory and the catering theory. However, the results are not always consistent among different authors, which means that these theories cannot be used universally. Results vary widely among different samples and different time periods. The classic theories assume that all agents on the market are fully rational, which is rather unrealistic since an agent’s actions cannot always be explained by financial theories. These two facts led to the development of the behavioral explanation for the payout policy choice. This approach focuses on the behavioral characteristics of managers that are responsible for the decision-making process in the company. Thus, the payout policy, according to this approach, is considered to be a function of the behavioral characteristics of managers (overconfidence, optimism, risk preferences, etc.) rather than a function of the financial variables. The main difficulty here is how to measure the behavior of managers.This particular article reviews the research that covers the classic and modern theories of payout policy. This article covers the logic of the development of different views on the payout policy. The authors cover articles that test different theories, analyze the main results and conclusions, and investigate the reasons for the development of these theories. The main focus has been on the behavioral approach, which is considered to be the most fruitful direction for future research. The authors also cover the methodology of the existing papers, the variables that measure behavioral characteristics and the results.


Introduction
The classic work of Modigliani and Miller [Modigliani, Miller, 1961], which substantiated the irrelevancy of the dividend policy for company value in the perfect capital market, gave rise to the research regarding the dividend policy in the academic literature. The work mentioned above was based upon some rigid assumptions: the absence of taxes, the absence of transaction costs, the absence of information asymmetry, and the rationality of economic agents, all of which are impossible in real life. Later, these prerequisites were relaxed and thus emerged the signaling theory, which studied the information asymmetry problem [Bhattacharya, 1979], the agency cost theory, which introduced transaction costs and the conflict of various interest groups within a company into the analysis [Easterbrook, 1984;Jensen, 1986], the clientele theory [Grullon, Michaely, 2002;Brav, 2005] and the catering theory, which considered the dividend policy from the point of view of tax benefits for various investor groups [Baker, Wurgler, 2004a, 2004bJiang, 2013]. These lines of research may be called "classic" as long as they are united under the presumption of the rationality of economic agents in the capital markets. In real life, agents do not always act rationally: their personal qualities, subjectivity and biases may influence their decisions [Kahneman, Tversky, 1979]. Particular attention to this fact resulted in the emergence of the behavioral explanation for decisions concerning payments to owners. Within the behavior-based approach, two trends may be distinguished: the influence of the management's attitude to the risk on payments to the shareholders [White, 2012;Caliskan, Doukas, 2015] and the influence of the degree of self-confidence or overconfidence on payments [Ben-David, Graham, Harvey, 2007;Deshmukh, Goel, Howe, 2013]. The more managers are risk averse and the less self-assured they are, the less they are willing to accept high risk and highly uncertain investment projects, which would consequently release resources for payment to the owners. Nevertheless, the results of the authors' research are different. It has been found that the degree of risk aversion has a positive [Caliskan, Doukas, 2015] as well as a negative effect [Sundaram, Yermack, 2007;White, 2012]. The answers to the question of influence on the level of self-confidence towards the dividend policy were also different: it may be negative [Ben-David, Graham, Harvey, 2007;Deshmukh, Goel, Howe, 2013] as well as positive [Wu, Liu, 2011]. Furthermore, there may be no influence at all [Bouwman, 2010]. The structure of the paper is as follows: the first section considers the classic approaches to the explanation of the dividend policy and their basic assumptions. The second section considers the logic of emergence and the main conclusions of the behavioral concept.

Classic Theories of Payout Policy Choice
Payout policy choice is one of the most important in corporate finance, along with investment decision and a resolution for the choice of finance sources. As long as an investor may benefit from possessing company shares by way of dividends or growth of its market value, such a choice largely defines the investment attractiveness for the shares of a certain company. The problem of defining an optimal structure of payment to owners gained the attention of finance researchers approximately 50 years ago. The issue of the importance of the dividend policy for a company's value and the factors defining such a policy were brought up in the work of Modigliani and Miller, where they made the conclusion that the resolutions concerning payments to owners do not influence enterprise value [Modigliani, Miller, 1961]. Instead, they emphasized that the welfare of a shareholder is defined by the cash flows from accepted investment projects rather than by the manner of their distribution among the shareholders. These conclusions were made on the basis of rigid assumptions that define an "ideal" market: the rationality of investors, the absence of transaction costs and taxes, the absence of agency conflicts, etc. In the years that followed, researchers tried to verify the sustainability of the result obtained by Modigliani and Miller. Researchers formed different conclusions about the influence of payout policy choice on owners regarding company value. Modigliani and Miller's hypothesis was proven true for US companies using research carried out on long-term segments from the mid-1980s [Black, Scholes, 1974;Miller, Scholes, 1982]. However, later research [Siddiqi, 1995] and cases from Australian companies [Ball, 1979] rejected this hypothesis. So, the research failed to uncontroversially confirm the results substantiated by Modigliani and Miller. If payout policy choice does influence company value, then which policy would be the best possible? How does payout policy influence company value? Further research into this sphere of corporate finance has focused on these issues. They have tried to explain how payout policy choice may influence company value and what the reasons for such influence are. The researchers have gradually made the above-mentioned assumptions less rigid and offered theories that explain the dividend policy. Here we shall elaborate on those theories.

The Signaling Theory
It is clear that in the real market, economic agents do not have equal access to information. First of all, this concerns the company managers and investors in the securities of such companies. The managers possess more precise information on the company's financial and economic performance and its prospects. Furthermore, they may Electronic copy available at: https://ssrn.com/abstract=3346275 Higher School of Economics 105 influence the information available to the investors. Thus, the presumption of the absence of asymmetrical information is not realistic, which is exactly the point studied by the signaling theory. The underlying concept of the signaling theory is that managers use dividends to signal the market about the present state of the company in order to mitigate the information asymmetry problem. As a general matter, an increase of payments to the owners may be indicative of the management's confidence in the future flow of income and, consequently, in the possibility to maintain a higher level of payments in the future [Bhattacharya, 1979;Miller, Rock, 1985]. The authors come to contradictory conclusions regarding the signals that are contained in the announcements of payments to owners. The research was carried out using the Event-Study methodology, which was based on US companies from the 1960-70s, and showed that announcing the dividend payment is a signal for the market about the future profits of the company [Aharony, Swary, 1980;Brickley, 1983;Michaely, Thaler, Womack, 1995]. Research workers who had applied the regression analysis [Woolridge, 1983;Bali, 2003] made the same conclusion. Nevertheless, later research based on data taken from the 1980-90s found no evidence in the USA [DeAngelo, DeAngelo, Skinner, 1996;Benartzi, Michaely, Thaler, 1997;Grullon, 2005] or in Japan [Conroy, Eades, Harris, 2000] that validates the signaling hypothesis This fact testifies that in the course of time during which the information component of the dividend payments was reduced, the market did not respond so actively to announcements of change or to the initiation of payments to owners. This tendency coincided with a reduction in the share of dividends in aggregate payments to owners and with an increase in the share of stock repurchase [Fama, French, 2001]. For that reason, it was deemed necessary to also check the signaling concept against that of stock repurchase. The researchers agree that stock repurchase does not carry the information on future company profits to the market [Ikenberry, Lakonishok, Vermaelen, 1995;Grullon, Michaely, 2004]. Unlike dividends, which may be a signal of sustainable profit growth that will be preserved in the future, the increase of payments via the stock repurchase is rather perceived as the signal of a one-time shock in the current profit by the market [Guay, Harford, 2000;Jagannathan, Stephens, Weisbach, 2000]. In this case, stock repurchase does not show any signals towards future profits. Instead, the information carried by stock repurchase signals that the management considers the shares to be underestimated by the market in comparison with their fundamental value [Ikenberry, Vermaelen, 1996;Grullon, Michaely, 2002].

Agent Explanation
Another unrealistic assumption is the idea of the absence of so-called agency conflicts -the conflicts of interest for the various groups of economic agents. Suppose that a company has a significant amount of free cash in its accounts, while at the same time, the set of investment projects to be implemented is limited. In such circumstances, a manager may try to use this money to implement some unsuccessful projects or to cover for non-productive operational expenses. In such an occasion, the shareholders would prefer to withdraw the free cash from the manager by paying them to dividends or by stock repurchase. Empirically, this theory has been substantiated for US companies from the 1970-80s. In these companies, the larger is the number of shareholders, the less is the number of shares at the insiders' (i.e. managers') disposal [Jensen, Solberg, Zorn, 1992;Saxena, 1999], while the higher are payments to the owners [Rozeff, 1982;Alli, Khan, Ramirez, 1993]. Thus, the shareholders have sufficient opportunities to influence the level of payments. In the 1990s, such dependencies maintained stability [La Porta, 2000;Gugler, 2003] but in the 2000s, the authors made different conclusions concerning the influence of shareholders on the dividend policy. Research conducted with Australian companies confirmed that a more efficient corporate governance system results in an increase of payments to owners [Yarram, Dollery, 2015] while research from Malaysian companies showed that the dependence is inverse where an efficient corporate governance system is a compensation for low dividends [Benjamin, Mat Zain, 2015]. This change in dependence may be explained by evidence that showed in times of crisis, managers reduced the payment of dividends in order to increase the amount of internal resources that were necessary for implementing investment projects due to the increased cost of borrowing [Bliss, Cheng, Denis, 2015;Floyd, Li, Skinner, 2015].

Clientele Theory
Various groups of investors, such as individuals, funds including investment and pension funds, other companies, etc., invest their money into the shares of a company. The revenues from various investor groups may be taxed at different tax rates. Additionally, income in the form of dividends and income in the form of market value growth are also taxed at different tax rates. Thus, a decline in the presumption of the absence of taxes resulted in the emergence of the clientele theory, which introduces various groups of investors who benefit from receiving income from owning shares in one form or another into the analysis. The companies then try to satisfy the demands of these groups. The clientele theory has been verified regarding dividend payments. Research carried out on US companies for the period from the 1960s-2000s confirms that the clienteles of different taxation schemes may influence the amount of dividend payments. Thus, for example, if the company shareholder is an institute with a more advantageous taxation dividend, such a company will increase their payments to the shareholders in exactly the form of dividends and the shares' dividend yield will then start growing Electronic copy available at: https://ssrn.com/abstract=3346275 Higher School of Economics [Dhaliwal, Erickson, Trezevant, 1999;Elton, Gruber, 1970;Pettit, 1977;Denis, Denis, Sarin, 1994]. This consistent pattern is also confirmed by research using companies from other countries [Ang, Blackwell, Megginson, 1991;Short, Zhang, Keasey, 2002]. In the case of the existence of taxes, stock repurchase becomes more beneficial for individual investors as long as the country's tax rate for dividends is higher than the tax rate for capital gains [Allen, Bernardo, Welch, 2000;Graham, Kumar, 2006;Dahlquist, Robertsson, Rydqvist, 2014]. Thus, the clientele theory has also been substantiated for stock repurchase because the company will increase payments to the shareholders in the form of stock repurchase if the growth of the asset market value of predominant shareholder's income is taxed at a lower rate.

Catering Theory
Relaxing the presumption of the absence of taxes also resulted in the clientele theory being supplemented with the catering theory. Investors may have demand for the shares of the companies that pay dividends [Baker, Wurgler, 2004a, 2004b. Consequently, it should be reflected as a difference in the value of the company shares which of them are dividend payers and which of them are dividend non-payers. This is how the "dividend premium" is formed -the difference in the value of shares of the dividend payers and the dividend non-payers; the authors defined it as the difference in the logarithms of the corresponding Market-to-Book Ratios. The managers, in their turn, try to satisfy this demand by paying dividends when investors evaluate the dividend-paying companies more and by not paying dividends when investors prefer dividend non-payers. Thus, the authors propose that the propensity to pay dividends depends on the "dividend premium" that is embedded in the share value, which was confirmed by empiric tests. It is important to emphasize that the authors studied the topic of dividend payment itself rather than the amount of dividend payments. By so doing, they have not found an explanation for the change in investor demand within the clientele theory. Baker and Wurgler's theory is confirmed for developed economies [Li, Lie, 2006;Ferris, Jayaraman, Sabherwal, 2009;Baker, 2013] as well as for emerging economies [Dong, Liu, 2016;Wang, 2016;Tangjitprom, 2013]. In fact, in the 1990-2000s, companies from the abovementioned countries monitored the quantity of investor demand for the shares of dividend payers and adjusted their payout policy in accordance to the change in demand. However, there is a group of research that overturns this theory on the basis of similar data from companies in the USA [Julio, Ikenberry, 2004;Denis, Osobov, 2008], China [Zhan, 2016] and South Korea . Perhaps such a difference in results is related to an imperfect methodology when verifying this theory. This theory was also verified later from the point of view of stock repurchase [Jiang, 2013;Kulchania, 2013]. The authors concluded that the existence of the repurchase premium indeed has a positive effect on the possibility of carrying out the repurchase and on the continuation of the tendency to pay dividends through the repurchase. Furthermore, the hypothesis of dividend payment and share repurchase interchangeability was proven true: the probability of repurchasing shares depends positively on the "repurchase premium" and negatively on the "dividend premium". The amount of repurchase also depends on these premiums. Next, we turn to the analysis of works devoted to the behavioral motivations for payout policies.

Behavioral Explanation of Payout Policy Choices
The behavior-based approach emerged with the paper by Kahneman and Tversky that was dedicated to finding various biases which economic agents may possess [Kahneman, Tversky, 1979]. This approach is focused on the behavioral characteristics of managers and investors who can influence the decision regarding the dividend policy, and who do not always act rationally. The previous theories focused on the financial characteristics of companies and are a thoroughly studied field of expertise. Two explanations for the dominance of these theories on the behavioral aspect may be distinguished. First, it is easier to measure, aggregate and research financial indicators, while behavior-based characteristics may be measured only indirectly. Second, according to different surveys of company managers [Baker, Powell, 2012], current financial indicators, in particular, are the driver for making payout decisions for shareholders. The reason for the emergence and development of this branch of corporate finance was the discontent with the assumption of the complete rationality of economic agents. By the 1990s, enough evidence had been accumulated proving that agents are not always rational. Experiments by Kahneman and Tversky showed that people may be subject to psychological biases including overconfidence, anchoring, conformism, etc. In financial markets and in the economy in general, it is expressed as periodic bubbles, the everyday losses of investors (in the case of complete rationality an investor gains income), the managers' failure to sometimes accept projects profitable from the point of view of the shareholders and so on. Such evidence resulted in the necessity to include agents' irrationality in the analysis. Thus, this approach considers the influence of the characteristics inherent in the managers who manage the companies on strategic decisions rather than the influence of the indicators typical of a company acting as an agent of economic activity as in the classic theories. The greatest challenge for the researchers of behavioral finance is the quantitative measurement of managers' and investors' behavioral characteristics. Nevertheless, for the several decades that this branch has been developing, several approaches to their measurement have been laid Electronic copy available at: https://ssrn.com/abstract=3346275 Higher School of Economics out. Furthermore, we are going to consider the behavioral characteristics of managers and then move on to some properties of the behavior of the investors. One of the behavioral characteristics that has been most thoroughly investigated is optimism or overconfidence. This characteristic may be measured in two ways. The first way consists in calculating the period during which managers owns a stock option for the shares of the company with whom they are employed [Malmendier, Tate, 2005;Deshmukh, Goel, Howe, 2013;De Cesari, Ozkan, 2015]. If a manager exercises the option within the year in which the option expires, despite the fact that for a long time the shares' value has been exceeding the exercise price, such a manager may be considered optimistic or overconfident. This logic is based on the fact that such managers count upon a steady growth of their company's share price and for this reason they do not exercise the option until the last moment. The second way to define a manager's optimism is to search for key words such as "overconfident" or "optimistic" along with their synonyms and antonyms in interviews with these types of managers or in mass media materials about the companies managed by them [Malmendier, Tate, Yan, 2011;Andreou, 2016]. This method may be more reliable than the first one, but it involves a greater effort of labor and a possibly subjective assessment of a specific manager. The authors of the research on overconfident managers share the opinion that such managers are prone to increasing investments, especially those of high risk and investments in research and development [Hirshleifer, Low, Teoh, 2012;Fenn, Liang, 2001;Deshmukh, Goel, Howe, 2013]. As a result, this type of manager has fewer funds to pay dividends to owners during a current period and thus the payments decrease. Also, such managers consider external financing as the more expensive one in comparison with internal financing and therefore, they do not attract borrowed capital for project implementation. In spite of all else being equal, overconfident managers who pay smaller cash dividends are more inclined to make stock repurchase because they think that at the present, the shares are underestimated and there is room for growth [Fenn, Liang, 2001]. Thus, the influence of a manager's overconfidence on aggregate payments may be both positive and negative. Another behavioral characteristic frequently occurring in research is a manager's attitude to risk. In academic literature, this characteristic is inseparably associated with the manner of payment to such a manager. So, the attitude to risk is measured as the ratio of a certain manner of payment to that of a general payment to a manager. Also, the delta coefficient of the options at the manager's disposal may be used as a measuring instrument that shows the option's rate of risk. If a manager's payment is pegged against the market value of the debt obligations of a company which depends negatively on the risk, the manager will aim to reduce the risk and will be more risk-averse [Caliskan, Doukas, 2015].
Based on research findings, companies with these types of managers pay more dividends because the managers accept a smaller number of projects and thus, receive the available funds. The companies may also establish a payment of dividends for the shares, distributed on the basis of management incentive programs (restricted stocks units, hereinafter referred to as "RSU") and this also slackens the CEO's appetite for risk and consequently results in the increase of dividend payments to shareholders [Minnick, Rosenthal, 2014]. If a manager's compensation is based on the market value of the company shares and the payment of dividends on RSU is not provided for (equity compensation, option programs), then they have no such risk limitations and the manager may accept higher risk projects in order to ensure growth of the share value [Burns, McTier, Minnick, 2015;Douglas, 2007;Geiler, Renneboog, 2016]. Such managers, on the contrary, make smaller payments in the form of dividends, but provide for larger payments in the form of stock repurchase. Nevertheless, growth of payments from stock buyback often does not cover a decrease in cash dividends, and, for this reason, the aggregate payments in such companies are smaller [Cuny, Martin, Puthenpurackal, 2009]. Other researchers also confirm that the dividend policy dependent on the type of the manager's compensation may be sensitive to the specification of the dividend policy [Core, Guay, Larcker, 2003]. A manager's confidence in their position in the company (confidence that they won't be fired) may also influence the payout policy choice. So, managers who are not confident in their position will try to smooth dividends, that is, to maintain them at a stable level or to provide a minimal growth [Wu, 2016]. In other words, they undertake "standard" obligations that, in their opinion, they are capable of fulfilling [Cyert, Kang, Kumar, 1996]. Nevertheless, some studies, on the contrary, show that managers who are apprehensive of uncertainty and losses try to pay greater sums as dividends for the current period [Breuer, Rieger, Soypak, 2014]. They act this way in order to avoid dismissal if, in case of an external shock, they have to reduce dividends. On the other hand, if managers are sure that they will not be fired, they will increase dividends at a faster pace [Jo, Pan, 2009]. Now let's consider some behavioral characteristics of investors which may influence the level of a dividend payout. One of the main characteristics of an investor is their preference. Some investors may prefer stable companies who pay dividends. This may be due to the fact that an investor aims at gaining income in equal parts within a long time horizon in order to be able to spread out its consumption and to avoid distortion when consumption is high in the current period and it is necessary to sharply reduce it in the next period [Shefrin, Statman, 1984]. Companies may monitor the investors' demands if the companies have dividend payers who trade at a higher price than non-payers, which would mean that there is an investor "demand for dividends" and then it would become beneficial for Higher School of Economics the companies to start paying dividends, or to organize the repurchase of shares [Baker, Wurgler, 2004a, 2004bLi, Lie, 2006]. Lintner, in his classic research, makes the conclusion that companies smooth dividends not in order to increase the company value, but rather to raise the investors' satisfaction [Lintner, 1956]. In order to smooth dividends in the case of positive or negative net profit shocks, investments and debt level are used (in the case of a positive shock, the company uses a major part of the additional funds for investments or debt repayment and just a small part is used to increase payments to the shareholders) [Lambrecht, Myers, 2012;Hoang, Hoxha, 2016]. Another behavioral characteristic of investors is their focus on the indicators of previous periods, or anchoring.
Investors react exceptionally adversely to the news of a decrease in dividends or of an increase that was less than expected. This is confirmed by numerous cases of a fall in company stock after the announcement of negative news that is related to the dividend policy [Aharony, Swary, 1980]. In this case, the investors may use not only the previous dynamics of payments made by the company, but also, the average data of the industry or the market as a reference point as well. Based on a completed analysis of the works dedicated to behavior determinants of the dividend policy, one can make some preliminary conclusions about the influence of the previously mentioned determinants on payments to shareholders (Table 1). Focus on previous results n/a n/a Given the information in Table 1, it can be assumed that further development of the behavioral concept may progress towards an explanation for the variation in the repurchase of shares. In addition, the focus can be switched from the behavioral characteristics of the management to that of the shareholders as the main beneficiaries of payment policies. In order for this to be achieved, a methodology for calculating such indicators should be developed. Another promising area for further research would be to test methods aimed at reducing the impact of behavioral characteristics on the adoption of strategic decisions in a company. This would help protect shareholders from the negative consequences of excessive self-confidence or an appetite for power from the side of the management.

Conclusion
Thus, the diminution of the basic assumptions used in the work of Modigliani and Miller led to the emergence of theories that prove the existence of the impact of payout policy on a company's value by giving signals to the market, through the resolution of agency conflicts, and by satisfying the interests of various investor groups. None-theless, all these theories are based on the same premise of the rationality of the economic agents who participated in the analysis. The weakening of this premise resulted in the emergence of the behavioral approach, which can be used to explain the payout policy.
The authors of these studies share the point of view that overconfidence, enhanced risk disposition and high productiveness may produce a negative effect on the payment of dividends. However, the influence on stock repurchase was not considered or it produced the opposite results as expected. Whereas the behavioral characteristics of investors, which are mainly focused on defining the demand for shares of the companies paying dividends, on the contrary, have a positive effect on payments: if investors have a demand for dividend payers the payments of such companies grow. At the same time there is a series of restrictions in the analyzed works: 1. The authors differ in opinion regarding the influence of behavioral indicators on aggregate payments (on the amount of dividends and stock repurchase) because the influence on certain components of payments is multidirectional.
Higher School of Economics 2. The influence of behavioral characteristics on the speed of adjustment of the level of dividend payments to the targeted level has not been studied. 3. The catering theory only studied the act of making payout decisions but it has not been used to explain the difference in the amount of payments for various companies. Thus, the inclusion of behavioral aspects has given further development of the payout theory because, as we have seen, the behavioral characteristics of managers and investors do indeed have a significant impact on the payout policy choice. With all else being equal, self-confident, optimistic and risk-prone managers pay smaller dividends to shareholders. In this case, the behavior of investors, characterized by an increased demand for shares of companies paying dividends, on the contrary, stimulates payments. The authors, in general, also share the point of view that a well-organized corporate governance system of a company encourages an increase in payments and may mitigate the adverse effects of behavioral characteristics on payout decisions. Nevertheless, the indicators of the quality of the corporate governance system were used only in respect to the dividend policy but not in relation to the behavioral characteristics.