A Study of Relation between Retained Earnings and Earnings Growth for Istanbul Stock Exchange Companies

Transcription

1 Ugur Demir A Study of Relation between Retained Earnings and Earnings Growth for Istanbul Stock Exchange Companies Dissertation submitted in partial fulfilment for the degree of MSc in Investment Banking, University of Salford 2008

2 Declaration and Acknowledgments I declare that no part of this dissertation has been taken from existing published or unpublished materials without due acknowledgement and attribution, and that all secondary material contained therein has been fully and appropriately referenced. Signed Dated I want to thank my supervisor Prof. Glen C. Arnold for his contributions that he showed in patience and belief.

3 Abstract A Study of Relation between Payout Ratio and Earnings Growth for Turkish Stock Exchange Companies Ugur Demir This research aimed to discover the relation between retained earnings and earnings growth of corporations listed in Istanbul Stock Exchange. While commonsense awaits positive relationship recent researches, made on this topic, yield interesting and conflicting results. Academic community found control function of dividends as the most reasonable explanation. In developed countries, for instance the US, dividends helped to decrease agency cost. For Istanbul Stock Exchange companies the result is no different. Retained earnings didn t have a significant positive contribution so far. Still there are some signs of agency costs and dividends seem to be an indication of good industry and good company.

4 List of Contents

5 Introductory Chapter Introduction Shareholders, as well as other things, own the net earnings of the company. Thus, after all liabilities been served, the remaining should be distributed to shareholders in the form of dividends. This is limited by law in order to protect creditors from unexpected circumstances. Companies listed in Istanbul Stock Exchange regulated to retain a small amount of their earnings at the end of every years and keep this amount for a couple of years. Further, dividends are defined as sticky in the academic literature - meaning they don t tend to change drastically over the years monetarily. Managers choose stable dividend payments to indicate a sustainable business. To obtain this, as the economic and business conditions are subject to change, the management keeps some part of earnings as a cushion that will decrease risk of not paying future dividends. Managements increased preference to keep earnings in the company is subject to discussion in financial community. Increased retained earnings and accumulated cash in companies caused investors to speculate about future earnings growth. Prior to credit crunch, in 2007, investors, even managed funds, were demanding more cash distribution from companies. Equity market companies were holding cash worth as much as 20 percent of market capitalization. In that financial climate, even where credits were cheap, companies were eager to keep earnings in the company instead of distributing them to shareholders. (Brown-Humes, 2007) This picture was interpreted different among investors. In the same period, before equity markets hit by another financial crisis, bulls [who expect equity markets to appreciate in value] and bears [who expect equity markets to depreciate in value] were discussing potential earnings growth. Bullish investors were advocating the future earnings growth, which will be boosted by cash in companies balance sheets that is accumulated by retention of earnings. On

6 the other hand bears were arguing with the same reason; accumulated unused cash is signalling lack of investment opportunities. (Cohen, 2007; Brown-Humes, 2007; Riley, 2007) We can tell future earnings growth is the main differentiation between investors. If retained earnings has something to say about growth it will be valuable in equity valuation. If so, what is the contribution of retentions to future earnings growth? Does cash in the balance sheet, accumulated through retention of earnings, really serve shareholders by investing it in more profitable businesses? Is this beneficial to the investors in the end? At first glance someone using commonsense expects a positive relation between retention and earnings growth. As the management has available funds accumulated from activities there should be more investment and there should be more profit that can be served to shareholder in the end. Theoretically there is a linear relationship between retentions and earnings growth (Miller and Modigliani, 1961). Therefore Black (1972) asked himself and to academic community Why do corporations pay dividends? In practice the evidence suggests the opposite. When Arnott and Asness (2003) examined US equity market for long term contribution of retentions to earnings growth they have found a negative relationship in real terms [when inflation is taken into account]. This meant higher dividend paying companies were in fact candidates for higher growth, which is a win-win scenario for investors. This study was followed by Gwilym and others (2006) research, which contributed Arnott and Asness s (A&A) study by including 10 more countries. By adding different countries they were planning to overcome other constraints such as tax issues and any fact that may affect company policies. They achieved similar results with A&A but they point out that retentions or payouts can t be used as a forecast. They found some inconsistency among countries. For some the relation was weak and for some there were other factors that had more effect on earnings growth.

7 Al-Twaijry (2007) has done a similar company based research for Kuala Lumpur stock exchange, Malaysia as a semi-developed country. Al-Twaijry s research results were statistically insignificant, which meant dividend policy and earnings growth was irrelevant. Therefore it was contrasting with both orthodoxy and empirical findings from other equity markets. Empirical studies made so far covers mainly developed countries. Al-Twaijry made a contribution from a country that has different economic constraints than developed ones. The Istanbul Stock Exchange (ISE) from Turkey is similarly an emerging market. It is important to cover much wider area to see if other markets show the similar symptoms or markets have their own unique characteristics. In that sense I will examine the picture for Istanbul Stock Exchange companies. As a developing country there should be lots of positive net present value (NPV) investment opportunities and therefore lots of need for capital. On the other hand, cost of capital is more expensive than in developed countries, so both earning s retention ratio and usage of it should be higher. I will focus my analysis on company basis, like Al-Twaijry s study and also study the result among industries to find if there is similarity within industries. I will use fundamental data in my analysis and ignore market data which is subject to market efficiency issues. This will simplify the analysis of measuring company performance. In my research I will apply regression analysis to measure the relation between retentions to earnings growth. Regression data will be lagged to measure the effect of a retention made in the past. Then I will measure statistical significance of outcomes by t-statistics. I will interpret the results of my analysis in the light of literature that is discussing dividend policy and earnings of companies. Evaluating the results by contrasting with the other studies made so far will help to understand if ISE companies will make any contribution to any theory developed so far. Therefore, it will be a contribution to the empirical literature as evidence from a developing country.

8 My research will be interesting for both academics and for practitioners. Academics may find it valuable as a demonstration of theories about dividend policy. In practice it can be valuable in equity valuation by helping to predict future earnings growth. It can also be valuable for managers when considering their dividend policies. Research Question Questions to be addressed: What is the correlation between payout ratio and growth in Istanbul Stock Exchange companies in Turkey, as a developing country? Do industries have a typical retention-growth relation? Can retentions be useful in equity valuation by shedding some light to growth for a particular company in a particular industry? Statement of Aims To improve my knowledge in relation between dividends retained earnings, and earnings growth. To reach a decision if retained earnings can be used in forecasting future earnings therefore equity valuation. To contribute recent studies that evaluates the relation between retained earnings, and earnings growth, and dividends. Statement of Objectives 1. A critical, rigorous and thorough review of the body of literature that is relevant to the posed research question. 2. A set of recommendations relating to the application project.

9 Research Methodology Dividends have been a focal point from many perspectives. Their usages have been discussed in valuation, as a source of information, and as a control mechanism. Latest studies focused on the control function of dividends. This studies revealed in developed countries dividend payout ratio is positively related with earnings growth. Therefore managers performed better when the funds were limited. (Arnott and Asness, 2003) I laid a chronological course to understand the development of the theory that is accepting dividends as a control utility. Then I looked for the relation between retained earnings [1 Dividends] and earnings growth in Istanbul Stock Exchange (ISE). By using univariate regression and Pearson correlation methods I searched for an answer if there is a similarity between ISE companies and the latest researches. To understand the contribution of retentions on earnings growth I also made an in depth analysis on industries that is covered by National Industrial Index. This aimed to reveal some information to investors in their valuation methodology. Major Findings I couldn t find a significant relation between retained earnings and earnings growth. Information in hand was insufficient to reject the null hypothesis. On the other hand in the in depth industrial analysis I found that companies whose earnings were shrinking were penurious in dividends and companies whose earnings were growing were more generous. Organisation This introductory chapter will be followed by the Literature Review which is followed by Application Review and will be finished with the Conclusion.

10 LITERATURE REVIEW In the academic literature dividends and retentions, have been discussed from many points and with many views. Orthodox financial consensus is that an unpaid dividend should cause an increase in earnings of the company. From a shareholder s point of view, retention of earnings should result an expectation of increase in future earnings. Importance of Dividends In 1956 Lintner published his pioneering research about the distribution of corporate incomes and started the discussion of dividends and their value for investors. This paper was covering empirical studies and included interviews with company management of selected companies. Twenty eight companies selected from a group of well established 600 companies. He chose companies to reflect different factors affecting dividend payments and policy. Lintner stated savings... are largely a by-product of dividend action taken in terms of pretty well established practices and policies. So companies were building their dividend policy first and according to this policy they were distributing dividends and keeping the remaining earnings in the company. While building dividend policy and making dividend payments following list of factors were taken into account: Gross plant and equipment expenditure; usage of external finance; industry; company size; frequency of changes in rates; average earnings on invested capital; average priceearnings ratios; balance-sheet and fund flow liquidity; stability of earnings; capitalization; use of stock dividends, extras, and splits; the size and importance of stock ownership by management and other control groups. Every one of these aspects has formed a focus for discussions, going on academically and professionally. That said, the dividend planned to be paid in a particular year was decided by taking existing rate, company and shareholder interests into consideration. If a change in

11 dividend is needed the new dividend was decided relative to existing rate. Company managers were trying to draw a consistent line of dividend payments over the years. Management believed shareholders prefer incrementally growing dividends and this is also awarded by a premium by the market. Thus management avoided dividend changes that may be reversed soon. They were trying to avoid any adverse reaction from investors resulted by a decrease in dividend in the future The reason to change dividends has had to be strong to give enough motivation to management. Also management has to believe a change in dividend will be appreciated by shareholders and the financial community. In that sense current net earnings after tax meet these requirements as they were announced and followed widely. Management of companies interviewed were feeling the obligation to distribute increases in earnings with compliance to their dividend policy. If there was a substantial decrease in earnings managers believed a decrease in dividends would be justified too. As a result dividends were decided relative to earnings and a significant change in earnings was reflected to dividends. Speed of adjustment was commonly used to draw a consistent path of dividends along with dividend policy. Management reflected only a portion of rise in earnings to dividend, and only if it is necessarily big enough. If achieved performance kept in the following periods, payout rate gets closer to the previous rate over years. If the earnings occur same level they keep their adjustment year by year getting close to previous payout rate with a speed of adjustment depending on company. This way of adjustment gave management more comfort in dividend payments and in some cases the ability to sustain dividends when earnings got lower than before. On the other hand as a result of speed of adjustment the amount of earnings retained increased and gave more slack for managers. Companies were divided into two groups. The first group of companies- forming two third of the list had a definite policy about the ratio of dividends and earnings, and the second hadn t. In this first group most of the companies also defined the speed of adjustment of the ratio of

12 dividends to earnings. They were gradually reflecting increase in earnings to dividends by years till it gets to previous state. Two companies in the sample were paying fixed rate of earnings as dividends. Pay-out ratios were defined by taking companies requirements and shareholders interests into account. They were changing from a minimum of 20 percent to 80 percent and 50 percent was the common payout ratio in general. Except two companies reflecting every change in earnings to dividends in full, other companies were practicing speed of adjustment. Speed of adjustment was changing from company to company and unlike pay-out ratio there was no common figure in this one. Payout ratios and speed of adjustment were shaped by several factors depending on company experience, objectives, and operations. Can be listed in detail as: the growth prospects of the industry and, more importantly, the growth and earnings prospects of the particular company; the average cyclical movement of investment opportunities, working capital requirements, and internal fund flows, judged by past experience; the relative importance attached by management to longer term capital gains as compared with current dividend income for its stockholders, and management s views of its stockholders preference between reasonably stable or fluctuating dividend rates, and its judgment of the size and importance of any premium the market might put on stability or stable growth in the dividend rate as such; the normal pay-outs and speeds of adjustment of competitive companies or those whose securities were close substitutes investment wise; the financial strength of the company, its access to the capital market on favourable terms, and company policies with respect to the use of outside debt and new equity issues; and management s confidence in the soundness of earnings figures as reported by its accounting department, and its confidence in its budgets and projections of future sales, profits, and so on.

13 Companies managements were keen on dividend payments in a way they could cut working capital to meet such requirements. After dividends distributed, if investment opportunities couldn t met with the internal funds available management reassessed remaining opportunities before going for an external financing. Unless their return is sufficient to raise more capital, by issuing stock or using debt, management postpones the investment plan. If this scenario repeats in following years, management retained more of the earnings and omit the dividends even it is against the policy of the company. As a result of managements self protection and conservatism in dividends gap between net earnings and dividends, therefore retained earnings, have increased. Dependence on internal finance had also contributed to the result. The second group of companies, which didn t have well defined dividend policies, in general possessed similar characteristics with the other party. They had similar payout ratios and speed of adjustment systematic that had resulted similar. In an effort to build a predictive model Lintner also measured the correlation between dividends paid with net current earnings and with dividends paid last year. Correlations were with net current earnings and with last terms dividends paid. The model also suggested a fixed positive figure that showed companies tendency to pay dividends. Results are in line with surveys and study reflected conservatism about dividends paid. What did Investors Think about Dividends? Another contribution to dividend literature came from Gordon (1959) in an effort to understand market s view about dividends and earnings. In his paper Gordon was reflecting orthodoxy by stating the most important and predictable cause of growth in a corporation s dividend is retained earnings. He had stood this idea to the algebraic proof of a linear relationship between return on investments and retained earnings of the company. He expected companies invest retained earnings till they reach company s required rate of return

14 and therefore achieve growth. He was expecting managers to act the best way they could on shareholders behalf. He theoretically bound retained earnings with company growth. He accepted an investor can only achieve dividends as a shareholder. Price of the equity is not important in that sense as the next investor would also only receive dividends from her investment. Therefore dividends and their growth were important. He accepted presence of growth is more important than the degree of growth. Amount of earnings growth was not important. His argument was depending on reversing to the mean of the growth rates. Whether the amount of growth is big or small he thought it will be reversed to the mean in the long run. To understand market s view he measured correlation of dividends, earnings, and both dividends and earnings on share price in a multivariate analysis. With this equation Gordon aimed to understand which element is more effective in pricing of equities. His research covered companies from chemical (32), food (52), steel (34), and machine tools (46) industries for years 1951 and Investors didn t appreciated retained earnings as much as they did dividends. He found weak positive correlation between retained earnings and price. Coefficients have changed from time to time but dividend coefficients were always bigger than retained earnings coefficients. Dividend coefficients were subject to change in years. Still they were consistent and didn t bear much error that would diminish their soundness. On the other hand retained earnings coefficients were not consistent with each other. They were differing from industry to industry and with the high amount of errors they bear they appeared to be negative in some cases. These showed even investors didn't accept retained earnings positive every time. Therefore we can accept this as an indication of investors scepticism retained earnings and they were also rewarding dividends by pricing it higher.

15 Gordon found results discouraging as they showed investors preferred dividends instead of investment opportunities. Nevertheless, these results not accepted as disapproval to retained earnings contribution to growth of earnings by academic community. There was still a positive correlation with retained earnings and price of equity. Furthermore, they were reflecting market s opinion, which is subject to market efficiency, and the study had a limited coverage of years and industry. A research (Harkavy, 1953) referenced by Gordon studying the relation between retained earnings and price appreciation draws the same picture as a conclusion. Companies that paid bigger portions of their earnings had higher prices. Companies that retained bigger portions of their earnings on the other hand had more complex relationship. They did not have linear relationship but it was depending on the result of these retentions. These showed investors scepticism about retentions. Dividends in Theory Miller and Modigliani (1961) acknowledge previous empirical studies made by Lintner and Gordon but also found them inadequate in explaining results. Therefore they focused on framing the subject with theoretical explanations. Starting point of Miller and Modigliani was the effect of dividend policy to a share s current price. They showed algebraically, under such assumptions, the dividend policy should not affect the company prospects. Perfect capital markets No transaction is big enough to affect market price, all traders have equal and costless access to information, there are no transaction costs, and no difference in taxing any income such as between dividends and capital gains. Rational behaviour Investors are indifferent between sources of wealth and chooses the one with the most return. Perfect certainty there is complete assurance in any investment s return whether it is bonds or shares. Miller and Modigliani showed under these assumptions whether all earnings of the company is distributed as a cash dividend or retained as a capital for new investment opportunities it has the same contribution to investors wealth.

16 Under these assumptions, every security in the market would have the same return as investors would easily shift to another one with better return. Again with same assumptions dividend policy wouldn t have any effect on share price. The company would be indifferent between distributing earnings as dividends and issuing shares to raise capital or retain earnings. As a result of same reasons dividend policy wouldn t have any effect on return for shareholders too. That is if a company s investment policy is known, with the help of rational and perfect economic environment; there wouldn t be any financial illusions. Investors would come to same conclusion with assessing real facts. Authors proved their claims about valuation taking different valuation methods and founding the same algebraic result in each with the help of basic assumptions agreed. These valuation methods were: the discounted cash flow approach, the current earnings plus future investment opportunities approach, the stream of dividends approach, the stream of earnings approach. Miller and Modigliani emphasized difference between dividend policy and investment policy of the company: A company is not necessarily bound to its earnings for its investments. They argue with Gordon s findings as he oversaw the effect of investments made and he related pricing of the shares only with dividends and earnings. Miller and Modigliani have gone further in proving irrelevance of dividend policy to valuation in idealised circumstances. They argued, even investors have had their own expected returns value of any company would be irrelevant of the dividend policy. Investors expected return would be a distribution covering and valuation of a company would be a distribution too but still be irrelevant to dividend policy. The authors noted a perfect rational investing which will avoid any sort of speculation by using imputed rationality and symmetric market rationality. With imputed rationality authors meant while an investor is a rational investor that chooses more wealth to less in any form, also accepts the market does the same and act accordingly. Symmetric market rationality names every investor in the market as a rational investor and

17 they imputes rationality to the market it s self. In that way all valuations would be rational and avoids any bubbles created in a speculative way. Under uncertainty of future streams of profits and dividends (holding investment policy is not bound to dividend policy) they prove valuation of a company is not relevant to dividend policy. In a world of uncertainty debt financing would be an alternative for the company and investors. If it is more convenient to invest in debts of a company an investor may do so, instead of investing company s shares. An investor buying shares of the company and an investor giving debt to company would be indifferent. If investment policy is known, the return would be known and valuation wouldn t differ because of risk and would be same as an interest return as a return expected from buying shares of a company. They noted the informational content of dividends in their uncertainties too. They gave credit to this reasoning but pointed to the underlying fact of this indication shows investors an expectation of increase in return. Therefore they view this reaction not to dividend change but expected investment or change on return. They didn t support the idea of dividends being an indicator as they may be manipulative or a change in dividend policy only. Authors also relieved the perfect capital markets to a degree that tax issues wouldn t affect investors already as they could choose companies that will deliver their returns in their favour. Some investors might desire dividends and others might capital gains and with the increasing amount of institutional investors without any tax differential importance of the issue were getting smaller. Finally with relief of some assumptions Miller and Modigliani showed dividend irrelevance theorem was not distant from the current situation of market. When we turn Miller and Modigliani s findings upside down we can see why researchers think dividends and dividend policy is still important: Transaction costs, asymmetric information, and tax - depending on investor s situation

18 In Practice Miller and Modigliani s dividend relevance theorem suggests dividend policy and investment policy is not related, at least under perfect market conditions. Fama (1974) made an empirical research to find the degree of relation of the two for companies. In that content he also tested Lintner s model to estimate dividend if it would still work after two decades. Fama noted investment decisions affecting dividend decisions is not against Miller and Modigliani s theorem as company management should decide about optimal investment level. Therefore he accepted any outcome that is confirming optimal investment level is in Miller and Modigliani s theorem s favour. He studied 298 firms for 23 years period covering that is extending Lintner s empirical research and is a good contribution for Miller and Modigliani s theorem. In his research he created two groups of regression models named dividend and investment. In dividend side he used Lintner s model to test the strength of the model by putting different variables besides fore-year s dividend to predict change in dividend. Namely profit available for common share holders, depreciation, and gross national product. On investment side he used a similar investment model and tested it as an estimator for investments made. In investment side he used total output (sales plus change in inventories) of the company, depreciation, gross national product, and profit available for common share holders besides previous year s net investments made. In both groups he used data from years 1946 to 1966 to predict years 1967 and Fama assessed regression results on the amount and strength of error term in regression equation he used. Dividend group results were stronger than investment overall. In dividend group the strongest one was Lintner s original equation with two variables which including preceding dividend and current profit. This equation proved itself in both years and showed consistency.

19 On the contrary investment models were inaccurate. None of the models was superior as none showed consistency. This showed there was a pattern in dividends while there was no significant pattern in investments. Fama continued his research by assessing the relation of two groups. From dividend group he picked the superior Lintner s model. From investment group he have chosen one model including former year s plant and equipment and total output of current year, and the other one with former year s change in plant and equipment and current year s change in total output of the company. He had chosen the investment models not based on superiority but for evaluating a former research. Changes in dividend and investment averages show a trend over time, showed decline and incline in same terms. Fama looked for cross sectional correlations between residuals of the models to evaluate these two variables. He couldn t find strong correlation between dividend model and investment models. Therefore he concluded there is enough independency between these two to support Miller and Modigliani s theorem. The Question In their previous paper Miller and Modigliani (1958) showed that marginal cost of capital is the key in choosing an investment to be undertaken or not. In this sense management of the company may see the earnings of the company as a free capital that can be invested in any project that seems suitable for management. Even its net present value is lower than cost of capital. Black (1976) explains the dividend as a cheap way of funding the company as there is no cost such as underwriting fee. But he also points dividends should kept paying if the company has no good investments to make as it will cause managers to make unwise decisions. He accepts these unwise situations as rare but recent empirical studies shows the other way.

20 In his criticizing article Black (1976) searched for an answer to his question. He asked; Why do corporations pay dividends? He took Miller and Modigliani s theorem as fundamental point of his article and he looked for dividends benefits to evaluate the idea suggested by the theorem. In theory company and investors should be indifferent to dividends as it is independent with company s investment policy and wouldn t affect company s prospects. Under current market conditions the situation is not the same. Black points if a firm pays no dividend it can still transfer funds to its shareholders in other ways like share buybacks and also evade tax [because there is no income tax for share buybacks] thus benefit shareholder in that sense. Therefore value of the company increase as it means more funds to shareholders in need. He supports non-dividend policy as it can be used in the company as capital. It would be a low cost capital as it doesn t bear transaction costs unlike new debt or equity. In this concept he discussed control function of dividends - as it limits managers spending and save investors from unwise investments. He accepted these kinds of situations as relatively rare. Pecking-order theory argues managers prefer to use internal equity [retained earnings] to finance investments (Myers, 1984). We can take it as a sound base to believe in less dividend payments shows more investment opportunities lays ahead. Therefore managers who believes in prospects of the company makes less dividend payments not to miss coming opportunities and this is contributing to the classical view of retention s contribution to growth. Peckingorder theory doesn t conclude anything about earnings growth but as it tries to explain corporate financing in steps of internal equity, external borrowing [debt], and external equity [issuing new shares] we can conclude that retentions give the ability to make new investments easily to management.

21 The Answer A possible answer to dividend question came from Rozeff (1982). Rozeff argued with the idea of restricting dividends only to residuals of investments. In his paper Rozeff suggested agency costs as a determinant of dividend policy. The optimal dividend payout is determined by the demands of the firm s shareholders. Agency costs arise in an owner and agent position. That is the gap between the value of the firm when wholly owned by the insider and partly owned by the insider. When insider sells a portion of the firm - goes public the owner agent conflict starts. In that situation to maximize wealth, optimization of agency costs and monitoring, and bonding costs needed. Insider and outsider owner of the company is determined by the voting rights of the shareholder. (Jensen and Meckling, 1976) Monitoring costs includes auditing facilities running in a company. Bonding costs includes rewarding and contractual facilities made with the management in a company. Rozeff claimed, like monitoring and bonding costs, dividend is another type of facility that reduces agency costs. If the retained earnings are not enough to make an investment due to investment policy external finances have to be used. External finance sources come with their own monitoring and transaction costs. On the other hand dividends decrease agency costs by distributing funds to outsider owners of the company. Hence the optimum level of dividend payout ratio would be the result of comparison of these costs. Rozeff tested this hypothesis in a multivariate regression model including percentage of common stock held by insiders, average growth rate of revenues ( ), forecast of average growth rate of revenues ( ), beta coefficient, natural logarithm of number of common stock holders with average payout ratio ( ) as the dependent variable. He expected natural logarithm of number of common stock holders to have a positive variable and others to have negative in the regression model.

22 Regression model was highly significant and explained 48 percent of the variation in dividend payout ratio. If we focus on insiders and number of common stock holders these results were as expected in the hypothesis. Insiders had decreasing effect on payout ratio and number of common shareholders had positive effect in four of five observed years and they were statistically significant. In an effort to contribute agency cost discussion, Jensen points out the conflict in dividends - paying out cash to shareholders reduce agent s resources. If the management needs additional capital to make an investment and management has to go external resources, this will result management to encounter additional monitoring from these resources. If this is right issues company and its management will be evaluated again, and if this is debt a bond between lender and borrower will setup. Growth of a company, as a result of bonding cost, gives manager s more power, reward and compensation. Thus he claims managers attracted to grow beyond optimal size. He defined free cash flow as the cash flow that is excess of all projects that has positive net present value. He found the main conflict between owners and agents in free cash flow. As a remedy he offered debt creation and buying back company shares. He suggested creating debt and a creating another bond with lender will establish another monitoring and also cause management to become more efficient. He argued with the idea of increasing dividend paid, he taught an increase in dividend payout ratio may be reversed in future and company may face inefficiency again. Optimum debt to equity ratio would be where marginal cost of debt is equal to marginal benefits.

23 He indicated control function of debt would be useful in organizations producing large cash flows but have low growth prospects. He expected free cash flow occur companies and in industries that must shrink rather than grow- companies with lots of opportunities and no free cash flow. He brought his evidence from oil industry that made investments outside the industry such as retailing, manufacturing, office equipment, and mining by several companies. These acquisitions resulted in failure and value lost to shareholders. The Proof Arnott and Asness (A&A) started their research to find a predictive model of growth with relation to retained earnings in order to help practitioners in share valuation. Authors pointed falling dividend payout ratios since 1995 and dropping to a historical low level between 1999 and 2001, meaning retention ratio was historically high. On the other hand price to earnings ratios were historically high too. Meaning investors, in general, were expecting higher rate of growth than happened before. Investors were expecting the high retention ratio should point new profitable investments in the horizon. This was similar to the situation prior to credit crunch, when bullish investors were pointing to accumulated cash in company books that should yield more growth in earnings in the future. Transaction cost to achieve external financing is a fact, therefore points to a non-perfect capital market in that sense. So we can expect investment policy of companies can be affected by capital accumulation and give credit to orthodox expectation of growth. A&A looked back to historical data whether it supports this expectation. They used S&P 500 index data starting from 1871 to They used earnings per share of the index to payout ratio of the index in the regression model to find if the payout ratio can be an estimator of earnings growth. While they analyzed all the years starting from 1871 they focused on the period after the World War II named as the modern period between 1946 and 2001.

24 Authors were surprised with the out comings of research. They first made a regression test between payout ratio and 10 year subsequent real earnings growth. Their analysis was noncovering meaning none of 10 periods were overlapping with each other. As a result they had 7 period from pre-war and 5 for post-war period. Payout ratio coefficient was strongly positive. Meaning, if a company pays a significant amount of its earnings as dividend you can expect real growth in earnings in the long run. Therefore the result was contradicting with the theory built so far. They divided post-war period years to four quartiles in order to study in detail the relation between payout and growth. They figured out the lowest payout is followed by least growth and highest payout is followed by highest growth and the study was also drawing a consistent pattern in between quartiles. In order to find an explanation for the result of the research they evaluated Lintner s research and Jensen s free cash flow. Their first potential explanation is managers attitude about dividends from Lintner s study (1956). As managers were reluctant to change dividends and they tried to be proactive in dividend decision they might have foreseen coming low earning periods using public and private information. Second potential explanation combines Lintner s (1956) sticky dividends and reversion to the mean. In US dividends were decided monetarily so when the earnings are low the payout ratio will be high and will be followed by high earnings period as a result of reversion to the mean. Therefore this will result a sequence of high payout ratio is followed by earnings growth and low payout followed by the opposite way. Third potential explanation is Jensen s (1986) agency cost explanation as a result of managers desire to build inefficient empires.

25 Their fourth potential explanation was the weakness or wrongness of their test. In order to test the strength of their research they also included 5 year horizon tests. They were assessing 5 year test economically less significant as it is a shorter term, but to test the hypothesis s strength they made the same test in a bigger sample. 5 year overlapping sample had the same result with 10 year overlapping significant positive relation between payout ratio and real earnings growth. In an effort to clear potential explanations they focused on mean reversion. They added 10 year lagged earnings growth and 20 years moving average growth in the regression model of 10 year overlapping. The outcome showed that mean reversion has some effect on earnings growth but insignificant while pay-out ratio was still significant in the model. The regression models overall and pay-out ratio becomes more significant in modern period ( ) while mean reversion elements stay the same. Authors point out that the outcome of tests doesn t mean payout ratio versus mean reversion as a predictor. They interpret the result earnings revert to the mean but with a scale depending on pay-out ratio. Higher the pay-out ratio means stronger reversion to mean, especially in the modern era. To test the second potential explanation, of free cash flow hypothesis of Jensen (1986), they acquired private investment rate in the economy by dividing gross private investment by gross domestic production. As a result they got a trend of investment made by companies over time. Comparing private investment rate with retention ratio yield supporting result their correlation was Meaning high retention periods were also high investment periods for companies. This result supports Jensen s free cash flow hypothesis as he argued retained earnings motivates agents to make investment whether they are efficient or not. They used private investment ratio as a predictor for 10 year overlapping earnings growth and the variable coefficient s sign came out negative, which was significant but weak as a model. Then they build another regression model including payout ratio and private investment ratio. Second model was stronger in explaining earnings growth and was again putting pay-out ratio forward. These models were consistent with the pay-out ratio model stating pay-out is positively related

26 with 10 year earnings growth. As a robustness check for mean reversion they measured the correlation between private investment ratio with lagged earnings growth and moving average growth, which used before, and found that correlation with these were less than half the correlation with pay-out ratio. Thus the result was not supporting mean reversion but was putting free cash flow forward. Outcomes of private investment ratio with pay-out ratio is also conflicting with dividend and investment policy irrelevance of Miller and Modigliani as it showed that retention and investment is positively and significantly correlated. Following Arnott and Asness, Gwilym et.al (2006) published their paper extending A&A s research. They made their research about pay-out ratio s relation with real earnings growth and real dividend growth in 11 international markets including US S&P500. Countries included with data collection starting periods were as follow; France (1973), Germany (1973), Greece (1990), Italy (1986), Japan (1973), Netherlands (1973), Portugal (1990), Spain (1987), Switzerland (1973), the United Kingdom (1965), and the United States (1965). They acquired dividend and earnings yield for each country and divided these with consumer price index same as A&A to find real growth in each. They also accounted dividends as reinvested in the market which was a substantial difference with A&A. Depending on the data they divided countries into three groups, period , period , period For period 1 countries they built 10 years and 5 years lagged regression models, for period 2 countries they built 5 year lagged regression models, and for period 3 countries they built 1 year lagged regression models. For further analysis they also built four aggregate indices totalling every country in the period. For period they built equally weighted and value weighted world indices including seven countries. For period between 1990 and 2004 they built equally weighted and value weighted world indices including eleven countries. These tests designed to analyse different

27 countries that have different managerial cultures, financial market histories, and corporate and individual tax regimes (Gwilym et.al, 2006) Results of the regression models, listed above, were positive for all countries but Italy. Italy s pay-out ratio and real earnings growth sign is negative but its adjusted R 2 value is -0.3, which shows the weakness of the model. Greece s model s explanatory power is also negative (-0.5) but the conflicting results are limited with these two. Explanatory power (adjusted R 2 ) of other regression models varies significantly among countries. But coefficient of the pay-out ratio is significant in general. Therefore we can generalize the results as supporting A&A s research. Tests ran on four world indices yield similar results. Especially value weighted index including seven countries that looked for 5 year subsequent real earnings growth yield positive coefficient that is highly significant and the model had 55.9 percent explanatory power. This result showed the similarity between France, Germany, Japan, Netherlands, Switzerland, United Kingdom, and United States. Authors examined the relation between payout ratio and real earnings growth by scaling countries by their pay-out ratios. As a result they achieved higher pay-out countries yield higher real earnings growth and lower pay-out countries yield lower real earnings growth. Result is not strong but significant. That calls adverse effect of corporate savings. Bernstein and Arnott (2003) states the corporate earnings must grow smaller than GDP - as they only create a percentage of GDP. Jensen (1986) on the other hand argues free cash flow supports and sometimes increases inefficiencies in companies. Maybe corporations by being penurious, decreases overall wealth growth, thus decreases their own growth potential too. They also made a detailed study in real dividend growth, that won t be summarized in here, and found a positive relationship between retained earnings and real dividend growth. This gives a confusing picture for practitioners as it makes valuation of shares confusing. If we focus on real earnings growth, this research s outcome seems to support agency cost hypothesis and gives credit to free cash flow too.

28 Published in the same year with Gwilym and other s research Zhou and Ruland (2006) extended A&A s research in another way. Like A&A it was about US but it was company based unlike A&A. They suspect S&P 500 data might mislead the research as companies covered by the index is changing over time. Also S&P 500 includes companies with a certain size, which may result similarity among them. As S&P 500 is capitalization based group of large companies showing particular feature might affect the result of the test significantly. They gathered data for companies listed in NYSE, Amex, and NASDAQ form industries other than financial services and utilities from years 1950 to They collected earnings growth for subsequent years one, three and five; dividend payout ratio; market value of equity (lessened with natural logarithm); return on assets; leverage; earnings yield (Earnings/Price); past earnings growth again lagged one, three, and five years; and asset growth. They built a multivariate regression model testing above elements. They put other elements besides payout in order to understand effect of size, leverage, market reaction, mean reversion, and investments. The other elements sought to find more explanation for the relation between earnings growth and payout. Apart from main regression model, which included elements above, they ran a univariate analysis between payout ratio and future and past earnings growth. Result of the analysis very strongly (99% confident) supports A&A in all years from one to five. Univariate model also indicates mean reversion between past earnings growth and future earnings growth. A&A found the same mean reversion in their test but it was not significant as in this test. Result of main regression model suggests the same relation between payout with earnings growth and again with a high confidence level (99%). Coefficient decreases when we look further in subsequent years from for one year, for three years, and for five years. This can be explained with managers confidence in earnings for subsequent years (Lintner, 1956). But surprising positive relation between payout and earnings growth is common even over such variety of companies and over such a long period. For short term (one

29 year) past earnings growth had a positive coefficient but for other terms it had negative, which supports mean reversion. They further tested power of payout coefficient if it is depending on different time periods or depending on the industry company belongs. Coefficient varied from time to time especially in the short term. Before 1960 coefficient was smaller and less significant compared to recent periods. After 1960 coefficient was positive for all period and strongly significant in 99% confidence. Industry adjusted payout ratio didn t made any difference and proved the model is robust in any industry high pay-out companies realize higher growth. After 1980 share buybacks has became a new means of distributing cash along with and dividends in US (Skinner, 2006). While after 1980 and before 1980 shows consistency in the model for payout coefficient, effect of share buybacks might be a question. Zhou and Ruland s research was able to test this as its data based on companies. They tested total payout (dividend + repurchase), dividend payout, and repurchase payout for specific period. Result of each model was consistent with previous ones, even share buybacks coefficient was significantly positive in all terms. Share buyback coefficient in five year subsequent model was only significant in 90% confidence level. This gives the idea that share buyback occurs as a result of management confidence. Managers information and confidence in business is more possible to sign positive growth in shorter terms. In longer terms external effects, which they don t have control over, on their business can create resistance in growth. More plausible definition seems free cash flow hypothesis. Positive pay-out ratio and growth relation in log-term puts free cash flow forward. Also mean reversion in earnings growth shows managerial culture of distributing earnings to shareholders even the company enjoy growth in the past. On the other hand if the managers would stick to speed of adjustment they should lower the payout ratio which might decrease the significance of the coefficient.