TAXES AND FIRMS'DIVIDEND POLICIES - National Tax Association

National Tax Journal, Vol. 43, no. 4, (December, 1990), pp. 491-96

TAXES AND FIRMS'DIVIDEND STEPHANM

ABRUTYN*

POLICIES: AND ROBERT

Introduction

SURVEY RESULTS W. TURNER**

shareholders, while if the funds are retained by- the firm, they result in capital gains which receive prefir@ntia-l-tax treatment. This argument leads to the conclusion that no dividends should ever be distributed, a clearly counterfactual prediction. The primary explanation for why dividends are paid even though they lead to higher tax liabilities for shareholders is that dividends send a signal to shareholders about the future prospects of the firm (see Bhattacharya (1979), Mller and Rock (1985), and John and William (1985)). According to this view, "the raising and lowering of dividends communicates information [to shareholders] over and beyond what is provided by earnings reports, forecasts, and other announcements" (Hakansson (1982), p. 415). A major drawback to this explanation, however, is that no one has been able to determine why dividends are better signals than other, less costly alternatives. It is also not clear that the signals dividends send about the firm's expected return are unambiguous (see Black and Scholes (1974), Black (1976), and Ambarish et al. (1987)). A second "old view" explanation for dividends combines asymmetric information with agency costs. As explained by Crockett and Friend (1988), "shareholders are not sufficiently well informed to know whether or not management is acting in their best interests." Shareholders evaluate managerial behavior very little because the information necessary for such evaluation is costly to obtain. If shareholders demand a high dividend payout then managers must generate funds extemally. If the firm remains in the market for new capital, market mechanisms ensure that managers are acting in the best interests of shareholders (see Easterbrook (1984)). Ofer and Thakor (1987) add that, if managers are shareholders, they personally prefer dividends to share repurchases since most companies forbid managers from tendering their shares. Thus, the only way managers can get a

RIOR to the Tax Reform Act of 1986 Pthere appeared to be a large tax penalty to shareholders if dividends were paid by corporations. Why firms paid dividends in such an environment has been one of the most important puzzles in finance; there are at least four distinct explanations in the literature, with different implications for calculating the cost of capital and predicting the effects of changes in capital gains tax treatment. Almost all the empirical evidence regarding these explanations looks at firms' dividend payout behavior in indirect ways. The survey reported here instead asks questions directly of firms' managers in an attempt to identify the motives underlying their payout behavior. There have been a few other surveysi of corporate financial policy, dating back at least to Lintnees (1956) study of dividend policy which resulted in his oft-cited finding that firms tried to maintain a steady dividend rate. A recent survey (Baker, Farrelly, and Edehnan (1985)) had as its major objective to see if Lintner's behavioral model of dividend policy still holds, and also asked whether managers agreed with various statements regarding corporate dividend policy. While there are other surveys regarding financial policy (see the series of articles in recent issues of Financial Management) we know of no others that deal specifically with dividend policy. Our survey attempts to get direct evidence of the importance of tax factors in firms' dividend payout decisions and to distinguish between competing theories of how taxes affect dividends. The "old vieme'of dividend taxation holds that taxes affect shareholders' optimal dividend payout ratios because dividends result in an immediate tax liability for *University of Pennsylvania Law School, PhiladelPhia, PA 19103. **Colgate University, Hwnilton, NY 13323.

491


492

NATIONAL TAX JOURNAL

cash disbursement from their shares is through dividends, and they may be willing to impose a tax cost on other shareholders (and themselves) to get the cash. Perhaps the most widely tested explanation of dividends is known as the clientele hypothesis (see Gordon and Brafford (1980)). Some important groups of shareholders may prefer dividends to capital gains because dividends provide cash flow and, for these shareholders, there is little or no tax advantage to capital gains. The most important group is non-taxable institutions, but individuals with low marginal tax rates and other corporate shareholders are also in the "low-tax clientele." These shareholders will own stock in flrms with high dividend payout ratios while other shareholders (the "high-tax clientele") will invest in firms with low payout ratios. Empirical evidence regarding the clientele hypothesis has been mixed. Studies that find clientele effects include Pettit (1977), Gordon and Brafford (1980), and Scholz (1989). Studies finding contradictory evidence include Long (1978), Litzenberger and Ramaswamy (1979, 1982), Hess (1982), Auerbach (1983), Poterba and Summers (1984), Poterba (1987), and Blume and Friend (1987). The "new view" of dividend taxation holds that taxes are irrelevant because dividend taxes get capitalized into the value of the firm (see King (1977), Auerbach (1979), and Brafford (1981)). The major drawback to this theory is that it fails to deal adequately with the possibility of periodic share repurchases. There is also contradictory empirical evidence (see Poterba and Summers (1984)). The controversy between the "old view" and "new view') continues to be the main focus of research on dividend taxation. The Survey In 1988 a survey was sent to the chief executive offiem (CEOs) of 550 of the biggest 1000 corporations in the United States. The survey contained questions about the dividend payout ratio, shareholder demographics, and other factors. The survey instrument is shown in Exhibit 1. The numbers in brackets are discussed below. The top 1000 corporations

(Vol. XLIII

were identified from the November 1987 issue of Business Week. The name of the company, the name of the CEO, and the city in which the corporate headquarters is located were all listed in the magazine. Specific addresses and zip codes were found in Moody's Handbook (Esposito (1987)) for almost 550 of the companies; a few more addresses were found elsewhere. Of the 550 surveys sent out, 208 were returned. 163 of the returned surveys were complete. 2 The numbers in brackets in Exhibit 1 show the results of the survey. For questions 1 and 5 the numbers in brackets show the percentages of respondents choosing each option. For question 2 the number shows the mean response. 3 For question 3, a full 58 percent of the respondents claim not to have any idea of who their shareholders are; they had all 100 percent of their stock in the "don't know" category. The numbers in brackets next to each option for question 3 are the mean responses of the 42 percent of respondents who indicated a knowledge of the firm's shareholders. For example, of the respondents who estimated the fraction of their shareholders in each of the listed groups, an average of 5 percent of the stock is held by individuals in the 15 percent tax bracket while an average of 21 percent of the stock is held by unknown shareholders. For question 4 there are two numbers separated by a slash. The first number relates what percentage of the respondents ranked each explanation as first in importance. The second number indicates what percentage ranked each explanation either first or second in importance. Of the 18 percent who indicated that an explanation other than those listed was important, no two gave the same alternative. This indicates that many firms pay dividends for reasons unique to that firm. None of the proposed theories of dividend behavior are likely to explain their behavior. Analysis of the Survey Responses The first step in analyzing the results was to group the firms into five categories based on their payout ratios. The intervals were chosen on the basis of the mean


No. 41

TAXES AND FIRMS'DIVIDEND

and standard deviation of the respondents' payout ratios. 4 Table 1 shows the intervals and the percentage of companies falling in each interval. The distribution of payout ratios is approximately normal. Thus few of the respondents had payout ratios as high as the- l@iidugtff-average, which Feldstein and Green (1983) estimated to be 45 percent. This indicates that the corporations surveyed may not be representative of the general corporate population. The survey responses cast doubt on the notion that dividend policy is based on shareholders' tax rates. Only 42 percent of the firms claim to know who their shareholders are. If these claims are to be believed, it is hard to see how the tax laws can play a significant role in the determination of the payout ratio; these managers cannot know whether dividends will result in large tax liabilities for the firm's shareholders, and so cannot be tailoring dividend policy to their shareholders' tax status. Of the firms that know who their shareholders are, an average of only 25 percent of stock is held by "other individuals," a group assumed to have relatively high marginal tax rates. 4 If firms' payout ratios were determined primarily by stockholders' tax status, then the small percentage of stockholders facing high marginal tax rates would be expected to result in high average payout ratios in the sample. This is not the case, casting further doubt on tax-based explanations of dividend policy. Finally, only two of the suggested responses to question 4 of the survey entail taxes. Just 18 percent of the firms listed either of these two as important explanations of their dividend policy.

To see whether firms' answers regarding the determinants of their dividend policy explain their actual payout ratios, firms' reported payout ratios were regressed on a set of dummy variables based on their responses to question 4 on the suiVe-y.- Tti6 -rerr6sgion-" results are reported in Table 2. Each dummy variable equalled one if the corresponding explanation was ranked either first or second in importance, and zero otherwise. The dummy variable associated with the view that shareholders have high marginal tax rates and therefore prefer retained earnings was left out of the regression. The coefficients on the other dummy variables therefore measure the estimated difference between the payout ratio of a firm that considers the high tax rate/prefer capital gains argument as the only important explanation and the payout ratio of a firm that considers only one of the other explanations as important. Only 12.5 percent of the variation in the payout ratio can be explained by the set of independent variables."fhe firms who gave a rank of one or two to the explanation that shareholders prefer retained earnings due to high marginal tax rates do have signffimntly lower payout ratios than the rest of the sample, though, as seen by the high t-ratios on the other dummy variables. If this were a firm's only important explanation, its predicted payout ratio would TABLE 2 ORDINARY LEAST SQUARES REGRESSION RESULTS (T-Ratios in Parentheses) Dependent Variable: Payout Ratio Mean Payout Ratio: 32.4 R-Squared: Explanation Constant

PERCENTAGE Interval less than 2% 2-17% 17-32% 32-47% 47_62% at least 62%

TABLE1 DISTRIBUTION RATIOS

OF PAYOUT % of Firms 5 7 34 39 11 3

493

POLICIES

Dividends act as signals Shareholders prefer dividends b/c of low tax rates Shareholders prefer dividends b/c of non-tax reasons Availability of worthwhile investment projects Other reasons

125

Coefficient Estimate 15.034 (3.24) 10.772 (3.86) 14.067 (2.96) 11.228 (3.35) 6.786

(1.91)

10.947 (2.55)


494

NATIONAL TAX JOURNAL

be 15 percent as seen by the constant term. Other explanations yield increases in the predicted payout ratio of between 7 and 14 percent. Implications for Theories The survey results have a number of implications for the various theories of why corporations pay dividends. The theory that dividends are paid out because they serve as a signal to shareholders is supported by the 63 percent of respondents ranking that explanation as first or second in importance. The signalling theory is also consistent with the statistically significant coefficient on the signalling dummy variable in Table 2 and the 85 percent of respondents who stated that they do not plan to change their payout ratios as a result of the Tax Reform Act of 1986. The regression coefficient indicates that the fims who think signalling is important have significantly higher payout ratios than firms who believe their shareholders prefer retained earnings. While other explanations would also lead to the prediction that the Tax Reform Act would have no effect on payout ratios, the signalling theory does as well. However, CEOs who believe in signalling should also believe that a dollar of dividends will bring about a higher rise in share price than a dollar of retained earnings. Only 21 percent of all CEOs claimed to hold this belief. Agency cost explanations for dividends also receive some support from the survey results. About 44 percent of the respondents indicated that shareholders prefer dividends for non-tax reasons other than signalling. More indirect evidence is provided by looking at the correlations between firinspayout ratios and managers' views on whether dividends or retained earnings increase share prices by more. If managers were acting in shareholders' interests, then firms whose managers thought that dividfft yielded the highest increase in share price would be expected to have high payout ratios, while firms whose managers thought that retained earnings were more valuable to the firm should have low payout ratios. While the correlation coefficients between the

[Vol. XLIII

payout ratio and answers to question 5 of the survey have the predicted signs, they are both less than .15 in absolute value. The survey results contain three pieces of evidence damaging to the clientele hypothesis. First, the fact that so few of the firms know the tax status of their shareholders makes it unlikely that firms are tailoring their dividend policy to any particular tax clientele. Second, only 18 percent of the firms indicated that the marginal tax rates of shareholders was an important factor determining their dividend policy. Third, it is clear that most firms have payout ratios near the mean value. Third, it is clear that most firms have payout ratios near the mean value. The clientele hypothesis predicts that there will be many firms at the extremes of the payout ratio distribution and relatively few in the intermediate range. The tax capitalization view also receives some support from the survey results. Thirty-six percent of the respondents claimed that the availability of worthwile investment projects was an important factor in their dividend policy. This is consistent with the tax capitalization theory's view of dividends as a residual after the firms finance profitable investment projects out of their after-tax profits. The fact that 58 percent of the firms do not know the tax status of their shareholders is also consistent with the tax capitalization view, since the tax status of a fir&s current shareholders is irrelevant. But the most direct evidence about the tax capitalization hypothesis is provided by answers to question 5 of the survey. If the hypothesis is correct, firms should think that there is no difference in terms of effects on share prices between paying dividends and retaining earnings. While more respondents chose this answer than either of the others, the percentage of respondents was only 40. Conclusions Economists are hesitant to pay attention to the reasons firms give for their actions. Theories are most often tested by observing how the firms respond to exogenous changes in their environment. But such tests have been inconclusive regard-


No. 41

TAXES AND FIRMS'DIVIDEND

ing the competing theories of corporate dividend payout behavior. In this note a more direct approach is taken. Surveys were sent to the Chief Executive Officers of 550 of the biggest 1000 corporations in the UnitedC,--States to try to d isti nguis @ '-h among -- --fo@ our om@eeiiig -e@cp 1'4U nations of why firms pay dividends. The 163 complete responses do not provide unambiguous support for any of the four explanations. When asked to rank possible explanations for their dividend payout ratio, 63 percent of the firms ranked a signalling explanation either first or second. Forty-four percent gave a high rank to an explanation consistent with an agency cost explanation. Thirty-six percent gave a high rank to an explanation consistent with the "new" view of taxes and dividends. Only 18 percent of the firms included any explanation based on shareholders' tax rates in their top two explanations; a full 58 percent of the respon-

POLICIES

495

dents claimed not to know the tax status of their shareholders. Thus the tax clientele hypothesis received the weakest support. Firms' answers to a question about whether dividends or retained earnings leicct-t@ hiki@ei'ghhre -pii6i6s - -*@re fnd-@@onsis-* tent with any of the proposed explanations of dividend policy. The answers were nearly evenly divided among the three choices offered. Eighty-five percent of the respondents expected no change in their dividend payout ratio due to the Tax Reform Act of 1986, a result consistent with most of the theories. The survey results strongly suggest that no single theory consistently explains the behavior o.^all firms. Calculations of the cost of capital or the effect of capital gains tax changes that are based on an assumption that one particular theory of dividend behavior is correct should be viewed with caution. Empirical investi-

Exhibit 1 SURVEY OF PAYOUT RATIO BEHAVIOR* 1.

What effect do you think the Tax Reform Act of 1986 will have on your firiwa dividend payout ratio?

[51 ---Retain a higher percentage of after-tax profits [ill ay out a higher percentage of after-tax profits as dividends [851 ----No effect 2. 3.

What is your firies current payout ratio?@-[32.4%] What percentage of your fir&s stock is owned by the following?

[51 Individuals who would fall into the 15% tax bracket for 1988 [251 -Other individuals [371 ----Non-taxable institutions [131 -Other corporations [211 on't know [58% of firms put all their stock in this category] 4.

Please rank the following according to their importance in your firm's decision as to size of the payout ratio (if not considered, leave blank). A belief that shareholders view dividends as a signal of the firnfs financial strength [27/631 A belief that, because of their relatively high marginal tax rates, shareholders prefer that the [3/9) corporation retain earnings [4/91 A belief that, because of their relatively low marginal tax rates, shareholders prefer dividends [25/36] -The availability of worthwhile investment projects (26/441 A belief that shareholders, for reasons other than the tax laws, prefer dividends (or retained earnings) [16/181 -Other (please specify) 5.

Do you think that the share price of your firriys stock will increase more if a dollar of after-tax profits is retained by the firm or if the dollar is paid out to shareholders in the form of a dividend?

[391 ----retained by the firm [211 -paid out as a dividend [401 ---no difference between the two *Numbers in brackets are discussed in the text


496

NATIONAL

TAX JOURNAL

gations attempting to identify the single (,correct" explanation of dividend behavior are unlikely to succeed, since no single explanation is correct for all firms. ENDNOTES ***This paper is based on Abrutyn's undergraduate honors project in economics at Colgate University. The comments of two anonymous reviewers have greatly improved the paper. The authors retain responsibility for any remaining errors. 'We are indebted to an anonymous referee for tellini us of this literature. A number of surveys were returned completely blank accompanied by a letter stating that, due to the large number of similar requests received, the firm has a general policy against filling out surveys. This indicates that these non-responding firms do not have a specific common characteristic in their dividend policy that would bias the results through their exclusion. Other surveys were returned blank except for an indicated payout ratio of zero. Since the survey responses were anonymous, it is impossible to in vestigate fully the characteristics of those firms which did not returm complete responses. 'Thm mean is based on the complete responses. Many of the incomplete surveys indicated a payout ratio of zero. Adding these to the sample yields a mean payout4jt ratio of 2fl.9 percent. is assumed that little stock is owned by individ_ uals whose taxable income is low enough that they face a marginal tax rate of zero. 'Regressions using dummy variables representing whether each explanation was listed as first in importance had virtually - predictive power. REFERENCES Ambarish, R. et al. "Efficient Signaling with Dividends and Investments." Journal of Finance 42 (June 1987) 321-343. Auerbach, A. "Wealth Maximization and the Cost of Capital." Quarterly Journal of Economics 93 (August 1979) 433-446. Auerbach, A. "Stockholder Tax Rates and Firm Attributes." Journal ofPubli4cEconomics 21 (July 1983) 107-127. Baker, H. K., G. FarreUy and R. Edelman. "A Survey of Management Views on Dividend Policy." FinancW Management 14 (Autumn 1985) 78-84. Bhattacharya, S. "Imperfect Information, Dividend Policy, and the Bird in the Hand Fallacy." Bell Journal of Economics 10 (Spring 1979) 259-270. Black, F. "The Dividend Puzzle." Journal of Portfolio Management 2 (Winter 1976) 5-8. Black, F. and M. Beholes. "The Effects of Dividend Policy and Dividend Yield on Common Stock Prices and Returns." Journal of Financial Economics 1 (May 1974) 1-22. Blume, M. and I. Friend. "Institutions in NASDAQ: A Rapidly-Growing Presence." The NASDAQ Handbook (Chicago: Probus Publishing Company, 1987). Bradford, D. "The Incidence and Allocation Effects of

[Vol. XLIII

a Tax on CorporateDistributions."Journalof PublicEconomics15 (Fib. 1981)1-22. Crockett, J. and 1. Friend. "Dividend Policy in Perspective: Can Theory Explain Behavior?" Review of Economics and Statistics 70 (Nov. 1988) 603-613. Easterbrook, F. "Two Agency-Cost Explanations of Dividends." American Economic Review 74 (Sept. 1984) 650-659. Esposito, J., ed. Moody's Industrial Handbook of Common Stocks, Summer 1987 (NY: Mwdy's Industrial Service Service, 1987). Feldstein, M. and J. Green. "Why Do Companies Pay Dividends?" American Economic Review 73 (March 1983) 17-30. Gordon, R. and D. Bradford. "Taxation and the Stock Market Valuation of Capital Gains and Dividends: Theory and Empirical Results." Journal of Public Economics 14 (Oct. 1980) 109-136. Hakansson, N. "To Pay or Not to Pay Dividends." Journal of Finance 37 (May 1982) 415-428. Hess, P. "The Ex-Dividend Day Behavior of Stock Returns: Further Evidence on Tax Effects." Journal of Finance 37 (May 1982) 445-456. John K. and J. Williams. "Dividends, Dilution and is-;es: A Signalling Equillibrium." Journal of Finance 40 (Sept. 1985) 1053-1070. King, M. Public Policy and the Corporation (Undon: Chapman and Hall, 1977). Lintner, J. "Distribution of Incomes of Corporations Among Dividends, Retained Earnings and Taxes." American Economic Review 46 (May 1956) 97-113. Litzenberger, R. and K. Ramaswamy. "The Effect of Personal Taxes and Dividends on Capital Asset Priem. Theory and Empirical Evidence." Journal of Financial Economics 7 (June 1979) 163-195. Litzenberger, R. and K. Ramaswamy. "The Effects of Dividends on Common Stock Prices: Tax Effects or Information Effects?" Journal of Finance 37 (May 1982) 429-443. Long, J. "The Market Valuation of Cash Dividends: A Case to Consider." Journal of Financial Economics 6 (June/Sept. 1978) 236-264. Miller, M. and K. Rock. "Dividend Policy Under Asymmetric Information." Journal of Finance 40 (Sept. 1985) 1031-1051. Miller, M. and M. Scholes. "Dividends and Taxes: Some Empirical Evidence." Journal of Political Economy 90 (Dec. 1982) 1118-1141. Ofer, A. and A. Thakor. "A Theory of Stock Price Responses to Alternative Corporate Cash Disbursement Methods: Stock Repurchases and Dividends." Journal of Finance 42 (June 1987) 365-394. Pettit, R. "Taxes, Transactions Costs and the Clientele Effect of Dividends." Journal of Financial Economics 5 (Dec. 1977) 419-436. Poterba, J. "Tax Policy and Corporate Savings." Brookings Papers onEconomic Activity (1987:2) 455515. PoWtba, J. and L. Summers. "New Evidence That Taxes Affect the Valuation of a Dividend." Journal of Finance 39 (Dec. 1984) 1397-1415. Scholz, J. K. "The Effect of the Relative Tax Treatment of Dividends and Capital Gains on Aspects of Corporate and Individual Behavior." Proceedings of the Eighty-First Annual Conference. 1988 (Columbus, OH: National Tax Association-Tax Institute of America; 1989) 114-120.