2005-17_shef_ac_uk

This series is intended to promote discussion and to provide information about work in progress. The views expressed are those of the authors and should not be quoted without permission.
DISCUSSION PAPER SERIES
Dividends and Politics
* Marc Goergen, University of Sheffield, Management School
Steven Bank, U.C.L.A. School of Law.
Brian Cheffins, Faculty of Law, University of Cambridge
Discussion Paper No 2005.17
July 2005
*Address for correspondence:
Professor M Goergen
University of Sheffield Management School
9 Mappin Street
Sheffield, S1 4DT
Tel: + 44 (0) 114 222 3488
Fax: + 44 (0) 114 222 3348
M.Goergen@
Copies of discussion papers can be obtained by contacting the address below
Mandy Robertson
University of Sheffield Management School
9 Mappin Street, Sheffield, S1 4DT
Tel: + 44 (0) 114 222 3380
Email: M.Robertson @
The authors would like to thank Alan Auerbach, Stephen Bainbridge, Jeremy Edwards, Ian Garrett, Paul Guest, Luc Renneboog, David Skeel and Lynn Stout for helpful discussion and comments. We are also grateful for feedback received at a seminar at Tilburg University and at a panel session at the 2004 annual conference of the American Society of Legal History.
Abstract
Influential contributors to debates concerning corporate governance assert that it is impossible to understand key trends without taking politics into account. This proposition has, however, remained largely untested. This paper therefore offers an empirical study of the relation between politics and corporate governance, with the focus being on the determinants of dividend policy in publicly quoted United Kingdom (U.K.) companies between 1950 and the present. The departure point is the well-known “partial adjustment” model of dividend policy, which we augment to take into account the ideological orientation of the party in power and other potentially salient proxies for politics (e.g. tax policy and dividend controls). The model is tested by reference to aggregate annual data on earnings and dividends. The results indicate that the political placement of the party in office lacks explanatory power. Moreover, even when politics manifests itself in regulation explicitly designed to regulate corporate behaviour, political variables generally do not correlate in the predicted direction with dividend pay-outs. The evidence therefore is inconsistent with the proposition that politics shape corporate governance.
I. INTRODUCTION
Is corporate governance affected substantially by the political environment in which large firms operate? Certainly, in countries suffering from political repression and related civil strife, the economic and institutional pre-conditions for the development of large, privately-owned enterprises are unlikely to be satisfied.1 But what about rich, stable democracies? Are prevailing explanations for corporate institutions, based upon market forces, technological developments and the quality of the legal system, necessarily incomplete because no explicit allowance is made for the role of politics? Influential contributors to ongoing debates concerning corporate governance answer “yes”, asserting that it is impossible to get the full story on the modern corporation without taking the political angle into account.2
Despite the assertions made that politics affects corporate governance, few empirical tests have been conducted. Moreover, those done have focused on a single variable – ownership patterns in large companies in rich countries3 – and have to be treated with caution because of small sample sizes.4 This paper therefore offers an empirical study of the relation between politics and corporate governance, with the focus being on the determinants of dividend policy in publicly quoted United Kingdom (U.K.) companies between 1949 and 2002.
1Mark J. Roe, Corporate Law’s Limits, 31 J. Legal Stud.233-71, (2002).
2Mark J. Roe, Political Determinants of Corporate Governance ( 2003); Peter A. Gourevitch, The Politics of Corporate Governance Regulation, 112 Yale L.J. 1829 (2003).
3Roe, supra note 2, at 49-61; Peter Gourevitch & James Shinn, Explaining Corporate Governance: The Role of Politics, 112-14, 126-27, 185-87, 195 (2004, draft manuscript on file with the authors).
4Marc Goergen, Review of Political Determinants of Corporate Governance, by Mark J. Roe, 12 Corporate Governance: An International Review 116 (2004).
Why dividends? Dividend policy and the governance of corporations have traditionally been treated as two separate issues. A strong case can be made, however, that they should be analyzed jointly. For instance, dividend policy can act as a “bonding” mechanism that commits managers to maximize share values. When executives are not residual claimants to a firm’s cash flows, the managers’ interests can diverge from those of shareholders’. Dividends can play a corrective role in this context. A policy of making sizeable and regular cash distributions to shareholders can constrain executive discretion since managers worried about the dividend will want to generate net cash flow and ensure that profits are not invested in projects with poor risk-adjusted returns.5 Dividend policy can also play an important “signaling” function.6 Since dividend cuts can be interpreted by the market as powerful signals of bad news about a company, the failure to meet an anticipated dividend level can activate alternative corporate governance mechanisms that address poor performance or financial distress.
Dividends also constitute a suitable testing ground for a study of the relationship between corporate governance and politics because the economics of dividend policy has been studied extensively. There remain open research questions; the impact dividends have on share prices remains something of a “puzzle”.7 Nevertheless, the consensus is that managers set dividend policy with targets based on profitability in mind but only move partially towards these over time because they are reluctant to cut existing dividends and will
5Frank Easterbrook, Two Agency-Cost Explanations of Dividends, 74 Am. Econ. Rev. 650 (1984); Zohar Goshen, Shareholder Dividend Options, 104 Yale L.J. 881, 887-93 (1995).
6Merton Miller & Franco Modigliani, Dividend Policy, Growth and the Valuation of Shares 34 J. Fin. 411 (1961).
7Fischer Black, The Dividend Puzzle, 2 J. Portfolio Mgmt. 5 (1976); Terry A. Marsh & Robert C. Merton, Dividend Behavior for the Aggregate Stock Market, 60 J. Bus. 1, 1-2 (1987).
only increase pay-outs if a change can be justified in terms of long-run sustainable earnings.8 Correspondingly, with dividends there is a firm economic foundation that can provide a departure point for an assessment of the impact of politics.
The dramatic decline in dividend payments by U.S. public companies during the 1980s and 1990s might lead one to wonder whether dividends in fact do constitute a meaningful corporate governance metric.9 Miller and Modigliani’s well-known theory that the rate at which dividends are paid does not affect returns to investors might do likewise.10 Any such misgivings, however, seem misplaced. In the U.S., dividends have been reappearing since 2000, which implies that the decline in cash pay-outs was a temporary phenomenon.11 With Miller and Modigliani’s “irrelevance” proposition, it relies on restrictive assumptions associated with perfect capital markets, such as the symmetry of information between managers and shareholders and the absence of tax, transaction costs and bankruptcy costs. These assumptions rarely hold fully and finance scholars have shown that under real world conditions dividend policy can indeed affect share prices and firm value.12 Why the United Kingdom? Availability of evidence is one reason. Those who study the behavior of U.K. companies are, by international standards, very fortunate in the quality
8The foundation for the consensus is work done by John Lintner, Distribution of Incomes of Corporations Among Dividends, Retained Earnings, and Taxes, 46 Am. Econ. Rev. 97 (1956). On the degree of acceptance of this model, see Marsh & Merton, supra note 7, at 5-6.
9Eugene F. Fama and Kenneth R. French, Disappearing Dividends: Changing Firm Characteristics or Lower Propensity to Pay?, 60 Journal of Financial Economics 3 (2001).
10Miller & Modigliani, supra note 6.
11David L. Ikenberry & Brandon R. Julio, Reappearing Dividends (July 2004),
/abstract=585703, visited November 29, 2004.
12See Franklin Allen & Roni Michaely, Payout Policy, in Handbook of the Economics of Finance 339, 387 (George M. Constandtinides et al., eds., 2003); James M. Poterba & Lawrence H. Summers, New Evidence that Taxes Affect the Valuation of Dividends, 39 J. Fin. 1397, 1412 (1984).
of accounting data available to them. 13 This is because detailed annual financial data covering most of Britain’s publicly quoted companies is available back to 1949.
Also important is that Britain’s political system is well-suited for testing the impact that politics potentially has on corporate governance. This is because its “Westminster Model,” exemplified by the fusion of the executive and the legislature and by majority governments elected under a “first past the post” electoral system, gives the party in control substantial leeway to implement policies it prefers.14 This feature would not be of great significance if a single party had had a monopoly on power between 1949 and 2002. In fact, however, neither of Britain’s major political parties (the “right wing” Conservatives and “left wing” Labour) dominated exclusively,15 meaning that there was considerable potential for significant shifts in policy over time.
The U.K. also provides a good case study because of its experience with dividends. British public companies have traditionally been more generous in paying dividends to shareholders than their counterparts in other major industrial countries.16 Even during the
13 A.W. Goudie & G. Meeks, Company Finance and Performance: Aggregated Financial Accounts for Individual British Industries, 1948-82 2 (1986).
14On the distinctions between consensus and majoritarian political systems and why these might influence corporate governance, see Gourevitch & Shinn, supra note 3, at 181-85.
15The Conservatives were in office three times, covering from 1951 to 1964, 1970 to 1974 and 1979 to 1997. Labour was the governing party four times, covering from 1945 to 1951, 1964 to 1970, 1974 to 1979 and from 1997 to the present.
16Li-Chin Jennifer Ho & Chao-Shin Liu, An Agency Cost/Corporate Control Explanation of Cross-National Differences in Dividend Pay Out Ratios: Evidence in the 1980s, 12 Int’l J. Mgmt.437, 437-38 (1995); Stephen Bond, Lucy Chennells & Michael Devereux, Company Dividends and Taxes in the U.K., 16 Fiscal Stud. 1, 4-5 (1996); Rafael La Porta et al., Agency Problems and Dividend Policies Around the World 55 J. Fin. 1-33, 14 (2000).
1990s, when only about one out of four U.S. public companies paid dividends,17 more than four out of five of their U.K. counterparts did so.18
Moreover, whatever the predictions made by financial theory, managers of U.K. companies have not thought of dividend policy as being “irrelevant.” Instead, most have believed that a cut in dividends jeopardizes a company’s share price and will make it more difficult to raise new equity finance.19 Shareholders have agreed that dividends are important. The Economist observed in 1979 that the “preoccupation with (dividend) yields can reduce investment analysis to a simple question of whether a dividend is likely to be held or not.”20 The chairman of a powerful fund manager said in 1990 that dividends “are the core of the relationship between management and owners.”21 And a leading investment bank’s top U.K. analyst observed in 2001 that “offering a decent dividend has been essential for raising equity, especially in attracting and maintaining institutional ownership.”22
British lawmakers have also considered dividends important. From the late 1940s onwards, they have at various times used devices such as tax law and dividend controls to try to regulate cash distributions.23 In Britain, in sum, the consensus has been that dividends
17Fama & French, supra note 3, at 9.
18Grzegorz Trojanowski, Ownership Structure and Payout Policy in the UK, (2004), (working paper, Univ. Exeter, School of Business and Economics 2004, Table 4).
19Jeremy Edwards & Colin Mayer, An Investigation into the Dividend and New Equity Issue Practices of Firms: Evidence from Survey Information, (Working paper No. 80, Institute for Fiscal Studies 1986), Table 2. 20To Cut or Not to Cut, Economist, June 9, 1979, at 118.
21L.E. Linaker, Dividend Cuts No Help on Cyclical Trade Difficulties (Letter to the Editor), Fin. Times, December 12, 1990, at 16.
22Eric Uhlfelder, U.K. Tradition of Paying Out Hefty Dividends Proves a Bright Spot in Gloomy Global Equity Scene, Fin. Times(U.S. edition), July 24, 2001, Global Investing, at 28 (quoting H.S.B.C.’s U.K. strategist).
23For examples, see the discussion in sections 4 and 5, infra.
“matter.” Correspondingly, examining the dividend policy adopted by the country’s public companies provides a legitimate test of the proposition that politics constitutes a determinant of corporate governance.
Our departure point is a “partial adjustment” model of dividend policy used regularly as the foundation for empirical analysis of dividend pay-outs. We augment the model so as to take into account the ideological orientation of the party in power and other potentially salient proxies for politics. The model is then tested by reference to aggregate annual data on the earnings and dividends of U.K. public companies covering from 1949 to 2002.
Our results generally contradict the thesis that politics influences corporate governance. The political placement of the party in office, as measured by positions taken on a wide range of issues in their campaign platforms, does not account in any statistically meaningful way for dividend pay-outs. The result is the same even if only economic policies advocated by the governing party are taken into account. Our test of various “secondary” political variables reveals similarly that they lack explanatory power. The results are not uniform, but for reasons to be elaborated upon, the findings that do appear to support the proposition politics dictate dividend policy need to be carefully qualified.
The paper is organized as follows. Section II describes our use of the “partial adjustment model” of dividends. Section III outlines how the ideology of governing political parties is brought into play. Section IV discusses tax. Section V deals with other “secondary” political variables. Section VI summarizes our results. Section VII concludes.
II. THE PARTIAL ADJUSTMENT MODEL OF DIVIDENDS Even if politics has an impact on dividends, managers can be expected to consider the financial logic involved. Correspondingly, a model of dividend behavior that takes politics
into account should have as its foundation well-accepted economic parameters. The obvious departure point is work done in the mid-1950s by John Lintner that laid the foundation for current understandings of dividend policy.24
Lintner conducted a series of interviews with U.S. corporate managers to ascertain the rationale behind the determination of dividend policy.25 His findings from the interviews can be distilled into a series of “stylized facts”:26
1) Managers have some long-term target pay-out ratio in mind.
2) Managers focus on the change in existing pay-out rather than the dividend level.
3) There is “dividend smoothing”: A dividend change is normally caused by substantial and persistent changes in earnings rather than transitory fluctuations.
4) Managers avoid changing the dividend if such a change might have to be reversed in the short term.
Lintner, using his findings as a departure point, formulated a simple “partial adjustment” model of dividends, employing as a working assumption the use of target pay-out ratios by firms. He then tested the model on aggregate data for all U.S. corporations for 1918 to 1954 and found that it fitted the data well.
24Alon Brav et al., Payout Policy in the 21st Century, (Working paper, Duke Univ., School of Business 2003, at 1).
25Lintner, supra note 8.
26On this characterization of Lintner’s work, see, for example, Richard A. Brealey & Stewart C. Myers, Principles of Corporate Finance (7th ed. 2003), 437; Terry A. Marsh & Robert C. Merton, Dividend Behavior for the Aggregate Stock Market, 60 J. Bus. 1, 5-6 (1987).
Over time the basic Lintner model has been redefined in numerous ways. Fama and Babiak made perhaps the most influential refinement by extending it to incorporate a lagged earnings variable.27 The logic involved is that the past matters. Fama and Babiak
acknowledge that the probability of a rise (fall) in the dividend rate should be greatest when current earnings have increased (decreased). But they make allowances for the fact that
managers might change dividends as a result of earnings fluctuations occurring in prior years, even if the likelihood of this diminishes over time.28 Fama and Babiak used firm level data for a sample of 392 U.S. public companies covering a 19 year period to test their lagged
earnings model. They found that their addition of the lagged profits variable did lead to some improvement in the predictive power of the standard Lintner model.
Lintner’s partial adjustment model, as amended to take into account Fama and
Babiak’s lagged earnings variable, can be formulated as this empirically testable equation:29
t t t t t u E cT E cT cD a D +−+Δ+−=Δ−−11)1(λ
(1)
where:
D t is the dividend in period t; ΔD t is the annual change in dividends, measured by reference to the total for a given year minus the total for the previous year;
a is a constant;
27 Eugene F. Fama & Harvey Babiak, Dividend Policy: An Empirical Analysis, 63 J. Am. Stat. Ass’n 1132-61 (1968).
28
Brealey & Myers, supra note 26, at 438. 29 On the derivation of this equation, see Appendix A.
c is the speed-of-adjustment coefficient, with 0 < c ≤ 1;
T is the target pay-out ratio
E t is annual earnings and
u t is an error term that will absorb the impact of all considerations not systematically reflected in the values assigned to the target pay-out ratio and the speed-of-adjustment coefficient.
Testing this equation requires figures on year-by-year changes to dividends and annual earnings. Correspondingly, two sources, a Cambridge/DTI Databank of Company Accounts and Datastream, have been relied upon to assimilate aggregate data for 1949 to 2002. The Cambridge/DTI Databank offers a wide range of financial data for most U.K. publicly quoted companies covering from 1948 to 1977.30 The sample covered companies with shares listed for trading on the London Stock Exchange but excluded firms operating outside the U.K. or engaged in agriculture, mining, shipping, insurance, property, banking and finance.31
Datastream begins providing firm-level data for U.K. public companies as of 1969 but coverage through 1972 is patchy. By 1973, however, Datastream provides figures for 1217 companies, which exceeds the 1116 firms that comprise the Cambridge/DTI Databank
30The raw data in the sample was collated, standardized and refined in Goudie & Meeks, supra note 13. We have relied on their “all companies” figures, set out in Tables 216, 217. For background on the Cambridge/DTI Databank, see G. Meeks, J.M. Wheeler & G. Whittington, The Cambridge/DTI Databank of Company Accounts 9-15 (1998).
31Also, from 1960 onwards, some companies were excluded on the basis of size. For instance, between 1969 and 1974 companies were only included in the sample if they had net assets of at least £2 million or gross annual income of at least £200,000.
sample for the same year. Correspondingly, for 1973 onwards Datastream’s firm-level data has been relied upon to compile aggregate figures.32
To make due allowance for the industrial exclusions in the Cambridge/DTI Databank figures, we compiled alternate samples based on Datastream. One included all companies listed by Datastream for the 1973-2002 period. The other was restricted to firms falling within the industrial sectors encompassed by the Cambridge/DTI data (i.e. manufacturing, distribution, construction, transport and certain other services). Our results did not vary materially depending on whether the “all companies” sample or the sample focusing on the Cambridge/DTI industries was used. As a result, only the results generated from aggregate data based on all companies reported by Datastream are discussed here.33
Lintner, in the empirical study he carried out to verify his partial adjustment model of dividends, relied on “net” (i.e. after-tax) figures for dividends and earnings. This approach is perfectly defensible for U.S. data, given the tax law regime in place. The U.S. has what is called a classical system of tax. Under this regime retained earnings and distributed earnings are treated the same for tax purposes, which means the amount of dividends distributed does not alter a company’s total tax burden and, by extension, post-tax corporate income. Hence, with the data Lintner relied upon it was appropriate to rely on after-tax figures to test his model.
By contrast, other than between 1965 and 1972, when the U.K. had a U.S.-style classical corporate regime, the amount of dividends distributed under the British tax system
32Due allowance was made for the fact that the Cambridge/DTI data is organized around financial years running from April 6th of the year shown to April 5th of the following year whereas the Datastream figures are based around the financial years of individual companies. This has been done by reorganizing the Datastream data to conform with the Cambridge/DTI definition of the financial year. Hence, the aggregate data for “1984” in fact covers firms whose financial year ends between April 6th 1984 and April 5th 1985.
has affected the calculation of profits. For instance, as Appendix B describes, between 1947 and 1957 the U.K. imposed a profits tax under which distributed profits were explicitly taxed at a considerably higher rate than undistributed profits. By virtue of this scheme, the amount of dividends distributed directly affected the tax companies paid and hence the corporate “bottom line”.
Between 1973 and 1997 the situation was the same due to the operation of the Advance Corporation Tax ("ACT"). Under the ACT regime, a company paid a “gross” dividend but withheld a portion on behalf of U.K. tax authorities (the Inland Revenue). This amount paid in tax could be credited against the firm's corporation tax liability for the current year and carried forward to future years if the credit was not used fully. Nevertheless, if a company’s corporate tax liability persistently lagged behind the amount of its ACT payment, the option expired. For this sort of company, to the extent it paid dividends, its aggregate tax liability would have been higher than if it had retained all profits, and its net earnings lower.
To filter out distortions caused by dividend pay-outs, profits have been recalculated on the basis that companies did not distribute any dividends. The resulting figure is known as “zero-distribution profits.”34 For the purposes of this study, the precise manner in which the figure is calculated varies over time, depending on the particular tax rules in place. Table 1 provides a detailed breakdown.
The U.K.’s tax regime also requires an adjustment of the measurement of dividends. Other than the years between 1965 and 1972 Britain had an “imputation” system under which companies paid out more in dividends than ended up in shareholders’ hands. The remainder
33The results for the Cambridge/DTI industries are available on request.
34Luis Correia da Silva, Marc Goergen & Luc Renneboog, Dividend Policy and Corporate Governance 70 (2003).
was deducted at source and paid as tax to the Inland Revenue. Shareholders, in turn, could seek at least partial reimbursement for the amount the company had deducted. With respect to dividends, for most years the Cambridge/DTI Databank and Datastream report net rather than gross dividends, which means the focus is on what shareholders actually received. Still, since the purpose of this study is to ascertain determinants of corporate dividend policy, what is of interest is the amount companies have paid out, including to tax authorities. Correspondingly, the data has been revised to translate net into gross dividends.35
III. BRINGING POLITICS INTO THE EQUATION
Mark Roe is a forceful proponent of the proposition that political conditions affect corporate governance.36 He argues that left-wing countries favor employees over investors, backed by regulation designed to increase the leverage workers possess. Corporate executives, in this milieu, will tend to cater to employee preferences and give shareholders short shrift, thus ensuring that a U.S.-style stock market economy is unlikely to evolve. In making this point Roe acknowledges the standard agency cost problem affecting relations between managers and investors, but draws attention to a worker-related twist.37 Roe asserts that senior executives want to run big firms since doing so increases the resources under their control, thereby yielding greater prestige and power. Still, because executives in effect have all their eggs in one basket, the continued operation of their companies matters greatly to them. As a result, senior managers tend to avoid changes that might put the survival of their firms at risk. As Roe points out, this is an agenda that tallies
35The equation that is used to translate net into gross dividends is D/(1-t
d ) with t d being th
e tax rate on
distributed profits over time.
36See, for example, Roe, supra note 2.
37Roe, supra note 2, 543, 545-60; Mark J. Roe, Rents and Their Corporate Consequences, 53 Stan. L. Rev. 1463, 1468-75 (2001).
with the objectives of incumbent employees. This is because for staff being associated with a large company can create numerous promotion opportunities and “a steady as she goes” ethos will bolster job security. Correspondingly, under appropriate political conditions there is a foundation for an alliance between managers and employees that could leave shareholders out in the cold.
To find out if politics in fact constitutes a determinant of corporate governance, Roe carried out an empirical test covering ownership structure of large companies in sixteen wealthy nations. He borrowed from an index based on ratings compiled by political scientists that ranked governing political parties from left to right on a numerical scale.38 Roe then tested whether political rankings assigned to each country covered in his study were correlated with ownership concentration and found out they were.39 His results thus provide support for the proposition that politics constitute a determinant of corporate governance.40 Roe’s conjectures can be readily applied to dividends. His characterization of managerial preferences implies that corporate executives should prefer to retain rather than distribute profits. This is because executives will have incentives to accumulate undistributed funds (referred to as “free cash flow” in the finance literature) so as to foster corporate growth without being subjected to unwelcome scrutiny via capital markets.41 Also, executives apprehensive of a potential downturn in their company’s fortunes will tend to be
38See Roe, supra note 2, at 563.
39Roe, supra note 2,at 563.
40The sample size Roe used does mean, however, that his results have to be treated with caution. See Goergen, supra note 4.
41Goshen, supra note 5, at 887-88; Victor Brudney, Dividends, Discretion and Disclosure, 66 Va. L. Rev. 85, 95-96 (1980); Michael C. Jensen, Agency Costs of Free Cash Flow, Corporate Finance and Takeovers, 76 Am. Econ. Rev.(AEA Papers and Proceedings) 323, 323 (1986).。