ARTICLE IN PRESS

Volume 88, Issue 1, April 2008ISSN 0304-405X

Managing Editor:Contents lists available at ScienceDirect

JOURNAL OFFinancialECONOMICS

G. WILLIAM SCHWERT

Founding Editor:MICHAEL C. JENSEN

Advisory Editors:EUGENE F. FAMA

KENNETH FRENCHWAYNE MIKKELSON

JAY SHANKENANDREI SHLEIFER

CLIFFORD W. SMITH, JR.RENÉ M. STULZ

Associate Editors:HENDRIK BESSEMBINDER

JOHN CAMPBELLHARRY DeANGELO

DARRELL DUFFIEBENJAMIN ESTY

RICHARD GREENJARRAD HARFORD

PAUL HEALYCHRISTOPHER JAMES

SIMON JOHNSONSTEVEN KAPLANTIM LOUGHRAN

MICHELLE LOWRYKEVIN MURPHYMICAH OFFICERLUBOS PASTORNEIL PEARSON

JAY RITTERRICHARD GREENRICHARD SLOANJEREMY C. STEIN

JERRY WARNERMICHAEL WEISBACH

KAREN WRUCK

Journal of Financial Economics

Journal of Financial Economics 89 (2008) 347–374

0304-40

doi:10.1

$ We

Philip M

this pro

Millela

also be

Vanderb

the Uni

of Asma

and the

from su

Christof

Hertzel

Raaball

Scapens

Pradeep

script w

Distingu

Univers� Cor

E-m

Published by ELSEVIERin collaboration with theWILLIAM E. SIMON GRADUATE SCHOOL OF BUSINESS ADMINISTRATION, UNIVERSITY OF ROCHESTER

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/jfec

Dividends and share repurchases in the European Union$

Henk von Eije a, William L. Megginson b,�

a Faculty of Economics and Business, University of Groningen, Landleven 5, 9747AD Groningen, The Netherlandsb Price College of Business, University of Oklahoma, Norman, OK 73019, USA

a r t i c l e i n f o

Article history:

Received 10 December 2006

Received in revised form

26 November 2007

Accepted 29 November 2007Available online 3 July 2008

JEL classification:

G35

G15

F15

Keywords:

Payout policy

Dividends

Share repurchases

International financial markets

Economic integration

5X/$ - see front matter & 2008 Elsevier B.V.

016/j.jfineco.2007.11.002

wish to thank Chuck Chahyadi, Veljko Fo

ebius, and Mark Schoenaker for their resea

ject, and Andrea Beltratti, Bernardo Bortol

for sharing their data on EU privatized comp

nefited from comments offered by semin

ilt University, Concordia University, Florid

versity of Oklahoma, the University of Gronin

ra, the VU University Amsterdam, the Univer

2006 Australian Banking and Finance Confer

ggestions offered by Maria Boutchkova, B

fersen, Chitru Fernando, Marc Goergen, H

, Anjo Koeter-Kant, Albert Menkveld, Igor

e, an anonymous referee, Luc Renneboog

, Bill Schwert, Mehari Tewolde, Vahap Uysal

Yahad, and, especially, Harry and Linda DeA

as finalized while Bill Megginson was the Fu

ished Chair in American Studies and a vi

ite Paris Dauphine.

responding author.

ail address: wmegginson@ou.edu (W.L. Megg

a b s t r a c t

We examine cash dividends and share repurchases from 1989 to 2005 in the 15 nations

that were members of the European Union before May 2004. As in the United States, the

fraction of European firms paying dividends declines, while total real dividends paid

increase and share repurchases surge. We also show that financial reporting frequency is

associated with higher payout, and that privatized companies account for almost one-

quarter of total cash dividends and share repurchases. Our regression analyses indicate

that increasing fractions of retained earnings to equity do not increase the likelihood of

cash payouts, whereas company age does.

& 2008 Elsevier B.V. All rights reserved.

All rights reserved.

tak, Ajay Kongera,

rch assistance with

otti, and Valentina

anies with us. We

ar participants at

a State University,

gen, the University

sity of New Orleans

ence (Sydney), and

ill Christie, Susan

alit Gonenc- , Mike

Osobov, Johannes

, Jesus Salas, Bob

, Wim Westerman,

ngelo. This manu-

lbright Tocqueville

siting professor at

inson).

1. Introduction

Dividends are both pervasive and perplexing. They arepervasive in that companies have been paying regularcash dividends since the dawn of the modern limitedliability company over three centuries ago, and publiclytraded companies in all market economies have beenpaying out large fractions of their earnings ever since.1

Dividends are perplexing (especially to financial econo-mists) because it is not obvious why investors shoulddemand cash dividends. Since the seminal paper by Millerand Modigliani (1961), a vast literature has examined thepayout policies of U.S. companies, and much of the recentresearch has studied both cash dividends and share

1 Dewing (1953, p. 93) notes that an act of Parliament in 1697

formalized the restriction that cash dividends could only be paid out of a

company’s profits, not its capital, clearly indicating that dividends were

important even then. A similar provision appeared in the Massachusetts

Land Bank’s charter, granted in 1739.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374348

repurchases.2 However, relatively little research has yetbeen published examining the payout polices of non-U.S.companies.

This study examines cash dividend payments andshare repurchases, over the period 1989–2005, of compa-nies headquartered in nations that were members of theEuropean Union before May 2004. Besides being theworld’s second largest economy, the EU during this eraprovides an ideal laboratory for studying payout policies,for four reasons. First, it is a grouping of 15 mostlysovereign and highly developed nations that over ourstudy period is becoming increasingly unified economic-ally and politically, but whose corporate governancesystems, taxation regimes, and financial markets remainlargely segmented. Second, while most of the EU countrieshave commercial codes based on civil law, two key nations(Britain and Ireland) have common law codes. Third, thereis significant cross-sectional and time-series variation inthe taxation of dividends and share repurchases over thestudy period, which allows us to directly examine how taxpreferences influence payout probabilities and amounts.Finally, a large majority of the EU nations adopt a newcommon currency, the euro, after 1999, but three membercountries—including the nation with the largest numberof listed firms (Britain)—retain their own currenciesthroughout the study period.

Despite the EU’s global importance, very little pub-lished research has examined cash dividend payments orshare repurchases on a continent-wide basis. Denis andOsobov (2008) examine the dividend policies, but notrepurchases, of companies headquartered in three Eur-opean countries in their sample of six large economies,while LaPorta, Lopez-de-Silanes, Shleifer, and Vishny(2000) include European firms in their global study ofdividend policies. Pinkowitz, Stulz, and Williamson (2006)similarly include many European companies in theirexamination of how cash holdings and dividend paymentsare valued in 35 countries with varying corporategovernance regimes. The only other recent studies wefind are single-country analyses. This relative dearth ofcontinental research primarily results from difficulties inobtaining comprehensive data prior to the 1999 adoptionof the euro.

2 The theoretical and empirical literature from the first 40 years of

modern dividend policy research is summarized in Allen and Michaely

(2003). More recently, Koch and Sun (2004) show that investors

interpret dividend changes as credible signals of the sustainability of

past earnings changes, Morck and Yeung (2005) explore how American

tax policy relating to inter-corporate dividend payments has promoted

dispersed stock ownership in the United States, Grinstein and Michaely

(2005) examine how corporate payout policy (dividends and re-

purchases) impacts institutional share ownership, and Brav, Graham,

Harvey, and Michaely (2005) present survey evidence on the factors that

drive managerial decisions regarding dividend payments and share

repurchases. Intriguingly, DeAngelo and DeAngelo (2006) reexamine the

original M&M dividend irrelevance theorem—which has served for four

decades as the theoretical touchstone for payout research—and find that

it does not hold, even in frictionless markets. In DeAngelo and DeAngelo

(2007), these same authors advocate an alternative method of teaching

dividend policy to students that does not begin with an irrelevance

assumption.

We ask and answer a series of basic questions aboutEuropean payout policies. Are EU dividend payout ratiosand likelihood to repurchase shares higher or lower thanin the United States, and are these ratios the same fordifferent European countries? Do European companiesshow the same declining propensity to pay dividends asFama and French (2001) show for U.S. listed firms? Haveshare repurchases become as, or more, important thancash dividends among European firms, as Skinner (2008)shows has occurred in the United States, and are EUrepurchases driven by the same firm-specific factors ascash dividends? Are dividends and earnings also concen-trating in EU firms, as DeAngelo, DeAngelo, and Skinner(2004) show is happening in American companies? Arecash payouts as highly correlated with the fraction ofretained earnings (versus contributed equity) in theequity capital structures of European firms as DeAngelo,DeAngelo, and Stulz (2006) find for U.S. dividend-payingfirms? Are total payouts becoming more responsive toearnings in companies that regularly pay cash dividendsand repurchase shares, as Skinner (2008) finds for theUnited States? Finally, do the patterns observed in Europesupport any of the competing theories explaining divi-dend payments—especially the agency cost model, whichis the current mainstream favorite, or the catering theoryproposed by Baker and Wurgler (2004a)?

We find that dividend and share repurchase policies ofEU companies are similar in many ways to those ofAmerican firms. For example, the fraction of Europeanfirms paying dividends has also declined in recent years,while total real dividends paid have increased. The totalvalue of share repurchases has also surged in Europe, andnow accounts for over half of the total value of cashdividends—although only one-fourth as many Europeancompanies repurchase shares as pay cash dividends.Additionally, dividends are also concentrating amongEuropean firms and, while the overall propensity of allEU firms to pay cash dividends has declined continuouslywith time, the propensity to repurchase shares hassteadily increased. We find that company characteristicslike size, market-to-book, and profitability that explainthe likelihood to pay dividends or repurchase shares in theUnited States also influence the likelihood that EU firmswill distribute cash to investors. While fewer Europeanthan American firms repurchase shares, dividend pay-ments and repurchases are also complements in Europeand repurchases are significantly more sensitive to earn-ings in the most recent sub-period (2001–2005).

In addition to showing how American and EU firms aresimilar, we also make several new contributions by eitherreporting differences between Europe and America or byexamining factors not considered in the U.S. context. Mostcritically, we show that large-scale share repurchasesstarted much later in Europe than in the United States, buthave grown even more rapidly over the past decade.Additionally, this is the first academic study to examinethe relationship between financial reporting frequencyand corporate payout policies. We find that the averagereporting frequency of EU companies has steadily in-creased over 1989–2005, from 1.2 to 2.4 times per year,and is associated with higher amounts of cash dividends

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 349

paid and shares repurchased. We also examine the impactof privatization on the payout policies of divested firms,and find massive individual and aggregate effects;privatized companies account for almost one-quarter ofthe total value of EU cash dividends and share repurch-ases—but represent barely 2% of the number of listedfirms. Our final result is surprising. We find that thefraction of retained earnings in a European firm’s totalequity is not significantly correlated with the likelihood topay cash dividends, which contrasts with U.S. evidenceshown in DeAngelo, DeAngelo, and Stulz (2006) andinternational evidence presented in Denis and Osobov(2008).

We also present other, equally informative, results forEU payout policies. Older EU companies are more likely topay cash dividends than are younger companies, and olderfirms also pay higher cash dividends. Companies head-quartered in a common law country are more likely to paycash dividends—but are no more likely to repurchaseshares—than are companies headquartered in a civil lawcountry. Higher leverage reduces the likelihood to paycash dividends. Larger cash holdings reduce the prob-ability of paying cash dividends, but increase the prob-ability to repurchase and increase the amount ofdividends paid if the company is a cash dividend payer.Companies with more liquid stocks repurchase more ifthey do repurchase, while companies that have a majorityshareholder repurchase less. Finally, we find no systematiceffects of a country-specific catering variable in EUcompanies, which suggests that catering is not animportant factor influencing European payout policies.

This study is organized as follows. Section 2 motivatesour research based on recent theoretical and empiricaladvances in corporate dividend policy, while Section 3presents our data and sample selection criteria. Section 4describes patterns we observe in European payoutpolicies, and traces the evolution of these policies overtime. Section 5 presents multivariate analyses of thelikelihood that EU firms will pay dividends and repurch-ase shares, while Section 6 does the same for the amountof dividends paid and shares repurchased. Section 7provides robustness tests using alternative (or rearranged)variables, and tests whether the major findings change ifwe add information on new variables that are onlyavailable for 2001–2005. This section also formally testsfor the speed of adjustment to earnings for companiesthat regularly pay out cash through dividends and/orrepurchases. Finally, Section 8 concludes.

2. Motivation and literature review

2.1. What we have recently learned about U.S. payout policy

Recently published research on dividends and re-purchases in the United States has shown seven importantnew findings. First, Fama and French (2001) show that thefraction of U.S. industrial firms paying cash dividends hasdropped sharply over the past five decades, from 66.5% oflisted firms in 1978 (and over 80% during the 1950s) to20.8% in 1999. Fama and French show this decline is due

to two influences: changing firm characteristics and adeclining propensity to pay. In the first instance, thefinancial characteristics of the ‘‘typical’’ publicly tradedcompany have changed since 1978, with many newlistings of firms with low (or negative) profits, highgrowth opportunities, and an asset base tilted heavilytowards intangible rather than fixed assets. These com-panies are most prevalent on Nasdaq, but the character-istics of the typical NYSE-listed firm have also changedsimilarly. The second, roughly equally important influencehas been the declining propensity of all types ofcompanies to pay dividends.

Second, Grullon and Ikenberry (2000), Grullon andMichaely (2004), and Skinner (2008) show a massiveincrease in the number (and total value) of U.S. industrialfirms repurchasing their own shares since 1982, when akey Securities and Exchange Commission (SEC) ruling firstprovided a legal safe harbor for managers implementingopen-market repurchases. This method of distributingcorporate cash to shareholders is both tax-favored and farmore flexible than paying regular cash dividends, andmultiple academic studies—including Dann (1981), Ver-maelen (1981), Ikenberry, Lakonishok, and Vermaelen(1995), Guay and Harford (2000), Jagannathan, Stephens,and Weisbach (2000), Weston and Siu (2003), Maxwelland Stephens (2003), and Grullon and Michaely(2004)—show positive repurchase announcement effects,comparable to dividend increases or initiations. Further-more, Boudoukh, Michaely, Richardson, and Roberts(2007) show that including share repurchases as part ofcorporate payout significantly improves the performanceof asset pricing models. Most strikingly, Skinner (2008)shows that the total annual value of share repurchasesnow usually exceeds that of cash dividends in the UnitedStates, and reports that repurchases have become thepreferred method of distributing cash to investors.

The third, seemingly aberrant major recent finding isthat the total (nominal and real) value of cash dividendpayments and repurchases by U.S. corporations has beenrising inexorably for several decades, and now oftenapproaches 100% of aggregate corporate profits. Westonand Siu (2003) show that the U.S. corporate sector’s cashdividend payout ratio increased from 40% in 1971 toaround 60% in 1990—where it remained throughout the1990s—and finally to 81% in 2001. Including repurchaseswith dividends had little effect on the aggregate payoutratio in 1971, but adding in repurchases brings the payoutratio to 105% in 1998 and to 116% in 2001.

The fourth major dividend policy finding essentially‘‘squares the circle’’ by explaining how the three resultscited above—declining fractions of listed firms payingdividends, an increasing inclination of dividend payingcompanies to repurchase shares, and rising aggregate cashdistributions—can all occur simultaneously. DeAngelo,DeAngelo, and Skinner (2004) show that both dividendsand earnings have become increasingly concentratedamong a relative handful of U.S. corporations over thepast quarter-century. A mere 25 firms now account forover 50% of industrial earnings and dividends, and thehugely increased dividends of these ‘‘high payers’’ swampthe declining tendency of small and mid-sized firms to

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374350

pay dividends. Further, there has been a decline only inthe number of industrial payers since 1978; the number offinancial and utility payers has increased, as have theirtotal real dividend payments.

Fifth, DeAngelo, DeAngelo, and Stulz (2006) show aheretofore unconsidered influence on dividend policy—

the mix of earned versus contributed capital in a firm’sequity capitalization. Rapidly growing young companiesthat have recently executed an Initial Public Offering (IPO)have mostly ‘‘contributed’’ equity capital and pay few orno dividends. On the other hand, the equity capitalizationof mature, highly profitable firms will consist mostly ofretained earnings, and these public companies pay thebulk of cash dividends each year. This life-cycle explana-tion of dividend payments, while intuitive, is nonethelessa radical departure from received theory, which has longbeen based on dividend irrelevance theorems. DeAngelo,DeAngelo, and Stulz also show that, had these 50 high-dividend companies not paid dividends over the pastquarter-century, their cash holdings would equal one-seventh of America’s GDP, they would be debt-free, andtheir managers would be totally insulated from capitalmarket discipline.

Sixth, there is some evidence that cash dividends maybe ‘‘reappearing’’ in the United States. Julio and Ikenberry(2004) show a small, but significant, five percentage pointincrease in the fraction of U.S. industrial firms paying cashdividends since 2001. They also describe a greatertendency for large firms to pay dividends since 1999. Thisrebound in dividend payments is partly explained by the2003 Bush Tax Cut, and partly by the natural maturationof IPO firms that went public during the 1990s. It isunclear whether the dividend reappearance Julio andIkenberry report is permanent or temporary.

Finally, Brav, Graham, Harvey, and Michaely (2005)present survey and interview findings showing thatfinancial executives have a strong desire to maintain thelevel of dividends, but that the link between cashdividends and earnings has weakened. They also showthat many managers now favor repurchases, becausethese provide more flexibility than cash dividends. Thesefindings are corroborated by Skinner (2008), who findsthat the relatively small number of companies that bothpay cash dividends and repurchase shares regularlyaccount for over 50% of total cash payouts. Skinner alsoshows that aggregate payouts have become more sensitiveto earnings for these firms, while the sensitivity of cashdividends has become less significant.3

3 In addition to these published studies, two unpublished working

papers are worth noting. Banerjee, Gatchev, and Spindt (2005) suggest

that the Fama and French model of a firm’s ability to pay dividends fails

to predict future dividend payers, especially when it is applied to larger,

more profitable, and more liquid firms. When several measures of firm

liquidity are included, the predictive ability of the FF model improves

and the improvements are particularly prominent for large firms and

firms with more liquid shares. They conclude that improved liquidity in

the U.S. stock market may have made investors in need of income less

sensitive to dividends as they might now more easily sell their own

stocks in liquid firms. Officer (2006) shows that dividend payments can

be a substitute for weak internal and external governance by document-

ing that predicted U.S. dividend-payers with weak governance are

In addition to the empirical contributions discussedabove, there has been a major addition to the theoreticalliterature on dividends. Baker and Wurgler (2004a, b)develop a catering theory of dividends to explain observedU.S. payout patterns. They assert that companies supplydividends to meet investor demand. Their measure ofdemand, the dividend premium, is the logarithmof the ratio of average market-to-book ratios of divi-dend-payers to non-payers. When this premium ishigh, non-paying firms try to cater to the demand ofinvestors by initiating dividends. Baker and Wurgler(2004a, b) and Li and Lie (2006) find empirical supportfor the catering model, but unfortunately this modeldoes not explain why investors demand dividends in thefirst place.

2.2. What we know about dividend policies of non-U.S.

companies

Compared to the success researchers have achieved inexplaining U.S. payout patterns, the international researchoutput is fairly meager. We are unaware of any multi-national studies examining share repurchases, and findonly three studies presenting truly cross-sectional analysesof global dividend payments. First, LaPorta, Lopez-de-Silanes, Shleifer, and Vishny (LLSV, 2000) offer a ‘‘Lawand Finance’’ (LLSV, 1997, 1998, 2002) explanation forobserved dividend patterns, predicting that dividendsexist to overcome agency problems. In common lawcountries, with legal systems that protect investors,shareholders are able to force managers to pay out freecash flow, whereas they can do this less easily in civil lawcountries. LLSV examine payout policies using cross-sectional data for 4,000+ companies from 33 countries,during the single year 1995, and find strong support forthe agency cost model. Dividends are economically andstatistically significantly higher in common law countries.Second, Denis and Osobov (2008) examine the dividendpolicies of companies headquartered in six major coun-tries using data from 1989 to 2002. They study the UnitedStates, United Kingdom, Canada, France, Germany, andJapan—and show declining propensities to pay in all sixcountries. These authors also test whether the patternsobserved support either the agency cost or the cateringtheory models of dividend payments. They find thelikelihood of paying is negatively related to growthopportunities in common law countries, but positivelyrelated in civil law countries, and conclude this supportsthe agency cost model over the catering theory model.Third, Pinkowitz, Stulz, and Williamson (2006) examinehow cash holdings and dividend payments are valued in35 countries with weak to strong corporate governanceregimes. They find that a dividend payment rate corre-sponding to 1% of a firm’s assets increases firm value by9.80% in countries with weak corporate governance, butby only 4.07% in other countries.

(footnote continued)

significantly more likely to pay dividends than are predicted payers with

strong governance.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 351

While truly multi-national payout research is sparse,several single-country studies have been produced. Threepublished articles (Ang, Blackwell, and Megginson, 1991;Lasfer, 1997; Bell and Jenkinson, 2002) and two workingpapers (Bank, Cheffins, and Goergen, 2004; Renneboogand Trojanowski, 2006) examine how British tax policiesaffect dividend payouts of U.K. firms, while Ferris, Sen, andYui (2006) examine whether the evolution in Britishpayout ratios is as predicted by catering theory. Ferriset al. conclude their results are consistent with cateringtheory, though Denis and Osobov (2008) dispute thisinterpretation. Lee, Liu, Roll, and Subrahmanyam (2004),Cannavan, Finn, and Gray (2004), Christoffersen, Geczy,Musto, and Reed (2005), and McDonald (2001) analyzetaxation’s impact on dividend policies of, respectively,Taiwanese, Australian, Canadian, and German firms.Additionally, Dewenter and Warther (1998) contrastdividend policies of Japanese and U.S. companies andGoergen, Renneboog, and Correia da Silva (2005) examinewhen and why German companies change their dividendpayout. The only non-U.S. studies examining sharerepurchases we are aware of are Ginglinger and L’Her(2006) and Ginglinger and Hamon (2007), both of whichstudy French firms.

Finally, a small corner of the empirical privatizationliterature also examines how ownership changes influ-ence the likelihood of firms paying dividends. Megginson,Nash, and van Randenborgh (1994), Boubakri and Cosset(1998), and D’Souza and Megginson (1999) show thatformerly state-owned companies typically pay muchhigher dividends after their first share issue privatizationthan they did under state ownership. In most cases, thefirst, partial privatization prompts firms to initiate divi-dend payments. Given the size and importance of manyprivatized firms, this tendency to initiate dividend pay-ments appears to have a bell-weather effect on other firmsin the national market, though it is unclear whetherprivatized companies adopt national or global payoutstandards. Our study is the first to directly examine theimpact of privatization on cash dividend payments andrepurchases within a broader study of payout likelihood.

4 We also retrieve data from Worldscope on earnings reporting

frequency, retained earnings in relation to total equity, and the date of

incorporation. We moreover add to our database information on the date

of incorporation from the Amadeus database, if the date of incorporation

is not available through Worldscope. Unless otherwise indicated, all

other data come from Datastream.5 We decide not to differentially convert into euros income

statement data (which ideally should be converted using yearly average

exchange rates) versus balance sheet data (that should be converted

using end-of-the-year exchange rates). In later regressions we combine

income statement data with balance sheet data and we do not want our

regression outcomes to be influenced by distortions caused by a

differential conversion of the two types of financial variables.6 The consumer price indexes for all items are from the Organisation

for Economic Co-operation and Development (OECD)’s Main Economic

Indicators. For Greece, we also use the consumers’ price index for the

period 1999 and 2000, because Greece adopted the euro only in January

2001.

3. Data and sample selection

We examine the evolution of payout policies in theEuropean Union over the period 1989–2005. We start ouranalysis in 1989 for two reasons. First, this is a historicallylogical date to begin, since this was when the Plan Delorswas adopted. This plan presaged creation of the EuropeanMonetary Union (EMU), which became official EU policywith the signing (1991) and ratification (1992) of theMaastricht Treaty. The second reason for beginning in1989 is empirical; prior to that date, the database we use(described below) provides less than comprehensivecoverage of European listed firms. We conclude the studywith data from 2005, which is the most recent year withavailable data.

We study the 15 historically capitalist, democraticnations that belonged to the European Union before May2004 (EU15) and were members of the EU throughout the

study period. We exclude the 12 mostly central andeastern European countries that joined the EU during theMay 2004 and January 2007 expansions. We search theDatastream database for active as well as dead andsuspended listings in order to avoid survivor bias, andselect companies with usable International SecuritiesIdentifying Number (ISIN) and industry codes. We elim-inate companies with similar ISIN codes and similarnames, companies that give error codes in downloadingdata, and companies that report in U.S. dollars. Becausemost existing empirical studies delete utilities andfinancials, we also focus primarily on industrial firms(including transportation). These screens yield 6,946companies from the transport and industry sectors.Within this group, 2,301 companies have cash dividenddata for 1989, of which 2,017 are cash dividend payers.This grows to 4,153 companies in 2005, of which 2,112 arecash dividend payers. Using the ISIN codes, we thenretrieve share repurchase data from Worldscope.4 Thereare 1,707 repurchase observations in 1989, with 108repurchasing companies, and these numbers grow to3,148 and 532, respectively, in 2005. The final samplecontains 60,729 firm-year observations on cash dividends,of which 39,731 firm-years are for cash dividend payersand 20,998 firm-years are for companies that do not paycash dividends. There are 42,398 firm-year observationson share repurchases, of which 3,830 firm-year observa-tions are for repurchasers and 38,568 are for non-repurchasers.

Most of the dividend payments and share repurchasesafter 1998 are available in euros, except for those paid bycompanies headquartered in Great Britain, Denmark, andSweden. For these countries we recalculate all amounts ineuros using end-of-year exchange rates from Datastream.5

All nominal values are then transformed to real values forthe year 2000. We deflate the data with the consumers’price index for all items in the individual countries duringthe period 1989–1998. Starting in 1999, we use theconsumers’ price indexes for all items in the EU15.6 Basedon the evidence presented in Joumard (2001), weconclude that there were no major changes in continent-wide cash dividend taxation policies during our studyperiod, though Bell and Jenkinson (2002) show thatBritain did reduce the attractiveness of cash dividends to

ARTICLE IN PRESS

0

10

20

30

40

50

1989

Perc

enta

ges

UK Germany France Benelux Southern Europe Other

20052003200119991997199519931991

Fig. 1. Percent of industrial cash dividend observations by country/region, EU15, 1989–2005. Percentages are the number of cash dividend observations of

listed industrial companies by country or region divided by the total number of cash dividend observations for listed industrial companies and the

outcome multiplied by 100. Companies from the United Kingdom are indicated by U.K., while German and French companies are indicated by Germany

and France. Benelux represents companies from Belgium, The Netherlands and Luxembourg, ‘‘Southern Europe’’ represents companies from Portugal,

Spain, Italy and Greece, and ‘‘Other’’ represents companies from Austria, Denmark, Finland, Ireland and Sweden.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374352

institutional investors in 1997. On the other hand, both theEU collectively and specific countries individually changedtheir policies regarding share repurchases during thestudy period, invariably relaxing restrictions on andreducing tax burdens on repurchases.

The single largest cohort of firms (40% of the total) inour sample comes from the United Kingdom, while 16%come from France, 13% from Germany, 5% come from theBenelux countries (Belgium, The Netherlands, and Lux-embourg), 12% from Southern Europe (Greece, Italy,Portugal, Spain), and 13% from other countries (Austria,Denmark, Finland, Ireland, and Sweden).7 Fig. 1 shows thegeographical distribution of companies in the overallsample of cash dividend observations over time. Britishcompanies are again the largest single cohort in everyyear’s sample, representing 42% of the 2,301 cash dividendobservations in 1989, falling to 32% in 2000, but risingback to 39% of the 4,153 cash dividend observations in2005. French companies are the second largest cohortevery year, accounting for 14–17% of the total during1989–1995 and 2000–2005. During the 1997–1999 gloryyears of the Nouveau Marche, when several hundredyoung French companies executed IPOs, their nation’sshare reached 18% of the total, but the virtual collapse ofthe Nouveau Marche (and many of the newly publiccompanies) after March 2000 accounted for the latesample contraction. Companies from Germany and South-ern Europe have a slightly increasing share over time:German companies account for 11% of cash dividendobservations in 1989 and 14% in 2005. Southern Europeancountries increase their share from 10% in 1989 to 13% in2005. Benelux companies represent a slightly decreasingshare of the sample over time (falling from 8% to 6%) whilecompanies headquartered in Scandinavia, Austria, and

7 These percentages are rounded and for that reason do not add to

100%.

Ireland (grouped together as ‘‘Other’’ Europe) are a stable13–15% of the sample throughout the study period. Itbears noting that while privatization IPOs considerablyincreased the total market capitalization of all continentalEuropean markets during 1989–2005, there are but 108issues in our sample, so these do not materially impactaverage or national values in an equally weighted sampleof listed companies. However, we will soon find that thesefirms do account for a disproportionate share of the value

of dividend payments and share repurchases.

4. Cash dividend payment and share repurchase patternsof EU15 companies

4.1. Declining fractions of dividend paying companies and

increasing fractions of repurchasing firms

Fig. 2 indicates the total number of industrial compa-nies with cash dividend data available for the period1989–2005, showing the rise from 2,301 companies in1989 to the peak of 4,454 in 2001, and 4,153 in 2005. Thisfigure also presents the number of cash dividend payersand non-cash-dividend payers, as well as the number ofcompanies that once paid cash dividends but havestopped (Did-pay) and those that have never paid cashdividends at all (Never-paid). The figure, finally, alsoshows the growing number of companies that repurchaseshares over 1989–2005. There is a large increase in thenumber of cash dividend observations over time, duemostly to new listings but also partly to improved datacoverage. The total number of cash dividend observationsrises from 2,301 in 1989 to a peak of 4,454 in 2001, thendeclines slowly to 4,153 in 2005.8 The number of cash

8 This assertion that the decline in the fraction of EU firms that pays

dividends began after 1990 is supported by work we performed

previously using data from a longer time period, 1980–2002. As noted

ARTICLE IN PRESS

0

1000

2000

3000

4000

5000

1989

Num

ber

of com

panie

s

CD observations Payers Did-pay Never-paid Non-payers Repurchasers

20052003200119991997199519931991

Fig. 2. Number of companies with cash dividend data available, dividend payers, non-dividend payers, never-payers, did-payers, and share repurchasers,

listed industrial EU15 companies, 1989–2005. ‘‘CD observations’’ represents the number of observations on cash dividend payers (‘‘Payers’’) summed with

the number of companies that pay no (zero) dividends (‘‘Non-payers’’). The companies that do not pay dividends are split into companies that at least paid

dividends once since 1989 but do not so in the relevant year (‘‘Did-pay’’) and companies that never paid dividends at all since 1989 (‘‘Never-paid’’). The

figure, moreover, shows the number of companies that repurchase shares (‘‘Repurchasers’’).

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 353

dividend payers rises from 2,017 in 1989 to a peak of 2,939in 1998 then falls to 2,025 in 2004 before reboundingslightly to 2,112 in 2005. These trends imply that thefraction of European listed companies paying dividendsdeclines from 88% in 1989 and 1990 to 51% only 15 yearslater. While this is the same pattern that Fama and French(2001) show for listed U.S. companies, the decline inEuropean dividend payments started much later in time(1990 versus 1952) and was proportionately more rapidonce it began. The increase in non-payers over this periodis sharpest for companies that once paid dividends buthave now stopped (Did-pay), though the number ofcompanies that have never paid dividends is still larger.Finally, the number of firms repurchasing their ownshares increases monotonically over the study period,from a mere 109 in 1989 to 543 in 2005.

Table 1 presents an annual summary of the number ofobservations for cash dividend payments and sharerepurchases, as well as the number of firms that do anddo not pay dividends, that do and do not repurchaseshares, that both pay dividends and repurchase shares,and that do neither. The three most interesting series inthis table describe the evolution in the number ofcompanies that neither pay cash dividends nor repurchaseshares (column 9), that repurchase shares but do not paycash dividends (column 10), and that both pay cashdividends and repurchase shares (column 12). Thenumber of firms that do not distribute cash in any formincreases almost by an order of magnitude over the17-year study period, from 166 in 1989 to 1,466 in 2004and 1,403 in 2005. The number of pure repurchasers and‘‘switch-hitters’’ (firms that both pay dividends and

(footnote continued)

earlier, we decided to begin our study period in 1989 for data consistency

reasons. However, all of our cohort-year and aggregate samples from

1980 to 1990 showed relatively stable payout ratios, and the fraction of

listed companies that were dividend payers fluctuated in a narrow range

throughout the 1980s.

repurchase shares) also increases, but more modestly,from 5 to 124 and from 102 to 406, respectively, between1989 and 2005. These patterns suggest that progressivelyfewer EU companies choose to distribute cash to investorsin any way, and that those which distribute cashincreasingly choose to employ share repurchases to do so.

Fig. 3 presents a geographical breakdown of thepercentage of listed industrial companies that pay cashdividends. The fraction of industrial companies payingcash dividends declines steadily—if not always conti-nuously—in every country or region from 1989 through2005. The decline is especially large in the UnitedKingdom and Germany. In the United Kingdom, 92% oflisted companies pay cash dividends in 1989, but this thendeclines to 42% in 2005, while the fraction of listedGerman companies paying regular cash dividends fallsfrom a maximum 84% in 1991 to only 37% in 2004, butthen increases to 40% in 2005. The reason for thesedisproportionate declines in the two largest EU economiesis not obvious, though the rapid increase in the number ofentrepreneurial growth companies going public via theU.K.’s Alternative Investment Market (and its predecessorexchanges) and Germany’s Neuer Markt probably plays animportant role. In all countries or regions except Britainthe relative number of companies that pay dividendsbounces back in 2004 or 2005, which is similar to thepattern Julio and Ikenberry (2004) observe for U.S.companies. Since there are no major EU dividend taxchanges during this period, the most likely explanation forthis rebound in the fraction of EU firms paying cashdividends is the improving economic outlook after 2003.

4.2. Declining propensity to pay, but rising total real

dividend and share repurchase amounts

As noted in Section 2, two striking features of recentU.S. payout policies are the sharp rise in the total realvalue of cash dividends paid over the past quarter-century,

ARTIC

LEIN

PRESS

Table 1Number of observations for cash dividend payments and share repurchases by listed EU15 industrial companies, 1989–2005

All observations are presented by year (column 1). The number of observations on cash dividends (column 2) is split into companies that pay cash dividends (column 3) and companies that pay no (zero) cash

dividends (column 4). The number of observations on share repurchasers (column 5) is split into companies that repurchase shares (column 6) and companies with no (zero) repurchase amounts (column 7).

The number of companies with both types of observations (column 8) is split into companies that made no disbursement at all (column 9: both zero cash dividends and zero repurchase amounts), companies

that repurchase, but do not pay cash dividends (column 10), companies that only pay cash dividends, but have zero repurchases (column 11) and companies that both paid cash dividends as well as repurchase

shares (column 12).

Number of annual observations relating to cash dividends Number of annual observations relating to share

repurchases

Number of annual observations relating to both cash dividends and share repurchases

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Year Firms with

cash dividend

data available

Firms that

pay cash

dividends

Firms that do

not pay cash

dividends

Firms with

repur-chase

data available

Firms that

repur-chase

shares

Firms that do

not repur-

chase shares

Firms with

data available

on both cash

dividends and

share

repurchases

Firms that

neither pay

cash dividends

nor repur-

chase shares

Firms that

repurchase

shares, but do

not pay cash

dividends

Firms that pay

cash

dividends, but

do not

repurchase

shares

Firms that pay

cash and

repur-chase

shares

1989 2301 2017 284 1707 109 1598 1698 166 5 1425 102

1990 2452 2150 302 1847 105 1742 1839 190 3 1547 99

1991 2542 2129 413 1908 71 1837 1898 274 5 1554 65

1992 2630 2110 520 1985 96 1889 1976 369 12 1513 82

1993 2753 2068 685 2067 100 1967 2061 463 18 1503 77

1994 2872 2140 732 2188 94 2094 2181 498 8 1591 84

1995 2927 2279 648 2280 84 2196 2269 444 10 1741 74

1996 3764 2701 1063 2339 130 2209 2325 459 17 1740 109

1997 4196 2886 1310 2428 157 2271 2412 496 10 1762 144

1998 4286 2939 1347 2610 256 2354 2592 571 16 1769 236

1999 4248 2753 1495 2609 254 2355 2585 681 22 1657 225

2000 4445 2652 1793 2866 376 2490 2846 947 58 1538 303

2001 4454 2482 1972 3058 376 2682 3045 1185 66 1500 294

2002 4318 2207 2111 3092 372 2720 3080 1337 81 1380 282

2003 4241 2081 2160 3150 390 2760 3136 1422 101 1331 282

2004 4147 2025 2122 3116 388 2728 3110 1466 78 1258 308

2005 4153 2112 2041 3148 543 2605 3142 1403 124 1209 406

1989–2005 60729 39731 20998 42398 3901 38497 42195 12371 634 26018 3172

H.

vo

nE

ije,W

.L.M

egg

inso

n/

Jou

rna

lo

fFin

an

cial

Eco

no

mics

89

(20

08

)3

47

–3

74

35

4

ARTICLE IN PRESS

0

20

40

60

80

100

1989

Pe

rce

nta

ge

s

UK Germany France Benelux Southern Europe Other

20052003200119991997199519931991

Fig. 3. Relative percentages of listed industrial EU15 cash dividend payers, by country or region, 1989–2005. The graph shows the number of cash

dividend payers in the total number of cash dividend observations in the relevant countries or regions in percentages. Companies from the United

Kingdom are indicated by U.K., while German and French companies are indicated by Germany and France. Benelux represents companies from Belgium,

The Netherlands and Luxembourg, ‘‘Southern Europe’’ represents companies from Portugal, Spain, Italy and Greece, and ‘‘Other’’ represents companies

from Austria, Denmark, Finland, Ireland and Sweden.

0

20

40

60

80

100

120

140

160

180

1989

Bill

ion e

uro

s (

prices o

f 2000)

Cash dividends Share repurchases Total payout

20052003200119991997199519931991

Fig. 4. Cash dividends, repurchases and total payout by listed industrial EU15 companies, h real billions (2000 prices), 1989–2005.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 355

even as the fraction of all listed industrial companies thatpay cash dividends has been cut more than in half, andrapidly increasing annual share repurchase amounts.Fig. 4 shows similar patterns are observed for EUcompanies, though in Europe the changes have occurredmuch more rapidly and recently. First, total real cashdividend payments by EU listed firms increase modestlyfrom h35 billion in 1989 to h42 billion in 1994 then surgemore than 170% between 1994 and 2001, to peak at h114billion. Total cash dividends then fall to h89 billion in2003 before rebounding to h113 billion in 2005 asEurope’s economic recovery gathers strength. The sharprise in dividends during the late-1990s is caused partly byan increase in the number of dividend paying companies,but is mostly due to increased dividend payments by theoriginal dividend payers of 1989. This is in line withDeAngelo, DeAngelo, and Skinner’s (2004) finding ofincreasing concentration of dividends amongst high

payers in the United States. Second, Fig. 4 also revealsthe large increase in real share repurchases in theEuropean Union after 1992. Starting from h7 billion in1989 (representing 17% of total payout), repurchases fallto only h1 billion in 1992 (3% of total payout), butsubsequently rise rapidly and almost monotonically to avalue of h58 billion in 2005. This level represents slightlymore than 50% of real cash dividends and 34% of totalpayout.

Table 2 details the total value of share repurchases, bycountry, each year between 1989 and 2005. This tableclearly shows that U.K. firms are the most prolificrepurchasers of their own shares, accounting for almostexactly half of the total h253 billion value of repurchasesfor the entire 1989–2005 period and of the h59 billion2005 total. However, share repurchases become increas-ingly important in several other EU countries, particularlyafter 1997—when the total value of EU repurchases jumps

ARTIC

LEIN

PRESS

Table 2Total value of share repurchases by industrial companies listed in the specific countries of the EU15, h real millions (2000 prices), 1989–2005

The table presents the real value of the sum of share repurchases made by individual companies in a country in prices of 2000 for all the 15 countries in the EU15 for all the years between 1989 and 2005. Due

to rounding errors the sum of the values over the years per country may differ from the totals reported in the final line of the table. Rounding errors may also make the sum of the repurchase values over all

countries per year different from the values reported for all countries in the final column of the table.

Year Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom All countries

1989 0 0 8 129 577 0 0 0 260 0 114 6 46 40 4965 6146

1990 0 0 27 134 278 54 0 0 366 0 19 0 37 3 1543 2461

1991 0 0 0 22 99 0 0 0 99 0 19 0 2 17 316 575

1992 0 0 0 83 �11 210 0 5 202 0 18 3 6 35 462 1013

1993 0 5 1 9 91 45 0 58 1 0 1 �14 33 14 1049 1293

1994 0 0 16 5 151 70 0 6 0 0 432 11 23 0 1100 1813

1995 0 0 2 40 252 2 0 71 0 5 5 7 59 2 898 1344

1996 0 0 0 104 86 1 0 1 0 3 140 16 82 363 1152 1946

1997 1 0 16 20 472 71 0 13 27 2 540 17 67 1743 6726 9717

1998 0 0 331 388 888 4 0 66 182 3 1405 60 218 610 13569 17723

1999 24 0 463 247 5019 1601 0 29 153 1 1464 36 177 408 6706 16329

2000 105 107 665 542 4291 4583 4 16 697 0 1443 38 173 2874 10518 26056

2001 160 190 176 689 11875 2289 162 43 1797 6 1639 39 168 1685 9695 30611

2002 19 341 449 351 5883 1164 0 65 824 2 1176 28 257 415 13409 24382

2003 4 57 492 1595 5917 1424 4 13 658 35 837 54 189 477 10123 21880

2004 0 363 704 2727 8419 2234 43 34 182 265 1091 32 382 1001 13335 30814

2005 35 1076 984 4428 6184 3508 256 349 1355 48 7383 31 1126 2733 29344 58841

Total 349 2138 4336 11515 50473 17259 468 767 6804 370 17726 363 3047 12420 124909 252943

H.

vo

nE

ije,W

.L.M

egg

inso

n/

Jou

rna

lo

fFin

an

cial

Eco

no

mics

89

(20

08

)3

47

–3

74

35

6

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 357

from h9.7 billion in 1997 to h17.7 billion in 1998 and h30.6billion in 2001. French companies repurchased h50 billionworth of their own shares, with over 96% of this occurringafter 1998, as described in Ginglinger and L’Her (2006)and Ginglinger and Hamon (2007). German and Dutchcompanies executed repurchases exceeding h17 billion intotal value over the study period (again, mostly after1997), while Swedish and Finnish companies repurchasedh12.4 billion and h11.5 billion, respectively.

4.3. Privatization’s impact on dividend payments and share

repurchases

One of the most significant differences between thecapital markets of the United States and the EuropeanUnion is the prominence of privatized companies in thelatter, and their near total absence in the former. A wave ofshare issue privatizations in the United Kingdom duringthe 1980s and in continental Europe during the 1990screated a relatively small number of extremely large listedcompanies that almost always became the most valuableand most actively traded firms in every European stockmarket (see Jones, Megginson, Nash, and Netter, 1999;Megginson, 2005). Empirical research shows that priva-tized companies usually initiate (or increase) dividendpayments after being listed, and we examine their impacton aggregate EU dividend payments and share re-purchases using a listing of privatized companies pre-sented in Beltratti, Bortolotti, and Milella (2007). Theirinfluence is hugely disproportionate to their numbers:even though privatized companies never represent morethan 2.5% of the total number of companies, the vastmajority (always over 75%) of privatized firms paydividends and their payments equal between 11% (in1994) and 23% (in 2005) of total EU cash dividends paid.The 57 privatized companies listed in 1989 pay an averageh108 million in cash dividends, whereas the 2,244 non-privatized (never state-owned) companies pay an averageh13 million. By 2005, average cash dividend payments riseto h308 million for the 83 privatized companies and to h21million for the 4,070 non-privatized firms. Fig. 5 shows

0

20

40

60

80

100

1989

Bill

ion e

uro

s (

prices o

f 2000)

Cash dividends paid by privatized companies

Cash dividends paid by non-privatized companie

1199519931991

Fig. 5. Cash dividends paid and repurchases made by privatized and non-privat

(2000 prices), 1989–2005.

the evolution of cash dividends and share repurchases forprivatized and never-state-owned companies over theperiod 1989–2005.

The relative growth in share repurchases for privatizedversus non-privatized firms is even greater than fordividend payments. Disregarding 1989, the year a priva-tized British Petroleum repurchased shares worth h4.6billion, the average amount repurchased by privatizedfirms is h2 million in 1990, and this rises one-hundredfold

to h222 million in 2005. This massive increase in averagerepurchase amounts by privatized companies is notmatched by non-privatized companies, even though theaverage share repurchase amount for these companiesincreases elevenfold from h1.3 million in 1990 to h14.6million in 2005. Privatized firms account for a low of 2.7%of total repurchase amounts in 1996, but this rises to anaverage 22.2% of repurchases during 2000–2005. Combin-ing dividends and repurchases, privatized firms accountfor 19.1% of total payout by EU companies for the full1989–2005 period, and for 22.1% of payout during2000–2005. Despite the larger average amounts paid outby privatized companies, non-privatized companies stilldominate both cash dividend and share repurchaseamounts. Nevertheless, the amounts paid by the relativelysmall number of privatized companies cannot be ne-glected, especially since Fig. 5 shows that privatizedcompanies have a more stable pattern of payouts thannon-privatized companies, particularly with respect tocash dividends.

4.4. Financial reporting frequency

Not all listed EU companies must report earnings everyquarter. Some of the largest listed companies must nowreport financial results quarterly, but these requirementshave been adopted only recently and key countries (suchas the United Kingdom) still require only semi-annualfinancial reports. However, companies have always hadthe option of reporting more frequently than legallyrequired, and many have chosen to do just that in orderto acquire a reputation for transparency. As investors are

Repurchased by privatized companies

s Repurchased by non-privatized companies

2005200320011999997

ized (never state-owned) listed industrial EU15 companies, h real billions

ARTICLE IN PRESS

1.00

2.00

3.00

4.00

1989

Avera

ge r

eport

ing fre

quency

All companies Civil law companies

Common law companies Privatized companies

20052003200119991997199519931991

Fig. 6. Average financial statement reporting frequencies in all listed industrial EU15 companies and in companies from common law (United Kingdom

and Ireland) and civil law (other 13) countries, and in privatized firms in all countries, 1989–2005. Companies that present financial statements once a

year have a reporting frequency of 1, companies that do so twice a year have a frequency of 2 and companies that report on a quarterly basis have a

reporting frequency of 4.

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

1989

Bill

ion

eu

ros (

price

s o

f 2

00

0)

Cash dividends paid by annual reporters Cash dividends paid by quarterly reporters

Repurchases by annual reporters Repurchases by quarterly reporters

20052003200119991997199519931991

Fig. 7. Value of cash dividends and share repurchases for listed industrial companies in the EU15 that report financial statements on a quarterly basis and

for companies that report financial statements on an annual basis respectively, h real billions (2000 prices), 1989–2005.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374358

informed more often, asymmetric information may bereduced and Wood (2001) suggests that improvedreporting (and corporate governance) in Europe makesinvestors focus less on dividends and more on earnings.We study the evolution of financial reporting frequency inEU15 countries over 1989–2005 and examine its impacton both cash dividend payments and share repurchases.Fig. 6 shows that companies report on average 1.2 times ayear in 1989, with 1,850 companies reporting once a year,371 companies reporting twice a year, eight companiesreporting three times a year, and only 36 companiesreporting on a quarterly basis. By 2005, the averagenumber of times that companies report doubles to 2.4,with a majority (2,183) of companies reporting semi-annually, 1,299 companies reporting on a quarterly basis,740 reporting annually, and one company reporting threetimes a year. Companies headquartered in common lawcountries report earnings in 1989 more often (1.3 times)

than companies headquartered in civil law countries (1.2times) but, surprisingly, common law firms reportsignificantly less frequently than civil law firms in 2005(1.9 times versus 2.8 times, t-value of difference ¼ 30.65).Privatized companies, however, persistently show higheraverage earnings reporting frequencies than any othergroup during the study period, and their average earningsreporting frequency rises from 1.6 to 3.4 times per yearbetween 1989 and 2005.

Fig. 7 shows the amounts of cash dividends and sharerepurchases for the two extreme reporting frequencygroups, annual and quarterly reporters. This reveals thatannual reporters hardly ever repurchase shares, but areimportant cash dividend payers in the first half of thestudy, 1989–1997. From 1998 on, these companies aresurpassed by quarterly reporters with respect both tocash dividends and share repurchases, and the dominanceof quarterly reporters becomes extreme after 2003.

ARTICLE IN PRESS

0

0.2

0.4

0.6

0.8

1989

Tota

l payout ra

tio

Mean total payout ratio for all observations Median total payout ratio for all observations

Mean total payout ratio for payers Median total payout ratio for payers

20052003200119991997199519931991

Fig. 8. Mean and median total (cash dividends and repurchases) payout ratios for all observations and for payers (companies that either pay cash

dividends or repurchase shares), listed industrial companies in the EU15, 1989–2005.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 359

Quarterly reporters pay h76 billion in cash dividends (67%of the EU total) and repurchase h44 billion worth of shares(75% of the EU total) in 2005, with annual reporterspaying only h3 billion in cash dividends and repurchasingshares worth less than h1 billion. The decline in cashdividends by annual reporters is caused by a decrease inaverage cash dividends amongst annual reporters fromh7.8 million in 1989 to h4.8 million in 2005 and a declinein the relative numbers of companies reporting on anannual basis (declining from 1,803 cash dividend observa-tions in 1989 to 712 in 2005). The increase of cashdividends paid by quarterly reporters is solely caused byan increase in numbers: on average, the 35 companieswith cash dividend data available in 1989 pay h161million, while the 1,239 quarterly reporters with cashdividend data pay h61 million in 2005.

4.5. Payout ratios of EU15 listed companies

Thus far, we have shown that the fraction of EU15listed industrials paying cash dividends declines byroughly one-third between 1989 and 2005, even as totalreal dividends, share repurchases, and average dividendsand repurchases by dividend payers all increase. But whattrend is observed for payout ratios? Fig. 8 details meanand median total payout ratios (cash dividends and sharerepurchases) for all listed industrials and all payers.9

Except for showing that payout ratios peak naturallyduring recessions (1991–1993 and 2001–2003) anddecline during expansions, we find no significant changesbetween the mean payout ratios of all industrial payers

9 The payout ratios are calculated as described in Julio and Ikenberry

(2004, p. 92, footnote 16), who set the payout ratio equal to one if it is

actually negative because of negative income. Moreover, they also set the

payout ratio to one if firms pay more than 100% of their earnings as

dividends. We apply their approach to Figs. 8 and 9. The results are

qualitatively similar—though more volatile—using a standard payout

ratio.

between 1989 and 2005. The mean (median) payout ratiofor listed payers increases from 40% (33%) of profits in1989 to 61% (55%) in 1992, then falls to 48% (40%) in 1994and remains close to 50% (40%) for payers until the nextrecession begins in 2001. On the other hand, the mean(median) payout ratio for all listed firms—payers and non-payers—falls almost continuously after peaking at 49%(43%) in the recession year of 1992. By 2005, mean payoutratios for all listed firms have fallen to 29% with only avery slight uptick in payout ratios during the 2001–2002recession. The decline in median payout ratios of allcompanies between 1989 and 2005 is very obvious, with alowest level in 2004 of 2%. The mean (median) payoutratio of payers is much larger than the similar ratio of allcompanies in 1989, but this increases much further in2005. These structural developments are a signal of aconcentration of dividend payouts in the EU15 compar-able to the developments in the United States.

Fig. 9 presents a geographic breakdown of the trends inmedian total payout ratios of EU15 industrial companiesthat either pay dividends or repurchase shares, or both.The most notable is the reduced dispersion of payoutratios over time. The range in median payout ratios is 28percentage points in 1989 (between 21% for France and49% for Germany), but declines to a much narrower 15percentage point range in 2005, with French industrialpayers distributing a median 37% of profits and Britishpayers distributing 52% of earnings. The reduction in thedispersion of medians of total payout ratios also holds forall individual EU15 countries as an F-test over the 15countries finds significantly smaller variance of mediansin 2005 than in 1989. Mean payout ratios (not reported) ofEU15 payers show generally similar reductions in thedispersion of total payout ratios. This suggests that theEU15 countries’ payout ratios are starting to march tothe beat of the same drummer.

For a final take on trends in EU15 payout ratios, weexamine the evolution in median total payout ratios of alllisted industrial companies, payers and non-payers, by

ARTICLE IN PRESS

0

0.2

0.4

0.6

0.8

1989

Media

n p

ayout ra

tio

UK Germany France Benelux Southern Europe Other

20052003200119991997199519931991

Fig. 9. Median total (cash dividends and repurchases) payout ratios for listed industrial companies in the EU15, that either pay cash dividends or

repurchase shares, by country/region, 1989–2005. Companies from the United Kingdom are indicated by U.K., while German and French companies are

indicated by Germany and France. Benelux represents companies from Belgium, The Netherlands and Luxembourg, ‘‘Southern Europe’’ represents

companies from Portugal, Spain, Italy and Greece, and ‘‘Other’’ represents companies from Austria, Denmark, Finland, Ireland and Sweden.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374360

country or region. These trends (not reported) tell astriking story. Median payout ratios of all industrialcompanies fall after the recession year of 1992, whenthey range between 22% and 51%, and—with the im-portant exceptions of Britain and Germany—have appar-ently converged to a range of 15–29% in 2005. In 1992, themedian German industrial firm paid out the EU’s highesttotal payout ratio, 51%, but after 2000 the median Germanfirm does not distribute cash at all! After 2001, the medianU.K. listed company also neither pays dividends norrepurchases shares.

10 If all value was concentrated in one decile, the Herfindahl Index

would be 1.000. If all deciles instead accounted for 10% each, and

therefore the distribution was fully equal, the Herfindahl Index would be

0.100.

4.6. Concentrating earnings and dividends in EU15

companies

The evidence presented above suggests that cashdistributions—dividends and share repurchases—may beconcentrating among a small number of companies. Weexamine this further by breaking down the distribution oftotal real dividends and real assets, by deciles, for listedEU15 companies in 1989 and 2005. Panel A of Table 3,which sorts firms by real dividends paid, shows dividendsare highly concentrated, and that concentration increasesbetween 1989 and 2005. The same patterns are observedfor repurchases. In 1989 the smallest 10% of the dividendpayers pay no dividends and the smallest 40% do notrepurchase shares. In the same year the largest decilefirms pay 79.0% of the cash dividends, make 97.5% of theshare repurchases, and distribute 81.8% of total cashpayouts. By 2005, the smallest 50% of listed companiespay virtually no cash dividends and the smallest 60% donot repurchase, whereas the largest decile of listedcompanies pay 92.1% of cash dividends, 98.6% of sharerepurchases, and 94.3% of total cash distributions. Thesefigures show that share repurchases are even moreconcentrated among larger firms than are cash dividends.Moreover, both cash dividends and share repurchases

become more concentrated, as shown by the increasingHerfindahl Index between 1989 and 2005.10

Panel B of Table 3 also shows that dividends andrepurchases are highly concentrated among the largestfirms, measured by total real assets, and that theconcentration increases between 1989 and 2005. In1989, the smallest 10% of companies distributes 0.2% ofcash dividends, 0.0% of repurchases and 0.1% of alldividends. The largest decile firms pay 72.9% of cashdividends, 87.9% of share repurchases, and 75.2% of totalcash distributions. In 2005, firms in the smallest two assetdeciles make no cash payments whatsoever, while thelargest decile pays 86.0% of all cash dividends, makes93.3% of the share repurchases, and distributes 88.5% of allcash payments. Also, if total cash distributions aremeasured in asset deciles, we find that share repurchasesare concentrated most, and that all three types ofdistributions become more concentrated between 1989and 2005. The growing concentration is confirmed by thedevelopment in the Herfindahl Index based on assetdeciles. This measure increases for cash dividends from0.554 to 0.748 between 1989 and 2005, while it increasesfor share repurchases and total cash distributions from0.777 and 0.583, respectively, in 1989 to 0.873 and 0.789in 2005.

5. Measuring the likelihood to pay cash dividends and torepurchase shares

5.1. Expected and actual changes

Since we have shown that the relative number of EU15cash dividend payers declines steadily after 1989, and the

ARTICLE IN PRESS

Table 3The distribution of real dividends of EU15 listed industrial companies in deciles of real total dividends (cash dividends and repurchases) in Panel A and in

deciles of real assets in Panel B, 1989 and 2005

The table shows the relative real amounts of cash dividends (Cash dividends), the relative real amounts used for repurchasing shares (Share

repurchases), and the relative real amounts of both cash dividends and/or share repurchases (Cash dividends and/or share repurchases) for companies

ranked in deciles in 2 years (1989 and 2005). In Panel A the first decile holds 10% of the companies with the smallest amounts and the tenth decile

contains the companies with the highest amounts spend on both types of dividends together. Companies with no information on cash dividends, but with

information of share repurchases, or vice versa are included in calculating these deciles and in calculating the two final columns on cash dividends and/or

share repurchases. In Panel B the first decile holds 10% of the companies with the smallest amounts of assets and the tenth decile contains the companies

with the highest amounts of assets. Herfindahl is the value of the Herfindahl index, which represents an equal distribution over the deciles if its value is

0.1, while the highest possible concentration is indicated by a Herfindahl value of 1. Amount represents the amounts spent on either cash dividends,

repurchases or both. Due to a rounding error the total amount spent for cash dividends and/or share repurchases differs from the sum of the two

constituencies in 2005. Number of observations gives the number of observations used in the calculations.

Panel A: The distribution of real dividends of EU15 listed industrial companies in deciles of real total dividends (cash dividends and repurchases), 1989

and 2005

1989 2005 1989 2005 1989 2005

Total dividend

deciles

Cash dividends Cash dividends Share repurchases Share repurchases Cash dividends

and/or share

repurchases

Cash dividends

and/or share

repurchases

1 0.0 0.0 0.0 0.0 0.0 0.0

2 0.1 0.0 0.0 0.0 0.1 0.0

3 0.3 0.0 0.0 0.0 0.2 0.0

4 0.5 0.0 0.0 0.0 0.4 0.0

5 0.9 0.0 0.1 0.0 0.8 0.0

6 1.4 0.2 0.0 0.0 1.2 0.1

7 2.4 0.6 0.3 0.1 2.1 0.4

8 4.5 1.6 0.9 0.3 3.9 1.2

9 10.9 5.6 1.0 1.1 9.4 4.0

10 79.0 92.1 97.5 98.6 81.8 94.3

Total 100.0 100.0 100.0 100.0 100.0 100.0

Herfindahl 0.639 0.851 0.952 0.972 0.680 0.891

Amount (h billion) 34.8 112.5 6.1 58.8 40.9 171.4

Number of

observations

2301 4153 1707 3148 2310 4159

Panel B: The distribution of real dividends of EU15 listed industrial companies in deciles of real assets, 1989 and 2005

1989 2005 1989 2005 1989 2005

Total asset deciles Cash dividends Cash dividends Share repurchases Share repurchases Cash dividends

and/or share

repurchases

Cash dividends

and/or share

repurchases

1 0.2 0.0 0.0 0.0 0.1 0.0

2 0.4 0.0 0.2 0.0 0.3 0.0

3 0.6 1.2 0.5 0.0 0.6 0.8

4 0.8 0.2 0.2 0.1 0.7 0.2

5 1.2 0.3 0.1 0.2 1.1 0.3

6 2.4 0.8 0.2 0.2 2.0 0.6

7 3.0 1.1 3.5 0.6 3.1 1.0

8 5.6 2.3 3.6 1.2 5.3 1.9

9 12.9 8.1 3.8 4.4 11.5 6.8

10 72.9 86.0 87.9 93.3 75.2 88.5

Total 100.0 100.0 100.0 100.0 100.0 100.0

Herfindahl 0.554 0.748 0.777 0.873 0.583 0.789

Amount (h billion) 34.8 112.5 6.1 58.8 40.9 171.4

Number of

observations

2301 4152 1705 3146 2308 4157

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 361

number of share repurchasers increases, we now examinewhether this results from changes in company character-istics or from changes in the propensity to pay. We use the

portfolio approach of Fama and French (2001) and form atotal of 27 portfolios at the beginning of the observationperiod (1989). These portfolios are derived from three size

ARTICLE IN PRESS

Table 4The effects of changing characteristics and the propensity to pay cash dividends (Panel A), and the effects of changing characteristics and the propensity to

repurchase shares (Panel B), listed industrial companies in the EU15, based on the group approach of Fama and French (2001)

Companies is the number of companies with cash dividend observations (Panel A) and share repurchases (Panel B) in the relevant year (averaged for

1989–1993). Actual is the number of cash dividend payers (Panel A) and share repurchasers (Panel B) divided by the number of companies multiplied by

100 (averaged for 1989–1993). Expected is the percentage of cash dividend payers (Panel A) and share repurchasers (Panel B) calculated from the number

of companies in each of 27 portfolios for the year times the proportion of cash dividend payers (Panel A) and share repurchasers (Panel B) in the portfolio

during the period 1989–1993 summed over the 27 portfolios, divided by the total number of companies in the 27 portfolios for the year and multiplied by

100. Portfolios were based on size, earnings ratio, and the market-to-book value of the company. For each year the companies were classified into small,

medium, and large companies based on whether a company belonged to the first 20 percentiles, the 21st to the 50th percentile, or the 50th to the 100th

percentile. For the earnings ratio we divided the earnings before interest but after tax (E) by the assets (A) of the companies and measured the average

values of the 33.3 percentile and the 66.7 percentile from the 5-year period 1989–1993. Then we used the average values of the (E/A) breakpoints as cut-

off values for classifying the companies in low, middle, or high earning companies in the years 1994–2005. The breakpoints used for the earnings ratio

were 0.047 and 0.084 respectively. The same procedure was also used for the market-to-book value of the company (V/A), defined as assets minus book

value of equity plus market value of equity (V) divided by assets (A). The breakpoints for the market-to-book value were 1.072 and 1.417 respectively.

Panel A: Propensity to pay cash dividends Panel B: Propensity to repurchase shares

Year(s) Number of

companies

Actual

(%)

Expected

(%)

Actual (%) �Expected

(%)

Number of

companies

Actual

(%)

Expected

(%)

Actual (%) �Expected

(%)

1989–1993 2315 83.5 1764 5.0

1994 2716 75.3 83.0 �7.7 2084 4.3 4.8 �0.5

1995 2771 79.1 83.6 �4.5 2172 3.7 5.0 �1.2

1996 2804 79.1 85.2 �6.1 2208 5.7 5.0 0.6

1997 2797 79.1 86.5 �7.4 2266 6.8 5.1 1.7

1998 2914 78.6 84.6 �6.0 2426 10.2 5.1 5.1

1999 2811 75.4 83.5 �8.1 2384 10.0 5.1 4.9

2000 3429 65.4 80.9 �15.4 2677 13.4 5.1 8.3

2001 3520 60.8 78.6 �17.8 2836 12.6 5.0 7.6

2002 3795 54.2 77.0 �22.8 2787 12.5 5.0 7.5

2003 3693 52.6 77.0 �24.4 2735 13.3 5.0 8.3

2004 3657 52.3 78.1 �25.8 2740 13.4 5.0 8.4

2005 3816 52.8 78.1 �25.4 2915 17.8 5.0 12.8

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374362

classes, each divided into three earnings ratio classes, andthe resulting nine portfolios subdivided into three (low,middle, and high) categories based on growth character-istics.11 The resulting portfolios define the originalcompany characteristics and these define the portfoliosof companies in the subsequent years. We then estimateprobabilities of paying cash dividends as well as repurch-asing shares for all 27 portfolios over the period1989–1993, and apply these estimations to the proportionof companies found in the same classes for everysubsequent year through 2005. Changes in this variableindicate a change in the likelihood to pay caused by achange in firm characteristics. We then compare theresulting expected percentages to pay with the actualpercentages of firms paying dividends and repurchasingshares. If the difference between the actual and expectedpercentages becomes systematically more negative weconclude that there is an autonomous decline in thepropensity to pay or vice versa. The results of theseestimations are presented in Table 4.

It is clear that the declining propensity to pay cashdividends observed in the United States by Fama andFrench (2001) is also observed for listed companies in theEuropean Union. First, the characteristics of the compa-

11 Fama and French (2001) distinguish between two growth

variables, the relative change in assets and the market-to-book value

of the firm. We use only the market-to-book ratio as this ratio yields

results that are more significant in logistic regressions of the probability

to pay (not reported here).

nies change over time and result in a relatively smalldecline in the expected percentage of firms paying cashdividends, from an average 83.0% of all listed firms in 1994to 78.1% in 2005. Second, the actual percentage of cashdividend payers declines more, from an average 83.5% in1989–1993 to 75.4% in 1999, before plunging to 52.8% in2005. The change in actual payers is more pronouncedthan the decline based on changing characteristics, andthis difference increases sharply (from �7.7 to �25.4percentage points) between 1994 and 2005.

Interestingly, applying the Fama and French (2001)approach to the propensity to repurchase gives oppositeresults. Using the same 27 portfolios as before, but nowestimating the expected percentage that will repurchase,we find no systematic evidence of a changing expectedrepurchase likelihood after 1993, but the actual number ofcompanies repurchasing shares more than triples. The 5%of companies that actually repurchase shares in the period1989–1993 equals the 5% of the companies that areexpected to be repurchasing shares in 2005. However, thefraction of companies that actually do repurchase sharesincreases over time, from 5% in 1989–1993 to 17.8% in2005, so the excess of share repurchasing companies risesby 12.8 percentage points after 1993.12

12 Because of space limitations we do not discuss in detail our

logistic regressions using the Fama and MacBeth (1973) methodology to

estimate the probability of paying cash dividends (also used by Fama and

French, 2001; Denis and Osobov, 2008) or to repurchase shares.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 363

5.2. Further explanations for the likelihood to pay cash

dividends and repurchase shares

To further investigate the probability of paying cashdividends and repurchasing shares, we include in ourestimations an additional eleven variables suggested byrecent academic research.13 First, we follow Wood (2001)in assuming that improved standards of financial dis-closure—proxied by earnings reporting frequency (ERF)—may reduce the need to make cash distributions, thoughimproved disclosure may also influence the amount ofdividends paid or shares repurchased. Second, we assumethat privatized companies are more likely to pay cashdividends, as suggested by Megginson, Nash, and vanRandenborgh (1994), Boubakri and Cosset (1998), andD’Souza and Megginson (1999), so we use a dummyvariable (PRIV) to signify divested companies. Third, wefollow Jensen (1986) in assuming that leverage, proxied bythe book value of debt divided by the book value of assets(LR), may help control agency costs, thus reducing theneed to distribute cash to shareholders through dividendsor repurchases. According to this view, leverage and cashdistributions will be substitutes for curbing managerialincentives to predate, and a negative relationship betweencash dividends and debt ratios is predicted.14 Alterna-tively, higher leverage might simply proxy for older, larger,more stable, and more profitable companies that arebetter able to afford paying dividends and buying backshares. This view implies that leverage and cash payoutswill be complements so a positive relationship betweencash dividends and leverage should be observed.

Fourth, we assume that higher income risk, measuredas the standard deviation of net income during the past5 years divided by each year’s sales (SDS), negativelyinfluences the probability of paying. Fifth, we assume thatolder companies may be better able to accumulate fundsthan younger firms, while at the same time having fewergrowth opportunities, so we include AGE measured asyears since incorporation. This is in line with Salas andChahyadi (2006) who attribute 5% of the decline in theproportion of U.S. firms paying dividends to firm char-acteristics related to age. Sixth, a somewhat similar life-cycle variable is suggested by the model of DeAngelo,DeAngelo, and Stulz (2006), who use the fraction ofretained earnings relative to total equity (RETE). We alsoexamine RETE, so incorporating both this and the AGE

(footnote continued)

Nevertheless, such an approach corroborates the declining propensity to

pay cash dividends and the increasing propensity to repurchase.13 Because the older literature does not distinguish between the

probability of paying and the amount of cash dividends paid by payers,

we use identical variables to explain the likelihood to pay cash dividends

as to explain the amount of cash dividend paid by cash dividend payers.

Similarly, we use the same variables for an estimation of the probability

of repurchasing shares and the amounts repurchased by repurchasing

companies. We expect that this will deepen the insight regarding what

variables are empirically relevant for the four different aspects of

dividend policy.14 A negative relation might also exist if a company takes on

additional debt and then is prohibited by debt covenants from paying

additional dividends.

variable allows us to study which of the two life-cyclevariables explains more cross-sectional variation in thelikelihood to pay dividends or to repurchase shares.

Seventh, we test whether shareholder protectioninfluences the probability of paying dividends or repurch-asing shares by including a proxy for the company beingheadquartered in a common law versus a civil lawcountry. We use a dummy variable (COM ¼ 1) for thecommon law countries and predict that a companyoperating under common law is more likely to paydividends and repurchase shares than a company head-quartered in civil law countries.15 Eighth, we follow Bakerand Wurgler (2004b) in calculating a country-specificdividend-catering measure for each year. A countrycatering dummy variable (CCD) takes the value of one ifthe log of the median market-to-book value of a minimumof five payers in each country is larger than the log of themedian market-to-book value of a minimum of five non-payers in the same country, and zero otherwise.16 Cateringtheory predicts this variable will be positively related topayment likelihood.

Ninth, we introduce a dummy variable (EURO) for thecountries that adopt the euro as a currency in 1999 (plusGreece, which adopted it in 2001), to examine whethermonetary union has promoted financial convergenceamong firms from EMU countries. We do not have strongpriors regarding the sign of this variable. Tenth, we studywhether transportation companies show different prob-abilities of paying, compared to industrial firms, byintroducing a dummy variable (T) for firms in thetransportation sector. We again assume no pre-specifiedsign. Finally, we include a time variable to account for apossible trend. The existence of such a trend wouldsuggest a secular change in the likelihood to pay,unrelated to any of the variables presented above.

To test the impact of all the variables above, we makeuse of the longitudinal character of our data set byemploying panel regressions. Thus, we do not use thetechnique applied by Fama and French (2001). Instead,panel regression allows different intercepts for individualfirms and thus can account for unobserved variables thatdetermine these intercepts. Since we have known sinceLintner (1956) that dividend payments are sticky, compa-nies that paid dividends or repurchased shares last yearmay be much more likely to do so again this year thancompanies that did not distribute cash to investors. Panelregression techniques account for company-related un-observed variables, like the tendency to be sticky inpaying dividends, by allocating each company its ownintercept.

15 Though we do not apply the terminology of LaPorta, Lopez-de-

Silanes, Shleifer, and Vishny (2000), a positive sign would be in line with

their outcome model and a negative relation with their substitute model.16 We are aware that this variable differs from the catering variable

of Baker and Wurgler (2004a, b), who use the difference of the log of the

weighted and unweighted mean (rather than median) market-to-book

values of payers and non-payers. Because of the small number of

country-specific observations of payers and/or non-payers both means

are dominated by extreme outliers. For this reason we select the median

and set the requirement of a minimum of five observations on both

payers and non-payers.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374364

A second advantage of panel regressions is that wemay apply a statistical test to the time variable andevaluate whether the sheer passage of time is significant,and whether the likelihood to distribute cash is changing,irrespective of the other explanatory variables. A thirdadvantage of using this approach is that the Fama andFrench methodology uses independent variable coeffi-cients that are invariant over time. This approach is wellsuited to explaining the reduction in the propensity to payover a long time period, but one may question whetherthe coefficients of the independent variables do notchange over time. Our panel approach allows us tomeasure the changing impact of several variables overtime, while still studying whether—now with changedcoefficients—the likelihood to pay cash dividends is stilldeclining. We therefore estimate panel regressions forboth the whole period 1991–2005, with the variablespertaining to these 15 years, and for the sub-periods1991–1995, 1996–2000, and 2001–2005 with the corre-sponding sub-period variables.

We do not use estimation techniques that require theuse of several lagged variables, because the lack ofavailable data in the beginning of the period and theattrition of the database by companies that merge, delist,or go bankrupt (‘‘dead’’ companies) prohibits the use of abalanced panel, unless researchers accept a large reduc-tion in the number of observations. Nevertheless, in orderto reduce possible endogeneity problems, we lag allindependent variables once (unless they are time invar-iant or represent time itself). The lagged variables arepreceded by the capital letter L. The results of the logisticrandom effects panel regressions for the probability ofpaying are presented in Table 5, where Panel A providesthe results for the likelihood to pay cash dividends andPanel B presents results for the probability of repurchas-ing shares.

Panel A presents nine key findings about the prob-ability to pay cash dividends. First, the four traditionalvariables used in explaining the probability to pay in theUnited States also prove to be significant with theexpected signs in the European Union. The lagged sizevariable (LSIZE) is significant, with the expected positivesign, for the full period and all sub-periods, while thelagged earnings-to-assets ratio (LEA) always has a largeand significantly positive impact. The lagged market-to-book value of the firm (LMBF) always has a negative andsignificant sign, and the lagged asset growth rate (LDAA) issignificant with the expected negative sign, but only in thefinal sub-period, 2001–2005.

Second, the lagged earnings reporting frequency vari-able (LERF) has the expected negative coefficient for thefull study period and for two sub-periods, but issignificant only in the final sub-period. This suggests thatthe increase in earnings reporting frequency over time (asreported in Fig. 6) indeed explains a part of the reducedprobability to pay cash dividends. Third, privatizedcompanies (PRIV) are not likely to pay cash dividendssignificantly more (or less) frequently than other compa-nies. Fourth, the age variable proves to be significant andpositively related to the likelihood of paying dividends inall periods (except for the first sub-period), while the

lagged variable that captures the earned versus contrib-uted capital of the company (LRETE) is not significant. Thissuggests that the actual age of the company is a strongervariable in explaining the probability to pay cashdividends in the EU15 than is its ‘‘financial age.’’ Fifth,lagged leverage (LLR) proves to be highly significant andnegatively related to the probability to pay cash dividendsin all periods, implying that agency costs and/or financialcovenants are a major factor in explaining dividend policy.Sixth, the lagged country-specific catering dummy (LCCD)has a persistently negative sign in the EU15, suggestingfirms are less likely to pay dividends when the marketassigns a relative valuation premium to dividend-payingcompanies, though this variable is significant only for thewhole period (1991–2005) and after the year 2000.Seventh, a company that is headquartered in one of thetwo common law countries, Ireland or Britain, is morelikely to pay cash dividends, though the coefficient of theCOM dummy is not significant in the latest two sub-periods. This suggests that the integration of the EuropeanUnion may have reduced the differences between com-mon law and civil law countries with respect to thelikelihood of paying cash dividends. Eighth, the companiesthat belong to the euro area differentiated themselvesfrom the non-euro based companies by paying lessdividends in the first two periods, but this effect islessened and is insignificant for the latest period when theeuro was fully introduced. Finally, the other variables arenot significant in explaining the probability to pay cashdividends, except for the YEAR variable, which is sig-nificant and negative for the full period and the final sub-period. The persistently declining coefficient on thisvariable from insignificantly positive in the first periodto significantly negative in the final period confirms theresults of Panel A in Table 4. Even after the inclusion ofless traditional explanatory variables and after allowingfor variation in the coefficients of these variables, theinclination of EU companies to pay cash dividendsdeclines.

Panel B of Table 5 examines the impact of theexplanatory variables on the probability to repurchaseshares. In general, the explanatory power of the repurch-ase likelihood equations are smaller than those for theprobability to pay cash dividend regressions, thoughthe repurchase equations are still significant accordingto the Wald tests. Similarly, a smaller number of variablesare significant in the equations for the repurchaseequations. The only variable with a similar (negative)and significant sign for the whole period and all sub-periods is the lagged market-to-book value of the firm(LMBF): future growth prospects reduce the likelihood ofpaying cash dividends and the likelihood to repurchaseshares. For the entire 1991–2005 period and for the latesttwo sub-periods (1996–2005), lagged size (LSIZE) hassignificant and positive coefficients for the probability torepurchase, and this applies to lagged earnings (LEA) forthe whole period and the final sub-period. The laggedleverage ratio (LLR) is significantly negatively related tobuy-back likelihood for the whole period and after 1996.Two of the variables that distinguish Europe from theUnited States (LERF and PRIV) are not significant, while

ARTIC

LEIN

PRESS

Table 5Logistic random effects panel regression coefficients for the likelihood to pay cash dividends (Panel A), and the likelihood to repurchase shares (Panel B), listed industrial companies in the EU15 for 1991–2005

and three 5-year sub-periods

The dependent variable applies to cash dividends (Panel A) and to repurchases (Panel B) and is a dummy that takes the value of 1 if a company pays dividends, and zero if its dividends are zero. C indicates the

coefficient of the variable in the logistic regression equation, and P indicates the level of significance of the z-value of the coefficient based on bootstrapping 500 times. Variables preceded by the capital letter L

are 1-year lagged variables. ERF is the earnings reporting frequency, which takes the values between 1 and 4, representing the number of times per year earnings are reported. PRIV is a dummy variable that

takes the value of 1 if a company is privatized, and zero otherwise. SIZE is calculated as the percentile ranking of a company in the range of market values in the respective years, EA is the earnings ratio of a

company defined as the earnings before interest but after tax divided by the book value of assets, MBF is the market-to-book value of the firm, DAA is the relative change in assets at constant prices of a

company, RETE is retained earnings to total equity, LR is the leverage rate calculated from book values, SDS is the standard deviation of net income during the past 5 years related to actual annual sales, AGE is

the age of the company. T is a dummy variable equal to 1 if the company is a transportation company, and otherwise zero. CCD is a dummy that takes the value of one if the median market-to-book value of cash

dividend payers in the relevant country is larger than the median market-to-book value of no cash payers, and zero otherwise. COM is a dummy indicating whether a company originates from a common law

country (COM ¼ 1), otherwise zero. EURO is a dummy that takes the value 1 if a country originates from a country that accepted the euro as its currency, and zero in the case of the United Kingdom, Denmark,

and Sweden. YEAR is the observation year (1991, 1992,y, 2005). CONST is the intercept of the regression equation. Observ. is the number of observations for which all information was available, Groups is the

number of groups for which observations were available, Wald is the Wald chi-square of the equation for 15 independent variables, and Prob. indicates the significance of the equation.

Panel A: Likelihood to pay cash dividends Panel B: Likelihood to repurchase shares

1991–2005 1991–1995 1996–2000 2001–2005 1991–2005 1991–1995 1996–2000 2001–2005

C P C P C P C P C P C P C P C P

LERF �0.040 0.460 �0.175 0.184 0.142 0.188 �0.181 0.005 0.080 0.155 �0.149 0.421 0.212 0.062 0.044 0.564

PRIV �0.200 0.619 �1.233 0.058 �0.182 0.788 0.476 0.378 0.575 0.060 0.220 0.962 0.735 0.064 0.631 0.106

LSIZE 0.057 0.000 0.053 0.000 0.045 0.000 0.062 0.000 0.017 0.000 0.005 0.224 0.017 0.000 0.022 0.000

LEA 10.323 0.000 16.059 0.000 9.106 0.000 11.178 0.000 2.253 0.000 0.556 0.643 0.060 0.923 3.694 0.000

LMBF �0.378 0.000 �0.740 0.000 �0.374 0.000 �0.513 0.000 �0.164 0.000 �0.490 0.040 �0.242 0.001 �0.140 0.014

LDAA 0.001 0.986 �0.021 0.873 0.019 0.826 �0.345 0.000 �0.120 0.134 �1.076 0.045 �0.026 0.860 �0.269 0.106

LRETE 0.005 0.104 0.028 0.561 0.001 0.795 0.007 0.499 0.002 0.495 0.004 0.867 0.003 0.759 0.001 0.879

LLR �3.612 0.000 �6.495 0.000 �4.621 0.000 �1.981 0.000 �1.201 0.000 �0.163 0.784 �1.288 0.016 �1.662 0.000

LSDS �0.195 0.734 �0.044 0.970 �0.730 0.395 �1.065 0.438 �0.002 0.979 0.039 0.836 �0.789 0.113 �0.039 0.700

AGE 0.017 0.000 0.006 0.090 0.007 0.018 0.019 0.000 0.002 0.109 0.003 0.338 0.002 0.364 0.001 0.530

T �0.080 0.874 �0.531 0.575 �0.351 0.640 �0.191 0.732 0.486 0.138 0.606 0.681 �0.084 0.877 0.463 0.280

LCCD �0.421 0.000 �0.130 0.400 �0.151 0.426 �0.509 0.000 �0.253 0.003 0.519 0.015 �0.243 0.266 �0.166 0.157

COM 0.783 0.000 1.167 0.004 0.474 0.208 0.387 0.177 �0.021 0.915 1.407 0.002 0.447 0.104 �0.496 0.049

EURO �0.416 0.047 �0.832 0.042 �0.920 0.015 �0.441 0.084 �0.125 0.499 1.198 0.006 �0.565 0.035 �0.279 0.206

YEAR �0.161 0.000 0.043 0.504 �0.072 0.140 �0.182 0.000 0.182 0.000 0.089 0.225 0.442 0.000 0.147 0.000

CONST 320.890 0.000 �82.679 0.515 146.003 0.135 364.080 0.000 �367.389 0.000 �182.025 0.211 �886.943 0.000 �297.017 0.000

Observ. 22917 5478 7043 10396 19507 4788 6156 8563

Groups 3377 1514 1811 2903 2948 1348 1669 2408

Wald 1586.2 388.7 376.4 963.5 444.6 26.95 169.2 241.8

Prob. 0.000 0.000 0.000 0.000 0.000 0.029 0.000 0.000

H.

vo

nE

ije,W

.L.M

egg

inso

n/

Jou

rna

lo

fFin

an

cial

Eco

no

mics

89

(20

08

)3

47

–3

74

36

5

ARTIC

LEIN

PRESS

Table 6Panel A Random effects panel regressions for the real amounts paid by listed industrial cash dividend payers in the EU15 for the whole period (1991–2005) and three sub-periods of 5 years

The dependent variable is the natural logarithm of the amount of cash dividends paid by cash dividend payers in million euros in prices of 2000. C indicates the coefficient of the variable in the regression

equations. P represents the level of significance of the z-value calculated from robust estimates. Variables preceded by the capital letter L are 1-year lagged variables. ERF is the earnings reporting frequency,

which takes the values between 1 and 4, representing the number of times per year earnings are reported. PRIV is a dummy variable that takes the value of 1 if a company is privatized, and zero otherwise. SIZE

is calculated as the percentile ranking of a company in the range of market values in the respective years, EA is the earnings ratio of a company defined as the earnings before interest but after tax divided by the

book value of assets, MBF is the market-to-book value of the firm, DAA is the relative change in assets at constant prices of a company, RETE is retained earnings to total equity, LR is the leverage rate calculated

from book values, SDS is the standard deviation of net income during the past 5 years related to actual annual sales, AGE is the age of the company. T is a dummy variable equal to 1 if the company is a

transportation company, and otherwise zero. CCD is a dummy that takes the value of one if the median market-to-book value of cash dividend payers in the relevant country is larger than the median market-to-

book value of no cash payers, and zero otherwise. COM is a dummy indicating whether a company originates from a common law country (COM ¼ 1), otherwise zero. EURO is a dummy that takes the value 1 if a

country originates from a country that accepted the euro as its currency, and zero in the case of the United Kingdom, Denmark, and Sweden. YEAR is the observation year (1991, 1992,y, 2005). CONST is the

intercept of the regression equation. Observ. is the number of observations for which all information was available, Groups is the number of groups for which observations were available, Wald is the Wald

chi-square of the equation for 15 independent variables, and Prob. indicates the significance of the equation. R2 within is the within group R-square, R2 between is the between groups R-square, and R2 overall is

the overall R-square.

1991–2005 1991–1995 1996–2000 2001–2005

C P C P C P C P

LERF 0.027 0.028 0.064 0.001 0.104 0.000 0.041 0.053

PRIV 1.416 0.000 0.521 0.012 1.237 0.000 1.273 0.000

LSIZE 0.047 0.000 0.059 0.000 0.048 0.000 0.057 0.000

LEA 1.573 0.000 1.281 0.005 0.810 0.000 1.830 0.000

LMBF �0.125 0.000 �0.339 0.000 �0.147 0.000 �0.193 0.000

LDAA 0.002 0.000 �0.011 0.108 0.004 0.000 �0.101 0.001

LRETE 0.001 0.442 �0.009 0.567 �0.001 0.318 0.015 0.008

LLR �0.435 0.000 �0.516 0.002 �0.342 0.001 �0.586 0.000

LSDS �0.002 0.821 0.007 0.000 �0.070 0.029 �0.039 0.283

AGE 0.005 0.000 0.000 0.916 0.002 0.001 0.005 0.000

T 0.225 0.162 �0.182 0.442 0.001 0.994 0.194 0.333

LCCD �0.204 0.000 �0.113 0.000 �0.029 0.302 �0.182 0.000

COM 0.126 0.022 0.174 0.002 0.166 0.016 0.020 0.806

EURO �0.129 0.027 �0.097 0.086 �0.127 0.082 �0.175 0.022

YEAR 0.003 0.104 �0.026 0.000 0.062 0.000 �0.080 0.000

CONST �8.148 0.041 51.042 0.001 �125.234 0.000 158.469 0.000

Observ. 17321 4614 5920 6787

Groups 2611 1350 1627 2032

Wald 10133.0 6531.1 4405.3 4015.5

Prob. 0.000 0.000 0.000 0.000

R2 within 0.287 0.192 0.235 0.090

R2 between 0.741 0.854 0.798 0.726

R2 overall 0.744 0.825 0.792 0.713

H.

vo

nE

ije,W

.L.M

egg

inso

n/

Jou

rna

lo

fFin

an

cial

Eco

no

mics

89

(20

08

)3

47

–3

74

36

6

ARTIC

LEIN

PRESS

Panel B. Random effects panel regressions for the real amounts repurchased by listed industrial repurchasers in the EU15 for the whole period (1991–2005) and three sub-periods of 5 years

The dependent variable is the natural logarithm of the repurchased amounts in million euros by repurchasing companies in prices of 2000. C indicates the coefficient of the variable in the regression equations.

P represents the level of significance of the z-value calculated from robust estimates. Variables preceded by the capital letter L are 1-year lagged variables. ERF is the earnings reporting frequency, which takes

the values between 1 and 4, representing the number of times per year earnings are reported. PRIV is a dummy variable that takes the value of 1 if a company is privatized, and zero otherwise. SIZE is calculated

as the percentile ranking of a company in the range of market values in the respective years, EA is the earnings ratio of a company defined as the earnings before interest but after tax divided by the book value of

assets, MBF is the market-to-book value of the firm, DAA is the relative change in assets at constant prices of a company, RETE is retained earnings to total equity, LR is the leverage rate calculated from book

values, SDS is the standard deviation of net income during the past 5 years related to actual annual sales, AGE is the age of the company. T is a dummy variable equal to 1 if the company is a transportation

company, and otherwise zero. CCD is a dummy that takes the value of one if the median market-to-book value of cash dividend payers in the relevant country is larger than the median market-to-book value of

no cash payers, and zero otherwise. COM is a dummy indicating whether a company originates from a common law country (COM ¼ 1), otherwise zero. EURO is a dummy that takes the value 1 if a country

originates from a country that accepted the euro as its currency, and zero in the case of the United Kingdom, Denmark, and Sweden. YEAR is the observation year (1991, 1992,y, 2005). CONST is the intercept of

the regression equation. Observ. is the number of observations for which all information was available, Groups is the number of groups for which observations were available, Wald is the Wald chi-square of the

equation for 15 independent variables, and Prob. indicates the significance of the equation. R2 within is the within group R-square, R2 between is the between groups R-square, and R2 overall is the overall

R-square.

1991–2005 1991–1995 1996–2000 2001–2005

C P C P C P C P

LERF 0.016 0.815 0.636 0.020 0.206 0.116 0.035 0.688

PRIV 1.420 0.000 �0.987 0.131 1.256 0.007 1.259 0.000

LSIZE 0.061 0.000 0.044 0.000 0.061 0.000 0.068 0.000

LEA 0.711 0.186 �3.940 0.000 1.229 0.291 1.776 0.002

LMBF �0.168 0.001 �0.012 0.967 �0.306 0.000 �0.162 0.013

LDAA �0.218 0.000 0.338 0.639 �0.318 0.001 �0.250 0.005

LRETE 0.000 0.966 �0.081 0.037 0.008 0.490 �0.021 0.245

LLR �0.557 0.102 �0.341 0.760 �0.657 0.209 �0.700 0.122

LSDS �0.013 0.001 �0.015 0.000 0.853 0.314 0.030 0.565

AGE 0.005 0.012 �0.001 0.902 0.000 0.973 0.005 0.018

T �0.111 0.751 �1.295 0.040 �0.213 0.664 �0.121 0.777

LCCD �0.256 0.030 0.624 0.126 �0.011 0.971 �0.272 0.073

COM �0.089 0.676 0.624 0.455 �0.644 0.018 0.028 0.914

EURO �0.743 0.000 0.114 0.885 �1.091 0.001 �0.641 0.007

YEAR �0.033 0.032 �0.101 0.494 0.060 0.310 �0.105 0.003

CONST 64.411 0.040 197.053 0.502 �121.702 0.302 206.762 0.004

Observ. 2083 178 619 1286

Groups 985 128 393 724

Wald 963.9 140.5 459.8 750.5

Prob. 0.000 0.000 0.000 0.000

R2 within 0.008 0.076 0.001 0.006

R2 between 0.456 0.450 0.514 0.469

R2 overall 0.445 0.398 0.483 0.470

H.

vo

nE

ije,W

.L.M

egg

inso

n/

Jou

rna

lo

fFin

an

cial

Eco

no

mics

89

(20

08

)3

47

–3

74

36

7

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374368

the euro area has a significantly positive sign in the firstsub-period and a significantly negative sign in the secondsub-period. Companies in common law countries (Britainand Ireland) are more likely to repurchase shares in thefirst sub-period than are companies headquartered in civillaw countries, but this tendency reverses in the2001–2005 sub-period. Company age is never signifi-cantly related to the probability to repurchase. The laggedcountry-specific catering dummy used here (LCCD) has asignificantly negative coefficient over the whole periodand the first sub-period, which casts doubt on theapplicability of the catering theory to repurchases.17

Finally, the YEAR variable proves to be significant for thewhole period and for the final two periods, but now thesign is positive. This implies there is a secular increase inthe likelihood to repurchase, as suggested earlier by Fig. 2and Table 4, Panel B.

6. Explaining dividend amounts

We now investigate factors influencing the amountsof real dividends paid and shares repurchased, usingthe same explanatory variables as used for the probabil-ities to pay. We only study the amounts paid by dividendpayers and repurchasers, corrected for inflation, andexclude non-payers. Panel A of Table 6 describes real cashdividends paid by dividend payers and Panel B presentsthe real euro amounts of stock repurchases. We makefull use of our data set and once more employ panelregressions.

Perhaps the most interesting findings presented inPanel A are the highly significant effects of the twovariables that most differentiate the European Union fromthe United States—changing financial reporting frequencyand privatization history. The lagged earnings reportingfrequency (LERF) proves to be significantly related to theamount of cash dividends paid (except for the latestperiod where it is insignificant) though, intriguingly, thesign of the coefficient is positive. Wood (2001) suggeststhat dividends may be less relevant in signaling companyvalue because of improved reporting, so we expected anegative sign. However, if a company signals its quality byfrequent reporting, it does not pay cash dividends asfrequently nowadays (see Table 5, most recent period)—but if it does, it can afford to pay more.

Privatized companies (PRIV ¼ 1) that pay cash divi-dends also pay more than comparable firms that werenever state owned (PRIV ¼ 0). Once again, the distinctionbetween the probability to pay and the amount of

17 If our catering measure would have provided correct signs for the

probability to pay cash dividends, we might have considered that the use

of a catering measure based on cash dividend catering should be blamed

for the wrong sign and that we should use a separate catering measure

for repurchases. This, however, is not very likely, as we also find that

repurchases and cash dividends are strongly positively correlated in the

EU15. Moreover, the fact that we use a lagged catering variable may be

affecting our results for both the probability to pay cash dividends and

for the probability to repurchase. Overall, we conclude that if the

catering theory is relevant in the EU15 countries, it requires a separate

study that goes into the intricacies of measuring catering effects more

deeply than we are able to do here.

dividends is relevant: privatized companies are not morelikely to pay dividends than always-private companies(see Table 5), but they pay more if they do pay.

Lagged company size (LSIZE), lagged relative companyearnings (LEA), and lagged market-to-book value of thefirm (LMBF) are important determinants of cash dividendpayment amounts with the expected signs for theircoefficients. However, the need to finance assets mea-sured by the lagged relative change in assets (LDAA) isonly significant, with a negative coefficient, during thelatest sub-period. As we find a negative sign for thisvariable in the first sub-period, a significantly positivesign for this variable in the second, and a significantlynegative sign again in the third sub-period, the need tofinance assets by reducing dividend amounts seems to becyclical. To some extent this is corroborated by the factthat we find a similar cyclical pattern for the LDAAvariable in our probability to pay cash dividend equations.

Company age is again significantly positively related tocash dividends paid by payers (except for the first period),while the lagged retained earnings to total equity (LRETE)variable is insignificant, except for the final period. Duringand after the shake-out of companies in the internetbubble (during the 2001–2005 sub-period), the fraction ofearned to contributed equity of companies may havebecome more significant in determining how much to payif a company pays cash dividends. Lagged leverage (LLR)also seems to have similar effects on the amounts of cashdividends as it has on the probability to pay suchdividends. Its sign is always significantly negative andmore leverage may allow or require companies to retainmore cash rather than pay dividends. Once again, laggedincome risk (LSDS) does not have a systematic relationwith the amount of dividends paid, though it has theexpected significant negative effect in the period1996–2000. Also the lagged country-specific cateringdummy (LCCD) has a significant negative sign for the fullperiod and for the first and third sub-periods. This is the‘‘wrong’’ sign, if we extend the catering theory outside itsoriginal realm of probabilities to pay to cash dividendamounts of payers. Generally, common law countries(COM) pay higher cash dividend amounts, but the effect isnot significant in the latest period. This suggests againthat the unification of the European Union may graduallymitigate the original relevance of law systems fordividend policy. However, companies from the euro areapay lower cash dividends, significantly so in the finalperiod. This suggests that the introduction of the euroarea may have become a (new) distinguishing factor individend amounts paid. Whether a company is from thetransportation industry is not significant. Finally, the YEARvariable is significant for the entire study and in all threesub-periods, though the coefficient signs differ fromobservation period to observation period.

Panel B of Table 6 presents results from estimating theamount that share repurchasing companies pay out. Thesignificance levels of the equations are again lower for theamount of repurchases than for the amount of cashdividends. The lagged earnings reporting frequency (LERF)only has a significant (positive) impact in the first sub-period (1991–1995). The increase in reporting frequencies

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 369

presented in Fig. 6 may have reduced the impact of thisvariable on amounts of share repurchases. However,privatized companies (PRIV) tend to repurchase largeramounts after 1995 than they did previously, as indicatedby the significant positive coefficients in the latest twosub-periods. Again, the lagged company size variable(LSIZE) is always significantly positive, which is notsurprising if one considers the results of Table 3, wherewe find that large companies also distribute more throughrepurchases. Very interestingly, the lagged earningsvariable flips sign, from significantly negative in the firstsub-period to significantly positive in the last. This is inline with the findings of Brav, Graham, Harvey, andMichaely (2005) and Skinner (2008), who find that sharerepurchases have become increasingly sensitive to earn-ings. Moreover, the ability of repurchases to cater flexiblyto company needs is confirmed by the development of thecoefficients of the two lagged growth variables (LMBF andLDAA), which both give significant negative coefficients inthe latest two periods.

We find that company age is significantly positivelyrelated to amount of shares repurchased for the fullstudy period and for the final sub-period, while compa-nies headquartered in countries that adopt the euro astheir currency buy back significantly lower share amountsafter 1995 than do firms in countries that retain theirown currencies. There is an overall declining trend inthe amount of money spent by individual repurchasingcompanies, though the YEAR variable is significantonly for the full study period and for the final sub-period (2001–2005). The lagged income risk variable(LSDS) is significantly negative for the full study periodand for the first sub-period but not after 1995, suggest-ing that income volatility becomes less of an issueover time. This is indirectly in line with the findingspresented in Brav, Graham, Harvey, and Michaely (2005)and Skinner (2008): if firms consider the amount ofmoney spent on stock repurchases as a flexible instrumentfor distributing cash, they may experience less fear forincome risk and the risk itself may no longer be asignificant variable.

18 We use the same formula as LaPorta, Lopez-de-Silanes, Shleifer,

and Vishny (2000, pp. 28–29) for all 15 EU countries and all years

2001–2005. We calculate the values in each year’s implied income tax

bracket by inflating the figures with the consumer price index in order to

have a real income of 50,000 euro in the year 2000. As there were no

exact data available for Spain we use a dividend tax rate between 45%

and 48% in the calculations for that country.

7. Robustness and tests on earnings responsiveness

7.1. Robustness checks for the probability to pay dividends

and repurchase shares

This section examines more closely how the factorsdiscussed above and in other empirical studies impactpayout policies of EU firms. We focus on the final sub-period 2001–2005, since only this period offers compre-hensive data on several variables. We add five newvariables as robustness checks on our previous results.First, we add a variable indicating the cash dividend taxpreference (DTP) in each country. This variable isoriginally presented in LaPorta, Lopez-de-Silanes, Shleifer,and Vishny (2000), but we update the variable for themost recent period (2001–2005) using data for therespective years from the Worldwide Corporate Tax Guide

and the Global Executive, both published by Ernst and

Young.18 We assume that the probability to pay anddividend amounts paid will be higher in countries wherethe tax laws favor cash dividends (or penalize paymentsless) relative to earnings retention. We do not use a laggedvariable here, as it is unlikely that the tax system will beinfluenced by dividend payments. Moreover, it is likelythat companies will know the tax system and if they basetheir payments on the cash dividend tax preference, it islikely that they will do so for the system that is relevant inthe year of dividend payment. To our knowledge, this isthe first multi-year, multi-national test of the impact ofdividend taxation on the likelihood to pay dividends andrepurchase shares, and on the amounts paid and repurch-ased.

Second, we test the impact of share liquidity byincluding the number of stock market transactions perday in December of the previous year, divided by thenumber of outstanding shares at the end of that year,multiplied by 1,000 (LRELTM). This variable, taken fromthe Amadeus database, characterizes how easily investorscan make their dividends ‘‘at home,’’ as suggested byBanerjee, Gatchev, and Spindt (2005). In line with theirarguments, we hypothesize that relative liquidity willreduce the probability to pay cash dividends, though wealso test whether this variable influences the probabilityto repurchase as well as the amounts of cash dividendsand share repurchases by payers. Third, we study theeffect of two additional variables from the literature onshare repurchases as used most recently by Skinner(2008). The first is the availability of cash, measured hereas cash to total assets, lagged one period (LCASHA). Thesecond variable is the annual change in stock pricesmeasured at the end of the previous year (LDPP). Largercash holdings should make dividends and repurchasesmore likely, and the same effects are predicted fordecreases in share prices. Finally, we add a variable fromthe Amadeus database measuring the dependence of acompany on a large shareholder (DEP). We classify acompany as dependent if there is a shareholder with totalownership of over 50%. This variable is invariant overtime. We have no strong priors regarding the sign of thisvariable, which is suggested by Johnson, La Porta, Lopez-de-Silanes, and Schleifer (2000) as a measure of tunneling.Since dividends are generally paid to outsiders, as well asto blockholders, more dividend payments by moreblockholder dependent companies may indicate that EUcorporate governance systems effectively discouragemajor shareholders from tunneling cash out of the firmsthey control.

The second and third columns of Table 7 indicate theimpact of the independent variables on the probabilityto pay cash dividends. Overall, the significance of most ofthe original coefficients presented in Table 5 (the final

ARTICLE IN PRESS

Table 7Robustness tests on the significance of the coefficients of dividend policy for listed industrial companies in the EU15 for 2001–2005

The dependent variables are respectively (1) a dummy that takes the value of 1 if a company paid cash dividends, and zero otherwise; (2) a dummy that

takes the value of 1 if the company repurchases, and zero otherwise; (3) the natural logarithm of the amount of real cash dividends paid by payers; (4) the

natural logarithm of the amount of real repurchases of repurchasing companies. C indicates the coefficient of the variable in the regression equations.

P indicates the level of significance of the z-value of the coefficient, where the z-values are calculated from bootstrapping 500 times with the likelihood to

pay equations and from robust estimates for the real dividends equations. The variables refer to the period 2001–2005. Variables preceded by the capital

letter L are 1-year lagged variables. ERF is the earnings reporting frequency, which takes the values between 1 and 4, representing the number of times per

year earnings are reported. PRIV is a dummy variable that takes the value of 1 if a company is privatized, and zero otherwise. SIZE is calculated as the

percentile ranking of a company in the range of market values in the respective years, EA is the earnings ratio of a company defined as the earnings before

interest but after tax divided by the book value of assets, MBF is the market-to-book value of the firm, DAA is the relative change in assets at constant

prices of a company, RETE is retained earnings to total equity, LR is the leverage rate calculated from book values, SDS is the standard deviation of net

income during the past 5 years related to actual annual sales, AGE is the age of the company. T is a dummy variable equal to 1 if the company is a

transportation company, and otherwise zero. CCD is a dummy that takes the value of 1 if the median market-to-book value of cash dividend payers in the

relevant country is larger than the median market-to-book value of no cash payers, and zero otherwise. COM is a dummy indicating whether a company

originates from a common law country (COM ¼ 1), otherwise zero. EURO is a dummy that takes the value 1 if a country originates from a country that

accepted the euro as its currency, and zero in the case of the United Kingdom, Denmark, and Sweden. YEAR is the observation year (1991, 1992,y, 2005).

DTP is the dividend tax preference calculated for each country every year according to the method of LLSV (2000, pp. 28–29). RELTM is the relative

number of transactions in percentage of outstanding shares. DPP is the relative change in the end of year stock price. CASHA is the amount of cash related

to total book value of assets. DEP is a dummy variable that takes the value of 1 if the company has a shareholder that owns at least 50% of the shares.

CONST is the intercept of the regression equation. Observ. is the number of observations for which all information was available, Groups is the number of

groups for which observations were available, Wald is the Wald chi-square of the equation for 15 independent variables, and Prob. indicates the

significance of the equation. R2 within is the within group R-square, R2 between is the between groups R-square, and R2 overall is the overall R-square.

Likelihood to Amounts paid by

Pay cash dividends Repurchase Cash dividend payers Repurchasers

C P C P C P C P

LERF �0.188 0.010 �0.004 0.960 0.038 0.094 �0.037 0.712

PRIV 0.576 0.290 0.452 0.306 1.276 0.000 0.950 0.007

LSIZE 0.060 0.000 0.024 0.000 0.058 0.000 0.073 0.000

LEA 11.571 0.000 4.867 0.000 1.803 0.000 2.116 0.009

LMBF �0.433 0.000 �0.139 0.037 �0.192 0.000 �0.142 0.046

LDAA �0.252 0.001 �0.476 0.003 �0.107 0.003 �0.299 0.143

LRETE 0.037 0.190 0.000 0.995 0.012 0.001 �0.026 0.102

LLR �3.079 0.000 �0.820 0.076 �0.412 0.005 0.052 0.922

LSDS �3.382 0.005 �0.068 0.810 �0.411 0.056 �0.007 0.914

AGE 0.017 0.000 �0.001 0.522 0.005 0.000 0.005 0.065

T �0.353 0.559 0.119 0.793 0.174 0.421 0.026 0.957

LCCD �0.451 0.001 �0.181 0.194 �0.144 0.000 �0.217 0.205

COM 0.729 0.037 �0.625 0.030 0.124 0.154 0.014 0.962

EURO �0.465 0.138 �0.293 0.256 �0.084 0.325 �0.362 0.170

YEAR �0.134 0.001 0.121 0.007 �0.071 0.000 �0.089 0.043

DTP 3.177 0.000 �0.518 0.568 �0.587 0.020 �0.155 0.883

LRELTM �0.033 0.366 0.025 0.142 0.003 0.768 0.071 0.001

LDPP �0.065 0.287 �0.090 0.221 �0.003 0.875 �0.038 0.765

LCASHA �2.319 0.001 1.853 0.000 0.606 0.002 1.100 0.066

DEP �0.059 0.780 �0.367 0.058 �0.113 0.126 �0.596 0.011

CONST 265.901 0.001 �246.201 0.006 141.265 0.000 173.905 0.048

Observ. 7198 5881 5076 994

Groups 1954 1595 1485 562

Wald 640.5 178.3 3442.4 645.9

Prob. 0.000 0.000 0.000 0.000

R2 within 0.103 0.013

R2 between 0.737 0.481

R2 overall 0.711 0.490

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374370

sub-period) are unchanged, except that now the laggedincome risk (LSDS) is significant with the expectednegative sign and the common law countries (COM) nowhave a significantly higher probability to pay. With respectto the new variables, dividend tax preference (DTP) ishighly significant and positively related to the probabilityto pay cash dividends. While it may be unsurprising thattax preferences in fact encourage cash dividend payments,this finding is an important new result. Contrary to

expectations, however, the lagged relative amount of cashto assets (LCASHA) has a significant negative sign.

When we study the effect of adding additionalvariables to the probability to repurchase estimations,presented in the fourth and fifth columns of Table 7, wefind that the lagged change in assets (LDAA) becomessignificant with the expected negative sign, while thelagged leverage rate (LLR) becomes insignificant. Thesubstitution effect of leverage on the probability to

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 371

repurchase shares may thus be smaller than previouslysuggested. The persisting significant negative coefficienton COM indicates that share repurchases are less likely forcompanies headquartered in common law countries thanfor those in civil law countries—despite the fact thatBritish companies typically buy back roughly half thevalue of EU shares repurchased each year. The lagged cashto assets ratio (LCASHA) is significant and positive, asexpected.

7.2. Robustness checks for the amount of dividends paid and

shares repurchased

Table 7 also provides robustness checks for the amountof cash dividends paid (columns 6 and 7) by cash dividendpayers and for the amount of shares bought back(columns 8 and 9) by repurchasers. The addition of thefive new variables has hardly any effect on the originalvariables that determine the amount of cash dividendspaid by cash dividend payers. Only the impact of the euromaterially changes, becoming insignificant in explainingthe amounts paid by cash dividend payers. Furthermore,the dividend tax preference (DTP) is significant, but withan unexpected negative sign, suggesting that the same taxpreferences which increase the likelihood of paying cashdividends also decrease the actual amount of dividendspaid by these firms. Of the other additional variables, thelagged relative amount of cash to assets (LCASHA) has apositive influence on the amount of cash dividends paidby payers.

Finally, the inclusion of five additional variablesreduces the significance of three coefficients for theamounts repurchased by share repurchasers. The impactof the lagged relative change in assets (LDAA) becomesinsignificant at the 5% level, as do the dummy variableproxying for countries that adopt the euro (EURO) and theage (AGE) variable. Two of the newly introduced variablesare significantly related to the amount of shares repurch-ased. The lagged relative number of transactions (LRELTM)is positively related to the amount of repurchases,indicating that companies with more liquid stock re-purchase more, while firms that are more dependent on amajor shareholder (DEP) repurchase less.

7.3. Are total payouts becoming more responsive to

earnings?

Finally, we test whether the linkage between cashdividends and earnings has weakened, as Brav, Graham,Harvey, and Michaely (2005) and Skinner (2008) find forU.S. data, and whether the total payout of companies thatboth pay cash dividends and repurchase shares regularlyhas become more sensitive to earnings. We first searchfirms that are regular payers as well as regular repurcha-sers. We find 291 industrial companies that pay cashdividends at least every 5 years and that repurchase atleast three times in the 10-year period 1996–2005. Thesecompanies represent 46% of all cash dividends paidbetween 1989 and 2005 and 72% of all repurchases,though they only represent 7% of the companies that pay

cash dividends and 10% of the companies that repurchaseduring the period 1989–2005. Then we calculate thefollowing Lintner model regressions for cash dividends,repurchases, and total payouts:

DDt ¼ a0 þ a1 EBIt � a2 Dt�1 þ �t (1)

where DDt is the change in the dividend variable (cashdividends, repurchases and total payout) for the alter-native regressions at time t, EBIt the earnings beforeinterest but after tax, Dt�1 ¼ the lagged dividend variable,namely lagged cash dividends (LCD), lagged repurchases(LREP), and lagged total payouts (LTOT), respectively, andet ¼ the error term of the equation.

In Eq. (1) a1 represents the sensitivity to earnings anda2 the speed of adjustment. We estimate Eq. (1) for twosub-periods (1996–2000 and 2001–2005) in order to testwhether a1 and a2 change over time. Moreover, likeSkinner (2008), we apply these regression equations toboth annual and biennial data.

In Table 8, Panel A we find a smaller effect of earningson cash dividends for annual data in the 2001–2005 sub-period compared with 1996–2000, but the difference isnot significant. However, from the biennial data we find asignificantly larger coefficient for the earnings variable inthe equation for cash dividends in the EU15 in the mostrecent sub-period (0.212 in the 2001–2005 sub-periodversus 0.109 in the 1996–2000 sub-period). This isunexpected, as Brav, Graham, Harvey, and Michaely(2005) and Skinner (2008) find a weakened relationbetween earnings and cash dividends over time. More-over, the speed of adjustment of cash dividends toearnings increases significantly with time, from 0.098 in1996–2000 to 0.393 in 2001–2005. For repurchases wefind similar increases for the earnings coefficient of sharerepurchases in the second period. For both the annual andthe biennial data the increase is significant (from 0.058 to0.114 for annual data and from 0.033 to 0.065 for biennialdata). However, the speed of adjustment of sharerepurchases to earnings declines significantly for bothtypes of data (from 0.754 to 0.288 for regressions based onannual data and from 0.325 to 0.083 for regressions basedon biennial data). So share repurchases become increas-ingly dependent on earnings, but the companies in theEU15 that regularly distribute become more reluctant tolet repurchases follow earnings quickly.

Finally, the regressions based on annual data, pre-sented in Panel C, show an increase in the impact ofearnings on total payout, from 0.327 to 0.396 for annualdata, though the increase is not significant. This increasedsensitivity of payout to earnings changes is similar to theU.S. results in Skinner (2008), though his findings arestatistically significant. In the EU15 the effect is entirelydue to repurchases following earnings significantly moreclosely in the 2001–2005 sub-period. Overall, however,earnings still influence cash dividends significantly morethan repurchases, as is indicated by the coefficient of 0.211for cash dividends during 2001–2005, compared to thecoefficient of 0.114 for repurchases for the same sub-period. Biennial data shows a similar increase in thesensitivity of total payout to earnings (from 0.165 to0.339), but now with a significant coefficient. In line with

ARTICLE IN PRESS

Table 8Lintner model regressions of cash dividends, repurchases and total

corporate payouts for 291 listed industrial companies in the EU15 that

paid cash dividends during at least 5 years and repurchased at least three

times in the period 1996–2005

Pooled regression analysis is shown with robust standard errors

clustered at the company level for observations on an annual level and

on a biannual level. The dependent variable is the change in cash

dividends, repurchases, or total company payout. C is the coefficient of

the explanatory variable and P is the probability of the t-value. The

independent variables are earnings before interest but after tax (EBI), the

lagged value (indicated with L) of cash dividends (CD), repurchases

(REP), and total company payout (TOT). N and (Adj R2) indicate the

number of observations and the adjusted R2 for the equation respec-

tively. The 291 companies in this group represent 46% of all cash

dividends paid during the period 1989–2005 and 72% of all repurchases

(50% and 73% respectively in the year 2005), though they only represent

7% of the companies that paid cash dividends and 10% of the companies

that repurchased during the period 1989–2005 (5% and 7% respectively

for 2005).

Annual data Biennial data

C P C P

Panel A: Cash dividends 1996–2000

EBI 0.253 0.000 0.109 0.000

LCD �0.547 0.001 �0.098 0.123

Intercept �4.754 0.227 �8.091 0.056

N (Adj R2) 1333 (0.310) 1306 (0.165)

Panel A: Cash dividends 2001–2005

EBI 0.211 0.000 0.212b 0.000

LCD �0.392 0.000 �0.393a 0.000

Intercept �8.877 0.026 �28.132 0.007

N (Adj R2) 1201 (0.320) 1195 (0.425)

Panel B: Repurchases 1996–2000

EBI 0.058 0.002 0.033 0.002

LREP �0.754 0.000 �0.325 0.000

Intercept 13.033 0.015 14.623 0.006

N (Adj R2) 1322 (0.330) 1290 (0.143)

Panel B: Repurchases 2001–2005

EBI 0.114a 0.000 0.065a 0.000

LREP �0.288a 0.015 �0.083b 0.404

Intercept �9.066a 0.238 �16.587a 0.149

N (Adj R2) 1165 (0.238) 1157 (0.169)

Panel C: Total payout 1996–2000

EBI 0.327 0.000 0.165 0.000

LTOT �0.608 0.000 �0.168 0.005

Intercept 5.670 0.282 �0.382 0.952

N (Adj R2) 1342 (0.346) 1317 (0.188)

Panel C: Total payout 2001–2005

EBI 0.396 0.000 0.339a 0.000

LTOT �0.502 0.000 �0.439b 0.000

Intercept �13.742b 0.094 �33.110a 0.007

N (Adj R2) 1204 (0.445) 1199 (0.439)

a Indicates the coefficients found in the period 2001–2005 differ

significantly from the coefficients from the period 1996–2000 at the 1%

significance level.b Indicates the coefficients found in the period 2001–2005 differ

significantly from the coefficients from the period 1996–2000 at the 5%

significance level.

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374372

the U.S. results of Skinner (2008), we find that the speed ofadjustment coefficient on total payout also becomessignificantly larger in the latest period for biennial data.

As we find a decline in responsiveness of share re-purchases, the effect in this case is due to faster responsein cash dividend payments.

8. Summary and conclusions

Using a database of over 4,100 listed industrialcompanies, we examine the evolution of payout policyfrom 1989 to 2005 in the 15 nations that were members ofthe European Union before May 2004. We find thatdividend and share repurchase policies of EU companiesare similar in many ways to those of American firms. Forexample, the fraction of European firms paying dividendshas also declined significantly in recent years, while totalreal dividends paid have increased. Additionally, divi-dends are concentrating as sharply among European asamong American companies and, while the overall like-lihood of all EU firms paying cash dividends has declinedcontinuously with time, the probability of repurchasingshares has steadily increased.

However, we also present several important newresults showing how payout policies differ betweenEurope and America. Most critically, we show that large-scale share repurchases started much later in Europe thanin the United States, but have grown even more rapidlyover the past decade. We also find that the averagereporting frequency of EU companies has steadily in-creased over 1989–2005, from 1.2 to 2.4 times per year,and is associated with higher amounts of cash dividendspaid and shares repurchased. We examine the impact ofprivatization on the payout policies of divested firms, andfind that privatized companies account for almost one-quarter of the total value of EU cash dividends and sharerepurchases—but represent barely 2% of the number oflisted firms. We also find that the fraction of retainedearnings in a European firm’s total equity is not signifi-cantly correlated with the likelihood to pay cash divi-dends, but that company age (in years) is. This contradictsthe U.S. findings of DeAngelo, DeAngelo, and Stulz (2006)and the international evidence of Denis and Osobov(2008). Additionally, we find significantly negative effectsof a lagged country-specific catering variable for cashdividends in EU companies, which suggests that thecatering theory is not relevant for European dividendpolicies.

Our panel regressions of the likelihood of distributingcash to investors through dividend payments and sharerepurchases, and of the amounts distributed, revealseveral significant influences on EU payout policies.Companies headquartered in a common law country aremore likely to pay cash dividends—but are no more likelyto repurchase shares—than are companies headquarteredin a civil law country. Higher leverage reduces thelikelihood of paying cash dividends. Larger cash holdingsreduce the probability of paying cash dividends, butincrease the probability of repurchasing and increase theamount of cash dividends paid if the company is a cashdividend payer. Companies with more liquid stocksrepurchase more if they do repurchase, while companiesthat have a majority shareholder repurchase less.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374 373

References

Allen, F., Michaely, R., 2003. Payout policy. In: Constantinides, G., Harris,M., Stulz, R. (Eds.), Handbook of Economics. North Holland,Amsterdam, pp. 337–430.

Ang, J., Blackwell, D., Megginson, W., 1991. The effect of taxes on therelative valuation of dividends and capital gains: evidence from dual-class British investment trusts. Journal of Finance 46, 383–399.

Baker, M., Wurgler, J., 2004a. A catering theory of dividends. Journal ofFinance 59, 1125–1165.

Baker, M., Wurgler, J., 2004b. Appearing and disappearing dividends: thelink to catering incentives. Journal of Financial Economics 73,271–288.

Banerjee, S., Gatchev, V., Spindt, P., 2005. Stock market liquidity and firmdividend policy (February 25, 2005), Working paper. Available atSSRN: /http://ssrn.com/abstract=391663S.

Bank, S., Cheffins, B., Goergen, M., 2004. Dividends and politics.Working Paper #24/2004. European Corporate Governance Institute,Brussels.

Bell, L., Jenkinson, T., 2002. New evidence of the impact of dividendtaxation and on the identity of the marginal investor. Journal ofFinance 57, 1321–1346.

Beltratti, A., Bortolotti, B., Milella, V., 2007. State ownership, political risk,and asset prices. Working paper. Fondazione Eni Enrico Mattei,Milan.

Boubakri, N., Cosset, J-C., 1998. The financial and operating performanceof newly-privatized firms: evidence from developing countries.Journal of Finance 53, 1081–1110.

Boudoukh, J., Michaely, R., Richardson, M., Roberts, M., 2007. On theimplications of measuring payout yield: implications for empiricalasset pricing. Journal of Finance 62, 877–915.

Brav, A., Graham, J., Harvey, C., Michaely, R., 2005. Payout policy in the21st century. Journal of Financial Economics 77, 483–527.

Cannavan, D., Finn, F., Gray, S., 2004. The value of dividend imputa-tion tax credits in Australia. Journal of Financial Economics 73,167–197.

Christoffersen, S., Geczy, C., Musto, D., Reed, A., 2005. Crossborderdividend taxation and the preferences of taxable and nontaxableinvestors: evidence from Canada. Journal of Financial Economics 78,121–144.

Dann, L., 1981. Common stock repurchases: An analysis of returns tobondholders and stockholders. Journal of Financial Economics 9,113–138.

DeAngelo, H., DeAngelo, L., 2006. The irrelevance of the MM divi-dend irrelevance theorem. Journal of Financial Economics 79,1415–1431.

DeAngelo, H., DeAngelo, L., 2007. Payout policy pedagogy: what mattersand why. European Financial Management 13, 11–27.

DeAngelo, H., DeAngelo, L., Skinner, D., 2004. Are dividends disappear-ing? Dividend concentration and the consolidation of earnings.Journal of Financial Economics 72, 425–456.

DeAngelo, H., DeAngelo, L., Stulz, R., 2006. Dividend policy and theearned/contributed capital mix: a test of the life-cycle theory. Journalof Financial Economics 81, 227–254.

Denis, D., Osobov, I., 2008. Disappearing dividends, catering incentivesand agency costs: international evidence. Journal of FinancialEconomics, forthcoming.

Dewenter, K., Warther, V., 1998. Dividends, asymmetric information,and agency conflicts: evidence from a comparison of the divi-dend policies of Japanese and U.S. firms. Journal of Finance 53,879–904.

Dewing, A., 1953. The Financial Policy of Corporations. The Ronald PressCompany, New York.

D’Souza, J., Megginson, W., 1999. The financial and operating perfor-mance of newly privatized firms in the 1990s. Journal of Finance 54,1397–1438.

Fama, E., French, K., 2001. Disappearing dividends: changing firmcharacteristics or lower propensity to pay? Journal of FinancialEconomics 60, 3–43.

Fama, E., MacBeth, J., 1973. Risk, return, and equilibrium: empirical tests.Journal of Political Economy 81, 607–636.

Ferris, S., Sen, N., Yui, H., 2006. God save the Queen and her dividends:Corporate payout in the United Kingdom. Journal of Business 79,1149–1173.

Ginglinger, E., Hamon, J., 2007. Actual share repurchases, timing andliquidity. Journal of Banking and Finance 31, 915–938.

Ginglinger, E., L’Her, J-F., 2006. Ownership structure and openmarket stock repurchases in France. European Journal of Finance12, 77–94.

Goergen, M., Renneboog, L., Correia da Silva, L., 2005. When do Germanfirms change their dividends? Journal of Corporate Finance 11,375–399.

Grinstein, Y., Michaely, R., 2005. Institutional holdings and payout policy.Journal of Finance 60, 1389–1426.

Grullon, G., Ikenberry, D., 2000. What do we know about stockrepurchases? Journal of Applied Corporate Finance 13, 31–49.

Grullon, G., Michaely, R., 2004. The information content of sharerepurchase programs. Journal of Finance 59, 651–680.

Guay, W., Harford, J., 2000. The cash-flow permanence and informationcontent of dividend increases versus repurchases. Journal ofFinancial Economics 57, 385–415.

Ikenberry, D., Lakonishok, J., Vermaelen, T., 1995. Market underreactionto open market share repurchases. Journal of Financial Economics 39,181–208.

Jagannathan, M., Stephens, C., Weisbach, M., 2000. Financial flexibilityand the choice between dividends and stock repurchases. Journal ofFinancial Economics 57, 355–384.

Jensen, M., 1986. Agency costs of free cash flow, corporate finance andtakeovers. American Economic Review 76, 323–329.

Johnson, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2000. Tunneling,American Economic Review 90 (2), 22–27.

Jones, S., Megginson, W., Nash, R., Netter, J., 1999. Share issueprivatizations as financial means to political and economic ends.Journal of Financial Economics 53, 217–253.

Joumard, I., 2001. Tax systems in European Union countries. WorkingPaper #301. Organization of Economic Cooperation and Develop-ment, Paris.

Julio, B., Ikenberry, D., 2004. Reappearing dividends. Journal of AppliedCorporate Finance 16, 89–100.

Koch, A., Sun, A., 2004. Dividend changes and the persistence of pastearnings changes. Journal of Finance 59, 2093–2116.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1997. Legaldeterminants of external finance. Journal of Finance 52, 1131–1150.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law andfinance. Journal of Political Economy 106, 1113–1150.

LaPorta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 2000. Agencyproblems and dividend policies around the world. Journal of Finance55, 1–33.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 2002.Investor protection and corporate valuation. Journal of Finance 57,1147–1170.

Lasfer, M., 1997. On the motivation for paying scrip dividends. FinancialManagement 26, 62–80.

Lee, Y-T., Liu, Y-J., Roll, R., Subrahmanyam, A., 2004. Taxes and dividendclientele: evidence from trading and ownership structure. Workingpaper, University of California, Los Angeles.

Li, W., Lie, E., 2006. Dividend changes and catering incentives. Journal ofFinancial Economics 80, 293–308.

Lintner, J., 1956. Distribution of incomes of corporations amongdividends, retained earnings, and taxes. American Economic Review46, 97–113.

Maxwell, W., Stephens, C., 2003. The wealth effects of repurchases onbondholders. Journal of Finance 58, 895–919.

McDonald, R., 2001. Cross-border investing with tax arbitrage: the caseof German dividend tax credits. Review of Financial Studies 14,617–657.

Megginson, W., 2005. The Financial Economics of Privatization. OxfordUniversity Press, New York.

Megginson, W., Nash, R., Van Randenborgh, M., 1994. The financial andoperating performance of newly privatized firms: an internationalempirical analysis. Journal of Finance 49, 403–452.

Miller, M., Modigliani, F., 1961. Dividend policy, growth, and the valuationof shares. Journal of Business 34, 411–433.

Morck, R., Yeung, B., 2005. Dividend taxation and corporate governance.Journal of Economic Perspectives 19, 163–180.

Officer, M., 2006. Dividend policy, dividend initiations, and governance.Working paper, Marshall School of Business. The University ofSouthern California, Los Angeles.

Pinkowitz, L., Stulz, R., Williamson, R., 2006. Does the contribution ofcorporate cash holdings and dividends to firm value depend ongovernance? A cross-country analysis. Journal of Finance 61,2725–2751.

Renneboog, L., Trojanowski, G., 2006. Patterns in payout policy andpayout channel choice of UK firms. Working paper. TilburgUniversity, Tilburg, The Netherlands.

Salas, J., Chahyadi, C., 2006. A decomposition of the dividend payout ratiotrend in the 20th century. Working paper. University of Oklahoma,Norman.

ARTICLE IN PRESS

H. von Eije, W.L. Megginson / Journal of Financial Economics 89 (2008) 347–374374

Skinner, D., 2008. The evolving relation between earnings, divi-dends, and stock repurchases. Journal of Financial Economics 87,582–609.

Vermaelen, T., 1981. Common stock repurchases and market signaling.Journal of Financial Economics 9, 139–183.

Weston, F., Siu, J., 2003. Changing motives for share repurchases. FinancePaper 3, Anderson Graduate School of Management. UCLA, LosAngeles.

Wood, A., 2001. Death of the dividend? CFO Europe Research Report. TheEconomist Group, London.