DIVIDEND POLICY DETERMINANTS OF AUSTRALIAN MCS AND DCS

Transcription

1 DIVIDEND POLICY DETERMINANTS OF AUSTRALIAN MCS AND DCS Shumi Akhar* School of Finance and Applied Saisics College of Business and Economics Ausralian Naional Universiy Phone: (61) Fax: (61) Shumi.Akhar@anu.edu.au * The auhor would like o express a very special graiude o Prof. Tom Smih for his excellen supervision. I also wish o hank Dr Barry Oliver and Professor Michael Marin for heir splendid advice. All errors and omissions are my own. 1

2 PREFACE Tile of Thesis: A STUDY OF CAPITAL STRUCTURE AND DIVIDEND POLICY DETERMINANTS IN MULTINATIONAL AND DOMESTIC CORPORATIONS A CROSS-COUNTRY COMPARISON Supervisor: Professor Tom Smih This hesis uses a unique daa se o assess wheher he dynamics of financial srucure can be capured hrough he specificaion and esimaion of capial srucure and dividend payou models for domesic corporaions (DCs) and mulinaional corporaions (MCs) across counries wih differen characerisics. Therefore, his sudy aemps o provide a comparaive and comprehensive undersanding of he capial srucure and dividend policy deerminans a boh he micro (single counry) and macro (muliple counry) levels for 6038 DCs and 5918 MCs across Ausralia, he Unied Saes (U.S.), he Unied Kingdom (U.K.), Japan and Malaysia over he 1995 o 2004 period. These counries are characerised by differen financial, axaion and legal sysems which may have an impac on capial srucure and dividend paymen deerminans. Boh cross-secional and imevariaion in capial srucure and dividend policy are explored. This hesis includes four empirical essays. Essay I is focused on capial srucure deerminans for DCs and MCs in Ausralia. Essay II expands his o a muli-counry comparison of capial srucure deerminans for DCs and MCs. Essay III is cenered on dividend policy deerminans for DCs and MCs in Ausralia while Essay IV expands his analysis o a comparison of dividend policy deerminans for DCs and MCs in muliple counries. Linner s (1956) speed of adjusmen coss for capial srucure and dividend payou raios over ime is also esed for DCs and MCs across counries. Finally, resuls of sensiiviy of differen definiions of dependen variables and inerdependen (endogeneous) relaionship beween long-erm deb and dividend policy is presened. The hesis is srucured as follows:: Inroducion Daa The Dynamics of Ausralian MCs and DCs Capial Srucure Deerminans Capial Srucure of MCs and DCs An Inernaional Comparison Dividend Policy of Ausralian MCs and DCs Dividend Policy of MCs and DCs An Inernaional Comparison Sensiiviy Summary and Conclusion This paper is based on Chaper 5 2

3 ABSTRACT This paper invesigaes he deerminans of dividend policy for Ausralian DCs and MCs. In considering dividend payou raios we adop wo definiions cash dividends and oal dividends. The cash dividend payou raios consiss of cash dividends o ne earnings while oal dividend payou raios consiss of boh cash dividends and share repurchases o ne earnings. The deerminans of foreign exchange, ax clienele and financial slack are significan in explaining he difference of oal dividend payou raios beween Ausralian DCs and MCs. Diversificaion, profiabiliy, firm-specific risk, size, collaeral value of asses and financial slack are he significan variables in explaining he difference in cash dividend payou raios beween Ausralian DCs and MCs. These resuls remain unaffeced irrespecive of conrolling for indusry and ime effecs. 3

4 1. INTRODUCTION Theoreically, i is ofen argued ha inernaional diversificaion of earnings should enable mulinaional corporaions (MCs) o mainain higher dividend payous han domesic corporaions (Hines, 1996). Corporaions use cash dividends and share repurchases as he wo main mehods o disribue cash o shareholders. While each mehod has received considerable aenion in he academic lieraure, fewer sudies have examined he deerminans of cash dividends and share repurchases wihin he mulinaional corporaions framework. In paricular, lile is known abou wha facors deermine he choice beween cash dividends and cash dividends and share repurchases across MCs and DCs and wha, if any, informaion invesors infer from hese decisions. Facors ha have been idenified as deerminans of dividend policy could differ subsanially in boh domesic and mulinaional corporaions. In a world where here is uniform axaion, fixed and sable exchange raes, perfec capial markes, and no barriers o he ransfer of capial, hen he financial decision for MCs should be he same as hose for DCs. In such a siuaion, funds would be raised and disribued by he cheapes sources and flow o wherever hey are needed. Such a world does no exis in pracice. Given MCs operae in more han one counry, financial manager of MCs should be aware of he impac of differen ax srucure, legal srucure, changing exchange raes, barriers o capial inflows and segmened capial markes on dividend paymens (Vinso, 1982). Thus he MCs manager should be concerned abou he availabiliy of profi and facors ha influence he disribuion of profi as a dividend o he shareholders. Secondly, managers express a srong desire o avoid dividends cus, excep in exraordinary circumsances. However, beyond mainaining he level of dividends per share, payou policy is a second-order concern; ha is, increases in dividends are considered only afer invesmen and liquidiy needs are me (Brav e al., 2005). In conras o Linner (1956), hey find ha he arge payou raios is no longer he preeminen decision variable affecing payou decisions because of he recen developmen of share repurchase ypes of dividend paymens across corporaions and heir growing imporance over he las wo decades. Given he dividend payou raios have become a growing concern in he las decades, especially in he U.S., and given Ausralia closely follows U.S. in many aspec of financial markes and no research has been conduced in Ausralia, i demands an invesigaion o es he deermining facors for boh cash and cash and share repurchase in Ausralia. Share repurchases were virually nonexisen when Linner (1956) and Modigliani and Miller s (1961) conduced heir research, so i is no surprising ha hese auhors ignore repurchases. Consisen wih Modigliani and Miller s irrelevance heorem, and in conras o decisions abou preserving he level of he dividend, Brav e al. (2005) find ha managers make repurchases decisions afer invesmen decisions. Many execuives view share repurchases as being more flexible han dividends, and hey use his flexibiliy in an aemp o ime he marke by acceleraing repurchases when hey believe heir 4

5 sock price is low. Companies are likely o repurchase when good invesmens are hard o find, when heir sock s floa is adequae, and when hey wish o offse opion diluion. While some evidence exiss ha share repurchases are used o reduce excess cash holdings (consisen wih Jensen s (1986) free cash flow hypohesis), Brav e al. (2005) do no find evidence ha managers use payou policy o arac a paricular invesor clienele ha could monior heir acions (Allen e al. 2000). Execuives believe ha dividends are aracive o individual invesors bu ha cash dividends and share repurchases are equally aracive o insiuions. In general, mos execuives say ha hey do no use payou policy as a ool in an aemp o aler he proporion of insiuions among oher invesors. We could argue ha if we jus look a he sources and uses of funds ideniy, share repurchases and cash dividends should be a subsiue payou mehod, holding all oher hings consan. However, firms can always adjus heir sources of funds, and herefore i is possible ha dividends and share repurchases are deermined independenly. For example i is possible ha dividends are deermined ogeher wih invesmen, as Miller and Rock (1985) sugges, and ha repurchases are deermined independenly. In summary, curren heories do no provide a unique predicion on wha he relaion should be beween dividends and share repurchases and herefore he deerminans. I is clear ha he quesion of he exen o which dividends and repurchases are subsiue and heir deerminans and mulinaionaliy in Ausralia are he cenral issues, which have imporan implicaions for many of he exising heories. The above discussion provides us an inspiraion o invesigae he deerminans of cash dividends and share repurchase wihin he frame work of DCs and MCs as his issue has never been considered before. The above issues will be addressed by considering he facors ha deermine boh cash dividends (a narrow definiion of dividend) and cash and share repurchase (a broader definiion of dividend) for MCs and DCs which includes firm-specific and inernaional facors. This paper invesigaes he above issues in a sample of Ausralian mulinaional and domesic corporaions by incorporaing he suggesed deerminans in he prior dividend policy lieraure and mulinaional heory. In considering dividend payou raios we adop wo definiions cash dividends and oal dividends. The cash dividend payou raios consiss of cash dividends o ne earnings while oal dividend payou raios consiss of boh cash dividends and share repurchases o ne earnings. The deermining facors found o be imporan for DCs are diversificaion, sock reurn, ax clienele, cash flow variaion, profiabiliy, company-specific risks, firm size and financial slack. The deermining facors o explain MCs dividend payou raios are ax clienele, profiabiliy, collaeral value of asses and leverage. Furher, he common facors ha explain he difference beween DCs and MCs dividend payou raios vary beween cash dividend paymens and cash and share repurchase paymens. 5

6 In explaining cash dividend payou raios, diversificaion, profiabiliy, firm-specific risk, size, collaeral value of asses and financial risks are he explanaory facors, while for ax clienele, cash flow variabiliy and financial slack explain he difference of cash and share repurchase payou raios beween DCs. Indusry and ime facors are considered and are found ha he only indusry ha has any significan relaionship wih dividend payou raios is he ransporaion and communicaion indusry, while ime effecs show ha on average dividend payou for Ausralia decreased over ime. The res of he paper proceeds as follows: secion 2 describes he sample and presens summary saisics. Secion 3 explains he heories of dividend payou. Secion 4 discusses indusry and ime issues affecing dividend payou raio. Secion 5 presens daa and models descripions. Secion 6 discusses he findings of he regression resuls and he las secion 7 concludes he chaper. 2. IMPORTANCE OF MULTINATIONALITY AND DIVIDEND PAYOUTS To idenify he deerminans of capial srucure for DCs and MCs i is necessary o caegorise corporaions as eiher DCs or MCs. This can be achieved by considering wha consiues a mulinaional corporaion. The remainder can hen be regarded as domesic corporaions. Several crieria have been suggesed in he lieraure o define MCs, including foreign sales raio (Geyikdagi, 1981; Errunza & Senbe, 1984; Faemi, 1984, 1988; Kim & Lyn, 1986; Shaked, 1986), foreign ax raio (Lee & Kwok, 1988) and he number of counries in which he firm operaes hrough subsidiaries (Errunza & Senbe, 1984; Kim & Lyn, 1986; Michel & Shaked, 1986; Shaked, 1986). Burgman (1996) and Chkir and Cosse (2001) use he foreign ax raio o classify firms as MCs. The foreign sales raio has been popular bu i does no differeniae beween firms ha earn income hrough expor or firms ha generae foreign source income hrough subsidiary operaions. Thus, in his hesis MCs are corporaions ha have a leas one subsidiary in anoher counry (oher han he domicile counry) in which a firm operaes and earn income (Tallman & Li, 1996). Domesic corporaions, being he remainder. 2.1 Uniqueness of Operaional Aciviies Holding all oher hings consan, MCs operaional risks are deemed o be differen from DCs as hey operae in an inernaional environmen. One can argue ha foreign exchange risk and poliical risk of he MCs subsidiaries may lead he oal business risk o be higher (Burgman, 1996). For example, MCs are exposed o addiional risk including poliical risk and foreign exchange risk, which would no exis in a domesic marke (Bae and Noh, 2001). Monioring, bonding and audiing coss are agencyrelaed and are higher for MCs because of he diversiy of geographical locaions, culural differences, higher audiing coss, differing legal sysems, and language differences. Addiionally, hese naional 6

7 differences increase he complexiy of such sandard asks as generaing muli-counry financial saemens, hiring muli-counry audiors and/or muliple audiors, and compleing consolidaed balance and income saemens (Burgman, 1996; Reeb, Saar & Allee, 1998) and here are complexiies of heir operaions as compared o DCs. According o Wrigh, Madura and Wian (1997), hese coss are due o he disance and he difference in he corporae and naional culure beween he paren and he subsidiaries as well as he difference in he level of economic developmen beween he paren and he subsidiary hos counries. Hence, hese characerisics of MCs may limi he amoun of dividend ha MCs can pay ou o shareholders. On he oher hand, Michel and Shaked (1986) evaluae he differences in financial characerisics and performance beween MCs and DCs. Their resuls show ha while DCs have superior risk-adjused, marke-based performance o MCs, MCs are more capialized and less riskier han DCs; he average sandard deviaion of sock reurns and he average sysemaic risk (bea) of DCs are significanly higher han hose of MCs. Hines (1996) argues ha in he 1980s U.S. corporaions paid dividends a very high raes ou of heir aferax profis, and ha an unusually high fracion of hose profis came from foreign profis (non-u.s.) sources. Furher, Hines (1996) argued ha mulinaionals pay higher dividends han domesic counerpars and heir findings sugges MCs pays hree imes higher han DCs. So, he risk and he benefi of diversificaion may have more of an impac on MCs dividend policy han heir domesic counerpars. 2.2 Differen Tax Regime MCs operae in muliple counries which means is income is earned in differen ypes of ax environmen. For example, impuaion ax sysem encourages companies o pay more dividends o he shareholders since i is less axable han he capial gains. As for MCs, if hey are paying higher ax for heir earnings in overseas counries hen paying dividends ou of profi migh be cosly. Therefore, MCs will pay fewer dividends. However, alernaively, i can also be argued ha if subsidiaries pay less ax from he profi hey make overseas which makes hem beer able o pay dividends, an MC will have a higher dividend policy. 2.3 Easier Capial Marke Access MCs are relaively in a beer posiion o ge access o inernaional capial markes o raise deb han he DCs due o heir inernaional operaion. This means ha in a bad economic year in he home counry, MCs can borrow money overseas a a favourable rae o mainain he dividend policy wih he shareholders, which indicaes MCs should have higher dividends han DCs (Hines Jr., 1996). 7

8 3. THE DETERMINANTS OF DIVIDEND POLICY For nearly four decades, researchers have been grappling wih he dividend puzzle o undersand he deerminans of dividend policy. Mos sudies focus on U.S. firms. We expand he invesigaion by sudying dividend policy for Ausralia where he axaion sysem is so differen from he U.S. and also he advancemen of he capial marke and he economy advancemen are comparaively differen. We measure he relaionship beween dividend payous and 18 differen inernaional and firm-specific variables. We also invesigae indusry and iming differences. 3.1 Mulinaionaliy Michael and Shaked (1986) evaluae he differences in financial characerisics and performance beween MCs and DCs. Hines (1996) argues ha in he 1980s U.S. corporaions paid dividends a very high raes ou of heir afer-ax profis, and ha an unusually high fracion of hose profis came from foreign profis (non-u.s.) sources. Mulinaionaliy of a firm is comprised of muliple aspecs of risks and benefis in operaing muliple counries. There is no universal measuremen o capure all hese inernaional risks and benefis for MCs and heir impac on dividend payou raios. Therefore his chaper will use a dichoomous variable o confine he impac of mulinaionaliy on dividend payou raios. 3.2 Diversificaion The diverse operaions in differen counries ofen assis MCs o minimise operaional cash flow risks and all oher financial risks due o imperfecions ha exis across counries around he world. Therefore, inuiively, MCs should have relaively higher diversificaion benefis han DCs and herefore i should enable MCs o pay higher dividend payous relaive o DCs Hines (1996). Diversificaion in his conex indicaes he breadh of firms operaion in inernaional environmen. While diversificaion enables MCs o pay higher dividend payous, in conras, MCs are exposed o addiional risk including poliical risk and foreign exchange risk, which would no impac on a domesic marke o ha exen (Bae and Noh, 2001). These characerisics of mulinaionals may limi he amoun of dividend ha MCs can pay ou o shareholders. 3.3 Foreign Exchange Risk MCs and DCs ha are exposed o foreign exchange risk which will affec he demand and supply of he firm s producs and prices and coss for he firm (Adler & Dumas, 1984). The more sensiive he MCs are o foreign exchange rae flucuaions, he greaer he chance of price flucuaions, which ulimaely shocks he profi level figures and cash flows from which dividends usually ge disribued. 8

9 The greaer he flucuaions of cash flows, he more he expeced cos of bankrupcy risk increases and consequenly leads o generaing less profi and herefore he lower he dividend paymen o is shareholders. 3.4 Poliical Risk Jodice (1985) sugges ha poliical risk can be defined as changes in he operaing condiions of a firm ha arise ou of a poliical process, eiher hrough war, insurrecion, or poliical violence, or hrough changes in governmen policies ha affec he behaviour of firms and heir financial decisions. Poliical risk can be concepualised as evens in he naional and inernaional environmens ha can affec he profi level, physical asses, personnel, and operaions of firms. Such adverse effecs ofen ake place hrough consrains on he way in which he MCs operae in foreign counries. Kim and Mei (1994) sugges ha poliical risk has a significan impac on firms profiabiliy. Marke volailiy increases during poliical elecion and ransiion periods. Bailey and Chung (1995) also documen ha poliical risk can have a significan effec on firms profi level and profi disribuion o is shareholders. This means ha firms wih significan foreign financing, foreign suppliers or cusomers, or oher inernaional ransacions or asses are relaively exposed o adverse changes in currency conrols, capial flow barriers and oher laws and regulaions ha consiue poliical risk. Depending on firms earnings exposure o poliical risks in he domicile counry and foreign counries, i wil affec firms dividend paymens. 3.5 Agency Coss The principal agen problem is ha managers may pursue heir own goals a he cos of obaining lower profis for he owners (sockholders). Consequenly, agency coss are incurred by shareholders o insure ha he manager acs in he bes ineress of shareholders who are no acively involved in he managemen of he firm (Jensen & Meckling, 1976, 1986; Fama, 1980; Cruchley & Hansen, 1989). There are many ways of reducing agency coss. Dividend paymens serve as one means of monioring or bonding managemen performance. Greaer dividend paymens o shareholders may force he firm boh o raise capial by selling new shares and o go o he capial marke more frequenly. Agency coss are reduced as a resul of he increased scruiny he capial marke places on he firm (Easerbrook, 1984; Jensen, 1986). Rozeff (1982) aemp o finds an empirical relaionship beween agency coss and dividend policy and he findings suggess ha: i) if a firm has a high percenage of insider sock ownership, i will pay a small dividend (a lower dividend payou raio) o reduce agency coss; and ii) if a firm has a greaer 9

10 number of shareholders, i will pay high dividends (a higher payou raio) o reduce agency coss. Furher, i is also argued ha an opimal dividend policy may exis even afer ax consideraions are ignored. According o his argumen, increased equiable dividend paymens reduce he volume of funds over which managemen has discreionary conrol; hus he coss of agency are reduced, bu he ransacion coss of exernal financing increase. Therefore, here is an opimal dividend payou which minimises he sum of hese opposing coss. Lloyd, Jahera and Page (1986), Dempsey and Laber (1992), Schooley and Barney (1994), Hansen e al. (1994), Rao and Whie (1994), Moh d e al. (1995), and Holder e al. (1998) suppor Rozeff s (1982) original findings. As Jensen and Meckling (1976) sugges, managers can allocae resources o aciviies ha benefi hem privaely, bu ha are no in shareholders bes ineres. Easerbrook (1984) views dividend paymens as a poenial soluion o agency conflics. Dividend paymens force managers o raise funds in he exernal financial markes and hus subjec managers o scruiny by ouside professionals such as invesmen bankers, lawyers, and public accounans. Recognising he monioring value of exernal financial markes, shareholders will insis ha managers pay dividends. Also, Jensen (1986) poins ou ha managers have incenive o grow heir firms beyond opimal size as growh increases managers power by increasing he resources under heir conrol. Dividend payous can be used o reduce discreionary cash under managers conrol ha could be wased in negaive NPV projecs. 3.6 Free Cash Flows Pruden managers working in he shareholders bes ineress should inves in all profiable opporuniies. Managemen and owner separaion affords corporae managers he empaion, however, o consume or oherwise wase surplus funds. The inefficien use of funds in excess of profiable invesmen opporuniies by managemen was firs recognised by Berle and Means (1932). Jensen s (1986) free cash flow hypohesis updaed his asserion, combining marke informaion asymmeries wih agency heory. The funds remaining afer financing all posiive ne presen value projecs cause conflics of ineres beween managers and shareholders. Dividend and deb ineres paymens decrease he free cash flow available o managers o inves in marginal ne presen value projecs and managers perquisie consumpion (Myers, 1987, 1990). 3.7 Pas and Fuure Growh In general, he growh of a firm depends on: (i) he amoun of resources reained and reinvesed in he firm; and (ii) he rae of reurn ha is earned on he reained resources. If a firm were experiencing subsanial success and rapid growh, he firm would require large addiions of capial. Therefore, 10

11 growh firms may expec o pursue a low dividend payou policy since invesmens and dividends are linked hrough he firm s cash-flow ideniy. 1 Many sudies have suggesed a negaive relaionship beween he growh (invesmen) opporuniies of a firm and is dividends. For example, Rozeff (1982) and Schooley and Barney (1994) find ha dividend payou raio is negaively relaed o boh pas growh revenues and prediced growh revenues of he firm. I is found ha higher growh raes in pas and fuure revenues require more funds o susain growh; hus firms wih more growh opporuniies pay lower dividends in order o finance furher growh. Using indusry-level daa, Smih and Was (1992) also discover ha various measures of growh opporuniies are negaively relaed o corporae dividend policy. Meanwhile, Gaver and Gaver (1993), using firm-level daa, find an inverse relaionship beween invesmen (growh) opporuniy and dividend policy. Using firm-level daa in he U.K., Adedeji (1998) observes a negaive ineracion beween dividend payou and invesmen. La Pora e al. (2000), using firm-level daa from 33 counries, also find ha high-growh firms make lower dividend payous han low-growh firms. Using he over-invesmen argumen, Jensen (1986) argues ha if a firm has more growh opions han oher firms, i will have lower free cash flow and will pay lower dividends. 2 Consisen wih he overinvesmen argumen, Lang and Lizenberger (1989), Howe e al. (1992), and Denis e al. (1994), using Tobin s Q as an indicaor of a firm s growh (invesmen) opporuniies, find a negaive relaionship beween Tobin s Q and dividend paymen. Thus, heir resul is consisen wih he findings of previous sudies, which repor a negaive relaionship beween growh opporuniies and dividends. Firms wih many good invesmen opporuniies have high cash needs, which may lead hem o payou a low fracion of earnings o shareholders as dividends. If his is he case, a negaive relaionship is expeced beween invesmen opporuniies and payous. Alernaively, some ohers argue ha he relaionship beween invesmen opporuniies and dividend is in fac posiive (Brav e al., 2005). Survey evidence suggess ha firms are highly relucan o cu dividends, and increase dividends only when susainable higher earnings are expeced (Linner, 1956; Brav e al., 2005). This argumen suggess ha only firms wih a variey of good invesmen projecs pay high dividends oday because he cash flows earned from fuure projecs suppor high dividends in he fuure. Given hese wo conflicing hypoheses on he relaion beween invesmen opporuniies and dividend paymens, i is lef o empirical ess o deermine which is indeed he more accurae. 1 In general, he greaer he amoun of invesmen during he period, he smaller he dividend or he greaer he new equiy issued. 2 According o Jensen s (1986) argumen (or over-invesmen hypohesis), a dividend increase by a firm wih free cash flow problems will reduce he exen of over-invesmen, and will increase he marke value of he firm. Similarly, a dividend decrease by such a firm will have he opposie resul. 11

12 3.8 Sock Reurn In he prior lieraure i has been documened ha higher sock reurns are associaed wih higher dividends, independenly of wheher income is axed more or less heavily han capial gains (Lizenberger & Ramaswamy, 1979, 1982; Bajaj & Vijhs, 1990; Morgan & Thomas, 1998). Recenly, his evidence is furher documened by McManus, Gwilym & Thomas (2004). Lamon (1998) finds ha sock reurn has a significan posiive relaionship wih dividend payou raios. We follow he Black and Scholes (1974) approach where hey saed he effec in erms of he price of he company s shares, or in erms of he expeced reurn on he company s shares, where reurn is defined as boh capial gains and dividends. For example, if we believe ha increasing a company s dividend will increase he price of is shares, hen we can say his in eiher of wo ways eiher increasing he dividend will increase he price of a company s shares or increasing he dividend will reduce he expeced reurn on a company s shares. From MCs poin of view i remains an empirical quesion as o how sock reurn explains he dividend payou raios for MCs. 3.9 Average Tax Rae Ausralia operaes under an impuaion ax sysem, where dividends may have franking credis aached o hem ha allow shareholders o claim he ax paid a he company level as a credi agains heir personal income ax liabiliy. The sysem effecively removes he double axaion of dividend income ha occurs under he classical ax sysem. Companies engaging in off-marke (someimes referred o as equal access) repurchase programs in Ausralia have generally sough a ruling from he Tax Commissioner, he resul of which has been ha a proporion of he buyback price can be designaed as a fully-franked dividend, and he remainder is defined as a capial amoun. This srucure may have ax advanages for shareholders such ha hey are willing o sell shares back o he company a a price ha is below he marke price; shareholders gain from he impuaion credis aaching o he dividend componen of he buyback and may also gain from capial gains ax credis. I is an empirical quesion o see he impac of impuaion ax on he dividend policy for Ausralian MCs and DCs Tax Clienele The ax clienele argumen posulaes ha invesors in low ax brackes prefer high dividend- paying socks when compared o invesors in high ax brackes. (Brennan, 1970; ; Elon & Gruber, 1970; Long, 1978; Lizenberger & Ramaswamy, 1979; DeAngelo & Masulis, 1980; among ohers). Early invesigaions of he ax clienele effec were indirec ess of he ax clienele argumen and have been criicised by Miller and Scholes (1982) because of heir exreme sensiiviy o he definiion of dividends. 12

13 Mos recen sudies have documened ha clieneles may no depend on axes alone. While Chaplinsky and Seyhun (1990) find ha ax-deferred and ax-exemp dividend recipiens accouned for half of all dividends in 1979, significan dividends were sill subjec o axes. Serk and Vandenberg (1990) find a preference for cash dividends despie he eliminaion of differen ax raes beween capial gains and dividend income in DeAngelo (1991) argues ha an equilibrium consisen wih dividend payou may exis even in he presence of ax sysems ha favour capial gains. Brennan and Thakor (1990) also presen an equilibrium model where dividends exis for small disribuions despie he preferenial ax reamen of capial gains. In his paper, he clienele effec is esed differenly by incorporaing he ax saus of he firm s major sockholders. Insiuional invesors are eiher ax-exemp, can defer axes on dividend received, or pay axes on he dividends received from anoher corporaion. Consequenly, if he ax clienele argumen is valid, a posiive relaionship is expeced beween insiuion holdings, and dividend payou raio. There is no heory ha has been developed in regards o how MCs should respond o he ax clienele effec bu i may be argued ha since MCs operae in more han one ax regime, depending on he exploiaion of ax advanage a posiive sign will indicae ha MCs invesors are beer off han DCs invesors in maximising he ax advanage, and herefore would prefer high dividends Volailiy of Cash Flows Firms facing high levels of cash-flow uncerainy are likely o pay low dividends, fearing cash shorfalls in he fuure. In order o fund profiable fuure invesmen projecs, firms wih high cashflow uncerainy will choose o hoard cash oday by keeping dividends paymens low. Moreover, firms wih unsable cash flows may have o resor o exernal financing more ofen. In general, exernal funds are more expensive han inernal funds, bu his is ye ruer for firms wih high cashflow uncerainy; hese firms generally have low credi raings, a resul of a volaile cash-flow sreams, and hus mus pay premium ineres o raise money. Firms wih high cash-flow uncerainy will hus pay lower dividends because of heir greaer need o rely on inernal funds (Pei, 1972; Asquih & Mullins, 1983; Miller & Rock, 1985). Based on his discussion, i is expeced ha dividend payous will be negaively relaed o cash-flow volailiy. This argumen echoes survey evidence by Brav e al. (2005). They repor ha more han wo-hirds of chief financial officers of dividend-paying firms say ha he sabiliy of fuure cash flows is an imporan facor affecing dividend decisions Profiabiliy Fama and French (2001) documen ha he probabiliy ha a firm pays dividends is posiively relaed o profiabiliy and size and negaively relaed o growh. The inuiion is ha higher profiabiliy and greaer size imply a greaer capaciy o disribue cash, whereas greaer growh indicaes superior 13

14 invesmen opporuniies, hus a sronger incenive o reain cash. More profiable firms are expeced o hold less deb, since i is easier and more cos effecive o finance inernally and consequenly pay higher dividends. MCs have beer opporuniies han DCs o earn more profi mainly due o having access o more han one source of earnings and beer chances o have favorable business condiions in paricular counries (Kogu, 1985; Barle & Ghoshal, 1989). Consequenly, MCs being more profiable han DCs are expeced o have higher dividend payou han DCs afer conrolling for he oher variables Firm-specific Risk I is ofen argued ha a firm wih more sysemaic risk (bea) 3 ends o adop a policy of seing a relaively low payou raio. Consisen wih his argumen, Rozeff (1982), Lloyd (1985), Bajaj and Vijh (1990), Schooley and Barney (1994), Dyl and Weigand (1998) and ohers discover an inverse relaionship beween dividend and securiy bea. For example, Rozeff (1982) argues ha, since higher bea is a reflecion of he presence of higher operaing and financial leverage, a firm will end o pay lower dividend when i has a higher bea coefficien. Bajaj and Vijh (1990) argue ha dividend increases are associaed wih increases in sock reurn because he percenage change in a sock price is ypically much smaller han he percenage change in is dividend. Thus, he auhors expec and find a decrease (increase) in securiy bea afer a dividend increase (decrease). Dyl and Weigand (1998) also find ha boh he oal risk and he sysemaic risk of a sample of dividend-iniiaing firms are significanly lower he year following he announcemen of dividend paymens. The decrease in oal risk is more pronounced especially for he large firms, and he decrease in bea is more pronounced for he smaller firms. Similar o he sysemaic risk, Beaver e al. (1970), Michael and Shaked (1986), Bar-Yosef and Huffman (1988), Glen e al. (1995), and ohers argue ha he uncerainy of a firm s earnings may lead i o pay lower dividends because he exisence of large flucuaions in earnings maerially increase he risk of defaul. Furher, if firm s follow a policy of dividend sabilizaion, firms wih greaer volailiy in earnings will se a low payou raio, which can be mainained even in he face of a relaively serious or prolonged decline in earnings. 3 A firm s securiy bea represens is level of sysemaic risk. 14

15 3.14 Size Research by Lloyd, Jahera and Page (1985) and Vog (1994) indicaes ha firm size plays a role in explaining he dividend payou raio of firms. They find ha larger firms end o be more maure and hus have easier access o he capial markes, which reduces heir dependence on inernally-generaed funding and allows for higher dividend payou raios. I is argued ha firms ha are large have greaer access o capial markes and hey can easily able o swich beween deb and equiy and ake advanage of lower ransacion coss, which allows for more sable and possibly higher dividend paymens of he firm. Hence a posiive relaionship is expeced beween he size and dividend payou raio (Ali, Khan & Ramirez, 1993) Collaeral Value of Asses Shareholders may expropriae wealh from bondholders by paying hemselves dividends. Bondholders ry o conain his problem hrough resricions on dividend paymens in he bond agreemen. However, fewer resricions are placed on he firm if deb can be collaeralised as he borrower is resriced o use he funds on specific projecs. Hence, a posiive relaionship is expeced beween dividend and dividend payou raios Firm Mauriy Dividends end o be paid by maure, esablished firms, plausibly reflecing a financial lifecycle in which young firms face relaively abundan invesmen opporuniies wih limied resources so ha reenion dominaes disribuion, whereas maure firms are beer candidaes o pay dividends because hey have higher profiabiliy and fewer aracive invesmen opporuniies. Fama and French (2001), Grullon e. al (2002), and DeAngelo and DeAngelo (2006) all advance lifecycle explanaions for dividends ha rely, implicily, on he rade-off beween he advanages (e.g. floaion cos savings) and he coss of reenion (e.g. agency coss of free cash flow). The rade-off beween reenion and disribuion evolves over ime as profis accumulae and invesmen opporuniies decline, so ha paying dividends becomes increasingly desirable as firms maure. The lieraure offers only a rough empirical idea of he characerisics ha differeniae firms ha pay dividends from hose ha do no. Mos noably, Fama and French (2001) find ha firms wih curren high profiabiliy and low growh raes end o pay dividends, while low profi/high growh firms end o reain any profis Long-erm Deb There are wo alernaive hypoheses ha predic no abnormal dividend paymens afer deb issue. Firs, he deb covenan hypohesis predics ha firms will decrease or no increase dividends afer 15

16 deb issues. Kalay (1982) shows ha deb conracs resric dividend paymens direcly and indirecly. I is found ha he sockholders of leveraged firms choose o pay dividends under deb conracs. Consisen wih he deb covenan hypohesis, DeAngelo and DeAngelo (1990), Jensen e al. (1992), and Long e al. (1994) find an inverse relaionship beween dividends and debs. However, Denis (1990), Gupa and Rosenhal (1991), Smih and Was (1992), Mougoue and Mukherjee (1994), and Adedeji (1998) find posiive relaionship beween he leverages of a firm and is dividends. The second hypohesis, he repuaion hypohesis, also predics no abnormal dividend paymens afer deb issues. On he basis of asymmeric informaion beween he insiders of he firm and he ousider deb-holders, John and Nachman (1985) develop a model which shows he imporance of repuaion in conrolling coss. They sugges ha firms are willing o mainain heir repuaion for he following reasons: (i) alhough a firm may aemp o ransfer wealh from shareholders o bondholders, his can occur only in one period, no in a muli-period world; (ii) he higher-raed deb (bonds of repuable firms) carry less resricive consrains in he bond covenans on maximum payous or minimum invesmen han hose of lower-raed deb. Handjinicolaou and Kalay (1984) examine he repuaion hypohesis, and find no evidence of wealh redisribuion from bondholders o sockholders, bu do find evidence in favour of he repuaion argumen. Long e al. (1994) es boh he deb covenan and repuaion hypoheses. Using a sample of he firms dividend payou raios, sraigh deb, and converible deb from 1964 o 1977, heir findings do no suppor he deb covenan argumen, bu do suppor he repuaion argumen. They conclude ha repuaion is he mos plausible explanaion why firms do no ransfer wealh from bondholders o sockholders hough dividend policy. Firms rade-off dividend paymens wih fixed financial charges. A highly leveraged firm would end o lower is dividend payou raio because of high fixed financial commimens. On he oher hand, increased indebedness leads o increased conacs wih exernal financing sources, which resuls in closer monioring and an increased dividend payou Financial Slack This chaper also considers examining he relaionship beween financial slack and dividend payou. In order o underake profiable invesmens, firms may prefer o increase heir financial slack raher han pay higher dividends. A high financial slack causes payou raio o become lower, exacerbaing he cash flow problems. The pecking order heory seeks o explain why managemen prefers o build financial. Asymmery of informaion beween invesors and managemen makes exernal financing more difficul o obain. Thus managemen prepares for fuure financial risks and business 16

17 opporuniies by accumulaing inernal funds whose use is no resriced. Therefore, financial slack is expeced o be inversely relaed o dividend payou raio. 4. THE DIVIDEND POLICY AND OTHER ISSUES 4.1 Indusry Indusry-specific facors affec a firm s dividend policy (Baker & Powell, 2000). Alhough variaion in dividend payous among firms appear o be affeced by firm-specific variables such as invesmen requiremens and earnings variabiliy, Linner (1953) hypohesises ha dividend policy also is influenced by an indusry effec. This effec could be inerpreed as common correlaions wih deerminans of dividend payou by firms in he same indusry, bu Linner suggess an effec of dividend leadership analogous o price leadership or wage leadership. Such an indusry effec, if i exiss, presumably sands apar from oher firm-specific variables ha affec payou decisions of he member firms wihin an indusry and causes hem o have varying dividend policies (Dempsey, Laber and Rozeff, 1993). Some evidence suggess ha here is significan variaion in dividend payou raios among indusries (Michael, 1979; Baker, 1988). Richardson, Tuna and Wysocki (2002) confirm ha indusry affiliaion is a srong deerminan of corporae cash holdings, acquisiions, R&D and capial expendiures, leverage policy and dividend and share repurchase policy. Several sudies specifically examine he poenial effec of indusry affiliaion on dividend policies. Research by Michel (1979) and Baker (1988) among ohers suggess a posiive relaionship exiss beween indusry classificaion and dividend policy. 4 Rozeff (1982) concludes, however, ha a company s indusry does no help o explain is dividend payou raio. Alhough firms wihin he same indusry end o have similar dividend payous, Rozeff aribues he apparenly significan indusry effec found in oher sudies o he fac ha oher variables are ofen similar wihin a given indusry. His conclusion does no apply o uiliies since he inenionally excluded regulaed companies because heir regulaors saus may affec heir dividend policies. Survey research by Baker, Farrelly and Edelman (1985) examine differences in deerminans of dividend policy among hree indusry groups manufacuring, wholesale/reail, and uiliies. Their resuls show ha he responses of he hree groups differ significanly among eigh of he fifeen deerminans of dividend policy examined in heir sudy. Soer, Brigham and Evanson (1996) noe he economic environmen for uiliies has been changing over ime. Due o regulaors acions, he uiliy indusry has become a riskier place in which o operae and inves. In he curren world, uiliies find hemselves increasingly subjec o compeiion. For he purpose of our analysis, I exclude financial 4 Rozeff (1982) concluded ha a company s indusry does no help o explain is dividend paymen raio. This conclusion may no apply o uiliies because he inenionally excluded regulaed companies from his sample. 17

18 companies as hese firms may have differen consideraions in esablishing heir invesmen and dividend policies. A zero one dummy variable will be used in he proposed models o capure he effec of indusry in boh domesic and mulinaional corporaions dividend payou policy. Furher, indusry dummies are included in he regression o accoun for differences in asse srucure, accouning pracices, governmen regulaion, and compeiiveness each may affec corporae governance of dividend paymens and firm valuaions (Durnev and Kim, 2005). Given he prior sudies have been conduced in mos U.S. and oher European counries, i would be worhwhile o see he impac of indusry in he Ausralian corporaions conex. 4.2 Timing I is imporan o examine he effecs of ime-varying informaion and deerminans on he cash dividend payou and cash dividend and share repurchases. Firsly, ime series evidence of dividend policies is relaively more plausible han cross-secional evidence as ime series analysis capures he dynamic changes of dividend payou policy across ime. Secondly, i would provide evidence on he robusness of he disappearing puzzle. Thirdly, i allows conducing a furher es of he ime dependence variable changes. For example, when invesmen and dividend paymen is primarily financed wih inernal funds, worsening condiions should no have as larger an impac as when exernal funds accoun for he bulk of financing. Since his only happens in financial markes which are imperfec (i.e. if inernal and exernal funds are no perfec subsiues), he differenial impac should be sronger when financing fricions are more prevalen (Braun and Larrain, 2005). In order o capure year effec (e.g. business cycle shocks or economic downurn), a dichoomous one and zero variable will be used for each year. 5. DATA AND METHODOLOGY We gaher daa from he Osiris daabase and Compusa-Global for he period of 1995 o The sample consiss of 1254 Ausralian MCs and 994 Ausralian DCs. The following hree main regression models are proposed o es he deerminans and mulinaionaliy effec on differen measure of dividend payou raios. For example, Model I aemps o explain he deerminans of DCs and MCs independenly so ha a direc comparison can be made in erms of he explanaory power of each deermining facors beween hose wo ypes of firms. Model II aemps o isolae he mulinaionaliy effec which is believe o capure any unique facors ha exis bu we could no capure wihin he capaciy of his research even afer considering he facors ha is believe o explain o he bes (e.g. diversificaion, foreign exchange risk, poliical risks). Thirdly, Model III 18

19 endeavours o disinguish he difference across he explanaory variables beween DCs and MCs. The es of indusry influence and ime influence on dividend paymen behaviour is also invesigaed by exending model II. The main hree models are presened below: 5.1 Model I DIVC = β DIVER FX PR AGC FCF GROW _ MB Where: DIVC* DIVER FX PR AGC FCF GROW_MB GROW_PT SR ATR TAX_CLTL CFV PROF BETA NDTS SIZE CVA AGE LTD SLACK * i, 0 1 i, 2 i, 3 i, 4 i, 5 i, 6 i, 5.2 Model II GROW _ PT SR ATR TAX _ CLTL CFV PROF BETA SIZE CVA AGEi, 17 LTDi, 18 SLACKi, + ε i, 7 i, 8 i, 9 i, 10 i, 11 i, 12 i, 13 i, 14 i, 15 i, 16 = Cash dividend (* represens ha oal dividend (comprised of cash dividend and share repurchase) is also used as he dependen variable) = Diversificaion = Foreign exchange risk = Poliical risk = Agency Coss = Free cash flow = Growh for marke o book value = Growh for pas = Sock reurn = Average ax rae = Tax Clienele = Cash flow variaion = Profiabiliy = Firm specific risk (De-geared bea) = Non-deb ax shield = Toal asses = Collaeral value of asses = Firm s age = Long-erm deb = Financial Slack DIVC = β MULT DIVER FX PR AGC FCF GROW _ MB * i, 0 1 i, 2 i, 3 i, 4 i, 5 i, 6 i, 7 i, GROW _ PT SR ATR TAX _ CLTL CFV PROF 8 i, 9 i, 10 i, 11 i, 12 i, 13 i, BETA SIZE CVAi, 17 AGEi, 18 LTDi, 19 SLACK i, + ε i, 14 i, 15 i, Model III DIVC = β MULT DIVER FX PR AGC FCF * i, 0 1 i, 2 i, 3 i, 4 i, 5 i, 6 i, GROW _ MB GROW _ PT SR ATR TAX _ CLTL 7 i, 8 i, 9 i, 10 i, 11 i, CFV PROF BETA SIZE 16 CVA, 17 AGE, 18 LTD, 12 i, 13 i, 14 i, 15 i, i i i SLACK M _ DIVER M _ FX M _ PR M _ AGC 19 i, 20 i, 21 i, 22 i, 23 i, M _ FCF M _ GROW _ MB M _ GROW _ PT M _ SR 24 i, 25 i, 26 i, 27 i, 28M _ ATRi, 29 M _ TAX _ CLTLi, 30 M _ CFVi, 31 M _ PROFi, M _ BETA M _ SIZE M _ CVA M _ AGE M _ LTD 32 i, 33 i, 34 i, 35 i, 36 i, M _ SLACK + ε 37 i, i, Where M_i, represens he MULT Dummy for 18 independen variables i. 19

20 6. REGRESSION RESULTS 6.1 Ausralian DCs and MCs Dividend Payou Deerminans The analysis is based on levels which allows for a more direc inerpreaion of he resuls and avoids some measuremen issues ha exis in he change analysis (Nissim and Ziv, 2001). The regression analysis presened in Table 1 indicaes four ses of regression resuls across DCs and MCs for boh cash dividend payou raios (DIVC i, ) and oal dividend payou raios (DIVR i, ). Resuls show ha he firs deermining facor of diversificaion (DIVER i, ) of firms has a negaive and highly significan impac on DCs cash dividend payou decisions (=-4.40). Similarly, he DIVER i, facor for DCs also show a significan and negaive (=-2.76) relaionship in explaining cash and share repurchase payou raios (broader definiion of dividend). This suggess ha when Ausralian DCs expand in erms of number of acive subsidiaries in differen geographical locaions wihin he local counry, i decreases boh cash and oal dividend paymens o shareholders. However, no significan relaionship is found wih diversificaion of MCs and ype of dividend paymen modes (e.g. cash dividend or oal dividend). This suggess ha MCs expansion boh naionally and inernaionally has no significan impac on dividend payou decision. A weak and significan posiive relaionship is found for foreign exchange risk (FX i, ) for DCs in explaining oal dividend payou raios (=1.69). These resuls sugges ha when he exchange rae flucuaes, he Ausralian DCs are in a beer posiion relaive o MCs, as i appears ha he variaion in foreign exchange rae favours he DCs increase in he dividend paymen mehod hrough share repurchase raher han jus cash dividends. And his may be due o DCs mainaining he value of heir share wihin he firm and no leing heir share value unnecessarily inflae in he marke. No apparen significan relaionship is observed in MCs FX i, variaion and is impac on cash dividend paymens or share repurchases aciviies. Prior lieraure suggess ha dividends can be used in reducing agency problems beween managers and sockholders. For example, he paymen of dividends reduces he discreionary funds available o managers for perquisie consumpion and helps address he manager-sockholder conflic (Jensen & Meckling, 1976; Easerbrook, 1984; Cruchley & Hansen, 1989). While he direc measure of agency coss is used here (raio of number of shareholders o oal ousanding shares) o capure he agency problem beween sockholders and managers, no percepible evidence is found in eiher of Ausralian DCs or MCs. Similarly, he free cash flow (FCF i, ) of agency coss shows ha here is an inefficien use of funds in Ausralian firms which decreases significanly heir cash dividend paymens capaciy across DCs and MCs; however, his relaionship is no significan. This resul sugges ha he funds remaining afer financing all posiive ne presen value projecs cause conflics of ineres beween managers and shareholders almos idenically beween DCs and MCs. This implies ha he managers 20

21 of Ausralian DCs and MCs use he excess cash flow o maximize heir own wealh by spending on unnecessary managerial expenses a he cos of shareholders by no paying cash dividends o heir shareholders on ime. This resul is inconsisen wih Jensen (1986) who finds ha free cash flow posiively and significanly affecs dividend payou raios higher levels of free cash flow will have higher agency coss and need higher dividends payou raios o reduce hose agency coss. Table 1 The deerminans of cash dividends and oal dividends for Ausralian DCs and MCs This able repors he resuls of OLS regression (Model I) o obain he parameers and -saisics accordingly for he sample of 2248 (994 DCs and 1254 MCs separaely) lised companies in he Ausralian Sock Exchange over 10 years o The adjused R squared indicaes he model s goodness of fi. DIVC = β DIVER FX PR AGC FCF GROW _ MB * i, 0 1 i, 2 i, 3 i, 4 i, 5 i, 6 i, GROW _ PT SR ATR TAX _ CLTL CFV PROF 7 i, 8 i, 9 i, 10 i, 11 i, 12 i, BETA SIZE CVA AGEi, 17 LTDi, 18 SLACKi, + ε i, 13 i, 14 i, 15 i, 16 * indicaes ha his model uses wo differen measuremen of dividend payou raios which is explained below. Two dependen variables have been employed and hey are: DIVC i, (cash payou raios) is calculaed using cash dividend paid o ne earnings and DIVR i, (cash and share repurchase dividend payou raios) is calculaed as cash plus change in reasury sock scaled by ne earnings. DIVER i, (diversificaion) is he oal number of subsidiaries ha any firm has wihin heir domicile counry and overseas. FX i, (foreign exchange risk) is calculaed by he raio of overseas sales o oal sales revenue. PR i, (poliical risk) is he sum of all he MCs subsidiaries counries poliical risk raings exposed o he proporion of each sale ha a subsidiary makes overseas. AGC i, (agency cos) is he naural logarihm of oal shareholders. FCF i, (free cash flow) measured afer Lehn and Poulsen (1989) is he sum of earnings before ineres and ax plus depreciaion plus amorizaion plus ax paid minus dividend paid and minus ineres expense and hen i was scaled by so ha he magniude of his variable is compaible wih oher measuremens. Two growh measures have been used in his sudy namely GROW_MBB i, (marke value of equiy o book value of asse) which is measured as marke value of asse o book value of asse. And secondly, GROW_PT i, (pas growh) which is measured as he change in oal asses over oal asses. SR i, (sock reurn) is measured as expeced reurn of individual corporaions by using CAPM. ATR i, (average ax raios) is calculaed as oal ax paid per annum scaled by ne profi. TAX_CLTL i, (ax clienele) is he raio of number of common shares ousanding o number of oal shares ousanding. CFV i, (cash flow variaion) is he sandard deviaion of he firs difference in earnings before ineres and axes, scaled by mean value of ineres expense. SIZE i, is he naural logarihm of oal asse. CVA i, (collaeral value of asses) is he raio of fixed asses o oal asses. AGE i, is he naural logarihm of he age of he firm in years from dae of incorporaion. LEVERAGE i, is he raio of long-erm deb o long-erm deb plus marke value of equiy. SLACK i, (financial slack) is measured as he sum of cash balances and markeable securiies scaled by he marke value of equiy. DCs MCs DIVC i, DIVR i, DIVC i, DIVR i, Coeff -Sa Coeff -Sa Coeff -Sa Coeff -Sa C DIVER i, a a FX i, c PR i, AGC i, FCF i, GROW_MB i, GROW_PT i, SR i, a a ATR i, TAX_CLTL i, a b a a CFV i, a a PROF i, a a b c BETA i, a a SIZE i, a a c b CVA i, b AGE i, LTD i, b b SLACK i, a a b Adj R-sqr No. of Obs a, b and c are he saisical significance level for 1%, 5% and 10% respecively. 21