2.3 S ample S election and D ata F iltering
2.3.3 Selection of Countries
Financial theory would suggest that in an efficient global capital market the dividend payout and capital structure of identical firms in different nations would be the same. Empirically, similar firms would trend towards similar financial structures, unless there are still fundamental differences in the national capital markets in which they operate and barriers exists to the efficient flow of information and capital across countries. Since country differences exist, dividends (both cash dividends and total dividends (cash and share repurchase)) and capital structures may be different among similar firms in different nations. Therefore, a cross country comparison is useful to investigate to what extent the dividend payout and capital structure vary across countries. The reasons for selecting the five sample countries for this thesis are briefly discussed below.
Five countries were selected because they represent one of five different geographical segments namely, Australia from the South Pacific-continent; the U.S. from the American- continent; Japan from the Asia-continent; the U.K. from the Europe-continent; and Malaysia from the East Asian-continent, which are quite different in their economic and business activities.
Firstly, the motivation of choosing Australia was mainly due to the lack of previous studies investigating the determinants of dividend payout and capital structure on DCs and MCs, and also because Australia is regarded as one of the most active and innovative markets for securitised debt in the world (World Investment Report, 2005). The fact that corporate debt issue increased from $A10 billion in 1996 to $A15 billion in 20 004 certainly demands a closer look at Australian firms’ debt and its determinants. Further, the Australian imputation tax system impacts both dividend payout and capital structure decision both from firm and shareholders’ point of view, which makes Australia an attractive candidate to include in the study. Secondly, the reason for selecting the U.S. was purely because the few studies that have investigated the U.S. in an older data set with limited periods do not necessarily give a whole picture of U.S. DCs’ and MCs’ capital structure and dividend payout. However, it is at least a minimum reference to compare the results with the rest of the countries chosen. In addition, U.S. capital structure is more sensitive to default risk than that of Japan (Rajan & Zingales, 1995). This suggests that the expected bankruptcy costs of U.S. firms may be larger than Japan. Further, retentions and bond markets are the major sources of finance for U.K. and U.S. firms. Thirdly, Japan is chosen since the Japanese firm characteristics are unique in the sense that the banks are the dominant source of finance. Further, their legal system and commercial law is modelled on the German system. For example, Japan follows the German system of using reserves to increase the financial strength of the company. Finally, the sample country of Malaysia is chosen because its corporate financial structure is relatively different from the other four countries. For example, Malaysian corporations use their income proceeds to finance their regular investments. Further, the Malaysian government established the Corporate Debt Restructuring Committee (CDRC). The CDRC aims to facilitate voluntary corporate restructuring by coordinating voluntary negotiations and responsibilities between creditors and corporate debtors. The CDRC also intends to minimise losses to creditors, shareholders and other stakeholders, preserve viable business, and implement a comprehensive framework for debt restructuring (World Investment Report, 2005). In addition, the Malaysian government
4 Source: Axiss Australia - Australia’s Debt Securities Market, Executive Briefing, a2a Newsletter, 2003.
plays a more substantial role in stock market formation and development, by pursuing aggressive pro-equity financing policies and placing limitations on debt financing of firms, especially from abroad (Singh, 1995).
Choosing these countries highlights the differences in their capital structure and dividend payout determinants - if there are any. For each country, the main stock exchange website was used to obtain a sample of all listed companies for each year from 1995 to 2004. The stock exchanges selected for each country are:
• Australia: Australian Stock Exchange
• U.S.: New York Stock Exchange
• Japan: Tokyo Stock Exchange
• U.K.: UK Stock Exchange
• Malaysia: Kuala Lumpur Stock Exchange.
From the sample of firms selected from each exchange for each year, both the Osiris database and Compustat-Global database are searched for annual report information. From annual report information, segment information is used to determine if the firms reported business activity from another country. If business activity was reported from more than the domiciled country, it was coded as an MC; if it did not, it was coded as a DC. The number of firms satisfying these requirements for each country is represented as the initial sample in Table 2.2.
MCs could have subsidiaries listed on overseas exchanges. For example, a U.S. multinational also could be listed on the Australian Stock Exchange. Including this firm as an MC in Australia and the U.S. would result in double counting. The technique that is used to identify such MCs in foreign multinationals listed on a non-domiciled exchange is to report financial statements in the domiciled currency and not the currency of the country where they are listed. Therefore, to avoid double counting, firms reporting in a non-domicile currency are excluded as they are foreign MCs. The number of firms excluded on this criterion is shown at (a.) in Table 2.2.
Firms in the financial and regulated industries have dividend payout and capital structures that are determined by levels of deposits and financial regulation. Determinants of dividend payout and capital structure for these firms are considerably different from other firms, and as a result are excluded (Fama & French, 2002; Flannery & Rangan, 2006). Financial organisations excluded under this criterion are shown at (b.) of Table 2.2.
A minimum of three years of data was necessary for some variables. Firms with less than three years of data for estimation of these variables were excluded. This is shown at (c.) in Table 2.2. Also some firms were excluded as the reported figure duration is less than 12 months. Since some of the proxies required a full year of observation to be consistent across other variables, a full year of reporting was important. This is indicated by (d.) in Table 2.2. Further, we define annual observations on the basis of fiscal time (as opposed to calendar time) because sample firms use a variety of fiscal year ends across countries.
Finally, two important statistical conditions are also applied so that the final sample size is statistically valid to use in the multivariate regressions. The conditions include a reasonably large sample selected at random from large populations which is, on average, representative of the characteristics of that population. Secondly, it is statistically advisable that large groups of data show a higher degree of stability than a smaller data set. Since there are a large number of independent variables, a reasonable amount of observations are required to produce reliable and unbiased estimates (eg. degrees of freedom). For example, the sample selection was such that it allows enough observations in each year (minimum 30 observations to meet the central limit theorem) to do a cross-sectional analysis of dividend payouts and capital structure determinants between DCs and MCs across five countries across each year. Also, the comparability of data does suffer from large differences in observations among the countries. This should not bias the results since each country is analysed and compared separately, and for each country the number of observations is sufficiently large for parametric comparisons.
A breakdown and stratification of the sample on a yearly basis is presented in Table 2.3 (Panel A) while Table 2.3 (Panel B) illustrates industry distribution. Table 2.3 shows that, overall, the numbers of DCs and MCs have doubled from the mid-1990s to the early 2000s. The U.K. has considerably higher numbers of MCs than the other countries in each year (1469). It is also clear that the U.K. has less DCs than other countries (719). Japan and Malaysia have less MCs than the other countries. Also Malaysia has twice as much DCs than MCs listed on their stock exchange.
Table 2.3
The structure of the final sample over 10 years across 5 countries
Table 2.3 Panel A provides a description of the sample in detail including the number of DCs (Domestic Corporations) and MCs (Multinational Corporations) that is available for each country across five countries. The country acronyms of AU, US, JP, UK, and ML are Australia, U.S., Japan, U.K. and Malaysia respectively. Panel B presents Industry distribution while Panel C provides the distribution of sampled countries MCs’ geographic location of subsidiaries.
Panel A - Sample Distribution across Years
P a n e l A AU U S J P UK ML T o ta l G r a n d DCs MCs DCs MCs DCs MCs DCs MCs DCs MCs DCs MCs T o ta l 1995 6 2 72 9 9 8 9 52 9 9 58 157 82 52 353 469 822 1996 67 9 7 103 127 5 6 6 9 63 109 73 6 6 362 468 830 1997 73 105 105 144 5 9 7 9 61 164 9 0 73 388 565 953 1998 88 112 120 150 88 75 6 6 121 129 100 491 558 1049 1999 101 125 141 137 105 7 7 6 9 142 2 4 0 108 656 589 1245 2000 109 133 142 146 131 75 7 6 152 2 3 7 106 695 612 1307 2001 133 150 161 119 142 78 72 177 2 5 7 9 7 765 621 1386 2002 121 155 163 162 163 73 79 164 2 5 8 119 784 673 1457 2003 125 159 175 173 157 91 82 147 261 122 800 692 1492 2004 115 146 162 170 140 98 93 136 2 3 4 121 744 671 1415 T o ta l 994 1254 1371 1417 1093 814 719 1469 1861 964 6038 5918 11956 T o ta l DCs & MCs 2248 2788 1907 2188 2825 11956
Table 2.3 Panel B reports that, generally, both DCs and MCs are either manufacturing or retail industries across the five sampled countries. The proportion of DCs that fall in the manufacturing sector ranges from 15% to 29%, while MCs range from 15% to 25%. Similarly, the proportion of DCs that fall in the retail sector range from 19% to 34% while MCs range from 15% to 31% across the countries. However, there is a slight difference in the industrial distribution of Australian firms. It is clearly visible that in our total sample of 2248 Australian firms, the proportion of mining DCs and MCs are 21% and 25% respectively. In the UK however, the majority of the DCs and MCs are in the retail (33% and 31%) and manufacturing (23% and 38%) sectors. U.S. DCs and MCs are generally more stratified across different
industries relative to the other countries. Malaysian sample firms’ industry distribution is
similar to Australian firms (with an exception o f mining industry) where the majority o f the
DCs and MCs fall in the agricultural, forestry and fishing; building construction and heavy
construction; manufacturing and retail sectors. It is clearly visible that in our chosen sample
range, there is an approximate equal distribution o f sample selection across industries between
DCs and MCs.
Panel B - Sample Distribution across Industries
Table 2.3 Panel B provides the US Standard Industrial Codes classification for five countries' industry distribution of Multinational Corporations (MCs) and Domestic Corporations (DCs), including the proportion of total sample. The acronyms for the industries are: There are ten industries in the sample and a dichotomous variable is used to capture each of these industries’ effect (except INDJi'. Finance, Insurance and Real Estates) on capital structure and dividend payout ratios. The industries are: IND_A_AGR1F1SH (agricultural, forestry and fishing); IND B MIN1NG (metal, coal, oil and gas); INDjCjCONSTRUCTN (building constructions and heavy constructions); IND_D_MNFCTRNG (manufacturing, food, Tobacco, Textiles, Furniture and Fixtures and Papers); IND E TRNSPT CMCTN (Transport, Communication, Electric, and utilities); 1ND_F^WHOLESALE (wholesale trade and durable goods); 1ND_G RETAIL (retails), IND_H\ Finance, Insurance and Real Estates) and IND I SERVIC (health, legal, educational, engineering and social).
P a n e l B A U U S J P U K M L U S S I C D C s % M C s % D C s % M C s % D C s % M C s % D C s % M C s % D C s % M C s % I N D _ A 1 0 7 1 1 % 121 1 0 % 6 5 5 % 7 5 5 % 61 6 % 3 8 5 % 18 3 % 5 6 4 % 3 0 4 1 6 % 9 2 1 0 % I N D _ B 2 1 1 2 1 % 3 0 8 2 5 % 1 9 8 1 4 % 1 8 8 1 3 % 8 1% 10 1% 3 6 5 % 4 2 3 % 3 0 2 % 5 1% I N D _ C 1 0 5 1 1 % 1 7 8 1 4 % 1 3 3 1 0 % 1 3 6 1 0 % 1 9 3 1 8 % 3 2 4 % 61 8 % 1 1 3 8 % 2 6 8 1 4 % 1 2 9 1 3 % I N D J ) 1 4 8 1 5 % 2 0 3 1 6 % 1 3 8 1 0 % 2 1 8 1 5 % 3 1 3 2 9 % 2 0 6 2 5 % 1 7 9 2 5 % 3 0 7 2 1 % 2 8 4 1 5 % 1 9 0 2 0 % I N D _ E 3 3 3 % 4 8 4 % 1 6 5 1 2 % 2 0 3 1 4 % 1 4 5 1 3 % 17 1 2 1 % 7 3 1 0 % 1 5 0 1 0 % 2 2 1% 3 4 4 % I N D _ F 1 1 5 1 2 % 1 0 5 8 % 2 0 1 1 5 % 171 1 2 % 7 5 7 % 1 4 5 1 8 % 7 8 1 1 % 2 7 0 1 8 % 2 6 6 1 4 % 1 9 0 2 0 % I N D _ G 1 8 8 1 9 % 191 1 5 % 3 6 4 2 7 % 2 6 6 1 9 % 2 2 9 2 1 % 1 5 2 1 9 % 2 3 7 3 3 % 4 5 8 3 1 % 6 3 6 3 4 % 2 9 1 3 0 % I N D J I 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % 0 0 % I N D J 3 6 4 % 4 2 3 % 41 3 % 8 8 6 % 1 4 1% 3 2 4 % 4 1% 2 7 2 % 9 0 % 2 6 3 % I N D J 51 5 % 5 8 5 % 6 6 5 % 7 2 5 % 5 5 5 % 2 8 3 % 3 3 5 % 4 6 3 % 4 2 2 % 7 1% T o t a l 9 9 4 1 0 0 % 1 2 5 4 1 0 0 % 1 371 1 0 0 % 1 4 1 7 1 0 0 % 1 0 9 3 1 0 0 % 8 1 4 1 0 0 % 7 1 9 1 0 0 % 1 4 6 9 1 0 0 % 1 861 1 0 0 % 9 6 4 1 0 0 %
Further, MCs’ subsidiaries geographical location is also investigated. The approximate distribution of the sampled countries MCs’ subsidiaries are presented in Panel C. It shows that
the geographic location o f subsidiaries o f the sampled countries MCs are not distributed evenly
across continents around the world.
Panel C - MCs’ Distribution across Continents
S a m p l e d C o u n t r i e s M C s ’ D i s t r i b u t i o n a c r o s s C o n t i n e n t s N o r th A m e r ic a | S o u t h A m e r ic a | P a c if ic R e g io n E u r o p e | A s ia A f r ic a A u s t r a l i a 1 5 % 6 % 3 0 % 1 0 % 3 5 % 1 2 % U . S . 1 0 % 6 % 8 % 2 5 % 4 6 % 5 % J a p a n 2 5 % 3 % 2 0 % 4 0 % 1 2 % 0 % U . K . 3 2 % 1 4 % 6 % 3 0 % 1 4 % 4 % M a l a y s i a 2 7 % 6 % 2 2 % 3 2 % 1 4 % 2 % 25
The distributions of Australian MCs subsidiaries locations are different from Malaysia. It shows that the Australian MCs subsidiaries’ locations are heavily concentrated in Pacific region and Asia while most of the Malaysian MCs are located in Europe, North America and Pacific Region. Also, a high proportion of the countries in Pacific region and Asia are less developed nations and countries in Europe, North America are more developed. It is also visible that Malaysian MCs subsidiaries are located in more developed nations while Australian MCs are distributed in between developed and developing nations around the world. Finally, U.S. MCs are predominantly concentrated in Asia and Europe while Japanese MCs more crowded in Europe and Asia.
Given the distribution of the sampled countries MCs’ subsidiaries have no strong uniformity, the variation in the results will be driven by the MCs’ parent country as well as subsidiaries clustering effect in certain continents.