Determinants of Capital Structure for Japanese 2009

Determinants of Capital Structurefor Japanese Multinational and

Domestic Corporationsn

SHUMI AKHTAR AND BARRY OLIVER

School of Finance and Applied Statistics, Faculty of Economics and Commerce,

Australian National University, Canberra, Australia

ABSTRACT

Our study examines whether there are systematic differences in standardleverage determinants for a sample of Japanese multinational (MNCs) anddomestic corporations (DCs). We find that on a univariate basis JapaneseMNCs differ significantly on most variables relative to Japanese DCs. Thesevariables include leverage, age, collateral value of assets, free cash flows,foreign exchange risks, growth, non-debt tax shields, political risks, profit-ability and size. Business risks are not found to be significantly differentbetween the two groups of organizations. When modeling capital structureand the determinants of capital structure we find that Japanese multi-nationals have significantly less leverage than Japanese DCs, and thatmultinationality is an important aspect of leverage for Japanese firms. Wefind that business risks are not significant for modeling capital structure ofdomestic firms but they are for multinationals and foreign exchange risks arenot significant for multinationals but are significant for domestic firms.Business risks are negatively related to leverage for multinationals and wedocument that significant positive leverage effects of foreign exchange risksand size are subsumed by the negative effect of business risks to explain thelower leverage experienced by Japanese multinationals relative to JapaneseDCs. The lack of significance of foreign exchange risks for DCs can beexplained by economies of scale in risk management, such as derivatives.Domestic firms seem to manage increased foreign exchange risks throughlower leverage rather than derivative use. On the other hand, the largermultinationals can take advantage of economies of scale in risk manage-ment. Consequently, foreign exchange risks of multinationals can bemanaged through derivatives and other risk management operations andnot reduced leverage.

I. INTRODUCTION

Corporate capital structure remains a controversial issue in modern corporatefinance. Since the seminal work by Modigliani and Miller (1958), a plethora of

n Special thanks to Tom Smith, participants at the 2005 AFAANZ conference and BruceGrundy (editor) for comments and suggestions on earlier drafts.

r 2009 The Authors. Journal compilation r International Review of Finance Ltd. 2009. Published by BlackwellPublishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

International Review of Finance, 9:1–2, 2009: pp. 1–26DOI: 10.1111/j.1468-2443.2009.01083.x

research has been undertaken in attempting to identify the determinants ofcapital structure. This research has resulted in a variety of different theories thatpredict why firms have a particular capital structure. Support for one theoryover another has been determined by empirical evidence. However, one areathat has received only little attention is whether multinational corporations(MNCs) differ in capital structure and capital structure determinants relative todomestic corporations (DCs).1 In the majority of empirical research nodistinction is made between these two different forms of corporations. This issurprising given the anecdotal evidence of the differences between the two.

Theoretically, it is often argued that the international diversification ofearnings should enable MNCs to sustain higher levels of debt than DCs.Empirically, Fatemi (1988), Lee and Kwok (1988), Burgman (1996), Chen et al.(1997), Homaifar et al. (1998), Chkir and Cosset (2001) and Doukas andPantzalis (2003) investigated capital structures between US-based MNCs andDCs and all reported US MNCs as having less debt than US DCs. Burgman (1996)and Chkir and Cosset (2001) argue that capital and labor market imperfectionsand complexity of international operations for MNCs leads to the lower debtlevels.

Mittoo and Zhang (2008) find Canadian MNCs to have higher leveragerelative to Canadian DCs. This is found to be the result of Canadianmultinationals accessing US bond markets. Singh and Nejadmalayeri (2004)find that multinationality is positively associated with higher leverage for asample of French corporations. No substantive reason is given for their results.Akhtar (2005) does not find any significant difference in leverage betweenAustralian DCs and MNCs.

The existence of capital and labor market imperfections can lead to higheragency costs for MNCs (Chkir and Cosset 2001). Capital and labor marketimperfections include increased complexity of international operations, whichimpose higher monitoring costs for MNCs. Often capital markets in hostcountries are less developed than in domicile countries, this may result inconflicts between debtholders and equityholders, particularly if host countrynorms are not followed by the MNC. Barnea et al. (1985) argue that severalmechanisms through which capital and labor markets can mitigate agencycosts.2 In practice, however, significant international capital and labor marketbarriers exist which could limit the efficiency of these market mechanisms inreducing the agency costs faced by MNCs. These barriers include restrictions onthe degree of foreign ownership of firms, direct controls on the export andimport of capital, reserve requirements on bank deposits held by foreign firmsand the possibility of expropriation of foreign holdings.

1 The words ‘firm’ and ‘corporation’ are used interchangeably.

2 These mechanisms include both internal and external mechanisms such as the leverage, market

for corporate control, economic and legal institutions and external managerial labor markets

(Shleifer and Vishny 1997; Cremers and Nair 2005).

International Review of Finance

r 2009 The AuthorsJournal compilation r International Review of Finance Ltd. 20092

Generally, MNCs are expected to operate in a relatively more complexenvironment than DCs (Burgman 1996). MNCs may face higher auditing costs,language differences, sovereignty uncertainties, and varying legal and account-ing systems. Further, investors in MNCs are confronted with increasedinformation gaps and higher costs of investigation. Hence, MNCs are likely toface significantly higher monitoring costs than DCs.

The inconsistency in levels of leverage for MNCs relative to DCs warrantsfurther investigation from other countries to help further ascertain if suchdifferences are due to unique operating environments. One country which has aunique corporate system with control mechanisms different from other westerncountries is Japan (He and Ng 1998). However, Japan is of interest for otherreasons. First, Japanese MNCs control substantial amounts of wealth and theyrepresent some of the largest corporations worldwide.3 Their rapid expansionglobally, their capability to integrate business across national frontiers, andtheir economic power has led significant repercussions in the world economy.Understanding the determinants of capital structure for both MNCs and DCs inJapan is important in its own right given the amount of wealth involved.Secondly, determining if Japanese MNCs have different determinants of capitalstructure to Japanese DCs helps to further our understanding of capitalstructure choice. Finally, to the best of our knowledge, no study has yetconducted a comprehensive analysis of the capital structure determinants ofJapanese MNCs. Therefore, we aim to identify the determinants of capitalstructure for a sample of Japanese corporations, to identify if being a JapaneseMNC is important in explaining capital structure and if the capital structure isdifferent for MNCs and DCs, what explains the difference.

The analyses first considers the significance of standard determinants ofcapital structure over a 10 year period to 2003 for a sample of Japanese DCs andMNCs. Determinants of capital structure that are considered include businessrisks, age, collateral value of assets, free cash flows, foreign exchange risks,growth opportunities, non-debt tax shields, political risks, profitability and size.The results are broadly consistent with other studies on capital structure in thefull sample. We then introduce a dichotomous variable for multinationality toascertain if being a MNC is a determinant of capital structure. Then interactivedichotomous variables are introduced to ascertain the relative importance ofbeing an MNC for the determinants of capital structure relative to DCs.

On a univariate basis Japanese MNCs have significantly less leverage thanJapanese DCs.4 This is consistent with US studies (Lee and Kwok 1988; Burgman1996; Homaifar et al. 1998), but inconsistent with studies from Canada (Mittooand Zhang 2008), France (Singh and Nejadmalayeri 2004) and Australia (Akhtar2005). Japanese MNCs are vastly different from Japanese DCs on most othercapital structure determinants. Japanese MNCs have significantly lowercollateral value of assets but are significantly older and larger, and have

3 This includes firms such as Toyota Motor Corporation, Sony, Kyocera and Hitachi Ltd.

4 The results are insensitive to a range of measures of leverage.

Determinants of Capital Structure for Japanese MNCs and DCs

r 2009 The AuthorsJournal compilation r International Review of Finance Ltd. 2009 3

significantly higher agency costs, free cash flows, foreign exchange risks, growthlevels, non-debt tax shields, political risks and profitability relative to JapaneseDCs. Business risks are not significantly different between the MNCs and DCs.

The univariate results only tell part of the MNC versus DC story. Most of thewell-established determinants of capital structure are significant in the fullsample of Japanese firms. When the sample is divided between MNCs and DCsbusiness risks are significant for MNCs but not DCs and foreign exchange risksare significant for DCs but not for MNCs. However, in explaining the differencebetween capital structure for MNCs and DCs, firm size is significant along withbusiness risks and foreign exchange risks. These results suggest that whenempirically testing theories on capital structure the impact of multinationalfirms in the sample needs to be considered.

The paper is divided into four sections. The next section reviews the relevantliterature and defines the variables. The third provides discussion of the methodand data respectively. Section four discusses the results and section fivesummarizes the key findings and concludes the paper.

II. CAPITAL STRUCTURE DETERMINANTS AND VARIABLEDEFINITIONS

Any paper considering capital structure would be incomplete if no reference wasmade to the work of Franco Modigliani and Merton Miller. Their substantialtheoretical and empirical investigations on capital structure from Modiglianiand Miller (1958, 1966) and Miller and Modigliani (1963) onwards, hereafterM&M, have provided one of the most significant contributions to corporatefinance. A review of this work from 1958 until 1988 can be found in Miller(1988). Grundy (2002) also provides a compilation of selected works of Miller(which includes work with Modigliani).

The variables that are now generally regarded as determinants of capitalstructure include business risks, age, collateral value of assets, free cash flows,foreign exchange risks, growth opportunities, non-debt tax shields, politicalrisks, profitability, and size (Myers 1984; Jensen 1986; Rajan and Zingales 1995;Burgman 1996; Chen et al. 1997; Homaifar et al. 1998; Chkir and Cosset 2001;Doukas and Pantzalis 2003; Frank and Goyal 2009). Generally, these variablesrelate to value and risks of the firm as faced by bondholders, equityholders andmanagers. Each variable can be traced back to one or more of the many theorieson capital structure.

Apart from identifying the determinants of capital structure, an importantissue is defining what is meant by capital structure or leverage.5 Rajan andZingales (1995) provide an overview of the different definitions of leverage. Inrelation to research on the capital structure of MNCs, two main definitions ofleverage have been applied. Doukas and Pantzalis (2003) and Mittoo and Zhang

5 The terms ‘leverage’ and ‘capital structure’ are used interchangeably.



(2008) among others, define leverage as long-term debt scaled by total debt plusmarket value of equity. This measure seems inappropriate in our study becauseit is often argued that short-term debt, which is included in total debt, has ahigh variability throughout the year. As a result, since our data are based onannual data, the addition of short-term debt based on the reporting date onlywould unnecessarily inflate the debt ratio. On the other hand, Lee and Kwok(1988), Burgman (1996), Chen et al. (1997) and Chkir and Cosset (2001)amongst others define leverage as long-term debt scaled by long-term debt plusmarket value of equity. This definition of leverage is chosen to be appropriatefor our analysis.6 Therefore, we define leverage for firm i at time t, as

LEVERAGEi;t ¼Long-term debti;t

ðLong-term debti;t þMarket value of equityi;tÞ:

In this definition of leverage we do not include lease commitments. This isbecause there is considerable variation in reporting practices regarding leasecommitments in Japan. This was also noted by Rajan and Zingales (1995).Given that the implications of this for capital structure research are unclear,we chose not to include lease commitments in our definition of leverage.The independent variables are discussed below.

A. Business risks

Business risks refer to the risk associated with the future operations of thebusiness. Firms with higher business risks face higher expected costs of financialdistress. Generally, it is expected that there is an inverse relation betweenleverage and business risks due to the associated increase in bankruptcy risks(Burgman 1996; Chen et al. 1997). According to tradeoff theory the increase inbankruptcy risk competes with the tax deductibility of interest. Japan follows aclassical tax system. Under a classical tax system there is a tax incentive tofinance with debt relative to equity due to the tax deductibility of interest (thework of M&M). However, higher levels of business risks reduce the probabilitythat interest tax shields are going to be utilized and increase the risk ofbankruptcy. Therefore, according to the tradeoff theory higher risks shouldresult in lower levels of leverage. However, firms that are risky are more likely tosuffer from information asymmetries. According to pecking order theory, firmswith higher information asymmetries are expected to have higher levels ofleverage.

In relation to MNCs, if they have better ability than DCs to diversify acrossless than perfectly correlated markets, business risks for MNCs should be lessthan DCs. According to tradeoff theory MNCs should therefore be able to

6 We conducted all our tests using the alternative definition for leverage and the results were

substantially similar to those reported. The correlation between the two measures of leverage is

90%. The results were also insensitive to other measures of leverage including total debt scaled

by total assets and long-term debt scaled by total assets.



support relatively more leverage than DCs (Reeb et al. 1998). However, if wetake a pecking order perspective then the lower information asymmetryassociated with the lower risk will result in MNCs having lower levels of debtthan DCs. From the point of view of a debtholder, the relevant measure of risk isthe standard deviation of the returns from the assets backing their debt claim. Astandard measure of risk is the standard deviation of equity returns. However,this measure incorporates both business risks and financial risks faced by theequityholders. Financial risk is the increase in risk associated with leverage.Therefore, a de-geared equity standard deviation is used as a proxy for businessrisks.

When de-gearing the equity standard deviation, the Japanese corporate taxrate and its effect on the value of debt must be considered. Therefore, we definebusiness risks in the following way:

BUSINESS RISKSi;t ¼si

1þ Di;tð1�tcÞEi;t

� �

where si is the Standard deviation of equity returns (Ri) for firm i over theprevious 52 weeks, Ri5Ln(Pi,w/Pi,w�1) and Pi,w is the price of stock i in week w,Di,t is the book value of total debt at end of year t, Ei,t the market capitalizationof ordinary equity at end of year t, and tc the corporate tax rate.7

B. Age

Age is usually seen as a proxy for a range of issues relevant to capital structurechoice. This includes agency costs, default risks and information asymmetries.Older firms are expected to face lower debt-related agency costs (Frank andGoyal 2009). Lower debt-related agency costs result in greater access to debt andhence a higher leverage ratio for older firms is expected. Older firms generallyface lower default risks due to more stable earnings and this also implies ahigher leverage ratio for older firms.

Age may also proxy for lower information asymmetries. In the presence ofinformation asymmetries firms should finance with relatively value-insensitivesecurities like debt, rather than by issuing value sensitive securities like equity.According to pecking order theory managers will prefer debt over equity andtherefore with less information asymmetry older firms are expected to have lessleverage. Due to these conflicting theories it is not clear whether age is expectedto have a positive or negative relation to leverage. However, we expect MNCs tobe older than DCs since firms are generally likely to begin as DCs and expandover time to become multinationals.

7 The corporate tax rate in Japan over the sample period is 37.5% from 1994 to 1998, 34.5% in

1999 and 30% from 1999 to 2003. Source: http://www.mof.go.jp/english/zei/report2/

zc001d01.htm.



http://www.mof.go.jp/english/zei/report2/zc001d01.htm

http://www.mof.go.jp/english/zei/report2/zc001d01.htm

The variable for firm age is measured following Michaelas et al. (1999) amongothers:

AGEi;t ¼ lnðage of firm i; in years from date of incorporationÞ:

C. Collateral value of assets

The collateral value of assets or tangibility of assets, held by a firm has beenfound to be a determinant of leverage (Rajan and Zingales 1995). The expecteddirection of this relation is unclear. Firms with higher tangible assets can beexpected to have higher leverage due to lower default costs and less debt-relatedagency problems. Tangible assets are likely to have an impact on the borrowingdecisions of a firm because they have a greater value than intangible assets incase of bankruptcy. Additionally, the moral hazard risks are reduced when thefirm offers tangible assets as collateral, because this constitutes a positive signalto the debtholders. If tangible assets lower information asymmetries then thismakes equity relatively less costly, thereby lowering leverage ratios. Generally,we would expect tangible assets to have lower information asymmetry thanintangible assets. However, this may not always be the case and therefore wecannot predict if information asymmetries are always going to be lower andhence providing an inverse relation between tangibility and leverage.

In relation to MNCs and DCs, it is uncertain whether the level of tangibleassets is higher or lower for MNCs relative to DCs.

Following Friend and Lang (1988) we define the collateral value of assets foreach firm as

COLLATERALi;t ¼Book value of tangible assetsi;t

Book value of total assetsi;t:

D. Free cash flows

Easterbrook (1984) and Jensen (1986) argue that excess free cash flow is cashflow in excess of that required to fund all projects that have positive net presentvalue. Firms with excess free cash flow face conflicts of interest betweenstockholders and managers. The problem is how to motivate managers todistribute excess funds rather than investing them below the costs of capital oron organizational inefficiencies, such as management perks. Leverage is aneffective means for addressing the free cash flow problem, because contractualpayments of interest and principal must be made. Bondholders can take thefirm to bankruptcy if managers do not maintain loan repayments. Accordingly,debt reduces the agency cost of free cash flows by reducing the cash flowavailable for spending at the discretion of managers (Stulz 1990; Harris andRaviv 1991). Therefore, firms with higher free cash flow are expected to havehigher leverage. In relation to MNCs and DCs, it is uncertain whether the levelof free cash flow is higher or lower for MNCs relative to DCs.



We define free cash flow as per Lehn and Poulsen (1989), however, someadjustment has been made to fit in within our study purpose. For example,firms with net cash flows from operations can mitigate a free cash flow problemassociated with managers investing in projects that increase firm size (or theirown perquisite consumption) but do not increase the shareholder’s wealth, bycommitting to distribute future net cash flows from operation as interest anddebt payments. Therefore, Lehn and Poulsen’s (1989) measurement isinappropriate to implement in this study because it nets out dividend andinterest payments. Further, we also standardize free cash flows by the bookvalue of total asset so that it is consistent with the other ratio variables:

FREE CASH FLOWSi;t ¼EBITi;t þDEPi;t þ AMOi;t � TAXi;t

Book value of total assetsi;t

where EBITi,t is the earnings before interest, tax and abnormal items for firm iin year t, DEPi,t the depreciation expense for firm i in year t, AMOi,t theamortization expense for firm i in year t, and TAXi,t the total tax paid for firm iin year t.

E. Foreign exchange risks

Krainer (1972), examining the applicability of M&M to a multinational firm,argues that the existence of foreign exchange risks are sufficient to causeotherwise identical firms to belong to different risk classes. Since M&M assumefirms are of similar risk class, Krainer (1972) concludes that such principlescannot be applied in the international case. The more sensitive a firm’s cashflows and earnings are to foreign exchange rates, the lower the expected level ofdebt due to increases in default risk. Burgman (1996) analyzed the relationbetween foreign exchange risks and corporate financing decisions and reportedthat foreign exchange risks significantly affects financing decisions. Further,exchange rate movements affect both the cash flows of a firm’s operations anddiscount rates employed to value the cash flows (Bartov et al. 1996).8 In relationto MNCs, we expect MNCs to have a higher exposure to foreign exchange risksthan DCs. This implies that MNCs would have relatively lower leverage, aftercontrolling for all other variables. However, foreign exchange risks arecommonly hedged by firms. Although MNCs may have greater exposure toforeign exchange risks, the risks may be hedged. Of course DCs also have theopportunity to hedge any foreign exchange risks. However, evidence ofeconomies of scale in risk management, particularly the use of derivatives,may result in limited derivative use by DCs (Smith and Stulz 1985; Nance et al.1993; Martin and Mauer 2004). Therefore, although we expect foreign sales to

8 Bartov et al. (1996) shows that there is an increase in the variability of equity returns for MNCs

following a period of increased exchange-rate variability. The results suggest that the increase in

exchange-rate fluctuations is an indication of an increase in the riskiness of cash flows.



be greater for MNCs relative to DCs we make no prediction as to what effect thismay have on capital structures.

Foreign exchange risk exposure of each firm is defined as per Wright et al.(2002):

FX RISKSi;t ¼Total foreign salesi;t

Total salesi;t:

F. Growth

Higher agency costs are expected to result in lower debt levels (Jensen 1986;Doukas and Pantzalis 2003). Significant agency costs arise from the conflict ofinterest between stockholders and bondholders. The consistent message ofagency models is that these conflicts create incentives for stockholders to takeactions that benefit themselves at the expense of bondholders. Myers (1977)argues that agency problems are especially serious for assets that give the firmthe option to undertake growth opportunities in the future. The greater thefirm’s investment in such assets the less it would be debt financed. However,firms with more growth options are expected to have higher informationasymmetries, and according to pecking order theory these firms are expected tohave higher leverage.

The value of future growth opportunities can be created through exploitationof imperfections in product, factor or capital markets. MNCs tend to be in abetter position than DCs for taking advantage of such market imperfections(Homaifar et al. 1998). This implies that MNCs are expected to have highergrowth opportunities than DCs.

Adam and Goyal (2007) show that the market-to-book ratio of commonequity is the most reliable proxy for growth opportunities. Accordingly, we usethe market-to-book ratio as a proxy of growth opportunities9:

GROWTHi;t ¼MVi;t

BVi;t

where, MVi,t is the market value of common equity for firm i in year t and BVi,t

the book value of common equity for firm i in year t.

G. Non-debt tax shields

Taxation and its relation to capital structure are explicitly linked to theapplicable tax regime. Under the Japanese classical tax system the taxdeductibility of interest is expected to induce a preference for debt. De Angeloand Masulis (1980) formalized a framework whereby tax deductions that are notassociated with debt act as substitutes for interest deductions. These non-debt

9 The market-to-book ratio is also used to proxy for market timing effects. Higher market-to-book

ratios make equity relatively more attractive than debt and as such we would expect a negative

relation between market-to-book and leverage.



tax shields compete with interest as a tax deduction. Firms with higher non-debt tax shields are expected to have lower leverage, as the tax benefits ofleverage are relatively less valuable. MNCs should be better equipped than DCsto exploit taxation regulations across different countries for the purpose ofreducing their tax liabilities (Homaifar et al. 1998). Therefore, we expect thatMNCs should have higher levels of non-debt tax shields leading lower levels ofleverage than DCs.

Following Bradley et al. (1984) and Titman and Wessels (1988) we definenon-debt tax shield as

NDTSi;t ¼Total annual depreciation expensei;t

Book value of total assetsi;t:

H. Political risks

Political risk is the chance that political events will have an adverse effect on theoperations of the firm. Political risks include expropriation of assets, tradecontrols, institutional ineffectiveness, threat of war, social unrest, disorderlytransfers of power, political violence, international disputes, regime changesand regulatory restrictions (Jodice 1985). This means that firms with significantforeign financing, foreign suppliers or customers, or other internationaltransactions or assets, are relatively exposed to adverse changes in currencycontrols, capital flow barriers and other laws and regulations that constitutepolitical risk. DCs also are exposed to their own country’s political instability.However, it can be argued that MNCs are relatively more exposed to politicalrisks than DCs (Burgman 1996). MNCs operating overseas subsidiaries areexpected to have higher political risks than DCs. Therefore, given a negativerelation between political risk and leverage we expect that MNCs would haveless leverage relative to DCs after controlling for other variables.

Political risk of each firm is measured as follows. Let C be the sample ofcompanies and R the set of different countries that companies in the sampleoperate from. Let Ic,r be the revenue of company c coming from a particularcountry, r. The sum of sales revenues for company c, from different countries isthe sum of Ic,i. Pc,r is the proportion of revenue from a particular country relativeto the total revenue of company c. In notation, this is as follows:

Pc;r ¼Ic;rPi2R Ic;i

:

For each country in which a company operates a political risk rating isobtained from the PRS group Handbook of Country and Political Risk Analysis(Howell 2005). This is denoted as lambda (li). The political risk rating is thenmultiplied by the proportion of revenue from that particular country relative tothe total revenue of the company. This provides a measure of political risk facedby the firm. The maximum value of this political risk rating variable is 100. Thisindicates the lowest level of political risk. A minimum value of zero indicates



the riskiest political risk rating. To ease interpretation we subtract the rating (gc)from 100 to end up with the most risk as 100 and the least risk as zero. Innotation this is as follows:

POLITICAL RISKSi;t ¼ 100� gc ¼ 100�Xi2R

liPc;i ¼ 100�P

i2R liIc;iPi2R Ic;i

:

I. Profitability

Profitable firms generally are less likely to face bankruptcy and value morehighly the tax deduction of interest payments than less profitable firms. Thisprovides incentives to utilize more debt.10 Debt provides a discipline onmanagers and therefore from an agency perspective, a positive relation betweenprofitability and leverage is expected. Myers (1984) pecking order theory ofcapital structure shows that if a firm is profitable then it is more likely thatfinancing would be from internal sources rather than external sources.Therefore, a negative relation between profitability and leverage is expectedunder the pecking order theory.

MNCs have better opportunities than DCs to earn more profit mainly due tohaving access to more than one source of earnings and a better chance toexploit favorable business conditions in particular countries (Kogut 1985;Bartlett and Ghoshal 1989; Chkir and Cosset 2001). In addition, Lee et al.(1996) find that multinational diversity has a positive impact on profitabilityfor US firms. Therefore, although the expected relation between profitabilityand leverage is unclear it is expected that Japanese MNCs will be more profitablethan Japanese DCs.

The variable chosen to measure profitability for each firm follows Doukas andPantzalis (2003):

PROFITABILITYi;t ¼ Average net income to total sales for the past 4 years for firm i:

J. Size

Firm size has been found to be a positive determinant of capital structure(Agrawal and Nagarajan 1990; Rajan and Zingales 1995; Chkir and Cosset 2001).Rajan and Zingales (1995) argues that larger firms tend to be more diversifiedand therefore less likely to go bankrupt. Consequently, size should have apositive relation with leverage. However, larger firms are expected to have lowerinformation asymmetries making equity issues more attractive. This implies anegative relation between size and leverage. In relation to MNCs and DCs, it isexpected that MNCs are larger than DCs.

10 The tradeoff theory comes in several forms and it is not clear what drives the relation between

leverage and profits (Frank and Goyal 2009).



The natural logarithm of total assets is commonly considered a proxy for thesize of each firm:

SIZEi;t ¼ Lnðbook value of total assetsi;tÞ:

In summary, although there are many determinants of capital structure,generally it is not clear what relation is expected between each determinant andleverage. Of the main theories such as tradeoff, pecking order or market timingthey generally offer inconsistent expected relations between leverage and itsdeterminants. In relation to differences between MNCs and DCs on individualdeterminants we can offer some expected results, but whether these lead todifferences between capital structures between them is an empirical issue.

III. DATA AND METHOD

A. Data

Three hundred and sixty Japanese corporations that existed over the period1994–2003 were initially randomly selected from the Osiris database.11 Thisdatabase reports geographical segmental sales information and details thelocations of firm subsidiaries, which is required in identifying MNCs. Weexclude banks, finance and insurance companies because their operations anddebt specifications being unique to their industry. The final sample of firms isobtained after meeting the sample selection criteria. For example, all firms arerequired to have at least 4 years of observations for the profitability measure.Firms are also required to have relevant accounting data items to constructother variables. If a firm reported having at least one overseas subsidiary in 2003and reported foreign sales in any year it is coded as an MNC.12 Of the 360 firms,209 were coded as MNCs and 147 were coded as DCs.

B. Method

Our data is cross-sectional time series. To avoid the problems of using cross-sectional proxies for time-sequenced variables (Gul 1999), we conduct poolcross-sectional time series analysis.13 A pooled cross-sectional time seriesregression model (Model 1) is used to analyze capital structure determinants:

11 The Osiris database is provided by Bureau Van Dijk (http://www.bvdep.com/en/OSIRIS.html).

12 Different studies have used different definitions of multinationality. Information on Osiris for

the number of overseas subsidiaries was only available in 2003. We do not consider this a

significant problem as anecdotal evidence suggests that the number of overseas subsidiaries

remains relatively constant over time. MNCs had foreign sales on average in five of the 10 years

and DCs had foreign sales on average in one of the 10 years.

13 We found that our results were robust to different regression techniques.



http://www.bvdep.com/en/OSIRIS.html

LEVERAGEi;t ¼ b0 þ b1BUSINESS RISKSi;t þ b2AGEi;t þ b3COLLATERALi;t

þ b4FREE CASH FLOWSi;t þ b5FX RISKSi;t þ b6GROWTHi;t

þ b7NDTSi;t þ b8POLITICAL RISKSi;t

þ b9PROFITABILITYi;t þ b10SIZEi;t þ ei;t ð1Þ

where all variables are as previously defined.Model 1 will be estimated using the full sample of both Japanese MNCs and

DCs and then separately for MNCs and DCs. This will provide evidence of thesignificance of the determinants of capital structure for Japanese firms generallyand then for MNCs and DCs.

In order to assess the significance of multinationality in explaining capitalstructure, a dichotomous variable for multinationality will be augmented intoModel 1 using the full sample and a log likelihood ratio (LR) test will beconducted. The LR test enables us to determine whether multinationality makesa significant contribution to explaining variation in leverage.

Model 2 is relevant for identifying the significance of capital structuredeterminants for MNCs relative to DCs. Again it is a pooled cross-sectional timeseries regression model but now with a dichotomous variable and dichotomousinteraction variables for multinationality:



þ b7NDTSi;t þ b8POLITICAL RISKSi;t þ b9PROFITABILITYi;t

þ b10SIZEi;t þ b11MULTi þ b12MULTi � BUSINESS RISKSi;t

þ b13MULTi � AGEi;t þ b14MULTi � COLLATERALi;t þ b15MULTi

� FREE CASH FLOWi;t þ b16MULTi � FX RISKSi;t þ b17MULTi

� GROWTHi;t þ b18MULTi �NDTSi;t þ b19MULTi

� POLITICAL RISKSi;t þ b20MULTi � PROFITABILITYi;t

þ b21MULTi � SIZEi;t þ ei;t ð2Þ

where MULTi is a dichotomous variable equal to unity if firm i is an MNC andzero otherwise; and all other variables are as previously defined.

The purpose of the interaction dichotomous variables (MULTi) is to allow thesignificance of the determinants of leverage for MNCs relative to DCs to bedetermined. In other words, if any of the coefficients b12 through b21 aresignificant then the associated variable is significantly explaining differences incapital structure between MNCs and DCs.14 An LR test will be conductedto ascertain the significance of augmenting the dichotomous variable intoModel 2 as a separate independent variable. Appropriate adjustments for

14 The magnitude each variable has on explaining leverage for MNCs is the sum of both

coefficients. For example the relation SIZE of MNC has on explaining leverage consists of

summing b10 and b21.



Heteroskedasticity will be made using White (1980) Heteroskedasticity-consistent standard errors and covariance.

IV. RESULTS

Table 1 (panel A) contains descriptive statistics of the variables used in theregression models. Table 1 (panel B) contains correlations between the variables.

Panel A of Table 1 provides some interesting results. Characteristics of MNCsand DCs differ significantly on all variables except business risks. In relation toleverage, Japanese MNCs have significantly less leverage than Japanese DCs.This is similar to the results of studies on US multinationals (Lee and Kwok1988; Burgman 1996; Homaifar et al. 1998), but different to studies ofmultinationals from other countries including France, Australia, and Canada(Singh and Nejadmalayeri 2004; Akhtar 2005; Mittoo and Zhang 2008,respectively).

Business risks, as proxied by de-geared equity standard deviation, arerelatively lower for MNCs but this is not significant. It is the only variablewhere on a univariate analysis we found no significant difference betweenMNCs and DCs. We expected MNCs to have significantly lower business risksrelative to DCs due to their perceived ability to diversify across less thanperfectly correlated markets. However, this is not evident in our sample.

MNCs are on average approximately 10 years older than DCs.15 We expectedMNCs to be older than DCs simply due to the development that is required for afirm to become an MNC, so this result is expected.

The collateral value of assets is significantly lower for MNCs relative to DCs,indicating that Japanese DCs have relatively higher levels of tangible assets thantheir MNC counterparts. Due to lack of theoretical justification we were notprepared to hypothesize on differences between MNCs and DCs on the level oftangible assets.

MNCs have significantly higher levels of free cash flows indicating greaterpotential conflicts of interest between stockholders and managers. Again wecould not establish any theoretical priors for differences between free cash flowsfor MNCs and DCs.

MNCs have on average approximately five times the level of sales in foreigncurrency relative to DCs. Although we made no prediction on the actualdifference in foreign sales we expected MNCs to have relatively higher foreignsales than DCs. Although this may translate to potential foreign exchange risks,the foreign currency exposures may be hedged and the risks transferred.

MNCs have significantly higher growth opportunities than DCs as expected,reflecting better opportunities of MNCs to exploit imperfections in markets.Also as expected MNCs have significantly higher non-debt tax shields implyingthat they are better equipped to exploit tax regulations across different

15 e4.21567 years and e4.04557 years.



Ta

ble

1D

escr

ipti

ve

stati

stic

so

fsa

mp

levari

ab

les

Pan

elA

:D

escr

ipti

ve

stati

stic

s

Vari

ab

leM

ean

Med

ian

Max.

Min

.SD

t-St

ati

stic

Wil

coxo

n/

Man

n–W

hit

ney

:Z

-sco

re

LEV

ER

AG

EM

NC

0.2

10.1

50.8

20.0

00.1

9D

C0.2

90.2

30.9

20.0

00.2

4A

ll0.2

80.1

70.9

20.0

00.2

1�

4.0

8nnn

�3.2

7nnn

BU

SIN

ESS

RIS

KS

MN

C0.0

70.0

316.0

0�

1.5

10.7

3D

C0.0

80.0

34.9

0�

0.3

70.3

8A

ll0.0

70.0

316.0

0�

1.5

10.6

4�

0.1

3�

0.3

6A

GE MN

C4.2

14.2

35.0

93.4

70.3

1D

C4.0

44.0

44.7

22.6

40.5

0A

ll4.1

64.1

65.0

92.6

40.3

84.2

8nnn

3.7

2nnn

CO

LLA

TER

AL

MN

C0.3

00.2

90.6

20.0

50.1

0D

C0.3

50.2

90.8

90.0

30.1

8A

ll0.3

10.2

90.8

90.0

30.1

3�

3.9

3nnn

�3.5

8nnn

FR

EE

CA

SHFLO

WS

MN

C0.0

70.0

61.0

8�

0.0

80.0

6D

C0.0

50.0

50.1

7�

0.0

90.0

4A

ll0.0

60.0

61.0

8�

0.0

90.0

53.9

2nnn

5.0

2nnn

FX

RIS

KS

MN

C0.2

60.2

10.9

50.0

00.2

6D

C0.0

50.0

00.6

00.0

00.1

2A

ll0.2

00.1

20.9

50.0

00.2

514.5

7nnn

11.5

4nnn

GR

OW

TH

MN

C1.7

11.5

69.0

00.0

01.0

0D

C1.3

91.1

210.0

00.0

01.1

6A

ll1.5

91.4

110.0

00.0

01.0

87.4

8nnn

12.5

1nnn



Ta

ble

1(c

onti

nued

)

Pan

elA

:D

escr

ipti

ve

stati

stic

s

Vari

ab

leM

ean

Med

ian

Max.

Min

.SD

t-St

ati

stic

Wil

coxo

n/

Man

n–W

hit

ney

:Z

-sco

re

ND

TS

MN

C0.0

40.0

40.1

00.0

00.0

2D

C0.0

30.0

30.1

00.0

00.0

3A

ll0.0

40.0

40.1

00.0

00.0

25.5

2nnn

5.7

0nnn

PO

LIT

ICA

LR

ISK

SM

NC

16.9

315.9

331.4

913.0

02.9

4D

C15.8

614.5

022.2

913.0

03.1

0A

ll16.6

315.5

831.4

913.0

03.0

24.2

5nnn

5.7

2nnn

PR

OFIT

AB

ILIT

YM

NC

0.0

20.0

20.4

8�

0.1

40.0

5D

C0.0

10.0

10.4

2�

0.1

40.0

5A

ll0.0

20.0

10.4

8�

0.1

40.0

52.4

0nn

3.9

1nnn

SIZ

E MN

C12.9

512.9

516.1

110.2

31.3

1D

C12.1

012.0

616.5

00.0

01.7

0A

ll12.7

112.7

416.5

00.0

01.4

87.0

4nnn

6.7

5nnn



Pan

elB

:C

orr

elati

on

s

BU

SIN

ESS

RIS

KS

AG

EC

OLLA

TER

AL

FR

EE

CA

SHFLO

WS

FX

RIS

KS

GR

OW

TH

ND

TS

PO

LIT

ICA

LR

ISK

SPR

OFIT

AB

ILIT

YSI

ZE

LEV

ER

AG

E0.0

10.2

10.3

5�

0.2

4�

0.1

7�

0.2

6�

0.0

2�

0.1

0�

0.4

20.1

4B

USI

NESS

RIS

KS

0.0

30.0

5�

0.0

1�

0.0

2�

0.0

2�

0.0

00.0

7�

0.0

10.0

0

AG

E�

0.2

0�

0.0

30.0

1�

0.0

9�

0.1

0�

0.0

5�

0.0

80.2

5C

OLLA

TER

AL

0.0

9�

0.2

0�

0.0

80.4

2�

0.0

8�

0.1

5�

0.0

4FR

EE

CA

SHFLO

WS

0.1

50.2

10.3

6�

0.2

30.3

80.1

0

FX

RIS

KS

0.1

60.3

30.2

20.0

20.2

1G

RO

WT

H0.1

16

0.1

00.2

50.2

3N

DT

S0.0

9�

0.0

40.2

3PO

LIT

ICA

LR

ISK

S0.0

30.0

1

PR

OFIT

AB

ILIT

Y0.0

1

Th

ista

ble

rep

ort

svari

able

des

crip

tive

stati

stic

san

dco

rrel

ati

on

so

fth

esa

mp

levari

able

sfo

r360

Jap

an

ese

mu

ltin

ati

on

al

corp

ora

tio

ns

(MN

Cs)

an

dd

om

esti

cco

rpo

rati

on

s(D

Cs)

over

10

yea

rsto

2003.

Pan

elA

:D

escr

ipti

ve

stati

stic

sfo

rM

NC

s,D

Cs

an

dth

efu

llsa

mp

le-m

ean

,m

edia

n,

max

imu

m,

min

imu

man

dst

an

dard

dev

iati

on

are

rep

ort

edin

colu

mn

s2–6

,re

spec

tivel

y,an

dt-

test

san

dW

ilco

xo

n/M

an

n–W

hit

ney

test

sare

rep

ort

edin

colu

mn

s7

an

d8,re

spec

tivel

y.LEV

ER

AG

Eis

lon

g-t

erm

deb

tsc

ale

db

ylo

ng-t

erm

deb

tp

lus

mark

etvalu

eo

feq

uit

y.T

he

pro

xy

for

BU

SIN

ESS

RIS

KS

ises

tim

ate

das

the

de-

gea

red

stan

dar

dd

evia

tio

no

fo

rdin

ary

equ

ity.

Fir

mA

GE

isth

en

atu

rallo

gari

thm

of

the

age

of

the

firm

inyea

rsfr

om

date

of

inco

rpo

rati

on

.T

he

pro

xy

for

CO

LLA

TER

AL

valu

eo

fass

ets

ises

tim

ate

das

the

rati

oo

fta

ngib

leass

ets

toto

tal

ass

ets.

Th

ep

roxy

for

FR

EE

CA

SHFLO

WS

ises

tim

ated

as

earn

ings

bef

ore

inte

rest

an

dta

xad

just

edfo

rd

epre

ciati

on

,am

ort

izati

on

an

dta

xati

on

scale

db

yto

tal

ass

ets.

Th

ep

roxy

for

fore

ign

exch

ange

(FX

)ri

sks

ises

tim

ate

das

the

rati

oo

ffo

reig

nsa

les

toto

talsa

les.

Th

ep

roxy

for

GR

OW

TH

op

po

rtu

nit

ies

faci

ng

the

firm

ises

tim

ated

as

the

mark

etto

bo

ok

rati

oo

fco

mm

on

equ

ity.

Th

ep

roxy

for

no

n-d

ebt

tax

shie

lds

(ND

TS)

ises

tim

ate

das

tota

lan

nu

al

dep

reci

ati

on

exp

ense

scale

db

yto

tal

ass

ets.

Th

ep

roxy

for

PO

LIT

ICA

LR

ISK

Sfa

ced

by

the

firm

ises

tim

ated

as

afu

nct

ion

of

the

pro

po

rtio

no

fre

ven

ue

fro

md

iffe

ren

tco

un

trie

sw

eigh

ted

by

ap

oli

tica

lri

skfa

cto

r.T

he

pro

xy

for

PR

OFIT

AB

ILIT

Yis

esti

mate

das

the

aver

age

net

inco

me

toto

tal

sale

sfo

rth

ep

ast

fou

ryea

rs.

Th

ep

roxy

for

firm

SIZ

Eis

esti

mate

das

the

nat

ura

llo

gar

ith

mo

fto

tal

ass

ets.

nnn

an

dnn

Sign

ifica

nt

at

the

1%

an

d5%

level

s,re

spec

tivel

y.Pan

elB

:R

epo

rts

corr

elat

ion

sbet

wee

nin

dep

end

ent

var

iab

les.



countries. As we expected political risk faced by Japanese MNCs is significantlyhigher than those faced by DCs. MNCs are also significantly more profitable andsignificantly larger than DCs as we expected.

Panel B of Table 1 indicates that multicollinearity in Models 1 and 2 isunlikely to be a major problem with all correlations less than 50%. On aunivariate basis on the full sample leverage is positively related to business risks,age, collateral value of assets and size and negatively related to free cash flows,foreign exchange risks, growth, non-debt tax shields, political risks andprofitability. These results are consistent with many other studies of capitalstructure.

Particularly important is our finding that Japanese MNCs have significantlyless leverage than Japanese DCs. Japanese MNCs are very different from DCs onall determinants of capital structure analyzed with the exception of businessrisks.

Table 2 provides regression results for Model 1 for the full sample, only MNCsand only DCs. Significant F-statistics and adjusted R2 of approximately 40%across all three groups suggests that the model fits well statistically.

The results of model 1 indicate that business risks are negative and significantfor the full sample of firms and for MNCs but not DCs. MNCs with higherbusiness risks (as proxied by the de-geared equity standard deviation) have lessleverage. This is an interesting result given there was no significant difference inbusiness risks between MNCs and DCs as shown in Table 1. For the full sampleand the sample of MNCs the negative relation between leverage and businessrisk supports the work of Burgman (1996) and Chen et al. (1997). Firms withhigher business risks are expected to face higher expected costs of financialdistress which results in the lower leverage. Further, higher levels of risk reducethe probability that interest tax shields are going to be utilized and as a result ofthe tradeoff with increasing bankruptcy risk, leverage is lower. So even thoughbusiness risks are relatively lower for MNCs it has a significant impact on theircapital structure. This is not the case for DCs.

The results show that age is a significant variable in explaining leverage for allJapanese firms regardless of whether they are MNCs or DCs. Firms that are oldergenerally have higher leverage. This could be due to older firms having lowerdebt related agency issues or lower information asymmetries (Frank and Goyal2009).

Collateral value of assets is also positive and significantly related to leverage.This is the case for both MNCs and DCs. Lower expected costs of distress andfewer debt-related agency problems can also explain this relation. This result isconsistent with Rajan and Zingales (1995).

The free cash flow variable is not significant in either the full sample or forMNCs or DCs. Although in general we expected firms with higher free cash flowto have higher leverage we could not predict if there would be any differencebetween MNCs and DCs on this variable.

The foreign exchange risks variable has a negative coefficient across allgroups of firms; however, it is only significant at the 10% level for DCs.



A possible explanation of the insignificant impact of foreign exchange risks inexplaining capital structure of MNCs is that this risk may be hedged by thesefirms. The greater foreign exchange risks faced by DCs is significantly correlatedwith lower leverage indicating that foreign exchange risks may be increasingthe expected bankruptcy risks for DCs and reducing the reliance on debt.Possibly DCs do not have a hedging strategy in place and their leverage ismore sensitive to foreign exchange exposures through increased risk of distress.

Table 2 Capital structure determinants

Variable Full sample MNCs DCs

Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic

Constant �0.46 �4.78nnn �0.59 �6.93nnn �0.35 �2.93nnn

BUSINESS RISKS �0.01 �4.13nnn �0.01 �12.57nnn 0.01 0.97AGE 0.05 3.39nnn 0.05 2.63nnn 0.07 4.23nnn

COLLATERAL 0.57 10.55nnn 0.52 12.89nnn 0.65 8.06nnn

FREE CASH FLOWS 0.04 0.17 �0.22 �0.48 0.14 1.04FX RISKS �0.02 �0.71 �0.01 �0.25 �0.16 �1.94n

GROWTH �0.04 �6.56nnn �0.03 �5.88nnn �0.04 �4.97nnn

NDTS �2.53 �10.67nnn �3.08 �4.66nnn �1.46 �2.46nn

POLITICAL RISKS �0.00 �0.49 �0.00 �0.36 0.00 0.01PROFITABILITY �1.72 �4.77nnn �1.74 �3.92nnn �1.58 �5.52nnn

SIZE 0.04 20.78nnn 0.05 14.85nnn 0.02 4.69nnn

Adjusted R2 0.40 0.41 0.38F-statistic 82.30nnn 57.96nnn 27.94nnn

This table reports the results of the following pooled cross-sectional time series regression (Model1) on the full sample of 360 Japanese multinational corporations (MNCs) and domesticcorporations (DCs) over 10 years to 2003 as well as on the sample of domestic and multinationalcorporations separately:

LEVERAGEi;t ¼b0 þ b1BUSINESS RISKSi;t þ b2AGEi;t þ b3COLLATERALi;t


þ b7NDTSi;t þ b8POLITICAL RISKSi;t

þ b9PROFITABILITYi;t þ b10SIZEi;t þ ei;t ð1Þ

where LEVERAGEi,t is the ratio of long-term debt to long-term debt plus market value of equity,BUSINESS RISKSi,t is estimated as the de-geared standard deviation of ordinary equity, AGEi,t is thenatural logarithm of the age of the firm in years from date of incorporation, COLLATERALi,t aproxy for collateral value of assets and is estimated as earnings before interest and tax adjusted fordepreciation, amortization and taxation scaled by total assets, FREE CASH FLOWS is estimated asearnings after tax adjusted for depreciation and amortization scaled by total assets, FX RISKSi,t is aproxy for foreign exchange risks and is estimated as the ratio of foreign sales to total sales,GROWTHi,t a proxy for growth opportunities facing the firm and estimated as the market-to-bookratio of common equity, NDTSi,t a proxy for non-debt tax shields and estimated as total annualdepreciation expense scaled by total assets, POLITICAL RISKSi,t a proxy for political risk faced bythe firm and estimated as a function of the proportion of revenue from different countriesweighted by a political risk factor, PROFITABILITYi,t a proxy for profitability and is estimated asthe average net income to total sales for the past four years, and SIZEi,t a proxy for firm size and isestimated as the natural logarithm of total assets.Sample t-statistics are White (1980) adjusted for heteroskedasticity.nnn, nn and n Significant at the 1%, 5% and 10% levels, respectively.



There is evidence of economies of scale in risk management practices,particularly derivative use (Smith and Stulz 1985; Nance et al. 1993; Martinand Mauer 2004). Possibly DCs are not sufficiently large to utilize economies ofscale in derivatives and carry lower debt instead. MNCs are sufficiently large anduse derivatives (and other risk management practices) and hence capitalstructure is invariant to foreign exchange exposure.

The growth variable is negative and significant for the full sample and forMNCs as well as DCs. This result is similar to Barclay and Smith (1995), Titmanand Wessels (1988) and Rajan and Zingales (1995). As a proxy for agency coststhe negative coefficient on growth supports the expected lower leverage.

Similarly, non-debt tax shields are also significantly negative for all groups offirms. De Angelo and Masulis (1980) show that non-debt tax shields are asubstitute for the tax deduction of interest. Therefore, firms with higher non-debt tax shields gain less tax benefits from leverage resulting in the negativevalue we find in the sample of Japanese firms.

Political risk is not a significant explanatory variable for leverage in thesample of Japanese corporations, nor is it significant for leverage for MNCs orDCs. The insignificant coefficient for Japanese DCs is expected since thepolitical risk they face in Japan is not expected to impact on their capitalstructure choice. A justification for the insignificant finding of political risks as adeterminant for leverage for Japanese MNCs is that foreign subsidiaries of MNCsmay borrow from the local market. When debt is borrowed from the localmarket, expected losses due to expropriation may be partly offset by acorresponding expected reduction in local debt liability (Lee and Kwok 1988).This can result in political risks being irrelevant in capital structure decisions.

A significant negative coefficient on profitability across all three groups isidentified. This result is consistent with many other studies and supports thepecking order theory of capital structure (Myers 1984; Kogut 1985; Bartlett andGhoshal 1989; Chkir and Cosset 2001).

Firm size is a significant determinant of leverage irrespective of whether thefirm is an MNC or a DC. This result supports the tradeoff theory. Since Japanesefirms are larger in size, they can afford to raise debt with less concerns regardingbankruptcy (Rajan and Zingales 1995).

Generally, the results of Model 1 on the full sample support many of thestandard determinants of capital structure. However, political risk and free-cashflow is not significant individually or in the full sample. When the sample isdivided into MNCs and DCs differences in some of the determinants appear.For example, for Japanese DCs, business risks are not significant and foreignexchange risks are significant while for Japanese MNCs business risks aresignificant but foreign exchange risks are not significant.

To assess the impact of multinationality, the dichotomous variablefor multinationality (MULTi) was introduced into Model 1. A LR test wasperformed to assess the significance of this variable. The results indicate thatmultinationality is a significant variable in capital structure determination(LR F-stat 5 5.24; prob 5 0.020; coefficient 5 �0.03; t-stat 5 �2.28; prob-value 5



0.022). The negative sign on the coefficient indicates that MNCs have significantlylower leverage than DCs, confirming the univariate results on leverage (see Table1). Model 2 is estimated to identify the causes of the difference in leverage betweenMNCs and DCs. This model was also used to assess the impact of individualdeterminants on the leverage of MNCs.

The results of Model 2 are shown in Table 3. The adjusted R2 is 42% and theregression is highly significant (F-stat 5 43.16; prob-value 5 0.00). Three variablesexplain the significant difference in leverage between MNCs and DCs. These arebusiness risks, foreign exchange risks and firm size. The negative coefficient oninteraction variable for business risks supports the results that were shown inTable 2. Although business risks for MNCs are not significantly different from DCs(as shown in Table 1, panel A) it has a significant effect not only on the leverage ofMNCs (as shown in Table 2) but on explaining the difference in leverage betweenMNCs and DCs. The impact of the difference in business risks between MNCs andDCs can be explained as follows16:

When MULT 5 1:

LEVERAGEi;t ¼ b0 þ b1BUSINESS RISKSi;t þ b11MULTi þ b12MULTi � BUSINESS RISKSi;t

¼ b0 þ b1BUSINESS RISKSi;t þ b11 þ b12BUSINESS RISKSi;t

¼ ðb0 þ b11Þ þ ðb1 þ b12ÞBUSINESS RISKSi;t

¼ ð�0:35� 0:23Þ þ ð0:01� 0:03ÞBUSINESS RISKSi;t

¼ � 0:58� 0:02� BUSINESS RISKSi;t

:

When MULT50:

LEVERAGEi;t ¼ b0 þ b1BUSINESS RISKSi;t þ b11MULTi þ b12MULTi � BUSINESS RISKSi;t

¼ b0 þ b1BUSINESS RISKSi;t

¼ � 0:35þ 0:01 BUSINESS RISKSi;t :

Since b1 is not significantly different from zero, leverage for DCs is invariantto business risks. The above result indicates that a one-unit increase in businessrisks will have a greater significant negative effect on leverage of MNCs relativeto DCs. The results show that the tradeoff theory of capital structure helps notonly to explain the capital structure of MNCs (Table 2) but to also explain thedifference in leverage between MNCs and DCs (Table 3). This implies that as thevolatility of de-geared stock returns increase (our measure of business risks) forMNCs, the benefits (costs) of interest as a tax shield diminish (increase) leadingto lower debt, which is significantly different than for DCs. The increasedinformation asymmetry expected from increased business risks does nottranslate into lower equity finance (higher leverage) as predicted by the peckingorder theory. Although DCs have a positive coefficient on business risks,providing some support for the pecking order theory domestically, the result isinsignificant.

16 The following is only included for explanatory purposes. We assume that all the other variables

are held constant.



Table 3 Interaction effects in capital structure determinants for multinationalcorporations

Variable Coefficient t-statistics

Constant �0.35 �2.94nnn

BUSINESS RISKS 0.01 0.97AGE 0.07 4.24nnn

COLLATERAL 0.65 8.09nnn

FREE CASH FLOWS 0.14 1.04FX RISKS �0.16 �1.95n

GROWTH �0.03 �4.99nnn

NDTS �1.46 �2.46nn

POLITICAL RISKS 0.00 0.01PROFITABILITY �1.58 �5.54nnn

SIZE 0.02 4.71nnn

MULT �0.23 �1.68n

MULT�BUSINESS RISKS �0.03 �1.92nn

MULT�AGE �0.02 �1.02MULT�COLLATERAL �0.12 �1.53MULT�FREE CASH FLOWS �0.36 �1.09MULT�FX RISKS 0.16 2.05nn

MULT�GROWTH 0.00 0.54MULT�NDTS �1.61 �1.42MULT�POLITICAL RISKS �0.00 �0.24MULT�PROFITABILITY �0.16 �0.56MULT�SIZE 0.03 4.87nnn

Adjusted R2 0.42F-statistic 43.16nnn

This table reports the results of a pooled cross-sectional time series regression (Model 2) on thesample of 360 Japanese multinational corporations (MNCs) and domestic corporations (DCs) over10 years to 2003:


þ b4FREE CASH FLOWSi;t þ b5FX RISKSi;t þ b6GROWTHi;t þ b7NDTSi;t

þ b8POLITICAL RISKSi;t þ b9PROFITABILITYi;t þ b10SIZEi;t

þ b11MULTi þ b12MULTi � BUSINESS RISKSi;t þ b13MULTi � AGEi;t

þ b14MULTi � COLLATERALi;t þ b15MULTi � FREE CASH FLOWi;t

þ b16MULTi � FX RISKSi;t þ b17MULTi �GROWTHi;t

þ b18MULTi �NDTSi;t þ b19MULTi � POLITICAL RISKSi;t

þ b20MULTi � PROFITABILITYi;t þ b21MULTi � SIZEi;t þ ei;t ð2Þwhere LEVERAGEi,t is the ratio of long-term debt to long-term debt plus market value of equity,BUSINESS RISKSi,t is estimated as the de-geared standard deviation of ordinary equity, AGEi,t is thenatural logarithm of the age of the firm in years from date of incorporation, COLLATERALi,t aproxy for collateral value of assets and is estimated as earnings before interest and tax adjusted fordepreciation, amortization and taxation scaled by total assets, FREE CASH FLOWS is estimated asearnings after tax adjusted for depreciation and amortization scaled by total assets, FX RISKSi,t is aproxy for foreign exchange risks and is estimated as the ratio of foreign sales to total sales,GROWTHi,t is a proxy for growth opportunities facing the firm and estimated as the market-to-book ratio of common equity, NDTSi,t is a proxy for non-debt tax shields and estimated as totalannual depreciation expense scaled by total assets, POLITICAL RISKSi,t is a proxy for political riskfaced by the firm and estimated as a function of the proportion of revenue from differentcountries weighted by a political risk factor, PROFITABILITYi,t a proxy for profitability and isestimated as the average net income to total sales for the past four years, SIZEi,t a proxy for firmsize and is estimated as the natural logarithm of total assets, MULTi a dichotomous variable formultinationality and equals unity if the firm is a multinational and zero otherwise.Sample t-statistics are White (1980) adjusted for heteroskedasticity.nnn, nn and n Significant at the 1%, 5% and 10% levels, respectively.



Model 2 (Table 2) shows foreign exchange risks are positively significant inexplaining differences between leverage of MNCs and DCs. Although MNCshave significantly higher foreign exchange risks than DCs (as shown in Table 1,panel A), foreign exchange is invariant to MNCs capital structure, but it issignificant for DCs (as shown in Table 2). As per the interaction interpretationof this variable from Table 3, when foreign exchange risks are zero MNCs areexpected to have significantly less leverage than DCs.

When there is an increase in foreign exchange exposure, through increasedsales in foreign currency, the leverage of MNCs is relatively invariant (the slopecoefficient on foreign exchange risks is close to zero). This result supports theconclusion that MNCs are active in foreign exchange risk management, such ashedging, while DCs manage this increased risk through reducing leverage. Thisis further supported by the evidence that hedging and risk managementespecially through the use of derivatives is often only undertaken by largerfirms, due to scale economies. Multinationals are significantly larger thandomestic firms and can thereby take advantage of economies of scale in riskmanagement.

Japanese MNCs are significantly larger than Japanese DCs (as shown in Table1, panel A) and firm size is a significant explanatory variable for leverage forboth MNCs and DCs (as shown in Table 2). However, Table 3 shows that firmsize is also a significant variable in explaining the differences between leverageof MNCs and DCs. In other words, as size of the MNC increases it has anincreasingly greater positive relation with leverage relative to DCs.

V. CONCLUSION

MNCs operate in an international environment encountering risks andopportunities which are not faced by their domestic counterparts. It is thereforelikely that managers of MNCs must consider additional factors in determiningtarget capital structures. Nevertheless, the majority of the current literatureeither ignores international factors completely or implicitly assumes that theyare adequately proxied by the standard variables. Our study examines whetherthere are systematic differences in standard leverage determinants for a sampleof Japanese MNCs and DCs.

We find that on univariate basis Japanese MNCs differ significantly on mostvariables relative to Japanese DCs. These variables include leverage, age,collateral value of assets, free cash flows, foreign exchange risks, growth, non-debt tax shields, political risks, profitability and size. Business risks are notfound to be significantly different between the two groups of organizations.When modeling capital structure and the determinants of capital structure wefind that Japanese multinationals have significantly less leverage than JapaneseDCs, and that multinationality is an important aspect of leverage for Japanesefirms. We find that business risks are not significant for modeling capitalstructure of domestic firms but they are for multinationals and foreign



exchange risks are not significant for multinationals but they are for domesticfirms. Business risks are negatively related to leverage for multinationals and wedocument that significant positive leverage effects of foreign exchange risks andsize are subsumed by the negative effect of business risks to explain the lowerleverage experienced by Japanese multinationals relative to Japanese DCs. Thelack of significance of foreign exchange risks for DCs can be explained byeconomies of scale in risk management, such as derivatives. Domestic firmsseem to manage the increased foreign exchange risks through lower leveragerather than derivative use. On the other hand, the larger multinationals cantake advantage of economies of scale in risk management. Consequently,foreign exchange risks of multinationals can be managed through derivativesand other risk management operations and not reduced leverage.

Barry OliverSchool of Finance and Applied StatisticsFaculty of Economics and CommerceAustralian National UniversityCanberra, [email protected]

REFERENCES

Adam, T., and V. Goyal (2007), ‘The Investment Opportunity Set and its ProxyVariables’, SSRN working paper no: 298048.

Agrawal, A., and N. Nagarajan (1990), ‘Corporate Capital Structure, Agency Costs andOwnership Control: The Case of All-equity Firms’, Journal of Finance, 45, 1325–56.

Akhtar, S. (2005), ‘The Determinants of Capital Structure for Australian Multinationaland Domestic Corporations’, Australian Journal of Management, 30, 321–4.

Barclay, M., and C. Smith (1995), ‘The Maturity Structure of Corporate Debt’, Journal ofFinance, 50, 609–31.

Barnea, A., R. Haugen, and L. Senbet (1985), Agency Problem and Financial Contracting.Englewood Cliffs, NJ: Prentice Hall.

Bartlett, C., and S. Ghoshal (1989), Managing Across Borders: The Transnational Solution.Cambridge, MA: Harvard Business School Press.

Bartov, E., G. Bodnar, and A. Kaul (1996), ‘Exchange Rate Variability and the Riskinessof US Multinational Firms: Evidence from the breakdown of Bretton Woods’, Journalof Financial Economics, 42, 105–37.

Bradley, M., G. Jarrell, and E. Kim (1984), ‘On the Existence of an Optimal CapitalStructure: Theory and Evidence’, Journal of Finance, 39, 857–928.

Burgman, T. (1996), ‘An Empirical Examination of Multinational Corporate CapitalStructure’, Journal of International Business Studies, 27, 553–608.

Chen, C., A. Cheng, N. He, and J. Kim (1997), ‘An Investigation of the RelationshipBetween International Activities and Capital Structure’, Journal of InternationalBusiness Studies, 23, 563–640.

Chkir, I., and J.-C. Cosset (2001), ‘Diversification Strategy and Capital Structureof Multinational Corporations’, Journal of Multinational Financial Management, 11,17–37.



mailto:[email protected]

Cremers, M., and V. Nair (2005), ‘Governance Mechanisms and Equity Prices’, Journalof Finance, 60, 2859–94.

De Angelo, H., and R. Masulis (1980), ‘Leverage and Dividend Irrelevancy UnderCorporate and Personal Taxation’, Journal of Finance, 35, 453–517.

Doukas, J. A., and C. Pantzalis (2003), ‘Geographic Diversification and Agency Costs ofDebt of Multinational Firms’, Journal of Corporate Finance, 9, 59–92.

Easterbrook, F. (1984), ‘Two Agency Cost Explanations of Dividends’, AmericanEconomic Review, 74, 650–9.

Fatemi, A. (1988), ‘The Effect of International Diversification on Corporate FinancingPolicy’, Journal of Business Research, 16, 17–30.

Frank, M., and V. Goyal (2009), ‘Capital Structure Decisions: Which Factors areReliably Important’, Financial Management, 38, 1–37.

Friend, I., and H. Lang (1988), ‘An Empirical Test of the Impact of Managerial Self-interest on Corporate Capital Structure’, Journal of Finance, 43, 271–352.

Grundy, B. (ed.), (2002), Selected works of Merton H. Miller – A Celebration of Markets. VolsI and II: Finance. Chicago: University of Chicago Press.

Gul, F. (1999), ‘Growth Opportunities, Capital Structure and Dividend Policies inJapan’, Journal of Corporate Finance, 5, 141–68.

Harris, M., and A. Raviv (1991), ‘The Theory of Capital Structure’, Journal of Finance,49, 297–355.

He, J., and L. Ng (1998), ‘The Foreign Exchange Exposure of Japanese MultinationalCorporations’, Journal of Finance, 53, 733–5.

Homaifar, G., J. Zietz, and O. Benkato (1998), ‘Determinants of Capital Structure inMultinational and Domestic Corporations’, Economia Internazionale, 51, 189–210.

Howell, L. (2005), Handbook of Country and Political Risk Analysis, 3rd edn. PRS Group,New York.

Jensen, M. (1986), ‘Agency Costs of Free Cash Flow, Corporate Finance and Takeovers’,American Economic Review, 76, 323–62.

Jodice, D. (1985), Political Risk Assessment: An Annotated Biography. Westport: Green-wood Press.

Kogut, B. (1985), The Multinational Corporations in 1980’s. Cambridge, MA: MIT Press.Krainer, R. (1972), ‘The Valuation and Financing of the Multinational Firm: Reply’,

Kyklos, 25, 553–626.Lee, J., E. Shim, and R. Lee (1996), ‘Product Diversification, Multinational Diversity

and Corporate Performance. US and Japanese Multinationals’, Journal of Multi-national Financial Management, 6, 195–207.

Lee, K., and C. Kwok (1988), ‘Multinational Corporations vs. Domestic Corporations:International Environmental Factors and Determinants of Capital Structure’,Journal of International Business Studies, 19, 195–217.

Lehn, K., and A. Poulsen (1989), ‘Free Cash Flow and Stockholder Gains in GoingPrivate Transactions’, Journal of Finance, 3, 771–858.

Martin, A., and L. Mauer (2004), ‘Scale Economies in Hedging Foreign Exchange CashFlows’, Global Finance Journal, 15, 17–27.

Michaelas, N., F. Chittenden, and F. Poutziouris (1999), ‘Financial Policy and CapitalStructure Choice in U.K. SMEs: Empirical Evidence from Company Panel Data’,Small Business Economics, 12, 113–5.

Miller, M. (1988), ‘The Modigliani–Miller Propositions after 30 Years’, Journal ofEconomic Perspectives, 2, 99–120.

Miller, M., and F. Modigliani (1966), ‘Some Estimates of the Cost of Capital to theUtility Industry, 1954–57’, American Economic Review, 56, 333–91.

Mittoo, U., and Z. Zhang (2008), ‘The Capital Structure of Multinational Corporations:Canadian versus US evidence’, Journal of Corporate Finance, 14, 706–20.



Modigliani, F., and M. Miller (1958), ‘The Cost of Capital, Corporation Finance and theTheory of Investment’, American Economic Review, 53, 261–358.

Modigliani, F., and M. Miller (1963), ‘Corporate Income Taxes and the Cost of Capital:A Correction’, American Economic Review, 53, 433–76.

Myers, S. (1977), ‘Determinants of Corporate Borrowings’, Journal of FinancialEconomics, 13, 187–221.

Myers, S. (1984), ‘The Capital Structure Puzzle’, Journal of Finance, 39, 575–92.Nance, D., C. Smith, and C. W. Smithson (1993), ‘On the determinants of corporate

hedging’, Journal of Finance, 48, 267–84.Rajan, R., and L. Zingales (1995), ‘What Do We Know About Capital Structure? Some

Evidence from International Data’, Journal of Finance, 50, 1421–8.Reeb, D., C. Kwok, and H. Baek (1998), ‘Systematic Risk of the Multinational

Corporation’, Journal of International Business Studies, 29, 263–79.Shleifer, A., and R. Vishny (1997), ‘A Survey of Corporate Governance’, Journal of

Finance, 52, 737–83.Singh, M., and A. Nejadmalayeri (2004), ‘Internationalization, Capital Structure, and

Cost of Capital: Evidence from French Corporations’, Journal of MultinationalFinancial Management, 14, 153–69.

Smith, C., and R. Stulz (1985), ‘The Determinants of Firms’ Hedging Policies’, Journal ofFinancial and Quantitative Analysis, 20, 391–405.

Stulz, R. (1990), ‘Managerial Discretion and Optimal Financing Policies’, Journal ofFinancial Economics, 26, 3–27.

Titman, S., and R. Wessels (1988), ‘The Determinants of Capital Structure Choice’,Journal of Finance, 43, 1–18.

White, H. (1980), ‘A Heteroskedasticity-Consistent Covariance Matrix and a DirectTest for Heteroskedasticity’, Econometrica, 48, 817–38.

Wright, F., J. Madura, and K. Wiant (2002), ‘The Differential Effects of Agency Costs onMultinational Corporations’, Applied Financial Economics, 12, 347–406.



Determinants of Capital Structure for Japanese 2009

Documents