The impact of combination effects of agency cost of debt and internationalization on capital structure: in respect of international evidence

The impact of combination effects of agency cost of

debt and internationalization on capital structure: in

respect of international evidence

Master's Thesis International Financial Management

EBM022A

Student number: S3363635

Name: Yiyang Dong

Study Programme: MSc IFM

Field Key Words: Capital structure, Agency cost of debt,

Internationalization, Governance index, Domestic corruption level,

Domestic financial development

1

_{Supervisor(s)}

Abstract

This thesis examines how the combination effect of agency cost of debt and internationalization impacts the capital structure. The empirical results confirm the combination effect on capital structure is negative but weak, and results show that governance index and domestic corruption level could negatively moderate the relationship between internationalization and capital structure. Moreover, domestic financial development has been found to have negative moderating effects between internationalization and capital structure after excluding firms of United States. Specifically, the results imply that better shareholder rights protection leads to a reduction of agency costs of debt for MNCs and managers should know how to maintain debtholders’ confidence under a corruption circumstance.

1. Introduction

MNCs are operating in a complicate and highly diverse international environment and there are lots of firm-specific and non-firm-specific factors affecting MNCs’ operation. Capital structure is one of the most important factors that MNCs managers should think about. Lots of previous studies show that MNCs could have lower debt ratio compared to that of DCs, and most of these studies confirm that MNCs’ lower debt ratio is subject to high agency cost (Burgman 1996; Chen and Yu 2011; Doukas and Pantzalis 2003).

Based on the above arguments, this study will establish theory on Doukas and Pantzalis (2003)’s article in two ways. First, this study will base on the Doukas and Pantzalis (2003)’s assumption to hypothesized that internationalization could have a negative impact on the capital structure, as the main relationship in this article. Second, the firm-level and country-level interaction effects are assumed to have impact on agency cost of debt and thus affect the capital structure.

However, results of this article have found that Doukas and Pantzalis (2003)’s assumption in the main relationship has been weakly supported. Nevertheless, this article still makes contributions on confirming the moderating effects of governance score and domestic corruption level. Although the main relationship is weak, this article still contributes to extending previous articles’ theory and provides international aspects for MNCs’ operation. As such, the contributions in this article are to verify the negative moderating roles of governance score and domestic corruption index and propose some potential managerial implications for MNCs’ management. Moreover, the negative moderating effects of domestic financial development has been found after excluding firms of United States.

structure or not. Finally, relevant managerial implications on the findings will be introduced and some limitations will be explained as well. The following sections are organized as follows: literature, hypothesis, methodology, results, and conclusion.

2. Literature

Previous researchers have argued that agency cost is an important factor determining the capital structure while agency cost of equity and agency cost of debt are critical aspects in a firm’s optimal capital structure. According to the tradeoff theory, managers maintain an optimal debt-equity ratio to maximize the firm’s value by reducing costs due to market imperfections (Shyam-Sunder, 1999). Burgman (1996) analyzes several determinants of capital structure, through which agency cost could exert a significant effect on the capital structure. Shapiro (1978) selects some major determinants of a firm's capital structure and suggests that agency costs of debt have negative effects on a firm's debt ratios. Similarly, Kim and Sorensen (1986) provide empirical findings suggesting that agency costs of debt and agency costs of equity explain debt ratios. Before this study further analyzes the combined effects of agency cost of debt and internationalization on the capital structure, this study first demonstrates how agency cost of debt is formed.

2.1. Sources of agency cost of debt 2.1.1. Asset substitution effect

principles’ interests. Jensen and Meckling (1976) suggest the following arguments to demonstrate where the agency cost of debt financing comes from: There are conflicts between debtholders and shareholders since issuing debt provides shareholders an incentive to invest in risky projects. If those projects succeed, most of the returns belong to the shareholders. If those projects fail, then the debtholders suffer most of the losses. Such an incentive to invest in projects with poor returns is created by shareholders who issue the debt. The effect above is referred as the "asset substitution effect" or the agency cost of debt financing (Harris & Raviv, 1991).

Continuing with Jensen and Meckling’s (1976) arguments, debtholders must use incentives for agents to regulate their actions by spending on monitoring costs. As such, the monitoring costs ensure that the agents do not harm the debtholders’ interests or compensate for the debtholders’ interests (Jensen and Meckling, 1976).

2.1.2. Underinvestment

Therefore, the above two sources of the agency cost of debt confirm that the relation between the agency cost of debt and capital structure is negative.

2.2. Agency cost of debt and MNCs’ leverage

Previous studies have provided evidence to explain the level of agency cost of debt associated with MNCs. Lee and Kwok’s (1988) arguments suggest that MNCs have higher agency cost of debt than DCs do because of the complicated international environment. Lee and Kwok (1988) argue that the agency cost of debt arises due to the monitoring and bonding activities of MNCs in foreign countries, with Lee and Kwok (1988) listing several events that might cause such monitoring and bonding activities. Since MNCs have foreign subsidiaries and DCs don’t, different accounting standards makes it costly to prepare consolidated financial statements. Moreover, serving foreign subsidiaries leads to additional auditing costs, such as travel cost, subcontracting cost, and oversea office operation cost, which induces more cost for MNCs than for DCs. Lee and Kwok (1988) also demonstrate that international market imperfections explain why the agency costs of debt for MNCs are higher than that of DCs. Several mechanisms could reduce costly agency problems to ensure there are no barriers to international investment, while such agency cost reduction effects are more limited in MNCs than in DCs. Therefore, it can be expected that MNCs have higher agency costs of debt than DCs do. In summary, the above arguments providence that the complexity of the international environment can cause MNCs to have higher agency costs of debt than DCs do.

causes underinvestment problems. Therefore, MNCs tend to have higher agency costs of debt resulting from underinvestment problems.

Similarly, Chen et al. (2011) believe that the “asset substitution” problem above becomes more severe for MNCs due to the complicated international business environment. In such a complex international environment, monitoring the shareholders becomes harder for debtholders. As such, debtholders might assume higher risks of the “asset substitution” problem occurring. Thus, capital structure is correlated with the extent of internationalization and the agency cost of debt.

According to Burgman (1996), MNCs need to deal with international gaps and differences, which causes them to incur more monitoring costs than DC firms, making the capital structures of MNCs differ from that of DC firms.

2.3. The impact of internationalization on capital structure

The above evidence recognizes the problems with the agency cost of debt and the following arguments in this section will discuss what the possible impacts of internationalization on capital structure are and why the capital structures of MNCs can differ from those of DCs. Since MNCs implement internationalization in foreign countries with very different business environments, MNCs require higher funding to engage in such an unfamiliar business environment than DCs do. If MNCs are concerned about more debt financing, then managers must consider the balance between debt financing and equity financing. MNC managers should think of the debt financing and equity financing percentages when internationalizing (Chen & Yu, 2011). Given the importance concerning capital structure when internationalizing as it has been verified above, the following items from previous studies explain the detailed differences in the capital structure between MNCs and DCs.

2.3.1. Literatures advocate MNCs have more debt than DCs

External financing sources As argued by Doukas and Pantzalis (2003), internationalization provides MNCs with more capital resources than DCs. With these capital resources, MNCs could issue more debt financing and conduct more favorable activities than DCs. Interest tax shields are increased because the loans of MNCs’ foreign subsidiaries are exposed to high tax rates. Therefore, external financing sources can be one of the reasons MNCs have higher debt ratios than DCs.

hedge against exchange rate risk results in MNCs having higher debt ratios than DCs, as discussed by Doukas and Pantzalis (2003). Since MNCs could have greater exposure levels to foreign exchange rate risk than DCs do, MNCs are expected to use more debt financing for hedging purposes than DCs. Due to their large amount of debt captured by foreign currency, MNCs can be motivated to hedge against political risks in foreign countries, while political risks are not a concern for DCs (Doukas and Pantzalis, 2003). Therefore, hedging purposes can be one of the reasons resulting in the higher debt ratios of MNCs than DCs.

Lower bankruptcy cost Lewellen (1971) believes that internationalization leads to reduced probability of debt default. Debtholders could become more satisfied and more willing to lend more since the risk of default is reduced due to the internationalization of MNCs. Moreover, MNCs can diversify their cash flows through different foreign subsidiaries, which reduces their overall bankruptcy risk (Lee and Kwok, 1988). Hence, the increased confidence of debtholders can explain why MNCs have more debt than DCs.

2.3.2. Literatures advocate MNCs have less debt than DCs

Financial constraints Since the operations of MNCs span across different industries and geographic boundaries, the business and financial risks of MNCs are relatively lower than that of DCs. Since MNCs have a greater ability to overcome financial distress than DCs do, the effects of financial constraints are greater on DCs than on MNCs as the debt ratios of MNCs should be positively associated with internationalization (Doukas and Pantzalis, 2003; Desai et al., 2008).

their parent companies, while DCs don’t receive such funding. Hence, MNCs are assumed to depend on the internal capital market instead of the external market and have lower debt ratios than DCs do. -

Availability of capital The international operation environment of MNCs provides better availability of capital than for DCs, which maintain MNCs’ a target leverage level. DCs that do not have access to international capital must maintain an optimal debt ratio. DCs might have to issue more debt and exceed their optimal debt ratio when equity capital is not available and therefore, DCs might have higher debt than MNCs do (Lee, K.C., and Kwok, 1988).

The above literature review discusses the possible impacts of internationalization on capital structure, which can be seen to be either positive or negative. Therefore, it can be expected that the capital structures of MNCs are different from that of DCs.

3. Hypotheses

Based on the above arguments, the first hypothesis is proposed as follows and the regression will be tested with consideration of several critical control variables and country-level variables as shown in the following equations:

Hypothesis 1: The effect of internationalization on capital structure is negative

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼₇TOBINQ + ε (1) LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂DIV + 𝛼₃PROF + 𝛼₄SIZE + 𝛼₅TAN_ASSET +

𝛼₆TOBINQ + ε (2) LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼7TOBINQ + 𝛼8DOM_CREDIT + 𝛼9DOM_CREDIT + ε (3)

3.1. Moderating effects

The above literature review argues the combined effects of agency cost of debt with internationalization on capital structure and discusses the differences between the capital structures of MNCs and DCs. The following sections will introduce several important country- and firm-level factors for MNCs and investigate whether these country-level and firm-level factors can change the combined effects of agency cost of debt and internationalization on capital structure.

3.1.1. Institutional ownership

How does institutional ownership indicate the quality of governance and affect leverage level? Previous articles have demonstrated several lines of reasoning behind the answer to these questions.

Doukas and Pantzalis (2003) believe institutional ownership is the extent of external monitoring of managerial behavior and demonstrate that institutional ownership is positively associated with the effectiveness of the monitoring mechanism.

production since more transparency results in lower disagreement among analysts and lower information collecting costs (Boone and White, 2015). Borochin and Yang (2017) provide further evidence on how dedicated institutional ownership influences governance. As argued by Borochin and Yang (2017), dedicated investors who take concentrated positions and low portfolio turnover tend to invest in long-term projects and gather firm-specific information. As such, Borochin and Yang’s (2017) results demonstrate that dedicated institutional investors can reduce the misevaluation and overvaluation of firms.

Burns et al.’s (2010) empirical findings elaborate in detail on how different types of institutional ownership affect governance and leverage. Burns et al. (2010) argue that the misreporting of firms’ earnings is positively and significantly related to aggregate institutional ownership, while this relation becomes negative when the institutional ownership is concentrated. As for its relationship with leverage, the results show that leverage is positively related to the likelihood of misreporting. Assuming that good governance could reduce the likelihood of misreporting, it can be assumed that good governance could lead to lower institutional ownership and lower debt.

As discussed above, greater institutional ownership could lead to higher level of misreporting and the relationship between institutional ownership and debt is positive. Since the main relationship in this article is the relationship between internationalization and capital structure, consequently, it is a question of whether the effects of institutional ownership can affect the relationship between internationalization and the capital structure of MNCs. Therefore, this paper has formulated the following hypothesis:

Hypothesis 2: Relationship between the level of internationalization and leverage level will be positively affected by institutional ownership

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼7TOBINQ + 𝛼8INSTIT + 𝛼9INSTIT ∗ FOREIGN_S + ε (5)

3.1.2. Firm-level governance index

Previous articles have clearly elaborated on the governance index and show the relation between corporate governance and capital structure.

As noted by Chen et al. (2009), firm-level corporate governance can reduce agency cost. Thus, firms with sound corporate governance have higher firm values. As referenced by Chhaochharia and Laeven (2009), they elaborate on firm-level governance as the extent to which firms adopt governance methods that are not in line with “corporate norms” in all firms. Chhaochharia and Laeven (2009) argue that sound corporate governance restricts the abilities of managers to gain personal benefits from the shareholders. When governance attributes are adopted, potential investors are informed that firms are under sound governance because firms can easily gain access to external funds. Chhaochharia and Laeven (2009) introduce the ISS (Institutional Shareholder Services) index since the ISS index captures most governance attributes. They use the ISS governance index as a proxy for firm-level governance and their findings show that firms without appropriate governance have more concentrated ownership and large cash flows, which reflects the agency theory based on the self-interest of managers and shareholders.

are restricted by the regulations, which suggests that firms with weak shareholder rights incur higher agency costs and thus more debt. Under a well protection of shareholders, managers are less likely to reap the benefits from shareholders, resulting in less agency costs. Finally, the empirical results of Jiraporn and Gleason (2007) show that there is an inverse relationship between shareholders’ right and debt.

Therefore, the role of the governance index is clear: a firm-level governance index reflects how well corporate governance can protect shareholders and a higher governance index means shareholder rights are more limited.

The governance index used by Jiraporn and Gleason (2007) show that higher governance index leads to more limited shareholder rights, while the CGV governance score in this article based on ASSET4 ESG DATA GLOSSARY is to ensure that a firm could act in the interests of its shareholders, which indicates that higher CGV governance score leads to better shareholder rights. Hence firms with higher CGV governance score are expected to have a less debt ratio.

Based on the main relationship examined in this article, it is thus a question of whether corporate governance can affect the relationship between internationalization and the capital structure of MNCs. Therefore, this paper has formulated the following hypothesis:

Hypothesis 3: Relationship between the level of internationalization and leverage level will be negatively affected by corporate governance index.

The regression will be tested as follows:

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼₇TOBINQ + 𝛼₈FOREIGN_S ∗ GOV + ε (6)

3.1.3. Corruption level in MNC’s home country

indicate the extent of the corruption in a country. Fan et al. (2012) argue that in corrupt countries, long-term debt can be issued more frequently than term debt because short-term debt is more difficult to expropriate. Hence, according to Fan et al. (2012), it can be expected that debt is used more than equity in more corrupt countries. On the other hand, Daher (2017) argues that a weak legal environment is associated with corruption as it might lead to debt covenant violations, which might weaken the debtholder’s rights.

Bae and Goyal (2009) connect the corruption level with creditor rights. As argued by Bae and Goyal (2009), the protection of creditors rights affects future activities and significantly influences whether lenders will re-contract. The re-contract activities of lenders result in a decrease of credit quality, higher interest rates, shorter loan maturity terms, and other future activities. Bae and Goyal (2009) state that the probability of re-contract is more likely to happen when creditor rights are poorly protected. To measure whether creditors are protected, Bae and Goyal (2009) use the index of ICRG (International Country Risk Guide database) to measure corruption level with their empirical results showing that countries with a high degree of corruption tend to have higher expropriation risks and contract repudiation risks.

As discussed above, previous articles have proven that corruption can lead to weak lender rights and thus lower debt. It is necessary to question whether the effects of the country’s level of corruption can affect the relationship between internationalization and capital structure for MNCs. Therefore, this paper has formulated the following hypothesis:

Hypothesis 4: Relationship between the level of internationalization and leverage level will be negatively affected by corruption level of home country

The regression will be tested as follows:

3.1.4. Emerging market based firms and developed market based firms

Previous articles provide evidence on how the extent of the domestic financial market’s development can influence the firm’s leverage level. Following Kwok and Reeb’s (2000) argument, MNCs located in less developed markets are more likely to gain access to debt that was not available in their domestic markets, while MNCs from developed markets already have access to debt in their domestic markets. Since MNCs from less developed markets gain access to more debt than in their domestic markets, they are therefore likely to issue more debt than MNCs from developed markets.

Antzoulatos et al. (2016) provide further evidence to explain how capital market development can influence the firm’s leverage level. As argued by Antzoulatos et al. (2016), informed lenders tend to monitor borrowers to prevent moral hazard of borrowers. However, such monitoring is costly, while borrowers could easily access the capital of uninformed lenders since it is cheaper. Hence, the implication is that firms will be less constrained and issue more debt when monitoring costs are reduced. Antzoulatos et al. (2016) further connects monitoring cost with capital market development. Antzoulatos et al. (2016) believe that monitoring cost decreases as the financial system develops, because financial intermediaries develop advanced methods to identify the information of potential borrowers and organize a mature monitoring system under a developed capital market.

Hypothesis 5: Relationship between the level of internationalization and leverage level will be negatively affected by domestic financial market development.

Since the domestic financial development FIN_DIV equals to the sum of MARKET_CAP and DOM_CREDIT, interaction terms of MARKET_CAP and DOM_CREDIT will be tested and compared in different models:

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼7TOBINQ + 𝛼8FIN_DIV + 𝛼9FIN_DIV ∗ FOREIGN_S + ε (8)

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼₇TOBINQ + 𝛼₈MARKET_CAP + 𝛼₈DOM_CREDIT + 𝛼₉MARKET_CAP ∗ FOREIGN_S + 𝛼₉DOM_CREDIT ∗ FOREIGN_S + ε (9) LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼7TOBINQ + 𝛼8MARKET_CAP + 𝛼9MARKET_CAP ∗ FOREIGN_S + ε (10)

LTD = 𝛼₀+ 𝛼₁FOREIGN_S + 𝛼₂GOV + 𝛼₃DIV + 𝛼₄PROF + 𝛼₅SIZE + 𝛼₆TAN_ASSET + 𝛼₇TOBINQ + 𝛼₈DOM_CREDIT + 𝛼₉DOM_CREDIT ∗ FOREIGN_S + ε (11)

4. Methodology

variables. To test the above hypotheses, this paper covers a time period of 2010 - 2017. The time frame is chosen to avoid the financial crisis, which ends at the end of 2009 in many countries. In addition, countries with a few number of observations and inactive firms without accessible data are excluded. All of the regression models include year, country and industry fixed effects. Finally, a total of 29114 observations across 42 countries from 2010 to 2017 are selected as total data sample.

5. Results

5.1. Results of summary statistics and the correlation coefficients

Table 1 Summary statistics

Variable Name N Mean Median Min Max SD

LTD 29114 0.204 0.183 0.000 0.867 0.172 FOREIGN_S 29114 0.350 0.261 0.000 1.0000 0.350 DIV 29114 0.023 0.013 0.000 0.199 0.033 PROF 29114 0.107 0.110 -0.796 0.452 0.135 TAN_ASSET 29114 0.323 0.265 0.002 0.917 0.249 INSTIT 29114 0.062 0.000 0.000 0.400 0.086 GOV 29114 0.523 0.570 0.017 0.963 0.304 CORRUPTION 29114 0.706 0.740 0.280 0.910 0.145 SIZE 29114 15.147 15.187 9.644 18.623 1.569 TOBINQ 29114 1.968 1.464 0.575 11.216 1.551 DOM_CREDIT 23739 1.463 1.590 0.273 2.079 0.447 MARKET_CAP 23121 1.129 1.140 0.232 2.632 0.487

Table 2 Correlation coefficients

LTD FOREIGN_S DIV PROF TAN_ASSET INSTIT GOV SIZE TOBINQ CORRUPTION MARKET_CAP DOM_CREDIT

LTD 1.000 FOREIGN_S -0.075* 1.000 DIV -0.081* 0.000 1.000 PROF -0.017* 0.050* 0.403* 1.000 TAN_ASSET 0.195* -0.130* -0.066* 0.00 1.000 INSTIT 0.047* -0.037* -0.068* -0.066* -0.133* 1.000 GOV 0.174* 0.132* 0.027* 0.052* 0.043* 0.227* 1.000 SIZE 0.234* 0.155* -0.095* 0.147* 0.103* -0.250* 0.103* 1.000 TOBINQ -0.129* -0.030* 0.335* 0.174* -0.212* 0.121* -0.010 -0.344* 1.000 CORRUPTION 0.023* 0.155* -0.047* -0.092* -0.021* 0.162* 0.307* -0.137* 0.00 1.000 MARKET_CAP 0.050* 0.000 0.000 -0.020 -0.054* 0.245* 0.334* -0.147* 0.133* 0.224* 1.000 DOM_CREDIT 0.059* -0.037* -0.126* -0.036* -0.054* 0.237* 0.251* 0.01 0.027* 0.476* 0.543* 1

Table 3 Coefficients of Regression Models for Main Relationship

Model 1 Model 2 Model 3 Model 4

FOREIGN_S -0.004 -0.004 -0.008* -0.006 [0.003] [0.003] [0.004] [0.011] DIV -0.135*** -0.130*** -0.098* -0.097* [0.043] [0.043] [0.057] [0.057] PROF -0.059*** -0.059*** -0.061*** -0.062*** [0.011] [0.011] [0.015] [0.015] SIZE 0.022*** 0.023*** 0.023*** 0.023*** [0.001] [0.001] [0.001] [0.001] TAN_ASSET 0.115*** 0.116*** 0.123*** 0.123*** [0.006] [0.006] [0.007] [0.007] TOBINQ -0.005*** -0.005*** -0.005*** -0.005*** [0.001] [0.001] [0.001] [0.001] GOV -0.013** -0.01 -0.01 [0.005] [0.006] [0.006] MARKET_CAP 0.002 [0.008] DOM_CREDIT 0.037** [0.019] COUNTRY_DIV -0.001 [0.004] Constant -0.245*** -0.255*** -0.221*** -0.212*** [0.033] [0.033] [0.037] -0.006 R-squared 0.269 0.27 0.276 0.276 Observations 29114 29114 19060 19060

Country fixed effect Yes Yes Yes Yes

Year fixed effect Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes

5.2. Coefficients interpretation

5.2.1 Coefficients interpretation of table 3

Model 1 in table 3 only contains the dependent variable, the independent variable as well as control variables except for GOV, and the model is controlled by country, industry and year fixed effects. The insignificant coefficient of FOREIGN_S indicates that there is no relationship between foreign sales ratio and long term debt ratio. This model, however, does not include any country-level variables and governance score as defined in table AI.

Compared to model 1, model 2 includes the control variable GOV, while the coefficient of FOREIGN_S still demonstrates insignificant. Other control variables do not change their significant level and coefficients either remain constant or change slightly. Still, it cannot make any conclusion of the main relationship yet without the consideration of country-level variables.

In Model 3, the independent variable FOREIGN_S changes from insignificant in model 2 to negative and significant model 3. In model 3, coefficient of DOM_CREDIT shows positive and significant, while coefficient of MARKET_CAP shows insignificant. Control variable GOV changes from negative and significant in model 2 to insignificant in model 3. Control variable DIV change its significant level into 10% level compared to model 2. All other control variables do not change their significant level and coefficients either remain constant or change slightly.

In Model 4, the independent variable FOREIGN_S changes from negative significant in model 3 to insignificant model 4. Coefficient of COUNTRY_DIV does not show its significant level. All control variables do not change their significant level and coefficients either remain constant or change slightly compared to model 3.

Table 4 Coefficients of Regression Models for Firm-level Factors

Model 1 Model 2 Model 3

FOREIGN_S 0.013** -0.005 0.012** [0.005] [0.004] [0.005] GOV -0.002 -0.013** -0.033*** [0.006] [0.005] [0.008] DIV -0.127*** -0.126*** -0.002 [0.043] [0.044] [0.006] PROF -0.059*** -0.058*** -0.123*** [0.011] [0.011] [0.044] SIZE 0.023*** 0.023*** -0.059*** [0.001] [0.001] [0.011] TAN_ASSET 0.115*** 0.116*** 0.023*** [0.006] [0.006] [0.001] TOBINQ -0.005*** -0.005*** 0.115*** [0.001] [0.001] [0.006] GOV * FOREIGN_S -0.031*** -0.005*** [0.008] [0.001] INSTIT * FOREIGN_S 0.021 0.043 [0.032] [0.033] INSTIT 0.023 0.016 [0.018] [0.019] Constant -0.256*** -0.260*** -0.262*** [0.033] [0.033] [0.033] R-squared 0.27 0.27 0.27

Country fixed effect 29114 29114 29114

Country fixed effect Yes Yes Yes

Year fixed effect Yes Yes Yes

Industry fixed effect Yes Yes Yes

5.2.2 Coefficients interpretation of table 4

In Model 1, the control variables except for GOV maintain the same significant level as in model 2 of table 3 and the coefficients of remaining control variables either keep constant or change slightly. The interaction term GOV*FOREIGN_S demonstrates negative and significant at 99% confident level (coefficient=-0.0031, p-value <0.01). Operating with the coefficient of interaction term GOV*FOREIGN_S and independent variable FOREIGN_S, results reveal that foreign sales of firms with a high governance index have significant and negative effect on long term debt ratio (coefficient=0.013-0.031=-0.018), and foreign sales of firms with a low governance index have significant and positive effect on long term debt ratio (coefficient=0.013). Such result supports the assumption in hypothesis 3. The managerial implications of such result will be discussed in the conclusion section.

In Model 2, the control variables except for GOV remain the same significant level as in model 1 and the coefficients of remaining control variables either keep constant or change slightly. The interaction term of institutional ownership does not show its significant level as predicted by previous studies (Burns et al. 2010; Doukas & Pantzalis 2003). As such, the interpretation with the interaction term can be interpreted as: foreign sales of firms with a high institutional ownership have insignificant effect on long term debt ratio (coefficient=0.013-0.031=-0.018), and foreign sales of firms with a low institutional ownership have insignificant effect on long term debt ratio (coefficient=0.013).

Table 5 Coefficients of Regression Models for Domestic Corruption Model 1 FOREIGN_S 0.059*** [0.013] GOV -0.012** [0.005] DIV -0.129*** [0.043] PROF -0.057*** [0.011] SIZE 0.023*** [0.001] TAN_ASSET 0.115*** [0.006] TOBINQ -0.005*** [0.001] CORRUPTION* FOREIGN_S -0.086*** [0.018] CORRUPTION 0.143*** [0.036] Constant -0.310*** [0.036] R-squared 0.27 Observations 29114

Country fixed effect Yes

Year fixed effect Yes

Industry fixed effect Yes

5.2.3 Coefficients interpretation of table 5

Again, all control variables in model 1 remain the same significant level as in model 2 of table 3 and the coefficients of all control variables either remain constant or change slightly compared to model 2 of table 3.

The positive and significant coefficient of CORRUPTION (coefficient=0.143, p-value<0.01) indicates that the results of domestic corruption level are in line with Fan et al. (2012)’s expectation that firms could issue more long term debt in a more corrupted country.

Table 6 Coefficients of Regression Models for Country-level Factors

Model 1 Model 2 Model 3 Model 4

FOREIGN_S -0.01 -0.004 -0.002 -0.003 [0.008] [0.011] [0.013] [0.012] FIN_DIV * FOREIGN_S -0.002 [0.004] FIN_DIV 0.01 [0.007] GOV -0.010* -0.011* -0.01 -0.01 [0.006] [0.006] [0.006] [0.006] DIV -0.076 -0.134*** -0.097* -0.097* [0.052] [0.049] [0.057] [0.057] PROF -0.056*** -0.070*** -0.061*** -0.062*** [0.012] [0.014] [0.015] [0.015] SIZE 0.023*** 0.022*** 0.023*** 0.023*** [0.001] [0.001] [0.001] [0.001] TAN_ASSET 0.111*** 0.126*** 0.123*** 0.123*** [0.007] [0.006] [0.007] [0.007] TOBINQ -0.006*** -0.005*** -0.005*** -0.005*** [0.001] [0.001] [0.001] [0.001] MARKET_CAP 0.007 0.002 [0.008] [0.008] DOM_CREDIT 0.014 0.039** [0.012] [0.019] MARKET_CAP * FOREIGN_S 0.005 0 [0.007] [0.007] DOM_CREDIT * FOREIGN_S -0.002 -0.004 [0.007] [0.009] Constant -0.223*** -0.251*** -0.221*** -0.214*** [0.036] [0.034] [0.037] [0.037] R-squared 0.266 0.272 0.276 0.276 Observations 23121 23739 19060 19060

Country fixed effect Yes Yes Yes Yes

Year fixed effect Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes

5.2.4 Coefficients interpretation of table 6

All control variables except for DIV in model 1 are significant and the coefficients of control variables either keep constant or change slightly compared to the control variables of model 2 in table 3.

In respect of interaction term, none of the coefficient of MARKET_CAP nor the interaction term of MARKET_CAP *FOREIGN_S demonstrate significant coefficients. Accordingly, operating with independent variable FOREIGN_S and the interaction term MARKET_CAP *FOREIGN_S, the results can be interpreted as: For firms located in a poor stock market development country, foreign sales ratio have insignificant effect on long term ratio. For firms located in a well-developed stock market country, foreign sales ratio has insignificant effect on long term ratio.

In model 2, the coefficient of foreign sales ratio still indicates insignificance. All control variables except for DIV in model 2 remain the same significant level as in model 1 and the coefficients of control variables either keep constant or change slightly. Control variable DIV changes from insignificant to negative and significant. None of the coefficient of DOM_CREDIT nor the interaction term of DOM_CREDIT*FOREIGN_S demonstrate their significant level. Accordingly, operating with independent variable FOREIGN_S and the interaction term DOM_CREDIT*FOREIGN_S, the results can be interpreted as: For firms located in a poor bond market development country, foreign sales ratio have insignificant effect on long term ratio. For firms located in a well-developed bond market country, foreign sales ratio has insignificant effect on long term ratio.

MARKET_CAP *FOREIGN_S and DOM_CREDIT *FOREIGN_S do not demonstrate their significant level. Therefore, the interpretation of these two interaction terms are the same as the interpretations in model 1 and model 2. The positive and significant coefficient (p-value<0.05) of DOM_CREDIT performs as Antzoulatos et al. (2016)’s expectation that domestic credit market promotes the use of debt.

Table 7 Coefficients of Regression Models for Models Excluding US Firms

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

[0.007] [0.007] DOM_CREDIT * FOREIGN_S 0.031*** 0.007 [0.011] [0.008] Constant -0.245*** -0.247*** -0.247*** -0.237*** -0.203*** -0.213*** -0.196*** -0.247*** [0.033] [0.033] [0.033] [0.037] [0.037] [0.037] [0.036] [0.034] R-squared 0.306 0.307 0.306 0.307 0.34 0.342 0.33 0.311 Observations 20220 20220 20220 20220 11982 11982 14227 16661

Country fixed effect Yes Yes Yes Yes Yes Yes Yes Yes

Year fixed effect Yes Yes Yes Yes Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes Yes Yes Yes Yes

5.3. Regression test excluding firms of United States

As shown in table 7, models are repeated after excluding firms from United States. Since United States firms occupy a large number of firms in the world and a majority of US firms could have higher debt ratio than non-US firms, excluding US firms could better observe the relationships of the above models. Models related to the main relationship, interaction terms of governance index, domestic corruption level do not change their significant level. Surprisingly, interaction terms of domestic financial development and domestic stock market develop become negative and significant, and domestic bond market development become positive and significant after excluding firms of US. Since US firms could have higher long term debt and occupy a large number of firms in the whole world, the results are still able to confirms the moderating effects of domestic financial development.

As such, operating with independent variable FOREIGN_S and the interaction term FIN_DIV* FOREIGN_S, the results can be interpreted as: For firms located poor financial development countries, foreign sales ratio do not have impact on long term debt ratio. For firms located in well-developed financial development countries, foreign sales ratio has negative impact on long term debt ratio. Such results are in line with hypothesis 5 when firms of United States are excluded. Therefore, hypothesis 5 is partially supported. The effect of domestic financial development in this article is similar to Chen and Yu(2011)’s empirical results. According to Chen and Yu(2011), MNCs located in poor financial developed countries have lower debt ratio than MNCs located in well-developed countries.

5.4. Results interpretation summary

is negative, as predicted by Doukas & Pantzalis (2003). Park et al. (2013)’s empirical results find that there is no relationship between foreign sales ratio and the market debt ratio. Therefore, the weak relationship in this article could also reflect the empirical results of Park et al. (2013).

Considering control variables, Doukas & Pantzalis (2003)’s expectation of dividend suggests that firms with high dividend could make higher profits and thus issue less debt, while the negative and significant coefficients of DIV act totally opposite to Doukas & Pantzalis (2003)’s findings. The coefficients of SIZE demonstrate positive and significant across all models. Such results could meet with Smith and Watts (1992)’s expectation that large firms have higher debt due to more diversified strategy and less risk, while the results totally contradict to Doukas & Pantzalis (2003)’s expectation of firm size. The negative and significant coefficients of PROF in all models accord with Chen and Yu(2011)’s conclusion that firms are not willing to borrow when firms make profits. Control variable GOV is significant in most of models. Such results are in line with Jiraporn and Gleason (2007)’s findings on corporate governance, while it can be assumed that the introduction of country financial development could lead to the insignificant coefficients. The demonstration of positive and significant coefficients of TAN_ASSET is in line with Chen and Yu(2011) and Myers (1977)’s findings that higher tangible assets result in higher debt financing when tangible assets act the role as pledges. Control variables TOBINQ are negative and significant in all models, which consistent with findings of Doukas & Pantzalis (2003) that firms with higher growth opportunities tend to issue less debt. The positive and significant coefficient of CORRUPTION copes with Fan et al. (2012)’s expectation that debt is used more than equity in more corrupted countries.

FIN_DIV* FOREIGN_S demonstrates negative and significant after excluding firms from United States Such results could be partly supported because US firms could have higher debt than other non-US firms, which could affect the effect of domestic financial development. Institutional ownership does not reveal its moderating effects due to the insignificant coefficients of INSTIT*FOREIGN_S.

Therefore, it can be concluded that hypothesis 1 is weakly supported. Hypothesis 3 and hypothesis 4 can be accepted and hypothesis 5 can be partly supported, while no evidence could support hypothesis 2.

6. Conclusion

Previous articles provide various reasons that lead to more expensive monitoring costs for MNCs and MNCs could have sever agency problems and thus MNCs could issue less debt. Based on such assumption, this article conducts several models to investigate the combination effects of internationalization and agency cost of debt on long term debt ratio, and several firm-level and country–firm-level moderating factors as extensions. This article confirms that the negative relationship between internationalization and long term debt ratio is weakly supported and verifies the moderating effects of corporate governance and domestic corruption level.

The effect of corporate governance in this article performs in line with Jiraporn and Gleason (2007)’s findings that better protections of shareholders’ rights could lead to less debt issuance. Accordingly, this article expends Jiraporn and Gleason (2007)’s findings on corporate governance and this article authenticates the moderating role of corporate governance index between internationalization and capital structure.

previous literatures concern the moderating effect of domestic corruption level, one of the contributions in this article is that this article confirms the negative moderating effects of domestic corruption level (Corruption perceptions index) on the relationship between internationalization and long term debt ratio.

The effect of domestic financial development in this article is partially in line with Chen and Yu (2012)’s empirical results that MNCs located in high financial developed countries have lower debt ratio than non-MNCs. This article extends Chen and Yu (2012)’s results and confirms the moderating effects of domestic financial development without US firms.

The appearance of weak negative effects of internationalization on long term debt ratio does not totally cope with previous studies’ assumption. One of the possible limitations is that this article does not include variables to control agency cost of debt. There is only one measure of foreign involvement in this article, while this article does not use an alternative foreign involvement. Moreover, the mixed effects of foreign sales ratio on leverage suggest that there could exist other potential factors that could influence the relationship between internationalization and long term debt ratio. Moreover, this article fails to find the moderating effects of institutional ownership. To further improve future investigation, researches on possible factors influencing agency cost of debt in MNCs should be conducted. One of the restrictions in this article is that this article does not classify institutional ownership as done by Boone and White (2015). To investigate the role of institutional ownership in the relationship between foreign sales ratio and long term debt, classification and more detailed definition of institutional ownership should be conducted.

The evidence of corruption index implies that firms located in a corrupted are highly possible to inherit and extend the corruption factors into their internationalization activities. Such actions could destroy debtholders’ confidence and increase the possibility of debt violation. Thus, when managing a frim from a highly corrupted country, mangers should focus more on controlling the corruption culture and prevent the influence of corruption from affecting firm’s debt financing. Accordingly, MNCs managers should be familiar with the investor protection law as well as culture in both home country and host country to gain the confidence of debtholders.

The findings of governance index imply that better shareholder protection induce a reduction of agency costs and come out with a lower debt level for MNCs. As first proposed by Doukas & Pantzalis (2003) in this article, the more foreign involvement for MNCs, the more severe the agency problems. Improving shareholders’ right and monitoring managers will be tools for MNCs to mitigate the agency problems. Thus, MNCs should depend on the extent of their foreign involvement and decide to what extent they should control managers’ abilities to reap shareholder’s benefits.

Bibliography

Antzoulatos, A.A., Koufopoulos, K., Lambrinoudakis, C., Tsiritakis, E., 2016. Supply of capital and capital structure: The role of financial development. Journal of Corporate Finance 38: 166–195.

Anderson, R.C., Mansib, S.A., Reeb, D.M., 2003. Founding family ownership and the agency cost of debt. Journal of Financial Economics, 68: 263- 285.

Anwar, S., & Sun, S., 2015. Can the presence of foreign investment affect the capital structure of domestic firms? Journal of Financial Economics, 30: 32-43.

Bae, K., and Goyal, V.K., 2009. Creditor Rights, Enforcement, and Bank Loans. Journal of finance, 64(2): 823-857.

Boone, A.L., and White, J.T., 2015. The effect of institutional ownership on firm

transparency and information production. Journal of Financial Economic, 117:508-533.

Borochin, P., and Yang, J., 2017. The effects of institutional investor objectives on firm valuation and governance. Journal of Financial Economics, 126: 171-199.

Burgman, T.A., 1996. An Empirical Examination of Multinational Corporate Capital Structure. Journal of International Business Studies, 27(3): 553- 570.

Chen, K.C.W., Chen, Z., Wei, K.C.J., 2009. Legal Protection of Investors, Corporate Governance, and the Cost of Equity Capital. Journal of Corporate Finance, 15:273-289.

Chen, C.J., Yu, C.J., 2011. FDI, Export, and Capital Structure: An Agency Theory Perspective. Management International Review, 51(3):295-320

Chhaochharia, V., and Laeven, L., 2009. Corporate governance norms and practices. Journal of Financial Intermediation, 18:405-431.

Daher, M., 2017. Creditor control rights, capital structure, and legal enforcement. Journal of Corporate Finance, 44: 308-330.

Desai, M.A., Foley, C.F., and K.J. Forbes, 2008. Financial Constraints and Growth: Multinational and Local Firm Responses to Currency Depreciations, Review of Financial Studies, 21: 2857–2888.

Desaia, M.A., Foleya,C.F., Hines Jr, J.R., 2008. Capital structure with risky foreign investment. Journal of Financial Economics, 88: 534–553.

Doukas, J.A., & Pantzalis, C., 2003. Geographic diversification and agency costs of debt of multinational firms. Journal of Corporate Finance, 9: 59–92.

Harris, M., & Raviv, A., 1991. The theory of Capital Structure. Journal of finance, 46(1):297-355.

Jensen, M.C., & Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4):305-360.

Jiraporn , P., and Gleason K.C., 2007. Capital Structure, Shareholder rights, and Corporate Governance. The Journal of Financial Research, 1:21-33.

Kim, W.S., & Sorensen, E.H., 1986. Evidence on the Impact of the Agency Costs of Debt on Corporate Debt Policy. The Journal of Financial and Quantitative Analysis, 21(2): 131-144.

Kwok, C.C.Y., & Reeb, D.M., 2000. Internationalization and Firm Risk: An Upstream-Downstream Hypothesis. Journal of International Business Studies, 31(4): 611- 629.

Lee, K.C., & Kwok, C.C.Y., 1988. Multinational Corporations vs. Domestic Corporations: International Environmental Factors and Determinants of Capital Structure. Journal of International Business Studies, 19(2):195- 217.

Lewellen, W.G., 1971. A pure financial rationale for the conglomerate merger. Journal of Finance, 26:521-537.

Oesterle, M., Richta, H.N., Fisch, J.H., 2013. The influence of ownership structure on internationalization. International Business Review, 22: 187–201.

Park, S.H., J. Suh, and B. Yeung., 2013. Do Multinational and Domestic Corporations Differ in their Leverage Policies? Journal of Corporate Finance, 20: 115-139.

Shapiro, A.C., 1978. Financial Structure and Cost of Capital in the Multinational Corporation. The Journal of Financial and Quantitative Analysis, 13(2): 211-226.

Shyam-Sunder, L., 1999. Testing static tradeoff against pecking order models of capital structure. Journal of Financial Economics, 51(2):219-244.

Smith, C. Jr., & Warner, J.B.,1979. On Financial Contracting An Analysis of Bond Covenants. Journal of Financial Economics, 7: 117- 161.

Appendix A

Table AI Variables definition and source

Variables Definition and description Article source and/or data source

LTD LTD is the dependent variable, which is defined as long-term debt ratio. LTD is calculated as the ratio of long term debt over total asset.

Doukas & Pantzalis (2003): long term debt ratios are proxies to measure leverage level.

Data source: Datastream FOREIGN_S Independent variable, is defined as the degree of internationalization. FOREIGN_S

is calculated as firm’s foreign sales to total sales ratio. Chen and Yu(2011), Kwok and Reeb (2000): foreign sales ratio can be used to measure the degree of internationalization and the relationship between leverage and internationalization.

Data source: Datastream

SIZE Firm size is calculated as the log of book value of asset (currency of US dollar). Doukas & Pantzalis (2003): Large firms are assumed to have a larger internal capital market and thus larger firms are expected to have less debt, while Smith and Watts (1992) find that larger firm could have higher debt because large firms are less affected by financial risk and more diversified than smaller firms. Data source: Datastream

DIV DIV is referred as dividend payout ratio, which is calculated as the total cash dividend payment.

Doukas & Pantzalis (2003): higher dividend payout ratios might suggest that firms could make higher profits and tend to borrow more debt.

Data source: Datastream

PROF Profitability is calculated as the ratio of EBITA over total asset. Myers(1984), Chen and Yu(2011): When firms’ internal profits are available, firms are not willing to borrow external debt. It is unnecessary for profitable firms to borrow.

Data source: Datastream TOBINQ Tobin’s q is referred as the proxy for growth opportunity, and tobin’s q can be

calculated as (total asset - common equity + market value of company)/total asset.

As found by Doukas & Pantzalis (2003), value gains will belong to debtholders when there is risky debt. As such, firms with more growth opportunities could be more easily affected by agency cost of debt. Thus, firms with greater growth opportunity should have less debt.

Data source: Datastream

TAN_ASSET Tangible asset is calculated as the ratio of fixed asset (PP&E)over total asset. Chen and Yu(2011), Myers (1977): higher tangible assets lead to higher debt financing when tangible assets act the role as pledges. Data source: Datastream

SIC 4 digits SIC codes to identify industries. Data source: Datastream

ISO 2 digits country codes to identify countries. Data source: Datastream

Year Year variable to identify years.

INSTIT Institutional ownership of a company. Doukas & Pantzalis (2003)

Data source: Datastream GOV The CGV score is referred as corporate governance Index to measure the

effectiveness of corporate governance. According to the definition in Datastream(ASSET4 ESG DATA GLOSSARY), the CGV score is to ensure that firm’s board members and executives could act in the interests of its long term shareholders.

Since the importance of corporate governance is necessary for every firm and such importance has been confirmed from above literature review, the variable GOV will be included in every regression model.

Data source: Datastream

FIN_DIV FIN_DIV is referred as Domestic financial development, which is equal to sum of Domestic credit to private sector + Market capitalization of listed domestic companies.

Antzoulatos et al. (2016),

Data source: World Bank’s Financial Structure Database DOM_CREDIT DOM_CREDIT is referred as Domestic credit to private sector (% of GDP), which

is a measure of domestic bond market development.

As defined by World Bank’s Financial Structure Database, DOM_CREDIT is the financial resources which are received by private sector from financial institutions. According to World Bank’s Financial Structure Database, full definition of DOM_CREDIT is stated as follows: “Domestic credit to private sector refers to financial resources provided to the private sector by financial corporations, such as through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment. For some countries these claims include credit to public enterprises. The financial corporations include monetary authorities and deposit money banks, as well as other financial corporations where data are available (including corporations that do not accept transferable deposits but do incur such liabilities as time and savings deposits).”

Data source: World Bank’s Financial Structure Database

MARKET_CAP MARKET_CAP is referred as Market capitalization of listed domestic companies (% of GDP), which is a measure of domestic stock market development.

According to the definition of World Bank’s Financial Structure Database, market capitalization contains share price times the number of shares outstanding of domestic listed companies. The market capitalization does not contain: investment funds, unit trusts, and companies only with a goal to hold shares of other listed companies.

Data source: World Bank’s Financial Structure Database

CORRUPTION Corruption perceptions index, Corruption index of home country.

According to Transparency International, the index measuring corruption level across 180 countries and territories and the measurement is based on the business and public areas. The corruption index, referred as CPI (Corruption Perceptions

Fan et al. (2012)

Table AII Mean value of main variables clustered by countries Country No. of Obs Mean of LTD

Mean of

FOREIGN_S Mean of INSTIT Mean of GOV

Switzerland 372 0.155 0.745 0.019 0.508 0.861 2.093 1.669

Taiwan 905 0.107 0.513 0.007 0.152 0.611 . .

Turkey 142 0.194 0.161 0.001 0.296 0.440 0.272 0.604

United Arab Emir 27 0.181 0.406 0.000 0.290 0.686 0.531 0.757

United Kingdom 2103 0.198 0.489 0.087 0.781 0.783 . 1.508

United States 8894 0.260 0.289 0.108 0.685 0.738 1.390 1.872

Total 29114 0.204 0.350 0.062 0.523 0.706 1.129 1.463