Although the hallmark of foreign direct investment (FDI) is that a company holds a sizeable share of equity in a business registered in another country, much of the capital employed in such a business will not be equity; it will be borrowed, either from external creditors (typically local sources) or in the form of intra-corporate loans. (1) According to the World Investment Report (UNCTAD 2013) 11% of the total FDI income over the period 2005-2011 was interest paid on debt between parent companies and subsidiaries or between subsidiaries of multinational enterprises (MNE). Non-parent debt represented about 80% of the debt of US MNE subsidiaries (Desai et al. 2004). What determines the capital structure decisions that MNE make for their FDI? In this theoretical paper, we take a transaction cost theory (TCT) perspective on FDI capital structure.
Companies make FDIs to ensure control over business activities that are most efficiently conducted abroad. Control is achieved by means of establishing an equity-based link (typically, a 10% equity share is required for the investment to qualify as an FDI) to a business entity--a foreign subsidiary--in another country.
According to transaction cost/internalization theory FDIs are made when internal governance (the firm) is considered as superior to external governance (markets and contracts) as a way of carrying out international business (IB) activities, due to (comparatively) inefficient or even missing markets (Buckley and Casson 1976; Hennart 1982; Rugman 1986; Teece 1986; Williamson 1981). An extensive empirical literature has largely corroborated the arguments that specific assets (referring to a lower value of the assets in alternative uses), knowledge (which is often tacit and hence difficult to transfer across organizational boundaries), and reputation assets (which are vulnerable to free-riding), lead companies to choose FDI over non-equity alternatives such as licensing; for an overview, see Zhao et al. (2004). (2)
Although Buckley and Casson (1976) also mentioned internalization of financial markets, the financial aspects of MNEs have received much less attention from the perspective of transaction cost and internalization theory. With an early proponent in Rugman (1980), such research has tended to focus on benefits from operating internal capital markets (e.g., Aulakh and Mudambi 2005; Nguyen and Rugman 2014). In particular, despite the increased focus on subsidiaries in IB research (Rugman et al. 2011) no study so far has considered to what extent transaction cost factors may explain subsidiary capital structure. (3) MNE subsidiary capital structure has so far been the domain of financial economists demonstrating that the benefits of internal capital markets (Gertner et al. 1994) are augmented in the international context, where MNEs use capital structure strategically to overcome the limitations of host country capital markets, to mitigate political risk, and to reduce the MNE's overall tax burden (e.g., Buettner et al. 2009; Desai et al. 2004).
However, there are good reasons to also consider subsidiary capital structure from a governance perspective. Although internalization of capital markets may reduce transaction costs compared to external markets, the fundamental factors driving transaction costs such as asset specificity, bounded rationality and uncertainty (Williamson 1975) do not disappear. Moreover, internalization can lead to new types of costs such as blunted incentives and intra-MNE rent-seeking. Hence, headquarters (HQ) can improve the functioning of internal capital markets by appropriate choices for internal governance.
Our key argument in this paper is that the use of debt in a subsidiary's capital structure, represents a partial reintroduction of market mechanisms inside the boundaries of a MNE, that can help strengthen subsidiary incentives and limit influence activities and rent-seeking. However, extensive use of debt may not be appropriate for subsidiaries characterized by specific assets and substantial external uncertainty. Our analysis builds on the transaction cost argument of Williamson (1988) that debt approximates a market mechanism, based on contractually specified repayments, providing strong incentives to project managers, but which is inappropriate when projects involve specific assets in the use of which adaptation to new circumstances is important and which do not serve well as collateral. In contrast, equity allows better monitoring and control, and thus more favorable terms of financing and better adaptation in the presence of non-redeployable specific assets which do not serve well as collateral.
In an intra-MNE setting, HQ formally have control rights to all assets, and may in principle intervene at any time to save the subsidiary's assets from liquidation. In this setting the question thus becomes under what conditions a commitment by HQ to rely on rules and enforce debt payments and liquidation is credible. We argue that each form of subsidiary financing is suitable under a specific set of circumstances in terms of the specificity of subsidiary assets.
First, using external debt to finance subsidiaries, the MNE essentially "outsources" some of the governance to external debt holders. However, if the subsidiary's assets are specific to the MNE, they have less value as collateral for external debt holders who would then offer less favorable financing terms--exactly as suggested in Williamson's (1988) original argument.
Second, the new instrument of internal debt that is available in the intra-MNE context could on its hand play a particular governance role in financing assets that are MNE-specific (precluding the use of external debt), but not subsidiary-specific (i.e., that they can be easily redeployed within the MNE) (Gertner et al. 1994).
Third, however, unlike previous studies of internal capital markets we argue that even within a company, not all assets may be easily redeployed. Hence, assets may be subsidiary-specific (Birkinshaw and Hood 1998; Rugman and Verbeke 2001). A high share of equity financing may then be preferred rather than debt to promote adaptation and avoid costly liquidation implying loss of subsidiary-specific assets. These arguments thus lead us to the main propositions that non-specific assets can be financed with external debt, MNE-specific assets can be financed with internal debt, while subsidiary-specific assets must be financed with equity.
Another contribution of the paper is to introduce and discuss the concept of system asset specificity, whereby the value of assets may depend on multilateral relationships rather than the bilateral relationships traditionally studied in TCT. System asset specificity is likely to be a characteristic of many modern MNEs with highly interdependent subsidiaries.
Our arguments are illustrated by a discussion of a particularly important form of specific assets, namely knowledge. While traditionally considered a public good with associated problems in using markets (Arrow 1962), recent theory also emphasizes the potential specificity of knowledge for instance due to complementary assets and human capital (Helfat 1994). Subsidiaries not only exploit knowledge assets from HQ as traditionally assumed, but also develop their own subsidiary-specific knowledge assets (Birkinshaw and Hood 1998). Knowledge assets are therefore an illustrative case of how subsidiary-specific assets can have implications for internal governance based on subsidiary capital structure.
While our general arguments also apply to domestic firms, the MNE provides a particularly fertile analytic context allowing us to push the arguments towards their limit (Roth and Kostova 2003). Spanning national borders, MNEs likely face higher internal intervention costs than domestic firms (Richter 2014). MNEs may also face particular challenges to redeploy assets, as some assets may be location-bound (Rugman and Verbeke 2001). Host-country regulations may restrict transfers of assets within MNEs, while some assets (e.g., local product adaptations) may have little value for other units of the MNE outside their original setting. Finally, the extreme specialization and interdependence between units characterizing many modern MNEs (Bartlett and Beamish 2014) is likely to imply substantial system asset specificity.
There is a rich, yet largely untapped, common ground between finance and IB (Agmon 2006; Bowe et al. 2010). Our paper contributes to the finance literature on internal capital markets (Kolasinski 2009; Scharfstein 1998; Scharfstein and Stein 2000) by exploring the implications of subsidiary-specific assets, a concept developed in IB (Rugman and Verbeke 2001). We also contribute to the IB literature on intra-MNE governance and subsidiaries (see e.g., Birkinshaw and Hood 1998; Bowe et al. 2010; Ghoshal and Bartlett 1990; Hennart 1991; Martinez and Jarillo 1989) and to the growing corporate governance literature within IB (Strange et al. 2009) by exploring capital structure as a new internal governance mechanism from a transaction cost perspective. Finally, our analysis also answers calls for management research with direct managerial implications (cf. Oesterle and Wolf 2011): According to our analysis subsidiary capital structure is a decision variable managers can use to improve intra-MNE governance.
Section 2 briefly reviews previous literature on capital structure decisions in MNE subsidiaries and positions our perspective in this literature. Section 3 presents Williamson's (1988) original TCT of financing, which is applied to FDI capital structure in Sect. 4, leading to a set of empirically testable propositions. In Sect. 5, we illustrate some of the arguments by considering MNE knowledge assets. Section 6 provides some remarks about empirical testing and reports some suggestive evidence. Section 7 provides a discussion and conclusion.
2 Capital Structure...