The Effect of Firm and Country Level Corporate Governance and Multinationalism on U.S. Multinational Enterprise Performance
Master Thesis
Reinier Volgers S2188759
r.l.a.volgers@student.rug.nl
University of Groningen Faculty of Economics and Business
Groningen
Supervisor: Prof. dr. H. van Ees Co-Assessor: Prof. dr. T. Kohl
Final Version: 19-06-2017
Abstract: This study investigates the effect of corporate governance characteristics on the performance of U.S. multinational enterprises (MNE) from the S&P 500. First, the effect of the location of the MNEs subsidiaries (CLCG) is researched. Second and third the moderating effect of the MNEs own level of corporate governance (FLCG) and the MNEs level of
multinationalism on the earlier researched effect are taken into account. The results suggest that the lower the score of CLCG the better is the performance of the MNE. The second and third moderating effect of FLCG and level of multinationalism on CLCG are both not
significant and not as suggested. Further research should focus more on the ambiguous results about FLCG evident in the literature and on qualitative research concerning
managerial aspects on the effect of corporate governance and the location of subsidiaries.
Table of Contents
1) Introduction ... 4
2) Theory and hypotheses development ... 8
Multinationalism: ... 16
3) Sample, Variables and Method ... 20
Sample ... 20
Variables ... 20
Method ... 23
4) Results ... 25
5) Discussion and Conclusions ... 28
6) Limitations and further research ... 33
7) References ... 35
8) Appendices ... 40
1) Introduction
The influence of corporate governance on firm performance is extensively researched (Bhagat and Bolton (2008), Klapper and Love (2004), Bai, Liu, Lu, Song and Zhang (2003)).
Corporate governance refers to the relations, mechanisms and processes of by which the firm is controlled and directed (Shailer, 2004). In addition, Shleifer and Vishny (1997) describe corporate governance as the mechanism that insures investors a fair return on their investment. Governance systems are different across firms and countries and potential investors consider the corporate governance of a firm before investing. The governance of a firm (FLCG) is also determined by country governance characteristics (CLCG), so the location of a firm may matter. (Doidge, Karolyi and Stulz, 2007)
After a series of corporate scandals in 2002, the Sarbanes-Oxley Act (SOX) passed the U.S.
Congress to strengthen corporate governance. The SOX was set in place to restore investor confidence in U.S. corporations and is the most important piece of corporate legislation in recent U.S. history. President Bush stated ‘The era of low standards and false profits is over.
No boardroom in America is above or beyond the law’ after the enactment of the SOX (Lucci, 2003). As studies have shown, the benefits of the improved corporate governance after fraudulent behaviour is significant (Farber, 2005), and the success of the SOX is not only of influence in the U.S. The SOX act is of importance in all international markets where the U.S.
corporations invest (Ribstein, 2003), and for cross-listed non- US companies in the U.S.
(Litvak, 2007). As the SOX act can be seen a benchmark for countries to construct a governance framework, success must be generated before countries experience the benefits of international corporate standards and until then little incentive arises to abandon national regulatory frameworks. But the SOX is also seen as controversial. (Li, 2014) Critics state that the SOX reforms have been not well-considered, as U.S. congress rushed the implementation, whilst reforms in the U.K. and the rest of Europe have been well-considered. (Lucci, 2003). However, after the success of the SOX act became clear, many countries (e.g. the U.K., South Korea, Australia and South Africa) used key provisions of the SOX as a model to implement an oversight model, which is comparable to the SOX (Lucci, 2003). This leads to increased corporate governance in the foreign country and a higher country score concerning CLCG.
For U.S. firms, multiple changes occurred after the implementation of the SOX, such as changing national accounting standards, board members could be individually sentenced for fraud and the creation and independence of audit committees.(Linck, Netter and Yang, 2008) Some firms, such as General Electric, went beyond the required SOX provisions by setting more stringent internal control standards. (Lucci, 2003). The implementation of the SOX in the U.S. further implied several aspects for firms of other countries. First of all, firms who trade on an American exchange were left in uncertainty of which provisions applied to
them. Moreover, many cross listed firms and firms whose securities trade in the U.S. are affected by the SOX. The rules enacted by the SOX are duplicative of regulations in their respective countries and found too strict by the firms who are affected.
Another aspect commonly considered with corporate governance is sustainability, as a mean to improve the firm’s transparency and accountability. Originally sustainability focused on environmental issues, nowadays social issues for employees and community matter, as well as the organizational structure that is in place to control risk management and governance. (Kolk, 2008) These developments cannot be seen apart from financial crises and scandals, as is the same for the SOX. As of today, firm performance is dependent on measures such as sustainability, as for example Unilever has launched the Unilever Sustainable Living Plan, which aims at creating a sustainable world1. The SOX
implementation led towards improvement of shareholder insights in and influence on corporate behavior on the whole range of business matters for firms operating in the U.S.
Internal mechanisms that were targeted, next to the well-known aspects regarding boards, managers and auditors, are ethical aspects related to employee behavior, complaint mechanisms and remuneration. (Kolk, 2008) This shows corporate governance covers a substantial part of the internal framework of the firm, including social aspects. Also, companies nowadays pay attention to board supervision, codes of ethics and auditor
involvement in relation to sustainability. The subject of corporate governance is intertwined with multiple aspects within the modern firm. Therefore, reasons can be found that
corporate governance has an increased influence on the performance of the firm. With sustainable and moral issues becoming more and more of influence on the behavior of the firm and higher or ‘better’ corporate governance will lead to higher firm performance, this is an interesting subject to research.
The relevance of corporate governance is of great influence nowadays. After the scandal of Enron at the beginning of this century, the role of corporate governance was highlighted in the academic literature. After the more recent financial crises starting in 2008, the literature focused more on the role of corporate governance on bank failures, such as Berger,
Imbierowicz and Rauch (2016), Haan and Vlahu (2016) and Hopt (2013). However, these studies do have influence on other corporate enterprises then financial enterprises, as the entire mechanism of corporate governance is on debate. It is difficult to prove that
corporate governance is more relevant as of today than before, but with evolving business mechanisms, the corporate governance literature will stay of interest, for example on block chains (Yermack, 2017) and international ownership structures (Kumar and Zattoni, 2017)
As the world becomes more globalized, so does the complexity of the international firm.
Studies, such as performed by Sanders and Carpenter (1998), argue that this increased
1 www.unilever.com
complexity is accompanied with a higher degree of corporate governance within the firm (e.g. FLCG). This argument of Sanders and Carpenter is a crucial element in the development one of the three hypotheses in the following of this article. Moreover, better governed firms are found to be more profitable, higher valued and pay out more cash to shareholders.
(Brown and Caylor, 2004) Interesting to research therefore is the effect of the level of internationalization of the MNE, labeled as multinationalism in the remainder of this study, on the overall performance of the MNE.
This thesis tries to address the individual influence of CLCG on MNE performance, and the combined effect of CLCG with FLCG and multinationalism (CLCG-Mult), respectively on MNE performance. These kind of indicators become more and more important for firms as the general public becomes more aware of negative activities conducted by firms and potential financial fraudulent activities and crises started by firms. As will become more clear below, opposing results are published by authors, considering the similar studies performed as this one. This study further aims to contribute to the literature by obtaining results that indicate that the more multinational firms are more able to deal with different and changing
governance legislation around the globe. Furthermore, the arguments considering CLCG will be contradictory towards the general belief of the literature about whether or not higher CLCG leads to better performances.
CLCG-multinationalism is an interaction variable, that tries to capture the benefits of a larger MNE, in number of subsidiaries, and the experience the MNE obtains from this multinationalism level and reflected in the MNEs FLCG. A direct influence of FLCG is not included in the last research question, but is believed to be incorporated in the level of multinationalism. This is described in more detail in section 2. Governance researchers, such as Aguilera, Filatotchev, Gospel and Jackson (2008) and Bruna and Claessens (2007) already recognize the need for a combination of CLCG and FLCG and CLCG in combination with multinationalism. Aguilera and Cuervo-Cazzura (2009) state that the voluntarily codes of governance are relatively young, and international corporate governance is not well researched. International corporate governance can be seen as the international codes of governance that exists, which can guide firms in setting their framework. A twofold of literature exists: one focusing on a particular codes influence on firms in the same country the other on the existence and subject of codes in multiple countries. An example of converging governance codes is International Corporate Governance Network (ICGN), who encourage uniform global equitable shareholder voting procedures. (Porter and Kramer, 2002) The study by Bell et al. (2014) investigates home country legitimation on IPO firms to the U.S., and is treated as a one of the building blocks in order to develop the arguments in this study. This thesis investigates the effect on firm performance of U.S. firms while benefitting from their home country U.S. institutional characteristics in host country
environments. As Filatotchev and Wright (2011) argue, a minimum amount of literature has been devoted to governance across national contexts.
As intangible assets, which corporate governance can be described as of being part of, become more and more important, firms can internationalize for other reasons than increased competitiveness, market penetration, production possibilities, etcetera. Dyer (1996) researched the U.S. and Japanese automotive market and concluded that the Japanese had a competitive advantage due to their corporate governance. His findings support transaction cost theories, which suggest aligning governance structures with the firms’ transactions and will result in efficiency advantages. These lower transaction costs than competitors is, again, confirmed by Dyer and Singh (1998). Moreover, these authors state that ‘’effective governance mechanisms can permit the realization of rents through the synergetic combination of assets, knowledge, or capabilities.’’ Aguilera, Filatotchev, Gospel and Jackson (2008) suggest other concepts which will benefit the firm and their
environment: cost, contingencies and complementarities. Costs refer to the cost-benefit analysis for firms who operate in different environments, which this studies addresses as well. Contingencies refers to the FLCG that interrelate with the firms’ external and internal strategic resources and its connection with market, regulatory, sectoral and institutional environment. The example given by Aguilera et al. (2008) is that older firms focus on their deep pool of external stakeholders whether younger firms focus on the internal
development of codes of governance. Different kind of firms can therefore look for different value added activities. The concept of complementarities refer to the existing practices of governance rules within the firm, and their bundled capacity to enhance the best practices to achieve effective corporate governance. These three concepts suggested by Aguilera et al. (2008) help to achieve effective governance, and create relational benefits with the environment.
Multiple other authors address the issues of corporate governance, such as Bauer, Guenster and Otten (2004). These authors state that better corporate governance leads to efficiency gains. They emphasize that these efficiency gains from operations lead to higher expected future cash flows and a higher firm value. From this perspective, corporate governance can be a competitive advantage, especially in an environment where the governance codes do not require firms to adopt certain practices, but firms can voluntarily adopt these. This is an interesting subject for me, as little evidence exist for the this relation between corporate governance and MNE performance. In order to research this subject, data is collected from multiple sources, such as the Worldbank, Wharton Research Data Services and Orbis. This is needed in order to construct a valid and representing sample of large U.S. MNEs. The U.S. is chosen because the country has large MNEs, a considerable amount of data availability, a large number of academic material dedicated towards corporate governance and
ambiguous conclusion are drawn considering the effect of corporate governance on the performance of the MNE. In order to write a relevant and updated research the time period researched is that of 2012 until 2015.
The outline of the thesis is as follows: in section two the related theory of the subject will be discussed and analyzed. This will be the framework for the development for the hypotheses, and these hypotheses are stated in chapter two as well. Section three includes the methods and variables used in order to analyze the suggested hypotheses. The successive section displays the results of the performed analysis and the outcomes of these analyses are described. Section 5 will discuss the outcomes of preceding section and conclusions will be drawn from those results. In section 6 limitations of this particular study are described and future research subjects are proposed. At the end of this article the used references are displayed and an appendix is provided.
2) Theory and hypotheses development
Gompers, Ishii and Metrick (2003) argue that firms with weaker shareholder rights earn significantly lower returns, have a lower firm value, poorer operating performance and are more active in capital expenditure and takeover activities. Contradictory results are
obtained by and Bauer et al. (2004). These authors perform a similar study as Gompers et al.
(2003) but retrieve opposing results considering corporate governance and firm
performance. The difference between the two studies is Gompers et al. (2003) collect data from U.S. firms and Bauer et al. (2004) collect data from the U.K. and Europe. Therefore, the results retrieved by Gompers et al. (2003) suit better in this study and therefore an increase in FLCG would suspect an increase in firm performance. However, both the studies indicate that ambiguous evidence for corporate governance is obtained and supported and
therefore is interesting to research. Cremers and Ferrel (2014) confirm this idea developed by Gompers et al. (2003). They investigate FLCG by measuring shareholder rights by a G- index; the higher the index means more restrictions at the firm level on shareholders rights or a greater number of anti-takeover measures. The article concludes that an increase in shareholders rights (less restrictions, i.e. lower G-index) leads to higher firm value, but only after the year 1985, when the Household shock (which increased the importance of
shareholder rights) was enacted. Core, Holthausen and Larcker (1999) investigate this matter and conclude that indeed this line of reasoning is the case. Firms with weak FLCG have greater agency problems, CEOs at these firms receive higher compensation and firms with larger agency problems perform worse. This leads to my argumentation that indeed FLCG has a positive impact on the performance of the MNE. However, this is not a direct research question, as the article is focusing on the location of the MNEs subsidiaries.
Together with the location of subsidiaries, interesting is whether the number of subsidiaries is of influence on the performance of the MNE and if multinationalism is of influence at all.
Aybar and Ficici (2009) analysed the effect of international acquisitions on firm value, which indicate whether larger firms are more valued by investors. The findings of this study
indicate that cross border expansions of emerging-market multinationals did not add value to the firm, rather led to value destruction of this firm. Also, Morck and Yeung (1991) find that investors do not value MNEs cross-border mergers or acquisitions as a means of diversifying their portfolio, but they do support the view of the internationalization theory;
intangible assets are necessary in order to rectify foreign direct investment. This is in particular interest of this study, as corporate governance is seen as an intangible asset.
Expanding MNEs are not always valued by investors, as described in the previous paragraph.
On this subject, Garrod and Rees (1998) discuss the differences in valuation of domestic and multinational firms, using a valuation method including net assets and profit before taxes, combined with an estimation of investments of MNCs in certain areas of the world. The authors conclude that the application of their basic valuations method show significant support that multinational firms have a higher capitalization of earnings then their domestic counterparts. Moreover, the differences in value lies not solely in the MNC’s foreign
operations, but in all their operations. The MNCs can use arbitrage opportunities for restrictions imposed by governments and prosper from economies of scale. This indicates that not every scholar has the same view on the relationship between multinationalism and firm value, which creates opportunities for further research. The interesting parts to me, however, are not the productivity improving factors of internationalization, but rather the institutional gains of going abroad. It is well known that, e.g. Nike locates their factories in cheap labour countries to minimize the costs, but do firms look at the level of governance in the country? Or differently asked: do firms benefit from codes of governance abroad? As will become clearer, MNEs might benefit from their FLCG in countries with low CLCG, to take advantages from this created gap between FLCG and CLCG. As Filatotchev and Wright (2011) argue, MNEs who internationalize towards countries with low CLCG are associated with higher information asymmetry. MNEs with high FLCG then in turn deal with this issue by employing the high FLCG within the firm, and make use of the opportunities. Together with the growth opportunities present in the low CLCG countries, better performances are achieved in low CLCG countries.
Firms which intend to seek opportunities to increase equity financing have the opportunity to issue an initial public offering (IPO). Hence, firms who seek money abroad can issue a foreign IPO. The article of Bell, Filatotchev and Aguilera (2014) discus those foreign IPOs by firms who not only seek equity financing, but also try to achieve marketing, political and employee relational objectives. Those firms try to comply with the host (mostly U.S.) country’s corporate governance expectations, but the home country’s institutional
environment decreases those foreign firms’ valuation by U.S. investors. This indicates firms look at expanding abroad to profit from institutional characteristics, such as CLCG. Those firms, from mostly developing countries, list their firm as an U.S. company to benefit from the institutional advantages the U.S. has to offer and the firm is valued higher. Another finding of Bell et al. (2014) indicates that firms from developing countries are valued on the
strength of the firms’ home country regulative and governance related institutions. Why should this not be the case for firms from developed countries? As firms are valued higher because of their domestic institutions, this can be the case for both developing and
developed home based firms. And as the developing country home based firms try to gain from the institutions in the U.S., U.S. firms can gain from their strong home based
institutions and employ those advantages in a foreign environment. For this study, it is interesting to look beyond the value of the MNE and see whether operating performance in the entire MNE is affected by the locations of the MNEs subsidiaries. The performance in those foreign locations should be positively affected by the quality of FLCG embedded in the MNE.
As firm value is not researched in this article, it can be applied as a guiding principle why firms and potential investors care about corporate governance. Investors care about shareholder value maximization as the guiding principle for top managers considering strategic decisions. Other primary features of importance to shareholders are governmental organizations, legislation and court decisions (Scott, 1998) which all can be seen as
institutions or CLCG. These components of the nation’s institutional framework are there to hold managers accountable to shareholders, ensure shareholder voting rights, prevent self- enhancement by managers and rule enforcement. Bell et al. (2014) argue that U.S. investors might suspects that managers from firms in countries with weak regulatory frameworks for this matter are, for example, more difficult to be hold accountable for their actions and are less eager on investing in this company. By itself this might increase the opportunities for acting on self-interest. The other way around: foreign investors might suspect that top executive of U.S. may lay higher value on the shareholder value maximization of the firm in question as, on average U.S. firms have higher FLCG scores than their foreign matching counterparts (Aggarwal, Erel, Stulz and Williamson, 2007). The firms that do have higher governance scores than their U.S. counterparts are mostly from the U.K and Canada, which are also highly developed countries with high CLCG scores. Moreover, Aggarwal et al.
(2007) show that with a larger governance gap, defined as a larger difference between the quality of the foreign MNE’s governance and the governance of the comparable U.S. firm, the value of the foreign firms increases with the governance gap. This, indicates that firms are rewarded for better FLCG practices. Lastly, not every aspect of governance is valued at the same level; board and audit committee independence are valued high, separation of chairman and board low. What is also interesting on this matter, is the link between CLCG and FLCG. Firms from high CLCG environments are firms who employ a high level of FLCG quality. This because the high CLCG environment leads towards a strict set of rules and regulations which the firms must comply to, leading towards higher quality FLCG.
In this line of thinking, Klapper and Love (2004) find that higher FLCG is correlated with better operating performance. Evidence is provided that FLCG features are more important in countries with weak legal environments, which indicates that strong corporate
governance within the firm in an environment which do not demand for it, leads towards better performances. However, their results are valid for emerging markets. The authors do state, that many firms have the options to not fully comply with country set regulations, because firms have flexibility to opt-out and decline some provisions or adopt extra provisions not set in the countries’ legal codes. This way, firms can differentiate from each other, with a well-known example of various degrees of investor protection amongst firms.
These findings are of large interest for this research. Firms cannot completely recreate the country level governance themselves, but firms that make an effort are performing better than firms that do not. Especially investor protection and minority shareholders rights can be addressed by the firm. (Shleifer and Wolfensohn, 2002). As Aguilar and Cuervo-Cazurra (2009) discuss; the corporate governance codes’ voluntary nature is under criticism, but despite this criticism good codes of governance appear to lead towards ‘better’ governance codes in countries. Two different approaches are used of implementing codes; mandatory (SOX legislation/U.S.) and voluntarily (comply or explain approach/ U.K. combined code of 2003). The mandatory approach is rarely used in codes of good governance as it more associated with law. The U.S. firm sample in this study is legally required to implement the rules in their home country, and when internationalizing they can voluntarily adopt codes.
This way, firms can employ governance frameworks which suits the firms’ interest and structure the best.
As the study is about subsidiaries, these subsidiaries are not legally required to adopt to the U.S. legislations, but best practices are commonly taken abroad and transferred from the parent company to the subsidiary. Subsidiaries with independent boards simultaneously deal with answering to shareholders and stakeholders, while communicating and integrating with the parent firm. Moreover, as Luo (2005) describes, subsidiaries with their own board of directors make adaptations to the host country legal requirements or the parents
company strategic considerations for establishing such boards. These boards of subsidiaries actively channel back to the parent level for governance sharing. Filatotchev and Wright’s (2011) article states the parent firm has substantial influences on a subsidiary’s operations, decision-making process and strategy within the MNE’s global network.
As is described by Gompers et al. (2003) and Cremer and Ferrel (2014) FLCG quality leads towards better firm performances in developed countries. For developing countries such influences might be different. Research from Korea (Black, Jang and Kim, 2006) show that, for emerging countries, greater board independence correlates with higher share prices.
Black (2001) finds similar results for the Russian economy. Cheung, Connelly, Jiang and Limpaphayom (2011) report that firms who exhibit improvements in the quality of FLCG display subsequent higher levels in future value. The results indicate that good corporate governance level is (or will be) correlated with a higher the market valuation of the firm in developing countries as well. For the effect of a country’s legislation on firm value,
Chhaochharia’s and Grinstein’s (2007) study investigates the effect of the SOX on firm value.
The results show that large firms who are less compliant with governance rules have
positive economic significant abnormal returns compared to large firms who are compliant.
These results might be odd at first thought, but such new rules are costly to implement and therefore economically significant to the firm. The study does implicate that the SOX does not benefit the value of the firm, and a country’s CLCG legislation might not be beneficial for MNEs in general.
As a country’s institutional environment might not be beneficial for MNEs, the questions arises: do firms on purpose locate subsidiaries in ‘corporate governance havens’ in order to avoid strict regulations? Wan and Hoskisson (2003) investigate corporate diversification by incorporating the importance of home country environments, including terms such as judiciary efficiency. They argue that firms draw on their home country environments’
factors and environments to do business abroad and certain diversification is performed because of home country superiority performances, as argued by Luo (2005) earlier. In this context, home country environment can be described as the institutions which set the ‘rules of the game’ that oversee the country’s economic specialization and incentive structure together with a country’s traditional set of production factors. (North, 1990) Moreover, Wan and Hoskisson (2003) state that firms’ actions and success are dependent on country level environmental factors and institutions; firms can make use of this institutions by pursuing strategic actions and improving their competitiveness. This is especially true, considering this paper, for outbound international diversification. This can be described as the diversification of a firm across geographic boundaries by entailing their capabilities, developed in the home country. Additionally, to the usual benefits of internationalization (e.g. economies of scale and scope) other important benefits are the accumulation of valuable international experience and share core competencies across the firm. Most importantly, the results listed by Wan and Hoskisson (2003) show that, in regard of outbound international activity, firms from strong CLCG countries improve their
performance when international diversification occurs. As Wan and Hoskisson (2003) argue that institutions in general represent important factors in influencing a MNEs activities, and MNEs draw on this environment and institutions to help produce goods or services,
diversification strategies are likely to be associated with good performance in the home country environment.
As Wan and Hoskisson’s (2003) main argument is competition driven, Yiu, Lau and Bruton (2007) focus on firms from emerging markets, ruling out that the firms are the best
competitive firms. Those firms are often latecomers in the global competition, their results at home might be impressive, but probably not enough to compete with host countries’
domestic firms. Firms who persevere outward international expansion can undertake an entrepreneurial organizational transformation process that transform the firm into an international competitive MNE. In conventional FDI frameworks, institutional characteristics are not embedded, which according to Dunning (2006) should be among analysing
internationalisation activities. This shows that not only competition driven forces are giving a head start towards MNEs, but institutional advantages should be incorporated in the internationalization frameworks. Lien, Piesse, Strange and Filatotchev (2005) describe the FDI decisions for Taiwanese firm on a governance base. The results indicate that the kind of governance framework matters for international actions, such as deciding where to invest.
These thoughts are confirmed by Hoskisson, Hitt, Johnson and Grossman (2001) who discuss the issue of owners issuing their voice in corporate strategy decisions and is related to the agency problem. In order to monitor managerial actions, shareholders can use different mechanisms to align managerial actions with shareholders interest, such as board of
directors incentivized contracts, to obtain profit maximizing behavior, or as I argued earlier, whether or not to internationalize.
The argument I would like to research is that companies from the U.S who internationalize do better in a low CLCG foreign environment than in a high CLCG foreign environment and that high quality FLCG U.S. MNEs perform better than their low quality FLCG counterparts in this foreign environment. The first part of this argument is confirmed by Alshammari, Hammoudeh and Pavlovic (2015). These authors find that regulation quality, CLCG, can play a negative role on the relationship between the openness towards FDI. This means that countries with high openness and low quality regulation receive more attention from MNEs because of the potential regulations gaps and flaws. This is an indication that MNEs use the possibility of additional regulation benefits for their firm performance. This thought is confirmed by Mitton (2002), who states that individual firms have some control over the amount of protection offered to shareholders. This can be achieved through multiple ways, such as a more focused corporate organization, improved transparency, higher disclosure quality or a change in the ownership structure. This way, shareholders can be offered more protection, even beyond their legal rights. More support for these arguments is given by Claessens (2006), who states that firms benefit from good governance and should reform more voluntarily. Globerman and Shapiro (2002) prove that the governance structure of a country is an important determinant for both inward and outward FDI and good CLCG trigger outflows. The second part of the argument made at the beginning of this paragraph, is supported in previous paragraphs, as is established that a high level of FLCG quality increases MNE performance.
The question arises in why countries do differ in corporate governance. As the U.S. was one of the first country to install corporate governance codes in 1978 (Aguilera and Cuervo- Cazurra), most ‘Western’ countries only followed in the period 1992-2005. Issues
concerning efficiency and legitimacy are triggering to form those kind of codes, but a strong presence of foreign institutional investors is definitely a reason for countries to employ such codes. Enrione, Mazza and Zerbone (2006) find that codes of governance are
institutionalized and corporate governance is implicitly globalized. Still, rules differ, as in the
process of developing the codes, different stakeholders are in place, such as law, model, market and governance makers.
In order to establish sufficient theoretical argument for the development of this research, the article by Doidge, Karolyi and Stulz (2007) particularly contributed. In this article, a distinction is made between investor protection in twofold: granted by the state and protection adopted by the firm. The amount of protection adopted by the firm depends on the cost of doing so. In countries with a weak regulatory framework adopting those
practices are expensive, but might be fruitful because high FLCG leads to high value or performance, as seen in the discussion of theory above. Their results show, that for high S&P rated firms, FLCG does matter, but less than CLCG. But in contrary globalization has reduced the need for country characteristics, increasing the incentives for good firm governance. A difficult task in this paper is to identify the firms’ level of corporate
governance. Klapper and Love (2004) already found that there are firms with strong FLCG in weak governance environments (e.g. weak legal systems or minimum good codes of
governance) and firms with weak FLCG in strong governance environments. Doidge et al.
(2007) find that for the former firm characteristics are less important, but for the latter FLCG is highly important for strong governance environments (read: developed countries)
An indicator of strong CLCG is investor protection. La Porta, Lopez-de-Silanes, Shleifer and Vishny (2000) find that strong investor protection is associated with effective corporate governance and starting by describing investor protection is a fruitful way in starting to differ in countries regulatory environment. As the view of La Porta et al. (2000) is seen as common, Becht, Bolton and Röell (2005) indicate that there is no unique set of optimal rules that are universally applicable and adjustments have to be made. This is comparable with politics; no single political constitution is universally applicable. Bhasa (2004) approves this line of thinking by stating that due to different legal and cultural structures corporate governance can never converge universally. Considering investor protection, one must define the agency theory: an economic view on risk-sharing between two parties, agents (managers) and principles (shareholders/investors), who possess different approaches to solve a problem. (Jensen and Meckling, 1976) The heart of the agency problem lies in the concern of self-interest of agents to act not in behalf of the principal. The phenomenon of limiting agent’s self-serving behavior is linked to the positivist agency theory, while other features can be linked to the principal-agent research. Two of those are: divergence in risk- sharing leads to 1: information asymmetry, which then leads to less ability to monitor from the principal’s perspective and 2: a lack of ideal contract negotiation. (Bendickson,
Muldoon, Ligouri, Davis, 2016). Filatotchev and Wright (2011) argue that the agency theory is complementary with the earlier described internationalization theory. The
internationalization theory is the design of governance structures which minimizes the cost of transactions. The agency theory is the design of the contract to optimize alignment between principal and agent. Achieving optimal forms of these two concepts leads to a
higher profitability of the MNE, which indicates that higher FLCG increases performance.
Filatotchev and Wright (2011) conclude with endorsing the need for a governance structure contract which aligns the manager and shareholders, rather than simply achieve governance structures that minimize transaction costs.
In the light of the agency problem and corporate governance, Singh and Davidson (2003) find that in large publicly traded corporations the ownership structure alleviates those principal-agent conflicts. On the other hand, Brown, Helland and Smith (2006) find that larger boards tend to become more prone to social interaction and become less connected to the managerial process. Moreover, larger boards have more free-rider problems leading to less effective monitoring. This indicates that there is no optimum governance mode.
In order to confirm that good FLCG matters, an earlier step has to be taken. As argued, the voluntarily adaptation of governance rules leads to better performances of the firm. This creates a gap between the CLCG and FLCG scores. However, firms can opt to locate their subsidiaries in low CLCG countries, which automatically creates a large gap between the (mostly) high U.S. MNEs FCLG score. Argued can be that U.S. S&P 500 MNEs have more
‘playing ground’ in countries with a low CLCG score: the ability to increase competitiveness due to a lack of regulations. As Maher and Anderson discuss (2000), OECD countries have developed a wide variety of systems to deal with the agency problem arising from the dispersed ownership and control. The authors show how a corporate governance
framework can have a negative effect on the development for firms in the area of R&D and innovative activity, a slower development of equity markets in the country and therefore negatively affect firm performance. Moreover, Klapper and Love (2004) state that firms could voluntarily increase the corporate governance framework, for example increasing investor protection by increasing disclosure and impose disciplinary mechanisms that prevent management and controlling shareholders to discriminate against minority shareholders. I argue that in lower CLCG environments MNEs have a higher flexibility in adjusting those mechanisms towards their own approach and perform better. Such
thoughts are further confirmed by Mata and Freitas (2012) who argue that subsidiaries from across borders perform better than local competitors because of advantages they possess due to general ownership, which are prior of entering the foreign market. Also, they choose a specific strategy of entering the market and, more importantly for this study, foreign firms are less constrained by the host country local institutions. This because those firms are better in dealing with periods of institutional changes and, as will be argued below further, better in recognizing inefficiencies in local institutions. This leads to hypothesis 1, which states that firms operating in countries with low CLCG quality have a higher firm
performance.
Considering the information above a hypotheses I would like to research is:
Hypothesis 1: MNE performance of US firms in host countries decreases with an increase in quality of CLCG in these host countries.
Hypothesis 2: The intensity of relationship between US firms performance in relatively low CLCG countries is higher for US firms with high FLCG then US firms with low FCLG.
As this article’s sample is from the largest U.S. MNEs, their level of governance ought to be high. As Klapper and Love (2004) argue, FLCG matters more in emerging markets, which therefore derives hypothesis 2 of this article. The hypothesis test that the assumption that high FLCG MNEs have better performances in low CLCG countries (e.g. less developed).
Hypothesis 2 assumes that this effect is stronger in countries where the governance environment is underdeveloped, and a large gap can be created between the mandatory regulations and voluntarily regulations. FLCG is a moderator on the connection between CLCG and MNE performance, as a higher level is ought to increase the performance. This implies a large difference between the firms FLCG and CLCG score, which are further discussed in chapter 3.
Multinationalism:
In the previous section, aspects of multinationalism, which is defined as being a large cooperation with subsidiaries in several countries, is already highlighted, such as the difference in valuating domestic and multinational firms and if expanding internationally leads to value destruction or an increase in value. Those concepts are related to the corporate governance literate and needed to establish hypothesis 1 and 2. A large part of the literature is focused on the MNE and corporate governance, which automatically leads towards discussing the MNE literature, whilst the following part focusses on the effect of being an MNE on the performance of the firm.
For years on end, the traditional American view of foreign expansion included FDI, focused on firm specific capabilities developed in the U.S. and a slow country per country expansion strategy dominated the global economy. (Guillén and García-Canal, 2009) In more recent times. MNEs have in general started to pay attention towards board supervision, structuring of sustainability responsibilities, compliance, external verification and ethics. (Kolk, 2008)
The case of Enron, which was one of the scandals leading to the SOX, was an example of a complex MNE, which led to an apparent lack of oversight of the decision-making process.
(Muchlinski, 2005). Sanders and Carpenter (1998) confirm this view, as they argue that firms face pressure to internationalize in order to survive, but this increases the ability to cope with complexity. A large international organization can benefit from corporate governance
on accounting and disclosure practices to minimize a repeat of scandals and raising investor trust. Some authors even describe corporate governance as a critical determinant in the firms’ ability to successfully deal with such complexity. (Bartlett and Ghoshal, 1989. Child, 1972)
As firms tend to increase their multinationalism, far away countries are targeted, which lead to an increase in institutional distance. This institutional distance can be described as the degree of difference or similarity between regulatory, cognitive and normative institutions (Kostova, 1996). Institutional distance influences many decisions, such as choice of location, entry mode, ownership strategy and subsidiary performance analysis (Shirodkar and Konara, 2016). The difficult part for MNEs is to tackle institutional distance and Shirodkar and
Konara (2016) find that institutional distance negatively affects subsidiary performance in emerging markets. This effects stems from the fact that regulatory differences between countries will increase the liabilities of being multinational and increase the cost of learning.
This learning curve is necessary to adopt to local institutions to achieve strong performance.
Gaur and Lu (2007) argue that small institutional differences will lead to the ability for MNEs to gain from institutional arbitrage opportunities, such as better ownership structures. The changes to be made to overcome large institutional differences, however, will outweigh the benefits, and will lead to disadvantages in comparison with firms from similar institutions.
Shirodkar and Konara conclude that, indeed, the greater the institutional distance, the lesser the subsidiaries’ performances.
As Sanders and Carpenter argue that a firm’s degree of internationalization has significant effect on its choice of corporate governance arrangements, I argue that the level of internationalization leads to a better governance structure within the MNE. As the
complexity of the firm increases, so does the level on governance employed within the firm.
Aggarwal, Erel, Ferreira and Matos (2011) investigate the role of the institutional investor on FLCG and conclude that an international portfolio promotes good governance. Moreover, host country institutions influence the level of FLCG in the firm outside the U.S. This will lead towards better performances for U.S. firms in foreign countries that are more international than firms that are less international (i.e. better results for higher level of multinationalism).
This argument is backed by Goerzen and Beamish (2003). In their article, they conclude that being international is significantly related to firm economic performance. However, many diversity in host nation environments can lead to a decline in the firms’ performance. MNEs should be encouraged to expose the firm more to international markets for positive forces within the firm. The article of Henisz (2000) supports Sanders and Carpenter with his article, which states that expanding abroad does face complex institutional objectives. Those problems are enlarged by political hazards and potential contractual hazards with host- country firms. A combination of host country institutional characteristics and potential joint ventures should be considered during the foreign expansion. By acquiring host-country
operations, MNEs can safeguard themselves from political hazards and further can employ their own governance mechanisms.
More international firms are those firms who have a higher presence in foreign countries, which can be described as a higher level of multinationalism. This multinationalism is a positive moderator for the in hypothesis 1 researched effect of CLCG. With an increase in multinationalism and everything else ceteris paribus, the performance of the MNE will be less negative or even positive, because the more multinational firm should have a higher governance framework. An increase in multinationalism will generate MNEs which are better in dealing with higher complexity than their low multinational counterparts. They are better in transferring institutional advantages across borders. Institutional advantages are the advantages the firms have obtained and employ those in the foreign environment. The larger the gap between the governance rules enforced by a host country (CLCG) and the voluntarily employed rules by the MNE (i.e. MNEs entail better governance because they believe this will lead to higher firm performance), the higher the firm performance for the MNE.
With a current trend of globalization, it can be expected that different corporate
governance system will converge. Khanna, Kogan and Palepu (2006) investigate this issue and find that such convergence does not exist, except for developed countries with proximity and similar economies. Therefore, large differences between countries exist which MNEs can use in arbitrage opportunities. These large differences in turn automatically provide the MNEs with large institutional distances, but the more multinational firm should overcome this issue by employing a higher level of FLCG. Moreover, Khanne et al. (2006) argue that firms might exceed the rules of the book in order to achieve economic
advantage. If the U.S. MNEs do benefit from this arbitrage opportunity (i.e. the opportunity that arises in the host country of exceeding the necessary rules) this will lead to better economic performance. The same line of reasoning is applied in the development of hypothesis 3 as was before in hypothesis 2: the more multinational MNEs face more complexity by their presence in foreign territories, which is dealt with by incorporating better FLCG. Even without measuring the FLCG the number of foreign subsidiaries will indicate this indirect effect of better FLCG.
By expanding into foreign territories the number subsidiaries increases automatically, leading towards a higher level of multinationalism. This increases complexity (Sanders and Carpenter, 1998), which is dealt with by the MNEs ability by employing a high level of FLCG.
Filatotchev and Wright (2011) argue that the decision to internationalize should also depend on the governance mechanisms (FLCG) of the MNE.
Hypothesis 3: The intensity of relationship between US firms performance in relatively low CLCG countries is higher for US firms with high multinationalism then US firms with low multinationalism
As Aguilar and Cuervo-Cazurra discuss, not every countries’ code of governance (e.g.
German personal liability code) is enforced by firms. More extensively, emerging countries’
code of governance tends to be less developed as such, and the stock markets development determines a large deal of strength of the codes. Even more, the compliance with the existing codes in developing countries is scarce. This leads to my thought that in countries were the codes are underdeveloped and enforcement is little, firms who voluntarily employ
‘good’ governance will have a better performance. The more multinational firm makes us of their experience in employing extensive corporate frameworks and will in turn perform better. Aguilera and Cuervo-Cazurra (2009) further stretch the fact that the link between the firms governance structure and performance should be investigates into greater depth.
As all hypotheses are focused on U.S.-based firms, I will elaborate more about the environmental setting in the U.S. Institutional investors have the power to exercise coordinated and collective power, as they control a large share of the equity market.
Considering the board of directors in the U.S., the CEO is also the chairman of the Board, which lays an enormous amount of responsibility and power on one single person and therefore is likely to inhibit effective monitoring. (Aguilera, Williams, Conley and Rupp, 2006) Other general information includes dispersed ownership, few mechanisms for stakeholder relationship handling are in place, high merger and acquisition activity and hostile takeover laws are highly regulated.
In order to research the developed hypotheses empirical methods must be developed. A benchmark for such a process is provided by Dowel, Hart and Yeung (2000), as their research question is constructed in a similar manner. However, their study focuses on an entire different topic. The dependent variable is market value, which is comparable measurement as this study’s dependent variable, and the independent variable is environmental standards. This differs substantially from the researched subject of this study, but the research method is viable as benchmark. Dowell et al. (2000) investigate the role of environmental issues employed by the MNE, whereas this study focusses on
corporate governance, and the method employed is frequently seen in the literature.
Another article of influence in the choice of empirical methods, is De Jong, Phan and van Ees (2011). A combination of the two articles is used as framework in the employed method, which is further discussed in section three.
3) Sample, Variables and Method
Sample
The sample researched in this study is drawn from within the borders of the U.S. The discussion about corporate governance has been severe in the U.S., which initiated the global discussion about corporate governance. A part of this study focusses on the effect of multinationalism on the MNEs performance. For this reason, data of large global U.S. MNES must be collected, which is obtained from the S&P500. In order to perform a conclusive research, an homogenous and representative sample must be used. Therefore I have selected the firms between the Standard Industrial Classification (SIC) codes 0-3999. This sample of firms provide a sufficient amount for statistical analysis (for amount of unique firms, N=212) and are fairly homogenous. The firms embedded in this region are active in the Agriculture, Forestry & Fishing (01-09), Mining (10-14), Construction (15-17) and Manufacturing (20-39) industry2. The sample includes firms with a large number of subsidiaries >600 but also firms with a small amount of subsidiaries <10.
The sample is from the period 2012-2015. This is the most recent data available. These four years are weighted equally in the calculated scores. For the total number of subsidiaries, the latest information for the locations is taken. The information provided by the U.S. Securities and Exchange Commission is inconsistent considering the years provided, which rules out the option to gather the same year data for every firm. However, the data on the location of the subsidiaries is from within the 2012-2015 period. Firms who do not provide this
information are deleted from the sample. The sample period 2012-2015 is used as the corporate governance discussion is becoming stronger and firms have more option to differentiate themselves after the 2008 financial crisis. In this period, the U.S faced similar growth rates as pre-crisis3. The literature about corporate governance increased extensively after the Enron scandal of the beginning of this century, which was re-aroused after the financial crisis, which started in 2008 and lasted several years.
Variables
Dependent Variable
The dependent variable for measuring firm performance, as is done by (De Jong et al. 2011) is MNE performance, which can be described as the earnings before interest and taxes divided by the total assets. MNEs that are embedded in this study must be listed as stock
2 www.siccode.com
3 http://data.oecd.org
firm in the U.S. The dependent variable is of particular interest because it is widely used as an indicator of the performance of the firm. Return on assets would be another
measurement for performance, but as subsidiaries are located in different tax regimes, the EBIT is more of interest. MNE performance is an interesting dependent variable, as the research is about the effect of the location of the subsidiaries on the MNE overall. A large part of this overall performance is derived from foreign territories, as the MNE operates in multiple countries. Firm value is of less interested, as this variable focusses on investor perception of corporate governance and not on the level of multinationalism. An overview of all variables is provided in Table 1.
Independent variables
FLCG
As stated by Krafft, Qu and Ravix (2008), corporate governance is not an end point, but rather best practices and are most of the time a journey that is time-bound. This is
illustrated by the fact that during the time of the Enron scandal (more companies involved), which caused the SOX, the firms, industries and countries where those scandals occurred were perceived as the ‘best practice model’ of corporate governance. More clearly; Enron was indicated as doing better than 41.1% of comparable companies on the S&P financial index.4 For studying the effect of FLCG and CLCG in combination with multinationalism, a clear line must be drawn on what ‘good’ and ‘bad’ governance rules are exactly. This is provided by CVGSCORE data from Asset4, retrieved from the Datastream database on the Zernike campus. The measure for FLCG is described by the Thomson Reuters Asset4 ESG Data Glossary as:
‘The corporate governance pillar measures a company's systems and processes, which ensure that its board members and executives act in the best interests of its long term shareholders. It reflects a company's capacity, through its use of best management practices, to direct and control its rights and responsibilities through the creation of
incentives, as well as checks and balances in order to generate long term shareholder value’.
The original data is ranked as percentage score, with 100 = 100%.
CLCG
The data for CLCG is much more complicated and time-consuming to construct. For all the foreign subsidiaries of the S&P500 firms with SIC code between 0-3999 an individual value CLCG is assigned. Exhibit 21 of The U.S. Securities and Exchange Commission provides all locations for the firms. In combination with data from the Worldwide Governance Indicators from the Worldbank5 an average value from the six dimensions is calculated per country.
Those dimensions are: Voice & Accountability, Political Stability & No Violence, Government
4 www.issproxy.com
5 http://info.worldbank.org/governance/wgi/index.aspx#home
Effectiveness, Regulator Quality, Rule of Law and Control of Corruption. The underlying data is weighted by the original constructors of the data. Then, the combination location value is assigned and the average value for all foreign subsidiaries is calculated. For example: Firm X has four subsidiaries: in The Netherlands, Japan, South Africa and China. The corresponding scores are (minima = -2.5, maximum = 2.5) 1.69, 1.32, 0.2 and -0.49. The CLCG value for Firm X is therefore 0.68. If Firm X has multiple subsidiaries in a country those are taken into account as well, by multiplying the CLCG score with the number of subsidiaries. As The U.S.
Securities and Exchange Commission does not provide the number of foreign subsidiaries for all years, the number of foreign subsidiaries is calculated by using average percentages of change per year.
Multinationalism
The total number of foreign subsidiaries is added to derive a multinationalism score. This is calculated by simply add up the total number of foreign subsidiaries. As the U.S. Securities and Exchange Commission does not provide data for consecutive years and average
percentages change are not provided, the variable multinationalism is time invariant. As the rest of the data is panel data, adding observations for multinationalism is preferable.
However, as these are not found, and broadening the sample of units or lengthening the time-period under observation does not provide a solution, method described below in the method section is validated (Plümper and Troeger, 2005)
Control variables
As many other factors influence the dependent variable, control variables need to be included. MNE performance is furthermore related to MNE size (natural logarithm of revenues), MNE sales growth (SG = (Sit – Sit-1)/Sit-1 (%)), MNE age and MNE subsidiaries (foreign) (de Jong, Phan and van Ees, 2011). MNE subsidiaries is defined as the variable multinationalism, as it is the same measure. Those variables are derived from Datastream, Orbis and Wharton Research Data Services and a summary is provided in Table 1. The variable sales growth assumes that capacity is used more fully and therefore fixed costs are divided over a larger number of products or services delivered. Hence, the higher the sales growth the better the MNE performance. The variable MNE size, measured by the natural logarithm of revenues, includes the effect of large MNEs who are better in exploiting economies of scale, which allows larger return on assets and sales. Therefore, the expected effect is positive on MNE performance. A third control variable included is MNE age, as older firms might have lower performance than their younger counterparts, as the MNE might embed outdated management, obsolete technologies or resilience towards new approaches. This variable is calculated by subtracting the founding year of the MNE of the year of interest (e.g. between 2012-2015). The last control variable included (not in all regression models) is the level of multinationalism. In Model 5 this variable is excluded as control variable and added as independent variable, as this measure is also the variable for hypothesis 3. As De Jong et al. (2011) describe this effect as a restraint towards MNE
performance, I argued earlier and extensively above that MNE performance is positively related with the level of multinationalism.
The data is gathered from different sources, which indicates that any bias derived from a singular source would be dismissed. However, to cross check the different data sources and validate them, a good indicator is the percentage of foreign sales and compare them with the total number of foreign subsidiaries, with the assumption that foreign sales are performed through foreign subsidiaries. Both those variables are derived from a different source. 10 MNEs out of the sample do not have any foreign subsidiaries and the average percentage of foreign sales for those firms is around 1.5%, which is most lower than the average. This indicates that the sources provide similar data. Table 1 provides an overview of the variables described in this section.
Variable Measurement Source
MNE Performance EBIT/ Total Assets Wharton Research Data Services
FLCG CVGSCORE Asset4 Datastream
CLCG Score constructed on basis of
location of MNE’s subsidiaries
Score from The Worldwide Governance Index
(Worldbank) and location from The Securities and Exchange Commission Multinationaliism Natural Logarithm of
Number of foreign subsidiaries
The Securities and Exchange Commission
Control Variables
MNE Size Revenues (log) Wharton Research Data
Services Sales Growth SG = (Sit – Sit-1)/Sit-1 (%) Orbis
MNE Age Number of years since
incorporation
Orbis
Table 1 - Variables
Method
When using panel data, several tests are needed to validate the models employed. In order to perform a panel regression, first The Hausman test is used, which compares the
coefficient estimates from the random effects model to those from the fixed model and
assumes consistency. The chi square score of 64.70 and small p-value leads towards the rejection of H0 which assumes equal estimates to another.6 This difference suggest that the random effects model is inconsistent. However, as Plümper and Troeger (2005) argue, a random effects model should be adopted over a pooled OLS regression, which is another option in performing the analysis, in the presence of time-invariant variables. The variable multinationalism is time-invariant, as the observations are equal over the years, as
discussed earlier in the previous section. Plümper and Troeger (2005) state: ‘’Random effects should perform better than pooled OLS even if the Hausman test suggests (Hausman 1978) that random effects are inconsistent and a fixed effects specification is required. When time invariant variables preclude the estimation of unit fixed effects, random effects may serve as a viable second best option.’’ For this research, the ‘second best option’ is applied as the first best option, as the fixed effects model excludes an important variable:
multinationalism.
To test for heteroscedasticity, a likelihood-ratio (LR) test is performed between generalized least square regressions, which are part of the random effects model, between (1) model with panel-level heteroscedasticity and (2) a model without heteroscedasticity. The results of the LR test assume a better fit of model (1) over model (2), which suggest the model which allows for heteroscedasticity fits better.7 To test for autocorrelation in the panel data, a serial test is performed.8 This Woolridge test for autocorrelation in panel data is significant (Prob > F = 0.0041) and we fail to reject the null hypothesis of no first-order autocorrelation.
In order to analyse the hypotheses, models need to be constructed. Hypothesis 1 can be analysed with a simple bivariate regression model, Hypothesis 2 and Hypothesis 3 include an interaction variable able to measure the moderating effect of FLCG and Multinationalism, respectively and multivariate regressions are performed.
For hypothesis 1:
𝑀𝑁𝐸 𝑃𝐸𝑅𝐹𝑂𝑅𝑀𝐴𝑁𝐶𝐸+ = 𝑎.+ 𝛽1𝐶𝐿𝐶𝐺+4+ 𝛽5𝐶𝑂𝑁𝑇𝑅𝑂𝐿+4 + 𝜀+,4
For hypothesis 2:
𝑀𝑁𝐸 𝑃𝐸𝑅𝐹𝑂𝑅𝑀𝐴𝑁𝐶𝐸+ = 𝑎.+ 𝛽1𝐶𝐿𝐶𝐺+4+ 𝛽5𝐹𝐿𝐶𝐺+4 + 𝛽9𝐶𝐿𝐶𝐺+4 𝑋 𝐹𝐿𝐶𝐺+4 + 𝛽;𝐶𝑂𝑁𝑇𝑅𝑂𝐿+4 + 𝜀+,4
For hypothesis 3:
𝑀𝑁𝐸 𝑃𝐸𝑅𝐹𝑂𝑅𝑀𝐴𝑁𝐶𝐸+ = 𝑎.+ 𝛽1𝐶𝐿𝐶𝐺+4+ 𝛽5𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑠𝑚+4 + 𝛽9𝐶𝐿𝐶𝐺+4 𝑋 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑠𝑚+4 + 𝛽;𝐶𝑂𝑁𝑇𝑅𝑂𝐿+4 + 𝜀+,4
6 Stata results and calculations available upon request.
7 Heteroskedasticity is of a problem when the residuals do not have the same variance. When performing White’s test, we reject the hypothesis of homoscedasticity and accept HYPOTHESIS 2 of unrestricted heteroskedasticity. Chi-square of 1625.58, p-value of 0.0000.
This allows for a Generalized Least Squares estimator, which corrects for heteroskedasticity across panels.
8 Downloaded and installed manually for Stata.
Table 2 - Descriptive Statistics and Correlations, observations = 623. * = p < 0.10. ** = p< 0.05. and *** = p < 0.01.
The regression method applied in this study needs to control for both heteroscedasticity as autocorrelation, which can be done by a random effects model; the Generalized Least Squares (GLS) estimator. A study with a similar subject and research questions (Shirodknar and Konara, 2016). As Hoechle (2007) describes in his article, the GLS option allows for the standard errors being robust to both heteroscedasticity and autocorrelation. This validates the use of this regression method.
4) Results
Summary statistics and correlations are presented in Table 2. All correlations are below the 0.80 threshold, except for the interaction variable (5) CLCG-FLCG with (2) CLCG. This result is not shocking as the stand-alone variable CLCG is part of the interaction variable CLCG-FLCG.
The other variables are below 0.8, which is the common threshold value for multicollinearity. (De Jong et al., 2011)
To analyze the data, bivariate and multivariate analyses are performed. By performing multiple regression, the dependent variable is predicted by the independent variables, and controlled for by the control variables. As the variable multinationalism is only collected for one year in the 2012-2015 time frame, the same score is given to all the years. By applying a random effects model and not a fixed effect model, as suggested by the Hausman test, these variables remain in the sample. The amount of subsidiaries remaining the same over
Variable Mean SD (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) MNE Performance
0.309 0.557 1.00
(2) CLCG 0.943 0.357 -.036 1.00
(3) FLCG 81.34 4
12.758 -.042 -.0012 1.00
(4) Multinat ionalism
3.790 1.294 -.029 -.265*** 0.094** 1.00
(5) CLCG- FCLG
76.70 4
31.260 -.050 .909*** .374*** -.201*** 1.00
(6) CLCG- MULT
3.623 1.448 -.087** .428*** .012 .709*** .392*** 1.00
(7) MNE Size 4.007 0.500 .026 .0107*** 0.270*** .172*** .193*** .094** 1.00 (8) MNE Age 44.33
0
33.804 .029 -.084** .149*** .263*** -.017*** .136*** .183*** 1.00 (9) Sales
growth
3.538 18.294 .309*** .023 -0.178*** -.086** -.013 -.013 -.151*** -.103*** 1.00
