The Influence of the Meta-environment of MNEs on the Convergence of Excess Returns

Master Thesis

MSc Finance and MSc International Economics & Business

The Influence of the Meta-environment of MNEs on the Convergence of Excess Returns

By Willem Sieben

Author: Willem Sieben Student number: 1891456

Place and date: Groningen, 9

of January 2015 1

supervisor: Prof. Dr. H. van Ees

2

supervisor: Dr. Ing. N. Brunia

1

The Influence of the Meta-environment of MNEs on the Convergence of Excess Returns

Willem Sieben

Abstract

This study entails the influence of the meta-environment of multinational enterprises (MNEs) on the convergence of excess returns. Whereas the meta-environment is defined as the configuration of all MNE’s home-country and host-country environments. Investigating which factors determine future excess returns, is important for valuating firms. Therefore, this model employs value-based excess returns and takes a closer look into how meta-environments support MNEs in their competitive efforts to sustain their excess returns. Though, strong evidence of converging excess returns is found, the effects of various dimensions of the MNEs’ meta-environment on the speed of convergence of excess returns are more ambiguous.

Keywords Mean reversion, abnormal performance, competition, meta-environment, environmental munificence

JEL Classification G17, G30, F37

1 Introduction

This paper examines whether mean reversion in MNEs’ abnormal performance is influenced by their meta-environments. The munificence of several advanced resources and formal institutions in the meta-environments of MNEs might affect the convergence of their excess returns.

According to Stigler (1961) and Aghion et al. (2001) economic theory argues economic rents

will perish due to competition. Firms with superior or abnormal economic rents will face more

competition, as the abnormal economic rents attract new entrants perishing them away. Therefore,

abnormal performance will mean revert in time (i.e. Fama & French, 2000). Though, competitive

efforts may enable firms to persist their abnormal performance for a longer period of time (Porter,

2008; Porter, 2003). In other words, firms obtaining sustainable competitive advantages have the

opportunity to resist the convergence of their excess returns. These sustainable competitive advantages

may be induced by concepts such as economies of scale, entry barriers and innovativeness (i.e. Koller

et al., 2010). Moreover, the mean reversion of excess returns may also be determined by country-

environments. More specifically, competition on product, capital and labor markets is determined by

the home-country environments of firms according to Healy et al. (2014), thereby impacting the mean

reversion of excess returns. Consequently, more competition in home-country environments induces

faster mean reversion of firms’ excess returns.

2 However, the abnormal performance of MNEs is not only determined by the competition on several markets in its home-country environment. According to Healy et al. (2014) host-country environments also influence the mean reversion of MNEs’ excess returns. This follows from how a MNE can be characterized and operates. MNEs can be characterised as firms with an internal network structure over multi-countries in which they operate. These MNEs may obtain several competitive advantages from their unique network structures throughout their home-country and host-countries (De Jong, et al., 2011; Nachum, Zaheer & Gross, 2008; George & Zaheer, 2006). Through leveraging the opportunity set of advanced resources and institutions MNEs can achieve abnormal performance levels (Wan & Hoskisson, 2003). As MNEs operate in their meta-environment of multiple countries, MNEs may be able to leverage the opportunity sets from multiple countries by utilizing their internal network structure. Though, more munificent country-environments in terms of advanced resources and institutions may attract more competition (Wan & Hoskisson, 2003), the various advanced resources and formal institutions in the meta-environment employed in this paper may induce for example economies of scale, innovation and protection of innovative products and other entry barriers, which are at the source of sustaining excess returns (Dickinson & Sommers, 2012; Koller et al., 2010; Cheng, 2005; Mauboussin & Johnson, 1997). Moreover, the meta-environment is unique for every MNE as their symbiosis of all country environments in which it operates differs (De Jong et al., 2011). Hence, the meta-environments of MNEs may induce unique or sustainable competitive advantages, thereby enabling MNEs to persist excess returns over a longer period of time. Accordingly, this paper argues the meta-environment in terms of munificence of advanced resources and formal institutions is a more complete metric as a determinant of MNEs’ excess performance and their mean reversion as was previously employed by Healy et al (2014). This paper closes that gap.

Examining what determines abnormal performance, defined as the return spread between the return on invested capital (ROIC) and the weighted average cost of capital (WACC) in this paper, in terms of their persistence and magnitude is important for the valuation of firms as it helps predicting future cash flows. Consequently, this paper examines the influence of MNEs’ meta-environments on the convergence of excess returns. More specifically, the meta-environments of nonfinancial firms in the S&P 500 index from 2004-2013 are constructed to measure their influence on the excess returns on enterprise level of the corresponding firms. The hypotheses denote more munificent meta- environments in terms of advanced resources and institutions have a negative influence on the convergence of MNEs’ excess returns. In other words, more munificent meta-environments may have a positive effect on MNEs trying to persist their excess returns over time. The hypotheses correspond with the following general research question:

“What is the influence of MNEs’ meta-environment on the convergence of their excess returns?”

3 Thus, the main contributions of this paper is taking an important step towards a more appropriate environmental metric influencing the excess performance of firms and its mean reversion. Country- environments in which firms operate are conceptualized in terms of advanced resources and institutions as in Wan & Hoskisson (2003) and Castrogiovanni (1991) in this paper, as has been neglected in literature concerning the mean-reversion of excess returns. Subsequently, a more complete firm-environment metric, the meta-environment, is used as a factor in determining excess returns taking into account both home-country and host-country environments and the unique network structures of MNEs within these environments (De Jong et al., 2011; Nachum et al., 2008; George &

Zaheer, 2006).

Furthermore, the other contributions of this paper are threefold. Firstly, the results complement economic theory and previous research of Koller et al. (2010), Fairfield et al. (2009), Hawanini et al.

(2003) and Nissim and Penman (2001) in the sense that strong evidence is found of converging excess returns. However, this paper contributes to the existing literature by including a more realistic cost of capital in the excess return measure as it changes over time. A changing cost of capital assumes a changing capital structure over time, whereas previous literature assumed a constant cost of capital (i.e. Koller et al., 2010; Nissim & Penman, 2001), or no cost of capital at all in the convergence of excess returns (Fairfield et al., 2009). Moreover, this paper complements Fairfield et al. (2009) taking into account both an economy-wide and an industry-specific model. The economy-wide model assumes similar competition structures between industries as well as similar returns and cost of capitals. This seems not to be a very realistic assumption, as industrial organization economics argue that structural characteristics of industries are important determinants of performance (Porter, 1980).

Therefore, an industry-specific model is estimated as well, whereas some previous scholars found evidence of convergence to an economy-wide average (i.e. Nissim & Penman, 2001). Thirdly, this paper employs a value-based measure of excess returns (ROIC minus WACC), whereas previous literature often used return on assets (ROA) or ROA equivalents that do not measure cash flows and adjust the returns for risk (i.e. Healy et al., 2014). Thus, these accounting ratios do not represent a clear link with value creation. Notably many firms employ a strategy to deliver sustainable value creation, which makes it important to employ a value-based measure of excess performance (Hawawini et al., 2003). Alternatively, a return on equity (ROE) could be used as a metric within MNEs’ excess returns as was employed in Fairfield et al. (2009) and Nissim & Penman (2001).

However, this paper employs ROIC as the performance metric as it is a measure on enterprise level and includes performance from operations only (Koller et al., 2010). This paper closes all of the above mentioned gaps.

The remainder of this paper is structured as follows: Section 2 consists of a literature review and

subsequently develops hypotheses. Section 3 takes a closer look at the data, variable construction and

methodology. Section 4 entails the empirical results, whereas Section 5 provides conclusions,

limitations and possibilities for future research.

4 2 Literature review and hypothesis development

Economic theory predicts firms’ excess returns will perish, as competition will reduce future excess returns (Aghion et al., 2001; Stigler, 1961). In other words, economic theory states competitive forces may cause returns to diminish towards the cost of capital over time (Mauboussin & Johnson, 1997).

Firms try to resist their abnormal performance from being perished (Porter, 2008; Porter, 2003), whereas the period of abnormal performance is also known as the competitive advantage period (Mauboussin & Johnson, 1997). This is the period in which firms create value. Whereas firm value can be seen as the discounted future expected cash flows of a firm (Koller et al., 2010; Damodaran, 2007).

Additionally, in literature firms’ excess returns have by definition been strongly related to the value of firms (i.e. Koller et al. 2010; Damodaran, 2007; Hawanini et al., 2003). Early valuation models underline the relationship between growth and firm value. That is, firms with higher growth rates were assigned higher values. Though, more recent valuation models stress that value creation through growth is conditional to excess returns (Damodaran, 2007). This is in line with the value driver formula of Koller et al. (2010)

and with Mauboussin and Johnson (1997) both stating value is being created in periods of excess returns. In other words, firms create value by investing capital to obtain future cash flows with returns that exceed the cost of the invested capital (Koller et al., 2010).

Hence, forecasting excess returns has become of greater interest. Subsequently, explaining the mean reversion of excess returns is more relevant as it is to some extent a source of forecasting corporate returns (Fama & French, 2000), meaning that more accurate predictions of mean reversion of excess returns are important for valuating a firm. Additionally, firms can sustain their excess returns if they have strong competitive advantages. This links competitive advantage, one of the core concepts of business strategy, to value creation (Koller et al., 2010).

2.1 Mean reversion of abnormal performance and competition

In line with economic theory, Fama and French (2000) find profitability (measured by a return on assets equivalent) is mean reverting in general. Their results also show the rate of mean reversion is higher when profitability is below its average and when profitability is further from its average.

Additionally, Fairfield et al. (1996), Cubbin, & Geroski (1987) and Freeman et al. (1982) show results of mean reverting firm performance as well due to competition. Competition erodes abnormal performance as firms generating superior returns will face competition from new entrants, reducing future returns (e.g. Aghion et al., 2001). However, according to Mauboussin and Johnson (1997) firms with higher ROICs within an industry are generally more competitive and are able to sustain longer periods of competitive advantage, due to factors such as economies of scale, entry barriers, and good

1Continuing valuet

=

5 management. Firstly, economies of scale lead to lower costs, i.e. fixed costs can be spread over larger revenues. Secondly, barriers-to-entry induced through for example product differentiation and innovation, scarce resources and patents, reduce the threats of outside competition thereby sustaining abnormal performance (Dickinson & Sommers, 2012; Koller et al., 2010; Damodaran, 2007; Cheng, 2005; Mauboussin & Johnson, 1997). Thus, the main concept of the mean reversion of excess returns is as follows: Competition causes convergence of abnormal performance or excess returns, whereas (sustainable) competitive advantages may induce the persistence of excess returns. This main concept is fundamental for the results of previous research (i.e. Koller et al., 2010; Fairfield et al., 2009;

Hawawini et al., 2003; Nissim & Penman, 2001).

Furthermore, the convergence of excess returns has also been extensively covered in the literature (i.e. Healy et al., 2014; Koller et al., 2010; Fairfield et al., 2009; Hawanini et al., 2003;

Nissim & Penman, 2001). Previous research has had a strong focus on role of industry and firm characteristics in the mean reversion of abnormal firm performance and earnings forecasting (Dickinson & Sommers, 2012; Brown & Kimbrough, 2011; Koller et al., 2010; Fairfield et al. 2009;

Bou & Sattora, 2007; Cheng, 2005; Hawawini et al., 2003; Waring, 1996; Lev, 1983). Koller et al.

(2010) find that firms’ return on invested capital (ROIC) converge gradually towards industry medians due to competition, whereas their research entails an implicit constant cost of capital. Hawawini et al.

(2003) also finds abnormal firm performance is industry-specific (rather than firm-specific), that is for

an average company. They employ a value-based measure for abnormal performance, economic profit,

which is the spread between the ROIC and the WACC multiplied by the capital employed as this

focuses on the value of operations. This is in contrast to accounting ratios as a performance metric,

such as ROA, which is not consistent with what truly drives value. Moreover, Fairfield et al. (2009)

investigates the mean-reversion of performance on equity level instead of on enterprise level (Return

on equity; ROE vs. ROIC). They find firms’ ROE converge to an economy-wide mean instead of an

industry-specific level whilst assuming no cost of capital. However, Fairfield et al. (2009) also obtains

results indicating industry-specific models are generally better in predicting firm growth than

economy-wide models. Industry-specific patterns, magnitudes and signs of excess returns may stem

from the structural characteristics of industries being the primary enablers of performance (Porter,

1980). Thus, profit differentials may be sustained through the presence of different circumstances

(Caves & Porter, 1977), leading to for example economies of scale or innovation. Nissim and Penman

(2001) analyse excess returns both on enterprise and on equity level in an economy wide model whilst

assuming a constant cost of capital. They find return on assets (ROA) and return on equity (ROE)

converge towards an economy-wide average. All of the above do not include a varying cost of capital

over time or a cost of capital at all. A constant cost of capital assumes a constant capital structure,

which might not be true in reality. This implies a gap in the literature on the convergence of excess

returns.

6 Furthermore, firm-specific characteristics influencing the mean reversion of excess returns stem from the resource-based view (RBV). The RBV proposes that idiosyncrasies among firms in the aggregation and leverage of unique and durable resources and capabilities are the source of (sustainable) competitive advantage (i.e. Barney, 1991; Wernerfelt, 1984). These sustainable competitive advantages may lead to firms resisting the decay of their excess returns through time.

Several firm-specific characteristics have been tested on the mean reversion of accounting returns in prior literature, including firm size, which may induce economies of scale (i.e. Lev, 1983), future investment opportunities, as these might generate positive NPVs (Nissim & Penman, 2001) and intangible investments, which could stimulate innovation (Brown & Kimbrough, 2011). Additionally, Cheng (2005) finds that firm abnormal return on equity (ROE) depends on market share, firm size, firm-level barriers to entry, and firm conservative accounting factors and industry abnormal ROE on industry concentration, industry-level barriers of entry, and industry conservative accounting factors.

Most of these factors induce (sustainable) competitive advantages and hence excess returns. The latter is in line with Waring (1996), suggesting the persistence of firm performance is partly determined by a firm’s industry structure. Dickinson and Sommers (2012) extends the studies of Waring (1996) and Cheng (2005) using a larger set of competitive efforts. However, in contrast to previous studies they find no evidence of product differentiation, innovation, and capital requirements resulting in the persistence of firm performance adjusted for operational risk and industry.

Most recently, Healy et al. (2014) examined the degree of competition in the environment of firms on the mean reversion of excess returns. More specifically, Healy et al. (2014) investigated the influence of firms’ home-country environments in terms of market competition on the convergence of excess returns on enterprise level (ROA). They showed product, capital, and labor market competition have a significant effect on the rate of mean reversion of profitability. Their results are in line with economic theory as more competitive home-country environments have a positive effect on the convergence of excess returns. Prior to Healy et al. (2014) country environments were neglected as a factor in the mean reversion of (abnormal) firm performance. Thus, this provides research opportunities in terms of the influence of country-environments on the mean reversion of excess returns. Home-country environments might not be a complete environmental metric influencing performance for MNEs as they operate in multi-country environments. Healy et al. (2014) indicates their research has stronger effects for domestic firms than for MNEs as these domestic firms are only influenced by their home-country environments. On the other hand, the sustainability of MNEs excess performance is also determined by host-country environments. This line of theory is also hinted by an additional test they performed showing MNEs’ ROA mean reversion is lower than domestic firms’

ROA mean reversion. This is in line with the fact that geographic diversification strategies may be beneficial for MNEs (De Jong et al., 2011; Dunning & Lundan, 2008a; Dunning & Lundan, 2008b;

Nachum et al., 2008; George & Zaheer, 2006). Hence, their abnormal performance may also depend

7 on host-country environments and the unique network structures employed by these MNEs through their home- and host-countries creating unique firm-specific competitive advantages.

2.2 MNEs’ meta-environments

Though, Healy et al. (2014) took an important step in showing (home) country-environments have a role in the mean reversion of excess returns, it is important to determine a more complete environmental metric influencing the mean reversion of MNEs’ excess returns. In terms of country- environments, the abnormal performance of MNEs is not only determined by the competition on several markets in its home-country environment as in Healy et al (2014). MNEs can be characterised as firms with an internal network structure over multiple countries in which they operate. These MNEs may obtain several competitive advantages from their unique network structures throughout their home-country and host-countries (De Jong et al., 2011; Nachum, Zaheer & Gross, 2008; George &

Zaheer, 2006). Every country possesses an opportunity set of production factors and institutions firms seek to capture (North, 2006; North, 1990). Through leveraging the opportunity set in MNEs’

environments, MNEs can achieve abnormal performance levels (Wan & Hoskisson, 2003). As MNEs operate in their meta-environment of multiple countries, MNEs may be able to leverage the opportunity sets from multiple countries by utilizing their internal network structure. Whereas a MNE’s meta-environment is defined as the symbiosis of all country environments where a MNE operates (De Jong et al., 2011). Additionally, MNEs can benefit from being active in multiple countries, when being able to utilize country-specific advantages of more munificent countries in other countries where resources or institutions are lacking (De Jong et al., 2011; Giroud & Scott-Kennel, 2009; Dunning & Lundan, 2008a; Dunning & Lundan, 2008b; George & Zaheer, 2006).

Therefore, the business strategy of MNEs includes the location choice of subsidiaries as it can be at the source of competitive advantages. The location choices of MNEs and locational diversification strategies have been addressed in relation to firm performance (i.e. Qian et al., 2010;

Nachum et al., 2008; George & Zaheer, 2006). Although, firms being active in more distant markets

may suffer from the liability of foreignness, which is the economic, political, physical and cultural

distance with foreign countries causing operational difficulties. However, MNEs may be able to

overcome these liabilities. A more scattered ‘geographic signature’, which is a firm’s national

presence, enhances performance as it allows firms to tap into a broader set of knowledge bases and

diverse sources of innovation (George & Zaheer, 2006). Thus, a MNE’s local presence allows it to tap

into a broader and different opportunity set. Subsequently, Qian et al. (2010) argue an advantage of

being active in multiple countries is that it allows MNEs to more flexibly build, incorporate, or

rearrange different international resources and capabilities similar to the resources in Wan and

Hoskisson (2003) (see Appendix A). This raises opportunities for MNEs to gain competitive

advantages. Additionally, the meta-environment is unique for every MNE as their symbiosis of all

country environments in which it operates differs (De Jong et al., 2011). Hence, the meta-

8 environments of MNEs may induce unique or sustainable competitive advantages by leveraging their unique opportunity sets. Therefore, MNEs may be enabled to persist excess returns over a longer period of time. This is evidence of the importance of both home- and host-country environments in determining the excess performance of MNEs. Hence, a more complete environmental metric should include both home- and host-country environments, as was already indicated by Healy et al. (2014).

Moreover, Healy et al. (2014) configure country-environments on the basis of the competition on different markets. However, this configuration does not include the competitive advantages that can be obtained from country-specific advantages. The unique or sustainable competitive advantages induced by country-environments stem from the munificence of production factors and institutions in combination with the uniqueness of MNEs’ meta-environments. Firms that are being able to use this opportunity set have a greater chance of competitive success (Wan & Hoskisson, 2003). Though, more munificent country-environments in terms of advanced resources and institutions may attract more competition (Wan & Hoskisson, 2003), various advanced resources and formal institutions in the meta-environment may induce several unique or sustainable competitive advantages which are at the source of sustaining excess returns (i.e. Dickinson & Sommers, 2012; Koller et al., 2010; Cheng, 2005; Mauboussin & Johnson, 1997). This is related to the RBV, as the unique meta-environments of MNEs are a source of idiosyncrasies among firms in the aggregation and leverage of unique and durable resources and capabilities inducing (sustainable) competitive advantages (i.e. Barney, 1991;

Wernerfelt, 1984). Many scholars have specifically addressed the munificence of country- environments and its influence on firm strategy and performance from a resource-based view (i.e.

Sirmon et al., 2007; Verbeke & Yuan, 2007). Generally, high munificence in terms of resources should lead to increased firm performance, as it creates advantages such as improved interaction between firms and customers and facilitating new product development. Moreover, high quality institutions may enable several advantages for MNEs, including economies of scale, lower transaction costs and the protection of competitive advantages. (De Jong et al., 2011). This is in line with country- specific factors determining MNE performance as was addressed by several authors (i.e. De Jong et al., 2011; Giroud & Scott-Kennel, 2009; Dunning & Lundan, 2008a; Dunning & Lundan, 2008b;

Sirmon, Hitt, & Ireland, 2007; George & Zaheer, 2006; Wan & Hoskisson, 2003).

A stepping stone in environmental munificence literature were Wan and Hoskisson (2003)

who examined the influence of a firm’s home country environment on the relationship between

corporate diversification strategies and firm performance. Whereas their country-environment

distinguished endowed, advanced and human factors and political, legal and societal institutions (see

also Appendix A). They found that different home country environments causes the relationship

between corporate diversification strategies and firm performance to differ. Moreover, firms are able

to obtain abnormal returns due to resources and the quality of institutions available to the firm as they

enhance transformational and transactional efficiency. More specifically, firms use their home-country

environments’ advanced resources and institutions to produce goods and services at lower costs or

9 higher quality (transformational efficiency) and to facilitate transactions at lower cost (transactional efficiency). However, the environmental metric applied by Wan and Hoskisson (2003) lacks for MNEs as it does not recognize the influence of host-country opportunity set munificence and the network structures employed by MNEs.

In conclusion, a more complete environmental metric as a determinant of the mean reversion of excess performance lays with the meta-environments of MNEs incorporating the whole network of the MNE and the munificence advanced resources and institutions. Thus, ideally one should move to one holistic measure of the meta-environment of a MNE including the interactions between production factors and institutions in terms of munificence throughout the network of the MNE. However, due to the availability of data, this paper follows the approach of De Jong et al. (2011) incorporating the effects of standalone advanced resources and institutions influencing the mean reversion of excess returns. De Jong et al. (2011) tested the effects of MNEs’ meta-environments on firm performance.

Their meta-environments are determined by advanced production resources in the form of technological capabilities and quality of the infrastructure and formal institutions in the form of international trade promotion policies, flexibility of labor regulations, investment promotion policies, and efficiency of the law enforcement system of where the MNE operates. Though, this paper does not include the flexibility of labor regulation as this data is unavailable as well. Moreover, they argue meta-environments differ across firms, thereby allowing the firms to have access to a unique set of resources and institutions providing the firm a possible source of competitive advantage. Their results show a positive association between all advanced resources and institutions present in the meta- environment and firm performance, except for international trade promotion policies, which showed no significant results. Since the unique opportunity set of resources and institutions can be a source of competitive advantage this paper hypothesizes a more munificent meta-environment will reduce the decay factor of excess returns. In other words, the speed of excess return convergence of MNEs will be reduced for MNEs operating in a richer meta-environment in terms of resources and institutions.

2.3 Hypotheses development 2.3.1 Advanced resources

The first aspect of the meta-environment is advanced resources. The importance of advanced resources

stems from the RBV (i.e. Barney, 1991; Wernerfelt, 1984). The availability of advanced resources in

the meta-environment is of great importance for MNEs in order to create and sustain competitive

advantages (De Jong et al., 2011; Verbeke & Yuan, 2007; Hawawini et al., 2003). Moreover, the

availability of advanced resources may enable firms to generate abnormal returns (Castrogiovanni,

1991; Grant, 1991). The firm-specific idiosyncrasies in the accumulation and capitalization of unique

availability of resources may create sustainable competitive advantages, which are important factors in

the sustainability of excess returns (Hawawini et al., 2003). As the meta-environment is more or less

10 unique for every MNE, MNEs have access to a firm-specific pool of advanced resources that can be exchanged between subunits of the organization. This is in line with Qian et al. (2010) who argue an advantage of being active in multiple countries is that it allows MNEs to more flexibly build, incorporate, or reallocate different international resources and with George and Zaheer (2006) who argue being active in multiple countries creates a broader set of knowledge bases and sources of innovation. Whereas for example innovation can be a source of excess returns (i.e. Koller et al., 2010).

Additionally, MNEs have the advantage to be able to overcome scarcity of advanced resources in countries with less abundance of these advanced resources (De Jong et al., 2011). In summary, the firm-specific availability of advanced resources in MNEs’ meta-environments is relevant in terms of resisting to the mean reversion of excess returns.

Technological capabilities

In line with De Jong et al. (2011), technological capabilities of countries includes the availability of technologies and penetration of information and communication technologies. These facilities are of great importance for the effectiveness of intangible investments, innovation and product differentiation through new product development (Nooteboom, 2009). Additionally, innovation and product differentiation are distinguished as traditional barriers-to-entry (i.e. Caves & Porter, 1977), which is a potential source of sustaining excess returns (Koller et al., 2010; Damodaran, 2007; Cheng, 2005;

Mauboussin & Johnson, 1997). Moreover, meta-environments with high technological capability enable MNEs to obtain long-term competitive advantages, since MNEs in these environments are being empowered to develop unique competencies (De Jong et al., 2011; Nooteboom, 2009).

Obtaining these long-term competitive advantages through high levels of technological capability may support MNEs in slowing down the decay rate of their excess returns. Furthermore, intangible investments, in example to stimulate innovation, can be important as drivers of differentiation strategies allowing MNEs to create sustainable competitive advantages and excess returns (Brown &

Kimbrough, 2011). As more munificent meta-environments of MNEs enhance the sustainability of excess returns, the following hypothesis is formulated:

Hypothesis 1: Technological capabilities in the meta-environment of a MNE have a negative influence on the convergence of their excess returns on enterprise level.

Infrastructure

In literature the quality of infrastructure consisted often of one or more dimensions in terms of transport, communication and energy availability (i.e. De Jong et al., 2011, Khadaroo & Seetanah, 2010; Calderon & Chong, 2004). The availability of infrastructure is an important country environment variable that determines the minimization of costs of doing business (Khadaroo &

Seetanah, 2010). In other words, better quality of infrastructure in the meta-environment of MNEs

improves operational efficiency. For example transport infrastructure is believed to enhance both types

11 of FDI, horizontal and vertical, as it helps to reduce costs (Khadaroo & Seetanah, 2010). High-quality infrastructure in the meta-environment of a MNE enhances the interaction with other actors in the supply chain such as suppliers and customers by reducing interaction costs (De Jong et al., 2011;

Khadaroo & Seetanah, 2010). Furthermore, Giroud and Scott-Kennel (2009) argue high quality infrastructure facilitates the support between units within MNEs, which increases efficiency of the MNE. Hence, high quality infrastructure in the unique meta-environment of a MNE can be a source of sustainable competitive advantages, supporting the resistance of excess returns being perished towards the mean. Moreover, the facilitation of inter-unit support and exchange through high quality infrastructure in meta-environments allows MNEs to concentrate their production activities, thereby inducing economies of scale (Krugman, 1990). According to Koller et al. (2010), Damodaran (2007), Waring (1996) among others, economies of scale enable firms to persist their abnormal returns.

Additionally, Mauboussin and Johnson (1997) argue economies of scale are a potential source to sustain and extend the competitive advantage period, and are thus a potential source to persist the excess returns of the firm. In conclusion, the following hypothesis is formulated:

Hypothesis 2: High quality infrastructure in the meta-environment of a MNE have a negative influence on the convergence of their excess returns on enterprise level.

2.3.2 Institutions

The other aspect of the meta-environment is institutions. North (2006) and North (1990) define institutions as the rules of the game, which provide the terms for interaction between organizations.

Thus, these rules of the game prescribe an incentive structure that allow more efficient transactions for

firms (Wan & Hoskisson, 2003). Efficient transactions are especially important for MNEs as they

operate in multiple countries (i.e. De Jong et al., 2011; Chao & Kumar, 2010; Gatignon & Anderson,

1988). As the meta-environment is unique for every MNE, each MNE adapts to institutions present in

its meta-environment in its own way (De Jong et al., 2011; Dunning & Lundan, 2008). Low

munificent meta-environments in terms of institutions increases transaction costs. Consequently,

MNEs incurring lower transaction cost gain a competitive advantage (Hill, 1995). Thus, competitive

advantages can be obtained by MNEs by leveraging their own unique meta-environments in terms of

institutions. Hence, a more munificent institutional meta-environment of a MNE can be leveraged to

persist the convergence of excess returns as a more munificent MNE’s institutional meta-environment

can be a unique competitive advantage. However, this argument would not necessarily hold for MNEs

for emerging markets, as these MNEs are generally better at dealing with lower quality institutions

(i.e. Khanna et al., 2005). The latter is not the case in this paper as it only includes US MNEs.

12 International trade promotion policies

The first institutions being addressed in this paper are the international trade promotion policies. A high degree of international trade promotion policies is especially beneficial for MNEs, as these policies lower trade barriers. Consequently, lower trade barriers facilitate inter-country trade of inputs and intermediate products within a MNE (Dunning, 1998). As inter-country exchanges of inputs and intermediate products within a MNE are facilitated, MNEs foster competitive advantages and incur lower costs (De Jong et al., 2011). Moreover, lower trade barriers within MNEs’ meta-environments allow them to concentrate their production activities, thereby inducing economies of scale (Krugman, 1990). According to Mauboussin and Johnson (1997) economies of scale are a potential source to sustain and extend the competitive advantage period, and are thus a potential source to persist the excess returns of the firm. Hence, given the uniqueness of MNEs’ meta-environments, more munificent environments in terms of international trade promotion policies may support MNEs in slowing the convergence of their excess returns. Therefore, the following hypothesis is formulated:

Hypothesis 3: International trade promotion policies in the meta-environment of a MNE have a negative influence on the convergence of their excess returns on enterprise level.

Investment promotion policies

Investment promotion policies or incentive policies are introduced by governments in order to attract foreign direct investment by foreign MNEs (Blomström et al., 2003). Incentive policies can be categorized into fiscal and financial incentives amongst others (UNCTAD, 1996). These incentives may include for example tax holidays and reliefs, preferential loan provisions and subsidies (Blomström et al., 2003). Subsidies may lower in example capital, production and credit costs of MNEs (De Jong et al., 2011), thereby lowering for example the WACC. Thus, investment promotion policies can obviously have significant benefits for MNEs and its excess returns. Moreover, a MNE can reallocate these benefits, directly or indirectly through the proceeds of increased profitability by means of their internal capital markets, to other units in other countries and to positive NPV projects, which add value. Thus, a more munificent meta-environment in terms of investment promotion policies may support MNEs in resisting the diminishing of their excess returns. Hence, the following hypothesis is formulated:

Hypothesis 4: Investment promotion policies in the meta-environment of a MNE have a negative influence on the convergence of their excess returns on enterprise level.

Law enforcement system

Law enforcement systems are of vital importance for MNEs for a number of reasons. Effective law

enforcement systems support MNEs to sustain their competitive advantage, as it protects them from

imitation by its local and international competitors (Javorcik, 2004). This is in line with Brown and

Kimbrough (2011), as they argue the protection of property rights may also lead to earnings non-

13 commonality since firms are better protected against the risks of expropriation by its competitors. In other words, it allows firms to better protect innovative products and technologies that produce abnormal returns from imitation. This is important as competitors will eventually mimic these products, thereby perishing the abnormal returns (Fama & French, 2000). Since effective law enforcement systems can help MNEs in sustaining their competitive advantages, one can argue it may facilitate the persistence of MNEs’ excess returns as well. Moreover, adequate law enforcement systems that enforce property rights and settle disputes and stimulate contracting and trading ensure lower transaction costs (North, 2006; Javorcik, 2004). Furthermore, weak law enforcement systems may also shut out the option of outsourcing as the spill-over risk of valuable information or technologies is much higher, even if outsourcing is more effective and adds value (De Jong et al., 2011). Consequently, an effective law enforcement system in the meta-environment of a MNE may support the resistance of mean reversion of excess returns. Therefore, the following hypothesis is proposed:

Hypothesis 5: The effectiveness of the law enforcement system in the meta-environment has a

negative influence on the convergence of excess returns on enterprise level.

14 3 Methodology

3.1 Data

The hypotheses are tested based upon a panel dataset of US firms listed in the S&P 500 index in the period of 2004-2013. The S&P 500 includes mostly large MNEs from different industries, which operate in multiple-country environments, thereby being an appropriate sample to test my hypotheses.

Following Fairfield et al. (2009) and Hawawini et al. (2003) the dataset is classified into ten different industry classes by the 3-digit US Standard Industrial Classification (SIC) system to account for industry specific effects in the analyses. The database is constructed incorporating (1) measures of excess returns, (2) various dimensions of the meta-environment, and (3) industry dummy variables to account for industry-specific effects. Data to construct the measures of excess returns and industry dummy variables were extracted from the Reuters Datastream and the Bureau van Dijk Orbis database (Orbis) and the OECD Tax Table. Furthermore, the various dimensions of the meta-environment were constructed based upon data from Orbis, World Bank, ITU and Heritage Foundation. These data sources have frequently been used in institutional research (De Jong et al., 2011).

The sample is restricted to the years 2004-2013 as the data on subsidiary locations is not available over a longer period of time. The data on subsidiary locations was extracted from Orbis, which was required to construct the meta-environments for each MNE. Furthermore, in line with common practice, only non-financial firms were included in the database, as financial firms’ financial assets and liabilities are operating assets and liabilities. In line with Healy et al. (2014) also firms with missing observations for two consecutive years are excluded from the sample. Lastly, outliers are removed from the sample as these may bias the results. Firms are removed by truncating the dataset.

Observations below the 1

and above the 99

percentile are removed from the sample. These outliers include all firms with negative cost of capital, which are economically unrealistic. Finally, 140 firms with complete observations remain in the final sample.

3.2 Model and variable construction 3.2.1 Model construction

The effects of MNEs’ meta-environments on the mean reversion of excess returns are tested by employing the economy-wide model, equation (1) and the industry-specific model, equation (2):

(𝑟

− 𝑘

) = 𝛼

+ 𝛽

(𝑟

− 𝑘

) + ∑ 𝛾

[(𝑟

− 𝑘

) × 𝑀𝐸] + 𝜀

(1)

(𝑟

− 𝑘

) = 𝛼

+ 𝛽

(𝑟

− 𝑘

) + ∑ 𝛾

[(𝑟

− 𝑘

) × 𝑀𝐸] + ∑ 𝛿

𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦

+ (2)

Where r

is the return on invested capital (ROIC) of company i in year t, k

is the cost of

capital measured by the weighted average cost of capital (WACC) of company i in year t, r

is the

15 ROIC of company i in year t-1, k

is the WACC of company i in year t-1, ME is a vector of the various meta-environment dimensions, and industry

is a vector of industry dummy variables.

Equation (1) and (2) are derived from the model of Fairfield et al. (2009) testing the mean reversion of firm profitability both on economy-wide and industry-specific level, the model of De Jong et al.

(2011) testing the effects of MNEs meta-environments on firm performance and the model of Healy et al. (2014) testing the effects of home-country environment effects on accounting profits.

The interaction term γ

denotes the term, which tests the main predictions of the various meta- environment dimensions slowing down the convergence of excess returns. In other words, the interaction coefficient γ

shows the effect of meta-environment dimensions on the decay factor of excess returns. Moreover, β

denotes the general coefficient to determine the excess returns’ decay factor. Additionally, the model includes a constant α. When α is significant this could indicate MNEs’

excess returns converge to a long-run level of excess return instead of to the WACC as was indicated by Mauboussin and Johnson (1997). The industry dummy variables account for industry-specific effects, whereas the magnitude of their effect is denoted by coefficient δ

.

The models account only for the moderating effects of the meta-environment denoted by the

interaction effects of the various meta-environment dimensions with the lagged excess return variable,

because this paper is only interested in the influence of the meta-environment on the decay rate of

excess returns. Furthermore, there are several advantages to the model employed in this paper in

comparison to models that were previously employed in literature related to the convergence of excess

returns. Firstly, this model takes a varying cost of capital into account; the WACC changes year-by-

year for each individual company. Thus, the model does not assume a constant capital structure, which

is one of the drawbacks when you keep the WACC constant. A constant capital structure may for

example under- or overestimate the interest tax shield incorporated in the WACC, which represents a

part of the value of a firm. (Koller et al., 2010). Nissim and Penman (2001) only take a constant cost

of capital into account, whereas Koller et al. (2010) implicitly takes a constant cost of capital into

account throughout the whole sample. Moreover, Fairfield et al. (2009) did not take the cost of capital

into account at all. Secondly, these models complement Fairfield et al. (2009) including both an

economy-wide and an industry-specific model. However, many other authors did not take both models

into account simultaneously. Thirdly, the model employs a value-based measure of performance

instead of using accounting ratios (such as ROA). ROIC minus WACC is also clearly linked to the

value-driver formula of Koller et al. (2010) as was previously defined. If the ROIC exceeds the

WACC a firm creates value. The opposite is true when the ROIC is smaller than the WACC.

16 3.2.2 Variable construction

All variables employed in models (1) and (2) are presented in table 1. An overview of all raw inputs to construct the variables in this paper (including the various dimensions of the meta-environment) and its corresponding data sources are presented in Appendix B.

Table 1: Variables and measures

Variable Measure Data source(s)

Excess returns 𝑟

− 𝑘

Reuters Datastream, Orbis,

OECD Tax Table

Technological capability 𝑇𝐶

= ∑ (

𝑛_𝑖

𝑇𝐶

)

𝑗

World Bank, ITU

Infrastructure 𝐼𝐹

= ∑ (

𝑛_𝑖

𝐼𝐹

)

𝑗

World Bank

Investment promotion policies 𝐼𝑃

= ∑ (

𝑛_𝑖

𝐼𝑃

)

𝑗

Heritage Foundation

International trade promotion

policies 𝐼𝑇

= ∑ (

𝑛_𝑖

𝐼𝑇

)

𝑗

Heritage Foundation

Law enforcement system 𝐿𝐸

= ∑ (

𝑛_𝑖

𝐿𝐸

)

𝑗

World Bank

MNE industry Dummy variables per industry Orbis

Note: i denotes the company, t the year and j the country.

The MNE’s excess return measure used in this model contains two main components the ROIC and the WACC (see equation 3).

𝐸𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛

= 𝑅𝑂𝐼𝐶

− 𝑊𝐴𝐶𝐶

(3)

The excess returns in this model is a return spread which represents the ability of the MNE to create value per dollar of invested capital (IC). By multiplying the excess return in equation (3) by the IC the economic profit of a company can be obtained (Hawawini et al., 2003). Furthermore, the ROIC, which is defined as the return a MNE earns per dollar invested in its business, is calculated as follows:

𝑅𝑂𝐼𝐶

=

𝑅)

𝐷_𝑖,𝑡 +𝐸_𝑖,𝑡

(4)

17 Where EBIT

denotes the earnings before interest and taxes of company i in year t and 𝑇

denotes the reported tax rate of company i in year t, D

represents the book value of debt minus cash or net debt of company i in year t and E

represents the sum of the book values of preferred stock and common shareholders equity or total shareholders’ equity of company i in year t. Moreover, the WACC is commonly defined as a measure of the risk of operations of firms (Dickinson & Sommers, 2012) and denotes the opportunity cost that investors incur for investing their funds in a particular firm instead of others with similar risk. Hence, the WACC is defined as the cost of capital that mixes the required rates of returns of both debt and equity holders and is calculated following Hawawini et al.

(2003):

𝑊𝐴𝐶𝐶

= d

× 𝑟

× (1 − 𝑇

) + e

× 𝑟

(5)

Where d

represents the share of debt based on the book value of net debt, r

represents the cost of debt of company i in year t and r

represents the cost of equity of company i in year t. e

represents the share of equity. In finance the share of equity is normally determined by the target capital structure of industry peers based on the market value of equity (i.e. Koller et al., 2010).

However, the sample of companies employed in this paper does not contain sufficient companies to conduct an adequate peer analysis to determine the target capital structure including market values of equity. The companies are classified according to the US SIC system, but the companies are not all comparable in terms of for example products, geographical scope, capital structure, margins etc.

Therefore, this paper follows Hawawini et al. (2003) and Shil (2009) who employed a capital structure based on book values of equity and debt based on the methodology and Economic Value Added (EVA) performance metric of Stewart (1991). Employing a capital structure based on book values of equity and debt also improves the comparability of this paper with Hawawini et al. (2003). Moreover, T

denotes the marginal tax rate, which is equal to 35 percent in the US

. The T

term is incorporated in the WACC to take into account the interest tax shield. The interest tax shield lowers the net cost of debt as the interest on debt is generally tax deductible (Koller et al., 2010). Additionally, the cost of debt is calculated as follows:

𝑟

= 𝑟

+ 𝜏

(6)

Where r

represents the risk-free rate in year t and τ

represents the credit spread of company i in year t. The risk-free rate is proxied by the 10 year US Treasury Bond rate in year t as the sample only includes US firms

. This is an appropriate estimation of the risk-free rate as it is a highly liquid, long-term government security in the same currency as used by the firms in the sample. Moreover, US

2 As all firms included in the sample are from the US, the US marginal tax rate is used, which was retrieved from the OECD Tax Tables.

3 The 10 year US Treasury Bond rates are extracted from the Federal Reserve website:

http://www.federalreserve.gov/releases/h15/data.htm

18 Treasury Bonds are considered to be default-free (Koller et al., 2010). The credit spread of company i in year t is determined based on credit spreads corresponding to credit ratings, which are determined based on the combination of data of Moody’s Aaa and Baa corporate bond credit ratings

and default spreads provided by Damodaran

. The default spreads provided by Damodaran correspond to a synthetic rating based on the interest coverage ratio of each company. The default spreads are used to estimate the corporate bond spreads of the other rating classes not provided on the Federal Reserve website. Finally, each company is rated in correspondence to its interest coverage ratio in year t

. Thus, the credit spreads, and thereby the cost of debt vary between companies and over years. The interest coverage ratio denotes the ability of a firm to pay interest on outstanding debt or to meet short- term obligations (Koller et al., 2010) and is calculated as follows

:

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑐𝑜𝑣𝑒𝑟𝑎𝑔𝑒 𝑟𝑎𝑡𝑖𝑜

=

𝑖_𝑖,𝑡

(7)

Where i

is the interest expense on debt of company i in year t. Furthermore, the cost of equity is calculated according to the CAPM model:

𝑟

= 𝑟

+ 𝛽

× 𝑀𝑅𝑃 (8)

Where 𝛽

is a Blume adjusted beta, which is a measure of systematic risk induced by company i in comparison to the market as a whole in year t, and MRP is the market risk premium calculated by the return of the market portfolio minus the risk-free rate. In this paper a MRP of 5 percent is employed following Koller et al. (2014) and the implied MRP of the S&P 500

. In addition, a robustness test is executed using other values for the MRP. A Blume adjusted beta is used in this paper as betas tend to regress to the mean value of all betas over time, which is one. Thus, extreme values of the beta in a period tend to have less extreme values in the next period (Blume, 1975). The Blume adjusted beta is calculated in a few steps. First, the raw beta needs to be calculated:

𝛽

=

𝑀𝑡)

(9)

Where 𝐶𝑜𝑣(𝑟

, 𝑟

) represents the covariance between the monthly returns of company i and the monthly returns of the market portfolio, which is in this case the S&P 500, and 𝜎

(𝑟

)

4 The Moody’s corporate bond ratings are extracted from the Federal Reserve website:

http://www.federalreserve.gov/releases/h15/data.htm

5 The default spreads corresponding to synthetic ratings are extracted from the website:

http://people.stern.nyu.edu/adamodar/

6 http://people.stern.nyu.edu/adamodar/

7 Alternative methods to calculate the interest coverage ratio are to replace EBIT by EBITA or EBITDA (Koller et al., 2010).

8 http://people.stern.nyu.edu/adamodar/

19 respresents the variance of the return of the S&P 500. Secondly, the beta is smoothed according to Blume (1975):

𝛽

= 0.343 + 0.677 × 𝛽

(10)

Furthermore, the meta-environments of all MNEs need to be constructed for the whole sample period. This paper follows the approach of De Jong et al. (2013) to construct the MNE meta- environments. Firstly, country-indices for each meta-environment dimension are constructed over the period of 2004-2013. The country-indices contain observations for 189 to 215 countries depending on the meta-environment dimension at hand. Whereas every country-environment reflects an opportunity set for a MNE.

The country indices of the technological capability of each country is constructed following an approach of Archibugi and Coco (2004). Their index contains elements of technology creation (i.e. the degree of knowledge creation and innovativeness), technological infrastructures (strongly associated with production knowledge) and the development of human skills (complementing the first two dimensions of technological capabilities). However, the data for all inputs of the index of Archibugi and Coco (2004) is not available for 2004-2013 for a large amount of countries. Therefore, a part of this index, the telephone penetration, is used as a proxy to measure technological capability as it has a high correlation with the overall index of approximately 90 percent. Telephone concentration consists of the sum of the number of main telephone lines per 1000 people (log) and the number of mobile telephones per 1000 people (log). In line with Archibugi and Coco (2004) the degree of telephone concentration was benchmarked against the OECD average as follows:

𝑇𝐶_𝑗,𝑡

=

ln (𝑂𝐸𝐶𝐷 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑜𝑓 𝑡𝑒𝑙𝑒𝑝ℎ𝑜𝑛𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛)_𝑗,𝑡

(11)

Where TC

is the value of the technological capability index of country j in year t.

Furthermore, data measuring the quality of infrastructure was limited in terms of data that explicitly encompasses transport facilities (i.e. roads, ports and airports), communication facilities (i.e. network and information infrastructure) and energy availability. Therefore, the country indices of the infrastructure of each country is constructed using the amount of gross capital formation (% of GDP) of each country as a proxy as was previously done by i.e. Asiedu (2004). This proxy also explicitly incorporates investments in important infrastructure components such as roads and railways.

Furthermore, following De Jong et al. (2011) investment promotion policy (IP

) of countries

is proxied by the Investment Freedom Index of the Heritage Foundation. The index ranges from 0 to

100, whereas a high value indicates a more favourable investment promotion policy in a country for

the MNE. Similarly, trade promotion policy (TP

) is measured by the Trade Freedom Index of the

Heritage Foundation. The index ranges from 0 to 100, whereas a high value indicates a more open

trade policy in a country. Lastly, the law enforcement system (LE

) is also measured in line with De

20 Jong et al. (2011). Thus, the effectiveness of the law enforcement is measured by the rule-of-law index of the World Bank. The index measures for example the quality of enforcement of contracts, property rights protection and courts and ranges from -2.5 to 2.5.

Subsequently, the different country-indices are used as inputs to construct the meta- environments of the MNEs for 2004-2013. In general, the values of the meta-environments for company i in year t are established based on the weighted average of the country-index values. The weights are assigned to the country-indices based upon the number of subsidiaries present in that country relative to the total number of subsidiaries

(also see table 1).

3.2.3 Industry dummies

The industry-specific model includes industry dummy variables to account for industry-specific effects. Previous literature has shown industry effects may influence the convergence of firms’ excess returns (i.e. Koller et al., 2010; Fairfield et al., 2009; Hawawini et al, 2003). Additionally, Dickinson and Sommers (2012) and Cheng (2005) control for industry-specific effects on excess returns as well.

The industry-specific effects may stem from concepts such as industry-specific entry-barriers (i.e.

capital intensity or innovation) and industry-specific economies of scale. The model includes industries classified according to the US SIC system. However, the sample does not contain any firms within the public administration industry. Moreover, the agriculture, forestry, or fishing industry is considered as the base case in the model and is thus not included as a dummy variable.

3.1.2 Descriptive statistics

The correlation matrix of the variables employed in the econometric models are shown in table 2.

According to inspection of this table there are no signs of potential multicollinearity, except between trade promotion policies and investment promotion policies. Moreover, as the econometric model in this paper involves the interaction terms between the lagged excess return variable, (r-k)

, and the variables of the meta-environment dimensions, more multicollinearity problems can be expected. The interaction variables show correlation coefficients higher than 90 percent with the lagged excess return variable. Multicollinearity may lead to errors in statistical inference. The biased variable coefficients in the form of their different weights and signs, as well as influencing the significance of the results by affecting the standard errors in the model. This makes it hard to interpret the results. To solve the multicollinearity problem the independent variables used in the interaction terms are centered by subtracting the firm-specific (within group) mean values of the observed values of the independent variables. Centering variables enhances the interpretability of coefficients and reduces the numerical

9 An alternative and more appropriate weighting method would be to use the sales of company i in countryj relative to the total sales of company i. In this way there is generally a clearer link to where the company generates cash flows. Moreover, the amount of investments of company i in country j relative to the total investments of company i could be used as a weighting measure. However, data is not available for both methods.