The effect of the degree of internationalization on the convergence and permanence of excess returns

1

The effect of the degree of internationalization on the

convergence and permanence of excess returns

This paper examines the effect of the degree of internationalization of firms on the convergence and permanence of excess returns. Excess returns converge to a long-term level as a result of competition. Internationalization can affect the pace by which excess returns converge and the level it eventually reaches. The degree by which internationalization, measured by a firm’s relative share of foreign sales and foreign assets, affects excess returns is analyzed in an economy-wide and an industry-specific model. Both models show a convergence of excess returns over time to positive long-term excess return. The economy-wide model delivers strong evidence that internationalization speeds up the convergence of excess returns to long-term excess returns. The industry-specific model, however, generates inconclusive results about the effect of the degree of internationalization on the convergence and permanence of excess returns. These findings are based using a sample of 3,187 firms listed on US stock exchanges in the period 1998-2014.

Field Keywords: valuation, convergence, excess returns, long-term excess returns,

internationalization.

Data availability: data is available from the sources listed in the paper.

2 1. Introduction In literature excess returns are defined as the return on capital in excess of the cost of capital (Damodaran, 2007; Jacobsen, 1988; Koller et al., 2010). Markets, in which firms face high levels of excess returns, tend to attract competitors as a consequence of which downward pressure on excess returns emerges (Damodaran, 2007). Competition is therefore a factor that makes excess returns converge to long-term excess returns (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Fairfield et al., 2009; Healy et al., 2014; Mueller, 1977).

Factors that shield firms from competition (and thus create a competitive advantage) can slow down the speed of convergence and increase the level of long-term excess returns (Jacobsen, 1988). The existing literature examines firm, industry and country specific factors that may be of influence to the competitive position of a firm. Factors that influence competition at firm level are, among others, the possession of unique resources, the size of the firm and entry barriers at firm level (Cheng, 2005; Dickinson and Sommer, 2012; Mueller, 1977; Porter, 1980; 1985). At industry level, factors that influence competition are, among others, industry barriers of entry and industry economics of scale (Bou and Satorra, 2007; Cheng, 2005; Fairfield et al., 2009). At a country level, differences between countries in the labour market, capital market and product market can influence competition (Healy et al., 2014). Previous research provides evidence that competitive factors affect both the speed of convergence and the existence of long-term excess returns (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Fairfield et al., 2009; Healy et al., 2014; Mueller, 1977).

This paper studies the degree of internationalization as a factor creating a competitive advantage. Theories of internationalization claim that internationalization enhances value and creates competitive advantages due to the unique value of resources for internationalization purposes, the lower risk as a consequence of international diversification and cost advantages resulting from the possibility of tax avoidance, and lower cost of production inputs. However, internationalization can also create a competitive disadvantage due to agency problems (Morck and Yeung, 1991) and barriers-to-entry.

3

4

There are five main findings, based on both the economy-wide model (including all firms) and industry-specific model.

First, in the economy-wide model and industry-specific model, excess returns converge over time to long-term excess returns, which is in line with prior research (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Fairfield et al., 2009; Healy et al., 2014; Mueller, 1977).

Second, as described in literature, both models show that long-term excess returns exist (Bou and Satorra, 2007; Cheng, 2005).

Third, growth in net operating assets is negatively related to future excess returns. This can be due to two reasons: firms first initiate their most profitable investments, and pursue less profitable investments at a later stage and/or investments become more profitable over time (Fairfield et al., 2003). Fourth, the results in the economy-wide model show that internationalization, as measured by the relative share of foreign sales and assets, speeds up the convergence of excess returns to long-term excess returns. This negative effect of internationalization can be due to agency costs or trade barriers. However, there are no significant results for the effect of internationalization on long-term excess return. Fifth, the industry-specific models yield inconclusive results. Internationalization in the manufacturing sector speeds up the convergence of excess return, while in the personal and business services industry it slows down the speed of convergence. For the mining industry there is a positive effect of the relative share of foreign sales on long-term excess returns however a negative effect of the relative share of foreign assets.

In conclusion, the findings regarding the effects of internationalization are mixed. These findings, proven to be robust against different assumptions, are important for management and valuation practitioners.

5

Valuation practitioners should, when analyzing, forecast and valuing the profitability of the firm, take into account the effect of competition (Dickinson and Sommers, 2012) and internationalization on excess return. Ignoring these effects can make valuation practitioners overvalue a firm. Selling an overvalued firm is a disadvantage for the buyer. However, for the other parties involved (seller and stockholders) this undervaluation is beneficial.

This paper continues as follows. Section 2 gives an overview of the relevant literature on the basis of which hypotheses are developed. Section 3 explains how the sample of firms is selected, the model specified and the variables are measured. Section 3 also provides the descriptive statistics and describes the research method applied. Section 4 presents the results on the basis of which the hypotheses are accepted or rejected. Section 5 gives the conclusions and several limitations of the outcomes. This section also provides options for further research.

2. Literature overview and hypotheses development

This section reviews literature on the realization of competitive advantages, convergence of excess returns, permanence of excess returns, growth in net operating assets, and control variables. Furthermore, this section develops hypotheses about the possible effect of the degree of internationalization on the convergence and permanence of excess return.

6 2.1 Literature review Competitive advantage According to Porter (1980; 1985) there are five forces that determine the level of competition between firms. Porter’s five forces model gives insight into the strength of the competitive position of a firm by taking into account the following forces: threat of new entrants, threat of substitutes, bargaining power of customers, bargaining power of suppliers and industry rivalry. First, the threat of new entrants measures the easiness for competitors to enter the market. The threat of new entrants is influenced by among others barriers to entry, government policy and economies of scale. Second, the threat of substitutes explains the possibility that other products or services can replace the firm’s offerings. Switching costs can influence the threat of substitutes. If it is expensive to switch to another product or service, the threat of substitutes is low. Third, the bargaining power of customers explains the dependency of customers on the services or products of the firm, which can among others also be explained by switching costs. Fourth, the bargaining power of suppliers measures the ease of switching from supplier, which can depend of the uniqueness of the materials the supplier delivers. Fifth, industry rivalry explains the intensity of the competition within the industry. This can be explained by among others the degree of transparency within the industry and the number of firms within the industry (industry concentration).

Porter’s five forces are being reflected in both the industry view and firm-efficiency view on how firms may gain a competitive advantage (McGahan and Porter, 1999).

The industry view claims that firms create a competitive advantage due to industry characteristics such as industry economies of scale and barriers-to-entry, which also impacts industry concentration.

7

substitutes available (Barney, 1991). According to the RBV this also holds for tangible assets.

Moreover, Porter (1980, 1985) claims that in order to sustain a competitive advantage there are two important factors: low costs and differentiation. Low costs can result from for instance economies of scale. In order to differentiate the firm from competitors, Porter and Millar (1985) suggest that a firm can differentiate their scope by exploring new segments and industries, by vertical integration and by geographic differentiation. Firms that serve more segments, industries or geographic areas, create a competitive advantage over their domestic rivals. Vertical integration, where firms take on more activities internally instead of outsourcing, can create a competitive advantage due to the independency from suppliers. Moreover, product differentiation, due to innovation, can also yield a competitive advantage. Innovation, through research and development (R&D), can create superior products that increase barriers-to-entry for new entrants (Asthana and Zhang, 2006; Cheng, 2005; Dickinson and Sommers, 2012).

8 Table 1 Literature overview (1) (2) (3) (4) (5) (6) (7) (8) Research Convergence / permanence Factors that influence excess returns Speed of convergence Firm level control variables Cost of capital Growth in NOA Main findings Research focussing on the convergence of excess returns Dechow et al. (1999) Convergence Firm factors: Quality of the earnings, non-recurring special items, extreme levels of operating accruals and dividend policy. Industry factors: Industry structure 0.62 - 12% (Long-run return US equities) no Convergence is due to firm characteristics. Convergence slows down due to quality of earnings and increasing dividend payout. Convergence of firms within the same industry is correlated. Dickinson and Sommers (2012) Convergence Firm factors: Economies of scale, product differentiation, innovation (R&D), capital requirements, power of suppliers, power of customers, credible threat of new entrants/expansion existing competitor. Industry: 0.79; Firm: 0.78 Firm age and firm size 7.1% (WACC for portfolios of firms based on book-to-market values and market value of operations) yes The power of suppliers and credible threat of expected retaliation increase the persistence of excess returns. Fairfield et al. (2009) Convergence Industry factors: Product demand, industry barriers to entry and business risk

9

Mueller (1977) Convergence Firm factors:

10 Convergence of excess returns Table 1 shows that the convergence of excess returns is well documented in the existing literature (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Fairfield et al., 2009; Healy et al., 2014; Mueller, 1977). All research concludes that due to competition excess returns tend to convergence over time. If the entry to and exit from a market is relatively easy than excess returns converge quickly (Mueller, 1977). However, factors that create a competitive advantage slow down the speed of convergence. According to the third column of table 1, three competitive factors influence the speed of convergence: firm-specific factors (Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Mueller, 1977), industry-specific factors (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Fairfield et al., 2009) and country-specific factors (Healy et al., 2014).

At firm level, several factors can influence the convergence of excess returns (Cheng, 2005; Dechow et al., 1999; Dickinson and Sommers, 2012; Mueller, 1977). Cheng (2005) and Mueller (1977) analyse the competitive advantage of two of Porter’s five forces: barriers-to-entry and product differentiation. Cheng shows that firm-level barriers to entry lower the speed of convergence of excess returns. More recently, Dickinson and Sommers (2012) extend the literature by taking into account barriers to entry, product differentiation and other competitive factors of the five forces model. They show that the power over suppliers and the ability to signal the threat of new entrants or the expansion of existing competition decreases the speed of convergence.

At industry level, the structure of the industry may slow down the convergence of excess returns (Bou and Satorra, 2007; Cheng, 2005; Dechow et al., 1999; Fairfield et al., 2009). This is in line with the literature about competition predicting that firms can yield a competitive advantage within an industry due to the industry structure (McGahan and Porter, 1999; Porter, 1980; 1985). Table 1 shows that Bou and Satorra (2007), Cheng (2005) and Dechow et al. (1999) conclude that industry excess returns increase with the industry concentration and industry barriers to entry.

11

enforcement. They show that country-level factors impact competition between firms in different countries and therefore the speed of convergence of excess returns. They additionally find that multinationals face lower home country competition and thus have a competitive advantage over their domestic competitors. Firms are classified as multinationals, when having 50% of the sales or assets outside abroad. The results present a slower convergence of ROA for multinationals due to the fact that multinationals operate in more countries.

To summarize, the speed of convergence of excess returns to long-term excess returns is influenced by firm, industry and country level factors.

Permanence of excess returns / existence of long-term excess returns

Few research examines the existence of long-term excess returns (see table 1). Economic theory suggests that over time excess returns erode completely due to competition. However, literature shows that long-term excess returns can exist due to a sustainable competitive advantage at firm and/or industry level (Bou and Satorra, 2007; Cheng, 2005).

Bou and Satorra (2007) study long-term excess returns using a database of Spanish firms. They find that long-term excess returns exist at industry and firm level. They claim that at industry level, industry barriers to entry cause different levels of long-term excess returns. In addition, Cheng (2005) shows that besides industry barriers to entry, industry concentration is also a variable that can explain the level of long-term excess returns.

At firm level, the observed existence of long-term excess returns implies that firms are heterogeneous (Bou and Satorra, 2007; Cheng, 2005). Different resource endowments and different levels of efficiency among firms can explain the heterogeneity (Bou and Satorra, 2007). Resources create, according to the RBV, a competitive advantage (Wernerfelt, 1984).

12

of the firm. Second, firm level barriers reduce competition and therefore explain long-term excess returns (Porter 1980; 1985). Third, market share and firm size are considered to have a positive impact, which is explained in the control variables section. In conclusion, the level of long-term excess returns is influenced by firm and industry factors that can create a sustainable competitive advantage.

Growth in net operating assets

In line with the comparable literature, this paper takes into account growth in net operating assets (NOA). Future excess returns depend on growth in NOA and current excess returns (Ohlson, 1995). The seventh column of table 1 shows that Dickinson and Sommers (2012) and Fairfield et al. (2009) take growth in NOA into account. Growth in NOA is an indicator for the level of net investments. Net investments can create a stream of future profits (Miller and Modigliani, 1961; Palepu et al., 2004; Schultze, 2005) and therefore impact excess returns positively. Conversely, more recent literature (Dickinson and Sommers, 2012; Fairfield et al., 2003; Fairfield et al., 2009; Fairfield and Yohn, 2001) shows that growth in NOA negatively impacts the excess returns one year ahead. The literature gives two explanations for the negative impact: conservative accounting principles and diminishing marginal returns (Fairfield et al., 2003). Conservative accounting factors explain that investments lead to lower profits one year ahead while the investments increase profitability after several years. Diminishing marginal returns of investments can be explained by assuming that firms undertake their most profitable investments first, and exploit the less profitable investments later.

Control variables

13

due to the bureaucratic structure within large organization. This bureaucratic structure can create problems within large organizations that results in suboptimal performance and therefore in a competitive disadvantage. On the other hand, Cheng (2005) claims that firm size is positively correlated with excess returns. Large firms can profit from the economies of scale in the product and financing market leading to a competitive advantage.

Research shows that market share is positively related to excess returns due to the fact that firms with a large market share can profit from economies of scale and bargaining power (Cheng, 2005; Healy et al., 2014). According to Porter (1985), economies of scale and bargaining power increase the competitive advantage over competitors, which leads to higher excess returns.

2.2 Hypotheses development Influence of internationalization

According to the literature, the level of competition influences the convergence of excess returns and existence of long-term excess returns. Competitive advantages can slow down the speed of convergence of excess returns and sustainable competitive advantages increase the level of long-term excess returns. The following literature describes the reasons behind (sustainable) competitive advantages or disadvantages of multinationals.

14

Second, according to Porter and Millar (1985), geographic expansion can create a competitive advantage. Multinationals face when going international lower risks compared to their domestic competitors, due to international diversification (Rugman, 1976). This lower risks, result in lower a cost of capital. Excess returns increase if the cost of capital decreases, and therefore increases long-term excess returns (Morck and Yeung, 1991; Rugman, 1976).

Third, in line with the RBV, firms can create a competitive advantage if the firm has certain resources (tangible and intangible assets) that may help firms internationalizing (Morck and Yeung, 1991; Riahi-Belkaoui, 1999; Wernerfelt, 1984). Resources that may be of competitive advantage when internationalize are superior production skills and managerial skills. If the resources that create a competitive advantage are valuable, rare, difficult to duplicate and not substitutable, the competitive advantage sustains (Barney, 1991). Fourth, trade barriers can make it difficult and costly to enter the foreign market (Healy et al., 2014; Porter, 1980; 1985). This can result in a competitive disadvantage for multinationals in the foreign market. Tariff and non-tariff barriers can be distinguished. Tariff barriers can be in the form of import duties. Non-tariffs barriers can be in the form of quotas, licensing, regulation or be cultural (Barkema et al., 1996).

In conclusion, the majority of the theories of internationalization (Healy et al., 2014; Morck and Yeung, 1991; Wernerfelt, 1984) predict that internationalization creates a competitive advantage.

Measurement of internationalization

15 determining the effect of internationalization, measured by the relative share of foreign sales and foreign assets. Hypotheses on the effect of the degree of internationalization on the convergence of excess returns As concluded from the literature, the level of competition influences the speed of convergence. Due to competitive advantages, the convergence of excess returns slows down. The literature shows that a competitive advantage can be created due to heterogeneous, unique resources and low cost. Firms that internationalize can create a competitive advantage if they have the right resources, for instance managerial skills. Low arising from cheap foreign production and tax avoidance can also create an advantage. Therefore, internationalization can slow down the convergence of excess returns. This results in the following two hypotheses: H1: A higher relative share of foreign sales slows down the speed of convergence of excess returns. H2: A higher relative share of foreign assets slows down the speed of convergence of excess returns. Hypotheses on the effect of the degree of internationalization on long-term excess returns As concluded from the literature, the sustainability of the competitive advantage influences the level of long-term excess returns. If the competitive advantage is sustainable, it increases the level of long-term excess returns. The literature shows that valuable, rare, imperfectly imitable and non-substitutable resources and low costs can create a sustainable competitive advantage. Firms that internationalize can, if they have the right sustainable resources (that valuable, rare, imperfectly imitable and non-substitutable), gain from lower costs due to cheap foreign production and tax avoidance. Moreover, international diversification lowers risks and therefore increases the long-term level of excess returns. Therefore, internationalization creates a sustainable competitive advantage and increases the level of long-term excess returns. This results in the following two hypotheses:

16 H3: A higher relative share of foreign sales increases the level of long-term excess returns. H4: A higher relative share of foreign assets increases the level of long-term excess returns. 3. Research Design 3.1 Sample selection The data used in this research is obtained from the S&P Capital IQ database. In order to be able to create a large sample, data from firms listed on the New York Stock Exchange (NYSE), the American Stock Exchange (AMEX) and the National Association of Securities Dealers Automated Quotations (NASDAQ) is collected for the years 1998 - 2014. The choice for this time period depends on data availability. Before 1998, data availability was limited for the variables foreign sales and foreign assets. This creates a first sample of 5256 firms. Firms that meet the following criteria are included in the sample:

- Firms having a valid exchange ticker. This criterion excludes 32 firms, resulting in a sample of 5224 firms.

- Firms having values for at least one year over the sample period for the items growth in net operating assets and excess returns. This criterion reduces the sample to 4743 firms.

- Firms being non-financial institutions and non-utility firms (firms not in SIC1

codes 4900-4999 and 6000-6999). Financial institutions and utility firms are excluded since firms in these industries face different risks and their operations are different from other firms (Nwaeze, 2000). This final criterion creates the final sample of 3187 listed firms.

In order to determine the industry effect on the convergence and permanence of excess returns, firms are classified using the four-digit SIC codes. The final sample includes only 12 firms in the agriculture, forestry and fishing industry (SIC codes 0-999) and 4 in the public administration industry (9000-9999). Due to this underrepresentation of firms, these two industries are excluded in the industry-specific model.

17 In principle, three problems can occur within the sample of selected firms. First, firms can report their income and balance sheet in different currencies. This problem is solved by S&P capital IQ reporting all items in US dollars. Second, differences in fiscal years could create problems but by exporting the data to the end of each firms’ fiscal year avoids this problem. Third, differences in accounting systems can also create problems due to differences in financial reporting standards. However, since all firms are listed in the US, no differences in financial reporting standards exist.

In order to cope with the possible influence of outliers, this research follows previous research by not removing the outliers, but replacing them (Dechow et al., 1999; Dickinson and Sommers, 2011; Cheng, 2005; Healy et al., 2014; Fairfield et al., 2009). Data above the 95th _{percentile are replaced by the value of the 95}th _{percentile. Data}

below the 5th _{percentile are replaced by the value of the 5}th _{percentile. For the}

internationalization variables (foreign assets and foreign sales) only the values above the 95th _{percentile are considered to be outliers and thus replaced. Values below the 5}th

percentile mean that a firm does not have, or only has minimal, international exposure and cannot be considered to be an outlier.

3.2 Model specification

In line with previous studies (i.e. Dechow et al., 1999; Cheng, 2005; Riahi-Belkaoui, 1999), the model used in this paper is based on Ohlson’s (1995) and Feltham and Ohlson’s (1996) empirical valuation model in order to test the impact of the degree of internationalization on the convergence and permanence of excess returns. The start of this research is equation 3.12 , describing the convergence of excess returns to long-term excess returns: X_"#$% = (1 − δ)X_"#+ ε_"# (3.1)

Where Xit+1 is the excess return of firm i at time t+1; 1 − δ is the speed of

18

X_"#= 1 − T_0,"# × EBIT_"#− NOA_"#×WACC_; (3.2)

Where, Xit is equal to the excess return of firm i at time t; Te,it is the effective tax rate of

firm i at time t; EBITit are the earnings before interest and taxes of firm i at time t; NOAit

the net operating assets of firm i at time t and WACCj the weighted average cost of

capital in industry j.

In line with prior research (Fairfield et al., 2003; Fairfield et al., 2009; Fairfield and Yohn, 2001; Dickinson and Sommers, 2012) growth in net operating assets (NOA) is added to the empirical model since growth in NOA influences future excess returns (Ohlson, 1995). Please see appendix A for the derivation of the final model given in equation 3.3: X_"#$% NOA"#<% = δα + 1 − δ X_"# NOA"#<%+ i"#+ α − WACC; g"#+ 1 − δ X_"# NOA"#<%×FS"#+ αFS"# + 1 − δ X"#

NOA_"#<%×FA"#+ αFA"#+ ε"# 3.3

Where, Xit+1 is the excess return of firm i at time t+1; NOAit-1 are the net operating assets

of firm i at time t-1; 1 − δ is the speed of convergence, in which δ is a fixed non-negative parameter less than 1; a is the long-term excess return; iit is the return on net

investments; Xit is the excess return of firm i at time t; WACCj is the weighted average

cost of capital in industry j; git is growth in NOA of firm i at time t; FSit are the foreign

sales over total sales of firm i at time t; FAit are the foreign assets over total assets of firm

19

3.3 Variable measurement

Table 2 and 3 documents the relevant data. Table 2 contains an overview of all the raw data obtained from the S&P Capital IQ database. Table 3 contains the constructed variables.

Table 2 Raw data

Variable Symbol Remark Source

Accounting data

Book value net debt DBVit - Capital IQ [4364]

Book value total equity EBVit - Capital IQ [1275]

Earnings before interest and taxes

EBITit - Capital IQ [400]

Interest expense IEit - Capital IQ [82]

Market value total equity EMVit - Capital IQ

[IQ_MARKETCAP] Cost of capital Industry beta βi - Capital IQ [IQ_CUSTOM_BETA] Effective tax rate Te Income tax expense divided by EBT. Capital IQ [4376] Corporate tax rate Tc 35% Ibbotson (2015) Market risk premium MRP 5% Koller et al. (2012) Risk-free rate rf Average 10year US government bonds Oxford economics International variables Locations per geographic Segments Lgeoit - Capital IQ [IQ_GEO_SEG_NAME] Headquarter country HQ - Capital IQ [IQ_COUNTRY_NAME] Revenue per geographic segment

Rgeoit Excludes inter-segment revenues Capital IQ [3515] Assets per geographic

segment

Ageoit - Capital IQ [3510]

Notes: i: of firm i; t: end fiscal year t; j: per industry

20

Table 3 Constructed variables

Variable Symbol Formula

Accounting data

Net investments I"# I"#= NOA"#− NOA"#<%

Growth in net operating assets g"# _g "#= NOA"#− NOA"#<% NOA"#<% = I"# NOA"#<%

Net operating assets NOAit NOA"#= EBV"#+ DBV"#

Excess returns at time t+1 divided by NOA t-1

X"#$%

X"#$%=

1 − T0,"#$% × EBIT"#$%− NOA"#×WACC;

NOA"#<%

Excess returns divided by NOA t-1

X"#

X"#=

1 − T0,"# × EBIT"#− NOA"#<%×WACC;

NOA"#<% Cost of capital After tax cost of debt cDj cG;= (rI+ cJ)(1 − TK) Interest coverage ratio per firm over time cov","# _cov ","#= EBIT"# IE"# Interest coverage ratio per firm covi,i cov","= median(cov","#)

Interest coverage ratio per industry

covi,j cov_",;= median cov_","∈;

Beta per firm β"= median(β"#) Unlevered beta βU," _β U,"= β" 1 + D_E "

Unlevered industry beta Βu,j βU,;= median(βU," ∈ ;)

21 Table 3 Constructed variable (continued) International variables Revenue headquarter (HQ) country and production country RZ[,"# RZ[,"#= RZ[,"#+ R\],"# Revenue in country with production R\],"# _R \],"#= R^0_,"#if(A^0_,,"#> 0 and ≠ HQ) Total assets _A "#= A^0_,,"# Total revenues _R "#= R^0_,"# Assets headquarter country AZ[,"# AZ[,"#= if (L^0_,"#= HQ) Revenue headquarter country RZ[,"# RZ[,"#= if (L^0_,"#= HQ) Relative share of foreign sales FS"# _FS "#= R"#− RZ[,"# R"# Relative share of foreign assets FA"# _FA "#= A"#− AZ[,"# A"# Notes: i: of firm i; t: end fiscal year t; j: per industry 3.3.1 Growth in net operating assets

As presented in table 3, growth in net operating assets is measured by the difference between the net operating assets at year t and year t-1 divided by net operating assets of year t-1. The net operating assets are equal to the sum of the book value of net debt and the book value of total equity _{(Koller et al., 2010).} 3.3.2 Excess returns Table 3 and appendix A show how excess returns for firm i at time t and at time t+1 are determined. Excess returns are calculated using the following four variables: the effective tax rate, earnings before interest and taxes (EBIT), net operating assets (NOA) and the weighted average cost of capital (WACC). The determination of the effective tax rate, EBIT and NOA is clear from table 2 and 3. However, for the determination of the WACC, several assumptions are made.

22

assumption by taking a time-dependent cost of equity per industry, based on Fama and French (1997). Fama and French (1997) claim that the cost of equity per industry is more precise than the cost of equity per firm. In line with Cheng (2005) and Fama and French (1997), this paper uses a WACC per industry. To summarize, a time-independent WACC per industry is used in order to calculate the levels of excess returns. Table 4 presents the WACC and related information for all firms and per industry, classified by SIC codes as the first column shows. The second and third column present the name of the industry and number of firms respectively. Table 4 WACC for all firms and per industry (1) (2) (3) (4) (5) (6) (7)

Sic code Industry Firms Cost of _equity Credit _spread Debt-to -_equity WACC

23

The second item to determine, on behalf of the WACC, is the cost of debt. According to table 3, the after tax cost of debt requires estimations of a risk-free rate (same as used to calculate the cost of equity), a tax rate and a credit spread. The corporate tax rate for WACC purposes of the United States equals 35% (Ibbotson, 2015). The fifth column of table 4 presents the credit spread for all firms and per industry. The interest coverage ratio determines the credit spread. A higher interest coverage ratio means that for the firm the risk of getting into difficulties repaying its short-term debt is higher than for firms with a lower interest coverage ratio (Koller et al., 2010). Table 2 and 3 show the determination of the interest coverage ratio. The coverage ratio per firm equals the median of the coverage ratios of the years 1998-2014.

Damodaran3 _{provides a table to translate the interest coverage ratio into a credit}

spread. However, the information to determine the credit spread, is only available for the year 2016 for firms located in the United States. On the one hand, this information is applicable to the sample used in this research, since the sample consists out of firms headquartered in the US. On the other hand, a problem occurs due to the fact that the coverage ratio per firm used in this paper is based on the coverage ratio for the years 1998-2014. However, credit spreads are relatively stable over the period 1990-2014 (Callegan et al., 2015, pg. 248) and therefore I assume that using the credit spread of 2016 for determining the 1998-2014 average coverage ratio does not cause large differences. According to Callegan et al. (2015), pg. 248, credit spreads at the time of the financial crisis increased enormously. Therefore a robustness test at the end of this paper, takes into account the possible effects of the financial crisis.

24 The final column of table 4 shows the WACC per industry and for all firms. The table presents that the WACC per industry ranges from 7.9% for the transportation and wholesale and retail trade industry to 12.9% for the mining industry. 3.3.3 Internationalization variables To determine the impact of the degree of internationalization on excess returns two proxies are used: the relative share of foreign sales (foreign sales over total sales) and the relative share of foreign assets (foreign assets over total assets). Foreign sales The determination of foreign sales involves three steps. The first step is to obtain the sales per geographic segment and the headquarter country per firm. The second step involves the determination of total sales. Total sales are the sum of the sales in all geographic segments together. The third step considers that all foreign sales are all sales outside the headquarter country. Several firms report the United States and Canada as one geographic segment. If the headquarter of the firm is located in the United States or in Canada, foreign sales are the sales located outside the United States and Canada together. This is in line with Chen et al. (2005) estimating foreign revenue of firms headquartered in the US as the revenue outside the United States and Canada combined4_.

25

Foreign assets

In order to determine the amount of foreign assets, the assets per geographic segment and the headquarter country per firm are obtained. Total assets are the sum of the assets of all geographic segments together. Foreign assets are considered to be all assets outside the headquarter country. In line with the determination of foreign sales, foreign assets of firms headquartered in the US are all assets outside the United States and Canada. 3.3.4 Control variables As discussed in the literature review, the control variables are selected based on previous research. Please see appendix B for documentation of the data and the descriptive statistics of the control variables. The control variables are measured as follows: firm age is measured by the logarithm of the years a firm exists (Dickinson and Sommers, 2011); firm size is measured by the logarithm of the total assets of the firm (Cheng, 2005) and market share is measured by the total sales of the firm divided by the total sales of the industry the firm is operating in (Cheng, 2005; Healy et al., 2014).

In order to control for industry characteristics in the industry-specific model, dummy variables per industry are used. To overcome the dummy variable trap, one of the industries is used as base case (Brooks, 2014). This paper uses the manufacturing industry as a base case. 3.4 Descriptive statistics Table 5 and 6 report the descriptive statistics of the dependent and independent variables for the economy-wide model (all firms) and the industry-specific model (per industry) respectively. The total number of firm-observations equals 37,471. The first column of table 6 provides an overview of the number of firm-observations per industry.

26

due to export. The mean of the relative share of foreign assets equals 20%, meaning that 20% of the production is located outside the headquarter country.

The Jarque-Bera statistic makes clear whether the data follows a normal distribution. Table 5 shows high values for the Jarque-Bera statistic for all variables, meaning that the sample does not have a normal distribution. However, due to the large number of observations the non-normality can be assumed not to cause issues for the analysis.

Table 6 shows the descriptive statistics of the variables per industry. All of the descriptive statistics of the variables per industry are in line with table 5, except for the statistics of the excess returns in the construction and manufacturing industry. The construction industry shows a high mean of excess returns at time t+1 and t of 100% and 76% respectively. The manufacturing industry shows a low mean of excess returns at time t and t+1 of in both cases -3%.

Table 5 Descriptive statistics economy-wide model

Xit+1 Xit git FSit FAit

27

Table 6 Descriptive statistics industry-specific model

Industry Statistic Xit+1 Xit git FSit FAit

Mining Mean 0.01 0.00 0.25 0.05 0.18 (n=2661) Median 0.00 0.01 0.12 0.00 0.00 Maximum 0.58 0.41 1.75 0.68 1.00 Minimum -0.61 -0.52 -0.39 0.00 0.00 Std. Dev. 0.25 0.20 0.50 0.16 0.36 Construction Mean 1.00 0.76 0.16 0.09 0.17 (n=8151) Median 0.05 0.05 0.03 0.00 0.00 Maximum 18.25 13.89 2.97 1.00 1.00 Minimum -8.10 -6.57 -1.58 0.00 0.00 Std. Dev. 5.20 4.04 0.93 0.25 0.33 Manufacturing Mean -0.03 -0.03 0.14 0.19 0.26 (n=11761) Median 0.02 0.02 0.04 0.00 0.00 Maximum 0.84 0.72 1.39 1.00 1.00 Minimum -1.43 -1.31 -0.55 0.00 0.00 Std. Dev. 0.46 0.41 0.43 0.34 0.39 Transportation Mean 0.02 0.02 0.17 0.08 0.12 (n=2713) Median 0.00 0.00 0.05 0.00 0.00 Maximum 0.54 0.45 1.41 0.93 1.00 Minimum -0.47 -0.34 -0.41 0.00 0.00 Std. Dev. 0.21 0.17 0.42 0.22 0.30 Wholesale and Mean 0.08 0.07 0.12 0.01 0.11 retail trade Median 0.05 0.04 0.06 0.00 0.00 (n=4279) Maximum 0.52 0.46 0.87 0.20 1.00 Minimum -0.22 -0.19 -0.28 0.00 0.00 Std. Dev. 0.17 0.15 0.27 0.05 0.28 Personal and Mean 0.04 0.05 0.24 0.11 0.25 business services Median 0.01 0.01 0.05 0.00 0.00 (n=5921) Maximum 2.59 2.25 2.90 1.00 1.00 Minimum -2.48 -2.07 -1.05 0.00 0.00 Std. Dev. 0.98 0.84 0.85 0.27 0.40 Health, legal and Mean 0.06 0.03 0.18 0.04 0.16 education Median 0.04 0.03 0.06 0.00 0.00 (n=1738) Maximum 1.79 1.48 1.79 0.51 1.00 Minimum -1.93 -2.05 -0.71 0.00 0.00 Std. Dev. 0.70 0.66 0.56 0.12 0.34 Notes: the descriptive statistics without outliers on the variable used in the industry-specific model, based on observations in the period 1998-2014. The number of observations (n) per industry is indicated in the first column. Where, Xit+1 is the excess return for firm i at time t; Xit the excess return of firm i at time t; git growth in NOA of firm i at time t; FSit the relative share of foreign sales of firm i at time t and FAit the relative share of foreign assets of firm i at time t. See table 2 and 3 for the documentation of the data.

28 3.5 Research method In order to determine the appropriate research method, we need to consider the following. First, the sample is considered to be dated and unbalanced panel data since the sample includes observations of different firms (cross-sectional data) over time (time series data) and for each firm the dataset does not contain the same number of observations. Eviews does account for the missing observations, by excluding the firm from the sample for the years that it has missing observations (Brooks, 2014).

Second, it is important to establish whether a random effects model or a fixed effects model is more appropriate by using the Hausman and redundant fixed effect test. Random effects models give a good representation of the sample, when the sample is randomly selected from a population. Fixed effects models give a good representation of the data when the sample is a representation of the entire population, which is for instance the case when the sample includes all firms listed on an exchange (Brooks, 2014). The latter is the case in this paper and therefore a fixed effects model is seen as more appropriate. The p-values generated by both the Hausman and redundant fixed effect test are smaller than 0.05 implying that a cross-sectional and period fixed effects model as part of an OLS regression is appropriate for testing the hypotheses.

Third, multicollinearity, non-normality and serial correlation can create problems when analyzing the data. In order to detect multicollinearity, the correlation between the independent variables is analyzed. Please see appendix C for the correlation matrix. The lowest correlation (-0.07) is between foreign assets and foreign sales. The highest correlation (0.13) is between excess returns at time t and growth in NOA. According to Brooks (2014) these correlations are sufficiently small to conclude that multicollinearity is an issue. Non-normality, as already discussed in section 3.4 is considered not to be an issue. Serial correlation is not present as proved by Durbin-Watson statistics of around 2.

In sum, to determine the effect of the degree of internationalization on the convergence and permanence of excess, this research applies a cross-sectional and period fixed effect model as part of an OLS regression. In order to cope with possible heteroscedasticity, White diagonal standard errors haven been used when running the economy-wide model regression and the industry-specific model regression:

29 X"#$%= βg+ β%X"#+ βh (X"# × FS"#) + βi (X"# × FA"#) + βjg"#+ βk FS"#+ βlFA"#+ ε"# (3.4) X_;#$%= β_;,g+ β_;,%X_;#+ β_;,h (X_;# × FS_;#) + β_;,i (X_;# × FA_;#) + β_;,j g_;#+ β_;,k FS_;#+ β_;,lFA_;#+ ε_;# (3.5)

Where, Xi/jt+1 is the excess return of firm i/industry j at time t+1, Xi/jt is the excess return

of firm i/industry j at time t, FSi/jt are the foreign sales of firm i/industry j at time t,

Xi/jt*FSi/jt is the interaction term between the excess returns and the foreign sales, FAi/jt

are the foreign assets of firm i/industry j at time t, Xi/jt*FSi/jt is the interaction term

between excess return and foreign sales, gi/jt is growth in NOA of firm i/industry j at

time t, βX is the coefficient of the variable of firm i where x is a value between 0 and 6; βj,x is the coefficient of the variable of industry j where x is a value between 0 and 6 and ε_"/;# the error term of firm i/industry j at time t. The coefficient of each variable presents the following: β0 shows the coefficient of the long-term excess return; β1 shows the speed of convergence of the excess return to long-term excess returns; β2 shows the influence of foreign sales on the convergence of

excess returns to test hypothesis 1; β3 shows the influence of foreign assets on the

convergence of excess returns in order to test hypothesis 2; β4 shows the impact of

growth in NOA on future excess returns; β5

shows the impact of foreign sales on long-term excess returns in order to test hypothesis 3 and β6 shows the impact of foreign

sales on the long-term excess return in order to test hypothesis 4.

This research performs for both regression equations an univariate and multivariate analysis. The univariate analysis shows the impact of excess returns at time t on the independent variable. The multivariate analysis presents the impact of the dependent variables on the independent variable.

4. Results

30 Panel A of table 7 presents the multivariate results of the economy-wide model. The results in panel A are analyzed from the top to the bottom. First, table 7 shows the existence of long-term excess returns in line with Bou and Satorra (2007) and Cheng (2005). The intercept shows that the long-term excess returns equal 4% significant at a 99% confidence level. This is in line with table 5, showing that the median of the excess returns equals 3%.

Second, the results show a convergence of excess returns significant at a 99% confidence level and equal to 92%5_{. According to table 1, convergence of excess returns}

to long-term excess returns is in line with Bou and Satorra (2007), Cheng (2005), Dechow et al. (1999), Dickinson and Sommers (2012), Fairfield et al. (2009), Healy et al. (2014) and Mueller (1977). However, previous research finds a higher speed of convergence than the 92% mentioned above. Dechow et al. (1999) find a speed of convergence of 62%. More in line with the findings in this paper is Fairfield et al. (2009) with a speed of convergence between 83% and 85%.

Third, table 7 shows that growth in NOA equals -12% and denotes significance at a 99% confidence interval, which is in line with literature (Dickinson and Sommers, 2012; Fairfield et al., 2009; Fairfield et al., 2003; Fairfield and Yohn, 2001). This means that excess returns are negatively associated with growth in NOA of the previous year. This can be due to conservative accounting principles and diminishing marginal returns (Fairfield et al., 2003).

Fourth, the economy-wide model shows negative effects of the degree of internationalization on the convergence of excess return. The interaction term of foreign sales (-11%) and foreign assets (-6%) are both negative and significant at a 99% confidence level. This means that the relative share of foreign sales and foreign assets speed up the convergence of excess returns, rejecting hypotheses 1 and 2. The economy-wide model provides evidence for the agency theory. Due to agency costs or trade barriers, internationalization may negatively impact the profitability of the firm. Further, the analysis does not provide significant results, see appendix E, for the effect of the relative share of foreign sales and foreign assets on long-term excess returns, thereby rejecting hypotheses 3 and 4.

31 Table 7 Multivariate results excluding control variables Economy-wide model Industry-specific model

Mining Construction Manufacturing Transportation Wholesale and retail trade Personal and business services Health, legal and education

Panel A Panel B Panel C Panel D Panel E Panel F Panel G Panel H

Intercept 0.04*** 0.02*** 0.19*** 0.01*** 0.02*** 0.03*** 0.04*** 0.05*** (17.03) (3.34) (8.99) (4.20) (6.79) (15.08) (4.45) (5.19) X_"# 0.92*** 0.53*** 1.11*** 0.80*** 0.69*** 0.69*** 0.75*** 0.87*** (85.86) (11.91) (64.43) (39.65) (15.74) (23.49) (26.20) (25.26) g_"# -0.12*** -0.04*** -0.18*** -0.11*** -0.06*** -0.06*** -0.15*** -0.07*** (-14.52) (-3.04) (-3.61) (-11.25) (-5.75) (-6.20) (-8.23) (-2.97) FS_"# 0.06* (1.77) FS_"#× X_"# -0.11*** -0.10** (-3.48) (-2.43) FA_"# -0.03* (-1.31) FA_"#× X_"# -0.06*** -0.15*** 0.15*** (-2.36) (-3.72) (3.30) R2 0.73 0.54 0.87 0.68 0.61 0.66 0.62 0.76 Adj. R2 0.70 0.50 0.85 0.65 0.57 0.63 0.59 0.73 F-statistic _{29.06*** 11.58*** 60.63*** 25.62*** 15.72*** 21.21*** 15.79*** 30.54***} Observations 37471 2661 8151 11761 2713 4279 5921 1738 Notes: this table shows the multivariate results of a cross-section and period fixed effects OLS model. The corresponding univariate results are presented in appendix D and the results including control variables in appendix H. Panel A uses the following regression equation: X"#*+= β.+ β+X"#+ β0 (X"# × FS"#) + β3 (X"# × FA"#) + β4 g"#+ β5 FS"#+ β6FA"#+ ε"#.

Panels B-H use the following regression equation: X8#*+= β8,.+ β8,+X8#+ β8,0 (X8# × FS8#) + β",3 (X8# × FA8#) + β8,4 g"#+ β8,5 FS"#+ β8,6FA8#+ ε8#. The value within the parenthesis

32 4.1.2 Results from the industry-specific models Panels B to H present the multivariate results excluding control variables of the industry-specific models. From these panels we can conclude the following. First, the results show for all industries significant long-term excess returns at a 99% confidence level. All of the results are above 0%, meaning that long-term excess returns exist. These long-term excess returns can be the result of sustainable competitive advantages of industries (i.e. entry barriers and product differentiation). The construction industry (panel C) shows a high coefficient (19%) for the long-term excess returns significant at a 99% confidence level. This is in line with table 6 where the mean of excess returns for the construction industry is extremely high.

Second, panels B and D-H show that excess returns converge in line with the finding for the economy-wide model. All the results, relating to the convergence, denote significance at a 99% confidence level. Differences between industries exist in the speed by which excess returns converge, which can be caused by differences in industry structures (Cheng, 2005; Dechow et al., 1999; Fairfield et al., 2009). The mining industry (panel B) shows that excess return converge with a speed of 53%, while panel H shows a speed of convergence of 87% for the health, legal and education industry. Panel C shows that excess returns in the construction industry diverge instead of converge. This would imply that over time excess returns in the construction industry tend to grow. In principle, this is not a logical result but may possibly be caused by general limited competition in this industry due to government regulations. This divergence in panel C is in line with the descriptive statistics of the construction industry in table 6.

Third, for all industries excess returns are negatively correlated with growth in NOA. This is in line with the findings of the economy-wide model.

Fourth, panels B, D and G show mixed results for the effect of the degree of internationalization on the convergence and permanence of excess returns. Panel B, the mining industry, shows a positive effect (6%) of foreign sales on long-term excess returns significant at a 90% confidence level, accepting hypothesis 3. However, panel B shows a negative effect (-3%) of the relative share of foreign assets on the long-term excess returns significant at a 90% confidence level, rejecting hypothesis 4.

33

The results presented in panel G support hypothesis 2 as it shows a positive effect of foreign assets on the convergence of excess returns significant at 99% confidence level. This implies that in personal and business services internationalization can be profitable as the relative share of foreign assets slows down the speed of convergence.

4.1.3 Conclusions based on the regression results for the economy-wide and industry-specific models

In conclusion, both the economy-wide model and industry-specific model for the manufacturing sector present overall a negative effect of the degree of internationalization on the convergence of excess returns. For the personal and business services industry a positive effect of internationalization on the convergence is found. For the mining industry the results between the degree of internationalization and the existence of long-term excess returns are ambiguous. The effect of foreign sales on long-term excess returns is positive, but the effect of foreign assets negative. The differences in outcomes on the speed of convergence, the level of long-term excess returns, and the effects of internationalization for the different industries may be caused by differences in industry structure such as industry entry barriers and industry concentration (Cheng, 2005; Dechow et al., 1999; Fairfield et al., 2009). 4.2 Additional tests In order to determine the robustness of the results, three additional tests are performed. 4.2.1 Robustness test for the financial crisis and credit spread The first test is related to the robustness of the credit spreads used in this paper. As described before, Callegan et al. (2015), pg. 248, show that credit spreads are relatively stable over time. However, in the years around the financial crisis the credit spreads increased enormously. In order to examine the impact of this change in credit spreads on the results, the firm-observations for the years 2007-2009 are removed from the sample.

34

share of foreign assets speeds up the convergence of excess returns. Second, panel D finds a significant effect for the manufacturing sector suggesting that the relative share of foreign sales negatively impacts long-term excess returns. Third, panel H shows for the health, legal and education sector that the interaction term of foreign sales slows down the convergence of excess returns. The relative share of foreign assets in this sector negatively influences long-term excess returns.

However, the new results are significant at a lower confidence level (90%) and excluding the years 2007-2009, barely change the value of the coefficients. Hence, the results of this test do not change the earlier conclusions about the effect of internationalization on the convergence and permanence of excess returns.

4.2.2 Market risk premium and risk-free rate

This second robustness test looks into the effect of using different market risk premiums. Appendix G presents the regression results using market risk premiums of 4% (table 13) and 6% (table 14) instead of 5% (Koller et al., 2010). The results in table 13 and 15 of appendix G compared with the results in table 7 indicate that the results do hardly change and are therefore robust against the market risk premiums of 4% and 6%.

4.2.3 Control variables

Appendix H presents the results of the analysis including control variables in order to check the correctness of the model. Panels A and B show the results of industry control variables and panels C and D show the results of the firm specific control variables.

35

Panels C and D respectively show the univariate and multivariate results of the regression including firm characteristic control variables. Size is negatively associated with excess returns implying that the bureaucratic structure within large organizations can negatively impact excess returns (Dickinson and Sommers, 2012). Firm age is only significant in the univariate analysis, predicting a positive effect on excess returns. This positive effect is due to the learning effect that improves when firms become older (Dickinson and Sommers, 2012).

5. Conclusion and discussion

This paper studies the effect of the degree of internationalization on the convergence and permanence of excess returns. The empirical model of Ohlson (1995) and Feltham and Ohlson (1996) is used and adjusted in order to allow for the effect of the degree of internationalization and net investments on excess returns. Next to the findings with respect to the effect of internationalization (see below), this paper finds that in line with prior research showing that excess returns converge over time to long-term excess returns. Moreover, growth in NOA is negatively correlated with the excess returns of the following year. This is in line with previous research (Dickinson and Sommers, 2012; Fairfield et al., 2009; Fairfield et al., 2003; Fairfield and Yohn, 2001).

The findings regarding the effect of internationalization are mixed. Following an economy-wide model and an industry-specific model for the manufacturing sector, internationalization increases the speed of convergence of excess returns. This is in contradiction with most of the internationalization theories predicting that internationalization yields a competitive advantage. Possible explanations for the increased speed of convergence of excess returns can be agency problems resulting from internationalization and a negative impact of trade barriers. Agency problems create costs that result in a competitive disadvantage over more domestically oriented competitors. Trade barriers make it for multinationals more difficult to operate in a foreign country, creating a competitive disadvantage over firms in the foreign market. For the personal and business services industry evidence is found that, in line with internationalization theories, internationalization, as measured by the relative share of foreign assets, slows down the convergence of excess returns

36

An effect of internationalization on the permanence of excess returns has not been determined using the economy-wide model. For all industries, except the mining industry, no relation between the degree of internationalization and the permanence of excess returns is determined. The results for the mining industry are not conclusive as the relative share of foreign sales increases the level of long-term excess returns while the relative share of foreign assets decreases the level of long-term excess returns. The results are important for business management and valuation practitioners. The importance for management is in threefold. First, management can decide to enter industries where the excess returns converge slower over time and the level of long-term excess return is higher. Second, management can influence the convergence and permanence by deciding to go international or not. Third, management should try to minimize agency costs in order to make internationalization more profitable.

Moreover, valuation practitioners should, when analyzing, forecast and valuing the profitability of the firm, take into account the effect of competition (Dickinson and Sommers, 2012) and internationalization on excess returns.

These findings are subject to several limitations, which should be taken into account when reading the conclusions and considering future research.

First, the industry-specific results rely on different numbers of observations. For instance, the results for the health, legal and education industry rely on 1,737 observations, while the results for the manufacturing industry are based on 11,759 observations. This can impact the comparability of the results due to the fact that more observation gives a better view.

Second, the simplified WACC assumptions can impact the correctness of the results. Since the WACC directly impacts the level of excess returns, a more precise WACC can lead to different outcomes.

Third, the approach used to examine the level of export is disputable. For the determination of export (foreign sales over total sales), all sales in countries with a foreign production facility are excluded. It may however be the case that part of the foreign sales are not recognized as foreign sales when the firm exports goods to the country where also foreign assets are located. In order to solve this problem, future research needs to find a database that reports export as a separate variable.

37

convergence and permanence of excess returns. For instance, Hutzschenreuter et al. (2016) argue that the speed of internationalization is an important dimension of the internationalization process. Moreover, future research can try to explain whether there are more specific aspects of internationalization that can negatively impact the convergence and permanence of excess returns.

References

Asthana, S., Zhang, Y., 2006. Effect of R&D investments on persistence of abnormal earnings. Review of accounting and finance 5, 124-139.

Barkema, H., Bell, J., Pennings, J., 1996. Foreign entry, cultural barriers and learning. Strategic management journal 17, 151-166.

Barney,J., 1991. Firm resources and sustained competitive advantage. Journal of management 17 (1), 99-120.

Bou, J., Satorra, A., 2007. The persistence of abnormal returns at industry and firm levels: evidence from Spain. Strategic management journal 28 (7), 707-722. Brooks, C., 2014. Introductory econometrics for finance. The ICMA Centre, Henley

Business School, University of reading, third edition.

Buckley, P., Casson, M., 2009. The internationalization theory of the multinational enterprise: a review of the progress of a research agenda after 30 years. Journal of international business studies 40, 1563-1580.

Callegan, J., Murphy, A., Qian, H., 2015. Third international conference on credit analysis and risk management. Cambridge Scholars Publishing.

Chen, S., Tortoriello, R., Patton, M., Falk, R., 2015. Foreign exposure of S&P 500 companies and implications for earnings. S&P capital IQ, McGraw Hill Financial

April 2015, 9-12-2016, 14:13 PM, http://marketintelligence.spglobal.com/documents/our-thinking/ research/foreign-exposure-of-sp-500-companies-and-implications-for-earnings.pdf Cheng, Q., 2005. What determines residual income? The accounting review 80, 85-112. Conconi, P., Sapir, A., Zanardi, M., 2016. The internationalization process of firms: from exports to FDI. Journal of international economics 99, 16-30.

38

Dechow, P., Hutton, A., Sloan, R., 1999. An empirical assessment of the residual income valuation model. Journal of accounting and economics 26, 1-34.

Dickenson, V., Sommers, G., 2012. Which competitive efforts lead to future abnormal economic rents? Using accounting ratios to assess competitive advantage. Journal of Business Finance & Accounting 39, 360-398.

Fama, E., French, F., 1997. Industry costs of equity. Journal of financial economics 43, 153-193.

Fairfield, P, Whisenant, S., Yohn, T., 2003. Accrued earnings and growth: implications for future profitability and market mispricing. The accounting review 78 (1), 353-371.

Fairfield, P., Ramnaht, S., Yohn, T., 2009. Do industry-level analyses improve forecasts of financial performance? Journal of accounting research 47 (1), 147-178.

Fairfield, P., Yohn, T., 2001. Using asset turnover and profit margin to forecast changes in profitability. Review of accounting studies 6, 371-385.

Feltham, G., Ohlson, J., 1996. Uncertainty resolution and the theory of depreciation measurement. Journal of accounting research 34 (2), 209-234.

Hutzschenreuter, T., Kleindienst, I., Guenther, C., 2016. Speed of internationalization of new business units: the impact of direct and indirect learning. Management International Review 56 (6), 849-878. Ibbotson, 2015. Worldwide corporate tax guide. Jacobsen, R., 1988. Persistence of abnormal returns. Strategic Management Journal 9(5), 415-430. Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. The American economic review 76 (2), 323-329. Johanson, J., Vahlne, J., 1977. The internationalization process of the firm – a model of knowledge development and increasing foreign market commitments. Journal of international business studies 8 (1), 23-32. Koller, T., Goedhart, M., Wessels, D., 2010. Valuation measuring and managing the value of companies. John Wiley & Sons, Inc., Hoboken, New Jersey. McGahan, A., Porter, M., 1999. The persistence of shocks to profitability. The review of economics and statistics 81 (1), 143-153.

39 Morck, R., Yeung, B., 1991. Why investors value multinationality. The journal of business 64 (2), 165-187. Mueller, D., 1977. The persistence of profits above the norm. Economica 44, 369–380. Nwaeze, E., 2000. Deregulation of the electric power industry: the earnings, risk, and

return effects. Journal of regulatory economics 17 (1), 49-67.

Ohlson, J., 1995. Earnings, book values, and dividends in equity valuation. Contemporary accounting research 11 (2), 661-687.

Palepu, K.G., Healy, P.M., Bernard, V.L., 2004. Business analysis & valuation: using financial statements. Thomson/South western college pub.

Porter, M., 1980. Competitive strategy: techniques for analyzing industries and competitors. Free Press, New York, 1980.

Porter, M., 1985. Competitive strategy: creating and sustaining superior performance. Free Press, New York, 1985.

Porter, M., Millar, E., 1985. How information gives you competitive advantage. Harvard Business Review, July-August 1985, 149-160.

Rugman, A., 1976. Risk reduction by international diversification. Journal of international business studies 7 (2), 75-80.

Schultze, W., 2005. The information contact of goodwill-impairments under FAS 142: Implications for external analysis and internal control. Schmalenbach Business Review 57, 276-297.

Riahi-Belkaoui, A., 1999. The degree of internationalization and the value of the firm: theory and evidence. Journal of international accounting, auditing and taxation 8 (1), 189-196.

Sullivan, D., 1994. Measuring the degree of internationalization of a firm. Journal of international business studies.