The Multinational Advantage - SMU School of Accountancy · for measuring valuation effects in a multinational context. 5 This method, in effect, compares the value of the firm as

The Multinational Advantage

Presented by

Dr Jonathan Rogers

Associate Professor University of Chicago

Booth School of Business

201314-07

The views and opinions expressed in this working paper are those of the author(s) and not necessarily those of the School of Accountancy Singapore Management University

We thank Phil Berger Andy Bernard Marianne Bertrand Alan Bester Chris Hansen Chang-Tai Hsieh Rafael La Porta Christian Leuz Jon Lewellen Abbie Smith Doug Skinner and Jerry Zimmerman for helpful discussions and comments This paper has benefited from the comments of workshop participants at Chicago Cornell Dartmouth Florida Illinois at Chicago INSEAD LBS Michigan MIT Northwestern Rochester Syracuse University Texas UCLA and the EAA Annual Congress We thank William Steciak of KPMG for helpful discussions related to segment reporting We gratefully acknowledge the financial support of the University of Chicago Booth School of Business the Fama-Miller Center and the Neubauer Family Foundation The statistical analysis of firm-level data on US multinational firms was conducted at the International Investment Division Bureau of Economic Analysis US Department of Commerce under arrangements that maintain legal confidentiality requirements The views expressed are those of the authors and do not reflect official positions of the US Department of Commerce

Drew D Creal

University of Chicago Booth School of Business

Leslie A Robinson Tuck School of Business at Dartmouth

Jonathan L Rogers

Sarah L C Zechman University of Chicago Booth School of Business

February 2013

Chicago Booth Research Paper No 11-37

Abstract Using a proprietary dataset we evaluate whether the degree of foreign operations affects firm value by comparing actual value to imputed value for US multinational corporations (MNCs) We argue that using benchmark firms operating in the same country and industry as each MNC segment controls for differences in discount rates and expected growth rates across countries and industries This allows us to isolate the value effects of organizing a set of otherwise independent activities within a multinational network We find robust evidence that multinational networks trade at a premium relative to a benchmark portfolio of independent firms JEL Classification F23 G32 G34 M41 Keywords firm valuation multinational corporations diversification

1 Introduction

We investigate whether the degree of international diversification affects firm value for US

multinational corporations (MNCs) Specifically we determine whether the relative size of the

foreign operation has a positive negative or no association with the difference between the

actual observed firm value and an imputed value of the firm (called excess value)1 Our analysis

points to a statistically and economically significant positive excess value or premium to

increased foreign operations This is akin to finding that the value of the MNC is greater than the

sum of its individual parts We subject this finding to a battery of robustness tests to help

determine whether it is simply an artifact of endogeneity or measurement error problems We fail

to find evidence that this is the case The finding of a premium is robust across all specifications

examined In terms of magnitude a one percent increase in the size of the foreign operation is

associated with an increase of between 019 and 037 in excess value Collectively the

evidence suggests that on average MNCs create value by organizing a set of otherwise

independent activities within a multinational network

A large literature examines whether firms that are industrially diversified trade at a discount

or premium to non-diversified firms This literature typically finds that industrially diversified

firms trade at a discount2 A popular method to quantify this valuation effect was developed by

Berger and Ofek (1995) who compare the actual value of an industrially diversified firm with a

hypothetical firm whose value is the sum of the imputed values of its individual industrial

segments As the individual segments of the diversified firm are not traded the imputed value of

each segment is the observable median firm value of a single-segment (non-diversified) firm

1 As we study US-based multinational firms throughout the paper we refer to the US as lsquodomesticrsquo and any other country as lsquoforeignrsquo 2 Examples include Lang and Stulz (1994) Berger and Ofek (1995) Laeven and Levine (2007) Schmid and Walter (2009) Ammann Hoechle and Schmid (2012) and Hoechle Schmid Walter and Yermack (2012)

operating in the same industry The difference between the actual value and the imputed value is

an estimate of the premium (if positive) or discount (if negative) Evidence of a negative

association between this excess value measure and a measure of industrial diversification is

consistent with industrial diversification destroying firm value on average

While the finding of an industrial diversification discount is quite robust there is

considerable debate about the interpretation Some argue that the discount is evidence that the

costs of operating in multiple industries outweigh the benefits (eg Berger and Ofek 1995)

Others argue that the discount is driven by the types of firms that choose to diversify or the types

of businesses they invest in when diversifying (Campa and Kedia 2002 Villalonga 2004b) The

literature that uses this methodology continues to evolve improving our understanding of the

forces that drive the industrial diversification discount For example the recent study by Hoechle

et al (2012) finds that a substantial proportion of the discount can be explained by variation in

corporate governance proxies

In contrast to the literature on the industrial diversification discount comparatively few

studies examine how foreign operations affect firm value3 This contrast is surprising because

descriptive data for US firms indicates that industrial diversification has stagnated while foreign

expansion continues at a rapid pace We develop a method to quantify the value effects of

foreign operations along the lines of Berger and Ofek (1995) but that remains applicable to the

multinational setting While Berger and Ofek (1995) divide each firm into industry segments we

divide each MNC into geographic-industry segments (ie separate country-industry

components)4 We then compare the actual value of the firm to the imputed value of the firm

3 Denis Denis and Yost (2002) is a notable exception 4 In our setting we refer to a lsquosingle-segmentrsquo firm as a firm that operates in only one country and one industry Accordingly we use the term lsquoexcess valuersquo to denote the difference between the actual (observed) firm value and the hypothetical firm value computed as the sum of the imputed values of the individual country-industry segments

Using a method similar to Berger and Ofek (1995) we determine the imputed value for each

country-industry component by using the median single-segment firm operating exclusively in

the same country and industry (ie single-segment foreign (domestic) firms in the same industry

and country are used as benchmarks for the foreign (domestic) operations of US MNCs) In

other words our approach uses data on the observed firm values of single-segment US firms

(from Compustat) and foreign firms (from Worldscope)

This approach to measuring excess value differs from previous methods to impute the firm

value of MNCs Denis et al (2002) use single-segment domestic firms (ie US firms) in the

same industry as a benchmark for both domestic and foreign MNC operations Similarly studies

using Tobinrsquos Q to investigate valuation effects of MNCs rely only on domestic benchmarks

(eg Morck and Yeung 1991 Morck and Yeung 2001 Gande Schenzler and Senbet 2009)

As growth rates and discount rates vary by country we believe our method is more appropriate

for measuring valuation effects in a multinational context5 This method in effect compares the

value of the firm as a whole to the sum of the parts In addition our method is conceptually

consistent with theories on foreign direct investment (FDI) which note that an MNC exists when

a firm seeks to exploit its advantages and remove conflict arising in external market transactions

by combining a firm of one nationality that might otherwise exist independently under the

ownership of a firm of a different nationality (Dunning and Rugman 1985)

Our approach leverages data maintained by the Bureau of Economic Analysis (BEA) which

provides detailed accounting information about FDI allowing measurement of sales for each

5 Denis et al (2002) contemplate the use of single-segment foreign firms as a benchmark noting three constraints to implementing such a procedure (1) the country location of firmsrsquo foreign operations cannot be reliably identified using Compustat data (2) the industry membership of firmsrsquo foreign operations cannot be identified reliably using Compustat data and (3) valuation ratios may differ across countries due to differences in accounting standards As described in our study we overcome each of these constraints and find that the different benchmark is important and reverses prior findings

country-industry in which US-based MNCs operate These data provide a substantial advantage

over the Compustat database used by prior research While Compustat provides some

information about the countries and industries in which these firms operate it is not as detailed

about either of these dimensions (or their interaction) and heavily relies on managerial disclosure

choices which could induce measurement bias (Villalonga 2004a) In addition to the new

method of estimating the value effects of foreign operations we provide evidence that MNCs

trade at a premium which stands in contrast to the discount documented in Denis et al (2002)

This difference is primarily attributed to our use of foreign benchmarks to construct the imputed

values of foreign operations6

Our finding of a premium better reconciles with the findings in the international trade

literature By focusing on nonpublic establishment level (eg factory store or office) data the

trade literature is able to generate relatively precise proxies for total factor productivity (TFP)

and provides robust evidence that firms engaging in international trade are more productive than

those that do not (see Helpman 2006 and Syverson 2011 for reviews) Furthermore

productivity differences tend to be highly persistent even within narrowly defined industries

(Syverson 2011) Due to data limitations it is difficult to accurately measure TFP at the firm

level especially for firms which operate in multiple industries andor countries Nevertheless

Helpman Melitz and Yeaple (2004) provide some evidence that US firms engaging in FDI

have labor productivity advantages over firms that do not Reconciling the discount to

multinational operations found by Denis et al (2002) to the trade literature requires that the

persistent productivity advantages of firms engaged in international trade must be more than

offset by some other cost (eg agency costs) While this relation is possible it is difficult to

6 We repeat the analysis of Denis et al (2002) for our sample and confirm that differences in value effects of foreign operations are driven by the different methods for measuring the imputed value of foreign operations

conjecture why the most productive firms would have the largest agency costs In contrast the

finding of a premium easily reconciles with the trade literature

Given our result we investigate whether the premium is simply an artifact of innate

characteristics of firms that choose to invest abroad For example Helpman et al (2004) present

a simple model with heterogeneous productivity endowments The firms that receive the highest

productivity endowments are the ones capable of paying the fixed costs to establish a foreign

subsidiary To the extent productivity advantages create excess value one would expect the type

of firms that choose to invest abroad would be valued at a premium even in the absence of their

foreign investments After including firm fixed effects to control for time invariant firm

characteristics (eg innate productivity advantages) we continue to find economically and

statistically significant evidence of a premium to foreign operations

To further account for the endogenous nature of FDI we estimate a dynamic panel data

model using the generalized method of moments (GMM) estimator developed by Arellano and

Bond (1991) Arellano and Bover (1995) and Blundell and Bond (1998) This estimator allows

us to simultaneously account for the potential endogeneity of FDI industrial diversification and

firm value This method also yields a significant premium which is of similar magnitude and

significance to that estimated using fixed effects

We also perform a series of additional robustness tests First we examine whether the

measured excess value premium is driven by operations in countries with large control premiums

(Dyck and Zingales 2004) In countries with large control premiums using stock price in the

calculation of total firm value could underestimate the true value of the benchmark firms and

thereby induce an excess value premium We fail to find evidence that the excess value premium

is driven by such countries

Second we examine whether access to a low cost of capital is a meaningful source of the

US MNC advantage We incorporate foreign benchmarks in our method of estimating firm

excess value as it allows us to incorporate attributes of operating in those foreign countries ndash a

key attribute being the cost of capital Relative to the local benchmarks in foreign countries US

MNCs have the ability to obtain funds outside the local operating environments either by

borrowing in the US market or transferring capital internally When competing against firms in

shallow capital markets with high costs of capital MNCs may have a competitive advantage

over local foreign companies due to these alternative sources of capital However investors and

lenders are likely to expect higher rates of return from the additional risk associated with

operating in these environments making it unclear whether a multinational network will enhance

value in countries with a high cost of capital In addition foreign firms have some ability to

access international capital markets which mitigates the advantages to US firms To examine

whether the cost of capital plays a role in our finding of a premium we include a proxy for the

extent of operations in countries with higher costs of capital Our results are unaffected by the

inclusion of the proxy

Finally we control for several corporate governance proxies which Hoechle et al (2012)

find to be correlated with excess values (and industrial diversification) While magnitude of the

premium is virtually unaffected by including these proxies the significance level does decline

due to the reduced sample size

Our study contributes to the literature on multinational firms Our new method of estimating

excess values for multinational firms provides strong evidence of a statistically and economically

significant premium associated with increased international operations The premium is robust to

a variety of specifications and controls designed to evaluate alternative explanations In addition

this result easily reconciles with the international trade literature that establishes a link between

greater productivity and international operations though we do not find results supporting this as

the only source of the premium As segment disclosures improve and as new data sources

become available on multinational activity (eg Bureau Van Dijk) our empirical approach can

be used to investigate the extent to which the home country of the parent might influence the

extent with which foreign operations affect firm value (ie by examining non-US based

MNCs)7 Future studies could also examine the underlying forces affecting the excess value

premiums across countries

We also contribute to the understanding of international accounting standards and their

comparability Many studies seek to compare firms using different accounting standards These

comparisons generally rely on information generated by a particular accounting regime By

comparing commonly used metrics (total assets net income and sales) reported for the same

firm and year across different regimes we provide insights as to the relative comparability across

standards In particular we find that sales is reported significantly more consistently across

accounting regimes than the other metrics examined This finding should aid researchers in

reducing measurement error when comparing accounting information in multinational settings

The paper proceeds as follows Section 2 discusses the motivation for finding a premium or

discount to multinational operations Section 3 describes our data and how the excess value of a

firm is measured A discussion of our independent variables and primary findings is presented in

Section 4 with additional robustness tests provided in Section 5 Finally Section 6 concludes

7 As MNCs are ultimately bound by the tax regulatory and legal frameworks of their home country the valuation effect of foreign operations that we document for US-based MNCs may be different in a sample of non-US based MNCs with the same country-industry footprint Theorists refer to these as lsquolocational advantagesrsquo enjoyed by all multinational firms of a given nationality and obtained irrespective of skills or capabilities unique to a particular firm (Yamin 1991) There is also evidence of locational advantages in the market for corporate control (Huizinga and Voget 2009)

2 Premium versus discount

In a frictionless world where managers maximize firm value and markets are efficient there

should be no discount or premium to operating in foreign countries However there are a number

of forces that may cause actual firm value to deviate from the implied firm value

On one hand MNCs may have comparative advantages that generate premiums (ie positive

excess values) relative to a similar footprint of stand-alone firms Having operations in a variety

of locations that access different supplies of inputs and customers allows the firm to take

advantage of changing market conditions by shifting operations or products to maximize firm

value more so than a firm operating in a single country Similarly having access to a variety of

institutional settings such as different tax codes legal regimes and financial markets can provide

MNCs with options beyond those available to firms located in a single country

Another possible factor associated with a premium to multinational operations is the cost of

capital For operations located in countries with shallow capital markets or weaker creditor

rights operating within a conglomerate can provide the benefit of a lower cost of capital for

investments and expansion for at least two reasons First as our MNCs are domiciled in the US

their foreign segments have greater access to US capital markets relative to local competitors in

the foreign jurisdiction Second cash rich segments with few investment opportunities can

finance investments in cash poor segments with positive net present value projects (eg Myers

and Majluf 1984) Consequently diversified firms should be less liquidity constrained and better

able to shift resources to the most valuable investment opportunities Research has found that

internal financing is used more often when diversified firms have operations in countries with

more costly external financing (Desai Foley and Hines 2004) Multinational firms may also be

relatively more protected than single-country firms when negative shocks hit external capital

markets (Foley Desai and Forbes 2008) In a recent study Matvos and Seru (2011) find that

industrially diversified firms perform better than non-diversified firms when external capital

markets are impaired

While these arguments suggest that MNCs enjoy certain advantages when the foreign cost of

capital is high these benefits should be reduced by the additional risk associated with operating a

business in a high cost of capital environment For example Verizon is a multinational firm

operating in a number of countries One of the higher cost of capital locations in which they

operated was Venezuela While the multinational structure of Verizon may have mitigated some

of the risks associated with operating in a higher cost of capital location the firms was unable to

eliminate all of these risks In fact in 2007 Verizonrsquos operations in that country were

nationalized resulting in a net extraordinary loss of $131 million that year8 In other words

MNC investors are likely to expect higher rates of return to compensate for higher risk which

likely offsets some (if not all) of the potential excess value premium

On the other hand multinational firms may incur a discount (ie negative excess values)

relative to a similar footprint of stand-alone firms Potential reasons for a discount largely rely on

the existence of agency costs Multinational firms tend to be larger more complex and less

transparent The combination of complexity and opacity inherent in MNCs reduces a firmrsquos

reliance on external capital as the cost of such capital increases due to agency costs (Desai et al

2004) As a result of the reduced reliance on external funds MNCs face a reduced level of

monitoring As noted by Jensen (1986) managers have incentives to increase firm size beyond

that which is optimal with overinvestment or misallocation of funds to pet projects If this lack of

monitoring is in combination with available internal capital empire building becomes easier

(Hope and Thomas 2008) Furthermore the current segment reporting standards under US 8 See the 10-K of Verizon Communications Inc dated December 31 2007 (footnote 2)

GAAP only require highly aggregated disclosures of foreign operations and allow for

considerable managerial discretion As a result operating in multiple countries makes it easier

for management to hide poor performance ndash either their own or that of a division ndash such that low

quality managers are more likely to be retained and overcompensated and poorly performing

divisions retained longer than optimal Other reasons one might expect a discount to be applied

to multinational firms include higher coordination costs (eg coordinating across different

cultures and languages) as well as additional risks These risks include exposure to multiple

political regimes legal regimes economic regulations and currency fluctuations

Past studies investigate whether foreign operations of US firms enhance or reduce firm

value However all of these studies benchmark the value of the foreign operations of an MNC to

US domestic firms For instance Denis et al (2002) determine the implied value of a US

MNC using domestic (US) single-segment firms in the same industry as a benchmark Studies

using Tobinrsquos Q to investigate valuation effects of MNCs also rely on domestic benchmarks

However using a US domestic firm as a benchmark for the value of foreign operations

implicitly makes two assumptions First this method assumes that the risks and expected growth

rates of foreign operations are equivalent to those of domestic operations This assumption is

inconsistent with the findings of Hail and Leuz (2009) which provides evidence that the cost of

capital varies substantially across countries Second this method assumes that there is excess

domestic capacity to allow for expansion with profitability similar to current domestic

operations We relax these assumptions by measuring the imputed value of a foreign component

of a US firm as if that component operated autonomously within that foreign country (rather

than in the US) Specifically our multiples allow the cost of capital and expected growth rates

to vary across not just industry but also country Furthermore our counterfactual does not

require the assumption that operations be relocated mitigating the production and sales capacity

concerns

3 Measuring excess value

31 Description of excess value

Figure 1 shows the average percent of foreign sales (based on segment disclosures under US

GAAP) across firms through time and illustrates that firms are continuing to expand

internationally9 This expansion highlights the importance of understanding whether the decision

to operate internationally is on average a value enhancing corporate strategy We contribute to

this understanding by evaluating whether the firm as a whole is worth more or less than its

imputed value (ie the firm is valued in excess of the sum of its individual components)

We measure the excess value of a firm as the logarithm of the ratio of actual firm value to

imputed value based on the method used in Berger and Ofek (1995) The imputed value is the

hypothetical value of the MNC under the assumption that its country-industry components

operate as independent entities A key innovation of our study is an alternative approach to

imputing the value of the foreign operations of an MNC Components of MNCs operating in a

given industry and country are imputed using the value of single-segment firms operating in the

same industry and country Thus we ensure the discount rates and expected growth rates are

applicable to each given industry and location in estimating the implied values of the MNC

segments

9 The discontinuity in industrial diversification (the percent of sales outside the primary industry) between 1997 and 1998 is the result of a change in segment reporting rules

The actual firm values are observable for all firms in our sample The hypothetical firm

values for multi-segment firms are imputed using market value to sales ratios herein referred to

as lsquomultiplesrsquo of the median single-segment firm operating in the same country-industry Thus

these single country-industry firms serve as benchmarks for an MNCrsquos operations within that

same country-industry The imputed value of the country-industry component can be viewed as

an estimate of what that component would be worth if it operated independently The sum of the

imputed values across all country-industry components represents the hypothetical value that the

multi-segment firm would be worth if it were broken apart

32 Data

Four primary data sources play a distinct role in carrying out our analysis i) Bureau of

Economic Analysis (BEA) data ii) Compustat Segment data iii) Worldscope data and iv)

Compustat Fundamentals Annual data Broadly speaking BEA data provide information about

the diversified nature of MNCs Compustat Segment data provide information about the

industrial diversification of US domiciled firms Worldscope data are used to compute

multiples for single-segment foreign firms that serve as benchmarks for the foreign operations of

MNCs (ie foreign benchmarks) and Compustat Fundamentals Annual data are used to

compute multiples for single-segment domestic firms that serve as benchmarks for the domestic

operations of multinational firms We also use the latter data source to construct our primary

control variables

Key to the execution of our study is our access to BEA data Federal law requires US-

domiciled MNCs to report certain financial and operating data to the BEA10 With regard to

observing the country-industry operations of MNCs use of the BEA data overcomes two

important limitations of Compustat Segment data First Compustat does not consistently identify

the specific location or scale of MNCsrsquo foreign operations11 Second Compustat does not

identify the industry activities in each country of operation This latter point is particularly

limiting for MNCs because domestic industry activity need not mirror foreign industry activity

An MNC generating an equal proportion of sales in two industries in the domestic market need

not generate industry sales in equal proportions in every foreign market12 The lack of detail and

consistency in Compustat Segment data arises because segment reporting is highly aggregated

and firms exercise substantial discretion in defining segments under Accounting Standards

Codification 280 (ASC 280) (eg Villalonga 2004a and Bens Berger and Monahan 2011)

The BEA defines an MNC as the combination of a single US entity called the US parent

and at least one foreign affiliate ndash that is these firms have a physical presence outside the US

due to FDI Since 1982 MNCs have completed a mandatory and confidential lsquoSurvey of US

Direct Investment Abroadrsquo for the domestic operation and the operations of each foreign

affiliate defined as a foreign entity in which the US parent holds at least a 10 percent equity

interest (directly or indirectly) The data are reported in US dollars on a fiscal year basis and in

10 BEA surveys are conducted pursuant to the International Investment and Trade in Services Survey Act (PL 94-472 90 Stat 2059 22 USC 3101-3108) See httpwwwbeagovsurveysdiasurvhtm and Mataloni (2003) for more detailed information on the BEA data 11 Exhibit 21 disclosures of material subsidiaries and their locations in SEC 10-K filings could potentially help overcome the inability to observe the location of the firmrsquos foreign operations from segment disclosures However at least three limitations remain (1) for industrially diversified multinational firms the researcher would be required to make an assumption about the proportion of industrial diversification in each country (2) Exhibit 21 provides no information about the scale of activity in each jurisdiction and (3) managerial discretion is used to determine a lsquomaterialrsquo subsidiary under Section 601 of SEC Regulation S-K 12 In fact the expectation that a multinational firm would exhibit different industry membership in different countries is consistent with the theory of vertical foreign direct investment where firms separate their various value chain activities and locate them in the most favorable country (see eg Hanson Mataloni and Slaughter 2005)

accordance with US Generally Accepted Accounting Principles (US GAAP) For each year

we observe the sales industry composition and location of not only the parent but also each

affiliate13

The procedures we use to construct our sample are similar to those used by Berger and Ofek

(1995) and Denis et al (2002) Using firm-years appearing in Compustat Fundamentals Annual

and Compustat Segment data we obtain a sample of firms domiciled in the US with total sales

exceeding $20 million We exclude firms in the financial service (SIC codes 6000 to 6999) and

utility (SIC codes 4000 to 4999) industries and eliminate firm-years when the sum of segment

sales (Compustat geographic or Compustat business) is not within 1 of total firm sales for that

year We restrict our sample period to include fiscal years beginning after December 15 1997 as

ASC 280 Segment Reporting substantially altered the definition of a reporting segment under

US GAAP This restriction ensures that Compustat Segment data reflects the use of a consistent

accounting standard Finally we require that our MNCs appear in both Compustat and BEA

data The requirement that our sample firms appear in the BEA database ensures that the firms

have observable international operations In sum these procedures result in 1166 multinational

firms incorporated in the US with 4950 observations between 1998 and 2008 that meet all of

our data requirements

The dependent variable in our study is excess firm value (Excess Value) defined as the

logarithm of the ratio of actual firm value to imputed firm value (Berger and Ofek 1995)

13 We obtain sales industry membership and country location for affiliates from one of two sources When affiliates exceed the BEA-determined reporting thresholds (ie if their assets sales or net income (loss) exceed $7 million in 1999 $30 million in 2000 through 2003 $10 million in 2004 and $40 million in 2005 through 2008) they must file their own detailed survey When affiliates fall below the thresholds we obtain our data from an attachment to the parent survey

Appendix A provides the definition of this and all other variables used in our analysis We

observe actual firm value equal to a firmrsquos market value of equity plus book value of liabilities

(ie book value of total assets minus book value of total equity) using Compustat Fundamentals

Annual data We calculate imputed firm value as the sum of the imputed values of a firmrsquos

operations in each country-industry Our method for imputing the value of the separate

components of a firm can best be described in three steps First we obtain total sales generated

by a firm for each country-industry in which it operates14 Second we obtain multiples (market

value to sales ratios) for benchmark firms operating in those same country-industries Third we

multiply a firmrsquos country-industry sales by the applicable median country-industry multiple to

obtain the imputed market value for each country-industry operation We perform each step

annually

Conceptually imputed values can be based on assets (ie Tobinrsquos Q) net income or sales

multiples We restrict ourselves to sales multiples for two reasons First using value to sales

ratios maintains consistency with the prior research on the excess value implications of

multinational operations (eg Denis et al 2002) Second as our method uses accounting data

for foreign companies to compute country-industry multiples the accounting numbers we use

need to be consistently measured across firms using different accounting standards We find that

sales data are measured most consistently

To assess the comparability of sales relative to either net income or assets across various

accounting standards we examine all firms listed on Worldscope as changing to or from US

14 Specifically we allocate total sales from Compustat to each component based on the proportion of BEA sales (after eliminating intercompany sales) occurring in each country-industry This step ensures that our total firm sales do not deviate from a firmrsquos total sales as reported in the 10-K

GAAP between 2005 and 201115 We obtain accounting data for the year of the change and the

prior year where electronically available in English We are able to obtain data for 66 firms

changing standards as detailed in Table 1 The distribution of years for which accounting data

are available under two different accounting standards for the same fiscal year is shown in Panel

A The years shown are the year prior to the change as these accounting numbers were provided

under the original standard and then subsequently restated in the following year under the new

standard for comparative purposes The majority of the shifts were to or from IFRS (71)

though seven other accounting standards are noted (Panel B) We examine the percent that the

non-US GAAP reported number differs from the US GAAP reported number in Panel C to

assess the comparability of these three summary accounting numbers across various accounting

standards Examining both the full sample as well as the subset that overlaps our study we find

that sales are more consistently measured than net income or assets

The sum of the imputed values across the country-industry components of an MNC is an

estimate of the value of a portfolio of unrelated businesses that mirror the related businesses of

the firm Consequently a comparison of the actual firm value with the imputed firm value is a

measure of how a multinational network affects firm value We provide descriptive statistics for

Excess Value for our sample of 4950 MNC firm-years in Table 2 Panel A This table also

includes univariate comparisons of the values between the MNC firms and the domestic single-

segment and foreign single-segment benchmark firms The Excess Value measure is significantly

higher for MNCs suggesting a premium in firm value for multinational firms relative to the

benchmark firms in total as well as the domestic and foreign subgroups (p-values for tests of

15 Due to the hand collection efforts required and the limitation on electronic filings we began our search with changes documented in 2005 which generally require filings from 2004

means and medians are lt 001)16 These results provide some initial evidence that operating in

multiple countries is positively associated with firm value

To estimate country-industry multiples we rely on Worldscope financial data on foreign

firms and Compustat Fundamentals Annual for domestic firms We restrict the firms to include

only those with at least 90 of sales income and assets inside the country of domicile (ie

those that do not report significant multinational activity) and that operate in a single industry

We refer to these firms as benchmark firms (either foreign or domestic depending on the country

of domicile) these firms do not appear in any of our regressions We report the number of

benchmark firms during our sample period in Table 2 Panel A The industry criterion is based on

the two-digit SIC code17 For every country that has at least five firms in the respective industry

and year we use the median ratio of market value to sales18 This ratio is the country-industry

multiple ndash an input required to compute imputed values

To determine the country-industry composition of MNCs we rely on BEA data In Table 3

we report the aggregate number of foreign affiliates and total sales by country as provided by

the BEA data for our sample of 4950 MNCs The specific countries tabulated are those that

represent at least 02 percent of total firm sales (pooled across all years) in our sample The top

five foreign countries in which MNCs generate sales are Canada United Kingdom Germany

France and Japan Table 3 also provides a comparison of the number of foreign benchmark

firms available in Worldscope in the specific countries in which our sample of MNCs generate

16 Similar to Berger and Ofek (1995) the median values differ slightly from zero for the benchmark single-segment firms due to the truncation of extreme excess values 17 BEA reporting constrains us to the two-digit SIC code level 18 When five firms are not available for a given country and industry we follow Berger and Ofek (1995) and move to a more general industry definition (eg one-digit SIC code) This procedure results in a mean (median) 630 (757) of MNC sales matched at the two-digit level

the majority of their sales Country coverage in Worldscope is not available for countries

representing 004 percent of total firm sales (per BEA)

4 Empirical results

41 Independent variables

Recall that our objective is to examine the overall relation between excess firm value and

multinational operations Our proxy for the extent of multinational operations is the percentage

of total firm sales made by foreign operations (Foreign Sales) By this measure the extent of

multinational activity in our sample of MNCs is non-trivial ndash Table 2 Panel B shows that

approximately 24 percent of sales are generated by foreign operations on average Our first

analysis estimates ordinary least square regressions of Excess Value on Foreign Sales and

control variables (discussed below)

We include variables in our regression to control for other potential determinants of excess

value These are the percentage of sales made by the firm outside its primary industry (Industry

Other) to control for any relation between industrial diversification and excess value as in Berger

and Ofek (1995) For this purpose we obtain industry sales using Compustat Segment data19 A

firmrsquos primary industry is the industry in which the firm generates the majority of its sales and

we determine industry sales at the business segment level We set Industry Other equal to zero

for firms that operate in a single business segment

Consistent with prior research we also include controls for firm size (Log Size) the ratio of

long term debt to total market value (Debt Firm Value) the ratio of capital expenditures to total

sales (CAPX Sales) the ratio of earnings before interest and taxes to total sales (EBIT Sales)

the ratio of research and development expenses to total sales (RampD Sales) and the ratio of 19 Our results are unaffected by using BEA data to measure Industry Other

advertising expenses to total sales (Ad Sales) Table 2 Panel B provides the distribution of the

control variables All independent variables are Winsorized at 1 In our OLS regressions we

use heteroskedasticity and autocorrelation consistent robust standard errors with clustering at the

firm level

42 Primary regression analysis

In this section we examine the relation between firm excess value and the amount of foreign

activity at US MNCs Table 4 contains the results of regressions of excess value on Foreign

Sales for multiple specifications In column (1) Excess Value is regressed on both the percent of

foreign sales (Foreign Sales) and the percent of sales outside the firmrsquos primary industry

(Industry Other) We add the control variables to the regression in column (2) and year

indicators in column (3) In all three specifications the coefficient on the percent of foreign sales

remains significantly positive (with a p-value lt 001) For a sense of economic magnitude in

column (3) a one standard deviation (197) increase in the percent of foreign sales (Foreign

Sales) is associated with a 41 (equal to (e0349-1) times 0197) increase in firm excess value The

effect is larger for the other two columns ndash 82 increase in firm excess value in column (1) and

46 in column (2)20

Our finding of a premium reconciles with the international trade literature which finds that

firms engaging in international trade are more productive than those that do not (see Helpman

2006 and Syverson 2011 for reviews) Due to data limitations it is difficult to accurately

measure TFP at the firm level especially for firms which operate in multiple industries andor

20 Using the alternative strategy that employs US domiciled single-segment firms as the benchmark for the imputed value of the foreign operations Denis et al (2002) find that greater foreign activity results in a discount to firm value Using the Denis et al (2002) method to calculate excess value in our sample we obtain a significantly negative regression coefficient on Foreign Sales using the models in Table 4 columns (2) and (3) (the coefficient is not significant using the model in column (1) results untabulated)

countries As an alternative method to evaluate whether our finding of a premium is simply the

result of more productive firms also having more multinational operations we investigate

whether the premium is simply an artifact of innate characteristics of firms that choose to invest

abroad For example Helpman et al (2004) present a simple model with heterogeneous

productivity endowments The firms that receive the highest productivity endowments are the

ones capable of paying the fixed costs to establish a foreign subsidiary

To the extent that productivity advantages create excess value one would expect the type of

firms that choose to invest more abroad would be valued at a premium even in the absence of

their foreign investments To control for time invariant firm characteristics (eg innate

productivity advantages) and hence a potential source of endogeneity we add firm-level fixed

effects in column (4) of Table 4 We continue to find that the extent of foreign operations

(Foreign Sales) is positively associated with Excess Value Across all specifications the sign

on all significant control variables is consistent with prior research (eg Berger and Ofek 1995

and Denis et al 2002)

The finding of a premium makes a significant contribution to the relatively sparse literature

on the value effects of multinational activity Our measure of imputed value expands upon the

methods used by Berger and Ofek (1995) and Denis et al (2002) but allows us to answer a

different question than Denis et al (2002) Holding constant the extent of industrial

diversification we ask whether MNCs are more valuable than a portfolio of benchmark firms of

a similar geographic footprint In contrast Denis et al (2002) ask whether holding constant the

extent of industrial diversification MNCs are more valuable than a portfolio of domestic (US)

benchmark firms We believe that asking whether an MNCrsquos foreign (domestic) operations are

more valuable than those of a foreign (domestic) firm is a more appropriate way to assess the

value implications of a multinational network

5 Robustness

51 Endogeneity related to dynamic relations

Prior research argues that the relation between industrial diversification and excess firm

value could be endogenous through dynamic relationships based on observing past outcomes

Managers of firms likely choose to enter or exit industries or geographic regions based upon

their previous performance For example Campa and Kedia (2002) find that firms are more

likely to enter new industries when prospects in their current lines of business are deteriorating

Similar dynamics could also drive a relation between excess value and FDI We assess the

robustness of the previous results by estimating a dynamic panel data model which allows us to

control for dynamic endogeneity in addition to firm level unobserved heterogeneity We employ

the GMM estimator for dynamic panel models developed by Arellano and Bond (1991) Arellano

and Bover (1995) and Blundell and Bond (1998) Our results from this analysis confirm our

previous findings that firms with a larger degree of foreign operations have positive excess

values

Dynamic panel data models are generalizations of the traditional fixed effects model where

lags of the dependent variable are added to the right hand side of the equation These additional

lags control for the impact a firmrsquos past performance has on its current performance through the

channels discussed above The GMM estimator for dynamic panel models developed by

Arellano and Bond (1991) starts by differencing the data at the firm level in neighboring time

periods to eliminate any firm specific time invariant unobserved heterogeneity that may be

present Eliminating the firm specific fixed effect by differencing the data with lags of the

dependent variable on the right hand side introduces correlation between the new error term and

the differenced explanatory variables This problem is solved by applying the standard GMM

estimator using lags of both the original explanatory variables and excess value as instruments

We note that the dynamic panel data framework allows us to treat both international and

industrial diversification (as well as the other firm-specific control variables) as endogenous

Following Hoechle et al (2012) we also allow contemporaneous values of these variables to be

treated as endogenous

Lagged values of the explanatory variables are valid instruments in a dynamic panel data

model when the model is dynamically complete (eg Wooldridge 2010) A dynamic panel data

model is dynamically complete conditional on the unobserved effect if enough lags of both the

explanatory and dependent variable are included to make the resulting error term uncorrelated

with all the right hand side variables Under this assumption any lags for the explanatory

variables beyond this can be used within the GMM estimator as instruments Like Hoechle et al

(2012) we add two years of lags of a firmrsquos excess value to the right hand side assuming that

this is enough lags to control for dynamic feedback effects We then use past values of all

variables beyond the second lag up to a total of seven years to act as instruments

The results of the dynamic panel data estimator are reported in Table 5 These results confirm

the previous finding of a positive relation between the degree of international operations and

excess value We note that the size and significance level of the coefficient on Foreign Sales

diversification are robust to changes in the number of lags used as instruments The Arellano and

Bond (1991) estimator includes testing diagnostics for correct specification of the model If the

model is correctly specified differences in the residuals from the levels equation should be

serially uncorrelated at lag one but correlated at lag two The p-values for tests of serial

correlation of the residuals at lags 1 and 2 are reported in Table 5 and these indicate that the

model is not rejected We also report in Table 5 the Hansen test for overidentification which is a

test of the null hypothesis that the instruments are valid The p-value for this test is 030

indicating that we fail to reject this null hypothesis Finally we report another test for exogeneity

known as the difference-in-Hansen test which tests the null hypothesis that the instruments in

the original equation are valid We fail to reject this null hypothesis as well

52 Identifying value effects of multinational operations via cross-border acquisitions

Another alternative to investigating a causal link between foreign direct investment and firm

value is an event study of foreign acquisitions21 Acquired foreign targets represent only a small

fraction of MNCsrsquo worldwide sales in the year of the acquisition ndash less than 1 percent in our

sample Additionally cross-border acquisitions are not the primary source of international

expansion for the average firm We find that foreign acquisitions account for between 18 percent

and 55 percent of the year-over-year change in foreign sales with the remainder of the growth

coming from either newly established entities (ie greenfield investments) or growth in existing

operations22 These results suggest that acquisitions are not a material source of growth for the

21 Doukas and Travlos (1988) find positive returns when the acquirer is not operating in the target firmrsquos country and insignificant returns when the acquirer is operating in the target firmrsquos country or expanding internationally for the first time Morck and Yeung (1991) find marginal evidence of positive returns for firms with intangible assets Finally Doukas (1995) finds positive returns only for firms with average Tobinrsquos q ratios greater than one More recent work also using an event study methodology provides mixed evidence on valuation effects of global diversification through acquisitions Dos Santos Errunza and Miller (2008) find insignificant returns for firms with already established foreign operations firms establishing initial foreign operations and firms acquiring targets in related industries Doukas and Kan (2006) and Moeller and Schlingemann (2005) find reductions to shareholder value while Francis Hasan and Sun (2008) find positive returns for acquirers of targets in segmented financial markets 22 New foreign affiliates to an MNC group each year report on the BEA survey whether they entered the group as an acquisition or a newly established entity (ie greenfield investment) Affiliates under the lowest reporting threshold do not report this information resulting in our inability to precisely identify the source of 33 percent of the increase in firm sales as either acquisition or greenfield

typical US MNC suggesting that an event study would lack power and generalizability For

these reasons we do not implement an event study

53 Control premium

A control premium or private benefits of control occurs when ldquosome value whatever the

source is not shared among all the shareholders in proportion of the shares owned but it is

enjoyed exclusively by the party in controlrdquo (Dyck and Zingales 2004 p 541) Using stock

prices of traded shares held by minority shareholders in determining the market value to sales

ratios of our benchmark firms could underestimate the true value of our benchmark firms if those

firms face large control premiums This in turn could induce an excess value premium because

the imputed value of these country-industry segments would be artificially low The existence of

such benefits is more prevalent in some countries than others and could affect the value of an

MNCrsquos operations in such domains We use the country-level control premiums from Dyck and

Zingales (2004) and using each firmrsquos country sales as a percent of total firm sales we obtain a

weighted average control premium that each MNC faces (Control Premium) For countries not

included in the Dyck and Zingales (2004) analysis we set the control premium equal to zero23 In

Table 6 we show that Foreign Sales remains positive and highly significant (p-value lt 001)

when we include the proxy for control premium As a result it does not appear that the excess

value premium is driven by MNCs operating in countries where majority shareholders are able to

extract large control premiums

23 Results are robust to replacing missing values with the mean control premium of the respective GDP quartile

54 Cost of capital

As discussed in Section 2 when countries have shallow capital markets a multinational

network may provide the benefit of a lower cost of capital for investments through an internal

capital market as well as through better access to the US capital market Prior research

establishes that there is substantial variation in the cost of capital across countries (Erb Harvey

and Viskanta 1996 and Hail and Leuz 2009) To determine whether the documented premium

is due to cost of capital advantages we include a control variable to capture the difference

between the MNCrsquos cost of capital and the cost of capital faced by its competitors in the global

markets in which it operates

We compute our Cost of Capital proxy in two steps First we obtain the country-year credit

rating published by Institutional Investor24 The rating ranges from 0 to 100 where higher values

imply that a country has a higher default risk Country-level credit risk is a reasonable predictor

of expected equity market returns and volatility (Erb et al 1996) and exhibits a highly

significant correlation with international accounting-based estimates of imputed cost of capital

(Hail and Leuz 2009) Second using each firmrsquos country sales as a percent of total firm sales

we obtain a weighted average credit risk rating across the countries in which each firm operates

A higher value implies an MNC operates in countries with higher cost of capital In Table 7 we

do not find a result consistent with MNCs gaining an advantage from their access to low cost

capitalndash Cost of Capital is not significantly associated with Excess Value25

24 We thank Cam Harvey for providing these data 25 Results are consistent using cost of capital as defined in Erb et al 1996

55 Corporate governance

In light of the recent finding that corporate governance mitigates the diversification discount

we examine whether the premium to multinational activity is sensitive to controls for corporate

governance Hoechle et al (2012) find that the magnitude of the industrial diversification

discount is decreased and in several specifications no longer significant in the presence of

governance proxies In the first column of Table 8 we include the five governance proxies that

are consistently significantly associated with excess value in Hoechle et al (2012) ndash the percent

of CEO ownership (CEO Ownership and CEO Ownership squared) the percent of institutional

ownership (Institution Ownership) whether the firm has a powerful or influential CEO

(Powerful CEO) and the governance index from Gompers Ishii and Metrick (2003)

(Governance Index)

As documented in Table 8 including these control variables reduces our sample size by

almost 50 (to 2550 observations) and tends to exclude the smaller firms from the analysis

None of the governance proxies are significant Relative to column (4) of Table 4 the coefficient

on Foreign Sales is similar in magnitude but is no longer statistically significant at

conventional levels To investigate whether the decreased significance is due to the smaller

sample or the inclusion of the governance variables we re-estimate the regression for the same

2550 observations after removing the governance controls The coefficient and t-statistic are

virtually identical suggesting that the loss of significance is due to the reduced sample and not

the governance controls We do not find evidence to suggest that variation in corporate

governance practices drives variation in excess value for US MNCs

6 Conclusion

Using a new method to quantify the valuation effects of foreign direct investment by US

domiciled firms we find robust evidence of a premium in excess value associated with the

degree of multinational operations Our method builds on that used in previous studies

examining firm excess value (eg Berger and Ofek 1995 and Denis et al 2002) by

incorporating the location in addition to the industry of each corporate affiliate This allows

important factors such as growth and cost of capital to vary across countries as well as industries

Examining a sample of multinational firms we find that a one-standard deviation (197)

increase in foreign operations is associated with an increase of between 41 and 82 in excess

value minus both a statistically and economically significant finding

This result complements the finding of the international trade literature that more productive

firms are more likely to operate internationally However the premium remains after including

firm fixed effects to control for persistent firm traits (eg productivity)

The finding of a premium stands up to a succession of robustness tests designed to evaluate

whether the result arises from alternative methods or explanations First we estimate a dynamic

panel data model using the generalized method of moments (GMM) estimator to simultaneously

account for the potential endogeneity of FDI industrial diversification and firm value Second

we include a control variable to proxy for whether the firm operates in countries with larger

control premiums which could result in understated multiples and mechanically induce a

premium Third we include a control variable to measure the cost of capital across the countries

in which an MNC operates to evaluate whether a relatively low cost of capital borne by

multinational firms is a meaningful source of the premium Fourth we control for several

corporate governance proxies which have been found to be correlated with excess values (and

industrial diversification) We fail to find evidence supporting any of these alternative

explanations ndash the premium in excess value arising at firms with greater multinational operations

remains significant

Overall our new method of estimating excess values for multinational firms provides strong

evidence of a statistically and economically significant premium associated with increased

international operations The premium is robust to a host of specifications and controls designed

to evaluate alternative explanations We leave to future research the questions of whether the

premium to US multinational firms applies more broadly to other countries of domicile and

what factors influence the premium

References

Ammann M Hoeschel D and M Schmid 2012 Is there really no conglomerate discount

Journal of Business Finance and Accounting 39(12) 264-288 Arellano M and Bond S R 1991 Some tests of specification for panel data Monte Carlo

evidence and an application to employment equations Review of Economic Studies 58 (2) 277-298

Arellano M and Bover O 1995 Another look at instrumental variables estimation of error

components models Journal of Econometrics 68 (1) 29-61 Bens D A Berger P G and S J Monahan 2011 Discretionary disclosure in financial

reporting An examination comparing internal firm data to externally reported segment data The Accounting Review 86(2) 417-449

Berger P G and E Ofek 1995 Diversificationrsquos effect on firm value Journal of Financial

Economics 37 39-65 Blundell R and Bond S R 1998 Initial conditions and moment restrictions in dynamic panel

data models Journal of Econometrics 87(1) 115-144 Campa J M and S Kedia 2002 Explaining the diversification discount The Journal of

Finance 57 (4) 1731-1762 Denis D J Denis D K and K Yost 2002 Global diversification industrial diversification

and firm value The Journal of Finance 57 (5) 1951-1979 Desai M A Foley C F and J R Hines 2004 A multinational perspective on capital structure

choice and internal capital markets The Journal of Finance 59 (6) 2451-2487 Dos Santos MB V Errunza and D Miller 2008 Does corporate international diversification

destroy value Evidence from cross-border mergers and acquisitions Journal of Banking and Finance 32(12) 2716-2724

Doukas JA 1995 Overinvestment Tobinrsquos q and gains from foreign acquisitions Journal of

Banking and Finance 19 1285-1303 Doukas J and N G Travlos 1988 The effect of corporate multinationalism on shareholdersrsquorsquo

wealth evidence from international acquisitions Journal of Finance 43(5) 1161-1175 Doukas JA and O B Kan 2006 Does global diversification destroy firm value Journal of

International Business Studies 37(3) 352-371

Dunning J and A Rugman 1985 The influence of Hymerrsquos dissertation on the theory of foreign direct investment The American Economic Review 75(2) 228-232

Dyck A and L Zingales 2004 Private benefits of control An international comparison The

Journal of Finance 59 (2) 537-600 Erb C Harvey C and T Viskanta 1996 Expected returns and volatility in 135 countries The

Journal of Portfolio Management 46-58 Foley C F Desai M A and K J Forbes 2008 Financial constraints and growth

Multinational and local firm responses to currency depreciations Review of Financial Studies 21 (6) 2857-2888

Francis B I Hasan and X Sun 2008 Financial market integration and the value of global

diversification Evidence for US acquirers in cross-border mergers and acquisitions Journal of Banking and Finance 321522-1540

Gande A Schenzler C and L Senbet 2009 Valuation effects of global diversification

Journal of International Business Studies 1515-1532 Gompers P Ishii J and A Metrick 2003 Corporate governance and equity prices Quarterly

Journal of Economics 118(1) 107-156 Hail L and C Leuz 2009 Cost of capital effects and changes in growth expectations around

US cross-listings Journal of Financial Economics 93 428-454 Hanson G Mataloni R and M Slaughter 2005 Vertical production networks in multinational

firms Review of Economics and Statistics 87(4) 1-15 Helpman E 2006 Trade FDI and the Organization of Firms Journal of Economic Literature

44(3) 289-630 Helpman E Melitz M J and S R Yeaple 2004 Export versus FDI with Heterogeneous Firms

American Economic Review 94(1) 300-316 Hoechle D Schmid M Walter W and D Yermack 2012 How much of the diversification

discount can be explained by poor governance Journal of Financial Economics 103 41-60 Hope O-K and W B Thomas 2008 Managerial empire building and firm disclosure Journal

of Accounting Research 46 (June) 591-626 Huizinga H P and J Voget 2009 International taxation and the direction and volume of cross-

border MampAs Journal of Finance 64 (3) 1217-1249 Jensen M 1986 Agency cost of free cash flow corporate finance and takeovers American

Economic Review Papers and Proceedings 76 323-329

Laeven L and R Levine 2007 Is there a diversification discount in financial conglomerates

Journal of Financial Economics 85 331-367 Lang L and R Stulz 1994 Tobinrsquos Q corporate diversification and firm performance Journal

of Political Economy 102 1248-1280 Mataloni R 2003 US multinational companies Operations in 2001 Survey of Current

Business 85-105 Matvos G and A Seru 2011 The dynamic tradeoff between corporate socialism and financial

market dislocation Estimates from internal capital markets Working paper University of Chicago Booth School of Business

Moeller SB and F P Schlingemann 2005 Global diversification and bidder gains A

comparison between cross-border and domestic acquisitions Journal of Banking and Finance 29 533-564

Morck R and B Yeung 1991 Why investors value multinationality Journal of Business 64

165-186

Morck R and B Yeung 2001 Why firms diversify Internatization vs agency behavior In Intangibles B Lev (ed) Oxford University Press pp 269-302

Myers S C and N S Majluf 1984 Corporate financing and investment decisions when firms have information that investors do not have Journal of Financial Economics 13 187-221

Schmid M and I Walter 2009 Do financial conglomerates create or destroy economic value

Journal of Financial Intermediation 18 193-216 Syverson C 2011 What determines productivity Journal of Economic Literature 49(2) 326-

365 Villalonga B 2004a Diversification discount or premium New evidence from the Business

Information Tracking Series Journal of Finance 59 479-506 Villalonga B 2004b Does diversification cause the diversification discount Financial

Management 33 5-27 Wooldridge JM 2010 Econometric Analysis of Cross Section and Panel Data (Second

Edition) Cambridge MA MIT Press Yamin M 1991 A reassessment of Hymerrsquos contribution to the theory of the transnational

corporation C Pitelis and R Sugden (eds) The Nature of the Transnational Firm London Routledge

Appendix A Variable Definitions

We obtain information regarding firmsrsquo operations by country and industry from the Bureau of Economic Analysis (BEA) unless otherwise noted Acronyms used below reflect Compustat mnemonics and thus indicate when we obtain data from the Compustat North America Fundamentals Annual database All independent variables are winsorized at 1

Dependent variable

Excess Value = The log of the ratio of actual firm value to imputed firm value Actual firm value is equal to market capitalization (MVE) + book value of total assets (AT) ndash book value of shareholdersrsquo equity (SEQ) Imputed firm value is calculated as the sum of the imputed values of a firmrsquos operations in each country-industry The imputed value for each country-industry is equal to sales generated in that country-industry times a multiple equal to the ratio of a benchmark firmrsquos actual value to total sales A benchmark firm is defined as a median single-segment firm operating in the same country-industry All multiples require at least five firms listed in the respective country-industry on Compustat (domestic) or Worldscope (foreign) with less than 10 of sales income and assets outside the country of domicile Similar to Berger and Ofek (1995) we eliminate extreme values where the ratio of actual to imputed firm value exceeds 400 or is less than 025

Independent variables

Foreign Sales = Sales made by the foreign operations of US MNCs as a percent of total firm sales

Industry Other = Sales outside a firmrsquos primary industry as a percent of total sales (SALE) Equal to zero for firms operating in a single business segment

Log Size = Log of total sales (SALE)

Debt Firm Value = Ratio of long-term debt (DLTT) to total market value (market capitalization (MVE) + book value of total assets (AT) ndash book value of shareholdersrsquo equity (SEQ))

CAPX Sales = Ratio of capital expenditures (CAPX) to total sales (SALE)

EBIT Sales = Ratio of earnings before interest and taxes (EBIT) to total sales (SALE)

RampD Sales = Ratio of research and development expenses (XRD) to total sales (SALE)

Ad Sales = Ratio of advertising expenses (XAD) to total sales (SALE)

Control Premium = A weighted-average (based on the sales in a given country as a percent of total firm sales) of the country-level control premium from Dyck and Zingales (2004)

Cost of Capital = A weighted-average (based on the sales in a given country as a percent of total firm sales) of the country-level credit rating from Institutional Investor (httpwwwinstitutionalinvestorcom) The country-level rating is available for 160 countries and ranges from 0 to 100 Higher values imply greater credit risk

CEO Ownership = Percent of shares owned by the CEO from Execucomp

Institution Ownership = Percent of shares owned by institutional investors from Thomson Reuterrsquos Institutional Holdings database

Powerful CEO = Indicator variable equal to one if the CEO is the only insider on the board of directors and also serves as the chairman and president based on data from RiskMetrics Directors database Equal to zero otherwise

Governance Index = Count measure of takeover defenses used in Gompers Ishii and Metrick (2003) from RiskMetrics Governance database As the governance data are not available for every year we follow Gompers Ishii and Metrick (2003) and assume that the firmsrsquo governance attributes are unchanged until publication of the subsequent series

Figure 1 Trends in multinational activity and industrial diversification from 1984-2008 This figure presents a plot of the mean percent of foreign sales and the mean percent of sales outside the firmrsquos primary industry (industrial diversification) from 1984 through 2008 Dotted lines above and below the primary lines represent 95 confidence intervals To compile the plot we include the full population of firms in Compustat

Table 1 Analysis of accounting data for firms reporting under multiple accounting standards This table provides a comparison of sales net income and assets for firms that changed accounting standards to or from US GAAP Comparisons are provided for the 66 firms that changed to or from US GAAP and for which we could obtain accounting data electronically and in English under both standards for the same fiscal year (eg the firm presented the accounting numbers originally under one set of standards and then restated the prior year numbers in the subsequent year under the new set of standards) Panel A shows the distribution of firms by year Panel B shows the other accounting standards to which US GAAP is compared Panel C shows the mean and median percent that the other accounting standard differs from US GAAP for the same fiscal year and accounting number Two samples are shown in Panel C The first is the full sample of firms with a change in accounting standards to or from US GAAP for which data was available The second is the subset of these firms that occurred during the sample window used in this study (ie excluding changes post-2008) The percent difference in sales is compared to the percent difference in net income and assets Tests of means are based on matched-pair t-tests Tests of medians are based on Wilcoxon matched-pair signed-rank tests Panel A Accounting standard changes by year

Panel B Accounting standards compared to US GAAP

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

(Table 1 continued) Panel C Tests of differences between US GAAP and other accounting standards

Percent different from US GAAP

Comparison to sales (p-values)

N Mean Median

Mean Median

Full Sample (2001-2011)

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

Study Sample (2001-2008)

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Table 2 Descriptive statistics Panel A reports the mean and medians of Excess Value for the multinational firms and single-segment benchmark firms as well as univariate tests across the subsamples Panel B reports summary statistics for the independent variables for the multinational firms The p-values of the tests of differences in means are based on t-tests The p-values of tests of differences in the medians are for a Wilcoxon rank-sum test All variables are defined in Appendix A Panel A Excess Value by sample Multinational All Benchmark Benchmark Subgroups Sample Firms US Foreign

Value Value Test of Diff (p-value) Value

Test of Diff (p-value) Value

Test of Diff (p-value)

Mean 0067 0006 lt001 -0028 lt001 0014 lt001 Median 0076 0000 lt001 -0025 lt001 0000 lt001 N 4950 33074 5998 27076 Panel B Distribution of independent variables for the multinational firms N Mean 25th 50th 75th StDev Foreign Sales 4950 0238 0074 0190 0363 0197 Industry Other 4950 0212 0000 0166 0414 0215 Log Size 4950 6955 5823 6989 8140 1687 Debt Firm Value 4950 0136 0009 0097 0212 0145 CAPX Sales 4950 0054 0020 0033 0054 0101 EBIT Sales 4950 0065 0031 0077 0121 0138 RampD Sales 4950 0049 0000 0013 0059 0083 Ad Sales 4950 0008 0000 0000 0005 0018

Table 3 BEA and Worldscope samples

This table reports details on affiliate sales for significant countries pooled from 1998-2008 Significant countries are defined as those in which the operations of US MNCs represent at least 02 of total firm sales (per the BEA) For significant countries the table compares their representation in terms of MNC affiliate sales in the BEA data to the distribution of firm observations in Worldscope US Foreign Affiliates (BEA) Foreign Firms (Worldscope)

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Australia 1344 73305 056 859 255865 210 Belgium 894 52755 040 169 26136 021 Brazil 959 60245 046 1143 683167 560 Canada 2730 536767 407 1181 526750 432 China 922 75907 057 822 276541 227 France 1852 183942 139 540 105447 086 Germany 2182 275577 209 480 103888 085 Hong Kong 867 48100 036 157 34094 028 Ireland 478 36754 028 34 3467 003 Italy 1178 82116 062 111 47948 039 Japan 1494 148032 112 2281 1119223 917 Korea South 659 34500 026 3530 2752791 2255 Malaysia 386 26027 020 837 168683 138 Mexico 1405 113623 086 362 237173 194 Netherlands 1149 85105 064 47 22631 019 Singapore 1098 94917 072 139 32913 027 Spain 871 63094 048 129 37419 031 Sweden 712 32682 025 235 23090 019 Switzerland 726 60288 046 74 20242 017 Taiwan 592 30313 023 2949 787049 645 United Kingdom 3054 423934 321 1997 882423 723 United States (domicile) 4950 10460880 7923 na na na Other - Covered by WorldScope 6893 199394 151 9000 4057894 33 Other -Not Covered by WorldScope 713 5352 004 - - - Total 38108 13203609 1000 27076 12204834 100

Table 4 OLS regressions of firm excess values on the percent of foreign sales This table reports the results of OLS regressions of firm excess value on the percent of foreign sales (Foreign Sales) the percent of sales outside a firmrsquos primary industry (Industry Other) and control variables All variables are defined in Appendix A T-statistics and significance levels are computed using clustered standard errors with firm level clustering Significance levels are indicated by and representing 1 5 and 10 levels respectively (2-tailed) Dep Variable = Excess Value Coefficient t-stat Coefficient t-stat Coefficient t-stat Coefficient t-stat (1) (2) (3) (4) Foreign Sales 0349 419 0209 293 0189 266 0278 210 Industry Other -0134 -186 -0047 -072 -0053 -081 -0326 -325 Log Size 0066 634 0065 624 -0044 -096 Debt Firm Value -0196 -213 -0241 -260 -0685 -588 CAPX Sales 0901 425 0893 435 0488 227 EBIT Sales 1154 542 1170 547 0408 189 RampD Sales 2908 1242 2932 1237 1544 379 Ad Sales 1869 263 2003 280 -1696 -119 Year indicators No No Yes Yes Firm indicators No No No Yes R2 0017 0212 0226 0768 Observations 4950 4950 4950 4950

Table 5 Dynamic panel GMM model regressing firm excess values on the percent of foreign sales This table reports the result from a regression of the excess value on the percent of foreign sales (Foreign Sales) the percent of non-primary industry sales (Industry Other) and control variables using a dynamic panel GMM estimator All control variables are considered to be endogenous with the exception of the year indicator variables AR(1) and AR(2) are tests for first-order and second-order serial correlation in the first differenced residuals with the null hypothesis of no serial correlation The null hypothesis of the Hansen test of overidentification is that all instruments are valid The null hypothesis of the difference-in-Hansen test of exogeneity is that the instruments used for the equation in levels are exogenous All variables are defined in Appendix A T-statistics and significance levels are computed using clustered standard errors with firm level clustering Significance levels are indicated by and representing 1 5 and 10 levels respectively (2-tailed) Dep Variable = Excess Value Coef t-stat Excess Value (one lag) -0018 -082 Excess Value (two lags) -0048 -272 Foreign Sales 0371 270 Industry Other -0163 -171 Log Size 0010 123 Debt FirmValue -1135 -820 CAPX Sales 0590 342 EBIT Sales -0047 -300 RampD Sales 1376 531 Ad Sales -0751 -040 AR(1) test (p-value) 000 AR(2) test (p-value) 017 Hansen test of over-identification (p-value) 030 Difference-in-Hansen test of exogeneity (p-value) 093 Year indicators Yes Number of firms 1016 Number of observations 4090

Table 6 OLS regression of firm excess values on the percent of foreign sales controlling for a control premium This table reports the result of the OLS regression of firm excess value including firm and year fixed effects with an additional control variable to proxy for a country-level ownership control premium (Control Premium) All variables are defined in Appendix A T-statistics and significance levels are computed using clustered standard errors with firm level clustering Significance levels are indicated by and representing 1 5 and 10 levels respectively (2-tailed) Dep Variable = Excess Value Coef t -stat Foreign Sales 0303 224 Industry Other -0327 -325 Log Size -0045 -100 Debt FirmValue -0683 -585 CAPX Sales 0490 226 EBIT Sales 0409 189 RampD Sales 1545 379 Ad Sales -1689 -119 Control Premium -1004 -062 Year indicators Yes Firm indicators Yes R2 0768 Observations 4950

Table 7 OLS regression of firm excess values on the percent of foreign sales controlling for the cost of capital This table contains the results of an OLS regression of firm excess value including firm and year fixed effects with an additional control variable to proxy for a country-level cost of capital (Cost of Capital) All variables are defined in Appendix A T-statistics and significance levels are computed using clustered standard errors with firm level clustering Significance levels are indicated by and representing 1 5 and 10 levels respectively (2-tailed) Dep Variable = Excess Value Coef t -stat Foreign Sales 0255 178 Industry Other -0326 -324 Log Size -0043 -094 Debt FirmValue -0687 -589 CAPX Sales 0485 225 EBIT Sales 0406 188 RampD Sales 1539 379 Ad Sales -1703 -120 Cost of Capital 0004 039 Year indicators Yes Firm indicators Yes R2 0768 Observations 4950

Table 8 OLS regression of firm excess values on the percent of foreign sales controlling for aspects of corporate governance This table reports the results of two OLS regressions In the first column are the results of the firm excess value regression including firm and year fixed effect with five proxies for corporate governance included ndash the percent of shares owned by the CEO and the percent squared (CEO Ownership and CEO Ownership squared) the percent owned by institutional investors (Institution Ownership) whether there is a powerful CEO (Powerful CEO) and the Gompers Ishii and Metrick (2003) governance index (Governance Index) In the second column the regression includes the same sample and variables with the exception of the governance proxies which are excluded All variables are defined in Appendix A T-statistics and significance levels are computed using clustered standard errors with firm level clustering Significance levels are indicated by and representing 1 5 and 10 levels respectively (2-tailed)

Governance

No Governance (same sample)

Dep Variable = Excess Value Coef t-stat Coef t -stat Foreign Sales 0259 159 0251 154 Industry Other -0099 -080 -0099 -079 Log Size -0105 -185 -0102 -180 Debt FirmValue -0813 -478 -0817 -482 CAPX Sales 0440 252 0438 249 EBIT Sales 0457 247 0457 246 RampD Sales 1717 284 1716 286 Ad Sales -2210 -143 -2190 -143 CEO Ownership 0010 007

CEO Ownership squared -0029 -017 Institution Ownership 0047 067 Powerful CEO -0003 -014 Governance Index 0009 103

Year indicators Yes Yes Firm indicators Yes Yes R2 0811 0810 Observations 2550 2550

Cover for paperpdf

crrz_20130603_SMUpdf

We thank Phil Berger Andy Bernard Marianne Bertrand Alan Bester Chris Hansen Chang-Tai Hsieh Rafael La Porta Christian Leuz Jon Lewellen Abbie Smith Doug Skinner and Jerry Zimmerman for helpful discussions and comments This paper has benefited from the comments of workshop participants at Chicago Cornell Dartmouth Florida Illinois at Chicago INSEAD LBS Michigan MIT Northwestern Rochester Syracuse University Texas UCLA and the EAA Annual Congress We thank William Steciak of KPMG for helpful discussions related to segment reporting We gratefully acknowledge the financial support of the University of Chicago Booth School of Business the Fama-Miller Center and the Neubauer Family Foundation The statistical analysis of firm-level data on US multinational firms was conducted at the International Investment Division Bureau of Economic Analysis US Department of Commerce under arrangements that maintain legal confidentiality requirements The views expressed are those of the authors and do not reflect official positions of the US Department of Commerce

Drew D Creal

Leslie A Robinson Tuck School of Business at Dartmouth

Jonathan L Rogers

Sarah L C Zechman University of Chicago Booth School of Business

February 2013

Chicago Booth Research Paper No 11-37

Abstract Using a proprietary dataset we evaluate whether the degree of foreign operations affects firm value by comparing actual value to imputed value for US multinational corporations (MNCs) We argue that using benchmark firms operating in the same country and industry as each MNC segment controls for differences in discount rates and expected growth rates across countries and industries This allows us to isolate the value effects of organizing a set of otherwise independent activities within a multinational network We find robust evidence that multinational networks trade at a premium relative to a benchmark portfolio of independent firms JEL Classification F23 G32 G34 M41 Keywords firm valuation multinational corporations diversification

1 Introduction

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

1 Introduction

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

concerns

segments

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

32 Data

control variables

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

affiliate13

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

annually

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

4 Empirical results

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

firm level

46 in column (2)20

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

5 Robustness

values

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

53 Control premium

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

54 Cost of capital

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

(Governance Index)

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

6 Conclusion

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

remains significant

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

References

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

165-186

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Dependent variable

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Canadian GAAP

Israeli GAAP

Dutch GAAP

Indian GAAP

Japanese GAAP

Norwegian GAAP

Swiss GAAP

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

N Mean Median

Mean Median

66 236 000

Net Income

66 10653 647

005 lt001

Assets

66 1162 126

lt001 lt001

55 257 000

Net Income

55 10303 589

011 lt001

Assets

55 1047 099

003 lt001

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Country

Sales (millions)

Total Sales (3)

Sales (million)

Total Sales (6)

Governance

Cover for paperpdf

Governance

Cover for paperpdf

Governance

Cover for paperpdf

Governance

Cover for paperpdf

Governance

Cover for paperpdf

Governance

Cover for paperpdf

The Multinational Advantage - SMU School of Accountancy · for measuring valuation effects in a multinational context. 5 This method, in effect, compares the value of the firm as

Documents