Leverage-Mihov-2019-JIBS

Corporate internationalization, subsidiary locations, and the cost of equity capital
Author(s): Atanas Mihov and Andy Naranjo

Source: Journal of International Business Studies , DECEMBER 2019, Vol. 50, No. 9
(DECEMBER 2019), pp. 1544-1565
Published by: Palgrave Macmillan Journals on behalf of Academy of International Business
Stable URL: https://www.jstor.org/stable/10.2307/48686780
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Palgrave Macmillan Journals is collaborating with JSTOR to digitize, preserve and extend access to
Journal of International Business Studies
This content downloaded from

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
All use subject to https://about.jstor.org/terms
Journal of International Business Studies (2019) 50, 1544–1565
ª 2019 Academy of International Business All rights reserved 0047-2506/19
www.jibs.net
Corporate internationalization, subsidiary

locations, and the cost of equity capital
Atanas Mihov1 and Abstract

This study examines the relationship between corporate internationalization
Andy Naranjo2 and the cost of equity capital. We find that international diversification reduces
1
the cost of equity. The diversification benefits are particularly strong during the
Quantitative Supervision and Research, Federal
2008 financial crisis and for financially constrained firms. We also find that
Reserve Bank of Richmond, Charlotte, NC, USA;
2
Finance, Insurance, and Real Estate Department,
market-specific factors serve as important channels through which the
University of Florida, Gainesville, FL, USA corporate internationalization effects amplify or attenuate. Overall, our study
provides support for theories that multinational companies perform valuable
Correspondence: diversification functions to investors in a world with segmented and imperfect
A Mihov, Quantitative Supervision and financial markets.
Research, Federal Reserve Bank of Richmond, Journal of International Business Studies (2019) 50, 1544–1565.
Charlotte, NC, USA.
https://doi.org/10.1057/s41267-018-00207-3
Tel: +1 704 358 2464;
e-mail: atanas.mihov@rich.frb.org
Keywords: corporate internationalization; cost of equity capital; financial constraints;
subsidiary locations
INTRODUCTION
Research shows that the international operations of MNCs have
significant effects on firm value (e.g., Errunza & Senbet, 1981).
From a valuation perspective, such effects reflect the capitalization
of future profits and, thus, are derived primarily from two sources:
expected future earnings, and the required return of investors used
to discount future earnings back to the present. Although prior
research has hypothesized profitability effects of internationaliza-
tion (e.g., Fatemi, 1984; Ragazzi, 1973), the evidence for the effect
of internationalization on discount factors equity investors use, or
the cost of equity capital, remains indirect and inconclusive.1 This
is a critical issue as the cost of equity plays a central role in a firm’s
capital structure with first-order implications for corporate invest-
ment policies, financing of operations, and company growth.
In analyzing the effects of internationalization, many researchers
have focused on the real dimension of multinationality (e.g.,
Electronic supplementary mate- Caves, 1971; Morck & Yeung, 1991; Ragazzi, 1973). Specifically,
rial The online version of this article firms engage in international operations to diversify their real asset
(https://doi.org/10.1057/s41267-018-00207- portfolio and/or exploit managerial and technical expertise and
3) contains supplementary material, which is
available to authorized users.
economies of large-scale foreign production. Capital market theory,
Received: 20 January 2017 on the other hand, suggests little reason for shareholder wealth-
Revised: 10 September 2018 maximizing corporations to diversify their direct investments in
Accepted: 10 November 2018 plant and equipment – individual investors can diversify their
Online publication date: 4 January 2019

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
Corporate internationalization and the cost of equity capital Atanas Mihov and Andy Naranjo
1545
portfolios themselves. Thus, in a world with com- market development, and cultural proximity, serve
plete capital markets, the multinational firm does as important channels through which the cost of
not provide services to investors that they could equity effects of corporate internationalization
not otherwise provide for themselves. The empiri- amplify or attenuate. In addition, we show that
cal evidence in international finance, however, multinationals operating in foreign markets that
suggests that financial markets still face significant experience financial and economic turbulence
segmentation and are not fully integrated (e.g., could be susceptible to stress in those host markets.
Karolyi & Stulz, 2003; Lewis, 2011). Therefore, as Taken together, we provide unique evidence that
Errunza and Senbet (1981, 1984) advance, a glob- corporate subsidiary locations are important in
ally diversified company helps to ‘‘complete mar- affecting a firm’s cost of equity capital. By diversi-
kets’’ by allowing its investors indirect access to fying into markets with particular characteristics,
countries with restrictions on portfolio holdings, corporate managers can further reduce financing
thereby allowing investors access to multinational costs or reverse diversification benefits. Our find-
portfolio diversification. This should be reflected in ings are also helpful in reconciling theories in the
the multinational firm’s cost of equity capital. literature that internationalization exposes MNCs
The purpose and contributions of this study are to market-specific risks and, thus, have risk-increas-
twofold. First, this study provides direct and novel ing effects with potentially adverse effects on the
evidence for the effect of international operations cost of equity capital (e.g., Burgman, 1996; Cuervo-
on the required rate of return of equity investors at Cazurra, Maloney, & Manrakhan, 2007).
corporations and the channels through which the An important advantage of our analysis is that we
linkages occur. Using cost of equity estimates from use a granular, accurate, and comprehensive mea-
a discounted-dividend valuation model that cali- sure of corporate internationalization. Specifically,
brates current prices and analysts’ earnings fore- we use data from regulatory disclosures (Exhibit 21
casts, we find that internationalization lowers the in form 10-K), which the U.S. Securities and
financing costs of corporations.2 The channels Exchange Commission (SEC) enforces, and we
support the ‘‘incomplete world financial market measure corporate internationalization by the
theory,’’ in which market segmentation and imper- number of distinct foreign, non-U.S. countries
fections give rise to MNC comparative advantages where a firm has material operations. Our measure
in providing diversification benefits to investors. overcomes significant limitations of the most
In addition to documenting an overall risk- widely used measures of corporate international-
reducing effect of internationalization, we present ization constructed from geographic segment sales
direct evidence that operations in less integrated data, in which the extent of disaggregation in
markets with lower financial market correlations segment financial reporting is substantially lower
are important for risk diversification and lowering than the true extent of a firm’s segment diversifi-
financing costs. We also advance the notion that cation and the reporting structure across firms is
MNCs have better access to equity financing nonuniform (Villalonga, 2004).
because local presence and operations reduce fric- We have organized the remainder of the paper as
tions in segmented international equity markets. follows. The next two sections discuss related
Consistent with such arguments, using the 2008 research and develop our empirical hypotheses.
financial crisis as a natural experiment, we show We then describe our data sample and lay out our
that more internationally diversified firms are bet- main results. The last two sections check for
ter able than less diversified firms to weather local robustness and present our conclusions.
adverse economic conditions and capital market
dislocations in their home country. Firm charac-
teristics play an important role regarding the RELATED LITERATURE
benefits of diversification. We show that financially This paper contributes to several important
constrained firms, which are to a greater extent research streams. First, we advance the ongoing
forced to finance their investment internally and debate on the effects of corporate internationaliza-
are more likely to incur deadweight costs from low tion on the cost of equity capital. Although early
cash-flow states of the world, benefit proportion- studies suggest that firm internationalization low-
ately more from the effects of internationalization. ers the cost of equity capital (e.g., Fatemi, 1984;
Second, we show that market-specific character- Hughes, Logne & Sweeney, 1975), other studies
istics, such as the quality of governance, financial find no relationship or suggest that firm

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1546
internationalization raises the cost of equity capital capital. Errunza, Hogan, and Hung (1999) suggest
through increased risk channels (e.g., Amihud, that portfolios of domestically traded securities
Bartov, & Wang, 2015; Bartov, Bodnar, & Kaul, can mimic foreign market indices so that invest-
1996; Brewer, 1981; Reeb, Kwok, & Baek, 1998). We ment in assets only traded abroad is not required
contribute to the literature by providing additional to obtain international diversification gains. Con-
and more direct evidence for the effects of interna- sistent with the claims in Agmon and Lessard
tional operations on firms’ equity financing costs. (1977), Jang (2017), and Manova, Wei, and Zhang
Our results suggest that one of the channels for the (2015) that (1) U.S. investors, seeking diversifica-
positive valuation effects of internationalization tion but facing relatively less efficient capital
(Errunza & Senbet, 1981, 1984; Gande, Schenzler, & markets abroad and barriers imposed by legal
Senbet, 2009) is lower discount rates used by equity restrictions, could diversify by purchasing U.S.-
investors. Our direct and informative evidence on based multinationals with claims on foreign oper-
the mechanisms behind this linkage supports the ations, and (2) MNCs have greater access to
interpretation that MNCs provide valuable diversi- capital, we find that multinationals with foreign
fication services in a world with incomplete capital operations face lower costs of equity capital. In
markets. This research also provides unique evi- addition, we show direct evidence that globally
dence that corporate subsidiary locations are diversified firms with operations in countries with
important factors amplifying or attenuating the more developed equity markets have lower costs
cost of equity consequences of international of equity.
operations. Finally, our study contributes to the literature
Second, our research also contributes to an streams linking foreign market characteristics and
important literature stream on the deadweight economic outcomes. By showing a direct link
costs of external finance and how such costs can between a firm’s cost of equity capital and country
be mitigated. Livdan, Sapriza, and Zhang (2009) governance, financial market development, and
show financially constrained firms face greater risk cultural proximity, we contribute to the literature
and earn higher expected returns. Kuppuswamy on the effects of institutional quality on financial
and Villalonga (2016) show that the value of markets (e.g., Bartram, Brown, & Stulz, 2012;
diversified conglomerates increased during the Fauver, Houston, & Naranjo,2003; La Porta,
2008 financial crisis due to both financing and Lopez-de-Silanes, Shleifer, & Vishny, 1998; Pastor
investment advantages. Yan (2006) suggests that & Veronesi, 2013), market segmentation, financial
diversified firms can substitute external with inter- development and access to capital (e.g., Jang, 2017;
nal capital markets when financing costs in exter- Karolyi & Stulz, 2003; Lewis, 2011), and cultural
nal markets are high. Our study complements this differences and economic decision making (e.g.,
literature by showing differentially stronger cost of Ahern, Daminelli, & Fracassi, 2015, Giannetti &
equity effects of internationalization during periods Yafeh, 2012, Kogut & Singh, 1988).
of capital market stress. Specifically, we show that
diversified companies with foreign operations faced
lower cost of equity and return volatility relative to HYPOTHESES
geographically focused firms over the 2008 crisis We base our hypotheses of the effects of interna-
period. As a moderating effect, however, we also tionalization on firms’ cost of equity on the coin-
show that multinationals are susceptible to finan- surance models of Lewellen (1971) and Hann,
cial and economic stress in the host markets in Ogneva, and Ozbas (2013). The models assume
which they operate. In addition, consistent with that companies experience deadweight costs (e.g.,
the Livdan et al. (2009) findings that financially foregone business value due to defection of key
constrained firms are riskier and more likely to stakeholders, financial distress or external finance
incur deadweight costs from financial distress, we costs, reputation impairment) when their invest-
find stronger diversification effects of foreign oper- ments yield low cash-flow outcomes. The imperfect
ations for financially constrained firms. correlation of geographic units allows resource
Third, our research contributes to the literature reallocation from cash-flow-rich to -poor units in
on market segmentation and the cost of capital. certain states of nature. Similarly, diversified fund-
Bekaert and Harvey (2000), Errunza and Miller ing sources should help companies dissipate fund-
(2000), and Henry (2000) argue that international ing shocks to specific geographic units and reduce
market segmentation is relevant for the cost of the impact of local economic storms. As a result,

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1547
MNCs could avoid deadweight losses that more countries of poor governance have higher costs of
focused firms are unable to avoid.3 equity due to exposure to various forms of risk (e.g.,
regulatory, legal, political).
Hypothesis 1: Diversified MNCs, on average,
have lower risks and costs of equity than purely Hypothesis 4: MNCs with operations in coun-
domestic firms and less diversified MNCs. tries with weak governance have higher cost of
equity capital.
Coinsured cash flows and funding sources miti-
gate adverse economic conditions and capital mar- In perfectly integrated and frictionless global
ket dislocations in specific local markets and markets, the locations of MNCs’ assets would be
especially MNCs’ domestic market.4 irrelevant. However, the empirical evidence indi-
cates that financial markets, and especially equity
Hypothesis 2: Diversified MNCs have incre-
markets, still face significant segmentation across
mentally lower risk and cost of equity during
countries, with local capital markets continuing to
domestic crises.
play important roles in capital raising (e.g., Karolyi
Financially constrained firms find it difficult to & Stulz, 2003; Lewis, 2011). Financial market
finance projects (even worthy ones) with external segmentation suggests obtaining funding from
capital. With banks and outside investors reluctant international capital markets is often difficult and
to fund them, such firms are to a greater extent costly. To the extent that foreign operations reduce
forced to internally finance their corporate invest- the impact of frictions in international equity
ment. Financially constrained firms are more likely markets, firms’ local presence can affect their cost
to incur deadweight costs from low cash-flow states of equity capital. Further, these effects should be
of the world and should benefit more from the especially pronounced for MNCs with presence in
coinsurance effects of corporate countries with well-developed equity markets.
internationalization.
Hypothesis 5: MNCs with operations in coun-
Hypothesis 3: The effects of corporate inter- tries with more developed equity markets have
nationalization on the cost of equity capital are lower cost of equity capital.
more pronounced for financially constrained
The relevance of local markets for capital financ-
firms.
ing also suggests that dislocations in such markets
While cash-flow coinsurance through interna- could have adverse consequences for MNCs.
tionalization can on average decrease the cost of
Hypothesis 6: Host market financial shocks
equity for diversified firms, the specific character-
increase MNCs’ cost of equity.
istics of diversification destinations could also play
an important role in financing costs. For example, Additional local market factors could also affect a
risk-enhancing, market-specific factors could firm’s cost of capital. Culture, for example, could
reduce or even eliminate cost of equity reductions have effects through multiple channels. First, local
that could have otherwise accrued from the cash- culture spawns risk for MNCs (e.g., operations and
flow coinsurance dimension of multinationality. product demand risks) as they learn and adapt to
We hypothesize cost of equity effects for the the local markets’ languages, preferences, and
following country-level characteristics: governance informal institutions (e.g., Li & Guisinger,
quality, financial market development, financial 1991, 1992). Second, culture influences invest-
and economic crises, and cultural proximity. ments (Grinblatt & Keloharju, 2001). With the
Prior research suggests that a country’s gover- relevance of local capital markets in fund-raising,
nance quality is related to MNC risk. Desai, Foley, cultural differences could affect the supply of
and Hines (2008) argue that MNC investments in equity capital to MNCs due to investor cognitive
politically risky countries are riskier than invest- biases that favor the culturally familiar. And third,
ments elsewhere. Lin, Mihov, Sanz, and Stoyanova different cultural norms obstruct financial contract
(2018) show that MNCs operating in countries that negotiations (e.g., Giannetti & Yafeh, 2012) and
experience expropriations incur negative valuation could also inhibit information transfer and increase
shocks related to spikes in perceived risk. We asymmetry around MNCs, limiting their access to
hypothesize that firms with subsidiaries in external capital in local markets.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1548
Hypothesis 7: MNCs with operations in cul- realized and expected returns is low. Consequently,
turally distant countries from the U.S. have we follow Ohlson and Juettner-Nauroth (2005) and
higher cost of equity capital. estimate an implied cost of equity measure as the
discount rate that equates stock prices to expected
future cash-flows6 :
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DATA, VARIABLE DEFINITIONS Et ðEPStþ1 Þ
AND DESCRIPTIVE STATISTICS CoE ¼ A þ A2 þ ðg2 glt Þ; where
Pt
ð1Þ
Data Sources and Sample Selection DPStþ1
Our measure of internationalization is based on A ¼ 0:5 glt þ
Pt
SEC form Exhibit 21, which reports a list of the
names and countries of incorporation (or organi- Pt is the market price of a firm’s common stock at
zation) for all significant foreign subsidiaries of a time t. DPStþ1 is the expected dividend per share at
U.S. corporation at the annual frequency. SEC Rule time t þ 1, calculated as ordinary dividends (Com-
S-K, Item 601(b)(21) mandates the inclusion of pustat data item DVC) divided by common shares
Exhibit 21 in companies’ annual 10-K filings. Under outstanding (Compustat data item CSHO).
Rule 1-02(w) of Regulation S-X, a subsidiary is Et ðEPStþ1 Þ is the analyst forecasted (at time t)
deemed ‘‘significant’’ if it accounts for more than earnings per share at time t þ 1 from IBES. g2 is the
10% of the parent company’s investment, assets, or short-term earnings growth rate implied in EPStþ1
income. According to Item 601, ‘‘the names of and EPStþ2 . Finally, the long-term earnings growth
particular subsidiaries may be omitted if the rate, glt , is assumed at 4% (our results are also robust
unnamed subsidiaries, considered in the aggregate to alternative earnings growth rate assumptions).
as a single subsidiary, would not constitute a The implementation of this model requires that
significant subsidiary.’’ Alternatively stated, Exhibit EPStþ1 [ 0 and EPStþ2 [ 0. To account for changes
21 provides comprehensive coverage as the omitted in risk-free rates, we adjust the cost of equity capital
subsidiaries cannot exceed 10% of the parent’s by subtracting the prevailing yield on the 10-year
investment, assets, or income in the aggregate. T-bond. We name this variable CoE.
We start with all firms in Compustat between
1996, when most firms begin to have electronic 10- Corporate Internationalization
K filings, and 2012. We require non-missing CIK Based on the data collected from firms’ Exhibits 21,
identifiers so that we can link firms to data from the we extract the number of distinct non-U.S. coun-
SEC’s website. We then match Compustat firm- tries where firms operate, which we then log-
years to annual data gathered from Exhibit 21 in transform to reduce the influence of extreme values
firms’ 10-Ks. We use a text search algorithm to scan and call it Ln(NC).7 We identify 36,823 firm-year
Exhibit 21 tables, identify, and count all distinct observations with requisite data, of which 21,279
foreign countries in which firms report operations.5 have foreign operations. Ln(NC) overcomes limita-
We then obtain additional data from several tions of the most widely used internationalization
sources, including data used for the estimation of measures constructed from Compustat Segment.
implied cost of equity from IBES, stock return data Denis, Denis, and Yost (2002) point out that
from CRSP, financial statement data from Compus- regulators do not mandate or enforce a uniform
tat, and foreign exchange and interest rate data standard for reporting foreign operating segments
from the Federal Reserve. The intersection of these (SFAS No 131). Indeed, while some firms may
databases yields our main sample that consists of report segment data for different countries, others
267,149 firm-subsidiary-year observations (36,823 report data for different geographic areas (e.g.,
firm-years and 6424 unique firms). We winsorize ‘‘Eastern Europe’’) or even continents. Further, the
variables at their 1st and 99th percentiles to mitigate data classifications are not necessarily consistent or
outlier effects and data input errors. comparable across firms. Consequently, using the
number of reported geographic segments to proxy
Implied Cost of Equity Capital Estimates for geographic concentration has substantial limi-
Elton (1999) and Hann et al. (2013), among others, tations. The granularity, consistency and compara-
argue that inferring risk premium from realized bility across companies of Ln(NC) overcome such
returns is difficult as the correlation between limitations. In addition, the identification of the

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1549
countries where firms operate allows us to docu- exchange volatility, product line and customer
ment heterogeneous effects of internationalization concentrations. Table 1 provides variable defini-
on firms’ cost of equity based on country tions. Our regressions include Fama–French 48
characteristics. industry and year fixed effects. We also use two-
dimensional clustering of standard errors.
Descriptive Statistics and Pairwise Correlations Table 3 presents the results. Consistent with
Table 1 provides variable definitions and Table 2, Hypothesis 1, Column (1) shows that geo-
Panel A presents descriptive statistics. The average graphic diversification decreases the cost of
cost of equity capital in excess of the risk-free rate, equity: a one standard deviation increase in
CoE, for the firms in our sample is 10.08%, while Ln(NC) decreases cost of equity by 24 basis
the 25th; 50th and 75th percentiles are 5.35, 7.54 and points (significant at the 1% level). Given an
11.56%, respectively. The standard deviation is unconditional CoE mean of 10.08% points, this
8.48%. Close to 58% of the observations in our constitutes a decrease of 2.4%. Examining the
sample indicate foreign presence. On average, firms coefficients of control variables, we confirm
have operations in 7 foreign countries during a their signs are consistent with recent literature
given year. (e.g., Dhaliwal et al., 2016).
Next, we examine pairwise correlations. Table 2, We next test Hypothesis 2 by specifically exam-
Panel B shows a strong negative association ining whether geographic diversification has differ-
between the cost of equity, CoE, and corporate ential effects on the cost of equity during the 2008
internationalization, Ln(NC). The correlation coef- financial crisis. FinCrisis equals 1 if the fiscal-year
ficient is - 0.07, significant at the 1% level. This end is between July 1, 2008 and January 1, 2010 (0
relation is also visually confirmed by Figure 1, otherwise). Interaction terms of FinCrisis and
which provides equally-weighted mean/median Ln(NC) measure the incremental effects of interna-
plots split by ownership of foreign subsidiaries. tionalization during the crisis. Providing support
There is also a negative and significant association for Hypothesis 2, Table 3, Columns (2) and (3)
between firm risk, measured as total stock return show that internationalization effects are especially
volatility (RetVol), and internationalization. While strong during the 2008 crisis period. The coefficient
suggestive, the correlations reflect confounding estimate of Ln(NC) * FinCrisis is negative and sig-
effects due to uncontrolled factors. We address nificant at the 1%. A one standard deviation
these with multivariate regressions next. increase in internationalization is associated with
incremental 74 basis points lower cost of equity
capital during the crisis.
REGRESSION RESULTS
Cost of Equity Capital and Corporate Risk Diversification and Corporate

Internationalization Internationalization
We test the hypothesized relationship between We next examine the risk implications of interna-
internationalization and the cost of equity using a tionalization. We provide two sets of tests. First, we
least squares panel regression specification: show firms that operate in countries with less
correlated stock markets have lower cost of equity.
CoEitj ¼ at þ cj þ b1 LnðNCÞitj þ b2 Controlsitj þ itj Second, we show a direct link between internation-
alization and firm risk.
ð2Þ
Diversification tenets hold that operating in less
where CoEitj is the cost of equity capital for firm i in correlated markets would result in greater diversi-
year t and industry j, and LnðNCÞitj is the log- fication benefits.8 We start with a measure at the
transformed number of distinct foreign countries in firm-year level, which captures the degree of stock
which a firm has operations. Controlsitj is a set of market correlations of the countries where an MNC
control variables similar to ones employed in prior has material operations. StkMktCorr is computed as
research (e.g., Chen, Chen, & Wei, 2011; Dhaliwal, the difference between the standard deviation of
Judd, Serfling, & Shaikh, 2016). These controls average returns across countries where a firm
include firm size, book leverage, market-to-book reports subsidiaries and the standard deviation of
ratio, return on assets, analyst forecast dispersion, returns across the same countries but assuming a
long-term growth forecast, past return, foreign pair-wise correlation of one between all countries9 :

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1550
Table 1 Variable definitions
Variable Definition
AFDisp Analyst forecast dispersion, calculated as the ratio of the standard deviation of analysts’ estimates and the consensus
forecast for a firm’s next period earnings
Age An indicator variable equal to 1 if a firm’s age is below the median firm age (0 otherwise). A firm’s age is defined as the
number of years the firm is listed on Compustat
AsianCrisis An indicator variable equal to 1 if an MNC had material operations in a country affected by the Asian crisis for fiscal year
ends between July 1, 1997 and July 1, 1999 (0 otherwise)
BLev Financial leverage, calculated as the sum of book value of long-term debt and short-term debt divided by book value of
assets
BorderState An indicator variable equal to 1 if a firm’s headquarters is in a border state, 0 otherwise
CC Customer sales concentration, measured as a Herfindahl–Hirschman Index
CoE Implied cost of equity capital in excess of the yield on 10-year Treasury bonds constructed following Ohlson and
Juettner-Nauroth (2005) assuming 4% long-term earnings growth
CoE* A firm’s cost of equity capital constructed following Ohlson and Juettner-Nauroth (2005) and assuming 4% long-term
earnings growth
CredRating An indicator variable equal to 1 if a firm does not have a Standard & Poor’s credit rating or if the credit rating is below
investment grade, 0 otherwise
CtrlCorrupt A country-level governance index, which captures the extent to which public power is exercised for private gain as well
as ‘‘capture’’ of the state by elites and private interests
CtryGov A country-level governance index, calculated as the first principal component of control of corruption, rule of law,
government effectiveness, and regulatory quality
CultDist An index of the cultural distance between the U.S. and a given country, measured as the Mahalanobis distance
integrating dimensions of three cultural frameworks
DemoDist An index of the demographic distance between the U.S. and a given country, measured as a Mahalanobis distance
based on life expectancy, birthrate and population age
EqtyMktDev A country’s equity market development, calculated as the first principal component of the market capitalization of
listed domestic companies (% of GDP) and the number of stock exchange listed domestic companies
FinCrisis A indicator variable equal to 1 if a firm’s fiscal year end is between July 1, 2008 and January 1, 2010 (0 otherwise)
ForExVol Foreign exchange rate volatility, calculated as the range of a daily trade-weighted U.S. Dollar exchange rate index over
the 250 trading days prior to a firm’s fiscal-year end
FSales A firm’s percentage of sales from international operations over a given fiscal year
FSub An indicator variable equal to 1 if a firm reports international operations, 0 otherwise
GeoDist The average great circle distance between Washington, D.C. and the capital of a country
GlblSalesHHI An Herfindahl–Hirschman Index of companies’ geographic segment sales using country weights based on Exhibit 21
data
GDPGr A country’s annual GDP growth
GovEffect A country-level governance index, which captures the quality of public and civil services, and the quality of policy
formulation and implementation
IndFSales The fraction of firms with international sales in a firm’s Fama–French 48 industry
IndFSubs The fraction of firms with international subsidiaries in a firm’s Fama–French 48 industry
Language An indicator variable equal to 1 if the U.S. has the same official language (English) as a given country, 0 otherwise
Ln(GDP) A natural log transformation of a country’s annual GDP
Ln(NC) A natural log transformation of NC, calculated as Ln(1 þ NC)
Ln(NListed) A natural log transformation of the total number (in thousands) of domestic companies listed on a country’s stock
exchange
LTG The median analyst forecast of the long-term earnings growth rate of a company
MktCap A country’s market capitalization of listed domestic companies (% of GDP)
MtB The ratio of market equity to book equity
NC The number of distinct foreign, non-U.S. countries (as counted in Exhibit 21 of form 10-K) in which a firm discloses
operations over a given fiscal year
PLC Product line concentration, measured as the Herfindahl–Hirschman Index of sales in all reported industry segments
(product lines)
PRet Cumulative return over the 250 trading days prior to fiscal-year end
RegQuality A country-level governance index, which captures government’s ability to formulate and implement sound policies
and regulations that promote private sector development
Religion An indicator variable equal to 1 if the U.S. has the same primary religion (Christianity) as a given country, 0 otherwise

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1551
Table 1 (Continued)
Variable Definition
RetVol Stock return volatility, calculated as the variance of daily returns over 250 trading days prior to fiscal-year end
RfR The risk-free rate, defined as the yield on 10-year Treasury bonds
ROA Return on assets, calculated as the ratio of income before extraordinary items and beginning of year book value of
assets
SA An indicator variable equal to 1 if a firm’s Hadlock and Pierce (2010) SA financial constraints index is above its median
value in the sample, 0 otherwise
Size Firm size, calculated as the natural log transformation of a firm’s sales at fiscal year end
StkMktCorr A firm-level measure of risk diversification based on country stock market correlations, computed as the difference
between the standard deviation of average returns across countries where a firm reports subsidiaries and the standard
deviation of returns assuming a pair-wise correlation of one between all countries
Trd-to-GDP The ratio of annual trade (imports þ exports) between the U.S. and a given country, and the GDP of that country
CultZone An indicator variable equal to 1 if the U.S. is in the same cultural zone as a given country based on the Ronen and
Shenkar (2013)’s global cultural clusters definition
Panel A Panel B
18
14
16
12
14
10
12
8
10
8
1995 2000 2005 2010 2015 1995 2000 2005 2010 2015
Data Year - Fiscal Data Year - Fiscal
Mean CoE w\ FS Mean CoE w\out FS Med CoE w\ FS Med CoE w\out FS
Figure 1 Cost of equity and corporate internationalization over time. The sample consists of CRSP/Compustat/IBES firms from 1996
to 2012. Panels A and B plot equally-weighted mean and median cost of equity (CoE) splitting the sample by whether firms report
foreign subsidiaries. Foreign subsidiaries are identified through Exhibit 21 in firms’ annual financial reports (10-Ks). CoE is a firm’s cost
of equity capital (in excess of the yield on 10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005)
assuming long-term earnings growth rate of 4%
! vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi captures the return volatility of a portfolio of stock

X uX
m u m X m
StkMktCorrit ¼ r xi rit t xi xj rit rjt market indexes of the countries in which an MNC
i¼1 i¼1 j¼1 has material operations. The second term captures
a ‘‘no-diversification’’ return volatility of the same
ð3Þ
portfolio of stock market indexes if we assume a
where ri and ri are the returns and standard pair-wise correlation of one between all country
deviation of returns of country i during the past stock markets. By construction, StkMktCorr is
36 months, m is the number of countries where a always less than or equal to zero. A more negative
firm reports subsidiaries (including the U.S.) dur- StkMktCorr indicates that the correlations between
ing a given year, and x is the weight of each the stock markets of the countries where an MNC
country, assumed to be proportional to a country’s operates are lower, and thus diversification bene-
GDP.10 Intuitively, the first term in Eq. (3) fits of internationalization are higher. We next test

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1552
Table 2 Descriptive statistics and variable correlations
Panel A: Summary statistics (firm-year level)
Mean SD P25 P50 P75 n
CoE 14.79 8.38 10.24 12.15 16.04 36,823

RfR 4.71 1.32 3.80 4.90 5.80 36,823
CoE 10.08 8.48 5.35 7.54 11.56 36,823
NC 6.69 11.92 0.00 1.00 8.00 36,823
Ln(NC) 1.18 1.26 0.00 0.69 2.20 36,823
FSub 0.59 0.49 0.00 1.00 1.00 36,823
RetVol 12.02 14.36 3.82 7.32 14.48 36,823
StkMktCorr - 0.33 0.45 - 0.62 - 0.10 0.00 35,264
Size - 0.38 1.82 - 1.64 - 0.45 0.81 36,823
BLev 0.21 0.20 0.04 0.18 0.33 36,823
MtB 3.07 3.94 1.42 2.17 3.55 36,823
ROA 0.03 0.15 0.01 0.04 0.09 36,823
AFDisp 0.11 0.21 0.02 0.04 0.08 36,823
LTG 0.16 0.18 0.10 0.15 0.20 36,823
PRet 0.18 0.61 - 0.17 0.09 0.38 36,823
PLC 0.83 0.25 0.62 1.00 1.00 36,823
CC 0.05 0.10 0.00 0.00 0.04 36,823
ForExVol 9.74 3.50 6.84 9.65 12.40 36,823
FSales 0.11 0.18 0.00 0.00 0.18 36,823
GlblSalesHHI 0.86 0.22 0.75 1.00 1.00 36,823
Age 0.49 0.50 0.00 0.00 1.00 36,823
SA 0.46 0.50 0.00 0.00 1.00 36,823
CredRating 0.80 0.40 1.00 1.00 1.00 36,823
Panel B: Correlations
Variables CoE Ln(NC) FSub Ret Vol Stk MktCorr FSales
CoE 1.000
Ln(NC) - 0.07 1.00
(0.00)
FSub - 0.04 0.84 1.00
(0.00) (0.00)
RetVol 0.44 - 0.17 - 0.15 1.00
(0.00) (0.00) (0.00)
StkMktCorr 0.07 - 0.87 - 0.70 0.13 1.00
(0.00) (0.00) (0.00) (0.00)
FSales 0.03 0.44 0.37 - 0.02 - 0.40 1.00
(0.00) (0.00) (0.00) (0.00) (0.00)
Panel C: Summary statistics (country-year level)
CtryGov 0.00 1.94 - 1.51 - 0.41 1.35 2537

CtrlCorrupt 0.08 1.03 - 0.73 - 0.18 0.72 2537
RuleLaw 0.04 0.99 - 0.75 - 0.13 0.82 2537
GovEffect 0.13 0.97 - 0.61 - 0.06 0.83 2537
RegQuality 0.15 0.93 - 0.49 0.02 0.89 2537
EqtyMktDev 0.00 1.14 - 0.71 - 0.37 0.28 1263
MktCap 0.61 0.97 0.16 0.36 0.73 1276
Ln(NListed) 0.26 0.35 0.04 0.14 0.30 1516
AsianCrisis6C 0.00 0.06 0.00 0.00 0.00 2537
AsianCrisis10C 0.01 0.08 0.00 0.00 0.00 2537
CultZone 0.08 0.27 0.00 0.00 0.00 1146
CultDist 6.01 0.84 5.53 6.20 6.53 699

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1553
Table 2 (Continued)
Panel C: Summary statistics (country-year level)
Language 0.28 0.45 0.00 0.00 1.00 2523

Religion 0.66 0.47 0.00 1.00 1.00 2519
GeoDist 8.50 3.66 6.24 7.96 11.19 2523
DemoDist 5.19 4.43 2.34 4.40 6.55 2490
Ln(GDP) 3.41 2.13 1.95 3.15 4.99 2537
GDPGr 0.04 0.06 0.02 0.04 0.06 2537
Trd-to-GDP 0.09 0.11 0.02 0.04 0.12 2537
This table presents descriptive statistics and variable correlations of the major variables in our analysis. The sample consists of CRSP/Compustat/IBES
firms from 1996 to 2012. Panels A and B report descriptive statistics and variable correlations of our cost of equity, internationalization and control
variables. Panel C provides descriptive statistics of country-level variables related to subsidiary locations. The definitions of all variables are reported in
Table 1.
Table 3 Cost of equity capital
(1) (2) (3) (4)

CoE CoE CoE CoE
Ln(NC) - 0.19*** - 0.13** - 0.15***

(3.49) (2.21) (3.05)
FinCrisis 2.16*** 2.97***
(3.63) (5.06)
Ln(NC) 9 FinCrisis - 0.59***
(6.66)
RetVol 0.20***
(9.64)
Size - 0.71*** - 0.78*** - 0.71*** - 0.40***
(9.71) (9.89) (9.69) (5.25)
BLev 4.38*** 4.38*** 4.37*** 4.31***
(8.74) (8.76) (8.75) (7.31)
MtB - 0.06*** - 0.06*** - 0.06*** - 0.08***
(3.44) (3.43) (3.43) (3.41)
ROA - 19.26*** - 19.19*** - 19.27*** - 14.06***
(15.93) (16.01) (15.97) (18.22)
AFDisp 6.39*** 6.38*** 6.33*** 5.14***
(11.58) (11.64) (11.73) (14.98)
LTG - 0.94* - 0.94* - 0.93* - 1.99***
(1.90) (1.91) (1.89) (5.38)
PRet - 2.26*** - 2.26*** - 2.26*** - 2.23***
(6.70) (6.68) (6.65) (9.29)
ForExVol 0.14*** 0.11*** 0.11*** 0.02
(3.63) (2.76) (2.63) (0.26)
PLC 1.13*** 1.16*** 1.13*** 0.75***
(3.95) (4.05) (3.94) (2.94)
CC 1.29** 1.46*** 1.29** 1.16**
(2.30) (2.64) (2.29) (2.37)
Observations 36,823 36,823 36,823 36,823
Adjusted R2 0.37 0.37 0.37 0.44
This table reports coefficients from panel regressions of a firm’s cost of equity capital on corporate internationalization and control variables. The sample
consists of CRSP/Compustat/IBES firms from 1996 to 2012. The dependent variable is CoE, a firm’s cost of equity capital (in excess of the yield on
10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005). Ln(NC) is a natural log transformation of the number of distinct
foreign countries in which a company has operations. FinCrisis is a time dummy variable taking the value of 1 if a firm’s fiscal year end is between July 1st,
2008 and January 1st, 2010 (0 otherwise). RetVol is a firm’s return volatility, defined as the daily return variance over the 250 trading days prior to fiscal-
year end. The definitions of all variables are reported in Table 1. We use two-dimensional clustering of standard errors to account for within-firm and
within-year correlation of the error terms. We also include Fama–French 48 industry fixed effects and time fixed effects. Absolute values of t-statistics are
presented in parentheses.
Asterisks *, **, *** denote significance at the 10, 5, and 1% level, respectively.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1554
Table 4 Risk diversification
Panel A: Cost of equity and country stock market correlations
(1) (2) (3)

CoE CoE CoE
StkMktCorr 0.45*** 0.46*** 0.36**

(3.41) (3.67) (2.44)
FinCrisis 2.14*** 2.54***
(3.99) (4.83)
StkMktCorr 9 FinCrisis 1.69***
(4.64)
Observations 35,264 35,264 35,264
Adjusted R2 0.37 0.37 0.37
Panel B: Firm risk and corporate internationalization
(1) (2) (3) (4)

RetVol RetVol RetVol RetVol
Ln(NC) - 0.36*** - 0.36*** - 0.20

(2.65) (2.71) (1.25)
FinCrisis 12.88*** 14.63***
(7.69) (8.57)
Ln(NC) 9 FinCrisis - 1.48***
(7.68)
StkMktCorr 0.95**
(2.04)
Observations 52,644 52,644 52,644 50,602
Adjusted R2 0.39 0.39 0.39 0.40
This table reports coefficients from panel regressions of a firm’s cost of equity capital and a firm’s risk on corporate internationalization and control
variables. The sample consists of CRSP/Compustat/IBES firms from 1996 to 2012. The dependent variable in Panel A is CoE, a firm’s cost of equity capital
(in excess of the yield on 10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005). The dependent variable in Panel B is
RetVol, the daily return variance estimated over the 250 trading days prior to fiscal-year end. StkMktCorr is a measure of risk diversification based on
country stock market correlations, computed as the difference between the standard deviation of average returns across countries where a firm reports
subsidiaries and the standard deviation of returns assuming a pair-wise correlation of one between all countries. Ln(NC) is a natural log transformation of
the number of distinct foreign countries in which a company has operations. FinCrisis is a time dummy variable taking the value of 1 if a firm’s fiscal year
end is between July 1st, 2008 and January 1st, 2010 (0 otherwise). Control variables (Size, BLev, MtB, ROA, AFDisp, LTG, PRet, and ForExVol) are included,
but their coefficient estimates are omitted for brevity. The definitions of all variables are reported in Table 1. We use two-dimensional clustering of
standard errors to account for within-firm and within-year correlation of the error terms. We also include Fama–French 48 industry fixed effects and time
fixed effects. Absolute values of t-statistics are presented in parentheses.
StkMktCorr’s relation to the cost of equity capital point decrease in a firm’s cost of equity during that
in a multivariate regression framework similar to period. Since not all countries’ economies were
Eq. 2. Table 4, Panel A presents the results. affected by the crisis equally or simultaneously,
Consistent with our diversification conjectures, MNCs’ operations in less financially integrated
Column (1) shows that foreign operations in less countries helped to mitigate crisis-induced shocks
correlated financial markets are associated with that adversely impacted corporate America.
lower cost of equity. The coefficient of StkMktCorr is While the evidence from countries’ financial
positive and significant at the 1% level. A one integration is suggestive, the finding of a negative
standard deviation decrease in StkMktCorr (i.e., relation between internationalization and the cost
lower country correlations) is associated with a 20 of equity capital requires a direct link to company
basis point decrease in the cost of equity. Column risk. Based on traditional asset pricing theory,
(3) further shows the effects of operating in less corporate internationalization should be related to
correlated markets are magnified during the 2008 a firm’s systematic risk that is non-diversifiable
crisis. A one standard deviation decrease in (e.g., Lintner, 1965; Sharpe, 1964). However, inves-
StkMktCorr is associated with an additional 76 basis tors often do not diversify their portfolios due to

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1555
market imperfections and holding biases, which The interaction terms of financial constraints and
results in the pricing of idiosyncratic risk (e.g., Fu, internationalization are all negative and significant
2009). We thus focus on firm total risk, which we at least at the 5% level. For financially constrained
measure by the total stock return volatility of daily firms, a one standard deviation increase in Ln(NC)
returns over the 250 trading days prior to a firm’s results in an incremental 28–42 basis point decrease
fiscal-year end. Our regression tests are similar to in the cost of equity. Consistent with Hypothesis 3,
the specification in Eq. 2. Table 4, Panel B presents the internationalization effects are more pro-
these results.11 nounced for financially constrained firms with
A more diversified geographic base decreases high expected deadweight costs from financial
stock return volatility on average – the coefficient distress.
of Ln(NC) in Column (1) is negative and significant
at the 1% level. Relative to mean RetVol, a one Subsidiary Location Effects
standard deviation increase in Ln(NC) is associated In our hypotheses section, we conjectured that,
with 3.77% lower return volatility. Column (3) while foreign operations can decrease the cost of
indicates that the risk mitigation effects of inter- equity through diversification mechanisms, foreign
nationalization are driven by the 2008 crisis period. market-specific factors could also be important.
Lastly, Column (4) shows that operating in less This section examines the effects of country gover-
correlated financial markets is associated with nance quality, financial market development, for-
lower return volatility. These results are consistent eign financial crises, and cultural proximity on the
with Hypotheses H1a and H1b. However, while the cost of MNC equity. To test our hypotheses, we
effect of internationalization on the cost of equity utilize Exhibit 21 information to identify the
is at least partially attributed to risk diversification, foreign countries in which firms report operations.
a singular risk explanation for this relation is This set of results uses the subset of our data that
unlikely [e.g., Table 3, Column (5)]. We argue only includes MNCs.
internationalization could decrease the cost of Because our data has multiple instances per firm-
equity through channels other than risk as well year (one observation for every country a firm
(e.g., better access to capital as in our Hypothesis 5). operates in) and Exhibit 21 does not explicitly
provide information on subsidiary sizes, we use
Corporate Internationalization and Financial three approaches to weight observations in our
Constraints regressions: (1) we weight each country exposure
We expect that the cost of equity effects of inter- within a firm-year equally, firm-years having equal
nationalization should be particularly pronounced weight among each other; (2) we weight each
for financially constrained firms (Hypothesis 3). To country exposure within a firm-year proportionately
test this hypothesis, we use three proxies of finan- to the number of times the country’s name is
cial constraints: age (Rauh, 2006), the size and age recorded in a company’s annual Exhibit 21 (which
(SA) index (Hadlock & Pierce, 2010), and Standard we use as a proxy for the number of subsidiaries a
& Poor’s credit rating. We define Age as an indicator firm has in a country), then weighting firm-years
variable equal to 1 if a firm’s age is below the equally; and (3) we weight each country exposure
median firm age in our sample, 0 otherwise. A within a firm-year proportionately to the GDP of the
firm’s age is defined as the number of years since country in a given year, then weighting firm-years
listing on Compustat. SA is an indicator equal to 1 equally. We run weighted least squares (WLS) using
if a firm’s Hadlock and Pierce (2010) financial the above weighting approaches, with Fama–French
constraints index is above its median value in the 48 industry fixed effects and year fixed effects, and
sample, 0 otherwise. CredRating is an indicator clustered standard errors at the firm level. We
equal to 1 if a firm does not have a Standard & control for previously used firm-level determinants
Poor’s credit rating or if its credit rating is below of the cost of equity and include country-level
investment grade, 0 otherwise. We then interact controls (e.g., host country size, economic growth,
the financial constraints variables with Ln(NC) and and trade with the U.S.). For brevity, we present
test them in a regression specification with two- results only using weighting scheme 2 (with results
dimensional clustering of errors that includes our from schemes 1 and 3 presented in our Online
previous control variables, industry and time fixed Appendix). Overall, our results are robust across
effects. Table 5 presents our results. different weighting schemes.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1556
Table 5 Financial constraints
(1) (2) (3) (4) (5) (6)

CoE CoE CoE CoE CoE CoE
Ln(NC) - 0.19*** - 0.10* - 0.22*** - 0.11 - 0.19*** 0.05

(3.48) (1.89) (3.42) (1.56) (3.51) (0.81)
Age 0.42*** 0.67***
(3.80) (3.90)
Age * Ln(NC) - 0.22**
(2.56)
SA 0.70*** 1.05***
(3.87) (3.93)
SA * Ln(NC) - 0.32**
(2.40)
CredRating 0.60*** 1.14***
(3.99) (5.55)
CredRating * Ln(NC) - 0.33***
(3.83)
Observations 36,823 36,823 36,823 36,823 36,823 36,823
Adjusted R2 0.37 0.37 0.37 0.37 0.37 0.37
This table reports coefficients from panel regressions of a firm’s cost of equity capital on corporate internationalization, financial constraints proxies and
their interactions. The sample consists of CRSP/Compustat/IBES firms from 1996 to 2012. The dependent variable is CoE, a firm’s cost of equity capital
(in excess of the yield on 10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005). Ln(NC) is a natural log transformation of
the number of distinct foreign countries in which a company has operations. Age, SA and CredRating are proxies for firms’ financial constraints. Age is an
indicator variable, which takes a value of 1 if a firm’s age is below the median firm age. A firm’s age is defined as the number of years the firm is listed
with a non-missing stock price on Compustat. SA is an indicator variable, which takes a value of 1 if a firm’s Hadlock and Pierce (2010) SA financial
constraints index is above its median value. CredRating is an indicator variable taking the value of 1, if a firm does not have a Standard & Poor’s credit
rating or if a firm’s Standard & Poor’s credit rating is below investment grade. All three variables are increasing in a firm’s financial constraints. Control
variables (Size, BLev, MtB, ROA, AFDisp, LTG, PRet, ForExVol, PLC, and CC) are included, but their coefficient estimates are omitted for brevity. The
definitions of all variables are reported in Table 1. We use two-dimensional clustering of standard errors to account for within-firm and within-year
correlation of the error terms. We also include Fama–French 48 industry fixed effects and time fixed effects. Absolute values of t-statistics are presented
in parentheses.
Country Governance the previously described WLS regression approach

To test Hypothesis 4, we use country-level indica- for estimations. Table 6, Panel A presents our
tors from the World Bank developed by Kaufmann, results.
Kraay, and Mastruzzi (2009). These indicators are The coefficient of CtryGov is negative and signif-
along several dimensions. CtrlCorrupt captures the icant at the 5% level. A one standard deviation
exercise of public power for private gain and the increase in country governance quality suggests a 8
‘‘capture’’ of the state by private interests. RuleLaw basis point decrease in the cost of equity. Relative
summarizes the contract enforcement and property to mean CoE, this amounts to 0.8% decrease. The
rights quality, and the likelihood of crime and coefficient signs of CtrlCorrupt, RuleLaw, GovEffect
violence. GovEffect captures the quality of public and RegQuality are also directionally consistent and
and civil services, and the quality of policy formu- significant at least at the 10% level. Consistent with
lation and implementation. RegQuality summarizes Hypothesis 4, better governance quality in host
government ability to formulate and implement countries lowers the cost of equity for MNCs.
sound regulations supporting private sector devel-
opment. We also synthesize a composite gover- Equity Market Development
nance measure, CtryGov, by extracting the first To test Hypothesis 5, we measure countries’ equity
principal component of the four indices. All mea- market development in two ways using data from
sures are increasing in the quality of governance the World Bank. First, we employ the market value
(i.e., higher index values denote lower corruption, of equity for stock-exchange-listed host-country
stronger property rights, higher government effec- ‘‘domestic’’ companies as a proportion of the host
tiveness and better regulatory quality). Our final country’s gross domestic product (MktCap). Second,
sample consists of 235,930 firm-year-subsidiary we use the log-transformed total number of listed
country observations with available data. We use firms on the country’s stock exchanges

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1557
Table 6 Subsidiary location effects. This table reports coefficients from weighted least squares regressions of a firm’s cost of equity
capital on corporate internationalization, firm subsidiary country characteristics, and control variables. The sample includes a panel of
firm-year-subsidiary country observations with requisite data during the period [1996–2012] of CRSP/Compustat/IBES multinational
companies, which report foreign operations in their 10-K reports. The dependent variable is CoE, a firm’s cost of equity capital (in
excess of the yield on 10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005). Panel A presents
regressions of a firm’s cost of equity on subsidiary country governance (measured by CtryGov, CtrlCorrupt, RuleLaw, GovEffect, and
RegQuality). Panel B presents regressions of a firm’s cost of equity on subsidiary country equity market development (measured by
EqtyMktDev, MktCap, and Ln(NListed). Panel C presents regressions of a firm’s cost of equity on subsidiary country Asian financial crisis
effects (measured by AsianCrisis6C and AsianCrisis10C). Finally, Panel D presents regressions of a firm’s cost of equity on subsidiary
country cultural distance from the United States (measured by CultZone, CultDist, Language, Religion, GeoDist, and DemoDist). As a
regression weighting scheme, we weight each country exposure within a firm-year proportionately to the number of times a country’s
name is recorded in a given company’s annual Exhibit 21, then weighting firm-years equally. Firm-level control variables (Size, BLev,
MtB, ROA, AFDisp, LTG, PRet, ForExVol, PLC, and CC) are included, but their coefficient estimates are omitted for brevity. The definitions
of all variables are reported in Table 1. We include Fama-French 48 industry fixed effects and time fixed effects, and cluster standard
errors at the firm level. Absolute values of t-statistics are presented in parentheses
Panel A: Governance
(1) (2) (3) (4) (5)

CoE CoE CoE CoE CoE
CtryGov - 0.04**
(1.99)
CtrlCorrupt - 0.07**
(2.06)
RuleLaw - 0.08*
(1.89)
GovEffect - 0.08*
(1.76)
RegQuality - 0.10**
(2.11)
Ln(GDP) - 0.03 - 0.03 - 0.03 - 0.03 - 0.03
(1.21) (1.25) (1.20) (1.21) (1.31)
GDPGr 0.53 0.53 0.55 0.70 0.45
(0.59) (0.60) (0.61) (0.79) (0.49)
Trd-to-GDP 0.38 0.35 0.31 0.35 0.36
(1.43) (1.46) (1.29) (1.45) (1.49)
Ln(NC) - 0.12* - 0.12* - 0.12* - 0.12* - 0.12*
(1.72) (1.73) (1.71) (1.70) (1.72)
Observations 235,930 235,930 235,930 235,930 235,930
Adjusted R2 0.41 0.41 0.41 0.41 0.41
Panel B: Equity market development
(1) (2) (3)

CoE CoE CoE
EqtyMktDev - 0.09***
(2.79)
MktCap - 0.05**
(2.09)
Ln(NListed) - 0.21***
(2.59)
Ln(GDP) - 0.05* - 0.07*** - 0.02
(1.71) (2.73) (0.80)
GDPGr 2.15** 1.93* 2.37**
(2.03) (1.82) (2.37)
Trd-to-GDP 0.32 0.23 0.38
(1.22) (0.89) (1.50)
Ln(NC) - 0.15** - 0.14* - 0.14*
(1.99) (1.87) (1.92)

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1558
Table 6 continued
Panel B: Equity market development
(1) (2) (3)

CoE CoE CoE
Observations 199435 200707 217668

Adjusted R2 0.41 0.41 0.41
Panel C: Asian financial crisis
(1) (2)
CoE CoE
AsianCrisis6C 0.94***
(3.02)
AsianCrisis10C 0.82***
(3.12)
Ln(GDP) - 0.03 - 0.03
(1.60) (1.58)
GDPGr 2.14** 2.10**
(2.43) (2.41)
Trd-to-GDP 0.25 0.22
(1.03) (0.92)
Ln(NC) - 0.12* - 0.12*
(1.71) (1.73)
Observations 235930 235930
Adjusted R2 0.41 0.41
Panel D: Cultural distance
(1) (2) (3) (4) (5) (6)

CoE CoE CoE CoE CoE CoE
CultZone - 0.37***
(3.51)
CultDist 0.13***
(2.85)
Language - 0.21***
(2.82)
Religion - 0.31***
(3.52)
GeoDist 0.02*
(1.86)
DemoDist 0.03**
(2.02)
Ln(GDP) 0.00 - 0.02 - 0.04* - 0.04* - 0.03 - 0.03
(0.09) (0.47) (1.78) (1.82) (1.58) (1.21)
GDPGr 1.96 1.54 1.19 - 0.05 0.58 1.17
(1.41) (1.18) (1.38) (0.05) (0.64) (1.23)
Trd-to-GDP 0.42 0.23 0.54** 0.22 0.47* 0.30
(1.27) (0.68) (2.09) (0.91) (1.86) (1.17)
Ln(NC) - 0.21** - 0.17* - 0.13* - 0.13* - 0.13* - 0.13*
(2.29) (1.89) (1.88) (1.86) (1.82) (1.76)
Observations 211,557 183,216 235,712 235,760 235,712 231,848
Adjusted R2 0.42 0.41 0.41 0.41 0.41 0.41

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1559
(Ln(NListed)). In addition, we also synthesize the internationalization helps MNCs weather domestic
information in MktCap and Ln(NListed) into a single storms, financial and economic turmoil in host
measure by extracting their first principal compo- countries can moderate internationalization
nent (EqtyMktDev). Depending on the specification, benefits.
our estimation sample is between 199,435 and
235,712 firm-year-subsidiary country observations. Cultural Proximity
We use the previously described WLS regression To test Hypothesis 7, we proxy for cultural distance
approach for estimations. Table 6, Panel B presents in a number of ways. Our first measure is based on a
the results. clustering approach that groups countries into
The coefficient of EqtyMktDev is negative and supra-national cultural groups.13 We use Ronen
significant at the 1% level. A one standard devia- and Shenkar (2013)’s empirically derived ‘‘cultural
tion increase in EqtyMktDev suggests a 10 basis mapping,’’ where the authors classify countries into
point decrease in the cost of MNC equity. Relative 11 global clusters: ‘‘Arab,’’ ‘‘Near East,’’ ‘‘Latin
to mean CoE, this amounts to 1.0% decrease. The America,’’ ‘‘East Europe,’’ ‘‘Latin Europe,’’ ‘‘Nordic,’’
coefficients of MktCap and Ln(NListed) are also ‘‘Germanic,’’ ‘‘African,’’ ‘‘Anglo,’’ ‘‘Confucian,’’ and
directionally similar and significant. Consistent ‘‘Far East.’’ We define countries that belong to the
with Hypothesis 5, operations in countries with same cluster as the U.S., the Anglo cluster, to be
more developed financial markets, and the conse- culturally similar to the U.S. This information is
quent diversification of capital sources and captured by a binary variable, CultZone, equal to 1 if
increased access to capital, lowers MNCs’ cost of a host country belongs to the Anglo cluster, and 0
equity. otherwise.
For our second measure, we follow Beugelsdijk,
Foreign Crises: Evidence from the Asian Financial Kostova, and Roth (2017) and compute a Maha-
Crisis lanobis-based measure for cultural distance, Cult-
A countervailing effect to the diversification bene- Dist, integrating multiple cultural frameworks: the
fits of foreign operations is that MNCs operating in six-dimensional Hofstede framework with
foreign markets that experience financial and eco- Schwartz’s seven-dimensional values framework
nomic crises could be susceptible to these host and GLOBE’s nine-dimensional system of national
market stresses. We test Hypothesis 6 by focusing cultural values.14 Next, moving away from com-
on the 1997 Asian financial crisis – the second posite cultural distance measures, we consider three
major financial crisis over our sample period. The major antecedents of cultural differences: religion,
crisis erupted in July 1997 in Thailand, spreading language, and geography (Ronen & Shenkar, 2013;
rapidly to other countries in the region. Stulz & Williamson, 2003). We define the following
To test whether U.S. MNCs with material opera- variables: Religion is an indicator variable equal to 1
tions in countries affected by the Asian financial if the U.S. has the same primary religion (Chris-
crisis had their equity costs rise, we define an tianity) as a given country, 0 otherwise. Language is
indicator variable equal to 1 if an MNC had an indicator variable equal to 1 if the U.S. has the
material operations in an affected (by the crisis) same official language (English) as a given country,
country for fiscal years ending between July 1, 1997 0 otherwise. GeoDist is the average great circle
and July 1, 1999, 0 otherwise. We use two varia- distance from Washington, D.C. to a country’s
tions of the variable. AsianCrisis6C equals 1 for a capital city. As a final check, because cultural
firm-year-subsidiary country observation if the distance might be also related to the demographic
given country is Thailand, Indonesia, South Korea, profiles of countries, we use the demographic
Hong Kong, Malaysia, or the Philippines during the distance between the U.S. and other countries.
defined period, 0 otherwise. Second, AsianCrisis10C We borrow a measure from Berry, Guillen, and
uses a broader ten-country definition additionally Zhou (2010), calculated as a Mahalanobis-based
including India, Taiwan, Singapore and Brunei in distance over four demographic indicators (life
the list of affected countries. Table 6, Panel C expectancy, birth rate, population under 14, pop-
presents results. ulation under 65), which we call DemoDist.
Consistent with Hypothesis 6, both specifications Depending on the measure, the number of firm-
suggest that MNCs with material exposures to the year-subsidiary country observations in our estima-
countries plagued by the Asian financial crisis had tions varies between 183,216 and 235,760. Table 6,
higher costs of equity during that period.12 While Panel D presents WLS regression results.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1560
Table 7 Foreign sales and alternative internationalization measures
(1) (2) (3)

CoE CoE CoE
FSales 0.67** 1.10***

(2.26) (3.48)
Ln(NC) - 0.24***
(4.48)
GlblSalesHHI 0.60**
(2.13)
Observations 36,823 36,823 36,823
Adjusted R2 0.37 0.37 0.37
This table reports coefficients from panel regressions of a firm’s cost of equity capital on corporate internationalization and control variables. The sample
consists of CRSP/Compustat/IBES firms from 1996 to 2012. The dependent variable is CoE, a firm’s cost of equity capital (in excess of the yield on
10-year Treasury bonds) constructed following Ohlson and Juettner-Nauroth (2005). FSales is a firm’s percentage of foreign (non-domestic) sales
relative to the firm’s total sales. Ln(NC) is a natural log transformation of the number of distinct foreign countries in which a company has operations.
GlblSalesHHI is a corporate internalization measure combining Compustat Segment data and Exhibit 21 data, calculated as a Herfindahl–Hirschman
Index of companies’ geographic segment sales from both domestic and international operations. Compustat Segment clearly reports domestic sales.
We allocate Compustat Segment foreign sales to countries in which firms report operations in their Exhibit 21 proportionately to the number of times a
country’s name is recorded in a given company’s annual Exhibit 21. Control variables (Size, BLev, MtB, ROA, AFDisp, LTG, PRet, and ForExVol) are
included, but their coefficient estimates are omitted for brevity. The definitions of all variables are reported in Table 1. We use two-dimensional
clustering of standard errors to account for within-firm and within-year correlation of the error terms. We also include Fama–French 48 industry fixed
effects and time fixed effects. Absolute values of t-statistics are presented in parentheses.
Asterisks *, **, *** denote significance at the 5, and 1% level, respectively.
Specifications (1) and (2) suggest that firms with demand is the only form of exposure that an
subsidiaries in countries from the same cultural MNC has to foreign markets. However, MNCs often
zone as and culturally closer to the U.S. have lower have foreign operations that do not generate sales
cost of equity. Operations in countries outside of (e.g., production or resource extraction facilities)
the Anglo cluster increase the cost of equity by 37 that are relevant for financing costs (e.g., through
basis points. Alternatively, a one standard deviation access to capital markets). The internationalization
increase in cultural distance from the U.S. is measure used in our study, Ln(NC), is only moder-
associated with an 11 basis point increase in the ately correlated (44%) with the proportion of
cost of equity. Specifications (3)–(6) further support foreign sales constructed from Compustat Segment,
Hypothesis 7 that cultural factors play significant FSales, which suggests a substantial disparity
roles for the cost of MNC equity. between the two measures.
In this section, we check the cost of capital
implications of FSales as a measure of corporate
ADDITIONAL ANALYSES AND CHECKS internationalization. Table 7 suggests that firms
with more foreign sales have higher costs of equity
Foreign Sales and Measuring Internationalization
– FSales has a positive and significant (at the 5%
Prior research (e.g., Denis et al., 2002; Schmid &
level) coefficient. Such evidence is consistent with
Walter, 2012) has measured global diversification
Amihud et al. (2015), who argue that foreign sales
with the fraction of a firm’s foreign to total sales.
concentration could expose firms and investors to
While somewhat informative, this measure con-
risks that are difficult to hedge as well as Denis et al.
tains limited information on the degree of foreign
(2002) who argue that increases in international
diversification. Total foreign sales do not provide
sales reduce a firm’s excess value. The results
information on the number of foreign locations
reinforce our earlier observations that the propor-
where an MNC operates, which is crucial for risk
tion of foreign sales does not accurately capture the
diversification and cost of equity. In the extreme, a
number of foreign locations where MNCs operate,
company with significant sales to a single interna-
and thus does not properly measure diversification.
tional market will be mistakenly classified as diver-
Column (2) shows that the coefficient of Ln(NC)
sified based on a proportion of foreign sales metric,
remains directionally and statistically robust to
while it is in fact concentrated in only one foreign
including FSales as an additional explanatory vari-
market (with questionable implications depending
able in our regressions – the cost of equity decreases
on market-specific risk factors). Critically, using
with the diversification of foreign operations.
foreign sales implicitly assumes that product

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1561
Table 8 Endogeneity: neighbor matching and Heckman selection estimations
Panel A: Matching estimations
NN-1M NN-3M NN-5M
SATE - 0.33*** - 0.44*** - 0.52***

(3.44) (4.90) (5.88)
Observations 36,823 36,823 36,823
Panel B: Heckman selection model
(1) (2) (3)
CoE
Ln(NC) - 0.15*** - 0.19*** - 0.18***
(2.76) (3.45) (3.15)
Lambda 2.05*** 3.76*** 2.19***
(4.12) (8.09) (4.38)
FSub
BorderState 0.08***
(5.51)
IndFSubs 1.95***
(13.85)
IndFSales 1.01***
(8.60)
Observations 36,823 36,823 36,823
This table presents estimates of the average treatment effects from nearest neighbor matching and Heckman (1979) selection model estimations. The
sample consists of CRSP/Compustat/IBES firms from 1996 to 2012. In Panel A, we compare cost of equity outcomes between firms that have subsidiaries
in foreign countries and those that do not. We use exact matching on Fama–French 48 industry classification and fiscal year, and nearest neighbor
matching across previously known cost of equity determinants. These include Size, BLev, MtB, ROA, AFDisp, LTG, PRet, ForExVol, PLC and CC. Columns
(1), (2) and (3) present results based on the closest 1, 3, and 5 matches, respectively. SATE denotes the sample average treatment effect. Panel B reports
coefficient estimates from a Heckman (1979) selection model of a firm’s cost of equity capital on corporate internationalization and control variables.
We use three different instruments to calculate inverse Mills ratios – an indicator variable equal to 1 if a firm is headquartered in a border state
(BorderState), the fraction of firms with foreign subsidiaries in a firm’s industry (IndFSubs), and the fraction of firms with foreign sales in a firm’s industry
(IndFSales). Control variables (Size, BLev, MtB, ROA, AFDisp, LTG, PRet, ForExVol, PLC, and CC) are included, but their coefficient estimates are omitted for
brevity. All specifications include Fama–French 48 industry fixed effects and time fixed effects. The definitions of all variables are reported in Table 1.
Absolute values of t-statistics are presented in parentheses. Asterisks *, **, *** denote significance at the 10, 5, and 1% level, respectively.
While the above discussion and results suggest segment sales based on sales from both domestic
caution with using foreign sales, such sales should and international operations. Table 7, Column (3)
nonetheless contain useful information. We thus confirms that corporate internationalization is
combine Compustat Segment and Exhibit 21 data negatively associated with a firm’s cost of equity
to define an alternative corporate internationaliza- capital. The coefficient of GlblSalesHHI is positive
tion measure and check the sensitivity of our and statistically significant at the 5% level (Gl-
results. Compustat Segment contains information blSalesHHI is decreasing in corporate international-
on firms’ sales from geographic segments, both ization), showing that more globally diversified
domestic and non-domestic segments. While sales reduce the cost of equity. In unreported
domestic sales are clearly reported, foreign sales results, we also confirm that the results are robust
cannot be easily attributed to specific countries in a alternative calculations of HHI using GDP and
uniform manner that is consistent and comparable equal weights in allocating foreign sales to
across firms. We thus use Exhibit 21 data to allocate countries.
sales to specific foreign countries. We allocate
foreign sales proportionately to the number of Endogeneity and Selection Bias
times a country’s name is recorded in a company’s A potential endogeneity concern with our findings
annual Exhibit 21. Then, we calculate a Herfindahl– is that firms without foreign subsidiaries are sub-
Hirschman Index (HHI) of companies’ geographic stantially different from firms with foreign

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1562
subsidiaries along unobservable dimensions, which results presented in Table 8 assuage endogeneity
could be a driving factor behind our documented and selection concerns.
cost of equity effects. One way to address this issue
is to construct (nearest neighbor) matched samples Other Robustness Tests
based on observable factors that predict a firm’s We also ran a number of additional robustness tests
cost of equity (Abadie, Drukker, Herr, & Imbens, accounting for alternative ways of constructing cost
2004). Specifically, we use exact matching on of equity variables, alternative estimation method-
industry classification and fiscal year, and nearest ologies of choice, sub-period and sub-sample anal-
neighbor matching across other cost of equity ysis, and alternative regression weighting schemes.
determinants [per Eq. (2)] to pair domestic and The presentation and discussion of these additional
multinational firms. Table 8, Panel A presents the robustness checks are shown in our Online
sample average treatment effect (SATE) across the Appendix.
closest 1, 3 and 5 matches. Consistently, the
nearest neighbor matching results indicate that
MNCs have lower cost of equity capital than CONCLUSION
domestic firms.15 Our study documents the effects of foreign opera-
While the matching estimations mitigate endo- tions on the required rates of return of equity
geneity concerns, the endogenous choice of firms investors and the channels through which the
to grow internationally might additionally be a linkage occurs. This is important given firms’
source of selection bias (Dastidar, 2009). We con- increased international exposure, the pertinence
sequently estimate Heckman (1979) selection mod- of corporate equity capital and its costs, and the
els, where the selection equation of the model indirect and mixed evidence on the cost of equity
estimates a firm’s propensity to have foreign effects of corporate internationalization. We use
operations. detailed data from 1996 to 2012 on the foreign
In general, to have a well-identified Heckman countries where U.S. MNCs have material opera-
(1979) model, we need an instrumental variable in tions, which we collect from Exhibit 21 in firms’
the first-stage equation that is correlated with a 10-K filings. Such data allow us to overcome
firm’s decision to be multinational, but has no significant limitations in the most widely used
effect on its cost of equity directly. Following measures of international diversification from prior
Berger, Ghoul, Guedhami, and Roman (2017), we literature and provide us with granular location
use an instrument, BorderState, which is a variable information to examine previously untested
equal to 1 if a company is headquartered in a hypotheses. Our evidence supports the idea that
border state (i.e., a state having a border with MNCs perform valuable diversification functions to
Canada or Mexico), and 0 otherwise. BorderState investors in a world with segmented and imperfect
should be positively correlated with international- financial markets, and this value is reflected in
ization as companies in border states are more companies’ lower costs of capital. Importantly, not
likely to have foreign operations, but should not all foreign markets are ‘‘equal’’ – market-specific
have a direct effect on the cost of equity other than characteristics are important venues that amplify or
through internationalization. In addition, in the reduce the cost of capital benefits of foreign
spirit of Campa and Kedia (2002), we use the operations.
fraction of firms with foreign subsidiaries and the Our study implies that managers, and particu-
fraction of firms with foreign sales in a firm’s larly managers of firms with higher expected dead-
industry as instruments. The idea is that benefits of weight costs of distress (e.g., financially constrained
internationalization are industry-specific and the firms), can lower equity financing costs by diversi-
proportion of multinational firms in the same fying globally. Moreover, channeling such diversi-
industry is a proxy for such benefits. To control fication into culturally similar countries with good
for industry-specific cost of equity effects, we governance and developed financial markets is
include industry fixed effects in all estimations. likely to further decrease equity costs. Our findings
Table 8, Panel B reports the results. The second also have potential implications for financial
stage estimates confirm a negative and statistically reporting policy setters such as the Financial
significant relationship between internationaliza- Accounting Standards Board and the Securities
tion and the cost of equity capital. Overall, the and Exchange Commission. In line with Hope,
Kang, Thomas, and Vasvari (2009), we show that

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1563
the extent of international exposure and specific more than it does today in order to balance local
foreign market attributes are important factors in economic storms.’’ See Financial Times: ‘‘Car makers
the cost of equity capital and bear valuable infor- flock to new southeast Asian growth frontier’’
mation to investors. Further enhancing reporting (Bland, October 6, 2013).
5
standards regarding foreign operating segments Several other studies have also used Exhibit 21
would improve investors’ information environ- data (e.g., Black, Dikolli & Dyreng, 2014; Dyreng &
ment and make capital markets more efficient. Lindsey, 2009; Dyreng, Lindsey & Thornock, 2013;
Lin et al., 2018). As an additional data integrity
check, we also cross-validate our data. For example,
using Compustat Segment, we find that the firms
ACKNOWLEDGEMENTS
we identify as having operations in more countries
The authors thank Azamat Abdymomunov, Ray Bras-
are larger in terms of foreign sales. For additional
tow, Filippo Curti, Jeff Gerlach, Nada Mora, Leming
information on the text search algorithm and an
Lin, Leandro Sanz, Jennifer Tucker, Pinar Uysal and
illustrative example of Exhibit 21, see our Online
Mihail Velikov for valuable comments and suggestions.
Appendix.
Jeffrey Cheng provided able research assistance. All 6
Our results are robust across alternative implied
remaining errors are those of the authors alone. The
cost of equity methodologies. See our Online
views expressed in this paper do not necessarily reflect
Appendix.
the position of the Federal Reserve Bank of Richmond 7
Before taking the natural log transformation, we
or the Federal Reserve System.
first add 1 to the number of foreign countries to
address zero value problems.
8
For example, consider the case where multina-
NOTES tional company MNC1 operates in Country1 and
1
Some studies have examined the effects of Country2, whose stock markets have q12 correlation.
international operations on realized equity returns. In contrast, MNC2 operates in Country1 and Coun-
However, ex-post realized returns are poor proxies try3, whose stock markets have q13 correlation, and
for expected returns and thus using them to infer q12 \q13 . All else equal, MNC1 should be more
risk premiums is difficult (e.g., Elton, 1999; Hann diversified than MNC2 due to lower stock market
et al., 2013). correlation of the countries where it operates.
9
2
An important consideration in interpreting our Similar diversification measurement approaches
empirical results is the issue of endogeneity and have been used in Duchin (2010), among others.
10
self-selection. Underlying firm differences, not the We gather country stock return index data from
location of assets, could, thus, explain differences Compustat Global.
11
in the cost of equity. Moreover, concerns also exist This set of results does not require the calcula-
about the endogenous managerial decision of for- tion of cost of equity measures. As a result, the
eign market presence. We carefully address the availability of data increases the number of obser-
endogeneity and self-selection issue and examine vations from 36,823 to 52,644.
12
the robustness of our results in later sections of our Both AsianCrisis6C and AsianCrisis10C have the
empirical analysis. hypothesized signs across all three regression
3
Anecdotally, we find numerous firm-level exam- weighting schemes. While AsianCrisis6C is statisti-
ples consistent with coinsurance effects related to cally significant at conventional levels in all cases,
geographic diversification. For instance, General AsianCrisis10C is significant in two out of the three
Motors, Toyota, Volkswagen, and Renault-Nissan weighting schemes. See our Online Appendix for
use growth in emerging market regions, such as details.
13
China and South America, to subsidize losses in Recent research suggests that cultural differ-
Europe. See Financial Times: ‘‘Ford predicts European ences are more region- than country-specific, as
sales to pull ahead slowly’’ (Foy, October 24, 2013). countries cluster at the supra-national level in well-
4
Consistent with this hypothesis, Xavier Mos- known cultural zones (e.g., Beugelsdijk et al., 2017).
quet, a managing director at BCG, claims that Moreover, a clustering approach to measuring
‘‘geographic diversification has never mattered cultural differences at the country level also

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1564
summarizes cultural commonalities and distinc- á-vis high degree of internationalization. Second,
tiveness well, such that ‘‘information loss is mini- we estimate specifications that use firms’ geo-
mal’’ (Ronen & Shenkar, 2013). graphic diversification events, where we compare
14
We employ various cultural distance constructs firms that become multinational or domestically
as a robustness check on concerns regarding mea- focused to similar firms that do not. In both cases,
suring cultural distance (e.g., Brouthers, Marshall & internationalization is consistently associated with
Keig, 2016; Shenkar, 2001). lower cost of equity. See our Online Appendix for
15
We estimate two additional sets of matching details.
estimations. First, using only firms with foreign
operations, we compare MNCs with low degree vis-
REFERENCES
Abadie, A., Drukker, D., Herr, J. L., & Imbens, G. W. 2004. Dastidar, P. 2009. International corporate diversification and
Implementing matching estimators for average treatment performance: Does firm self-selection matter? Journal of
effects in Stata. The Stata Journal, 4(3): 290–311. International Business Studies, 40(1): 71–85.
Agmon, T., & Lessard, D. R. 1977. Investor recognition of Denis, D. J., Denis, D. K., & Yost, K. 2002. Global diversification,
corporate international diversification. The Journal of Finance, industrial diversification, and firm value. The Journal of Finance,
32(4): 1049–1055. 57(5): 1951–1979.
Ahern, K. R., Daminelli, D., & Fracassi, C. 2015. Lost in Desai, M. A., Foley, C. F., & Hines, J. R. 2008. Capital structure
translation? The effect of cultural values on mergers around with risky foreign investment. Journal of Financial Economics,
the world. Journal of Financial Economics, 117(1): 165–189. 88(3): 534–553.
Amihud, Y., Bartov, E., & Wang, B. 2015. The pricing of corporate Dhaliwal, D., Judd, J. S., Serfling, M., & Shaikh, S. 2016.
foreign sales risk. Working Paper. Customer concentration risk and the cost of equity capital.
Bartov, E., Bodnar, G. M., & Kaul, A. 1996. Exchange rate Journal of Accounting and Economics, 61(1): 23–48.
variability and the riskiness of U.S. multinational firms: Duchin, R. 2010. Cash holdings and corporate diversification.
Evidence from the breakdown of the Bretton Woods system. The Journal of Finance, 65(3): 955–992.
Journal of Financial Economics, 42(1): 105–132. Dyreng, S. D., & Lindsey, B. P. 2009. Using financial accounting
Bartram, S. M., Brown, G., & Stulz, R. M. 2012. Why are U.S. data to examine the effect of foreign operations located in tax
stocks more volatile? The Journal of Finance, 67(4): 1329–1370. havens and other countries on U.S. multinational firms’ tax
Bekaert, G., & Harvey, C. R. 2000. Foreign speculators and rates. Journal of Accounting Research, 47(5): 1283–1316.
emerging equity markets. The Journal of Finance, 55(2): Dyreng, S. D., Lindsey, B. P., & Thornock, J. R. 2013. Exploring
565–613. the role delaware plays as a domestic tax haven. Journal of
Berger, A. N., Ghoul, S. E., Guedhami, O., & Roman, R. A. 2017. Financial Economics, 108(3): 751–772.
Internationalization and bank risk. Management Science, 63(7): Elton, E. J. 1999. Presidential address: Expected return, realized
2283–2301. return, and asset pricing tests. The Journal of Finance, 54(4):
Berry, H., Guillén, M. F., & Zhou, N. 2010. An institutional 1199–1220.
approach to cross-national distance. Journal of International Errunza, V., Hogan, K., & Hung, M. 1999. Can the gains from
Business Studies, 41(9): 1460–1480. international diversification be achieved without trading
Beugelsdijk, S., Kostova, T., & Roth, K. 2017. An overview of abroad? The Journal of Finance, 54(6): 2075–2107.
Hofstede-inspired country-level culture research in interna- Errunza, V. R., & Miller, D. P. 2000. Market segmentation and
tional business since 2006. Journal of International Business the cost of capital in international equity markets. The Journal
Studies, 48(1): 30–47. of Financial and Quantitative Analysis, 35(4): 577–600.
Black, D. E., Dikolli, S. S., & Dyreng, S. D. 2014. CEO pay-for- Errunza, V. R., & Senbet, L. W. 1981. The effects of international
complexity and the risk of managerial diversion from multi- operations on the market value of the firm: Theory and
national diversification. Contemporary Accounting Research, evidence. The Journal of Finance, 36(2): 401–417.
31(1): 103–135. Errunza, V. R., & Senbet, L. W. 1984. International corporate
Brewer, H. L. 1981. Investor benefits from corporate interna- diversification, market valuation, and size-adjusted evidence.
tional diversification. The Journal of Financial and Quantitative The Journal of Finance, 39(3): 727–743.
Analysis, 16(1): 113–126. Fatemi, A. M. 1984. Shareholder benefits from corporate
Brouthers, L. E., Marshall, V. B., & Keig, D. L. 2016. Solving the international diversification. The Journal of Finance, 39(5):
single-country sample problem in cultural distance studies. 1325–1344.
Journal of International Business Studies, 47(4): 471–479. Fauver, L., Houston, J., & Naranjo, A. 2003. Capital market
Burgman, T. A. 1996. An empirical examination of multinational development, international integration, legal systems, and the
corporate capital structure. Journal of International Business value of corporate diversification: A cross-country analysis. The
Studies, 27(3): 553–570. Journal of Financial and Quantitative Analysis, 38(1): 135–157.
Campa, J. M., & Kedia, S. 2002. Explaining the diversification Fu, F. 2009. Idiosyncratic risk and the cross-section of expected
discount. The Journal of Finance, 57(4): 1731–1762. stock returns. Journal of Financial Economics, 91(1): 24–37.
Caves, R. E. 1971. International corporations: The industrial Gande, A., Schenzler, C., & Senbet, L. W. 2009. Valuation
economics of foreign investment. Economica, 38(149): 1–27. effects of global diversification. Journal of International Business
Chen, K. C. W., Chen, Z., & Wei, K. C. J. 2011. Agency costs of Studies, 40(9): 1515–1532.
free cash flow and the effect of shareholder rights on the Giannetti, M., & Yafeh, Y. 2012. Do cultural differences between
implied cost of equity capital. Journal of Financial and Quan- contracting parties matter? Evidence from syndicated bank
titative Analysis, 46(1): 171–207. loans. Management Science, 58(2): 365–383.
Cuervo-Cazurra, A., Maloney, M. M., & Manrakhan, S. 2007. Grinblatt, M., & Keloharju, M. 2001. How distance, language,
Causes of the difficulties in internationalization. Journal of and culture influence stockholdings and trades. The Journal of
International Business Studies, 38(5): 709–725. Finance, 56(3): 1053–1073.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00
1565
Hadlock, C. J., & Pierce, J. R. 2010. New evidence on measuring Ohlson, J. A., & Juettner-Nauroth, B. E. 2005. Expected EPS and
financial constraints: Moving beyond the KZ index. Review of EPS growth as determinants of value. Review of Accounting
Financial Studies, 23(5): 1909–1940. Studies, 10(2–3): 349–365.
Hann, R. N., Ogneva, M., & Ozbas, O. 2013. Corporate Pastor, L., & Veronesi, P. 2013. Political uncertainty and risk
diversification and the cost of capital. The Journal of Finance, premia. Journal of Financial Economics, 110(3): 520–545.
68(5): 1961–1999. Ragazzi, G. 1973. Theories of the determinants of direct foreign
Heckman, J. J. 1979. Sample selection bias as a specification investment. Staff Papers (International Monetary Fund), 20(2):
error. Econometrica, 47(1): 153–161. 471–498.
Henry, P. B. 2000. Stock market liberalization, economic reform, Rauh, J. D. 2006. Investment and financing constraints: Evidence
and emerging market equity prices. The Journal of Finance, from the funding of corporate pension plans. The Journal of
55(2): 529–564. Finance, 61(1): 33–71.
Hope, O.-K., Kang, T., Thomas, W. B., & Vasvari, F. 2009. The Reeb, D. M., Kwok, C. C. Y., & Baek, H. Y. 1998. Systematic risk
effects of SFAS 131 geographic segment disclosures by US of the multinational corporation. Journal of International
multinational companies on the valuation of foreign earnings. Business Studies, 29(2): 263–279.
Journal of International Business Studies, 40(3): 421–443. Ronen, S., & Shenkar, O. 2013. Mapping world cultures: Cluster
Hughes, J. S., Logue, D. E., & Sweeney, R. J. 1975. Corporate formation, sources and implications. Journal of International
international diversification and market assigned measures of Business Studies, 44(9): 867–897.
risk and diversification. The Journal of Financial and Quantita- Schmid, M. M., & Walter, I. 2012. Geographic diversification
tive Analysis, 10(4): 627–637. and firm value in the financial services industry. Journal of
Jang, Y. 2017. International corporate diversification and financial Empirical Finance, 19(1): 109–122.
flexibility. The Review of Financial Studies, 30(12): 4133–4178. Sharpe, W. F. 1964. Capital asset prices: A theory of market
Karolyi, G. A., & Stulz, R. M. 2003. Are financial assets priced equilibrium under conditions of risk. The Journal of Finance,
locally or globally? In G. M. Constantinides, M. Harris, & R. 19(3): 425–442.
M. Stulz (Eds.), Handbook of the economics of finance, financial Shenkar, O. 2001. Cultural distance revisited: Towards a more
markets and asset pricing (pp. 975–1020). Amsterdam: Elsevier. rigorous conceptualization and measurement of cultural
Kaufman, D., Kraay, A., & Mastruzzi, M. 2009. Governance differences. Journal of International Business Studies, 32(3):
matters VIII—Aggregate and individual governance indicators 519–535.
1996–2008. Working Paper. Stulz, R. M., & Williamson, R. 2003. Culture, openness, and
Kogut, B., & Singh, H. 1988. The effect of national culture on finance. Journal of Financial Economics, 70(3): 313–349.
the choice of entry mode. Journal of International Business Villalonga, B. 2004. Diversification discount or premium? New
Studies, 19(3): 411–432. evidence from the business information tracking series. The
Kuppuswamy, V., & Villalonga, B. 2016. Does diversification Journal of Finance, 59(2): 479–506.
create value in the presence of external financing constraints? Yan, A. 2006. Value of conglomerates and capital market
Evidence from the 2007–2009 financial crisis. Management conditions. Financial Management, 35(4): 5–30.
Science, 62(4): 905–923.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 1998.
Law and finance. Journal of Political Economy, 106(6): 1113–1155.
Lewellen, W. G. 1971. A pure financial rationale for the
conglomerate merger. The Journal of Finance, 26(2): 521–537. ABOUT THE AUTHORS
Lewis, K. K. 2011. Global asset pricing. Annual Review of Financial Atanas Mihov is a Financial Economist in the
Economics, 3: 435–466. Quantitative Supervision and Research group at the
Li, J., & Guisinger, S. 1991. Comparative business failures of
foreign-controlled firms in the United States. Journal of Federal Reserve Bank of Richmond. He joined the
International Business Studies, 22(2): 209–224. Federal Reserve after completing his Ph.D. degree in
Li, J., & Guisinger, S. 1992. The globalization of service finance at the University of Florida. Atanas is
multinationals in the ‘‘triad’’ regions: Japan, Western Europe
and North America. Journal of International Business Studies, interested in a wide spectrum of research areas
23(4): 675–696. including international finance and banking.
Lin, L., Mihov, A., Sanz, L., & Stoyanova, D. 2018. Property
rights institutions, foreign investment, and the valuation of
U.S. MNCs. Journal of Financial Economics, Forthcoming. Andy Naranjo is the John B. Hall Professor of
Lintner, J. 1965. The valuation of risk assets and the selection of Finance and Chairman of the Finance, Insurance,
risky investments in stock portfolios and capital budgets. The and Real Estate Department in the Warrington
Review of Economics and Statistics, 47(1): 13–37.
Livdan, D., Sapriza, H., & Zhang, L. 2009. Financially constrained College of Business at the University of Florida. He
stock returns. The Journal of Finance, 64(4): 1827–1862. previously served as the Director for the Center for
Manova, K., Wei, S.-J., & Zhang, Z. 2015. Firm exports and International Business Education and Research at
multinational activity under credit constraints. Review of
Economics and Statistics, 97(3): 574–588. UF. His research interests encompass international
Morck, R., & Yeung, B. 1991. Why investors value multination- asset pricing and corporate finance issues.
ality. The Journal of Business, 64(2): 165–187.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
Supplementary information accompanies this article on the Journal of International Business Studies website (www.palgrave.com/journals).
Accepted by Lemma Senbet, Area Editor, 10 November 2018. This article has been with the authors for three revisions.

202.28.16.254 on Tue, 26 Sep 2023 13:38:52 +00:00

Leverage-Mihov-2019-JIBS

Uploaded by

Copyright:

Available Formats

Leverage-Mihov-2019-JIBS

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Leverage-Mihov-2019-JIBS

Uploaded by

Copyright:

Available Formats

Corporate internationalization, subsidiary locations, and the cost of equity capital

Author(s): Atanas Mihov and Andy Naranjo

Stable URL: https://www.jstor.org/stable/10.2307/48686780

This content downloaded from

Corporate internationalization, subsidiary

Atanas Mihov1 and Abstract

This content downloaded from

Journal of International Business Studies

This content downloaded from

Journal of International Business Studies

This content downloaded from

Journal of International Business Studies

This content downloaded from

Journal of International Business Studies

This content downloaded from

Cost of Equity Capital and Corporate Risk Diversification and Corporate

Journal of International Business Studies

This content downloaded from

Table 1 Variable definitions

Journal of International Business Studies

This content downloaded from

! vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi captures the return volatility of a portfolio of stock

Journal of International Business Studies

This content downloaded from

Table 2 Descriptive statistics and variable correlations

Panel A: Summary statistics (firm-year level)

Mean SD P25 P50 P75 n

CoE 14.79 8.38 10.24 12.15 16.04 36,823

Variables CoE Ln(NC) FSub Ret Vol Stk MktCorr FSales

Mean SD P25 P50 P75 n

CtryGov 0.00 1.94 - 1.51 - 0.41 1.35 2537

Journal of International Business Studies

This content downloaded from

Panel C: Summary statistics (country-year level)

Mean SD P25 P50 P75 n

Language 0.28 0.45 0.00 0.00 1.00 2523

Table 3 Cost of equity capital

(1) (2) (3) (4)

Ln(NC) - 0.19*** - 0.13** - 0.15***

Journal of International Business Studies

This content downloaded from

Table 4 Risk diversification

Panel A: Cost of equity and country stock market correlations

(1) (2) (3)

StkMktCorr 0.45*** 0.46*** 0.36**

Panel B: Firm risk and corporate internationalization

(1) (2) (3) (4)

Ln(NC) - 0.36*** - 0.36*** - 0.20

Journal of International Business Studies

This content downloaded from

Journal of International Business Studies

This content downloaded from

Table 5 Financial constraints

(1) (2) (3) (4) (5) (6)

Ln(NC) - 0.19*** - 0.10* - 0.22*** - 0.11 - 0.19*** 0.05

Country Governance the previously described WLS regression approach

Journal of International Business Studies

This content downloaded from

(1) (2) (3) (4) (5)

(1) (2) (3)

Ln(NC) - 0.19* - 0.13 - 0.15***

StkMktCorr 0.45* 0.46* 0.36**

Ln(NC) - 0.36* - 0.36* - 0.20

Ln(NC) - 0.19*** - 0.10* - 0.22* - 0.11 - 0.19* 0.05

FSales 0.67 1.10*

SATE - 0.33* - 0.44* - 0.52***