International Studies Quarterly (2015) 59, 330343

Lingua Mercatoria: Language and Foreign Direct Investment*

Moonhawk Kim
University of Colorado Boulder

Amy H. Liu
University of Texas at Austin

Kim-Lee Tuxhorn and David S. Brown

University of Colorado Boulder

and

David Leblang
University of Virginia
Does language choice attract foreign direct investment (FDI), and if so, how? We argue that languagea dynamic instru-
ment for reducing transaction costscan influence investors decision to allocate capital. Potential host countries attract
investments by coordinating their domestic language policiesespecially those in educationto match the language of
the potential FDI investor. We subject our argument to three different tests: (i) a cross-sectional sample of all global
Organization for Economic Co-operation and Development investments that employs a newly constructed language-in-
education measurement; (ii) a newly assembled time-series cross-sectional data set of all Chinese FDI abroad; and (iii) a
detailed case study that uses process tracing to explain Chinese FDI in Indonesia. The results from these tests demon-
strate a significant and robust relationship between language and FDI.

Moonhawk Kim is an assistant professor of political science at the Univer-
sity of Colorado Boulder. His research examines the evolution and the conse-
quences of the international trading system in terms of both the multilateral
institution of the GATT/WTO and preferential trade agreements (PTAs). His
broader research interests encompass international political economy and
international institutions. His work has appeared in World Politics, Interna-
tional Studies Quarterly, and International Theory.
Amy H. Liu is an assistant professor of government at the University of
Texas at Austin. Her research interests include language politics and the Chi-
nese diaspora. She has done fieldwork in Southeast Asia and Eastern Europe.
Her book, The Political Economy of Language Regimes (2014, University of Penn-
sylvania Press), examines language regime choice and the economic effects of
this choice. She has also published in Comparative Political Studies, Journal
of Politics, and World Politics.
Kim-Lee Tuxhorn is a PhD candidate in the Department of Political Sci-
ence at the University of Colorado Boulder. His research interests include
political economy, international organizations, and international bargaining.
David S. Brown (University of California, Los Angeles, Ph.D. 1995). Recent
articles by him have appeared in The American Political Science Review
(2009), World Development (2011), Latin American Politics and Society, and
Studies in Comparative International Development (2014). His primary
research and teaching interests are in the areas of comparative politics, eco-
nomic development, and Latin American Politics.
David Leblang is Professor of Politics at the University of Virginia and is a Fac-
ulty Associate at the Miller Center where he is the J. Wilson Newman Professor of
Governance. As a scholar in the area of international political economy, he has
expertise in international immigration, capital flows, financial crises, and
exchange rate arrangements. Recent articles have appeared in International
Organization, The American Political Science Review, and World Politics. He is
the editor of SSRNs International Political Economys Migration eJournal.
*Earlier versions of this article were presented at the 2011 Annual Meet-
ing of the American Political Science Association (APSA) and at Emory Uni-
versity Political Science Brown Bag in 2011. We thank the attendees at both
presentations for questions and comments. In particular, we acknowledge
valuable feedback from the following scholars: Claire Adida, Clark Gibson,
David Laitin, and Tom Pepinsky at the 2011 APSA conference; Richard
Doner, Jennifer Gandhi, Michael Giles, Sabrina Karim, Jane Lawrence, Rick
Morgan, Eric Reinhardt, Tom Remington, and Josh Strayhorn at the Emory
Brown Bag.
Kim, Moonhawk et al. (2015) Lingua Mercatoria: Language and Foreign Direct Investment. International Studies Quarterly, doi: 10.1111/isqu.12158
2014 International Studies Association
Although Spain once colonized the Philippines, Spanish
has largely vanished from everyday use throughout the
archipelago (Inquirer 2010). Yet in February 2010, Spain
and the Philippines signed an agreement for Spain to
help reincorporate the Spanish language into the Filipino
education curriculum and make this language available
on a large scale. Not coincidentally, this agreement came
2 months before Spanish business executives visited Man-
ila to explore investment ventures (Philippines Depart-
ment of Trade and Industry 2010). This pattern of
language policy changes preceding foreign investment
decisions did not happen only in the Philippines. A num-
ber of other countries actively adopted language policies
in hopes of attracting foreign capital. In this article, we
ask: Does language attract foreign direct investment
(FDI), and if so, how?
Our theory builds on an economic argument about
transaction costs. Potential receiver-states attract FDI by
decreasing transaction costs for foreign investments. We
argue that language policiesspecifically language-in-edu-
cation onesare instrumental to achieve this objective.
Potential receiver-states strategically choose to teach the
official language of an FDI sender-state. FDI from the
sender-state can increase as a result of this policy change.
Scholars recognize the international economic effect of
sharing a common language. In fact, language is featured
prominently in FDI models (B
enassy-Qu
er

e, Coupet, and
Mayer 2007; Leblang 2010). Often, the variable measures
whether two countries share a common official language
using data from Centre dEtudes Prosepctives et dInfor-
mations Internationales (CEPII).
We have three concerns with the CEPII measure. First,
if language is an instrument to reduce transaction costs,
then we cannot focus solely on whether two countries
share an official language. Doing so ignores the large set

Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct

by guest
on 09 October 2017
 Moonhawk Kim et al.
of non-official languages that these countries citizens
may know and use. In other words, governments do not
have to designate a language as official for it to function
as a prominent vehicle for communication. In this study,
we shift focus from official languages to the languages
taught in schools, which the government can more easily
manipulate. Such language-in-education policies can be
contentious because they reflect the governments vision
of the future (Kaplan and Baldauf 1997; Albaugh 2009).
Second, language remains a static measurement.
CEPIIs common official language variable is time invari-
ant. But, language is a dynamic instrument. Governments
use language policies to achieve various political objec-
tives over time. The Philippines is one example of a gov-
ernment actively changing language-in-education policies
to facilitate international economic exchange. A related
third problem concerns data construction. CEPII uses
Ethnologue and the CIA World Factbook. While both
sources are valid, CEPII relies on the most current
editions of each publication. To assume that current
language(s) and language policies match those from the
past is inaccurate and problematic. To address all three
concerns, we introduce a new language-in-education
variable focused on the teaching of foreign/second
languages over time.
This article proceeds as follows. We first review existing
theories explaining FDI flows. Then, highlighting the
approach that states reduce transaction costs to increase
their attractiveness to foreign investors, we articulate a the-
ory linking language-in-education policies to FDI. In the
following sections, we analyze this theory using three dif-
ferent tests. The first is a cross-sectional sample of all Orga-
nization for Economic Co-operation and Development
(OECD) investments globally (2002) that employs a newly
constructed language-in-education measurement. The sec-
ond is a new time-series cross-sectional data set of all Chi-
nese FDI abroad (20052009). The third is a detailed case
study of Chinese FDI in Indonesia. The results from these
tests demonstrate that language policiesspecifically lan-
guage-in-education policiesare dynamic instruments for
attracting FDI. We highlight the contributions of the arti-
cle and discuss the implications of our findings in the last
section.
Determinants of FDI
Originally designed to model trade flows (Anderson and
Van Wincoop 2003), the gravity model has become a reg-
ular fixture in the FDI literature (B
enassy-Qu
er

e et al.
2007). In this model, the principal factors that account
for bilateral investment flows are the economic size of
the two states and the geographical distance between
them. Larger economies are more likely to send and
receive direct investments. The distance between states
can also impede economic exchanges. The gravity model
stands in contrast to the traditional framework for analyz-
ing FDI flows, which focuses on ownership, location, and
internalization (OLI) (Dunning 1977).
Scholars have incorporated other factors into the grav-
ity model that facilitate or impede FDI flows. These extra-
gravity factors fall into two categories. The first, mirroring
research in the economic development literature, focuses
on the quality of governance institutions (Globerman
and Shapiro 2003; B
enassy-Qu

er

e et al. 2007). Demo-
cratic institutions can increase FDI flows by lowering the
political risk of investment (Jensen 2003, 2008; Li and
Resnick 2003; also see Addison and Heshmati 2004).
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
331
International institutions such as bilateral investment
treaties (Elkins, Guzman, and Simmons 2006) and prefer-
ential trade agreements (PTAs) can also promote FDI by
creating a mechanism for credible commitment (B uthe
and Milner 2008).
The second category captures the relationship between
sending and receiving states. The geographical distance
between two statesa key component of the gravity
modelis an example. Neighboring states are expected
to have close economic relations. Distance measures rep-
resent neither the value nor the cost of foreign invest-
ments but the barriers to carrying out economic
transactions. Accordingly, these measures of bilateral rela-
tionships frequently capture the transaction costs of FDI.
In addition to physical distance, scholars have examined
the cultural distance between states (Bandelj 2002).
Scholars frequently operationalize this concept as a com-
mon language or common legal heritage shared by two
states. States with the same languages or legal structures
are more likely to engage in economic exchanges (Linne-
mann 1966).
Contrary to domestic and international institutions
(which politicians can influence), physical distance and
cultural distance appear to be largely exogenous. For
example, common language is often regarded as a static
political fact (Hsiao and Hsiao 2004). In one exception,
Lien et al. examine the effects of Chinese language insti-
tutes on international economic flows (Lien, Oh, and Sel-
mier 2012). Leblang also identifies a dynamic
relationship between culture and economics: Diaspora
networks can influence FDI by providing familiarity and
information flows between two countries (Leblang 2010).
On the whole, however, scholars have overlooked how
states can manipulate this cultural affinity. Education is
an effective tool for increasing affinity between states. In
particular, language-in-education policies can decrease
transaction costs by producing entrepreneurs and workers
that are competent in the language of FDI-sending states.
In the next section, we develop this argument in greater
detail.
Language as a Dynamic Instrument
FDI across Languages
States manipulate economic policies and participate in
international institutions to reduce the monetary costs of
investments, thereby increasing the net value of FDI
(Hines 1995; Blonigen and Figlio 1998; De Mooij and
Ederveen 2001). States can also lower costs of investments
by increasing their commitment to liberal investment
polices and reducing the uncertainty of net investments
to foreign investors (B uthe and Milner 2008; Froot and
Stein 1991:12041205). In a similar manner, we argue
that states can pursue particular language policies to
reduce the transaction costs of economic investments.
FDI-receiving states can alter their citizens proficiency in
the languages of prominent FDI-sending states. When a
greater portion of the receiver-states population is famil-
iar with the language of the sender-state, more invest-
ments are likely. Language is far more dynamic than
previously credited in the FDI literature and can be an
effective tool for attracting FDI. Citizens can learn new
languages, and governments can adjust language-in-
education policies over time to minimize transaction costs.
Language differences between entrepreneurs and
laborers in FDI-sending and FDI-receiving states can
 332
create friction at two stages in their economic transac-
tions: (i) search and contracting and (ii) management
and operations. First, foreign investors need to search for
potential firms and opportunities in the receiver-state.
They also need to secure a contract that embodies the
negotiated terms of a transaction. Language differences
can lead domestic entrepreneurs to communicate ineffec-
tively with foreign counterparts and prevent the parties
from finding mutually beneficial opportunities. Second,
once a contract is in effect, the investing firm needs to
oversee and manage everyday operations. Foreign firms
typically import managers from their home countries to
manage the workers in host countries, and language dif-
ferences between managers and workers can hamper
smooth operations. The first type of friction (at the
search and contracting stage) is generally more common,
as FDI typically seeks out countries with abundant endow-
ment in unskilled labor (Yeaple 2003), implying lesser
need for language abilities of workers.
The parties can resolve these frictions in a number of
ways. First, the foreign investor and the domestic partner
may choose not to take any extraordinary measures. In
any FDI-receiving state, a subset of the population likely
already knows the languages of FDI-sending countries.
Second, the parties can hire translators to facilitate com-
munication. Third, the parties can rely on a third-party
language (for example, English) to communicate. Fourth,
foreign investors can expend resources and train their
personnel to learn the FDI-receiving states language.
Fifth, entrepreneurs and workers in the potential recei-
ver-state can learn the languages of the FDI-sending state.
From the perspective of states seeking to attract FDI,
the last option is the most attainable. By increasing the
domestic population familiar with the FDI-sending states
language, FDI-seeking states can reduce the associated
transaction costs. Lower transaction costs increase the
likelihood and the amount of investment. In this argu-
ment, however, we do not assume that the existence of
transaction costs from language frictions is the underlying
motivation for FDI. A transaction cost approach antici-
pates that firms will engage in investing abroad to lower
particularly high or persistent costs (Williamson 1975:84).
By contrast, we assume that other motivations for invest-
ments exist, such as the benefits of OLI (Dunning 1977),
but that language differences constitute frictions that FDI
senders seek to minimize whenever possible.
For example, consider Singapore where four languages
are considered official: Chinese, Malay, Tamil, and Eng-
lish. Despite this linguistic diversity, the increasing use of
English has empowered the city-state to be a competitive
economic hub. In the initial years after gaining indepen-
dence, with only 21% of the population claiming literacy
in the English language (Dixon 2005), top government
officials traveled to cities such as London and New York
to lobby foreign investors. Along with strategic port
access, the potential for an English-speaking workforce
proved pivotal in their sales pitch (Lee 2000:150152;
Wang 2007). Today, English is the native household lan-
guage of more than 40% of the population (Dixon
2005). As a reward for this increase in English profi-
ciency, Singapore is the worlds top logistics center with
an expanding economy at 14.5% GDP growth (Adam and
Tan 2011).
If teaching a potential FDI-sending states language
attracts FDI, then we might also believe the converse:
The intentional discontinuation of a sender-states
language should hinder FDI. Prior to 1970s, in Malaysia,
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
Lingua Mercatoria
students could learn both Chinese (as a minority lan-
guage) and English (as the colonial language). In fact,
English was a common medium of educational instruc-
tion. But with the adoption of the New Economic Policy
(19701990s), Malay became the singular language of
education, Chinese faced substantial reductions, and Eng-
lish was replaced in the curriculum (Ganguly 2003).
These changes had negative implications for foreign
investments (Esman 1987:406). Many middle class profes-
sionalsespecially the Chinesepermanently moved
abroad (Lee 2000:142). This brain drain paralleled a
decreased proficiency in English among Malays. By the
1990s, Malaysia found itself struggling to stay relevant
and competitive in the increasing globalized knowledge
economy (Lee 2007:139).
Language Policies and FDI
In general, shifts in language-in-education policies are
more common than changes in official languages (see
Marschan-Piekkari, Welch and Welch 1999). For instance,
South Korea officially introduced English as a second lan-
guage into its education curriculum in 1997. But, the ini-
tial scope was limited to the third grade. This changed in
2002, when English learning was expanded through the
sixth grade (Butler 2004). The development of English in
Taiwans education curriculum was also incremental. In
1998, it was introduced in select areas. By 2001, it had
expanded nationwide (Marschan-Piekkari et al. 1999).
There are two types of language-in-education policy
(Liu 2011). The first involves language as a medium of
instruction. When a language is a medium, it is used to
teach non-language classes (for example, history and
math) and to carry out the broader educational curricu-
lum. The status of English in Singaporean schools is one
example. With the exception of the mother tongue clas-
ses (Chinese, Malay, and Tamil), all subjects are taught in
English (Ganguly 2003; Tan 2007). By contrast, the sec-
ond type of policy involves languages taught as a foreign/
second language. A language is considered foreign or
second if it is recognized as a class subject by the educa-
tion ministry. The reincorporation of Spanish into the
Philippine curriculum is an example of a foreign/second
language. Although Filipino students now have the
option to take Spanish classes, the larger curriculum
remains in Filipino and English (Tupas 2007).
While both policy types are important, we focus on for-
eign/second languages. Because of lower political and
financial barriers, states are more likely to offer a foreign
language as a subject matter than to employ it as a med-
ium of instruction. Moreover, the choice of a foreign lan-
guage as a medium of instruction is highly contentious
and often incites divisive domestic political conflicts. In
short, the foreign/second language policy more closely
represents the theoretical logic examined in this study.
Following the hypothesis that language-in-education pol-
icy is a strategic instrument for attracting FDI, we predict
the following:
FDI-receiving states teaching a foreign/second lan-
guage that matches the FDI-sending states official lan-
guage(s) receive more FDI from the sender-state than
receiver-states that do not teach matching foreign/sec-
ond languages.
Moreover, we expect that longer temporal lags in profi-
ciency of the matching language will have greater effects
 Moonhawk Kim et al.
on FDI flows. First, since subject matter instruction can
take place as early as the primary level of education, stu-
dents receiving language training will require some time
before they enter the workforce with the requisite lan-
guage familiarity. Second, as each year passes, a new
cohort of students having received instruction graduates.
As the lags increase, the proportion of the population
familiar with the language of concern also increases.
Empirical Analysis
Ideally, we would test the proposed mechanism of trans-
action cost reduction as well as the observed outcome of
FDI flows. However, there is no systematic data on all FDI
opportunities that firms explored irrespective of their
actual investments. We would need these data to directly
examine change in investment costs. In the absence of
such data, we test the hypothesis using three different
types of evidence on observed FDI flows. First, we employ
a cross-sectional analysis using FDI outflows from OECD
member states to other states around the world. This
analysis confirms an overall pattern consistent with our
argument. Second, we run an analysis of FDI outflows
from China to all other states. This test shows that our
argument is more than simply an English language effect
and demonstrates the longitudinal effects of language
policy variation. Lastly, a case study of Indonesias efforts
to attractand increaseChinese FDI through language-
in-education policy reforms illustrates the causal mecha-
nism underlying our argument.
The main empirical challenge is the possibility of
simultaneous causation: Language policies may follow FDI
inflows from sender-states. We address this concern in
two ways. First, we empirically examine the lagged effects
of language policies. More specifically, we employ lags
that far exceed the conventional statistical approach. If
proficiency in the foreign language is indeed the mecha-
nism at play, we would expect FDI effects to be substan-
tively larger after sufficient time has passed for educated
students to integrate into the workforce. Second, theoreti-
cally, even if states implemented language-in-education
policies in response to extant FDI levels from a particular
sender-state, it is important to observe whether there is
an increase in future FDI inflows. It is reasonable to
assume that states do not begin teaching a foreign lan-
guage at random and that existing economic activity
motivates changes in educational practices. We focus on
whether and how language policy choices retain extant
investments and attract more FDI. This would also dem-
onstrate that having a population conversant in language
X can make a state more attractive to FDI-sending states
that speak X and also that states can strategically manipu-
late their language-in-education policies to influence
subsequent FDI flows.
Test 1: Cross-Sectional FDI Outflows from OECD States
Research Design
The unit of analysis is the sender-statereceiver-state dyad.
The sample contains all dyads of 26 OECD member states
(sender-states) and 143 receiver-states that received non-
negative outflows from the sender-state. There are three
reasons to focus on OECD members. First, OECD FDI
flows are the most important cases since its members con-
tribute the most outward FDI globally. For example in
2004, of the total US dollars (USD) 407 billion in FDI
flows, USD 261 billion originated from an OECD
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
333
member state.1 Second, for empirical analysis, the
inclusion of non-relevant dyads (for example, Hungary
Uruguay) would magnify the number of cases where FDI
flows are zero. The inclusion of these negative cases
when there is no theoretical reason for the outcome to
be positivecan bias the estimates (Mahoney and Goertz
2004). Put differently, not every state can invest in other
states, but every state can receive FDI. Consequently, we do
consider all states in our list of receiver-states. Third,
systematic cross-national bilateral FDI data are available
only for investments from OECD countries.
The dependent variable is FDI flow. FDI data were
compiled from the OECD database. Foreign investments
are measured in dollars and logged. We add one to
observations with a zero value before taking the loga-
rithm. The migrant-related variables, which critically
influence FDI flows according to Leblang (2010), are
available only for limited time periods and restrict our
analysis to the year of 2002.
The main explanatory variable measures the teaching
of foreign/second languages (hereafter referred to as
foreign/second language variable). A language is con-
sidered a foreign/second language if the states educa-
tion ministry designates it as a subject. Students learn this
language for some allotted period in the educational cur-
riculum. Conceptually, this parallels other substantive
classes in the curriculum, such as history and math.
Foreign/second language is coded as one if the sender-
states official language matches one of the receiver-
states foreign/second languages, and zero otherwise.
The foreign/second language variable is coded using
various sources. Jacques Leclercs (2011) database on lan-
guage policies (Lam e nagement linguistique dans le monde),
which covers all countries over time, provided the initial
information for most entries. We cross-checked the infor-
mation from this database and in some cases supple-
mented it with older editions of International Handbook of
Education Systems and Ethnologue. This second step was crit-
ical for capturing any differences between current and
past language policies.
As discussed above, we expect the lags of language-in-
education policies to be critical to the outcome of
increased FDI. Accordingly, we create lags of varying
lengths. The models contain 1-, 10- and 20-year lags.
While we do not have specific hypotheses about each lag
length, we anticipate that longer lags will have greater
effects on the FDI flows. The foreign/second language
variable is partially correlated with CEPIIs common offi-
cial language variable (0.42 and 0.45). The relatively low
correlation highlights that this measure of language
familiarity is substantively different from the prevalent
and static official language measures. Table 1 presents
correlations between the foreign/second language vari-
ables and other variables that may be correlatedthe
common official language measure, the common civiliza-
tion measure, and the migrant stock measures.
A cross tabulation of the foreign/second language and
the CEPII common official language variables illuminates
the empirical contribution of our variable. In 1,444 obser-
vations, both variables are zero, and in 147 observations,
both are one. By contrast, in 69 observations, common
official language is zero, while foreign/second language is
one. These are cases in which the official language
variable fails to capture the ability of citizens in an
1
OECD Annual Report 2005.
 334 Lingua Mercatoria

TABLE 1. Table of Correlations, Test 1 GDPGDP for both sender- and receiver-states was
sourced from the World Bank database. Larger states
(1) (2) (3) (4) (5) (6)
both send and receive more foreign investments. Given
Foreign/Second the non-normal distribution of GDP across countries, we
Language (t 1) log GDP of both states in dyads.
Foreign/Second 0.989
Language (t 10) Bilateral tradeTrade within the dyad captures existing
Foreign/Second 0.943 0.954 economic exchange levels between the states. States that
Language (t 20) trade more with each other are more likely to engage in
Common Official 0.420 0.426 0.451 FDI, even after controlling for gravity model factors. The
Language measure is logged. The data are from the IMFs Direction
Common Civilization 0.242 0.250 0.229 0.0566 of Trade Statistics (DOTS), updated by Goldstein, Rivers,
Sender Migrant 0.218 0.222 0.213 0.0973 0.483 and Tomz (2007).
Stock in Receiver
Receiver Migrant 0.342 0.342 0.328 0.155 0.289 0.601 Common Official LanguageThis is the language
Stock in Sender measure that scholars predominantly use in the
(Note. (1) Foreign/Second Language (t 1); (2) Foreign/Second Language literature. It is coded as one if an official language for
(t 10); (3) Foreign/Second Language (t 20); (4) Common Official the states in a dyad matches and zero otherwise. The data
Language; (5) Common Civilization; (6) Sender Migrant Stock in Receiver.) are from CEPII, and we implicitly consider this measure
as a competing hypothesis (that is, the relationship
between the common official language and FDI flows).
FDI-receiving state to use the foreign language in ques-
tion. Moreover, in 250 observations, while the official lan- Common civilizationWhen two countries are culturally
guage is coded as one, foreign/second language is not. divided, there are larger transaction costs, higher
These are cases in which receiving states citizens are not uncertainty about business practices, and greater unease
taught the foreign language in question, despite the fact about the prospects for doing business (Kogut and Singh
that the two countries share an official language. Because 1988; Habib and Zurawicki 2002; Siegel, Licht, and
an official language might simply be the language of a par- Schwartz 2013). By contrast, when two countries are
ticular ethnic group, the majority of the countrys popula- culturally similar, their citizens are more likely to engage
tion may not be conversant in it without explicit training. in economic transactions (Guiso, Sapienza, and Zingales
We also account for the lack of effects that a common offi- 2005; White and Tadesse 2008; Siegel et al. 2013). To
cial language has on FDI flows in our analysis. control for this cultural proximity, we use Russett, Oneal,
A simple correlation reveals that the foreign/second and Coxs (2000) civilization categories.
language variable is more closely associated with interna-
tional economic flows than the common official language Common legal heritageLa Porta, Lopez de Silanes,
variable. The correlation between foreign/second lan- Shleifer, and Vishny show a strong correlation between the
guage and bilateral trade flows is 0.20, whereas the corre- historical origins of a countrys judicial system and
lation between common official language and bilateral international economic outcomes (La Porta, Lopez-de-
trade flows is 0.02. Similarly, the correlation between Silanes, Shleifer, and Vishny 1999). Common legal heritage
foreign/second language and FDI flows is 0.29, whereas may also account for colonial factors related to economic
the correlation between common official language and cooperation. The data were compiled from La Porta et al.
FDI flows is 0.08. This variable is dichotomous and coded as one if the states
We also include three broad sets of control variables: share a common legal heritage, otherwise coded as zero.
factors based on the gravity model (distance, contiguity,
and economic size); measures of pre-existing affinity Migrant stockLeblang demonstrates that migrant
between states (bilateral trade, common official language, stocks meaningfully improve information flows and thus
common civilization, common legal heritage, and migrant investment flows between states (Leblang 2010). We
stock in one state from the other); and institutions that include his measures of migrant stocks, incorporating
can decrease transaction costs between states (shared both the stock in the FDI-receiving state from the sender-
GATT/WTO membership, dual taxation treaty, preferen- state and the stock in the FDI-sending state from the
tial trade agreement, common currency peg, and alli- receiver-state. Both measures are logged.
ance). Lastly, we control for the level of democracy in
receiver-states. Common WTO membershipBlomstrom and Kokko note
a direct relationship between World Trade Organization
DistanceThe data for geographic distance come from (WTO) membership and FDI. [T]he WTO regulates FDI
the CEPII database and measure the distance between incentives in its agreements on Subsidies and
two states capitals at 100-kilometer increments. The Countervailing Measures (SCMs) and Trade-Related
variable is logged. Like GDP, distance is a key term in the Investment Measures (TRIMS) (Blomstrom and Kokko
gravity model and typically considered a key measure for 2003:19). WTO membership status can be found on
capturing transaction costs in both the FDI and trade WTOs Web site. The variable is coded as one if both
literature (Bevan and Estrin 2004). states are member of the WTO, otherwise coded as zero.

Contiguous bordersFollowing the logic of distance in Dual tax treatyTaxes can inhibit FDI flows (De Mooij
gravity models, a dichotomous variable is included to and Ederveen 2001). To redress this issue, states have
capture whether the senderreceiver dyads are neighbors. passed domestic laws, created special tax-free economic
The variable is coded as one for neighboring dyads and zones, and signed international agreements to reduce
zero otherwise. transaction costs. Dual tax treaties are signed to minimize
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
 Moonhawk Kim et al. 335

FDI-related costs and to overcome the problem of double foreign/second language variable and FDI levels. We esti-
taxation. The data are from the United Nations mate the model using ordinary least squares (OLS) and
Conference on Trade and Development (UNCTAD). The control for unobserved receiver-state-specific variation
variable is coded as one if both states are signatories to a with fixed effects. To deal with the potential bias of FDI-
dual taxation treaty, and otherwise coded as zero. receiving states inflows from multiple sender-states, we
cluster the standard errors on the receiver-states.
Shared PTAFDI may covary with trade flows and trade
agreements (Makki and Somwaru 2004). The data come Results and Discussion
from Goldstein et al. (2007). The variable is dichotomous Table 3 summarizes the results. As discussed above, we
and coded as one if the sender-state and receiver-state vary the lags for the foreign/second language variable
share an agreement, otherwise coded as zero. from 1 year (Model 1) to 10 years (Model 2) to 20 years
(Model 3). Several variables are statistically significant
Common exchange rate pegExchange rate volatility has across all three models. Distance has the expected negative
been identified as a transaction cost. States can use fixed effect. States invest less in other states that are farther
exchange rates to stabilize a state currency against that of away. Interestingly, contiguity also has a negative effect. We
another state. Fixed exchange rates typically make trade believe, however, that this captures a geographical trend
and investment easier and more predictable between two rather than an economic one: Dominant sender-states
states. The data are from KleinShambaugh Nature of tend not to be contiguous with receiver-states, rather than
Exchange Rate Regimes Data (Klein and Shambaugh firms being resistant to invest in neighboring states. Recei-
2008). The variable is coded as one if the states share a ver-states with larger economies receive more FDI,
common exchange rate peg, otherwise as zero. although sender-states economic size does not systemati-
cally affect their FDI outflows. Supporting Leblangs
AllianceSecurity externalities may exist between allied (2010) findings, when a greater migrant stock from recei-
states, which facilitate investment as well as trade flows. ver-states exists in sender-states, more FDI occurs from the
Using the Alliance Treaty Obligations and Provisions latter to the former. In terms of international institutions,
(ATOP) data (Leeds, Ritter, Mitchell, and Long 2002), the presence of shared WTO membership, dual taxation
we code dyads that belong to a common alliance as one treaties, and a common peg for fixed exchange rates
and those that do not as zero. increases FDI flows from sender- to receiver-states.
The foreign/second language variable is statistically sig-
DemocracyLastly, scholars have identified democracy nificant across all three models. Although the 10-year lag
as a key factor for attracting FDI inflows because of the shows a slightly smaller effect than the 1-year lag, there is
regime types ability to commit to sound economic an overall increase in the effect of the foreign/second
policies (Jensen 2008). To measure regime type, we use language education over time. Note that the common
POLITY data (Marshall and Jaggers 2008). We adopt the official language variable is not significant in any of these
convention of combining the two scales into one single models. Because it is such a static measure, the official
index, which ranges from 10 to 10, with 10 being the languages match in only 129 dyads in the analysis. By
most democratic. contrast, the language-in-education policy of a receiver-
Table 2 presents summary statistics of all the variables. state matches the official language of a sender-state in
The theory posits a direct, positive relationship between 345 dyads. The common official language variable fails to
account for variation in FDI flows even in the absence of
TABLE 2. Summary Statistics, Test 1 the foreign/second language measure. This suggests that
education policy can manipulate economic actors knowl-
Mean SD Min/Max edge of languages, which in turn has a large impact on
foreign economic relations.
(logged) Foreign Direct 10.7 8.65 0/24.9
Moreover, the foreign/second language education vari-
Investment flow
Foreign/Second Language (t 1) 0.252 0.434 0/1
able is substantively significant. Because the dependent
Foreign/Second Language (t 10) 0.247 0.432 0/1
variable is log-transformed, we interpret exponentiated
Foreign/Second Language (t 20) 0.229 0.420 0/1 coefficient estimates as percentage changes in FDI flows.
(logged) Distance 8.26 1.07 4.09/9.88 With the 1-year lag, a state that provides instruction in the
Contiguous Borders 0.0489 0.216 0/1 FDI-sending state language state receives 177% more
(logged) GDP: Sender 26.6 1.56 22.7/29.9 investments than a state that does not provide such
(logged) GDP: Receiver 25.2 1.89 20.4/29.9 instruction. With the 20-year lagthat is, a state that
(logged) Bilateral Trade 18.2 3.72 0/26.1 began offering instructions of the foreign/second lan-
Common Official Language 0.0941 0.292 0/1 guage 20 years agothe state receives 219% greater invest-
Common Civilization 0.350 0.477 0/1 ments than a state that has not provided such instruction.
Common Legal Heritage 0.335 0.472 0/1 Admittedly, language-in-education policy changes can pro-
(logged) Sender Migrant 4.23 4.32 0/14.8 duce some immediate benefits by signaling that an FDI-
Stock in Receiver receiving state intends to provide favorable investment
(logged) Receiver Migrant 7.04 3.00 0/16.2 conditions. However, the largest impact of these policies
Stock in Sender occurs in the long run as the proportion of entrepreneurs
Common WTO Membership 0.867 0.339 0/1 and workers familiar with the foreign language increases
Dual Tax Treaty 0.412 0.492 0/1 over time. Including all three lags of the variable main-
Shared PTA 0.341 0.474 0/1
tains the size and the significance of the 20-year lag while
Common Exchange Rate Peg 0.070 0.255 0/1
rendering the other two lags non-significant.2
Alliance 0.429 0.495 0/1
Receiver-State Democracy 5.97 5.46 10/10
N 1371 2
The results are available from the authors.

Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct

by guest
on 09 October 2017
 336 Lingua Mercatoria

TABLE 3. Languages and Foreign Direct Investment (FDI)

Model 1 (1-Year Lag) Model 2 (10-Year Lag) Model 3 (20-Year Lag)

Foreign/Second Language 1.019 (0.464)* 0.989 (0.466)* 1.159 (0.497)*

(logged) Distance 1.327 (0.481)** 1.320 (0.481)** 1.302 (0.481)**
Contiguous Borders 2.514 (0.768)** 2.501 (0.767)** 2.497 (0.768)**
(logged) GDP: Sender 0.128 (0.196) 0.132 (0.197) 0.133 (0.197)
(logged) GDP: Receiver 1.516 (0.393)*** 1.502 (0.394)*** 1.338 (0.390)***
(logged) Bilateral Trade 0.142 (0.150) 0.142 (0.150) 0.144 (0.150)
Common Official Language 0.145 (0.650) 0.134 (0.646) 0.263 (0.656)
Common Civilization 0.389 (0.765) 0.393 (0.765) 0.404 (0.762)
Common Legal Heritage 0.364 (0.428) 0.361 (0.428) 0.351 (0.429)
(logged) Sender Migrant Stock in Receiver 0.061 (0.106) 0.0621 (0.106) 0.0600 (0.106)
(logged) Receiver Migrant Stock in Sender 1.292 (0.137)*** 1.296 (0.138)*** 1.293 (0.138)***
Common WTO Membership 7.815 (1.931)*** 7.787 (1.933)*** 7.576 (1.918)***
Dual Tax Treaty 3.589 (0.727)*** 3.590 (0.726)*** 3.566 (0.723)***
Shared PTA 1.374 (1.351) 1.399 (1.352) 1.403 (1.350)
Common Exchange Rate Peg 1.740 (0.607)** 1.743 (0.607)** 1.760 (0.608)**
Alliance 0.536 (0.805) 0.551 (0.806) 0.551 (0.807)
Receiver-State Democracy 1.426 (1.038) 1.403 (1.041) 1.083 (1.027)
Constant 23.97 (10.57)* 23.72 (10.57)* 21.98 (10.62)*
N 1371 1371 1371
(Note. Robust standard errors reported and clustered by receiver-states. Receiver-state fixed effects are not reported for space constraints. *p < .100, **p < .050,
***p < .010.)

Despite the control variables we include, the findings
may result from an omitted variable. Given that the lon-
gest lag of the language variable is 20 years, if an omitted
variable indeed exists, it is likely a time-invariant factor.
The only reasonable dyad-specific factor we exclude from
our model is colonial legacy. This variable is collinear
with other more important variablesGATT/WTO mem-
bership, dual taxation treaty, and the receiver democracy
scorebut these likely constitute post-treatment variables
with respect to colonial legacies. To test for the effect of
colonial legacy between two states, we exclude these three
variables. We report the results in a supplementary
appendix (Table S1), and the results support our find-
ings. The coefficient on the 20-year lagged language vari-
able is indeed larger than in Model 3, although the
colonial measure is negative and significant.
These results withstand a variety of other robustness
checks. While we do not report these checks due to space
constraints, the results all support our argument. First,
rather than creating long lags of the dichotomous explan-
atory variable, we test the argument with a continuous
measure that tracks the number of years that language-in-
education matches the official language of FDI senders.
The result is positive and significant, indicating that
countries with longer years of matching foreign/second
language instruction receive greater FDI. Second, FDI
sender-states can also be receiver-states in the data set.
This captures the general pattern that OECD FDI out-
flows generally go to other OECD countries. Excluding
OECD receivers of FDI does not substantively alter the
results. Third, the United States might be a special case
on the basis of its large economy and its usage of English
the de facto international lingua franca. Excluding the
United States as both an FDI-sending and receiving state
does not alter the results, suggesting that the English lan-
guage is not driving our argument. Last, pushing this
logic further, we exclude the main English-speaking
OECD states from our sample (Australia, Canada, Great
Britain, New Zealand, and the United States). The effects
of the foreign/second language variable become larger
after this exclusion. The coefficient estimate doubles in
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
size, which suggests that non-English foreign/second lan-
guage instruction has a stronger effect on attracting FDI
flows from OECD states that use the languages in ques-
tion.
While these results support our argument, there are
still two concerns. Theoretically, English language instruc-
tion in schools is common in most countries regardless of
their colonial origins. As such, it is possible that while a
potential receiver-state (for example, Hungary) does not
learn the official language of a sender-state (for example,
Japan), their shared knowledge of English might be suffi-
cient to reduce transaction costs, thereby facilitating FDI.
Empirically, the analysis excludes some of the larger non-
OECD economies, for example, Brazil, Russia, India, and
China (BRIC). If potential receiver-states use language
strategically to attract FDI, we should observe an increas-
ing educational emphasis on the official languages of
these non-OECD countries. In the next section, we
address both of these concerns.
Test 2: Chinese Language and Chinese FDI
In this section, we shift our analysis to China. Specifically,
we examine whether potential receiver-states attract more
Chinese FDI when they incorporate Mandarin Chinese
into their educational curricula. We focus on China
because it is the only BRIC country where the official lan-
guage is not a former colonial language. This allows us to
eliminate colonial connections (for example, Brazil
Mozambique, RussiaUkraine, and IndiaSouth Africa) as
a reason for investment decisions. Moreover, outside of
China, only one state has Chinese as an official language:
Singapore. These conditions ensure that educational
incorporation of the Chinese language is driven by eco-
nomics, rather than cultural affinity, and that the esti-
mated effects of language policies are not confounded by
deeper cultural factors.
Research Design
The Chinese Ministry of Commerce has data on Chinese
FDI (excluding Hong Kong) by receiver-states from the
 Moonhawk Kim et al. 337

years 2004 to 2010 (2010 Statistical Bulletin of Chinas Out- as with the first analysis, is measured using the combined
ward Foreign Direct Investment). All FDI outflows are 21-point POLITY index.
recorded in millions and in current USD. All numbers There are a few distinctions between this model and
have been transformed and standardized to constant the previous one. First, we leave out the GDP of the sen-
USD. In general, Chinese FDI levels have exponentially der-state, as China is the only sender in this sample. Simi-
increased with each passing year. In 2004 (the first year lar logic dictates the omission of common official
of the sample), the average FDI across the 158 receiver- language. Second, we exclude the migrant stock variables
states is 7.65 million. By 2010 (the last year of the sam- and the alliance variables because they are unavailable
ple), the same statistic had increased to 110.66 million. for this sample. Third, we do not control for common
Given the non-normal distribution of the variable, we log exchange rate peg because the Chinese RMB is pegged
FDI. against a basket of currencies. Finally, we drop common
The explanatory variable is Chinese language instruc- civilization given the rarity of the event. Only four coun-
tion in FDI receiver-states. The Chinese language case is tries share a common civilization with China: North and
unique in another sense: Compared to English or South Korea, the Philippines, and Vietnam (Russett et al.
French, there is a dearth of Mandarin Chinese teachers. 2000). This relationship is already captured by distance
While the global Chinese population is large, the majority and contiguity measures. Table 4 presents summary statis-
of native Chinese speakers with proficiency in Mandarin tics of all the variables.
reside predominantly in China (inclusive of Hong Kong,
Macau, and Taiwan). Moreover, the Chinese diaspora Results and Analysis
population is dialectically heterogeneous, and not all We estimate the model using GLS with standard errors
individuals are fluent in Mandarin. This is most evident clustered by receiver-state. The results are presented in
among immigrant populations from Hong Kong (Canton- Table 5. Since we expect the effects of teaching Manda-
ese), Taiwan (Taiwanese), and Southeast Asia (Canton- rin Chinese (Chinese Language) may take time to take
ese, Hokkien, etc.). Put differently, a Chinese heritage place, we run a number of models with a different lag
does not automatically translate into the ability to speak specification, starting with no lag (Model 4) and ending
or teach Mandarin Chinese. This suggests that the global with a 5-year lag (Model 9). Across all six models, the
supply of native Mandarin Chinese teachers is quite lim- coefficient for Chinese Language is in the expected direc-
ited. In contrast, a country that needs native English or tion: positive. With one exception, the coefficient is
French teachers has many potential countries from which always statistically significant. Moreover, as shown in
to recruit these individuals. This is not to say that China Figure 1, the effects of teaching Chinese have the most
is the only country that produces Chinese language teach- sizable impact in the long term. For instance, the pres-
ers, but it is most certainly the largest and primary sup- ence of a CI 5 years before can increase present-day Chi-
plier of these teachers. nese investment levels by a shocking ninefold (from USD
Recognizing this situation, the Chinese Ministry of 23.18 million to USD 213 million). These results strongly
Education has overseen the development of the Chinese corroborate those from the cross-sectional OECD test:
National Office for Teaching Chinese as a Foreign Lan- When some portion of the population can speak the FDI-
guage (herein referred to as Hanban). Hanban has sending states language, this reduces the need for costly
established a number of Confucius Institutes (CI) inter- translation services. All else being equal, FDI-sending
nationally to promote Mandarin Chinese and help for- states prefer minimal transaction costs.
eigners to learn the language. Using the Hanban These results indicate a strong and robust relationship
database, we can determine whether there is a CI in a between countries with CIs and Chinese FDI. There are,
receiver-state, and if so, we can determine the year the however, three alternative explanations that we need to
first institute was established. We are interested in address. The first concerns the possibility of a spurious
whether a CI exists and not the number of CIs. Although relationship: CIs are established in countries where China
important, the number of CIs alone cannot capture their wants to invest. While China may strategically target coun-
heterogeneous influence on language learning. For tries for future FDI, this cannot theoretically explain why
example, Hungary has three CIs. The largest, the one in
Budapest, employs thirty native Mandarin Chinese teach-
ers, and some of these instructors teach fulltime in other TABLE 4. Summary Statistics, Test 2
cities. This is not the case for the other Hungarian CIs in Mean SD Min/Max
Miskolc and Szeged. Additionally, the Budapest CI is
poised to become the central institute for all CI activities (logged) FDI flow 16.3 2.18 9.14/22.2
in Central and Eastern Europe. Given these measurement Chinese Second Language (t) 0.426 0.495 0/1
validity concerns, we operationalize Chinese Language as Chinese Second Language (t 1) 0.323 0.468 0/1
dichotomous. The variable is coded as one if the receiver- Chinese Second Language (t 2) 0.266 0.442 0/1
state has a CI in year t. Chinese Second Language (t 3) 0.199 0.400 0/1
With a few exceptions, the list of control variables Chinese Second Language (t 4) 0.112 0.316 0/1
remains the same as that in our previous analysis. Distance Chinese Second Language (t 5) 0.0565 0.232 0/1
is measured as the distance between Beijing and the (logged) Distance 8.97 0.533 7.07/9.87
receiver-states capital. GDP for the receiver-state is Contiguous Borders 0.0717 0.258 0/1
reported in constant USD. As with the OECD sample, (logged) GDP: Receiver 24.4 2.16 19.9/30.1
both Distance and GDP have been logged. The next five Common Legal Heritage 0.175 0.381 0/1
Common WTO Membership 0.871 0.336 0/1
variables are all dichotomous: Contiguous Borders (Depart-
Dual Tax Treaty 0.612 0.488 0/1
ment of State), Common Legal Heritage (La Porta et al.
Shared PTA 0.137 0.345 0/1
1999), Common WTO Membership (World Trade Organiza-
Receiver-State Democracy 3.71 6.20 10/10
tion), Dual Tax Treaty (Lane and Milesi-Ferretti 2004), N 502
and Shared PTA (World Trade Organization). Democracy,
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
 338 Lingua Mercatoria

TABLE 5. Chinese Language and Chinese Foreign Direct Investment

Model 5 Model 6 Model 7 Model 8 Model 9

Model 4 (No Lag) (1-Year Lag) (2-Year Lag) (3-Year Lag) (4-Year Lag) (5-Year Lag)

Chinese Language 0.423 (0.132)*** 0.393 (0.144)*** 0.401 (0.207)* 0.304 (0.185)* 0.426 (0.295) 1.075 (0.631)*
(logged) Distance 0.093 (0.231) 0.073 (0.224) 0.026 (0.256) 0.170 (0.262) 0.199 (0.259) 0.282 (0.294)
Contiguous 0.898 (0.359)** 0.892 (0.353)** 0.623 (0.401) 0.660 (0.389) 0.506 (0.448) 0.748 (0.769)
Borders
(logged) GDP 0.115 (0.054)** 0.122 (0.053)** 0.143 (0.055)*** 0.176 (0.054)*** 0.162 (0.060)*** 0.259 (0.080)***
Common Legal 0.214 (0.286) 0.186 (0.282) 0.076 (0.301) 0.170 (0.326) 0.264 (0.339) 0.668 (0.466)
Heritage
Common WTO 0.209 (0.252) 0.195 (0.248) 0.047 (0.265) 0.206 (0.268) 0.153 (0.279) 0.175 (0.351)
Membership
Dual Tax 0.012 (0.271) 0.009 (0.268) 0.107 (0.275) 0.234 (0.286) 0.245 (0.300) 0.107 (0.388)
Treaty
Shared PTA 0.227 (0.239) 0.247 (0.243) 0.390 (0.252) 0.234 (0.286) 0.523 (0.270)* 0.608 (0.304)**
Democracy 0.019 (0.014) 0.019 (0.014) 0.025 (0.015)* 0.015 (0.016) 0.016 (0.016) 0.017 (0.023)
Dependent 0.546 (0.038)*** 0.545 (0.038)*** 0.525 (0.044)*** 0.527 (0.050)*** 0.549 (0.054)*** 0.454 (0.082)***
Variable
(lagged)
Constant 3.998 (2.462) 4.057 (2.425) 4.833 (2.645)* 5.587 (2.742)** 5.822 (2.962)** 5.835 (3.033)*
N 502 502 421 341 259 177
(Note. Robust standard errors reported and clustered by receiver-states. *p < .100, **p < .050, ***p < .010.)

FIG 1. Effects of Teaching Chinese on Chinese Foreign Direct Investment

it would initially develop CIs if language were not an
important consideration.
The second alternative explanation involves the effects
of signaling. In the OECD analysis, the foreign/second
language education coefficient was significant in both the
1- and 20-year lagged models. The magnitude of the lat-
ter model suggested that the effects of coordinating lan-
guage-in-education policies with the sender-states official
language are more pronounced in the long run. How-
ever, it is possible that potential receiver-states use their
foreign/second language education to signal favorable
investment conditions. The results from the China model
indicate that this mechanism may be at work. Since no
other country (except Singapore) uses Chinese as an offi-
cial language, the choice to teach Chinese sends a clear
signal to Chinese firms that a potential receiver-state
wishes to increase its affinity to them. To take a hypothet-
ical example, consider Hungary passing legislation to
teach Koreana language spoken in only the Koreas.
Hungary would be sending a definitive signal that it seeks
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
greater interactions with Korean investors. Yet, if Hungary
adopted similar legislation to teach English, this signal
would be noisier, as English is the official language of
multiple countries. Consequently, this mechanism seems
more relevant to China and other countries (for exam-
ple, Korea) whose official languages are not globally
common.
To examine which of these two mechanismssignaling
versus transaction costshas a greater impact on FDI lev-
els, we run a model that includes both the contempora-
neous and the 5-year lag of Chinese Language. The results
in Model 10 (Table 6) show that when controlling for
the two periods simultaneously, the coefficient for the
contemporaneous coefficient is in the negative direction
and loses significance. In contrast, the coefficient for the
5-year lag remains positive (b = 1.083) and statistically sig-
nificant. These results suggest that signals to Chinese for-
eign investors are important, but their effects are muted
when we take the long-term benefits of reduced transac-
tion costs into consideration. This finding supports our
 Moonhawk Kim et al. 339

TABLE 6. Assessing Alternative Explanations

Model 10 (Signaling Model 11 (Instrument: Model 12 Model 13 (Number of

versus Transaction Costs) Lagged Chinese) (Number of Institutes) Years Matching)

Confucius Institute 0.007 (0.287) 0.353 (0.183)*

Confucius (t 5) 1.083 (0.640)*
(logged) # Confucius (t 1) 0.430 (0.093)***
# Years Confucius (t 1) 0.197 (0.054)***
(logged) Distance 0.285 (0.293) 0.086 (0.202) 0.174 (0.245) 0.094 (0.247)
Contiguous Borders 0.755 (0.774) 0.792 (0.316)** 0.954 (0.363)*** 0.827 (0.375)**
(logged) GDP 0.261 (0.080)*** 0.101 (0.049)** 0.091 (0.054)* 0.114 (0.054)**
Common Legal Heritage 0.671 (0.478) 0.201 (0.263) 0.187 (0.292) 0.194 (0.295)
Common WTO Membership 0.172 (0.352) 0.221 (0.222) 0.196 (0.259) 0.181 (0.261)
Dual Tax Treaty 0.106 (0.393) 0.016 (0.239) 0.067 (0.285) 0.017 (0.281)
Shared PTA 0.611 (0.307)** 0.230 (0.211) 0.293 (0.256) 0.265 (0.257)
Democracy 0.018 (0.023) 0.014 (0.014) 0.024 (0.015) 0.021 (0.015)
Dependent Variable (lagged) 0.450 (0.082)*** 0.606 (0.037)*** 0.525 (0.041)*** 0.527 (0.041)***
Constant 5.874 (3.060)* 3.539 (2.233) 4.158 (2.490)* 4.327 (2.561)*
N 177 502 502 502
Wald v2 182.63*** 57.73*** 566.87*** 451.73***
R2 0.5139 0.5114 0.5134 0.5107
(Note. Robust standard errors reported and clustered by receiver-states. *p < .100, **p < .050, ***p < .010.)

claim that receiver-states canand docoordinate lan-
guage-in-education policies to attract FDI.
The third alternative explanation that merits attention
is endogeneity. Countries may manipulate their language-
in-education policies in response to existing FDI. To
address this issue, we employ an instrumental variable
regression estimator. Specifically, we use the lagged Chi-
nese Language variable as an instrument. While the corre-
lation between current and lagged Chinese Language is
high, the results from a simple regression including both
variables suggest that the lagged value is not a significant
determinant of FDI flows (p = .64). This confirms the suf-
ficiency of lagged Chinese Language to meet the exclusion
criterion. As shown in Model 11, using an instrumental
variable allows the results for Chinese Language to remain
consistent with those in Table 5. Taking into consider-
ation any possible endogeneity, the presence of a CI has
a significant and positive effect on FDI levels.
Admittedly, the use of a dichotomous measure ignores
the variation across CIs. As an alternative, we use two dif-
ferent count measures. The first measure, consistent with
Lien et al. (2012), is the logged number of CIs (plus
one) in country i in year t. We log this number due to its
non-normal distribution: The United States in 2010 has
more than 60 CIs. By contrast, the rest of the world has a
median of one CI and a mean of two in the same year.
The second count measure examines the number of years
that a country has had a CI. This measure not only pro-
vides another test of the overtime transaction-cost-reduc-
tion mechanism, but it also allows us to disaggregate the
effects of older CIs from those of the younger institutes.
The results in Models 12 and 13 confirm the overall
claims of this article: FDI levels increase when languages
match, and FDI increases as the proportion of population
proficient in the matched language increases.3
Test 3: Chinese FDI in Indonesia
We now process-trace the ChinaIndonesia dyad from the
previous section and examine the link between teaching
Mandarin Chinese and Chinese FDI. We chose Indonesia
3
The results are robust to different lags and a raw count measure.
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
for two reasons. First, Indonesia is an example of what
Seawright and Gerring (2008:301302) would call an
extreme case: Chinese FDI levels in Indonesia are high.
From a relative standpoint, Indonesia has generally
attracted more than one standard deviation of the aver-
age global Chinese FDI. From an absolute standpoint,
Chinese FDI in Indonesia has steadily increased in recent
years (2010 Statistical Bulletin of Chinas Outward Foreign
Direct Investment). Figure 2 shows both trends. While the
economic sectors involved have been diverse, some of the
largest investments have been in energy. Chinese compa-
nies are heavily involved with Indonesias energy develop-
ment program, which is currently in its second iteration.
In the first program, nine out of the ten projects were
Chinese-managed (Xinhua 2011).
Second, Indonesia is a deviant case (Seawright and
Gerring 2008:302303). We should observe low levels of
Chinese FDI, and yet, we observe substantial investments.
For instance, despite both countries location in East
Asia, the geographical distance between Beijing and
Jakarta is approximately the same as Beijing and Tehran.
Augmenting this distance is the absence of a shared bor-
der, a lack of common legal heritage, and incongruence
in regime type. In addition to not sharing a common civi-
lization, IndonesianChinese ethnic relations have also
been hostile for a long time.
During Suhartos tenure in Indonesia (19651998),
ChinaIndonesia relations were antagonistic. Domesti-
cally, the Indonesian government repressed the Chinese
minority. In 1965, following a failed coup allegedly mas-
terminded by the Indonesian Communist Party (PKI),
the Suharto-led military pursued an anti-communist
purge. Since many Chinese in Indonesia were supposedly
PKI members, the purge took on an ethnic dimension.
The exact numbers vary, but as many as 1.5 million were
killed in this purge. When Suharto assumed legal power
in 1968, his fear of the possibility of [communist] sub-
version and infiltration characterized many of his domes-
tic policies (Soeharto 1991:230). His government created
specific intelligence security bodies to monitor the Chi-
nese and any possible communist activities.
Additionally, the Indonesian government tried to elimi-
nate any semblance of a distinct Chinese culture. The
 340
FIG 2. Chinese Foreign Direct Investment in Indonesia
government required Chinese surnames to be replaced
with Indonesian names, curtailed Chinese New Year cele-
brations, and prohibited public displays of the Chinese
language. Most importantly for our analysis, the Indone-
sian government shut down Chinese schools and banned
the teaching of the Chinese language. In the wake of the
19971998 East Asian financial crisis, the Chinese com-
munity was targeted once again. There were reports of
arson against Chinese-owned businesses and sexual
attacks on Chinese women. With escalating violence, eth-
nic Chinese fled Indonesia en masse with their assets,
which were in excess of USD 20 billion (Napier 1999).
Internationally, ChinaIndonesia relations were just as
cold. Although Indonesia was technically non-aligned
during the Cold War, the general global perception was
that Indonesia was in the Western camp. Moreover, many
Indonesians believed that China had supported the PKI
in the 1960s (Soeharto 1991). Given Indonesias historic
hostility toward its own Chinese minority and toward
China itself, the choice of Chinese firms to invest in Indo-
nesia is very puzzling. We assume that a country lacking
many of the positive attributes for Chinese investment
such as geographical proximity, border contiguity, and
historical/cultural affinitywould have very low levels
of FDI from China, yet Chinese FDI is quite high in
Indonesia.
What explains Chinese FDI levels in Indonesia? The
answer, we argue, is newer language-in-education policies,
specifically the teaching of Mandarin Chinese. Since Su-
hartos removal from office (1998), the Indonesian gov-
ernment has re-introduced Chinese into its educational
curricula. Currently, Chinese is an optional class taught
two or three times a week at the kindergarten-grade 12
(K-12) levels. Similar language instruction exists at the
university level. Today, one out of every five universities,
including Indonesia University in Depok and Gadjah
Mada University in Yogyakarta, offers Chinese language
and culturecompared to only 5% one decade ago (Tsai
2010). At Batam International University in Riau Prov-
ince, proficiency in Chinese or English is strongly encour-
aged. And at Sin Tjong University, Chinese, English, and
Indonesian are all mediums of instruction (Handoko
2008). Outside formal educational institutions, many pri-
vate language centers offer Chinese and English (Hand-
oko 2008). The first CI was established in Jakarta in 2007.
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
Lingua Mercatoria
Since then, six other CIs have opened throughout the
country (Confucius Institute 2011).
For the ethnically ChineseIndonesians, learning Man-
darin Chinese is not simply a cultural attempt to redis-
cover or salvage their Chineseness (Handoko 2008:15).
There is also an economic component that makes learn-
ing Chinese akin to learning English. For instance, Chi-
nese is growing in importance in terms of business
activities in Indonesia with more and more companies
listing ability to speak Mandarin Chinese as one of their
requirements for employment (Handoko 2008:14). Com-
panies like to recruit Chinese majors (Tsai 2010). Simi-
larly, in a rice-growing region of Java, Chinese is an
obligatory subject because Indonesian authorities now
realize that there is huge economic advantage in reaching
and learning from its economic compatriots. . . [W]ithout
it and a global reach [Indonesian is] not going to make
it (Money Morning 2012).
There is a clear link between teaching Chinese and
Chinese FDI in Indonesia. But are the language-in-educa-
tion policies preceding or following the investments?
While the two go hand-in-hand, evidence supports the
former theory. In 2000, when Indonesias aggregate FDI
inflow was negative (Thee 2006), President Abdurrahman
Wahid reversed many of his predecessors anti-Chinese
policies, including the no-Chinese-in-the-classroom ban.
Since 2005, Chinese FDI in Indonesia has steadily
increased from USD 11.8 million to 205 million (see Fig-
ure 2). The effect of the first CI (2007) is even more dra-
matic. Prior to 2007, Chinese FDI levels were largely
within one standard deviation of the global average, but
since 2007, investment levels have skyrocketed. This sug-
gests that language policies have been instrumentaland
successfulin attracting Chinese FDI. Consequently, in
2012, Indonesian President Susilo Bambang Yudhoyono
(SBY) and Chinese President Hu Jintao met in Beijing to
discuss increasing bilateral partnership. They negotiated
improving this partnership by expanding cultural and
people-to-people exchanges by increasing cooperation in
the area of education. . . (Xinhua 2012). In turn, SBY sta-
ted that Indonesia would welcome Chinese companies to
increase investment in the country and participate in the
major infrastructure projects (Xinhua 2012). Cai Jinch-
eng, director of Bahasa Indonesia and Malay Language
Study Program at Guangdong University, echoed this
 Moonhawk Kim et al. 341

position, arguing that by mastering the languages of

25
each country [China and Indonesia] through cultural

(logged) FDI in constant USD

exchange program, people in both countries would take
the benefit from smooth communications aimed at boost-
ing up the economic ties (Xinhua 2010).

20
Language-in-education policies are clearly important
and useful instruments to attract FDI. When a potential
receiver-state teaches the sender-states language, this
reduces transaction costs, thereby making the former

15
more attractive for investments. However, how does the
importance of language compared to alternative explana-
tions? In the large-N analysis of Chinese FDI, only one
control variable was statistically significant across all
model specifications: GDP of the receiver-state. The size

10
of Indonesias economymuch of it due to natural -1 0 1
resourceshas been a large factor in FDI inflows. It is, Conflict Non-Matching Matching
after all, the fifteenth largest economy in the world today

25
(Central Intelligence Agency 2012). Note, however, that

(logged) FDI in constant USD

the substantive effect of receiver-state GDP is smaller than
that of the foreign/second language variable in the large-
N analysis. In other words, Chinese Language alone

20
accounts for 0.10 of the variance; in contrast, logged GDP
accounts for 0.09. Second, a model that focuses only on
GDP explains less variance than one that includes Chi-
nese language learning. In short, our objective is not to

15
trivialize the effects of GDP, but to demonstrate that
teaching Chinese has had a positive effect on Chinese
FDI flow in Indonesia independent of its economic size.
We also consider the effects of ethnic politics. The
10

social and political climate in Indonesia under Suharto 20 22 24 26 28 30

was hostile to Chinese investment. As discussed above,
the Chinese minority in Indonesia faced political repres-
sion, and ChinaIndonesia relations were tepid at best.
Poor cultural relations and a repressed diaspora network
should have deterred investments. Given these conditions,
specific policies that improved Indonesias climate for
Chinese investmentparticularly teaching the Chinese
languagehave been important. Figure 3(a) provides
the evidence. In spite of the historical and cultural
clashes, Indonesia has attracted Chinese FDI levels far
beyond theoretical expectations. And while Indonesias
GDP is sizeable, there are simply too many residual
effects that cannot be explained by GDP alone (as
illustrated in Figure 3b). Here, we argue that Chinese
language instructioncaptured by the presence of
CIshas played an important role in reducing long-
term transaction costs.
Conclusion
The Trinidad government passed a bill in 2005 that made
Spanish a mandatory class in schools and basic Spanish
proficiency a requirement for all civil servants. Economic
factors drove this choice to elevate Spanish. There were
two major considerations. First, by adopt[ing] the lan-
guage of Venezuela, the hemispheres biggest oil pro-
ducer, Trinidad hope[d it would be able] to strengthen
its own oil and natural gas industries (Davies 2005). Sec-
ond, once Trinidad secured a seat on the Free Trade
Area of the Americas, having a workforce linguistically
competent in Spanish would yield sizable economic ben-
efits (Davies 2005). This development in Trinidad is one
of many examples of governments using language to
attract FDI. Specifically, when governments coordinate
their language-in-education policies to match the official
language of a sender-state, this can reduce economic
transaction costs, thereby encouraging FDI.
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
(logged) GDP in constant USD
FIG 3. (a) Effects of Civilization Clash on (logged) Foreign Direct
Investment. (b) Effects of (logged) GDP on (logged) Foreign Direct
Investment
In addition to explaining the behavior of governments
like Trinidads, this article makes a twofold contribution.
Theoretically, we argue that language constitutes a
dynamic instrument rather than a static policy. A lan-
guage need not be official to reduce transaction costs.
What matters instead is that a states future workforce
learns this language in school and becomes proficient in
it to some degree. In this manner, the predominant focus
on countries official languages in the literature has been
misplaced. Empirically, we introduce a new variable that
improves the widely employed yet grossly static CEPII
measurement. Analyses of international economic flows
will likely find more significant effects of languages when
they use our language-in-education measure rather than
the official language measure that does not vary over
time.
The findings in this article also provide an impetus for
future research. The current analysis can benefit from
extending the coverage of countries. While our analysis
was limited by the availability of bilateral FDI data, such
data are becoming available. Greater availability of south
south FDI data will enable a wider analysis of the role
that BRIC languages play in the international economy.
Another avenue might consider the reverse mechanism
of the one we advance in this article. Rather than using
language-in-education policies to attract FDI, do govern-
ments also use similar policies to retain extant FDI? Put
differently, do the linguistic origins of FDI already in the
country shape language-in-education policies? As
addressed above, the possibility of this reverse mechanism
 342 Lingua Mercatoria

does not pose an endogeneity challenge to our argument De Mooij, Ruud A., and Sjef Ederveen. (2001) Taxation and Foreign Direct
and analysis. Moreover, unlike our analysis, analyzing this Investment. CESifo Working Paper No. 588. Munich: CESifo Group.
mechanism will require data on FDI stocks rather than Dixon, L. Quentin. (2005) Bilingual Education Policy in Singapore: An
Analysis of its Sociohistorical Roots and Current Academic
flows. This question is an important one to address on its
Outcomes. International Journal of Bilingual Education and
own right, especially given the possibility of foreign influ- Bilingualism 8 (1): 2547.
ences on domestic policies. Dunning, John. (1977) Trade, Location of Economic Activity and the
Lastly, the implications of this research extend beyond MNE: A Search for an Eclectic Approach. In The International
analyses of international economic exchanges. The role of Allocation of Economic Activity, edited by Bertil Ohlin, Per-Ove
languages and language policies is an understudied phe- Hesselborn, and Per Magnus Wijkman. London: Macmillan.
nomenon in international relations. Proficiency in a Elkins, Zachary, Andrew T. Guzman, and Beth A. Simmons. (2006)
shared language is a basic necessity for political actors Competing for Capital: The Diffusion of Bilateral Investment
across countries to interact smoothly, even in the presence Treaties, 1960-2000. International Organization 60 (4): 811846.
Esman, Milton J. (1987) Ethnic Politics and Economic Power.
of interpreters. For central decision makers, shared lan-
Comparative Politics 19 (4): 395418.
guage proficiency can shape international diplomacy. For Froot, Kenneth A., and Jeremy C. Stein. (1991) Exchange Rates and
bureaucrats, it can affect international bargaining and Foreign Direct Investment. The Quarterly Journal of Economics 106
negotiations. And for the broader public, it can influence (4): 11911217.
peoples level of understanding and perception of other 
Ganguly, Sumit. (2003) The Politics of Language Policies in Malaysia
countries. Further research into these phenomena will and Singapore. In Fighting Words: Language Policy and Ethnic
enhance our understanding of international relations as Relations in Asia, edited by Michael E. Brown and Sumit  Ganguly.
well as of the role that language and language policies play. Cambridge, MA: MIT Press.
Globerman, Steven, and Daniel Shapiro. (2003) Governance
Infrastructure and U.S. Foreign Direct Investment. Journal of
References International Business Studies 34 (1): 1939.
Goldstein, Judith L., Douglas Rivers, and Michael Tomz. (2007)
Adam, Shamim, and Andrea Tan. (2011) Singapore Raises 2011 Inflation
Institutions in International Relations: Understanding the Effects of
Forecast on Record Expansion. Bloomberg, February 17. Available at
the GATT and the WTO on World Trade. International Organization
http://www.bloomberg.com/news/2011-02-17/singapore-raises-
61 (1): 3767.
2011-inflation-forecast-to-3-4-after-record-expansion.html. (Accessed
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. (2005) Cultural
September 3, 2014.)
Biases in Economic Exchange. NBER Working Paper 11005.
Addison, Tony, and Almas Heshmati. (2004) The New Global
Cambridge, MA: NBER.
Determinants of FDI Flows to Developing Countries. In Monetary
Habib, Mohsin, and Leon Zurawicki. (2002) Corruption and Foreign
Integration, Markets, and Regulation, edited by Michele Bagella,
Direct Investment. Journal of International Business Studies 33 (2):
Leonardo Becchetti, Iftekhar Hasan, and William Curt Hunter.
291307.
Amsterdam: Elsevier.
Handoko, Francisca. (2008) Mandarin in the educational systems in the
Albaugh, Ericka A. (2009) The Colonial Image Reversed. International
Indonesian Archipelago. Paper presented at the Language Change
Studies Quarterly 53 (2): 389420.
in Post New Order Indonesia Workshop, Depok, Indonesia.
Anderson, James E., and Eric Van Wincoop. (2003) Gravity with
Hines, James R. (1995) Forbidden Payment. NBER Working Paper 5266.
Gravitas: A Solution to the Border Puzzle. American Economic Review
Cambridge, MA: NBER.
93 (1): 170192.
Hsiao, Frank S. T., and Mei-Chu W. Hsiao. (2004) The Chaotic
Bandelj, Nina. (2002) Embedded Economies: Social Relations as
Attractor of Foreign Direct InvestmentWhy China? A Panel Data
Determinants of Foreign Direct Investment in Central and Eastern
Analysis. Journal of Asian Economics 15 (4): 641670.
Europe. Social Forces 81 (2): 411444.
Inquirer. (2010) Spain, Philippines Sign Agreement on Spanish
B

enassy-Qu
e, Agnes, Maylis Coupet, and Thierry Mayer. (2007)
er

Language. February 24. Available at http://globalnation.inquirer.
Institutional Determinants of Foreign Direct Investment. The World
net/news/breakingnews/view/20100224-255038/Spain-Philippines-
Economy 30 (5): 764782.
sign-agreement-on-Spanish-language. (Accessed September 3, 2014.)
Bevan, Alan A., and Saul Estrin. (2004) The Determinants of Foreign
Jensen, Nathan M. (2003) Democratic Governance and Multinational
Direct Investment into European Transition Economies. Journal of
Corporations: Political Regimes and Inflows of Foreign Direct
Comparative Economics 32 (4): 775787.
Investment. International Organization 57 (3): 587616.
Blomstrom, Magnus, and Ari Kokko. (2003) The Economics of Foreign
Jensen, Nathan M. (2008) Political Risk, Democratic Institutions, and
Direct Investment Incentives. NBER Working Paper 9489. NBER:
Foreign Direct Investment. Journal of Politics 70 (4): 10401052.
Cambridge, MA.
Kaplan, Robert B., and Richard B. Baldauf. (1997) Language
Blonigen, Bruce A., and David N. Figlio. (1998) Voting for
Planning from Practice to Theory. Clevedon, UK: Multilingual Matters
Protection: Does Direct Foreign Investment Influence Legislator
Limited.
Behavior? The American Economic Review 88 (4): 10021014.
Klein, Michael W., and Jay C. Shambaugh. (2008) The Dynamics of
uthe, Tim, and Helen V. Milner. (2008) The Politics of Foreign
B
Exchange Rate Regimes: Fixes, Floats, and Flips. Journal of
Direct Investment into Developing Countries: Increasing FDI
International Economics 75 (1): 7092.
through International Trade Agreements? American Journal of
Kogut, Bruce, and Harbir Singh. (1988) The Effect of National
Political Science 52 (4): 741762.
Culture on the Choice of Entry Mode. Journal of International
Butler, Yuko Goto. (2004) What Level of English Proficiency Do
Business Studies 19 (3): 411432.
Elementary School Teachers Need to Attain to Teach EFL? Case
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and
Studies from Korea, Taiwan, and Japan. TESOL Quarterly 38 (2):
Robert W. Vishny. (1999) The Quality of Government. Journal of
245278.
Law, Economics, and Organization 15 (1): 222279.
Central Intelligence Agency. (2012) World Factbook. Available
Lane, Philip R., and Gian Maria Milesi-Ferretti. (2004) International
at https://www.cia.gov/library/publications/the-world-factbook/.
Investment Patterns. Working Paper 06/290. Washington, DC:
(Accessed February 7, 2012.)
International Monetary Fund.
Confucius Institute Headquarters: Hanban. (2011). Available at
Leblang, David. (2010) Familiarity Breeds Investment: Diaspora
https://www.hanban.org. (Accessed September 22, 2011.)
Networks and International Investment. American Political Science
Davies, Elizabeth. (2005) Hola! Trinidad Drops English and Learns to
Review 104 (3): 584600.
Speak Spanish. The Independent, September 1. Available at http://
Leclerc, Jacques. (2011) Lam
enagement linguistique dans le monde.
www.independent.co.uk/news/world/americas/hola-trinidad-drops-
Available at http://www.tlfq.ulaval.ca/axl/. (Accessed September 3,
english-and-learns-to-speak-spanish-505008.html. (Accessed September
2014.)
3, 2014.)

Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct

by guest
on 09 October 2017
 Moonhawk Kim et al.
Lee, Kuan Yew. (2000) From Third World to First: The Singapore Story 1965
2000. Singapore: Times Media Private Limited.
Lee, Hock Guan. (2007) Ethnic Politics, National Development and
Language Policy in Malaysia. In Language, Nation and Development in
Southeast Asia. Edited by Hock Guan Lee, and Leo Suryadinata.
Singapore: Institute of Southeast Asian Studies.
Leeds, Brett Ashley, Jeffrey M. Ritter, Sara Mc Laughlin
Mitchell, and Andrew G. Long. (2002) Alliance Treaty
Obligations and Provisions, 18151944. International Interactions 28
(3): 237260.
Li, Quan, and Adam Resnick. (2003) Reversal of Fortunes: Democratic
Institutions and Foreign Direct Investment Inflows to Developing
Countries. International Organization 57 (1): 175211.
Lien, Donald, Chang Hoon Oh, and W. Travis Selmier. (2012)
Confucius Institute Effects on Chinas Trade and FDI: Isnt it
Delightful When Folks Afar Study Hanyu? International Review of
Economics and Finance 21 (1): 147155.
Linnemann, Hans. (1966) An Econometric Study of International Trade
Flows. Amsterdam: North-Holland.
Liu, Amy H. (2011) The Linguistic Effects of Political Institutions.
Journal of Politics 73 (1): 125139.
Mahoney, James, and Gary Goertz. (2004) The Possibility Principle:
Choosing Negative Cases in Comparative Research. American
Political Science Review 98 (4): 653669.
Makki, Shiva S., and Agapi Somwaru. (2004) Impact of Foreign Direct
Investment and Trade on Economic Growth: Evidence from
Developing Countries. American Journal of Agricultural Economics 86
(3): 795801.
Marschan-Piekkari, Rebecca, Denice Welch, and Lawrence Welch.
(1999) In the Shadow: The Impact of Language on Structure,
Power, and Communication in the Multinational. International
Business Review 8 (4): 421440.
Marshall, Monty G., and Keith Jaggers. (2008) POLITY IV Project.
Available at http://www.systemicpeace.org/polity/polity4.htm.
(Accessed February 2, 2012.)
Money Morning. (2012) Investing in the New China with this
Telecom Market Stock. April 23. Available at http://
moneymorning.com/2012/04/23/investing-in-the-new-china-with-this-
telecom-market-stock. (Accessed September 3, 2014.)
Napier, Catherine. (1999) Wahid Aims to Bring Back the Billions. BBC
News, November 9. Available at http://news.bbc.co.uk/2/hi/asia-
pacific/512754.stm. (Accessed September 3, 2014.)
OECD. (2005) Annual Report. Available at http://www.oecd.org/
dataoecd/34/6/34711139.pdf. (Accessed September 3, 2014.)
Philippines Department of Trade and Industry. (2010) Spanish
Firms Keen on Investment Opportunities in Manila. April 22.
Philippines Department of Trade and Industry Website. Available at
http://www.dti.gov.ph/dti/index.php?p=154&type=2&sec=2&aid=347.
(Accessed October 25, 2010.)
Russett, Bruce, John R. Oneal, and Michaelene Cox. (2000) Clash
of Civilizations, or Realism and Liberalism D
eja Vu? Some
Evidence. Journal of Peace Research 37 (5): 583608.
Downloaded from https://academic.oup.com/isq/article-abstract/59/2/330/1791838/Lingua-Mercatoria-Language-and-Foreign-Direct
by guest
on 09 October 2017
343
Seawright, Jason, and John Gerring. (2008) Case Selection
Techniques in Case Study Research: A Menu of Qualitative and
Quantitative Options. Political Research Quarterly 61 (2): 294308.
Siegel, Jordan, Amir Licht, and Shalom H. Schwartz. (2013)
Egalitarianism, Cultural Distance, and Foreign Direct Investment: A
New Approach. Organization Science 24 (4): 11741194.
Soeharto. (1991) My Thoughts, Words and Deeds. Trans. Sumadi. Jakarta:
Citra Lamtoro Gung Persada.
Tan, Eugene K. B. (2007) The Multilingual State in Search of the
Nation: The Language Policy and Discourse in Singapores Nation-
Building. In Language, Nation and Development in Southeast Asia,
edited by Hock Guan Lee, and Leo Suryadinata. Singapore:
Institute of Southeast Asian Studies.
Thee, Kian Wie. (2006) Policies for Private Sector Development in Indonesia.
ADB Institute Discussion Paper 46. Tokyo: ADB Institute.
Tsai, Michelle. (2010) Indonesia Learns Chinese. Foreign Policy, May 20.
Available at http://www.foreignpolicy.com/articles/2010/05/20/
indonesia_learns_chinese. (Accessed September 3, 2014.)
Tupas, T. Ruanni F. (2007) Go Back to Class: The Medium of
Instruction Debate in the Philippines. In Language, Nation and
Development in Southeast Asia, edited by Hock Guan Lee, and Leo
Suryadinata. Singapore: Institute of Southeast Asian Studies.
Wang, Gungwu. (2007) Keynote Address. In Language, Nation and
Development in Southeast Asia, edited by Hock Guan Lee, and Leo
Suryadinata. Singapore: Institute of Southeast Asian Studies.
White, Roger, and Bedassa Tadesse. (2008) Cultural Distance and the
US Immigrant-Trade Link. The World Economy 31 (8): 10781096.
Williamson, Oliver E. (1975) Markets and Hierarchies: Analysis and
Antitrust Implications. New York: The Free Press.
World Trade Organization. (2013) Regional Trade Agreements
Information System (RTA-IS). Available at http://rtais.wto.or.
(Accessed February 7, 2012.)
Xinhua. (2010) Cultural Exchange Essential to Nurture Indonesia-China
Relationship: Chinese Scholar. September 30. Available at http://
news.xinhuanet.com/english2010/china/2010-09/30/c_13537885.
htm. (Accessed September 3, 2014.)
Xinhua. (2011) Chinese Firms Power on with Indonesia Plan. October
18. Available at http://news.xinhuanet.com/english2010/china/
2011-10/18/c_131197899.htm. (Accessed September 3, 2014.)
Xinhua. (2012) Chinese, Indonesian Presidents Eye Stronger Links.
March 23. Available at http://news.xinhuanet.com/english/china/
2012-03/23/c_131485660.htm. (Accessed September 3, 2014.)
Yeaple, Stephen Ross. (2003) The Role of Skill Endowments in the
Structure of U.S. Outward Foreign Direct Investment. Review of
Economics and Statistics 85 (3): 726734.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Table S1. Colonial Legacy and FDI.