UC San Diego
UC San Diego Electronic Theses and Dissertations
Title
Capital and chaos : fragile states, political risk and foreign direct investment
Permalink
https://escholarship.org/uc/item/8112z3dm
Author
Graham, Benjamin A. T.
Publication Date
2012
Peer reviewed|Thesis/dissertation
eScholarship.org
Powered by the California Digital Library
University of California
UNIVERSITY OF CALIFORNIA, SAN DIEGO
Capital and Chaos:
Fragile States, Political Risk and Foreign Direct Investment
A dissertation submitted in partial satisfaction of the requirements for the degree of
Doctor of Philosophy
in
Political Science
by
Benjamin A.T. Graham
Committee in Charge:
Professor Miles Kahler, Chair
Professor Lawrence Broz
Professor Edmund Malesky
Professor Craig McIntosh
Professor Kaare Strom
2012
!
Copyright
Benjamin A.T. Graham, 2012
All rights reserved.
!
!
The Dissertation of Benjamin A.T. Graham is approved, and it is acceptable in quality
and form for publication on microfilm and electronically:
Chair
University of California, San Diego
2012
iii!
!
!
!
!
For Lynn.
iv!
!
!
!
!
TABLE OF CONTENTS
Signature Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iii
Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iv
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
v
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vi
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vii
List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
x
Vita . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xvii
Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
19
Chapter 2: Capital and Chaos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
67
Chapter 3: Political Risk and Diaspora Direct Investment . . . . . . . . . . . . . . . . . . . 71
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Chapter 4: Diaspora-Owned Firms and the Value of Social Networks . . . . . . . . . 118
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Chapter 5: Diaspora-Owned Firms and Social Responsibility . . . . . . . . . . . . . . . 170
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Chapter 6: Conclusion and Directions for Future Research . . . . . . . . . . . . . . . . .
v!!
!
!
221
!
LIST OF FIGURES
Figure 1.1: FDI into Fragile States: Top 10 Home States by Volume . . . . . . . . .
5
Figure 1.2: Top 10 States by Volume (Omitting Mexico) . . . . . . . . . . . . . . . . . .
6
Figure 1.3: FDI into Fragile States: Top 10 Home States by Share . . . . . . . . . . .
7
Figure 2.1: Scree Plot From Principal Factor Analysis . . . . . . . . . . . . . . . . . . . .
44
Figure 3.1: Marginal Effects of Political Risk on FDI . . . . . . . . . . . . . . . . . . . . .
90
Figure 4.1: Demographic Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
140
Figure 5.1: Demographic Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
195
Figure 5.2: Social Responsibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
206
vi!
!
!
!
!
LIST OF TABLES
Table 2.1: Types of Political Risk: Factor Loadings . . . . . . . . . . . . . . . . . . . . . .
45
Table 2.2: Pairwise Correlations Between Risk Types . . . . . . . . . . . . . . . . . . . .
46
Table 2.3: Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
48
Table 2.4: Political Risk Similarity and FDI . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
Table 2.5: Similarity in Bureaucratic Risk (Disaggregated) . . . . . . . . . . . . . . . .
53
Table 2.6: Similarity in Policy Risk (Disaggregated) . . . . . . . . . . . . . . . . . . . . .
55
Table 2.7: Similarity in War Risk (Disaggregated) . . . . . . . . . . . . . . . . . . . . . . .
57
Table 2.8: Bureaucratic Risk (ICRG Corruption): Firm Level Data . . . . . . . . . .
59
Table 2.9: Policy Risk: Firm Level Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
61
Table 2.10: Risk of War and Political Violence: Firm Level Data . . . . . . . . . . . .
63
Table 3.1: Political Risk and Migrant FDI: Dyad Fixed Effects . . . . . . . . . . . . . .
91
Table 3.2: Political Risk and Migrant FDI: Specification in Changes . . . . . . . . . .
95
Table 3.3: Political Risk and Migrant FDI: Cross Sectional Regression . . . . . . . .
99
Table 3.4: Political Risk and Migrant FDI: Risk Increases vs. Risk Decreases . .
103
Table 3.5: Descriptive Statistics From Annual Data . . . . . . . . . . . . . . . . . . . . . . .
116
Table 3.6: Descriptive Statistics From 3-Year Blocks of Data . . . . . . . . . . . . . . .
117
Table 4.1: Georgia in Comparison to Other Fragile and Developing Countries .
134
Table 4.2: Number of Respondents by Diaspora Ownership . . . . . . . . . . . . . . . .
138
Table 4.3: Distribution of Ownership by Home Region . . . . . . . . . . . . . . . . . . . .
139
Table 4.4: Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
143
Table 4.5: Friendship and Family Ties in Firm Entry Decisions . . . . . . . . . . . . . .
145
vii!
!
!
!
!
LIST OF TABLES (CONTINUED)
Table 4.6: The Subjective Importance of Family Relationships . . . . . . . . . . . . .
148
Table 4.7: The Subjective Importance of Friendships . . . . . . . . . . . . . . . . . . . . .
150
Table 4.8: The Use of Family Relationships to Buy or Purchase Real Estate . . .
152
Table 4.9: The Use of Friendships to Buy or Purchase Real Estate . . . . . . . . . .
154
Table 4.10: Ways Respondents Report That Their Firms Use Social Networks
157
Table 4.11: Home Country By Response Type . . . . . . . . . . . . . . . . . . . . . . . . . .
167
Table 4.12: Comparing the Random Sample and the Supplement . . . . . . . . . . .
169
Table 5.1: Georgia in Comparison to Other Fragile and Developing Countries .
189
Table 5.2: Survey Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
191
Table 5.3 Number of Respondents by Diapsora Ownership . . . . . . . . . . . . . . . .
193
Table 5.4: Distribution of Diaspora Ownership by Home Region . . . . . . . . . . . .
194
Table 5.5: Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
198
Table 5.6: Diaspora-Ownership and Social Responsibility: Labor . . . . . . . . . . .
201
Table 5.7: Contributions to Development and Donations to Charity . . . . . . . . . .
204
Table 5.8: Response Types by Home Country . . . . . . . . . . . . . . . . . . . . . . . . . . .
219
Table 5.9: Comparing the Random Sample and the Supplement . . . . . . . . . . . . .
220
viii!
!
!
!
!
LIST OF ABBREVIATIONS
BEA
Bureau of Economic Analysis
CDIS
Coordinated Direct Investment Survey
CEPII
Centre d'Etudes Prospectives et d'Informations Internationales
CIRI
Cingrinelli and Richards
DBI
Doing Business Indicators
DPI
Database of Political Institutions
FDI
Foreign Direct Investment
GDP
Gross Domestic Product
ICRG
International Country Risk Guide
IMF
International Monetary Fund
OECD
Organization for Economic Co-operation and Development
ONDD
Office National Du Ducroire
PITF
Political Instability Task Force
TI
Transparency International
WDI
World Development Indicators
WGI
World Governance Indicators
ix!
!
!
!
!
ACKNOWLEDGEMENTS
I would first like to thank my chair, Miles Kahler, for his support and guidance
throughout my graduate career. He has read the papers, the proposals and the chapters,
provided the detailed feedback, written the letters, and even provided the funds for
much of my fieldwork. Similarly, the members of my committee, Lawrence Broz,
Edmund Malesky, Craig McIntosh and Kaare Strom have been incredible in both the
volume of time they have contributed and the insightfulness of their comments. I
would also like to thank Ronald Suny and Liesl Riddle for their mentorship
throughout this process.
Through the generosity of these professors, I have incurred what I think of as a
large advising debt, which can only be repaid to future graduate students. I hope I can
do for these future students what has so kindly and effectively done for me.
I would also like to thank the Rohr Chair of Pacific International Relations for
funding fieldwork in Georgia, and the Georgian Center for Strategic and International
Studies and the Caucasus Research Resource Center for valuable in-country support.
I would like to thank the members of the Capital and Conflict: Georgia survey
team: Maya Baramidze, Amiran Chanchibadze, Anna Gogokhia, David Kokiashvili,
Alexandre Kukhianidze, Anastasia Laitadze, Giorgi Mekerishvili, Maia Mestvirishvili,
Anna Sekowska-Livny, and Giorgi Tarkhan-Mouravi.
Timothy Blauvelt, Hans Gutbrod, Alexandre Kukhianidze, Salome Kikvadze,
Jonathan Kulick, Tata Tsereteli, also provided valuable advice and support on the
survey project.
x!!
!
!
!
Numerous additional people have given comments on drafts and presentations
of this work. Without their contributions, this project would be much less than it is.
These people include Megan Becker, Cesi Cruz, Chris Dawes, Jesse Driscoll, Daniel
Enemark, Chris Fariss, Charles Graham, Ellie Graham, Steph Haggard, Daniel
Hidalgo, Erik Gartzke, Scott Gates, Dalia Ghanem, Peter Gourevitch, Steph Haggard,
Erin Hartman, Nathan Jensen, Allison Kingsley, David Lake, Quan Li, Dali Ma, Mat
McCubbins, Jonathan Markowitz, Steven Oliver, Kelly Paulson, Christina Schneider,
Douglas Spencer, Håvard Strand, Lynn Ta, Abby Wood, and Langzhe Zeng.
xi!
!
!
!
!
VITA
Current Position
Assistant Professor, School of IR, University of Southern California, August 2012.
Education
Ph.D., Political Science, University of California, San Diego, June 2012.
M.A., Political Science, University of California, San Diego, 2008.
B.A., Government, Dartmouth College, 2004.
Dissertation
Capital and Chaos: Fragile States, Political Risk, and Foreign Direct Investment.
Committee: Miles Kahler (Chair), Lawrence Broz, Kaare Strom, Edmund Malesky,
Craig McIntosh.
Publications
“Democracies Only: When do FDI Agreements Serve as a Seal of Approval.” (with
Molly Bauer and Cesi Cruz). Review of International Organizations, forthcoming.
“Nagorno-Karabakh in Limbo.” Middle East Quarterly, Vol. 16 (4), 2009.
“India’s Emergence as a World Power,” (with George Perkovich). In Devin T.
Hagerty (Ed.), South Asia in World Politics. Lanham, MD: Rowman and Littlefield
Publishers, 2005.
Current Working Papers
“Unrecognized States: Self Determination and Foreign Influence” (with Ben Horne).
Under Review.
“Doing Business in the Homeland: Diaspora Owned Firms and the Value of Social
Networks.” Under Review.
“Allying to Win: Regime Type, Alliance Size, and Victory” (with Chris Fariss and
Erik Gartzke). Under Review.
“Diaspora-Owned Firms and Social Responsibility.” Under Review.
“Political Risk and Diaspora Direct Investment.” Under Review.
“The Political Risk of Repatriating Profits: How Transfer Risk Affects Foreign
Investment” (with Allison Kingsley and Noel Pereyra-Johnston).
xii!
!
!
!
!
Selected Conference Presentations
“Doing Business in the Homeland: Diaspora Owned Firms and the Value of Social
Networks.” Presented at the International Business Research Forum on Diaspora
Investment and Entrepreneurship, Philadelphia, PA, October 2011.
“Unrecognized States: Self Determination and Foreign Aggression” (with Ben Horne).
Presented at the annual meeting of the American Political Science Association,
Seattle, WA, September 2011.
“Diaspora-Owned Firms and Social Responsibility.” Presented at the annual meeting
of the Academy of Management. San Antonio, Texas, August 2011.
“The Diaspora Difference: Firm Level Evidence From Georgia.” Presented at the
annual meeting of the International Studies Association. Montreal, Quebec, March
2011.
“Disciplining Area Studies: The Place of Comparative Politics.” Panelist, annual
meeting of the Association for Slavic, East European, and Eurasian Studies. Los
Angeles, CA, November 2010.
“Political Risk and Diaspora Direct Investment.” Presented at the annual meeting of
the American Political Science Association. Washington, DC, September 2010.
“Diaspora Direct Investment as a Means of Achieving Political Voice,” with Tjai
Nielsen and Liesl Riddle. Presented at the GW-CIBER Summer Doctoral Institute,
George Washington University. Washington, DC, August 2009.
“A Bargaining Model Applied: Prospects for a Negotiated Resolution of the Status of
Nagorno-Karabakh.” Presented as an invited speaker at the conference After Kosovo:
Whither Karabakh? University of Michigan. Ann Arbor, MI, January 2009.
“A Bargaining Model Analysis of the Stalemate in Abkhazia.” Presented at the
annual meeting of the Western Political Science Association. San Diego, CA, March
2008.
xiii!
!
!
!
!
Research Experience
Field Work
Capital and Conflict: Georgia, 2009-2010.
• Wrote, pre-tested, and revised a two-language survey instrument.
• Recruited, trained and managed a team of six enumerators to conduct a
survey of foreign firms in Georgia.
Georgian Foundation for Strategic and International Studies, Research Intern, 2007.
• Conducted research on professionalism in the Georgian bureaucracy.
Research Assistant
• Kaare Strom (UC San Diego ) and Scott Gates (Peace Research Institute of
Oslo), 2008-present.
• Miles Kahler (UC San Diego), 2007-2008.
• George Perkovich (Carnegie Endowment for International Peace), 2003.
• Stephen Brooks (Dartmouth College), 2002.
Datasets
Powersharing, Agency, and Civil Conflict, Operations Manager. National Science
Foundation Grant #SES-0819507b. Principal Investigators: Kaare Strom and Scott
Gates.
• Drafted the codebook & managed 6 researchers coding a 179-country database.
Teaching
Teaching Associate, University of California, San Diego, 2010-present.
Courses Taught:
• Introduction to International Relations (Department of Political Science,
Winter 2012).
• Practice of Social Research (Department of Sociology, Fall 2011).
• Analytic Writing (International Relations and Pacific Studies Graduate School,
Summer 2010).
Teaching Assistant, University of California, San Diego, 2007-2008.
• Introduction to Law and Society (Law and Society Program, Winter 2008).
•
Marine Policy (Department of Political Science, Spring 2007).
Teaching English as a Foreign Language Volunteer, United States Peace Corps,
Turkmenistan, 2004-2006.
Undergraduate Writing Tutor, Dartmouth College Composition Center, 2001-2004.
xiv!
!
!
!
!
Fellowships, Grants and Awards
Department of Political Science Teaching Fellowship, UC San Diego, 2011-2012.
Tuition and Fee Fellowship, Department of Political Science, UC San Diego, 2010–
2011.
American Political Science Association Travel Grant, 2010.
Dissertation Research Grant, Rohr Chair of Pacific International Relations, UC San
Diego, 2009-2010.
GW-CIBER Summer Doctoral Institute, George Washington University, 2009.
Edmund G. Brown Sr. Teaching Fellowship, Department of Political Science, UC San
Diego, 2006–2010.
War and Peace Studies Fellow, Dartmouth College, 2003–2004.
Rockefeller Center Public Policy Intern Grant, 2003.
Leadership
Mentor, Next Step Mentoring Program, UC San Diego, September 2011-present.
• Served as a one-on-one faculty mentor for undergraduate students.
Graduate Council Faculty Liaison, Department of Political Science, UC San Diego,
2007-2008.
• Served as liaison between graduate students and department faculty, with
particular focus on coordinating graduate student input to a revision of the
methods curriculum.
Team Leader, Habitat for Humanity Global Village, Casablanca, Chile, 2008.
• Co-lead a team of 16 people on a two-week trip to Chile to build emergency
housing.
Ad Hoc Reviewer
Demokratizatsiya, Academy of International Business, Nationalities Papers
Professional Memberships
American Political Science Association
International Studies Association
Association for Slavic, East European, and Eurasian Studies
Academy of International Business
Academy of Management
xv!
!
!
!
!
Language and Methodological Training
Foreign languages: Turkmen (Intermediate High), Russian (Intermediate Mid).
Methodological Training: Completed research design, field research methods and
econometrics courses in three different graduate departments at UC San Diego
(Political Science, Economics, and International Relations and Pacific Studies).
References
Professor Miles Kahler, Rohr Chair of Pacific International Relations, International
Relations and Pacific Studies Graduate School, University of California, San Diego.
Relationship: Chair of dissertation committee.
E-mail: mkahler@ucsd.edu, Phone: 858-534-3078.
Professor J. Lawrence Broz, Associate Professor of Political Science, University of
California, San Diego.
Relationship: Committee member and former Director of Graduate Studies.
E-mail: jlbroz@ucsd.edu, Phone: 858-822-5750.
Professor Kaare Strom, Distinguished Professor of Political Science, University of
California, San Diego.
Relationship: Committee member and employer.
E-mail: kstrom@ucsd.edu , Phone: 858-534-0793.
Professor Ronald Grigor Suny, Charles Tilly Collegiate Professor of Social and
Political History and Director of the Eisenberg Institute of Historical Studies,
University of Michigan.
Relationship: Mentor and panel co-chair at the 2011 Annual Meeting of the
Association for Slavic, East European, and Eurasian Studies.
E-mail: rgsuny@umich.edu, Phone: 734-615-7420.
Professor Scott Gates, Research Professor and Director of the Centre for the Study of
Civil War, Peace Research Institute of Oslo.
Relationship: Mentor, employer and collaborator.
E-mail: scott@prio.no, Phone: 011-47-22-54-77-32.
Professor Liesl Riddle, Associate Professor of International Business and
International Affairs and Associate Dean for MBA Programs, George Washington
University.
Relationship: Mentor and collaborator.
E-mail: lriddle@gwu.edu, Phone: 202-994-1217.
xvi!
!
!
!
!
ABSTRACT OF THE DISSERTATION
Capital and Chaos:
Fragile States, Political Risk, and Foreign Direct Investment
by
Benjamin A.T. Graham
Doctor of Philosophy in Political Science
University of California, San Diego, 2012
Professor Miles Kahler, Chair
Fragile states are trapped in cycles of poverty, violence, and instability. War
and instability deter investment. Low investment retards growth, and lack of growth
engenders further conflict and instability. One path out of this equilibrium is for
fragile states to succeed in attracting foreign direct investment (FDI) while political
risk remains high. My dissertation explores firm-level variation in how investors
experience and respond to political risk, identifying types of investors who are, and are
not, willing to invest in post-conflict and other fragile states. I then explore the
mechanisms through which these investors manage political risk. I focus specifically
on foreign firms that specialize in political risk management, and on diasporans (i.e.
migrants and their descendents), a group of potential investors that is theorized to be
particularly willing to, and capable of, investing in fragile states. Empirically, I
exploit both time-series-cross-sectional data on dyadic FDI between states, as well as
firm-level data from an original survey of foreign firms in the post-conflict country of
Georgia.
xvii!
!
!
!
!
Chapter 1:
Introduction
Fragile states are trapped in a cycle of poverty, violence, and instability. War
and instability deter investment, low investment retards growth, and lack of growth
engenders further conflict and instability. One path out of this equilibrium is for
fragile states to succeed in attracting foreign direct investment (FDI) while political
risk remains high.
The pattern of FDI flows into fragile states suggests that wars, coups, and
political violence contain both risk and opportunity to foreign firms. While political
risk deters investment, it does not deter all investors equally, and does not deter some
investors at all. This dissertation asks three related questions: What types of foreign
firms are willing and able to invest in fragile states and other states with high levels of
political risk? What types of political risk can foreign firms develop capabilities to
manage and what types of experience allow them to do so? Among foreign firms that
invest in fragile and high-political-risk states, what types of firms contribute more or
less to the stabilization and development of these countries?
These three questions all revolve around a core puzzle of political economy:
how institutions shape economic behavior. Due both to reasons of data availability and
the substantive importance of larger, more advanced, economies, most study of the
impact of institutions and economic behavior is conducted in the context of relatively
high-performing institutions. However, analyzing economic behavior in the absence of
institutions offers a unique window into this relationship as well, establishing a sort of
baseline against which institutional effects can be measured. Fragile states often
!
1!
!
2!
contain just such an institutional void, and represent a unique opportunity to expand
the scope of our understanding.
The study of fragile states also offers a unique opportunity to conduct
normatively important and policy relevant research. State fragility produces both
suffering for local residents and security problems for foreign states as havens for
terrorists, pirates, and black-market traders. As noted above, foreign investment –
perhaps particular kinds of foreign investment – offers a potential path out of a
normatively bad equilibrium. While most academic work on political risk focuses on
the type of institutions that limit political risk effectively, this research agenda focuses
on the type of investors that can function in countries in which political risk is not
effectively limited, and how those investors can contribute to stabilization and
development.
I begin this chapter by describing and defining fragile states and providing
some basic description of patterns of foreign direct investment into these states. I then
introduce the broad theoretical approach of the dissertation, and outline the four standalone papers that comprise it.
Defining and Describing Fragile States
Fragile states can be defined either as states at risk of collapse in the near
future or, as I do in this project, as states in which government control is limited
and/or violent intrastate conflict is already a reality.1 Because loss of government
control generally occurs gradually and the outbreak of intrastate violence generally
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
Goldstone et al. (2003, p. 3) write, “States that are ‘failing,’ ‘in failure,’ or ‘recovering from failure,’
may be considered as all—in varying degrees—fragile states.” This matches my definition well.
!
!
!
!
3!
begins prior to full state collapse, measures of state fragility based on these two
definitions have substantial overlap: limited government control and the outbreak of
state violence are among the best predictors of state collapse. Because intrastate
violence, partial loss of government control, and full state collapse are all outcomes of
interest, many projects, such as the Political Instability Task Force, move back and
forth between these definitions as specific research questions dictate.2
Violence and government incapacity are not, of course, the only useful
predictors of eventual state failure, and scholars who attempt to measure fragility as
the probability of failure often include many other predictors as part of their
definitions, including economic, social and political factors, that are correlated with
state failure, but which may also be present in states that characterized by peace and
high-capacity governance. The most prominent example is Foreign Policy Magazine’s
Failed States Index, which includes everything from the economic growth rate and
income inequality to demographics, elite factionalization, and foreign intervention
(Baker 2006).
In the context of this project, which focuses on the impact of state fragility and
political risk on foreign investment, I define state fragility narrowly as a set of current
conditions, rather than as a predisposition to future collapse. I draw on a variety of
data sources, but especially the Political Instability Task Force to measure whether a
state is currently experiencing, or has recently experienced, violent conflict or adverse
regime change.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2
The project has produced a series of reports and articles, all of which are available at
http://globalpolicy.gmu.edu/pitf/ (Accessed May 7, 2012).
!
!
!
!
4!
I define fragile states as states that meet any of the following conditions:
1. A major3 war, genocide, or politicide on its territory in the past three years.
2. A successful coup or regime-change-by-force in the past three years.4
3. Occupation by a foreign power (interruption) or absence of any central
governing authority (interregnum) in the past three years.5
Who Invests in Fragile States? Cross-National Patterns
Descriptive statistics on investment in fragile states are difficult to compile
because state fragility is an effective predictor of data missingness. The International
Monetary Fund (IMF) recently began the Coordinated Direct Investment Survey
(CDIS), which aims to collect bilateral FDI data for all country pairs in the world, but
the coverage remains limited. Figure 1.1 shows the ten home-countries that reported
the largest total FDI outflows to fragile states in 2009. The United States tops the chart
at $126 billion.6
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
3
For this component I use the Political Instability Task Force data on wars, genocides, and politicides.
Major events are: events in any of these categories in which one or more of the following conditions are
met: 1. Over 5,000 fatalities; 2. At least 25% percent of the territory and/or most major cities are
affected; 3. At least 10% of the territory (or some major cities) AND at least 1000 fatalities.
4
I draw this data from both the Powersharing, Agency, and Civil Conflict dataset and the Center for
Systemic Peace. If either data source records a successful coup or regime-change-by-force, a country is
coded as fragile for the next three years.
5
These codings are drawn from Polity IV.
6
Mexico is coded as fragile in 2009 because of cartel-related violence. However, even if Mexico is
coded as not-fragile, the U.S. remains the single largest source of FDI to fragile states. With flows to
Mexico omitted, the US does fall out of the top 10 in percentage terms.
!
!
!
!
120
100
80
60
40
20
0
Total FDI Outflows to Fragile States (Billions of USD)
140
5!
Hong Kong
Canada
Switzerland Germany
France
U.K.
Japan
Netherlands
Spain
U.S.
Figure 1.1: Top 10 States by Volume of Outward FDI Into Fragile States (2009)
Figure 1.1: Top 10 States by Volume of Outward FDI Into Fragile States (2009)
Of the U.S. outward FDI that went to fragile states in 2009, approximately
78% of that went to Mexico, which is coded as fragile in 2009 because of the violence
associated with drug cartels. Figure 1.2 presents the same information as Figure 1.1,
while omitting FDI flows to Mexico. Mexico is an outlier among fragile states in its
level of economic development, and it attracts much more FDI than most fragile
states, but the list of major investment-sending countries remains similar with Mexico
removed.
!
!
!
!
25
20
15
5
10
*Note: FDI Flows to Mexico Omitted
0
Total FDI Outflows to Fragile States (Billions of USD)
30
6!
Malaysia
Germany
Switzerland
Hong Kong
Spain
France
U.K.
Netherlands
Japan
U.S.
Figure 1.2: Top 10 States by Volume of Outward FDI Into Fragile States (2009)
Figures 1.1 and 1.2 are dominated by large, wealthy states that send large
volumes of FDI to both fragile and non-fragile states. Figure 1.3 shows the ten home
countries that send the highest percentage of their total FDI outflows to fragile states.
!
!
!
!
80
60
40
20
0
% of a Country's Total FDI Outflows That Goes to Fragile States
100
7!
U.S.
Mexico
Malaysia
Philippines
Spain
Pakistan
Bangladesh
Slovenia
Croatia
Mauritius
Figure 1.3: Top 10 States by Share of FDI to Fragile States
Seven out of ten of these countries come from outside the Organization for
Economic Co-operation and Development (OECD). Mauritius tops the table, sending
83% of its outward FDI to fragile states. This investment flows primarily to countries
nearby: Madagascar, the Seychelles, and Mozambique. South Africa is the only nonfragile recipient of Mauritian FDI. Mauritius is incredibly stable, having experienced
no violent conflict since decolonization, and consistently scoring the highest possible
value on the Polity IV democracy scale since 1982. The only other two African
countries that report data on outward flows report either zero FDI to fragile states
(Mozambique) or nearly zero (South Africa), despite numerous fragile states nearby.
South Africa, in particular, sends a great deal of FDI to OECD countries and to more
stable African states such as Botswana, Lesotho, Malawi and Tanzania.
!
!
!
!
8!
Who Invests in Fragile States? Firm-Level Theory
State fragility, and high levels of political risk more broadly, deter investment,
but do not deter all investors equally. Some firms are capable of operating profitably
in the face of violence, low institutional function, corruption, and other forms of
extreme political risk. But what type of firms? This question is important for
understanding why some fragile states succeed in attracting FDI and others fail, but
also for understanding what effect FDI is likely to have on the development and
stabilization of the fragile states that succeed in attracting it.
Foreign investment not only responds to political conditions, but also affects
those conditions. The stock of foreign direct investment in a country affects the costs
governments pay for violating property rights, and affects incentives for policy reform
(Malesky 2006; Egan 2010). Different types of firms have different preferences over
the institutional environment in which they operate. While most firms benefit from
lower levels of political risk and higher levels of institutional development, this is not
the case for all firms. If a firm is highly capable at managing a particular type of
political risk, and if that risk deters competitors from entering the market, the firm
may actually benefit from perpetuating a status quo in which political risk remains
high and competition remains low.
In understanding the implications of firms’ sensitivity to political risk, it is
useful to think of three archetypical firms that are capable of investing profitably in
fragile states.
!
!
!
!
9!
Firm A is a firm that has no special risk-management capabilities, but invests
in projects for which the returns are high enough to offset substantial expected losses
due to political risk. Firm A invests in projects where the direct costs are low relative
to the value of the product, allowing the firm to absorb high levels of indirect costs
from political instability, war, contract insecurity, etc. If the indirect costs fall, Firm
A’s profitability increases and it may expand to exploit additional business
opportunities. Firm A’s presence in a host country increases the likelihood of peace
and political stabilization by aiding economic development and by creating a
constituency for political stabilization – Firm A prefers higher levels of institutional
function.
Firm B has unusual capabilities for managing a particular type of political risk,
or managing political risks in a particular country. Because political risk imposes
smaller costs on Firm B than it imposes on Firm B’s competitors, Firm B can
profitably exploit business opportunities that would not be profitable to its
competitors. Political risk serves as a barrier to entry for Firm B’s competitors, and the
presence of political risk increases Firm B’s profitability by restricting competition.
Firm B represents a constituency opposed to political stabilization – Firm B prefers
lower levels of institutional function.
Firm C is a firm willing to absorb high indirect costs from political risk and
accept a below-market rate of return on investment in deference to patriotism or
related non-pecuniary incentives. Firm C, like Firm A, both contributes to
development through direct promotion of economic development and through
!
!
!
!
10!
development of a constituency for political stabilization. Investors motivated by nonpecuniary incentives may even contribute to development in ways that exceed the
contributions of strictly profit-motivated firms.
All three of these archetypical firms are willing and able to invest in fragile
states, but they vary in the degree to which their presence is expected to contribute to
the development and stabilization of these states.
The current political science literature, which treats political risk as a countrylevel variable, makes the tacit (and sometimes explicit) assumption that most firms
investing in fragile states are like Firm A. Investment projects whose returns are large
enough to outweigh large political risks generally derive from the natural endowments
of the host country, and are primarily in natural-resource extraction industries, and
occasionally in agriculture.
The existence of firms like Firm C has been proposed in the business literature,
but empirical evidence for their existence is limited. Chapters 3 and 5 of the
dissertation examine evidence at both the cross-national level, using aggregate flows
of FDI between countries, and at the sub-national level, looking at the behavior or
foreign firms investing in the post-conflict country of Georgia, regarding whether such
firms exist.
Firms like Firm B are similarly theoretically plausible, but under-theorized and
empirically under-explored. Guidolin and Ferrara (2007) find evidence that some
diamond-mining firms in Angola benefitted from that country’s civil war because it
kept competing firms from entering. However, little is know about how common
!
!
!
!
11!
similar dynamics are in other countries and other industries, or what allowed some
firms to operate successfully in Angola during the war while others could not.
In this dissertation I develop and test theory regarding two types of firms with
unusual capabilities for managing political risks: political risk specialists, who develop
capabilities for managing a particular type of political risk, and diaspora-owned firms,
who have social-networked based abilities specific to a particular country.
Investors with Country-Specific Advantages
Diasporans, i.e. migrants and their descendents, are a group of investors who
potentially possess motivations and advantages specific to their homelands. A number
of recent programs introduced by the US Agency for International Development and
other organizations promote diaspora direct investment as a development tool in
fragile states where various forms of political risk are high. I argue that social
networks provide diasporans with competitive advantages in their homeland, but that
diasporans’ access to information actually causes them to be more sensitive to overtime changes in political risk than are other foreign investors. Diaspora investment can
serve as a complement to, and incentive for, improvements in governance, but is no
more likely than other forms of FDI to precede these investments.
At the cross-national level, I demonstrate that flows of migrant-induced FDI
respond more strongly than other FDI flows to over-time changes in the level of
political risk, and particularly to decreases in the level of risk. At the firm level, I
exploit original survey data to show that diaspora-owned firms rely on social networks
more heavily than do other foreign firms and that the owners and managers of
!
!
!
!
12!
diaspora-owned firms perceive social networks to be more important to firm
profitability than do their counterparts at other foreign firms. The informational and
social network advantages of diasporans alter both the volume and the nature of
diaspora direct investment. Notably, high levels of information allow diasporans to
respond quickly and effectively to seize new investment opportunities that emerge in
the homeland when political risk increases. This implies that diaspora investment can
complement and incentivize improvements in governance and the rule of law, but it
cannot substitute for, and generally does not precede, these improvements.
Firms with Risk-Specific Advantages
I also examine investors with unusual capabilities for operating in countries
with specific political risk profiles. I argue that firms with experience managing
certain types of political risk develop capabilities in managing those specific risks, and
can exploit these capabilities across countries. Firms specialize in managing the types
of political risks associated with unreliable institutions: e.g. corruption, bureaucratic
dysfunction, policy instability, and political violence. Political risks associated with
adverse regime change, such as coups and revolutions, are less amenable to
specialization. I argue that the risks associated with the (potential) overturn of a
political order are not something the firms can learn to manage effectively – indeed it
is usually the firms that were most adept at managing the bureaucratic and policy risks
associated with the previous regime that are most vulnerable to expropriation and
negative treatment by the new regime.
!
!
!
!
13!
Firms develop political risk management specializations based on the political
environments in both their home country and in the host countries in which they
invest. Firms with experience in managing a specific type of political risk are more
likely to enter new countries with high levels of that specific political risk.
Empirically, I show that similarity in the level of bureaucratic risk between the home
and host country is associated with greater bilateral FDI flows in that dyad. I also
show that, among Japanese manufacturing firms, firms tend to enter new countries that
share the political risk profile of countries in which they already do business. I find no
evidence that firms benefit from prior experience managing the risk of adverse regime
change.
Challenges in the Study of Fragile States
The empirical study of investment in fragile states is plagued most centrally by
paucity of data. Fragile states are plagued by low institutional capacity, and many
governments of fragile states have reasons to avoid transparency. These two factors
combine to insure that vary little data about the economies of fragile states emerges;
even the most basic facts, such as gross domestic product, are often unknown.
One of the simplest descriptive questions posed in this dissertation, who
invests in the world’s most-fragile states, is one that remains largely unanswered at its
end. Where the dissertation succeeds, however, is in gaining an understanding of
investment in the face of high political risk more broadly. Pointed empirical questions
about the most-fragile states remain an important part of the research program of
which this dissertation is part, but by expanding the range of inquiry to include states
!
!
!
!
14!
with high levels of political risk more broadly, the dissertation gains both greater
empirical traction and broader policy relevance.
However, despite the empirical difficulties of examining these most extreme
cases and the benefits of expanding analysis to encompass a wider range of political
risk climates, the study of fragile states enables important theoretical and policy
contributions. Therefore, I keep a portion of the dissertation, and of the research
agenda in which it is embedded, focused on these most extreme cases.
An Outline of the Dissertation
The body of this dissertation consists of four stand-alone papers on the theme
of FDI in weak and fragile states. Two of the papers draw on cross-national evidence,
including bilateral flows of foreign direct investment at the national level; two of the
papers draw on evidence from an original firm-level survey conducted in the postconflict country of Georgia in 2009.
Chapter 2: Capital and Chaos
The pattern of FDI flows into weak and fragile states suggests that wars, coups,
corruption, inefficient courts, and unconstrained executives represent both risk and
opportunity for foreign firms. While political risk is often conceived of as a countrylevel phenomenon, firms vary widely in their ability to manage these risks. I argue
that, based on prior experience managing political risks in the home country and
previously-entered host countries, firms develop strategy, structures, and capabilities
tailored to the management of particular political risks. I also present a five-part
typology of political risks and argue that these types of risk vary in the degree to
!
!
!
!
15!
which prior home- and host-country experience predisposes firms to seek out future
host countries with familiar political risk profiles. Empirical analysis on home-country
effects is conducted using data on bilateral flows of FDI from the IMF’s Coordinated
Direct Investment Survey, while firm-level analysis of host-country effects is
conducted using data from the Japanese Overseas Investment database. Results show
that prior experience predisposes firms to enter future host-countries with similar
levels of bureaucratic risk, policy risk, and risk of war and political violence.
However, consistent with the hypotheses presented, no such effects are found with
regard to the risk of adverse regime change.
Chapter 3: Political Risk and Diaspora Direct Investment
I argue that diasporans are better informed about the political and economic
situation in the homeland than are other potential foreign investors; that higher levels
of information allows diaspora investors to anticipate and respond to changes in risk
more effectively; and that this makes migrant-induced foreign direct investment (FDI)
more sensitive to political risk than is other FDI. I conduct empirical analysis on a
dataset of bilateral FDI flows from 30 OECD countries to 105 developing countries
from 1994-2008: I find that migrant-induced FDI is more sensitive to political risk
than is other FDI, and particularly that migrant-induced FDI is more sensitive to
decreases in political risk. This has three central implications. First, diasporans’ access
to information is a major factor driving the causal relationship between migrant stocks
and FDI flows. Second, the effect of political risk on investment flows depends on the
mix of investor types (particularly diaspora vs. non-diaspora) in the pool of potential
!
!
!
!
16!
investors from which a state is drawing. Third, diaspora direct investment has the
potential to motivate and sustain governance reform and stabilization in fragile states,
but it is no more likely than other FDI to substitute for, or precede, these
improvements.
Chapter 4: Diaspora Owned Firms and the Value of Social Networks
Do diaspora-owned foreign firms enjoy a competitive advantage in the
homeland, and if so, what accounts for this? This paper uses data from an original
survey of 174 foreign-owned firms in the post-conflict country of Georgia to explore
mechanisms of diaspora difference, focusing particularly on how firms use social
networks in business. I find that diaspora-owned firms use social networks more than
other foreign firms in the acquisition of real estate, and that social networks are more
important to firm location decisions and to overall profitability of diaspora-owned
firms. Firms report using social networks to access information, but not to secure
contracts or deter opportunistic behavior. This indicates that social networks provide
important competitive advantages to diaspora-owned firms, but that diasporans, and
foreign firms in general, do not use social networks to substitute for formal contracting
institutions (i.e. courts).
The data for Chapters 4 and 5 was gathered in the Capital and Conflict:
Georgia survey, the first firm-level survey (to my knowledge) that compares diasporaowned firms to other foreign firms. The survey was conducted in person and covers a
representative sample of all firms in Georgia meeting the following conditions: 1) a
for-profit enterprise; 2) at least 10% foreign ownership; 3) registered as active and
!
!
!
!
17!
paying taxes as of June 1, 2009; 4) obtained its first registration in Georgia after the
year 2000. Diaspora-owned firms are those in which at least one of the owners
considers him/herself to be Georgian and was living outside of Georgia when the firm
first entered the country.
Chapter 5: Diaspora-owned Firms and Social Responsibility
A growing literature in economic sociology and business suggests that
diaspora investors are motivated by patriotism and other social and emotional factors,
endowing them with unique potential as a force for international development. The
existing literature argues that diaspora-owned firms are more socially responsible than
other foreign firms, and engage in a range of development-promoting behaviors when
investing in the homeland: hiring more local labor, paying higher wages, and taking
better care of the local environment. I test this theory at the firm level, using data from
the Capital and Conflict: Georgia survey. Across a range of self-reported behaviors
and priorities, I find no evidence that diaspora-owned firms are more socially
responsible than other foreign firms, and some evidence that they are less socially
responsible. I argue that diaspora investors are uniquely capable, but not uniquely
philanthropic, when doing business in their homelands.
Chapter 6: Conclusion
These four stand-alone papers are followed by a short conclusion that serves
both to highlight the joint contributions and implications of the papers, and to outline
directions for further research. While this dissertation is not structured as a protobook, it contains theory and empirical analysis that will be included in two separate
!
!
!
!
18!
book projects, each of which will require additional development beyond the scope of
the four papers presented here.
!
!
!
!
19!
References
Baker, Pauline. 2006. Conflict Assesstment System Tool (CAST): An Analytical Model
for Early Warning and Risk Assessment of Weak and Failing States.
Washington DC: The Fund For Peace.
Goldstone, Jack A., Robert H. Bates, David L. Epstein, Ted Robert Gurr, Michael B.
Lustik, Monty G. Marshall, Jay Ulfelder, and Mark Woodward. 2010. A
Global Model for Forecasting Political Instability. American Journal of
Political Science 54 (1): 190-208.
Guidolin, Massimo, and Eliana La Ferrara. 2007. Diamonds Are Forever, Wars Are
Not: Is Conflict Bad for Private Firms? The American Economic Review 97
(5): 1978-1993.
!
!
!
!
Chapter 2:
Capital and Chaos
Abstract
The pattern of FDI flows into weak and fragile states suggests that wars, coups,
corruption, inefficient courts, and unconstrained executives represent both risk and
opportunity for foreign firms. While political risk is often conceived of as a countrylevel phenomenon, firms vary widely in their ability to manage these risks. I argue
that, based on prior experience managing political risks in the home country and
previously-entered host countries, firms develop capabilities tailored to the
management of particular political risks. I present a five-part typology of political
risks and argue that these types of risk vary in the degree to which prior home- and
host-country experience predisposes firms to seek out future host countries with
familiar political risk profiles. Empirical analysis on home-country effects is
conducted using data on bilateral flows of FDI from the IMF’s Coordinated Direct
Investment Survey, while firm-level analysis of host-country effects is conducted
using data from the Japanese Overseas Investment database. Results show that prior
experience predisposes firms to enter future host-countries with familiar levels of
bureaucratic risk, policy risk, and risk of war and political violence. However,
consistent with the hypotheses presented, no such effects are found with regard to the
risk of adverse regime change.
!
20!
!
21
Introduction
Foreign direct investment (FDI) contributes to economic growth, produces
technological spillovers, and increases human capital. However, it is often the
countries that need FDI the most that are least capable of attracting it: the world’s least
developed countries (LDC’s) received less than 5% of the world’s FDI flows in 2010
(UNCTAD 2011). While poverty and under-development themselves are a major
deterrent to investment, political risk also plays a major role in limiting developing
countries’ access to FDI.1 Poor countries fail to attract investment not only because
they are poor, but also because political risk in these countries is high.
While political risk deters most investors, it does not deter all investors
equally, and does not deter some investors at all. I argue that this heterogeneity must
be taken seriously and that political risk must be analyzed as a firm-level phenomenon
as well as a country-level phenomenon. Firms possess varying capabilities for
mitigating different types of risk, and firms with high capabilities for mitigating a
particular risk may not view the presence of this risk as a major deterrent to
investment. In extreme cases, firms that can mitigate a particular type of risk
effectively may find that risk to be beneficial, in that it deters competing firms and
reduces competition.2 Identifying firms that are relatively risk insensitive is both
intellectually compelling and normatively important, as it informs and assists
investment promotion in the world’s most capital-starved economies.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
See, for example, Pfefferman, Kisunko, and Sumlinki (1999).
For example, Guidolin and La Ferrera (2007) demonstrate that diamond-mining companies operating
in Angola during the civil war benefitted from the presence of the war and lost value when the war
ended suddenly.
2
!
!
22
In the broader research agenda within which this paper is nested, I identify two
categories of factors that affect firms’ ability to manage political risk: factors that
allow a firm to mitigate political risk in a particular country, and factors that allow a
firm to mitigate a particular type of political risk across multiple countries. This paper
explores the latter category, analyzing the degree to which risk-mitigation expertise is
transferable across countries.
I divide political risk into five categories: bureaucratic risk, or risks related to
corruption, low bureaucratic capacity, and the governance of the economy by informal
institutions; transfer risk, or risks associated with the movement of capital across
borders; policy risk, or risks related to policy instability and opportunistic government
behavior at the macro level; risk of war or political violence, including government
violations of physical integrity rights; and risk of adverse regime changes, including
coups and revolutions.
The role of prior experience in the development of risk management
capabilities varies by risk type. On one extreme, I expect that firms with experience
operating in countries with high levels of corruption, low-quality court systems, and
inefficient government bureaucracies develop transferable skills for managing the
risks associated with those country characteristics. On the other extreme, I expect that
firms with experience operating in countries plagued by frequent adverse regime
changes do not develop skills that are useful in mitigating the same type of risk in
other countries. The risk of regime overthrow involves the risk that the entire
institutional order is overturned, and is not amenable to management through the same
!
!
23
type of strategies that firms can use to navigate the risks associated with the flaws in a
particular institutional order.
Firms develop risk mitigation strategies in response to both the risks they face
in their home country, as well as the risks they face in host countries they have entered
previously. As firms develop a repertoire of effective strategies for mitigating specific
types of risk, they become more willing to enter additional countries that are
characterized by high levels of this risk. They become, in effect, risk management
specialists with competitive advantages based on their capability to manage certain
types of political risk.
The empirical portion of this paper is divided into three parts. The first uses
factor analysis to assess the degree to which the five categories of political risk I
distinguish theoretically can also be distinguished empirically. The second analyzes
bilateral flows of FDI to assess whether similarity in political risk profile between
home and host country is an important determinant of FDI. The third uses a dataset of
overseas investments by Japanese manufacturing firms to assess the degree to which
previous experience operating in a host country with a given political risk profile
makes a firm more likely to enter additional host countries with similar risk profiles.
Results are consistent with the theory, but a variety of revised and additional tests are
planned.
Literature Review
In the political science literature, the relationship between governance and
foreign direct investment is viewed primarily as a monadic relationship, and political
!
!
24
risk itself is viewed as a country-level variable. The quality of governance in the host
country affects the costs, and particularly the risk, facing foreign firms who choose to
invest there. Host country characteristics, such as democracy (Li and Resnick 2003;
Jensen 2008); constraints on the executive (Henisz 2000); federalism (Jensen and
McGillivray 2005), and corruption (Habib and Zurawicki 2003; Javorcik and Wei
2009), have all been found to affect aggregate FDI inflows to a given host country.
When the literature addresses governance in a dyadic context, the focus is generally on
relationships between governments: bilateral investment treaties (Büthe and Milner
2009; Allee and Peinhardt 2010), bilateral trade agreements (Büthe and Milner 2008),
or violent interstate conflict (Li and Vashchilko 2010).
In the last several years, an emerging stream of literature has begun to break
from this tradition and explore variations across types of firms in how they experience
and respond to political risk. Wellhausen (2011) demonstrates that a firm’s nationality
(and the number of co-national firms in a given market) affects a firm’s vulnerability
to breach of contract by the host government. Kerner and Lawrence (2012) argue that
vulnerability to political risk is concentrated among firms with investments in fixed
capital rather than liquid assets in foreign affiliates. Graham, Kingsley, and Johnson
(2012) argue that firms can insulate themselves from transfer risk based on how they
finance their direct investments. This paper contributes to this emerging research
program.3
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
3
Other examples of work in this vein include Tan and Meyer (2011) and Kingsley and Noordewier
(2011).
!
!
25
Within the business literature, several authors find that firms based in corrupt
home countries are less deterred by host-country corruption (Habib and Zurawicki
2002; Cuervo-Cazurra 2006). Holburn and Zelner (2010) argue that firms learn to
influence policy and manage policy risk in weakly institutionalized host countries by
operating in a weakly institutionalized home country. They find that firms in the
electric power generation industry tend to enter countries with high levels of policy
risk if they are based in home countries where the level of policy risk is also high.
These findings suggest that firms can learn to manage at least some types of political
risk, and that these risk-management capabilities are transferable across countries.
What the literature lacks, however, is a detailed theory of political risk management
capabilities, including the types of political risk that firms can manage effectively and
the types of experience (in home vs. host countries) that produce and select for those
capabilities.
Theory: Political Risk Management Capabilities as Mobile Assets
Political risk is a firm-level phenomenon as well as a country-level
phenomenon. Firms vary in their abilities to mitigate the risks and costs associated
with corruption, policy instability, and political violence.4 The political risk
environment(s) in which a firm has previously operated are an important determinant
of the firm’s ability to mitigate various types of political risks in any new host country
it enters.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4
Political risk management capabilities fit well within the resource based view of the firm, in that they
are rare, valuable, inimitable, and non-substitutable.
!
!
26
This argument is based on both learning and survival. Firms invest in
developing risk-management capabilities relevant to the types of risk they face in the
countries in which they are currently operating (learning), and firms that best succeed
in developing these capabilities are most likely to survive in those markets and
proceed to enter new markets (survival). For example, if a given home country is
characterized by an inefficient judiciary, firms based there face incentives to invest in
developing resources for enforcing contracts and resolving civil disputes outside the
court system. Firms that succeed in developing alternative means to resolve disputes
and enforce contracts will continue to operate in that country, while firms less able to
avoid reliance on the court system will not be able to survive. Therefore, firms based
in countries with poorly functioning courts are more likely to be skilled in informal
dispute resolution than firms based in countries with high quality judiciaries, for
reasons of both learning and survival. If firms are able to transfer their informal
dispute resolution capabilities from one country to another, we should see the firms
based in countries with low-quality courts willing to invest in countries with similar
judicial deficiencies. The same logic applies to other host countries in which a firm
operates: the firm develops the skills necessary to manage the political risks in one
host country, and can then transfer those skills to operations in any future countries in
enters that share a similar risk profile.
!
!
27
Expectation 1: Firms develop capabilities (e.g. knowledge, skills, strategy, staffing)
appropriate to minimizing the costs and risks associated with the political environment
in the firms’ home country and the host countries in which they operate.
Expectation 2: Firms who possess or develop capabilities to minimizing the costs and
risks associated with the home and host countries in which they operate are more
likely to survive than firms that do not possess or develop these capabilities.
These capabilities are not costless to develop, and firms do not develop
capabilities for managing risks they do not face. For example, firms operating in
corrupt countries will invest resources and structure their operations to deter and
appease bribe-seekers in a way that firms in less corrupt countries do not. In a firm
operating in a corrupt country, permits must be secured by politically connected senior
managers who are selected for their skill in navigating the bureaucracy, while firms in
a less corrupt country may treat the securing of permits as a primarily clerical task
handled by low-level staff.
The level of political risk in a given country is an important factor affecting
firms’ decisions of whether or not to invest in that country. However, the cost of
mitigating the political risks in a given country are higher for some firms than for
others. Firms that are already engaged in mitigating a particular source of risk in one
country will face lower costs in developing the capabilities to mitigate that risk in a
new host country.
!
!
28
A Typology of Political Risks
I distinguish between five types of political risk: bureaucratic risk, transfer
risk, policy risk, risk of war and political violence, and risk of adverse regime change.
While I expect that the skills firms develop to manage a specific type of political risk
are transferable across countries, I do not expect that skills developed to manage one
type of risk are useful in managing a separate type of political risk. I also expect that
some types of risk political risk are more amenable to learned management than
others.
The business literature distinguishes between exogenous hazards, which cannot
be resolved once they materialize, and endogenous hazards, which can be at least
partially resolved after the fact. This literature asserts that political risks are exogenous
hazards: “Governance hazards … typically take the form of irreversible governmental
decisions and political and societal turmoil, and are hence generally impossible for
MNCs to resolve once they become reality (Slangen and Beugelsdijk 2009, p. 983).”5
The theory of political risk management capabilities as mobile assets
challenges this belief directly: I argue that most political risks are endogenous hazards,
and that firms with the proper capabilities can mitigate, and not just avoid, these risks.
In the following sections I articulate this logic with regard to bureaucratic risk, transfer
risk, policy risk, and risk of war and political violence. However, I also argue that the
risk of adverse regime change is an exogenous hazard, and it is precisely for this
reason that this category of political risk is not amenable to learned management.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5
See also Cuypers and Martin (2010) and Li and Rugman (2007).
!
!
29
Bureaucratic Risks
Bureaucratic risks are risks related to corruption, low bureaucratic capacity,
and the governance of the economy by informal institutions. Among the five
categories of political risk I identify, bureaucratic risk is most clearly an endogenous
hazard – one that firms can develop capabilities to manage, even after the risk has
become realized, i.e. even after the firm has been asked for a bribe, or had a shipment
of goods delayed at customs. Firms require specialized skills and strategies to mitigate
bureaucratic risks effectively, and there is reason to expect capabilities developed in
one country to be useful in another. In the previous example, in which a firm hires
senior executives based on their political connections and ability to secure permits and
manage relationships with government officials, this strategy should be useful across a
range of countries in which arms-length relationships with officials are not conducive
to business success, even if the precise nature of the necessary political connections
varies from country to country.
I expect both that firms are more likely to enter host countries with similar
levels of bureaucratic risk to their home country and with similar levels of
bureaucratic risk to host countries in which they are already operating.
Hypothesis 1a: Similarity in bureaucratic risk between home and host country is
positively correlated with flows of bilateral FDI.
!
!
30
Hypothesis 1b: The average level of bureaucratic risk in the previous host countries a
firm has entered is positively correlated with the level of bureaucratic risk in the next
host country selected for entry.
Transfer Risks
Transfer risks are those related to the movement of capital across borders.
Transfer risks are less clearly endogenous hazards – they often arise from central
government actions, like the imposition of capital controls, that individual firms are
unlikely to be able to influence. However, there are strategies that firms can employ,
such as structuring project financing to hedge against these risks, that allow some
firms to mitigate these risks more effectively than others (Graham, Kingsley, and
Johnston 2012). Engaging in hedging and other transfer-risk-mitigation strategies
requires firms to acquire costly knowledge and capabilities, and I expect these
capabilities to be transferable across countries. Transfer risk is most often conceived
of as a threat to foreign firms (e.g. Cosset and Suret 1995; Jensen 2008) rather than as
a threat to multinational firms in their home market. However, while purely domestic
firms that are not involved in foreign markets are immune from the effects of transfer
risk, multinationals move capital in and out of their home country in much the same
way they move capital in and out of the host countries in which they operate.
Therefore, while a firm based in a high-transfer-risk home country may not experience
a prior-learning benefit in its first foreign adventure, it will experience these benefits
in future foreign investments, even if its early foreign entries are all in countries with
!
!
31
low levels of transfer risk. The firm develops transferable capabilities for navigating
transfer based on its experience moving capital in and out of its home country.
Hypothesis 2a: Similarity in transfer risk between home and host country is positively
correlated with flows of bilateral FDI.
Hypothesis 2b: The average level of transfer risk in the previous host countries a firm
has entered is positively correlated with the level of transfer risk in the next host
country selected for entry.
Policy Risk
Policy risks are risks related to policy instability and opportunistic behavior by
government at the macro level. Scholars and risk insurers often group policy risk with
bureaucratic risk as “risk of adverse government action,” but I distinguish between
risks associated with low bureaucratic quality and opportunistic behavior by individual
government officials and the risks associated with government policy. These policy
risks include changes in laws, regulations and tax rates, nationalizations of private
property, and government breach of contract. The two key differences between policy
risk and bureaucratic risk are the degree to which individual firms can affect the
behavior of the political actors in question, and the degree to which each risk affects
domestic firms vs. foreign firms – i.e. the degree to which there is a significant
liability of foreignness with regard to each risk type.
!
!
32
Bureaucratic risk is particularly amenable to effective mitigation by capable
firms because it more often involves local government actors, such as inspectors,
police officers, and clerks, while policy risk involves actors at the national level:
presidents, legislatures, and supreme courts. Large firms may have the ability to
engage in effective non-market strategies at this level, such as lobbying, campaign
contributions, or bribery, but most firms cannot individually affect national policy. As
with transfer risk, however, mitigation is still possible. Firms can invest in
information-gathering capabilities that allow them to anticipate policy changes before
they are announced, or they can partner with domestic firms and other stakeholders to
reduce their vulnerability to nationalization or other firm-specific government
actions.6
Allying with domestic firms can reduce a firm’s exposure to policy risk
because the cost of government expropriation of assets, creeping or outright, is higher
with regard to foreign firms than domestic firms. Costs are higher because domestic
firms are constituents of the government, and because they are more likely than
foreign firms to source products from other domestic firms, making their failure more
economically damaging (Henisz 2000, pp. 337-339). This implies that domestic firms
are more protected from policy risks than are foreign firms.
These two factors lead me to predict both that the host-country effects of
policy risk are likely smaller than those for bureaucratic risk, and that the home
country effects are likely to be non-existent. To the degree that host-country effects do
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
6
Markus (2012) has an excellent discussion of how firms can ally with diverse stakeholder groups to
reduce their vulnerability to government violations of property rights.
!
!
33
exist, they are likely to exist primarily for industries that are heavily regulated,
inducing even domestic firms to develop capabilities in lobbying and campaigning for
particular policies, and for very large firms, who are capable of engaging in these sorts
of activities effectively.
Hypothesis 3a: Similarity between home and host country with regard to policy risk
is NOT positively correlated with higher flows of bilateral FDI.
Hypothesis 3b: The average policy risk in the previous host countries a firm has
entered is positively correlated with the policy risk in the next host country selected
for entry.
Hypothesis 3a directly contradicts the findings of Holburn and Zelner (2010).
There are two reasons for this. First, I argue that the Holburn and Zelner finding is
likely driven by the fact that policy risk is highly collinear with bureaucratic risk,
which Holburn and Zelner do not control for in their analysis. I expect (and find) large
home-country effects regarding bureaucratic risk, effects which are large enough to
explain the Holburn and Zelner finding, even if there is no independent home country
effect of policy risk.
Second, Holburn and Zelner conduct their analysis in the electric power
generation industry, an industry that is extremely heavily regulated, and populated
primarily by very large firms. If home country effects for policy risk are, in fact,
!
!
34
present within this industry, it is still likely that such effects do not exist in other
industries that are either more lightly regulated or populated by smaller firms
incapable of effective lobbying at the national level.
Risk of Adverse Regime Change
All four of the categories of risk discussed above are risks based on the nature
of the current political order and can be managed, to some degree, by firms that have
developed the necessary skills to navigate that political order. The risks posed by
adverse regime change are different; the risk of adverse regime change is the risk that
the entire institutional order may be overturned. The nature of the new political order
that will replace the current order if it is overthrown is difficult to predict, and the very
firms that were successful in mitigating various political risks under the previous
regime are likely to be precisely those firms targeted by the new regime. This is
particularly true when firms contract directly or engage in joint-ventures with the
government or are otherwise perceived to be allies of the current regime.
The risk of adverse regime change is a purely exogenous hazard that firms
cannot substantially mitigate once it has been realized, i.e. once a regime is
overthrown.
Hypothesis 4a: Similarity between home and host country with regard to the risks of
adverse regime change (i.e. coups and revolutions) are NOT positively correlated with
higher flows of bilateral FDI.
!
!
35
Hypothesis 4b: The average risk of adverse regime change (i.e. coup or revolution) in
the previous host countries a firm has entered is NOT positively correlated with the
risk of adverse regime change in the next host country selected for entry.
There is an important caveat when assessing the effect of risk of adverse
regime change on investment behavior. When a regime is overthrown, this hurts most
MNCs that are already operating in the country at the time of the overthrow, but it also
opens up new opportunities for investment. These post-overthrow opportunities
include privatization sales, new opportunities for contracts and joint ventures with the
government, and the exploitation of economic opportunities opened up by the new
regime more generally. Firms may become specialized in seizing these opportunities,
even if they have no particular advantage in managing the risk of future regime
turnover. Therefore, even though recent experience of a coup is a good predictor of
risk of future coups, I do not include this variable in the analysis.
Risk of War and Political Violence
War and political violence pose a range of risks to foreign firms: employees
may be killed or injured, property may be destroyed, transport may be disrupted, and
the local economy is often crippled in both the short and long-term (e.g. Collier 1999;
Ghobarah, Huth, and Russett 2003). The risk of war and political violence are closely
related to the risk of adverse regime change – regime overthrow is often accompanied
by war and/or political violence. However, the risks posed by violence are distinct
from the risks posed by adverse regime change and many cases of civil war and
!
!
36
political violence, particularly secessionist conflicts, occur in situations in which the
risk of overthrow is quite low but the risk of business disruption is substantial.7
While adverse regime change overturns the entire political order, civil war and
political violence are more accurately conceived of as aspects of the current political
order. As with bureaucratic risk, transfer risk, and policy risk, firms can invest in
capabilities to manage these risks. They can employ private security forces, invest in
gathering information, minimize their geographic dispersion and dependence on intracountry transport, and rely more heavily on local, rather than expatriate, employees.
Therefore, I expect firms that have experience operating in countries plagued by war
or political violence, either in their home country or in previously-entered host
countries, to develop skills in managing these risks that give them advantages in
managing similar risks in other countries.
One complicating factor is that war and political violence often trigger capital
flight. While the level of corruption in a country tends to move slowly and to be
stable over time, acute episodes of war and political violence often begin and end
abruptly. Firms and private individuals that are living and operating in a peaceful
country in year t, may find themselves living in war zone in year t+1, and they may
respond to this sudden shift by transferring assets abroad, either into banks, or into
direct investments overseas. This type of FDI-as-capital-flight will tend to result in
flows from war-torn countries to countries with low levels of war risk.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
7
Goldstone et al. make this distinction well (2010: pp. 191-192).
!
!
37
This FDI-as-capital-flight flows to precisely the opposite type of countries as
flows of FDI generated by firms moving to exploit their learned capabilities in
managing war risk. Learned capabilities induce firms based in countries that are or
have recently experienced war or political violence to invest in other similarly
violence-plagued states; FDI-as-capital-flight creates flows of investment from wartorn states to stable, non-violent destinations. The relative expected magnitude of
these two competing effects is unclear, but if they are roughly similar, similarity in
war risk between home and host country will have no effect on bilateral FDI flows.
Hypothesis 5a: Similarity between home and host country regarding the presence of
ongoing war or political violence does NOT have a clear relationship with bilateral
flows of FDI.
Hypothesis5b: The proportion of previous host countries a firm has entered that are
currently experiencing civil war or severe political violence is positively correlated
with the probability that the next host country selected for entry is currently
experiencing civil war or severe political violence.
Dependent Variable and Sample for Analysis of Home Country Effects
The dependent variable, FDI Inflows, is measured bilaterally as net flows of
FDI from the home country to the host country in 2009. These data are taken from the
International Monetary Fund’s (IMF) 2009 Coordinated Direct Investment Survey
!
!
38
(CDIS); the raw values are measured in millions of USD.8 I use a logged dependent
variable, which is created as Log_FDI = log(FDI Inflows + 1).9
The CDIS was released in June 2011 with data for 2009 only. Prior to this
release, the only bilateral FDI data available was from the Organization for Economic
Co-operation and Development (OECD), and included only OECD members as home
countries. The CDIS data is well suited to this analysis because it includes SouthSouth dyads, and data on high-risk states as both home and host countries. While
unique in providing bilateral data on South-South FDI flows, the CDIS data is plagued
by the same problem as other bilateral FDI sources, namely nonrandom missing data
where data is most likely to be missing in dyads where the reporting countries are poor
and/or have limited bureaucratic capacity.10
Dependent Variable and Sample for Analysis of Previous Host Country Effects
To assess the effect of previous host-country experience on the selection of
new host countries, I use a firm-level dataset of 4,743 entries into 73 foreign countries
by 512 Japanese manufacturing firms from 1928-1993, with almost all entries
occurring between 1965 and 1991. I restrict the analysis to a firm’s first entry into a
given host country, which limits the sample to 1,988 entries.11 The dependent variable
is the level of political risk in the host country being entered. For most types of risk
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
8
While some of the related literature (e.g. Ahlquist 2006, Buthe and Milner 2008) uses FDI/GDP as a
dependent variable, this would be in appropriate here because the concept of interest is the ability of a
country to attract FDI, not the dependence of a country’s economy on FDI (Li 2009).
9
The raw data on bilateral FDI inflows includes approximately 100 negative observations out of 6200
total observations. Results are robust to an alternate specification of Log_FDI = ± log|(FDI inflows +1)|
10
In future versions of the paper I intend to test the robustness of these results to multiple imputation of
missing values.
11
Davidson (1980) and others have shown that previous experience in a given host country is an
important determinant of future entries.
!
!
39
this is continuous, for others it takes the form of a dummy variable for whether the
host country being entered has experienced a particular type of event, such as a war, in
the recent past.
The dataset of firm entries is limited to firms based in a single home country
(Japan), and a single broad sector (manufacturing). This holds many potentially
confounding variables constant. Unfortunately, it also limit the external validity of the
results, but this is somewhat offset by the fact that Japan was an early and prolific
investor in developing markets (Belderbos and Sleuwagen 1996; Delios and Henisz
2000), making foreign entries by Japanese firms a compelling universe of cases in its
own right. Also appropriate for this study, Japanese firms have been described as
taking a capabilities-based approach to investing (Chang 1995).
Independent Variables of Interest
I distinguish theoretically between five types of political risk: bureaucratic
risk, transfer risk, policy risk, risk of war and political violence, and risk of adverse
regime change. I treat each type of risk is a latent variable, for which I have gathered
various observable measures. In the section that follows I analyze the effects of both
these individual measures of risk, and indices created to capture each category of risk.
Individual Measures
Bureaucratic risk and policy risk, in particular, are substantially collinear.
Countries in which opportunistic behavior by policymakers is effectively constrained
are often characterized by similarly effective constraints on bureaucrats. Therefore, in
measuring each of these types of risk, I look for narrow and specific measures of
!
!
40
constraints on specific actors, rather than looking at broad measures of government
effectiveness or the rule of law.
For bureaucratic risks, I use measures of corruption, judicial efficiency, and
customs efficiency. The measures of corruption are the Corruption Perceptions Index
from Transparency International (TI) and the corruption component of the
International Country Risk Guide (ICRG) assessments. The TI measure is based on
expert opinions and survey responses, while the ICRG measure is based solely on
expert assessments. The measures of judicial efficiency are the cost of enforcing a
contract using the courts (as a % of the claim) and the speed of those courts, measured
as the number of days (logged) from the filing of the plaintiffs suit until payment is
made. The measures of customs efficiency are the (average) costs per container to
import and export goods, as well as the number of days (logged) needed to complete
the necessary legal steps to do so.12 All four of these measures come from the World
Bank’s Doing Business Indicators.
An expert assessment of transfer risk comes from the Office National Du
Ducroire (ONDD). ONDD is the worlds largest political risk insurer and the price
leader in the industry. Its assessments of risk reflect not only profit-motivated expert
attempts to assess risk, but also the actual insurance costs paid by firms who wish to
be protected from these risks (Graham 2010; Jensen 2008). The ONDD measure of
transfer risk measures the risk that action by foreign authorities, such as the
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
12
The World Bank measures the time and cost for importing and exporting separately: I sum the import
and export measures to get measures of overall customs efficiency.
!
!
41
introduction of capital controls or other constraints, prevents the transfer of money
necessary to repay creditors.
Policy risk is measured both as constraints on the executive, and by the level of
political competition. Political Constraints is a measure of constraints on the
executive based on a spatial model of the preferences of veto players, which comes
from Witold Henisz (2000). Political Constraints considers the number of branches of
government, the preference alignment between each legislative house and the
executive, and the presence of constraints from the judiciary and sub-national
governments.13 Executive Constraints and Political Competition are component
variables of the Polity IV democracy score.
The risk of war and political violence is captured by expert assessments and by
binary variables measuring whether war is currently occurring or whether war has
ended in the past five years. The measures of current and recent war are taken from
the Political Instability Task Force, whose severity codings encompass both the
number of fatalities and the geographic spread of a conflict, both of which affect the
level of economic impact from a conflict (Goldstone et al. 2010). Expert assessments
of war risk are taken from ONDD and ICRG.14 I also use a measure of the security of
physical integrity rights from the CIRI human rights datasets, which measures the
government’s engagement use of torture, extra-judicial killings, political
imprisonment, and disappearances.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
13
For political constraints, I use data from 2007, the most recent year available.
I create a single measure of ICRG war risk as the sum of the risk of internal conflict and the risk of
external conflict.
14
!
!
42
To assess the risk of adverse regime change, I use a measure of political
stability from the World Governance Indicators that captures the likelihood that a
regime is “destabilized or overthrown by unconstitutional or violent means.”15
The independent variables used in the analysis are measures of the similarity
between home and host country across each of the dimensions discussed above. These
similarity measures are created by taking the absolute value of the difference in a
given variable between home and host country and subtracting this value from zero.
Control Variables
Gravity model variables are strong determinants of flows of FDI (e.g. Benassy,
Coupet and Mayer 2007), leading FDI to flow primarily to states near the home
country. Relevant elements of distance include not only geographic distance, but also
cultural distance, economic distance, and political and administrative distance, which
includes the elements of political risk at the core of this study (Ghemawat 2004). In
order to evaluate the effect of similarity in political risk effectively, I need to control
for other aspects of distance that may both affect FDI flows and be correlated with
similarity in political risk. For geographic distance I use the logged distance between
the capitals of the home and host country and a dummy variable for a shared border.
For cultural distance I use the logged number of migrants from the host country living
in the home country and dummy variables for a common language, a previous colonial
relationship, and a common colonizer. For economic distance I measure similarity in
GDP per capita and similarity in educational attainment; both are continuous. I also
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
15
This quote is drawn from data documentation available online at
http://info.worldbank.org/governance/wgi/pdf/pv.pdf (Accessed May 3, 2012). See also Kaufmann,
Kraay, and Mastruzzi (2009).
!
!
43
construct a dummy variable for whether both home and host country are members of
the European Union. For an element of political and administrative distance that does
not fall directly under the heading of political risk, I also control for common legal
origin, a dummy variable.16
The measures of distance, common border, common language, common
colonizer, common legal system and prior colonial relationship are taken from Head,
Mayer, and Ries (2010), courtesy of the Centre d'Etudes Prospectives et
d'Informations Internationales (CEPII). Data on GDP per capita is taken from the
World Development Indicators (WDI). Education data are taken from Barro and Lee
(2010) who measure attainment for each country in both 2005 and 2010. Data for
2009 is constructed via linear interpolation between these two values.
In the firm-level analysis I control for the number of prior entries by the
subsidiaries of the same firm. Firms with more international experience are expected
to be more tolerant of novel environments than less-experienced firms (Davidson
1980). I also include dummy variables for 14 industry categories. All the firms in the
sample are manufacturing firms, but firms may vary in their risk tolerance across
industries.
Factor Analysis: Creating Indices for Each Type of Political Risk
My theory identifies five distinct types of political risk: bureaucratic risk,
policy risk, transfer risk, risk of war and political violence, and risk of adverse regime
change. These types of risk are theoretically distinct, but expected to be highly co!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
16
Common legal origin relies on data from La Porta, Lopez-de-Salinas, and Shleifer (2008).
!
!
44
linear, making them difficult to distinguish empirically.17 I employ factor analysis as a
means of both producing a single measure for each type of risk and of verifying that
the various measures of political risk cluster as they are expected to. The correlations
are expected to be higher between variables measuring the same risk type than
between variables measuring distinct risk types.
The section above identifies 15 measures of political risk, each of which I
argue is associated with one of these five risk types. Two of these types of risk are
captured by only a single measure: transfer risk by the ONDD expert assessment of
transfer risk, and risk of adverse regime change, captured only by the political stability
measure from the World Governance Indicators. These variables are omitted from the
factor analysis, leaving 13 measures that are theoretically associated with three latent
variables, bureaucratic risk, policy risk, and risk of war and political violence.18
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
17
Many of the political risk measures in the literature capture more than one of these risk types, and
measures of theoretically distinct risks are often highly correlated with one another.
18
If these variables are included in the analysis, the measure of transfer risk loads on the bureaucratic
risk factor and the measure of political stability loads on the violence factor. Including these measures
in the indices of these factors does not substantively change the regression results presented – the
alternative indices are correlated with the original indices at >0.96.
!
!
45
Figure 2.1: Scree Plot From Principal Factor Analysis
Figure 2.1 presents a scree plot drawn from a principal factor analysis of all 13
measures in a sample of 121 countries in 2009. The plot shows that, as expected, most
of the variance in these measures can be accounted for by three primary factors; the
eigenvalues for each remaining factor are less than 0.5. These three primary factors are
then rotated, using a varimax rotation, and these rotated results are shown in Table
2.1.19 Factor loadings of greater than 0.4 are shown in bold.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
19
The decision to rotate three factors is made based on the scree plot; remaining factors are left
unrotated and are omitted from Table 2.1.
!
!
46
Table 2.1: Types of Political Risk: Factor Loadings for Three Primary Factors
Bureaucratic
Policy
War and Uniqueness
Risk
Risk
Violence
Corruption (TI)
0.76
0.27
0.41
0.11
Corruption (ICRG)
0.70
0.25
0.28
0.27
Court Costs (DBI)
0.41
-0.03
0.06
0.79
Court Delays (DBI)
0.17
-0.12
0.38
0.71
Trade Delays (DBI)
0.88
0.22
0.14
0.14
Trade Cost (DBI)
0.68
0.03
-0.03
0.39
Political Constraints (Henisz)
0.36
0.61
0.18
0.39
Executive Constraints (Polity)
0.23
0.87
0.12
0.16
Political Competition (Polity)
0.13
0.85
0.11
0.24
War Risk (ONDD)
0.53
0.34
0.54
0.26
Civil War Risk (ICRG)
0.30
0.17
0.74
0.32
Post-Conflict Status (PITF)
-0.03
-0.13
0.21
0.72
Current War (PITF)
0.01
-0.05
0.65
0.47
Physical Integrity Violations
(CIRI)
0.31
0.31
0.72
0.25
The factor loadings on the various measures of political risk match well with
theoretical expectations. While several measures load on more than one factor, all
measures that load heavily on any factor load most heavily on the measure with which
they are theoretically associated, confirming the appropriateness of the measures in the
study.
Among the variables that load substantially (>0.4) on more than one measure
of risk, the TI measure of corruption loads primarily on bureaucratic risk (0.76), as
expected, but also loads on policy risk (0.41). The ICRG measure of war risk loads
most heavily on War and Violence (0.54), but almost as heavily on bureaucratic risk
(0.53). Court Delays (i.e. the log of the number of days it takes to resolve a contract
dispute through the legal system) and Post-Conflict Status, (i.e. a dummy variable for
!
!
47
whether a country is in the first five years following the end of a war that has not restarted), do not load heavily (>0.4) on any factor, though court delays loads almost
this heavily on the risk of war and violence. This indicates that court delays and postconflict status are probably not good measures of bureaucratic risk and risk of war and
violence. However, results are robust to the exclusion of these measures from the
indices.20
Table 2.2: Pairwise Correlations Between Risk Types
Table 2.2: Pairwise Correlations Between Risk Types
Variables
Bureaucratic Risk Transfer Risk Policy Risk Overthrow Risk War Risk
Bureaucratic Risk
1.000
Transfer Risk
0.393
1.000
Policy Risk
0.031
0.171
1.000
Overthrow Risk
0.132
0.289
0.076
1.000
War Risk
-0.030
0.106
0.046
0.713
1.000
Table 2.2 shows the pairwise correlations between the five types of risk.21
Overall, these five types of risk are less collinear than one might expect, with most of
the correlations at less than 0.3. Risk of adverse regime change (i.e. overthrow risk)
and risk of war and political violence (i.e. war risk) are correlated at 0.7, indicating
that there is some difficulty distinguishing these two concepts empirically. Transfer
risk and bureaucratic risk are correlated at 0.4, indicating some potential for multicollinearity in the analysis for these variables as well.
Table 2.3 gives summary statistics for all variables used in the analysis.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
20
If the results in Table 2 are reproduced using indices based on factor analyses from which these two
variables are excluded, the results are unchanged, with z-scores >4 on Similarity in Bureaucratic Risk in
all models.
21
This anlaysis is conducted on the same sample as the factor analysis.
!
!
48
Table 2.3: Summary Statistics
Table 2.3: Summary Statistics for Dyads with Non-missing FDI Data
Variable
Mean
Std. Dev. Min.
Max.
Bilateral FDI Flows (Billions of USD)
2.785
20.473
-2.961 592.624
Bilateral FDI Flows (Logged)
8.192
9.409
0
27.108
Sim. in Bureaucratic Risk (Index)
-1.085
0.784
-4.1
0
Sim. in Adverse Gov Action (Index)
-0.857
0.856
-3.586
0
Sim. in Violence & Overthrow (Index)
-1.047
0.896
-4.712
0
Sim. in Corruption (TI)
-2.718
1.939
-8.300
0
Sim. in Corruption (ICRG)
-1.402
1.139
-6
0
Sim. in Court Delays (DBI)
-0.547
0.408
-2.263
0
Sim. in Court Costs (DBI)
-17.005
24.097
-154.4
0
Sim. in Import/Export Costs (DBI)
-1380.152 1499.026 -10317
0
Sim. in Import/Export Delays (DBI)
-0.804
0.59
-3.011
0
Sim. in Transfer Risk (ONDD)
-2.567
1.858
-6
0
Sim. in Political Constraints (Henisz)
-0.335
0.285
-0.895
0
Sim. in Executive Constraints (Polity)
-2.144
2.024
-6
0
Sim. in Political Competition (Polity)
-3.171
2.98
-9
0
Sim. in Internal Conflict Risk (ICRG)
-1.672
1.267
-8.5
0
Sim. in War Risk (ONDD)
-1.606
1.469
-6
0
Sim. in Current Conflict (PITF)
-0.203
0.402
-1
0
Sim. in Post-Conflict Status (PITF)
-0.012
0.108
-1
0
Sim. in Political Stability (WGI)
-1.035
0.820
-4.755
0
Sim. in Physical Integrity Violations (CIRI)
-2.518
1.994
-8
0
Sim. in Wealth (GDP Per Capita, Logged)
-13.208
12.842
-103.84 -0.004
Sim. in Educational Attainment (Barro & Lee)
-3.117
2.429
-11.302 -0.003
Bilateral Trade Flows (logged)
3.857
3.124
0
12.974
Common Border
0.029
0.169
0
1
Colonial Relationship
0.03
0.172
0
1
Common Colonizer
0.046
0.209
0
1
Common Legal Origin
0.27
0.444
0
1
Common Language (10% of Pop)
0.129
0.335
0
1
Distance (logged)
8.542
0.913
4.088
9.894
Both Countries in EU (dummy)
0.089
0.285
0
1
N
6159
6055
3805
3805
3805
5651
4438
5720
5720
5720
5720
6040
5293
4969
4969
4438
5873
6159
6159
6126
5790
5780
4511
3070
6011
6011
6011
6011
6011
6011
6159
Specification of the Model Analyzing Home-Country Effects
The effect of similarity in political risk between home and host country is
assessed using an OLS regression with dummy variables for host country ID and home
country ID and robust standard errors. The model estimated takes the form of:
Yij = α i + α j + β1 Xsimij + βk controlsij + ε ij
!
(2.1)
!
49
where Yij is the (logged) FDI flow from country i to country j in 2009; αi denotes a
dummy variable for the home country (from which the FDI originates); αj denotes a
dummy variable for the host country (into which the FDI flows); Xsim is the difference
in the level of political risk between country i and country j in year t; controlsij is a
vector of k covariates, and εij is the error term.
Specification of the Model Analyzing Host-Country Effects
To analyze the effect of previous host-country experience on a firm’s
propensity to enter states with a given level of political risk, I use the mean level of
political risk in the countries a given firm has previously entered to predict the level of
risk in the country currently being entered. The unit of analysis is the firm-entry, and I
compare risk in the country being entered with risk in the other host countries in which
the firm is operating. For entries in year t, I assess the level of risk in previously
entered host countries also in year t. As noted above, the sample is restricted to first
entries by a given firm in a given country. The later addition of new subsidiaries into a
previously-entered country are not considered, as the firm already has country-specific
knowledge for managing political risk in that country.
The model estimated takes the form of:
Riski
) + β2 X 2 + βk X k + ε ijt
n
1→ n
Yit = α + β1 * ( ∑
(2.2)
Yit is the level of political risk in country i, entered by firm j in year t. The
independent variable of interest is the mean level of risk across n host countries that
firm j has entered previously. X2 is the (logged) number of previous entries by firm j,
!
!
50
Xk is a vector of industry dummy variables and εijt is the error term, which is clustered
by firm ID.
An alternative specification involves conditional logit, the same technique used
to model an individual’s choice been modes of transportation. In a dataset of firm
entries, the question is why a firm chooses one country as opposed to any other
country, or more tractably, why a firm chooses to enter a country in one risk category
instead of a country in any other risk category. This alternative analysis is not yet
completed.
Results: Analyzing Home-Country Effects
Table 2.4 shows the effect of similarity in political risk profile between home
and host country and bilateral flows of FDI. Hypotheses 1a predicts a positive
relationship between similarity in bureaucratic risk between home and host countries
and bilateral FDI flows. Hypothesis 2a makes a parallel prediction regarding
similarity in transfer risk. Hypotheses 3a, 4a, and 5a predict no significant
relationship with regard to similarity in policy risk (3a), risk of adverse regime change
(4a) and risk of war and political violence (5a).
Consistent with H1a, similarity in the level of bureaucratic risk across
countries is an important predictor of bilateral FDI across all models. Consistent with
Hypotheses 3a, 4a, and 5a, similarity in policy risk, risk of war and political violence,
and risk of adverse regime change (i.e. overthrow) are not significant predictors of
bilateral FDI flows. Contrary to Hypothesis 2a, however, similarity in transfer risk is
also not a strong predictor of bilateral FDI flows.
!
!
51
Table 2.4: Political Risk Similarity and FD
Table 2.4: Political Risk Similarity and Bilateral FDI
Sim. in Bureaucratic Risk (Index)
(1)
0.824***
(0.183)
Sim. in Policy Risk (Index)
(2)
(3)
(4)
(5)
(6)
0.847***
(0.184)
(7)
1.433***
(0.284)
0.152
(0.279)
0.253
(0.722)
0.037
(0.270)
0.317
(0.457)
-0.141
(0.135)
0.302
(0.272)
0.013
(0.163)
0.201
(0.270)
0.385
(0.442)
0.067
(0.279)
Sim. in War and Pol. Violence (Index)
0.116
(0.198)
Sim. in Transfer Risk (ONDD)
-0.018
(0.118)
Sim. in Adv. Regime Change (WGI)
Distance (logged)
-3.073***
(0.193)
-3.122***
(0.193)
-3.117***
(0.194)
-3.121***
(0.169)
-3.182***
(0.168)
-3.044***
(0.194)
-0.978***
(0.365)
Common Border
-0.502
(0.595)
-0.489
(0.598)
-0.489
(0.598)
-0.319
(0.552)
-0.302
(0.549)
-0.478
(0.594)
0.160
(0.651)
Colonial Relationship
3.324***
(0.532)
3.254***
(0.527)
3.254***
(0.528)
3.413***
(0.517)
3.245***
(0.521)
3.306***
(0.534)
1.807***
(0.591)
Common Colonizer
2.361***
(0.708)
2.218***
(0.709)
2.195***
(0.711)
2.054***
(0.508)
1.985***
(0.506)
2.391***
(0.711)
0.979
(0.956)
Common Language (10% of Pop)
0.664*
(0.376)
0.664*
(0.374)
0.664*
(0.375)
0.569*
(0.318)
0.659**
(0.315)
0.659*
(0.375)
-0.098
(0.485)
Both Countries in EU (dummy)
0.563
(0.479)
0.366
(0.481)
0.344
(0.481)
0.524
(0.449)
0.349
(0.440)
0.603
(0.499)
0.089
(0.655)
Sim. in Wealth
-0.239
(0.152)
0.087
(0.137)
0.083
(0.138)
0.267*
(0.154)
0.228*
(0.128)
-0.182
(0.198)
-0.623*
(0.350)
Sim. in Educational Attainment
0.176**
(0.087)
0.150*
(0.088)
0.141
(0.089)
0.162**
(0.073)
0.168**
(0.073)
0.160*
(0.089)
-0.051
(0.153)
Common Legal Origin
1.111***
(0.326)
Bilateral Trade Flows (logged)
1.646***
(0.165)
Bilateral Migrant Stock (Logged)
0.074
(0.088)
Home and Host Country Dummies
Constant
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
YES
YES
YES
YES
YES
YES
YES
29.252***
(3.663)
3210
0.748
34.011***
(3.695)
3210
0.746
29.630***
(3.504)
3210
0.746
30.366***
(1.893)
4168
0.757
30.922***
(3.305)
4227
0.756
28.767***
(3.756)
3210
0.748
12.633***
(3.302)
1416
0.808
All models are estimated using OLS regression with dummy variable fixed
effects for both home and host countries and Huber/White standard errors. The fixed
effects pick up all country-level characteristics, such as size, wealth, and political
conditions, that affect the volume of FDI flows into a given host country or out of a
given home country. Additional controls include standard gravity model variables
(distance, common language, etc) as well as similarity between home and host
!
!
52
countries in terms of both wealth and educational attainment. These last two controls
are important because similarity in these factors is likely to be correlated both with
greater FDI flows, as firms enter countries with markets similar to their home market,
and because these similarities are correlated with similarities in political risk. Results
are also robust to the inclusion of common legal origin and bilateral trade flows as
controls.22
In substantive terms, a one-standard-deviation (0.78 point) increase in
similarity in bureaucratic risk is associated with a 108% increase in bilateral FDI
flows.23
Bureaucratic Risk: Disaggregated Results
Hypothesis 1a states that similarity between home and host country with regard
to the level of bureaucratic risk is positively correlated with higher flows of bilateral
FDI. Table 2.4 uses an index of bureaucratic risk to show that similarity in
bureaucratic risk between two countries is associated with higher bilateral FDI flows.
Table 2.5 examines the effect of similarity across individual component measures of
bureaucratic risk. The results in Table 2.5 provide evidence consistent with Hypothesis
1a and the findings in Table 2.4. While not all results are statistically significant, many
are, and all estimated effects are positive, i.e. similarity between home and host
country across each individual measure of bureaucratic risk is associated with higher
bilateral FDI flows.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
22
I do not consider these variables as appropriate for inclusion in the main model because common
legal origin overlaps conceptually with similarity in bureaucratic risk (deriving from efficient judicial
resolution of civil disputes) and because political risk affects importers and exporters in ways similar to
how it affects foreign investors (Gillespie 1990).
23
Based on the coefficient in Model 6.
!
Table 2.5: Similarity in Bureaucratic Risk (Disaggregated)
Table 2.5: Similarities in Bureaucratic Risk: Disaggregated
Sim. in Corruption (TI)
(1)
0.113
(0.101)
(2)
0.418***
(0.144)
Sim. in Corruption (ICRG)
(3)
(4)
0.094
(0.136)
0.255
(0.205)
Sim. in Court Delays (DBI)
(5)
(6)
0.494
(0.322)
0.682
(0.437)
Sim. in Court Costs (DBI)
(7)
(8)
0.043***
(0.015)
0.061***
(0.019)
Sim. in Import/Export Costs (DBI)
(9)
(10)
0.330**
(0.149)
0.751***
(0.260)
Sim. in Import/Export Delays (DBI)
(11)
(12)
0.618**
(0.274)
1.326***
(0.417)
Sim. in Policy Risk (Index)
0.056
(0.282)
0.129
(0.739)
0.075
(0.280)
0.168
(0.738)
0.074
(0.280)
0.144
(0.735)
0.120
(0.280)
0.243
(0.740)
0.065
(0.280)
0.100
(0.738)
0.078
(0.279)
0.226
(0.732)
Sim. in War and Pol. Violence (Index)
0.059
(0.273)
0.298
(0.473)
0.039
(0.272)
0.162
(0.472)
0.012
(0.273)
0.126
(0.471)
0.028
(0.272)
0.142
(0.468)
0.001
(0.273)
0.136
(0.466)
0.040
(0.272)
0.186
(0.464)
Sim. in Transfer Risk (ONDD)
-0.150
(0.135)
0.187
(0.270)
-0.155
(0.136)
0.128
(0.271)
-0.143
(0.135)
0.151
(0.271)
-0.136
(0.136)
0.154
(0.269)
-0.171
(0.137)
0.163
(0.273)
-0.155
(0.135)
0.207
(0.272)
Sim. in Political Stability (WGI)
0.063
(0.275)
0.122
(0.459)
0.120
(0.269)
0.447
(0.453)
0.139
(0.269)
0.510
(0.453)
0.153
(0.268)
0.546
(0.449)
0.197
(0.270)
0.514
(0.453)
0.147
(0.269)
0.447
(0.447)
Sim. in Wealth
0.031
(0.229)
-0.453
(0.380)
0.134
(0.200)
0.032
(0.336)
0.177
(0.185)
0.155
(0.316)
0.140
(0.185)
0.075
(0.313)
0.179
(0.185)
0.155
(0.319)
0.024
(0.198)
-0.280
(0.348)
Sim. in Educational Attainment
0.145
(0.090)
-0.088
(0.151)
0.148
(0.090)
-0.069
(0.153)
0.134
(0.090)
-0.092
(0.153)
0.140
(0.090)
-0.120
(0.153)
0.150*
(0.090)
-0.070
(0.154)
0.139
(0.090)
-0.091
(0.152)
Common Legal Origin
1.216***
(0.327)
1.223***
(0.328)
1.209***
(0.327)
1.240***
(0.326)
1.262***
(0.327)
1.178***
(0.328)
Bilateral Trade Flows (logged)
1.599***
(0.166)
1.618***
(0.168)
1.620***
(0.167)
1.640***
(0.167)
1.643***
(0.167)
1.644***
(0.166)
Bilateral Migrant Stock (Logged)
0.102
(0.089)
0.106
(0.089)
0.111
(0.089)
0.111
(0.088)
0.099
(0.088)
0.093
(0.089)
Gravity Variables
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
Home and Host Country Dummies
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
28.924***
(3.658)
3210
0.746
11.329***
(3.346)
1416
0.805
29.243***
(3.581)
3210
0.746
12.549***
(3.340)
1416
0.804
29.384***
(3.702)
3210
0.747
12.394***
(3.330)
1416
0.804
36.621***
(4.057)
3210
0.747
13.024***
(3.320)
1416
0.806
29.644***
(3.617)
3210
0.747
12.809***
(3.352)
1416
0.805
28.935***
(3.664)
3210
0.747
12.378***
(3.334)
1416
0.806
Constant
53
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
!
54
The models estimated in Table 2.5 are similar to those estimated in Table 2.4
The most robust results in this table come from the measures of similarity in
import/export delays and similarity in the costs of resolving civil disputes through the
court system; however, significant results are also obtained for similarity in corruption
(TI) and similarity in import/export delays.
Policy Risk: Disaggregated Results
In contrast to Hypotheses 1a and 2a, Hypothesis 3a predicts no positive
correlation between similarities in policy risk and bilateral FDI flows. The results in
Table 2.6 show positive raw correlations between similarity and policy risk and
bilateral FDI flows, but no robust results.
!
Table 2.6: Similarity in Policy Risk: Disaggregated
Table 2.6: Similarity in Policy Risk: Disaggregated)
Sim. in Political Constraints (Henisz)
(1)
1.201***
(0.357)
(2)
-1.091*
(0.637)
(3)
-1.747
(1.383)
Sim. in Executive Constraints (Polity)
(4)
(5)
(6)
0.102*
(0.052)
-0.023
(0.082)
-0.043
(0.144)
Sim. in Political Competition (Polity)
(7)
(8)
(9)
0.104*
(0.054)
3.257***
(0.536)
0.030
(0.085)
1.840***
(0.591)
Colonial Relationship
3.717***
(0.490)
3.302***
(0.533)
1.717***
(0.587)
3.622***
(0.509)
3.262***
(0.536)
1.818***
(0.589)
0.132***
(0.035)
3.610***
(0.508)
Common Colonizer
2.195***
(0.430)
0.837***
(0.294)
2.302***
(0.711)
0.635*
(0.376)
0.920
(0.956)
-0.074
(0.483)
2.229***
(0.458)
0.771**
(0.301)
2.426***
(0.723)
0.699*
(0.378)
0.887
(0.954)
-0.112
(0.486)
2.211***
(0.457)
0.783***
(0.300)
2.478***
(0.725)
0.745**
(0.378)
0.891
(0.955)
-0.113
(0.487)
-3.036***
(0.166)
0.369
(0.547)
-3.069***
(0.194)
-0.499
(0.593)
-1.005***
(0.366)
0.059
(0.635)
-3.195***
(0.170)
0.394
(0.553)
-3.057***
(0.194)
-0.485
(0.594)
-0.975***
(0.365)
0.187
(0.649)
-3.191***
(0.170)
0.338
(0.552)
-3.035***
(0.194)
-0.514
(0.596)
-0.975***
(0.365)
0.178
(0.650)
1.106***
(0.427)
0.682
(0.497)
0.161
(0.657)
0.747*
(0.437)
0.593
(0.498)
0.005
(0.653)
0.633
(0.433)
0.527
(0.498)
0.003
(0.656)
0.873***
(0.183)
0.015
(0.270)
1.480***
(0.286)
0.267
(0.451)
0.815***
(0.186)
0.006
(0.274)
1.379***
(0.286)
0.192
(0.454)
0.843***
(0.186)
-0.008
(0.275)
1.389***
(0.286)
0.197
(0.457)
-0.125
(0.136)
0.258
(0.272)
0.293
(0.274)
0.527
(0.456)
-0.138
(0.136)
0.240
(0.273)
0.299
(0.276)
0.458
(0.446)
-0.159
(0.136)
0.251
(0.273)
0.299
(0.275)
0.453
(0.446)
Sim. in Wealth
-0.144
(0.198)
-0.594*
(0.354)
-0.170
(0.200)
-0.609*
(0.353)
-0.210
(0.199)
-0.624*
(0.352)
Sim. in Educational Attainment
0.172*
(0.089)
-0.027
(0.154)
1.109***
(0.325)
0.170*
(0.090)
-0.042
(0.155)
1.065***
(0.332)
0.151*
(0.090)
-0.049
(0.154)
1.074***
(0.330)
Common Language (10% of Pop)
Distance (logged)
Common Border
Both Countries in EU (dummy)
Sim. in Bureaucratic Risk (Index)
Sim. in War and Pol. Violence (Index)
Sim. in Transfer Risk (ONDD)
Sim. in Political Stability (WGI)
Common Legal Origin
Bilateral Trade Flows (logged)
1.634***
(0.166)
1.671***
(0.165)
1.670***
(0.165)
Bilateral Migrant Stock (Logged)
0.086
(0.088)
0.077
(0.088)
0.076
(0.088)
Home and Host Country Dummies
Constant
YES
28.996***
(3.626)
3210
0.748
YES
12.678***
(3.289)
1416
0.809
YES
27.289***
(2.310)
4806
0.748
YES
29.053***
(3.633)
3172
0.747
YES
5.489
(3.476)
1403
0.807
YES
27.656***
(2.300)
4806
0.748
YES
33.377***
(3.822)
3172
0.747
YES
5.485
(3.477)
1403
0.807
55
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
YES
29.582***
(1.927)
5140
0.749
!
56
While all three of the measures of similarity in policy risk are strongly
correlated with bilateral FDI flows, these correlations are not robust to the inclusion of
additional measures of risk similarity. This suggests that the correlations are likely
spurious, arising only because of the correlation between policy risk and other risk
types, such as bureaucratic risk, and not because of any independent effect of
similarity in policy risk on FDI. These results are not, in and of themselves, grounds
on which I can firmly reject the null hypothesis that similarity in policy risk is
positively correlated with bilateral FDI flows. However, these results are consistent
with Hypothesis 2a, and lend credibility to the theory.
Risk of War and Political Violence: Disaggregated Results
Hypothesis 5a predicts that similarity between home and host country in the
risks of war and political violence have no clear relationship with bilateral FDI flows.
In Table 2.3, I show evidence consistent with this predication. In Table 2.7, I show
disaggregated results for the component measures and again find no evidence of a
positive relationship.
!
Table 2.7: Similarity in the Risk of War and Political Violence (Disaggregated)
Sim. in Internal Conflict Risk (ICRG)
(1)
0.115
(0.085)
(2)
0.143
(0.196)
Sim. in War Risk (ONDD)
(3)
(4)
-0.033
(0.089)
0.323
(0.333)
Sim. in Current Conflict
(5)
(6)
-0.283
(0.377)
0.595
(1.266)
Sim. in Recent War
(7)
(8)
-0.262
(0.355)
0.885
(1.070)
Sim. in Physical Integrity Violations
(9)
(10)
-0.022
(0.042)
-0.065
(0.115)
Common Legal Origin
1.082***
(0.327)
1.133***
(0.325)
1.119***
(0.326)
1.128***
(0.327)
1.168***
(0.332)
Sim. in Bureaucratic Risk (Index)
1.412***
(0.285)
1.441***
(0.286)
1.413***
(0.284)
1.416***
(0.284)
1.450***
(0.289)
Sim. in Policy Risk (Index)
0.219
(0.729)
0.123
(0.731)
0.289
(0.724)
0.359
(0.730)
0.204
(0.738)
Sim. in Transfer Risk (ONDD)
0.321
(0.272)
0.315
(0.273)
0.288
(0.276)
0.277
(0.275)
0.377
(0.279)
Sim. in Political Stability (WGI)
0.446
(0.412)
0.436
(0.398)
0.577*
(0.327)
0.525
(0.332)
0.617*
(0.344)
Sim. in Wealth
-0.625*
(0.348)
-0.706**
(0.355)
-0.621*
(0.352)
-0.607*
(0.351)
-0.730**
(0.350)
Sim. in Educational Attainment
-0.055
(0.153)
-0.052
(0.154)
-0.042
(0.155)
-0.039
(0.153)
-0.007
(0.154)
Bilateral Trade Flows (logged)
1.652***
(0.165)
1.672***
(0.168)
1.652***
(0.165)
1.650***
(0.165)
1.611***
(0.167)
Bilateral Migrant Stock (Logged)
0.074
(0.089)
0.069
(0.088)
0.072
(0.088)
0.072
(0.088)
0.072
(0.089)
Gravity Variables
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
Home and Host Country Dummies
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
28.646***
(3.402)
4348
0.742
12.891***
(3.290)
1416
0.808
30.253***
(3.456)
5711
0.746
5.083
(3.632)
1416
0.808
31.831***
(3.534)
5926
0.745
12.759***
(3.298)
1416
0.808
31.840***
(3.533)
5926
0.745
12.732***
(3.298)
1416
0.808
26.392***
(2.712)
5537
0.747
26.688***
(2.720)
1390
0.809
Constant
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
57
!
58
Across a range of measures of risk of war and political violence, similarity
between home and host country is not a significant predictor of bilateral FDI flows,
and we see a mixture of positive and negative signs. This provides evidence consistent
with the prediction of no correlation.
Results: Analyzing Host-Country Effects
Three of the predictions regarding host-country effects mirror those of homecountry effects while two conflict. Bureaucratic risk and transfer risk are expected to
have both home and host country effects, while the risk of adverse regime change is
expected to have neither home nor host country effects. Policy risks are expected to
have host country effects, but not home country effects. Policy risks are expected to
affect foreign firms, who are not constituents of the regime in the host country, to a
much greater degree than domestic firms, who wield greater influence over the
government and are less often the subject of government expropriation. Risk of war
and political violence is expected to have both home and host country effects, but
because of the offsetting effects of capital flight discussed above, I predict that homecountry effects are offset and not observed. Therefore, the expectation is that only
host-country effects are observed in the data.
I will analyze the host-country effects of each category of risk separately.
Bureaucratic Risk
Hypothesis 1b predicts that the mean value of bureaucratic risk in the host
countries in which a firm is already operating is a good predictor of the bureaucratic
risk in the host countries that firm will enter next. Therefore, I expect to find that the
!
!
59
mean value of corruption in previously-entered host countryis predicts the level of
corruption in the host country currently being entered. I do not find a strong effect;
the analysis presented in Table 2.8 fails to find evidence in support of Hypothesis 1b.
Table 2.8: Bureaucratic Risk (ICRG Corruption): Firm Level Data
Table 2.8: Bureaucratic Risk (ICRG Corruption): Firm Level Data
(1)
(2)
(3)
Corruption (ICRG) in Prior HCs
-0.000
0.043
0.113
(0.049)
(0.076)
(0.076)
Number of Prior Foreign Entries (logged)
-0.012
(0.053)
-0.020
(0.058)
-0.036
(0.062)
Wealth of Prior HCs
-0.118
(0.125)
-0.205
(0.145)
Education in Prior HCs
0.035
(0.043)
0.063
(0.043)
Political Constraints in Prior HCs
-0.950*
(0.491)
Current War in Prior HCs
-1.526***
(0.504)
Industry Dummies YES
Constant
Observations
R2
YES
YES
3.473***
(0.397)
732
0.027
3.856***
(0.817)
698
0.034
4.879***
(0.937)
698
0.055
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Only one of the component measures of bureaucratic risk included in the index
in Table 2.3 is analyzed here because of limitations on data availability. Even the
measure that is used, the ICRG measure of corruption, covers only half of the countryyears in which firm entries occur. Most of the firm entries in the dataset occurred in
!
!
60
the 1970s and 1980s, years for which the better measures of bureaucratic risk are not
available and in which cross-sectional coverage of the ICRG measure is limited. The
next step in this research is to test this hypothesis on more recent firm entry data,
allowing for a more definitive test of the theory. The results in Table 2.8 do not allow
us to reject the null hypothesis of no host-country effects for bureaucratic risk.
Transfer Risk
Due to a lack of data on transfer risk prior to 1992, Hypothesis 2b cannot be
tested at this time.
Policy Risk
In contrast with Hypothesis 3a, which predicts no home-country effects for
policy risk, Hypothesis 3b predicts that previous experience managing policy risk in
other host countries should increase firms’ willingness to enter future host countries
with high levels of policy risk (though these effects are not expected to be as strong as
those for bureaucratic risk).
The results presented in Table 2.9 show no evidence of robust host country
effects for policy risk. Models 1, 4 and 7 show positive and significant host country
effects when the level of experience with bureaucratic risk (i.e. corruption) is not
controlled for. However, these results disappear, and in fact reverse sign, when
corruption is controlled for. This suggests that previous findings of learning effects
with regard to policy risk are likely driven by the confounding correlation between
policy risk and bureaucratic risk, and not by the independent effect of policy risk.
!
Table 2.9: Policy Risk: Firm Level Data
Political Constraints in Prior HCs
(1)
DV=PolCon
0.228***
(0.033)
(2)
DV=PolCon
0.137**
(0.058)
(3)
DV=PolCon
-0.203**
(0.095)
Executive Constraints in Prior HCs
(4)
DV=XConst
(5)
DV=XConst
(6)
DV=XConst
0.132***
(0.031)
0.040
(0.066)
-0.030
(0.087)
Political Competition in Prior HCs
# of Prior Foreign Entries (logged)
0.009
(0.008)
0.011
(0.008)
-0.035***
(0.011)
Wealth of Prior HCs
0.020
(0.015)
Education in Prior HCs
0.003
(0.007)
0.149***
(0.056)
(7)
DV=PolComp
(8)
DV=PolComp
(9)
DV=PolComp
0.181***
(0.031)
0.075
(0.058)
-0.001
(0.071)
0.128
(0.084)
0.183**
(0.086)
-0.149
(0.108)
0.187***
(0.056)
0.095
(0.070)
-0.040
(0.027)
0.185*
(0.103)
-0.488***
(0.171)
0.305*
(0.157)
-0.454
(0.278)
0.008
(0.008)
0.016
(0.052)
0.039
(0.060)
0.039
(0.081)
0.024
(0.096)
Corruption (ICRG) in Prior HCs
0.033*
(0.017)
0.130
(0.113)
0.064
(0.167)
Current War in Prior HCs
-0.172*
(0.098)
-2.998***
(0.722)
-2.881***
(1.079)
Industry Dummies
Constant
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
YES
YES
YES
YES
YES
YES
YES
YES
YES
0.415***
(0.077)
1338
0.043
0.166
(0.127)
1247
0.040
0.818***
(0.159)
689
0.058
3.698***
(0.641)
1334
0.024
2.187***
(0.817)
1239
0.030
8.058***
(1.129)
682
0.071
5.089***
(0.851)
1334
0.031
2.518**
(1.143)
1239
0.033
10.341***
(1.836)
682
0.039
61#
!
62
Risk of Adverse Regime Change
Due to a lack of data on political stability (WGI) prior to 1996, Hypothesis 5b,
which predicts that experience in managing the risks associated with regime overthrow
does not predispose firms to enter other similarly risky countries, is not tested at this
time.
Risks of War and Political Violence
The results regarding war and political violence are mixed. Firms with
investments in countries currently at war are more likely to invest in future host
countries that are also at war (Models 1 and 2); however, this does not seem to apply
to post-conflict countries (Models 3 and 4), and the average war risk score of prior
host countries is not a good predictor of the war risk of future entries (Models 5 and
6).
!
!
!
!
63
Table 2.10: Risk of War and Political Violence: Firm Level Data
Table 2.10: Risk of War and Political Violence: Firm Level Data
Current War in Prior HCs
(1)
DV=War
2.970***
(0.640)
(2)
DV=War
2.263*
(1.303)
(3)
DV=PostWar
(4)
DV=PostWar
-2.626
(2.534)
-3.206
(3.724)
Postconflict Status in Prior HCs
War Risk in Prior HCs
(5)
DV=WarRisk
(6)
DV=WarRisk
-0.082
(0.064)
-0.065
(0.074)
# of Prior Foreign Entries (logged)
0.012
(0.094)
0.379**
(0.155)
0.024
(0.101)
0.007
(0.173)
-0.043
(0.128)
-0.060
(0.138)
Wealth of Prior HCs
0.047
(0.177)
0.107
(0.340)
0.392**
(0.183)
0.687*
(0.372)
-0.047
(0.251)
-0.095
(0.308)
Education in Prior HCs
-0.069
(0.066)
0.024
(0.110)
-0.149**
(0.066)
-0.203**
(0.099)
-0.108
(0.094)
-0.104
(0.093)
Corruption (ICRG) in Prior HCs
-0.688**
(0.290)
-0.074
(0.235)
0.118
(0.234)
Political Constraints in Prior HCs
3.415*
(1.887)
-0.665
(1.020)
-0.143
(1.073)
Industry Dummies
Constant
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
YES
YES
YES
YES
YES
YES
-1.729
(1.439)
1301
-4.391*
(2.375)
714
-4.364***
(1.488)
1301
-5.687**
(2.835)
692
-18.607***
(1.621)
698
0.036
-18.208***
(1.933)
698
0.037
The correlation between the proportion of previously entered host countries in
which there is ongoing war and the probability that the host country currently being
entered is experiencing war is strong (Models 1 and 2). This suggests that firms who
are currently operating in war-torn host countries are more likely to enter additional
war-torn host-countries. There is no evidence of similar host-country effects for
expert assessments of war risk (Models 5 and 6) or post-conflict status (Models 3 and
4). One interpretation consistent with this evidence is that operating in countries at risk
of future war or in economies recovering from war does not lead firms to invest in
capabilities that allow them to manage war risk (and does not eliminate war-
!
!
!
!
64
vulnerable firms). Only operating in countries actually experiencing war induces these
learning and selection effects.
These results are consistent with Hypothesis 5b, which predicts host-country
effects for risk of war and political violence, but are not conclusive.
Interpreting Results
Taken together, these results are supportive of the theory. First, the factor
analysis provides empirical evidence supportive of the theoretical distinction between
the five categories of political risk.24 Second, consistent with the theory, I find strong
evidence of home-country effects for bureaucratic risk, and no evidence of homecountry effects for policy risk, risk of adverse regime change, and risk of war and
political violence. The host-country results are sharply limited by lack of data, but do
provide evidence consistent with host-country effects for risk of war and political
violence (regarding current war specifically).
However, three hypotheses are unsupported by the data: I fail to find evidence
of expected home-country effects for transfer risk or of expected host-country effects
for policy risk (which are expected to be weak) and bureaucratic risk (which are
expected to be strong). The failure to find host-country effects for bureaucratic risk
may be, in part, due to a lack of data on key measures of bureaucratic risk, and further
testing is necessary before this hypothesis is rejected. Similarly, the number of
observations is limited on which I could run a fully specified model of host-country
effects for policy risk.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
24
This analysis could not speak to the uniqueness of transfer risk and risk over adverse regime change
specifically, given that each is captured by only a single measure.
!
!
!
!
65
The failure to find evidence of home-country effects for transfer risk is
somewhat more damning. It seems likely that transfer risk may, in fact, affect firms
very differently in home and host countries, and that the skills developed to manage
transfer risk in the home country are not transferable to host-countries. However,
prior to embarking on a revision of theory, it is critical to test for the presence of
expected host-country effects for transfer risk, one of two hypotheses that is not tested
at this time due to a lack of data.
Conclusion and Next Steps
The results presented here are broadly consistent with the theory. I present a
five-part typology of political risks and show that, for the three types of risk for which
I have multiple measures, these measures cluster appropriately around the risk types
with which they are theoretically associated. I am also find empirical support for
several key hypotheses regarding the ability of firms to employ risk management
capabilities developed in one country to manage similar risks in other countries. These
findings constitute important progress in understanding both variations across firms in
their ability to manage political risk, and variation across types of political risk in their
amenability to learned management.
However, the firm level tests are limited by the age of the firm-level data; most
of the firm entries in the dataset occur during years for which detailed political risk
data is not available. More thorough testing of the hypotheses requires more recent
firm level data. I have recently passed the penultimate hurdle in accessing data from
the Bureau of Economic Analysis on foreign direct investment by US Multinationals
!
!
!
!
66
abroad. The US is the world’s top supplier of FDI to fragile states (by volume), and
this firm-level data should provide an excellent means for testing the hypotheses
regarding the effect of prior host-country experience on firms’ willingness to enter
fragile states. I am also exploring other firm-level data options, including the OSIRIS
database from Bureau Van Dijk. This additional testing is necessary before the
findings presented here are sufficiently strong to justify more definitive statements
regarding their theoretical and policy implications.
!
!
!
!
67
References
Ahlquist, John. 2006. Economic Policy, Institutions, and Capital Flows: Portfolio and
Direct Investment Flows in Developing Countries. International Studies
Quarterly (50): 681-704.
Allee, Todd, and Clint Peinhardt. 2010. Delegating Differences: Bilateral Investment
Treaties and Bargaining Over Dispute Resolution Provisions. International
Studies Quarterly 54 (1): 1-26.
Baker, Pauline. 2006. Conflict Assesstment System Tool (CAST): An Analytical Model
for Early Warning and Risk Assessment of Weak and Failing States.
Washington DC: The Fund For Peace.
Barro, Robert J, and Jong-Wha Lee. 2010. "A New Data Set of Educational
Attainment in the World, 1950-2010." 15902. National Bureau of Economic
Research.
Belderbos, Renè, and Leo Sleuwaegen. 1996. Japanese Firms and the Decision to
Invest Abroad: Business Groups and Regional Core Networks. The Review of
Economics and Statistics 78 (2): 214-220.
Bénassy Quéré, A, M Coupet, and T Mayer. 2007. Institutional determinants of
foreign direct investment. The World Economy 30 (5): 764-782.
Büthe, Tim, and Helen V. Milner. 2008. The Politics of Foreign Direct Investment
into Developing Countries: Increasing FDI through International Trade
Agreements? American Journal of Political Science 52 (4): 741-762.
Chang, Sea Jin. 1995. International Expansion Strategy of Japanese Firms: Capability
Building through Sequential Entry. The Academy of Management Journal 38
(2): 383-407.
Collier, Paul. 1999. On the Economic Consequences of Civil War. Oxford Economic
Papers 51 (1): 168-183.
Cosset, Jean-Claude, and Jean-Marc Suret. 1995. Political Risk and the Benefits of
International Portfolio Diversification. Journal of International Business
Studies 26 (2): 301-318.
Cuervo-Cazurra, Alvaro. 2006. Who Cares about Corruption? Journal of International
Business Studies 37 (6): 807-822.
!
!
!
!
68
Cuypers, Ilya R. P., and Xavier Martin. 2009. What makes and what does not make a
real option[quest] A study of equity shares in international joint ventures. J Int
Bus Stud 41 (1): 47-69.
Delios, Andrew, and Witold J. Henisz. 2000. Japanese Firms' Investment Strategies in
Emerging Economies. Academy of Management Journal 43 (3): 305-323.
Ghemawat, Pankaj. 2004. Distance Still Matters: The Hard Reality of Global
Expansion. Harvard Business Review.
Ghobarah, Hazem Adam, Paul Huth, and Bruce Russett. 2003. Civil Wars Kill and
Maim People-Long after the Shooting Stops. The American Political Science
Review 97 (2): 189-202.
Gillespie, Kate. 1990. US Corporations and Iran at the Hague. Middle East Journal 44
(1): 18-36.
Goldstone, Jack A., Robert H. Bates, David L. Epstein, Ted Robert Gurr, Michael B.
Lustik, Monty G. Marshall, Jay Ulfelder, and Mark Woodward. 2010. A
Global Model for Forecasting Political Instability. American Journal of
Political Science 54 (1): 190-208.
Graham, Benjamin A.T. 2010. "Political Risk and Diaspora Direct Investment."
Presented at the Annual Meeting of the American Political Science
Association, Washington, DC.
Graham, Benjamin A.T., Allison F. Kingsley, and Noel P. Johnston. 2012. "The
Political Risk of Repatriating Profits: How Transfer Risk Affects Foreign
Investment."
Guidolin, Massimo, and Eliana La Ferrara. 2007. Diamonds Are Forever, Wars Are
Not: Is Conflict Bad for Private Firms? The American Economic Review 97
(5): 1978-1993.
Habib, Mohsin, and Leon Zurawicki. 2002. Corruption and Foreign Direct Investment.
Journal of International Business Studies 33 (2): 291-307.
Head, Keith, Thierry Mayer, and John Ries. 2010. The Erosion of Colonial Trade
Linkages After Independence. Journal of International Economics 81 (1): 114.
Henisz, Witold J. 2000. The Institutional Environment for Multinational Investment.
Journal of Law and Economic Organization 16 (2): 334-364.
!
!
!
!
69
Holburn, Guy L. F., and Bennet A. Zelner. 2010. Political Capabilities, Policy Risk,
and International Investment Strategy: Evidence From the Global Electric
Power Generation Industry. Strategic Management Journal 31 (12): 12901315.
Javorcik, Beata Smarzynska. 2004. Does Foreign Direct Investment Increase the
Productivity of Domestic Firms? In Search of Spillovers through Backward
Linkages. The American Economic Review 94 (3): 605-627.
Jensen, Nathan. 2008. Political Risk, Democratic Institutions, and Foreign Direct
Investment. The Journal of Politics 70 (04): 1040-1052.
Jensen, Nathan, and Fiona McGillivray. 2005. Federal Institutions and Multinational
Investors: Federalism, Government Credibility, and Foreign Direct Investment.
International Interactions: Empirical and Theoretical Research in
International Relations 31 (4): 303 - 325.
Kingsley, Allison F., and Thomas G. Noordewier. 2011. "Microdeterminants of
Foreign Direct Investment into Developing Countries: Evidence from the
Telecom Sector." American Political Science Annual Meeting.
Li, Jing, and Alan M. Rugman. 2007. Real Options and the Theory of Foreign Direct
Investment. International Business Review 16 (6): 687-712.
Li, Quan. 2009. Outlier, Measurement, and the Democracy-FDI Controversy.
Quantitative Journal of Political Science (4): 167-181.
Li, Quan, and Adam Resnick. 2003. Reversal of Fortunes: Democratic Institutions and
Foreign Direct Investment Inflows to Developing Countries. International
Organization (57): 175-211.
Li, Quan, and Tatiana Vashchilko. 2010. Dyadic Military Conflict, Security Alliances,
and Bilateral FDI Flows. Journal of International Business Studies 41 (5): 765782.
Markus, Stanislav. 2012. Secure Property as a Bottom-Up Process: Firms,
Stakeholders, and Predators in Weak States. World Politics 64 (02): 242-277.
Noordewier, Thomas G., George John, and John R. Nevin. 1990. Performance
Outcomes of Purchasing Arrangements in Industrial Buyer-Vendor
Relationships. The Journal of Marketing 54 (4): 80-93.
!
!
!
!
70
Pfefferman, Guy P., Gregory V. Kisunko, and Mariusz A. Sumlinki. 1999. "Trends in
Private Investment in Developing Countries and Perceived Obstacles to Doing
Business." Discussion Paper No. 37. World Bank.
Porta, Rafael La, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2008. The
Economic Consequences of Legal Origins. Journal of Economic Literature 46
(2): 285-332.
Slangen, Arjen H. L., and Rob J. M. van Tulder. 2009. Cultural Distance, Political
Risk, or Governance Quality? Towards a More Accurate Conceptualization
and Measurement of External Uncertainty in Foreign Entry Mode Research.
International Business Review 18 (3): 276-291.
UNCTAD. 2011. Foreign Direct Investment in LDCs: Lessons Learned From the
Decade 2001-2010 and the Way Forward. New York, New York: United
Nations Commission on Trade and Development.
!
!
!
!
Chapter 3:
Political Risk and Diaspora Direct Investment
Abstract
I argue that diasporans are better informed about the political and economic
situation in the homeland than are other potential foreign investors; that higher levels
of information allows diaspora investors to anticipate and respond to changes in risk
more effectively; and that this makes migrant-induced foreign direct investment (FDI)
more sensitive to political risk than is other FDI. I conduct empirical analysis on a
dataset of bilateral FDI flows from 30 OECD countries to 105 developing countries
from 1994-2008: I find that migrant-induced FDI is more sensitive to political risk
than is other FDI, and particularly that migrant-induced FDI is more sensitive to
decreases in political risk. This has three central implications. First, diasporans’ access
to information is a major factor driving the causal relationship between migrant stocks
and FDI flows. Second, the effect of political risk on investment flows depends on the
mix of investor types (particularly diaspora vs. non-diaspora) in the pool of potential
investors from which a state is drawing. Third, diaspora direct investment has the
potential to motivate and sustain governance reform and stabilization in fragile states,
but it is no more likely than other FDI to substitute for, or precede, these
improvements.
!
!
71!
!
72!
Introduction
A causal link between bilateral migrant stocks and bilateral flows of foreign
direct investment (FDI) has been established empirically (Docquier and Lodigiani
2010; Javorcik et al. 2011; Kugler and Rapoport 2007; Leblang 2010), and the
amounts in question are not small. By some estimates, diaspora direct investment (i.e.
investment by migrants and their descendents) accounted for over 50% of FDI inflows
to China during the 1990s (Huang 2003) and 20-30% of FDI flows into India during
the same time period (Ye 2010).1 A related literature has also shown that at least some
subset of diaspora investors are socially and emotionally motivated (Gillespie, Sayre,
and Riddle 2001; Nielsen and Riddle 2010). This raises the possibility that diasporans
are capable of providing capital to their troubled homelands at times of high political
risk when other foreign investors are scarce.
I argue that diaspora investors have access to privileged channels of
information via social networks and face lower costs of monitoring publicly available
information sources. Higher levels of information cause diaspora investors to respond
both more strongly and more quickly to over-time changes in the level of political risk
than do other foreign investors. Consistent with this theory, I present evidence that
FDI flows in high-migrant dyads are more sensitive to over-time fluctuations in
political risk than are flows in low migrant dyads, particularly with regard to decreases
in political risk, i.e. stabilizations, conflict terminations and institutional
improvements.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
In the Chinese case, these estimates include a substantial amount of “round-tripping” to Taiwan and
Hong Kong, but the impact of diaspora investment remains substantial.
!
!
!
!
73!
In the current economics and political science literature, political risk is
generally treated as a country-level variable, ignoring variation across types of firms
(e.g. Busse and Hefeker 2007; Jensen 2008; Kolstad and Villanger 2007). I argue that
diasporans differ from other foreign investors in terms of both their access to private
information and their attention to publicly available information, both of which affect
their response to political risk. If diaspora direct investment responds differently to
political risk than other types of FDI, then the effect of political risk on investment
flows is moderated by the mix of investors types (i.e. diaspora vs. non-diaspora) in the
pool of potential investors from which a country is drawing.
In the empirical sections of this paper, I present evidence consistent with my
argument that diaspora direct investment is more sensitive to variations in the level of
political risk than is other FDI. These findings imply that, while diaspora direct
investment is likely not a reliable source of capital for the world’s most fragile states,
it can serve both as a complement to, and as an incentive for, improvements in
governance: diaspora investors will be among the investors with the strongest response
to governance improvements, giving diaspora investment the potential to motivate and
sustain important reforms, and contribute to the overall stabilization and growth of the
economy.
The paper is organized as follows: First, a brief review of the relevant literature
including the theory and evidence in economic sociology that supports the hypothesis
that diaspora-induced FDI is less sensitive to political risk than is other FDI. I then
present a theoretical argument for why I expect diaspora direct investment to be more
!
!
!
!
74!
sensitive to political risk than other FDI, and proceed to test these competing argument
empirically using data on dyadic (pairwise) FDI flows from 30 OECD countries to 105
developing countries from 1994-2008. This is followed by a discussion of theoretical
implications, policy implications and conclusions.
The Current Literature
The empirical literature on political risk and foreign investment generally
treats political risk as a country-level variable, and ignores important variation
between types of foreign firms. This literature finds that political risk in a developing
country negatively affects its ability to attract FDI (Schneider and Frey 1985;
Globerman and Shapiro 2003; Busse and Hefeker 2007; Desbordes 2009), and
identifies a variety of risk-reducing institutional characteristics and arrangements that
help attract FDI to developing countries.2 These include democracy (Li and Resnick
2003; Jensen 2008);3 constraints on the executive (Henisz 2000); federalism (Jensen
and McGillivray 2005); bilateral investment treaties (Elkins, Guzman, and Simmons
2006; Kerner 2009; Allee and Peinhardt 2010); and multilateral and bilateral trade
agreements (Büthe and Milner 2008).
I argue, however, that the tacit assumption that political risk affects all types of
foreign firms equally is a faulty and costly one. If different types of firms vary in their
ability to manage political risk, then these types firms can be expected to vary in their
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2
Bevan and Estrin (2004) and Sethi et al. (2003) do not find a relationship between political risk and
FDI.
3
ONeal (1994) fails to find a relationship between regime type and FDI inflows, and Choi and Samy
(2008) suggest that the relationship is weak if it exists at all.
!
!
!
!
75!
sensitivity to risk. Diaspora-owned firms are one type of firm that I expect to
experience and respond to political risk differently than other foreign firms.
The economic sociology literature was the first to argue that diasporans may
respond differently to political risk than other foreign firms. This literature argues that
investment in the homeland by a member of the diaspora is not just an economic act,
but an emotional, social, and political act (Bandelj 2008). Diasporans may invest for
social reasons, such as raising their social standing in the diaspora community in the
country of residence, or for emotional reasons, such as patriotism (Aharoni 1960;
Schulte 2008; Nielsen and Riddle 2010). These motivations may be sufficient to
induce diasporans to accept a lower risk-adjusted rate of return in their homeland than
they would elsewhere, in turn making their investment decisions less dependent on
factors like political risk that moderate the expected rate of return. This theory of nonpecuniary motivation and risk insensitivity has catalyzed substantial interest in the
policy community, inducing a 2009 report written for USAID to refer to the “bravery
and resilience” of diaspora direct investment (Debass and Ardovino 2009: p. 17).
Survey work with diaspora populations in the US confirms that some
diasporans are motivated to invest in their homeland by non-pecuniary incentives,
such as emotional ties to the homeland and social ties to the diaspora community in
the country of residence (Gillespie et al. 1999; Nielsen and Riddle 2010). A
longitudinal survey of Palestinians living in the United States even found evidence
that diaspora interest in homeland direct investment does not decrease when
diasporans’ perceptions of political and economic risk increases (Gillespie, Sayre and
!
!
!
!
76!
Riddle 2001). These surveys use diasporans’ interest in investing as the dependent
variable, rather than actual investment behavior. Therefore, one of the important
empirical contributions of this paper is to test theories of diaspora difference using
data on actual investment flows.
While this literature deserves credit for theorizing about variations in risk
sensitivity across firm types, it overlooks the implications of key elements of diaspora
difference affecting investor capabilities. In particular, while diasporans’ levels of
information about the homeland are theorized to be an important cause of diaspora
direct investment (Leblang 2010; Javorcik et al. 2011; Doquier and Lodigiani 2010),
the implications of high levels of information for political risk sensitivity are never
considered. In the following section, I demonstrate that, while motivation-based
theories of diaspora difference predict that diaspora direct investment is less sensitive
to political risk than is other FDI, informational advantages imply the opposite. In later
sections, I conduct an empirical test that allows us to pit the predictions of this theory
against those derived from a theory of diaspora difference based on motivations.
Theory
Incomplete and asymmetric information is a major cause of the international
immobility of capital and an impediment to international trade (Gordon and
Bovenberg 1996, Portes and Rey 2005).4 The existing literature on diaspora
investment argues the diaspora investors enjoy a competitive advantage in their
homeland that derives from their social networks, their linguistic abilities, and their
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4
For a review of the literature on “home bias” in investing, see Lewis 1999.
!
!
!
!
77!
cultural familiarity (Javorcik et al. 2011, Leblang 2009). I accept this general premise
and argue more specifically that diasporans have two informational advantages over
other foreign investors: 1) diasporans’ have access to social networks that provide
privileged channels of information; 2) diasporans have language skills, cultural
literacy, and non-investment related reasons for monitoring political, social, and
economic conditions in the country-of-origin, and therefore face lower costs to
maintaining a high level of knowledge about these conditions.
Social Networks and Access to Information
Social networks alleviate cross-border information asymmetry by increasing
both the quantity and the quality of information that firms have access to (Uzzi 1996,
1999; Yli-Renko, Autio and Sapienza 2001). Transnational ethnic and social networks
can provide privileged channels of information that allow efficient international
matching between potential business collaborators and between capital and
opportunity (Rauch and Casella 2003). Diasporans, and particularly first-generation
migrants, retain access to social networks in their homelands that other potential
foreign investors generally do not have. Especially when members of diasporans’
social networks are directly involved in business or politics, they may have private
information about conditions in the homeland and provide this information to
members of the network.
Attention to Publicly Available Information
Some of the informational advantages enjoyed by diasporans have nothing to
do with private or privileged access to information, but rather with greater attention to
!
!
!
!
78!
publicly available information. Most foreign firms actively monitor publicly available
information about political risk in the countries in which they already have
investments. However, firms may not always choose to bear the cost of closely
monitoring conditions in countries where they could potentially make profitable new
investments. The list of countries in which profitable opportunities could arise is long,
and the cost of monitoring political and economic conditions in each country, even
when relying exclusively on publicly available information, is not trivial. Diasporans,
however, may have an interest in monitoring information about the homeland because
of the friends and family they have living in the homeland, or because of patriotism or
emotional connection to the homeland.
Language skills and cultural familiarity are also expected to lower diasporans’
costs of accessing and interpreting publicly available information, particularly
information in the local language of the homeland. With both lower costs of acquiring
publicly available information and greater non-investment related motivations for
doing so, diasporans are likely to acquire more publicly available information about
their homeland than other potential foreign investors.
From Information to Investment: Hypotheses
FDI flows respond to the level of political risk in the receiving country,
increasing when political risk drops and decreasing when political risk rises. However,
not all investors respond to fluctuations in political risk to the same degree or with the
same speed. I expect that investors with more information respond to changes in
political risk both more strongly and more quickly than do investors with less
!
!
!
!
79!
information. If diasporans are better informed, they should have greater ability to
identify and exploit profitable business opportunities in the homeland and to anticipate
risks that may arise there.
Changes in the level of political risk in a country either close off or open up
profitable investment opportunities. The more information an investor has, the more
likely she is aware that a particular investment opportunity has just become more or
less attractive, and the more likely she will increase or decrease investment in
response to the change in risk. Therefore, I expect better-informed investors to
respond more strongly to changes in the level of political risk than less-informed
investors. Because diaspora investors have accessed to privileged channels of
information via social networks, and because they face lower costs of acquiring
publicly available information about the homeland, I expect migrant-induced FDI to
be more sensitive than other FDI to changes in political risk.
Hypothesis 1: Migrant-induced FDI flows respond more strongly to changes in the
level of political risk than do other FDI flows.
If we find evidence that diaspora direct investment is more political-risk
sensitive than other FDI, it becomes important to understand whether this greater
sensitivity applies to both increases and decreases in political risk. Investment flows
that drop dramatically in response to a rise in political risk increase economic
volatility and can turn political crises quickly into economic crises, begetting a vicious
!
!
!
!
80!
cycle in which instability begets capital flight, which begets economic crisis, which
begets instability. Conversely, economic flows that rise quickly in response to
decreases in political risk can generate a virtuous cycle in which political reforms
begets an increase in investment, which begets economic growth and stabilization of
the post-reform political order. I expect that better-informed investors will both be
more likely to reduce their investments when risk increases as well as to increase
investment when risk falls.
Hypothesis 2A: Migrant-induced FDI flows respond more strongly to increases in the
level of political risk than do other FDI flows.
Hypothesis 2B: Migrant-induced FDI flows respond more strongly to decreases in the
level of political risk than do other FDI flows.
Social and Emotional Motivations for Investment
Hypotheses 1, which is derived from a theory of diaspora difference based on
access to information, makes a prediction diametrically opposed to the predictions that
emerge from the theory of diaspora difference based on motivation in the economic
sociology literature. This allows us to test a central implication of each of these
theories simultaneously simply by evaluating the risk sensitivity of migrant-induced
FDI to political risk. If Hypothesis 1 is correct and migrant-induced investment is
more sensitive to political risk than is other FDI, it provides evidence consistent with a
!
!
!
!
81!
theory of diaspora difference based on access to information and in opposition to a
central prediction derived from theories of diaspora difference based on investor
motivations.
Empirical Strategy
The central empirical task of this paper is to establish whether diaspora
investors (and therefore migrant-induced FDI) are more or less sensitive to political
risk than are other foreign investors. This is made challenging by the fact that data on
FDI flows do not distinguish investments made by diasporans from investments made
by non-diasporans. Because I cannot compare flows of diaspora and non-diaspora FDI
directly, I compare FDI flows in country pairs in which migrants from a given
homeland (investment-receiving country) make up varying size shares of the
population in the country-of-residence (investment-sending country). If diaspora
investors are more sensitive to political risk than other investors, and if diaspora direct
investment is a substantial portion of overall FDI flows in dyads with large bilateral
migrant stocks, I should observe that FDI flows are more sensitive to political risk in
country pairs in which migrants from the investment-receiving country make up larger
proportions of the pool of potential investors. I use cross-sectional data (1990) on
bilateral migrant stocks as a share of the country-of-residence population as a proxy
for diaspora investors as a share of potential investors in the period 1994-2008.
Relying on a comparison of risk sensitivity across country pairs with varyingsize migrant stocks makes establishing a statistically significant result challenging.
Diaspora direct investment (i.e. investment by diaspora-owned firms) is a subset of
!
!
!
!
82!
migrant-induced investment, which in turn accounts for only a portion of dyadic FDI
flows even in country pairs in which migrants are numerous. Any difference between
diaspora-owned and non-diaspora-owned FDI is diluted, increasing the likelihood of a
finding of no difference. The smaller the proportion of total FDI flows that is made up
of diaspora capital, the greater the dilution. Therefore, a significant finding indicates
not only that the difference between the two types of investors is large, but also that
diaspora direct investment represents a non-trivial portion of total FDI flows in highdiaspora country pairs.
The sensitivity of diaspora direct investment to political risk is evaluated by an
interaction term between the level of migrant stocks in a dyad (as a share of the
country-of-residence population) and the level of political risk in the homeland. In
terms of main effects, I expect the effect of migrant stocks on FDI flows to be positive
– more FDI should flow in country pairs with higher diaspora investment potential. I
expect a negative effect of political risk on FDI flows: less FDI is expected to flow
when risk in the homeland is high. The term of interest, however, is the interaction
between these two. If diaspora direct investment is more sensitive to political risk than
are other types of FDI, the interaction term between migrant stocks and homeland
political risk should be negative: when migrants make up a large proportion of
potential investors, the (negative) impact of political risk should be amplified. This
would be consistent with Hypothesis 1.
Conversely, if non-pecuniary motivations reduce diasporans risk sensitivity
and migrant-induced FDI is less sensitive to political risk than are other types of
!
!
!
!
83!
foreign direct investment, the sign on the interaction term between risk and migrant
stocks should be positive. The main effect of political risk should still be negative and
the main effect of migrant stocks should still be positive.
Evaluating Hypothesese 2A and 2B requires exploring increases and decreases
in political risk separately, which I do simply by substituting dummy variables for risk
increases and risk decreases for a continuous measure of political risk.
Dependent Variable and Sample
The data used in this analysis consist of dyads pairing countries of residence in
the OECD with developing country homelands, with annual observations from 19942008. Due primarily to increases in the availability of political risk data over time, the
panel is unbalanced, with more observations in later years.
The dependent variable, FDI Inflows, is measured bilaterally as net annual FDI
inflows into 105 developing countries5 from 30 OECD countries. These data are taken
from stats.oecd.org; the raw values are measured in millions of USD.6 I use a logged
dependent variable in all specifications. For specifications analyzing levels of FDI,
this logged DV is created as ln_FDI = ± ln|(FDI inflows + 1)|.7
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5
While data are available on flows between OECD countries, only flows from the OECD to countries
on the IMF’s list of developing countries (some of which are OECD members) are considered. Because
the data is collected by the OECD, data on bilateral flows between pairs of non-OECD countries is not
available.
6
While some of the related literature (e.g. Ahlquist 2006, Buthe and Milner 2008) uses FDI/GDP as a
dependent variable, this would be in appropriate here because the concept of interest is the ability of a
country to attract FDI, not the dependence of a country’s economy on FDI (Li 2009).
7
The raw data on bilateral FDI inflows includes a substantial number of negative values, including
some very large negative values. Taking the log of the absolute value of FDI plus one and then restoring
the negative sign to negative FDI flows retains the appropriate location of zero while reducing the
impact of both extremely large negative and positive values.
!
!
!
!
84!
For specifications in changes, the dependent variable is ln_ΔFDIijt where ln_ΔFDIijt =
± ln|(FDIijt – FDIij(t-1))+1|and FDIijt is the net inflow of FDI to Country j from Country
i in period t.
The full sample contains a number of country pairs in which the annual flows
of FDI are zero in all years – some of the smaller OECD economies have FDI in only
a handful of developing countries. To analyze the responsiveness of FDI flows to
political risk in these perpetually zero-FDI country pairs is illogical: FDI will be zero
regardless of the level of risk, making FDI in these dyads appear completely riskinsensitive. Therefore, I remove from the samples all country pairs in which FDI is
zero for all years. 8
Independent Variables
The primary measure of political risk, political risk, is drawn from the Belgian
export credit agency Office National Du Ducroire (ONDD). ONDD is the world’s
largest political risk insurer and the price leader in the industry – its assessments of
risk reflect not only profit-motivated expert attempts to assess risk, but also the actual
insurance costs paid by firms who wish to invest without shouldering the burden of
political risk themselves. ONDD data are used in a similar context by Jensen (2008)
and Jensen and Young (2008).
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
8
While the average number of migrants is lower among the zero-FDI dyads than other dyads, more than
half of zero-FDI dyads have recorded migrant stocks greater than zero. Results are robust to the
inclusion of these dyads in the main specification of the analysis on levels; the results fall from
significance, but retain their sign, when these dyads are added to the analysis on changes.
!
!
!
!
85!
ONDD scores each country in three categories: war risk; government risk,
which is the risk of expropriation or adverse government action;9 and transfer risk,
which refers to risk that action by foreign authorities, such as the introduction of
capital controls or other constraints, prevents the transfer of money necessary to repay
creditors.10 From 2002-2010 these three classifications of risk are all coded annually
on a 7 point scale. From 1992-2001, war and government action risks were combined
and measured annually on a single 5 point scale (wargovrisk). From 1994-2010
transfer risk is measured on a 7 point scale. The ONDD data is available for 126
countries in 1994, but this expands to 186 countries by 2010, producing an unbalanced
panel.11
The primary political risk variable used in this analysis is political risk. From
2002-2008 this variable is a simple sum of war risk + government risk + transfer risk.
From 1994-2001, however, some rescaling is necessary. For this period: political risk
= transfer risk + (wargovrisk * 14/5). To ease the interpretation of coefficients on the
interaction term, this variable is demeaned before it is interacted.12
While risk data are issued annually, a team at ONDD meets four times per year
to update risk evaluations, addressing ¼ of countries (by region) in each meeting.
However, if events justify it, a country’s risk rating may be revised at a meeting in
which it is not otherwise scheduled to be discussed, allowing the potential for multiple
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
9
ONDD defines government action as “local authorities’ decisions, deficiencies, and impairments that
are of an arbitrary and discriminatory nature.”
10
The ONDD’s lexicon can be accessed online at
http://www.ondd.be/webondd/Website.nsf/weben/Documentation_Lexicon?OpenDocument#NT000040
8A (accessed April 20, 2010).
11
Only developing host countries are analyzed, limiting the sample to 147 countries in 2010.
12
This is done in both the specification in levels and the specification in changes.
!
!
!
!
86!
revisions during a year (Jensen 2008). Therefore, the score given to a country for any
given year will reflect any major changes that occurred during that year prior to
ONDD’s last meeting of the year.
Bilateral Migrant Stocks is measured as the number of migrants from Country j
living in Country i in 1990 divided by the total population of Country i in that year.
Migrant stocks are divided by the population of Country i because this variable is
being used to proxy for diaspora investors as a share of potential investors.13 Stocks
are measured as of 1990 (prior to the first year of the analysis) to reduce endogeneity.
The raw data on bilateral migrant stocks is taken from Docquier, Lowell, and
Marfouk (2007). Bilateral migrant stocks is measured cross-sectionally and does not
vary over time.14 Like political risk and Δ political risk, bilateral migrant stocks is
demeaned before it is interacted.
The variable other FDI outflows from Country i is the total FDI outflows from
Country i in year t, minus the value of FDI flows from i to j in that year. Data on
bilateral investment treaties (BITs) are taken from Allee and Peinhardt (2010). The
variable BIT to date takes a value of 1 if a BIT has been signed by the two countries in
the dyad and zero otherwise. Average years of schooling in the homeland, education,
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
13!Leblang!(2010)!uses!logged!bilateral!migrant!stocks,!without!dividing!by!country!i"population.!
In!the!context!of!this!study,!this!variable!is!inappropriate!because!its!use!implies!that!flows!of!FDI!
from!large!home!countries!(with!larger!migrant!populations!in!absolute!terms)!contain!a!larger!
proportion!of!diaspora!direct!investment!than!flows!from!smaller!countries!(whose!migrant!
stocks!are!small!in!absolute!terms).!Not!surprisingly,!using!this!alternate!variable!of!interest!
causes!results!fall!from!statistical!significance!(though!the!direction!of!estimated!effects!remains!
consistent).!
14
I have discovered what I believe to be an error in this data systematically overstating migrant stocks
in Finland, perhaps by a factor of 10. The main results in both levels and changes are robust to dropping
the dyads involving Finland, though in models with the most additional controls, the main effect
sometimes fall from significance when these dyads are omitted.
!
!
!
!
87!
is drawn from Barro and Lee (2010). They estimate the average number of years of
schooling attained by members of the population aged 25 and older. Their estimates
are provided at 5-year intervals – for use in this panel I have annualized their data via
linear interpolation.
Oil Reserves is time-invariant. It is measured as the maximum value of proven
oil reserves 1960-1993 and is constructed from data collected by Macartan Humphreys
(2005).15 GDP per capita is the per capita GDP of the homeland. Population is the log
of the homeland population. Distance is the logged distance between the capitals of
the homeland and country of residence; border is a dummy variable for a shared
border; common language is a dummy variable for a common official language and
colony is a dummy variable for a previous colonial relationship.16 The data on
population is taken from the World Development Indicators, and the distance, border,
common language, and colony measures are taken from CEPII.
For specifications using changes, the independent variables are transformed in
the same manner as the dependent variable.17 Those independent variables that do not
vary over time are retained in levels.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
15
The level of oil reserves is preferable to the level of oil exports because it is less dependent on the
level of investment. However, discovering and proving reserves also requires investment, making it
endogenous to FDI. Therefore, I take the maximum value of reserves prior to the start of the study
period, rather than using panel data on the variable.
16
While some gravity models include a dummy for strict customs unions, I exclude it here. The only
two dyads in the sample in which a customs union was present as of 1997 are Australia/Kiribati and
US/Panama. No OECD/developing country dyads share a common colonizer, so this is also excluded.
17
In the case of the Other FDI Outflows from Country i, I log this variable in the same manner as the
dependent variable.
!
!
!
!
88!
A Specification in Levels and a Specification in Changes
I test Hypothesis 1 using two different specifications: one in levels and one in
changes. The model specified in levels is a least squares dummy variable regression
(LSDV) with dyad fixed effects.18 I run this model on annual data. The model
estimated takes the form of:
y
ijt
= α ij + Β1 y ij(t −1) + Β2 politicalrisk jt + Β3 ( politicalrisk jt * migrantsij ) + Β4 X ijt + ε ijt
(3.1)
where yijt indicates the log of FDI flows from Country i (the country of residence in
the OECD) to Country j (a developing-country homeland) in year t; political riskjt is
the level of political risk in Country j in year t; migrantsij is the stock of migrants from
Country j living in Country i in 1990 (as a share of the Country i population, which
proxies for the share of FDI from Country j to Country i that is made up by diaspora
investment; Xijt is a vector of covariates, αij denotes dyad fixed effects, which may
correlate with the covariates, and εijt is the error term.
For the specification in changes, I collapse the fifteen years of annual
observations for which political risk data is available into five three year periods,
taking the mean value of FDI Inflows in each period. FDI is slow-moving relative to
other forms of investment (e.g., Ahlquist 2006), and the effects of a change in the level
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
18
While the panel dataset covers 15 years, the average number of observations per dyad is 5.5. In short
panels with a lagged dependent variable and fixed effects, the estimated coefficient on the lagged
dependent variable (as well as the coefficients on the fixed effects dummies) is biased (Nickell 1981;
Beck and Katz 2004). However, because I do not have a substantive interest in interpreting the
estimated dynamics, I do not view this as a large drawback to this model – however, I would caution
others against drawing substantive conclusions for the estimated coefficients reported on the lagged
dependent variable. Results are robust to an alternative specification using differenced data, which does
not suffer from Nickell bias.
!
!
!
!
89!
of political risk may be spread out over several years.19 Descriptive statistics for both
the annual data in levels the 3-year blocks of data in differences are in the appendix.
The specification in changes is given as:
Δy
ijt
= α + Β1Δpoliticalrisk jt + Β2 (Δpoliticalrisk jt * migrantsij ) + Β3 X ijt + ε ijt
(3.2)
Δyijt is the (logged) change in FDI flows from Country i to Country j between
years t-1 and t. Δ political riskjt is the change in political risk in Country j from period
t-1 to period t. Xijt is a vector of covariates, some specified in changes, some in levels.
Results for Specifications Analyzing Levels of FDI
In testing Hypothesis 1, I find that diaspora direct investment is much more
sensitive to political risk than are other types of FDI. Figure 3.11 is a plot of the
marginal effect of political risk on bilateral FDI flows in dyads with migrant stocks of
varying sizes. The full results of that regression are presented in Table 3.1, Model 2 on
the following page.
Consistent with Hypothesis 1, the negative relationship between bilateral FDI
flows and Political Risk is much stronger in dyads with large migrant stocks than
dyads with small migrant stocks.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
19
Panel regression would be inappropriate on 3-year blocks of data because the lower number of time
periods would increase the salience of Nickell bias.
!
!
!
!
-4
-6
-8
-10
Marginal Effect of Political Risk on FDI
-2
0
90!
-12
Marginal Effect
95% Confidence Intervals
0
2
6
4
8
10
Migrant Stocks (as % of country i population)
Figure 3.1: Marginal Effect of Political Risk
on Bilateral FDI as Migrant Stocks Varies
Figure 3.1 shows the marginal effect of political risk on bilateral flows of FDI
(logged) as bilateral migrant stocks varies along the x-axis. The y-axis values are
marginal effects. In dyads where migrant stocks are zero, a one-unit increase in
political risk is associated with a small decrease in FDI flows (about -0.5 on a log
scale), while in dyads with large migrant stocks, the negative effect of political risk
increases on FDI flows is more severe.
!
!
!
!
91!
Table 3.1: Political Risk and Migrant FDI: Dyad Fixed Effects
Table 3.1: Political Risk and Migrant FDI: Dyad Fixed Effects
(1)
(2)
(3)
Main Effect Primary Model All Dyads
Political Risk
-0.420***
-0.474***
-0.486***
(0.086)
(0.108)
(0.177)
1 year lag of DV
-0.126
(0.130)
-0.463***
(0.159)
-0.477***
(0.168)
Other FDI outflows from country i
0.037*
(0.020)
0.037
(0.023)
0.042
(0.025)
-875.988***
(257.555)
-766.348***
(242.470)
Education
-60.913
(49.831)
-59.057
(54.234)
BIT to Date
-164.098
(108.339)
-156.962
(120.120)
Political Risk∗ Migrant Stock
Political Risk∗ Distance
-359.280***
(134.678)
Political Risk∗ Common Language
742.802
(470.419)
Political Risk∗ Common Border
4.357
(404.827)
Political Risk∗ Colony
-1186.210**
(531.581)
Political Risk∗ BIT to Date
34.727
(202.776)
Political Risk∗ Oil Reserves
1.957
(1.341)
Constant
Observations
R2
6.466***
(0.472)
13193
0.004
11.703***
(3.542)
9196
0.008
12.045***
(3.810)
8077
0.010
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
The following independent variables are dropped because they do not vary over time:
Migrants, Natural Resources, Distance, Common Language, Common Border, Colony.
!
!
!
!
92!
In Table 3.1 the sign on the interaction term between political risk and migrant
stocks, Political Risk*Migrant Stocks, is strongly negative and robust to controls. This
indicates that migrant-induced FDI is indeed more sensitive to political risk than is
other FDI. All the models in Table 3.1 include dyad fixed effects. The independent
effect of political risk is negative, as expected. Because both political risk and bilateral
migrant stocks have been demeaned before they are interacted, the coefficient on
political risk in Model 2 can be interpreted as the independent effect of political risk
on FDI inflows when bilateral migrant stocks is at its mean.
The dummy gravity variables are not demeaned before interaction, while the
continuous measures, distance and oil reserves, are demeaned. Therefore, the
coefficient of political risk in Model 3 can be interpreted as the marginal effect of
political risk on FDI inflows when bilateral migrant stocks, oil reserves, and distance
are at their means and common border, colonial relationship, common language and
BIT to date are equal to zero. Higher political risk in the homeland is associated with
lower FDI inflows.
Because bilateral migrant stocks, oil reserves, colonial relationship, common
border, common language, and distance are all time-invariant, no estimate of their
main effect is produced by these fixed effects regressions. These main effects are
examined in a cross-sectional regression in Table 3.3.
The most important control variable is other FDI outflows from Country i,
which measures FDI outflows from the country of residence to all countries except the
homeland in a particular year. The coefficient is positive and significant in all
!
!
!
!
93!
specifications. This effectively controls for all supply-side factors affecting the overall
availability of FDI outflows from each Country i, allowing us to focus narrowly on the
decision of investors from Country i to invest in Country j (the homeland) as opposed
to other possible destinations.
It is possible that some form of confounding variable causes both large migrant
stocks and high FDI flows. In particular, it is plausible that a variable that causes
migration, such as distance between the two countries, a common border, common
language, or colonial tie might be driving the reduced sensitivity to political risk,
rather than the migrant population itself. Model 3 includes interaction terms between
all four of these gravity variables and political risk, as well as an interaction between
BIT to date and political risk. Results are robust to the inclusion of these interactions.
Like migrant stocks, a colonial relationship and geographic proximity are
associated with greater risk sensitivity in FDI flows. This is consistent with the
information-based argument put forward in this paper. Multinational firms based in
countries that are near a particular developing country or that were former colonizers
of that developing country are likely to have greater access to information about that
developing country than are multinationals based in more distant or unrelated
countries. Higher levels of information allow firms to respond quickly and effectively
to changes in the level of political risk, making FDI flows in these dyads more
sensitive to changes in political risk in County j.
!
!
!
!
94!
Results for Specifications Analyzing Changes in FDI
The expectations regarding the specification in changes are the same as with
regard to the specification in levels. If Hypothesis 1 is correct, the interaction between
Δ political risk and bilateral migrant stocks will be negative. I expect the main effect
of Δ political risk on changes in FDI flows also to be negative.
Because this model does not contain dyad-fixed effects, the main effect of
bilateral migrant stocks does not drop out of the regression, and a coefficient is
estimated. However, it is important to note that bilateral migrant stocks is measured in
levels as of 1990 – it does not vary over time. There is no theoretical reason to expect
changes in FDI flows to be different in high-migrant dyads as opposed to low migrant
dyads, so the expectation is that the coefficient on bilateral migrant stocks is close to
zero. This expectation is borne out. Similar expectations apply to the main effects of
control variables added in Model 3: these variables are measured in levels, not
changes, and there is no theoretical reason to expect their coefficients to be
significantly different from zero. The inclusion of these terms serves only to allow
interpretation of the interaction effects. Their main effects, along with the main effect
of bilateral migrant stocks, are estimated in a cross-sectional regression in Table 3.3.
!
!
!
!
95!
Table 3.2: Political Risk and Migrant FDI: Specifications in Changes
∆ Political Risk
Table 3.2: Political Risk and Migrant FDI: Specifications in Changes
(1)
(2)
(3)
Main Effects Primary Model Added Controls
-0.739***
-0.701***
-0.558***
(0.156)
(0.156)
(0.180)
Bilateral Migrant Stock (1990)
0.046
(0.075)
0.032
(0.061)
0.003
(0.059)
0.025
(0.067)
∆ Other FDI Outflows from Country i
0.097***
(0.013)
0.098***
(0.013)
0.094***
(0.014)
0.083***
(0.016)
∆ BIT To Date
0.952
(1.924)
0.958
(1.925)
1.231
(2.016)
1.750
(2.171)
∆ Education
0.548
(1.361)
0.582
(1.358)
0.668
(1.512)
1.399
(2.105)
-0.171***
(0.064)
-0.174***
(0.060)
-0.167***
(0.057)
∆ Political Risk * Migrant Stock
Oil Reserves Per Capita
2.493
(2.720)
∆ Political Risk*Oil Reserves
0.540
(2.057)
1.592
(3.023)
-1.428***
(0.364)
-1.056
(0.711)
∆ Political Risk * Distance
-0.388*
(0.226)
-0.466*
(0.244)
Common Language Dummy
1.559
(1.052)
1.863
(1.281)
∆ Political Risk * Common Language
1.320*
(0.737)
1.018
(0.770)
Shared Border Dummy
-0.779
(2.339)
-0.110
(2.487)
∆ Political Risk * Shared Border
1.112
(1.558)
0.862
(1.655)
Colonial Relationship Dummy
0.698
(1.516)
0.367
(1.685)
∆ Political Risk * Colonial Relationship
-1.160
(0.960)
-1.203
(0.962)
2.445***
(0.532)
3281
-2.078
(6.646)
3281
Distance (logged)
Constant
Observations
2.626***
(0.479)
3808
2.600***
(0.478)
3808
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
The following independent variables are dropped because they do not vary over time:
Migrants, Natural Resources, Distance, Common Language, Common Border, Colony.
!
!
!
(4)
Country FE
-0.507**
(0.210)
!
96!
These results match well with those from the specification in levels. The
interaction term of interest, Δ Political Risk * Migrant Stocks, is negative, significant,
and robust to a range of controls. This indicates that FDI flows are more risk sensitive
in dyads with larger migrant stocks, and this is consistent with diaspora investors
responding more strongly to political risk than non-diaspora foreign investors.
As with the regressions in levels, both Δ political risk and bilateral migrant
stocks are demeaned before they are interacted, so the coefficient on Δ political risk in
Model 2 can be interpreted as the independent effect of Δ political risk on FDI inflows
when bilateral migrant stocks is at its mean. The dummy gravity variables are not
demeaned before interaction, while the continuous measures, distance and oil reserves,
are demeaned. Therefore, interpretation of the coefficient of Δ political risk in Models
3 and 4 can be interpreted as the marginal effect of Δ political risk on FDI inflows
when bilateral migrant stocks, oil reserves, and distance are at their means and
common border, colonial relationship and common language are equal to zero. The
main effect of Δ political risk is negative and significant, as expected: increases in
political risk are associated with decreases in net FDI inflows. Among the control
variables, the level of FDI outflows from Country i remains positive and significant,
as expected. The interaction terms between colonial relationship and political risk and
between distance and political risk are not significant in this specification, though their
signs are the same as in the specification in levels.
Model 4 of Table 3.2 includes fixed effects for both home and host country.
This model does not allow the main effect of Oil Reserves Per Capita (in Country j) to
!
!
!
!
97!
be estimated because this variable does not vary over time, and hence its main effect is
picked up by the country fixed effects for Country j. The results regarding the
interaction of interest are robust to the inclusion of these fixed effects.
The results from Table 3.2 demonstrate the robustness of my findings to
starkly differently specifications. The results are the same when analyzing annual-data
specified in levels and 3-year increments of data specified in changes. These results
provide strong support for Hypothesis 1. They indicate that migrant-induced FDI is
actually more sensitive to political risk than is other FDI. This is consistent with the
theory that diasporans are better informed than other foreign investors and contradicts
central predictions derived from theories of diaspora difference based on nonpecuniary motivations.
Cross-Sectional Results
While dyad-fixed-effects modeling offers the clearest insight into the causal
relationship between political risk and diaspora direct investment, it is useful to
present a specification in which the main effect of bilateral migrant stocks and time
invariant controls can be estimated.20 Hausman test results indicate that a randomeffects specification would be biased; therefore, I opt for a simple cross-sectional
regression on a single period.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
20
Plümper and Troeger (2007) propose a three-stage estimator, which they refer to as fixed effects
vector decomposition and claim allows for efficient estimation of the effects of time invariant variables
in fixed effects panel regressions. Greene (2011) points out several critical flaws in the estimator,
leading me not to use it here. Using this estimator produces results that are consistent with the key
interaction effects in Tables 3.2 and 3.4, i.e. both the main effect of political risk and the coefficient on
the interaction between political risk and migrant stocks are negative and significant. However, the
main effect of migrant stocks on FDI is negative (and not significant), which I attribute to the nature of
the estimator, which relies on a misspecified random effects panel regression to assess the main effects
of time-invariant variables.
!
!
!
!
98!
I omit the interaction between political risk and migrant stocks from these
regressions. Any diasporan has only one homeland with regard to which her
knowledge and connections are relevant. Therefore, it is the over-time variation in
diasporans’ willingness to invest in their homeland, rather than their cross-sectional
decision about which country to invest in, on which the theories in this paper bear
most directly.
In the cross-sectional analysis I utilize bilateral FDI data from the International
Monetary Funds’s 2009 Coordinated Direct Investment Survey. This survey provides
data on FDI outflows for a wider range of FDI-sending countries – the OECD data
contains data of FDI outflows only from OECD members. However, the data is only
available for a single year, making it useful only in the cross-sectional context.
Because dyad fixed effects are not included in the following models, I include
standard gravity-model determinants of FDI (e.g., Benassy-Quere, Coupet, and Mayer
2007). I also use the most up-to-date measure of migrant stocks available, which
comes from the World Development Indicators and is measured as of 2000.21
Controls include dummy variable fixed effects for the FDI-sending country
(Country i) and economic characteristics of the FDI-receiving country (Country j),
including political risk. Due to data limitations the control variable BIT to date only
captures BITs signed in 2006 and earlier.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
21
In Tables 3.1 and 3.2, I use a measure of bilateral migrant stocks from 1990 to reduce endogeneity.
!
!
!
!
99!
Table 3.3: Political Risk and Migrant FDI: Cross Sectional Regression
Table 3.3: Political Risk and Migrant FDI: Cross Sectional Regression
(1)
(2)
(3)
(4)
Main Effects Controls More Controls Country i & j FE
Bilateral Migrant Stocks (2000)
155.6***
133.6***
43.09***
47.11***
(36.344)
(31.206)
(14.630)
(13.700)
Political Risk (2009)
-0.615***
(0.020)
-0.351***
(0.034)
-0.267***
(0.032)
Population of Country j (logged)
1.084***
(0.064)
1.276***
(0.079)
1.306***
(0.076)
4693.0***
(1637.737)
7052.3***
(1646.033)
Education in Country j
0.593***
(0.053)
0.474***
(0.049)
Oil Reserves in Country j
2.121***
(0.727)
1.283*
(0.728)
Political Risk * Oil Reserves
0.312
(0.217)
0.0436
(0.216)
GDP Per Capita in Country j
Common Language (10% of Pop)
1.381***
(0.347)
0.805**
(0.313)
Common Border
1.482**
(0.575)
1.244**
(0.571)
Colonial Relationship
2.436***
(0.575)
2.513***
(0.574)
Distance (logged)
-2.455***
(0.139)
-3.020***
(0.168)
BIT (2006)
2.396***
(0.325)
2.571***
(0.317)
Dummies for Country i
YES
YES
YES
YES
Dummies for Country j
NO
NO
NO
YES
-4.512***
(1.650)
5103
0.541
-17.59***
(1.509)
4085
0.581
1.880
(1.949)
4085
0.638
32.67***
(3.242)
5983
0.659
Constant
Observations
R2
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
!
!
!
!
100!
The purpose of this table is to show the main effects of bilateral migrant stocks
and time-invariant control variables. As expected, the main effect of bilateral migrant
stocks is positive and the main effect of political risk is negative. The effect of
bilateral migrant stocks is weakened when a sufficient range of controls is added, but
its coefficient remains significant at the 0.05 level in a one-tailed test.22 Substantively,
this is consistent with what is established in the literature, that migrants promote the
flow of FDI from their countries of residence to their homelands.
Models 4 and 5 in Table 3.3 include a variety of elements of the relationship
between Country i and Country j whose main effects could not be estimated in Tables
3.1 and 3.2. Not surprisingly, these characteristics have a strong influence on bilateral
FDI flows: FDI flows are largest between countries that are geographically proximate
and that share linguistic and cultural ties. Bilateral investment treaties are also
associated with larger FDI flows.
Increases vs. Decreases in Risk: Dummy Independent Variables
Hypotheses 2a and 2b state that the greater risk sensitivity of diaspora direct
investment observed above is expected to manifest itself with regard to both increases
and decreases in risk. To test this, I substitute dummy variables for political risk
increase and political risk decrease in place of the continuous measure of Δ political
risk. Risk up takes a value of 1 if Δ political risk > .5, risk down takes a value of 1 if Δ
political risk < -.5, while both variables are zero for observations with very small or no
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
22
A one-tailed test is appropriate here because the expected direction of effect (positive) is clear. Larger
(and more robust) main effects can be achieved using an alternative measure of migrant stocks, such as
the raw or logged number of migrants in a dyad. However, a ratio measure of migrants/population is
more theoretically appropriate for estimating the share of outward FDI attributable to migrants.
!
!
!
!
101!
changes in political risk. Because risk up and risk down are binary, they are not
demeaned before they are interacted with bilateral migrant stocks (which is
demeaned).
Equation 3.3 gives the equation estimated in model 4 of Table 3.4 below.
Δy
ijt
= α + Β1Δrisk _ up jt + Β2 (risk _ up jt * migrantsij ) + Β3 Δrisk _ down jt
+Β4 (risk _ down jt * migrantsij ) + Β5 X ijt + ε ijt
!
!
!
(3.3)
!
102!
Table 3.4: Political Risk and Migrant FDI: Decreases vs. Increases in Risk
Table 3.4: Political Risk and Migrant FDI: Risk Increases vs. Risk Decreases
(1)
Main Effects
0.094***
(0.013)
(2)
Risk Up
0.095***
(0.013)
(3)
Risk Down
0.095***
(0.013)
(4)
Up & Down
0.095***
(0.013)
(5)
Controls
0.088***
(0.014)
(6)
Controls & FE
0.079***
(0.016)
∆ BIT To Date
1.153
(1.953)
0.768
(1.922)
1.193
(1.938)
1.119
(1.948)
1.145
(1.981)
1.627
(2.166)
∆ Education
0.575
(1.207)
0.414
(1.389)
0.788
(1.194)
0.690
(1.204)
0.797
(1.318)
1.568
(2.110)
Bilateral Migrant Stock (1990)
0.047
(0.058)
0.049
(0.066)
-0.104***
(0.030)
-0.104***
(0.022)
-0.103***
(0.022)
-0.091
(0.089)
Risk Up
-0.640
(1.039)
-2.162**
(0.976)
-0.688
(1.029)
0.380
(0.986)
0.441
(0.984)
Risk Down
2.298***
(0.714)
2.165***
(0.709)
2.172**
(0.854)
2.642***
(0.869)
-0.056
(0.398)
-0.138
(0.288)
-0.023
(0.384)
0.411***
(0.120)
0.366***
(0.105)
0.364***
(0.117)
Distance (logged)
-0.250
(0.734)
-0.082
(1.100)
Risk Up * Distance
-3.462***
(1.089)
-3.391**
(1.322)
Risk Down * Distance
-0.598
(0.906)
-0.383
(1.065)
Common Language Dummy
2.687
(1.928)
3.481*
(2.054)
Risk Up * Common Language
0.907
(2.637)
-0.111
(2.912)
Risk Down * Common Language
-2.513
(2.681)
-3.171
(2.503)
Shared Border Dummy
-0.263
(2.961)
-0.645
(5.780)
Risk Up * Common Border
1.489
(6.173)
0.857
(8.464)
Risk Down * Common Border
-1.758
(3.534)
-0.281
(6.338)
Colonial Relationship Dummy
-1.102
(3.372)
-2.123
(3.005)
Risk Up * Colonial Relationship
-2.702
(6.297)
-1.894
(4.865)
Risk Down * Colonial Relationship
4.389
(3.838)
4.944
(3.459)
∆ Other FDI Outflows from Country i
Risk Up * Migrants
2.469***
(0.694)
-0.210
(0.393)
Risk Down * Migrants
Dummy Variables for Country i and Country j
Constant
0.412***
(0.118)
NO
NO
NO
NO
NO
YES
1.610**
(0.702)
3808
3.179***
(0.692)
3808
1.292**
(0.585)
3808
1.624**
(0.702)
3808
1.277*
(0.746)
3353
2.161
(6.105)
3353
Observations
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
In Tables 1-5, errors are clustered on country i’s country ID.
Hypothesis 2A states that migrant-induced FDI responds more strongly than
other FDI to increases in political risk, while Hypothesis 2B makes a parallel assertion
!
!
!
!
103!
with regard to decreases in political risk. The results in Table 3.4 allow us to reject the
null hypothesis with regard to Hypothesis 2B, but not with regard to 2A. The
coefficient on the interaction between risk down and bilateral migrant stocks is
positive and significant in all specifications, indicating that the positive effect of risk
decreases on FDI is amplified in dyads with large migrant stocks, consistent with
Hypothesis 2B.
The coefficient on the interaction between risk up and bilateral migrant stocks
is negative in all models, but not statistically significant. This indicates that the
negative effect of risk increases on FDI flows is perhaps somewhat larger in dyads
with large migrant stocks, but this interaction effect is much weaker and in some
models very near zero. Therefore, there is little evidence in support of Hypothesis 2A.
Across all levels of migrant stocks, decreases in risk have a substantially larger
effect on changes in FDI than do risk increases, despite the fact that each direction of
change has the same average magnitude.23 In other words, relative to a static level of
risk, institutional improvements prompt greater changes in investment behavior than
institutional deteriorations. This is a finding that justifies future research. It may be
that, while investment in established sectors is generally continued after risk rises, FDI
into new sectors generally takes off in response to institutional stabilizations and
improvements.
As in Table 3.2, the coefficient on bilateral migrant stocks is not expected to be
significantly different from zero. However, in Models 2 and 3 the coefficient is
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
23
The mean size of risk increases in the sample is 1.65, while the mean risk decrease is 1.68.
!
!
!
!
104!
negative and significant. While not significant once controls are added, analyzing
these coefficients in conjunction with the coefficients on the interaction terms suggests
that when political risk in Country j remains neutral or increases (i.e. risk down = 0),
dyads with larger migrant stock experience slightly negative (or less positive) changes
in FDI flows relative to similar dyads. This effect is reversed in dyad-periods with risk
decreases in Country j, when dyads with large migrant stocks see more positive
changes in FDI relative to similar dyads.
The control variables included in Table 3.4 are very similar to those in Table
3.2, however, gravity model variables like common language and shared border are
now interacted with dummy variables for both risk increase and risk decrease, instead
of with a continuous measure of change in political risk. Most of the coefficients on
these interactions are not significantly different than zero. However, FDI flows
between geographically proximate countries (i.e. dyads with smaller values for
distance) are more sensitive to risk increases, which is consistent with the informationbased theory I present. Potential investors are likely better informed about states in
their region than those far away, and hence are better able to seize on investment
opportunities that arise when political risk decreases.
Substantively, these results show that the difference in the risk sensitivity of
FDI inflows between high-migrant and low-migrant dyads is primarily a difference in
responsiveness to decreases in political risk in Country j, i.e. stabilizations and
institutional improvements. While Table 3.4 provides no evidence that FDI flows in
high-migrant dyads are more stable in the face of crises than flows in low-migrant
!
!
!
!
105!
dyads, there is also no evidence that flows in these dyads are significantly more
sensitive to crises. These finding are consistent with the information-based theory of
diaspora difference I propose, and suggest that diaspora investors are more likely than
other foreign investors to make or expand investments in response to improving risk
conditions. Diasporans constantly monitor conditions in the country of origin for nonbusiness reasons, and when the political risk climate improves, they have the
information they need to seize new opportunities associated with the improving
investment climate.
Theoretical Implications
The results presented in this paper provide evidence that migrant induced FDI
is more sensitive to changes in political risk, and particularly to decreases in political
risk, than is other FDI. This is consistent with the theory that diaspora-owned firms
experience and respond to political risk differently than other foreign firms, and
demonstrates that the way in which aggregate flows of investment respond to changes
in political risk depends on the mix of diaspora and non-diaspora investors on which
that country is drawing. Theses findings also contradict some of the more optimistic
theories in economic sociology, which claim that diaspora investors are less risk
sensitive than other investors and are more willing to make and sustain investments
during times of acute political crisis. While it is with regard to risk decreases that
migrant-induced FDI is much more sensitive, it is no less sensitive than other FDI
with regard to risk increases. These findings are consistent with the theory of diaspora
difference based on information I propose: that diasporans gain access to private
!
!
!
!
106!
information about the homeland via social networks, and that diasporans are more
attentive to publicly available information about the homeland than other potential
foreign investors.
My informational theory assumes that diasporans’ non-pecuniary motivations
do shape some behaviors. Diasporans bear the costs of monitoring the political climate
in the homeland because of the social and emotional ties they keep with the homeland
and its inhabitants. This attentiveness causes diasporans to be better informed about
the homeland than other potential foreign investors, and in turn positions them to seize
new investment opportunities that arise when political risk decreases. Therefore, while
I argue that some of the more optimistic expectations in the economic sociology
literature must be amended, I do not argue that diasporans’ social and emotional ties to
the homeland are irrelevant to their investment behavior. These ties, and the behaviors
they motivate, are important components of diaspora difference and underscore the
importance of taking diaspora identity seriously when studying the relationship
between political institutions and economic behavior.
Policy Implications
Fragile states often persist in a cycle where instability reduces investment,
investment reduces growth, and lack of growth begets further instability. The current
policy literature is optimistic that diasporans may provide a risk-insensitive source of
capital that continues to flow as risk increases and may help prevent or end this cycle
(e.g. Debass and Ardovino 2009). The theoretical argument and empirical results in
this paper belie these optimistic expectations.
!
!
!
!
107!
This has important policy implications. If the risk sensitivity of migrantinduced investment demonstrated here is, as I argue, driven by the risk sensitivity of
diaspora investors, efforts dedicated to recruiting diaspora investment into fragile
states during times of acute crisis and high political risk are likely to fare much less
well than efforts to recruit this investment made just after settlements have been
reached or reforms have been made. While I find no evidence to support previous
theories that diaspora investors are less sensitive than their peer firms to risk increases
(i.e. crises) in the homeland, my results do suggest that diaspora investors are much
more responsive than other investors to reductions in political risk in the homeland,
i.e. stabilizations and improvements. While diaspora investment cannot serve as a
substitute for political reform in fragile states, it may be capable of serving as both a
complement to, and incentive for, institutional reforms that limit political risk.
Political reforms that limit risk are often costly to actors who undertake them.
Take for example, executive constraints. Constraints on executive authority, such as
those provided by an opposition-controlled legislature or an independent supreme
court with the right of judicial review, are associated with a higher degree of policy
stability and lower levels of political risk (e.g. Keefer and Knack 1997; Henisz 2000).
However, when an executive accepts additional constraints on her authority, she
necessarily loses the ability to pursue its interests and preferences as freely as before –
this is a substantial cost. If the executive, or any other political actor, understands that
costly reforms that reduce political risk will likely be followed by a rapid influx of
diaspora capital, this provides some additional incentive to undertake such reforms.
!
!
!
!
108!
Similarly, if political actors know that actions that increase political risk will lead to a
reduction in FDI flows (even from diaspora investors), it provides incentives to avoid
such actions.
Political-risk insensitivity can be a desirable characteristic of capital flows in
countries plagued by a cycle of instability, low investment, and low growth – breaking
the cycle by providing investment in the absence of stability. However, risk sensitive
capital also provides a means out of the same cycle – incentivizing the creation of
stability in the absence of growth. A high volume, risk sensitive source of potential
capital can provide valuable incentives for political actors to initiate difficult reforms
before any investment is in place; the entrance of these investors can provide the
capital (and growth) to sustain these reforms.
Alternative Explanations, Causal Identification, and Future Research
The most substantial impediment to causal identification in this analysis is the
risk is that the observed interaction effects in Tables 1, 2, and 4 are driven by some
omitted factor that affects both migration and investment, and which affects
investment more in the countries in which its effect on migration is also largest. I have
mitigated this risk by controlling for both the main and interaction effects of a number
of variables expected to affect both migration and FDI: geographic proximity and
cultural, linguistic and historical connections, and oil reserves in the homeland.
Results are robust to these controls.
While the results presented here are consistent with the information-based
theory of diaspora difference that I have proposed, evidence at this level of
!
!
!
!
109!
aggregation cannot decisively rule out all alternative theories. For example, Kugler
and Rapoport (2007), argue that the presence of migrants provides information to all
the firms in the migrants’ country of residence about the labor market (and possibly
other conditions) in the homeland. This would imply that the informational advantages
provided by migrants are not limited to firms that are migrant-owned, or even to firms
that hire migrants.
It is also possible that the high political risk sensitivity of diaspora investors is
driven, at least in part, by political vulnerabilities unique to diaspora-owned firms.
Diaspora direct investors function inside the domestic political milieu of the homeland
in a way that other foreign direct investors do not; many still retain their citizenship in
the homeland and hence occupy an uncertain space between domestic and
international business. Some lack citizenship in their country of residence, depriving
them of the protections afforded to investors by the country-of-residence government
and its embassy. This may cause them to be uniquely vulnerable to certain types of
political risk including adverse government actions such as expropriation, extortion, or
limitations on the repatriation of profits.
The most promising path to solidify (or challenge) the causal inference made
in this paper and to parse between the information-based theory proposed here and
plausible alternatives involves data that differentiates directly between diaspora and
non-diaspora FDI. The next step in this research agenda is a firm-level comparison
between diaspora-owned firms and other foreign firms.
!
!
!
!
110!
Conclusion
I argue that diaspora investors differ from other foreign investors in both their
access to information and their costs of accessing publicly available information, and
that these information-related differences make diaspora direct investors more
sensitive to changes in political risk in the homeland. Empirically, I demonstrate that
FDI flows in high-migrant dyads are more sensitive to over-time changes in political
risk in the homeland than are FDI flows in low-migrant dyads, and that this
differential sensitivity manifests primarily with regard to decreases in political risk.
This suggests a need to revise downward some of the most optimistic expectations in
the policy literature regarding diaspora direct investment as a source of capital for
fragile states. It also challenges current practice in the international political economy
literature, where it is common to treat political risk as a country-level variable and
ignore variations across types of firms. These results indicate that disaggregating
investment flows by investor type would substantially improve our understanding of
how political risk affects investment behavior.
The political science and economics literature demonstrates a causal
relationship between diaspora populations and flows of FDI, I develop, and test key
implications of, an information-based theory of diaspora difference. The evidence that
migrant-induced investment is more sensitive than other FDI to risk-decreases is
consistent with the theory that diasporans are more attentive to information about the
homeland than are other potential investors. I argue that diasporans monitor political
and social conditions in the homeland for non-business reasons, and when the risk
!
!
!
!
111!
climate improves, they have the information necessary to exploit the new
opportunities that these improvements produce.
My findings suggest that diasporans’ non-pecuniary motivations do not reduce
their sensitivity to political risk, and overturn expectations of risk insensitivity that are
currently influential in US development policy. However, while diaspora direct
investment diaspora direct investment does not substitute for, and is no more likely
than other FDI to precede, basic institutional stabilization in fragile states, these results
suggest it can serve as a robust incentive for, and complement to, these improvements.
!
!
!
!
112!
References
Ahlquist, John. 2006. Economic Policy, Institutions, and Capital Flows: Portfolio and
Direct Investment Flows in Developing Countries. International Studies
Quarterly (50): 681-704.
Allee, Todd, and Clint Peinhardt. 2010. Delegating Differences: Bilateral Investment
Treaties and Bargaining Over Dispute Resolution Provisions. International
Studies Quarterly (54): 1-26.
Bandelj, Nina. 2008. From Communists to Foreign Capitalists. Princeton, NJ:
Princeton University Press.
Barro, Robert J, and Jong-Wha Lee. 2010. "A New Data Set of Educational
Attainment in the World, 1950-2010." 15902. National Bureau of Economic
Research.
Beck, Nathaniel, and Jonathan Katz. 2004. "Time-Series-Cross-Section Issues:
Dynamics."
Bénassy Quéré, A, M Coupet, and T Mayer. 2007. Institutional Determinants of
Foreign Direct Investment. The World Economy 30 (5): 764-782.
Bevan, Alan A., and Saul Estrin. 2004. The Determinants of Foreign Direct
Investment into European Transition Economies. Journal of Comparative
Economics 32 (4): 775-787.
Busse, Matthias, and Carsten Hefeker. 2007. Political Risk, Institutions and Foreign
Direct Investment. European Journal of Political Economy 23 (2): 397-415.
Buthe, Tim, and Helen V. Milner. 2008. The Politics of Foreign Direct Investment to
Developing Countries: Increasing FDI through International Trade
Agreements? American Journal of Political Science 52 (4): 741-762.
Choi, Seung-Whan, and Yiagadeesen Samy. 2008. Reexamining the Effect of
Democratic Institutions on Inflows of Foreign Direct Investment in Developing
Countries. Foreign Policy Analysis 4 (1): 83-103.
Debass, Thomas, and Michael Ardovino. 2009. Diaspora Direct Investment (DDI):
The Untapped Resource. Washington, DC: United States Agency for
International Development.
Desbordes, Rodolphe. 2009. Global and Diplomatic Political Risks and Foreign Direct
Investment. Economics and Politics 22 (1): 92-125.
!
!
!
!
113!
Docquier, Frédéric, and Elisabetta Lodigiani. 2010. Skilled Migration and Business
Networks. Open Economies Review.
Elkins, Zachary, Andrew T. Guzman, and Beth A. Simmons. 2006. Competing for
Capital: The Diffusion of Bilateral Investment Treaties, 1960-2000.
International Organization 60 (04): 811-846.
Gillespie, Kate, Liesl Riddle, Edward Sayre, and David Sturges. 1999. Diaspora
Interest in Homeland Investment. Journal of International Business Studies 30
(3): 623-634.
Gillespie, Kate, Edward Sayre, and Liesl Riddle. 2001. Palestinian Interest in
Homeland Investment. Middle East Journal 55 (2): 237-255.
Globerman, Steven, and Daniel Shapiro. 2003. Governance Infrastructure and US
Foreign Direct Investment. Journal of International Business Studies 34 (1):
19-39.
Gordon, Roger H., and A. Lans Bovenberg. 1996. Why Is Capital So Immobile
Internationally? Possible Explanations and Implications for Capital Income
Taxation. The American Economic Review 86 (5): 1057-1075.
Greene, William. 2011. Fixed Effects Vector Decomposition: A Magical Solution to
the Problem of Time-Invariant Variables in Fixed Effects Models? Political
Analysis 19 (2): 135-146.
Henisz, W. J. 2000. The Institutional Environment for Multinational Investment.
Journal of Law and Economic Organization 16 (2): 334-364.
Huang, Ye. 2003. Selling China: Foreign Direct Investment During the Reform Era.
Cambridge: Cambridge University Press.
Humphreys, Macartan. 2005. Natural Resources, Conflict, and Conflict Resolution.
Journal of Conflict Resolution 49 (4): 508-537.
Javorcik, Beata S., Çaglar Özden, Mariana Spatareanu, and Cristina Neagu. 2011.
Migrant Networks and Foreign Direct Investment. Journal of Development
Economics.
Jensen, Nathan. 2008. Political Risk, Democratic Institutions, and Foreign Direct
Investment. The Journal of Politics 70 (04): 1040-1052.
!
!
!
!
114!
Jensen, Nathan, and Fiona McGillivray. 2005. Federal Institutions and Multinational
Investors: Federalism, Government Credibility, and Foreign Direct Investment.
International Interactions 31 (4): 303 - 325.
Kerner, Andrew. 2009. Why Should I Believe You? The Costs and Consequences of
Bilateral Investment Treaties. International Studies Quarterly 53 (1): 73-102.
Kolstad, Ivar, and Espen Villanger. 2008. Determinants of Foreign Direct Investment
in Services. European Journal of Political Economy 24 (2): 518-533.
Kugler, Maurice, and Hillel Rapoport. 2007. International Labor and Capital Flows:
Complements or Substitutes? Economics Letters 94 (2): 155-162.
Leblang, David. 2010. Familiarity Breeds Investment: Diaspora Networks and
International Investment. American Political Science Review: 1-17.
Lewis, Karen. 1999. Trying to Explain Home Bias. Journal of Economic Literature
38: 571-608.
Li, Quan. 2009. Outlier, Measurement, and the Democracy-FDI Controversy.
Quantitative Journal of Political Science (4): 167-181.
Li, Quan, and Adam Resnick. 2003. Reversal of Fortunes: Democratic Institutions and
Foreign Direct Investment Inflows to Developing Countries. International
Organization (57): 175-211.
Nickell, Stephen. 1981. Biases in Dynamic Models with Fixed Effects. Econometrica
49 (6): 1417-1426.
Nielsen, Tjai, and Liesl Riddle. 2010. Investing in Peace: The Motivational Dynamics
of Diaspora Investment in Post-Conflict Economies. Journal of Business
Ethics.
Oneal, John R. 1994. The Affinity of Foreign Investors for Authoritarian Regimes.
Political Research Quarterly 47 (3): 565-588.
Plümper, Thomas, and Vera E. Troeger. 2007. Efficient Estimation of Time-Invariant
and Rarely Changing Variables in Finite Sample Panel Analyses with Unit
Fixed Effects. Political Analysis 15 (2): 124-139.
Portes, Richard, and Hélène Rey. 2005. The Determinants of Cross-Border Equity
Flows. Journal of International Economics 65 (2): 269-296.
!
!
!
!
115!
Rauch, James, and Allessandra Cassella. 2003. Overcoming Informational Barriers to
International Resource Allocation: Prices and Ties. Economic Journal 113: 2142.
Schneider, Friedrich, and Bruno S. Frey. 1985. Economic and Political Determinants
of Foreign Direct Investment. World Development 13 (2): 161-175.
Schulte, Bettina. 2008. "Second Generation Entrepreneurs of Turkish Origin in
Germany: Diasporic Identity and Business Engagement." 56. Center on
Migration, Citizenship, and Development.
Sethi, D., S.E. Guisinger, S.E. Phelan, and D.M. Berg. 2003. Trends in Foreign Direct
Investment Flows: A Theoretical and Empirical Analysis. Journal of
International Business Studies 34: 315-326.
Uzzi, Brian. 1996. The Sources and Consequences of Embeddedness for the Economic
Performance of Organizations: The Network Effect. American Sociological
Review 61 (4): 674-698.
Ye, Min. 2010. "Ethnic Investors vs. Foreign Investors: The Impact of Diasporas on
Economic Liberalization in China and India." Presented at the Annual Meeting
of the American Political Science Association, Washington DC.
Yli-Renko, Helena, Erkko Autio, and Harry J. Sapienza. 2001. Social Capital,
Knowledge Acquisition, and Knowledge Exploitation in Young TechnologyBased Firms. Strategic Management Journal 22 (6-7): 587-613.
!
!
!
!
116!
Appendix
Table 3.5: Descriptive Statistics From Annual Data
Variable
Name
FDI Inflows
(Millions of
USD)
Category
Mean
Standard
Deviation
Minimum
Maximum
Number of
Observations
Overall
Between
Within
65.8
468
168
393
-23700
15700
N=31100
N=3920
T-bar=7.94
FDI Inflows
(logged)
Overall
Between
Within
4.66
10.6
6.09
8.62
-23.9
23.5
N=31100
n=3920
T-bar=7.94
Other FDI
Outflows
from
Country i
(logged)
Political
Risk
Overall
Between
Within
21.2
9.38
6
7.72
-24.4
26.7
N=30900
N=3910
T-bar=7.91
Overall
Between
Within
11.5
4.45
4.17
1.54
3
21
N=34100
N=4040
T-bar=8.46
Migrant
Stocks
Overall
Between
Within
34.3
312
265
0
0
10200
N=48500
n=4590
T-bar=10.6
Education in
Country j
Overall
Between
Within
-0.116
2.01
1.55
0.453
-86.7
22.9
N=32600
n=3690
T-bar=8.84
BIT to Date
Overall
Between
Within
0.195
0.396
0.336
0.159
0
1
N=54083
n=6066
T-bar=8.92
Distance
(logged)
Overall
Between
Within
0.00245
0.0043
0.00358
0.00177
0
0.01
N=41400
n=5010
T-bar=8.27
Common
Language
Overall
Between
Within
0.00057
0.00366
0.00331
0.00121
-0.0159
0.0286
N=28300
n=3110
T-bar=9.11
Common
Border
Overall
Between
Within
0.000055
0.00143
0.00135
0.00049
-0.00851
0.00949
N=28300
n=3110
T-bar=9.11
Colonial
Relationship
Overall
Between
Within
-0.000037
0.000563
0.000485
0.000168
-0.00851
0.00949
N=28300
n=3110
T-bar=9.11
Oil
Reserves
Per Capita
Overall
Between
Within
-0.000336
0.00218
0.00173
0.000958
-0.00851
0.00949
N=26700
n=3930
T-bar=6.8
!
!
!
!
117!
Table 3.6: Descriptive Statistics From 3-Year Blocks of Data
Variable
Name
Category
Mean
Standard
Deviation
Minimum
Maximum
Number of
Observations
Δ FDI
Inflows
Overall
Between
Within
31.2
356
136
314
-7810
-1910
-5860
6320
2300
4900
6630
3340
1.99
Δ FDI
Inflows
(logged)
Overall
Between
Within
2.73
12.8
7.42
10.7
-22.8
-20.4
-29.5
22.6
21
34.8
6630
3340
1.99
Δ Other FDI
Outflows
from Country
i (logged)
Overall
Between
Within
11.6
19.6
13.3
15.3
-25.8
-25.4
-25.7
26.4
26.4
45.4
6620
3330
1.99
Political Risk
Overall
Between
Within
0.0186
0.114
0.0591
0.0927
0
0
-0.481
1
0.5
0.769
11400
4410
2.59
Migrant
Stocks
Overall
Between
Within
0.258
0.174
0.135
0.112
-0.537
-0.0741
-0.666
1.46
1.01
1.33
8430
3170
2.66
Education in
Country j
Overall
Between
Within
-0.459
1.59
1.11
1.24
-8.9
-4.3
-7.53
10.4
5.67
6.81
8680
3430
2.53
BIT to Date
Overall
Between
Within
Distance
(logged)
Overall
Between
Within
0.0224
0.0923
0.093
0
0
0
0.0224
0.77
0.77
0.0224
13300
3940
3.36
Common
Language
Overall
Between
Within
8.19
0.742
0.718
0
4.76
4.76
8.19
9.38
9.38
8.19
13000
3660
3.56
Common
Border
Overall
Between
Within
0.121
0.326
0.338
0
0
0
0.121
1
1
0.121
13000
3660
3.56
Colonial
Relationship
Overall
Between
Within
0.00975
0.0983
0.0916
0
0
0
0.00975
1
1
0.00975
13000
3660
3.56
!
!
!
!
!
!
Chapter 4:
Diaspora-Owned Firms and the Value of Social Networks
Abstract
A causal relationship between diaspora populations and bilateral foreign direct
investment has been established empirically, but the causal mechanism driving this
relationship remains unclear. Do diaspora-owned foreign firms enjoy a competitive
advantage in the homeland, and if so, what accounts for this? I argue that diasporans
have important social network based advantages over non-diaspora-owned firms, and
that these competitive advantages explain the relationship between diaspora
populations and investment flows. This paper uses data from an original survey of 174
foreign-owned firms in Georgia to explore mechanisms of diaspora difference,
focusing particularly on how firms use social networks in business. I find that
diaspora-owned firms use social networks more than other foreign firms in the
acquisition of real estate, and that owners and managers at diaspora-owned firms
perceive social networks to be more important to firm location decisions and to overall
profitability than do their counterparts at non-diaspora-owned firms. This indicates
that social networks provide important competitive advantages to diaspora-owned
firms, advantages that may explain the causal relationship between diaspora
populations and global flows of foreign direct investment.
!
!
!
118!
!
!
119!
Introduction
A recent literature in economics and political science has established a causal
relationship between the size of migrant populations and bilateral flows of foreign
direct investment (FDI) to the migrants’ homeland (Javorcik et al. 2011; Leblang
2010; Kugler and Rapoport 2007; Docquier and Lodigiani 2010). The amounts in
question are not small. While direct measures of the volume of diaspora investment
(investment by migrants and their descendents) are not available for most developing
countries, some estimates suggest that diaspora direct investment accounted for over
50% of FDI inflows to China during the 1990s (Huang 2003; Ye 2010) and 20-30% of
FDI flows into India during the same time period (Ye 2010). The survey discussed
below finds that 17% of foreign-owned firms in Georgia, including the largest firm in
the sample, have at least one diaspora owner.
Despite the substantive importance of migrant-induced investment, the causal
mechanisms driving the relationship have not been established empirically, and indeed
there is ambiguity as to whether these flows are attributable to diaspora-owned firms,
firms employing members of the diaspora, or other factors entirely (Kugler and
Rapoport 2007, Leblang 2010). This uncertainty remains because the existing
empirical literature examines aggregate cross-national flows of FDI, which do not
allow researchers to distinguish directly between flows of diaspora and non-diaspora
investment. To identify the causal mechanism linking migrant populations to flows of
investment, it is necessary to differentiate the behavior of diaspora-owned firms from
that of other foreign firms. Only at the firm level is it possible to determine whether
!
!
!
!
120!
diaspora-owned firms enjoy a competitive advantage in the country of origin and, if
so, to identify the source of that competitive advantage. This paper introduces the
results of a new firm-level survey designed that offers the first empirical comparison
between diaspora-owned firms and a comparable set of non-diaspora-owned foreign
firms.
Of primary interest is the use of social networks by diaspora-owned firms. I
argue that diasporans’ social networks in the homeland provide diasporans with two
important competitive advantages: access to privileged channels of information, and a
means of establishing trust with business counterparts. In concurrent work I use crossnational evidence on bilateral flows of FDI to show that migrant-induced FDI
responds more strongly and more quickly to changes in the level of political risk than
does FDI, a finding I attribute to diasporans’ access and attention to information
(Graham 2010). The most direct implications of this theory, however, are only testable
at the firm level. I test those implications here.
The core hypotheses of this paper are that diaspora-owned firms use social
networks more than their non-diaspora-owned peers and that firm owners and
managers at diaspora-owned firms perceive these networks to be more important to
firm profitability and firm location decisions that do their peers at non-diaspora-owned
firms. These hypotheses are tested using new survey data on 174 foreign-owned firms
operating in Georgia. This data allows direct comparison between diaspora-owned
!
!
!
!
121!
firms and non-diaspora-owned firms operating in the same risk environment.1 While
the analysis is based on firms operating in a single developing country, the theorized
differences between diaspora-owned and other foreign firms are expected to hold
across a wide range of diaspora communities. In terms of its investment climate in
particular, Georgia is a typical developing country, and hence a “typical case” in the
language of small-N case selection.
The paper is organized as follows. First, I review the relevant literature and
establish a set of hypotheses regarding the use of social networks by diaspora-owned
firms. Then I describe the details of the survey and the data it produced, and conduct
hypothesis tests. I finish by discussing open-ended responses in the survey, drawing
conclusions and outlining plans for future research.
A Review of the Current Literature
There is recent evidence that migrant populations cause flows of FDI from
their country of residence to their homeland (Javorcik et al. 2011, Leblang 2010,
Kugler and Rapoport 2007; Docquier and Lodigiani 2010). One possible explanation
is that diasporans have access to privileged channels of information about the
homeland via their social networks. Survey evidence from domestic firms in the
United States shows that social networks increase not only the quantity of information
firms have access to, but also the quality of information (Uzzi 1996, 1999; Yli-Renko,
Autio and Sapienza 2001). Similarly, transnational ethnic and social networks provide
privileged channels of information that allow efficient matching between potential
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
One hundred seventy-four firms responded to the survey. Due to item non-response, only 161 are used
in the analyses presented here.
!
!
!
!
122!
business collaborators and between capital and opportunity (Rauch and Casella 2003).
Migrant-induced FDI is more sensitive than other FDI to over-time changes in the
level of political risk in the homeland, which is consistent with the argument that
migrants’ levels of information about the homeland serves as an important mechanism
in this causal relationship (Graham 2010).
There is an extensive management literature that links the social networks of
firm owners and managers to a variety of benefits beyond mere access to information:
these include access to capital2 (Uzzi 1999; McMillan and Woodruff 1999), adaptation
of innovation (Abrahamson and Rosenkopf 1997; Moran 2005), and firm profitability
and survival (Hochberg, Ljungqvist and Lu 2007; Musteen and Francis 2010).3
Additionally, there is evidence that social networks are more important in
environments, like emerging markets, where market uncertainty is high (Podolny
1994). While networks are important to all these outcomes, simply adding more
connections to other firms may not increase positive outcomes in a linear fashion (e.g.,
Uzzi 1996; Gargiulo and Benassi 2000). However, the simple expectation presented in
this paper – that diaspora firm owners are able to use social networks to greater
advantage than non-diaspora foreign owners – is consistent with even these more
nuanced understandings of the effect of network structure.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2
Malesky and Taussig (2009) find that, while socially connected firms in Vietnam have greater access
to bank capital, relationship-based lending is not an efficient means of allocating capital. They find that
firms with access to bank capital are not more profitable than firms without access, and the most
successful investors forego bank capital entirely.
3
This literature overlaps with a literature in which the network position of firms in considered as an
independent variable explaining similar outcomes: because I am focused on differentiating on the
network access of firm owners in particular, I restrict this discussion to the literature references the
social networks of individuals rather than entire firms.
!
!
!
!
123!
Diasporans often have extended family members still living in the homeland,
and first-generation migrants in particular often still retain ties to friends in the
homeland as well (Gillespie et al. 1999; Sheffer 2006; Portes, Guarnizo and Haller
2002). While a literature on diaspora investment motivations has shown that
diasporans believe that their diaspora identity would give them a competitive
advantage in investing, it has yet to be demonstrated that this is the case (Gillespie,
Sayre and Riddle 2001). Anecdotal evidence suggests that many diasporans overestimate the degree to which their social ties will facilitate business success (Riddle,
Hrivnak and Nielsen 2008).
The questions that the current literature leaves unanswered are: do diasporans
actually retain the type of social networks that can enhance the profitability and
survival of diaspora-owned firms? Do diaspora-owned firms use social networks
more than their non-diaspora-owned counterparts, and do managers perceive these
networks to be important to the profitability of their firms? I answer those questions
here.
Theory: Diaspora-Owned Firms and Social Networks
Both the causal link between migrant stocks and bilateral FDI flows found in
the literature, and the sensitivity of diaspora direct investment to political risk are
consistent with diasporans using social networks to gain access to information. As
noted above, at least some diasporans believe that the social networks they retain in
the homeland are useful in business. However, it is possible that the types of social
networks diasporans retain are, in fact, of no business use at all. It may be that
!
!
!
!
124!
diasporans’ friendship and family ties, and the obligations that come with them, are an
impediment to running a profitable business rather than an advantage. Hence, the core
empirical tasks of this paper are to establish whether diaspora-owned firms use social
networks more or differently than their non-diaspora-owned peers, and whether firm
owners and managers at diaspora-owned firms report that social networks are more
important to firm profitability and firm location decisions than do their peers at nondiaspora-owned firms.
There are three primary ways in which I expect diaspora-owned firms to use
social networks to increase the profitability of their firms:
1) Social networks provide privileged channels of information about potential
risks and opportunities for investments in the homeland;
2) Social networks serve as an informal means of enforcing contracts with
other firms and individuals in the homeland;
3) Relational norms, such as trust, can operate within social networks as
substitutes for formal contracting (e.g., Granovetter 1985; Adler 2001).
These three uses for social networks are not mutually exclusive. These uses
may complement each other and overlap, and a single firm may use social networks in
all of these ways simultaneously. Indeed, it is plausible that social networks allow the
formation of trusting relationships between foreign firms and the firms and individuals
with which they interact, and that this trust facilitates the sharing of information and
collaboration to enforce contracts.
!
!
!
!
125!
Diasporans’ Use of Social Networks: Access to Information
Incomplete and asymmetric information is a major cause of the international
immobility of capital and an impediment to international trade (Frankel 1992; Gordon
and Bovenberg 1996).4 Social networks increase not only the quantity of information
that firms have access to, but also the quality of information (Uzzi 1996, 1999;
Premaratne 2001). Diasporans, and particularly first-generation migrants, retain access
to social networks in their homelands that other potential foreign investors generally
do not have access to. If diasporans are better informed, they should have greater
ability to identify and exploit profitable business opportunities in the homeland and to
anticipate risks that may arise there. While previous literature has asserted that
diasporans use social networks to access information (e.g. Freinkman 2002; Bandelj
2002), this has not been tested directly.5
One of the implications of diasporans’ use of social networks to access
information is that they should be better able to anticipate and avoid certain types of
political risk. Let us take as an example the type of transfer risk that can arise when a
country passes new capital control laws. Some diaspora-owned firms may become
aware that such laws are under consideration before deliberations are made public –
they may have friends or family members that work in government or otherwise have
access to classified or private information. Probably much more common, however, is
that early information about laws under consideration is reported only in the local
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4
For a review of the literature on “home bias” in investing, see Lewis (1999).
It is interesting that one of the few existing studies that explore diaspora business advantages at the
firm level, Schulte (2008), mentions a linguistic advantage but makes no mention of social networks at
all.
5
!
!
!
!
126!
language press. Furthermore, only a relatively small number of well informed
homeland residents may be able to accurately forecast the likelihood that the laws
being considered will ever be passed, much less take effect, and their opinions may
not be widely circulated. Therefore, regarding any change in the political or legal
system, local social networks provide access to the type of knowledge needed to
anticipate adverse (or favorable) changes in the political and legal situation before
they occur. Diasporans’ social networks may be useful in this role even if those
networks do not include highly placed government officials or others with access to
secret information.
Access to information, however, is useful in more contexts than simply
anticipating government behavior. We might expect social networks to be most
effective at providing information involving actors that enjoy less media coverage than
the central government. Particularly in less-developed markets, lack of information
often prevents firms from identifying trading opportunities and leads to inefficient
matching between buyers and sellers (Solinger 1989; Ronnas 1992; McMillan 1997).
If diasporans’ social networks are effective in delivering market information as well as
risk information, it gives them a competitive advantage even in countries where
political risk is low.
Access to market and risk information is an advantage that local firms
generally have over foreign firms, and that diaspora-owned firms may have over other
foreign firms.
!
!
!
!
127!
Diasporans’ Use of Social Networks: Trust and Contract Enforcement
The law and investment literature suggests that formal contracting institutions
and social networks are substitutes. Either firms can rely on formal institutions to
enforce contracts and protect them from opportunistic behavior, or they can rely on
social networks and personal relationships to play these same roles. The
transformation from kinship and social ties to formal institutions as an organizing
social principle is often viewed as a pivotal step in the process of economic
development (Berry 1993). When formal institutions are “weak, hostile, or
indifferent,” social structures can substitute for them (Woolcock and Narayan 2000).
Wang (2000) argues that informal Guanxi networks in China substitute for weak
formal institutions, and that members of the Chinese diaspora serve as “agents of
transmission” who connect foreign firms with these networks and thereby facilitate
FDI.6
There are two related mechanisms through which social networks may be used
to substitute for formal contracting institutions. The first is by providing a direct
means of enforcing contracts and sanctioning opportunistic behavior. The second is by
establishing trust between diaspora-owned firms and potential business counterparts,
reducing the likelihood that diaspora-owned firms are victimized by opportunistic
behavior.
First, social networks might be expected to facilitate direct enforcement of a
formal contract. Consider a hypothetical dispute between a diaspora-owned firm and a
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
6
Weidenbaum and Hughes (1996) make similar observations.
!
!
!
!
128!
domestic customer that has received a shipment of goods from the diaspora-owned
firm but refuses to pay. If the court system in the homeland is efficient and reliable,
the diaspora-owned firm can bring suit against the domestic customer and eventually
win payment of their claim. If the domestic courts are slow, expensive, or unreliable, it
might be more profitable for the diaspora-owned firm to rely on informal means of
enforcing the contract. If the owner of a diaspora firm has friendship or family ties
with a large number of other firms in the same industry, she can encourage these firms
to refuse to do business with the customer in question until the debt is paid. If the
diaspora-owned firm and the customer have friendship or family ties with each other,
they may opt for informal mediation of the dispute by another friend or family
member to avoid rupturing the social bond between them in an all-out dispute. In
either scenario, the social ties of the diaspora owner facilitate the informal, rather than
the judicial, resolution of the dispute. While potentially useful in any institutional
environment, these informal contract enforcement and dispute resolution mechanisms
are most valuable when formal contracting institutions are expensive, slow, or
unreliable.
It is also possible for diasporans to use social networks as a more indirect
substitute for, or complement to, formal contracting institutions. The role of relational
trust in reducing transaction costs and increasing economic efficiency is well
established at the cross-national level (e.g. Knack and Keefer 1996; Knack and Zak
1999) and at the firm level (Sako 1992; Barney and Hansen 1994; Chow and Holden
!
!
!
!
129!
1997).7 Reputation, trust, and norms of reciprocity are important factors determining
the duration and stability of exchange structures between firms and individuals (e.g.
Granovetter 1985; Larson 1992; Adler 2001).
If diasporans retain social networks in the homeland, they do not start from
scratch in forming important business relationships when they start a new business or
expand the operations of an existing business into the homeland. Consistent with the
theory outlined above, social ties are expected to provide a level of trust not only
between a diasporan and his/her family and friends: individuals with whom the
diasporan shares a social tie can serve as the diasporan’s advocate in forming trusting
relationships with people the diasporan has never met.8 Diasporans’ social networks
in the homeland increase the degree to which diaspora-owned firms can use social
trust to lower the likelihood that they are victimized by opportunistic behavior.
Testable Implications
If diaspora-owned firms have access to business-relevant social networks in
the homeland, and if they use these social networks to gain competitive advantage, this
should manifest itself in differences between diaspora- and non-diaspora-owned
foreign firms across a range of behaviors. The first implication addresses firms’ initial
decision to invest in Georgia. The initial decision to enter the country is crucial
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
7
Jeffries and Reed (2000) find that too much trust in a relationship, as well as too little trust, can
damage firms’ abilities to solve problems of adaptation in relational contracting. This is one of several
findings that complicate the relationship between trust and contracting efficiency, but the broad
consensus in the literature is that higher levels of trust in business relationships is generally beneficial to
both parties.
8
Wong and Boh (2010) examine managers’ intrafirm networks to demonstrate the ability of third
parties to serve in this capacity. I extend this logic to diasporans’ interfirm networks.
!
!
!
!
130!
because part of what we are seeking to explain is the impact of diaspora populations
on flows of FDI. Do social networks drive the decision of diaspora-owned firms to
invest in the homeland?
Hypothesis 1: Diaspora-owned firms are more likely than other foreign firms to cite
“your firm’s friendship and family ties to Georgia” as an important positive factor
influencing their decision to invest in Georgia.
Social networks will affect firm entry decisions if owners and managers expect
these networks to augment firm profitability in the new location. However, it is also
important to look at whether those expectations prove realistic. Are social networks
important to the profitability of diaspora-owned firms? Perhaps the most
straightforward means of assessing the importance of social network relationships to
firm profitability is to ask respondents to assess this importance subjectively. This has
the advantage of capturing a broad range of mechanisms through which social
networks might convey advantage in a single metric.
Hypothesis 2A: Diaspora-owned firms report that their owners’ and managers’ family
relationships are more important to the firm’s profitability than do their counterparts
at non-diaspora owned firms.
!
!
!
!
131!
Hypothesis 2B: Diaspora-owned firms report that their owners’ and managers’
friendships are more important to the firm’s profitability than do their counterparts at
non-diaspora owned firms.
A more objective metric is provided by exploring firms’ use of social networks
in the purchase/rental of real estate. Almost all firms, regardless of sector, require at
least some form of real estate – be it for offices, storefront, warehouses, or other use.9
Therefore, we might expect that the purchase/rental of real estate is a task facing most
firms and one in which they will deploy social networks if it is advantageous to do so.
If diaspora-owned firms are better able than other foreign firms to use social networks
to augment their profitability, this is one area where we would expect this difference to
manifest itself.
Hypothesis 3A: Diaspora-owned firms are more likely than non-diaspora owned firms
to report having rented or purchased real estate with the help of a family member of
one of the firm’s owners or managers.
Hypothesis 3B: Diaspora-owned firms are more likely than non_diaspora owned
firms to report having rented or purchased real estate with the help of a friend of one
of the firm’s owners or managers.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
9
When testing hypotheses 3A and 3B, I omit firms in the real estate sector on the grounds that this
question has a distinctly different meaning to firms whose business is the buying and selling of real
estate.
!
!
!
!
132!
These hypotheses allow us to test for differences in social network use between
diaspora- and non-diaspora-owned foreign firms with regard to firm entry decisions
and later firm behavior, in terms of concrete firms behavior as well as subjective
impact on profitability.
Why Georgia?
In the language of small-N case selection, Georgia is a “typical case” for a
developing country with regard to its wealth and political institutions, and an “extreme
case” regarding the size of its diaspora population.10 Because cross-national data do
not differentiate between diaspora and non-diaspora FDI, it is difficult to accurately
assess the volume of diaspora direct investment in a given country before a survey is
conducted. Georgia has a large diaspora population, receives substantial remittances,
and receives a large amount of FDI relative to the size of its GDP, making it likely ex
ante Georgia was a recipient of substantial diaspora investment. However, as the
figure below demonstrates, Georgia is a relatively typical developing country in terms
of its wealth and the quality of its institutions.
Only in terms of war risk does Georgia score substantially higher than the
median developing state, and its level of war risk is typical of fragile states. Georgia’s
score of 5 (out of 7) for war risk in 2009, the year after it experienced a small,
weeklong war with Russia, was the same as its level of war risk in 2003, the first year
for which data is available. I argue that, despite the 2008 conflict, Georgia continued
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
10
See Seawright and Gerring (2008) for a discussion of the use of “extreme cases” and “typical cases”
in hypothesis testing. As discussed in related work, the political relationship between the Georgian
government and the Georgian diaspora at the time of this writing was generally positive and
uncomplicated, with efforts made by the Georgian government to promote diaspora investment
(Graham 2011).
!
!
!
!
133!
throughout the 2000s, to be a business climate typical of developing states – one in
which political turmoil and the risk of political violence were ever-present, but only
occasionally acute, sources of risk for foreign investors.
!
!
!
!
134!
Table 4.1:
Georgia in Comparison to Other Fragile and Developing Countries
Georgia
Fragile states
Developing States
median
Median
(25th -75th
(25th - 75th
percentile)
percentile)
GDP
5,500
12,300
6,500
(millions of USD)
(5,500 – 55,000)
(1,700 – 24,000)
GDP per capita
1,300
1,200
1,100
(USD)
(280 – 2,200)
(380 – 2,000)
Population
4.3
20
9.7
(millions)
(7.3 – 41)
(3.1 – 27)
FDI Inflows
1,600
790
748
(Millions of USD)
(120 – 2,600)
(120 – 2,200)
FDI Inflows
12
3.6
4.3
(% of GDP)
(1.9 – 6.1)
(1.9 – 7.9)
Remittances
73
73
73
(Millions of USD)
(15 – 270)
(8.1 – 290)
Remittances
5.7
2.0
4.3
(% of GDP)
(0.7 – 5.7)
(0.4 – 5.6)
Emigrant Stock
48,000
110,000
88,000
(1990)
(24,000 – 300,000) (23,000 – 330,000)
Emigrant Stock
190,000
216,863
168,220
(2005)
(82,000 – 640,000) (36,000 – 490,000)
Emigrant Stock
4.3
1.9
2.0
(% of population) (2005)
(0.7 – 3.0)
(0.9 – 4.4)
Cost of Dispute Resolution
35
30
33
(% of claim)
(23 – 44)
(23 – 45)
War Risk (1-7)
5
5
3
(Higher = Riskier)
(3 – 6)
(2 – 4)
Government Risk (1-7)
4
4
4
(Higher = Riskier)
(4 – 5)
(3 – 5)
Transfer risk (1-7)
6
6
5
Higher = Riskier
(4 – 7)
(4 – 6)
Fragile countries n = 31; Developing countries n = 107
The range from 25th percentile to 75th percentile is given in parentheses.
All items are measured as of 2008, except for the measures of migrant stocks, whose
years are given.
I define developing countries as those with a 2008 GDP per capita of less than
$3,855 – those countries defined by the World Bank as low income and lower middle
income.
Fragile states are those that meet any of the following criteria: 1. A violent intrastate conflict in the last 3 years; 2). A change of executive leadership via coup or
military overthrow in the last three years; 3) failure to control all their territory 4)
occupation by a foreign power or not controlled by a unified entity (i.e. coded as
“Interregnum” or “Interruption” in the Polity IV data).
!
!
!
!
135!
The Survey
The Capital and Conflict: Georgia survey is the first firm-level survey (to my
knowledge) that compares diaspora-owned firms to other foreign firms. Surveys were
conducted, in person, with the owner or manager of 174 foreign owned firms between
February and June 2010. Each respondent was given a choice of taking the survey in
English or Georgian.11
The sampling frame was derived from a list of foreign-owned firms provided
by the Georgian Ministry of Finance. The list included all foreign firms that met the
following criteria: 1) a for-profit enterprise; 2) at least 10% foreign ownership; 3)
registered as active and paying taxes as of June 1, 2009; 4) obtained its first
registration in Georgia after the year 2000. This sample was supplemented with a
randomly drawn sample of 300 of the 450 firms that responded to the Ministry of
Finance’s Balance of Payments survey in 2009. These firms also met criteria 1-3, but
some were initially registered prior to 2000.12
Diasporans are defined in the literature as a subset of migrants and their
descendents: individuals that, in some way retain a relationship with their homeland
and identify themselves as part of a community associated with that homeland (e.g.
Safran 1991; Gillespie et al. 2001; Sheffer 2006). To identify diaspora-owned firms,
enumerators read the following text:
I’m going to refer to the “Georgian diaspora” in some of the
following questions. The Georgian diaspora includes all individuals
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
11
The survey was taken through three rounds of reverse translation and refinement to ensure
equivalence between the English and Georgian versions.
12
Some firms on the Ministry of Finance list also turned out to be registered before the year 2000, but
had been re-registered after that date.
!
!
!
!
136!
who live outside of Georgia but who consider themselves to be
Georgian. It includes people who were born in Georgia and emigrated
to other countries. It also includes people who were born in other
countries, but whose ancestors are from Georgia. If a Georgian owner
of the firm was living abroad when he decided to start this firm in
Georgia, we count this as diaspora investment even if the owner
moved back to Georgia.
Are any of the owners of your firm members of the Georgian
diaspora?
The answer to this question is used to categorize firms as either diasporaowned or non-diaspora-owned.13
Survey Non-Response
Attempts were made to contact a total of 1,024 firms between February and
June 2010, representing over 80% of the foreign-owned firms that officially entered
Georgia during the target period. Only 484 could be contacted and, of those, only 362
met the criteria listed above (foreign ownership, for-profit, and currently operating).14
Many of the firms that could not be contacted were closed – some had closed recently
and others had closed before 2009 but had not been purged from the tax rolls. Other
firms could not be contacted because they had changed addresses and phone numbers
since the lists were updated. Georgia has no up-to-date telephone or address directory,
and so only a minority of the firms that have relocated within the last several years
could be contacted. Few of the foreign firms operating in Georgia maintain websites.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
13
Respondents at 12 firms either did not know or chose not to answer this question, and these firms are
not used in analyses.
14
We only attempted to contact firms that the Ministry of Finance or State Department of Statistics
indicated met the three criteria for inclusion, but nonetheless many firms on the lists provided were
either non-profit, fully Georgian-owned, or were closed.
!
!
!
!
137!
Of the 362 firms that were successfully contacted and that met the basic
criteria for inclusion in the sample, surveys were successfully conducted with 167; 195
firms refused to participate. In most firms, enumerators were successful in scheduling
interviews with either the firm owner or a firm manager. In some cases, however, this
was not possible and a shorter version of the survey (10 of 53 questions) was asked of
front-desk or other available personnel. These “front desk” surveys asked questions
about firm demographics, such as sector, headquarters country, firm size, and diaspora
ownership. The front desk surveys do not include sufficient information to include
these firms in most hypothesis tests, but do provide valuable additional information
about firm demographics in the sampling frame.
The Sample
The sampling lists discussed above produced 167 respondents: 26 diasporaowned firms and 129 non-diaspora-owned firms (12 respondents did not answer the
question on diaspora ownership).15 Because of the low number of diaspora-owned
firms, a non-random supplement of seven additional diaspora-owned firms was added
to the sample, bringing the total number to 174.16 Table 4.12 in the appendix provides
a comparison between the diaspora-owned firms in the random sample and those in
the supplemental sample.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
15
All firms in the sample are privately owned, i.e. none are government-owned or publicly traded.
Two of these supplemental firms were identified by the State Office for the Diaspora; five more were
located via snowball sampling. To create the snowball supplement, enumerators asked respondents at
the end of the interviews if they knew of any diaspora-owned firms that we might be able to contact.
16
!
!
!
!
138!
Table 4.2: Number of Respondents by Diaspora Ownership
Survey Type
Owners/Manager’s
Survey
Front Desk Survey
Total
Diasporaowned firms
(random
only)
23
Diasporaowned firms
(supplement
included)
30
Non-diasporaowned firms
Diaspora
ownership
unknown
Total
121
10
161
3
26
3
33
8
129
2
12
13
174
The only demographic information available from firms that refused to
participate in the survey is the firm’s home country, which in most cases could be
drawn from the firm lists provided by the Georgian government.17 Therefore, I
examine the number of respondents and non-respondents from each region. The
following table shows that refusals are distributed evenly across regions. The
percentage of diaspora-owned firms is smaller in the West than in other regions, but
because this is the largest region in terms of number of firms, the number of diaspora
firms from the West is still substantial.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
17
Not all firms have home countries listed on the original sampling lists. For those firms that responded
to the survey but did not have a home country listed, I treat the location of the firm headquarters as the
home country. Some respondents did not report a headquarters location outside of Georgia – the home
country of those firms remains as missing data.
!
!
!
!
139!
Table 4.3:
Distribution of Diaspora and Non-Diaspora Ownership
by Home Region
Region
Former Soviet
Union
W. Europe,
U.S. &
Australia
Middle East
Other
Region
Unknown
Total
Number of
Qualifying
Firms
Contacted
69
Number of
Responses
Percentage
Non-Response
Percentage of
respondents who
are diaspora-owned
31
55%
23%
162
73
55%
13%
68
14
56
31
7
24
54%
50%
57%
27%
14%
33%
369
174
56%
20%
Within the set of firms that were successfully interviewed, I compare diaspora
and non-diaspora-owned firms across a range of covariates. By establishing that firm
demographics between the two groups are similar, I am then better able to attribute
any observed differences in firm behavior to diaspora ownership itself, rather than
alternative factors like firm size or sector. Firm size is captured by the number of
employees; I use dummy variables for minority, majority, and 100% foreign
ownership, and Greenfield investment, as whether the firm is located in the capital city
of Tbilisi. I also include dummy variables for primary, secondary, tertiary, and
quaternary sectors. Because of the 2008 war between Russia and Georgia, I also
include a dummy variable for whether the firm is headquartered in Russia.
With regard to the demographic characteristics of individual respondents, I
examine whether the respondent was an owner of the firm (as opposed to a manager)
!
!
!
!
140!
and whether the respondent self-identified as ethnically Georgian. I also examine the
language in which the survey was conducted.
Figure 1: Demographic Comparison
Diaspora Mean - Non-Diaspora Mean
Location = Tbilisi
HQ = Russia
Respondent = Owner
Tertiary Sector
Secondary Sector
Primary Sector
100% Foreign Ownership
Greenfield Investment
Firm Age (Years)
Small Firm
Firm Size (# of Employees)
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
NOTE: Units = standard deviations, horizontal lines give 95% confidence in
Figure 4.1: Demographic Comparison
*Note: Units = standard deviations, horizontal lines give 95% confidence intervals.
!
!
!
!
141!
The demographic characteristics for which I observe the large difference
between diaspora and non-diaspora-owned firms are the whether the respondent is the
firm owner and whether the firm is headquartered in Russia. At each firm,
enumerators asked for a meeting with the firm owner or manager. At diaspora-owned
firms, the firm owner was more likely to be available for interview. This makes sense
if diaspora owners are more likely than other foreign owners to live in Georgia at least
part time while running their business. All of the respondents at diaspora-owned firms
self-identified as ethnically Georgian, as did 89% of respondents at non-diasporaowned firms. A higher proportion of diaspora-owned firms are headquartered in
Russia, which can be attributed to the large size of the Georgian population living in
Russia.
Firm owners may have somewhat different perspectives on the behavior of
their firm than do managers. Similarly, firms based in Russia, which has a hostile
relationship with Georgia, may face unique political risks and have unique uses for
social networks. This necessitates a multiple-regression framework controlling for the
identity (owner vs. manager) of the respondent and whether or not the firm is
headquartered in Russia. Other firm demographic characteristics are included as
supplemental controls, but do not substantively affect results.
The range in firm size in the sample is substantial, with firms varying from a
single employee to as many as 1400. However, as Figure 1 demonstrates, the
difference in firm size between diaspora-owned and non-diaspora-owned firms is not
large. Three-quarters of firms in both groups qualify as “small” enterprises: 75% of
!
!
!
!
142!
diaspora-owned firms and 77% of non-diaspora-owned firms have 50 employees or
less. Similar consistency is observed in the distribution of firms between sectors,18 the
distribution of firms between regions (not pictured), the share of foreign ownership,
and firm age.19
Item Non-Response, Multiple Imputation, and Summary Statistics,
In dealing with them missing values that arise from item non-response, I
employ the multiple imputation by chained equations technique developed by Patrick
Royston (2004, 2009).20 Multiple imputation reduces the bias and eliminates the
sample-size reduction associated with list-wise deletion, allowing full use of the
information collected in the survey (King et al. 2001).21 In the raw data, the
independent variable of interest, diaspora ownership, is 6% missing, firm demographic
characteristics used as controls are between 2% and 18% missing, and the dependent
variables range from 9% to 11% missing.22
Item-non-response is relatively low in the data, and multiple imputation does
not substantively alter the results of the analysis. Similar coefficients of interest are
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
18
I also checked the balance in a composite of real estate and construction – the balance here is quite
even. I checked because these sectors were particularly hard-hit during the downturn.
19
It is worth noting that very few firms in the sample entered after the 2008 conflict with Russia: of the
18 that entered during this period, 5 were diaspora-owned and 12 non-diaspora-owned (1 unknown).
Most of the firms in the sample entered Georgia as Greenfield investments, rather than mergers,
partnerships, or acquisitions of Georgian firms: 76% of diaspora-owned and 78% of non-diasporaowned firms.
20
This is implemented using the ice and mim commands in Stata 10. I create 10 imputed datasets for
analysis.
21
The coefficients of interest are similar in analyses using list-wise deletion.
22
These percentages refer only to the 161 firms that completed an owners/managers survey. Those
firms where only a front-desk survey was completed are omitted from analysis (and imputation)
because none of the social network questions are included on the front desk survey.
!
!
!
!
143!
estimated using list-wise deletion. None of the substantive results reported are altered
by the decision to impute missing values.
Table 4.4: Summary Statistics
Variable
Diaspora Ownership
Respondent = Owner
Primary Sector
Secondary Sector
Tertiary Sector
Quaternary Sector
100% Foreign Ownership
HQ in West
Number of Employees
Tbilisi (location)
HQ in Russia
Real Estate Sector
Friendships (Importance)
Family (Importance)
Friendship (Real Estate)
Family (Real Estate)
Social Networks (Location)
Observations
Mean
151
157
156
156
156
156
157
132
144
161
132
158
148
147
143
144
148
0.199
0.242
0.0705
0.301
0.487
0.141
0.599
0.53
89.4
0.789
0.083
0.057
4.149
2.585
0.252
0.118
0.278
Standard
Deviation
0.4
0.43
0.257
0.46
0.501
0.349
0.492
0.501
218
0.401
0.28
0.232
2.183
1.944
0.436
0.324
0.449
Minimum
Maximum
0
0
0
0
0
0
0
0
1
0
0
0
1
1
0
0
0
1
1
1
1
1
1
1
1
1400
1
1
1
7
7
1
1
1
A Theoretical Note About Diaspora Ownership
One important issue when moving from findings regarding migrant-induced
FDI in the economics and political science literature to hypotheses regarding the
behavior of diaspora-owned firms is that, in addition to investments by firms they
own, migrants may induce investment in the homeland by firms from the country of
residence they do not own. In particular, foreign firms can hire diasporans, perhaps
gaining many of the advantages of diaspora ownership. The Capital and Conflict:
Georgia survey does not distinguish between firms that hire diasporans and those that
do not. By pooling across non-diaspora-owned firms that do and do not hire
diasporans, I potentially bias results toward a finding of no difference between
diaspora-owned and other foreign firms.
!
!
!
!
144!
Results: Testing Hypothesis 1
Hypothesis 1 asserts that diaspora-owned firms are more likely than foreign
firms to cite friendship and family ties to Georgia as an important reason for locating
their investment in Georgia. Each respondent was given a set of 12 factors on show
cards and asked to first sort the cards between factors that positively affected their
firm’s initial desire to invest in Georgia and those that negatively affected it.
Respondents were then asked to rank order the most important positive and negative
factors.
The dependent variable in Table 4.5 is whether or not a respondent ranked
“your firm’s family and/or friendship ties to Georgia” among the top three positive
factors influencing their firm’s initial decision to invest in Georgia.23
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
23
The mean number of positive factors identified by diaspora owned firms is 3.03, for non-diasporaowned firms the mean is 3.14. This similarity indicates that the difference in the probability of listing
social networks among the top 3 positive factors is not driven by the total number of positive factors
firms in each group identify.
!
!
!
!
145!
Table 4.5: Friendship and Family Ties in Firm Entry Decisions
Table 4.5: Friendship and Family Ties in Firm Entry Decisions
(1)
(2)
(3)
Diaspora-Owned Firm
0.808***
0.671**
0.647**
(0.265)
(0.282)
(0.300)
Respondent = Owner
0.718***
(0.262)
0.667**
(0.275)
Primary Sector
0.207
(0.571)
Secondary Sector
0.298
(0.443)
Tertiary Sector
-0.189
(0.449)
100% Foreign Ownership
-0.238
(0.244)
Employees (logged)
-0.115
(0.0845)
Greenfield
0.0520
(0.311)
Location = Tbilisi
-0.367
(0.298)
HQ = Russia
0.454
(0.424)
Constant
Observations
-0.785*** -0.960***
(0.134)
(0.146)
161
161
-0.319
(0.559)
161
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Consistent with Hypothesis 1, the results in Table 5 show that diaspora-owned firms
are more likely than other foreign firms to cite friendship and family ties to Georgia as
an important reason for their initial investment. This difference is significant at the
!
!
!
!
146!
0.05 level in all specifications. In models 2 and 3, I control for whether the respondent
is an owner (as opposed to a manager) because owners comprise a higher proportion
of respondents among diaspora-owned than among non-diaspora-owned firms. In
Model 3, I also control for sector,24 a dummy variable for 100% foreign ownership,
firm size (the logged number of employees), and a dummy variable for Greenfield
investment. I continue to use these control variables in subsequent hypothesis tests of
this form.
In substantive terms, 50% of diaspora-owned firms rank “your firm’s
friendship and family ties with Georgia” among the top three positive reasons
influencing their firm’s decision to enter Georgia, while only 21% of non-diasporaowned rank social networks this highly.
Results: Testing Hypothesese 2A and 2B
Hypothesis 2A states that diaspora-owned firms report that their owners’ and
managers’ friendships are more important to the profitability of the firm than do nondiaspora-owned firms. Hypothesis 2B makes the same prediction regarding family
relationships.
These questions ask respondents to answer the following questions on a sevenpoint scale from “not important at all” to “extremely important.” “How important are
your owners’ and managers’ family relationships to increasing the profitability of your
firm?” While there are some drawbacks to this approach – primarily that the
responses to these questions are not anchored – there is no particular reason to believe
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
24
The quaternary sector is the omitted category.
!
!
!
!
147!
that respondents at diaspora-owned firms use a Likert scale in a systematically
different way than their counterparts at non-diaspora-owned firms. As was established
earlier, the demographics of respondents in the two groups are quite similar –
nonetheless, the ordered logit models presented in Tables 4.6 and 4.7 include a range
of control variables.
The dependent variable in Tables 4.6 and 4.7 is respondents’ answers on a
seven-point scale.
!
!
!
!
148!
Table 4.6: The Subjective Importance of Family Relationships
Table 4.6: The Subjective Importance of Family Relationships
(1)
(2)
(3)
Diaspora-Owned Firm
0.973** 0.918** 1.014**
(0.413) (0.420) (0.447)
Respondent = Owner
0.280
(0.361)
Primary Sector
0.734
(0.834)
Secondary Sector
1.224*
(0.650)
Tertiary Sector
1.109*
(0.656)
100% Foreign Ownership
-0.122
(0.322)
Employees (logged)
-0.0332
(0.118)
Greenfield
-0.777*
(0.402)
Location = Tbilisi
0.592
(0.446)
HQ = Russia
-0.167
(0.578)
0.878
(0.939)
1.459
(0.942)
1.648*
(0.934)
2.495**
(0.943)
3.206***
(0.957)
3.994***
(0.975)
161
Cut 1
Cut 2
Cut 3
Cut 4
Cut 5
Cut 6
Observations
0.154
(0.180)
0.699***
(0.188)
0.878***
(0.194)
1.680***
(0.224)
2.365***
(0.278)
3.143***
(0.382)
161
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
!
!
!
0.224
(0.381)
0.211
(0.197)
0.758***
(0.203)
0.937***
(0.211)
1.738***
(0.242)
2.424***
(0.283)
3.204***
(0.389)
161
!
149!
The results in Table 4.6 support Hypothesis 2A: there is evidence that
respondents at diaspora-owned firms believe their owners’ and managers’ family
relationships to be more important to firm profitability than do their counterparts at
non-diaspora-owned firms. This result is significant at the 0.05 level. Among
diaspora-owned firms, 37% report that family relationships are more than “somewhat
important” to firm profitability (five or higher on a seven-point scale), while only 15%
of non-diaspora-owned firms report this.
It is worth noting that firms engaged in Greenfield investments report that
family relationships are less important to their profitability. This is consistent with the
expectation that firms with large and relevant social networks are more likely to
engage in joint ventures, but counter to the expectation that foreign firms’ social
networks are substitutes for partnerships with domestic firms.
!
!
!
!
150!
Table 4.7: The Subjective Importance of Friendships
Table 4.7: The Subjective Importance of Friendships
(1)
(2)
(3)
Diaspora-Owned Firm
0.790**
0.661*
0.756*
(0.367)
(0.386)
(0.418)
Respondent = Owner
0.801**
(0.349)
Primary Sector
0.964
(0.761)
Secondary Sector
0.537
(0.543)
Tertiary Sector
0.421
(0.502)
100% Foreign Ownership
0.421
(0.333)
Employees (logged)
0.00195
(0.104)
Greenfield
-0.544
(0.367)
Location = Tbilisi
1.291***
(0.446)
HQ = Russia
0.274
(0.609)
0.309
(0.918)
0.788
(0.916)
0.993
(0.916)
1.925**
(0.916)
2.406**
(0.938)
3.417***
(0.967)
161
Cut 1
Cut 2
Cut 3
Cut 4
Cut 5
Cut 6
Observations
-1.127***
(0.210)
-0.701***
(0.192)
-0.515***
(0.187)
0.342*
(0.179)
0.785***
(0.193)
1.729***
(0.245)
161
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
!
!
!
0.902**
(0.347)
-0.993***
(0.221)
-0.558***
(0.204)
-0.369*
(0.200)
0.505**
(0.198)
0.957***
(0.212)
1.928***
(0.268)
161
!
151!
The finding in support of Hypothesis 2B is also significant at the 0.05 level,
and the observed difference, though slightly smaller, is more remarkable with regard
to friendships than with regard to family-relationships. Among diaspora-owned firms,
49% report that friendships are more than “somewhat important” to profitability, while
40% of non-diaspora-owned firms report this.
While non-diaspora firm owners cannot create family relationships in the
homeland, they can form friendships, and one might expect that the type of nondiaspora owners who choose to invest in a given developing country are those with the
densest social networks in that country. The findings in Table 4.7 suggest that, even if
this type of selection occurs, diaspora-owned firms still rely more heavily on their
owners’ and managers’ friendships than do non-diaspora-owned firms.
Results: Testing Hypotheses 3A and 3B
In testing Hypotheses 3A and 3B, I turn to survey questions about whether
firms use their owners’ or managers’ friends and family members to assist them in
renting or purchasing real estate, and the type of assistance they report receiving.
Respondents are asked if they have ever rented or purchased real estate with the help
of a friend of one of the firm’s owners or managers, and are then asked the same
question about family members. Firms in the real estate sector are excluded from these
regressions.
Again, results support both hypotheses. We observe significant differences
between diaspora-owned and non-diaspora-owned firms.
!
!
!
!
152!
Table 4.8: The Use of Family Relationships to Buy or Purchase Real Estate
Table 4.8: The Use of Family Relationships to Buy or Purchase Real Estate
(1)
(2)
(3)
Diaspora-Owned Firm
1.060*** 0.957*** 0.934**
(0.344)
(0.339)
(0.373)
Respondent = Owner
0.397
(0.305)
0.652*
(0.358)
Primary Sector
-0.389
(0.821)
Secondary Sector
-0.121
(0.562)
Tertiary Sector
0.0783
(0.528)
100% Foreign Ownership
0.0617
(0.355)
Employees (logged)
0.102
(0.0973)
Greenfield
-0.255
(0.354)
Location = Tbilisi
-0.719**
(0.356)
HQ = Russia
0.546
(0.540)
Constant
Observations
-1.394***
(0.208)
151
-1.486***
(0.240)
151
-1.270*
(0.694)
151
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Table 4.8 shows that diaspora-owned firms are more likely to report having
rented or purchased real estate with the help of a family member of one of the firm’s
owners or managers. Table 4.9 shows the results with regard to the help of a friend of
one of the firms’ owners or managers; the difference is stronger with regard to family
!
!
!
!
153!
members, and only of borderline with regard to friendships. Both tables are produced
via probit regressions.
Just 6% of non-diaspora-owned firms reported using owners’ or managers’
family relationships to purchase real estate, compared to 34% of diaspora-owned
firms. With regard to friendships, the divide was 21% for non-diaspora-owned to 39%
for diaspora-owned.
!
!
!
!
154!
Table 4.9: The Use of Friendships to Buy or Purchase Real Estate
(1)
0.519*
(0.287)
Diaspora-Owned Firm
Respondent = Owner
(2)
0.443
(0.294)
(3)
0.471
(0.309)
0.306
(0.265)
0.418
(0.289)
Primary Sector
0.723
(0.538)
Secondary Sector
-0.163
(0.406)
Tertiary Sector
0.0486
(0.388)
100% Foreign Ownership
0.334
(0.296)
Employees (logged)
-0.0905
(0.0849)
Greenfield
0.271
(0.308)
Location = Tbilisi
0.375
(0.352)
HQ = Russia
0.424
(0.500)
Constant
Observations
-0.773*** -0.841***
(0.138)
(0.152)
151
151
-1.449**
(0.538)
151
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Discussion of Results
Taken together, the hypotheses above provide strong evidence for core
theoretical expectations of the paper. They show that diaspora-owned firms make
greater use of social networks than do their non-diaspora-owned counterparts and that
respondents at diaspora-owned firms perceive these networks to be more important to
!
!
!
!
155!
firm location decisions and profitability than respondents at non-diaspora-owned
firms, despite the fact that most respondents in both categories are local, ethnicGeorgian firm managers.
The magnitude and consistency of these differences is particularly dramatic in
light of the fact that all questions are asked about the friends and family members of
firm managers as well as of firm owners. If friends and family members are important
to foreign firms, it is striking that non-diaspora-owned firms do not make more
aggressive use of the social networks of their managers to compensate for their
owners’ lack of family in the homeland. This suggests one of several things: that the
social relationships of diaspora owners are particularly valuable; that the value of
social ties is not easily transferred to non-diaspora-owned firms when they hire local
managers; or that firm culture or strategy in diaspora-owned-firms is conducive to the
use of social networks in a way that it is not in non-diaspora-owned firms. Developing
tests that can distinguish between these possibilities represents a compelling avenue
for future research.
How Firms Use Social Networks
The empirical analysis above establishes that owners and managers at
diaspora-owned firms perceive social networks to be more important to firm
profitability and entry decisions than do their counterparts at other foreign firms, and
that diaspora-owned firms use social networks more often, at least in buying and
renting real estate. It is also useful, however, to understand in more detail exactly how
diaspora-owned firms are using social networks.
!
!
!
!
156!
To explore this we evaluate the open-ended responses of firm managers to the
following question: “Can you briefly describe the effects that your owners’ and
managers’ friendships and/or family relationships have on the way your firm does
business?” These open-ended responses were recorded by the enumerator in the
language in which the response was given, and then translated later if necessary.
Mentions of trust are coded in two ways. First, as a binary measure of
whether the respondent indicates that the trust generated by social relationships is
useful to his firm in some way. Second, among those respondents who indicate that
social networks generate trust that is useful to the firm, we code whether the use of
trust mentioned is inter-firm, intra-firm, or non-specific. A similar approach is taken to
coding enforcement and information. Enforcement is coded one if the respondent
mentions using social networks for any form of dispute resolution, contract
enforcement, or sanctioning of opportunistic behavior. Information is coded one if
respondents mention using social networks for gathering information, soliciting
advice, evaluating risk, or identifying customers, suppliers, or clients, or to make
“contacts” more generally.
Non-responses and responses such as, “No,” or, “These relationships are not
important to our business,” are coded as missing for these purposes. Answers where
some use of social relationships was mentioned, even vague generalities such as
“Friendship plays a positive role in the conduct of business,” but in which the concept
of trust is not explicitly mentioned, are coded zero.
!
!
!
!
157!
Consistent with the results from the closed-ended questions, 70% of
respondents at diaspora-owned firms described at least some positive use of social
relationships by the firm, while only 42% of non-diaspora-owned firms offered such a
description. It is also noteworthy that, despite a phrasing of the question that allowed
respondents to discuss any negative effects that social networks have on the function
of their business, no such negative effects were mentioned.
Table 4.10: Ways Respondents Report That Their Firms Use Social
Networks
Diasporaowned
firms
Trust (Any claim by respondent that social
networks generate trust that is useful to their firm)
8
(38%)
Nondiasporaowned
firms
19
(37%)
All
Respondents
Specifically inter-firm trust
1
(5%)
5
(10%)
6
(8%)
Specifically intra-firm trust
2
(10%)
7
(14%)
9
(13%)
28
(39%)
Information (i.e. Identifying potential business
7
11
20
counterparts, and receiving advice or risk
(33%)
(22%)
(28%)
information)
Enforcement (i.e. Dispute resolution, contract
0
0
0
enforcement, or deterrence of opportunistic
(0%)
(0%)
(0%)
behavior)
Number of respondents who provided at least
21
51
73
some description of their firm’s use of social
networks.
Note: The numbers in “all respondents” exceeds the sum of diaspora-owned and nondiaspora-owned firms because the diaspora/non-diaspora identity of one responding firm is
not known.
Table 4.10 depicts broadly similar patterns of social network use by diasporaowned and non-diaspora-owned firms. It seems that differences between these two
!
!
!
!
158!
groups of firms are more in the percentage of firms that use social networks actively in
their business strategy, rather than in the types of uses these networks are put to.
Strikingly, we see that there are zero cases where anything related to contract
enforcement or dispute resolution is mentioned, despite the variety of other uses
respondents choose to list. Of 87 responses indicating at least some use of social
networks in business, none mention anything related to these areas. If firms are using
social networks to substitute for formal contracting institutions, this use should be
important enough and salient enough to be mentioned by at least some firms; it is
mentioned by none. There is a possibility that social desirability bias or related factors
lead firms that use networks in this way not to mention doing so, and further empirical
examination with new data is necessary.25
Thirteen firms mentioned using social networks to identify clients, customers,
suppliers, or other contacts; seven made reference to more generic uses of social
networks to gather information. This establishes that at least some firms use social
networks to gather information, a finding consistent with Hypothesis 3B, but most
remarkable in contrast to absence of any references to enforcement.
At the high end of the spectrum, twenty-eight respondents explicitly mentioned
the usefulness of social networks in establishing trust. Respondents mentioned
primarily that they prefer to do business with people or firms they know and trust, but
in a few cases they were more detailed about the benefits trust conveys, such as stating
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
25
More direct questions about firms’ use of courts and relationship with formal political institutions
were stripped from the survey during the pilot stage because they made respondents uncomfortable. In
the Post-Soviet context, questions about business-government relations are generally interpreted as
questions about corruption. I expect these questions to be less sensitive in the Philippines, where the
next iteration of this survey will be conducted.
!
!
!
!
159!
that the trust generated by social ties allows them to extend credit to customers they
would not otherwise extend credit to.
Conclusions
Diaspora populations residing in rich countries have been identified as an
important cause of FDI flows into developing countries. However, the existing
literature lacks the firm-level data necessary to identify the causal mechanisms behind
these flows. This paper introduces a theory of social-network based diaspora
advantage, and uses a new firm-level survey from the developing country of Georgia
to directly compare the behavior of diaspora-owned firms and other foreign firms
operating in the same environment. I show that diaspora-owned firms use social
networks more than non-diaspora-owned firms and that the managers and owners of
diaspora-owned firms view social networks as more important to both their firm’s
profitability and investment-location decisions than do their counterparts at nondiaspora-owned firms.
These findings offer the first direct empirical comparison between the behavior
of diaspora-owned and non-diaspora-owned firms. The differences in behavior I
demonstrate are consistent with the social-network-based theory of diaspora difference
that I present: that diasporans have access to social networks in the homeland and that
these networks provide important competitive advantages to diaspora-owned firms.
Future iterations of this survey are planned in India, the Philippines, and
Vietnam that will provide the opportunity to both establish that the findings presented
here are not unique to the Georgian case, and to develop stronger tests to parse
!
!
!
!
160!
between the different mechanisms through which diaspora-owned firms use social
networks to increase profitability. Of particular interest is the degree to which
diasporans’ social networks can substitute for courts and other poorly functioning
formal political institutions.
!
!
!
!
161!
References
Abrahamson, Eric, and Lori Rosenkopf. 1997. Social Network Effects on the Extent of
Innovation Diffusion: A Computer Simulation. Organization Science 8 (3):
289-309.
Acemoglu, Daron, and Simon Johnson. 2005. Unbundling Institutions. Journal of
Political Economy 113 (5): 949-995.
Adler, Paul S. 2001. Market, Hierarchy, and Trust: The Knowledge Economy and the
Future of Capitalism. Organization Science 12 (2): 215-234.
Bandelj, Nina. 2002. Embedded Economies: Social Relations as Determinants of
Foreign Direct Investment in Central and Eastern Europe. Social Forces 81 (2):
411-444.
Barney, Jay B., and Mark H. Hansen. 1994. Trustworthiness as a Source of
Competitive Advantage. Strategic Management Journal 15 (S1): 175-190.
Berry, Sara. 1993. No Condition is Permanent: The Social Dynamics of Agrarian
Change in Sub-Saharan Africa. Madison, WI: University of Wisconsin Press.
Chow, Simeon, and Reed Holden. 1997. Toward an Understanding of Loyalty: The
Moderating Role of Trust. Journal of Managerial Issues 9 (3).
Debass, Thomas, and Michael Ardovino. 2009. Diaspora Direct Investment (DDI):
The Untapped Resource. Washington, DC: United States Agency for
International Development.
Docquier, Frédéric, and Elisabetta Lodigiani. 2010. Skilled Migration and Business
Networks. Open Economies Review 21 (4): 565-588.
Frankel, Jeffrey A. 1992. Measuring International Capital Mobility: A Review. The
American Economic Review 82 (2): 197-202.
Freinkman, Lev. 2002. "Role of the Diasporas in Transition Economies: Lessons From
Armenia." 10013. The World Bank.
Foreign Service Institute. Engaging with Diaspora Communities: Focus on EAP, EUR,
and NEA: Summary Report. 2010. Foreign Service Institute.
Gargiulo, Martin, and Mario Benassi. 2000. Trapped in Your Own Net? Network
Cohesion, Structural Holes, and the Adaptation of Social Capital. Organization
!
!
!
!
162!
Science 11 (2): 183-196.
Gillespie, Kate, Liesl Riddle, Edward Sayre, and David Sturges. 1999. Diaspora
Interest in Homeland Investment. Journal of International Business Studies 30
(3): 623-634.
Gordon, Roger H., and A. Lans Bovenberg. 1996. Why Is Capital So Immobile
Internationally? Possible Explanations and Implications for Capital Income
Taxation. The American Economic Review 86 (5): 1057-1075.
Graham, Benjamin A.T. 2010. "Political Risk and Diaspora Direct Investment."
Presented at the Annual Meeting of the American Political Science
Association, Washington, DC.
Graham, Benjamin A.T. 2011. "Diaspora-Owned Firms and Social Responsibility."
Presented at the Annual Meeting of the Academy of Management, San
Antonio, TX.
Granovetter, Mark. 1985. Economic Action and Social Structure: The Problem of
Embeddedness. The American Journal of Sociology 91 (3): 481-510.
Greif, Avner. 1989. Reputation and Coalitions in Medieval Trade: Evidence on the
Maghribi Traders. Journal of Economic History 49 (4): 8857-82.
Greif, Avner. 1993. Contract Enforcablity and Economic Institutions in Early Trade:
The Maghribi Traders' Coalition. American Economic Review 83 (3): 525-48.
Guidolin, Massimo, and Eliana La Ferrara. 2007. Diamonds Are Forever, Wars Are
Not: Is Conflict Bad for Private Firms? The American Economic Review 97
(5): 1978-1993.
Hochberg, Yael V., Alexander Ljungqvist, and Yang Lu. 2007. Whom You Know
Matters: Venture Capital Networks and Investment Performance. Journal of
Finance 62 (1): 251-301.
Huang, Ye. 2003. Selling China: Foreign Direct Investment During the Reform Era.
Cambridge: Cambridge University Press.
Inkpen, Andrew. C., and Eric. W. K. Tsang. 2005. Social Capital, networks, and
knowledge transfer. Academy of Management Review 30 (1): 146-165.
Jeffries, Frank L., and Richard Reed. 2000. Trust and Adaptation in Relational
Contracting. The Academy of Management Review 25 (4): 873-882.
!
!
!
!
163!
Karlan, Dean, Markus Mobius, Tanya Rosenblat, and Adam Szeidl. 2009. Trust and
Social Collateral. Quarterly Journal of Economics 124 (3): 1307-1361.
King, Gary, J. Honaker, A. Joseph, and K. Scheve. 2001. Analyzing Incomplete
Political Science Data: An Alternative Algorithm for Multiple Imputation.
American Political Science Review 95: 49-69.
Knack, Stephen, and Philip Keefer. 1997. Does Social Capital Have An Economic
Payoff? A Cross-Country Investigation*. Quarterly Journal of Economics 112
(4): 1251-1288.
Kugler, Maurice, and Hillel Rapoport. 2007. International Labor and Capital Flows:
Complements or Substitutes? Economics Letters 94 (2): 155-162.
Lane, P. J., and M. Lubatkin. 1998. Relative Absorptive Capacity and
Interorganizational Learning. Strategic Management Journal 19 (5): 461-477.
Larson, Andrea. 1992. Network Dyads in Entrepreneurial Settings: A Study of the
Governance of Exchange Relationships. Administrative Science Quarterly 37
(1): 76-104.
Leblang, David. 2010. Familiarity Breeds Investment: Diaspora Networks and
International Investment. American Political Science Review: 1-17.
Lewis, Karen. 1999. Trying to Explain Home Bias. Journal of Economic Literature
38: 571-608.
Malesky, Edmund J., and Markus Taussig. 2009. Where Is Credit Due? Legal
Institutions, Connections, and the Efficiency of Bank Lending in Vietnam.
Journal of Law, Economics, and Organization 25 (2): 535-578.
McMillan, John. 1997. "Markets in Transition." In Advances in Economics and
Econometrics: Theory and Applications, eds. David M. Kreps and Kenneth F.
Wallis. Cambrdge: Cambridge University Press. 210-239.
McMillan, John, and Christopher Woodruff. 1999. Interfirm Relationships and
Informal Credit in Vietnam *. Quarterly Journal of Economics 114 (4): 12851320.
Podolny, J. M. 1994. Market Uncertainty and the Social Character of Economic
Exchange. Administrative Science Quarterly 39 (3): 458-483.
Portes, Alejandro, Luis Eduardo Guarnizo, and William J. Haller. 2002. Transnational
Entrepreneurs: An Alternative Form of Immigrant Economic Adaptation.
!
!
!
!
164!
American Sociological Review 67 (2): 278-298.
Premaratne, S. P. 2001. Networks, Resources, and Small Business Growth: The
Experience in Sri Lanka. Journal of Small Business Management 39 (4): 363371.
Rauch, James, and Allessandra Cassella. 2003. Overcoming Informational Barriers to
International Resource Allocation: Prices and Ties. Economic Journal 113: 2142.
Royston, Patrick. 2004. Multiple Imputation of Missing Values. Stata Journal 4 (3):
227-241.
Royston, Patrick. 2009. Multiple Imputation of Missing Values: Further Update of ice,
with an Emphasis on Categorical Variables. Stata Journal 9 (3): 466-477.
Safran, W. 1991. Diasporas in Modern Societies: Myths of Homeland and Return.
Diaspora 1: 93-99.
Sako, Mari. 1992. Prices, Quality, and Trust: Inter-Firm Relations in Britain and
Japan. Cambridge University Press.
Sapienza, Paolo, Anna Toldra Simats, and Luigi Zingales. 2007. "Understanding
Trust."
Schulte, Bettina. 2008. "Second Generation Entrepreneurs of Turkish Origin in
Germany: Diasporic Identity and Business Engagement." 56. Center on
Migration, Citizenship, and Development.
Seawright, Jason, and John Gerring. 2008. Case Selection Techniques in Case Study
Research. Political Research Quarterly 61 (2): 294.
Sheffer, Gabriel 2006. Diaspora Politics: At Home Abroad. Cambridge, MA:
Cambridge University Press.
Uzzi, Brian. 1996. The Sources and Consequences of Embeddedness for the Economic
Performance of Organizations: The Network Effect. American Sociological
Review 61 (4): 674-698.
Uzzi, Brian. 1999. Embeddedness in the Making of Financial Capital: How Social
Relations and Networks Benefit Firms Seeking Financing. American
Sociological Review 64 (4): 481-505.
Wang, Hongying. 2000. Informal Institutions and Foreign Investment in China. The
!
!
!
!
165!
Pacific Review 13 (4): 525 - 556.
Wasko, Molly McLure, and Samer Faraj. 2005. Why Should I Share? Examining
Social Capital and Knowledge Contribution in Electronic Networks of
Practice. MIS Quarterly 29 (1): 35-57.
Weidenbaum, Murray, and Samuel Hughes. 1996. The Bamboo Network: How
Expatriate Chinese Entrepreneurs are Creating a New Economic Superpower
in Asia. New York: Free Press.
Wong, Sze Sze, and Wai Fong Boh. 2010. Leveraging the Ties of Others to Build a
Reputation for Trustworthiness Among Peers. Academy of Management
Journal 53 (1): 129-148.
Woolcock, Michael, and Deepa Narayan. 2000. Social Capital: Implications for
Development Theory, Research, and Policy. The World Bank Research
Observer 15 (2): 225-249.
Ye, Min. 2010. "Ethnic Investors vs. Foreign Investors: The Impact of Diasporas on
Economic Liberalization in China and India." Presented at the Annual Meeting
of the American Political Science Association, Washington DC.
Yli-Renko, Helena, Erkko Autio, and Harry J. Sapienza. 2001. Social Capital,
Knowledge Acquisition, and Knowledge Exploitation in Young TechnologyBased Firms. Strategic Management Journal 22 (6-7): 587-613.
Zak, Paul J., and Stephen Knack. 2001. Trust and Growth. The Economic Journal 111
(470): 295-321.
!
!
!
!
166!
Appendix
Diaspora-Owned and Non-Diaspora-Owned Firms by Home Country
Table 4.3 in the text shows the percentage of non-responses and diaspora
ownership by region. Table 4.11 shows the number of responses, number of refusals,
and number of diaspora-owned and non-diaspora-owned respondents for each
investing country. Firms from Azerbaijan, France, Israel, the Netherlands, and the
United States were particularly likely to refuse to participate in the survey, but there is
no obvious common characteristic across these countries that appears to drive nonresponse.
!
!
!
!
167!
Table 4.11: Home Country by Response Type
Refused
Country
Armenia
Australia
Austria
Azerbaijan
Belgium
Bulgaria
Canada
Cayman Islands
China
Cyprus
Czech Republic
Estonia
France
Germany
Greece
Iran
Israel
Italy
Japan
Jordan
Kazakhstan
Latvia
Lithuania
Luxemburg
Marshall Islands
Netherlands
Panama
Philippines
Poland
Russia
Sweden
Switzerland
Syria
Turkey
UAE
UK
USA
Ukraine
Virgin Islands
Unknown
Total
2
0
6
10
3
1
0
1
0
1
3
0
5
10
5
2
9
4
1
1
2
3
2
0
2
9
2
0
0
12
0
6
1
23
1
19
20
3
2
26
197
Responded
Diaspora
Non-Diaspora
0
2
0
1
1
3
0
2
0
1
0
1
0
1
1
0
0
3
0
4
0
2
1
1
0
2
1
9
0
0
1
3
2
3
1
1
0
0
0
0
0
2
0
3
0
1
0
2
0
1
2
4
0
1
0
1
0
1
5
7
0
1
0
3
0
0
5
14
0
2
4
19
0
10
1
1
0
0
6
10
31
122
!
!
!
Total
4
1
10
12
4
2
1
2
3
5
5
2
7
20
5
6
14
6
1
1
4
6
3
2
3
15
3
1
1
24
1
9
1
42
3
42
30
5
2
42
350
!
168!
Evaluating the Supplemental Sample
Table 4.12 provides the results of difference-in-means tests comparing
diaspora-owned firms in the random sample from those in the supplemental sample.
The demographic differences between these groups of firms is minimal, as are the
differences in the firms’ assessments of the importance of social networks to firm
profitability. With regard to the importance of social networks to firm entry decisions,
and firms’ use of social networks to procure real estate. The largest difference is with
regard to the use of social networks to acquire real estate. However, even if the
supplemental firms are dropped from the analysis, the substantive results presented in
Tables 8 and 9 remain the same. Diaspora-owned firms are more likely to report
having used family relationships to assist in renting or purchasing real estate
(significant at the 0.05 and robust to the exclusion of supplemental firms), and slightly
more likely to report having used friendships in the same way (not statistically
significant).
!
!
!
!
169!
Table 4.12: Comparing Diaspora-Owned Firms from the Random Sample
and the Supplement
Variable
Respondent = Owner
Primary Sector
Secondary Sector
Tertiary Sector
Quaternary Sector
100% Foreign
Ownership
HQ in West
Number of Employees
Tbilisi (location)
HQ in Russia
Real Estate Sector
Friendships
(Importance)
Family (Importance)
Friendship (Real
Estate)
Family (Real Estate)
Social Networks
(Location)
Random
Sample
Mean
Supplemental
Sample Mean
Standard
Error (of
difference)
t-statistic
0.43
0.04
0.28
0.52
0.16
0.58
0.43
0.17
0.5
0.17
0.17
0.71
0.006
0.13
0.22
0.35
0.007
0.14
0.22
0.11
0.22
0.22
0.17
0.21
0.02
0.1
1.0
1.6
0.04
0.64
0.33
145
0.81
0.19
0.08
5.1
0.67
124
0.71
0.33
0
4.6
0.33
21
0.09
0.14
0.08
0.5
0.30
148
0.18
0.36
0.11
0.9
1.0
0.14
0.52
0.55
0.70
0.6
3.4
0.29
3.6
0.71
0.2
0.42
1.0
.2
0.2
2.1**
0.27
0.43
0.57
0.71
0.29
0.28
0.2
0.22
1.5
1.3
!
!
!
Difference
in means
Chapter 5:
Diaspora-Owned Firms and Social Responsibility
Abstract
A causal relationship between diaspora populations and bilateral foreign direct
investment has been established empirically, but the question of which elements of
diaspora difference are responsible for this relationship, and what this implies for
development, remains unanswered. A growing literature in economic sociology and
business suggests that diaspora investors are motivated by patriotism and other social
and emotional factors, endowing them with unique potential as a force for
international development. This literature argues that diaspora-owned firms are more
socially responsible than other foreign firms, and engage in a range of economicdevelopment-promoting behaviors when investing in the homeland: hiring more local
labor, paying higher wages, and making more contributions to charity. I argue instead
that diaspora-owned firms enjoy competitive advantages in the homeland based on
access and attention to information. I test this theory at the firm level, using data from
an original survey of 174 foreign-owned firms in the post-conflict country of Georgia.
Across a range of self-reported behaviors and priorities, I find no evidence that
diaspora-owned firms are more likely to engage in a specific set of socially
responsible behaviors than are other foreign firms, and some evidence that they are
less likely to do so. I argue that diaspora investors are uniquely capable, but not
uniquely philanthropic, when doing business in their homelands.
170!
!
!
!
!
171!
Introduction
Diaspora direct investment, i.e. foreign direct investment (FDI) in the
homeland by migrants and their descendents, is a major source of capital in many
developing countries. While direct measures are not available, scholars estimate that
diaspora direct investment accounted for over 50% of FDI inflows to China during the
1990s (Huang 2003; Ye 2010) and 20-30% of FDI flows into India during the same
period (Ye 2010). The survey discussed below finds that 17% of foreign-owned firms
in Georgia, including the largest firm in the sample, have at least one diaspora owner.
The policy community in the United States has shown optimism regarding
diaspora direct investment as a tool of international economic development, recently
launching a number of programs targeting diaspora direct investors.1 The optimism
regarding the economic development potential of diaspora direct investment is based
not just on its volume, but also on its alleged unique pro-development characteristics.
Policymakers believe that diaspora-owned firms act with greater social responsibility
than other foreign firms, and that their presence produces unusually large prodevelopment spillovers (Foreign Service Institute 2010; Debass and Ardovino 2009.)
Unfortunately, enthusiasm has run ahead of the empirical evidence. To date there is no
firm-level research demonstrating that diaspora-owned firms actually behave
differently than other foreign firms, and therefore no direct evidence that warrants the
optimism on which policy is now based.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
The African Diaspora Marketplace project launched by USAID in 2009 is one example. See also
Foreign Service Institute, 2010.
!
!
!
!
172!
In political science, scholars generally analyze the effect of variations in host
country institutions on investment flows, while making the tacit assumption that the
foreign firms doing the investing are interchangeable. Diaspora-owned firms are
compelling subjects of study because they very clearly do not interact with and
respond to institutions in the host country (their homeland) in the same way as other
foreign firms. Economic sociologists argue that diasporans are motivated to invest in
their homeland by social and emotional factors, and that these motivational differences
manifest themselves in firm behavior (Gillespie et al. 1999). Specifically, this
literature contends that diaspora direct investors are less sensitive to the risks imposed
by political violence (Gillespie, Sayre, and Riddle 2001) and that diaspora investors
are more likely to hire local labor, pay higher wages, employ more environmentally
friendly business practices, and generally strive harder to contribute to the economic
development of the homeland (Nielsen and Riddle 2010). While these alleged
behavioral differences are sometimes used to justify a policy focus on promoting
diaspora investment (Foreign Service Institute 2010; Debass and Ardovino 2009),
there is recognition in the academic literature that rigorous empirical evaluation of this
potential is urgently needed (Nielsen and Riddle 2010, p. 442). I provide that here.
This work also contributes to a literature in political science and economics
that has identified bilateral migrant stocks as a cause of FDI, but has not clearly
identified a causal mechanism (Leblang 2010; Javorcik et al. 2011; Kugler and
Rapoport 2007). By conducting firm-level analysis directly comparing the behavior of
diaspora-owned firms and other foreign firms, this paper advances our understanding
!
!
!
!
173!
of the elements of diaspora difference that account for diasporans’ tendency to invest
in their homelands.
This paper draws on an original survey of 174 foreign firms operating in the
post-conflict country of Georgia. This survey provides the first data that allows for
direct comparison between diaspora-owned and non-diaspora owned foreign firms and
includes information regarding firms’ propensity to engage in a set of socially
responsible and pro-economic development behaviors. I find no evidence that diaspora
owned firms make greater effort to promote homeland economic development than
their non-diaspora-owned peers, and some evidence that they take less effort. This
failure to find evidence of diasporans engaging in more pro-development behavior
than their peers is all the more striking in light of other results from the same survey
reported in concurrent work, which show large and statistically significant differences
between diaspora and non-diaspora owned firms regarding their exploitation of social
networks for competitive advantage (CITATION REDACTED). Diaspora-owned
firms behave much differently than other foreign-owned firms but, contrary to current
expectations in the literature, these differences are born of pecuniary interest, not
benevolence.
This paper begins by surveying the existing literature on the development
potential of diaspora investment, including a summary of existing theories of diaspora
difference based on motivation. I then derive a series of testable hypotheses regarding
the social responsibility and pro-development behavior of diaspora-owned firms and
!
!
!
!
174!
test these hypotheses empirically. The paper concludes with a discussion of the
theoretical and policy implications of these results.
Diasporas and Development
Well into the 1990s, the dominant view in the literature on migration and
development was that emigration was a symptom of under-development and economic
insecurity2 and the literature emphasized the negative effects of brain drain on the
level of human capital in migrant-sending countries (e.g. Bhagwati and Hamada 1974;
Miyagiwa 1991). While several region- or country-specific studies during this period
pointed to specific cases in which labor emigration contributed to development, these
findings received little attention.3
The positive effects of emigration (and emigrants) on the homeland economy
gained more notice with later work that demonstrated the effect of migrant networks in
promoting trade between migrants’ homelands and their countries of residence (e.g.
Rauch and Trindade 2003). Beginning in the early 2000s, scholars also awoke to the
large scale of remittances, or money sent by migrants back to their families in the
homeland (e.g. Orozco 2002; Maimbo and Ratha 2005; Giuliano and Ruiz-Arranz
2009).4 Since then, the literature on migrants’ contributions to homeland development
has broadened rapidly to include diasporans’ role in promoting the transfer of
knowledge from countries of residence back to the homeland (e.g. Saxenian 2006;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2!See!de!Haan!(1999)!and!Faist!(2008)!for!reviews!of!this!literature.!Bakewell!(2009)!also!
documents!a!strong!antiAmigration!bias!among!Western!governments!and!development!NGOs.!
3!For!example,!Chirwa!(1997)!had!noted!the!positive!effects!of!labor!emigration!on!rural!sending!
communities!in!Malawi,!and!the!negative!effects!of!the!forced!repatriation!of!those!migrants.!
4!Some!excellent!scholarly!work!on!the!development!impacts!of!remittances!had!preceded!this,!
such!as!Oberai!and!Singh!(1980),!but!it!was!not!until!after!the!World!Bank!began!to!collect!data!on!
the!volume!of!these!flows!globally!that!academic!interest!in!the!topic!took!off.!
!
!
!
!
175!
Kapur 2010; Mullings 2011), the role of emigration opportunities in inducing
investment in education,5 diaspora philanthropy (Brinkerhoff 2008); diaspora direct
investment (Weidenbaum and Hughes 1996; Riddle, Hrivnak, and Nielsen 2010); and
diaspora venture capital (Vaaler 2011; Pandya and Leblang 2012).
This paper focuses on the channel of direct investment in the homeland as a
means through which diasporans contribute to economic development in the
homeland.
Diaspora Investment as a Tool for Economic Development
Developing countries generally, and states with high levels of political risk in
particular, often struggle to attract foreign direct investment. Diaspora investors are
heralded in the recent literature as unique among foreign investors in both their
willingness to tolerate political risk and their propensity to engage in socially
responsible and pro-development behavior when in investing in the homeland (e.g.
Brinkerhoff 2009). Diasporans are alleged to possess non-pecuniary motivations for
investment in the homeland (e.g. Gillespie, Sayre, and Riddle 2001; Riddle and
Nielsen 2010), and these motivations are expected to drive attention to social
objectives and homeland economic development by diaspora-owned firms.
Cross-national research has established a causal link between diaspora
populations and flows of foreign direct investment from diasporans’ countries of
residence to their homelands (Javorcik et al. 2011; Leblang 2010; Kugler and
Rapoport 2007; Docquier and Lodigiani 2010; Graham 2010). This literature theorizes
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5!The!work!of!Michael!Clemens!(2011)!and!others!has!also!highlighted!the!positive!effects!of!
emigration!on!migrants!themselves,!pointing!out!that!a!singular!focus!on!the!effects!of!emigration!
on!the!sending!economy!ignores!the!benefits!to!the!very!agents!creating!the!phenomenon.!!!
!
!
!
!
176!
that the propensity of diasporans to invest in their homelands (or otherwise channel
FDI from their country of residence to their homeland) is based on competitive
advantage. Migrants have familiarity with the culture and easier access to information
in the homeland (e.g. Javorcik et al. 2011; Leblang 2010), they may enjoy higher
levels of trust with co-ethnics, (Docquier and Lodigiani 2010; Rauch 2003), and firms
in the countries where migrants reside may glean information from them about their
homeland (Kugler and Rapoport 2007). In concurrent work that draws on the same
survey used in this paper, I demonstrate that diaspora-owned firms are more likely
than other foreign firms to exploit social networks in the homeland to gain competitive
advantage (Graham 2012).
A parallel, and not necessarily incompatible, literature on the motivations of
diaspora investors has also developed in economic sociology and business. This
literature perceives investment in the homeland by a member of the diaspora as not
just an economic act, but as an emotional, social, and political act (Bandelj 2008). It
argues that, in addition to the desire to earn a profit, diasporans may invest for social
reasons, such as raising their social standing in the diaspora community in their
country of residence, or for emotional reasons, such as patriotism (Aharoni 1960;
Schulte 2008; Nielsen and Riddle 2010). Empirically, surveys of US-based diasporas
have shown that diasporans’ self-reported interest in homeland investment is greatest
among those diasporans with the strongest emotional ties to the homeland, as well as
the strongest social ties to their diaspora community in the United States (Gillespie et
al. 1999; Gillespie, Sayre and Riddle 2001; Raveloharimisy et al. 2010).
!
!
!
!
177!
These findings of non-pecuniary investment motivations have motivated
further theorizing about the potential of diaspora investors as sources of social and
economic development in the homeland. If diaspora investment in the homeland is
motivated by the desire to increase social standing in the diaspora community and to
engage positively with a homeland to which the investor has strong emotional ties, it is
reasonable to expect that they will engage in behaviors consistent with these goals.
Specifically, the economic sociology literature theorizes that diaspora-owned firms
may rely more heavily on local labor and local inputs, pay above-market wages, and
strive to provide a high quality of life for their employees, protect the local
environment, and generally seek to promote development of the homeland (Nielsen
and Riddle 2010). However, these assertions remain almost entirely theoretical. The
evidence supporting these claims is anecdotal, including examples from places like
Afghanistan where diasporans have undertaken large, socially responsible investment
projects (Brinkerhoff 2004; Nyberg-Sørensen 2007; Riddle, Hrivnak, and Nielsen
2010). However, while the theoretical arguments in this literature are well developed,
the need for rigorous empirical testing is acute and acknowledged in the theoretical
literature itself (Nielsen and Riddle 2010, p. 442).
The rush to roll out diaspora investment promotion programs is not necessarily
driven by a lack of understanding among policymakers that these theories have not yet
been tested. The theories are difficult to test well, and if the theories are correct, the
potential upsides of effective diaspora direct investment promotion are very large,
!
!
!
!
178!
justifying the decision to proceed optimistically, at least until empirical evidence
becomes available.
Socially responsible investment is particularly desirable in developing
countries where the ability of the government to restrain rapacious firms and provide
social insurance is low. While exposure to the global economy is associated with
greater welfare spending in rich countries, the reverse is true in the developing world
(Wibbles 2006; Rudra 2002). Developing countries often open their economies to
foreign direct investment to spur development, but their very openness prevents large
social programs that might expand the benefits of growth to larger segments of the
population (Kaufman and Segura Ubiergo 2001; Reuveny and Li 2003). Against this
backdrop of limited government welfare capabilities, social responsibility among
foreign investors is all the more salient.
Defining Socially Responsible Behavior
Caroll (1979) divides firms’ social responsibilities into four categories:
economic, legal, ethical, and discretionary. In this paper, I focus empirically on
behaviors that foreign firms take to maximize their contributions to economic
development and social welfare in the host country. More specifically, I focus on the
treatment of employees, contributions to charity, and self-perceived contributions to
economic development in Georgia. These behaviors fall into Carroll’s final category
of discretionary behaviors, i.e. behaviors that contribute to the wellbeing of society,
but which are not legally required and the omission of which is not considered
inherently unethical. I do not suggest that these behaviors capture the full range of
!
!
!
!
179!
means through which a firm can be “socially responsible.”6 However, expectations
regarding these behaviors are critical in shaping investment promotion and
development policy, and theory predicts they should be observed at elevated levels in
firms with socially and emotionally motivated owners.
The argument that the motivations of diaspora owners might serve as causes of
socially responsible firm behavior, such as paying above-market wages and
contributing to charity, in their developing homelands is consistent with the general
literature on the reasons firms adopt corporate social responsibility (CSR). This
literature suggests that the personal motivations of top management are important
causes of CSR (Hemingway and Maclagan 2004; Juholin 2004), that small, privately
held firms are most likely to engage in CSR (Jones 1999), and that philanthropic,
rather than strategic, CSR is particularly prevalent in the developing world (Jamali and
Mirshak 2008).
Deriving Testable Hypotheses
Drawing on the theoretical work of Nielsen and Riddle (2010), I focus
empirically on diaspora-owned firms’ treatment of employees, contributions to
charity, and self-perceived contributions to economic development in the homeland.
These activities do not encompass the full range of ways in which non-pecuniary
motivations for investment may manifest themselves in socially responsible behavior,
but they represent some of the most likely manifestations. They also represent a core
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
6
For broad reviews of the corporate social responsibility literature, see Garriga and Melé (2004) and
Elms and Westermann-Behaylo (forthcoming).
!
!
!
!
180!
set of behavioral expectations on which optimism about the pro-development potential
of diaspora direct investment is based.
Fair or generous treatment of employees is both a core tenet of many
definitions of social responsibility (e.g. Hemphill 1997) and one that may be
particularly relevant for diaspora-owned firms in developing countries. Developing
countries, particularly those with weak formal institutions of economic governance,
are often characterized by a high proportion of family firms (Bertrand and Schoar
2006). In this context, one of the normal ways through which firm owners contribute
to the wellbeing of their family and community is through the provision of quality
employment. Providing good jobs to relatives, friends, and members of the community
whose respect they seek is therefore, among the most theoretically likely means
through which emotionally, and especially socially, motivated diasporans may act out
those motivations.7
Treatment of employees, however, captures but a narrow subset of the ways in
which diaspora-owned firms may engage in socially responsible behavior. The two
other empirical areas I examine are much broader in the types of behavior and
objectives they encompass. I also examine whether diaspora-owned firms are more
likely than other foreign firms to contribute to charity or to self-report that they make
above-average contributions to economic development in the homeland. The selfreport of contributions to economic development avoids limiting the means through
which diaspora-owned firms might seek to contribute to economic development in the
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
7!This!would!be!consistent!with!the!literature!on!family!firms!and!CSR!(e.g.!Uhlaner,!van!GoorA
balk,!and!Masurel!(2004).!
!
!
!
!
181!
homeland, capturing any perceived success toward that end. Broader still, I examine
whether diaspora-owned firms contribute to any charitable organizations. The goals of
such organizations are left entirely open, and may focus on economic development or
other non-economic objectives that diasporans may have. For almost any goals a
diaspora investor might have outside of a high return on investment, firm donations to
charitable organizations represent a logical means (though of course not the only
means) through which those ends might be achieved.
I begin by specifying two general hypotheses based on existing theory, and
then I proceed to derive a subset of specific and directly testable hypotheses.
Hypothesis 1: Diasporans’ social and emotional motivations for
investment lead diaspora-owned firms to engage in more socially
responsible labor practices.
Hypothesis 2: Diasporans’ social and emotional motivations for
investment lead diaspora-owned firms to make greater effort to
contribute to economic development in the homeland and to be more
likely to make charitable contributions.
Socially Responsible Labor Practices
One of the primary means by which foreign investment benefits the host
country is the employment of locals, which involves labor training and the transfer of
!
!
!
!
182!
knowledge (e.g., de Mello 1999). Hiring local labor also reduces unemployment and
injects wages directly into the local economy. If diaspora investors are motivated to
promote development in the homeland, this is perhaps the most obvious channel
through which they might do so. Particularly important in this mechanism is human
capital development, so one would expect diaspora-owned firms to both prioritize the
hiring of local labor and to provide their employees with opportunities for professional
development. One would also expect that socially responsible firms would pay abovemarket wages and provide the means for their employees to provide a high quality of
life for their families, and that they would pursue these ends even if it increased their
total labor costs.
Hypothesis 1A: Diaspora-owned firms are more likely than non-diaspora-owned firms
to report that their firm prioritizes the hiring of local labor over foreign labor.
Hypothesis 1B: Diaspora-owned firms are more likely than non-diaspora-owned
foreign firms to report that their firm offers higher salaries than other firms in their
sector.
Hypothesis 1C: Diaspora-owned firms are more likely than non-diaspora-owned
foreign firms to report that employees at their firm are able to provide a higher quality
of life for their families than are employees at other firms in their sector.
!
!
!
!
183!
Hypothesis 1D: Diaspora-owned firms are more likely than non-diaspora-owned
foreign firms to report that, compared to other firms in their sector, employees at their
firm have many opportunities for professional development.
Hypothesis 1E: Diaspora-owned firms are less likely than other foreign firms to report
that their firm keeps their total labor costs as low as possible.
Contributions to Development and Charitable Contributions
Firms have a variety of mechanisms for contributing to development in their
homeland, many of which fall outside the area of employment and may not be
captured by questions about specific firm behaviors. Therefore, I also assess some
more general questions about how firms balance the sometimes competing priorities of
profit maximization and development contribution.
While a firm’s contribution to development is distinct from its effort to make
such a contribution, both are interesting. From a policy perspective, it is the actual
impact that is of greatest interest. If diaspora-owned firms do not contribute more to
the development of the homeland than other foreign firms, it undermines key
arguments currently made for the targeted recruitment of diaspora investors into
developing countries.8
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
8!For!examples!of!these!arguments,!see!Debass!and!Ardovino!(2009)!and!Terrazas!(2010).!
!
!
!
!
184!
Hypothesis 2A: Diaspora-owned firms are more likely than non-diaspora-owned
foreign firms to report that their firm contributes to economic development in their
homeland more than other firms in their sector.
I also assess whether firms report having contributed to charity in the past year.
I acknowledge that firms may be able to reap marketing benefits from some types of
charitable contribution, and large firms might be able to reap sufficient benefits from
social improvements over the long run. For example, a charitable contribution to
education may produce a better educated workforce that, over the extremely long run,
may benefit the firm (Porter and Kramer 2002). However, charitable contributions
mark a firm’s most direct contributions to non-pecuniary ends, at the most obvious
cost to short-term profitability. As noted previously, charitable contributions also
represent a likely means toward a wide variety of objectives that diaspora investors
might have, including those far removed from economic development.
Hypothesis 2B: Diaspora-owned firms are more likely than non-diaspora-owned
foreign firms to report having made contributions to charity within the last year.
The testable implications of Hypotheses 1 and 2 encompass a range of ways in
which diaspora-owned firms may manifest a commitment to social responsibility and
development of the homeland. Some of these may serve as substitutes for one another,
but if the underlying theory is correct, and diasporans’ social and emotional
!
!
!
!
185!
motivations lead diaspora-owned firms to engage in more socially responsible and
pro-development behavior, I should find evidence in support of at least some of these
hypotheses. If I cannot find evidence of the expected differences between diasporaowned and non-diaspora-owned firms across this wide array of behaviors and
priorities, it is reasonable to conclude that diaspora-owned firms do not actually
behave in a more socially responsible manner than do other foreign firms.
In the following sections I discuss the structure of the survey used to test the
hypotheses outlined above.
Why Georgia?
In the language of small-N case selection, Georgia is a “typical case” with
regard to its economic characteristics, and an “extreme case” regarding the likelihood
of observing socially responsible behavior by diaspora investors.9 For reasons
outlined below, Georgia represents precisely the type of country in which I would
expect the social and emotional motivations of diaspora investors to be salient, and in
which I would expect to see diasporans using investment as a means to foster
development in the homeland. If the predictions made by theories of diaspora
difference based on motivations are not borne out in the Georgia case, this would
contradict the general theory from which these predictions are drawn.
Georgia!is!a!small!country!surrounded!by!volatile!and!sometimes!hostile!
neighbors.!It!has!only!modest!endowments!of!natural!resources!but!is!located!on!
a!major!transport!corridor!for!both!energy!(oil!and!gas)!and!goods!between!the!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
9
See Seawright and Gerring (2008) for a discussion of the use of “extreme cases” and “typical cases” in
hypothesis testing.
!
!
!
!
186!
Black!Sea!and!Central!Asia.!!The!amount!of!arable!land!is!not!vast,!but!the!climate!
allows!cultivation!of!some!high!value!crops,!including!citrus!and!nuts.!!While!
Georgia!cannot!become!wealthy!on!the!strength!of!its!natural!endowments!alone,!
with!political!stability!and!competent!institutions!it!could!certainly!become!
wealthy.!!!!!!
Following!Mikheil!Saakashvili’s!ascent!to!power!in!the!“Rose!Revolution”!
of!2003,!Georgia!embarked!on!a!series!of!freeAmarket!reforms!aimed!at!
increasing!foreign!investment!and!integrating!Georgia!with!the!West!both!
economically!and!politically.!!These!reforms!have!born!substantial!fruit:!Georgia!
rocketed!up!through!the!World!Bank’s!Ease!of!Doing!Business!rankings!and!FDI!
surged!from!$335!million!in!2003!to!$1.7!billion!in!2007.!!Georgia,!however,!has!
remained!a!highArisk!destination!for!foreign!investment.!!Large!protests!in!
November!2007!underscored!the!fragility!and!unpredictability!of!the!domestic!
political!situation;!corruption,!though!improved,!remains!substantial;!
Transparency!International!has!accused!the!government!of!practicing!“tax!
terrorism”!(Transparency!International!2010);!and!a!brief!war!between!Russia!
and!Georgia!in!August!2008!highlighted!the!international!instability!generated!by!
Georgia’s!secessionist!regions,!which!remain!outside!the!government’s!control.!!!
In addition to attracting substantial FDI, Georgia has a large diaspora
population relative to its size, making it an excellent venue for studying diaspora
!
!
!
!
187!
direct investment.10 The other relevant idiosyncrasy is the aforementioned 2008 war
between Georgia and Russia over the territory of South Ossetia. Two regions of
Georgia, Abkhazia and South Ossetia, fought wars of secession against the Georgian
government in the early 1990s and secured de facto territorial control. Fighting broke
out between South Ossetia and the Georgian government in 2008, with Russian troops
doing much of the fighting on behalf of the Ossetians. The war was brief, limited of
scope, and minor in terms of its economic impacts, but it may have affected the
relationship between segments of the Georgian diaspora and the Georgian
government.
Nielsen and Riddle (2010) theorize that non-pecuniary motivations may be
more salient to diasporans whose homeland is a post-conflict state. However, in cases
of civil conflict, the role of the diaspora is not always to promote peace and economic
development.11 The Abkhaz and Ossetian diasporas (primarily residing in Russia) are
very likely to oppose the Georgian government. Most of the Georgian diaspora, on the
other hand, is comprised of ethnic Georgians, and enjoys positive relationships with
the government.12 Reflecting these good relations, the Georgian government
established the Office of the State Minister of Diaspora in 2008 and has made
substantial efforts to attract diaspora investment. For logistical reasons, this survey
excluded firms in the contested regions of Abkhazia and South Ossetia, and all of the
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
10
Because of a paucity of cross-national data on diaspora investment, it is difficult to assess the volume
of diaspora investment in a given country ex ante.
11
Collier and Hoeffler (2004) find that diaspora populations in rich countries increase the risk of civil
conflict, presumably because they represent a potential source of rebel financing.
12
This relationship is most likely to be strained in the case of firms based in Russia. Neither dropping
Russian firms from the analysis nor including a dummy variable for Russian origin changes the
substantive results.
!
!
!
!
188!
diaspora investors covered in the survey self-reported their ethnic identity as
Georgian. Given these facts, existing theory suggests the 2008 war should have
increased the relevance of non-pecuniary motivations for the diaspora investors
captured in the survey.
This makes Georgia an apparently easy case in which to demonstrate
diasporans’ social responsibility: if diaspora-owned firms act with greater social
responsibility than other foreign firms, Georgia is exactly the type of case in which
such a difference should manifest itself. A failure to find such evidence in Georgia
would contradict the general theory.
!
!
!
!
189!
Table 5.1:
Georgia in Comparison to Other Fragile and Developing Countries
Georgia
Fragile states
Developing States
median
Median
(25th -75th
(25th - 75th
percentile)
percentile)
GDP
5,500
12,300
6,500
(millions of USD)
(5,500 – 55,000)
(1,700 – 24,000)
GDP per capita
1,300
1,200
1,100
(USD)
(280 – 2,200)
(380 – 2,000)
Population
4.3
20
9.7
(millions)
(7.3 – 41)
(3.1 – 27)
FDI Inflows
1,600
790
748
(Millions of USD)
(120 – 2,600)
(120 – 2,200)
FDI Inflows
12
3.6
4.3
(% of GDP)
(1.9 – 6.1)
(1.9 – 7.9)
Remittances
73
73
73
(Millions of USD)
(15 – 270)
(8.1 – 290)
Remittances
5.7
2.0
4.3
(% of GDP)
(0.7 – 5.7)
(0.4 – 5.6)
Emigrant Stock
48,000
110,000
88,000
(1990)
(24,000 – 300,000) (23,000 – 330,000)
Emigrant Stock
190,000
216,863
168,220
(2005)
(82,000 – 640,000) (36,000 – 490,000)
Emigrant Stock
4.3
1.9
2.0
(% of population) (2005)
(0.7 – 3.0)
(0.9 – 4.4)
Cost of Dispute Resolution
35
30
33
(% of claim)
(23 – 44)
(23 – 45)
War Risk (1-7)
5
5
3
(Higher = Riskier)
(3 – 6)
(2 – 4)
Government Risk (1-7)
4
4
4
(Higher = Riskier)
(4 – 5)
(3 – 5)
Transfer risk (1-7)
6
6
5
Higher = Riskier
(4 – 7)
(4 – 6)
Fragile countries n = 31; Developing countries n = 107
The range from 25th percentile to 75th percentile is given in parentheses.
All items are measured as of 2008, except for the measures of migrant stocks, whose
years are given.
I define developing countries as those with a 2008 GDP per capita of less than
$3,855 – those countries defined by the World Bank as low income and lower middle
income.
Fragile states are those that meet any of the following criteria: 1. A violent intrastate conflict in the last 3 years; 2). A change of executive leadership via coup or
military overthrow in the last three years; 3) failure to control all their territory 4)
occupation by a foreign power or not controlled by a unified entity (i.e. coded as
“Interregnum” or “Interruption” in the Polity IV data).
Data on GDP, GDP per capita, Population FDI inflows, migrant stocks, and
remittances are taken from the World Development Indicators. Data on war
risk, government, and transfer risk are taken from Office National Du Ducroire
!
!
!
!
190!
As noted above, Georgia is a “typical case” with regard to its economic
characteristics. For 2008-2010, Georgia’s risk profile matched the risk profile of the
median fragile state in all three categories of political risk: government risk, which is
the risk of expropriation and adverse government action; transfer risk, which is the
risk of capital controls or other regulations that might prevent the repatriation of
capital by an investing firm; and war risk. This profile also places it near the
developing states median in these categories, except for war risk, which is particularly
high in Georgia.
Across a range of other variables, such as wealth, total FDI inflows, and
judicial efficiency, Georgia is within one standard deviation of the median value for
both fragile states and developing states. The exceptions to this derive from the fact
that Georgia is quite small relative to the mean in either category of state, but has a
large diaspora and attracts a large amount of FDI relative to its size.
The Survey
The Capital and Conflict: Georgia survey is the first firm-level survey that
compares diaspora-owned firms to other foreign firms. Surveys were conducted, in
person, with the owner or manager of 174 foreign owned firms between February and
June 2010. Each respondent was given a choice of taking the survey in English or
Georgian.13
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
13
The survey was taken through three rounds of reverse translation and refinement to ensure
equivalence between the English and Georgian versions.
!
!
!
!
191!
The sampling frame was derived from a list of foreign-owned firms provided
by the Georgian Ministry of Finance. The list included all foreign firms that met the
following criteria: 1) a for-profit enterprise; 2) at least 10% foreign ownership; 3)
registered as active and paying taxes as of June 1, 2009; 4) obtained its first
registration in Georgia after the year 2000. This sample was supplemented with a
randomly drawn sample of 300 of the 450 firms that responded to the Ministry of
Finance’s Balance of Payments survey in 2009. These firms also met criteria 1-3, but
some were initially registered prior to 2000.14
Each of the testable hypotheses listed above are tested using a specific question
on the survey. All questions, except that used to test Hypothesis 2B, elicit answers on
a seven point Likert scale from “strongly agree” (7) to “strongly disagree” (1).
Table 5.2: Survey Questions
Hypothesis
H1A
H1B
H1C
H1D
H1E
H2A
H2B
Question
Please indicate how well each statement describes your company’s Georgian
operations:
This firm gives priority to hiring local staff over foreign staff.
This firm offers higher salaries than other firms in this sector.
Compared to employees at other firms in this sector, employees at this firm are
able to provide a higher quality of life for their families.
Compared to employees at other firms in this sector, employees at this firm have
many opportunities for professional development.
This firm keeps its total labor costs as low as possible.
This firm contributes to economic development in Georgia more than other firms
in this sector.
As a yes or no question:
Does your firm make any donations to charitable causes?
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
14
Some firms on the Ministry of Finance list also turned out to be registered before the year 2000, but
had been re-registered after that date.
!
!
!
!
192!
Non-Response
Attempts were made to contact a total of 1,024 firms between February and
June 2010, representing over 80% of the foreign-owned firms that officially entered
Georgia during the target period. Only 484 could be contacted and, of those, only 362
met the criteria listed above (foreign ownership, for- profit, and currently operating).15
Many of the firms that could not be contacted were closed – some had closed recently
and others had closed before 2009 but had not been purged from the tax rolls. Other
firms could not be contacted because they had changed addresses and phone numbers
since the lists were updated. Georgia has no up-to-date telephone or address directory,
and so only a minority of the firms that have relocated within the last several years
could be contacted. Few of the foreign firms operating in Georgia maintain websites.
Of the 362 firms that were successfully contacted and that met the basic
criteria for inclusion in the sample, surveys were successfully conducted with 167; 195
firms refused to participate. In most firms, enumerators were successful in scheduling
interviews with either the firm owner or a firm manager. In some cases, however, this
was not possible and a shorter version of the survey (10 of 53 questions) was asked of
front-desk or other available personnel. These “front desk” surveys asked questions
about firm demographics, such as sector, headquarters country, firm size, and diaspora
ownership. The front desk surveys do not include sufficient information to include
these firms in most hypothesis tests, but do provide valuable additional information
about firm demographics in the sampling frame.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
15
We only attempted to contact firms that the Ministry of Finance or State Department of Statistics
indicated met the three criteria for inclusion, but nonetheless many firms on the lists provided were
either non-profit, fully Georgian-owned, or were closed.
!
!
!
!
193!
The sampling lists discussed above produced 167 respondents: 26 diasporaowned firms and 129 non-diaspora-owned firms16; twelve respondents did not answer
the question on diaspora ownership.17 Because of the low number of diaspora-owned
firms, a non-random supplement of seven additional diaspora-owned firms was added
to the sample, bringing the total number to 174.18
Table 5.3: Number of Respondents by Diaspora Ownership
Survey Type
Diasporaowned firms
(random only)
Nondiasporaowned firms
Diaspora
ownership
unknown
Total
23
Diaspora-owned
firms
(supplement
included)
30
Owners/Manager’s
Survey
Front Desk Survey
Total
121
10
161
3
26
3
33
8
129
2
12
13
174
The only demographic information available from firms that refused to
participate in the survey is the firm’s home country, which in most cases could be
drawn from the firm lists provided by the Georgian government.19 Therefore, I
examine the number of respondents and non-respondents from each region. The
following table shows that refusals are distributed evenly across regions. The
percentage of diaspora-owned firms is smaller in the West than in other regions, but
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
16!I!treat!any!firm!with!at!least!one!diaspora!owner!as!“diasporaAowned.”!However,!all!the!
empirical!results!that!follow!are!robust!to!treating!only!majorityAdiasporaAowned!firms!as!
diasporaAowned.!
17
All firms in the sample are privately owned, i.e. none are government-owned or publicly traded.
18
Two of these supplemental firms were identified by the State Office for the Diaspora; five more were
located via snowball sampling. To create the snowball supplement, enumerators asked respondents at
the end of the interviews if they knew of any diaspora-owned firms that we might be able to contact.
19
Not all firms have home countries listed on the original sampling lists. For those firms that responded
to the survey but did not have a home country listed, I treat the location of the firm headquarters as the
home country. Some respondents did not report a headquarters location outside of Georgia – the home
country of those firms remains as missing data.
!
!
!
!
194!
because this is the largest region in terms of number of firms, the number of diaspora
firms from the West is still substantial.
Table 5.4: Distribution of Diaspora Ownership by Home Region
Region
Former Soviet
Union
Western
Europe,
US & Australia
Middle East
Other
Region
Unknown
Total
Number of
Qualifying Firms
Contacted
69
Number of
Responses
Percentage
Non-Response
31
55%
Percentage of
respondents who are
diaspora-owned
23%
162
73
55%
13%
68
14
56
31
7
24
54%
50%
57%
27%
14%
33%
369
174
56%
20%
Within the set of firms that were successfully interviewed, I compare diaspora
and non-diaspora-owned firms across a range of covariates. By establishing that firm
demographics between the two groups are similar, I am then better able to attribute
any observed differences in firm behavior to diaspora ownership itself, rather than
alternative factors like firm size or sector. Firm size is captured by the number of
employees: I compare diaspora and non-diaspora firms both in terms of average
number of employees and the likelihood of being a small firm, i.e. a firm with 50
employees or less.20 I use dummy variables for minority, majority, and 100% foreign
ownership, and a dummy variable for Greenfield investment. I also include dummy
variables for primary, secondary, tertiary, and quaternary sectors.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
20!Seventy!percent!of!firms!in!the!sample!have!50!employees!or!less.!The!median!number!of!
employees!is!12.!
!
!
!
!
195!
With regard to the demographic characteristics of individual respondents, I
examine whether the respondent was an owner of the firm (as opposed to a manager)
and whether the respondent self-identified as ethnically Georgian. I also examine the
language in which the survey was conducted.
Figure 1: Demographic Comparison
Diaspora Mean - Non-Diaspora Mean
Location = Tbilisi
HQ = Russia
Respondent = Owner
Tertiary Sector
Secondary Sector
Primary Sector
100% Foreign Ownership
Greenfield Investment
Firm Age (Years)
Small Firm
Firm Size (# of Employees)
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
NOTE: Units = standard deviations, horizontal lines give 95% confidence in
Figure 5.1: Demographic Comparison
*Note: Units = standard deviations, horizontal lines give 95% confidence intervals.
!
!
!
!
196!
The only demographic characteristic for which I observe a large difference
between diaspora and non-diaspora-owned firms is in the probability that the
respondent is the firm owner. At each firm, enumerators asked for a meeting with the
firm owner or manager. At diaspora-owned firms, the firm owner was more likely to
be available for interview. This makes sense if diaspora owners are more likely than
other foreign owners to live in Georgia at least part time while running their business.
All of the respondents at diaspora-owned firms self-identified as ethnically Georgian
(i.e. both owners and managers), as did 89% of respondents at non-diaspora-owned
firms (i.e. almost all managers but no owners).
Firm owners may have somewhat different perspectives on the behavior of
their firm than do managers; this necessitates a multiple-regression framework
controlling for the identity (owner vs. manager) of the respondent. Other firm
demographic characteristics are included as supplemental controls, but do not
substantively affect results.
The range in firm size in the sample is substantial, with firms varying from a
single employee to as many as 1400. However, as Figure 1 demonstrates, the
difference in firm size between diaspora-owned and non-diaspora-owned firms is not
large. Three-quarters of firms in both groups qualify as “small” enterprises: 75% of
diaspora-owned firms and 77% of non-diaspora-owned firms have 50 employees or
!
!
!
!
197!
less. Similar consistency is observed in the distribution of firms between sectors,21 the
distribution of firms between regions (not pictured), the share of foreign ownership,
and firm age.22
Summary Statistics and Multiple Imputation
In dealing with missing values that arise from non-response, I employ the
multiple imputation by chained equations technique developed by Patrick Royston
(2004, 2009).23 Multiple imputation reduces the bias and eliminates the sample-size
reduction associated with list-wise deletion, allowing full use of the information
collected in the survey (King et al. 2001).24 In the raw data, the independent variable
of interest, diaspora-ownership, is 6% missing, firm demographic characteristics used
as controls are between 2% and 11% missing, and the dependent variables range from
9% to 30% missing.25
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
21
I also checked the balance in a composite of real estate and construction – the balance here is quite
even. I checked because these sectors were particularly hard-hit during the downturn.
22
It is worth noting that very few firms in the sample entered after the 2008 conflict with Russia: of the
18 that entered during this period, 5 were diaspora-owned and 12 non-diaspora-owned (1 unknown).
Most of the firms in the sample entered Georgia as Greenfield investments, rather than mergers,
partnerships, or acquisitions of Georgian firms: 76% of diaspora-owned and 78% of non-diasporaowned firms.
23
This is implemented using the ice and mim commands in Stata 10. I create 10 imputed datasets for
analysis.
24
The coefficients of interest are similar in analyses using list-wise deletion.
25
These percentages refer only to the 161 firms that completed an owners/managers survey. Those
firms where only a front-desk survey was completed are omitted from analysis (and imputation)
because none of the social responsibility questions are included on the front desk survey.
!
!
!
!
198!
Table 5.5: Summary Statistics
Variable
Diaspora Ownership
Respondent = Owner
Primary Sector
Secondary Sector
Tertiary Sector
Quaternary Sector
100% Foreign
Ownership
HQ in West
Number of Employees
Greenfield Investment
Prioritize Local Labor
Pay Higher Salaries
Higher Quality of Life
Professional
Development
Contribute to Charity
Minimize Total Labor
Costs
Contribute to
Development
Focus on Profits
Focus on Efficiency
Observations
Mean
Minimu
m
0
0
0
0
0
0
0
Maximum
0.199
0.242
0.0705
0.301
0.487
0.141
0.599
Standard
Deviation
0.4
0.43
0.257
0.46
0.501
0.349
0.492
151
157
156
156
156
156
157
132
144
154
139
126
124
129
0.53
89.4
0.76
6.3
4.77
0
5.44
0.501
218
0.429
1.08
0
1.2
1.38
0
1
0
3
1
1
1
1
1400
1
7
7
7
7
145
138
0.586
4.01
0.494
1.66
0
1
1
7
112
4.96
1.3
2
7
126
135
5.42
5.77
0
0
1
3
7
7
1
1
1
1
1
1
1
Differential Item Response
Many of the questions in this survey rely on respondents’ subjective
assessments. One of the central concerns in questions of this nature is differential item
functioning, the problem that individuals may understand questions differently or that
the scales of their answers may not match. Many of the responses reported here are
subjective assessments comparing the behavior of a respondent’s firm to that of “other
firms in your sector.” This peer comparison creates a self-anchoring scale and reduces
!
!
!
!
199!
the problem of differential item functioning (King et al. 2004).26 Self-anchoring
scales are effective when respondents are referencing the same or similar anchor
points – in this case, when they are comparing their firm to the same or a similar group
of peer firms. Georgia is a small country, making it reasonable to assume that, within a
given sector, firms are referencing the same or similar firms in anchoring the scale of
their answers. A less restrictive, but still sufficient, assumption is that the referenced
peer firms do not vary systematically between diaspora-owned and non-diasporaowned firms.
The within-question comparison to peer firms does not address the possibility
that some firms may use the Likert scale differently than others. This potential
problem is explored by the inclusion of questions that use the Likert in opposing
directions. For example, respondents are asked the degree to which they agree or
disagree with the following statements: “This firm offers higher salaries than other
firms in this sector,” and six questions later, “This firm keeps its total labor costs as
low as possible.” There is evidence that the direction of the Likert scale does affect
answers. Responses to the statement that the firm in question pays higher salaries than
its peers and that the firm minimizes total labor costs, which use the Likert scale in
opposite directions, are correlated at -0.2. Responses to the salaries statement and the
statement “Compared to employees at other firms in this sector, employees at this firm
are able to provide a higher quality of life for their families,” which use the Likert
scale in the same direction, are correlated at 0.75. These three questions are not exact
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
!
!
!
200!
substitutes, but the correlations in responses to all three questions were expected to be
high, and the weak negative correlation is surprising. However, across all three
questions, diaspora owned firms are less likely than non-diaspora-owned firms to
report pro-labor behavior. Observing a theoretically consistent direction of difference
between diaspora-owned and non-diaspora-owned firms across similar statements in
opposing directions indicates that results are not driven by systematic differences in
use of the Likert scale between those two groups of firms.
Results for H1A-H1E: Local Labor, Wages, and Professional Development
Hypothesis 1 states that diasporans’ social and emotional motivations for
investment lead diaspora-owned firms to engage in more socially responsible and prodevelopment behavior than non-diaspora-owned foreign firms. In the questions used to
test this hypothesis, firm managers are asked, using a seven point Likert scale, to what
degree a set of statements reflects their firm’s Georgian operations.27 Table 5.6
presents results from tests of Hypotheses 1A-1E, each of which specifies a laborrelated means through which diaspora-owned firms might show themselves to be more
socially responsible than other foreign firms.
I also test the impact of diaspora ownership on an index of the question used to
test Hypotheses 1B and 1C. In factor analysis, it was these two questions that hang
together as measure of firms’ willingness to bear costs to provide additional employee
benefits.28
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
27
The questions referred to in this section were developed collaboratively with Professor Maia
Mestvirishvili of Tbilisi State University.
28
The Cronbach’s alpha for these two variables is 0.8.
!
!
!
!
201!
Table 5.6: Diaspora Ownership and Social Responsibility: Labor
Table 5.6: Diaspora Ownership and Social Responsibility: Labor
(1)
(2)
(3)
(4)
(5)
Local Labor Salaries Qaul. of Life Prof. Dev. Labor Costs
Diaspora-Owned Firm
0.405
-0.559
-0.342
0.043
1.014**
(0.533)
(0.504)
(0.487)
(0.474)
(0.438)
(6)
Labor Index
-0.882
(0.685)
Respondent = Owner
0.609
(0.491)
0.632
(0.417)
0.600
(0.430)
0.559
(0.427)
-0.269
(0.444)
0.977*
(0.559)
Primary Sector
0.178
(0.740)
0.762
(0.832)
0.579
(0.782)
0.288
(0.677)
-0.479
(0.626)
1.051
(1.017)
Secondary Sector
-0.360
(0.566)
0.051
(0.614)
0.071
(0.568)
0.270
(0.603)
0.037
(0.520)
0.259
(0.833)
Tertiary Sector
0.056
(0.588)
-0.289
(0.590)
-0.188
(0.540)
0.348
(0.580)
0.285
(0.478)
-0.171
(0.777)
100% Foreign Ownership
0.248
(0.375)
0.423
(0.366)
0.870**
(0.360)
-0.096
(0.352)
-0.407
(0.354)
0.940*
(0.461)
Western HQ
-0.320
(0.446)
0.106
(0.414)
-0.060
(0.437)
0.045
(0.370)
0.075
(0.380)
0.037
(0.558)
Employees (logged)
0.186
(0.135)
0.179
(0.131)
0.157
(0.132)
0.168
(0.141)
-0.135
(0.109)
0.257
(0.166)
Greenfield
0.394
(0.447)
-0.104
(0.383)
0.019
(0.421)
-0.559
(0.369)
-0.155
(0.428)
-0.082
(0.545)
Constant
Cut 1: Constant
Cut 2: Constant
Cut 3: Constant
Cut 4: Constant
8.061***
(1.038)
-2.152**
(1.046)
-1.037
(0.903)
-0.403
(0.884)
0.537
(0.904)
Cut 5: Constant
Cut 6: Constant
Observations
161
-2.271**
(0.861)
-1.924**
(0.784)
-1.162
(0.743)
0.385
(0.723)
1.507*
(0.752)
3.103***
(0.832)
161
-2.587**
(1.244)
-1.842*
(0.922)
-1.264
(0.834)
0.690
(0.752)
2.041**
(0.766)
3.247***
(0.828)
161
-2.650**
(0.986)
-2.309**
(0.918)
-1.954**
(0.878)
-0.587
(0.845)
0.222
(0.841)
1.655*
(0.845)
161
-2.645***
(0.774)
-1.728**
(0.765)
-1.262
(0.754)
0.187
(0.730)
0.917
(0.741)
2.039**
(0.821)
161
161
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Models 1-5 use ordered logit regression, model 6 uses OLS. All analysis is conducted on imputed data.
In the only statistically significant result (Model 5), diaspora-owned firms are
more likely than other foreign firms to report keeping their total labor costs as low as
!
!
!
!
202!
possible.29 This is consistent with results in Models 2 and 3, which show that diasporaowned firms are slightly less likely than other foreign firms to report devoting
resources to socially responsible employment practices, namely paying above-market
wages or providing a higher quality of life to their employees than peer firms. Model 6
shows a negative, but not significant effect of diaspora ownership on an additive index
of responses to the questions used as dependent variables in Models 2 and 3.
The coefficient in Model 4 is approximately zero, indicating no difference
between diaspora-owned and non-diaspora owned firms in the opportunities they
provide employees for professional development.
The only result in Table 5.6 suggesting a greater social responsibility from
diaspora-owned firms is Model 1, which indicates that diaspora-owned firms may be
somewhat more likely than other foreign firms to prioritize the hiring of local labor.
However, in light of the results in Models 2, 3, 5, and 6, this is more reasonably
interpreted as a strategy for limiting labor costs than as a means to maximize prodevelopment impact.
Taken together, these results run directly counter to Hypothesis 1. There is no
evidence that the labor practices of diaspora-owned firms are more socially
responsible than those of other foreign firms, and some evidence that their practices
are less socially responsible. While a larger sample size might provide us with
evidence that diaspora-owned firms are marginally more committed to hiring local
labor, there is no evidence that they treat the labor they do hire any better than other
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
29
Substantively, a respondent from a diaspora-owned firm is 2.8 times as likely as an otherwise
identical non-diaspora-owned firm to respond with a higher category response across any given cutpoint on a seven point scale from “strongly disagree” to “strongly agree.”
!
!
!
!
203!
foreign firms. In fact, my results show diaspora-owned firms are more committed to
minimizing total labor costs than are other foreign firms, and there is at least some
evidence that this manifests itself in lower wages and a correspondingly lower ability
of employees to provide a high quality of life for their families.
H2A and H2B: Development Impact and Efficiency vs. Social Responsibility
Hypotheses 2A and 2B posit more general ways in which diaspora-owned
firms might demonstrate their commitment to social responsibility and the
development of the homeland: by reporting that they contribute to the development of
the homeland more than their peer firms and by being more likely to report making
contributions to charity.
!
!
!
!
204!
Table 5.7: Contributions to Development and Donations to Charity
Table 5.7: Contributions to Development and Donations to Charity
(1)
(2)
Development Charity
Diaspora-Owned Firm
0.155
-0.314
(0.556)
(0.536)
Respondent = Owner
0.192
(0.512)
0.043
(0.440)
Primary Sector
-0.288
(0.648)
1.181
(0.898)
Secondary Sector
-0.514
(0.608)
0.125
(0.618)
Tertiary Sector
-0.129
(0.546)
0.045
(0.621)
100% Foreign Ownership
0.729*
(0.377)
-0.335
(0.406)
Employees (logged)
0.331**
(0.134)
0.527***
(0.169)
Greenfield
0.440
(0.482)
-0.389
(0.505)
Western HQ
0.061
(0.411)
-0.528
(0.426)
Constant
-0.427
(0.901)
Cut1: Constant
-0.952
(0.931)
-0.256
(0.845)
1.423*
(0.810)
2.222**
(0.805)
3.108***
(0.862)
161
Cut 2: Constant
Cut 3: Constant
Cut 4: Constant
Cut 5: Constant
Observations
161
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < .01
Model 1 uses ordered logit regression, model 2 uses OLS.
All analysis is conducted on imputed data.
Neither of the differences between diaspora and non-diaspora owned firms are
statistically significant, with the coefficient on diaspora ownership in Model 1 almost
!
!
!
!
205!
exactly equal to zero, indicating that diaspora-owned firms are no more likely than
their peers to report high contributions to the development of the homeland.
The positive sign on the coefficient in Model 2 indicates that diaspora-owned
firms might be slightly more likely to contribute to charity, but the magnitude of the
coefficient is less than half that of the standard error. The substantive size of this
estimated effect is moderate; the expected probability a firm with mean characteristics
will report contributing to charity increases from 42% if the firm is non-diasporaowned to 52% if it is diaspora-owned.30 Larger firms are more likely to contribute to
charity and more likely to believe that the contribute more to economic development
than their peers, which is unsurprising: the scale of their economic activity is larger as
is the pool of resources they can draw on to make charitable contributions.31
As with the results in Table 5.6, the results presented here fail to give any
evidence that diaspora-owned firms are more socially responsible than other foreign
firms.
Graphical Presentation of Results
Figure 5.2 provides a graphical supplement to Tables 5.6 and 5.7. I omit the
control variables and present a simple comparison of means between diaspora-owned
and non-diaspora owned firms. Because of the similarity in firm demographics
between diaspora-owned and non-diaspora-owned firms, these simple mean
comparisons depict the same substantive results as the more complex analysis above.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
30
The odds ratio for the diaspora dummy in Model 2 is 1.23.
31!Analysis!was!also!run!including!an!interaction!term!between!diaspora!identify!and!firm!size,!
but!no!statistically!significant!results!were!found.!
!
!
!
!
206!
Before the mean of each group is taken, responses to each question are rescaled to a
mean of zero and a standard deviation of one.
Coefficient on Diaspora Ownership
*Minimize Total Labor Costs
Professional Development
Higher Quality of Life
Pay Higher Salaries
Prioritize Local Labor
Contribute to Development
Contribute to Charity
-4
-2
0
2
4
Standard Errors Normalized to 1
Figure 5.2: Social Responsibility
*Note: Horizontal lines give 95% confidence intervals.
It is in the final category, in which a positive difference indicates less socially
responsible behavior by diaspora owned firms, that we observe the only statistically
!
!
!
!
207!
significant difference between the two groups. Diaspora owned firms are more likely
than other foreign firms to report that they minimize total labor costs. Across all other
categories, a positive difference would indicate more socially responsible behavior by
diaspora owned firms, and among these, it is only in the area of prioritizing local labor
that we see any evidence that diasporans might be more socially responsible than their
peers. As discussed above, other results suggest that this is more reasonably
interpreted as a strategy for limiting labor costs than as a means to maximize prodevelopment impact.
Interpretation and Implications of Results
These results indicate that, even if non-pecuniary motivations are an important
cause of diaspora investment, these motivations do not translate into socially
responsible behavior by diaspora-owned firms. The hypotheses tested here did not
capture all of the ways in which diasporans’ non-pecuniary motivations might
manifest themselves in socially responsible or pro-development actions, but if
diaspora-owned firms are more socially responsible or engage in more prodevelopment behaviors than their non-diaspora-owned peers, this should have
manifested itself with regard to at least some of the hypotheses tested here. It did not.
This allows us to reject Hypotheses 1 and 2, and provides evidence that diasporaowned firms are not more socially responsible than non-diaspora-owned firms.
Because the results presented here are from only a single country, it is possible that
diaspora-owned firms in other contexts are more socially responsible than their peers.
However, based on current theory, Georgia was an unusually likely case in which
!
!
!
!
208!
these behaviors should have been observed if current theory was correct. While levels
of social responsibility among diaspora-owned firms in other contexts has not been
empirically evaluated, the only extent theory that would lead us to expect these levels
to be higher than among foreign firms generally is inconsistent with the results
presented here.
The failure of diaspora-owned firms to report more socially responsible
behavior than their non-diaspora-owned peers is particularly striking in contrast with
the large and statistically significant behavioral differences between diaspora-owned
and non-diaspora-owned firms found regarding their use of social networks. In
concurrent work based on the same survey, I show that respondents at diaspora-owned
firms are more likely to report that social networks are important to the profitability of
their firm and to report that their firm has used friends and family members to help
rent or purchase real estate (Graham 2012). Diasporans’ social connections to the
homeland affect firm strategy and augment firm profitability, but they do not motivate
socially responsible firm behavior.
While these survey results come from a single, post-conflict developing
country, Georgia is a relatively typical developing country, and one whose
idiosyncrasies are expected to predispose diaspora-owned firms towards more socially
responsible behavior, not less. Georgia recently fought an interstate war with Russia,
which should have made the patriotic loyalties of diasporans more salient (Nielsen and
Riddle 2010), and its diaspora enjoys a generally positive relationship with the current
government. If diasporans’ social and emotional motivations cause diaspora-owned
!
!
!
!
209!
firms to engage in socially responsible and pro-development behaviors, Georgia is
among the countries in which those differences were most likely to be found.
While my empirical findings contradict the theory and expectations that drive
some of the current optimism regarding diaspora investment as a tool for development,
it is important to note that the findings presented here are not necessarily incompatible
with the argument that social and emotional motivations induce diaspora direct
investment. Even if diaspora-owned firms do not demonstrate greater social
responsibility than other foreign firms, their presence still contributes to the
development in the homeland. The implication of these findings is not that developing
country governments should not seek to attract diaspora direct investment, but rather
that they should not view these firms as inherently more socially responsible than
other foreign firms.
Directions for Future Study
Thus far, the literature on remittance has focused heavily on diasporans from
Latin America and, to a lesser extent, Sub-Saharan Africa, while much of the research
on diaspora direct investment has focused on diasporans from Asia, North Africa, and
the Former Soviet Union. Both literatures have focused almost exclusively on diaspora
populations living in the US and Europe. Comparatively little research has been done
on diaspora direct investment in Latin America and Sub-Saharan Africa, and still less
work on diaspora populations residing outside of the US and Western Europe. While
the Georgian case is ideal for the purposes of this paper (i.e. it is a “most likely” case
in which to observe socially responsible investment behavior by diaspora-owned
!
!
!
!
210!
firms), there is an urgent need for both theory and empirical evidence regarding the
behavior of diaspora populations outside of the regional contexts in which they are
currently being studied. Diaspora populations vary in their geographic concentration,
their social integration in the country-of-residence, and their intention/desire to
eventually return to their homeland.32 For example, South-East Asian migrants
working on oil rigs in the Middle East have very different social structures, and likely
very different remittance and investment behavior, than the diaspora populations that
have so far been the primary objects of study. Building and testing a complete theory
of diaspora investment motivation and reaching a full understanding of the
development potential of diaspora direct investment requires careful attention to these
sources of variation.
Building datasets that allow the direct comparison of diaspora-owned and nondiaspora-owned firms is a painstaking process, but I hope this paper demonstrates the
benefits of doing so. As theory in this area becomes more nuanced, more and better
data will be necessary to test it. As Newland notes, “The quality of information, much
less hard data, about Diaspora influences [on development] is in general very poor,
posing a serious challenge to policy development (2004, p. iv).”
Conclusion
Migrant populations are significant drivers of global flows of foreign direct
investment, and this article takes another step forward in identifying the elements of
diaspora difference that generate this causal relationship. Related research has
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
32!An!exciting!first!step!in!this!research!agenda!was!taken!by!Vaaler!(2012)!who!finds!that!
remittances!from!geographically!concentrated!diasporas!have!a!larger!impact!on!the!availability!
of!venture!capital!in!the!homeland!than!do!remittances!from!geographically!dispersed!diasporas.!!!
!
!
!
!
211!
provided evidence for one mechanism, that diaspora investors use greater access and
attention to information to gain competitive advantage in the homeland (CITATION
REDACTED). This paper tests, and fails to find evidence for, an alternative theory of
diaspora difference based on diasporans’ non-pecuniary motivations for investment.
The economic sociology literature has established the importance of social and
emotional motivations to diaspora investors at the ideational stage. The theoretical
connection between these motivations and socially responsible behavior by diasporaowned firms is clear and direct. Policymakers have launched programs to promote
diaspora investment based, in part, on expectations that these firms will engage in
more socially responsible and pro-development behavior than other foreign firms. This
paper provides the first empirical comparison of social responsibility between
diaspora-owned and non-diaspora-owned foreign firms. I find no evidence that
diaspora-owned firms are more socially responsible than other foreign firms, and I
find some evidence that they are less socially responsible (i.e. with regard to wages,
Table 6 and Fig. 2).
These results underscore the point that diaspora-owned firms are profit seeking
enterprises, not development NGOs. Diaspora-owned firms behave differently than
other foreign firms – in particular, they exploit social networks in the homeland to
gain information and secure trusting relationships with business counterparts – but
they still behave as firms. This does not imply that diaspora investment is bad for
development. The diaspora-owned firms in this survey have hired and trained workers,
!
!
!
!
212!
given to charity, and contributed to the development of the Georgian economy. But
they did not do more of these things than other foreign firms.
These findings should cause policymakers to revise expectations downward
regarding the development-promoting effect of diaspora direct investment. As I have
shown in concurrent work, diaspora-owned firms have important competitive
advantages in the homeland, which may lead them to be more willing than other
foreign firms to invest in fragile and developing countries, particularly as they
stabilize. Diaspora investors should be recruited as uniquely capable, rather than
uniquely socially responsible, investors in fragile and developing countries.
!
!
!
!
213!
References
Aharoni, Yair. 1966. "The Foreign Investment Decision Process." Harvard Graduate
School of Business Administration, Division of Research.
Bandelj, Nina. 2008. From Communists to Foreign Capitalists. Princeton, NJ:
Princeton University Press.
Bertrand, Marianne, and Antoinette Schoar. 2006. The Role of Family in Family
Firms. The Journal of Economic Perspectives 20 (2): 73-96.
Bhagwati, Jagdish, and Koichi Hamada. 1974. The Brain Drain, International
Integration of Markets for Professionals and Unemployment: A Theoretical
Analysis. Journal of Development Economics 1 (1): 19-42.
Brinkerhoff, Jennifer M. 2004. Digital Diasporas and International Development:
Afghan-Americans and the Reconstruction of Afghanistan. Public
Administration and Development 24 (5): 397-413.
Brinkerhoff, Jennifer M. 2008. Diaspora Philanthropy in an At-Risk Society: The Case
of Coptic Orphans in Egypt. Nonprofit and Voluntary Sector Quarterly 37 (3):
411-433.
Brinkerhoff, Jennifer M. 2009. Digital Diasporas: Identity and Transnational
Engagement. Cambridge: Cambridge University Press.
Carroll, Archie B. 1979. A Three-Dimensional Conceptual Model of Corporate
Performance. The Academy of Management Review 4 (4): 497-505.
Clemens, Michael A. 2011. Economics and Emigration: Trillion-Dollar Bills on the
Sidewalk? The Journal of Economic Perspectives 25 (3): 83-106.
de Mello, Luiz R. 1997. Foreign Direct Investment in Developing Countries and
Growth: A Selective Survey. Journal of Development Studies 34 (1): 1-34.
Debass, Thomas, and Michael Ardovino. 2009. Diaspora Direct Investment (DDI):
The Untapped Resource. Washington, DC: United States Agency for
International Development.
Docquier, F., and E. Lodigiani. 2010. Skilled Migration and Business Networks. Open
Economies Review: 21 (4): 565-588.
!
!
!
!
214!
Eriksson, Kent, Jan Johanson, Anders Majkgard, and D. Deo Sharma. 1997.
Experiential Knowledge and Cost in the Internationalization Process. Journal
of International Business Studies 28 (2): 337-360.
Foreign Service Institute. Engaging with Diaspora Communities: Focus on EAP, EUR,
and NEA: Summary Report. 2010. Foreign Service Institute.
Garriga, Elisabet, and Domènec Melé. 2004. Corporate Social Responsibility
Theories: Mapping the Territory. Journal of Business Ethics 53 (1): 51-71.
Gillespie, Kate, Liesl Riddle, Edward Sayre, and David Sturges. 1999. Diaspora
Interest in Homeland Investment. Journal of International Business Studies 30
(3): 623-634.
Gillespie, Kate, Edward Sayre, and Liesl Riddle. 2001. Palestinian Interest in
Homeland Investment. Middle East Journal 55 (2): 237-255.
Giuliano, Paola, and Marta Ruiz-Arranz. 2009. Remittances, Financial Development,
and Growth. Journal of Development Economics 90 (1): 144-152.
Graham, Benjamin A.T. 2010. "Political Risk and Diaspora Direct Investment."
Presented at the Annual Meeting of the American Political Science
Association, Washington, DC.
Graham, Benjamin A.T. 2012. "Doing Business in the Homeland: Diaspora-Owned
Firms and the Value of Social Networks." University of California, San Diego.
de Haan, Arjan. 1999. Livelihoods and Poverty: The Role of Migration ‐ a Critical
Review of the Migration Literature. Journal of Development Studies 36 (2): 147.
Hemingway, Christine A., and Patrick W. Maclagan. 2004. Managers' Personal Values
as Drivers of Corporate Social Responsibility. Journal of Business Ethics 50:
33-44.
Hemphill, Thomas A. 1997. Legislating corporate social responsibility. Business
Horizons 40 (2): 53-58.
Huang, Ye. 2003. Selling China: Foreign direct investment during the reform era.
Cambridge: Cambridge University Press.
Jamali, Dima, and Ramez Mirshak. 2007. Corporate Social Responsibility (CSR):
Theory and Practice in a Developing Country Context. Journal of Business
Ethics 72 (3): 243-262.
!
!
!
!
215!
Javorcik, Beata S., Çaglar Özden, Mariana Spatareanu, and Cristina Neagu. 2011.
Migrant Networks and Foreign Direct Investment. Journal of Development
Economics.
Jones, Marc T. 1999. The Institutional Determinants of Social Responsibility. Journal
of Business Ethics 20 (2): 163-179.
Juholin, Elisa. 2004. For Business of the Good of All? A Finnish Approach to
Corporate Social Responsibility. Corporate Governance 4 (3): 30-31.
Kapur, Devesh. 2010. Diaspora Development and Democracy: The Domestic Impact
of International Migration from India. Princeton, NJ: Princeton University
Press.
Kaufman, Robert, and Alex Segura-Ubiergo. 2001. Globalization, Domestic Politics,
and Social Spending in Latin America: A Time-Series Cross-Section Analysis
1973-97 World Politics 53: 553-87.
King, Gary, J. Honaker, A. Joseph, and K. Scheve. 2001. Analyzing Incomplete
Political Science Data: An Alternative Algorithm for Multiple Imputation.
American Political Science Review 95: 49-69.
King , Gary, Christopher J. L. Murray, Joshua A. Salomon, and Ajay Tandon. 2004.
Enhancing the Validity and Cross-Cultural Comparability of Measurement in
Survey Research. American Political Science Review 98 (01): 191-207.
Kugler, Maurice, and Hillel Rapoport. 2007. International Labor and Capital Flows:
Complements or Substitutes? Economics Letters 94 (2): 155-162.
Leblang, David. 2010. Familiarity Breeds Investment: Diaspora Networks and
International Investment. American Political Science Review: 1-17.
Maimbo, Samuel Muzele, and Dilip Ratha. 2005. Remittances: Development Impact
and Future Prospects. Washington, DC: The World Bank.
Miyagiwa, Kaz. 1991. Scale Economies in Education and the Brain Drain Problem.
International Economic Review 32 (3): 743-759.
Mullings, Beverley. 2011. Diaspora Strategies, Skilled Migrants and Human Capital
Enhancement in Jamaica. Global Networks 11 (1): 24-42.
Newland, Kathleen. 2004. Beyond Remittances: The Role of Diaspora in Poverty
Reduction in their Countries of Origin. Washington, DC: Migration Policy
Institute.
!
!
!
!
216!
Nielsen, Tjai, and Liesl Riddle. 2010. Investing in Peace: The Motivational Dynamics
of Diaspora Investment in Post-Conflict Economies. Journal of Business
Ethics.
Nyberg–Sørensen, Ninna. 2007. "Living Across Worlds: Diaspora, Development, and
Transnational Engagement." International Organization for Migration.
Oberai, A.S., and H.K. Manmohan Singh. 1980. Migration, Remittances and Rural
Development: FIndings of a Case Study in the Indian Punjab. International
Labour Review 119 (2).
Orozco, Manuel. 2002. Globalization and Migration: The Impact of Family
Remittances in Latin America. Latin American Politics and Society 44 (2): 4166.
Pandya, Sonal, and David Leblang. 2012. "Deal or No Deal: Explaining the Rise of
International Venture Capital Investment." University of Virginia.
Porter, Michael E., and Mark R. Kramer. 2002. The Competitive Advantage of
Corporate Philanthropy. Harvard Business Review 80 (12): 56-68.
Rauch, James E., and Vitor Trindade. 2002. Ethnic Chinese Networks in International
Trade. Review of Economics and Statistics 84 (1): 116-130.
Raveloharimisy, Joel, Liesl Riddle, and Tjai Nielsen. 2010. "Measuring Diaspora
Identity Among Liberians in the United States."
Reuveny, Rafael, and Quan Li. 2003. Economic Openness, Democracy, and Income
Inequality. Comparative Political Studies 36 (5): 575-601.
Riddle, Liesl, Jennifer M. Brinkerhoff, and Tjai M. Nielsen. 2008. Partnering to
Beckon Them Home: Public-Sector Innovation for Diaspora Foreign
Investment Promotion. Public Administration and Development 28 (1): 54-66.
Riddle, L, GA Hrivnak, and TM Nielsen. 2010. Transnational Diaspora
Entrepreneurship in Emerging Markets: Bridging Institutional Divides. Journal
of International Management.
Royston, Patrick. 2004. Multiple Imputation of Missing Values. Stata Journal 4 (3):
227-241.
Royston, Patrick. 2009. Multiple Imputation of Missing Values: Further Update of ice,
With an Emphasis on Categorical Variables. Stata Journal 9 (3): 466-477.
!
!
!
!
217!
Rudra, Nita. 2002. Globalization and the Decline of the Welfare State in LessDeveloped Countries. International Organization 56 (02): 411-445.
Saxenian, AnnaLee. 2006. The New Argonauts: Regional Advantage in a Global
Economy. Cambridge, MA: Harvard University Press.
Schulte, Bettina. 2008. "Second Generation Entrepreneurs of Turkish Origin in
Germany: Diasporic Identity and Business Engagement." 56. Center on
Migration, Citizenship, and Development.
Seawright, Jason, and John Gerring. 2008. Case Selection Techniques in Case Study
Research. Political Research Quarterly 61 (2): 294.
Terrazas, Aaron. 2010. "Diaspora Investment in Developing and Emerging Country
Capital Markets: Patterns and Prospects." U.S. Agency for International
Development and the Migration Policy Institute.
Transparency International. 2010. The Georgian Taxation System -- An Overview.
Tbilisi, Georgia: Transparency International.
Uhlaner, Lorraine M., H. J. M. van Goor-Balk, and Enno Masurel. 2004. Family
Business and Corporate Social Responsibility in a Sample of Dutch Firms.
Journal of Small Business and Enterprise Development 11 (2): 186-194.
Vaaler, Paul M. 2011. Immigrant Remittances and the Venture Investment
Environment of Developing Countries. Journal of International Business
Studies 42 (9): 1121-1149.
Vaaler, Paul M. 2012. "Diaspora Concentration and the Venture Investment Impact of
Remittances." University of Minnesota.
Weidenbaum, Murray, and Samuel Hughes. 1996. The Bamboo Network: How
Expatriate Chinese Entrepreneurs are Creating a New Economic Superpower
in Asia. New York: Free Press.
Wibbels, Erik. 2006. Dependency Revisited: International Markets, Business Cycles,
and Social Spending in the Developing World. International Organization 60
(02): 433-468.
Ye, Min. 2010. "Ethnic Investors vs. Foreign Investors: The Impact of Diasporas on
Economic Liberalization in China and India." Presented at the Annual Meeting
of the American Political Science Association, Washington DC.
!
!
!
!
218!
Appendix
Diaspora-Owned and Non-Diaspora-Owned Firms by Home Country
A table showing the percentage of non-responses and diaspora ownership by region is
included in the text. Table 5.8 shows the number of responses, number of refusals, and
number of diaspora-owned and non-diaspora-owned respondents for each investing
country. Firms from Azerbaijan, France, Israel, the Netherlands, and the United States
were particularly likely to refuse to participate in the survey, but there is no obvious
common characteristic across these countries that appears to drive non-response.
!
!
!
!
219!
Table 5.8: Response Types by Home Country
Refused
Country
Armenia
Australia
Austria
Azerbaijan
Belgium
Bulgaria
Canada
Cayman Islands
China
Cyprus
Czech Republic
Estonia
France
Germany
Greece
Iran
Israel
Italy
Japan
Jordan
Kazakhstan
Latvia
Lithuania
Luxemburg
Marshall Islands
Netherlands
Panama
Philippines
Poland
Russia
Sweden
Switzerland
Syria
Turkey
UAE
UK
USA
Ukraine
Virgin Islands
Unknown
Total
2
0
6
10
3
1
0
1
0
1
3
0
5
10
5
2
9
4
1
1
2
3
2
0
2
9
2
0
0
12
0
6
1
23
1
19
20
3
2
26
197
Responded
Diaspora Non-Diaspora
0
2
0
1
1
3
0
2
0
1
0
1
0
1
1
0
0
3
0
4
0
2
1
1
0
2
1
9
0
0
1
3
2
3
1
1
0
0
0
0
0
2
0
3
0
1
0
2
0
1
2
4
0
1
0
1
0
1
5
7
0
1
0
3
0
0
5
14
0
2
4
19
0
10
1
1
0
0
6
10
31
122
!
!
!
Total
4
1
10
12
4
2
1
2
3
5
5
2
7
20
5
6
14
6
1
1
4
6
3
2
3
15
3
1
1
24
1
9
1
42
3
42
30
5
2
42
350
!
220!
Evaluating the Supplemental Sample
Table 5.9 provides the results of difference-in-means tests comparing diasporaowned firms in the random sample from those in the supplemental sample. The
demographic differences between these groups of firms is minimal, as are the
differences in the firms’ responses to statements about social responsibility and
development contribution. The core results of the paper are robust to the exclusion of
the supplemental firms.
Table 5.9: Comparing Diaspora-Owned Firms from the Random Sample
and the Supplement
Variable
Respondent = Owner
Primary Sector
Secondary Sector
Tertiary Sector
Quaternary Sector
100% Foreign
Ownership
HQ in West
Number of
Employees
Tbilisi (location)
HQ in Russia
Pays Higher Salaries
Quality of Life
Labor Costs
Contribution to
Development
Prioritizes Local
Labor
Contributes to
Charity
Random
Sample
Mean
Supplemental
Sample Mean
Standard
Error (of
difference)
t-statistic
0.43
0.04
0.28
0.52
0.16
0.58
0.43
0.17
0.5
0.17
0.17
0.71
0.006
0.13
-0.22
0.35
-0.01
-0.14
0.22
0.11
0.22
0.22
0.17
0.21
0.02
0.1
-1.0
1.6
-0.04
-0.64
0.33
145
0.67
124
-0.33
21
0.30
148
-1.0
0.14
0.81
0.19
4.59
4.53
4.81
5.06
0.71
0.33
4
4.83
4
4.75
0.09
-0.14
0.59
-0.30
0.81
0.31
0.18
0.36
0.84
0.74
0.79
1.02
0.52
-0.55
0.70
-0.41
1.02
0.31
6.57
5.8
0.74
0.54
1.37
0.57
0.67
-0.10
0.24
-0.40
!
!
!
!
Difference
in means
!
Chapter 6:
Conclusion and Directions for Future Research
To explain why some fragile states succeed in attracting FDI and others fail,
and to craft policy approaches to help these states stabilize and achieve economic
growth, we must understand what type of investors are capable of investing in the face
of political risk. In Chapter 2, I argue that political risk specialists provide a reliable
source of capital for weak states plagued by problems of violence and regime
incompetence, but not for fragile states plagued by risk of adverse regime change. In
Chapters 3, 4, and 5, I argue that, while diasporans enjoy substantial competitive
advantages in their (sometimes fragile) homelands, there is no evidence that
diasporans’ social and emotional motivations for investment make them either risk
insensitive or more socially responsible than their peers. However, the high sensitivity
of migrant-induced FDI to improvements in political risk suggests that diaspora direct
investment can serve as a critical complement to, and incentive for, improvements in
governance in fragile and developing states.
These findings highlight the differences between FDI flows into fragile states,
where regimes are unstable and risk of adverse regime change tends to be high, and
weak states, where policy is unstable and the bureaucracy is corrupt or incompetent.
They also identify types of investors who can succeed in investing in states with high
levels of political risk, and the means by which they do so.
With regard to diaspora investors in particular, the expectations currently
guiding much policy are inaccurate. The expectations that diasporans are “brave and
resilient” investors who can be counted on to invest in the face of high levels of
221!
!
!
!
!
222!
political risk, and that diaspora-owned firms are more socially responsible than their
peers, are not supported by the evidence. The evidence presented here suggests instead
that diaspora investors represent a pool of attentive and informed investors who are
likely to respond quickly to improvements in governance. This implies that focusing
on governance improvements during the early stages of fragile-state-stabilization has
the potential to trigger inflows of diaspora direct investment. A wealthy and engaged
diaspora is a valuable asset for a developing country, but the investment potential of
this population cannot be effectively exploited without first limiting political risk.
One of the major strengths of this project is that I apply a mixture of methods
and test the implications of my theory at multiple levels of analysis. The results from
the firm-level survey in Georgia reinforce and unpack findings regarding bilateral
flows of FDI between countries and the location decisions of Japanese manufacturing
firms. However, the firm-level analyses are currently limited by a lack of data
availability, as is the analysis of the patterns of FDI flowing into fragile states. This
makes the acquisition and analysis of additional data the most important next steps in
pushing this research agenda forward.
Next Steps
There are three channels through which I am currently pursuing new data: I
have recently been granted access to classified firm-level data from the Bureau of
Economic Analysis (BEA); I will be rolling out a new firm-level survey of domestic
and foreign-owned firms (including diaspora-owned firms) in the Philippines in early
2013; and I am pursuing both an updated version of the Japanese Overseas Investment
!
!
!
!
223!
data and several possible substitute datasets through the University of Southern
California library system.
The U.S. is the single largest provider of FDI to fragile states and the BEA
data will allow me to look in detail at the composition of U.S. FDI into fragile and
poor countries, including the sector, size, and previous international experience of the
entrants. The data also includes information on firm growth and profitability, allowing
me to examine not only who invests in fragile states, but who succeeds there.
The survey on the Philippines will repeat many of the questions from the
Capital and Conflict: Georgia survey, but will ask additional questions about firms’
interactions with the government – especially their use of civil courts and interactions
with regulators, customs officials, and tax authorities. Two of the primary goals of the
survey are to confirm the core findings in Chapters 4 and 5, and to test whether
diaspora-owned firms can use their social networks to substitute for poorly functioning
formal political institutions, particularly civil courts. The survey will include both
foreign and domestic firms, allowing me to assess the role of diaspora capital and the
use of social networks in both categories of firm
An updated version of the Japanese Overseas Investment in necessary to re-run
the firm-level analyses from Chapter 5 with a wider range of risk variables. This will
allow me to test the two hypotheses left untested in the current version, and do a more
thorough job of testing those I was able to examine. I also am exploring the OSIRIS
database from Bureau Van Dijk and several other options. These databases contain
information on foreign entries by multinationals based in a range of home countries,
!
!
!
!
224!
which may allow me to test for effects of both home- and host- country political risk
experience at the firm level. However, it is not clear how easy it will be to convert
information from these data sources into a useable format for this purpose.
Two Book Projects
This dissertation contains the seeds of two separate book projects. One book
project will share its title with the dissertation and focus on foreign investment in
fragile states. This book will depend heavily on data from the BEA to provide
descriptive statistics regarding the type of U.S. investment taking place in fragile
states. It will also rely on firm-level data from the Japanese Overseas Investment data
and potentially other firm-level databases. The book will first introduce descriptive
statistics, then present theory about who invests in fragile states and who succeeds
there (hypotheses about home country, prior international experience, sector, size, and
ownership structure), and then test these hypotheses using several distinct firm-level
datasets.
The other book project is on the political economy of diaspora direct
investment. This book will primarily present a general theory of diaspora direct
investment, discuss the findings from the Georgia and Philippines surveys as well as
from Chapter 3. I am also exploring the possibility of including data from a third
survey – most likely in India.
As noted above, expanding the survey to additional countries will allow me
both to improve the external validity of my findings, as well as drill deeper into how
diasporans interact with the homeland government, including their use of courts, their
!
!
!
!
225!
interface with regulatory bodies, and their involvement in local politics. In India I
expect to be able to pre-stratify the sample by diaspora ownership, which I was not
able to do in Georgia. In the Philippines I will be conducting the survey in
collaboration with Cesi Cruz, who has already run a number of successful surveys
looking at vote buying in the Philippines and has the necessary experience asking
politically sensitive questions in that country.
These book projects provide the avenue for pursuing the empirical testing
begun in this dissertation, as well as an opportunity to further sharpen the theory to
derive additional testable implications. The theoretical and normative implications of
the project are substantial, and the current state of knowledge quite limited. I am
optimistic that a broad sustained research agenda in this area can be extremely
productive.
!
!
!