Foreign Direct Investment and Democracy: A Robust Fixed Effects Approach to a Complex Relationship
Author | Vincenzo Verardi,Rodolphe Desbordes |
Published date | 01 February 2017 |
DOI | http://doi.org/10.1111/1468-0106.12204 |
Date | 01 February 2017 |
FOREIGN DIRECT INVESTMENT AND DEMOCRACY: A
ROBUST FIXED EFFECTS APPROACH TO A COMPLEX
RELATIONSHIP
RODOLPHE DESBORDES*University of Strathclyde
VINCENZO VERARDI University of Namur and University of Brussels
Abstract. We develop a new robust-to-outliers dummy estimator that we subsequently apply to
investigate the impact of various democratic attributes on foreign direct investment in recent
years. We find that democracy has generally a positive impact on foreign direct investment,
once outliers are controlled for, but that this relationship is very specific to each host country’s
characteristics.
1. INTRODUCTION
The impact of democracy on foreign direct investment (FDI) in developing
countries is a hotly debated issue. Jensen (2003, 2008) and Azémar and
Desbordes (2009) provide an overview of the literature on FDI and democracy.
On the one hand, it has been argued that democratic forms of government at-
tract FDI because they tend to be predictable and credible, thanks to vertical
and horizontal accounting mechanisms. The combination of periodic elections
and widespread diffusion of information ensures that politicians make clear their
preferences, do not renege on their promises once in power, and, more broadly,
do not abuse their position for private gains. Strong checks and balances in-
crease policy stability, and, ultimately, the protection of property rights, because
the existence of several distinct veto players hinders any changes in the status
quo. When policy changes do occur, institutionalized channels of influence/
lobbying can allow foreign investors to mitigate any adverse effects. Consoli-
dated democracies also guarantee orderly and non-violent changes of govern-
ment within the current institutional framework.
On the other hand, the virtues of democracy can easily be seen as disadvan-
tages to countries wanting to attract multinational enterprises. Politicians’desire
to win elections may lead them to promise and adopt popular but economically
harmful policies. Political turnover creates uncertainty about the persistence of
policies, whereas an increase in the number of veto players has the potential to
result in long-term reform sclerosis. Foreign firms may suffer from a political li-
ability of foreignness and be discriminated against relative to their better
politically-connected domestic competitors. Finally, a related but more sinister
view is that freedom of expression and open media can prevent foreign firms
*Address for Correspondence: Department of Economics, Sir William Duncan Building, University
of Strathclyde, 130 Rottenrow, Glasgow G4 0GE, Scotland, UK. E-mail: rodolphe.
desbordes@strath.ac.uk. Vincenzo Verardi is associated researcher of the FNRS and gratefully ac-
knowledges their financial support
Pacific Economic Review, 22: 1 (2017) pp. 43–82
doi: 10.1111/1468-0106.12204
© 2017 John Wiley & Sons Australia, Ltd
bs_bs_banner
from colluding with officials in order to obtain generous entry deals or to de-
crease market competition. Hence, FDI may be attracted by autocracies.
This ambiguous theoretical impact of democracy on FDI is mirrored in the
empirical literature. As described in Table 1, studies find positive (Harms and
Ursprung, 2002; Jensen, 2003; Busse, 2004; Azémar and Desbordes, 2009), neg-
ative (Li and Resnick, 2003) or insignificant (Blanton and Blanton, 2007; Büthe
and Milner, 2008; Choi and Samy, 2008) impacts, depending on the econometric
model, the proxy for democracy and the sample used.
The presence of outliers (i.e. observations that are substantially different from
the bulk of the data) has sometimes been blamed for this variety of findings.
Some researchers have attempted to deal with these atypical observations
(Jakobsen and De Soysa, 2006; Choi, 2009; Li, 2009). However, it is unlikely
that they have solved this issue in a satisfactory manner. Popular robust estima-
tors (e.g. median regression estimator or M-estimator) are sensitive to outliers
present in the values of explanatory variables and outlier diagnostics (e.g. the
Cook distance) frequently fail to detect atypical observations as they fundamen-
tally rely on the extremely non-robust-to-outliers least-squares estimator
(Verardi and Croux, 2009). Problems of inadequate econometric identification
and treatment of outliers thus remain pervasive in the FDI literature.
A related issue is the increasing use of the fixed effects estimator to control for
unobserved time-invariant unit-specific effects that may be correlated with the
explanatory variables. While this estimator may reduce any omitted variable
bias, it does not protect against outlier contamination. An aggravating problem
is that the most resistant robust regression methods are ill-designed to analyse
econometric models, which include a large number of dummies. While some
robust panel data estimators have been proposed, they appear not to be consis-
tent when the number of time periods is small, and cannot deal with dummies
not related to the within transformation (Aquaro and Pavel, 2010).
In this paper, we revisit the link between FDI and democracy in develop-
ing countries, during the 1998–2006 period. We pay particular attention to
the multi-faceted nature of democracy, the influence of outliers, and the het-
erogeneous relationship that may exist between these two variables. We use
a variety of measures to explore how each particular dimension of democ-
racy can influence FDI. We develop a new robust dummy estimator, which
allows us to control for outliers in a fixed effects panel data model. Finally,
we investigate whether the impact of democracy on FDI depends on natural
resources, income inequality, and their interaction. Our results show that
democracy generally has a positive impact on FDI, once outliers are con-
trolled for, but that this relationship is very specific to each host country’s
characteristics.
The rest of the paper proceeds as follows. Section 2 introduces our new robust
dummy estimator and explains its advantages over existing robust regression
methods. Section 3 provides Monte-Carlo simulations highlighting the good
performance of our estimator. Section 4 replicates a key study on FDI and
democracy to highlight that the correct treatment of outliers is a key issue.
Section 5 describes the data used in our original empirical analysis and motivates
R. DESBORDES AND V. VERARDI44
© 2017 John Wiley & Sons Australia, Ltd
Table 1. Selective review of the impact of democracy on FDI
Reference Citations Sample Main estimator(s) Dependent variable Main proxy (proxies)
Explicit control
for outliers Findings
Harms and
Ursprung (2002)
258 62 developing
countries;
1989–1997
Between
(cross-sectional);
fixed effects
FDI/population Freedom house
indexes of
political
rights and
civil rights
Yes Positive but
statistically
significant only
with a between
estimator
Jensen (2003) 653 114 developed
and developing
countries;
1970–1997
Fixed effects;
dynamic
FDI/GDP Polity III score No Positive
direct impact
Li and
Resnick (2003)
587 53 developing
countries;
1982–1995
Pooled FDI Polity IV score No Negative direct
impact; indirect
positive impact
via strengthening
property rights
protection
Jakobsen and
De Soysa (2006)
88 99 developing
countries;
1984–2001
Pooled;
dynamic
ln(FDI) Polity IV score Yes; log
transformation
to attenuate
influence of
outliers in
dependent
variable; China
dummy variable
Positive direct
impact
Blanton and
Blanton (2007)
114 129 developing
countries;
1980–2003
Pooled FDI/GDP Polity IV score No Statistically
insignificant
impact
Busse and
Hefeker (2007)
576 83 developing
countries;
1984–2003
Fixed effects/
DIFF-GMM;
dynamic
ln(FDI + sqrt(FDI
2
+1)) ICRG
democracy
index
No but
dependent
variable is
transformed
Positive direct
impact
(Continues)
FOREIGN DIRECT INVESTMENT AND DEMOCRACY 45
© 2017 John Wiley & Sons Australia, Ltd
To continue reading
Request your trial