1
Economic Globalization and the Environment
Vally Koubi,1,2 Tobias Böhmelt,1,3 and Thomas Bernauer1
1 CIS, ETH Zurich
2 VWI, University of Bern 3 Department of Government, University of Essex
Abstract
The authors focus on the standard Factor Endowment Theory (FET) and the Pollution Haven
Hypothesis (PHH) for contributing to our understanding of the relationship between economic
globalization and the environment. One implication of the theoretical framework that
combines both approaches is that inward-investment flows are likely to decrease domestic
environmental quality in developing countries. The theory’s main contribution, however, is
the ability to distinguish between the FET and the PHH with the interactive effects of trade
and inward FDI. Empirically, the authors study the impact of economic globalization in the
form of trade openness and FDI inflows on SO2 emissions in 150 countries between 1971 and
2003. The findings suggest that inward FDI is indeed associated with higher levels of SO2
emissions in developing states, while increases in trade amplify the negative effects of FDI
inflows. This constitutes prima facie evidence in favor of the PHH over the FET.
Keywords: environmental quality; factor endowments; pollution haven; trade; foreign direct
investment; SO2 emissions
Paper to be presented at the annual meeting of the International Political Economy
Society (IPES), Stanford, CA, USA, November 13-14, 2015.
2
1. Introduction How do trade liberalization and foreign direct investment (FDI) affect the environment? Put
differently, what is the relationship between economic globalization and environmental
quality? A key development over the last few decades is the significant increase in the degree
of globalization (see, e.g., Dreher et al. 2008). While there are different forms, e.g., political
or social (Dreher 2006), economic globalization has important implications for numerous
issue areas and manifests itself via the following main channels: increased trade in goods and
services and a higher capital mobility. For example, reductions in trade barriers and the
relaxation or elimination of capital controls led to increases in trade and capital flows that
have outpaced the rate of economic growth. In fact, the levels of trade and asset openness1 are
much higher at the present time than, e.g., 25 years ago. To underline these claims, Figures 1
and 2 depict the development of FDI flows and trade openness, respectively, over time.2
_________
Figures 1 and 2 in here
__________
As indicated, economic globalization influences several policy areas in important ways,
ranging from living standards to the distribution of economic and political power. One such
issue that currently occupies center stage in scientific and political agendas is the
environment: the channels of economic globalization from above are thought to matter for
environmental quality, although different theories and approaches disagree how this actually
occurs and with what empirical implications. Specifically, according to standard trade theory
(i.e., the Heckscher-Ohlin model), trade in goods worsens environmental quality in those
1 We define the former as the sum of imports and exports divided by GDP. The latter is specified by
foreign assets as a percentage of GDP. 2 The graphs are based on the data sources we discuss in the research design section below.
3
countries that have a comparative advantage in the production of “polluting” goods. On one
hand, the comparative advantage may derive from the distribution of the world endowments
of the factors of production (i.e., the factor endowment theory, FET), in which case the
developed countries become dirtier with free trade due to their capital abundance. On the
other hand, the comparative advantage may come from policy-related differences in tolerance
of pollution (i.e., the pollution haven hypothesis, PHH), in which case the less developed
countries are expected to have worse environmental quality with more international trade
(Taylor and Copeland 1994; see also Taylor 2004).
Most importantly, however, static-trade theory abstracts from an important determinant of
environmental quality that is affected by international trade, namely income. In a pioneering
study that considers both the direct and indirect effects of trade, Antweiler et al. (2001)
demonstrate that such income (i.e., so-called technique) effects are sufficiently large in the
long run as to overcome the negative impact arising from the scale and the composition of
economic activity. Ultimately, it is found that trade has a positive effect on environmental
quality as measured by lower SO2 concentrations (see also Frankel and Rose 2005).
The relationship between international capital mobility, in particular Foreign Direct
Investment (FDI), and the environment has also received considerable attention. The
predominant theoretical framework underlying this relation is the PHH. In short, it is
postulated that polluting firms will find it profitable to relocate to countries with “lax”
environmental standards. Consequently, FDI flows will worsen the environment in the
receiving country, and improve it in the originating one (e.g., Mani and Wheeler 1998). The
empirical evidence on the relationship between FDI and environmental quality is rather
mixed, though (Copeland and Taylor 2004). For instance, several studies find support, albeit
only weak at times, for the PHH (e.g., Dean et al. 2009; Kellenberg 2009; Jorgenson 2009;
Wagner and Timmins 2009; Acharyya 2009; Cole and Fredriksson 2009; He 2006; Cole and
Elliott 2005; Millimet and List 2004; Eskeland and Harrison 2003; Keller and Levinson 2002;
4
Xing and Kolstad 2002). Others, however, provide evidence for the absence of any link (e.g.,
Spilker 2013; Perkins and Neumayer 2009; Elliott and Shimamoto 2008; Javorsik and Wei
2004) or for conditional effects (Lan et al. 2012; Manderson and Kneller 2012).
In light of this, we believe that the lack of consensus in the empirical results suggests that
the impact of trade and FDI on environmental quality is likely to differ according the
circumstances and, hence, intervening influences. Particularly interesting for this study, Lan et
al. (2012), for example, employ data on Chinese provinces and condition the relationship
between FDI and pollution on human capital levels. They show that the PHH holds only in
provinces with low human capital, whereas FDI is negatively associated with pollution
emissions in provinces with higher human capital due to a “technology effect.” In essence, the
core of our paper is a conditional, interactive effect as well as we seek to provide new insights
into the relationship between economic liberalization and pollution by developing an
argument that simultaneously considers the effects of trade and FDI in an interactive fashion.
Specifically, we pursue a twofold contribution. First, theoretically, we develop an
argument that theorizes on the join impact of free trade and FDI on environmental quality,
while encompassing the PHH and standard-trade theory effects. This argument implies, on
one hand, that FDI inflows are likely to have a negative, i.e., worsening effect on the
environmental quality in developing countries, independent of whether the mechanisms of the
FET or the PHH apply. On the other hand, our argument in fact helps to distinguish between
these two approaches. If the PHH plays the dominant role in production and trade patterns, we
expect a joint effect of FDI and trade to decrease environmental quality in developing
countries. If differences in factor endowments do matter more, the interaction of trade and
FDI should suggest that the joint effect actually improves environmental quality in developed
countries.
Second, we seek to make an empirical contribution by quantitatively examining the join
effect of international trade and FDI on SO2 emissions using time-series cross-section data on
5
150 countries between 1971 and 2003. Our analysis, while controlling for a series of
alternative determinants of environmental quality including year and country fixed effects,
highlights that more FDI inflows into less developed states are indeed associated with higher
SO2 emissions. We also obtain evidence that the interaction between trade and inward FDI
results in worse environmental quality. Particularly this last finding provides support for the
PHH over the FET. In fact, we believe that these results and, more accurately, their
underlying interactive specification between trade and FDI might be an explanation for the so
far mixed results of earlier studies on the PHH.
The next section briefly reviews the relevant literature pertaining to the effects of trade and
FDI on the environment and presents our theoretical argument. Afterwards, we discuss the
research design, before presenting the empirical results. The last section concludes with a
discussion of our findings and the implications for future research and policymakers.
2. Development, Trade, and FDI Flows and their Impact on Environmental Quality One of the most debated issues concerns the environmental impact of economic globalization,
i.e., free trade and free capital movement (FDI). However, the effect of economic
liberalization on the environment is both theoretically and empirically ambiguous in the
existing literature. One of the reasons for this ambiguity might stem from the fact that the
impact of economic liberalization on the environment for a given level of income operates at
least via three channels (Frankel 2005): system-wide effects that are either adverse or
beneficial, and effects that vary across countries depending on a local “comparative
advantage”.3
3 Economic liberalization also affects the environment via income and it has been shown that, at least
for local pollutants such as SO2 emissions, the net effect of trade is the reduction of pollution
(Antweiler et al. 2001; Frankel and Rose 2005; Damania et al. 2004).
6
Starting with the system-wide effects, several scholars argue that increased competition
between companies induced by economic liberalization causes a “race to the bottom” in
countries’ environmental standards: states will lower their environmental policy standards in
order to protect their industry from international competition (e.g., Esty and Geradin 1998;
Prakash and Potoski 2006b); similarly, governments could seek to attract foreign firms and
FDI with lower environmental protection (e.g., Sheldon 2006: Copeland and Taylor 2004).
While various studies attempt to provide empirical evidence about the nature and impact of
these linkages, there is actually little evidence that environmental regulation is one of the most
important determinants of firms’ ability to compete internationally. It seems that multinational
firms’ location decisions are determined more by issues such as labor costs and market access
rather than the stringency of local environmental regulations (Potoski 2001; Drezner 2001, Jaffe
et al 1995).
In contrast, the “race-to-the-top” argument suggests that economic liberalization induces
an international improvement of environmental standards and, in turn, has a positive effect on
the environment (e.g., Porter and van der Linde 1995). Empirically, Vogel (1995; 1997), for
example, shows that stricter regulations tend to diffuse among industrialized countries,
because companies that produce for multiple markets have an incentive to standardize
production processes for cost minimization. In addition, Porter and van der Linde (1995) state
that a tightening of environmental regulations stimulates technological innovation, which
helps to improve competitiveness. However, there exists only weak evidence that
environmental regulations actually stimulate innovation (Ambec and Barla 2006).
With regard to comparative advantage, which constitutes the starting point of our
theoretical argument,4 standard trade theory (i.e., the Hecksher-Ohlin model) asserts that trade
4 Our argument is based on three simple and justifiable assumptions. First, developed countries are
capital abundant and less developed countries are capital poor. Second, manufacturing is polluting
7
leads to more production of goods that are intensive in that factor, which is abundant in the
country concerned. The literature then provides two competing approaches to predict the
effect of international trade on the environment: the factor endowment theory (FET) and the
pollution haven hypothesis (PHH). On the one hand, if the comparative advantage derives
from the distribution of world endowments of the factor of production (FET)5, countries
where capital is relatively abundant will export capital-intensive (“dirty”) goods. This
stimulates production, while pollution increases (Antweiler et al. 2001). Given that developed
countries are capital abundant, they will be more polluted with international trade. On the
other hand, if the comparative advantage derives from policy-related differences across
countries in tolerance of pollution (PHH), developing countries are expected to have worse
environmental quality with international trade due to pollution-haven effects (Copeland and
Taylor 1994). Ultimately, this leads us to expect the effect of trade on pollution to be positive
–independent of which theory, the FET or the PHH, is the relevant one.
Hypothesis 1 (Trade-Environment Hypothesis): Environmental quality decreases with
higher levels of trade openness.
Developed and developing countries differ, however, with regard to the levels of income
which is also expected to have an effect on environmental quality. According to the well-
known argument underlying the environmental Kuznets curve (EKC), environmental quality
deteriorates at low levels of income and then improves at higher income levels after a tipping
capital-intensive. Third, manufacturing is more polluting than agriculture and agriculture is more
polluting than service (see also Copeland and Taylor 2003). 5 A comparative advantage could also be determined by endowments of natural resources. In such
case, a country with abundant natural resources, say oil or forests, will most likely export them, and
thus trade is likely to damage the environment.
8
point is reached (Grossman and Krueger 1995). The effects of economic scale, technology
and structure, as well as the income elasticity of environmental demand explain this inverted
U-shaped relation between income and environmental pollution. That is, over the course of a
country’s economic development, rising national income increases the scale of economic
activity, which leads to higher pollution levels (scale effect). Then, after a certain threshold of
national income has been reached, the production process of the economy becomes more
technology and information based (technology effect), and the service sector is boosted
(composition effect). This shift in the composition of production, combined with
improvements in technology result in a leveling-off and, eventually, a steady decline of
environmental degradation. In addition, at higher income levels, people pay more attention to
the quality of life and want to enjoy better environmental quality (see also Congleton 1992).
The government will respond to these demands by enacting and enforcing stricter
environmental regulations, which further improve environment quality. As a result, the EKC
argument posits that developed countries are likely to have stricter environmental regulations,
while the developing states are expected to have laxer environmental regulations relative to
the developed ones.
A number of empirical studies (Cole et al. 1997; Shafik and Bandyopadhyay 1997;
Grossman and Krueger 1995; Selden and Song 1994) find this non-monotonic, inverted U-
relationship between several local pollutants and income, suggesting a changing relationship
between environment and growth along the course of a country’s economic development.6
For example Grossman and Krueger (1995) examine data on air pollution levels in 42
6 The literature on the EKC has been criticized that it does not adequately capture all of the factors,
which are considered important for the relationship between income and pollution; and that most
statistical models are not correctly specified (see, e.g., Aklin 2015; Stern 2004; Cole and Elliot 2003;
Dasgupta et al 2002; Millimet et al 2003).
9
developed and developing countries and find support for the EKC in that pollution levels first
rise and fall with GDP per capita.7
Based on this discussion, we argue that ultimately the relationship between economic
globalization, that is the combined effect of trade and FDI, and environmental quality might
be conditional on a country’s income.
With higher income, developed countries enact stricter environmental regulations relative
to developing ones, which tend to increase the costs of production and erode some of their
industries’ competitiveness. Under such conditions, multinationals may have an incentive to
shift some or all of their (“dirty”) production to other countries with weaker environmental
regulations. Hence, according to the FET, “dirty” capital will flow out from the developed
countries. Given that developed countries have comparable stringent environmental
regulations, FDI can only go to developing countries that are assumed to have lax
environmental regulations. Cole and Elliott (2005) provide empirical evidence that the level
of pollution abatement costs in a US industry significantly determines that industry’s FDI
decisions. Along the same lines, Eskeland and Harrison (2003) report that industry pollution
abatement costs have a positive, albeit weak, impact on US outbound FDI into four
developing countries. Xing and Kolstad (2002) provide evidence that the laxity of
environmental regulations in a host country is a significant determinant of FDI from the US
for heavy polluting industries.
According to the PHH, laxer environmental regulations in a developing host country
attract FDI from pollution-intensive sectors that want to avoid costly environmental
7 Empirical evidence shows that global pollutants such as CO2 do not conform to the EKC predictions,
since the impact of global warming is (largely externalized to other countries and future generations.
Hence, these global pollutants are expected to increase monotonically with income, at least within an
observable income range (Cole et al. 1997; Shafik 1994).
10
compliance in their developed home country (e.g., Dasgupta et al. 1995).8 Dean et al. (2009)
examine whether weak environmental stringency affects FDI inflows into China. They report
that environmental stringency only influences certain types in highly polluting industries with
investment originating from Hong Kong, Macau, and Taiwan being attracted to provinces
with relatively weak environmental controls.On the other hand, investment from OECD is
attracted to high-regulation regions. Javorcik and Wei (2005) also analyze the relationship
between cross-country FDI flows and environmental stringency for 143 multinational firms in
25 Eastern European countries. They find some evidence for the PHH in regressions
employing treaties as a proxy for environmental standards in a host country, but the overall
evidence is relatively weak.
Given that both the FET and PHH argue that “dirty” FDI will flow from developed to
developing9 countries, the expectation is that the environmental quality will deteriorate in
these countries, and by implication it will improve in developed ones.
Hypothesis 2 (FDI-Environment Hypothesis): Environmental quality in developing countries
decreases with FDI inflows.
8 Opposite to the PHH, there is the pollution halo hypothesis, which states that FDI has a positive
effect on the environment through the use of environmental-friendly techniques of production as well
as the transfer of such technologies from developed countries to developing ones that rely mainly on
environment-damaging production techniques such as less efficient and polluting types of energy (e.g.,
Aklin 2015; Elliott and Shimamoto 2008; Gallagher and Zarsky 2007; Prakash and Potoski 2006a,
2007; Eskeland and Harrison 2003). 9 Given that capital is scarce in developing countries, we should not expect capital outflows from such
countries. However, developing countries rich in natural resources might be an exception. But even in
this case, FDI can flow only to other developing countries and not to developed ones due to the
stringency of their environmental regulations.
11
Coming now to the globalization effect on environmental quality, the FET suggests that the
joint effect of FDI outflows and free trade on the environment is positive for only the
developed countries. Recall that developed countries are capital abundant and thus specialize
in the production of manufacturing goods, which are capital-intensive and hence harmful to
their environment. Nevertheless the stringency of their environmental regulations will force
“dirty” industries via FDI to relocate (and it will also not allow any ‘dirty’ capital to enter
independently of the country of origin), and consequently developed countries will specialize
in the production and exports of clean goods. In other words, according to the FET, the
negative effect of trade will be mitigated by FDI outflows and hence their join effect should
lead to improved environmental quality in developed countries.
According to the PHH, countries with laxer environmental regulations will attract FDI
from firms that seek to escape stringent environmental regulations in the developed countries.
Since the developing countries have laxer environmental regulations, because the
environment is considered to be a luxury good at their low level of income, ‘dirty’ FDI is
more likely to flow into developing countries. FDI inflows to developing countries will
increase the scale of economic activity, i.e., more production, which will result in worse
environmental quality. That is, the joint effect of FDI inflows and free trade on the
environment in will be even “worse” than the negative trade-effect per se, since they will
expand production and exports of the dirty goods. In other words, according to PHH, the
negative effect of trade will be amplified by FDI inflows and hence their combined effect
should be detrimental to the environment of less developed countries.
Hypothesis 3 (FDI-Trade- Hypothesis): There is a joint effect of FDI inflows and trade on
environmental quality. Under the FET, the joint
impact improves environmental quality. Under the
12
PHH, the joint impact deteriorates environmental
quality.
3. Research Design
Data, Methodology
In order to test the empirical implications of the theoretical framework, we analyze a data set
that has the country-year as the unit of analysis and covers the time period from 1970 to 2003.
This time period is pre-determined by the availability of data for our dependent variable and
the explanatory items; moreover, since all our explanatory variables are lagged by one year,
the “effective” starting date of our analysis is 1971. Given the continuous nature of our
dependent variable (explained in the next paragraph) and the time-series cross-section format
of our data, we use OLS regression models that include country fixed effects, year dummies,
and a lagged dependent variable. The longitudinal nature of our data allows us to consider the
role of countries’ past emission levels on their current emissions.10 While this also captures
any existent time dependencies more generally, year fixed effects control for temporal shocks
that are common for all states in a given year. Finally, country-fixed effects control for
idiosyncratic path dependencies and other forms of cross-sectional heterogeneity.
Dependent Variable
In terms of our dependent variable, we measure environmental quality by sulfur dioxide (SO2)
emissions per capita (SO2 Emissions per capita (ln)).
10 The inclusion of a temporally lagged dependent variable also has implications for inference. Given
the structure of the data, serially correlated errors within countries might be possible. The lagged
dependent variable addresses this possibility (Beck 2001). We are aware of the arguments against the
inclusion of a temporally lagged dependent variable in fixed effects models (Plümper, Troeger, and
Manow 2005), but we opt to include it, since it yields more conservative estimates. Note, however,
that the substantive results we report below are identical (or even stronger) when the temporally
lagged dependent variable is omitted.
13
While there are several pollutants or emission types that could serve as indicators of
environmental quality, for the purposes of our study, a pollutant should fulfill the following
requirements. First, it must be produced by human activity. Second, it must have harmful
effects on humans, ecosystems, and the economy. Finally, for statistical purposes, there must
be available data for both developed and developing countries. Air pollution, and in particular
SO2 emissions fulfill all these requirements.
First, SO2 is one of the so-called criteria pollutants,11 used by the World Bank, the OECD,
and numerous other national and international authorities to measure air quality. Second, SO2
is arguably the most prominent form of air pollution worldwide, since it has direct and visible
effects on human health, ecosystems, and the economy. Third, SO2 emissions can be directly
affected (i.e., decreased or increased), if governments wish to, by altering the techniques of
production. For example, SO2 emissions can be curtailed by reducing consumption of fossil
fuels (especially high-sulfur coal), by using smoke-scrubbing equipment in power plants and
smokestacks, by reducing the sulfur content of fossil fuel, or by increasing energy
efficiency.12 Finally, emissions are more closely linked to economic activity than
concentrations (Stern et al., 1996).
The data for countries’ annual SO2 emissions are taken from Stern (2005; 2006), which
cover the time period until 2003. We divided this information by a state’s population to
standardize it, and consider its natural log as the final variable in order to scale down the
variance and reduce the effect of outliers.
Core Explanatory Variables: Trade and Foreign Direct Investment
We operationalize our first core explanatory variable, trade, with a country’s degree of
openness, i.e., the share of imports plus exports divided by GDP (Trade Openness (ln)). This 11 Carbon monoxide (CO), nitrogen oxide (NO2), ozone (O3), particulate matter (PM10 and PM2.5), and
lead (Pb) are other criteria pollutants. 12 Although these emission-reduction measures are readily available and effective, they are also costly.
14
item is based on Gleditsch’s (2002) trade and population data, and we lag and log it as well.
Imports and exports of this variable are measured in current US dollars.
The data for our second set of core variables are taken from Lane and Milesi-Ferretti
(2001; 2007) who recently updated their data on foreign direct investment to 188 states in
1970-2011. Specifically, it is distinguished between FDI inflows and outflows, with “capital
inflows measuring net purchases or sales by nonresidents of domestic assets, while outflows
measure net purchases or sales of foreign assets by residents” (Lane and Milesi-Ferretti 2007:
225). More generally, though, according to these authors, “the FDI category includes
controlling stakes in acquired foreign firms (at least 10% of an entity’s equity – in practice,
however, most FDI holdings reflect majority control), as well as green-field investments. In
addition, at least for some countries, an increasingly important component of FDI is foreign
property investment” (Lane and Milesi-Ferretti 2007: 227). In line with our theoretical
argument, FDI flows are disaggregated into inflows and outflows as we argued above that the
effect of these should go in opposite directions. The former is measured by FDI liabilities,
while we take FDI assets for the latter. Both variables are measured in current US dollars,
expressed as a percentage of a country’s GDP, and lagged by one year and logged.
Finally, in order to test our expectations regarding the intervening influence of GDP per
capita and trade, respectively, some of our models include multiplicative terms between FDI
Inflows / GDP (ln) and GDP per capita (defined below) and between FDI Inflows / GDP (ln)
and Trade Openness (ln).13 Since we cannot directly interpret the size, signs, and t-statistics of
the components of a multiplicative specification in an OLS regression, we calculated average
marginal effects (Brambor et al. 2006), which are displayed in graphical form below.
13 As argued above, the impact of trade does not vary across developed and developing states, and we
do find evidence for this. Hence, specifying a three-way interaction between income, trade openness,
and FDI inflows is not necessary to test our third hypothesis; a two-way specification between trade
and inward FDI is sufficient.
15
Control Variables
We also consider a series of control variables, which account for alternative determinants of a
country’s SO2 emission level. Hence, our control variables include standard domestic-level
economic and political variables and international-level influences such as the participation in
international organizations or the number of international environmental treaties a country has
ratified.
First, regarding the economic determinants, some forms of environmental degradation,
including SO2 emissions, follow a Kuznets curve pattern (e.g., Grossman and Kruger 1995;
Selden and Song 1994): pollution first deteriorates and then improves as economic
development in the form of GDP per capita increases. In light of this, we include GDP per
capita (ln) and its square term in our estimations. We also employ this variable to distinguish
between developed and developing states and, hence, to model the interaction between GDP
per capita (ln) and FDI Inflows / GDP (ln).The data are taken from Gleditsch (2002) and are
measured in current US dollars (million).
Second, environmental degradation may be the result of high rates of economic growth and
a larger scale of economic activity more generally (e.g., Grossman and Kruger 1995; Selden
and Song 1994). Moreover, a larger population usually consumes more natural resources and,
hence, produces greater environmental degradation (e.g. Spilker 2012). In order to address
these concerns, we include a state’s lagged yearly average percentage growth in GDP and
measure the scale of economic activity by the lagged and log level of a country’s GDP.
Moreover, we consider Population Density (ln), which is the lagged and logged midyear
population divided by land area in square kilometers. All of these three variables are taken
from the World Bank Development Indicators, while GDP (ln) and GDP Growth in % are
measured in current million US dollars.
Third, coming to a domestic-level political influence, it is frequently claimed that non-
democratic regimes are likely to underprovide public goods, including environmental quality
16
(Congelton 1992; Bueno de Mesquita et al. 2003; Fredriksson et al. 2005; Li and Reuveny
2006; Bernauer and Koubi 2009). The underlying rationale for this contends that non-
democratic regimes are typically ruled by small elites that use the resources of their respective
country to create personal wealth and to redistribute income from their populations towards
themselves. If the costs of stricter environmental policies are born disproportionately by the
elites (as it would be the case with restrictions on polluting industrial activities) while the
benefits are uniformly dispersed throughout the population, these elites would have little
incentive to implement such policies.
In contrast, in democracies, the median voter who decides on public policy faces a lower
cost from environmental policies relative to the economic and political elite. This makes the
adoption and implementation of stricter environmental policies more likely in democratic
regimes. In order to control for this mechanism, we rely on the polity2 item from the POLITY
IV data set (Marshall and Jaggers 2002). This variable is a composite index that includes the
following elements: presence of competitive political participation, guarantee of openness and
competitiveness of executive recruitment, and existence of institutionalized constraints on the
exercise of executive power. Our final variable, Democracy, ranges from -10 (highly
autocratic) to +10 (highly democratic).
Finally, we incorporate international determinants in the form of states’ membership in
intergovernmental organizations (IGOs) and their participation in international environmental
treaties. International policy cooperation is an important component of political globalization
(Dreher 2006). International policy cooperation can produce a globally cleaner environment
by limiting free riding/externality problems in the production of pollution, i.e., by reducing
the cross-country spillover effects of poor, national environmental policies. In addition, states
more strongly involved in the IGO network are also more likely to cooperate on
environmental issues, due to socialization, information exchange, and enhanced possibilities
for issue linkages across policy areas. Recent studies have shown that indeed membership in
17
intergovernmental organizations (IGOs) is robustly associated with a cleaner environment
(Böhmelt and Vollenweider 2015; Spilker 2012; Ward 2006). We thus include the number
IGOs a state is a member of (see, e.g., Ward, 2006; Bernauer et al., 2010; Spilker, 2012). The
data for this count variable are taken from Pevehouse et al. (2004).
Moreover, international environmental treaties oblige member countries more directly than
IGO Membership to cooperate on environmental problems such as air or water pollution,
climate change, trade in toxic waste, and endangered species (Ward, 2006). We employ the
degree of participation in international environmental treaties, i.e., the cumulative number of
ratified environmental treaties in a given year, in order to capture this mechanism. We use the
data from Spilker and Koubi (2014).
Table 1 summarizes the descriptive statistics of the variables included in our analysis. We
also present the variation inflation factors (VIFs) for each explanatory variable to see whether
multicollinearity constitutes a major problem in our setup. As demonstrated in the table,
however, all independent variables have VIFs well below the threshold level of 5 – the only
exception being, not surprisingly, GDP per capita (ln) and its square term due to the
construction of these variables. Furthermore, there seems to be not much overlap between,
e.g., IGO Membership and Env. Treaties Rat.
__________
Table 1 in here
__________
Empirical Results
Table 2 presents the results for our baseline models that focus on the impact of FDI flows and
trade on environmental quality either with or without the control variables. These
specifications seem important in light of Clarke (2005; 2009) who argues that control
variables may actually increase the bias in our estimates instead of decreasing it. Hence, while
18
Model 1 focuses on our core variables of interest only (although country fixed effects, year
dummies, and the lagged dependent variable are included), Model 2 also incorporates the
control variables we discussed above. Finally, Model 3 drops the GDP related variables (GDP
(ln) and GDP Growth in %) as well as the variable capturing the ratification of environmental
treaties as these might be overlapping with or are in fact very much related to the items that
are standardized by GDP (e.g., FDI Inflows / GDP (ln)) or are likely to be included in IGO
Membership, respectively. Either way, while the model-fit statistics demonstrate that the
control variables lead to an improvement of the model, there is not much difference between
Model 2 and Model 3.
_________
Table 2 in here
__________
Coming to our core variables of interest in Table 2, we see that our results are generally very
robust across specifications. More specifically, Trade Openness (ln) is positively signed
throughout Models 1-3 and statistically significant at conventional levels. Substantially, Table
2 shows that a 10% increase in Trade Openness (ln) induces on average a 0.4% rise in SO2
emissions per capita. Hence, our findings actually suggest that the stronger a country is
involved in the global trade, the more it will pollute. This means that we obtain support for
both the FET and PHH and our first hypothesis, which implies that higher economic activity
(scale effect) as captured by imports and exports comes with worse environmental quality – in
this unconditional model and only when focusing on Trade Openness (ln).
Contrary to the trade item, FDI Inflows / GDP (ln) is in fact negatively signed and
statistically significant at least at the 5% level regardless of model specifications. Therefore,
we obtain support for the pollution hallo argument, which states that environmental quality
19
improves with more inflows of FDI into a country14 (e.g., Aklin 2015; Elliott and Shimamoto
2008; Gallagher and Zarsky 2007; Prakash and Potoski 2006a; 2007; Eskeland and Harrison
2003). Substantively, when increasing FDI Inflows / GDP (ln) by 10%, SO2 emissions per
capita go down by about 5% on average. Somewhat unexpectedly, however, FDI Outflows /
GDP (ln) shows inconclusive results in Table 2: the variable’s coefficient is either
insignificant (and also changes signs) in Models 1-2 or positively signed and significant in
Model 3, which actually suggests that a state’s environmental quality as measured by SO2
emissions actually worsens with more outflows of FDI.
That said, the estimations presented in Table 2 rely on “unconditional” specifications and,
thus, do not allow for a direct test of our second and third hypotheses. Specifically, we need to
estimate the effect of FDI inflows conditional on a country’s income level and, in a second
step, conditional on its level of trade activity. Hence, we calculated multiplicative terms along
these lines and estimated “conditional” or interactive models, which rely on the specifications
used for Model 3.15 Table 3 summarizes our findings here, while we present the marginal
effects of the interactive relationships in Figures 3 and 4.
14 Since this variable captures total FDI inflows, i.e., FDI inflows to both developed and developing
countries, it seems that multinational companies (MNCs) not only export greener technologies, but
also conduct business in an environmentally friendly manner in host countries. The main reasons for
doing so might be that (1) MNCs use the same type of (advanced) technologies across countries to
minimize production costs, (2) MNCs possess newer and less polluting production methods as
compared to local firms, (3) MNCs use less polluting technologies in order to avoid criticism from
consumers in their home countries, and (4) host countries might require the same as local and/or
stricter environmental standards for foreign companies. 15 While Model 3 does not include all controls that are considered for Model 2, the model fit of the
former is better than for the latter. Hence, due to parsimony, we focus on the specification of Model 3
in the following, although our interaction results remain unchanged when using Model 2’s
specification.
20
_________
Table 3 and Figure 3 in here
__________
As demonstrated in Figure 3, FDI inflows into the developed countries do not matter for the
environment. However, FDI inflows play a major role for (highly) underdeveloped countries:
the marginal effect of FDI Inflows / GDP (ln) becomes statistically insignificant at a logged
GDP per capita level of about 6.5. As soon as we cross this threshold level, FDI Inflows /
GDP (ln) has a statistically insignificant effect. Before that threshold, however, FDI Inflows /
GDP (ln) is actually positively signed. Substantively, a 10% increase in FDI Inflows / GDP
(ln) leads to 1.5% increase in SO2 emissions per capita in the most underdeveloped countries
(i.e., GDP per capita (ln) level of 4; according to our data, Bhutan had such an income level
between 1971 and 1978). Overall, these findings are consistent with the PHH and the FET.
Specifically, according to the FET, FDI inflows into developed countries must go into non-
polluting activities, because the rate of return in those activities is higher than in the rest of the
world. The PHH has the same implication: clean capital will flow into those states that have
higher environmental standards, while dirty capital will go into the countries with the less
stringent environmental regulation. More generally, however, we not only find support for our
second hypothesis, but also show that the effect of FDI Inflows / GDP (ln) can change when
taking a conditional relationship into account. We do so in Models 4-5, but have estimated
unconditional models in Table 2 above.
__________
Figure 4 in here
__________
Turning to the third hypothesis and, hence, our second interactive effect, the combined
influence of trade openness and FDI inflows is also positive (Figure 4) – and thus “reverses”
the unconditional findings we obtained in Table 2. In more detail, FDI Inflows / GDP (ln) is
21
statistically insignificant for trade-openness levels of about 1.0 (logged) or less. For higher
values of Trade Openness (ln), however, FDI Inflows / GDP (ln) exerts a strongly positive
and significant impact on SO2 emissions: at the maximum of 4.0 of Trade Openness (ln) in
Figure 5, a 10% increase in FDI Inflows / GDP (ln) leads to 1.9% increase in SO2 emissions.
Most importantly, it is this finding that has discriminating power across the two theories of
the PHH and FET the positive impact of the interactive term favors the PHH, while a negative
one would have supported the FET.
Finally, coming to our control variables, we focus on the statistically significant items only
due to space limitations. First, the lagged dependent variable is, not surprisingly, positively
signed and highly significant. A higher level of SO2 emissions in the previous year leads to
more emissions in the current year. On average and across Models 1-5, our calculations
suggest that a 1% increase in the lagged dependent variable increases SO2 Emissions per
capita (ln) by about 0.8%. Second, we only find weak (at best) support for the environmental
Kuznets effect. Regardless of model specifications, the estimated coefficient of GDP per
capita on pollution is positive, while its square term is negatively signed. Note, however, that
the square term reaches conventional levels of statistical significance only in Model 5. Hence,
although the signs of the GDP per capita variables are consistent with the expectations of the
Kuznets curve literature (e.g., Grossman and Kruger 1995; Selden and Song 1994), the (in-)
significance levels of the square term actually point to a linear effect of income on SO2
Emissions per capita (ln). Third, the only remaining control item that is statistically
significant is population density. The variable’s coefficient estimate is consistent both in
terms of substance and significance across Tables 2 and 3. On average, a 10 % rise in
Population Density (ln) induces an increase of about 5.3% in SO2 Emissions per capita (ln).
All other control items have statistically insignificant coefficient estimates. For example,
participation in international organizations, either in terms of participating in IGOs more
generally or ratifying international environmental agreements, does not make it more or less
22
likely that a country will have a higher environmental quality. The poor performance of the
control variables in Tables 2 and 3 might be explained by the fact that fixed effects models
lack the ability to make inferences about time-invariant or slow-moving variables, because
those covariates are highly collinear with fixed effects and their coefficients are either not
identified or difficult to estimate with precision (Plümper and Troeger 2007). Thus, fixed
effects soak up most of the explanatory power of slowly changing variables (Beck 2001: 285).
We also changed a series of model specifications in order to examine the robustness of our
findings. We briefly describe these additional checks here, but omit tables or graphs as these
can be replicated with our replication material. First, and as demonstrated above, including or
excluding control variables does not affect the substance of our findings (Clarke 2005; 2009).
Second, we also estimated models that replaced fixed effects with random effects, substituted
the year dummies by a time (yearly) trend variable, or relied on a Prais-Winsten specification
with panel-corrected standard errors. Again, none of these modifications in our quantitative
setup altered our core findings. Third, SO2 emissions are just one type of pollutant that is
frequently used as a proxy for environmental quality. Other work also focuses on CO2
emissions as another, major determinant of climate change and, hence, environmental quality.
We thus re-estimated all models with the variable CO2 Emissions per capita (ln). While FDI
Inflows / GDP (ln) becomes insignificant in some models that are based on Table 2, the
substantive interpretation of our interactive relationships remains unchanged. Hence, our
findings are robust across two related, but different pollutants.
4. Conclusion
In this paper, we have developed a theoretical framework that allows us to discriminate
empirically between two competing theories, namely the FET and the PHH. This is due to two
innovations in our theoretical argument relative to the existing literature. The first is the joint
study of international trade and FDI. And the second is the inclusion of both PHH and FET
23
effects. The key discriminating implication is that the interaction between trade with inward
FDI is expected to be positive under the PHH, but negative under the FET. Our empirical
findings from a large set of countries indicate that the former is true.
Furthermore, we believe that the discrimination between the two theories is of particular
interest to policymakers and the scholarly community. While the FET and the PHH do not
have different implications about which type of countries (developed vs. less developed) will
become “dirtier” with higher levels of trade, both theories do have different implications
about the effects of the harmonization of environmental standards on pollution levels, though.
In particular, under the PHH, harmonization will eliminate the “comparative” advantage in
pollution enjoyed by the less-developed countries and, thus, lead to lower pollution there.
Under the FET, on the other hand, the harmonization of environmental standards will have no
such effects, as it is the developed countries (which already have high standards) that become
dirtier with globalization.
Consequently, a key policy implication of our analysis is that the globalization of
environmental standards (international harmonization) has the potential to reverse the increase
in pollution in less developed countries that derives from the globalization of trade and FDI.
24
References
Acharyya, Joysri. 2009. FDI, Growth, and the Environment: Evidence from India on CO2
Emissions during the Last Two Decades. Journal of Economic Development 34(1): 43-57.
Aklin, Michaël. 2015. Re-Exploring the Trade and Environment Nexus through the Diffusion
of Pollution. Environmental and Resource Economics: Forthcoming.
Ambec, Stefan and Philippe Barla. 2006. Can Environmental Regulations be Good for
Business? An Assessment of the Porter Hypothesis. Energy Studies Review 14(2): 42-62.
Antweiler, Werner, Brian Copeland and M. Scott Taylor. 2001. Is Free Trade Good for the
Environment? American Economic Review 91(4): 877-908.
Beck, Nathaniel. 2001. Time-Series-Cross-Section Data: What Have We Learned in the Past
Few Years? Annual Review of Political Science 4(1): 271-293.
Ben Kheder, Sonia and Natalia Zugravu. 2012. Environmental Regulation and French Firms
Location Abroad: An Economic Geography Model in an International Comparative Study.
Ecological Economics 77(1): 48-61.
Bernauer, Thomas, Anna Kalbhenn, Vally Koubi and Gabriele Spilker. 2010. A Comparison
of International and Domestic Forces of Global Governance Dynamics. British Journal of
Political Science 40(2): 509-538.
Bernauer, Thomas and Vally Koubi. 2009. Effects of Political Institutions on Air Quality.
Ecological Economics 68(5): 1355-1365.
Böhmelt, Tobias and Jürg Vollenweider. 2015. Information Flows and Social Capital through
Linkages: The Effectiveness of the CLRTAP Network. International Environmental
Agreements: Forthcoming.
Brambor, Thomas, William Clark and Matt Golder. 2006. Understanding Interaction Models:
Improving Empirical Analysis. Political Analysis 14(1) 63-82.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson and James D. Morrow.
2003. The Logic of Political Survival. Cambridge, MA: MIT Press.
25
Clarke, Kevin A. 2005. The Phantom Menace: Omitted Variable Bias in Econometric
Research. Conflict Management and Peace Science 22(4): 341-352.
Clarke, Kevin A. 2009. Return of the Phantom Menace: Omitted Variable Bias in
Econometric Research. Conflict Management and Peace Science 26(1): 46-66.
Cole, Mathew A. and Per G. Fredriksson. 2009. Institutionalized Pollution Havens.
Ecological Economics 68(4): 1239-1256.
Cole, Matthew A. and Robert J.R. Elliott. 2005. FDI and the Capital Intensity of “Dirty”
Sectors: A Missing Piece of the Pollution Haven Puzzle. Review of Development
Economics 9(4): 530-548.
Cole, Matthew A. and Robert J.R. Elliott. 2003. Do Environmental Regulations Influence
Trade Patterns? Testing Old and New Trade Theories. The World Economy 26(8): 1163-
1186.
Cole, Matthew. A., A.J. Rayner and J.M. Bates. 1997. The Environmental Kuznets Curve: An
Empirical Analysis. Environment and Development Economics 2(4): 401-16.
Congleton, Roger. 1992. Political Institutions and Pollution Control. Review of Economics
and Statistics 74(3): 412-421.
Copeland, Brian and M. Scott Taylor. 2004. Trade, Growth, and the Environment. Journal of
Economic Literature 42(1): 7-71.
Damania, Richard, Per Fredriksson and John List. 2004. The Multiplier Effect of
Globalization. Economic Letters 83(3): 285-292.
Dasgupta, Susmita, Benoit Laplante, Hua Wang and David Wheeler. 2002. Confronting the
Environmental Kuznets Curve. Journal of Economic Perspectives 16(1): 147-168.
Dean, Judith M, Mary E. Lovely and Hua Wang. 2009. Are Foreign Investors Attracted to
Weak Environmental Regulations? Evaluating the Evidence from China. Journal of
Development Economics 90(1): 1-13.
26
Dreher, Axel. 2006. The Influence of Globalization on Taxes and Social Policy: An Empirical
Analysis for OECD countries. European Journal of Political Economy 22(1): 179-201.
Dreher, Axel, Jan-Egbert Sturm and Heinrich Ursprung. 2008. The Impact of Globalization
on the Composition of Government Expenditures: Evidence from Panel Data. Public
Choice 134(3-4): 263-292.
Drezner, Daniel. 2001. Globalization and Policy Convergence. International Studies Review
3(1): 53-78.
Elliott, Robert J.R. and Kenichi Shimamoto. 2008. Are ASEAN Countries Havens for
Japanese Pollution-intensive Industry? World Economy 31(2): 236-254.
Eskeland, Gunnar and Ann Harrison. 2003. Moving to Greener Pastures? Multinationals and
the Pollution Haven Hypothesis. Journal of Development Economics 70(1): 1-23.
Esty, Daniel C. and Damien Geradin. 1998. Environmental Protection and International
Competitiveness. A Conceptual Framework. Journal of World Trade 32(3): 5-46.
Frankel, Jeffrey and Andrew Rose. 2005. Is Trade Good for the Environment? Sorting out the
Causality. Review of Economics and Statistics 87(1): 85-91.
Frankel, Jeffrey. 2005. Climate and Trade. Environment 47(7): 8-19
Fredriksson, Per G., Erik Neumayer, Richard Damania and Scott Gates. 2005.
Environmentalism, Democracy, and Pollution. Journal of Environmental Economics 49(2):
343-365.
Gallagher, Kevin and Lyuba Zarsky. 2007. The Enclave Economy. Cambridge, MA: MIT
Press.
Gleditsch, Kristian Skrede. 2002. Expanded Trade and GDP Data. Journal of Conflict
Resolution 46(5): 712-724.
Grether, Jean-Marie, Nicole A. Mathys and Jaime de Melo. 2010. Global Manufacturing SO2
Emissions : Does Trade Matter? Review of World Economics 145(4): 713-729.
27
Grossman, Gene M. and Alan B. Krueger. 1995. Economic Growth and the Environment.
Quarterly Journal of Economics 110(2): 353-377.
Jaffe, Adam B., Steven R. Peterson, Paul R. Portney and Robert N. Stavins. 1995.
Environmental Regulation and the Competitiveness of U.S. Manufactoring: What does the
Evidence Tell Us? Journal of Economic Literature 33(1): 132-163.
He, Jie. 2006. Pollution Haven Hypothesis and Environmental Impacts of Foreign Direct
Investment: The Case of Industrial Emission of Sulfur Dioxide (SO2) in Chinese
Provinces. Ecological Economics 60(1): 228-245.
Javorsik, Beata Smarzynska and Wei Shang-Jin. 2004. Pollution Havens and Foreign Direct
Investment: Dirty Secret or Popular Myth? Contributions to Economic Analysis and Policy
3(2): 1-32.
Jorgenson, Andrew. 2009. Foreign Direct Investment and the Environment, the Mitigating
Influence of Institutional and Civil Society Factors, and Relationships Between Industrial
Pollution and Human Health: A Panel Study of Less-Developed Countries. Organization &
Environment 22(2): 135-157.
Kellenberg Derek K. 2009. An Empirical Investigation of the Pollution Haven Effect with
Strategic Environment and Trade Policy. Journal of International Economics 78(2): 242-
255.
Keller, Wolfgang and Arik Levinson. 2002. Pollution Abatement Costs and Foreign Direct
Investment to U.S. States. Review of Economics and Statistics 84(4) 691-703.
Kim, Myeong H. and Nodir Adilov. 2012 The Lesser of Two Evils: An Empirical
Investigation of Foreign Direct Investment Pollution Tradeoff. Applied Economics 44(20):
2597-2606.
Lan, Jing, Makoto Kakinaka and Xianguo Haung. 2012. Foreign Direct Investment, Human
Capital and Environmental Pollution in China. Environmental and Resource Economics
51(2): 255-275.
28
Lane, Philip R. and Gian Maria Milesi-Ferretti. 2001. The External Wealth of Nations:
Measures of Foreign Assets and Liabilities in Industrial and Developing Countries. Journal
of International Economics 55(2): 263-294.
Lane, Philip R. and Gian Maria Milesi-Ferretti. 2007. The External Wealth of Nations Mark
II: Revised and Extended Estimates of Foreign Assets and Liabilities, 1970-2004. Journal
of International Economics 73(2): 223-250.
Li, Quan and Rafael Reuveny. 2006. Democracy and Environmental Degradation.
International Studies Quarterly 50(4): 935-956.
Mani, Muthukumara and David Wheeler. 1998. In Search of Pollution Havens? Dirty Industry
in the World Economy, 1960 to 1995. Journal of Environment and Development 7(3): 215-
247.
Manderson, Edward and Richard Kneller. 2012 Environmental Regulations, Outward FDI,
and Heterogeneous Firms: Are Countries Used as Pollution Havens? Environmental and
Resource Economics 51(3): 317-352.
Marshall, Monty G. and Keith Jaggers. 2002. Polity IV Dataset. College Park, MD: Center for
International Development and Conflict Management. University of Maryland.
Millimet, Daniel L., John A. List and Thanasis Stengos. 2003. The Environmental Kuznets
Curve: Real Progress or Misspecified Models? Review of Economics and Statistics 85(4):
1038-1047.
Millimet, Daniel L. and John A. List. 2004. The Case of the Missing Pollution Haven
Hypothesis. Journal of Regulatory Economics 26(3) 239-262.
Perkins, Richard and Eric Neumayer. 2009. Transnational Linkages and the Spillover of
Environment-Efficiency into Developing Countries. Global Environmental Change-
Human and Policy Dimensions 19(3): 375-383.
29
Porter, Michael E. and Claas van der Linde. 1995. Toward a New Conception of the
Environment-Competitiveness Relationship. Journal of Economic Perspective 9(4): 97-
118.
Potoski, Matthew. 2001. Clean Air Federalism: Do States Race to the Bottom? Public
Administration Review 61(3): 335-42.
Prakash, Aseem and Matthew Potoski. 2007. Investing Up: FDI and the Cross-Country
Diffusion of ISO 14001 Management Systems. International Studies Quarterly 51(3): 723-
744.
Prakash, Aseem and Matthew Potoski. 2006a. The Voluntary Environmentalists: Green
Clubs, ISO 14001, and Voluntary Environmental Regulations. Cambridge: Cambridge
University Press.
Prakash, Aseem and Mathew Potoski. 2006b. Racing to the Bottom? Trade, Environmental
Governance, and ISO 14001. American Journal of Political Science 50(2): 350-364.
Selden, Thomas and Daqing Song. 1994. Environmental Quality and Development: Is there a
Kuznets Curve for Air Pollution Emissions? Journal of Environmental Economics and
Management 27(2): 147-162.
Shafik, Nemat. 1994. Economic Development and Environmental Quality: An Econometric
Analysis. Oxford Economic Papers 46(1): 757-777.
Shafik, Nemat and Sushenjit Bandyopadhyay. 1997. Economic Growth and Environmental
Resources. Journal of Environmental Economics and Management 4(1): 1-24.
Sheldon, Ian. 2006. Trade and Environmental Policy: A Race to the Bottom? Journal of
Agricultural Economics 57(3): 365-392.
Spilker, Gabriele. 2012. Helpful Organizations: Membership in Inter-Governmental
Organizations and Environmental Quality in Developing Countries. British Journal of
Political Science 42(2): 345-370.
30
Spilker, Gabriele. 2013. Globalization, Political Institution and the Environment in
Developing Countries. New York: Routledge.
Spilker, Gabriele and Vally Koubi. 2014. The Effects of Treaty Legality and Domestic
Institutional Hurdles on Environmental Treaty Ratification. International Environmental
Agreements, DOI 10.1007/s10784-014-9255-4.
Stern, David I. 2004. The Rise and Fall of the Environmental Kuznets Curve. World
Development 32(8): 1419-1439.
Stern, David I. Michael S. Common, and Edward B. Barbier. 1996. Economic Growth and
Environmental Degradation: The Environmental Kuznets Curve and Sustainable
Development. World Development 24(7): 1151-1160.
Taylor, M. Scott. 2004. Unbundling the Pollution Haven Hypothesis. Advances in Economic
Analysis & Policy 4(2): 1-28.
Taylor, M. Scott and Brian A. Copeland. 1994. North-South Trade and the Environment.
Quarterly Journal of Economics 109(3): 755-787.
Vogel, David. 1997. Barriers or Benefits? Regulation in Transatlantic Trade. Washington,
D.C.: Brookings Institution Press.
Vogel, David. 1995. Trading Up: Consumer and Environmental Regulation in a Global
Economy. Boston, MA: Harvard University Press.
Xing, Yuqing and Charles Kolstad. 2002. Do Lax Environmental Regulations Attract Foreign
Investment? Environmental and Resource Economics 21(1): 1-22.
Wagner, Ulrich J. and Christopher D. Timmins. 2009. Agglomeration Effects in Foreign
Direct Investment and the Pollution Havens Hypothesis. Environmental and Resource
Economics 43(2): 231-256.
Ward, Hugh. 2006. International Linkages and Environmental Sustainability: The
Effectiveness of the Regime Network. Journal of Peace Research 43(2): 149-166.
31
32
33
34
35
36
37