1 s2.0 S0921800915301191 Main

Ecological Economics 131 (2017) 139151
Contents lists available at ScienceDirect
Ecological Economics
journal homepage: www.elsevier.com/locate/ecolecon
Analysis
Inequality, democracy, and the environment: A cross-national analysis

Prakash Kashwan
University of Connecticut, Storrs 365 Faireld Way, Storrs, CT 06269, USA
a r t i c l e
i n f o
Article history:
Received 12 July 2015
Received in revised form 8 August 2016
Accepted 16 August 2016
Available online xxxx
Keywords:
Inequality
Democracy
Environmental protection
Political choice
Social choice
Institutions
a b s t r a c t
This paper joins the debate on the relationship between inequality and the environment. Departing from the past
contributions, which focused either on the theories of environmental behavior or on economic interests, this
paper develops arguments about political choice mechanisms that help explain the linkages between inequality and national policymaking related to the establishment of protected areas. A cross-national analysis of the
interactions between inequality, democracy and the legal designation of protected areas in a global sample of
137 countries shows that, ceteris paribus, the effects of inequality vary depending on the strength of democracy:
in relatively democratic countries inequality is associated with less land in protected areas, whereas in relatively
undemocratic countries the reverse is true. The highly signicant effects of inequality undermine the democratic
dividend in the arena of nature conservation.
2016 Elsevier B.V. All rights reserved.
The inequality hypothesis posits that economic inequality hampers industrial regulation and leads to an oversupply of environmental
pollution. James Boyce, one of the most prominent scholars of environmental inequalities, noting that the rich often prefer private external
assets to public domestic assets, conjectures: Inequalities may fatten
foreign bank accounts, but they do not protect the environment at
home (Boyce, 2002, 43). More broadly, Boyce's thesis is that distribution of costs and benets of environmental intervention play an important role in determining environmental outcomes. As such, inequalities
can cut both ways. Under certain conditions, political and economic
elites may be able to impose a larger share of the costs of environmental
protection on the relatively poor and politically marginalized groups,
who often lack institutional representation (Torras and Boyce, 1998;
Clement and Meunie, 2010). In those cases, economic inequalities are
likely to contribute to some types of environmental protection.
This paper tests the inequality hypothesis in this broader sense to
examine how economic inequalities and political freedoms affect
cross-national variation in the percentage of national territory set
aside for nature conservation, via legal designation of terrestrial
protected areas (PAs). The choice of the outcome variable reects the
intent to draw on the available research on the subject and to respond
to methodological critiques of much of the scholarship on inequality
and environment (see, Gates et al., 2002; Berthe and Elie, 2015). The
empirical results and supplementary analyses presented in the paper
show that the effect of income inequality on designation of PAs is
contingent on the strength of democracy. Inequality has a positive effect
on designation of PAs in non-democratic countries, while the effect of
E-mail address: prakash.kashwan@uconn.edu.
http://dx.doi.org/10.1016/j.ecolecon.2016.08.018
0921-8009/ 2016 Elsevier B.V. All rights reserved.
democracy on designation of PAs is positive at low levels of inequality;

however, the democracy dividend for conservation diminishes with increasing economic inequality. Noticeably, the environmental Kuznets
curve is valid for cross-sectional analysis of PAs within the developed
countries subsample, though, as I explain below, none of the income
variables is signicant either in a full sample using developed country
interaction effects or in the subsample of developing countries.
The analytical strategy used in this paper proceeds in three steps.
First, the next section offers a concise review of the scholarship on inequality and environment and maps the mechanisms of inequalityenvironmental linkages to inform the hypotheses that this paper tests.
Section 2 discusses data and empirical methods, followed by a discussion
of the results of empirical analysis in Section 3. Section 4 supplements empirical analysis by drawing on two different strands of research related to
PAs: 1) extensive research on the political economy of PAs and 2) scholarship in biological sciences, which examine the extent of overlap between
the land under PAs and the areas rich in critically endangered biodiversity (gap-analysis). The main ndings of these research programs
help buttress the ndings of political economy of institutions analysis
presented below. The concluding section summarizes the contributions
that this paper makes to the debates on inequality-environment
linkages and outlines an agenda for further research on the topic.
1. Inequality, Democracy and the Environment
1.1. Inequality and the Environment: Taking Interests and Preferences
seriously
The contributors to this journal have been engaged in a productive
debate about the relationship between economic and political inequality
140
P. Kashwan / Ecological Economics 131 (2017) 139151
and environmental outcomes (Boyce, 1994; Scruggs, 1998; Torras and

Boyce, 1998; Roca, 2003; Berthe and Elie, 2015). Economic elites, who
often also command signicant political power, derive benets from
the polluting activity, while economically and politically weaker sections are forced to bear the costs of environmental degradation
(Boyce, 1994). Boyce argues that, because the rich stand to prot from
lower environmental regulations, they would not necessarily demand
a cleaner environment. An important counter-argument is that increasing income may be associated with either same level of preference for
degradation (if environment is considered a normal good) or a lower
level of preference for environmental degradation (when environment
is considered a superior good) (Scruggs, 1998). If the environment is
valued as a superior good, the rich are likely to prefer a clean environment and are willing to pay higher taxes or accept a greater amount
of environmental regulation, as implied in the research on environmental Kuznets curve.
The Kuznets curve argument has been criticized on a number of
grounds. Empirical studies show that income elasticity of environmental improvements is less than or very close to one, which undermines
the argument that environmental quality exhibits the characteristics
of a normal or superior good (Torras and Boyce, 1998; Martini and
Tiezzi, 2014). Even if increased incomes were associated with stronger
preferences for clean environment, such preferences may not lead to a
clean environment for everyone. While the wealthy may be willing to
pay for the cost of cleanup, as the literature on environmental justice
shows, waste disposal and landlls facilities tend to be sited disproportionately around neighborhoods inhabited by the predominantly
poor groups of people of color and racial minorities (Taylor, 2014).
Environmental protections may be accompanied by displacement of
the costs of clean-up and conicts related to competing resource use
and management strategies (Roca, 2003). Lastly, because environmental protection is a public good that individuals cannot buy in the market,
it is not proper to analyze environmental outcomes as an aggregation of
individual choices (Roca, 2003, 7; Wilkinson and Pickett, 2010).
Effective environmental protection requires public participation,
good government, effective regulation, and diffusion of technological
change each of which is often related to higher incomes (Magnani,
2000; Bimonte, 2002; Carson, 2010). Even so, it is likely, but not
inevitable, that a society will choose to reduce pollution levels as it
becomes wealthier (Carson, 2010, 3), as the present logjam in the
U.S. environmental policies suggests. The coincidence of increased
wealth and public action for environmental protection cannot be
taken for granted; neither can the institutional means through which
popular preferences translate into policy interventions. Public institutions rarely function autonomously as implied in formal analyses of institutions. Even in advanced democracies, the differences of political
and economic power within a society shape the emergence and functioning of political institutions, including those that are directly linked
to environmental policymaking (see, Magnani, 2000; Moe, 2005;
Steinberg, 2015). This may be one reason why empirical studies examining the effect of decision-making rules, such as Single Majority Rule
(SMR) or Power Weighted Rule (PWR), on environmental outcomes
produce mixed results (Berthe and Elie, 2015). Collective action
problems act as an important barrier against popular mobilization,
which reinforces the elite dominance of environmental policymaking.
The complex entanglement of collective action problems, enduring
power asymmetries, and the public good nature of environmental outcomes present theoretical and policy challenges against the efforts to
examine the effects of inequality on environment. Theoretical difculties discussed above are closely related to methodological problems.
Scott Gates and co-authors argue that the empirical literature on
democracy and environment focuses too much on environmental
outcomes instead of looking at environmental commitment (Gates
et al., 2002, 11). Somewhat counterintuitively, the focus on outcomes
is problematic because environmental outcomes are a product of complex interactions between social, economic, political, and environmental
forces, which most empirical studies rarely control for. Such omittedvariable biases are likely to be especially relevant in studies pertaining
to non-point sources of environmental outcomes, such as terrestrial
conservation. To address the methodological problems discussed
above, Gates et al. suggest that, instead of environmental outcomes,
scholars should examine environmental commitments which are directly linked to the policy process. Berthe and Elie (2015, 195) also articulate a related concern in their review of the inequality-environment
literature: because they mask the intermediary stages between inequality and environmental pressures, these tests are unable to validate
any particular theoretical explanation. Responding to these arguments,
this paper formalizes the notion of intermediary stages by introducing
the distinction between policy outputs and policy outcomes, commonly
employed in policy studies. Policy outputs are plans, projects, and
other tangible items that result from environmental policy process,
while policy outcomes are the effects of outputs on environmental
and social conditions (Koontz and Thomas, 2006, 113). Political scientist David Easton denes outputs as a stream of activities owing
from the authorities in a system (Wahlke, 1971, 282). Legal designation of PAs indicates a concrete policy commitment and constitutes an
important policy output, which is valued in international environmental
policy arena and contributes to a variety of local outcomes, as the next
sub-section explains.
This paper's focus on policymaking and institutional development in
the presence of inequalities provides a theoretically grounded alternative mechanism to supplement previous research anchored in theories
of economic behavior and environment preferences. The emphasis on
policies and institutions also speaks to James Buchanan's argument
that studies of social choice mechanisms cannot rely on assumptions
that are often made about individualistically oriented decisionmaking processes of markets (Buchanan, 1954, 118; Ostrom,
2011). The next subsection develops a framework related to political
decision-making at the national level, which is a response to Berthe
and Elie's (2015, 195) recommendation about the development of
political choice mechanisms relevant to environmental policy
area under examination. This paper focuses on how interests of
policymakers align vis--vis the legal designation of PAs, as well as
how the policy outputs may affect the material interests of forestdependent people.
Instead of relying purely on a deductive logic, the framework
outlined in the next section mirrors the methods of abduction
used by a number of contributors to this journal (for a review, see,
Forstater, 2004). The starting point in abduction is to identify a puzzling
empirical outcome, which does not conform to well-established theories. The next subsection begins by outlining a puzzling outcome related
to the designation of PAs the world over, which is difcult to explain on
the basis of available theories of environmental public goods. To explain
this puzzling anomaly, I propose and test a set of hypotheses based on
theories of political economy of institutions.
1.2. Political-Economy of Protected Area Designation
Designation of PAs has been a major focus of international conservation groups that have helped enact international agreements, such as
the United Nations' Convention on Biological Diversity, UNESCO's
World Network of Biosphere Reserves, and the World Congress on
National Parks and Protected Areas. The fourth Congress held in 1992
resolved to bring 10% of the planet's landmass under PAs. While it was
an ambitious goal at the time, approximately 210,000 PAs covered
15.4% of terrestrial areas as of 2014 (Juffe-Bignoli et al., 2014; see,
Fig. 1). Such an accelerated growth of PAs cuts across the developingdeveloped countries divide that tends to characterize most other
environmental policies. Moreover, unlike other environmental commitments on which governments often renege, no country has witnessed
a net reduction in the landmass brought under the legal designation
of PAs since 1990. It is evident that, quite contrary to theoretical
141
Fig. 1. Distribution of land under protected areas, year 2014.
expectation, the designation of PAs constitutes a case of provision in

abundance, if not the overprovision of global public goods. How should
this be explained?
Provision of public goods is relatively easy if the costs of provisions
are broadly diffused, such as the taxes that go into funding national
defense the benets of which are available without discrimination.
The provision of terrestrial PAs instead leads to concentration of costs,
often passed on to local populations that tend to be poor and politically
marginalized, while most important benets accrue to national
and global stakeholders. Equally important, regions most in need of
protection, such as the global biodiversity hotspots, are often social
and political hotbeds with longstanding forest and land rights conicts,
high levels of poverty, enduring power asymmetries, and histories of
state control and repression (Brechin et al., 2002, 42). Accordingly,
some scholars of political economy attribute the rapid expansion of
PAs to green imperialism led by conservation groups and scientists
doing God's work[with] a divine mission to save the Earth (Chapin,
2004, 21; see also, McNeely and Schutyser, 2003). While international
lobbying and nances are crucially important for the promotion of
PAs, it is inconceivable that national policymakers would follow such
leads, if it did not serve their political and economic interests. Often,
multiple layers of nonlocal political forces, including a variety of
commercial enterprises that have strong connections to the state or
political elites, exert signicant inuence over conservation policies
and programs (Brechin et al., 2002, 51).
In light of these theoretical insights and empirical ndings, this
paper examines whether elite dominance of PA-related policymaking
is systematically linked to levels of domestic inequalities. Accordingly,
the framework proposed here emphasizes the role of policymaker
within the context of the most commonly cited features of this specic
policymaking system.
Even though the main focus of this paper is on the inequality in access to policymaking, micro-level wealth inequalities are not irrelevant
to the question of PAs. The decreasing marginal utility of consumptiondriven life satisfaction bolsters demand for idyllic and pristine
surroundings, which motivates the elite and urban middle classes to
ask for establishment of nature reserves to satisfy the soul's thirst of
the town dweller (Guha, 2006, 7). Domestic inequalities that fuel
elite tendencies to undertake exotic vacations and wilderness safaris
as a prestige good (Wilkinson and Pickett, 2010), also help channel
elite support that international conservation groups need to promote
new wildlife preserves. Domestic policymakers act at the intersection
of these international and national processes, which foster a type of
elite environmentalism that often results in denial of local claims related
to subsistence uses of the park resources (Brechin et al., 2002;
Sanderson, 2005). This will be especially true of countries with a strong
rule of law framework in which legal designation of PAs is likely to impose effective restrictions on local resource use (Nesbitt and Weiner,
2001). On the other hand, in countries without a robust rule of law
framework, legal designation of protected area creates opportunities
for rent-seeking for local protected area ofcials (Brockington, 2002;
Kashwan, 2016). In either situation, PAs are likely to be unpopular
among populations in and around the host sites in the presence of
signicant inequalities.
The extent to which policymaking elites will successfully translate
their ideas and interests into concrete policies and programs will be
contingent on the level of domestic inequality. This leads to the rst
hypothesis related to the effect of inequality on designation of PAs:
Hypothesis 1. Higher level of income inequality is associated, ceteris
paribus, with a larger percentage of national territory set aside as nature
PAs.
To motivate the next set of hypotheses, Fig. 2 below illustrates the
effects of economic inequality and democracy on the designation of
PAs, as predicted by the theories of the provisions of public good. High
levels of inequalities in weak democracies (the bottom left cell in
Fig. 2) is likely to facilitate policy elites setting aside large areas of land
as PAs without risking lost elections or legal actions. Conversely, plans
to set aside large areas for nature protection are likely to provoke public
resentment under high levels of inequalities in strong democracies
(the bottom right cell). In these cases, popular resentment is likely to
translate into electoral consequences or formidable legal challenges,
which is likely to deter policymakers from setting aside large areas of
land as PAs. In some cases, though not always, stronger democracies
142
Democracy
Inequality
Weak
Strong
Low
Moderate Percentage of
Land Protected
High
Large Percentage of
Land Protected
Small Percentage of Land

Protected
Fig. 2. Effect of inequality on designation of PAs, contingent on democracy.
may also afford the poor better institutional representation in comparison to the relatively weaker democracies (Clment and Meuni
2010). Effective institutions in strong democracies with low levels of
inequalities (the upper right cell in Fig. 2) may ensure an optimum
provision of public goods (Boyce, 2002), which is likely to translate in
this case to a moderate percentage of land set aside as protected areas.
It is relatively difcult to predict the nature of outcomes pertaining
to inequality-democracy congurations in the upper left cell, which
correspondents to low inequality in weak democracies. These hypothesized interaction effects are depicted in a stylized way in Fig. 2 below,
with a ? signifying an uncertain effect.
Hypothesis 2. Ceteris paribus high levels of inequality in conjunction
with poor democratic institutions are likely to lead to larger areas of
protected area, as compared to countries with strong democratic institutions with high levels of inequality, which are likely to be associated
with smaller areas of land designated as PAs.
While the main arguments in this paper center on the effects of
inequality that are contingent on the level of democracy, following the
method of multiple working hypotheses (Chamberlin, 1965), in this
paper I also test a rival claim that designation of protected areas is
meant to protect critically endangered biodiversity and wildlife. In
that case, the World Bank biodiversity index should be positively associated with larger area of national territory designated as protected
areas. Lastly, because the effects of inequality and democracy do not
exist independent of demographic and economic factors, the empirical
analysis below also accounts for relevant socioeconomic variables as
discussed below.
2. Democracy, Inequality, and Designation of PAs: Data and
Empirical Methods
The main focus of the paper is to examine the effects of democracy
and inequality on variation in the percentage of national territory that
137 countries covered in this analysis have set aside for territorial
nature conservation. The data related to PAs for the year 2012 is
drawn from the comprehensive dataset compiled by the World Commission on Protected Areas and is included in the 2015 World Development Indicators data of the World Bank. The analysis in this paper is
restricted to terrestrial PAs and excludes marine PAs, because of significant differences in the political economy of resource management in
these two sectors. First, in the context of marine resources, international
environmental groups often emphasize sustainable harvesting, marketing, and consumption of sheries, as opposed to wildlife tourism in
which case international institutions emphasize the maintenance of
stock within protected areas (Constance and Bonanno, 2000). Second,
the designation of marine protected areas entails biophysical and institutional uncertainties that are far greater than those commonly
witnessed in the context of terrestrial protected areas (Young, 1998).
The effects of inequality discussed here would be best captured in a
composite economic and political inequality index. However, because
no such index exists for cross-national measurements of political and

economic inequalities, this paper uses Gini coefcient for economic inequality as the main measure of inequality. The use of Gini coefcients
in cross-national analysis has been contentious because of the differences in how income and income inequality are measured in different
countries. However, the development of the Standardized Income
Inequality Data by researchers at the United Nations University World
Institute for Development Economics Research (UNU-WIDER) addresses many of these concerns, including those related to international
and inter-temporal inconsistencies (Babones and Alvarez-Rivadulla,
2007). Even so, Gini coefcient data does not exist for all the countries
for the duration of this study. As a result, this paper uses average
of the UNU-WIDER income inequality data available for the period
19952005.
Lastly, I re-run the main model by using three alternative specications for the measurement of inequality based on the World Bank data
for the years 20032005 to ensure that the rst set of results are not
driven by specic data for specic period. Because of sporadic availability of World Bank data on inequality, the data from the most recent year
in the period 20032005 was used to for each of the three alternative
specications listed here: 1) World Bank Gini coefcient; 2) the share
of income held by those in the top 10 income percentile, such that a
larger share held by richest ten percentile in population indicates
greater inequality; and 3) the difference between the shares of income
held by the top and bottom 10 income percentiles, such that larger
the difference then the larger is the level of economic inequality in a
society. For facilitating an intuitive understanding of the results, Gini
scores have been rescaled so that the minimum Gini score corresponds
to zero, as shown in summary statistics presented in Table 2.
Other relevant measures of inequality not used in the analysis here
merit some discussion. National literacy rates, which Torras and Boyce
(1998) use as a proxy for power equality, did not turn out to be a significant predictor of cross-national variation in PAs. Horizontal inequality
a measure of inequality between different ethnic groups (Stewart,
2010) may also be a relevant measure considering that a majority of
forest-dependent people tend to be indigenous minority groups.
However, it is uncertain how this measure relates to the inequalities
that affect policymaking at the national level. It is likely, as Daudelin
(2003, 10) argues in the context of land-related conicts, that policies
affecting access to land, security of tenure, and the distribution of
holdings produce differential impact on various groups because of
joint effects of both horizontal and class-based inequalities. Future studies may benet from the use of other indicators of political inequalities
or composite political and economic inequality indicators.
The strength of democracy is measured by the Freedom House (FH)
Index, which combines political rights (including electoral process, political pluralism and participation, and functioning of government)
and civil liberties (including four subcategories: freedom of expression
and belief, associational and organizational rights, rule of law, and
personal autonomy and individual rights).1 While the use of FH index
has been criticized (see Gates et al., 2002), FH's a la carte measurement
accounts for legal entitlements, as well as the effectiveness with which
citizens are able to exercise their rights in practice. This thick notion of
democracy (Munck and Verkuilen, 2002) also conforms to the theoretical framework of political economy of institutions, as discussed in the
previous section.
The total of FH scores for political rights and civil liberties range between 2 and 14 in which the higher scores represent less democratic
countries, while the lower scores represent more democratic countries.
Following Torras and Boyce (1998), this paper subtracts the FH index
score from 14, so that the rescaled democracy variables takes values between 0 and 12. The main analysis uses the average of democracy scores
over the periods 19952005. The choice of periods reects two major
1
For further details about the methodology, see https://freedomhouse.org/report/
freedom-world-2012/methodology, or any of the annual Freedom in the World reports.
143
Table 1
The key variables, denition, and the data source.
Variables
Description
Source
Protected area
Percentage of national territory set aside as protected areas (2012); Data sourced from the World
Development Indicators dataset.
Average Freedom House Democracy Index 19952005
Average Freedom House Democracy Index 20032005
Average Household Income Inequality over the years 19952005, computed by the author from Gini
for household income compiled as part of the Standardized Income Inequality Database.
Gini index measures the extent to which the distribution of income (or, in some cases, consumption
expenditure) among individuals or households within an economy deviates from a perfectly equal
distribution. Based on the most recent year's World Development Indicators data available for the
period 20032005.
Percentage share of income that accrues to subgroups of population in the top 10% of income. Based
on the most recent year's World Development Indicators data available for the period 20032005.
Difference between the percentage share of income that accrues to subgroups of population in the
top and bottom 10% of income. Based on the most recent year's World Development Indicators data
available for the period 20032005.
GDP per capita, PPP (current international $, 1000) in the year 2000/2005, from the World
Development Indicators dataset
Binary variable created based on membership of the Organisation for Economic Co-operation and
Development (OECD), exclusive of Mexico and Turkey.
GEF benets index for biodiversity is a composite index of relative biodiversity potential for each
country based on the species represented in each country, their threat status, and the diversity of
habitat types in each country. The index has been normalized so that values run from 0 (no
biodiversity potential) to 100 (maximum biodiversity potential).
Forest area is land under natural or planted stands of trees of at least 5 m in situ; excludes tree stands
in agricultural production systems (for example, in fruit plantations and agroforestry systems) and
trees in urban parks and gardens.
Population density (people per sq. km of land area) (1000) for the years 2000/2005 from the World
Percentage of rural population with access to electricity in the year 1990 from the World
Life expectancy at birth indicates the number of years a newborn infant would live if prevailing
patterns of mortality at the time of its birth were to stay the same throughout its life. Source: World
World Bank
Democracy 19952005
Democracy 2005
Income Inequality 19952005
Economic inequality (GINI index) 2005
Top 10% (income share) 2005

Income inequality (top 10% bottom 10%) 2005
GDP 2000/2005
Developed (yes = 1; no = 0)
GEF Biodiversity Index 2005
Forest area 2000
Population density 2000/2005

Rural electrication 2000
Life expectancy 1990
considerations. First, 19952005 allows for the minimum time-lag

necessary for transition to democracy for a good number of former
Soviet Union states. Second, because of sporadic availability of data on
democracy and income inequality, averaging the data on these variables
over a ten-period time span allows for sufcient sample size. Table 1
Freedom House
Freedom House
UNU-WIDER
World Bank
World Bank
World Bank
World Bank
OECD
Global Environment
Facility, World Bank
World Bank
World Bank
World Bank
World Bank
explains the measurements and data sources for each of the variables
used in the analysis, Table 2 presents the summary statistics, and
Table 3 the correlation matrix.
To test rival hypotheses about environmental Kuznets curve, which
has been the central focus of many of the important contributions to
Table 2
Summary statistics.
Count
Mean
Standard deviation
Minimum
Maximum
Percentage land protected 2012

Average FH Democracy Index 19952005
FH Democracy Index 2005
Average Income Inequality 19952005
Income Inequality 2005
Percentage population in top 10% 2005
Percentage in top 10% bottom 10% 2005
GDP per-capita 2000
GDP per-capita squared 2000
Biodiversity Index 2005
Area under forest 2000
Population density 2000
Urban population 2000
Dummy: developed countries
Regional dummy: Europe
Regional dummy: Africa
Regional dummy: Latin
Regional dummy: Asia
180
163
176
148
106
106
106
171
171
175
180
180
180
180
180
180
180
180
0.1548906
6.75575
7.414773
14.70666
22.98642
31.28575
28.81972
11,036.8
3.25E+08
8.287253
31.99099
236.5544
53.24816
0.1777778
0.2777778
0.2777778
0.1666667
0.2777778
0.1147664
3.900106
3.940066
9.409952
9.500381
7.405742
8.257028
14,305.75
9.37E+08
17.45236
23.50535
1274.28
23.64991
0.383392
0.4491526
0.4491526
0.3737175
0.4491526
0.0008391
0
0
3.93E07
6.10E07
17.41
11.31
410.976
168,901.2
0.0042915
0
1.54319
8.246
0
0
0
0
0
0.545086
12
12
45.62234
52.83
55.39
55.36
86,725.54
7.52E+09
100
98.66026
16,040.5
100
1
1
1
1
1
Sub-sample: developed countries

Percentage Land Protected 2012
32
32
32
0.234786
11.4654
6.265863
0.106073
0.609809
5.888303
0.062686
10.18182
3.93E07
0.545086
12
29.03901
Sub-sample: developing countries

Percentage Land Protected 2012
148
131
116
0.137616
5.605302
17.03515
0.109434
3.474588
8.862603
0.000839
0
1.312339
0.52973
12
45.62234
1
0.127
0.0408
1
0.199
Population
density 2000
Forest
2000
1
0.0506
0.141
0.111
0.421
1
0.903
0.0106
0.116
0.186
0.647
1
0.148
0.069
0.119
Biodiversity
2005
GDP pc_sq
2000
GDP pc
2000
Economic
inequality 2005
Table 4
Percentage land protected 2012.
(M1)
(M2)
1
0.391
0.333
0.224
0.259
0.135
0.215
0.123
0.251
GDP 2000
GDP _sq. 2000
0.0278
0.0916
0.175
0.068
0.355
0.147
0.175
0.0136
0.228
Developed (Yes = 1;
No = 0)
(0.0262)
0.162
(0.0384)
0.00937
(0.00967)
0.0532
(0.0142)
0.0444
(0.0152)
0.00565 0.00607
(0.00200)
(0.00253)
0.00331
0.0882
0.194
0.00202
0.385
0.0000168 0.00000211
(0.0000294) (0.0000241)
2.90e10 1.14e10
(2.10e10) (5.04e10)
0.327
(0.219)
0.211
(0.324)
0.000401
(0.000246)
0.000214
(0.000115)
Urban population
2000
0.00307
(0.00423)
0.00417
(0.00540)
Biodiversity index
2005
0.00173
(0.00263)
0.00379
(0.00272)
0.01000
(0.00265)
0.0000123
(0.0000204)
4.64e11 1.62e10
(4.66e10) (4.58e10)
0.101
(0.273)
0.000310
(0.000175)
0.000261
(0.000129)
0.0106
(0.00343)
EUROPE
0.0820
(0.186)
AFRICA
0.443
(0.215)
LATIN
0.483
(0.297)
Constant
p b 0.05.
p b 0.01.
p b 0.001.
0.0000305
(0.0000168)
Population density
2000
Area under forests

2000
(M4)
0.162
(0.0385)
Democracy #
income inequality
0.0996
0.0257
0.265
0.0871
0.183
0.175
0.177
Income inequality
0.154
0.18
1
0.998
0.415
0.351
0.216
0.246
1
0.982
0.991
0.383
0.331
0.236
0.280
1
0.864
0.869
0.867
0.403
0.351
1
0.198
0.167
0.200
1
0.962
0.187
0.200
0.231
0.207
0.220
1
0.375
0.296
0.0698
0.1
0.0943
0.104
0.205
(M3)
0.0609
Democracy
Protected Areas 2012

Democracy 19952005
Democracy 2005
Gini 19952005
Gini 2005_WB
Top 10pc 2005
Economic Inequality 2005
GDP pc 2000
GDP pc_sq 2000
Biodiversity 2005
Forest 2000
Population Density 2000
Urban Population 2000
Top 10pc
2005
Gini 2005_WB
Gini
19952005
Democracy
2005
Democracy
19952005
Protected areas
2012
Table 3
Correlation matrix.
the debate on inequality and environment, this study also analyzes the
effects of GDP per capita and its squared variant. Similarly, one cannot
dismiss the claim that policymakers make a good faith effort to protect
areas that host critically endangered wildlife and biodiversity. To test
this rival hypothesis, the empirical analysis includes, among the independent variables, the biodiversity index of the global environmental
facility (GEF) of the World Bank, which measures the conservation potential of countries (Models M1 & M2, Table 4). However, the correlation between GEF biodiversity index and the percentage of national
territory designated as PAs is a mere 0.0257, which is why it is excluded
from the main models analyzed in the next subsection. The models tested in this study also incorporate a number of control variables potentially related to the variation in PA designation, including the area under
forests, population density, urban population, and regional dummies for
Asia, Africa, Latin America, and Europe (which includes Europe and
North America).
To examine the possibility that the effects being studied in this
research may vary according to the level of economic development, I
created a dummy variable for developed countries (n = 32), which
include the OECD countries other than Mexico and Turkey; these
two countries were included in the group of developing countries
Urban population
2000
144
Observations
AIC
BIC
2.105
2.471
3.015
3.167
(0.227)
(0.287)
(0.305)
(0.197)
167
135
137
137
0.7046915
0.7671886
0.7389739
0.7653187
805.2436
608.3709
629.2541
619.8049
Standard errors in parentheses.

p b 0.10.
p b 0.05.
p b 0.01.

Table 5
Percentage land protected 2012 effects in the sub-samples of developed & developing
countries.
(M5)
Income inequality
Democracy # income inequality
GDP 2000
GDP _sq. 2000
Population density 2000
Developed (Yes = 1; No = 0)
Dummy: developed countries = 1
# average FH Democracy Index
19952005
# average Income Inequality
19952005
# average FH Democracy Index
19952005 # average income
inequality 19952005
Constant
Observations
AIC
BIC
(M6)
0.149
(0.0449)
0.0519
(0.0153)
0.00488
(0.00245)
0.00000584
(0.0000225)
1.51e10
(3.45e10)
0.000434
(0.000237)
7.363
Democracy
(M7)
0.198
(0.209)
0.0574
(0.171)
0.00132
(0.0162)
0.0000633
(0.0000267)
1.04e09
(3.39e 10)
0.000353
(0.000793)
0.131
(0.0418)
0.0498
(0.0149)
0.00435
(0.00232)
0.0000220
(0.0000443)
2.79e09
(2.22e09)
0.000861
(0.000371)
(2.939)
0.626
(0.267)
0.196
(0.247)
0.0140
(0.0223)
3.026
(0.304)
137
0.779179
614.9861
2.214
(2.111)
32a
1.119656
88.74791
2.876
(0.333)
105b
0.7099868
452.6217
a the developed countries' sub-sample; b the developing countries' sub-sample.

p b 0.10
p b 0.05
p b 0.01.
(n = 101). First, learning from Torras and Boyce (1998), the dummy
variable for developed countries is put into a three-way interaction
with variables for democracy and inequality respectively (Model M5,
Table 5). Second, following the suggestion of a reviewer for this journal,
Models M6 and M7 (Table 5) show the results for split-samples for each
of the two categories of the developed country dummy. These models
contribute important insights, which I discuss in the next section.
The dependent variable is not normally distributed (Fig. 3). Instead
of OLS, this paper employs generalized linear models, which extend
the classical linear model to describe the relationship between one or
more explanatory variables and a variety of non-normally distributed

outcome variables that include binary, count, and positive-valued
variates (Neuhaus and McCulloch, 2011). Several endogeneity concerns
have been tested for. First, because the dependent variable represents a
stock-sensitive output, such that the percentage of national territory
set aside as PAs records all the past historical processes (Bimonte,
2002), it is likely that regions that have had signicant areas of
land under PAs for quite some time may have had relatively fewer opportunities for economic development and, therefore, relatively higher
level of inequalities, which makes it possible that income inequality is
endogenous to the area under PAs. Second, the area brought under PA
designation may have inuenced the area of national territory under
forests and the existence of valuable biodiversity and wildlife, as measured by the GEF biodiversity index. Results of regression diagnostics
show that endogeneity is not a concern vis--vis inequality and
biodiversity, but it is a potential concern vis--vis the forest area variable, which is excluded from the main analyses as explained in the
next section.
A series of GLM models, presented in Tables 4, 5, and 6, collectively
offer a ne-grained understanding of the effects that the inclusion of
specic variables produces, and this demonstrates the robustness of
these models to different types of specications.
3. Results
Table 4 presents the results of empirical analysis. In this discussion, I
consider results with a statistical signicance at 1% (two tails combined)
to be highly signicant, those with 5% signicance to be signicant,
and those with 10% signicance to be marginally signicant. The
baseline model (M1) uses variables related to economic development,
demographics, and natural resources, which one would expect to be
causally related to the legal designation of PAs. In this model, the
coefcient for income is marginally signicant, and the coefcient for
Table 6
Percentage land protected 2012 alternative specications of income inequality.
(M8)
0.294
(0.0976)
0.266
(0.0857)
Income inequality 2005
0.0745
(0.0197)
0.0850
(0.0211)
0.0788
(0.0200)
0.00678
(0.00243)
0.00728
(0.00282)
0.00693
(0.00263)

# income inequality 2005
0.0000119
(0.0000252)
0.0000106
(0.0000248)
gdp_sq_2005
7.72e11
(3.98e10)
1.19e10
(4.15e10)
1.05e10
(4.09e10)
popdensity_2005
0.000660**
(0.000363)
0.000739
(0.000379)
0.000722
(0.000377)
0.00000823
(0.0000241)
Density
2
Constant
Observations
AIC
BIC
Measure of income
inequality 2005
.2
.3
.4
.5
Percentage of National Territory under PAs (2012)
Fig. 3. Distribution of percentage of land under protected areas (PAs).
(M10)
0.225
(0.0675)
Dummy: developed
countries
.1
(M9)
gdp_2005
145

p b 0.10.
p b 0.05.
p b 0.01.
0.0843
(0.321)
0.0906
(0.328)
0.0887
(0.326)
3.962
(0.547)
102
0.7824408
432.9899
World Bank
Gini
4.905
(0.731)
102
0.7837091
432.8605
Percentage
population
in Top 10%
4.507
(0.653)
102
0.7833339
432.8988
Percentage in
top 10%
bottom 10%
146
.005
0
-.005
-.01
Effects on Predicted Mean % of Land under PAs (2012)
.01
Average Marginal Effects of Income Inequality (95% CIs)
.5
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12
Average Democracy Scores 1995-2005
Fig. 4. Average marginal effects of income inequality on land under PAs, contingent on democracy.
the area under forests is highly signicant, while the coefcients for
population density and biodiversity index are not signicant. Model
M2 adds the key variables for democracy and income inequality, after
which economic growth becomes non-signicant, forest area remains
highly signicant as before, and population density is marginally significant. In this model, democracy is statistically signicant, but inequality
has no effect. Even though area under forest is highly signicant again, it
is excluded from the next set of models (M3M10) in Tables 4, 5, and 6,
because of endogeneity concerns discussed above. Exclusion of forest
area does not change the overall results in any fundamental way,
while it improves model efcacy as indicated by lower values for Akaike
information criterion (AIC) and Bayesian information criterion (BIC).
The main model (M3, Table 4) tests the key hypotheses about effects
of interaction between democracy and inequality on legal designation of
PAs. Democracy has a highly signicant effect and is positively related
to percentage of national territory set aside as PAs. Note that because of
the presence of an interaction effect, the stand alone democracy coefcient reports the effect of democracy under conditions of perfect equality
i.e. when inequality takes a value of zero. This result, read in isolation of
its interaction with inequality, is supportive of the institutional functions
of democracy as a means of mediating social choice over provision of
public goods. Similarly, the stand-alone coefcient for inequality, which
captures the effect under the condition when democracy takes a value
of zero, is also highly signicant and has a positive effect on the percentage of land set aside as PAs. As expected under Hypothesis 1, in the absence of democratic constraints, inequality allows policymakers to set
aside large areas of land as PAs. The coefcient for interaction term
(Democracy Inequality) is highly signicant and has a negative sign
(Model M3, Table 4), which shows that inequality undermines the effects
of democracy and vice-a-versa. These results support the inequality
Hypothesis 2 above, which suggests that, ceteris paribus, high levels of inequality in conjunction with poor democratic institutions are likely to result in large percentage of land set aside as PAS and vice-a-versa.
The presence of interaction terms makes the interpretation of overall
effect of inequality and democracy slightly more challenging than usual.
One way of facilitating an easy reading of the overall effect is to compute
the marginal effects (Kam and Franzese, 2010), which are shown in
Figs. 4 and 5, Fig. 4 shows that inequality produces a highly signicant
and positive effect on the provision of PAs in the countries in the

lower half of the range of democracy; i.e. between the transformed democracy scores of 0 and 6, equivalent to the FH index scores of between
8 and 14, which includes all the countries that Freedom House ranks as
not-free and most countries the group ranks as partially free..2
To understand how inequality shapes the effect of democracy on PA
designation, Fig. 5 shows the average marginal effects of democracy
across the entire range of values that economic inequality takes in
this sample of 137 countries. At low levels of income inequality, the
marginal effect of democracy on the percentage of national territory set
aside as PAs is positive, and these effects are highly signicant for countries with inequality up to the average level of income inequality in the
global sample (as shown Table 2). However, with increasing income inequality, the democracy dividend diminishes consistently before being
rendered statistically insignicant at the value of inequality around the
75th percentile in the global sample. Overall, the coefcient for interaction effect is highly signicant. On the other hand, the economic
variables GDP per-capita and its square variant (used to measure the environmental Kuznets curve), as well as the developed country dummy,
are not signicant in the main model M3 discussed above.
A review of the literature on wildlife and biodiversity conservation
suggests that two other types of differences among groups of countries
are worth exploring: rst, the developingdeveloped country distinction; second, the regional differences between countries on the contingents of Asia, Africa, Latin America, and the industrially developed
countries on the continents of Europe and North America. The contestation over PAs in North America and Europe tend to be latent and rarely
escalates into violent struggles (Stevens, 1997). On the contrary, wildlife conservation often turns violent in Africa, increasingly so with the
nancialization of forest-based ecosystem services, which also rely on
approaches rst developed in the context of PA-based conservation
(Fairhead et al., 2012). To examine and analyze these regional differences, Model M4 replaces the developed country dummy by regional
dummies for Africa, Latin America, and Europe & North America with
Asia being the reference dummy excluded from the model.
2
For details of Freedom House Methodology, see Methodology: Freedom in the World
2016 https://freedomhouse.org/report/freedom-world-2016/methodology.
147
.02
0
-.02
-.04
Effects on Predicted Mean % of Land Under PAs 2012
.04
Average Marginal Effects of Democracy 1995-2005 (95% CIs)
12
15
18
21
24
27
30
33
36
39
42
45
Average Income Inequality 1995-2005

Fig. 5. Average marginal effects of democracy on land under PAs, contingent on income inequality.
Results of this model reinforce the previous ndings about importance of inequality, democracy, and the interaction of these two variables. Additionally, they show that the countries on the continent of
Africa put signicantly larger percentage of national territory under
PAs than do the countries on the Asian continent (the reference
category). The coefcients for regional dummies, which show the net
effects after accounting for the commonly cited reasons for interregional differences like population density, offer an answer as to why
so many of the conservation related conicts are concentrated in
Africa (Garland, 2008, 51). Model M4 also shows that the percentage
of land set aside as PAs in countries in Europe and North America is statistically indistinguishable from countries on the Asian continent. Even
so, it is possible that the interaction between inequality and democracy
is of a fundamentally different nature in developed countries in comparison to the developing countries. Models M5-M7 shown in Table 5 help
clarify these questions.
Model M5 uses a three-way interaction term including variables for
democracy, inequality, and the dummy for developed countries, which allows for a nuanced interpretation of the effects of the variables included in
the interaction term. Results show that the coefcients for democracy, income inequality, and the interaction between democracy and inequality
are highly signicant and of the same signs as in the main models included in Table 4, while the economy variables are not signicant. After accounting for a three-way interaction, the developed country dummy is
signicant and has a substantively large effect on the designation of PAs.
Democracy has a statistically signicant and negative effect, while inequality has a negative but statistically insignicant effect. This is one
way of understanding the differences of effects between the developed
and developing countries, as shown by Torras and Boyce (1998).
A second method of approaching the same problem is to test the
effects of inequality and democracy in the two sub-samples separately
(M6 developed countries; M7 developing countries). 3 The
inequality-democracy effects are not signicant in the subsample of
the developed countries, while Kuznets curve thesis holds true for the
designation of PAs in this subsample (models M6). These results are in

conformance with the prior research showing similar results vis--vis
other environmental indicators in the developed countries (Scruggs,
1998; Bimonte, 2002). One must bear in mind that within the subsample
of developed countries, there is very little variation on the democracy
variable and a signicantly smaller variation on the inequality variable.
This lack of variation is expected to undermine the regression
coefcients. Results in model M7 show that the contingent nature of
the effects of inequality on the provision of PAs remains valid within
the sample of developing countries, though the interaction effect is
marginally signicant.4
Lastly, models M8M10, presented via Table 6, use alternative
specications for income inequality data for the period 20032005. As
discussed in the previous section, Model M8 uses the World Bank's
Gini coefcient, and M9 employs the share of income held by
individuals in the top 10 income percentile, while M10 uses the
difference between the shares of income held by the top and bottom
10 income percentiles. Each of these model specications produce
results that are fully consistent with the ndings of the main models
analyzed in Table 4 above. The coefcients for democracy and inequality
are highly signicant and positive; the interaction effect is also highly
signicant and has a negative sign similar to the previous models.
Such consistency of results in the face of alternative specication of
inequality data drawn from the World Development Indicators attests
to the robustness of the results in the main model reported above.
3
I thank an anonymous reviewer for this journal for suggesting this split sample
strategy.
4
The coefcient for the interaction effect in this model is very close to being signicant
with a p-value of 6 percentage points.
4. Analysis: Relevance for Theory and Policy

Empirical results discussed above point to a set of complex
inequality-democracy-environment relationships in the context of the
legal designation of PAs. While the individual coefcient for democracy
is associated positively with the provision of PAs, the ability of
democracies to set aside PAs the democratic dividend in the arena of
conservation diminishes consistently with increasing levels of
148
inequality (Fig. 5). As such, inequality is bad for nature conservation in

democracies, which supports the general theses proposed by Boyce
(2002) and Wilkinson and Pickett (2010). Paradoxically, income inequality has a statistically signicant positive effect on the provision of
PAs in non-democratic countries. Understanding how inequality may
promote apparently environmental friendly policies and programs
requires further elaboration. The next two subsections deal with two
sets of issues relevant to theoretical and policy debates. First, what
specic mechanisms, in highly unequal societies that are also weak democracies, lead to designation of large areas of PAs? The second subsection outlines the conditions under which debilitating consequences of
high levels of inequalities may be overcome.
4.1. Political and Economic Drivers of Nature Conservation Amidst Inequality

The provision of public goods of wildlife and biodiversity conservation is likely to clash with the interests of local groups, unless they are
appropriately compensated for the costs they have to bear. While this
is applicable most clearly to the developing countries with large areas
of forests and poor rule of law, tensions are also visible in developed
countries. In Appalachia and other less developed regions of the
United States, natural resources are central to local economic development, while environmentalist outsiders often demand strict regulations and curtailment of resource use in favor of recreational and
conservation activities (Nesbitt and Weiner, 2001, 333; Hurley and
Halfacre, 2011). In the United States, policymaking in the arena of conservation has been shaped most prominently by the elite and urban
middle-class environmental activists (Warren, 1999). The distributional
consequences of elite-dominated policymaking are magnied several
times over, as well-funded international conservation organizations
push developing country governments to bring large territories under
increasingly stricter regimes of nature conservation (Chapin et al.
2004). The creation and expansion of PAs in developing countries, especially those on the continents of Africa and Asia, have led to the displacements of an estimated 10 million people, including numerous cases of
violent evictions (Brockington, 2002; West et al., 2006).5
The promotion of PAs occurs through inter-elite bargains in which
developing country leaders commit to set aside land under PAs in return
for international funding, political recognition, and reputational gains.
One of the earliest examples of an international inter-elite bargain
was the Arusha Manifesto, which Julius Nyerere (Tanzania's longest
serving president) signed at the 1961 Arusha Symposium on the Conservation of Nature and Natural Resources in Modern African States.6
Arusha Manifesto was an implicit contract that in return for technical
expertise and material resources, African states would allow outside
experts more or less free latitude in determining the nature and level
of protection (Neumann, 1998; Rogers, 2002). The relatively easy
availability of international assistance for nature conservation also
enables government ofcials and elected leaders to exploit conservation
programs for rent seeking (Nelson, 2011;). These transactions are often
linked to large scale nances of wildlife tourism, which was worth $740
million in the year 2000 an amount equal to 16% of Tanzania's GDP
(Brockington, 2006). By January 2015 and with annual earnings of
$2.05 billion, tourism had beat gold to become the top source of foreign
exchange for Tanzania.7 Dan Brockington, who has studied wildlife
conservation in Tanzania for decades, attributes the high percentage of
land dedicated to wildlife conservation to the existence of an environmental-conservation complex (Brockington, 2006, 105).
5
American environmental journalist Mark Dowie puts the number of conservation refugees at 40 million, though most scholars consider that to be an overestimate.
6
The symposium was organized by the Commission for Technical Cooperation in Africa
(CCTA) and the International Union for the Conservation of Nature and Natural Resources
(IUCN).
7
http://www.theeastafrican.co.ke/business/Tanzania-tourism-beats-gold-in-foreignexchange-earning/-/2560/2668656/-/11d7ndtz/-/index.html.
None of these effects of exclusionary nature conservation are limited

to Tanzania or to the continent of Africa either (Gibson, 1999; Kashwan,
2013). Several developing country leaders, including Indian Prime
Minister Indira Gandhi and Mexican Presidents Lzaro Crdenas and
Carlos Salinas de Gortari, used environmental conservation policies as
a means of securing international support (Haenn, 2005; Chhatre and
Saberwal, 2006; McAfee and Shapiro, 2010). Elite policymakers and
politicians at all levels use wildlife policy for their own political ends,
which may or may not include conservation (Gibson, 1999). High levels
of domestic inequality and weak democratic institutions are necessary
for these types of inter-elite pacts for the designation of PAs, which
often tend to benet the ruling elite (Neumann, 1998; Gibson, 1999;
Brockington, 2002).
The political and economic motivations for the designation of PAs, as
described above, may undermine the primary goals of wildlife and
biodiversity conservation. Evidence from gap analysis studies in
conservation biology lends support to such concerns. Conservation biologists show that even if one were to assume that every species present
in a legally designated area is effectively protected, 20% of threatened
species remain beyond the PA coverage, while many PAs are located
in areas that are without valuable biodiversity (Rodrigues et al., 2004;
Naughton-Treves et al., 2005). There is also some evidence that political
relationships and geostrategic considerations often inuence decisions
by donor governments about how to direct investments meant for biodiversity conservation (Naughton-Treves et al., 2005). Besides, even as
governments expand coverage of PAs in some places, they continue to
reallocate forestlands rich in wildlife and biodiversity to development
projects (Kashwan, 2016).
These multiple layers of evidence suggest that the conicts over
the designation of PAs may not be attributed solely, or even mainly,
to the standard problems linked to the provision of public goods.
Instead, many of these conicts emanate from the political economic
exploitation of an environmental policy agenda that relies on closing
off popular access to increasingly larger areas of hinterlands (Brechin
et al., 2002; Chapin, 2004). To paraphrase James Boyce, PAs seem to
demonstrate a case in which the beneciaries of environmental
conservation impose too much environmental protection on society
than would be the case in the absence of economic inequality
(Boyce, 2002, 37). Disentangling the effects of inequality from the
difculties inherent in the provision of public goods will help in
clarifying conditions that are likely to foster distinct social and
environmental goals.
4.2. Inequality and Nature Conservation: Policy Implications
International conservation groups often promote the PAs as
everyone's solution with promises of addressing agendas as diverse
as species conservation, ecosystem stabilization, and provision of environmental goods and services, such as biofuels, climate change adaptation, and forest-based carbon sequestration as part of climate change
mitigation strategies (Corson et al., 2014). This rush to nancialize
forest-based services has led to the appropriation of land and resources
for environmental ends, often referred to as green grab because of
the involvement of corporations and conservation groups from developed countries (Fairhead et al., 2012, 237). Developing country governments, which own N 85% of the world's forests, are directly involved in
such high-stake commercialization of the hinterlands (Agrawal et al.,
2011). Programs of wildlife conservation are increasingly becoming
more militarized, as evident by the establishment of forest police
stations, acquisition of conservation drones, and forest ofcials
authorized to shoot on sight (Duffy, 2014). Such is the appeal of
these policies that some scholars offer cost-benet analyses of shoot
on sight policies (Messer, 2010).
The proposals for the increase of PAs enjoy signicant support from
conservation advocates, as evident from the popularity of the proposal
to set aside half the planet as permanently protected areas for the ten
million other species put forth by evolutionary biologist E.O. Wilson.8

While Wilson's intention may simply be to provoke debates about nding an appropriate balance between development and conservation,
such arguments are likely to reinforce the status-quo of elitedominance of the conservation agenda (Holmes, 2011). This is not to
say that PAs are necessarily detrimental to the interests of local people.
Indeed, considering that we see signicant geographic overlaps of
poverty, inequality, and biodiversity, PAs may prove instrumental for
poverty alleviation (Brechin et al., 2002; Naughton-Treves et al.,
2005). This potential is directly linked to the magnitude of opportunity
costs. In the case of Ranomafana National Park in southeastern
Madagascar in 1991, these costs were negligible in relation to national
and global benets that the park would produce. At the same time,
these costs were substantial in relation to the average household income in the region (Ferraro, 2002). As such, the economics of PAs is
not insurmountable, and in the presence of appropriate policies, PAs
do benet local residents, who are often signicantly poorer than
their fellow citizens living outside the forested regions (Haenn, 2005;
Andam et al., 2010).
The main contention is over who should pay the opportunity costs
related to PAs. Developed countries pay a small fraction of the nances
needed for PAs, even though poor rural and forest-dependent groups
are often asked to bear the opportunity costs without any meaningful
compensation (Balmford and Whitten, 2003). Even the minimum
compensatory payments and benets are not delivered very often
to the affected populations (Brockington, 2002; Kashwan, 2016).
However, the picture is not as uniformly bleak as it may seem at rst;
PA outcomes and their benets for local residents vary quite signicantly. For instance, few cases of eviction and displacements are reported
from Latin America despite large areas of land being set aside as PAs
(West et al., 2006). Latin America provides important insights about
the simultaneous pursuit of the goals of social justice and nature
conservation.
Scholars who study the politics of conservation in Mexico, for instance, suggest that the country's leadership has employed conservation programs to redirect resources to rural communities that have
been left behind as a result of the country's turn to neoliberal economic
policies (McAfee and Shapiro, 2010). Mexico's land rights regime,
coupled with the political context that affords some bargaining power
to peasant groups, enables them to negotiate with ruling party leaders
and international conservation agencies the terms of local conservation.
Several other Latin American countries have taken leadership in according security of peasant land rights and establishing some of the most
successful programs of payment for environmental services (Pagiola
et al., 2005; Corbera et al., 2011). Lastly, more than in other regions of
the world, Latin American environmental groups have joined hands
with social justice advocates to create discourses of environmental
citizenship in Brazil, Costa Rica, Peru, and Mexico among others
(Hochstetler and Keck, 2007; Orlove and Brush, 1996). The studies
cited here suggest that the PAs benet local residents in Latin
American countries relatively more than they do in countries on the
continents of Asia and Africa (e.g. Steinberg, 2015). Even so, such
gains cannot be attributed to geography, as the non-signicance of regional dummy for Latin America suggests (Model M6, Table 4). Instead,
they must be attributed to the region's peculiar history of peasant mobilization and the active linkages that social movements have forged with
political elites and government agencies (Hochstetler and Keck, 2007;
Corbera et al., 2011; Orlove and Brush, 1996).
This brief summary of research ndings from Latin America also
points to the importance of political bargaining in determining PA
outcomes, which has important implications for the empirical ndings discussed above. Because the models that this paper uses do
8
Tony Hiss. Can the World Really Set Aside Half of the Planet for Wildlife? Smithsonian
Magazine. September 2014. http://www.smithsonianmag.com/science-nature/canworld-really-set-aside-half-planet-wildlife-180952379/.
149
not account for meso and macro-level political bargaining, it leaves

two types of possibilities open. First, in all likelihood, conservation
is relatively less coercive in the Latin American contexts than it is
elsewhere.9 Second, a broader point is that, in these cases, the establishment of PAs seem to entail some kind of political bargaining,
which facilitates a relatively better accounting of local concerns and a
relatively more pro-poor distribution of the benets from the PAs
(see, Haenn, 2005; McAfee and Shapiro, 2010). Such context-specic
bargaining models need to be studied for their implications for theories
linking inequality, democracy, and environment and the formulation of
PA policies.
5. Conclusion
This paper provided empirical evidence about the effects of inequality with respect to a policy domain that is quite salient to the contemporary environmental policy debates. The empirical analysis presented
above shows that the effects of inequality on the provision of PAs is
greatly contingent on the strength of democracy. These highly signicant results, which are valid for multiple specications of the inequality
variable and for different time spans, show the importance of accounting for the inequality-democracy interaction for a full analysis of
inequality-environment linkages. Even so, this study also nds some evidence supporting the validity of a Kuznets curve argument vis--vis the
designation of PAs in the developed countries. The models inclusive of
regional dummies also show that ceteris paribus, countries on the
African continent, tend to set aside a larger area of national territory
under PAs. These ndings are of great relevance in the context of
increasing reliance on PAs for multiple pursuits in the arenas of
environment, development, and climate change. The empirical analyses
presented in this paper make a number of contributions to the
scholarship on environment and inequality.
To the best of author's knowledge, this is the rst paper that focuses
explicitly on James Boyce's inequality hypothesis in its broader sense
that the effect of inequality on environmental outcomes depends on
how the costs and benets of environmental protection are distributed
between different social groups. This paper nds that relatively greater
area under PAs tend to emerge under two sharply opposed conditions:
1) in undemocratic settings with high inequality and 2) in democratic
settings with low inequality.10 These results are illuminated in Figs. 4
& 5 respectively. In line with the focus on inequalities, this paper has
focused mainly on the rather counter intuitive result that inequalities
in non-democratic societies lead to larger areas set aside for the stated
goals of environmental protection. The empirical analysis above showed
that in non-democratic countries or in the presence of weak democratic
institutions, inequality may lead to outcomes that international
agencies readily and routinely equate to pro-environmental outcomes.
Discussion in the previous section showed that such apparently
environmental-friendly outcomes may contribute to greater poverty
and dispossession, especially in non-democratic societies.
The results of this research suggest that international agencies
should exercise caution in pushing for ever larger areas to be brought
under PAs, unless it is possible to distinguish between two very different
types of PAs: those that serve the private interests of elites and those
that serve the public good. Future research projects in this area will
benet from further disentangling of such contrasting drivers of PA
designation. The current emphasis on landscape conservation as a
means of climate change mitigation and adaption may not serve those
goals, because PA designations are often driven by non-environmental
political and economic considerations. To the extent these ndings
hold true, inequality is likely to produce deleterious environmental
consequences in the long run (Wilkinson and Pickett, 2010).
9
I thank a reviewer at Ecological Economics for helping formulate this point.

The author owes this specic framing of study's conclusions to an anonymous reviewer for this journal.
10
150
This paper also contributes to the discussion of two important

methodological limitations identied in literature on inequality and
environment recently. First, responding to a suggestion by Gates et al.
(2002) to separate environmental outcomes from environmental commitments, this paper accounts for the outputoutcome distinction
discussed in the policy literature to analyze legal designation of PAs,
which is a policy output. The approach adopted in this paper also
contributes to the development of an analysis of political channels
related to the implementation of public policy aimed at protecting the environment, as recommended by Berthe and Elie (2015). To do so, this
paper combined quantitative and qualitative evidence to examine the
linkages between the designation of PAs and the interests of two main
stakeholders national policymakers and conservation advocates, who
dominate decision-making related to the designation of PAs, and the
largely poor groups of local residents, who bear signicant costs related
to the lost opportunities for building roads, schools, and health facilities
among others. The extensive qualitative evidence summarized briey in
this paper shows that, in the presence of economic and political inequalities, the interests and political positions of some groups may be excluded
from, as opposed to being represented in, policymaking processes.
The analyses presented here build on the political economy of environment framework proposed by Boyce (1994), while engaging with
the social choice arguments that Scruggs (1998) makes. Following
Buchanan's important insight about social choice function, this paper diverges from the reliance on theories of economic behavior and group
preferences for or against environmental protection (but, see, Gates
et al., 2002; Roca, 2003). Considering the multiple sources of preference
formation and the difculties associated with attributing preferences
to groups, this paper emphasizes the powers and interests of the
policymakers, who are involved directly in PA related policymaking at national level. These arguments rely on fundamental tenets of the political
economy of institutions in which elite political actors and policymakers
exercise signicant power (Moe, 2005). The political economy approach
outlined in this paper recognizes environmental policymaking as a venue
for the contestation of diverse interests and ambitions attached to environmental resources. To supplement the past research focused on either
environmental Kuznets curve or the interest-based explanation of environmental outcomes, future research should pay equal attention to the
policymaking processes that translate environmental goals to statutory
and legal mechanisms with specic distributional consequences.
To conclude, inequality and democracy intervene directly in the policy process pertaining to nature conservation at the national level. The
single-minded focus of international conservation groups and advocates
on pursuing the apparently conservation-promoting policy outputs
often create social conicts, which occur predominantly in, but are not
limited to, poor developing countries (Nesbitt and Weiner, 2001;
Brockington, 2002; Kashwan, 2016). Indeed, inequality greatly undermines the democratic dividend for environmental protection. More
research is needed to examine inequalities entrenched in national environmental policymaking systems (Berthe and Elie, 2015). As many of
the scholars of environmental inequalities show, environmental protection cannot be sustained without institution building, yet inequalities
entrenched within institutional and policy processes must be addressed
separately. The goal should be to improve transparency in policymaking
and implementation, an equitable sharing of benets, and ensuring
accountability among actors and agencies vested with the task of
environmental protection. Public action on this front is likely to help
alleviate the effects of institutionalized inequalities, especially if they
help redirect much needed economic development opportunities to
rural communities that inhabit natural landscapes.
Acknowledgements
An early draft of this paper was presented at the Stockholm Environment Institute's International Conference on Inequality and Sustainability November 910, 2012 at the Fletcher School of Law and Diplomacy
at Tufts University. More recent versions were presented in the summer

of 2015 at the Indian Institute of Forest Management, Bhopal and at the
Annual Public Policy Conference at the Indian Institute of Management,
Bangalore, and at the Political Economy Workshop at the University of
Connecticut in spring 2016. Constructive comments and suggestions
from audiences at each of these meetings contributed to the development of the article in its present shape. The author also gratefully
acknowledges original database shared by Lyle Scruggs, which this
author updated and developed further for this study. The full sample
data used in this article are available from the author upon request.
References
Agrawal, A., Nepstad, D., Chhatre, A., 2011. Reducing emissions from deforestation and
forest degradation. Annu. Rev. Environ. Resour. 36, 373396.
Andam, K.S., Ferraro, P.J., Sims, K.R.E., Healy, A., Holland, M.B., 2010. Protected areas
reduced poverty in Costa Rica and Thailand. Proc. Natl. Acad. Sci. 107, 999610001.
Babones, S.J., Alvarez-Rivadulla, M.J., 2007. Standardized income inequality data for use in
cross-national research. Sociol. Inq. 77, 322.
Balmford, A., Whitten, T., 2003. Who should pay for tropical conservation, and how could
the costs be met? Orynx 37.
Berthe, A., Elie, L., 2015. Mechanisms explaining the impact of economic inequality on
environmental deterioration. Ecol. Econ. 116, 191200.
Bimonte, S., 2002. Information access, income distribution, and the Environmental
Kuznets Curve. Ecol. Econ. 41, 145156.
Boyce, J.K., 1994. Inequality as a cause of environmental degradation. Ecol. Econ. 11,
169178.
Boyce, J.K., 2002. The Political Economy of the Environment. Edward Elgar Publishing,
Incorporated.
Brechin, S.R., Wilshusen, P.R., Fortwangler, C.L., West, P.C., 2002. Beyond the square
wheel: toward a more comprehensive understanding of biodiversity conservation
as social and political process. Soc. Nat. Resour. 15, 4164.
Brockington, D., 2002. Fortress Conservation: The Preservation of the Mkomazi Game
Reserve, Tanzania. International African Institute in association with Indiana
University Press, Bloomington.
Brockington, D., 2006. The politics and ethnography of environmentalisms in Tanzania.
Afr. Aff. 105, 97116.
Buchanan, J.M., 1954. Social choice, democracy, and free markets. J. Polit. Econ. 114123.
Carson, R.T., 2010. The Environmental Kuznets Curve: seeking empirical regularity and
theoretical structure. Rev. Environ. Econ. Policy 4, 323.
Chamberlin, T.C., 1965. The method of multiple working hypotheses. Science 148,
754759.
Chapin, M., 2004. A Challenge to Conservationists. World Watch Magazine, pp. 1731.
Chhatre, A., Saberwal, V., 2006. Democratizing Nature: Politics, Conservation, and
Development in India. Oxford University Press, New Delhi.
Clement, M., Meunie, A., 2010. Is inequality harmful for the environment? An empirical
analysis applied to developing and transition countries. Rev. Soc. Econ. 68, 413445.
Constance, D.H., Bonanno, A., 2000. Regulating the global sheries: the World Wildlife
Fund, Unilever, and the Marine Stewardship Council. Agric. Hum. Values 17, 125139.
Corbera, E., Estrada, M., May, P., Navarro, G., Pacheco, P., 2011. Rights to land, forests and
carbon in REDD+: insights from Mexico, Brazil and Costa Rica. Forests 2, 301342.
Corson, C., Gruby, R., Witter, R., Hagerman, S., Suarez, D., Greenberg, S., Bourque, M., Gray,
N., Campbell, L.M., 2014. Everyone's solution? Dening and redening protected
areas at the convention on biological diversity. Conserv. Soc. 12, 190.
Daudelin, J., 2003. Land and Violence in Post-Conict Situations. Unpublished Paper for
the North-South Institute and the World Bank. Carleton University, Ottawa.
Duffy, R., 2014. Waging a war to save biodiversity: the rise of militarized conservation. Int.
Aff. 90, 819834.
Fairhead, J., Leach, M., Scoones, I., 2012. Green grabbing: a new appropriation of nature?
J. Peasant Stud. 39, 237261.
Ferraro, P.J., 2002. The local costs of establishing protected areas in low-income nations:
Ranomafana National Park, Madagascar. Ecol. Econ. 43, 261275.
Forstater, M., 2004. Visions and scenarios: Heilbroner's worldly philosophy, Lowe's political economics, and the methodology of ecological economics. Ecol. Econ. 51, 1730.
Garland, E., 2008. The elephant in the room: confronting the colonial character of wildlife
conservation in Africa. Afr. Stud. Rev. 51, 5174.
Gates, S., Neumayer, E., Gleditsch, N.P., 2002. Environmental Commitment, Democracy
and Inequality: A Background Paper to the World Development Report 2003.
World Bank, Washington D.C.
Gibson, C.C., 1999. Politicians and Poachers : The Political Economy of Wildlife Policy in
Africa. Cambridge Univ. Press, Cambridge [u.a.
Guha, R., 2006. How Much should a Person Consume? : Environmentalism in India and
the United States. University of California Press, Berkeley, CA.
Haenn, N., 2005. Fields of Power, Forests of Discontent: Culture, Conservation, and the
State in Mexico. University of Arizona Press.
Hochstetler, K., Keck, M.E., 2007. Greening Brazil: Environmental Activism in State and
Society. Duke University Press.
Holmes, G., 2011. Conservation's friends in high places: neoliberalism, networks, and the
transnational conservation elite. Global Environ. Polit. 11, 121.
Hurley, P., Halfacre, A., 2011. Dodging alligators, rattlesnakes, and backyard docks: a
political ecology of sweetgrass basket-making and conservation in the South Carolina
Lowcountry, USA. GeoJournal 76, 383399.

Juffe-Bignoli, D., Burgess, N.D., Bingham, H., Belle, E.M.S., de Lima, M.G., Deguignet, M.,
Bertzky, B., Milam, A.N., Martinez-Lopez, J., Lewis, E., Eassom, A., Wicander, S.,
Geldmann, J., van Soesbergen, A., Arnell, A.P., O'Connor, B., Park, S., Shi, Y.N., Danks,
F.S., MacSharry, B., Kingston, N., 2014. Protected Planet Report 2014: Tracking Progress towards Global Targets for Protected Areas. The United Nations Environment
Programme World Conservation Monitoring Centre (UNEP-WCMC), Cambridge, UK.
Kam, C., Franzese, R., 2010. Modeling and Interpreting Interactive Hypotheses in
Regression Analysis. University of Michigan Press, Ann Arbor.
Kashwan, P., 2013. The politics of rights-based approaches in conservation. Land Use
Policy 31, 613626.
Kashwan, P., 2016. Power asymmetries and institutions: landscape conservation in
central India. Reg. Environ. Chang. 16, 97109.
Koontz, T.M., Thomas, C.W., 2006. What do We know and need to know about the environmental outcomes of collaborative management? Public Adm. Rev. 66, 111121.
Magnani, E., 2000. The Environmental Kuznets Curve, environmental protection policy
and income distribution. Ecol. Econ. 32, 431443.
Martini, C., Tiezzi, S., 2014. Is the environment a luxury? An empirical investigation
using revealed preferences and household production. Resour. Energy Econ. 37,
147167.
McAfee, K., Shapiro, E.N., 2010. Payments for ecosystem services in Mexico: nature,
neoliberalism, social movements, and the state. Ann. Assoc. Am. Geogr. 100,
579599.
McNeely, J.A., Schutyser, F., 2003. Protected Areas in 2023: Scenarios for an Uncertain
Future. World Congress on Protected Areas - The World Conservation Union
(IUCN), Gland.
Messer, K.D., 2010. Protecting endangered species: when are shoot-on-sight policies the
only viable option to stop poaching? Ecol. Econ. 69, 23342340.
Moe, T.M., 2005. Power and political institutions. Perspect. polit. 3, 215.
Munck, G.L., Verkuilen, J., 2002. Conceptualizing and measuring democracy. Comp. Polit.
Stud. 35, 534.
Naughton-Treves, L., Holland, M.B., Brandon, K., 2005. The role of protected areas in
conserving biodiversity and sustaining local livelihoods. Annu. Rev. Environ. Resour.
30, 219252.
Nelson, F., 2011. Blessing or curse? The political economy of tourism development in
Tanzania. J. Sustain. Tour. 20, 359375.
Nesbitt, J.T., Weiner, D., 2001. Conicting environmental imaginaries and the politics of
nature in Central Appalachia. Geoforum 32, 333349.
Neuhaus, J., McCulloch, C., 2011. Generalized linear models. Wiley Interdiscip. Rev.
Comput. Stat. 3, 407413.
Neumann, R.P., 1998. Imposing Wilderness: Struggles over Livelihood and Nature
Preservation in Africa. University of California Press, Berkeley.
151
Orlove, B.S., Brush, S.B., 1996. Anthropology and the conservation of biodiversity. Annu.
Rev. Anthropol. 329352.
Ostrom, E., 2011. Background on the institutional analysis and development framework.
Policy Stud. J. 39, 727.
Pagiola, S., Arcenas, A., Platais, G., 2005. Can payments for environmental services help
reduce poverty? An exploration of the issues and the evidence to date from Latin
America. World Dev. 33, 237253.
Roca, J., 2003. Do individual preferences explain the Environmental Kuznets Curve? Ecol.
Econ. 45, 310.
Rodrigues, A.S.L., Andelman, S.J., Bakarr, M.I., Boitani, L., Brooks, T.M., Cowling, R.M.,
Fishpool, L.D.C., da Fonseca, G.A.B., Gaston, K.J., Hoffmann, M., Long, J.S., Marquet,
P.A., Pilgrim, J.D., Pressey, R.L., Schipper, J., Sechrest, W., Stuart, S.N., Underhill, L.G.,
Waller, R.W., Watts, M.E.J., Yan, X., 2004. Effectiveness of the global protected area
network in representing species diversity. Nature 428, 640643.
Rogers, P.J., 2002. Global Governance/Governmentality, Wildlife Conservation, and
Protected Area Management: A Comparative Study of Eastern and Southern Africa,
African Studies Association 45th Annual Meeting, Washington, DC.
Sanderson, S., 2005. Poverty and conservation: the new Century's peasant question?.
World Dev. 33, 323332.
Scruggs, L.A., 1998. Political and economic inequality and the environment. Ecol. Econ. 26,
259275.
Steinberg, P.F., 2015. Who Rules the Earth?: How Social Rules Shape our Planet and our
Lives. Oxford University Press.
Stevens, S., 1997. Conservation through Cultural Survival: Indigenous Peoples and
Protected Areas. Island Press.
Stewart, F., 2010. Horizontal Inequalities and Conict: Understanding Group Violence in
Multiethnic Societies. Palgrave Macmillan, UK.
Taylor, D.E., 2014. Toxic Communities: Environmental Racism, Industrial Pollution, and
Residential Mobility. New York University Press.
Torras, M., Boyce, J.K., 1998. Income, inequality, and pollution: a reassessment of the
environmental Kuznets curve. Ecol. Econ. 25, 147160.
Wahlke, J.C., 1971. Policy demands and system support: the role of the represented. Br.
J. Polit. Sci. 1, 271290.
Warren, L.S., 1999. The Hunter's Game: Poachers and Conservationists in TwentiethCentury America. Yale University Press.
West, P., Igoe, J., Brockington, D., 2006. Parks and peoples: the social impact of protected
areas. Annu. Rev. Anthropol. 35, 251277.
Wilkinson, R., Pickett, K., 2010. The Spirit Level: Why Greater Equality Makes Societies
Stronger. Bloomsbury Publishing.
Young, O.R., 1998. Institutional uncertainties in international sheries management. Fish.
Res. 37, 211224.

1 s2.0 S0921800915301191 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0921800915301191 Main

Uploaded by

Copyright:

Available Formats

Ecological Economics 131 (2017) 139151

Contents lists available at ScienceDirect

Inequality, democracy, and the environment: A cross-national analysis

E-mail address: prakash.kashwan@uconn.edu.

democracy on designation of PAs is positive at low levels of inequality;

P. Kashwan / Ecological Economics 131 (2017) 139151

and environmental outcomes (Boyce, 1994; Scruggs, 1998; Torras and

P. Kashwan / Ecological Economics 131 (2017) 139151

Fig. 1. Distribution of land under protected areas, year 2014.

expectation, the designation of PAs constitutes a case of provision in

P. Kashwan / Ecological Economics 131 (2017) 139151

Small Percentage of Land

Fig. 2. Effect of inequality on designation of PAs, contingent on democracy.

no such index exists for cross-national measurements of political and

P. Kashwan / Ecological Economics 131 (2017) 139151

Top 10% (income share) 2005

Forest area 2000

Population density 2000/2005

considerations. First, 19952005 allows for the minimum time-lag

Percentage land protected 2012

Sub-sample: developed countries

Sub-sample: developing countries

P. Kashwan / Ecological Economics 131 (2017) 139151

GDP _sq. 2000

Area under forests

Protected Areas 2012

Standard errors in parentheses.

P. Kashwan / Ecological Economics 131 (2017) 139151

a the developed countries' sub-sample; b the developing countries' sub-sample.

more explanatory variables and a variety of non-normally distributed

Income inequality 2005

FH Democracy Index 2005

Percentage of National Territory under PAs (2012)

Fig. 3. Distribution of percentage of land under protected areas (PAs).

FH Democracy Index 2005

Standard errors in parentheses.

P. Kashwan / Ecological Economics 131 (2017) 139151

Effects on Predicted Mean % of Land under PAs (2012)

Average Marginal Effects of Income Inequality (95% CIs)

and positive effect on the provision of PAs in the countries in the

P. Kashwan / Ecological Economics 131 (2017) 139151

Effects on Predicted Mean % of Land Under PAs 2012

Average Marginal Effects of Democracy 1995-2005 (95% CIs)

Average Income Inequality 1995-2005

designation of PAs in this subsample (models M6). These results are in

4. Analysis: Relevance for Theory and Policy

P. Kashwan / Ecological Economics 131 (2017) 139151

inequality (Fig. 5). As such, inequality is bad for nature conservation in

4.1. Political and Economic Drivers of Nature Conservation Amidst Inequality

None of these effects of exclusionary nature conservation are limited

P. Kashwan / Ecological Economics 131 (2017) 139151

million other species put forth by evolutionary biologist E.O. Wilson.8

not account for meso and macro-level political bargaining, it leaves

I thank a reviewer at Ecological Economics for helping formulate this point.

P. Kashwan / Ecological Economics 131 (2017) 139151

This paper also contributes to the discussion of two important

at Tufts University. More recent versions were presented in the summer

P. Kashwan / Ecological Economics 131 (2017) 139151

You might also like