E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / e c o l e c o n

ANALYSIS

Taking the U out of Kuznets

A comprehensive analysis of the EKC and environmental degradation

Jill L. Caviglia-Harris a,, Dustin Chambers a , James R. Kahn b
a b

Salisbury University, 1101 Camden Ave., Salisbury, MD 21804, United States Washington and Lee University, Science AG-15, Lexington, VA 24450, United States

AR TIC LE D ATA
Article history: Received 13 August 2007 Received in revised form 5 August 2008 Accepted 5 August 2008 Available online 12 September 2008 Keywords: Environmental Kuznets Curve Ecological Footprint Development Growth Sustainability EKC JEL classification: Q0; Q01

ABSTR ACT
Unlike most Environmental Kuznets Curve (EKC) studies which focus on narrow measures of pollution as proxies for environmental quality, we test the validity of the EKC using the Ecological Footprint (EF), a more comprehensive measure of environmental degradation. We find no empirical evidence of an EKC relationship between the EF and economic development, and only limited support for such a relationship among the components of the EF. In addition, we discover that energy is largely responsible for the lack of an EKC relationship, and that energy consumption levels would have to be cut by over 50% in order for a statistically significant EKC relationship to emerge from the data. Overall, these results suggest that growth alone will not lead to sustainable development. 2008 Elsevier B.V. All rights reserved.

1.

Introduction

If the Environmental Kuznets Curve (EKC) is valid for all types of environmental degradation, then sufficient economic development alone will solve environmental problems in both developed and underdeveloped nations. Not surprisingly, this simple yet powerful implication has played an important role in the ongoing debate regarding appropriate economic growth and environmental policies (Ranjan and Shortle, 2007). Unfortunately, most of the empirical investigations of the EKC have focused on the narrow relationship between pollution output (as an inversely proportional proxy for environmental

quality) and economic growth. These particular pollutants are only a small part of environmental concerns at the global level. Consequently, the analysis performed in this paper tests the validity of the EKC using a much more comprehensive measure of environmental degradation, the Ecological Footprint (EF). Research on the validity, application, and measurement of the Environmental Kuznets Curve (EKC) has been prolific (Azomahou et al., 2006). Adapted from Kuznets' (1955) original study on the influences of economic development on income inequality, the EKC reflects the relationship between environmental quality and per capita income. The EKC asserts that

Corresponding author. E-mail address: jlcaviglia-harris@salisbury.edu (J.L. Caviglia-Harris). 0921-8009/$ see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolecon.2008.08.006

1150

E C O L O G IC A L E C O N O M IC S 6 8 ( 2 0 09 ) 11 4 9 1 15 9

environmental quality first declines (traditionally measured by an increase in pollution) in response to economic development, and improves (i.e. pollution levels decline) only after per capita income surpasses a critical threshold. This combination of falling then rising environmental quality (as measured by pollution output) during the course of economic growth and resulting development results in an inverted U shaped curve. Research on the EKC began with the analysis of panel data on 42 countries to identify an EKC effect for different measurements of air quality (Grossman and Krueger, 1993). In the same genre, Selden and Song (1994) found support for an EKC for SO2, while Grossman and Krueger (1995) andShafik and Banyopadhyay (1992) found water pollution to decline monotonically with income per capita while carbon emissions rise with income per capita. Since these initial studies, many have followed, focusing specifically on air pollution (i.e. List and Gallet, 1999; Heerink et al., 2001; Cole, 2003; Khanna, 2002; Bruvoll et al., 2003; Deacon and Norman, 2006; Merlevede et al., 2006; water pollution (Torras and Boyce, 1998; Paudel et al., 2005), deforestation (i.e. Culas, 2007; Rodriguez-Meza et al., 2003; Heerink et al., 2001; Barbier, 2001), hazardous waste and toxins (i.e. Gawande et al., 2001; Rupasingha et al., 2004), carbon dioxide (CO2) (Azomahou et al., 2006) among others (see Cavlovic et al., 2000; Dasgupta et al., 2002; Copeland and Taylor, 2004 for reviews). One result of this expansive literature is that no simple, predictable relationship between an aggregate measure of environmental quality and per capita income has been identified; instead the EKC has been found to hold only for a subset of environmental measures (Stern, 1998; Plassmann and Khanna, 2006). Several shortcomings along with inconsistencies in theoretical modeling have lead to strong criticisms of the EKC (Mller-Frstenberger and Wagnerb, 2007; Perman and Stern, 2003). Critics have challenged both the findings (especially those based on cross-sectional data) and policy implications of these studies (Dasgupta et al., 2002); pointing out that the results are often sensitive to the nations (or states) chosen, the pollutant measurement (emissions versus ambient concentrations), trade effects, functional form, and methodological choice (Harbaugh et al., 2002; also see Cavlovic et al., 2000). And, since much of the analysis on the EKC is derived from reduced-form models, a variety of (sometimes conflicting), theoretical explanations can apply. For example, several studies have proposed the new toxins scenario may exist in which the traditional pollutants exhibit an inverted Ushape in relation to increases in income; however the pollutants that replace these do not, leading to an overall increase in environmental degradation (Stern, 2004). In addition, an important conclusion that can be drawn from a summary of the literature is that greenhouse gasses, in particular CO2, exhibit an increasingand even U (not inverted) shapedrelationship with growth (Galeotti et al., 2006; Azomahou et al., 2006). Perhaps the greatest limitation of earlier EKC studies is their singular focus on one (or a small group of) pollutants as their measure of environmental quality. While the implications of single pollutants on health and the environment are important issues to address, the impact of individual decisions on the entire suite of pollutants along with potentially irreversible damage to ecosystems is of equal or greater

importance since the substitution possibilities between different pollutants could negate any positive impacts on the environment noted for a single source. Notable exceptions to these studies on single pollutants include Rupasingha et al. (2004), Jha and Murthy (2003), and Boutaud et al. (2006). Recently, greater effort has been made to construct comprehensive measures of environmental quality. For example, Jha and Murthy (2003) estimate global environmental degradation with an environmental degradation index (EDI) incorporating six environmental indicators: annual per capita fresh water withdrawal, annual fresh water withdrawal as a percentage of water resources, per capita paper consumption, per capita CO2 emissions, share of world CO2 emissions, and the average annual rate of deforestation. While broader than a single pollutant, the EDI is limited as a measurement of overall environmental quality by available data. Strong arguments could be made for the inclusion of a different or more inclusive set of environmental indicators. Finally, Boutaud et al. (2006) exam the relationship between the Ecological Footprint (EF) and Human Development Index (HDI) and growth. While Boutaud et al. (2006) include aggregate indices to test for an EKC, the authors rely on cross-sectional data for a single year and graphical representation of the data, resulting in analysis that is not conducive to hypothesis testing. This paper builds on this more inclusive approach with the development of a theoretical framework incorporating environmental capital into the carrying capacity of a nation and an empirical model utilizing a time series of 40 years of data on GDP and an aggregate measurement of environmental damage called the Ecological Footprint. More specifically, the goal of the analysis is to determine whether an EKC can be identified for this cumulative measurement of environmental degradation. The remainder of the paper is organized as follows: Section 2 discusses the Ecological Footprint; Section 3 derives necessary conditions if both strong sustainability and balanced economic growth are to be achieved; Section 4 describes the data used in the panel regressions; Section 5 describes the various EKC panel models and their estimation results; and Section 6 concludes.

2.

The Ecological Footprint

The Ecological Footprint (EF) was introduced by Rees (1992) and further developed in Wackernagel and Rees (1996) to determine how the environmental damage associated with human consumption compares to the biosphere regenerative capacity. The EF estimates the amount of natural capital (measured in biologically productive area) needed to support the resource demand and waste absorption requirements of a population and is expressed in global hectares or hectares of globally standardized bioproductivity (Wackernagel et al., 2004a,b). Specifically, the EF measures the human demand on nature by assessing how much biologically productive land and sea area is necessary to maintain a given consumption pattern (Wiedmann et al., 2006). In the basic calculation of the EF, consumption (categorized by food, services, transportation, consumer goods, and housing) is divided by the predetermined yield (biological productivity) by land type

E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

1151

including cropland, pasture, forest, built-up land, fisheries, and energy land. The ability of these areas to supply ecological goods and services (i.e. the predetermined yield) depends on the biophysical characteristics of the land (such as soil type, slope, and climate) in addition to socio-economic choices (such as management decisions and technological inputs). The EF requires that strong sustainability is maintained, as it assesses physical utilization of environmental resources (i.e. renewable factors of production and ecological services). However, the measurement is not all inclusive as it neglects atmospheric ozone levels, and does not account for pollutants that are difficult to convert to land or water ecosystem equivalents, such as methane and sulfur (Rees, 2000). The measurement, use, and interpretation of the EF have been extensively debated in the literature. A major strength of the EF is that it condenses a large array of environmental data into a single measure, which can be easily compared to a region's corresponding carrying capacity (Costanza, 2000). This is a relatively simple concept to understand and therefore can be used to explain issues of sustainability to the general public, and as a result of its rising notoriety has been increasingly applied within the literature. For example, the measurement has been used to evaluate resource use across nations (White, 2007), methods have been developed to improve its robustness across comparative means (Lenzen et al., 2007; Wiedmann and Lenzen, 2007), and to evaluate the impact of tourism and trade agreements (Hong et al., 2007; Patterson et al., 2007). While the ease of interpretation adds to the strengths of the EF, the assumptions that are made to convert this encompassing measurement into a single unit have lead to much of its criticism. Noted weaknesses include the many simplifying assumptions required to convert consumption data into land area. Specifically, Ayres (2000) faults the energy equivalence assumption used to convert energy flows into land area, while Van Kooten and Bulte (2000) note several weaknesses with aggregation, discounting, and sustainability, and take the position that the authors of the EF never present a clear and scientifically rigorous definition of the EF. Similar criticism focuses on the conversion of energy into land used to absorb CO2 emissions, as there are several ways to compensate for CO2 emissions outside of forest absorption. Finally, the EF, despite its ability to pinpoint areas under environmental pressure, provides no guidelines for environmental policy (Nijkamp et al., 2004). Despite these shortcomings, it is important to note that any aggregate indicator of environmental quality will have both strengths and weaknesses (as for example, measures of aggregate economic output or the price level suffer from specific problems). In this paper, we choose the EF as an aggregate measure of environmental quality because its limitations are well-known, it is a widely referenced measurement of sustainability (Nijkamp et al., 2004; Haberl et al., 2001), and has been adopted by a growing number of government authorities, agencies, and policy makers as a measure of ecological performance (Wiedmann et al., 2006). An alternative approach would be to devise our own index of environmental quality, but in so doing, we would introduce a measure that has not benefited from the normal scrutiny of the properties and limitations of the indicator.

3.

Necessary conditions for sustainability

One cannot analyze ecological degradation over time without addressing the issue of sustainability and building a working definition to incorporate in the analysis. The literature on sustainability has defined two concepts: weak and strong sustainability. Weak sustainability is an economic principle requiring that productive capacity not decline over time. On the other hand, strong sustainability, having evolved from the ecological economics perspective, requires that the total stock of natural capital not decline over time (Costanza and Daly, 1992; Hediger, 1999). The debate on whether weak or strong sustainability should provide the foundation for public policy reflects differences in opinion regarding the degree to which natural capital can be substituted for human and physical capital (Cabeza Gutes, 1996). Proponents of strong sustainability believe that natural capital is unique and plays an important role in human welfare, and thus cannot be replaced (Barbier, 2005). Furthermore, the concepts of growth and development are clearly outlined within this literature. Accordingly, time infinite growth cannot be sustainable on a finite planet as it is accomplished through the use of natural resources. On the other hand, development occurs through improvements in efficiency and therefore can lead the sustainable use of resources over time (Costanza and Daly, 1992).1 Regardless of one's preferred concept of sustainability, the literature on sustainable development and growth theory provides a number of interesting insights for the long-run path of the EF. While the sustainability of long-run economic growth subject to non-renewable resource constraints has interested economists for more than two centuries (see for example Malthus, 1789), it was not until the energy crises of the 1970s that economists rigorously analyzed the affects of natural resource scarcity on growth and development within the context of dynamic, general equilibrium models (see Solow, 1974; Stiglitz, 1974 among others). Their findings were straightforward: so long as the reproducible factor of production (i.e. physical or man-made capital) is sufficiently substitutable for the non-renewable factor, long-run balanced growth (i.e. per capita output growing en infinitum at a constant rate) is possible. The major drawback with this research is that it ignores the impact production has on the state of the environment. Addressing this deficiency, Stokey (1998) builds a model with pollution-generating output and a government that imposes progressively stringent emission regulations (achieved vis--vis costly abatement). She finds that even in this context, sustainable balanced growth is possible providing a sufficiently high rate of return on capital, giving rise to an output path of pollution that follows an inverted U shaped pattern consistent with the EKC. These results, despite their admittedly narrow focus, provide us with the key insight to understanding the long-run trajectory of the EF if both strong

Francheschi and Kahn (2003) separate natural and environmental resources and link sustainability to the continued ability of environmental resources to provide ecological services, for which human capital and human-made capital are not good substitutes.

1152

E C O L O G IC A L E C O N O M IC S 6 8 ( 2 0 09 ) 11 4 9 1 15 9

Table 1 Average per capita GDP and footprint by country Country

Afghanistan Albania Algeria Angola Argentina Armenia Australia Austria Azerbaijan Bangladesh Belarus Belgium & Luxembourg Benin Bolivia Bosnia and Herzegovina Botswana Brazil Bulgaria Burkina Faso Burundi Cambodia Cameroon Canada Central African Rep. Chad Chile China Colombia Congo, Dem. Rep. Congo, Republic of Costa Rica Cote d`Ivoire Croatia Cuba Czech Republic Denmark Dominican Republic Ecuador Egypt El Salvador Eritrea Estonia Ethiopia Finland France Gabon Gambia, The Georgia Germany

GDP
1713 2967 4866 1975 9827 3444 18,419 18,219 3060 1544 9487 17,411 1125 2770 2415 4345 5743 7385 778 829 537 2364 18,860 968 892 7291 1508 4658 906 1842 6378 2044 8241 5014 13,064 20,097 4066 3968 2812 3970 606 9973 661 15,785 17,697 13,674 878 3601 19,626

EF
0.22 1.26 1.28 0.82 2.81 0.98 6.85 3.92 1.53 0.51 3.40 4.71 0.99 1.23 1.69 1.44 1.88 3.07 1.03 0.88 0.74 0.99 6.78 0.94 1.12 1.66 1.19 1.25 0.72 0.91 1.88 1.02 2.23 1.57 4.95 5.15 1.22 1.28 1.15 0.98 0.77 5.07 1.30 5.37 4.49 1.34 1.13 1.14 4.90

Country
Ghana Greece Guatemala Guinea GuineaBissau Haiti Honduras Hungary India Indonesia Iran Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Korea, Dem. Rep. Korea, Rep. of Kuwait Kyrgyzstan Laos Latvia Lebanon Lesotho Liberia Libya Lithuania Macedonia Madagascar Malawi Malaysia Mali Mauritania Mauritius Mexico Moldova Mongolia Morocco Mozambique Namibia Nepal Netherlands New Zealand Nicaragua Niger

GDP
1071 10,641 3448 2488 623 2040 2142 9191 1606 2338 5378 2278 12,409 14,886 15,855 4243 16,201 4218 7197 1245 1222 7082 31,830 3154 1136 8370 5083 1155 1162 10,335 8737 4972 1072 662 5805 818 1330 8276 6320 2539 1456 3021 1043 5067 1033 18,468 16,595 4921 1041

EF
0.93 3.19 1.03 1.09 0.84 0.69 1.34 3.88 0.74 0.95 1.63 0.91 3.98 3.81 3.17 1.65 3.64 1.46 3.59 0.91 1.99 2.17 6.32 1.40 0.91 2.99 2.50 1.00 0.92 3.28 3.88 2.35 0.91 0.74 1.80 1.01 1.23 1.24 2.04 1.52 3.50 0.89 0.71 1.18 0.69 3.92 4.60 1.36 1.38

Country
Nigeria Norway Pakistan Panama Papua New Guinea Paraguay Peru Philippines Poland Portugal Romania Russia Rwanda Saudi Arabia Senegal Serbia and Montenegro Sierra Leone Slovak Republic Slovenia Somalia South Africa Spain Sri Lanka Sudan Swaziland Sweden Switzerland Syria Tajikistan Tanzania Thailand Togo Trinidad &Tobago Tunisia Turkey Turkmenistan Uganda Ukraine United Arab Emirates United Kingdom United States Uruguay Uzbekistan Venezuela Vietnam Yemen Zambia Zimbabwe

GDP
1051 20,842 1735 5484 3671 4199 4308 2981 6412 10,618 4441 9321 1044 20,806 1472 2353 1131 9057 16,449 1033 7119 12,825 2233 1080 6188 18,696 23,621 1603 1697 586 3603 1048 10,421 4399 4009 6811 894 5449 29,267 16,992 23,351 7804 3475 7761 1905 998 1135 2985

EF
1.23 4.69 0.64 1.74 1.71 1.98 0.98 0.97 4.26 2.94 2.92 4.48 0.87 3.80 1.42 2.38 0.92 3.03 2.85 0.57 2.60 3.20 0.82 1.04 1.20 5.25 4.56 1.42 0.73 0.85 1.06 0.99 2.40 1.31 1.97 2.85 1.34 2.94 7.92 4.91 8.16 2.80 1.83 2.44 0.69 0.77 0.85 1.06

Note: Footprints are measured in standardized global hectares (g ha); GDP in PPP-adjusted (2000) international dollars.

sustainability and balanced growth are be achieved: natural capital currently used in production must be replaced (in the long run) by reproducible factors of production. In order for strong sustainability to hold, total demands placed on the ecosystem or use of natural resources (N) in a given period cannot exceed the planet's regenerative capacity or the total stock of natural resources R(t) during the same period: NtVRt 3:1

capital used in environmental services, including necessary life support and ecological services, Nenv, and the natural capital required to produce current output (expressed as the product of NY (natural capital per real dollar of output) and Y(t) (GDP)). In other words, at any time period, natural capital can be divided as follows: Nt Nenv t NY tdY t 3:2

Building on this necessary condition, assume that natural capital can be partitioned into two components: natural

Holding natural capital used in non-production activities constant, Nenv(t) = Nenv (which is consistent with strong

E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

1153

Table 2 Global and income-specific footprint averages Period Ecological Footprint

19611965 19661970 19711975 19761980 19811985 19861990 19911995 19962000 1.70 1.85 2.02 2.05 1.97 2.04 2.13 2.13

World Per Capita GDP

3414 4014 4670 5127 5344 5904 6340 7065

Poor nations Population

2.54 2.89 3.48 3.83 4.19 4.60 5.33 5.83

Middle income nations Ecological Footprint

1.23 1.22 1.23 1.28 1.25 1.26 1.29 1.31

Rich nations Ecological Footprint

3.84 4.60 5.36 5.51 5.21 5.73 5.83 6.14

Ecological Footprint
1.17 1.11 1.09 1.07 1.04 0.97 0.96 0.97

Per Capita GDP

911 915 1,041 1064 986 982 1007 1067

Per Capita GDP

2430 2739 3153 3600 3773 3984 4184 4597

Per Capita GDP

10,320 12,701 14,895 16,622 17,810 20,337 21,944 24,656

Notes 1) Figures are period averages. 2) Ecological footprints are per capita g ha. 3) GDP is expressed in $I 2000. 4) Population is in billions.

sustainability), and expressing Eq. (3.2) in per capita terms yields the following identity:   Nt Yt Nenv NY td popt popt

4.

Data

3:3

It is straightforward to demonstrate that a necessary condition for strong sustainability and balanced economic growth is that NY (t) 0 as t approaches infinity. Suppose not, i.e. NY (t) N Y N 0 in the limit, then Eq.(3.3) becomes unbounded: 
tYl

lim

   Nt Y t Nenv NY td lim Yl tYl popt popt

3:4

Thus, strong sustainability and balanced growth require that progressively less natural capital be used per unit of output, a result which clearly echoes Solow (1974), Stiglitz (1974), and Stokey (1998). The only way that the need for natural capital in the production process could subside, of course, is if it is replaced by some other factor of production. While this finding is explicit in the Solow (1974), Stiglitz (1974), and Stokey, it is only implicit in our simple derivation. If we take it as given that natural capital usage has been historically rising, then the required decline in natural capital (if both strong sustainability and balanced growth are to be achieved) gives rise to an inverted U shaped relationship between the per capita levels of output and natural capital. The empirical implications are clear: the Ecological Footprint (EF) must follow an inverted U shaped pattern if both strong sustainability and sustained economic growth are being simultaneously achieved.2

The data used in this analysis include a panel of the Ecological Footprint (a proxy for environmental capital) for 146 countries covering 40 year s (1961 to 2000)3 as defined in our theoretical model, and real per capita GDP (expressed in chain-weighted, PPP-adjusted 2000 international dollars (denoted $I 2000)) from the Penn World Tables v 6.2 (see Heston et al., 2006 for more details).4 The list of countries included in the panel, along with sample-average values of real per capita GDP and the EF are provided in Table 1. The EF is measured in standardized global hectares (g ha) per capita and is constructed by aggregating seven ecological components.5 The first component, built-up land, measures the land area used for buildings and permanent structures and/or eroded and degraded land. The second measure, cropland, measures land area used to cultivate crops consumed by the population, in addition to those fed to poultry and pigs. The third component, fisheries, includes the ocean area used to generate current marine harvests. The fourth component, pasture, includes the land area used to maintain livestock. The fifth and sixth components include, forest land (managed and unmanaged forest), and land area used to produce timber, fuel wood, charcoal, paper, and pulp. Finally, the seventh component captures the environmental consequences of consuming fossil fuels, hydroelectricity, and other renewable energy sources. Specifically, this measure includes the land area required to absorb all CO2 emissions resulting from the direct use of coal, oil, gas, and the indirect use of the consumption of electricity, public transportation, manufactured goods or other services, and is divided between 1) CO2the area used

Note, an empirical Kuznets Curve between EF and per capita GDP is not a sufficient condition for sustainability (i.e. sustainability would require that EF shrink as fast as the population grows and that the level of the EF at an given point in time be below per capita biocapacity).

3 Global Footprint Network (2006). As is common practice in macroeconomic growth studies, we excluded major oil exporting nations from the analysis as such nations' economies are not driven by normal production sectors/industries, by rather by commodity exports. 4 Complete data are available upon request from the authors. 5 Precise definitions of each of the foregoing can be found in Haberl et al. (2001).

1154

E C O L O G IC A L E C O N O M IC S 6 8 ( 2 0 09 ) 11 4 9 1 15 9

to sequester carbon dioxide emissions and 2) Nuclearthe area needed to sequester carbon dioxide emissions from nuclear power plants if they produced emissions at the same rate (per KWH) as traditional fossil-fuel plants.6 The average global EF was 1.7 g ha per capita in the early 1960s, increasing to 2.13 g ha per capita by 2000. During the same time period, average global GDP per capita grew from $3414 to $7065 ($I 2000), and world population from 2.5 to 5.8 billion people (see Table 2). As clearly seen in Table 2, this increase in the global EF over roughly the last half century is the result of middle income and rich nations placing evergreater demands on the environment. Specifically, the EF increased from 1.23 to 1.31 g ha per capita in middle income nations, and ballooned from 3.84 to 6.14 g ha in wealthy nations. In contrast, poor nations experienced a decline in their ecological footprint (albeit small) over the same forty year timeframe.

have to be cut by 50% in order for the EF to display a statistically significant EKC relationship. Further details on these findings follow below.

5.1.

Baseline panel model

5. Environmental Kuznets Curve model and estimation

Following both the EKC and original Kuznets Curve literature, we first estimate a variety of baseline quadratic EKC models using OLS, and later re-estimate the same models using two stage least squares (2SLS) to correct for any endogeneity. The results are surprisingly robust, with only the agricultural component of the EF exhibiting any signs of an inverted U shaped relationship with output.7 Correcting for serial correlation in the baseline model's residuals, we next introduce and estimate a dynamic panel version of the EKC model using the Arellano and Bond (1991) estimation procedure (henceforth AB). Our AB estimation results support the general finding that there is little empirical evidence of an EKC in the ecological footprint or its constituent components. Interestingly, we discover that when energy is removed from the EF, a statistically significant EKC emerges with a turning point of $652 ($I 2000). To determine how much energy consumption would have to be cut in order for the overall EF to be consistent with sustainability, we conduct a sensitivity exercise whereby the energy component of the EF is not completely eliminated from the EF time series, but rather is scaled down by a constant proportion. We find that energy consumption would

To determine the functional form of our empirical model, we draw from both the EKC and original Kuznets Curve literature (e.g. Ahluwalia, 1976; Barro, 2000; among others). A majority of the papers on the EKC follow the original Kuznets Curve literature by including log GDP in quadratic form (see Bimonte, 2002; Mason and Swanson, 2003; Perman and Stern, 2003; Halkos, 2003; Martinez-Zarzoso and Bengochea-Morancho, 2003; Harbaugh et al., 2002 for additional sources); while others add cubic functions of log GDP to test for additional threshold effects (Cole, 2003; List and Gallet, 1999; Rupasingha et al., 2004); and still others add additional control factors such as energy intensity (Agras and Chapman, 1999), population density (Selden and Song, 1994), among others (Cavlovic et al., 2000; Khanna, 2002; Hill and Magnani, 2002). Following the traditional approach, we include log GDP in quadratic form in the following fixed-effects panel model: EFit b1 yit b2 y2 ai gt eit it 5:1

This is an odd component of the EF given that nuclear plants produce no CO2 emissions. The justification commonly given for this accounting rule is that nuclear energy produces radioactive waste, and thus nuclear power production is penalized in the CO2 accounting measure. 7 Critics of the traditional empirical techniques used in the EKC (see for exampleMller-Furstenberger and Wagnerb, 2007) point out that specifying a quadratic relationship between pollution and output imposes strong parametric restrictions and implies that all nations in the panel possess the same turning point. While we acknowledge these criticisms, the use of spline or semiparametric methods would not eliminate the common turning point problem. Moreover, the temporal dimension of our dataset is too short to estimate separate, nation-specific quadratic relationships between output and the EF. Thus, we adopt the commonly accepted practice of estimating a single quadratic relationship.

where EFit is the per capita ecological footprint (g ha) in country i during period t, yit is the log of real per capita GDP ($I 2000), i is a nation-specific fixed-effect, t is a period-specific effect, and it is initially assumed to be an i.i.d. stochastic shock. The unbalanced panel consists of 146 nations spanning eight, 5-year time periods (19611965, 19661970,, 1995 2000). Using standard, fixed-effect OLS estimation methods, we estimate Eq. (3.1) using various measures of the EF and its components (i.e. timber, pasture, etc.).8 One critical issue that must be addressed is the possible bias resulting from endogeneity between the EF and GDP. This stems from the fact that GDP is a function of natural capital (and other aggregate factors of production), and the EF is an imperfect proxy for the use of natural capital. To address this issue, we use two stage least squares (2SLS) estimation with lagged regressors as instruments, which is acceptable given the assumptions of our regression model (i.e. no lagged dependent variables and i.i.d. stochastic errors).9 In order to construct a benchmark for later comparison, we first estimate the EKC and its subcategories with OLS, with the
8

Alternative estimation techniques could have been employed at this stage, including random effects estimation. Random effects estimation may produce more efficient estimates, but carries the cost of greater likelihood of bias/inconsistency (e.g. if the country-specific effects are correlated with output (which they almost surely are), random effect estimation is biased and inconsistent). To demonstrate this, we conduct a Hausman specification test (under the null that the common random and fixed effects coefficient values are identical) and reject the null at any standard level of significance (the chi-squared distributed test statistic is 71.5). While not as efficient, fixed effect estimation is unbiased, consistent, and generally more robust than random effects. 9 In Section 5.2 we will relax this assumption and use a more appropriate dynamic panel model.

E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

1155

Table 3 Baseline Environmental Kuznets Curve estimates Baseline EKC model (5.1) estimates OLS coefficient estimates Dependent footprint variable: regressors
log rgdp log rgdp squared Observations R2 EKC Turning point ($I 2000)

Total
4.520 0.328 904 0.970 No

Built
0.015 0.0003 904 0.982 No

Crop
0.711 0.044 904 0.932 Yes 3028

CO2
3.959 0.280 904 0.944 No

Fish
0.023 0.003 904 0.788 No

Fuelwood
0.187 0.012 904 0.923 No

Pasture
0.193 0.011 904 0.870 No

Timber
0.273 0.021 904 0.971 No

2SLS coefficient estimates

log rgdp log rgdp squared Observations R2 EKC Turning point ($I 2000) 4.190 0.302 765 0.978 No 0.015 0.001 765 0.986 No 0.875 0.056 765 0.948 Yes 2647 3.477 0.247 765 0.962 No 0.006 0.001 765 0.803 No 0.214 0.014 765 0.940 No 0.136 0.007 765 0.905 No 0.057 0.006 765 0.972 No

Notes 1) , , refer to 1%, 5%, and 10% levels of significance respectively. 2) Each regression included time and cross-section dummies. 3) The 2SLS instruments consist of dummies and one-period lags of the log rgdp and log rgdp squared.

results reported in Table 3. Empirical evidence only supports the existence of an EF Kuznets Curve for agricultural land (i.e. Crops), with a turning points of $3028 ($I 2000). Because of potential endogeneity problems between GDP and the EF, Eq.(5.1) is re-estimated using 2SLS.10 The results, also reported in Table 3, change very little. The only EF series with a statistically significant Kuznets Curve relationship is agricultural land (i.e. Crops), with a turning point of $2647 ($I 2000). Thus, there is robust and strong empirical evidence that there exists a Kuznets Curve relationship between crop land and per capita output, but very little evidence in support of a broader Kuznets Curve. A glance at the global production (tones) and harvested area (ha) time series from United Nation's Food and Agriculture Organization (FAO) provides insight into this finding (see Fig. 1). Between 1961 and 2000, world agricultural production grew by an astounding 127%, while harvested area grew by only 20%.

i.i.d. in order to use lagged regressors as instruments). In order to overcome this problem, we introduce a dynamic panel model which explicitly captures the autocorrelation in the EF series: EFit b1 EFit1 b2 yit1 b3 y2 ai gt eit it1 5:2

Following the estimation procedure of Arellano and Bond (1991), unbiased and consistent estimates of the coefficients in Eq.(5.2) are obtained and reported in Table 4. The results are remarkably similar to both the OLS and 2SLS estimates, finding a statistically significant EKC in neither the overall EF series nor any of its components (save pasture and timber,

5.2.

Dynamic panel model

The baseline model (Eq. (5.1) appears to possess serially correlated residuals, as evidenced by Durbin Watson statistics from the OLS regressions. The Durbin Watson statistics range in value from a low of 0.35 (for the pasture component) to a high of 1.65 (for the CO2 component), all of which lie below the lower 5% critical value of 1.89. Endogeneity notwithstanding, the OLS estimates of Eq. (5.1) are still unbiased and consistent despite the presence of serial correlation, while the 2SLS estimates are biased (as we must assume that the error term is

10 One-period lagged values of GDP and GDP squared and the contemporaneous dummies serve as instruments. The lagged regressors are good instruments because GDP is a highly persistent process.

Fig. 1 FAO global production and harvested area time series.

1156

E C O L O G IC A L E C O N O M IC S 6 8 ( 2 0 09 ) 11 4 9 1 15 9

Table 4 Dynamic Environmental Kuznets Curve estimates Arellano and Bond coefficient estimates Dependent footprint variable: lagged regressors
Footprint log GDP log GDP squared Observations EKC Turning point ($I 2000)

Total
0.568 0.452 0.029 628 No

Built
0.415 0.075 0.004 628 No

Crop
0.755 0.057 0.007 628 No

CO2
0.332 1.577 0.108 628 No

Fish
0.448 0.124 0.002 628 No

Fuelwood Pasture Timber

0.769 0.046 0.004 628 No 0.779 0.075 0.004 628 Yes 8153 0.127 0.165 0.013 628 Yes 656

Notes 1) , , refer to 1%, 5%, and 10% levels of significance respectively. 2) GDP and GDP squared measured in real terms. 3) Each regression includes period dummies.

which have turning points of $8153 and $918 ($I 2000) respectively).11 Returning to the Total EF series, it appears that most of the growth in that series is due to energy consumption (CO2 accounts for up to 50% of the EF measurement). Interestingly, log GDP and log GDP squared are found to be significant, but in the opposite direction that the EKC would predict. In other words, the EF is found to be increasing at an increasing rate with growth and development. To illustrate this relationship, Fig. 2 plots the overall Total EF series for rich, middle income, and poor nations.12 Clearly, the overall EF increased sharply in rich countries (by approximately 60%), and increased more mildly in middle income nations (by approximately 7%), while poor nations' actually experienced a 17% decline in total EF between 1961 and 2000. This contrasts sharply with total EF when the energy series are removed, as reported in Fig. 3. In this case, all three income groups experienced a near continuous decline in the EF between 1961 and 2000. Rich nations total EF (less energy) declined 18%, while middle income and poor nations experienced a 26% and 27% decline respectively. The growth in energy consumption therefore explains the large increase in the ecological footprints in both rich and middle income countries. Consequently, the dynamic EKC model (Eq. (5.2)) is re-estimated using total EF (less energy) as the dependent variable (Table 5). According to the regression results, there is a EKC relationship between total EF (less energy) and per capita GDP, with a turning point of $652 ($I 2000). As of 2003, 176 of 182 (or 97%) of nations had per capita GDP levels in excess of this threshold, which is consistent with the overall downward trend over the past 40 years in this series across all income groups.13

interesting exercise would be to diminish the importance of energy in the EF and see if doing so would generate a traditionally shaped EKC. To accomplish this we, construct a new counterfactual dependent variable, denoted EF : it EFk uEFit it
energy

kEFit

energy

ka0; 1

5:3

where EF energy is the total ecological footprint less the energy it components, and EF energy consists of the energy components it of the ecological footprint. The scale factor, , is varied from 0.10 to 0.90, in increments of 0.10, producing nine footprint series (i.e. EF0.10, EF0.20,, EF0.90). Employing each of the nine it it it foregoing EF series, model(5.2) is re-estimated using AB methodology. According to the results, provided in Table 6, energy consumption levels would have to be cut by 50% across the board before a statistically significant EKC emerges from the estimation process. In other words, even if the impact of energy use was overstated in the EF by 100%, the EF would not conform to the traditional EKC hypothesis. The estimated turning points are also consistent, ranging from a low of $862 to $955 ($I 2000) of real per capita GDP. This is graphically represented in Fig. 4, which plots the estimated relationship

5.3.

Energy sensitivity analysis

It is clear that energy use dominates the measure of ecological footprint, subjecting the measure to potential criticism. An
11 The Crop series was marginally significant, with a negative and statistically significant coefficient on log GDP squared, but an insignificant coefficient on log GDP. 12 The rich nations consist of those nations in the top third of the global income distribution over the entire time span of the panel. Likewise, the middle income and poor nations consist of the nations in the middle-third and bottom third of the global income distribution over the entire time span of the panel. 13 Based on Penn World v 6.2 data , see Heston et al. (2006).

Fig. 2 Total footprint by income group. Note: The periods are 5 years in length, beginning with period 1 which spans the years 1961 to 1965, and continuing in this manner ends with period 8, which spans the years 1996 to 2000.

E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

1157

Table 6 Energy sensitivity analysis Energy sensitivity analysis Lamda () Arellano and Bond estimation model (5.2) coefficients log GDP
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.251 0.270 0.266 0.245 0.203 0.129 0.021 0.115 0.262

EKC

Turning point

log GDP squared

0.019 0.020 0.019 0.018 0.015 0.010 0.002 0.007 0.016 Yes Yes Yes Yes Yes No No No No 862 962 971 981 955

Fig. 3 Total footprint net of energy consumption by income group. Note: The periods are 5 years in length, beginning with period 1 which spans the years 1961 to 1965, and continuing in this manner ends with period 8, which spans the years 1996 to 2000.

Notes 1) , , refer to 1%, 5%, and 10% levels of significance respectively. 2) Lagged dependent variable and period dummy estimates not reported.

between EF0.10, EF0.50 and EF0.90 and log per capita output using the coefficient estimates provided in Table 6.

6.

Conclusions

Departing from the current Environmental Kuznets Curve literature which typically uses pollution as a measure of environmental quality, this paper empirically investigates the EKC using an aggregate index of environmental degradation called the Ecological Footprint. We find no evidence of an EKC relationship between per capita output and the EF or any of its subcomponents, with the exception of land in agriculture, and to a lesser extent land used in pasture or for timber. The EF is found to increase at an increasing rate for both rich and poor nations on average, with much of the increase attributed to energy use (i.e. CO2). When the energy components of the EF are removed, the resulting series yields a statistically signifi-

cant EKC with a turning point of approximately $652 ($I 2000). More importantly, we find that the energy use component must be discounted by a full 50% before a traditional EKC is found. The literature has found evidence of an EKC relationship between pollution and economic development, suggesting that comprehensive environmental policy is not necessary for developing nations, as growth is expected to improve environmental quality over time. Our empirical results show that if a broader measure of the loss of environmental capital is used, there is little evidence of an EKC. Going one step further, we find that energy consumption is, by and large, the major culprit behind this result, and that substantial cuts in CO2 emissions are necessary to bring about an EKC relationship between the ecological footprint and economic growth. Together these results suggest that growth alone will not serve as a solution to environmental problems. Instead, the

Table 5 Kuznets Curve Estimates for EF less energy components Dependent variable: total EF less energy components Regressors
log GDP log GDP squared Observations EKC Turning point ($I 2000)

Arellano and Bond coefficient estimates Model 5.2

0.207 0.016 628 Yes 652

Notes 1) , , refer to 1%, 5%, and 10% levels of significance respectively. 2) GDP and GDP squared measured in real terms. 3) Model 5.2 uses one-period lags of the reported regressors. 4) Coefficient estimates of the lagged dependent variable not.

Fig. 4 Estimated relationship between the EF and log per capita output.

1158

E C O L O G IC A L E C O N O M IC S 6 8 ( 2 0 09 ) 11 4 9 1 15 9

impacts of energy usage must be included in development policy if sustainability is to be achieved.

REFERENCES
Ahluwalia, M., 1976. Income distribution and development: some stylized fact. American Economic Review 66, 128135. Agras, Jean, Chapman, Duane, 1999. A dynamic approach to the Environmental Kuznets Curve hypothesis. Ecological Economics 28, 267277. Arellano, Manuel, Bond, Stephen, 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies 58, 277297. Ayres, Robert U., 2000. Commentary on the utility of the ecological footprint concept. Ecological Economics 32, 347349. Azomahou, Theophile, Laisney, Francois, Nguyen Van, Phu, 2006. Economic development and CO2 emissions: a nonparametric panel approach. Journal of Public Economics 90 (67), 13471363. Barbier, Edward, 2005. Natural Resources and Economic Development. Cambridge University Press, Cambridge, UK. Barbier, Edward B., 2001. The economics of tropical deforestation and land use: an introduction to the special issue. Land Economics 77 (2), 155171. Barro, R., 2000. Inequality and growth in a panel of countries. Journal of Economic Growth 5, 532. Bimonte, Salvatore, 2002. Information access, income distribution, and the Environmental Kuznets Curve. Ecological Economics 1, 145156. Boutaud, Aurlien, Natacha, Gondran, Christian, Brodhag, 2006. Local environmental quality versus (global) ecological carrying capacity: what might alternative aggregated indicators bring to the debates about Environmental Kunzites Curves and sustainable development. International Journal of Sustainable Development 9 (3), 297310. Bruvoll, Annegrete, Taran, Faehn, Birger, Strom, 2003. Quantifying central hypotheses on Environmental Kuznets Curves for a rich economy: a computable general equilibrium study. Scottish Journal of Political Economy 50 (2). Cabeza Gutes, Maite, 1996. The concept of weak sustainability. Ecological Economics 17 (3), 147156. Cavlovic, Therese A., Baker, Kenneth H., Berrens, Robert P., Gawande, Kishore, 2000. A meta-analysis of Environmental Kuznets Curve studies. Agricultural and Resource Economics Review 29 (1), 3242. Cole, Matthew, 2003. Development, trade, and the environment: how robust is the Environmental Kuznets Curve? Environmental and Development Economics 8, 557580. Copeland, Brain R., Taylor, M. Scott, 2004. Trade growth and the environment. Journal of Economic Literature 42, 771. Costanza, Robert, 2000. The dynamics of the ecological footprint concept. Ecological Economics 32, 341345. Costanza, R., Daly, H.E., 1992. Natural capital and sustainable development. Conservation Biology 6 (1), 3746. Culas, R.J., 2007. Deforestation and the environmental Kuznets curve: an institutional perspective. Ecological Economics 61 (2), 429437. Dasgupta, Susmita, Laplante, Beniot, Wang, Hua, Wheeler, David, 2002. Confronting the Environmental Kuznets Curve. Journal of Economic Perspectives 16 (1), 147168. Deacon, Robert T., Norman, Catherine S., 2006. Does the Environmental Kuznets Curve describe how individual countries behave? Land Economics 82 (2), 291315. Francheschi, Dina, Kahn, James R., 2003. Beyond strong sustainability. International Journal of Sustainable Development and World Ecology 10, 211-200.

Galeotti, M., Lanza, A., Pauli, F., 2006. Reassessing the environmental Kuznets curve for CO2 emissions: a robustness exercise. Ecological Economics 57 (1), 52163. Gawande, Kishore, Berrens, Robert, Bohara, Alok, 2001. A consumption based theory of the Environmental Kuznets Curve. Ecological Economics 37, 101112. Global Footprint Network (2006) National Footprint and Biocapacity Accounts, URL: www.footprintnetwork.org, accessed May 2007. Grossman, Gene M., Krueger, Alan B., 1993. Environmental impacts of a North American Free Trade Agreement. In: Garber, Peter M. (Ed.), The U.S.Mexico Free Trade Agreement. MIT Press, Cambridge MA, pp. 1356. Grossman, Gene M., Krueger, Alan B., 1995. Economic growth and the environment. Quarterly Journal of Economics 110 (2), 353377. Haberl, Helmut, Erb, Karl-Heinz, Krausmann, Fridolin, 2001. How to calculate and interpret ecological footprints for long periods of time: the case of Austria 19261995. Ecological Economics 38, 2545. Halkos, Geroge, 2003. Environmental Kuznets Curve for sulfur: evidence using GMM estimation and random coefficient panel data models. Environment and Development Economics 8, 581601. Harbaugh, William T., Levinson, Arik, Wilson, David Molloy, 2002. Rethinking the empirical evidence for an Environmental Kuznets Curve. Review of Economics and Statistics 84 (3), 541551. Hediger, Werner, 1999. Reconciling weak and strong sustainability. International Journal of Social Economics 26 (7/8/9), 11201143. Heerink, Nico, Mulatu, Abay, Bulte, Erwin, 2001. Income inequality and the environment: aggregation bias in Environmental Kuznets Curves. Ecological Economics 38, 359367. Heston, A., Summers, R., Aten, B., 2006. Penn World Table Version 6.2, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania. URL: http://pwt.econ.upenn.edu/, accessed May 2007. Hill, Robert, and Magnani, Elisabetta (2002) An Exploration of the Conceptual and Empirical Basis of the Environmental Kuznets Curve, Blackwell Publishers Ltd/ University of Adelaide and Flinders Univeristy of South Australia. Hong, Li, Pei Dong, Z., Chunyu, H., Gang, Wang, 2007. Evaluating the effects of embodied energy in international trade on ecological footprint in China. Ecological Economics 62 (1), 136148. Jha, Raghbendra, Murthy, K.V. Bhanu, 2003. An Inverse Global Environmental Kuznets Curve. Journal of Comparative Economics 31, 352368. Khanna, Neha, 2002. The income elasticity of non point source air pollutants: revisiting the environmental Kuznets curve. Economic Letters 77, 387392. Kuznets, S., 1955. Economic growth and income inequality. American Economic Review 45, 128. Lenzen, Manfred, Borgstrom Hansson, Carina, Bond, Stuart, 2007. On the bioproductivity and land-disturbance metrics of the Ecological Footprint. Ecological Economics 61 (1), 610. List, John, Gallet, Craig, 1999. The Environmental Kuznets Curve: does one size fit all. Ecological Economics 31, 409423. Malthus, T.R., 1789. An Essay on the Principle of Population. J. Johnson. URL: http://www.econlib.org/library/Malthus/malPop1.html, accessed May 2007. Martinez-Zarzoso, Inmaculada, Bengochea-Morancho, Aurelia, 2003. Testing for an Environmental Kuznets Curve in Latin American countries. Revista De Analisis Economico 18, 326. Mason, Robin, Swanson, Timothy, 2003. A Kuznets Curve analysis of ozone-depleting substances and the impact of the Montreal Protocol. Oxford Economic Papers 55, 124. Merlevede, B., Verbeke, T., De Clercq, M., 2006. The EKC for SO2: does firm size matter? Ecological Economics 59 (4), 451461. Mller-Frstenberger, Georg, Wagnerb, Martin, 2007. Exploring the environmental Kuznets hypothesis: theoretical and econometric problems. Ecological Economics 62 (34), 648660. Nijkamp, Peter, Rossi, Emilia, Vindigni, Gabriella, 2004. Ecological footprints in plural: a mete-analytic comparison of empirical results. Regional Studies 38 (7), 747765.

E CO L O G I CA L EC O NO M IC S 6 8 (2 0 0 9) 1 14 9 1 15 9

1159

Patterson, T.M., Niccolucci, V., Bastianoni, S., 2007. Beyond more is better: ecological footprint accounting for tourism and consumption in Val di Merse, Italy. Ecological Economics 62 (3), 747756. Paudel, Krishna P., Zapata, Hector, Susanto, Dwi, 2005. An empirical test of Environmental Kuznets Curve for water pollution. Environmental and Resource Economics 31 (3), 325348. Perman, Roger, Stern, David I., 2003. Evidence from panel unit root and cointegration tests that the Environmental Kuznets Curve does not exist. Australian Journal of Agricultural and Resource Economics 47 (3), 325347. Plassmann, Florenz, Khanna, Neha, 2006. Household income and pollution: implications for the debate about the Environmental Kuznets Curve hypothesis. Journal of Environment and Development 15 (1), 2241. Ranjan, R., Shortle, J., 2007. The environmental Kuznets curve when the environment exhibits hysteresis. Ecological Economics 64 (1), 204215. Rees, W.E., 1992. Ecological footprints and appropriated carrying capacity: what urban economics leaves out. Environment and Urbanization 4, 121130. Rees, W.E., 2000. Eco-footprint analysis: merit and brickbats. Ecological Economics 32, 371374. Rodriguez-Meza, Jorge, Southgate, Douglas, Gonzalez-Vega, Claudio, 2003. Rural poverty, household response to shocks, and agricultural land use: panel results for El Salvador. Department of Agriculture, Environmental and Developmental Economics, Ohio State University. Sept. 2003. Rupasingha, Anil, Goetz, Stephan J., Debertin, David L., Pagoulatos, Angelos, 2004. The Environmental Kuznets Curve for US counties: a spatial economic analysis with extensions. Papers in Regional Science 83, 407424. Selden, Thomas M., Song, Daqing, 1994. Environmental quality and development: is these a Kuznets Curve for air pollution emissions. Journal of Environmental Economics and Management 27 (2), 147162. Shafik, Nemut, Banyopadhyay, Sushenjit, 1992. Economic growth and environmental quality: time series and cross sectional evidence. Work Bank Paper, vol. 904.

Solow, R.M., 1974. Intergenerational equity and exhaustible resources. Review of Economic Studies 41, 2945. Stern, David I., 1998. Progress on the Environmental Kuznets Curve? Environment and Development Economics 3 (2), 173196. Stern, David I., 2004. The rise and fall of the Environmental Kuznets Curve. World Development 32 (8), 14191439. Stiglitz, J., 1974. Growth with exhaustible natural resources: efficient and optimal growth paths. Review of Economic Studies 41, 123137. Stokey, N.L., 1998. Are there limits to growth. International Economic Review 39, 131. Torras, M., Boyce, J.K., 1998. Income, inequality, and pollution: a reassessment of the environmental Kuznets Curve. Ecological Economics 25 (2), 147160. Van Kooten, G.C., Bulte, E.H., 2000. The ecological foot print: useful science or politics? Ecological Economics 32, 385389. Wackernagel, Mathis, Rees, William, 1996. Our Ecological Footprint, Reducing Human Impact on the Earth. New Society Publishers, Gabriola Island, Philadelphia. Wackernagel, Mathis, Monfreda, Chad, Schulz, Niels B., Erb, Karl-Heinz, Haberl, Helmut, Krausmann, Fridolin, 2004a. Calculating national and global ecological footprint time series: resolving conceptual challenges. Land Use Policy 21, 271278. Wackernagel, Mathis, Monfreda, Chad, Erb, Karl-Heinz, Haberl, Helmut, Schulz, Niels B., 2004b. Ecological footprint time series of Austria, the Philippines and South Korea for 19611999: comparing the conventional approach to an actual land area approach. Land Use Policy 21, 261269. White, T.J., 2007. Sharing resources: the global distribution of the Ecological Footprint. Ecological Economics 64 (2), 402410. Wiedmann, T., Lenzen, M., 2007. On the conversion between local and global hectares in Ecological Footprint analysis. Ecological Economics 60 (4), 673677. Wiedman, Thomas, Minx, Jan, Barrett, John, Wackernagel, Mathis, 2006. Allocating ecological footprints to final consumption categories with inputoutput analysis. Ecological Economics 56 (1), 2848.