Statistica Applicata - Italian Journal of Applied Statistics Vol.

24 (2) 173

DOMESTIC MATERIAL CONSUMPTION INDICATOR AND

NATURAL RESOURCES: A EUROPEAN ANALYSIS OF THE
ENVIRONMENTAL KUZNETS CURVE

Sabrina Auci
Department of European Studies and International Integration, University of
Palermo, Italy

Donatella Vignani1
Social and Environmental Statistics Department, Istat Italian National Institute of
Statistics, Italy

Received: 15 May 2013/ Accepted: 7 April 2014

Abstract. The study investigates the relation between per capita Domestic Material
Consumption indicator (DMC) and per capita income. Economic literature focuses mainly
on air, water and land pollution while we consider as environmental degradation the
consumption on natural resources extracted from the environment. Using a cross–
European panel of countries over the period 2000-2011, our results confirm the absence
of an Environmental Kuznets Curve (EKC) between per capita DMC and per capita GDP
both for EU-27 vs. 30 European countries and for Western vs. Eastern European countries.
The turning points are so high that it is present a monotonic increasing relation between
DMC indicator and GDP.
Keywords: Natural resource use, Environmental degradation, Environmental Kuznets curve,
Domestic material consumption and economy-wide material flows accounts.

1. INTRODUCTION
The increasing awareness about global warming, air and water pollution and other
environmental disasters is becoming the main issue for both people and policy
makers in these recent years. The stress on the environment has been carried on by
developed countries because of their industrial structure and by developing countries
because of the increase of their per capita income levels. The effects of the
production and consumption activities on the environment could be subdivided

1 Donatella Vignani, email: vignani@istat.it

174 Auci S., Vignani D.

into: i) economic effects; ii) social effects; iii) environmental effects. Country
actions regard mainly the improvement of natural resource management or the
adoption of sustainable development strategies.
The Environmental Kuznets Curve (EKC) analysis is the most important
empirical result within the end-of-pipe approach. The EKC relationship is based on
a rather intuitive assumption. In early stages of industrialization, pollution grows
rapidly while in the later stage, after a turning point, pollution declines as the level
of GDP increases, as the relationship between income inequality and economic
development described by Kuznets (1955). For comprehensive overviews of
theoretical and empirical studies on the EKC the reader can refer to Panayotou
(2000), Borghesi (2001), Dinda (2004), and Kijima et al. (2010). The EKC
literature does find different patterns for the same pollution indicators (see among
others Shafik and Bandyopadhyay, 1992; Grossman and Krueger, 1995; Selden and
Song, 1995; Azomahou et al., 2006; Vollebergh et al., 2009; Musolesi et al., 2010;
Ordás Criado et al., 2011; Mazzanti and Musolesi, 2013a and 2013b).
Even if early authors find some evidence on the existence of an inverted-U-
shaped pattern of the relationship between per capita income and emissions, some
recently studies have criticised the reduced-form of the EKC model. For example
Cole et al. (1997) have underlined that the reduced-form relationships reflect
correlation rather than causal mechanism. Due to the feedback effect of environmental
quality on income growth, Stern et al. (1996) and Pearson (1994) have concluded
that reduced–form models are not appropriate to describe the mechanism of
environmental degradation in terms of income.
Other authors, instead, are sceptical about the existence of EKC hypothesis
because of the lack of adequate econometric techniques used. The analysis of unit
root test, cointegration and related error correction model is implemented by Dinda
and Coondoo (2006), Auci and Becchetti (2006) and Lee and Lee (2009) while the
recent analyses of Mazzanti and Musolesi (2013a) and (2013b) focus on the
simultaneously handle of three econometric issues: functional form bias,
heterogeneity bias and omitted time related factors bias.
However, as underlined by Brock and Taylor (2010), economic growth and
pollution are closely related. In fact, for the growth of an economy, more inputs and
more natural resources are required in the production process. As a consequence,
more pollution and negative effects for the environment are yielded. As Azomahou
et al. (2006) underline, the study of this relationship is relevant for the definition of
an appropriate joint economic and environmental policy.
Having in mind all these critiques of the ECK literature, our paper is close in
spirit to the seminal works of Grossman and Krueger (1995) on the EKC analysis.
We consider this relationship by not controlling for possible determinants like
Domestic material consumption indicator and natural resources: A European … 175

technological change, policy events, etc. Even if we do not deny the role of these
determinants, our choice is based on comparability issue with the seminal studies.
The originality of our study, in fact, is in the use of a different indicator, drawn by
the Economy-Wide Material Flow Accounting (EW-MFA) belonging to Satellite
Accounts System of National Accounting (EUROSTAT, 2001; 2002). This indicator,
in line with result-oriented approach, putting emphasis on preventive and integrated
policies, can: i) provide an integrated view of the material used; ii) capture indirect
flows not embodied in products; iii) understand the effects on global environment
of material flows among countries.
We hope that using this comprehensive indicator we could deepen the
understanding of the ‘driving forces’ behind environmental burden. Within the EW-
MFA indicators, we have focused our attention on Domestic Material Consumption
(DMC) indicator that measures the total amount of natural resources directly used
in production and consumption processes of an economic system. This indicator
should be considered as a proxy for ‘potential environmental pressure’. Knowing
the methodological aspects, we are aware of the limits of the DMC, such as the high
level of aggregation, covering important disparity within its items.
An advantage of our indicator is related to consumers’ pressure. It is
undeniable that environmental degradation can be affected by consumers’ behaviour
both directly and indirectly and in EKC literature several authors, Stern et al.
(1996), Ekins (1997), Rothman (1998) and Suri and Chapman (1998), have
emphasised the consumption–based approach as the more appropriate measure of
global environmental impact.
Our empirical analysis is just another example within the EKC literature.
Using a cross–European panel of countries over the period 2000-2011, the aim of
our paper is to verify whether this measure of potential environmental pressure can
follow an EKC-like path among European countries. Given the relative lack of
empirical studies on the relationship between material consumption indicators and
the level of per capita GDP, our purpose is to delineate some stylized facts through
the use of a simple empirical model, whose results should highlight important
issues that merit further investigations.
Our indicator belongs to the family of sustainability indicators based on a
holistic approach. DMC related to GDP have been widely used in the decoupling
analysis. Decoupling2 occurs when the growth rate of an environmental pressure is

2 In literature, a distinction is often made between absolute and relative decoupling. In a

growing economy, relative decoupling means a positive environmental pressure growth rate
– in terms of resources required – but less than the growth rate of GDP. Absolute decoupling
is said to occur when environmental pressure growth rate is decreasing while GDP is
increasing.
176 Auci S., Vignani D.

less than GDP growth rate over a given period. Our results confirm a slight presence
of relative decoupling among European countries considered.
In our empirical analysis, an inverted–U shaped relationship between per
capita DMC indicator and per capita GDP within the EU-27 countries, the EU-30
enlarged to Norway, Switzerland and Turkey and the two macro-areas (West and
East of Europe) is not present. The results are the same both in the panel estimations
and in the robustness estimations. The turning points are so high that there is a
monotonic increasing relation between per capita DMC indicator and per capita
GDP as in the case of per capita CO2 emissions and per capita GDP (Holtz-Eakin
and Selden, 1995; Roberts and Grimes, 1997; Cole et al., 1997). These results are
in line with Arrow et al. (1995) who have emphasized that EKCs for global
pollutants with long-term costs and perhaps for some resource stocks have tended
to be monotonically increasing.
The remaining of the paper is organized as follows. In the second section, we
review Physical Measures of Environmental Impact developed by official statistics.
The third section underlines some stylised facts about the DMC indicator in order
to better understand its components and trend. In the fourth section, we describe the
empirical model and illustrate the main econometric findings, while in the fifth we
introduce the sensitivity analysis. Finally, in the last section, we conclude.

2. DOMESTIC MATERIAL CONSUMPTION INDICATOR

In the last fifteen years, official statistics have developed some indicators as
physical measures of potential environmental impact to better understand and
measure consequences of economic growth. Our analysis is based on Domestic
Material Consumption (DMC) included in the construction of Total Material
Requirement (TMR) indicator, that nowadays has gained in importance at
international level within European countries (EUROSTAT, 2002).
A satellite accounting system has been adopted in the last decade in which a
system of accounts are dedicated to measure raw resources extracted by nature and
reintroduced as waste materials. All items are expressed in units of weight. Within
Material Flow Accounts (MFA), EW-MFA is now (with air emission and
environmental taxation accounts) one of the three top-priorities in European official
statistics’ environmental accounting, which are object of the EU Regulation n. 691/
2011. Italy has very actively contributed to international statistic developments
through Istat3 (Femia and Vignani, 2005; 2010).

3 In the framework of EW-MFA, Istat compiles accounts and derived indicators on yearly basis:
the time series 1980-2011 are published in www.istat.continazionali
Domestic material consumption indicator and natural resources: A European … 177

EW-MFA is the methodology for the construction of a system of accounts

dedicated to measure natural resources based on a holistic approach. This system
gives synthetic information on physical exchanges of an economy and on potential
environmental pressures (Adriaanse et al, 1997). The anthropic system is looked at
as a living organism whose activities need flows of natural materials extracted by
the environment (inputs) to be processed in many ways such as manufacture, use
reuse and recycling, accumulated in stocks or returned back to natural environment
in an altered shape (outputs). Measuring inputs to socio-economic functioning
gives an indication of all potential “pressures” to nature. The EW-MFA analysis can
be carried out at different levels of detail, according to the set of activities and the
type of materials4. Compared to other methodologies EW-MFA offers three
advantages: i) certainty of the basis in units of weight; ii) accounting of actual
phenomena; and iii) use of the main national accounting concepts.
Similarly to national accounts, in EW-MFA accounts the cycle of matter is
divided into a series of steps, each illustrated by an account. The accounts are
structured in waterfall and are all linked together, as to show the relationships
among various measures of resources use. Material-flow-based indicators provide
an aggregate picture by describing the evolution of the demand for natural resources
by a country.
Among the EW-MFA indicators, highly significant is Total Material
Requirement (TMR) that accounts for all material flows required at global level to
satisfy domestic and foreign final demand of domestic and foreign products
(Bringezu and Schütz, 2001). TMR, indicating the Total Material Base, is obtained
by adding together the following items, all expressed in units of weight: i) Domestic
Extraction of Materials Used (DEU): all materials extracted from natural environment
in a country; ii) Imports of raw materials and goods; iii) Domestic Extraction of
Materials Unused (DEUnused): all materials flows extracted from nature
intentionally but not used in final products i.e. materials from mining and quarrying
activities; iv) Indirect Flows Associated to Imports (IF): quantities of materials,
used and unused, removed from other countries’ nature in order to produce goods
imported by a country. IFs account materials not actually embodied in goods,
because they are transformed in waste and emissions during the production process
carried out abroad.

4 Due to heterogeneous, fragmentary and low-quality statistical data and metadata, Istat
environmental accounting experts should have had reduce to coherence a large set of data
(Costantino, Femia and Vignani, 2008, Femia and Vignani, 2010).
178 Auci S., Vignani D.

As this paper intends to analyse a panel of European countries by using EKC

methodology, the lack of data on TMR forces to use another EW-MFA indicator
DMC. Thus, we focus on DMC indicator because it is available for a large number
of countries and years and is also one of the indicators used to assess the
achievement of EU policy targets.
DMC measures the total amount of materials directly used by an economy and
is defined as the annual quantity of raw materials extracted from domestic territory
of an economy (Domestic Extraction Used) plus all physical imports minus all
physical exports (Physical Trade Balance). This consumption-based indicator
refers to the domestic use of natural resources and provides an assessment of
consumption driven by domestic demand and by the export market. DMC can be
seen as an indicator reflecting all materials that physically remain in a country,
emitted by or accumulated in a given region (Bringezu and Schütz, 2001). All these
materials will eventually turn into emissions and wastes, thus the value of DMC can
also measure the potential waste of a given country (Moll et al., 2003). DMC and
TMR are only rough proxies for measuring the overall environmental potential
impact of resource use, as materials have very different impacts on the environment.
In this framework, EW-MFA is perceived as a very relevant decision tool by
policy makers. For example, both EU Sustainable Development Strategy (EU-
SDS) and EUROPE 2020 Strategy are based on DMC and a set of sustainable
development indicators. Defining the targets to put the EU on a path of sustainable
development, the EU-SDS puts incentives in managing efficiently natural resources,
in protecting the environment and in promoting social cohesion. The Europe 2020
strategy, putting emphasis on the efficient use of resources, promotes the target of
a low carbon economy.
EUROSTAT and all national statistical institutes have a leading role to play
in providing indicators and monitoring to what extent the EU is on track to achieve
the concrete goals for a sustainable growth based on a resource efficient economy.
The EU-SDS indicators are more than one hundred and have been organised within
a theme-oriented framework. One of them called “sustainable consumption and
production” contains the sub-theme “resource use and waste” whose headline
indicator is Resource Productivity (GDP/DMC) that measures efficiency in the use
of natural resources in an economic system.
All these considerations strengthen the selection of DMC as an appropriate
measure to analyse the relationship between environment pressures and economic
activities by using the EKC methodology. This type of indicator helps to underline
the change of the environment in which man lives due to production-consumption
activities.
Domestic material consumption indicator and natural resources: A European … 179

3. DESCRIPTIVE EVIDENCE
Before discussing our empirical results, we present some stylised facts on the DMC
indicator and GDP in order to better understand its components, trend and its
relationship with national production. In Figure 1, the first figure on EU-27 shows
the two components of DMC. The EU–27’s DEU is the first component because on
average accounts for the 85% of DMC. In fact, in 2010 the DEU value is equal to
5,937 million of tonnes. Within DEU, non-metallic minerals – including sand,
gravel and other inert materials – represent about the 56% of total with 3,343 million
of tonnes. The importance of construction and infrastructure activities can be easily
seen. Biomasses represent about the 27% of total while the remaining fossil energy
materials and metal ores extracted are respectively 14% and 3%. DEU’s trend
shows a decrease of about 3% from 2000 to 2003, while it raises until to 2007
reaching more than 6,967 million of tonnes (+6.9% compared to 2000) and finally,
in the last period (2007-2010) DEU diminishes by almost 14.7%.
As regard PTB, it presents a positive value in the period observed (2000-
2010), the EU–27 can be considered as a net importer of materials. PTB raises
almost constantly from 2000 to 2007 of about 27%, so it reaches the highest value
of the period observed just in 2007 when foreign imports of materials exceed
exports by 1,306 million of tonnes. Natural resources and products, needed to
satisfy EU-27 countries’ demand, are replaced by foreign sources.
Thus, as a whole, DMC undergoes a slight decreasing only 1.5% from 2000
to 2003. After 2003 it starts rising until 2007 up to almost 8,273 million of tonnes,
that represents an overall increase just over 9.5% compared to 2000. A structural
break takes place between 2008 and 2009 due to the financial and real crisis that
reduces production and consumption activities and domestic demand of natural
resources.
Focusing on the three NO–EU–27 countries - Norway, Switzerland and
Turkey - their DMC is available only for the period 2000-2009. The overall value
of DMC shows that the component PTB remains negative, meaning that aggregated
exports of material flows overcome aggregated imports by an annual average of
more than 100 million of tonnes (Figure 1). This is true especially for Norway and
Turkey but not for Switzerland whose PTB is positive in all years. The negative
value of PTB component contributes to reduce DMC in each year. However, DMC
shows a positive trend from 2000 to 2009 with an overall increase of 29%. In 2009
it reaches the highest value of 1,122 million of tonnes of natural resources.
180 Auci S., Vignani D.

Figure 1: Domestic Material Consumption (DMC) by components, EU–27 and NO-EU-27

(million tonnes)
Source: data processed by EUROSTAT database
Note: a) Eurostat dataset does not present data available for Norway and Turkey in year 2010 and
2011. b) For all three countries PTB is negative, even if Switzerland PTB is positive in all years.

As far as a dynamic comparison between per capita DMC (DMCpc) and per
capita GDP (GDPpc) of the EU-27 countries is concerned, as in Figure 2, only from
2000 to 2003 DMCpc seems to be relative decoupled from GDPpc (respectively
4.5% and +5%). Starting from 2003 up to 2007 GDPpc increases more quickly than
in the previous four years (+9.9%) and DMCpc shows a similar rise (+8.4%).
Figure 2 shows a structural break for the EU-27 countries in the DMCpc series
from 2007 onwards. The decreasing of DMCpc and GDPpc takes place (respectively
-19% and -4.7%) and a relative decoupling occurred. In absolute values, DMCpc,
that exceeds 20 tonnes of material flows in 2007, comes down to just over 16 tonnes
in 2010 while GDPpc changes from 20,973€ to about 20,163€. For the three NO-
EU countries a similar trend for GDPpc occurs. As regards DMCpc, in the period
2000-2003, is evident a decoupling with respect to GDPpc. What is interesting to
highlight is that in 2001 DMCpc decreases (-13% over the previous year). However,
Domestic material consumption indicator and natural resources: A European … 181

Figure 2: Per capita Domestic Material Consumption (DMC) and Gross Domestic Product
(GDP), years 2000-2010, index 2000=100
Source: data processed by EUROSTAT dataset

starting from 2003 up to 2009 there is a continuous increasing of DMCpc that

reaches almost 20 tonnes of materials, with an overall rise of 30% in the period
considered.

4. EMPIRICAL MODEL AND RESULTS

In our econometric analysis, we estimate a reduced form model, which, following
Auci and Becchetti (2006), we call the unadjusted EKC-like curve. As in the
previous analysis, we estimate the simple relationship between an environmental
indicator DMC and GDP. Drawing an unbalanced panel dataset of 30 European
countries by the European Statistical Office (EUROSTAT) dataset, whose data are
collected yearly from 2000 to 2011, we compare both the EU27 countries with
respect to the 30 European countries and the west European countries with respect
to the east European countries, using fixed and random effect panel models with
proper diagnostics.
The simple EKC curve is represented by a polynomial approximation in
logarithmic terms:

ln( E / P )it = α i + γ t + β1 ln(GDP / P )it + β 2 (ln(GDP / P ))i2t + ε it

(1)
with β1 > 0 and β 2 < 0
182 Auci S., Vignani D.

where E stands for domestic material consumption indicator, P represents total

population and GDP is gross domestic production. The subscripted indices i and t
indicate, respectively, countries and years while with αi and γt-intercepts we can
measure, respectively, country specific time invariant effects and time effects.
From this specification the turning point of income at which per capita
resource use is at its maximum level is easily derived as:

GDPmax = exp(− β1 / (2β 2 )) (2)

where β1 and β2 are the parameters of levels and square of per capita GDP in
Equation (1).
We are aware that this simple specification has been extensively criticised on
econometric and theoretical grounds, but our aim is to verify if this new indicator
shows a relation akin to the EKC curve. However, to improve our simple empirical
analysis, we conduct a sensitivity analysis.
Table 1 reports descriptive statistics of all variables included into the estimation
model. Descriptive statistics are reported not only for the whole data in the dataset
but also distinguishing between EU27 countries and NO–EU27 countries such as
Norway, Switzerland and Turkey and between western and eastern European
countries.

Table 1: Descriptive statistics of estimation variables - 2000–2011

Variable Mean SE(Mean) CV P25 P50 P75 N IQR
OVER ALL DMCpc 18.20 0.44 0.45 12.70 16.50 21.60 355 8.91
GDPpc 19,541 721 0.70 6,427 16,290 28,974 360 22,547
NO–EU27# DMCpc 18.03 2.07 0.64 10.90 12.01 34.78 31 23.87
GDPpc 29,646 2,925 0.59 6,014 40,239 42,645 36 36,630
EU27 DMCpc 18.25 0.44 0.43 13.53 16.77 21.59 324 8.06
GDPpc 18,418 707 0.69 6,459 15,823 27,690 324 21,231
WEST DMCpc 19.56 0.67 0.50 12.50 17.45 23.60 212 11.10
GDPpc 28,110 741 0.39 21,427 27,402 32,952 216 11,525
EAST DMCpc 16.26 0.41 0.30 13.17 16.03 19.03 143 5.86
GDPpc 6,689 315 0.57 4,492 6,008 7,131 144 2,639
Notes: # the NO–EU27 are 3 countries: Norway, Switzerland and Turkey; CV means coefficient of
variation (sd/mean); IQR means inter-quartile range = p75 - p25. Variable legend: DMCpc: per
capita Domestic Material Consumption (million tonnes per capita); GDPpc: unit of euro, chain-
linked volumes, reference year 2000 (at 2000 exchange rates).

We estimate our unadjusted EKC-like model compared to the linear

specification using fixed effects panel technique both all 30 countries versus EU-
27 countries and western versus eastern European countries. With this model we
Domestic material consumption indicator and natural resources: A European … 183

can control only for country specific constant terms which may account for the
portion of the dependent variable heterogeneity not explained by the considered
regressors and the time effects. The Hausman test confirms, in all cases, the absence
of orthogonality between the set of regressors and residuals suggesting that the
model should be estimated with fixed and not with random effects.
The results of our estimation model, as reported in Table 2, do not confirm the
existence of an EKC-like curve because the turning points are very high. While
coefficients of income are significant and signs are correct, the coefficients of
square income have the right sign but are not significant apart from the last
estimation in the last column of Table 2. The values of the coefficients of GDP are
not so high (between 1.36 and 1.77), the only exception being the estimates of western
and eastern European countries in the unadjusted EKC-like case (more than 3.5).
Moreover, we control for time effects as well introducing time dummies in the
econometric estimates. Our findings show that time effects are always negative but
they are significant only for 2003 and from 2005 up to 2011 in both specifications.
This result confirms the structural break observed in Figure 2, where since 2007
EU-27 countries have recorded a change in their trends because of the sharp
decrease of DMCpc.
The need of a government intervention by European countries on the
environmental pressure by human activities is confirmed by our results. Even if
Cole et al. (1997), Agras and Chapman (1999), Galeotti and Lanza (1999), Heil and
Selden (2001), Cole (2004) and Galeotti et al. (2006) find an inverted–U shaped
relationship and turning points so high that a de-linking of CO2 emissions from
economic growth is present, our analysis is in line with the recent analysis of
Mazzanti and Musolesi (2013b) that using a more sophisticated empirical model
confirms a positive monotonic the carbon-income relationship in most economically
advanced countries.
The role of policy makers in managing natural resources and human activities
with an impact on the environment is becoming more relevant. European countries
have to both encourage more sustainable use of resources and integrate the work of
national agencies in managing the environmental burden of human activity.
184 Auci S., Vignani D.

Table 2: Fixed effects estimates of linear and unadjusted EKC-like specifications - 27

countries vs. 30 countries and West countries vs. East countries
Unadjusted EKC-like Unadjusted EKC-like
Linear estimation Linear estimation
Dep. Variable: estimation estimation
Ln (DMCpc) Western Eastern Western Eastern
27 countries 30 countries 27 countries 30 countries
countries countries countries countries
lnGDPpc 1.36*** 1.39*** 1.62** 1.50** 1.77*** 1.44*** 3.61 3.47***
(15.49) (16.61) (2.06) (1.98) (6.6) (8.76) (1.03) (2.83)
lnGDPpc^2 -0.02 -0.01 -0.09 -0.13*
(-0.34) (-0.15) (-0.53) (-1.67)
Cons -10.03*** -10.32*** -11.06*** -10.77*** -14.99*** -9.57*** -24.27 -17.68***
(-12.16) (-13.12) (-3.48) (-3.54) (-5.54) (-6.92) (-1.36) (-3.51)
d_2001 -0.02 -0.03 -0.02 -0.03 -0.04 -0.02 -0.04 -0.02
(-0.79) (-1.02) (-0.77) (-1.00) (-1.09) (-0.50) (-1.08) (-0.45)
d_2002 -0.04 -0.05* -0.04 -0.05** -0.06* -0.05 -0.06* -0.04
(-1.40) (-1.75) (-1.35) (-1.71) (-1.72) (-1.03) (-1.71) (-0.91)
d_2003 -0.06* -0.06** -0.06* -0.06** -0.08** -0.07 -0.08** -0.06
(-1.95) (-2.36) (-1.87) (-2.30) (-2.35) (-1.31) (-2.34) (-1.11)
d_2004 -0.04 -0.05* -0.03 -0.05 -0.06* -0.07 -0.06* -0.05
(-1.23) (-1.71) (-1.10) (-1.60) (-1.71) (-1.24) (-1.68) (-0.93)
d_2005 -0.07** -0.07** -0.06* -0.07** -0.09** -0.1* -0.09** -0.08
(-2.10) (-2.53) (-1.86) (-2.31) (-2.39) (-1.65) (-2.35) (-1.24)
d_2006 -0.11*** -0.12*** -0.11*** -0.12*** -0.14*** -0.16** -0.14*** -0.13
(-3.42) (-3.83) (-2.95) (-3.36) (-3.24) (-2.35) (-3.17) (-1.79)
d_2007 -0.15*** -0.15*** -0.14*** -0.15*** -0.19*** -0.20** -0.19*** -0.15
(-4.20) (-4.71) (-3.46) (-3.94) (-3.96) (-2.53) (-3.87) (-1.83)
d_2008 -0.18*** -0.19*** -0.18*** -0.18*** -0.24*** -0.19** -0.24*** -0.14
(-5.07) (-5.61) (-4.25) (-4.74) (-5.23) (-2.36) (-5.15) (-1.65)
d_2009 -0.23*** -0.22*** -0.22*** -0.22*** -0.24*** -0.26*** -0.24*** -0.22***
(-6.91) (-7.31) (-6.34) (-6.74) (-6.20) (-3.63) (-6.16) (-3.04)
d_2010 -0.27*** -0.27*** -0.27*** -0.27*** -0.30*** -0.29*** -0.30*** -0.25***
(-8.10) (-8.45) (-7.30) (-7.61) (-7.22) (-3.89) (-7.15) (-3.23)
d_2011 -0.28*** -0.28*** -0.27*** -0.28*** -0.31*** -0.32*** -0.31*** -0.27***
(-7.96) (-8.59) (-7.08) (-7.69) (-7.30) (-3.93) (-7.21) (-3.22)
R-sq Within 0.5598 0.56 0.56 0.56 0.4178 0.6755 0.4187 0.683
R-sq Between 0.0937 0.12 0.0921 0.12 0.2733 0.1916 0.2777 0.1606
R-sq Overall 0.0968 0.12 0.0957 0.12 0.2658 0.1971 0.2702 0.1805
F test ++ 30.2*** 33.56*** 27.8*** 30.88*** 10.88*** 20.64*** 10.03*** 19.56***
F test (ui =0) +++ 180.54*** 194.21*** 172*** 186.93*** 239.98*** 59.34*** 228.51*** 48.96***
Hausman
89.37*** 109.10*** 58.37*** 69.91*** 11.8*** 41.85*** 9.44** 44.98***
Test++++
Number of obs 324 355 324 355 212 143 212 143
Turning point 8.24E+21 9.49E+45 3.88E+08 8.85E+05
b d h l d l d
Notes: # besides EU27, the estimation includes 3 countries: Norway, Switzerland and Turkey; in
parenthesis t-statistics are reported; * significant at 10%; ** significant at 5%; *** significant at
1%;++ H0: joint significance of the regressors; +++ H0: joint significance of the fixed effects; ++++
H0: the two estimates do not differ in coefficients systematically. Variable legend: see Table 1.

Furthermore, to understand the environmental impact and costs of resource use

throughout the entire life cycle of materials and products that embody them, there
could be useful to deepen the analysis of MFA data and in particular of the material
flows among countries. Some additional econometric information could be suitable
to summarize some implications for policy makers. Having in mind that different
empirical models can influence the way the burden is shared among countries, the
linear and the unadjusted EKC-like estimations have the implication of considering
Domestic material consumption indicator and natural resources: A European … 185

that the admissible level of pressure for each country should be conditional to their
level of economic development. In other words, at the lower levels of income the
marginal cost of reducing the environmental pressure should be higher than at the
high levels of income.
However, using the fixed effect estimator, we are assuming that some
unobserved heterogeneity across individuals, fixed over time, can influence the
DMCpc. In Figure 3, we report the country fixed effect values in terms of predicted
values for unadjusted EKC-like estimations in 2010. In the first graph, based on the
30 country estimation, the western European countries are under the zero line,
meaning that their effective domestic potential pressure on the environment is less
than the estimated level, while the vice versa is true for the eastern European
countries. The first kind of countries seems to use less their natural resources, even
if they present differences once you deepen the analysis focusing only on the
western countries (see the second graph). Several undefined reasons could be at the
basis of this choice and certainly they could be linked to the countries’ government
structure and its policy.

30 countries































































































Western countries Eastern countries

 

















































































































 


Figure 3: Fixed effects to predicted value ratios of domestic environmental potential

pressure based on unadjusted EKC-like estimations – 2010
186 Auci S., Vignani D.

Among the worst countries in the pressure of natural resources, we can find
Romania and Bulgaria and this is confirmed even in the third graph where the sample
is restricted to only the eastern European countries. The choice of these two countries
has been to boost economic growth on the detriment of natural resources this because
they consider too costly reducing the environmental pressure of human activities and
they give less value to the environment and natural amenities. In other words, for these
countries reducing the economic, social and environmental negative effects of production
and consumption activities is not desirable in terms of benefits and costs.
The fixed effect analysis could turn out to be stringent for the assumption of
a unique coefficient of independent variables. No parameter variation across
countries is permitted. To estimate if the heterogeneity of countries is beyond the
fixed effect we report in Table 3 the test of parameter constancy based on the
random-coefficients model. With this estimator the parameter vector is treated as
a realization (in each panel) of a stochastic process. The results do not support the
assumption of parameter constancy for all the specifications and sub-samples.
However, we could not report the evidence on a country basis due to data constraints
even if Mazzanti and Musolesi (2013b) show how countries differ more on carbon-
time relation than the carbon-income relation.
Table 3: Test of parameter constancy
Unadjusted EKC-like Unadjusted EKC-like
Linear estimation Linear estimation
estimation estimation
27 30 27 30 Western Eastern Western Eastern
countries countries countries countries countries countries countries countries
chi2 6242.42 16202.45 1070.55 1531.14 14434.51 1486.31 139.89 1241.2
Prob > chi2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

5. SENSITIVITY ANALYSIS
The limitation of our dataset does not prevent us from improving our empirical
analysis through a sensitivity analysis. First we conduct a linear regression with
panel-corrected standard error. The model assumption is that the disturbances are
assumed to be either heteroskedastic across panels or heteroskedastic and
contemporaneously correlated across panels. The disturbances may also be assumed
to be autocorrelated within panel, and the autocorrelation parameter may be
constant across panels or different for each panel. This empirical model is used as
a simple equation when all 30 countries are considered and is adapted to a multiple
equation, fixed-effect, panel-data equations when the sample is divided in western
and eastern countries. Secondly, due to the fact that past decisions could have an
impact on current choices, we estimate a dynamic panel-data model based on the
bias-correction of least-squares dummy variable (LSDV) with a small number of
observations and strictly exogenous covariates.
Domestic material consumption indicator and natural resources: A European … 187

In Table 4, the error term not only is assumed heteroskedastic and contemporaneously
correlated across panels but also follow aAR(1) process. The estimate of the autocorrelation
parameter is quite high (the range is between 0.41 and 0.95), meaning that the
autocorrelation is present in this sample. As regards the coefficients, we observe that the
significance is very high – in all estimations is at 1% – and the signs are as expected. In
particular, these results turn the situation around because in this model the unadjusted
EKC-like curve is present both for EU-27 countries and 30 countries. However, the test
for a structural difference between the two specifications does not reject the null
hypothesis so the addition of the 3 countries (Norway, Turkey and Switzerland) to the
27 members of European Union does not imply any differences unlike in the case of
linear estimations where the test rejects the H0.
When we estimate multiple equation, fixed-effect, panel-data equations,
subdividing the sample between western and eastern countries as in Table 5, we find
again that structural difference test does not reject the null hypothesis in the case of
unadjusted EKC-like while the opposite occur in the linear case. As regards the
existence of the EKC curve, the western countries do not show an inverted U-shaped
curve, the vice versa is verified in the eastern country case.
Finally, to consider the influence of past decisions on the current behaviour
we use the bias-corrected least-squares dummy variable (LSDV) estimators for the
standard autoregressive panel-data model using the bias approximations in Bruno
(2005). The outcomes, Table 6, confirm the high influence of the lagged per capita
DMC variable in all the specifications while the squared per capita GDP coefficients
of EKC estimations are not significant apart for the EU27 specification.
The puzzle results of our sensitivity analysis are in line with the empirical
EKC literature that underlines how the results are sensitive to the sample of
countries chosen and to the time period considered (Grossman and Krueger, 1995;
Hill and Magnani, 2002) or to the methodology used (Borghesi, 2001). As regards
the methodology, our analysis, as the majority of empirical studies, is based on a
reduced-form model estimated using a parametric function which does not account
for heterogeneity of time related factors (Mazzanti and Musolesi, 2013b).
Even if some authors have departed from this stringent methodology and have
estimated the environment-income relationship using non-parametric technique
(Azomahou et al. 2006, Azomahou and Mishra, 2008, and Millimet et al., 2003),
semiparametric methods (Bertinelli and Strobl, 2005) or spline interpolation
(Schmalensee et al., 1998), there is not a unique and definite result. Our opinion can
be summarized by Vollebergh et al. (2009), who underline how this state of the art
has left both researchers and policy makers with the idea that econometric
techniques are not able to explain the real long run relationship between economic
growth and the environment.
188 Auci S., Vignani D.

Table 4: Linear and unadjusted EKC-like regressions with panel-corrected standard errors -
27 countries vs. 30 countries
Linear regression with panel-corrected standard errors
Dep. Variable: Unadjusted EKC-like
Linear estimation
Ln(DMCpc) estimation
27 countries 30 countries 27 countries 30 countries
cons 3.07*** -10.74*** -17.44*** -14.32***
(172.07) (-14.93) (-8.22) (-3.94)
lngdppc 0.12 1.36*** 3.00*** 2.19**
(0.47) (19.51) (6.12) (2.27)
lngdppc2 -0.10*** -0.05
(-3.41) (-0.77)
d_2001 0.02** -0.02*** -0.02*** -0.02*
(2.50) (-3.19) (-2.61) (-1.91)
d_2002 0.03*** -0.04*** -0.04*** -0.04***
(3.50) (-3.92) (-3.39) (-2.67)
d_2003 0.05*** -0.05*** -0.05*** -0.05***
(5.04) (-4.14) (-3.70) (-2.89)
d_2004 0.12*** -0.04** -0.03* -0.03
(10.22) (-2.38) (-1.74) (-1.19)
d_2005 0.14*** -0.06*** -0.04** -0.05*
(11.28) (-3.27) (-2.50) (-1.91)
d_2006 0.16*** -0.09*** -0.07*** -0.09***
(12.35) (-4.72) (-3.87) (-2.79)
d_2007 0.19*** -0.12*** -0.10*** -0.12***
(14.57) (-5.66) (-4.71) (-3.04)
d_2008 0.17*** -0.16*** -0.13*** -0.15***
(12.80) (-7.03) (-6.29) (-3.92)
d_2009 0.04*** -0.19*** -0.17*** -0.19***
(3.35) (-9.51) (-9.39) (-6.70)
d_2010 0.03** -0.23*** -0.21*** -0.23***
(2.33) (-10.72) (-10.79) (-7.28)
d_2011 0.06*** -0.23*** -0.21*** -0.23***
(4.71) (-10.76) (-11.17) (-6.99)
turning point 6.34E+06 1.82E+10
rhos 0.41 0.95 0.64 0.48
(0.78) (0.32) (0.69) (0.58)
Wald chi2 8.32E+09*** 1.65E+06*** 1.50E+06 5.76E+06
Wald test on eu27coefficients vs. eu30 coefficients
chi2( 13) 8.00E+05***
Prob > chi2 0.00
chi2( 14) 7.51
Prob > chi2 0.9133
Notes: Prais-Winsten regression, where panels are correlated and standard errors are corrected for
heteroskedasticity and contemporaneous correlation across panels; a panel-specific AR(1)
autocorrelation structure is hypothesized; in parenthesis t-statistics are reported; * significant at
10%; ** significant at 5%; *** significant at 1%.
Domestic material consumption indicator and natural resources: A European … 189

Table 5: Linear and unadjusted EKC-like fixed-effect panel-data systems with panel-
corrected standard errors – western countries vs. eastern countries
Dep. Linear estimation Unadjusted EKC-like estimation
Variable: Western Eastern Western Eastern
Ln (DMCpc) countries countries countries countries
cons -1.76 -9.65*** 22.65 -26.03***
(-1.46) (-8.79) (1.01) (-3.80)
lngdppc 0.48*** 1.62*** -4.45 5.64***
(4.10) (11.08) (-0.98) (3.29)
lngdppc2 0.25 -0.25**
(1.09) (-2.27)
d_2001 -0.01 -0.02*** -0.02 -0.02**
(1.05) (-2.89) (-1.24) (-2.46)
d_2002 -0.02 -0.06*** -0.03* -0.05***
(1.49) (-3.78) (-1.75) (-2.84)
d_2003 -0.03* -0.08*** -0.04** -0.07**
(1.86) (-3.61) (-2.11) (-2.50)
d_2004 0.02 -0.09*** 0.01 -0.07*
(-0.86) (-2.84) (0.34) (-1.70)
d_2005 0.01 -0.13*** 0.001 -0.10*
(0.72) (-3.16) (0.03) (-1.78)
d_2006 0.01 -0.20*** -0.01 -0.15**
(0.59) (-3.88) (-0.23) (-2.13)
d_2007 0.01 -0.25*** -0.02 -0.18**
(0.34) (-4.05) (-0.52) (-2.10)
d_2008 -0.05** -0.25*** -0.07** -0.17*
(-2.48) (-3.68 (-2.45) (-1.87)
d_2009 -0.10*** -0.29*** -0.11*** -0.24***
(-5.32) (-5.10) (-4.79) (-3.20)
d_2010 -0.12*** -0.34*** -0.14*** -0.28***
(-6.45) (-5.44) (-5.38) (-3.39)
d_2011 -0.12*** -0.35*** -0.14*** -0.28***
(-6.62) (-5.37) (-5.15) (-3.25)
turning point 7525.82 94864.33
rhos 0.17 0.18
(-0.59) (0.74)
Wald chi2 2828.36*** 2922.77***
Wald test on west coefficients vs. east coefficients
chi2( 13) 1.09E+03***
Prob > chi2 0.00
chi2( 14) 331.36
Prob > chi2 0.9133
Notes: Multiple equation, fixed-effect panel-data equations specified as seemingly unrelated
regressions are based on Prais-Winsten regression, where panels are correlated and standard errors
are corrected for heteroskedasticity and contemporaneous correlation across panels; a panel-
specific AR(1) autocorrelation structure is hypothesized; in parenthesis t-statistics are reported; *
significant at 10%; ** significant at 5%; *** significant at 1%.
190 Auci S., Vignani D.

Table 6: Linear and unadjusted EKC-like regressions with dynamic panel-data models –
27countries vs. 30 countries and western countries vs. eastern countries
Dep. Unadjusted EKC-like Unadjusted EKC-like
Linear estimation Linear estimation
Var.: estimation estimation
Ln 27 30 27 30 Western Eastern Western Eastern
(DMCpc) countries countries countries countries countries countries countries countries
L1_lndm
0.67*** 0.67*** 0.68*** 0.67*** 1.00*** 0.62*** 1.02*** 0.65***
cpc
(10.12) (11.16) (10.31) (10.56) (11.60) (6.70) (11.83) (6.47)
lngdppc 0.47*** 0.52*** -0.91 -0.87 0.53** 0.72*** -2.39 -0.54
(3.66) (4.79) (-1.31) (-0.93) (2.12) (3.51) (-0.64) (-0.46)
lngdppc2 0.08** 0.08 0.14 0.08
(2.05) (1.56) (0.78) (1.11)
d_2001 0.08** 0.09*** 0.11*** 0.11*** 0.03 -0.01 0.06 -0.01
(2.36) (2.96) (3.02) (3.75) (0.67) (-0.31) (1.41) (-0.32)
d_2002 0.08** 0.09*** 0.11*** 0.11*** 0.03 -0.02 0.06* -0.02
(2.41) (3.24) (3.21) (3.97) (0.93) (-0.41) (1.78) (-0.45)
d_2003 0.07*** 0.08*** 0.10*** 0.10*** 0.02 -0.01 0.05 -0.02
(2.68) (2.72) (3.27) (3.36) (0.62) (-0.25) (1.54) (-0.35)
d_2004 0.11*** 0.12*** 0.13*** 0.13*** 0.08** -0.05 0.11*** -0.06
(4.10) (4.91) (4.68) (5.58) (2.09) (-1.08) (3.10) (-1.26)
d_2005 0.07** 0.08*** 0.08*** 0.08*** 0.01 -0.10* 0.04 -0.11*
(2.25) (3.25) (2.74) (3.57) (0.42) (-1.65) (1.35) (-1.86)
d_2006 0.05* 0.05** 0.05** 0.05** -0.003 -0.11 0.03 -0.13*
(1.81) (2.16) (2.11) (2.10) (-0.08) (-1.48) (0.75) (-1.75)
d_2007 0.04 0.05* 0.04 0.04 -0.02 -0.12* 0.01 -0.14**
(1.64) (1.93) (1.63) (1.43) (-0.67) (-1.71) (0.20) (-2.03)
d_2009 -0.09*** 0.004 -0.07** -0.004 -0.08** -0.27*** -0.05 -0.29***
(-2.93) (0.17) (-2.53) (-0.18) (-2.22) (-4.36) (-1.49) (-4.66)
d_2010 -0.03 -0.07*** -0.02 -0.07*** -0.07** -0.17** -0.04 -0.18***
(-0.94) (-2.85) (-0.54) (-2.66) (-2.46) (-2.43) (-1.35) (-2.63)
d_2011 -0.0004 -0.02 0.01 -0.01 -0.03 -0.17** 0.03 -0.18**
(-0.02) (-0.63) (0.33) (-0.56) (-0.98) (-2.28) (0.97) (-2.50)
Turning
232.34 170.99 4092.88 36.81
point
Note: Bias-correction of least-squares dummy variable (LSDV) in dynamic panel-data models with
a small N and strictly exogenous covariates; in parenthesis t-statistics are reported; * significant at
10%; ** significant at 5%; *** significant at 1%.

6. CONCLUSIONS
The economic literature on environmental pollution, adopting an end-of-pipe
approach, has focused mainly on damage measures regarding water and air
pollution and land waste. In our analysis, instead, we assume a more result-oriented
approach, putting much more emphasis on preventive and integrated models. We
investigate the negative effect of production and consumption process on the
environment through the analysis of the rate of extraction and depletion of
renewable and non-renewable resource stocks, measured by DMC and considered
as a good proxy for potential environmental pressure. The underlined assumption
Domestic material consumption indicator and natural resources: A European … 191

is that sooner or later material flows required can represent an environmental

burden.
Even if our empirical model is based on the comparison of linear and—
unadjusted EKC-like hypothesis, the results should highlight important issues that
merit further investigations. However, as first result, we do not confirm the
existence of an inverted U–shaped relationship between per capita income and per
capita resource use indicator. These outcomes are confirmed in all estimations (EU-
27, EU-30 and western and eastern European countries). The high level of turning
points could imply the need of a government intervention in each European country
to reduce the environmental pressure by human activities.
A limit of this empirical result, though, is related to the model we have tested.
A further development of this work could be to consider an adjusted EKC-like
relationship in which other variables are considered as control variables. To
overcome some methodological limitations, we performed some sensitivity analyses.
All the results confirm the previous estimations.
A brief and straightforward conclusion is that European countries have to both
encourage more sustainable use of resources and integrate the work of national
agencies in managing the environmental burden of human activity. Until now it is
impossible to widen our conclusions because of the limits of our analysis. So far,
the main important result has been the in-existence of a EKC path for this indicator
of potential environmental degradation. However, we have to underline that this
result, even if strong and significant, needs more detailed investigations.

ACKNOWLEDGMENTS
We would like to thank all the participants to the XXIII Edition of Villa Mondragone
International Economic Seminar “After the Crisis: Rethinking Global Governance,
Development and Growth”, 22nd -23rd June, 2011, Rome, Italy; the 8th Annual
Conference of the European Association of Environmental and Resource Economists
(EAERE) 29 June – 2 July 2011 Rome, Italy; the XXV Villa Mondragone
International Economic Seminar on “The long term policy view. Shifting Wealth,
Income Inequalities and Demography” University of Rome “Tor Vergata”‘– Tor
Vergata Foundation 25–27 June 2013 Rome, Italy and two anonymous referees for
the useful comments.
192 Auci S., Vignani D.

REFERENCES
Adriaanse, A., Bringezu, S., Hamond, A., Moriguchi, Y., Rodenburg, E., Rogich, D., and Schütz, H.
(1997). Resource Flows: The Material Base of Industrial Economies. World Resource
Institute, Washington.
Agras, J. and Chapman, D. (1999). A dynamic approach to the environmental Kuznets curve
hypothesis. In Ecological Economics, 28: 267–277.
Arrow, K., Bolin, B., Costanza, R., Dasgupta, P., Folke, C., Holling, C.S., Jansson, B.O., Levin, S.,
Maler, K.G., Perings, C., and Pimental, D. (1995). Economic growth, carrying capacity and the
environment. In Science, 268: 520-521.
Auci, S. and Becchetti, L. (2006). The instability of the adjusted and unadjusted environmental
Kuznets curves. In Ecological Economics, 60: 282–298.
Azomahou, T. and Mishra, T. (2008). Age dynamics and economic growth: Revisiting the nexus in
a nonparametric setting. In Economics Letters, 99(1): 67-71.
Azomahou, T., Laisney, F., and Nguyen Van, P. (2006). Economic development and CO2 emissions:
A nonparametric panel approach. In Journal of Public Economics, 90: 1347– 1363.
Bertinelli, L. and Strobl, E. (2005). The environmental Kuznets curve semi-parametrically revisited.
In Economics Letters, 88: 350–357.
Borghesi, S. (2001). The environmental Kuznets curve: a critical survey, In M. Franzini and A. Nicita,
editors, Economic Institutions and Environmental Policy, 201–224. Ashgate Publishing,
Farnham, UK.
Bringezu, S. and Schütz, H. (2001). Total material requirement of the European Union, Technical
Report, 55, European Environmental Agency, Copenhagen.
Brock, W. and Taylor, S. (2010). The green Solow model. In Journal of Economic Growth, 15: 127-
53.
Bruno, G.S.F. (2005). Approximating the bias of the LSDV estimator for dynamic unbalanced panel
data models,. In Economics Letters, 87: 361-366.
Cole, M.A. (2004). US environmental load displacement: Examining consumption, regulations and
the role of NAFTA. In Ecological Economics, 48(4): 439–450.
Cole, M.A., Rayner, A.J., and Bates, J.M. (1997). The environmental Kuznets curve: An empirical
analysis. In Environmental and Development Economics, 2: 401–416.
Costantino, C., Femia, A., and Vignani, D. (2008). Material resources use, In: Istat, Key Topics. APAT
Environmental Data Yearbook 2007, 85-102, April Rome.
Dinda, S. and Coondoo, D. (2006). Income and emission: A panel data-based cointegration analysis.
In Ecological Economics, 57: 167– 181.
Dinda, S. (2004). Environmental Kuznets curve hypothesis: A survey. In Ecological Economics, 49:
431– 455.
Ekins, P. (1997). The Kuznets curve for the environment and economic growth: Examining the
evidence. In Environment Planning, 29: 805-830.
EUROSTAT (2001). Economy-Wide Material Flow Accounts and Derived Indicators – Methodological
Guide, European Commission, Luxembourg.
EUROSTAT (2002). Material Use in the European Union 1980-2000: Indicators and Analysis,
Luxembourg.
Femia, A. and Vignani, D. (2005). The development of economy-wide material flow accounting, In
Contributed Papers for the Meeting SIS Statistics and Environment, 61-66, University of
Messina 21-23 September, CLEUP, Padua.
Domestic material consumption indicator and natural resources: A European … 193

Femia, A. and Vignani, D. (2010). Material Flows of the Italian Social-Economic System: Years
1981-2007. Italian National Institute of Statistics, Rome.
Galeotti, M. and Lanza, A., (1999). Richer and cleaner? A study on carbon dioxide emissions in
developing countries. In Energy Policy, 27, 565–573.
Galeotti, M., Lanza, A., and Pauli, F. (2006). Reassessing the environmental Kuznets curve for CO2
emission: A robustness exercise. Ecological Economics, 57: 152–163.
Grossman, G.M. and Krueger, A.B. (1995). Economic growth and the environment. In The Quarterly
Journal of Economics, 110(2): 353-77.
Heil, M.T. and Selden, T.M. (2001). Carbon emissions and economic development: future trajectories
based on historical experience. In Environment and Development Economics, 6: 63–83.
Hill, R.J. and Magnani, E. (2002). An exploration of the conceptual and empirical basis of the
environmental Kuznets curve. In Australian Economic Papers, 41(2): 239-54.
Holtz-Eakin, D. and Selden, T.M. (1995). Stoking the fires? Co2 emissions and economic growth. In
Journal of Public Economics, 57: 85-101.
Kijima, M., Nishide, K., and Ohyama, A. (2010). Economic models for the environmental Kuznets
curve: A survey. In Journal of Economic Dynamics & Control, 34: 1187–1201.
Kuznets, S. (1955). Economic growth and income inequality. In American Economic Review, 45(1):
1–28.
Lee, C.C. and Lee, J.D. (2009). Income and CO2 emissions: Evidence from panel unit root and
cointegration tests. In Energy Policy, 37: 413–423.
Mazzanti, M. and Musolesi, A. (2013a). The heterogeneity of carbon Kuznets curves for advanced
countries: Comparing homogeneous, heterogeneous and shrinkage/Bayesian estimators. In
Applied Economics, 45: 3827-42.
Mazzanti, M. and Musolesi, A. (2013b). Nonlinearity, heterogeneity and unobserved effects in the
CO2 relation for advanced countries, ENI Nota di Lavoro, 91, 2013.
Millimet, D., List, J., and Stengos, T. (2003). The EKC: Real progress or misspecified models? In The
Review of Economics and Statistics, 85: 1038-1047.
Moll, S., Bringezu, S., and Schütz, H. (2003). Resource Use in European Countries. An Estimate of
Materials and Waste Streams in the Community, Including Imports and Exports Using the
Instrument of Material Flow Analysis, European Environment Agency, Copenhagen.
Musolesi, A., Mazzanti, M., and Zoboli, R. (2010). A Bayesian approach to the estimation of EKC
for CO2. In Applied Economics, 42: 2275-87.
Ordás Criado, C., Valente, S., and Stengos, T. (2011). Growth and pollution convergence: Theory and
evidence. In Journal of Environmental Economics and Management, 62: 199–214.
Panayotou, T. (2000). Economic growth and the environment, environment and development, CID
Working Paper, 56(4).
Pearson, P.J. (1994). Energy, externalities and environmental quality: Will development cure the ills
it creates? In Energy Studies Review, 6(3): 199–216.
Roberts, J.T. and Grimes, P.E. (1997). Carbon intensity and economic development 1962-1991: A
brief exploration of the environmental Kuznets curve. In World Development, 25(2): 191-198.
Rothman, D.S. (1998),. Environmental Kuznets curves real progress or passing the buck? A case for
consumption-based approaches. In Ecological Economics, 25: 177-194.
Schmalensee, R., Stoker, T.M., and Judson, R.A. (1998). World carbon dioxide emissions: 1950–
2050. In Review of Economics and Statistics, 80(1): 15–27.
194 Auci S., Vignani D.

Selden, T.M. and Song, D. (1995). Neoclassical growth, the J curve for abatement and the inverted
U curve for pollution. In Journal of Environmental Economics and Environmental Manage-
ment, 29: 162-168.
Shafik, N. and Bandyopadhyay, S. (1992). Economic Growth and Environmental Quality: Time
Series and Cross-country Evidence, Background Paper for the World Development Report
1992, World Bank, Washington DC.
Stern, D.I., Common, M.S., and Barbier, E.R. (1996). Economic growth and environmental degradation:
The environmental Kuznets curve and sustainable development. In World Development, 24:
1151–1160.
Suri, V. and Chapman, D. (1998). Economic growth, trade and energy: Implications for the
environmental Kuznets curve. In Ecological Economics, 25: 195-208.
Vollebergh, H.R.J., Melenberg, B., and Dijkgraaf, E. (2009). Identifying reduced-form relations with
panel data: The case of pollution and income. In Journal of Environmental Economics and
Management, 58: 27–42.