Energy Policy 66 (2014) 419–427

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Energy consumption, human well-being and economic development

in central and eastern European nations: A cautionary tale of
sustainability
Andrew K. Jorgenson a,n, Alina Alekseyko a, Vincentas Giedraitis b
a
Department of Sociology, University of Utah, 380 South 1530 East, Room 301, Salt Lake City, UT 84112, USA
b
Faculty of Economics, Vilnius University, Sauletekio al. 9, (II building), Room 411, 10222 Vilnius, Lithuania

H I G H L I G H T S

We analyze the energy intensity of well-being in Central and Eastern European nations.

The effect of economic development is time-dynamic.

Other factors influence the energy intensity of well-being.

The results highlight possibilities for enhanced sustainability policies.

art ic l e i nf o a b s t r a c t

Article history: Sustainability is fundamentally a challenge of tradeoffs. In order to improve human well-being through
Received 17 September 2013 economic development we consume nonrenewable energy and other natural resources, relying on a
Received in revised form broad range of ecosystem services. Enhancing sustainability requires reducing the “energy intensity of
6 November 2013
human well-being (EIWB)”: the amount of energy used per unit of human well-being. In this study we
Accepted 8 November 2013
employ longitudinal analysis techniques to assess the temporally dynamic relationship between EIWB
Available online 26 November 2013
and economic development for a sample of 12 Central and Eastern European (CEE) nations for the 1992
Keywords: to 2010 period. These are nations that have recently transitioned, which is still an ongoing process, from
Sustainability socialist command economies to market demand economies. During this ongoing transition, many of
Development
them have experienced declines in energy intensity, coupled with increased energy efficiency, while
Energy
human well-being has improved considerably. The results of the analysis indicate that the relationship
Human well-being
Central and Eastern Europe between EIWB and economic growth in CEE nations is complex and has changed dramatically through
Environmental sociology time. Of particular importance, the later years of the study exhibit an increasingly sustainable relation-
ship between EIWB and economic development. The findings point to future possibilities for relatively
more harmonious relationships between development, human well-being, and the natural environment.
& 2013 Elsevier Ltd. All rights reserved.

1. Introduction enhances quality of life once societies have reached certain

Pathways to sustainability pose challenges for how to balance
ecological and environmental concerns with human well-being
and economic development. It is generally assumed that develop-
ment enhances human well-being (Brady et al., 2007; Gilpin,
2001), while ecosystems provide valuable services to humanity
(Liu et al., 2007; Ringold et al., 2012), which themselves enhance
quality of life in various ways (Prescott-Allen, 2001; Reid et al.,
2005). However, for decades many sustainability scholars have
questioned the extent to which heightened resource use further
n
Corresponding author. Tel.: þ 1 801 554 7393.
E-mail addresses: andrew.jorgenson@soc.utah.edu (A.K. Jorgenson),
alina.alekseyko@utah.edu (A. Alekseyko),
vincas.giedraitis@ef.vu.lt (V. Giedraitis).
thresholds of collective well-being (Mazur, 2011; Mazur and
Rosa, 1974; Steinberger and Roberts, 2010). At the same time,
lively debates across the environmental social sciences continue
regarding the complex relationships between environmental con-
ditions and economic development (Jorgenson and Clark, 2012;
Rosa et al., 2010; York, 2012).
Each of these smaller pieces of the overall sustainability
challenge demands continued theoretical consideration and
empirical investigation. For example, Mazur (2011), in a long-
itudinal analysis of energy consumption and multiple human well-
being indicators for a sample of high income nations, finds that
from 1980 to 2006, life expectancy rose in all nations in his sample
irrespective of whether they increased or decreased their per
capita energy consumption levels. However, Mazur's study does
not consider the broader causal mechanisms that explain this

0301-4215/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enpol.2013.11.020
420 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
variation, and perhaps the longitudinal relationship between
energy consumption and human well-being might differ for lower
income nations relative to high income nations. Turning to human
well-being and economic development, much past research sug-
gests that development increases life expectancy and reduces
infant mortality in nations throughout the world (e.g., Firebaugh
and Beck, 1994), but the effect of development on different well-
being outcomes has slightly decreased though time in less-
developed nations (Brady et al., 2007), and other studies suggest
that many factors besides economic development also influence
changes in the well-being of human populations (e.g., Preston,
1975). Even with the need for further analyses in each of these
specific areas, we contend that sustainability research needs to
advance integrative approaches that simultaneously focus on the
interconnections of human well-being and environmental condi-
tions, and how economic development affects them.
One of the more promising integrative approaches to sustain-
ability studies is the emerging body of research on the environ-
mental (or energy) intensity of well-being (EIWB), which refers to
pressure placed on the environment per unit of human well-being
(Dietz et al., 2009, 2012; Knight and Rosa, 2011). Scholars commonly
operationalize EIWB as the ratio of a measure of stress on the
environment, such as the ecological footprint, to a measure of human
well-being, typically life expectancy at birth. A central question in
this body of work is what role economic development plays in
shaping the EIWB of societies. Some prevalent theoretical traditions
in both environmental sociology and environmental economics
suggest that the effect of development on the environment may
change across levels of development, with some theories predicting
increased stress while others suggest the opposite (e.g., Grossman
and Krueger, 1995; Jorgenson and Clark, 2012; Mol, 2001). These
perspectives can also apply to the broader relationships between
EIWB and development (Dietz et al., 2012; Knight and Rosa, 2011).
However, the implications of research on EIWB and develop-
ment stretch well beyond academic boundaries and theoretical
considerations, and have potential policy implications. Since
economic growth is a common objective of governments and
international organizations, understanding the relationship
between economic development and EIWB is critical. Some
assume that development leads to reductions in the environmen-
tal (or energy) intensity of well-being. If that is so, then the pursuit
of economic growth will have the beneficial side effect of enhan-
cing sustainability. However, if economic growth has no effect on
EIWB, then the environment will be protected if and only if
policies that supplement growth promotion are enacted. If eco-
nomic growth increases EIWB, then development policies are
deleterious to sustainability, and must be reconsidered or balanced
with new approaches to counter the unintended environmental
harms of development (Banuri and Hallstrom, 2012).
Past research on EIWB and economic development includes
large samples of nations from most regions of the world (Dietz
et al., 2012; Knight and Rosa, 2011), a practice common in other
areas of sustainability research (Jorgenson and Clark, 2011; Roberts
and Parks, 2007; Rosa et al., 2004). While such an approach has
many strengths and advantages, a limitation common in cross-
national comparative research is the exclusion of nations from
particular regions of the world, most notably the nations of Central
and Eastern Europe (CEE). These are nations that have recently
transitioned from socialist command economies to market
demand economies and the attending changes are ongoing. With
their independence, CEE countries inherited an energy-intensive
industrial sector, coupled with high dependence on raw materials
and primary energy sources (Boeri, 2000). Further, energy ineffi-
ciency was common in early years of independence for a number
reasons dating back to the Soviet era, including older equipment
and outdated technologies, low thermal performance of
households and public buildings, the prevalence of older auto-
mobiles, and inadequate metering and control of energy con-
sumption (Miskinis et al., 2006). Since the collapse of the Soviet
Union, however, many CEE nations have experienced decreases in
energy intensity, coupled with relative increases in energy effi-
ciency (Alekseyko and Giedraitis, 2013). Turning to well-being,
many CEE nations' welfare has improved considerably relative to
the pre-independence period, partly due to infrastructural
improvements and economic development (Giedraitis et al., 2009).
Given the noteworthy changes taking place in these nations in
recent decades, rigorous assessments of the environmental and
human well-being impacts of their patterns and structures of eco-
nomic development would offer valuable insights for their future
sustainability efforts as well as having implications for other world
regions. The changes that these nations have made in recent decades
are unique in the period for which we have data and thus are of
considerable interest for understanding how transitions in political
and economic institutions influence the environment and human
well-being. Some scholars have conducted longitudinal analyses of the
relationships between carbon emissions or energy consumption and
economic development in CEE nations (Alekseyko and Giedraitis,
2013; Jorgenson, 2011; Jorgenson et al., 2012; York, 2008), while other
sustainability-oriented studies have focused on the predictors of
environmental concern for individuals within CEE nations relative to
other regions of the world (Marquart-Pyatt, 2012). Public health
research on the economic factors influencing human well-being in
these nations has also blossomed in recent decades (Beaglehole and
Bonita, 2009; Bobak et al., 2000; Hyns, 2005).
What is missing is research on CEE nations that focuses on the
connections between human well-being, the environment, and
economic development in general, and the effects of development
on EIWB in particular. The current study is an initial attempt to fill
this void. Drawing from past work, we calculate an annual
measure of EIWB for a sample of CEE nations from 1992 to 2010,
using the ratio of per capita energy consumption to average life
expectancy at birth (i.e., the energy intensity of well-being). We
analyze these data on EIWB as well as measures of economic
development and interactions between development and time to
assess their temporarily dynamic relationships. To assess the
relationships between EIWB and development we employ two-
way fixed effects panel modeling techniques. The estimated model
includes a number of additional controls commonly employed in
past research on related outcomes, including measures of domes-
tic income inequality, democratization, public health expenditures,
manufacturing, and world-economic integration. The results high-
light the temporally dynamic relationship between EIWB and
development, and point to future possibilities for enhanced
sustainability in CEE nations.
2. Materials and methods
2.1. The sample
We analyze a perfectly balanced dataset that consists of annual
observations from 1992 to 2010 for a sample of 12 CEE nations.
This results in a sample of 228 total observations. Data availability
precludes us from including time points before 1992 or after 2010.
The 12 nations in the dataset include Armenia, Azerbaijan, Belarus,
Estonia, Georgia, Kazakhstan, Kyrgyz Republic, Latvia, Lithuania,
Russia, Tajikistan, and Ukraine. These are the same nations
included in other recent sustainability-oriented studies of Central
and Eastern Europe (Jorgenson, 2011; Jorgenson et al., 2012),
although data availability has limited some other studies to a
smaller numbers of nations in the region (York, 2008).
 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
For this study, the sample of nations during the period of
investigation underwent significant transformations in institu-
tional conditions and practices. Such transformations could to
some extent influence the quality of data and their applicability for
analyses of changes between nations as well as changes within
nations through time. However, the employed model estimation
techniques, which we discuss below, include fixed effects for each
nation and time point that are included in the study. These fixed
effects techniques can partly help resolve the influence of case-
specific and time-specific anomalies in analyzed data (Jorgenson
et al., 2009), thereby lessening the likelihood of rejecting a type
one error (i.e., falsely rejecting the null hypothesis) and ultimately
allowing for more conservative analyses of relationships between
an outcome and predictors (Allison, 2009). Nonetheless, it is
important to recognize the potential limitations in the analyzed
data, given the unique and changing circumstances of the nations
in the study.
2.2. Dependent variable: Measuring the energy intensity of human
well-being
We quantify the energy intensity of human well-being (EIWB)
as a ratio between a per capita energy consumption measure and a
measure of human well-being. The most common EIWB approach
uses the ecological footprint, a comprehensive measure of stress
on the environment, as the numerator (Dietz et al., 2012; Knight
and Rosa, 2011). However, relatively limited ecological footprint
data are available for CEE nations (Jorgenson and Clark, 2011),
while energy consumption measures are readily available for the
CEE nations included in our analyses as recently as 2010 and as far
back as 1992. The ecological footprints of nations tend to be very
highly correlated with energy consumption (Rosa et al., 2004).
Further, recent cross-national sustainability studies of CEE nations
have focused on carbon emissions resulting from energy con-
sumption, and emissions and energy consumption tend to be
correlated well above.9 (Jorgenson, 2011; Jorgenson et al., 2012;
York, 2008). Considering that greenhouse gas emissions resulting
from energy use are the primary anthropogenic contributor to
climate change, the use of energy consumption as the numerator
for EIWB in the current study seems justified, although subsequent
work could fruitfully explore other measures. We note that the use
of per capita carbon emissions (from the burning of fossil fuels and
the manufacturing of cement) as a numerator yields results that
are entirely consistent with the reported findings, which is
expected given the close to perfect correlation between energy
consumption and carbon emissions.
More specifically, we employ energy use per capita as the
numerator and average life expectancy at birth as the denomi-
nator. Energy use refers to use of primary energy before transfor-
mation to other end-use fuels, which is equal to indigenous
production plus imports and stock changes, minus exports and
fuels supplied to ships and aircraft engaged in international
transport. These data are in thousands of kilograms of oil equiva-
lent per capita, and are obtained from the World Bank (2013),
which obtained them from the International Energy Agency
(http://www.iea.org/stats/index.asp). Such energy consumption
measures are commonly used in environmental and sustainability
studies (Clark et al., 2010; York, 2012).
Fig. 1 provides the panel-specific distributions for energy
consumption per capita for the sample of CEE nations from 1992
to 2010. The line plots for each panel indicate initial downward
trends in per capita energy consumption for all 12 nations,
followed by notably different temporal patterns in energy con-
sumption levels through 2010.
Life expectancy at birth indicates the average number of years a
newborn infant would live if prevailing patterns of mortality at the
421
time of its birth were to persist throughout its life. These data,
which we obtain from the World Bank (2013), are derived from
several sources including (1) United Nations Population Division
(World Population Prospects) and (2) United Nations Statistical
Division (Population and Vital Statistics Report, various years).
Average life expectancy is well measured in most countries, it is
widely accepted as an indicator of societal “good”, and it has been
the focus of much demographic and public health research. Life
expectancy is not the only appropriate, valid, or reliable measure
of human well-being, but it is well-established in various areas of
research and it is employed in recent EIWB studies (Dietz et al.,
2012; see also Mazur, 2011).
Fig. 2 provides line plots of each panel for average life
expectancy for the sample of CEE nations from 1992 to 2010. Like
the line plots for energy consumption, here we observe an initial
pattern of decreasing values for life expectancy, followed by
notably different time trends through 2010. The majority of
nations in the sample experienced overall increases (in varying
magnitudes) in average life expectancy during the period of study.
Calculating and employing a ratio as a dependent variable
creates a complication that must be resolved prior to using it as a
dependent variable. Since the variability and the range of the
numerator and the dominator can differ substantially, the variance
of the ratio can be dominated by one or the other. In the dataset
for the current study the coefficient of variation (standard devia-
tion/mean) for energy consumption per capita is .612 and the
range is .336 to 5.351. For life expectancy the coefficient of
variation is .039 and the range is 62.286 to 75.429. The coefficients
of variation indicate that the relative variation in energy con-
sumption per capita – the numerator – is substantially larger than
the variation in life expectancy—the dominator. Under such con-
ditions the variation in the numerator could drive variation in the
ratio. To resolve this complication we take the approach used in
Dietz et al. (2012), which according to them, was developed by
researchers at the New Economics Foundation and explained to
the lead author of their study. We constrain the coefficient of
variation for the numerator and denominator to be equal by
adding a constant to one of them, shifting the mean without
modifying the variance, thereby allowing for the equalization of
the coefficient of variation for the two initial measures. For our
6
4
2
0
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Armenia Azerbaijan Belarus Estonia
Georgia Kazakhstan Kyrgyz Republic Latvia
Lithuania Russia Tajikistan Ukraine
Fig. 1. Energy consumption in thousands of kilograms of oil equivalent per capita.
422 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427

75 some nations in the sample experiencing overall declines in EIWB

from 1992 to 2010, while other nations appear to exhibit values of
EIWB towards the end of the period of study at levels similar to the
beginning point. Overall, the line plots presented in Fig. 3 indicate
that the temporal dynamics of EIWB for each nation in the sample
are far from monolithic and exhibit nontrivial variation to be
70
explained.

2.3. Key independent variables

65 We include gross domestic product per capita (GDP per capita)

as a measure of a nation's level of economic development. These
data, which are in logarithmic form (ln) to minimize their
positively skewed distribution, are measured in 2000 constant
U.S. dollars and obtained from the World Bank (2013). We
calculate and use interactions between GDP per capita (ln) and a
60 dummy variable for each time point, where 1992 is the reference
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
year. Such interactions allow us to assess the extent to which the
effect of development on EIWB increases or decreases through
Armenia Azerbaijan Belarus Estonia
time in CEE nations. Interactions between time and quantitative
Georgia Kazakhstan Kyrgyz Republic Latvia measures are relatively common in cross-national panel studies
Lithuania Russia Tajikistan Ukraine within the environmental social sciences (Jorgenson and Clark,
2013; Jorgenson et al., 2012; York, 2012).
Fig. 2. Average life expectancy at birth.
Fig. 4 provides the line plots of GDP per capita (ln) for each
nation from 1992 to 2010. The plots indicate that the majority of
CEE nations in the sample experienced initial declines in GDP per
55
capita, followed by varying temporal patterns of change in levels
of development, with the majority of nations in the sample
exhibiting higher GDP per capita values at the endpoint relative
to the beginning year in the study.

50 2.4. Control variables

Even though the fixed effects in the employed panel model

estimation technique described below typically explain a high
proportion of variation in the dependent variable, we also include
5 time-variant control variables common is research on environ-
mental and human well-being outcomes.
45

9
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010

Armenia Azerbaijan Belarus Estonia

8
Georgia Kazakhstan Kyrgyz Republic Latvia
Lithuania Russia Tajikistan Ukraine

Fig. 3. Energy intensity of well-being.

7

employed data the coefficients of variation for the two variables

can be made equal by adding 30.957 to the measures of energy
consumption per capita. Thus, our employed measure of the 6
energy intensity of well-being is
EIWB ¼ ½ðECPC þ 30:957Þ=LE  100

where EIWB is the energy intensity of well-being, ECPC is energy 5

consumption in thousands of kilograms of oil equivalent per
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010

capita, and LE is average life expectancy in years. Consistent with

past research (Dietz et al., 2012), we multiple by 100 to scale
the ratio. Armenia Azerbaijan Belarus Estonia
Fig. 3 provides the panel-specific distributions for the energy Georgia Kazakhstan Kyrgyz Republic Latvia
intensity of well-being for each nation in the sample from 1992 to
Lithuania Russia Tajikistan Ukraine
2010. The line plots indicate substantial differences in the tem-
poral dynamics of the distribution of EIWB for each nation, with Fig. 4. GDP per capita (ln).
 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
We use the combined “POLITY2” score to control for level of
democratization, which we obtain from the Integrated Network
for Societal Conflict Research (http://www.systemicpeace.org/
inscr/inscr.htm, accessed March 2, 2013). We note that there is
some debate in the literature about how to measure democracy
(Chiebug et al., 2009), but this is the most commonly employed
measure of democratization in the environmental social sciences
(e.g., Jorgenson et al., 2009) and public health fields (e.g., Shandra
et al., 2004). It ranges from values of 10 to þ10, where a value of
10 represents strongly autocratic and þ10 represents strongly
democratic. Research in the public health field suggests the
importance of democracy for improving a country's average life
expectancy (Clark, 2011; Wickrama and Mulford, 1996). Environ-
mental social scientists (Ehrhardt-Martinez et al., 2002) commonly
argue that environmental protection and suppressed resource
consumption (including energy consumption) are enhanced where
both responsiveness and policy capacity are high, and such
conditions are more likely in nations with higher levels of
democratization. Knight and Rosa (2011) employ level of demo-
cratization as a control variable in their cross-sectional study of
the environmental intensity of well-being, yielding a nonsignifi-
cant relationship. Considering the potential for democratization to
increase life expectancy and suppress energy consumption levels
as well as its employment as a control in prior research on EIWB,
we consider it appropriate to include level of democratization as a
control in the current study.
Conventional wisdom suggests that increases in government
expenditures on health relative to the size of a national economy
should enhance human well-being, including increases in life expec-
tancy. However, the results of prior research yield mixed findings. For
example, Kennelly et al. (2003) find that government health expen-
ditures increases life expectancy in a sample of high income nations,
while Shandra et al. (2004) find that government health expendi-
tures does not have a significant effect on infant mortality rates in
developing nations. Nonetheless, the transition economy nations of
Central and Eastern Europe are largely absent from these past bodies
of comparative research. And consistent with conventional lines of
reasoning, one might expect that EIWB will be negatively associated
with government expenditures on health. Thus, we control for
general government expenditures on health as percent of total
GDP. These data are obtained from Transformative Monitoring for
Enhanced Equity (TransMonEE) database (http://www.transmonee.
org/index.html, accessed March 3, 2013), who gather them from the
National Health Accounts of the World Health Organization.
We include a Gini coefficient for domestic income inequality as a
control variable. These data are gathered from the TransMonEE
database (http://www.transmonee.org/index.html, accessed March 3,
2013), who obtain them from various governmental statistical sources
noted in their database. The Gini coefficients indicate distributions of
earnings within a given nation, and can range from a value of zero to a
value of one, where zero would refer to perfectly equal distributions of
earnings, while a value of one would refer to perfectly unequal
distributions of earnings. It is often found that nations with higher
levels of domestic income inequality exhibit lower levels of average
life expectancy (Babones, 2008; Clark, 2011). Prior work on the
environmental intensity of well-being, particularly Knight and Rosa
(2011), suggest that income inequality may affect EIWB by dispropor-
tionately distributing resource and energy consumption and whatever
well-being benefits it entails, and their findings do indicate that
income inequality increases the environmental intensity of well-
being. Further, Jorgenson (2003) finds that resource consumption
(measured as the ecological footprint) is negatively correlated with
income inequality. Thus, in the present study we would expect to
observe a significant positive effect of income inequality on EIWB.
We control for manufacturing as percent of total GDP in
logarithmic form (ln) to normalize its positively skewed
423
distribution. Relative levels of manufacturing are often found to
increase energy consumption levels in studies of developed and
developing nations (Clark et al., 2010). It is commonly argued that
growth in manufacturing—or industrialization more generally,
involves the use of greater amounts of energy and material inputs.
The impacts of manufacturing on life expectancy or other aspects
of human well-being remain understudied and under-theorized in
macro-comparative contexts. Thus, one might expect EIWB in CEE
nations to be positively associated with level manufacturing due to
its influence on energy consumption.
We control for exports as percent of total GDP, which we obtain
from the World Bank (2013). This is a commonly employed
measure of world-economic integration in general and an indica-
tor of export-oriented production in particular. Some recent
studies find modest evidence indicating that relative levels of
exports increase energy use and carbon emissions (Roberts and
Parks, 2007), while other studies of carbon emissions and energy
use yield inconclusive findings, including panel studies of CEE
nations (Jorgenson et al., 2012; York, 2008). Regarding human
well-being, some perspectives in development economics and the
sociology of development suggest that export-oriented production
stimulates economic development, raises wages, reduces unem-
ployment, leads to increased revenue for the state, and thus
enhances the well-being of nations. However, findings from
well-being research yield inconclusive results regarding the
impact of level of exports (Brady et al., 2007; Moore et al.,
2006). Nonetheless, we consider it crucial to control for world-
economic integration to allow for a more rigorous assessment of
the EIWB and development relationship.
Table 1 presents the descriptive statistics for the variables
included in the analyses, with the exceptions of the interactions
between GDP per capita and each yearly dummy variable as well
as the country-specific and year-specific dummy variables that
serve as the fixed effects (discussed below). These additional
statistics are available from the authors upon request.
2.5. Model estimation technique
We employ a time-series cross-sectional Prais–Winsten (PW)
regression model with panel-corrected standard errors (PCSE). We
allow for disturbances that are heteroskedastic and contempor-
aneously correlated across panels (Cameron and Trivedi, 2009).
We use this approach because the feasible generalized least-
squares estimator that is often used to analyze panel data
produces standard errors that can lead to overconfidence with
panel datasets that do not have many more time periods than
panels (Beck and Katz, 1995), and in our study we have 19 annual
observations for 12 nations. We correct for AR(1) disturbances (i.e.,
first-order autocorrelation) within panels, and since we have no
theoretical basis for assuming the process is panel specific, we
treat the AR(1) process as common to all panels (Beck and Katz,
Table 1
Descriptive statistics.
Mean Standard Skewness Kurtosis
deviation
Energy intensity of well-being 48.326 2.859 .238  .545
GDP per capita (ln) 7.066 1.004  .285  .688
Democratization 2.202 5.972  .371  1.432
Public health expenditures as % 2.987 1.385  .066  1.046
of GDP
Domestic inequality .409 .062 .083  1.093
Manufacturing as % of GDP (ln) 2.838 .433  .338 .226
Exports as % of GDP 45.887 17.302 .273  .439
Note: N ¼ 228.
424 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427

1995:638). We control for both period-specific and unit-specific 3. Results

disturbances. Our general model is as follows:
yit ¼ β xit þ ui þ wt þ eit
subscript i represents each unit of analysis (i.e., country), subscript
t represents the time period, and yit is the dependent variable for
each country at each time period. βxit represents the vector of
coefficients for predictor variables that vary over time, ui is the
unit-specific (i.e., country-specific) disturbance term, wt is the
period-specific disturbance term that is constant across all coun-
tries, and eit is the disturbance term unique to each country at each
point in time. We calculate and employ dummy variables to
control for ui and wt. The former controls for potential unobserved
heterogeneity that is temporally invariant within countries (unit-
specific intercepts), while the latter controls for potential unob-
served heterogeneity that is cross-sectionally invariant within
periods (period-specific intercepts). The unit-specific intercepts
approach is analogous to the dummy variable fixed effects model
(Allison, 2009), often referred to as one-way fixed effects (Baum,
2006). Similarly, the inclusion of the period-specific intercepts is
equivalent to modeling temporal fixed effects, and including both
period-specific and unit-specific intercepts is analogous to esti-
mating a two-way fixed effects model (Baum, 2006). With the
inclusion of the unit-specific (i.e., country-specific) fixed effects,
the focus is on modeling and explaining within-country variation
in the outcome. Including period-specific intercepts also lessens
the likelihood of biased model estimates resulting from outcomes
and predictors with similar time trends (Wooldridge, 2005). The
panel analyses are conducted with the “xtpcse” suite of commands
in Stata software.
Table 2 reports the findings for the estimated model. We
provide coefficients, flagged for statistical significance, as well as
corresponding panel corrected standard errors. We also report the
R2 value for the model, the number of estimated coefficients, the
overall sample size (i.e., N), sample size divided by number of
coefficients, and rho for the model. We remind readers that the
very high R2 value is largely a function of the employed country-
specific and year-specific intercepts, which also increase the
number of estimated coefficients for the model. We note that
results of sensitivity analyses available upon request indicate that
the reported findings are not overly influenced by any of the 12
nations included in the study, and diagnostics indicate that the
analysis is not influenced by multicollinearity.
The main effect of GDP per capita is negative and statistically
significant, while all interactions between GDP per capita and the
yearly dummy variables are positive and statistically significant
with the exception of the interaction between level of develop-
ment and the year 1999, which yields a p-value of approximately.
11 (one-tailed test), slightly above the marginal threshold of.10.
Thus, the relationship between EIWB and development in CEE
nations appears to vary in magnitude through time.
For ease of interpretation, Fig. 5 plots the estimated effects of
GDP per capita for each year, derived from the estimated model.
The estimated effect of GDP per capita on EIWB is  .995 in 1992
and decreases successively in absolute value until 1994, where the
Table 2
Findings for the Regression of the Energy Intensity of Well-Being, 1992–2010 in 12 CEE
Nations: Two-Way Fixed Effects PW Regression With PCSE and an AR(1) Correction.

Key independent variables Coefficients Panel corrected standard errors

2.6. The estimated model
GDP per capita  .995nn .447
In the panel analysis we estimate the following two-way fixed GDP per capita  1993 .347nnn .073
GDP per capita  1994 .668nnn .086
effects model:
GDP per capita  1995 .467nnn .101
 .230nn
EIWBit ¼ β 1GDP per capitait þ β 2 year 1993t þ⋯ þ β 19 year2010t GDP per capita 1996 .118
GDP per capita  1997 .172n .122
þ β20 GDP per capitait  year 1993t þ ⋯ GDP per capita  1998 .242nn .113
GDP per capita  1999 .144 .118
þ β37 GDP per capitait  year 2010t GDP per capita  2000 .240nn .128
GDP per capita  2001 .424nnn .125
þ β38 Democratizationit þ β39 Health Expendituresit GDP per capita  2002 .383nnn .124
GDP per capita  2003 .490nnn .124
þ β40 Inequalityit þ β41 Manufacturingit GDP per capita  2004 .447nnn .122
GDP per capita  2005 .434nnn .120
þ β42 Exportsit þui þ eit GDP per capita  2006 .481nnn .132
GDP per capita  2007 .577nnn .131
where the dependent variable, EIWBit, is the energy intensity of GDP per capita  2008 .412nnn .119
well being, and the model includes GDP per capita (β 1GDP GDP per capita  2009 .155n .108
 .200nn
per capitait Þ, the period-specific intercepts β6 year1965t þ ⋯ þ β14 GDP per capita 2010 .110

year2005t , and the interaction for GDP per capita and the dummy Control variables
variables for each year included in the study (β 20 GDPper Democratization  .031n .017
Health expenditures as % GDP  .061 .104
capitait  year1993t þ ⋯ þ β 37 GDP per capitait  year2010t ) where Domestic inequality 2.096n 1.560
1992 is the reference category. The model also includes the Manufacturing as % of GDP .339n .229
controls for democratization (β 38 Democratizationit Þ, public health Exports as % of GDP .003 .003
expenditures as % of GDP (β39 Health Expendituresit Þ, domestic R-sq overall .993
inequality (β40 Inequalityit Þ, manufacturing as % of GDP N 228
# estimated coefficients 54
(β 41 Manufacturingit Þ, exports as % of GDP (β 42 Exportsit ), the N/# estimated coefficients 4.222
country-specific intercepts (ui) and the disturbance term unique Rho .621
to each country at each point in time (eit). The coefficient for GDP
per capita is the unit change in the dependent variable in 1992 for Models include unreported country-specific and year-specific intercepts (2-way
fixed effects); GDP per capita and Manufacturing as % of GDP are logged (ln) to
each unit increase in GDP per capita for the same year. The overall minimize skewness; p value for GDP per capita  1999 is approximately .11 (one-
effect of GDP per capita for the other time points (e.g., 1993, 1994, tailed).
…, 2009, 2010) equals the sum of the coefficient for GDP per capita nnn
po .01 (one-tailed tests).
nn
(i.e., its effect in 1992) and the appropriate interaction term if the p o .05 (one-tailed tests).
n
latter is statistically significant (Allison, 2009). p o .10 (one-tailed tests).
 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
-.2
-.4
-.6
-.8
-1 1992
1993
1994
1995
1992 1993 1994 1995 1996 1997 1998
-0.995 -0.648 -0.327 -0.528 -0.765 -0.823 -0.753
Fig. 5. Estimated coefficients of GDP per capita (ln).
estimated coefficient for GDP per capita is  .327. This is followed
by increases in the absolute value of the effect, with an estimated
coefficient of  .528 for GDP per capita in 1995, .765 in 1996, and
.823 in 1997. The estimated effects of development on EIWB
remain relatively stable through 2000, with coefficients of  .753
in 1998,  .851 in 1999, and .755 in 2000, followed by a pattern
of decreases in absolute value, which lasts until 2007, the latter of
which yields a coefficient for GDP per capita of  .418. The years
2008 and 2009 exhibit sharp increases in absolute values, with
estimated coefficients of .583 in 2008 and  .840 in 2009,
followed by a slight decrease in absolute value in 2010, which
yields an estimated coefficient of  .795 for level of development.
For Fig. 5 we treat the interaction between GDP per capita and the
year 1999 as statistically significant, while recognizing that its p-
value is slightly above conventional benchmarks for statistical
significance. We realize that a more normative and conservative
approach would conclude that the estimated effect of GDP per
capita on EIWB in 1999 is not significantly different than the
estimated effect of GDP per capita for the year 1992. However, we
remind readers of our relatively modest sample size and large
number of estimated coefficients due to the two-way fixed effects,
characteristics both known to reduce the possibility of estimated
standard errors yielding small p-values.
Turning briefly to the control variables, the results presented in
Table 2 indicate that level of democratization negatively affects
EIWB in CEE nations, while domestic income inequality and
manufacturing as percent of GDP both increase EIWB. The esti-
mated coefficients for public health expenditures as percent of
GDP and exports as a percent of GDP are both nonsignificant.
While not the focus of this study, the findings for the control
variables point to fruitful avenues for future studies of EIWB in CEE
nations as well as for nations in other regions of the world.
4. Discussion and conclusion
This study was motivated by the need for integrated research on
sustainability in Central and Eastern European (CEE) nations that
considers the interconnections between human well-being and
environmental/natural resource conditions, and how economic devel-
opment affects them. Drawing from recent research, we designed and
calculated an annual measure of the energy intensity of well-being
(EIWB) for a sample of CEE nations that quantifies the adjusted ratio of
425
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
-0.851 -0.755 -0.571 -0.612 -0.505 -0.548 -0.561 -0.514 -0.418 -0.583 -0.840 -0.795
per capita energy consumption and average life expectancy rates for
the 1992 to 2010 time period. To assess the potentially changing
effects of economic development on our measure of EIWB, we
employed two-way fixed effects panel model estimation techniques
and statistical interactions between time and annual measures of
economic development, measured as GDP per capita, while controlling
for other potentially relevant factors.
The estimated effects of development on EIWB were negative for
each year in the period of study, which is generally desirable from a
broader sustainability perspective. However, the results indicated that
in the early years of the time period for the study, economic
development became relatively less-sustainable—at least in the con-
text of EIWB, with the estimated negative coefficient becoming
notably smaller for the year 1994 relative to 1992. This was followed
by a 5 year period of increasing sustainability with the estimated
negative effect of GDP per capita becoming increasingly larger through
the year 1999. In other words, the first seven years of the study
experienced a period of notably decreasing effects of development on
sustainability, followed by a trend of increasing effects of development
on sustainability. After this initial period, the CEE nations in the sample
experienced a general trend of decreasing impacts of development on
sustainability that lasted until 2007. However, from 2007 through
2010, the results suggest that development in CEE nations became
increasingly more sustainable (as the negative effect of GDP per capita
increased in value), at least in the context of the energy intensity of
human well-being. An unanswered question, which requires data
unavailable at the time of this study, is what has happened in CEE
nations since 2010? Has the more recent trend of enhanced sustain-
ability continued, has the relationship between EIWB and economic
development stabilized, or has it become less sustainable? This is a
critically important research question with significant policy implica-
tions that we will pursue as the necessary data become available.
Our results differ significantly from past cross-national research.
In particular, the findings for the current study are far more encoura-
ging from a sustainability perspective than prior work on development
and EIWB where economic growth is usually seen as having a
deleterious or at least not ameliorative effect on sustainability (e.g.,
Dietz et al., 2012; Knight and Rosa, 2011). There are many reasons why
this could be the case. First, the samples are quite different in past
research and tend to exclude most CEE nations, and our employed
measure of EIWB focuses on energy consumption instead of other
commonly used environmental stress measures. Second, unlike past
research, we consider how the relationship between EIWB and
development might change in magnitude through time. While Dietz
426 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
et al. (2012) conduct a longitudinal study and find evidence of a
quadratic relationship between their measure of EIWB and economic
development, which shows the relationship is nonlinear, their analysis
is not conducted in such a way that considers how the estimated effect
of development on EIWB might change through time within nations.
Overlooking the possibility of such temporal shifts can lead to
incomplete pictures of the potentially dynamic relationships between
societies and the environment. Third, it might simply be the case that
development, at least from 1992 to 2010, is relatively more sustainable
in CEE nations than in other regions of the world. In particular, while
the transition from the state socialist economy towards a market
economy was initially disruptive, in recent years the transition seems
to have allowed economic growth to improve the energy intensity of
human well-being.
The changing effects of development on EIWB, and especially
the trend of a reduced effect of development towards the end of
the time period under investigation suggests that perhaps the
nations of Central and Eastern Europe have increasingly experi-
enced, in general, various internal pathways of development that
involve reductions in the amount of energy required to maintain
and enhance the quality of life for their citizens. Such findings are
consistent with more optimistic approaches concerning the pos-
sibility for development to become increasingly sustainable
through time as internal components of societies transform in
meaningful ways, such as ecological modernization theory within
sociology (e.g., Mol, 2001).
As important, the estimated negative effect of development on
EIWB is largest and thus most sustainable in 1992, the first year of
the time period under investigation. This was followed by a notable
change the next two years in which the estimated effect of devel-
opment on EIWB in 1994 became the least negative and thus the
least sustainable for the entire 1992 to 2010 timeframe. These
findings suggest that the forms and structures of whatever levels
of development were taking place in the Central and Eastern
European nations in these initial years were perhaps quite different
than in the later years, which further highlights that the internal
characteristics of development as well as levels of development are
perhaps incredibly important to consider when seeking to under-
stand broader processes of and pathways towards greater sustain-
ability (e.g., Dietz et al., 2012; Jorgenson et al., 2012; York, 2008).
In the next steps of this research, besides including data for
years more recent than 2010 as they become available, we plan to
conduct similar analyses for samples of nations for many regions
in addition to CEE. If future stages of this research – like the
current study – point to CEE nations' unique potential relative to
other world regions for enhanced sustainability in the context of
economic development and EIWB, a closer examination of the
institutional conditions and structural mechanisms in these tran-
sition economy nations that lead to their recent pathways to
enhanced sustainability could provide valuable insights for similar
pathways of and policies for enhanced sustainability in societies
throughout the world. Such closer examinations might also
provide valuable insights that could help lead to even more
enhanced future sustainability outcomes in these nations as well.
References
Alekseyko, A., Giedraitis, V., 2013. Economic Growth and Energy Security: Economic
and Environmental Solutions in a Changing Europe. Lap Lambert Academic
Press.
Allison, P., 2009. Fixed Effects Regression Models. Sage, Thousand Oaks.
Babones, S., 2008. Income inequality and population health: correlation and
causality. Soc. Sci. Med. 66, 1614–1626.
Banuri, T., Hallstrom, N., 2012. A global programme to tackle energy access and
climate change. Dev. Dialogue 61, 264–279.
Baum, C., 2006. An Introduction to Modern Econometrics Using Stata. Stata Press,
College Station, TX.
Beaglehole, R., Bonita, R. (Eds.), 2009. Oxford University Press.
Beck, N., Katz, J., 1995. What to do (and not do) with time-series cross-section data.
Am. Polit. Sci. Rev. 89, 634–647.
Bobak, M., Pikhart, H, Rosa, R., Hertzman, C., Marmot, M., 2000. Socioeconomic
factors, material inequalities, and perceived control in self-rated health: cross-
sectional data from seven post-communist countries. Soc. Sci. Med. 51,
1343–1350.
Boeri, T., 2000. Structural Change, Welfare Systems, and Labour Reallocation:
Lessons from the Transition of Formerly Planned Economies. Oxford University
Press.
Brady, D., Kaya, Y., Beckfield, J., 2007. Reassessing the effect of economic growth on
well-being in less-developed countries, 1980–2003. Stud. Comp. Int. Dev. 42,
1–35.
Cameron, C., Trivedi, P., 2009. Microeconometrics Using Stata. Stata Press, College
Station, TX.
Chiebug, J., Gandhi, J., Vreeland, J., 2009. Democracy and dictatorship revisited.
Public Choice 143, 67–101.
Clark, B., Jorgenson, A., Kentor, J., 2010. Militarization and energy consumption: a
test of treadmill of destruction theory in comparative perspective. Int. J. Sociol.
40, 23–43.
Clark, R., 2011. World health inequality: convergence, divergence, and develop-
ment. Soc. Sci. Med. 72, 617–624.
Dietz, T., Rosa, E., York, R., 2009. Efficient well-being: rethinking sustainability as
the relationship between human well-being and environmental impacts. Hum.
Ecol. Rev. 16, 114–123.
Dietz, T., Rosa, E., York, R., 2012. Environmentally efficient well-being: is there a
Kuznets curve? Appl. Geogr. 32, 21–28.
Ehrhardt-Martinez, K., Crenshaw, E., Jenkins, C., 2002. Deforestation and the
environmental Kuznets curve: a cross-national investigation of intervening
mechanisms. Soc. Sci. Q. 83, 226–243.
Firebaugh, G., Beck, F., 1994. Does economic growth benefit the masses? Growth,
dependence, and welfare in the third world. Am. Sociol. Rev. 59, 631–653.
Giedraitis, V., Notten, T., Rastenienė., A., 2009. Business clusters and the potential of
the biotechnology sector in Lithuania. Ekonomika 87, 81–89.
Gilpin, R., 2001. Global Political Economy. Princeton University Press.
Grossman, G., Krueger, A., 1995. Economic growth and the environment. Q. J. Econ.
110, 353–377.
Hyns, B., 2005. Emerging inequalities in Central and Eastern Europe. Annu. Rev.
Sociol. 31, 163–197.
Jorgenson, A., 2003. Consumption and environmental degradation: a cross-national
analysis of the ecological footprint. Soc. Prob. 50, 374–394.
Jorgenson, A., 2011. Carbon dioxide emissions in Central and Eastern European
nations, 1992–2005: a test of ecologically unequal exchange theory. Hum. Ecol.
Rev. 18, 105–114.
Jorgenson, A, Austin, K., Dick, C., 2009. Ecologically unequal exchange and the
resource consumption / environmental degradation paradox: a panel study of
less-developed countries, 1970–2000. Int. J. Comp. Sociol. 50, 263–284.
Jorgenson, A., Clark, B., 2011. Societies consuming nature: a panel study of the
ecological footprints of nations, 1960–2003. Soc. Sci. Res. 40, 226–244.
Jorgenson, A., Clark, B., 2012. Are the economy and the environment decoupling?
A comparative international study, 1960–2005. Am. J. Sociol. 118, 1–44.
Jorgenson, A., Clark, B., 2013. The relationship between national-level carbon
dioxide emissions and population size: an assessment of regional and temporal
variation, 1960–2005. PLoS One 8 (2), e57107.
Jorgenson, A., Clark, B., Giedraitis, V., 2012. The temporal (in)stability of the carbon
dioxide emissions/economic development relationship in Central and Eastern
European nations. Soc. Nat. Resour. 25, 1182–1192.
Kennelly, B., O'Shea, E., Garvey, E., 2003. Social capital, life expectancy, and
mortality: a cross-national examination. Soc. Sci. Med. 56, 2367–2377.
Knight, K, Rosa, E., 2011. The environmental efficiency of well-being: a cross-
national analysis. Soc. Sci. Res. 40, 931–949.
Liu, J., et al., 2007. Complexity of coupled human and natural systems. Science 317,
1513–1516.
Marquart-Pyatt, S., 2012. Environmental concern in cross-national context: how do
publics in Central and Eastern Europe compare with other regions of the
world? Czech Sociol. Rev. 48, 441–466.
Mazur, A., 2011. Does increasing energy or electricity consumption improve quality
of life in industrial nations? Energy Policy 39, 2568–2572.
Mazur, A., Rosa, E., 1974. Energy and life-style. Science 186, 607–610.
Miskinis, V., Vilemas, J., Konstantinaviciute, I., 2006. Analysis of energy consump-
tion and energy intensity indicators in Central and Eastern European countries.
Energy Stud. Rev. 14, 171–188.
Mol, A., 2001. Globalization and Environmental Reform. MIT Press, Cambridge, MA.
Moore, S., Teixeira, A., Shiell, A., 2006. The health of nations in a global context:
trade, global stratification, and infant mortality rates. Soc. Sci. Med. 63,
165–178.
Prescott-Allen, R., 2001. The Well-Being of Nations: A Country-by-Country Index of
Quality of Life and the Environment. Island press, Washington, DC.
Preston, S., 1975. The changing relation between mortality and level of economic
development. Popul. Stud. 29, 231–248.
Reid, et al., 2005. Ecosystems and Human Well-Being: Synthesis. Island Press,
Washington DC.
Ringold, P., Boyd, J., Landers, D., Weber, M., 2012. What data should we collect?
A framework for identifying indicators of ecosystem contributions to human
well-being. Front. Ecol. Environ. 11, 98–105.
Roberts, T., Parks, B., 2007. A Climate of Injustice: Global Inequality, North–South
Politics, and Climate Policy. MIT Press, Cambridge, MA.
 A.K. Jorgenson et al. / Energy Policy 66 (2014) 419–427
Rosa, E., Diekmann, A., Dietz, T., Jaeger, C. (Eds.), 2010. Human Footprints on the
Global Environment: Threats to Sustainability. MIT Press.
Rosa, E., York, R., Dietz, T., 2004. Tracking the anthropogenic drivers of ecological
impacts. Ambio 33, 509–512.
Shandra, J., Nobles, J., London, B., Williamson, J., 2004. Dependency, democracy, and
infant mortality: a quantitative cross-national analysis of less developed
countries. Soc. Sci. Med. 59, 321–333.
Steinberger, J., Roberts, T., 2010. From constraint to sufficiency: the decoupling
of energy and carbon from human needs, 1975–2005. Ecol. Econ. 70,
425–433.
427
Wickrama, K., Mulford, C., 1996. Political democracy, economic development, disarticu-
lation, and social well-being in developing countries. Sociol. Q. 37, 375–390.
Wooldridge, J., 2005. Introductory Econometrics: A Modern Approach, third ed.
Thomson, Mason.
World Bank, 2013. World Development Indicators (CD ROM Version). World Bank,
Washington, D.C..
York, R., 2008. De-Carbonization in former Soviet Republics, 1992–2000: the
ecological consequences of de-modernization. Soc. Prob. 55, 370–390.
York, R., 2012. Do alternative energy sources displace fossil fuels? Nat. Clim. Change
2, 441–443.