Environ Sci Pollut Res DOI 10.1007/s11356-016-6982-9
RESEARCH ARTICLE
The environmental Kuznets curve and CO2 emissions in the USA Is the relationship between GDP and CO2 emissions time varying? Evidence across economic sectors Emilio Congregado 1 & Julia Feria-Gallardo 1 & Antonio A. Golpe 1 & Jesús Iglesias 1
Received: 15 February 2016 / Accepted: 25 May 2016 # Springer-Verlag Berlin Heidelberg 2016
Abstract In this paper, we analyze the existence of the environmental Kuznets curve as reported by Kuznets (Am Econ Rev 5:1–28, 1955) by using the methodology proposed by Kejriwal and Perron (J Econ 146:59–73, 2008, J Bus Econ Stat 28:503–522, 2010) and applying Jaunky’s (Energy Policy 39(3):1228–1240, 2011) specification using quarterly data from 1973:1 to 2015:2. We also allow different behaviors across time and identify it by economic sectors. Our results show the existence of the environmental Kuznets curve (EKC) in the USA only when we allow for structural breaks. Interestingly, the industrial sector shows a different pattern than do other economic sectors; with the beginning of the economic crisis, it appears to have abandoned the objective of the environmental stabilization found until then. Keywords CO2 emissions . Growth . Cointegration . Time varying . Structural breaks . USA
Introduction As in most developed countries, in the USA, there has been a growing concern since the early 1990s regarding the establishment of a production system with energy-efficient technologies and appropriate practices to preserve the environment. Although globally, following the publication of the report known as The Limits to Growth (1972) and the Brundtland Responsible editor: Philippe Garrigues * Antonio A. Golpe
[email protected]
1
Departamento de Economía, Universidad de Huelva, Huelva, Spain
report in 1987 by the World Commission of Environment and Development, there has been growing environmental awareness that has created a commitment by industrialized countries to adjust the emissions of polluting gases. Dinda (2004) argued that developed economies must forgo income growth and that developing countries must restrain their growth ambitions to reduce carbon emissions. Thus, the USA appears to be at the heart of the environment debate at the recent 2015 Paris talks, where countries re-committed to reducing CO2 emissions from 40 % to 70 % in 2050 compared with the 2010 levels. In that direction, the USA created a legal reform on August 3, 2015. President Obama, through the Environmental Protection Agency, announced the Clean Power Plan, which is a historic and important step in reducing carbon pollution from power plants and takes real action on climate change.1 The environmental economics literature has not ignored this issue, as illustrated by Grossman and Krueger (1991), who suggested that the relationship between economic development and environmental quality has the form of an inverted U-shape that can be explained by the idea that economies experience environmental degradation in their infancies with low per capita income levels. Once a determinate level of per capita income is achieved, economies are able to establish less polluting industries, as proposed by the environmental
1
In this action, the Environmental Protection Agency (EPA) established final emission guidelines for states to follow when developing plans to reduce greenhouse gas (GHG) emissions from existing fossil fuel-fired electric generating units (EGUs). Specifically, the EPA established subcategories of existing fossil fuel-fired EGUs, fossil fuel-fired electric utility steam generating units and stationary combustion turbines; and stateimplementation of state plans that establish emission standards emission performance rates, which may be accomplished by meeting the state goals.
Environ Sci Pollut Res
Kuznets curve (EKC hereafter). In other words, the EKC hypothesis sustained the idea that pollution transitions from a normal good to an inferior good as income increases. A large number of papers have been produced that attempt to explain the relationship between economic growth and pollution (Arrow et al. 1995; Grossman and Krueger 1995; Martínez-Alier 1995; Carson and Mccubbin 1997; Suri and Chapman 1998; Torras and Boyce 1998; Stern 2004; He and Richard 2010; Ajmi et al. 2015; Doda 2014, among others). Unfortunately, the validity of this relationship has not yet been established (see Aslanidis 2009; He and Richard 2010; Jordan 2010; Payne 2010, or Stern 2004). This ambiguity is partially due to the lack of robustness of the different studies because the results are highly sensitive to the econometric estimation technique used (Stern 2004 or recently Kaika and Zervas 2013a, 2013b). It is notable that although a number of studies have addressed this relationship, in recent years, they have analyzed the cointegration relationship between economic growth and CO2 emissions. Wagner (2008) noted that per capita gross domestic product (GDP) and CO2 emissions data series are often non-stationary, and this difficulty has not been adequately addressed in the EKC literature. The existence of unit roots in both variables makes it mandatory to test for the possible existence of cointegration relationships between them (regardless of the econometric specification). Furthermore, the amplitude of the temporal series indicates the potential existence of structural changes; therefore, results that are obtained by cointegration would show a linear relationship over the entire period, which would obscure the possibility that the relationship was different over time. In addition, the variation of the elasticity over different phases of economic development would impose certain uncertainties on emissions forecasting (Sheldon 2014). Consequently, finding the appropriate environmental policies is extremely important because of the relationship between GDP and CO2 emissions as a relationship. The main contribution of this paper consists of providing the know-how to vary the elasticity of the EKC for the USA over time and to further distinguish this behavior via testing by economic sectors. In accordance with the specification proposed by Jaunky (2011), we present a novel methodology to verify the EKC hypothesis in USA by testing the possible existence of cointegration relationships between CO2 emissions and economic growth and distinguishing by sectors. In particular, the specification proposed by Jaunky (2011)2 suggests the need to explore the polluter behavior relationship between CO2 emissions and income through the elasticity. This specification avoids some of the problems alerted by the previous literature concerning the functional form of the 2
The Jaunky (2011) specification is based on Narayan and Narayan (2010). We make references to the Jaunky specification because we also apply the DOLS analysis.
relationship. Specifically, it has been postulated that this functional form can be quadratic or cubic (see Stern 2004). Nevertheless, Narayan and Narayan (2010) support a significant problem of multi-collinearity among per capita income, per capita income squared, and per capita income cubed. Additionally, when this ratio is approached by a linear approximation, it is supposed that the relationship is identical throughout the period, a question that is not established by EKC hypothesis. Finally, and even more importantly, Dinda (2004) indicates that a key factor at play in the EKC is the structural changes. For all that, to avoid this rigid constraint and allow the relationship to be time variant, in our work, we propose following the Jaunky specification through the analysis of structural breaks, which presents the advantage to be able to discriminate between changes produced throughout the time series studied, in the long run, and observe what the elasticity is founded in each of the regimes. In sum, we propose to test through these elasticities if the functional form proposed by the EKC hypothesis is confirmed. After confirming the existence of and obtaining the elasticity between the variables, we study the possibility of structural breaks in the relationship using the methodology proposed by Kejriwal and Perron (2008, 2010). Finally, we analyze the different elasticities obtained for each period to test for the existence of the EKC in the USA and by sectors. The proposed results indicate that the US economy has, in recent years, begun to fulfill the environmental Kuznets curve. The remainder of this study is as follows. In the second section, we review the questions that the literature suggests regarding EKC analysis, the evidence for the US case and the theoretical framework. Subsequently, the data and methodology used appear in the third section. The fourth section presents the results that will ultimately inform the main conclusions.
A brief review of the EKC The hypothesis proposed by the EKC is that pollution emissions and income have an inverted U-shaped relationship. An economy in its initial stage generates high pollution levels, and as it develops, it is capable of reducing pollution levels (surveys of the EKC include those of Dinda 2004; or Stern 2004; and more recently, Pasten and Figueroa 2012). The empirical work of Grossman and Krueger (1991) was designed to measure the influence of free trade agreements in North America on the environment. As their most important conclusion, the researchers argued that the increase in international trade and therefore, economic growth also led to less pollution.3 From a theoretical perspective, the direct polluter 3
The works of Beckerman (1992) and Panayotou (1993) were the first to use the term EKC as an extension of the relationship between the level of inequality and income per capita as Kuznets (1955) proposed.
Environ Sci Pollut Res
relationship between growth and CO2 emissions set out in the EKC hypothesis has resulted in an extensive field of research in environmental economics (Grossman and Krueger 1995; Cole et al. 1997; Borghesi 1999; List and Gallet 1999; López and Mitra 2000; Hettige 2000; Andreoni and Levinson 2001; Harbaugh et al. 2002; Pfaff et al. 2004; Dinda 2004; Kijima et al. 2010; Bo 2010; Esteve and Tamarit 2012; Iglesias et al. 2013; Kaika and Zervas 2013a, b; Niu and Li 2014). The body of EKC literature is very large, and the results are, at best, mixed. However, recent studies indicate that the economic growth process does not reduce the CO2 emissions that are related to economic growth through energy consumption (Kaika and Zervas 2013a, b). This debate over the acceptance of the EKC hypothesis was raised by the work of Ekins (1997) and de Bruyn and Heintz (1999) because they did not find evidence for this hypothesis for several pollutants that they analyzed. Later, Dinda (2004) analyzed the curve interpretation and deviations from the methodology and techniques employed. As is well known, the literature offers many reasons to argue this lack of unanimity, in accordance with the cited works of Kaika and Zervas, the major critiques on the EKC speculation related to the normal distribution of world income, the feedback from environmental degradation to economic growth, the characteristics of the pollutants in question, various econometric issues, the evolution of consumption when income rises and the assumption regarding a common developmental pattern in all countries. Another factor at play to recognize the existence of the EKC is the geographical area. It can be argued that the data were not as comparable across countries as one may hope because different methods and procedures may have been used in each country; the best comparative data may be the time trends within a country (refer to World Resources Institute 1994, 1996). Conversely, it is also argued in the literature that missing factors may exist that would determine the environmental progress (see He and Richard 2010), public policies, transparency and participation (Carpentier 2006), problems caused in global terms such as longevity in the atmosphere (Niu and Li 2014), or how to establish the turning point of the relationship between income and pollution (Selden and Song 1994 and List and Gallet 1999). Thus, several studies have noted the influence of the energy crises as a breakdown in the relationship proposed in the EKC; this reinforces the importance of studying the over time rather than relying on a static analysis (refer to Moomaw and Unruh 1997 or Payne 2010). Soytas et al. (2007) and Jalil and Mahmud (2010) discussed the advantages of using time series data to test the proposed polluter relationship of the EKC hypothesis. Summarizing the deviations in the interpretation of the EKC, the work of Zhao et al. (2013) observes that the country degree of development featured differing EKC patterns with a very rapid increase in CO2 emissions; they concluded that
although the overall development of a country or region may not follow an EKC pattern, the EKC hypothesis may continue to be justified during different periods of economic growth. A more complete explanation is provided by Roca and Padilla (2003), who describe different arguments presented in the literature to explain the functional form of an inverted-U relationship between CO2 and growth. This is caused by technological advances that allow for reductions in the CO2 emissions that are caused by the substitution of the service sector for the industrial sector. The service sector is more productive, and it causes reductions in environmental pressures per unit of income. Alternatively, this change could occur because the income increase leads to the consumption of goods with high environmental quality (López 1994; Selden and Song 1994; McConnell 1997; Roca 2003).4 Undoubtedly, the examination of the Kuznets theory by sectors is critical in the EKC literature. The shift in the composition of output in the economy could affect the energy consumption-output relationship because different industries may have different energy intensities (Bowden and Payne 2010). Furthermore, some studies have shown that secondary industry is the sector that generates more CO2 emissions (see among others Cole 2008). Thus, to find an adequate response to this, it is necessary to include matters related to the capture of the GDP allocation to the three productive sectors. To consider services as less polluting than the industry sector requires caution because the services sectors in developed countries have occupied the largest proportion of total GDP since 1970; consequently, the empirical confirmation of the original Kuznets theory, which laid the foundation for the EKC concept, appears to be confirmed over a longer period (see Kaika and Zervas 2013b). Furthermore, Thoma (2004) describes several reasons to be interested in the observation of the energy consumption by sectors in the business cycles. First, the peak annual demand for electricity is useful for adjusting forecasts of the demand for electricity according to the expected changes in macroeconomic conditions. Second, the effect on energy usage from the expected changes in the composition of output such as a change in the percentage of total output explained by the industrial and commercial sectors to be estimated, which provides an additional means to improve the forecasts of electricity usage. Third, firms that operate in the futures markets for electricity usage have a keen interest in forecasting future energy usage. Therefore, forecasts of future macroeconomic conditions can be used to guide electricity 4 The argument over technological change, as detailed in the work of Roca and Padilla (2003), is ruled by the so-called Brebound effect^, by which the increase in environmental efficiency derived from technology leads to greater technological demands that nullify this effect. In addition, it would be unconvincing if environmental improvement resulted from the replacement of the service sector by industry, if we suppose that the environmentally most problematic sectors produce inferior goods, which is not likely (Torras and Boyce 1998).
Environ Sci Pollut Res
futures market decisions and to improve the assessment of the probability of a blackout. In the paper by Kaika and Zervas (2013b), they question the behavior of CO2 emissions variances according to the sectors studied. The researchers specifically argue that there are doubts regarding whether the GDPseries captures the transition of production to the three productive sectors in empirical estimations and whether services are less polluting than industry activities. All of the described controversy raises three alternative forms that the relationship between growth and CO2 emissions could take: (i) monotonically increasing, (ii) an inverted U-shaped, or (iii) an N-shape; this implies that the decrease in pollution is a temporary occurrence (Galeotti et al. 2006 or Kaika and Zervas 2013a). To elucidate the question regarding the interpretation of the EKC, in this paper, in accordance with the Jaunky (2011) approach, we check the polluter relationship derived by the EKC theory in accordance with the interpretation of the income elasticity of CO2 emissions. CO2 emissions by sectors in the USA and EKC The role of the US economy in the total world energy consumption and production is vital. According to the EPA, Carbon dioxide (CO2) emissions in the USA increased by approximately 7 % between 1990 and 2013. Because the combustion of fossil fuel is the largest source of greenhouse gas emissions in the USA, changes in emissions from fossil fuel combustion have historically been the dominant factor that affects the total US emission trends. Changes in CO2 emissions from fossil fuel combustion are influenced by many long- and short-term factors, including population growth, economic growth, changing energy prices, new technologies, changing behavior, and seasonal temperatures. Between 1990 and 2013, the increase in CO2 emissions corresponded to increased energy use by an expanding economy and population and an overall growth in emissions from electricity generation. Transportation emissions also contributed to the 7 % increase, largely due to an increase in miles traveled by motor vehicles. The energy market in the USA presents a different behavior across economic sectors. In this sense, reading the US Energy Information Administration in the Annual Energy Outlook 2015 (AEO 2015), we can observe that the average annual growth in electricity use in the USA has decreased from 9.8 % per year in the 1950s to 0.5 % per year over the past decade. However, the historical trend of transportation energy consumption grew by an average of 1.3 % per year from 1973 to 2007. As for the industrial sector, that represents 34 % of total US delivered energy consumption, and the projections supports growth at an annual rate of 0.7 % from 2013 to 2040. Finally, in the residential and commercial sector, the end-use energy intensity also decreased as a result of increases in the efficiency of equipment for many end uses. As a consequence, energy demand is generating a pollutant process also reflected
in CO2 emissions average by sector. For this reason, the CO2 emissions by sector reported by the US Environment Protection Agency also show that the scenario among sectors presents important differences. The Electricity production generates the largest share of greenhouse gas emissions, approximately 30 % of the total in 2014. The transportation sector accounted for 26 % of 2014 greenhouse gas emissions, while the industry sector accounted for 21 % of 2014 greenhouse gas emissions. Finally, the commercial and residential sector caused 12 % of total 2014 greenhouse gas emissions in the USA. Given this scenario, in the absence of a sectoral approach, conclusions about the behavior of the different EKC behaviors might be hiding among them, in other words, presenting biased conclusions. Accordingly, our work will provide specific economic policies for each sector. Although there is an extensive body of literature on CO2 emissions from an international perspective, very few authors have considered the study of the EKC in a macro perspective studying only the USA. However, there is ample evidence that has been used in attempts to test the theory indirectly through the relationship between CO2 emissions and economic growth from an analysis of Granger causality, i.e., the recent work of Ajmi et al. (2015), which reveals a bidirectional time-varying causality, and there is a growing interest in understanding how this relationship varies before economic shocks (Heutel 2012; York 2012; or Jotzo et al. 2012; Fischer and Heutel 2013; Doda 2014). 5 Perhaps one of the most cited papers in the literature is Soytas et al. (2007), who did not find evidence of causality between either income and carbon emissions or income and energy use in the USA. In an approach aiming to test the causality by incorporating possible asymmetries in the relationship, Shahiduzzaman and Layton (2015) analyzed the changes in CO2 emissions over business cycle recessions and expansions using yearly data from 1949 and monthly data from 1973 for the USA. The researchers found that energy consumption has a positive impact on CO2 emissions but provide poor evidence in support of the EKC hypothesis. Beyond the possible causality between growth and CO2 emissions and according to a specific check of the evidence on the EKC hypothesis in the USA, Carson et al. (1997), using a 1990 cross-section of state-level point-source emissions for some pollutants, found that per capita emissions of all pollutants monotonically declined as income increased; this supported the EKC hypothesis. However, no relationship between changes in the income and per capita emissions of a panel dataset of air toxic emissions over a 6-year period (1989–1994) was found, although they revealed that states 5
Jaunky (2009) previously surveyed the link between CO2 emissions and productivity growth for 27 rich countries over the period 1974– 2000. Unidirectional causality running from productivity growth to CO2 emissions is found in the short run, whereas bidirectional causality is revealed in the long run. Productivity growth is found to exert a positive impact on CO2 emissions in the long run.
Environ Sci Pollut Res
with high initial incomes were associated with larger reductions in per capita emissions. Furthermore, Aldy (2005) estimated CO2 emissions (based on fossil fuel use) across the forty-eight continental US states from 1960 to 1999 and found evidence supporting an EKC relationship for the USA. In addition, to explore the incidence of shocks in the relationship proposed in the EKC hypothesis, Huntington (2005) and Lanne and Liski (2004) found an early break in 1913; over those two periods (before and after 1913), Huntington estimated a stable income elasticity of 0.9. Both Lindmark and Huntington emphasized the importance of technological advance rather than smooth EKC transitions. Regarding the importance that the sectors play in the differences in the possible existence of the EKC hypothesis, Thoma (2004) explored it using a time series analysis with US monthly data from 1973 to 2000; he applied a Grangercausality approach in which he investigated the reaction of energy consume the variation of the elasticity over different phases of economic development would impose certain uncertainties on emissions forecasting across business cycles by industrial production to residential, commercial, industrial, other, and total energy usage. He found that commercial and industrial sectors experience a pro-cyclical energy usage similar to that felt by the economy. Undoubtedly, different industries may have different energy intensities in CO2 emission levels (Bowden and Payne 2010). Furthermore, some studies have shown that secondary industry is the sector that generates more CO 2 emissions (see, among others, Cole 2008). However, it is necessary to include matters related to the capture of the GDP allocation in the three productive sectors; to consider services as less polluting than the industry sector requires caution because the services sectors in developed countries have occupied the largest proportion of total GDP since 1970. However, the behavior of consumers may be different; they are more resilient in a recession than industry is in the use of energy. In addition, credit constraints on lowerincome families may induce them to reduce their consumption of costly environmental goods. Thus, Bin and Dowlatabadi (2005) revealed that more than 80 % of the energy used and the CO2 emitted in the USA is a consequence of consumer demands and the economic activities to support these demands. Although the direct influences due to consumer activities represent 4 % of the US GD but 28 and 41 % of US energy use and CO2 emissions, respectively, the indirect influences (such as housing operations, transportation operations, food, and apparel) involve more than twice the direct energy use and CO2 emissions. Bowden and Payne (2010) also examined the causal relationship between energy consumption and real GDP in a multivariate framework; they found that the Toda and Yamamoto (1995) long-run causality tests reveal that the relationship between energy consumption and real GDP is not uniform across sectors. However, in the Granger-causality approach, the authors
revealed by the causality test that in absence between total and transportation primary energy consumption and real GDP, respectively. Finally, a bidirectional Granger-causality is present between commercial and residential primary energy consumption and real GDP, respectively. Finally, the researchers’ results indicate that industrial primary energy consumption Granger-causes real GDP. In sum, the researchers provided an alert that prudent energy and environmental policies should recognize the differences in the relationship between energy consumption and real GDP by sector. Nonetheless, most of the existing evidence regarding the testing of the EKC hypothesis in the USA. case has focused on the study at the sub-national level, in which the empirical evidence suggests that the estimated pollution-income relationships vary across states; this lends support to our premise that differences in consumer preferences matter in the empirical analyses of the EKC (Aldy 2005; Plassmann and Khanna 2006; Carson 2010; Baek and Gweisah 2013 or Burnett et al. 2013). Hence, despite several approaches that have attempted to achieve consensus on the existence of EKC in the USA, it appears that multiple approaches have done nothing but highlight the complexity of conclusive results for this purpose.
Data and methodology EKC and the CO2-GDP elasticity A particular analysis to test the EKC hypothesis consists of estimating the CO2-GDP elasticity. In order to provide an adequate approach to test de EKC hypothesis we take into account the long-run and univariate approach. On the one hand the distinction between short and long-run effects of economic growth on environmental degradation is important therefore equations with explicit short and long-term dynamics should be preferred (see Egli 2004). Furthermore, because the empirical objective of this paper is testing the cointegration and the structural breaks between economic growth and CO 2 emissions in the USA, we do not control for possible determinants of CO2 emissions. In this sense, List and Gallet (1999) argue that a reduced-form model allows one to measure the direct and indirect relationship between economic growth and environmental degradation controlling, for instance, by foreign trade, energy consumption, manufacturing from GDP, education, regulatory intensities, or technological changes. We discard the multivariate analysis because inclusion of additional variables would distort our primary objective and also produce conflicting results on the existence of EKC on individual countries that cannot be extrapolated as evidence of similar results for all countries. This paper acts in accordance with the specification proposed by Jaunky (2011) to test the EKC hypothesis
Environ Sci Pollut Res
in the USA, which incorporates the analysis by sector. To conduct a thorough analysis of this relationship, we have proposed to analyze distinguishing by economic sector because there are breakdowns which can be diluted on total emissions as some sectors can register very high emissions while others, low levels of emissions. We propose an equation to determine the polluter relationship between CO2 emissions and GDP: LCO2t ¼ μo þ μ1 LGDPt þ εt
ð1Þ
where LCO2 is the natural logarithm of the CO2 emissions; LGDP is the natural logarithm of the GDP6; εt is the error term; μ0 is the constant term; and μ1 estimates the CO2-GDP elasticity. An elasticity of μ1 > 1 (stage 1, environmental degradation) changes in GDP generates a more proportional increase in CO2 emission, i.e., the country is in an early stage of environmental sustainability characterized by minimal environmental responsibility. If 0 < μ1 < 1, the economy is stage 2 (environmental stabilization), where a GDP increase leads to a less proportional increase in CO2 emissions. Finally, for μ1 < 0, GDP and CO2 have a negative relationship. This is the last stage of the EKC, where the economy is in a phase of environmental optimization. Only when the economy is in Stage 3 is the EKC confirmed. However, the correct interpretation of this parameter requires an understanding of the effect that changes in the productive factors can have on the environment depends on the assumptions under which are set to the specified model. In this sense, the work of López (1994) and López and Yoon (2014) develop the basis for establishing a proper interpretation of the elasticity between revenue and CO2 emissions. In these works, it is emphasized that the analysis focuses on exogenously determined growth rather than on endogenous growth. In the neoclassical specification of a production function (see López 1994), the macro production function is characterized by constant returns to scale in the productions factor. His work reveals the implications of assuming that the factors of capital and labor endowments are fixed and mobile across sectors. Thus, he argues that whether originating in increased level of factor endowments or technical change, it will be followed by a proportional increase in air pollution, and in consequence, the price of pollution, that is, the pollution taxes, is endogenous. In other words, the intuition behind his result implies that the preferences, measured in terms of risk aversion, of the consumer for the environment are key factors in the elasticity, and at 6 The exercise was also conducted using variables in per capita terms. The results are very similar to those presented in this work and are available to the reader upon request. We consider it more appropriate to show the results in aggregate terms that establish no comparison between countries.
that time, that for firms, it is less costly to reduce pollution by substituting it for more conventional inputs. However, and more importantly, a high level of the coefficient of relative risk aversion (a-Frisch coefficient) implies that consumers would be willing to give up proportionally greater additional income as they become richer to purchase a better environment. Additionally, López and Yoon (2014) reveal that the limits to growth can be eluded through a Kuznets type process only if there is a sufficient degree of substitution flexibility in either production technology or consumer preferences. If an economy is endowed with flexibility, then economic growth can be sustained at positive levels while pollution falls over the long run. Thus, under the assumption of a fixed composition with production technology also fixed, emissions increase in proportion to the increase in the scale of economic activity, while for a given scale and technology, emissions will increase or decrease depending on changes in the composition of output toward more or less intensive goods in the production of pollutants. Finally, emissions per unit of production, that is, the emission intensity, may be reduced by improved technology. Therefore, depending on the relative importance of these three elements, there can be some pattern of behavior between per capita income and per capita emissions. This issue has great importance from a point of view of economic policy; it presents the possibility of economic growth compatible with environmental improvement. Figure 1 describes these three stages proposed by Jaunky (2011) with their respective elasticities. As we can observe in the figure, the GDP and CO2 polluter is time varying, which confirms the existence of structural breaks and different phases in the relationship. Accordingly, when an economy achieves a higher degree of development, the relationship between economic growth and CO2 emissions change. Consequently, it would cast doubt on the existence of the curve if the elasticity obtained for each period had a similar pattern to that described in Fig. 1. Data In our empirical analysis, we use quarterly data from the USA for the 1973:1–2015:2 period. The variable definitions and the main sources are described as follows. GDP is defined as the real gross domestic product (measured in billions of chained 2009 US dollars); data are taken from the US Bureau of Economic Analysis (BEA). Total CO 2 emissions are expressed in million metric tons of carbon dioxide and are distinguished by economic sector, total, commercial, electrical, industrial, residential, and transport; this information is available from the US Energy Information Administration (EIA). Time series are seasonally adjusted by X12 ARIMA. Figs. 2 and 3 in the appendix plot the time series.
Environ Sci Pollut Res Fig. 1 Environmental Kuznets Curve, elasticities across stages. Source: Compiled from the elasticities assumption proposed by Jaunky (2011)
Stage 1; Environmental degradation : µ1 > 1
Stage2; Environmental stabilization: 0< µ1 <1
Etapa 3; Environmental optimization (EKC) µ1 <0
CO2pc
GDPpc
To complete our analysis, the methodology used the following three steps. First, we propose a test for unit roots. Second, we seek structural breaks to confirm or reject the stability of the long-term relationship between GDP and CO2 emissions. Third, to complete the analysis, we measure the elasticity estimations of both variables to show what type of relationship is present in each possible regime. Methodology: testing for unit roots Because the estimation of a linear cointegration model requires the series to be non-stationary, we begin by testing for a unit root in the GDP and CO2 emissions. We apply the class of unit root tests developed by Ng and Perron (2001), which solve several statistical problems associated with more ‘conventional’ unit root tests.7 All test statistics formally examine the unit root null hypothesis against the stationary alternative. Table 1 reports the results. As shown, the null hypothesis of non-stationarity in levels is clearly non rejected at the usual significance levels for all variables. Thus, according to the results of these tests, these series would be I(1). Methodology: looking for structural breaks Having confirmed the non-stationarity of both variables, we now apply the tests for structural change proposed by Kejriwal
and Perron (2008, 2010). We use a 15 % trimming, which limits the maximum number of breaks allowed under the alternative hypothesis to 3.8 Both the intercept and the slope are allowed to change. Table 2 shows the results of the stability tests and the number of breaks selected by the sequential procedure proposed by Bai and Perron (2003) as well as the Bayesian and modified Schwarz information criteria (BIC and LWZ, respectively). The supFT(1), supFT(2), and supFT(3) test is not significant, which suggests that the data do not support a three-break model. However, the BIC and LWZ select three breaks for all cases, except the residential sector, in which two breaks are found. Moreover, in the commercial sector, two breaks occur when the LWZ criterion is applied. Consequently, we provide evidence against the stability of the long-term relationship. Interestingly, we find clustering among the timing of the breaks. Among the breaks identified, an important concentration of breaks is observed. In all sectors, the first breaks (1979:3 until 1981:4) occurred during the period involving the years of the second oil crisis, whereas the last breaks in every sector occurred from 2007:4–2008:2, which consists of the dates of the beginning of the last economic crisis. The structural breaks found in a brief review of the previous literature confirm that this break occurs around 1986 and 2000. Mork
7
In general, the majority of conventional unit root tests, such as the Dickey-Fuller tests and the Phillips-Perron tests, suffer from three problems. First, many tests have low power when the root of the autoregressive polynomial is close to but less than one (de Jong et al., 1992). Second, most tests suffer from severe size distortions when the moving-average polynomial of the first-differenced series has a large negative autoregressive root (Schwert 1989 and Perron and Ng 1996). Third, the implementation of unit root tests often requires the selection of an autoregressive truncation lag k; however, as discussed by Ng and Perron (1995), there is a strong association between k and the severity of size distortions and/or the extent of power loss. Ng and Perron (2001) solved these problems, and we refer to their article for further details.
8 We use 15 % trimming so that the maximum number of breaks allowed under the alternative hypothesis is 3. However, a lower trimming than 15 % would mean that the required years of each regime is at least 2.25 or 4.25 years, perhaps some significant periods, so we opted for the broader trimming permitted by this technique, as 15 % involved at least 6.5 years. However, if a higher number of breaks states that three breaks (four regimes), in the case of the existence of more breaks, occurs, than the number of years of each regimen should be 8.4 years maximum, so again it seems more representative to opt for the largest possible amplitude by this technique, i.e., 10.5 years corresponding to 42 quarters.
Environ Sci Pollut Res Table 1
Ng and Perrona,b and ADFc tests for a unit root
I(1) vs. I(0)
MZGLS α MZGLS t MSBGLS α MPGLS T ADF Critical values: Variable
MZGLS α MSBGLS α MZGLS t MPGLS T ADF
−4.734
Case: p = 1, c = −13.5 CO2t Total Commercial −7.962 −2.550
Electrical −3.265
Industrial −8.087
Residential −1.031
Transport −6.309
−1.377
−1.912
−0.843
−0.910
−2.002
−0.488
−1.661
0.291
0.240
0.331
0.279
0.248
0.474
0.263
18.290
11.688
26.310
21.255
11.294
48.625
14.439
−0.813 Case: p = 1, c = −13.5 10 % −14.2
−0.923
0.198
0.471
−1.857
−0.173
−1.398
5% −17.3
1% −23.8
0.185
0.168
0.143
−2.62
−2.91
−3.42
6.67
5.48
4.03
−3.15
−3.45
−4.04
yt
Notes: a
*, **, and *** denote significance at the 10, 5, and 1 % levels, respectively;
b
The MAIC information criteria are used to select the autoregressive truncation lag, k, as proposed in Perron and Ng (1996). The critical values are taken from Ng and Perron (2001), Table 1.
c
ADF stands for Augmented Dickey-Fuller (1979)
(1989) verifies that the oil price–macroeconomy relationship breaks down after the oil price collapse of 1986. For its part, Cologni and Manera (2008) found that the failure of the 1986 oil price collapse to produce an economic boom led several authors to hypothesize the existence of an asymmetric relationship between oil prices changes and economic activity. Narayan and Smyth (2008) also reveal that for the USA, the break occurred in 1988, just after the stock market crash in the USA and just prior to the Gulf War. Most recently, Bataa et al. (2015) have suggested that the administered (or fixed) price system collapsed in 1985 was followed by a near-collapse of OPEC in 1986, with its members unable to coordinate their production levels until 1998. Regarding the 2000 break, according to the National Bureau of Economic Research (NBER), there was a recession in 2001 in the US economy. Overall, the results of the Kejriwal-Perron tests suggest a model with three breaks and four regimes for total, electrical, industrial, and transport as well as a model with two breaks for commercial and residential. Because the above stability tests reject the null coefficient stability when the regression is spurious, we need to confirm the presence of cointegration among the variables. We use the residual-based test of the null of cointegration against the alternative of cointegration with unknown multiple breaks pro posed by Kejriwal (2008), Vk ̂ λ̂ . Arai and Kurozoumi (2007) show that the limit distribution of the test statistic,
Vk ̂ λ̂ , depends only on the timing of the estimated break fraction λ̂ and the number of I(1) regressors m. In our case (three-break model for total, electrical, industrial and transport, and two-break model for commercial and residential), critical values are obtained by simulation using 500 steps and 2000 replications. The Wiener processes are approximated by partial sums of i.i.d. N(0,1) random variables. Table 3 shows the results of the Arai-Kurozumi test, allowing three breaks (or two depending of the sector). Again, the level of trimming used is 15 %. The results show that the test Vk ̂ λ̂ cannot reject the null of cointegration with three (two) structural breaks in any case (except for the total and at 10 % level only), i.e., the relationship is time varying in all cases. Once the presence of structural breaks has been confirmed, to compare the coefficients obtained from a threebreak (two-break) model with those reported from a model without any structural break, we proceed with a comparison of the estimates of the elasticity CO2-GDP obtained from a three-break model with those obtained from the full sample.
Results For the full sample, we estimate the long-term regression model using the Dynamic Ordinary Least Squares
Environ Sci Pollut Res Table 2 Kerjiwal-Perron tests for testing multiple structural breaks supFT(1) supFT(2) supFT(3) UDmax Sequential
Total
Commercial
Electrical
Industrial
Residential
Transport
4.572 8.247 7.458 8.247
18.406*** 13.358*** 10.928*** 18.406***
20.750*** 13.591*** 12.328*** 20.750***
3.230 5.621 6.066 6.066
10.214 19.325*** 14.217*** 19.325***
6.268 8.681 6.957 8.681
0
0
1
0
0
1
3
2
2005:2 3
3 1981:2
2 1981:4
3 1979:3
2000:4
1998:1
2008:2
2008:2
2007:4
Breaks BIC
3
3
2004:4 3
LWZ Breaks
3 1980:1
2 1996:3
3 1981:4
1986:2
2008:2
2007:4
1985:3 2007:4
Notes: *,**, and *** denote significance at the 10 %, 5 % and 1 % levels, respectively. The critical values are taken from Kejriwal and Perron (2008, 2010).
(DOLS)9 estimation method of Stock and Watson (1993) that was extended by Shin (1994).10 The Shin (1994) approach is similar to the KPSS tests, which, in this case, is implemented in two stages.11 Therefore, the first step in our estimation strategy consists of the estimation of a long-term dynamic equation that includes leads and lags of the explanatory variables in the long-term regression model, i.e., the so-called DOLS regression: InðCO2t Þ ¼ μ0 þ μ1 lnðGDPt Þ þ
q X
φ j ΔlnGDPt þ ε j ð2Þ
j¼−9
In the second step, we use the statistic Ccμ, a LM-type test designed by Shin (1994), to test the null of cointegration against the alternative of no cointegration in DOLS regression.12 In Table 4, we report the estimates from the DOLS regression and the results from Shin’s test. The results show that the null of deterministic cointegration is not rejected at the 1 % significance level. The critical values are taken from Shin (1994), Table 1, from m = 1: Because there is evidence of the presence of structural breaks in the cointegration relationship, we divide our sample into four (three, depending on the sector) subsamples (periods) to analyze whether the CO2-GDP elasticity changes across the periods. We estimate equation (5) for the four 9 DOLS approach allows a robust correction to the possible presence of endogeneity in the independent variables and the serial correlation in error terms of OLS estimation. 10 To overcome the problem of the low power of classical tests under the presence of persistent roots in the residuals of the regression, Shin (1994) suggested a test in which the null hypothesis is cointegration. 11 These tests are called the Kwiatkowski et al. (1992) tests and assume the null hypothesis of stationarity. 12 Cμ is the test statistic for deterministic cointegration, i.e., when no trend is present in the regression.
(three) subsamples. The estimates for the subsamples are reported in the last four (three) columns of Table 4. In all periods, we cannot reject the null of deterministic cointegration at the 1 % level of significance. Focusing on the full sample, we obtain significant estimates of μ1, i.e., the estimated values for μ1 ̂ = 0.246, 0.557, 0.542, −0.083, 0.335, and 0.377; these parameter estimates are the values of the CO2-GDP elasticity for the total, commercial, electrical, industrial, residential, and transport, respectively. It is remarkable that all series under analysis show a positive CO2GDP elasticity except the industrial sector. However, in the last regime (from the beginning of the last economic crisis until now), the CO2-GDP elasticity switches to negative in all sectors; the exception, again, is in the industrial sector, which shows that this sector has positive (but nonsignificant) elasticity. Thus, ignoring shifts creates an erroneous finding when testing the existence of the EKC. In summary, our results report elasticity that is consistent with stage 2 (environmental stabilization) for all sub-periods (except the last) and across sectors (excluding industrial sector). We also find that the elasticity from the electricity sector in the first period is consistent with Stage 1. During these periods, the USAwas in the second stage of the EKC characterized by environmental stabilization; a possible explanation for the first breaks is the second oil crisis occurred at the end of 1979 and the beginning of the 1980s. The second breaks could correspond with the crisis derived of the OPEC Oil price collapse. The last breaks clearly correspond to the beginning of the last economic crisis. Focusing on the results of the last regime, three conclusions clearly emerge. First, comparing the coefficient in the full sample with the last regime, the latter is not only much smaller than the full sample value, but it also has a sign switch from positive to negative. Second, the negative sign implies that from the beginning of the economic crisis, the USA has crossed the border from the BEnvironmental
Environ Sci Pollut Res Table 3
Arai-Kurozumi-Kejriwal cointegration tests with structural breaks Total
Commercial
Electrical
Industrial
Residential
Transport
0.088* 0.17
0.021 0.56
0.020 0.21
0.040 0.19
0.022 0.21
0.041 0.15
T ̂1
1980:1
1996:4
1981:4
1981:2
1981:4
1979:3
λ̂ 2
0.32
0.83
0.65
0.59
0.83
0.29
T ̂2
1986:2
2008:2
2000:4
1998:1
2008:2
1985:4
λ̂ 3
0.82
0.83
0.82
0.82
T ̂3 Critical values 10 % 5% 1%
2007:4
2008:2
2007:4
2007:4
Test Vk ̂ λ̂ λ̂ 1
0.070
0.092
0.059
0.055
0.091
0.072
0.093 0.155
0.119 0.201
0.077 0.123
0.070 0.108
0.117 0.188
0.097 0.154
Notes: a
*, **, and *** denote significance at the 10, 5, and 1 levels, respectively.
b
Critical values are obtained by simulation using 500 steps and 2000 replications.
The Wiener processes are approximated by partial sums of i.i.d. N(0, 1) random variables.
stabilization^ to BEnvironmental optimization^. Consequently, ignoring structural changes can lead to misleading conclusions. Third, and finally, the industrial sector is an outlier because its pattern is the opposite of the others sectors.
Conclusions and policy implications This paper has analyzed the relationship between income and polluter emissions in the USA using quarterly time series data from 1973:1–2015:2. The method offered by Jaunky (2011) has been applied to test the existence of the EKC hypothesis. Although the literature is inconclusive due to deviations of the strict principles of Kuznets, our analysis uses a broad time period and techniques that measure the cointegration of the variables and the study of the breaks. Consequently, this study elucidates the available empirical evidence for the USA because we allow different behaviors is the CO2 emissions and GDP relationship, and more importantly, we observe it when distinguishing by sectors. Our result recognizes that this polluter relationship is not stable over time by finding three (two) breaks and, therefore, four (three) periods during which this ratio has varied. In the first three periods, the USA has been in the environmental stabilization stage and because the beginning of the last economic crisis, the elasticity CO2-GDP has become negative, i.e., the USA has achieved the Environmental Stabilization stage; thus, the EKC hypothesis is confirmed (except in the industrial sector).
Finally, in relation to the industrial sector, policy makers likely need to promote a wide variety of activities to reduce CO2 emissions. The last economic crisis may have led to a relaxation in the targets in relation to the environmental stabilization. In June 2014, President Obama proposed the Clean Power Plan, which will require states to reduce carbon pollution from power plants, cutting emissions to 30 % of 2005 levels by 2030, promoting the use of a relatively environmentally friendly long-term driver of economic growth. However, our result suggests that the US economy seems to be on the right track with the exception of the industrial sector. In other words, in the absence of feasible alternatives to increase energy consumption from environmental friendly sources, cutting carbon dioxide emissions, particularly in this sector under the Kyoto Protocol, will have a negative effect on economic growth in the USA. Of course, one might argue that it is unlikely that the USA will risk their real GDP growth through CO2 emissions by sectors without adopting alternatives to minimize the impact of energy conservation on economic growth. In this sense, environmental policies aimed at reducing industrial energy intensity include developing a market for emissions trading, establishing mechanisms for the use of renewable energy through new technologies or using alternative cleaner energy sources. Although these alternatives require time to replace fossil fuels with renewable and environmentally friendly policies and drive further decarbonization of the energy system as the US economy recovers and grows. In sum, all these recipes to reduce CO2 emissions should be severe, particularly in the industrial sector. As reflected in
Environ Sci Pollut Res Table 4
Stock–Watson-Shin’s DOLSa,b,c,d parameters estimation of linear cointegration
Total μ0 μ1 R2 Test: Ccμ Commercial μ0 μ1 R2 Test: Ccμ Electrical μ0 μ1 R2 Test: Ccμ Industrial μ0 μ1 R2 Test: Ccμ Residential μ0 μ1 R2 Test: Ccμ Transport μ0 μ1 R2 Test: Ccμ Critical values: Cμ
Full sample 1973:1–2015:2 4.900*** (0.106) 0.246*** (0.012)
First regime 1973:1–1980:1 3.279*** (0.613) 0.437*** (0.071)
Second regime 1980:2–1986:2 6.700*** (0.063) 0.040 (0.086)
Third regime 1986:3–2007:4 3.382*** (0.097) 0.411*** (0.010)
Last regime 2008:1–2015:2 10.987*** (1.823) −0.391** (0.190)
0.785 0.135
0.842 0.079
0.486 0.174
0.958 0.197
0.565 0.111
Full sample 1973:1–2015:2 0.212* (0.110) 0.557*** (0.012) 0.948 0.151
First regime 1973:1–1996:3 0.239** (0.119) 0.554*** (0.014) 0.962 0.101
Second regime 1996:4–2008:2 1.018** (0.469) 0.479*** (0.049) 0.864 0.152
Last regime 2008:2–2015:2 12.008*** (2.947) −0.676** (0.307) 0.439 0.115
Full sample 1973:1–2015:2 1.156*** (0.152) 0.542*** (0.017)
First regime 1973:1–1981:4 −3.432*** (0.545) 1.068*** (0.063)
Second regime 1982:1–2000:4 −0.078*** (0.162) 0.682*** (0.017)
Third regime 2001:1–2008:2 3.137*** (0.063) 0.339*** (0.063)
Last regime 2008:3–2015:2 19.129*** (2.898) −1.336*** (0.302)
0.917 0.207 Full sample 1973:1–2015:2 6.774*** (0.143) −0.083*** (0.015) 0.334 0.135 Full sample 1973:1–2015:2 2.473*** (0.134) 0.335*** (0.015) 0.812
0.920 0.068 First regime 1973:1–1981:2 5.239*** (0.620) 0.101 (0.072) 0.652 0.114 First regime 1973:1–1981:4 3.798*** (0.650) 0.184** (0.074) 0.465
0.965 0.061 Second regime 1981:3–1998:1 3.104*** (0.231) 0.325*** (0.025) 0.810 0.168 Second regime 1982:1–2008:2 1.060*** (0.171) 0.488*** (0.018) 0.919
0.726 0.082 Third regime 1998:2–2007:4 8.745*** (0.032) −0.286*** (0.032) 0.867 0.069 Last regime 2008:3–2015:2 17.401*** (4.091) −1.224** (0.426) 0.465
0.647 0.116 Last regime 2008:1–2015:2 4.489*** (1.146) 0.150 (0.119) 0.837 0.103
0.126 Full sample 1973:1–2015:2 2.540**** (0.089) 0.377*** (0.010) 0.927 0.125 10 % 0.231
0.105 First regime 1973:1–1979:3 −0.571 (0.825) 0.738*** (0.096) 0.952 0.103 5% 0.314
0.055 Second regime 1979:4–1985:3 4.155*** (0.759) 0.193** (0.086) 0.496 0.181 1% 0.533
0.109 Third regime 1985:4–2007:4 1.581*** (0.070) 0.483*** (0.008) 0.986 0.073
Notes: We choose q = INT(T1/3 ) as proposed by Stock and Watson (1993) Cμ is an LM statistic for cointegration using the DOLS residuals from deterministic cointegration, as proposed Shin (1994) *, **, and *** denote significance at the 10, 5, and 1 % levels, respectively
Last regime 2008:1–2015:2 8.721*** (1.175) −0.269** (0.122) 0.608 0.117
Environ Sci Pollut Res
Appendix 1
9.8 9.6 9.4 9.2 9 8.8 8.6 8.4 8.2 8
Fig. 2 GDP (in logs)
5.8
6.6
5.6
6.4 6.2
5.4
6
5.2
5.8
5
5.6
4.8
5.4
4.6
5.2
Industrial
Residential
6.3
5.9
6.2
5.8 5.7
6.1
5.6
6
5.5
5.9
5.4
5.8
5.3
5.7
5.2
5.6
5.1
Transport
Total
6.3
7.4
6.2
7.3
6.1 6 5.9 5.8 5.7 5.6
7.2 7.1 7 6.9
5.5
6.8
5.4
6.7
Fig. 3 CO2 emissions by economic sector (in logs)
Environ Sci Pollut Res
our study, the services sector can be observed as a cleaner sector, and the measures to be applied to the industrial sector can be direct, such as reductions to specific levels by a certain date following the Kyoto Protocol and indirect, such as establishing renewable portfolio standards with public investment and efficiency targets or overseeing the profits generated in activities requiring high levels of CO2 emissions.
References AEO (2015): Annual energy outlook 2015 with projections to 2040. April 2015 U.S. energy information administration. Office of Integrated and International Energy Analysis. U.S. Department of Energy Washington, DC 20585. Ajmi AN, Hammoudeh S, Nguyen DK, Sato JR (2015) On the relationships between CO2 emissions, energy consumption and income: the importance of time variation. Energy Econ 49:629–638 Aldy JE (2005) An environmental Kuznets curve analysis of U.S. state-level carbon dioxide emissions. J Environ Dev 14(1):48–72 Andreoni J, Levinson A (2001) The simple analytics of the environmental Kuznets curve. J Public Econ 80:269–286 Arai Y, Kurozoumi E (2007) Testing for the null hypothesis of cointegration with a structural break. Econ Rev 26:705–739 Arrow K, Bolin B, Costanza R, Dasgupta P, Folke C, Holling CS, Jansson BO, Levin S, Maler KG, Perrings C, Pimentel D (1995) Economic growth, carrying capacity, and the environment. Science 268:520–521 Aslanidis, N. (2009). Environmental Kuznets curves for carbon emissions: a critical survey. Department of Economics, University Rovira Virgili FCEE, Avinguda Universitat, Department of Communication, Working Paper No 51. Baek J, Gweisah G (2013) Does income inequality harm the environment?: empirical evidence from the United States. Energy Policy 62:1434–1437 Bai J, Perron P (2003) Critical values for multiple structural change tests. Econ J 6:72–78 Bataa, E., Izzeldin, M., Osborn, D. R. (2015). Changes in the global oil market. economics working paper series 2015/004. The Department of Economics Lancaster University Management School Lancaster LA1 4YX UK. Beckerman W (1992) Economic growth and the environment: whose growth? Whose environment? World Dev 20(4):481–496, Elsevier Bin S, Dowlatabadi H (2005) Consumer lifestyle approach to US energy use and the related CO2 emissions. Energy Policy 33:197–208 Bo S (2010) A literature survey on environmental Kuznets curve. Energy Procedia 5:1322–1325 Borghesi, S. (1999). The environmental Kuznets curve: a survey of the literature. European University Institute November 1999. Bowden N, Payne JE (2010) Sectoral analysis of the causal relationship between renewable and non-renewable energy consumption and real output in the US. Energy Sources Part B 5:400–408 Burnett JW, Bergstrom JC, Dorfman H (2013) A spatial panel data approach to estimating U.S. state-level energy emissions. Energy Econ 40:396–404 Carpentier CL (2006) NAFTA commission for environmental cooperation: ongoing assessment of trade liberalization in North America. Impact Assess Proj Apprais 24(4):259–272 Carson R (2010) The environmental Kuznets curve: seeking empirical regularity and theoretical structure. Rev Econ Environ Policy 4(1):3–23
Carson RJ, Mccubbin D (1997) The relationship between air pollution emissions and income: US Data. Environ Dev Econ 2:433–450 Cole MA (2008) Industrial activity and the environment in china: an industry-level analysis. China Econ Rev 19(2008):393–408 Cole MA, Rayner AJ, Bates JM (1997) The environmental Kuznets curve: an empirical analysis. Environ Dev Econ 2:401–416 Cologni A, Manera M (2008) Oil prices, inflation and interest rates in a structural cointegrated VAR model for the G-7 countries. Energy Econ 30(3):8569–8888 de Bruyn S, Heintz R (1999) The environmental Kuznets curve hypothesis. Handbook of environmental and resource economics. Edward Elgar, Massachusetts Dickey DA, Fuller WA (1979) Distribution of the estimators for autoregressive time series with a unit root. J Am Stat Assoc 74:427–431 Dinda S (2004) Environmental Kuznets curve hypothesis: a survey. Ecol Econ 4(1):431–455 Doda B (2014) Evidence on business cycles and CO2 emissions. J Macroecon 40:214–227 Egli, H. (2004). The environmental Kuznets curve-evidence from time series data for Germany. Working Paper Series 04–33, Institute of Economic Research. Ekins P (1997) The Kuznets curve for the environment and economic growth: examining the evidence. Environ Plan A 29:805–830 Esteve V, Tamarit C (2012) Is there an environmental Kuznets curve for Spain? Fresh evidence from old data. Econ Model 29(6):2696–2703 Fischer C, Heutel G (2013) Environmental macroeconomics: environmental policy, business cycles, and directed technical change. Ann Rev Resour Econ 5(1):197–210 Galeotti M, Lanza A, Pauli F (2006) Reassessing the environmental Kuznets curve for CO2 emissions: a robustness exercise. Ecol Econ 57:152–163 Grossman GM, Krueger AB (1991) Environmental impacts of a North American free trade agreement. Woodrow Wilson School - Public and International Affairs, Princeton, Papers 158 Grossman GM, Krueger AB (1995) Economic growth and the environment. Q J Econ 110(2):353–377 Harbaugh W, Levinson A, Wilson DM (2002) Reexamining the empirical evidence for an environmental Kuznets curve. Rev Econ Stat 84:541–551 He J, Richard P (2010) Environmental Kuznets curve for CO2 in Canada. Ecol Econ 69:1083–1093 Hettige H (2000) Industrial pollution in economic development: the environmental Kuznets curve revisited. J Dev Econ 62(2) Heutel G (2012) How should environmental policy respond to business cycles? Optimal policy under persistent productivity shocks. Rev Econ Dyn 15(2):244–264 Huntington H (2005) U.S. carbon emissions, technological progress and economic growth since 1870. Int J Global Energy Issues 23(4):292–306 Iglesias J, Carmona M, Golpe AA, Martín JM (2013) La curva de Kuznets y la emisión de CO2 en España. 1850–2008. Rev Econ Ind 389:135–144 Jalil A, Mahmud SF (2010) Kuznets curve for CO2 emissions: a cointegration analysis for China. Energy Policy 37:5167–5172 Jaunky VC (2009) Is productivity growth driving higher CO2 emissions? Evidence from high-income economies. Int J Environ Cult Econ Soc Sustain 5(5):250–269 Jaunky VC (2011) The CO2 emissions-income nexus: evidence from rich countries. Energy Policy 39(3):1228–1240 Jordan, B.R. (2010). The environmental Kuznets curve: preliminary meta-analysis of published studies, 1995–2010. Georgia Tech School of Public Policy Workshop on Original Policy Research (WOPR). December 3, 2010.
Environ Sci Pollut Res Jotzo F, Burke PJ, Wood PJ, Macintosh A, Stern DI (2012) Decomposing the 2010 global carbon dioxide emissions rebound. Nat Clim Chang 2(4):213–214, 2012 Kaika D, Zervas E (2013a) The environmental Kuznets curve (EKC) theory-part A: concept, causes and the CO2 emissions case. Energy Policy 62:1392–1402 Kaika D, Zervas E (2013b) The environmental Kuznets curve (EKC) theory. Part B: critical issues. Energy Policy 62:1403–1411 Kejriwal M (2008) Cointegration with structural breaks: an application to the Feldstein-Horioka Puzzle. Stud Nonlin Dyn Econ 12(1):1–37 Kejriwal M, Perron P (2008) The limit distribution of the estimates in cointegrated regression models with multiple structural changes. J Econ 146:59–73 Kejriwal M, Perron P (2010) Testing for multiple structural changes in cointegrated regression models. J Bus Econ Stat 28:503–522 Kijima M, Katsumasa N, Atsuyuki O (2010) Economic models for the environmental Kuznets curve: a survey. J Econ Dyn Control 34:1187–1201 Kuznets S (1955) Economic growth and income inequality. Am Econ Rev 45:1–28 Kwiatkowski D, Phillips PCB, Schmidt P, Shin Y (1992) Testing the null hypothesis of stationarity against the alternative of a unit root: how sure are we that economic time series have a unit root? J Econ 54:159–178 Lanne M, Liski M (2004) Trends and breaks in per-capita carbon dioxide emissions, 1870–2028. Energy J 25(4):41–65 List J, Gallet C (1999) The environmental Kuznets curve: does one size fit all? Ecol Econ 31(3):473–480 López R (1994) The environment as a factor of production: the effects of economic growth and trade liberalization. J Environ Econ Manag 27:163–184 López R, Mitra S (2000) Corruption, pollution, and the Kuznets environment curve. J Environ Econ Manag 40:137–150 López RE, Yoon SW (2014) Pollution–income dynamics. Econ Lett 124(3):504–507, Elsevier Martínez-Alier J (1995) The environment as a luxury good or too poor to be green? Ecol Econ 13(1):1–10 McConnell KE (1997) Income and the demand for environmental quality. Environ Dev Econ 2(4):383–400 Moomaw WR, Unruh GC (1997) Are environmental Kuznets curve misleading us? The case of CO2 emissions. Environ Dev Econ 2:451–463 Mork KA (1989) Oil and the macroeconomy when prices go up and down: an extension of Hamilton’s results. J Polit Econ 97(3):740–744, Jun., 1989 Narayan PK, Narayan S (2010) Carbon dioxide emissions and economic growth: panel data evidence from developing countries. Energy Policy 38(1):661–666, Elsevier Narayan PK, Smyth R (2008) Energy consumption and real GDP in G7 countries: new evidence from panel cointegration with structural breaks. Energy Econ 30(5):2331–2341, react-text: 47 Ng S, Perron P (1995) Unit root tests in ARMA models with data dependent methods for the selection of the truncation lag. J Am Stat Assoc 90:268–281 Ng S, Perron P (2001) Lag length section and the construction of unit root tests with good size and power. Econometrica 69:1529–1554 Niu, H., Li, H. (2014). An empirical study on economic growth and carbon emissions of G20 group. International Conference on Education Reform and Modern Management. Panayotou T (1993) Economic growth and the environment. Spring Seminar of The United Nations Economic Commission For Europe, Geneva
Pasten R, Figueroa E (2012) The environmental Kuznets curve: a survey of the theoretical literature. Int Rev Environ Resour Econ 6(3):195–224 Payne JE (2010) Survey of the international evidence on the causal relationship between energy consumption and growth. J Econ Stud 37(1):53–95 Perron P, Ng S (1996) Useful modifications to some unit root tests with dependent errors and their local asymptotic properties. Rev Econ Stud 63:435–465 Pfaff A, Chaudhuri S, Nye H (2004) Household production and the environmental Kuznets curve. Environ Resour Econ 27:187–200 Plassmann F, Khanna N (2006) household income and pollution. Implications for the debate about the environmental kuznets curve hypothesis. J Environ Dev 15(1):22–44 Roca J (2003) Do individual preferences explain the environmental Kuznets curve? Ecol Econ 45:3–10 Roca J, Padilla E (2003) Emisiones atmosféricas y crecimiento económico en España. La curva de Kuznets ambiental y el protocolo de Kyoto. Economía Industrial 351:73–86. Schwert GW (1989) Tests for unit roots: a Monte Carlo investigation. J Bus Econ Stat 7:147–159 Selden TM, Song D (1994) Environmental quality and development: is there a Kuznets curve for air pollution emissions? J Environ Econ Manag 27:147–162 Shahiduzzaman M, Layton A (2015) Changes in CO2 emissions over business cycle recessions and expansions in the United States: A decomposition analysis. Appl Energy 150:25–35 Sheldon, T.L. (2014). Asymmetric effects of the business cycle on carbon dioxide emissions: a new layer of climate change uncertainty San Diego Manuscript. Department of Economics, University of California; 2014. Shin Y (1994) A residual-based test of the null of cointegration against the alternative of no cointegration. Econometric Theory 10:91–115 Soytas U, Sari R, Ewing BT (2007) Energy consumption, income, and carbon emissions in the United States. Ecol Econ 62:482–489 Stern, D. I. (2004). The environmental kuznets curve. internet encyclopaedia of ecological economics. International Society for Ecological Economics. S t o c k J H , Wa t s o n M W ( 1 9 9 3 ) A s i m p l e e s t i m a t o r o f cointegrating vectors in higher order integrated systems. Econometrica 61:783–820 Suri V, Chapman D (1998) Economic growth, trade and energy: implications for the environmental Kuznets curve. Ecol Econ 25(2):195–208 Thoma M (2004) Electrical energy usage over the business cycle. Energy Econ 26(3):463–485 Toda HY, Yamamoto T (1995) Statistical inference in vector autoregressions with possibly integrated processes. J Econ 66:225–250 Torras M, Boyce JK (1998) Income, inequality and pollution: a reassessment of the environmental Kuznets curve. Ecol Econ 25:147–160 U.S. Enviromental Protection Agency. Web site: https://www3.epa.gov Wagner M (2008) The carbon Kuznets curve: a cloudy picture emitted by bad econometrics? Resour Energy Econ 30:388–408 World Resources Institute (1994) World resources 1994–1995: a guide to the global environment. Oxford University Press, New York World Resources Institute (1996) World resources 1996–1997: a guide to the global environment. Oxford University Press, New York York R (2012) Asymmetric effects of economic growth and decline on CO2 emissions. Nat Climate Chang 2(11):762–764, 2012 Zhao F, Xu M, Zheng Y, Gi Wong M, Chi Y (2013) Improving the environmental Kuznets curve for evaluating the relationships between carbon dioxide emissions and economic development. J Food Agric Environ 11(2):1193–1199