This action might not be possible to undo. Are you sure you want to continue?
Another look at CO2 emissions modelling: the role of energy prices in developed countries
M. Rodríguez and Y. Pena-Boquete*, Applied Economics, Universidade de Vigo, Spain
inertia will prevent us from realise in population growth to slow down CO2 emission rates. That conclusion connects with the Malthusian tradition. The negative environmental impact caused by demographic pressure, as outlined previously by other authors such as Daily and Erlich (1992), Zaba and Clarke (1994) or Birdsall (1992), may arise through a double mechanism: the larger the population (i) the greater the energy demand from industry and transport services, and (ii) the greater the deforestation and land use changes. Following the framework from Dietz and Rosa (1997), Martínez-Zarzoso et al. (2007) considered additionally the energy intensity as a proxy variable to measuring the level of environmentally damaging technology. By applying panel data econometrics they reached different patterns for old and new EU members. For instance, the elasticity emissionpopulation is lower than unity for the former, whereas for the later is 2.73, which is in accordance with the higher marginal propensity to emit in less developed regions as reported in the literature. It remains unclear whether a demographic decline will curb CO2 emissions since there is an increase in the number of households simultaneously to a households size decrease in most European accession countries (MacKellar et al., 1995). As a result, Martínez-Zarzoso et al. (2007) argue that “a review of the Communitarian emissions policy, that takes into account the characteristics of the new EU members, would be desirable. [...] Several factors must be taken into account when establishing the allocation of emission quotas to each country, including population dynamics, income and productive structures and energy intensities”. Furthermore, they found a negative elasticity between emissions and urbanization for more developed countries which has important policy implications: policymakers and experts should recognize the potential value of cities for long-term sustainability. There are plenty of examples that analyse this sort of relationship, most of them for a cross section data on countries and more recently for panel data, but usually the population is included in the dependent variable (per capita emissions) instead of being treated as a predictor in the model. The underlying idea behind most of papers that analyse the IPAT framework is the Environmental Kuznets Curve (EKC)1. The theoretical explanations of the EKC hypothesis are based on three effects: the scale effect, the structure effect and the
Abstract--Climate change due to the emission of greenhouse gases (GHGs) is a critical global environmental problem. In particular, CO2 emissions represent about 80% of GHGs. For this reason, we need a good modelling approach capable to forecast in the short and medium term the evolution of CO2 emissions in order to: (i) design the best policies to reduce GHG emissions and (ii) allow businesses regulated by the European Union Emission Trading Scheme to anticipate the evolution of the European carbon emissions and its policy implications. This paper examines the relationship between CO2 emissions, GDP and the energy consumption in developed countries. A panel data analysis for the period 1980 to 2004 is applied. The findings reported in this piece of research confirm the suspicion in the literature that the relationship among CO2 emissions and GDP may be the result of spurious statistical correlations. Therefore that could be evidence against the EKC hypothesis. Thus the IPAT model and EKC based on GDP as an explicative variable can lead to misleading results.
Index Terms--CO2 Emissions, Econometric Analysis, Energy Consumption, Energy Prices, Environmental Kuznet Curve, GDP.
RLICH and Holdren (1971) suggested the IPAT framework as a way to analyse the determinants of environmental impact from economic development. Accordingly there is an important strand of the literature that tries to relate CO2 emissions with Gross Domestic Product (GDP) or with population growth, or with both variables simultaneously. For instance, Dietz and Rosa (1997) found that there is a linear relationship between the size of population and CO2 emissions for a cross section data on 111 countries. Additionally it includes the weight of the industry in economic activity as a proxy for the level of environmentally damaging technology apart from per capita GDP. This should not be surprising as long as it is capturing a scale effect. As a result, they concluded that demographic
* M. Rodriguez gratefully acknowledge the financial support from Spanish Ministry for Science and Education and ERDF (Projects SEJ200612939/ECON and ECO2009-14586-C02-01ECON), and the Galician government (Project INCITE08PXIB300207PR). M Rodriguez (e-mail: firstname.lastname@example.org); Y. Pena (e-mail: email@example.com); Facultade Empresariais e Turismo, 32004 Ourense, Spain.
See also Dinda and Coondoo (2006), Dinda (2004) and Verbeke and De Clercq (2006) for discussions about the EKC topic.
Thus income may well be endogenous. Shafik and Bandyopadhyay (1992).. Holtz-Eakin and Selden (1995). (2003). globalisation. among many others.). 1997) even though there are some exceptions addressing industrialized countries like De Bruyn et al. Studies for single countries most often address developing countries (e. Barrett and Graddy (2000) argues that the functional form of EKC may be conditioned by policy (environmental regulations. 2002. Liu (2005). Bradford et al. other studies have explored other underlying factors. The EKC literature leads to inconclusive results about this hypothesis. electoral processes. Martinez- Zarzoso and Maruotti (2011). Consequently. Despite this feature. the share of fossil fuels and the carbon intensity of fossil fuel combustion. measured as either policy variables or government regime type. Finally. Huang et al. some papers like Dijkraaf and Vollebergh (1998) indicate that the relationship between income and carbon emissions varies among nations. and demand effects). Most of the criticisms are related to the use of non-appropriated techniques and the presence of omitted variables bias. By taking a step further in this reasoning. 2011). the effect of government regime-type on environmental quality has been under studied in the EKC debate. structure of the economy. Halkos and Tsionas (2001). Islam et al. Holtz-Eakin and Selden (1995).g. economic growth in itself does not offer a solution to environmental problems as long as reductions in CO2 emissions will not occur during the normal course of development. Thus even though rich countries may have be experiencing a change in their production structure by outsourcing the production of pollution intensive products. But this idea may be controversial as Shen and Hashimoto (2004) find some evidence of an EKC in China over the last decade of the 20th century. Borghesi and Vercelli (2003) considered that the studies based on local emissions present acceptable results. additional research is called for. Moomaw and Unruh (1997) and Friedl and Getzner (2003). Vollebergh et al. and environmental quality. energy intensity.. 2005) and for a large number of countries (Lindmark. (1997). The vast majority of investigations regarding the existence of an EKC EKC for per capita CO2 emissions concentrate on cross-section and panel data. Tucker (1995). Aldy (2007). Furthermore. the component effects of income (scale. Schmalensee et al. See for instance. etc. (2006). An in deep scrutiny of model specification in the literature shows us that there are a considerable number of studies that consider only GDP and population growth. (2005). the results indicate that the EKC does not exist. Müller-Fürstenberger and Wagner (2007). Kriström and Lundgren. However. Galeotti et al. primary energy use in final energy consumption. Sweden (Lindmark. regulations and taxes influencing fossil fuel consumption. They find strong evidence for an EKC. their consumption structure remains unchanged (Vishal. Other variables. is the link between government action. (2005). 1995. Equally important. 2004). Marrero (2010). Brock and Taylor (2004. This situation is illustrated by the so-called Pollution Haven Hypothesis (PHH). Although that. Martinez-Zarzoso and Bengochea-Morancho (2004). level of democracy. Besides note that an EKC for CO2 emissions per capita does not imply an EKC for CO2 emissions within the wide range of data observations as showed in York et al. patterns of urbanization and sub-urbanization. Holtz-Eakin and Selden conclude that emissions will keep going with growth because output and population will expand faster in lower-income nations (with their high marginal propensity to emit) than higher ones (which eventually pass the turning point). Besides international trade may influence EKC as a consequence of stricter environmental regulations in rich countries (see for instance He. Most papers concentrate on crosssection and panel data. such as international trade. composition. etc. 1991. Stern 1998. Galeotri and Lanza (1999). This result is consistent with the 'inverted-U' shape for the relationship between per capita GDP and several air pollutants found in Grossman and Krueger (1991) and Selden and Song (1994). Galeotti et al.g. Patel et al. 2011). Friedl and Getzner (2003). But generally speaking papers that do not found evidence of EKC do found a positive relationship between per capita income and emissions. for instance. it remains unclear the impact of international trade from the empirical EKC literature (Stern. A good example is De Bruyn et al. (1998).. 2004). For instance. And that finding should raise some concerns about the distributional consequences of policies to reduce emissions. though even less explored. Aldy (2005). This has been done for USA (Tol et al. 2004).2 abatement effect (Grossman and Krueger. Moomaw and Unruh (1997). Wagner (2008). Cole et al. Put in other words. are not considered or . Roberts and Grimes (1997). Perman and Stern (2003) state that when diagnostic statistics and specification tests are taken into account and the proper techniques are used. Despite the important connection between government action and the provision of public goods such as environmental quality. Bertinelli and Strobl (2004). Shafik (1994). (2008). 2007). Thus as long as the evidence is rather diverse for panel data and the studies on single countries are rather rare. As a result of that (no homogeneous results in the EKC literature) the validity of this hypothesis has been questioned in some surveys (e... 2005). Vincent. Kahuthu (2006). For instance. Holtz-Eakin and Selden (1995) estimates suggest a diminishing marginal propensity to emit carbon dioxide as GDP per capita rises. the level of environmentally damaging technology. In fact there are many examples against this hypothesis. to explain CO2 emission growth. In fact. Dijkraaf and Vollebergh (1998). whereas those concerning global emissions do not offer the expected outcomes. some researches look at a single country rather than a group of countries but by taking a much longer period (around 100 years). Finally Porter and van der Linde (1995) assert that environmental protection can promote economic growth by increasing efficiency and stimulating technical progress. 1999). (1998). and therefore the EKC hypothesis cannot be generally accepted. (1998) who concluded that their results are consistent with the notion that an EKC estimated from pooled data need not hold for specific individual countries. (2009). Vincent (1997).
But nearly all of the studies described above omit energy prices. Metcalf (2008)2. Wang et al. Another approach used for look at the relationship between income and pollutants is the efficient frontier methods (frontier models) which have not been surveyed for this paper. through Granger causality test) between income and emissions in order to infer the direction of causality among variables. Shipper et al. Second. there are also exogenous. Coondoo and Dinda (2002). Hamilton and Turton (2002). 3 Within this class of models.T We will assume that the fixed effect uit follows a one-way error component model3.N.. If these omitted variables are also correlated with per capita GDP. energy consumption and output. Richmond and Kauffmann (2006) arrived to similar results concluding that “evidence for a turning point in the relationship between income and carbon/energy use disappears when real light fuel oil prices are included in the model” and thus previous findings are based on spurious statistical relationship.. In addition. in fact. Greening et al. and we incorporate also its lag in order to control for dynamics. subsidies. i = 1. The literature has follow two main approaches: (i) modified input-output modelling for environmental analysis and (ii) Index decomposition methodologies such as the Divisia and Laspeyres index. (iii) the energy intensity effect was fully exploited normally by the industrial sector. However Vishal (2011) does not provide such evidence of an EKC. and for the latter see for instance Cozzi and Di Giulio (1999). First. Illustrations for the former could be found in Chung (1998). the energy mix is driven for the energy prices and the production capacity. if the individual effect represents omitted variables. These techniques require detailed sectoral data and do not allow for stochasticity. (2011). However these conclusions may be subject to the short of countries being analysed (developed versus developing countries). For a good review of the literature see Löfgren & Muller (2010). (2009).. Ang (2007). a typical macro panel is not likely to be a random sample from a larger universe of countries. MODELLING CO2 DISCARGE AND THE DATA BASE As Marrero (2010) shows not just economic development is important for explaining CO2 discharge on the environment but also the energy mix. even more important than energy intensity. Nevertheless.. They suggest also that evidence for an EKC for carbon emission is weakened when energy prices and trade variables are included. since last decade there is a growing literature dealing with the dynamic causal relationships between pollutant emissions. II.σ υ ) . independent of each other and among themselves. As a general conclusion it could be said that there is evidence of an inverted U-shape pattern associated with the EKC in HsiaoTien and Chung-Ming (2010). They used Hanushek’s (1974) method to regress the estimated year effects from the emissions–GDP model on global energy prices. In this model. We also account for the aggregate energy use effect 2 Ang and Zhang (2000) provide a good methodological overview of the several studies that have used decomposition methodologies to track the sources of emissions growth. The first study to specify energy prices in regression models that test for an EKC for carbon emission is Agras and Chapman (1999). This literature examines the time series dynamics (i. Furthermore.. Their results indicate that the energy price is an important explanatory variable.. Liaskas et al. The trade variables are now insignificant while otherwise they have been shown to have considerable explanatory value. this energy mix is not a primary unit for policy-makers decisions. due to the wide range of energy taxes.e. Just to close this survey of the literature. (2007). cross-section estimates of the equations will be biased and inconsistent. As a result there is not clear and unambiguous evidence that income gains will reduce energy use at any level of income. and finally (iv) fuel substitution does usually not contribute much to lowering emissions. Akbostanci et al. resource endowments. They found that the magnitude of this effect is small but they draw our attention to the fact that their oil price measure is only imperfectly correlated with domestic energy prices. They conclude that the problem with such a diverse results may be an over-identified model.σ μ ) and υi ∼ IID (0. Apergis and Payne (2009). it is likely that these country-specific characteristics are correlated with the other regressors. Of course. (2000) and Schipper et al.3 dismissed by most researches. Dinda and Coondoo (2006).. there is an alternative methodology for analyzing the driving forces behind CO2 emissions growth by performing some kind of factorial decomposition breakdown. (ii) the energy and transport sectors are the main causes for the increase in CO2 emissions. Niu et al. and Lee and Lee (2009). (1998). Heil and Selden (2001) experiment with a method for incorporating oil prices into the model. Bataille et al. Finally.. land area. it causes other factors to become insignificant that were previously important. and other distortions that many countries impose. (2005). geography. The long-run price model shows that income is no longer the most relevant variable for environmental quality or energy demand but it is still a significant variable. country-specific factors that will impact on emissions like climate. Based on the Environmental Kuznetz Curve (EKC) we include the Gross Domestic Product per capita (yit) and its square in order to control for the possible inverted U-shape of the curve. the fixed effects specification is a common choice for macroeconomic analysis and it is believed to be more appropriate than a random effects model for two reasons. t = 1. etc. (2000). dit is CO2 discharge on the environment per capita. Some general conclusions emerge from this literature: (i) the level of economic activity and structural change play an important role on the behavior of CO2. but can play a significant role in single cases. 2 oil coal gas dit = δdit −1 + β1 y it + β2 y it + β3eit + β4 pit + β5 pit + β6 pit + β7o + uit . 2 uit = μi + υ it 2 where μi ∼ IID (0. . (1997).
70 Std.t −1 − di.4 (Marrero. The assumption means that there is no correlation between the GHG emissions and the country-specific effect in the absence of conditioning on other variables.07 2. reviewing developments to improve on the relatively poor performance of the standard one-step difference GMM estimator for highly autoregressive panel series.44 7. the within transformation wipes out the μi . since any arbitrary pattern in the time means is consistent with a constant mean of the transformed series for each country9.t-1 is correlated with the error term. 2007 edition” and the “Energy Prices & Taxes 2nd Quarter 2007” both published by International Energy Agency (IEA) and Organisation for Economic Co-operation and Development (OECD).57 0. Anderson and Hsiao (1981) suggested first differencing the model to get rid of the and then using Δ di.23 4. Nevertheless.89 -4. it immediately follows that di. biases could be as much as 20% of the true value of the coefficients of interest 8 8 Fisher Test for panel unit root using an augmented Dickey-Fuller test (2 lags) we are not able to reject that GHG per capita has a unit root at a level of 1%. 20 and 30.04 0. Belgium.49 -11. United Kingdom. In fact.t −1 − di. as long as the υ it themselves are not correlated. 0. Finland.48 0. 2010) including the total primary energy consumption per capita eit. Under these conditions. the GMM framework flexibly accommodates multiple endogenous variables. Since dit is a function of μi . These instruments will not be correlated with Δυ it = υ i.t-2 as an instrument for Δ di. Table 1.87 An alternative transformation that wipes out the individual effects is the first difference transformation.t −1 t =2 i.04 -5. system GMM is not in principle applicable. Hungary.t−1 . coal and gas instead of the shares energy consumption.t-1 is also a function of μi . Judson and Owen (1999) perform some Monte Carlo experiments for N = 20 or 100 and T = 5.74 0.28 -0.11 6.t − 2 = (di. This instrumental variables estimation method leads to consistent but not necessarily efficient estimates because it does not make use of all the available moment conditions (Ahn and Schmidt (1995)). 6 See Sevestre and Trognon (1985) for the magnitude of this asymptotic bias in dynamic error component 7 This is because is correlated with by construction (the latter average contains which is obviously correlated with ). He found that the system GMM estimator has a lower bias and higher efficiency than all the other estimators analysed. 5 Countries included are Austria. Our sample is a balanced panel of 15 European countries5 and 390 observations for the period 1980 to 2004. These authors also argue that including time dummies or transforming variables into deviation from time means should be enough to satisfy the stationary conditions. Our contribution to the model would be to incorporate the prices of primary energies (pit): oil products. only if T → ∞ the within estimator of δ and β will be consistent for the dynamic error component model. Japan.65 0.13 2. provided Monte Carlo simulation comparison between one-step difference and the estimator proposed in Blundell and Bond (1998). Arellano and Bond (1991) argue that additional instruments can be obtained in a dynamic panel data model if one utilizes the orthogonality conditions that exist between lagged values of dit and the disturbances υ it .75 0. (2001). Since it is not possible to calculate a price for renewable energy we include production of renewable energy showing the country capacity for using this kind of energy4.t −1 = (di. the within estimator will be biased by O(1/T) and its consistency will depend upon T being large (see Nickell 1981).18 Max 3. and showed that system GMM has substantial asymptotic efficiency gains.08 4.20 4. For the fixed effects estimator. The source for this data sample is the “Energy Statistics of OECD countries. Moreover.18 -7. 10. Blundell et al. including the standard one-step difference GMM estimators. We have applied several tests and we found that a GHG emission has a unit root8. France. Mean ln(GHGpc) ln(GDPpc) ln(GDPpc)2 ln(Epc) Lag_ln(Epc) ln(E prices) ln(Oil prices) ln(Coal prices) ln(Gas prices) ln(Renewables) 2. United States.70 5.41 0. as typical in country studies.t −2 ).59 12.84 8. (2001). ESTIMATION METHODOLOGY The inclusion of a lagged dependent variable in the model (1) renders the OLS estimator biased and inconsistent6.81 0. denoted system GMM.t −1 Details about the variables definitions and the database are showed in the Appendix. Later on. Check Republic. but ( di. Slovak Republic. Italy.t −1 ) where d = ∑ d (T −1) will still be correlated T with ( υ it − υ i ) even if υ it are not serially correlated7. and it does not take into account the differenced structure on the residual disturbances ( Δυ it ).16 5. 4 i.95 -7. Nevertheless.t − 3 ) or simply di.t − 2 − di.49 2. Descriptive statistic of the data sample.08 3. Denmark. even if the υ it are not serially correlated.01 5.44 1. 9 Note that the assumption E (μ Δd ) = 0 does not imply that the countryi i2 III. as it not only greatly improves the precision but also greatly reduces the finite sample bias. difference GMM is an even poorer choice. Biases increase with δ and decreases with T. T = 30.t − υ i. as explained in Bond et al. Soto (2007) analysed through Monte Carlo simulations the properties of various GMM and other estimators when the number of individuals is small. Poland.48 1.48 2. But even for specific effects play no role in GHG emissions.32 -5. Dev. Therefore.93 5. Table 1 summaries the main descriptive statistic.18 3. Arellano and Bond (1991) propose a generalized method of moments (GMM) procedure that is more efficient than the Anderson and Hsiao (1982) estimator. that relies on relatively mild restrictions on the initial condition process.47 0.02 4. Turkey. di. . Switzerland.06 Min 0.
We have examined the behaviour of the coefficients and the overidentification test when we reduce the number of included instruments. Therefore the CO2 dynamics are better explain by changes in per capita energy consumption instead of its level counterparts (current and lagged) which now turn to be not significant. the particular structure of the firstdifference model implies that an asymptotically equivalent GMM estimator can be obtained in one step. Only for the last one we could accept the EKC hypothesis. it may be more important from policy perspective to account for the prices that drive this energy mix. RESULTS As we have explained in the previous section. and the p-values of the Arellano-Bond AR(1). Besides. Additionally. we include the average price of the oil products (equation 3 and 4) as a order to obtain any valid lagged internal instruments in first-differenced or equations in levels. We will apply both techniques in order to assess the robustness of our results. too many instruments generated system GMM can lead to a problem of overfitting. Although this is a two-step estimator. We need to allow for this serial correlation in The first specification (columns 1 in Table 2) shows the basic approach to validate the Emissions Kuznets Curve (EKC) hypothesis in a dynamic framework according to the survey in section II. the increase of per capita energy consumption has a positive and significant effect on CO2. 2009b). In fact this shares could account for changes in sector composition due to other supply-demand reasons. The estimations for our model do not show autocorrelation of second order and Sargan test performed well. We consider that the shares of the sectors consumption are not a good proxy for changes in technology efficiency between sectors. reducing the power of the Sargan test (Roodman 2009a. the suitable estimator for our model is Blundell and Bond's (1998) systemGMM estimator. AR(1) residuals are not detrimental to estimation. Moreover. Nevertheless the influence of this variable on CO2 values is very tinny attending to the range of values for both in our sample data base. Simulation studies have suggested very modest efficiency gains from using the two-step version. To the contrary of Marrero (2010) we do not include the shares of final energy consumption to account for final energy composition effect. The validity of the assumptions used to obtain the moment conditions of System GMM can be assessed using a Sargan overidentification test under the null that these moment conditions are valid. First. Nevertheless. i. IV. we report the number of observations and instruments being used. Nevertheless. On the other hand. Table 2 shows the results for the CO2 discharge on the environment. Thus we suspect that those two cases where we did found a significant coefficient for GDP should be the result of spurious statistical relationship as reported in the survey. we found that per capita energy consumption is always an important element explaining CO2 discharged on the environment. the dependence of the two-step weight matrix on estimated parameters makes the usual asymptotic distribution approximations less reliable for the two-step estimator. Consequently. for example efficiency over time.5 As a consequence of all this. Instead we include the country-trend dummies to account for changes such as technological ones during the period 1979-2004 for each country. As a result. too. Neither the GDP per capita nor its square appears to be significant. For per capita GDP neither the current values nor the lag appear to be significant in most of the specifications. under homoscedasticity of the νit disturbances. There are two main techniques to limit the number of instruments: to use only certain lags instead of all available lags or to combine instruments through addition into smaller sets. . 10 Within the Arellano and Bond (1991) procedure. We also attempt to account for the dynamics in the adjustment of the energy consumption per capita and the GDP per capita (columns 2 and 3 in Table 2). we do not find evidence for the EKC. It is clear for any reader that the energy mix should be an important factor to determine the emission levels. the best available estimator for our equation is the system-GMM estimator by Arellano and Bond (1998) and we use the one-step estimator including Finally.e. We have applied some robustness test for all specifications. As we said before. And country-specific trend accounting for changes in the behaviour over time. GDP. our best available choice is the system-GMM estimator by Blundell and Bond (1998). we test for different dynamics adjustments for per capita CO2 according to different specifications for energy consumption. Arellano-Bond AR(2)10 and Sargan tests. we consider that using the shares of primary energies into the economy is not a proper approach. Instead. Both the current and the lag values appear to be significant but with opposite sign. Instrument can overfit endogenous variables. even in the presence of considerable heteroscedasticity. while AR(2) residuals are. Another further feature of our results is the importance of allowing for an AR(1) component in equation function error term. we also include country dummies to account for the particular factors inherent to each country not considered in the model. energy prices and renewables. For all specifications we get the same results independent of the number of the instruments being used. For this reason we attempt in equation nº9 to identify which effect dominates by including the increase (difference) experienced by this variable. failing to expunge their endogenous components and biasing coefficients estimates. We also incorporated specific-year dummies accounting for economic changes affecting all countries. all estimations have been checked using estimates based on OLS and within transformation procedures in order to appraise the robustness of the results.
gas and coal prices show a positive sign which may be subject to alternative interpretations.e.e. Therefore that could be evidence against the EKC hypothesis. This seems a reasonable assumption as oils products represent the main energy source in our economies and many other like natural gas are usually correlated to it.e. CONCLUSIONS The findings reported in this piece of research confirm the suspicion in the literature that the relationship among CO2 emissions and GDP may be the result of spurious statistical correlations. The results show that oil prices appear to be significant in all specifications and present a negative sign. This variable seems to be negative correlated with CO2 emissions and significant in all specification. so we include the prices for coal and gas too. Moreover.6 reference price. Finally the production of renewable energy should be important to reduce the emission intensity of the economies (unfortunately it is not possible to calculate a price for the renewable). V. As our preferred specification is the last column (nº9 in Table 2) again we are confident that we found a spurious statistical relationship for both prices and therefore they do not deserve our attention. (i) there is an unexpected positive price elasticity of emissions or (ii) they show a substitution effect respect to the oil products prices. using just the oil prices it is not possible to shed light into the substitution effect of the different energies. i. an increase in the oil prices lead to a decrease in the emissions. . i. the possibility of using renewable energy is an important tool to decrease the CO2 emissions. i. Nevertheless. Results show that gas and coal prices appear to be significant if we do not account for the adjustments in the total energy consumption (lagged and in differences).
016 (4) ‐0. Ln(GDPpc) Lag_ln(GDPpc) Ln(GDPpc)2 Ln(Epc) Lag_ln(Epc) D_ln(Epc) Ln(Oil prices) Ln(Coal prices) Ln(Gas prices) Ln(Renewables) Lag_ln(GHGpc) Constant Observations Instruments p value AR(1) p‐value AR(2) p‐value Sargan * p<.336 0. ** p<.010* ‐0.483* 0.295 0.276 ‐0.1.006* 0.917 0.336 (6) ‐0.02 1.891 (3) 0.358*** ‐1.394*** ‐1.027*** ‐0.046 390 46 0 0.070*** ‐0.037 390 47 0 0.016 0.047** 0.933*** ‐0.266 387 62 0.974*** ‐0.155* ‐0.57 0.071*** 0.778 0.006 0.767 387 55 0 0.008 0.761*** 1.563 0.458*** 4.009 0.49 (7) ‐0.101 ‐0.824*** ‐0.012 ‐0.268 0.007 0.200*** 0.802 0.597*** 390 53 0.021 0.018 (2) ‐0.004 0.7 Table 1: System-GMM estimates for the CO2 discharge on the environment.458*** 4.001 0.823*** 1.258 0.368*** 387 55 0.012 ‐0.472 0.967*** ‐0.01 0.018*** 0.008* 0.01 0.016*** ‐0.142 (1) 0.112*** ‐0.629 387 59 0.074 0.020*** 0.961 0.028*** 0.820*** 1 390 54 0 0.009 0.037* ‐0.001 0.499 (5) ‐0.042*** 0.45 0.452*** 387 51 0.003 0.083 0.27 0.788*** 1.05. *** p<.265 1.288*** 5.682 0.121*** ‐0.206 0.650*** ‐0.038*** 0.013** 0.019*** 0.003 0.011 1.001 0.584 0.34 0.659*** 0.555* 0.036*** year‐specific dummies and country trends included .863** ‐0.191 (8) ‐0.090*** ‐0.276 (9) ‐0.001 0.
21-40..I. Quarterly Journal of Economics 110.. L. and Cozzi. C. E.  Cole MA. (Eds.M. and nutrition policy research working paper. and Getzner. P.Weighted average price of coal (US$-using power purchase parity/unit): It is calculated as a weighted average of industry and household prices (we use the final consumption as weights). C.cesifo-group.. Payne.E. Land Economis 83 (3). the industry index includes representative steam coal and coking coal. FEEM Working Paper 63..  Friedl. Bond. Anderson. K. T. Lanza A(1999) Richer and Cleaner? Astudy on carbon dioxide emissions in developing countries..  Blundell. Divergence in state-level per capita Carbon Dioxide emissions.. Strobl (2004). Temple. Bengochea-Morancho... no. National Bureau of Economic Research.F. D.. T.H. 575–588.. S. W. Opschoor (1998). Decomposition of aggregate carbon intensity for the manufacturing sector: comparison of declining trends from 10 OECD countries for the period 1971–1991. Economic growth and the environment: a review of theory and empirics. Environmental and Resource Economics 19: 165–172..  Dinda. A. and Ehrlich. VII. 2001. Bengonchea-Morancho. Rivers N. 2007. How Malleable are the Greenhouse Gas Emission Intensities of the G7 Nations? The Energy Journal. Hoeffler.B. Lanza. (1995): “Efficient estimation of model for dynamic panel data”. Journal of Ecometrics... 4772–4778.2001  Dinda. Energy Policy 35. Tu. . J. geo.  Dietz. 1995. (Ed. Acad..  Birdsall. . Energy Policy 30. Science 171. Journal of Econometrics. USA. Levinson. at: /http://www. C. and Neck. Freedom. hydro. NBER Working Papers 10557.. A. . and Turton. 175-179. 431–455. 3048  Borghesi. 353–377. Decomposition methodology in energy demand and environmental analysis. van den Bergh and J. Barrett. 28(1). J.  Getzner. . Apergis. Ecological Economics. R. G... Ang. Coondoo.  Ehrlich. Determinants of emissions growth in OECD countries. In: Aghion. gas and renewable) (ktoe per person).B. Vollebergh HRJ (2001) A note on testing for environmental Kuznets curves with panel data. S. B. 2004. International trade and economic growth in a global environment. A. W. May 31-June 4. 2006. D. 2004. J.. S.. coal.  Dijkgraaf E. W..  Hamilton. Economic growth and the environment..  Bradford. Natl. ECEEE. E. 2005. Neumayer E(2004) Examining the impact of demographic factors on air pollution. Arrellano. Bataille.. Journal of Econometrics. Reassessing the environmental Kuznets curve for CO2 emissions: a robustness exercise. Population. 1749–1821.  Daily. S. 1997. J.  Coondoo.99  Galeotti. S. and Vercelli. Empirica 28 (1). Energy Policy 37. Cheltenham.). Pauli. 68 (1): 5-27.A.. Vol. R. Handbook of Environmental and Resource Economics. Ang. Aldy. A. Akbostanci. Sustainable globalisation. G.R. Determinants of CO2 emissions in a small open economy.. W. Re-examining the empirical basis for the environmental Kuznets curve. Tunc. N. vol. Washington. Bond... P. P. M.E. Population.W.. S. Martínez-Zarzoso I. Effects of population and affluence on CO2 emissions. J. Environment and Development Economics 8. Journal of the American Statistical Association. energy consumption..). 2006..  Harbaugh. D. CO2 emissions.  Chung. 2007. and Wilson.R.S. 2001. Krueger.C.. Durlauf. 2003. Elsevier. S. Popul Dev Rev 26(1):5–21.Gross domestic Product per capita (expressed in thousands 2000 $ US using PPPs per person). Graddy.. W. (1981): “Estimation of dynamic models with error components”...J. 581–601. Why do CO2 Emissions Differ in China. 87: 115-143. The carbon dioxide emission and income: a temporal analysis of cross-country distributional patterns. Income and emission: a panel data-based cointegration analysis.  Halkos. January. Journal of International Development 16. 2004.. Proc. 77: 598-606. N. 63–71. Schlieckert. Davis. S.P. Another look at population and global warming. 34–65.1971. 161175. Dinda. and output in Central America. health. Ecological Economics 45. 2008. 1992. The Environmental Kuznets Curve Semi-Parametrically Revisited. The relationship between income and environment in Turkey: is there an environmental Kuznets curve? Energy Policy 37 861–867. CEPR Discussion Paper No.. Determinants of CO2 emissions growth in Italy: a Laspeyres index analysis. M. 433–456. Ecological Economics 57.. M. 2003. T. APPENDIX Variables definitions and calculations: . .8  Bertinelli. Environmental Kuznets Curve hypothesis: a survey. Sci. 1999. WPS 1020. NBER. Economic Growth and CO2 Emission in the European Union. The Impact of Population Change on CO2 Emissions: Evidence from European Countries. 167–181. Ecological Economics 65. S. The household index includes steam coal. 8001. On the sustainability of Austrian budgetary policies. Taylor. Mandelieu. Nota di Lavoro 98. Taylor. Higón-Tamarit F.CO2 discharge on the environment: CO2 sectoral approach (mt of CO2). Oxford Institute for Energy Studies EV25. 761-771. G. Shore. and Rosa. growth. J. J. D. pp. L. A... 2000. 2007. 3282–3286.. (1998): “Initial conditions and moment restrictions in dynamic panel data models”. H. The household index includes representative gasoline and light fuel oil.. MA. Review of Economic Studies.M.. sustainability and earth’s carrying capacity. M. 2009. and Inmaculada MartínezZarzoso.. 1. (1982): “Formulation and estimation of dynamic models using panel data”. 2002. Vol. and Rafael MoralesLage. L. energy usage. .. M. REFERENCES             Ahn S. 2006.. NBER Working Paper. W. World Bank. 2000. 58: 277-297..  Galeotti M.  Grossman. P.. Environment and Development Economics 5 (4). Japan and Korea?.A. DC.  De Matteis. Ecological Economics 49. Handbook of Economic Growth. 2003. M. A. and E.. 152–163. Ecological Economics 44... Ecological Economics 57. C. 2000. solar and waste. E.Weighted average price of oil products (US$-using power purchase parity/unit): It is calculated as a weighted average of industry and household prices (we use the final consumption as weights). (1991): “Some test of specification for panel data: Monte Carlo evidence and an application to employment equations”. For coal. VI..pdfS.  Di Giulio. 77-89. 353–369. Fondazione Eni Enrico Mattei. 1998..  De Bruyn. G. light fuel oil and automotive diesel but not fuels used for electricity generation. Cambridge. A.. ‘Economic Growth and Emissions: Reconsidering the Empirical Basis of Environmental Kuznets Curves’. 145-169. WP. J. pp. 1998.. The environmental Kuznets curve: exploring a fresh specification.C. Mau. CORE Discussion Paper 2004/55. H. 133-148. and Schipper..  Bond.  Brock.Total primary energy consumption (it includes oil. in: van den Bergh..  Greening. and Holdren. Environmental Kuznets Curve for sulphur: evidence using GMM estimation and random coefficient panel data models.. Hsiao. Anderson. It includes nuclear. 2009. 1999. and output in France.  . 1992. R. S. Biosciences 42. S. A. P. 18: 47-82. Energy Economics 20.C. 25 (2).. Joseph. Schmidt. E. edition 1. S. Hsiao. Edward Elgar. CO2 emissions. B. Impact of Population Growth.  Brock. L. Glatzer. F. Universite Catholique de Louvain. (2001): GMM estimation of empirical growth models.de/DocCIDL/cesifo_wp367. Industry prices include representative heavy fuel oil. Turut-Asik. C. 1212-1217.S. FEEM Working Paper 87.. 94.Weighted average price of gas (US$-using power purchase parity/unit): It is calculated as a weighted average of industry and household prices (we use the final consumption as weights). 375–385. and the environment.Production of renewable energy per capita (Ktoe/person). The Green Solow Model.
G. Peterson. Journal of Econometrics. 41–54. B.  Wagner.  Kriström. and Sachs. Liège. at: /http://www2. D. .  Hsiao-Tien Pao. C..  Liaskas. M. The impact of population pressure on global carbon dioxide emissions. A. Econ Lett 82(1):121–126  Martínez-Zarzoso. Economic growth. Journal of Public Economics 57.. K.  Vishal 2012. growth and the energy mix in Europe. Development and Sustainability 8. and Clarke. Working paper 07-03. 46. Technology.. 2006.ac. 1419–1439.org/S. Ecological Economics.. Grimes PE(1997) Carbon intensity and economic development 1962–91: a brief exploration of the environmental Kuznets curve.. Presented at the University of Chicago.  Zaba. I.repec. and Morales-Lage. 2004. Nos. Ecological Economics 44. 33. Oxford Bulletin of Economics and Statistics. Ecological Economics 59.I. Panayotou. Derouaux Ordina Editions. M. A Note on the Theme of Too Many Instruments.W.R. 419–428.  Shafik.. Background Paper for the World Development Report 1992..osaka-u. 2004. D. Energy Policy (forthcoming). Economic growth and environmental quality: time series an cross... World Dev 25(2):191–198  Roodman D. Khrushch M.Income and CO2 emissions: evidence from panel unit root and cointegration tests. at: /www. Unander F (2001) Carbon emissions from manufacturing energy use in 13 IEA countries: long-term trends through 1995..I.. Environment. T. and Bandyopadhyay S. 29-42. T.N. N. Working Paper No 904.  Shi. A.  Shen.  Holtz-Eakin.pdfS. University of Tilburg. The World Bank. N. 9. Development Discussion Paper No. 1994. doi:10. 388–408. 417}432. (2002). J.11.  Roberts JT. 28: 231-245. 757-773. 2007. 1992. Lundgren (2005). 1997.. Explorations in Economic History. 2003.Luo. 2005.  Vincent. 426-447. 1994. J. A.. 2010. Y.  Panayotou.  Moomaw WR. . CAER II Discussion Paper 8. and Selden. (1985): “A note on autoregressive error component models”. Unruh GC(1997) Are environmental Kuznets curves misleading us? The case of CO2 emissions. 497–512.  Judson.edu/hiid/701. Causal relationship between energy consumption and GDP growth revisited: a dynamic panel data approach. 86136. (1981): “Biases in dynamic model with fixed effects”. 333-347. 1. Political Economy Workshop. Energy Policy 29:667–688. T. and T. The evolution of carbon dioxide emissions from energy uses in industrialized countries: an end-use analysis. 2011.  Porter. A. Harvard Institute for International Development. Stoking the fires? CO 2 emissions and economic growth. 2008. 2010. 1870-1997’. B. J. 2006.. Hashimoto. CentER Discussion Paper 2005-25. Li. 55–68. 2006.. Bengochea-Morancho A(2004)Testing for environmental Kuznets curves for CO2: evidence from pooled mean group estimates.  Lee.. Environmental Kuznets Curves for CO2: Heterogeneity versus Homogeneity. H. Vincent J.J.  Roodman D.  MacKellar L. S.no/publikasjoner/DP/pdf/dp447. The impact of population on CO2 emissions: evidencefrom European countries. Ting M. G. E. G. The rise and fall of the environmental Kuznets curve.pdfS. Murtishaw S. Lutz W.jp/library/global/dp/0409. 2008. Oxford Economic Papers.... 1999. C. 701. van der Linde. En Zaba y Clarke (ed. D. P. G. 849–865. 7-9. Economic and Political Determinants of Carbon Dioxide Emissions: Reexamining the Environmental Kuznets Curve. 2007. n.  Tucker M(1995) Carbon dioxide emissions and global GDP.. S. R. Owen. . Discussion Paper 04-09.) Environment and population change. (2004). CO2 emissions energy consumption and economic growth in BRIC countries. Karbuz S. ‘An EKC Pattern in Historical Perspective: Carbon Dioxide Emissions. A causality analysis on GDP and air Emissions in Norway.. Vol. October 11.  Lindmark.  Schmalensee R. Chen. Ecological Economics 67. Stoker TM. Energy 30 73–83. Decomposition of industrial CO2 emissions: the case of European Union. R. Energy Policy 37. The CO2 emissions-income nexus: Evidence from rich countries. Judson RA(1998)World carbon dioxide emissions:1950–2050..J... Hwang. J. Testing for environmental Kuznets curves within a developing country. Yang. Is the environmental Kuznets curve hypothesis valid for developing countries? A survey.  Soto :System GMM Estimation with a Small Number of Individuals. T. Decomposition of energy-relatedCO 2 emission in China: 1957–2000.  Wang. Energy Economics 22 (2000). Environ Resource Econ 38. Econometrica. 49: 1417-1426. 7850–7860.. Dijkgraaf and E.. Mandaraka. 2008. Energy Policy 39. How to Do Xtabond2: An Introduction to Difference and System GMM in Stata.. The carbon Kuznets curve: a cloudy picture emitted by bad econometrics? Resource and Energy Economics 30 (3). Ding.. pp. Harvard Business Review September–October.Lee. L.R. 1-26.  Marrero. M. B. Niu.  Shipper et al. Research Department of Statistics Norway. 383–394. pp. Vol.  Islam. 71. Environmental Kuznets curve on country level: evidence from China.1016/j. 135-158  Schipper L. 2009. 85-101.pdfS.harvard. at: /http://www. households and CO2 emissions. M. 2007  Stern. Ecol Econ 15:215–223  Verbeke. Rev Econ Stat 80(1):15–27  Sevestre. Bengochea-Morancho.. 25. 1995. 413–423.  Lindmark. G. Economic Growth and Environmental Degradation in a Global Context.country evidence. J. ‘Swedish CO2 Emissions 19002100 – An Exploratory Note’. Y. Environment and Development Economics Part 4. Energy Economics 32 (2010) 1356–1363  Martínez-Zarzoso I. C. M.9  He. A.2010.  Kahuthu.C.D. ‘Patterns of Historical CO2 Intensity Transitions among High and Low Income Countries’. Goujon A(1995) Population.. 1975-1996: evidence from pooled cross-country data. Mavrotas.. Zou. A.. 41. and Diakoulaki. J. J. mimeo. C. Energy Policy 38. Prinz C. Melenberg (2005)..econ. 2000.  Huang. Green and competitive: ending the stalemate. vol.ssb. 2005. A. vol. 120–134.  Liu. Institute for Economic Analysis. Y. 1995.  Metcalf.. 2. Energy Policy. M. Environ Dev Econ 2:451– 463  Nickell. Unveiling the income– environment relationship: an exploration into the determinants of environmental quality. at: /http://ideas. ‘Economic Development and Environmental Quality – An Econometric Analysis’.  Shafik N. Chung-Ming Tsai.. Discussion Papers 447. 1.. 2121–2131. Stata Journal. Energy Policy. (2009b). Economics Letters.. 29(3). Introduction: current directions in population-environment research. De Clercq. 2000. Is the Environmental Kuznets Curve driven by structural change? What extended time series may imply for developing countries. n. An Empirical Analysis of Energy Intensity and Its Determinants at the State Level The Energy Joumal.  Waldhoff S. Popul Dev Rev 21(4). 1223-1230.050  Vollebergh. 42. Trognon. (1999): “Estimating dynamic panel data models: a guide for macroeconomists”. J.. 1997. Y. 65: 9-15. (2009a).  Niu. Fuel Prices and Growth in Sweden. World Development 32 (8).cid. The income–environment relationship: evidence from a binary response model. M. Barcelona.enpol. Washington DC.. Greenhouse gases emissions. energy conservation and emissions reduction: A comparative analysis based on panel data for 8 Asian-Pacific countries.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.