Abstract
We address international convergence in carbon dioxide emissions per capita and per value added derived from emission inventories based on production and consumption patterns. We propose a Bayesian structural model that accounts for heteroscedasticity and endogeneity between emissions and economic growth, and tests for the existence of group-specific convergence via shrinkage priors. We find evidence for country-specific conditional convergence in all emission inventories, implying a half-life of 2.7–3.1 years for production-based emissions and 3.6–4.7 years for consumption-based emissions. When testing for global convergence without allowing for individual-specific convergence paths, the half-life of \(\hbox {CO}_2\) per capita increases to 15–26 years, whereas emission intensities show a half-life of 44–45 years. Our results highlight the current incompatibility between emission targets and economic growth and the need for faster diffusion of green technologies. Moreover, there is no evidence for specific convergence dynamics in the European Union, the OECD, or the countries that are subject to binding emission constraints specified in the Kyoto Protocol. The institutional frameworks implemented in industrialized countries did not induce faster convergence among developed economies.
Similar content being viewed by others
Notes
See Knutti et al. (2015) for a critical analysis of the 2 °C target.
Empirical studies that investigate the existence of an EKC in \(\hbox {CO}_2\) emissions usually fail to find such a relationship in samples covering a large group of countries (see e.g. Stern 2004, and Stern 2017, for exhaustive surveys, or Fernández-Amador et al. 2017, for a survey of empirical applications). Aslanidis and Iranzo (2009) and Fernández-Amador et al. (2017) provided evidence that the income elasticity of \(\hbox {CO}_2\) emissions decreases as income per capita rises above a threshold level though emissions continue growing, what challenges the sustainability of economic growth.
The Paris Agreement recognizes the need to support developing countries in order to facilitate the effective implementation of the objectives identified in the Agreement (Paris Agreement, Art. 2).
The findings of the literature range from evidence for convergence (Strazicich and List 2003; Nguyen 2005; Ezcurra 2007; Romero-Ávila 2008; Lee et al. 2008; Westerlund and Basher 2008; Lee and Chang 2009; Brock and Taylor 2010; Jobert et al. 2010; Huang and Meng 2013; Yavuz and Yilanci 2013; Anjum et al. 2014; Hao et al. 2015; Wu et al. 2016; Zhao et al. 2015) over the existence of convergence clubs (Nguyen 2005; Aldy 2006; Lee and Chang 2008; Panopoulou and Pantelidis 2009; Barassi et al. 2011; Ordás Criado and Grether 2011; Camarero et al. 2013; Herrerias 2013; Wang et al. 2014; Burnett 2016) to no evidence for convergence (Aldy 2007; Barassi et al. 2008; Nourry 2009).
Aldy (2007) investigated convergence of \(\hbox {CO}_2\) emissions across US states. This is the only study so far that also covers consumption inventories. The author did not find evidence for convergence for either \(\hbox {CO}_2\) production or for \(\hbox {CO}_2\) consumption per capita. In contrast to Aldy, our study covers economies at different development states, thus being the first one to evaluate global convergence patterns in \(\hbox {CO}_2\) consumption.
Salois and Balcombe (2015) proposed a related model in the context of cross-sections, where t-distributed errors in an IV-model are represented by weighted errors of normals. Their modelization shares with ours the use of a scale mixture of normals representation, though the authors do not perform Cholesky-rotation of the system to represent it as a recursive system of equations but condition the weighted errors on each other and use a Wishart prior for the variance–covariance matrix.
Earlier studies focused on unconditional convergence, while more recent studies tested for conditional convergence, i.e. convergence after allowing for heterogeneity across countries by accounting for additional determinants of economic growth. While unconditional convergence was often found for OECD countries, it was generally rejected for samples including non-OECD countries. If countries converge to different steady states, unconditional convergence models might result in biased coefficient estimates because the model used for estimation is miss-specified (Barro and Sala-i-Martin 2004). See for example Baumol (1986), Barro (1991), Barro and Sala-i-Martin (1992), Mankiw et al. (1992) and Barro and Sala-i-Martin (2004).
See e.g. Barro and Sala-i-Martin (1992), Quah (1993), Sala-i-Martin (1996) and Young et al. (2008). Phillips and Sul (2007b) developed a test for identifying club-convergence groups, which corresponds to a test for conditional \(\sigma \)-convergence (see Phillips and Sul 2007b). Phillips and Sul (2007a) provided a short empirical application of the test in the context of economic growth convergence.
See e.g. Ravallion (2003) who applied \(\beta \)-convergence tests to international income inequality.
Studies for OECD countries focused mainly on stochastic and \(\beta \)-convergence. For more details on the concept of convergence used by the respective studies, see Table A.1 in the Online Appendix.
Schmalensee et al. (1998) is an exception, finding support for an inverse-U relationship using non-parametric techniques. More recently, Aslanidis and Iranzo (2009) and Fernández-Amador et al. (2017) found that the income-elasticity of \(\hbox {CO}_2\) emissions decreases slightly after income per capita passes a certain threshold, such that relative decoupling increases with economic growth, though there is no evidence of absolute decoupling and an EKC relationship. Fernández-Amador et al. (2017) also provided evidence for a similar pattern in \(\hbox {CO}_2\) consumption-based inventories.
Anjum et al. (2014) reported that the negative correlation between initial emissions and subsequent emission growth is stronger for \(\hbox {CO}_2\) intensity than for \(\hbox {CO}_2\) per capita.
All these studies define \(\hbox {CO}_2\) intensity as \(\hbox {CO}_2\) per GDP. In our analysis we refer to \(\hbox {CO}_2\) intensity as \(\hbox {CO}_2\) per value added.
A description of the countries included in the composite regions is available in Fernández-Amador et al. (2016). The dataset has been recently extended by the authors to cover the year 2014.
This corresponds to calculating average annual growth rates. For a similar method see Ravallion (2003), who accounts for the unequal spacing in time between measures of income inequality for large-N, small-T panel data by regressing the difference in inequality between time t and the initial period \(t_1\) on a constant and initial inequality at time \(t_1\), both multiplied by a time-trend (\(t-1\)). In contrast to Ravallion’s data, our panel is balanced in the sense that for every individual we observe all variables at the same points in time. Thus, we can also exploit the variation of the data across time and use initial emissions in year \(t-s\) instead of in year \(t_1\) as a regressor.
Bernard and Durlauf (1996) pointed out that the power of time-series tests may be weak when the dynamics do not occur near the steady state. In this sense, time-series approaches to test for stochastic convergence may not be particularly suitable in our context, since data on \(\hbox {CO}_2\) emissions covering a global sample of countries are very likely to be driven by transition dynamics rather than being near the steady state.
Detailed information on the sector aggregation from the original GTAP sectoral disaggregation is available from the authors upon request.
Ordás Criado et al. (2011) tested for convergence in sulfur oxides and nitrogen oxides. Their theoretical model assumes optimal control of pollution emissions at the national level, making it particularly suited for applications to local air pollutants. Nevertheless, the structure of the empirical model they specify is compatible with the green Solow model by Brock and Taylor (2010), which the authors applied to \(\hbox {CO}_2\) emissions. Ordás Criado et al. (2011) regressed the average growth rates of emissions over the period \(t-5\) to t on the level of emissions at the initial period of the growth rate (\(t-5\)), the growth rate of GDP over \(t-5\) and t, GDP in \(t-5\), and time- and individual-dummies using OLS and a non-parametric model. The authors also addressed endogeneity between emissions and GDP by instrumenting GDP and its growth rate with their lagged values (following Barro and Sala-i-Martin 1992). Brock and Taylor (2010) developed a theoretical model that also predicts conditional \(\beta \)-convergence, which the authors applied to \(\hbox {CO}_2\) emissions. Although Brock and Taylor’s model is applicable to global pollutants, in our empirical approach we follow Ordás Criado et al. (2011) since their empirical analysis makes use of the panel structure of the data and accounts for the potential endogeneity of GDP per capita.
For the first period in our sample, 1997–2001, we use the average growth rate for a period of the same length, 1993–1997, as instrument.
In the original dynamic model of emissions, once we undo the average growth rate, the relevant parameter for the case of regular sampling every period is \((1+\beta )\). The precision elicited ensures that the hypothesis of an unit root in our autoregressive model with explanatory variables is not an extreme event in our prior for \(\beta \).
The precision is defined as the inverse of the variance. A precision of 0.2 implies a variance of 5.
Note that the precision of the individual-dummies is larger than the precision of the rest of the parameters. A uniform prior on the individual fixed effects would lead to improper posterior distributions for the parameters of interest, while very diffuse priors would lead to very slow convergence of the MCMC algorithm used for inference (see e.g. Lancaster 2008, Chap. 7).
Scale mixture of normals with the weights specified as in (9) are equivalent to a t-student distribution (see e.g. Andrews and Mallows 1974; West 1987; Ding 2016). The degrees of freedom of the t-student are equal to the hyperparameter governing the distribution of the weights \(\omega _{i}\), \(\nu \). With growing \(\nu \) the distribution converges to a normal distribution, as less probability mass is concentrated at the tails of the distribution. The prior for the weights in the scale mixture of normals, \(\omega _{i}\), together with the prior for the components of the variance matrix \(\Sigma \) that we will define below, imply a form of cross-sectional heteroscedasticity of the gamma type (Andrews and Mallows 1974; Geweke 1993; Koop 2003, Chap. 6; Lancaster 2008, Chap. 3). There are two main advantages of modeling the problem in terms of scale mixture of normals rather than as a t-student distribution. The first one is that the type of heteroscedasticity, cross-sectional in our case, can be explicitly stated. The second is that it is less computational demanding for the numerical algorithm to estimate the posterior distributions of the parameters.
We regard the priors for the parameters of interest \((\beta , \pi _0, \pi _1, \{\theta _r\}, \{\delta _t\}, \alpha _{iv}, \beta _{iv}, \{\alpha _i\},\{\beta _j\})\) as informative. Geweke (1993) shows that under informative (normal) priors for the slopes, both the first and the second moments of the slopes exist. When the priors of the slopes are uninformative, \(\nu >2\) ensures existence of the first moments, while \(\nu >4\) ensures existence of the second moments. Thus, the truncation defined contains roughly \(80\%\) of the density around the mean of the prior, while ensuring existence of first moments even in the case of noninformative priors for the parameters of interest.
We explain the derivation of the IV-prior in terms of covariance matrices because this is common in the literature, though the specification of the priors is in terms of precisions, as carried out in the software. Alternatively, we could use an inverted Wishart prior for \(\Sigma \), \(\Sigma \sim IW(v_0, \Sigma _0)\), with parameters \(v_0\) and \(\Sigma _0\). Priors for covariance matrices and variances have usually been addressed by means of inverted Wishart and inverted Gamma distributions, respectively, while Wishart or Gamma distributions have been used as priors for precision matrices and precisions. Wishart priors have been extensively used in the framework of Bayesian instrumental variable models under normal-distributed errors (see e.g. Kleibergen and Zivot 2003; Lancaster 2008, Chap. 8; Rossi et al. 2005).
Ding (2016) used the representation of a multivariate t-student distribution as a scale mixture of multivariate normals to derive the conditional distribution of the multivariate t-student, which can be represented by the conditional normal distribution times the conditional distribution of the weights.
That was sufficient for the chains to show mixing and the estimates of the coefficients to show convergence to their ergodic distribution.
The results from our simulations are available from the authors upon request. These simulations did not include the null of endogeneity between explained and explanatory variables and thus do not introduce an IV structure in the model. A more detailed simulation-based analysis of the Bayesian estimator in comparison with alternative dynamic panel estimators can be found in Fernández-Amador and Oberdabernig (2018).
Furthermore, we report the results of the DV-conditional and conditional homoscedastic models (normal-distributed errors with and without individual-dummies) in Tables A.4 and A.5 in the Online Appendix.
The underlying data cover 468 observations. Because we use the growth rate of emissions as dependent variable, our final sample includes 390 observations.
It should be noted that Bayesian estimation does not aim at minimizing the sum of square residuals and thus, it does not maximize the \(R^2\). However, we consider it together with the DIC when assessing how well our models fit the data and whether they can be regarded as consistent with our data. The DIC penalizes the number of parameters and is often regarded as a better measure of fit in the Bayesian context than the \(R^2\).
The half-life provides an indication of the speed of convergence. It is defined as the time required to eliminate half of the initial gap between actual emissions levels and the steady state. The half-life is calculated as \(\frac{-ln(0.5)}{-ln(1+\beta )}\) (see Allington and McCombie 2007, p. 206).
The half-life in their sample of developed countries was estimated to lie between 4.2 and 6.2 years; this is longer than the half-life estimated in their pooled sample including developing countries.
These figures correspond to estimates of conditional convergence. For unconditional convergence the authors reported a half-life between 4 and 8.5 years.
The negative effect of democracy on emissions growth is not robust to using alternative measures of democracy, such as the democracy measure sourced from the FSD1289 Measures of Democracy 1810–2014 database (see Finnish Social Science Data Archive 2018) or the average of the Freedom House indices of political rights and civil liberties (see Freedom House 2018). The main results are not sensitive to these alternative specifications and are also robust to the exclusion of the democracy variable.
The negative impact of the construction sector may be related to the low carbon intensity of this sector during the period analyzed. We take the value added share of agriculture as the benchmark sector and exclude it from the specifications in order to avoid multicollinearity.
The results of the models without individual-dummies could be affected by omitted variables and should be taken with care (see Barro and Sala-i-Martin 2004).
Tables A.4 and A.5 in the Online Appendix report the results for the models with homoscedastic errors. The main results do not change qualitatively. The only important qualitative change is that income growth becomes significant for emissions per value added in the DV-conditional models. The convergence coefficient only changes slightly, though this change is amplified in the half-lives, decreasing them in the specifications for emissions per value added in the conditional models.
A more detailed analysis of this mechanism is out of the scope of this paper.
References
Aichele R, Felbermayr G (2015) Kyoto and carbon leakage: an empirical analysis of the carbon content of bilateral trade. Rev Econ Stat 97(1):104–115
Aldy JE (2006) Per capita carbon dioxide emissions: convergence or divergence? Environ Resour Econ 33:533–555
Aldy JE (2007) Divergence in state-level per capita carbon dioxide emissions. Land Econ 83:353–369
Allington NF, McCombie JS (2007) Economic growth: new directions in theory and policy. Edward Elgar Publishing Limited, Ch. Economic growth and beta-convergence in the East European Transition Economies, pp 200–222
Andrews D, Mallows C (1974) Scale mixtures of normal distributions. J R Stat Soc Ser B 36:99–102
Anjum Z, Burke P, Gerlagh R, Stern D (2014) Modeling the emissions-income relationship using long-run growth rates. CCEP working paper 1403
Arellano M, Bond S (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Rev Econ Stud 58(2):277–297
Aslanidis A, Iranzo S (2009) Environment and development: is there a Kuznets curve for CO\(_2\) emissions? Appl Econ 41(6):803–810
Azomahou T, Laisney F, Van PN (2006) Economic development and CO\(_{2}\) emissions: a nonparametric panel approach. J Public Econ 90(6):1347–1363
Babiker MH (2005) Climate change policy, market structure, and carbon leakage. J Int Econ 65:421–445
Barassi M, Cole MA, Elliott RJ (2008) Stochastic divergence or convergence of per capita carbon dioxide emissions: re-examining the evidence. Environ Resour Econ 40:121–137
Barassi MR, Cole MA, Elliott RJ (2011) The stochastic convergence of CO\(_2\) emissions: a long memory approach. Environ Resour Econ 49:367–385
Barro R (1991) Economic growth in a cross section of countries. Q J Econ 106(2):407–443
Barro J, Sala-i-Martin X (1992) Convergence. J Polit Econ 100(2):223–251
Barro R, Sala-i-Martin X (2004) Economic growth, 2nd edn. The MIT Press, Cambridge
Baumol W (1986) Productivity growth, convergence, and welfare: what the long-run data show. Am Econ Rev 76(5):1072–1085
Bernard A, Durlauf S (1996) Interpreting tests of the convergence hypothesis. J Econom 71(1–2):161–173
Blundell R, Bond S (1998) Initial conditions and moment restrictions in dynamic panel data models. J Econom 87(1):115–143
Brock W, Taylor M (2010) The green Solow model. J Econ Growth 15:127–153
Bruno GS (2005) Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models. Econ Lett 87(3):361–366
Burnett J (2016) Club convergence and clustering of US energy-related CO\(_2\) emissions. Resour Energy Econ 46:62–84
Camarero M, Castillo J, Picazo AJ, Tamarit C (2013) Eco-efficiency and convergence in OECD countries. Environ Resour Econ 55:87–106
Carlino G, Mills L (1993) Are US regional incomes converging? a time series analysis. J Monet Econ 32(2):335–346
Dasgupta S, Laplante B, Wang H, Wheeler D (2002) Confronting the environmental Kuznets curve. J Econ Perspect 16:147–168
De Vos I, Everaert G, Ruyssen I (2015) Bootstrap-based bias correction and inference for dynamic panels with fixed effects. Stata J 15(4):986–1018
Ding P (2016) On the conditional distribution of the multivariate t distribution. Am Stat 70:293–295
Efron B (1982) The Jacknife, the bootstrap and other resampling plans. Vol. 38 of CBMS-NSF regional conference series in applied mathematics. Society for Industrial and Applied Mathematics (SIAM), Philadelphia
Efron B (2011) The bootstrap and Markov Chain Monte Carlo. J Biopharm Stat 21:152–162
Evans P, Karras G (1996) Convergence revisited. J Monet Econ 37(2):249–265
Everaert G, Pozzi L (2007) Bootstrap-based bias correction for dynamic panel. J Econ Dyn Control 31:1160–1184
Ezcurra R (2007) Is there cross-country convergence in carbon dioxide emissions? Energy Policy 35:1363–1372
Farzin Y, Bond C (2006) Democracy and environmental quality. J Dev Econ 81:213–235
Fernández-Amador O, Oberdabernig D (2018) Bayesian estimation of dynamic panels. Mimeo
Fernández-Amador O, Francois JF, Tomberger P (2016) Carbon dioxide emissions and international trade at the turn of the millennium. Ecolog Econ 125:14–26
Fernández-Amador O, Francois JF, Oberdabernig DA, Tomberger P (2017) Carbon dioxide emissions and economic growth: an assessment based on production and consumption emission inventories. Ecolog Econ 135:269–279
Finnish Social Science Data Archive (2018) Measures of democracy 1810–2018
Freedom House (2018) Country and territory ratings and statuses, 1973–2018
Friedman M (1992) Do old fallacies ever die? J Econ Lit 30(4):2129–2132
Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Bayesian data analysis. Texts in statistical science, 2nd edn. CRC Press, Boca Raton
Gelman A, Goodrich B, Gabry J, Ali I (2017) R-squared for Bayesian regression models. Unpublished manuscript, Department of Statistics, Columbia University
George EI, McCulloch RE (1993) Variable selection via Gibbs sampling. J Am Stat Assoc 88:881–889
Geweke J (1993) Bayesian treatment of the independent Student-t linear model. J Appl Econom 8:S19–S40
Grossman G, Helpman E (1995) Handbook of international economics. North-Holland, Amsterdam (Chap. 25: Technology and trade)
Hao Y, Liao H, Wei Y-M (2015) Is China’s carbon reduction target allocation reasonable? An analysis based on carbon intensity convergence. Appl Energy 142:229–239
Hastie T, Tibshirani R, Friedman J (2009) The elements of statistical learning. Series in statistics, 2nd edn. Springer, Berlin
Herrerias M (2013) The environmental convergence hypothesis: carbon dioxide emissions according to the source of energy. Energy Policy 61:1140–1150
Holtz-Eakin D, Selden TM (1995) Stoking the fires? CO\(_2\) emissions and economic growth. J Public Econ 1995:85–101
Hsiao C, Pesaran H, Tahmiscioglu K (1999) Analysis of panels and limited dependent variable models. Cambridge University Press, Cambridge (Chap. 11: Bayes estimation of short-run coefficients in dynamic panel data models, pp 268–296)
Huang B, Meng L (2013) Convergence of per capita carbon dioxide emissions in urban China: a spatio-tomporal perspective. Appl Geogr 40:21–29
Jobert T, Karanfil F, Tykhonenko A (2010) Convergence of per capita carbon dioxide emissions in the EU: legend or reality? Energy Econ 32:1364–1373
Kaika D, Zervas E (2010) The environmental kuznets curve (EKC) theory—part A: concept, causes and the CO\(_2\) emissions case. Energy Policy 62:1392–1402
Kiviet JF (1995) On bias, inconsistency, and efficiency of various estimators in dynamic panel data models. J Econom 68(1):53–78
Kleibergen F, Zivot E (2003) Bayesian and classical approaches to instrumental variable regression. J Econom 114:29–72
Knutti R, Rogelj J, Sedlace J, Fischer E (2015) A scientific critique of the two-degree climate change target. Nat Geosci 9:13–19
Koop G (2003) Bayesian econometrics. Wiley, London
Lancaster T (2000) The incidental parameter problem since 1948. J Econom 95:391–413
Lancaster T (2008) An introduction to modern Bayesian econometrics. Blackwell, Oxford
Lee C-C, Chang C-P (2008) New evidence on the convergence of per capita carbon dioxide emissions from panel seemingly unrelated regressions augmented Dickey–Fuller tests. Energy 33:1468–1475
Lee C-C, Chang C-P (2009) Stochastic convergence of per capita carbon dioxide emissions and multiple structural breaks in OECD countries. Econ Model 26:1375–1381
Lee C-C, Chang C-P, Chen P-F (2008) Do CO\(_2\) emission levels converge among 21 OECD countries? new evidence from unit root structural break tests. Appl Econ Lett 15:551–556
Lopes H, Polson N (2014) Bayesian instrumental variables: priors and likelihoods. Econom Rev 33:100–121
Maddala G, Hu W (1996) The econometrics of panel data: a handbook of theory with applications, 2nd edn. Kluwer, Boston (Chap. 13: The pooling problem, pp 307–322)
Mankiw N, Romer D, Weil D (1992) A contribution to the empirics of economic growth. Q J Econ 107(2):407–437
Moon H, Perron B, Phillips P (2015) The Oxford Handbook of Panel Data. Oxford University Press, Oxford, pp 111–148 (Chap. 4: Incidental parameters and dynamic panel modeling)
Newton M, Raftery A (1994) Approximate Bayesian inference with the weighted likelihood bootstrap. J R Stat Assoc Ser B 56:3–48 (With discussion and reply by the authors)
Neyman J, Scott E (1948) Consistent estimation from partially consistent observations. Econometrica 16:1–32
Nguyen PV (2005) Distribution dynamics of CO\(_2\) emissions. Environ Resour Econ 32:495–508
Nickell S (1981) Biases in dynamic models with fixed effects. Econometrica 49:1417–1426
Nourry M (2009) Re-examining the empirical evidence for stochastic convergence of two air pollutants with a pair-wise approach. Environ Resour Econ 44:555–570
Ordás Criado C, Grether J-M (2011) Convergence in per capita CO\(_2\) emissions: a robust distributional approach. Resour Energy Econ 33:637–665
Ordás Criado C, Valente S, Stengos T (2011) Growth and pollution convergence: theroy and evidence. J Environ Econ Manag 62:199–214
Panopoulou E, Pantelidis T (2009) Club convergence in carbon dioxide emissions. Environ Resour Econ 44:47–70
Pettersson F, Maddison D, Acar S, Söderholm P (2014) Convergence of carbon dioxide emissions: a review of the literature. Int Rev Environ Resour Econ 7:141–178
Phillips P, Sul D (2007a) Some empirics on economic growth under heterogenous technology. J Macroecon 29:455–469
Phillips P, Sul D (2007b) Transition modelling and econometric convergence tests. Econometrica 75(6):1771–1855
Quah D (1993) Empirical cross-section dynamics in economic growth. Eur Econ Rev 37(2–3):426–434
Quah D (1996) Empirics for economic growth and convergence. Eur Econ Rev 40:1353–1375
Ravallion M (2003) Inequality convergence. Econ Lett 80:351–356
Romero-Ávila D (2008) Convergence in carbon dioxide emissions among industrialised countries revisited. Energy Econ 30:2265–2282
Rossi PE, Greg A, McCulloch R (2005) Bayesian statistics and marketing. Wiley series in probability and statistics. Wiley, Chichester
Rubin D (1981) The Bayesian bootstrap. Ann Stat 9:130–134
Sala-i-Martin X (1996) Regional cohesion: evidence and theories of regional growth and convergence. Eur Econ Rev 40(6):1325–1352
Salois M, Balcombe K (2015) A generalized Bayesian instrumental variable approach under student \(t\)-distributed errors with application. Manch Sch 83:499–522
Schmalensee R, Stoker T, Judson R (1998) World carbon dioxide emissions: 1950–2050. Rev Econ Stat 80(1):15–27
Solow R (1956) A contribution to the theory of economic growth. Q J Econ 70(1):65–94
Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc Ser B 64:583–639 (with discussion)
Stern D (2004) The rise and fall of the environmental Kuznets curve. World Dev 32:1419–1439
Stern D (2017) The environmental Kuznets curve after 25 years. J Bioecon 19:1–22
Strazicich MC, List JA (2003) Are CO\(_2\) emission levels converging among industrial countries? Environ Resour Econ 24:263–271
Wang Y, Zhang P, Huang D, Cai C (2014) Convergence behavior of carbon dioxide emissions in China. Econ Model 43:75–80
West M (1987) On scale mixtures of normal distributions. Biometrika 74:646–648
Westerlund J, Basher SA (2008) Testing for convergence in carbon dioxide emissions using a century of panel data. Environ Resour Econ 40:109–120
Wu J, Wu Y, Guo X, Cheong TS (2016) Convergence of carbon dioxide emissions in Chinese cities: a continuous dynamic distribution approach. Energy Policy 91:207–219
Yavuz N, Yilanci V (2013) Convergence in per capita carbon dioxide emissions among G7 countries: a TAR panel root approach. Environ Resour Econ 54:283–291
Young A, Higgins M, Levy D (2008) Sigma convergence vs beta convergence: evidence from US country-level data. J Money Credit Bank 40(5):1083–1093
Zhao X, Burnett W, Lacombe D (2015) Province-level convergence of China’s carbon dioxide emissions. Appl Energy 150:286–295
Acknowledgements
The authors thank Carlos Ordás Criado for valuable insights and the participants of the SWSR for inspiring discussions. The authors acknowledge support of the NRP 73 project Switzerland’s Sustainability Footprint: Economic and Legal Challenges, grant No. 407340-172437, University of Bern, supported by the Swiss National Science Foundation (SNSF) within the framework of the National Research Programme “Sustainable Economy: resource-friendly, future-oriented, innovative” (NRP 73).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Fernández-Amador, O., Oberdabernig, D.A. & Tomberger, P. Testing for Convergence in Carbon Dioxide Emissions Using a Bayesian Robust Structural Model. Environ Resource Econ 73, 1265–1286 (2019). https://doi.org/10.1007/s10640-018-0298-9
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10640-018-0298-9