Skip to main content


Log in

Economic growth and pollutant emissions: new panel evidence from the union for the Mediterranean countries

  • Published:
Economic Change and Restructuring Aims and scope Submit manuscript


This paper investigates the existence of an environmental Kuznets curve (EKC) and its robustness using data from the Union for the Mediterranean (UfM) countries over the 1970–2020 period. Our methodology relies on four recent estimation methods for non-stationary panel data and includes four pollutants. Two main results emerge from our analysis. First, the EKC does not hold for most pollutants, and its validity crucially depends on the estimation techniques considered. Second, the Pooled-Mean Group method is the most favourable one and confirms the existence of an inverted U-shaped relationship for CO2 and SO2. Our results provide beneficial information for decision-makers. They suggest implementing proactive instruments based on both flexible regulations and tax incentives to stimulate ecological transition.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.


Source World Development Indicators (World Bank). (Monaco and Palestine had to be eliminated from the sample due to the unavailability of data)

Similar content being viewed by others

Data availability

Information on data source is provided as a supplementary material.


  1. Mediterranean Experts on Climate and Environmental Change (MedECC):

  2. For comparison purpose, we use the estimators of the Mean Group (MG) of Pesaran and Smith (1995), those of the augmented mean group (AMG) of Eberhardt and Eal (2010), and those of the correlated common effects mean group (CCEMG) of Pesaran (2006).

  3. The UfM initially included 44 countries before Syria suspended its participation on November 30, 2011, due to EU sanctions. Libya has the observer status at the UfM.

  4. Business development, improvement of the quality of higher education and scientific research, civil and social affairs, transport and urban development, energy and action for the climate and water and environment.

  5. The only econometric study testing for the EKC hypothesis in the UfM region is that of Sebri (2009), which analyses the trajectory of CO2 according to per capita income over the 1980–2005 period. Using three different models (linear, log-linear, and semi-log-linear), he confirms the EKC hypothesis for UfM countries.

  6. Countries are Bulgaria, Croatia, Estonia, Hungary, Iceland, Latvia, Lithuania, Macedonia, Malta, Romania, Slovak Republic, Czech Republic, Romania, Slovenia, and Turkey.

  7. Additional estimations including a larger number of control variables (such as Kaufman's governance indicators, urbanization, education, energy intensity, electricity consumption, agriculture (% of GDP), industry (% of GDP)) were also carried out without fundamentally changing the results. Therefore, not to overload tables, we limit ourselves thereafter to these four control variables.

  8. The GDP, ENERG, DPOP OP_TRA variables come from the World Bank database. FDI is taken from the database of the United Nations Conference on Trade and Development (UNCTAD).

  9. Note that the increase in temperatures is not directly considered in the estimated equation and could be a common factor modifying the impact of growth on pollution. The main reason is that we do not have data on the evolution of temperatures. In addition, it is likely that given our relatively short time series analysis, the temperature evolution is not important enough to significantly impact this relationship. However, this aspect is considered indirectly by introducing temporal effects in the econometric specification.

  10. Some UfM countries had to be eliminated from the sample during estimations due to the unavailability of data and geopolitical wars in some regions which suspended their membership, or if countries have had the observer status. See Table 9 for the list of selected countries.

  11. Other second-generation panel unit root tests (not reported here) have also been implemented, especially those of Bai and Ng (2004), Moon and Perron (2004), and Smith et al. (2004), without changing the conclusions.

  12. See Appendix 4 for a brief overview of each method.

  13. The EKC hypothesis is empirically verified only if the estimated coefficients of the income per capita (GDP) and the income per capita squared (GDP2) are respectively positive and negative (and significant).

  14. See for instance Dijkraaf and Vollebergh (2005), or Galeotti et al. (2006).

  15. Germany, Italy, France, Spain, United Kingdom, Netherlands, Poland, and Turkey.

  16. American Council for an Energy-Efficient Economy.


  • Ahmad M, Khan Z, Rahman ZU, Khattak SI, Khan ZU (2021) Can innovation shocks determine CO2 emissions (CO2 e) in the OECD economies? A new perspective. Econ Innov New Technol 30(1):89–109

    Article  Google Scholar 

  • AlmulaliSolarin USA, Ozturk I (2016) Investigating the presence of the environ-mental Kuznets curve (CEK) hypothesis in Kenya: an autoregressive distributed lag (ARDL) approach. Nat Hazards 80(3):1729–1747

    Article  Google Scholar 

  • Almulali U, Saboori B, Ozturk I (2015) Investigating the environmental Kuznets curve hypothesis in Vietnam. Energy Policy 76:123–131

    Article  Google Scholar 

  • Al-Mulali U, Ozturk I (2016) The investigation of environmental Kuznets curve hypothesis in the advanced economies: The role of energy prices. Renew Sustain Energy Rev 54:1622–1631

    Article  Google Scholar 

  • Apergis N, Ozturk I (2015) Testing Environmental Kuznets Curve hypothesis in Asian countries. Ecol Indic 52:16–22

    Article  Google Scholar 

  • Assamoi GR, Wang S, Liu Y, Gnangoin YTB (2020) Investigating the pollution haven hypothesis in Cote d’Ivoire: evidence from autoregressive distributed lag (ARDL) approach with structural breaks. Environ Sci Pollut Res 27(14):16886–16899

    Article  Google Scholar 

  • Atasoy BS (2017) Testing the environmental Kuznets curve hypothesis across the U.S.: evidence from panel mean group estimators. Renew Sustain Energy Rev 77:731–747

    Article  Google Scholar 

  • BaiNg JS (2004) A panic Attack on unit Roots and Cointegration. Econometrica 72(4):127–1177

    Google Scholar 

  • Belaid F, Youssef M (2017) Environmental degradation, renewable and nonrenewable electricity consumption, and economic growth: assessing the evidence from Algeria. Energy Policy 102:277–287

    Article  Google Scholar 

  • Belaïd F, Zrelli MH (2019) Renewable and non-renewable electricity consumption, environmental degradation, and economic development: evidence from Mediterranean countries. Energy Policy 133:110929

    Article  Google Scholar 

  • Bhattarai M, Hamming M (2001) Institutions and the environmental kuznets curve for deforestation; a cross-country Analysis for Latin America, Africa, and Asia. World Dev 29(6):995–1010

    Article  Google Scholar 

  • Brajer V, Mead RW (2007) Health Benefits of Tunnelling through the Chinese environmental Kuznets curve (EKC). Ecol Econ 66(4):674–686

    Article  Google Scholar 

  • Chudik A, Pesaran MH (2013) Large panel data models with cross-sectional dependence: A survey. SSRN Electron J.

    Article  Google Scholar 

  • Cole MA, Neumayer E (2004) Examining the impact of demographic factors on air pollution. Popul Environ 26:5–21

    Article  Google Scholar 

  • Dijkgraaf E, Vollegergn HRJ (2005) Environmental Kuznets Revisited: Time Series versus Panel Estimation: the CO2 Case. Erasmus University Rotterdam, OCFEB, Research Memorandum, p 9806

    Google Scholar 

  • Eberhardt M, Bond S. (2009). Cross-section dependence in non-stationary panel models: a novel estimator. Munich Personal Repec Arch (MPRA) Pap No 2009; 17692.

  • Eberhardt M. and Eal F. (2010). Productivity analysis in global manufacturing production. Univ Oxf Econ Ser. Working Papers 515.

  • Ferrantino MJ (1997) International trade, environmental quality and public policy. World Econs 20(1):43–72

    Article  Google Scholar 

  • Galeotti M, Lanza A, Pauli F (2006) Reassessing the environmental Kuznets curve for CO2 emissions: a robustness exercise. Ecol Econ 57:152–163

    Article  Google Scholar 

  • Gharnit S, Bouzahzah M, Soussane JA (2019) Foreign direct investment and pollution havens: evidence from African countries. Arch Bus Res 7(12):244–252

    Article  Google Scholar 

  • Grether JM, Mathys NA, de Melo J (2007) Is Trade Bad for the Environment? Decomposing World-Wide SO2 Emissions 1990–2000. Discussion Paper, University of Geneva

  • Grossman GM, Krueger AB (1991). Environmental impacts of a North American Free Trade Agreement. National Bureau of Economic Research, Working Paper No 3914.

  • Grossman G, Krueger A.B. (1994). Economic growth and the environment. In: National Bureau of Economic Research, working paper n° 4634, 21 p., in Quarterly Journal of Economics, May 1995, p. 353–377.

  • Hamit-Haggar M (2012) Greenhouse gas emissions, energy consumption and economic growth: a panel cointegration analysis from Canadian industrial sector perspective. Energy Econ 34:358–364

    Article  Google Scholar 

  • Hassan K, Salim R (2015) Population ageing, income growth and CO2 emission. Empirical evidence from high income OECD countries. J Econ Stud 42:154–167

    Article  Google Scholar 

  • He F, Chang KC, Li M, Li X, Li F (2020) Bootstrap ARDL test on the relationship among trade, FDI, and CO2 emissions: based on the experience of BRICS countries. Sustainability 12(3):1060

    Article  Google Scholar 

  • JalilFeridun AM (2011) The Impact of growth, energy and financial development on the environment in China: a cointegration analysis. Energy Econ 33(2):284–291

    Article  Google Scholar 

  • Jiang L, Zhou HF, Bai L, Zhou P (2018) Does foreign direct investment drive environmental degradation in China? An empirical study based on air quality index from a spatial perspective. J Cleaner Prod 176:864–872

    Article  Google Scholar 

  • Johansen S (1995) Likelihood-based inference in cointegrated vector autoregressive models. Oxford University Press on Demand

    Book  Google Scholar 

  • Kapetanios G, PesaranYamagata MHT (2011) Panels with non-stationary multifactor error structures. J Econ 160(2):326–348

    Article  Google Scholar 

  • Kasman A, Duman YS (2015) CO2emissions, economic growth, energy consumption, trade and urbanization in new E.U. member and candidate countries: a panel data analysis. Econ Model 44:97–103

    Article  Google Scholar 

  • Katsuya I (2017) CO2 emissions, renewable and non-renewable energy consumption, and economic growth: evidence from panel data for developing countries. International Economics 151:1–6

    Article  Google Scholar 

  • Magnani E, Tubb A. (2007). The Link Between Economic Growth and Environmental Quality: Does Population Ageing Matter? School of Economics, Discussion Paper: 2007/12.

  • Managi S (2004) Trade liberalization and the environment: carbon dioxide for 1960–1999. Economics Bulletin 17(1):1–5

    Google Scholar 

  • Martinez-Zarzoso I, Bengochea-Morancho A, Morales-Lage R (2006). The Impact of Population on CO2 Emissions: Evidence from European Countries. Fondazione ENI Enrico Mattei, Nota di lavaro N.98.

  • MEDECC (2019). Les risques liés aux changements climatiques et environnementaux dans la région Méditerranée.

  • Mendes CM, Junior SP (2012) Deforestation, economic growth and corruption: a non-parametric analysis on the case of Amazon Forest. Appl Econ Lett 19(13):1285–1291

    Article  Google Scholar 

  • Mongo M, Belaid F, 2021a. What are the short-and long-term impacts of eco-innovation on levels of CO2 emissions? Science for Environment Policy, p 565.

  • Mongo M, Laforest V, Belaïd F, Tanguy A (2021b) Assessment of the Impact of the Circular Economy on CO2 Emissions in Europe. J Innov Econ Manag 39(3):I107–I129

    Google Scholar 

  • Moon H, Perron B (2004) Testing for a unit Root in panels with dynamic factors. Of Economet 122(1):8–126

    Google Scholar 

  • NasirRehman MFU (2011) Environmental Kuznets curve for carbon emissions in pakistan: an empirical investigation. Energy Policy 39(3):1857–1864

    Article  Google Scholar 

  • O’Connell P (1998) The overvaluation of purchasing power parity. J Int Econ 44(1):1–19

    Article  Google Scholar 

  • OmriBelaid AF (2020) Does renewable energy modulate the negative effect of environmental issues on the socio-economic welfare? J Environ Manage 278:111483

    Google Scholar 

  • Oswald, A, Stern N (2019), Why does the economics of climate change matter so much, and why has the engagement of economists been so weak? Royal Economic Society Newsletter, October.

  • Pao HT, Tsai CM (2011) Multivariate Granger causality between CO2 emissions. Energy 36(1):685–693

    Article  Google Scholar 

  • Pedroni P (1999) Critical values for cointegration tests in heterogeneous panels with multiple regressors Oxf Bull. Rev Econ Stat 61:653–670

    Google Scholar 

  • Pesaran MH, Ullah A, Yamagata T (2008) A bias-adjusted LM test of error cross-section independence. Economics Journal 11(1):105–127

    Google Scholar 

  • Pesaran MH (2006) Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica 74(4):967–1012

    Article  Google Scholar 

  • Pesaran MH, Shin Y, SMITH R (1999) Pooled mean group estimation of dynamic heterogeneous panels. J Am Stat Assoc 94(446):621–634.

    Article  Google Scholar 

  • Pesaran MH (2007) A simple panel unit root test in the presence of cross-section dependence. J Appl Econ 22(2):265–312

    Article  Google Scholar 

  • Pesaran MH. (2004). General diagnostic tests for cross section dependence in panels, Cambridge Working Papers in Economics 0435, Faculty of Economics, University of Cambridge.

  • Pesaran MH, Smith R (1995) Estimating long-run relationships from dynamic heterogeneous panels. J Econ 68(1):79–113.

    Article  Google Scholar 

  • Sebri M (2009). La zone méditerranéenne face à la pollution de l’air : une investigation économétrique. Papier présenté au Quatrième Colloque International de l’Institut Supérieur de Gestion de Sousse.

  • Seyfettin E, Durmus ÇY, Ayfer G (2019) Investigation of causality analysis between economic growth and CO2 emissions: the case of BRICS – T Countries. Int J Energy Econ Policy 9(6):430–438.

    Article  Google Scholar 

  • Shahbaz M, SolarinSbia SAR, Bibi S (2015b) Does energy intensity contribute to CO2emissions? A trivariate analysis in selected African countries. Ecol Indic 50:215–224

    Article  Google Scholar 

  • Shahbaz M, Farhani S, Ozturk I (2015a) Do coal consumption and industrial development increase environmental degradation in China and India? Sci PolluT Res 22(5):3895–3907

    Article  Google Scholar 

  • Shao Q, Wang X, Zhou Q, Balogh L (2019) Pollution haven hypothesis revisited: a comparison of the BRICS and MINT countries based on VECM approach. J Cleaner Prod 227:724–738

    Article  Google Scholar 

  • Smith V, Leybourne S, Kim H (2004) More powerful panel unit Root tests with an Application to the Mean Reversion in Real Exchange Rates. J Appl Economet 19:147–170

    Article  Google Scholar 

  • Solarin SA, Al-Mulali U, Musah I, Ozturk I (2017) Investigating the pollution haven hypothesis in Ghana: an empirical investigation. Energy 124:706–719

    Article  Google Scholar 

  • STERN N. (2006), The Stern Review Report: The Economics of Climate Change. London, HM Treasury, 30 October, 603 p.

  • Sun C, Zhang F, Xu M (2017) Investigation of pollution haven hypothesis for China: an ARDL approach with breakpoint unit root tests. J Clean Prod 161:153–164

    Article  Google Scholar 

  • Tiba S, Belaid F (2020) The pollution concern in the era of globalization: do the contribution of foreign direct investment and trade openness matter? Energy Economics 92:104966

    Article  Google Scholar 

  • Tiba S, Belaid F (2021) Modeling the nexus between sustainable development and renewable energy: the African perspectives. J Econ Surv 35(1):307–329

    Article  Google Scholar 

  • Westerlund J (2008) Panel cointegration tests of the Fisher effect. J Appl Econ 23(2):193–233

    Article  Google Scholar 

Download references


'Not applicable' for that section.

Author information

Authors and Affiliations



The names of authors listed in metadata (1st page of the built PDF) and manuscript text are in the same order and are completely provided.

Corresponding author

Correspondence to Christophe Rault.

Ethics declarations

Conflicts of interest

We attest that in submitting our paper for your journal for publication, there is no potential conflict of Interest including financial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately influence, or be perceived to influence, their work.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

We are grateful to the editor, and two anonymous referees for very useful suggestions on a previous version of the paper.



Appendix 1 See.

Appendix 2 See.

Appendix 3 See.

Appendix 4 Methods for estimating parameters of the econometric specifications

We briefly review here the 4 methods implemented to estimate the short-run and long-run parameters of our 4 econometric specifications.

2.1 Appendix 4.1 The correlated common effects mean group estimator (CCEMG)

It was proposed by Pesaran (2006)) and Kapetanios et al. (2011). It is based on the Mean Group (M.G.) estimator and accounts for the dependence in cross-sectional units and parameters heterogeneity. The model is static and does not include the lagged dependent variable as an explanatory variable, as it is the case in Chudik and Pesaran (2013). Pesaran (2006) approximates common correlated effects by the cross-sectional means of the dependent variable and the explanatory ones. The model is specified as:

$${y}_{it}={\alpha }_{i}+{\beta }_{i}{x}_{it}+{\varphi }_{i}{f}_{t}+{\varepsilon }_{it}$$

, where yit and xit are, respectively, the dependent variable and the vector of explanatory variables. \({\alpha }_{1i}\) is the coefficient for country specificities, ft denotes a common unobservable factor, and \(\varepsilon_{it}\) is the error term.

Equation (3) is augmented by the cross-sectional means of the dependent and independent variables and can be rewritten as:

$$y_{it} = \alpha_{1i} + \beta_{i} x_{it} + \delta_{i} \overline{y}_{it} + \theta_{i} \overline{x}_{it} + \varphi_{i} f_{t} + \varepsilon_{it}$$

with \(\overline{y}_{t} = \frac{1}{N}\mathop \sum \limits_{i = 1}^{N} y_{it}\) and \(\overline{x}_{t} = \frac{1}{N}\mathop \sum \limits_{i = 1}^{N} x_{it}\)

The CCEMG estimator is given by:

$${\widehat{b}}_{CCEMG}=\frac{1}{N}{\sum }_{i=1}^{N}{\widehat{b}}_{it}$$

It is convergent under certain assumptions (e.g. Vogel, 2013).

2.2 Appendix 4.2 The Mean Group estimator (MG)

It was introduced by Pesaran and Smith (1995) and assume independence in cross-sectional units, which implies the absence of common correlated (unobservable) factors between countries. The M.G. estimator is calculated in two steps: first, a separate estimate is made for each country. Then, the average of the estimated coefficients is computed. This estimator can be obtained in a static model, or in a dynamic one. Under certain regularity conditions it is convergent (e.g, Pesaran and Smith 1995, for more details).

2.3 Appendix 4.3 The augmented mean group estimator (AMG)

It was proposed by Eberhardt and Bond (2009), and by Eberhardt and Eal (2010). Like the CCEMG estimator, it accounts for possible cross-sectional dependence among countries and parameters heterogeneity. The difference between them lies in the method for approximating common unobservable factors. The CCEMG estimator is based on linear combinations of the observed common effects cross-sectional means and the dependent and explanatory variables. The AMG estimator is obtained from the equation below, which is augmented by time dummies:

$$\Delta {y}_{it}={\alpha }_{1i}+{\beta }_{i}\Delta {x}_{i,t}+{\varphi }_{i}{f}_{t}+{\sum }_{t=2}^{T}{\tau }_{t}DUMM{Y}_{t}+{\varepsilon }_{it}$$

The AMG estimator is computed in the same way as CCEMG, i.e.

$$AMG={N}^{-1}{\sum }_{i=1}^{N}{\widetilde{\beta }}_{i}$$

, where \({\widetilde{\beta }}_{i}\) refers to the OLS estimators of the coefficients \({\widetilde{\beta }}_{i}\) of Eq. (5).

2.4 Appendix 4.4. The Pooled Mean Group estimator (PMG)

It was introduced by Pesaran et al. (1999). It relies on the estimation of an ARDL model which allows short-run and long-run parameters to be estimated. Besides, it accounts for individual heterogeneity, and permits to deal with economic series integrated of different orders (I (0) and I (1)).

Specifically, the PMG estimator is obtained from the following ARDL model (p, q):

$${y}_{i,t}={\sum }_{j=1}^{p}{\lambda }_{i,j}{y}_{i,t-j}+{\sum }_{j=0}^{q}{{\tau }^{^{\prime}}}_{i,j}{x}_{i,t-j}+{\mu }_{i}+{u}_{it}$$

, with \({u}_{it}={\rho }^{^{\prime}}{f}_{it}+{\varepsilon }_{it}\)

$${y}_{i,t}={\sum }_{j=1}^{p}{\lambda }_{i,j}{y}_{i,t-j}+{\sum }_{j=0}^{q}{{\tau }^{^{\prime}}}_{i,j}{x}_{i,t-j}+{\mu }_{i}+{u}_{it}$$

, where \({x}_{i,t}\) is a (k*1) matrix of explanatory variables, \({\mu }_{i}\) are individual fixed effects, \({f}_{it} is\) a vector of common unobservable factors, \({\varepsilon }_{it}\) is a white noise uncorrelated with regressors and common unobservable factors.

Equation 6 is equivalent to:

$$\Delta {y}_{i,t}=\phi {y}_{i,t-1}+{\beta }^{^{\prime}}{x}_{i,t}{\sum }_{j=1}^{p-1}{{\lambda }^{*}}_{i,j}\Delta {y}_{i,t-j}+{\sum }_{j=0}^{q-1}{{{\delta }^{*}}^{^{\prime}}}_{i,j}\Delta {x}_{i,t-j}+{\mu }_{i}+{u}_{it}$$

where \({\phi }_{i}=-(1-{\sum }_{j=1}^{p}{\lambda }_{ij})\), \({\beta }_{i}={\sum }_{j=0}^{q}{\delta }_{ij}\),\({{\lambda }^{*}}_{ij}=-{\sum }_{m=j+1}^{p}{\lambda }_{im}\)

and \({{\delta }^{*}}_{ij}=-{\sum }_{m=j+1}^{q}{\delta }_{im}\)

Equation 7 is stable if \(\phi\) < 0.

The cointegrating relationship is then given by.

\(y_{i,t} + \left( {\frac{{\beta_{i}^{^{\prime}} }}{{\phi_{i} }}} \right)x_{i,t} + \eta_{i,t}\), since long-run coefficients are assumed to be homogeneous across countries, i.e.

$$\theta ={\theta }_{i}=\left(\frac{{\beta }_{i}^{^{\prime}}}{{\phi }_{i}}\right)$$

, where \({\phi }_{i}\) refers to the speed of adjustment.

The existence of a long-run relationship is rejected if \(\theta =0\).

The PMG estimator assumes equal long-run coefficients for all countries, but short-run parameters may differ from a country to another.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ben Abdeljelil, M., Rault, C. & Belaïd, F. Economic growth and pollutant emissions: new panel evidence from the union for the Mediterranean countries. Econ Change Restruct 56, 1537–1566 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


JEL Classification