Do emissions and income have a common trend? A country-specific, time-series, global analysis, 1970–2008 (original) (raw)

Abstract

This paper analyzes the relation between income and emissions in the period 1970–2008, for all world countries. We consider time-series of CO2, SO2 and GWP100, and use Vector Autoregressive models that allow for nonstationarity and cointegration. At 5 % significance level, income and emissions are found to be driven by unrelated random walks with drift (respectively by a common random walk with drift) in about 70 % (respectively 25 %) of cases; in the remaining cases the variables are trend-stationary. Tests of Granger-causality show evidence of both directions of causality. For the case of unrelated stochastic trends, we almost never find income driving emissions, as predicted by a consumption-function interpretation. These causality results and the absence of a common trend challenge the main implications of the Environmental Kuznets Curve, namely that the dominant direction of causality should be from income to emissions, and that for increasing levels of income, emissions should tend to decrease.

Access this article

Log in via an institution

Subscribe and save

Buy Now

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

Instant access to the full article PDF.

Fig. 1

The alternative text for this image may have been generated using AI.

Fig. 2

The alternative text for this image may have been generated using AI.

Fig. 3

The alternative text for this image may have been generated using AI.

Fig. 4

The alternative text for this image may have been generated using AI.

Similar content being viewed by others

Notes

  1. Many non monotonic forms of this relation have been considered in the EKC literature; the inverted U shape is a representative of this class.
  2. See also Wagner (2008) and Stern (2010) on critical issues in the standard econometric practise on the EKC.
  3. ARMA models are widely used in modeling and forecasting time series of emissions, see e.g. Kim and Kumar (2005) and Gocheva-Ilieva et al. (2014).
  4. I(1) processes are often called ‘stochastic trends’; one such process is the random walk.
  5. Verbeke and De Clercq (2006) remarked that, “most empirical papers do not report whether the series are integrated or not. Basically this means that there is no way to tell if the reported results are due to the EKC or are spurious.”
  6. Several investigators, see Stern (2004, 2010), have favored a monotonic emissions-income relation instead of a non-monotonic one. Several studies, see the list of references on page 2181 in Stern (2010), do not find support for a non-monotonic emissions-income relation as predicted by the EKC hypothesis.
  7. The points in the right graph in Fig. 1 were obtained by selecting \(y\) values in the interval \([0,5]\) using a uniform distribution, and adding \(N(0,1)\) noise to the corresponding slopes.
  8. In fact when \(k=1\), the test of \(r=0\) is really a test that \(\Delta x_t \) is a white-noise process with no short-run dynamics.
  9. See the following Sect. 5.3 for a definition of the ECM.
  10. Given that the GDP series for the US is common to the system with GWP and SO2, the selected ranks of 1 and 2 are not consistent, because GDP either contains an I(1) component or not. This is a limitation of the inferential procedure. In the following we do not try to resolve these conflicts, but proceed in the analysis taking the selected cointegration rank as given.
  11. As noted in Lütkepohl (2005) and in Baek et al. (2009), \(\beta _{2}\) is not an elasticity due to the presence of dynamics in the system, see Johansen (2005) for an interpretation of identified cointegrating coefficients in terms of a counterfactual experiment involving the long-run of the system. Here we simply interpret \(\beta _{2}\) as the slope of the emissions-income relation.
  12. If the corresponding hypotheses implied both \(\beta _{1}=\beta _{2}=0\), we relaxed the restriction on the coefficient of \(x_{1t}\) or \(x_{2t}\) with highest \(p\)-value in the single-coefficient significance tests.
  13. The analysis for longer prediction horizons is more articulate, see Dufour et al. (2006), Omtzigt and Paruolo (2005) and Fanelli and Paruolo (2010), and goes beyond the scope of the present paper.
  14. \(P(t<t_{\zeta _i})\) in Table 8 indicates the \(p\)-value for a one-sided test against the alternative \(\zeta _i < 0\), while \(P(|t|>t_{\zeta _i})\) is the \(p\)-value against a two-sided alternative.

References

Download references

Acknowledgments

B. Murphy: Financial support from the EC-JRC is gratefully acknowledged.

Author information

Authors and Affiliations

  1. European Commission, Joint Research Centre, Via E. Fermi 2749, 21027, Ispra, VA, Italy
    Paolo Paruolo & Greet Janssens-Maenhout
  2. The Carbon Trust, Dorset House, 27-45 Stamford Street, London, SE1 9NT, UK
    Ben Murphy

Authors

  1. Paolo Paruolo
  2. Ben Murphy
  3. Greet Janssens-Maenhout

Corresponding author

Correspondence toPaolo Paruolo.

Electronic supplementary material

Rights and permissions

About this article

Cite this article

Paruolo, P., Murphy, B. & Janssens-Maenhout, G. Do emissions and income have a common trend? A country-specific, time-series, global analysis, 1970–2008.Stoch Environ Res Risk Assess 29, 93–107 (2015). https://doi.org/10.1007/s00477-014-0929-9

Download citation

Keywords