![Sample our Geography journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5937/TOC-Subject-Sample-2021-Geography.jpg"})
Abstract
We empirically test the decoupling of waste dynamics with regard to economic growth and the effectiveness of environmental and specific waste-related policies, by exploiting an integrated data set for the EU15 over 1995–2007. We find that absolute delinking for waste generation is far from being achieved in the European Union (EU) despite the fairly stringent and longstanding policy commitment, which was and is still probably too biased towards waste management and waste disposal targets, rather than towards waste prevention per se. On the other hand, policy action as well as country structural factors seems to have an impact on landfill diversion. Country heterogeneity fairly matters: the seemingly unrelated regression analyses that we adopt, including models that tackle policy endogeneity, show that EU average figures often hide high variance in decoupling performances. EU countries can be consistently grouped according to their waste sustainability performances. The results provide food for thought for setting comprehensive EU waste policy strategies jointly aimed at waste reduction and landfill diversion. This is a relevant outcome and food for thought within an EU framework that is strongly oriented towards allowing countries to decide about the implementation of EU Directives.
Notes
Then other works followed in providing theoretical explanations (Chimeli & Braden, Citation2005, Citation2009; Kelly, Citation2003).
Recycling is also an important target. We recall that as far as data availability is concerned, EUROSTAT data calculate recycling of waste as the residuals obtained by detracting landfilling and incineration from waste generation. Statistically speaking, it is not feasible and meaningful to analyse recycling under such premises.
National waste management strategies happen to be very different in Europe (Buclet & Goddart, Citation2001).
The literature includes examples of delinking studies that use SURE or a random-coefficients linear regression model (Swamy, Citation1970), in order to account for the presence of slope heterogeneity (Cole, Citation2005; List & Gallet, Citation1999).
For the Irish case, see Davies and O'Callaghan-Platt (Citation2008).
We do not use EU27 data given the current lower availability and reliability of waste data for ‘newcomers’ eastern EU countries.
EIONET is a partnership agency of the EEA and its member countries.
We can hypothesize that the backward effects of landfill policies and waste management actions on the amounts of MSW generated are difficult to be exerted. Nevertheless, since our synthetic policy index also captures the variety of waste measures implemented by a country in addition to landfill diversion actions, some effects may theoretically emerge.
More traditional fixed effects which are coherent with the nature of our panel have been tested as a preliminary exercise but not shown for brevity. It is known that SUR specifications deal with slope heterogeneity and/or cross-section correlation. When it is shown that correlation is an empirical fact, the adoption of constrained SUR improves the efficiency of fixed effect (FE). When slope homogeneity is rejected, unconstrained SUR deals with it. We discuss and tackle such issues throughout the article.
In the last step, we dropped three countries, in order to allow estimation of the SURE model with unconstrained slopes. This is because we have a relatively small T. We dropped all observations relating to the three countries with the lowest amounts of waste generation and the three countries with the lowest amounts of landfilled waste in the second phase of the analysis.
Fixed-effect model (FEM) results (tested as baseline specification, and not shown in the paper, but available upon request) show a relative delinking associated with quite a low elasticity (in the range 0.31–0.38 across all the specifications tested), with population density and the three policy variables being never significant.
Table A3 in the appendix shows an FEM estimation in the case of waste generation to highlight this point. Breusch–Pagan tests are available upon request.
The test can be performed with the Stata command xttest2, which tests the null hypothesis of cross-sectional independence, while the Pesaran test, performed with the Stata command xtcsd, tests for cross-sectional dependence following the methods described in Pesaran (Citation2004).
The countries with the lowest levels of waste production, that is, Luxembourg (330,473 kg), Finland (2,675,416 kg) and Ireland (3,389,645 kg), were dropped from the data set to allow SURE given the constraints. See Table A2 in the appendix.
See . We remark that we use the EUROSTAT data on ‘share of environmental taxation—net of energy—on GDP at constant prices’, wherein landfill taxation is an important pillar.
We refer to Tables A1–A2 in the appendix.
Denmark's performance may be influenced by the fact that construction and demolition waste was recently accounted as MSW. This shows how data commensurability is an issue in waste statistics.
Analogously, an unconstrained SURE with only consumption as the main economic driver shows a statistically significant and positive relationship between income and waste generated. These results confirm the absence of delinking for Belgium, Denmark and Finland.
The test follows an F-statistic and tests the hypothesis of slope homogeneity (under the null).
From a different but complementary perspective, we calculate the delinking indices following the OECD (Citation2002) formula, which we present in the appendix (Figure A1).
A possible explanation for the somewhat counterintuitive result for Denmark and the Netherlands might also be a consequence of forcing a quadratic specification, where we might actually observe reduction and stabilization in landfill use. The last few years of the panel nevertheless show an increase in waste that could be driven by a contingent economic growth. This shows the necessity to additionally research WKC through semi-parametric tools in the future. We thank one referee for this hint.