Skip to main content

How Do the Regions of the European Union Perform in the Sustainable Management of Municipal Waste? An Analysis of the Performance and Convergence of European Union Regions

  • Chapter
  • First Online:
Trash or Treasure

Abstract

This paper emphasizes the importance of Sustainable Waste Management for achieving a Circular Economy and the need to monitor the EU and its Member States’ transformation path. Despite positive findings in waste treatment studies, there is a lack of efficiency evaluation at the regional level. The research aims to benchmark municipal solid waste management in 167 NUTS-2 regions across 20 European member states from 2008 to 2013, assessing convergence between regions using four Data Envelopment Analysis (DEA) models, including the Benefit-of-Doubt (Bod) approach. The results reveal a reduction of the Coefficient of Variation equal to 3.6% per year. Although this convergence, differences in municipal solid waste management performance existed between 20 EU Members States NUTS-2 regions, even within the same country. The findings indicate a yearly reduction in the Coefficient of Variation, signalling progressive convergence in waste management efficiency. However, significant variations persist among the analysed regions, with better performance in Central and Northern regions and poorer performance in Eastern and Southern regions.

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

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 119.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD 159.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    The EU ‘waste hierarchy’, included in the EU2008/98/EC was inspired by the Dutch Lansnik’s ladder—1994 (Chioatto et al., 2023a).

  2. 2.

    Landfilling and incineration are equable disposal methods, while recycling is part of a circular process. However, this article focuses on the WM process rather than circular economy policies and achievements. Following the hierarchy, incineration is seen as an improvement over landfilling (as it include recovery), despite being akin to a “landfill in the sky”.

  3. 3.

    See Annex.

  4. 4.

    See Annex—A.1 for further details.

  5. 5.

    For further details about the methodologies adopted see Annex—A.2.

  6. 6.

    A resilient methodology has been employed specifically for the LP (3)–LP (4) models to fortify the outcomes. This is done to prevent the non-parametric methodology from being influenced by the supplementary assumptions, aiming to mitigate the impact of outliers and enhance the model's robustness. This is achieved through a Monte Carlo simulation involving B iterations (b = 1,…, B). In each iteration, a subset of m regions is randomly selected, and the LP is applied to calculate the CIk. Ultimately, for each NUTS 2-region k, B CIs are constructed. The final CI score is determined by averaging the values of the B CIs calculated.

  7. 7.

    The models have been used for their interconnectedness and their ability to complement each other's outcomes. It’s stressed that while β-convergence is necessary, it alone is insufficient for σ-convergence. There is a possibility that countries may display β-convergence without showing σ-convergence, as highlighted by Vojinović et al. (2009). Moreover, Lichtenberg (1994) has demonstrated that the two convergence definitions are not equivalent. The verification of σ-convergence, measured by the dispersion ratio, depends on not only confirming β-convergence but also assessing the coefficient of determination in the growth rate model.

  8. 8.

    See Annex A.3 for further information.

  9. 9.

    When conducting β-convergence testing, the natural logarithm of output growth across t-periods in k-regions is regressed against the natural logarithm of initial output values at t. This regression accounts for regional differences as it is based on cross-sectional data. The speed of convergence \((\lambda = \frac{ - \ln (1 + \beta_1 )}{T})\) is also calculated. The half-life, indicating the time needed to reduce regional disparities by half, is computed as 1/2 \(H = \frac{\ln (2)}{\lambda }\). Significance tests, specifically F-tests for the regression coefficients, are performed for the OLS model, gauging the significance of R2.

  10. 10.

    The determination of the direction vector in this analysis is subjective, guided by the aspiration that regions align with EU targets, particularly those outlined in Directive 2008/98/EC, which advocates for a 65% reduction in total Municipal Solid Waste (MSW). Directive 1999/31/EC sets a reference target of 10% landfill by 2035, while Directive 2000/76/EC lacks specific targets for MSW. The incorporation of the concept of Energy Efficiency (EE) in the 2008/98/EC directive, promoting incineration plants applying Waste-to-Energy (WtE), has led to a decision favoring energy recovery treatment. Consequently, recognizing that incineration, positioned in the waste hierarchy, is considered among the least preferable treatment in disposal operations, a decision was made to assess incineration at a ratio of 10% and energy recovery at 15%.

  11. 11.

    This aspect of the model has room for improvement but, at the moment, is the best model available according to the literature analysis.

  12. 12.

    Defined in the Green Paper on greenhouse gas emissions trading within the European Union COM (2000)/87.

  13. 13.

    See Appendix for the detailed results.

References

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Marco Ciro Liscio .

Editor information

Editors and Affiliations

Appendices

6 Annex

1.1 6.1 A.1

The Multiple Imputation approach, specifically the Amelia method in R, has been applied to all countries except Germany, ensuring robust predictions, reduced bias, and increased efficiency. The approach involves Expectation–Maximization with Bootstrapping (EMB) to address the absence of data for cross-sectional and time-series variables. This method assumes that the variables in the dataset are multivariate normally distributed, enabling the use of means and covariances for data summarization. Missing data is considered missing at random (MAR) and is treated using the Expectation Maximization (EM) algorithm with bootstrapped samples, drawing imputed values from each set of bootstrapped parameters. The imputations are combined using the univariate mean of predicted variables (Honaker et. al., 2010).

Eurostat’s original dataset had 35.7% missing data, leading to the exclusion of countries with over 45% missing observations. Data for Denmark, Spain, and Italy were integrated using the Multiple Imputation approach to predict missing values. The study considered seven EU Member States (Denmark, France, Germany, Italy, Poland, Romania, and Spain) and applied a consistent setting for imputations, defining cross-section patterns over time for all variables. The imputation model assumed a Multivariate Normal Distribution (MVN) and missing data at random.

Specific data treatments were conducted for each country. For Denmark, data were collected from the Danish National Statistical Institute and reclassified before applying the EMB algorithm. For France, data treatment was conducted twice, first for the period 2008–2011 and then for the entire timeframe using the previous imputation results. Germany’s missing data were completed using data from the German National Statistical Institute, requiring reclassification and rate calculations. Italy’s imputation utilized an integrated dataset with additional data from ISPRA for the period 2011–2013. Poland’s missing data for energy recovery were directly imputed using the EMB algorithm. Romania employed a heuristic approach for imputation due to unsatisfactory diagnostic results, involving increased priors, a polynomial relation, and adjusted confidence bounds. Spain’s missing data were integrated using the Spanish National Statistical Institute dataset, and residual values were addressed based on mean rates weighted on the coefficient of variation.

1.2 6.2 A.2

BoD methodologies

The model chosen for efficiency analysis in this study focuses on both inputs and outputs, with the specific orientation based on factors’ control. The selected approach is the DEA method, chosen for its capability to consider multiple outputs for CI determination without requiring additional information on inputs. The DEA method allows for an endogenous determination of variable weights used to measure efficiency, estimating the best result composed of best practices. Efficiency measures for DMUs are then assessed in reference to this best result using linear programming techniques (Fried et al., 2008).

CIs are recognized and increasingly applied as powerful tools to measure policies and programs, facilitating result communication (Nardo et al., 2005). The approach in this study constructs a CI incorporating various waste treatments specified in EU directives. Similar methodologies have been employed in WM by other authors, such as Rogge et al. (2017), providing valuable guidance to policymakers and stakeholders. In this study, the CI is built on five sub-indicators representing different waste treatments, and waste generation values are considered as exogenous variables.

The BoD model is considered suitable for this analysis, as it can be represented as a DEA-model in a pure output setting, focusing solely on achievements without considering the input side. Technically, the BoD-model is a DEA-model formulation treating all performance indicators as outputs and a ‘dummy input’ equal to one for all observations (Lovell et al., 1995). This dummy input serves as a ‘helmsman’ for different performance indicators (Cherchye et al., 2007).

The BoD model aims to derive weights to maximize a region’s CI-value subject to an upper bound. It compares regions relative to each other, identifying performance criteria of relative strength and weakness. The weights are DMU-specific, allowing for different sets of weights leading to the selection of different DMUs. The benchmark is DMU-dependent, and simple indicators must be comparable with the same unit of measurement (OECD, 2008a, 2008b).

Given the uncertainty in identifying the most accurate representation of reality, four different configurations of BoD models have been applied. This approach acknowledges the range of evaluations obtained by different computations and aims to verify which model best fits the available data and research objectives. The final analysis considers each model’s results operating on the outputs to achieve a comprehensive understanding.

Unrestricted BoD-model

In this model, two constraints are defined: the upper bound for each region, set as the weighted sum of that region’s performance indicator values, equal to 1, and the non-negativity of the weights assigned to each observed performance. If a region (k) receives a CI value of 1, it signifies that the region’s result becomes its own benchmark. Conversely, if region K’s result is outperformed by another region (x), then x region becomes the benchmark. The set of regions that outperform can be considered best-practice (Cherchye et al., 2007). Therefore, the region with the highest CI, indicating the best consideration under the BoD model, is deemed the best-performing one.

This approach serves two purposes: firstly, it overcomes the challenge of determining the optimal weight scheme due to the sample’s characteristic variety, and secondly, it ensures the assignment of stronger performance scores. It is important to note that any other weighting scheme would not yield the same aggregate performance score. Granting regions the benefit of the doubt on the applied importance of weights is a key objective of the research.

Constrained BoD-model

Various simple indicators (representing single types of treatment) often register null weights, but in certain regions, the data reveal that the best-performing simple indicators, in comparison to other regions, may have weights exceeding 1. Recognizing the need for greater result reliability, (Rogge et al., 2012) emphasized the significance of incorporating minor constraints.

The flexibility in determining weights endogenously has been affirmed, subject to specific conditions: imposing an importance weight of at least 5% to prevent null weights for some waste treatments and avoiding overrating the best results in the simple indicators. This condition is crucial as, without it, the model might excessively emphasize the superior performance of certain waste treatments in a region, potentially neglecting the poor performance of others. Consequently, restrictions are imposed, including an upper bound of 80% for the importance weight and ensuring the sum of importance weights for each region is set to 1, thereby achieving an equitable distribution of weights across all regions.

Bad BoD-model

In their work, Lavigne et al. (2019) introduced a comprehensive approach to MSW assessment, considering both desirable and undesirable factors in the evaluation process. Aligned with the waste hierarchy, the importance of weights was systematically ordered. Notably, unlike previous models, this approach incorporates landfill treatment as an integral part of the model without penalizing its percentage. Drawing from the WFD 2008/98/EC, a priority level was established among the simple indicators, integrating the concept of undesirable indicators into the LP model.

The imposed constraint reflects a five-stage priority inspired by the waste management hierarchy, indicating a clear preference order. Furthermore, the evaluation is conducted in the presence of undesirable simple indicators. Consequently, a set of restrictions is introduced to enforce a rank ordering of the simple indicators, aligning with the waste management hierarchy criteria. The objective function used to calculate the CI incorporates the consideration of both desirable and undesirable factors, providing a more comprehensive and nuanced assessment of MSW management performance.

Directional Distance BoD-model

In this model, the evaluation is centered on the output distance to the best performers in g-units, offering a distinctive approach to considering the preference structure among simple indicators. As Fusco (2014) notes, the BoD directional approach allows for the incorporation of a preference structure among simple indicators, steering clear of compensability through the application of “directional” penalties.

Nevertheless, implementing this approach necessitates additional assumptions. The directional distance models concurrently consider both desirable and undesirable performance indicators. Specifically, when seeking the regional CI, the objective is to contract undesirable indicators while expanding desirable outputs, thereby achieving the most favorable evaluation. The directional vector g signifies the precise direction in which improvements may be pursued. This nuanced approach provides a more refined and context-specific assessment of the waste management system’s performance.

1.3 6.3 A.3

The β-convergence represents a reliable means to understand whether EU regions are decreasing the disparities in MSW management. Initially, this approach was linked to the concept of neoclassical economic growth model states, an evaluation that explores the convergence of all regions to the same regional output level (steady state). It can detect if the poorer performers are moving with a higher growth rate compared to the good performers that le.d to a convergence between the regions.

In our case, \(Y_{k,t}\) represents the growth rate of the waste treatment under consideration for the region k in the period t; \(\beta_0 { }\) and \(\beta_1\) are the parameters to be estimated; \(X_{k,t}\) represents the initial level in the period t of the waste treatment for the region k; \(U_{k,t}\) represents the random residuals. \(Y_{k,t}\) is possible to express as \(Y_{k,t} = \frac{\Delta X_k }{{X_{k,o} }}\), the factor \(\Delta X_k\) shows the total variation in the k region between the period 0 and t meanwhile the factor \(X_{k,o}\) shows the initial level in the period 0 for the region k. The \({ }\beta_1\) parameters of interest represent the slope of the model.

In this model, the verification of the convergence hypothesis depends on the level of \(X_{k,o}\) being the point of inflection taking into consideration the function \(\frac{\partial Y_i }{{\partial X_{k,o} }} = 0\). The original convergence formula, presented by Barro apply an estimation through a nonlinear least squares (NLS) approach but in our case, it is used a linear transformation to allow the ordinary least squares (OLS) estimation (Dapena et al., 2016) thus the model can be written as

$$\ln Y_{k,t} = \ln \beta_0 + \beta_1 \ln X_{k,t} + \ln U_{k,t}$$

The model is taken into consideration as linear, but not non-linear in terms of variables in order to represent that the treatments with higher values should be more difficult to increase. We applied an estimation through the log–log relationship because is assumed the absence of negative trends over time.

The analysis is done taking into consideration all the periods included in the time frame under the study.

Appendices

See Tables A.1, A.2, A.3, and A.4.

Table A.1 Regional ranking over the period 2008–2013 using BoD model
Table A.2 Regional ranking over the period 2008–2013 using CBoD model
Table A.3 Regional ranking over the period 2008–2013 using BBoD model
Table A.4 Regional ranking over the period 2008–2013 using DBoD model

Rights and permissions

Reprints and permissions

Copyright information

© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this chapter

Check for updates. Verify currency and authenticity via CrossMark

Cite this chapter

Chioatto, E., Fedele, A., Liscio, M.C., Sospiro, P. (2024). How Do the Regions of the European Union Perform in the Sustainable Management of Municipal Waste? An Analysis of the Performance and Convergence of European Union Regions. In: Singh, P., Borthakur, A. (eds) Trash or Treasure . Springer, Cham. https://doi.org/10.1007/978-3-031-55131-4_12

Download citation

Publish with us

Policies and ethics