Literature Review

The body of literature on municipal waste management performance is extensive. Simões and Marques [10] conducted a comprehensive review of 107 studies published from 1965 to 2011, offering a thorough overview of the topic. In the subsequent decade, numerous studies have continued to explore various aspects of waste management performance.

The majority of these studies evaluate MSWM performance at the national or regional level, particularly within EU member countries or EU NUTS2 regions. These studies frequently use Eurostat or OECD datasets to compare geopolitical entities using partial or composite indicators. Examples of such studies include [11-19]. In addition to intra-EU comparisons, some studies extend the analysis to international contexts, comparing MSWM performance between the EU and other major economies such as China and the USA. These comparative studies highlight variations in waste management policies, efficiency levels, and technological approaches, reflecting differences in regulatory frameworks, economic conditions, and sustainability priorities. Notable examples of such research include [20-24].

While these studies provide valuable insights, the number of research studies focusing on MSWM performance at the municipal level is lower, primarily due to data limitations. Most studies at this level are based on samples of municipalities. Significant examples of such studies include [25-36]. Studies using data from entire populations of municipalities are rare, with [37] being a notable exception.

Published studies focused on MSWM efficiency employ a relatively wide range of methods for estimating efficiency measures. While waste treatment-specific indicators are frequently used, their ranking ambiguity has led to the adoption of more sophisticated methods that enable the estimation of composite indicators.

Multiple Criteria Decision Making (MCDM) methods are widely used across various fields, including waste management, due to their flexibility in handling multiple, often conflicting criteria. MCDM methods like AHP (Analytic Hierarchy Process), TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), and PROMETHEE (Preference Ranking Organization Method for Enrichment Evaluations) are commonly used to evaluate and rank municipal waste management practices, technologies, and policies. The advantage of MCDM methods lies in their ability to incorporate expert judgment and assign importance weights to waste treatment methods, reflecting the waste hierarchy. However, a primary limitation is that they typically produce only a ranking of the analyzed units, without the ability to express the performance improvement potential of the analyzed units. Valuable insights into the use of MCDM methods in waste management are provided in review works [38-41]. Additionally, a recent study on MSWM performance in Slovakia is available in [42].

A mathematical programming approach to estimate MSWM performance composite indicators is one of the most prevalent in recent studies. Several standard nonparametric methods are included in this approach. Data Envelopment Analysis (DEA) is frequently used for its focus on efficiency and benchmarking. DEA and its derivatives have become important efficiency estimators both in cross-sectional and intertemporal analyses. They enable work with both desirable and undesirable variables, representing desirable and undesirable waste treatment methods, distinguishing weak and strong disposability, and controllable and non-controllable variables when dealing with waste inputs and/or outputs according to their managerial controllability. Notable studies of this nature are studies [17, 18, 26, 29, 43-50].

When alternative technologies are used in MSWM, the DEA metafrontier methodology is a useful tool to estimate their net impact. Notable applications of this methodology include [19, 35, 47, 51].

Another valuable but less commonly used method is the Benefit-of-Doubt (BoD) method, derived from the DEA. It corresponds to the DEA model only with outputs, without considering the input side. In assessing MSWM, it was used in [19, 52].

Directional Distance Functions (DDF) is a useful methodology when dealing with undesirable outputs under alternative assumptions of their disposability. DDF can be estimated using DEA, and this approach was applied in [19, 26, 27, 35, 49, 51].

Free Disposal Hull (FDH) is less restrictive compared to DEA because it does not require a convexity assumption, which makes it suitable for complex processes. By using actual observed data points to construct the frontier, FDH provides realistic benchmarks based on best practices. In MSWM, FDH was employed in [18, 53].

All the aforementioned methods are non-parametric. However, a substantial body of research on MSWM efficiency also utilizes parametric approaches, primarily regression analysis and a key parametric alternative to DEA known as Stochastic Frontier Analysis (SFA). Regression analysis is frequently used in the second stage to examine the impact of endogenous factors on estimated efficiency measures [54]. In recent years, the two-stage DEA has gained popularity, particularly with the adoption of the double-bootstrap procedure introduced by Simar and Wilson [55], with applications in the MSWM sector found in [36, 56-59]. SFA, in contrast, integrates both stages into a single framework. It relies on a predefined functional form of the production function and is more complex to implement and interpret than DEA. Notable studies applying SFA in MSWM efficiency analysis include [60, 61].

Key discussions in modeling MSWM performance focus on incorporating the waste hierarchy through appropriate weighting of output variables [44, 62], classifying model variables as desirable or undesirable, free disposable or weak disposable [48], and distinguishing between controllable and non-controllable factors [26]. While previous studies have addressed these aspects individually, our research makes a significant contribution by integrating all these critical elements simultaneously. By applying the DEA AR-I model, as proposed by Thompson et al. [63], we provide a novel approach that embeds waste hierarchy principles directly into the efficiency assessment model via output weights. This advancement not only bridges a key gap in the existing literature but also offers a more comprehensive and robust framework for evaluating MSWM performance, setting a new standard for efficiency analysis in the field.

Methodology

In this study, we conceptualize the municipal solid waste management system as a dynamic process that involves the conversion of the total municipal solid waste (MSW) generated by the residents of a municipality into four primary waste treatment streams (Fig. 1). This transformation is a result of waste sorting activities conducted both at the household level and within the municipal infrastructure. Following the waste hierarchy and strategic goals outlined by the EU and Slovakia, we classify waste treatment processes as either desirable (all types of recycling, incineration with energy recovery) or undesirable (landfilling). Waste strategies aim for a minimum recycling rate of 60% and a maximum landfilling rate of 10%.

We posit that the performance of the municipal solid waste management system is determined by the waste sorting rates and costs, and influenced by external factors.

As a primary measure for evaluating the performance of municipal solid waste management, we utilize the composite indicator of technical efficiency, employing Data Envelopment Analysis (DEA). To assess changes in solid waste management performance between 2017 and 2021, we utilize Malmquist DEA-based metrics, including total performance (productivity).

In our DEA analysis, depicted in Fig. 1, we presume 2 inputs and 4 outputs. The first input encompasses the total municipal solid waste generated by the municipality, managed within the municipality’s waste management system. We assume that the total municipal waste generation is, from a short-term perspective, a non-controllable variable, meaning that municipality managers lack direct control over it. Management and processing of municipal waste incurs costs, which we classify as the second input, primarily consisting of expenses related to the collection and processing of residual waste. These costs are covered by fees paid by the municipality’s residents, directly influencing the level of waste sorting within the municipality.

The four DEA outputs represent the primary waste treatment methods for processing municipal solid waste (MSW): material recycling, composting and digestion (biological recycling), incineration with energy recovery, and landfilling. Each of these outputs is considered controllable, as municipal management can regulate waste distribution among these treatment streams. This controllability justifies the use of an output-oriented DEA model.

Fig. 1.

Model of the municipal solid waste management efficiency assessment.

According to the waste hierarchy outlined in the Waste Framework Directive 2008 [1], waste treatment operations are classified as either recovery (desirable) or disposal (undesirable). Following this classification, we categorize our outputs into desirable treatments - material recycling, biological recycling (composting and digestion), and incineration with energy recovery - and the undesirable treatment, landfilling.

Character of output variables led us to select a DEA model that combines desirable and undesirable outputs. In standard DEA, we assume the minimization of inputs and the maximization of outputs. In order to maximize only desirable outputs and to minimize undesirable outputs, the undesirable output of landfilling is subjected to a linear monotone decreasing transformation (Eq. (1)) as proposed by Seiford and Zhu [64].

where y̅ _r is the undesirable output vector, w is a proper translation vector that makes y_r>0. That is, we multiply each undesirable output by (-1) and find a proper translation vector w to convert all negative undesirable output’s data to positive data.

Since a reduction in undesirable output (landfilling) does not lead to a reduction in desirable outputs (recycling and incineration with energy recovery), we assume strong disposability of the undesirable output.

To evaluate the efficiency of waste management in municipalities with respect to EU and Slovakia waste targets, we introduce a novel approach that translates waste hierarchy preference information into the DEA output variable weights. For this purpose, we employ the DEA Assurance Region model, which combines the original multiplier CCR DEA model proposed by Charnes, Cooper, and Rhodes [65] with constraints for weight restrictions. In Eq. (2), we present the formal notation of the original CCR DEA model in its output-oriented form.

We assume that we have n municipalities (MUN), each using m inputs and generating s outputs. Let X∈ℜ₊^m×n (x_ij>0, i = 1, 2, …, m, j = 1, 2, …, n) and Y∈ℜ_*^s×n (y_rj≥0, r = 1, 2 ,…, s, j = 1, 2,…, n) be the matrices, consisting of positive and non-negative elements, containing the observed input and output quantities for the municipalities, respectively. We denote by x_j (the j^th column of X) the vector of inputs consumed by MUN_j, and by x_ij the quantity of input i consumed by MUN_j. A corresponding notation is used for outputs. Index “o” is associated with the municipality under observation (MUN_o).

To capture preference information on waste target rates for specific waste treatment methods, we incorporate weight restrictions for outputs into the DEA model. In absolute form, these weight restrictions are expressed in Eq. (3):

The first and second restrictions correspond to the EU 2030 target to recycle at least 60% of municipal waste. The third restriction considers the EU 2035 target to landfill a maximum of 10% of municipal waste [1]. Incineration with energy recovery is generally considered as a desirable waste treatment method, but is not restricted in the EU/Slovakia target rates.

Due to the likelihood of DEA models with absolute weight restrictions resulting in infeasible solutions [66, 67] in our analysis, we implemented weight restrictions in the form of weight relations, known as Assurance Regions Type I (AR-I), as proposed by Thompson et al. [63]. This approach is formally articulated in Eq. (4):

Here, u_h denotes multipliers that act as “numeraires” in establishing the upper and lower bounds, represented by L_r,h and U_r,h, for the multipliers associated with outputs r = 1, ..., s.

We utilize the AR-I constraints (4) rewritten to a linearized form, as expressed in Eq. (5):

The optimal values of technical efficiency scores, φ, in the CCR AR-I DEA model (Equations (1) and (4)), are greater than or equal to unity. In this study, we interpret their reciprocals, 1/φ, which vary within the range of 0 (indicating the worst-performing MUN) to 1 (indicating the best-performing MUN). When multiplied by one hundred, this measure indicates, in percentage terms, the extent to which the municipality under observation, MUN_o, is achieving MSWM performance compared to the best-performing municipalities and in relation to waste targets.

In our study we analyze MSWM performance development within the 5-year period 2017-2021. The performance change of each municipality is measured by Malmquist productivity index (MPI) introduced by Caves, Christensen, and Diewert [68], here referred as CCD, which is based on estimation of the Shephard [69] output distance functions.

We assume that for each time period t = 1, ..., T, the production technology S^t models the transformation of inputs, x^t∈ℜⁿ₊, into outputs, y^t∈ℜ^m₊, S^t = {(x^t, y^t): x^t can produce y^t}

The output distance function is defined at t as D_o^t (x^t, y^t) = (sup{φ: (x^t,φy^t)∈S^t})∈^–1

This distance function is defined as the reciprocal of the TE_o = φ, expressed in model (2).

In this study we employ output-oriented MPI developed by Färe et al. [70] as a geometric mean of two CCD-type indexes. For that purpose we estimate four distance functions with respect to two different time periods such as D_o^t (x^t, y^t), D_o^t+1 (x^t+1, y^t+1), D_o^t (x^t+1, y^t+1), and D_o^t+1 (x^t, y^t). Then Malmquist productivity index is given by the Eq. (6):

An equivalent notation of the Eq. (6) is given in the Eq. (7):

We decompose MPI into 2 components. The ratio outside the brackets measures the change in efficiency (i.e., the change in how far observed productivity is from maximum potential productivity) between years t and t + 1. Here, it is referred to as technical efficiency change (TEC). The geometric mean of the 2 ratios inside the brackets of Eq. (7) captures the shift in technology between the 2 periods. It is known as technical or technological change (TC) and refers to an improvement or worsening of the state of technology. It can be interpreted as providing evidence of innovation for the MUN considered.

A municipality has improved its productivity over time if MPI>1; whereas it has suffered deterioration in productivity if MPI<1. An MPI = 1 means that productivity is stagnating over time. For the components of the MPI, i.e., TEC and TC, the same interpretation applies.

In the second stage of our analysis, we examined the impact of several external factors on MSWM performance. We utilized nonparametric statistical techniques such as the Mann-Whitney test and Kruskal-Wallis test to test for differences, and regression analysis to examine relationships. The factors analyzed include: pricing method for waste collection, municipality size, regional MSWM programs, education level of the municipality population.

Data and Variables

In this study, we analyzed data from 2,887 municipalities out of a total of 2,890 municipalities in Slovakia (excluding three municipalities in military districts). We utilized two primary sources of data. The first source was the dataset created from annual reports on municipality waste “ŽP 6-01”, obtained upon request from the Statistical Office of the Slovak Republic (SOSR). These reports are outcomes of annual surveys conducted under the title “Annual questionnaire on municipality waste ŽP 6-01” [71], in accordance with Decree No. 291/2013 Coll. of the SOSR on the Program of State Statistical Surveys for the years 2015 to 2017 [72]. The second source comprises open data of the Slovakia 2021 Population and Housing Census, accessible on the website of the SOSR [73].

In the analysis, we utilized 7 variables from the Report ŽP 6-01 for the years 2017 to 2021, and 4 variables were extracted from the Slovakia 2021 Census. Table 1 presents all variables used in the analysis. To address missing data on the costs of 54 municipalities, we conducted a simulation, using panel data spanning the 5-year period, employing simple regression analysis. Table 2 displays the descriptive statistics of the variables used in the analysis.

Composite Indicators of MSWM Performance

First, we present the results obtained in the initial stage of our analysis, which focused on the calculation of composite indicators of MSW management performance of municipalities in Slovakia. All indicators align with strategic targets set for recycling and landfilling in the EU and Slovakia for the years 2025 and 2030. Table 3 presents descriptive statistics of the technical efficiency scores and Malmquist indices of the 2,887 municipalities under examination for the years 2017 and 2021.

The mean technical efficiency score in 2017 is relatively low (TE = 0.304), suggesting that municipalities, on average, had almost 70% potential for improvement to reach the level of efficient, or best-performing, municipalities in Slovakia. Within our sample, 30 municipalities were identified as efficient. By 2021, the mean technical efficiency score had increased to 0.473, and the number of efficient municipalities had risen by 5. Despite this improvement, municipalities on average still achieved less than 50% of their potential performance.

The Malmquist productivity index (MPI = 1.764) suggests an overall improvement in MSW management performance by an average of 76.4%, primarily driven by a technical efficiency improvement of 89.4% on average (TEC = 1.894). It is noteworthy that the majority of municipalities (78.5%) experienced total productivity improvement, while 21.4% experienced a decline. This improvement indicates that more waste was treated using desirable methods, resulting in less waste being sent to landfills.

However, the technological change index (TC = 0.925) reflects a regressive shift in the efficiency frontier, indicating that the top-performing municipalities experienced a 7.5% decline in productivity from 2017 to 2021.

The efficiency scores are significantly influenced by the unrestricted variable of incineration with energy recovery. As shown in Table 4, in 2017, municipalities with access to incineration technology had significantly higher mean technical efficiency scores (0.528) compared to those without access (0.296), as confirmed by the Mann-Whitney test (Z = 6.128, p<0.0001). Additionally, among the 30 technically efficient municipalities (TE = 1), 19 had access to incineration technology, while only 11 technically efficient municipalities were found among those without access.

In 2021, this disparity widened further. Municipalities with access to incineration technology displayed substantially higher mean technical efficiency scores (0.822) compared to those without access (0.460), with a significant Mann-Whitney test result (Z = 12.974, p = 0). Furthermore, the total number of technically efficient municipalities increased to 35, with 32 of these municipalities having access to incineration technology, whereas only 3 technically efficient municipalities were observed among those without access.

These findings suggest that access to waste incineration technology with energy recovery enhances municipal efficiency scores. However, this improvement may be influenced by the unrestricted weighting of the incineration variable in the analysis. According to the waste hierarchy outlined in the Waste Framework Directive [1], incineration without energy recovery, along with landfilling, is classified as “disposal” and subject to a maximum threshold of 10%. In contrast, incineration with energy recovery is categorized as “recovery” and is not explicitly restricted by EU regulations. Grosso et al. [74] argue that incineration with energy recovery should be positioned one step above landfilling in the waste hierarchy. Some studies, such as [38], even propose a specific target of 35% for incineration with energy recovery. However, the literature remains divided on its overall impact [75, 76]. While it often contributes to a reduction in landfill use [77], inefficient incineration technology can result in the release of harmful substances, potentially undermining circular economy principles and sustainable waste management [38].

Factors Influencing MSWM Performance

In the second-stage analysis, we explored several exogenous factors that could explain variations in waste management performance among the municipalities under examination. Based on the available data, we identified 4 key factors: (1) the pricing method for waste collection, (2) the size of the municipality, (3) the regional MSWM programs, and (4) the education level of the municipality population.