Introduction

The increase in the production of municipal solid waste (MSW) in recent decades has been mainly linked to the economic development of countries and has become one of the most serious problems facing modern society (Marques and Simoes 2009; Di Foggia and Beccarello 2018). In this context, several countries have developed waste management policies aimed at increasing resource efficiency and reducing the negative impact of waste on the environment and citizens’ health (Romano and Molinos-Senante 2020). Currently, the management of MSW is included as part of Goal 11 (Sustainable Cities and Communities) of the Sustainable Development Goals proposed by the United Nations (2015). As demonstrated by Schroeder et al. (2019) and Cavaleiro de Ferreira and Fuso-Nerini (2019), the ambitious Sustainable Development Goals can be accomplished by facilitating the practices of the circular economy. The European Commission defines the circular economy “as a system where the values of products, materials and resources are maintained in the economy for as long as possible, and the generation of waste is minimised” (European Commission 2015). Thus, in the circular economy context, improving the eco-efficiency of MSW management occupies a prominent role.

The concept of eco-efficiency was first defined by Schaltegger and Sturm (1989) as the ratio between the value added and the environmental impact. It entails producing more goods and services with fewer resources and a minimal environmental impact (Koskela and Vehmas 2012). In recent years, eco-efficiency has been popularised as a management philosophy that encourages companies and public services to balance their economic and environmental performances (WBCSD 2000). However, despite the usefulness of evaluating the eco-efficiency of MSW services, most previous studies benchmarking the performance of MSW service providers have focused on evaluating the cost efficiency by employing the total costs of the service as the input and tons of MSW collected as the output (e.g. Chen and Song 2008; Rogge and De Jaeger 2012; Vishwakarma et al. 2012; Yang et al. 2018). This approach considers both recyclables and unsorted wastes as outputs, despite the fact that they have notably different environmental impacts. In other words, this approach ignores the environmental performance of decision-making units (DMUs) in the provision of MSW services. In contrast, eco-efficiency assessment integrates both economic and environmental variables into the performance evaluation Gómez et al. (2018). Thus, to evaluate the eco-efficiency of MSW services, Courcelle et al. (1998) have suggested the use of recyclable waste as a desirable output and unsorted waste as an undesirable output. Nevertheless, to the best of our knowledge, only Díaz-Villavicencio et al. (2017), Sarra et al. (2017, 2019, 2020), Guerrini et al. (2017), Expósito and Velasco (2018) and Romano and Molinos-Senante (2020) have evaluated the eco-efficiency of waste services. It should be highlighted that the empirical applications conducted by these studies focused only on Italian and Spanish MSW service providers; i.e. there is no information regarding the eco-efficiency of MSW services outside of these two countries. An alternative approach was applied by Marques et al. (2012) who estimated the efficiency of a sample of Portuguese recycling companies. In this study, they applied several non-parametric models to compare the recycling and economic efficiency. In a first set of models, efficiency was estimated considering operational expenditure as input and amount of glass, paper and plastic collected as outputs. The economic efficiency assessment considered capital expenditure as input whereas outputs were the total economic revenue financial transfers from the Sociedade Ponto Verde.

To improve the management of MSW, scale-related economies are a relevant issue for waste service providers Sarra et al. (2020). In this context, some previous studies have analysed the presence of economies of scale in the provision of waste services (e.g. Bel and Fageda 2010; Bel and Mur 2009; Bartłomiej 2016; Caldas et al. 2019, Ferraresi et al. 2018; Sarra et al. 2019, 2020). With the exception of Sarra et al. (2019, 2020), these previous studies have focused on the economics of the provision of MSW services excluding the environmental performance of the service providers. Thus, the presence of economies of scale of the MSW services when eco-efficiency is employed as a metric of their performance remains as an open issue. Hence, for benchmarking the eco-efficiency of DMUs providing MSW services, investigation into whether they present economies of scale is needed.

In addition to the management carried out by the service providers, past research on benchmarking the performance of municipal waste services has found that there are some exogenous variables, such as population size, population density and tourism that have an impact on the efficiency of MSW services (e.g. Simões and Marques 2012; Carvalho et al. 2015; Caldas et al. 2019; Guerrini et al. 2017; Halkos and Petrou 2019), showing divergent results. Moreover, previous studies on eco-efficiency assessment and economies of scale analysis have mainly focused on European countries. Thus, the previous research is limited and inconclusive, warranting the need for further research with metrics that take into account environmental issues, i.e. considering eco-efficiency as a metric of performance.

The objectives of this study were threefold. The first was to evaluate the eco-efficiency of MSW services provided by a sample of municipalities, i.e. assessment of their performance considering not just economic costs and quantity of waste managed but differentiating recycled and unsorted wastes as they have different environmental impacts. The second objective was to investigate whether the eco-efficiency of MSW services is affected by economies of scale, i.e. evaluation of their scale eco-efficiency. This information is essential for local policymakers to develop policies that take advantage of inter-municipal cooperation with the aim of improving eco-efficiency in the management of MSW Silvestre et al. (2020). The third objective of this study was to explore some exogenous variables that could have an impact on the eco-efficiency of MSW services. This approach was used to benchmark the performance of a sample of Chilean municipalities providing MSW services to 62% of the Chilean population. Hence, this study is quite representative of the country. The Chilean example is very interesting because, unlike European countries, waste management policies are incipient in most Latin American countries. Hence, knowledge about the topics addressed in this study will be very useful for the design of effective policies for improving MSW management that are aligned with the principles of the circular economy and the Sustainable Development Goals.

Chile has the highest per capita Gross Domestic Product (GDP) among Latin American countries (World Bank 2019) and is part of the Organization for Economic Cooperation and Development (OECD) since 2010. Hence, municipal recycling services have been promoted in recent years, defining national goals for municipal recycling or taxes on landfills Valenzuela-Levi (2019). The main regulation that regulates waste management in Chile was implemented in 2016 and is known as Recycling and Extended Producer Responsibility Law. This law establishes that all producers or importers of “priority products” must take charge of the goods once their useful life ends. In other words, these “useless” products must return to the industries where they were manufactured, or to the warehouses where their distribution began. The Law establishes collection and recovery goals differentiated by types of waste (Ministerio de Medio Ambiente 2016). The standard defined by the Law focuses on preventing and recovering waste in all its aspects (Ministerio de Medio Ambiente 2016). However, before the adaptation of this Law, municipal recycling services emerged without a national recycling policy (a complete revision of the Chilean regulatory framework for waste management can be consult at Vásquez (2011)). In this context, in the past, recycling was introduced thanks to autonomous initiatives of the municipal authorities Valenzuela-Levi (2019).

The document is organized as follows. After the introduction, the “Methodology” section describes the methodology, where the estimation of eco-efficiency scores, analysis of economies of scale and factors affecting eco-efficiency are detailed. Then, in the “Sample description and data” section, the sample and data are described. The “Results and discussion” section presents and discusses the results. Finally, the “Conclusions” section highlights the conclusions of this paper.

Methodology

The methodological approach followed in this work was divided into three stages: First, the eco-efficiency of MSW services for each municipality analysed was evaluated. Second, the presence of economies of scale in the provision of MSW services for each municipality under study (Chilean municipalities) was assessed. Finally, the effects of external factors on the performance of municipalities providing MSW services were analysed using the Kruskal–Wallis non-parametric method.

Estimation of eco-efficiency scores

To evaluate the eco-efficiency of MSW services, the data envelopment analysis (DEA) method was employed (Boetti et al. 2012; Simões and Marques 2012; Guerrini et al. 2017; Sarra et al. 2017; Romano and Molinos-Senante 2020). It is a non-parametric technique based on linear programming that estimates the efficient production frontier, which is then used to measure the relative eco-efficiency of a set of DMUs. Hence, DEA methodology provides a measure of the relative performance of the DMUs based on the use of inputs to produce outputs (Banker et al. 1984; Charnes et al. 1978). The three main positive features of DEA to evaluate the eco-efficiency of MSW services are as follows: (i) it does not require information about input and output prices; (ii) it does not require the functional form of the efficient frontier to be defined Molinos-Senante and Guzmán (2018) and (iii) it allows desirable outputs (recyclable wastes), undesirable outputs (unsorted waste) and inputs (economic costs) to be integrated into a single index for the performance assessment Molinos-Senante et al. (2016). Hence, this methodological approach allows the joint economic and environmental performance of municipalities providing MSW services to be evaluated.

Subsequently, the DEA model employed to compute the eco-efficiency scores, i.e. the DEA method integrating unsorted waste as an undesirable output, is presented. The notation used is as follows: \( x\in {N}_{+}^N \) denotes the vector of inputs, \( y\in {N}_{+}^M \) is the vector of desirable outputs and \( b\in {N}_{+}^M \) is the vector of undesirable outputs.

The production possibility set of desirable and undesirable outputs given the level of inputs is defined as follows:

$$ P(x)=\left\{\left(y,b\right):x\ \mathrm{can}\ \mathrm{produce}\ \left(y,b\right)\right\} $$
(1)

The input distance function, including undesirable outputs, is defined as follows:

$$ D\left(x,y,b\right)={\min}_{\theta}\left\{\theta >0: x\theta \in P(x)\right\} $$
(2)

Following Fare et al. (1994), for each DMU (municipality in this study) j, the linear program to be solved to compute the eco-efficiency scores assuming constant returns to scale (CRS) is (Eq. 3):

$$ {\displaystyle \begin{array}{l}\mathit{\operatorname{Min}}\kern0.5em {\theta}^{CRS}\\ {}s.t.\\ {}\begin{array}{l}{\sum}_{j=1}^N{\lambda}_j{x}_{ij}\le {\theta x}_{i0}\kern0.5em 1\le i\le M\\ {}{\sum}_{j=1}^N{\lambda}_j{y}_{rj}\ge {y}_{r0}\kern0.5em 1\le r\le S\\ {}\begin{array}{l}\begin{array}{cc}{\sum}_{j=1}^N{\lambda}_j{b}_{zj}={b}_{z0}\kern0.75em & 1\le z\le H\end{array}\\ {}\ \begin{array}{cc}{\lambda}_j\ge 0& 1\le j\le N\end{array}\ \end{array}\end{array}\end{array}} $$
(3)

In the case of variable returns to scale (VRS) technology, the linear program to be solved to estimate the eco-efficiency scores for each unit j is (Eq. 4):

$$ {\displaystyle \begin{array}{l}\mathit{\operatorname{Min}}\kern0.5em {\theta}^{VRS}\\ {}s.t.\\ {}\begin{array}{l}{\sum}_{j=1}^N{\lambda}_j{x}_{ij}\le \theta {x}_{i0}\kern0.5em 1\le i\le M\\ {}\begin{array}{cc}{\sum}_{j=1}^N{\lambda}_j{y}_{rj}\ge {y}_{r0}\kern0.75em & 1\le r\le S\end{array}\\ {}\begin{array}{l}\begin{array}{cc}{\sum}_{j=1}^N{\lambda}_j{b}_{\mathrm{z}j}={b}_{z0}\kern0.75em & 1\le z\le H\end{array}\\ {}{\sum}_{j=1}^N{\lambda}_j=1\\ {}\begin{array}{cc}{\lambda}_j\ge 0& 1\le j\le N\end{array}\end{array}\end{array}\end{array}} $$
(4)

where θCRS and θVRS indicate the eco-efficiency scores of the DMUs evaluated, assuming constant and variable returns to scale technology, respectively. In both cases, θ ∈ (0, 1) and a DMU is eco-efficient if θ equals unity, whereas it is inefficient if 0 ≤ θ < 1. The difference between the eco-efficiency score and the value of 1 represents the input savings required by the unit to be eco-efficient; M is the number of inputs used; S is the number of desirable outputs generated; H is the number of undesirable outputs involved; N is the number of municipalities analysed; and λj is a set of intensity variables that represents the weighting of each DMU analysed.

Analysis of economies of scale

In order to analyse the economies of scale in the provision of MSW services, the direction of returns to scale was analysed as described by Marques and De Witte (2011). As illustrated in Eqs. 3 and 4, the eco-efficiency scores can be calculated by assuming either CRS or VRS technology. Under CRS technology, the eco-efficiency score is the product of scale eco-efficiency and pure technical eco-efficiency, since the CRS approach assumes that all DMUs operate at an optimal level (Charnes et al. 1978). In contrast, the VRS approach measures only eco-efficiency, since it compares DMUs with similar scales (Banker et al. 1984). According to Molinos-Senante and Guzmán (2018), when a DMU presents a CRS eco-efficiency score (θCRS) equal to the VRS eco-efficiency score (θVRS), its scale eco-efficiency (SEE) is equal to unity given that:

$$ SEE\ \left(x,y\right)=\frac{\theta^{CRS}}{\theta^{VRS}} $$
(5)

If SEE = 1, then the evaluated DMU operates under CRS technology. In contrast, if SEE < 1, then the evaluated DMU operates under VRS technology.

According to previous studies (Marques and De Witte 2011; Carvalho and Marques 2014; Molinos-Senante and Guzmán 2018), when a unit operates under VRS technology, i.e. SEE < 1, it can have either increasing returns to scale (IRS) or decreasing returns to scale (DRS). IRS is associated with positive economies of scale as desirable outputs increase by more than the proportional change in inputs Carvalho et al. (2015). In contrast, DRS involves negative economies of scale as it implies that desirable outputs increase by less than the proportional change in inputs (Guerini and Romano 2013).

Given that one of the objectives of this study was to evaluate the presence of economies in the provision of MSW services, whether DMUs (municipalities) operate under IRS or DRS technology was further evaluated. In doing so, an additional DEA model assuming non-increasing returns to scale (NIRS) was solved, providing an additional eco-efficiency score (θNIRS) (Cooper et al. 2011), which is as follows:

$$ {\displaystyle \begin{array}{l}\mathit{\operatorname{Min}}\kern0.5em {\theta}^{NIRS}\\ {}s.t.\\ {}\begin{array}{l}{\sum}_{j=1}^N{\lambda}_j{x}_{ij}\le \theta {x}_{i0}\kern0.5em 1\le i\le M\\ {}\begin{array}{cc}{\sum}_{j=1}^N{\lambda}_j{y}_{rj}\ge {y}_{r0}\kern0.75em & 1\le r\le S\end{array}\\ {}\begin{array}{l}\begin{array}{cc}{\sum}_{j=1}^N{\lambda}_j{b}_{zj}={b}_{z0}\kern0.75em & 1\le z\le H\end{array}\\ {}{\sum}_{j=1}^N{\lambda}_j\le 1\\ {}\begin{array}{cc}{\lambda}_j\ge 0& 1\le j\le N\end{array}\end{array}\end{array}\end{array}} $$
(6)

As stated by Hwang et al. (2016), on the one hand, a unit operates under DRS technology, i.e. presents negative economies of scale, when θNIRS = θVRS. On the other hand, a unit operates under IRS technology, i.e. exhibits positive economies of scale, when θNIRS ≠ θVRS.

Factors affecting eco-efficiency

The previous stages allowed the level of eco-efficiency in the provision of MSW by each municipality evaluated to be identified. However, there are some external factors affecting the management of MSW, also known as environmental variables, that could impact their eco-efficiency. These factors cannot be considered as inputs or outputs, since they are not directly controllable by the municipalities.

There are several alternative methodological approaches available to identify the factors that affect the eco-efficiency scores. Some previous studies (Fernández-Aracil et al. 2018) have employed parametric methodologies, i.e. econometric models in which the dependent variable is the performance index and the independent variables are the external factors evaluated (Simões and Marques 2011). However, this approach presents problems related to multicollinearity; therefore, a non-parametric methodology was applied in this study. The municipalities evaluated were grouped based on factors that were suspected to affect their eco-efficiency and were tested for statistically significant differences in the distribution of the eco-efficiency scores among the groups of municipalities (Guerrini et al. 2017; Sarra et al. 2017; Romano and Molinos-Senante 2020). The non-parametric Kruskal–Wallis test was applied. The null hypothesis tested was that the K samples are derived from the same population. The null hypothesis could be rejected at a 95% significance level when the p value was less than or equal to 0.05. The external factors were used to analyse the eco-efficiency of the MSW services. It should be noted that other studies (e.g. Simões et al. 2010; Romano et al. 2020; Ferreira et al. 2020) used alternative methodologies such as double bootstrap and order-m/order-α to evaluate the influence of exogenous variables in the performance of MSW service providers.

Sample description and data

The sample used for this empirical application corresponds to 142 out of 345 Chilean municipalities. These 142 municipalities correspond to 11,697,868 inhabitants (62%) of a total population of 18,729,160 people in Chile (Chilean Census 2017). Information about MSW management is available from the National System for the Declaration of Waste (SINADER) database. For 2018, which is the most recent year with available data, information regarding 310 municipalities was obtained. However, to conduct this study, only those municipalities that recycle some MSW were considered in the eco-efficiency assessment. Hence, according to the SINADER database, in 2018, 54.2% (168 out of 310) of Chilean municipalities did not recycle any MSW.

Based on past research on the performance of waste service management (e.g. Marques and Simoes 2009; Simões et al. 2010; Expósito and Velasco 2018; Romano et al. 2019), the total annual cost of the provision of MSW services by each municipality was integrated into the eco-efficiency assessment as the input. This variable was collected from the National Municipal Information System (SINIM) for 2018. It is one of the main Chilean databases and includes approximately 300 indicators for the 345 Chilean municipalities. The total cost indicator is defined as the expenditures made by each municipality, including cleaning services, waste collection and waste treatment. It is expressed in Chilean pesos/year. As this study evaluated the performance of municipalities providing MSW services using eco-efficiency as a metric, unsorted waste (tons/year) was considered as an undesirable output (Sarra et al. 2017; Expósito and Velasco 2018). On the other hand, recyclable wastes (tons/year) were integrated as desirable outputs and were categorised as follows: (i) paper and cardboard, (ii) glass, (iii) plastic, (iv) organic waste and (v) other inorganic recycled wastes (Marques et al. 2012; Díaz-Villavicencio et al. 2017). Hence, the eco-efficiency assessment conducted in this study involved one input, one undesirable output and five desirable outputs. Information about waste generation (unsorted water and recycled waste) was collected from the SINADER database for 2018.

To select the potential exogenous variables affecting the eco-efficiency scores, three criteria were considered: (i) the features of the MSW sector in Chile, (ii) the available information for the municipalities evaluated and (iii) the extant literature (e.g. Sarra et al. 2017; Calabrò and Komilis 2019; Romano and Molinos-Senante 2020). The following variables were considered: (i) the population served, which is defined as the number of inhabitants of the municipality; (ii) the municipality size, expressed in km2. This variable does not involve all areas of the municipality but focuses on the urban area where the MSW services are provided; (iii) the population density, which is expressed as the number of people per km2 of urban area of the municipality; (iv) the tourism ranking, proposed by the Studies and Territory Division of the Undersecretariat of Tourism together with the Destinations Unit, Territory and Environment, of the Sub-directorate of the Development of National Tourism Service (Sernatur, for its acronym in Spanish), and it considers 15 variables; (v) the amount of waste generated per capita, which was estimated as the ratio between the kg of waste generated and the number inhabitants of each municipality and (vi) the annual gross domestic product (GDP) per capita.

The main statistics of the variables used to estimate the eco-efficiency scores of the sample of Chilean municipalities and the exogenous variables affecting the eco-efficiency scores are shown in Table 1.

Table 1 Basic statistics of the variables considered to evaluate eco-efficiency
Full size table

Results and discussion

Evaluation of eco-efficiency and economies of scale

To evaluate the eco-efficiency of MSW services provided by each evaluated municipality, the first step was to investigate the technology they use in the management of their waste, i.e. whether they operate under CRS or VRS technology. In doing so and according to the methodological approach described in the “Methodology” section, the DEA-CRS and DEA-VRS models (Eqs. 3 and 4) were solved for each municipality evaluated. Figure 1 shows the SEE for the 142 Chilean municipalities evaluated. The eco-efficiency scores of each municipality, assuming CRS, VRS and NIRS technology, are shown in the supplemental material. Accordingly, 6 out of 142 (4.2%) of the municipalities operate under CRS as the SEE value equals unity, whereas 96% of the municipalities exhibit VRS technology (i.e. the SEE value is less than unity).

Fig. 1
figure 1

Scale eco-efficiency (SEE) in the provision of solid waste services for each Chilean municipality evaluated

Full size image

This finding indicates that only 4% of the municipalities evaluated present a scale of operation in which the inputs used and outputs generated are proportional. In contrast, 96% of the evaluated municipalities have economies or diseconomies of scale in their solid waste management. Therefore, the next step was to investigate whether the municipalities that operate under VRS technology present DRS or IRS technology, that is, if they exhibit negative or positive economies of scale, respectively. The eco-efficiency scores were estimated assuming NIRS. The difference between θNIRS and θVRS for each municipality is shown in Fig. 2. It was found that 55 out of 136 (40.4%) of the municipalities present IRS technology, indicating that they have positive economies of scale. From a management point of view, this means that these municipalities take advantage of their size of operation as the greater the quantity of wastes managed, the lower the cost per unit. On the other hand, 81 out of 136 (59.6%) of the evaluated municipalities operate under DRS technology, demonstrating that these municipalities present negative economies of scale in the management of MSW.

Fig. 2
figure 2

Difference of eco-efficiency scores in the provision of solid waste services for each Chilean municipality evaluated assuming variable returns to scale (VRS) and non-increasing returns to scale (NIRS)

Full size image

The geographic distribution of the municipalities operating under CRS, DRS and IRS technology is shown in Fig. 3. This figure illustrates that in northern Chile, very few municipalities recycle MSW; therefore, they were not considered in this study. Among the ones that do recycle waste, most of them operate under DRS technology, i.e. present negative economies of scale. In other words, the quantity of MSW recycled is not optimal. In the central zone of Chile, where most Chileans live, the municipalities exhibit both positive and negative economies of scale. Actually, some of the municipalities are continuous, suggesting that mergers or alliances among municipalities for the provision of MSW services will be beneficial in terms of the eco-efficiency. It should be noted that the area south of the metropolitan area of Santiago, which concentrates one-third of the total population of Chile, was not integrated into this study as municipalities located in this area formally do not recycle MSW. In this context, the municipalities evaluated within the metropolitan area of Santiago exhibit negative economies of scale, indicating that their operational size does not favour their eco-efficiency in the provision of MSW services; therefore, some policies should be implemented to change the current management of MSW. Finally, most of the municipalities located in southern Chile present positive economies of scale, which is very relevant as some of them are located in rather isolated areas.

Fig. 3
figure 3

Maps of Chilean municipalities exhibiting CRS, DRS and IRS technology in eco-efficiency assessment for the provision of MSW service. a North zone. b Central zone. c South zone

Full size image

Focusing on the eco-efficiency scores under VRS technology, Table 2 shows the main statistics of the eco-efficiency scores of the 142 Chilean municipalities evaluated. The mean eco-efficiency score is 0.58, indicating that the municipalities can save an average of 42% of the current cost by keeping the same amount of MSW managed (both recycled and unsorted wastes). The results estimated for Chilean municipalities are on the same order of magnitude as for the Tuscan municipalities reported by Romano and Molinos-Senante (2020). They reported mean eco-efficiency scores of 0.697, 0.677 and 0.523 for public, private and mixed firms providing MSW services. However, they are greater than the ones reported by Sarra et al. (2017) for a sample of Italian municipalities as well as a sample of Spanish municipalities Díaz-Villavicencio et al. (2017). In contrast, the average eco-efficiency of the MSW services provided by Chilean municipalities evaluated (0.58) is slightly lower than that estimated by Expósito and Velasco (2018) for Spanish regions, whose average value was 0.792. At the municipal level, the results illustrate that only 11 out of the 142 Chilean municipalities evaluated (7.74%) are eco-efficient; i.e. they comprise the best practice benchmark. Conversely, 131 out of 142 (92.25%) are not eco-efficient. Based on these results, most of the municipalities present room to improve their eco-efficiency in the management of MSW. Moreover, as shown in Fig. 4, most of the municipalities evaluated (108/142, 76.1%) present an eco-efficiency score ranging between 0.5 and 0.6, followed by 18 municipalities whose eco-efficiency is between 0.6 and 0.7. This finding indicates that the majority of the municipalities studied can improve their waste management process by at least 50% in comparison with the current status. On the other hand, only 11 out of the 142 municipalities evaluated are eco-efficient; that is, they are the best ones in terms of eco-efficiency in the management of MSW.

Table 2 Main statistics of eco-efficiency scores of the Chilean municipalities evaluated
Full size table
Fig. 4
figure 4

Histogram of the eco-efficiency scores of the municipalities evaluated

Full size image

Figure 5 shows the eco-efficiency scores for the Chilean municipalities evaluated for the northern, central and southern areas of the country. The 11 eco-efficient municipalities are not concentrated in a single area, but they are geographically distributed across Chile. Actually, of the few municipalities evaluated in the northern part of the country, only one of them is eco-efficient. As expected, due to the larger number of municipalities evaluated, most of the eco-efficient municipalities (7/11, 63.6%) are located in the central area of Chile; and four of these municipalities belong to the metropolitan area of Santiago. Finally, 3 out of the 11 eco-efficient municipalities are located in the southern area of the country. The geographical and environmental conditions of these eco-efficient municipalities are very variable. For example, the minimum and maximum populations are 880 and 400,869 people, respectively. Hence, it was important to conduct the second stage of analysis to identify potential exogenous factors affecting the eco-efficiency of MSW services.

Fig. 5
figure 5

Maps of Chilean municipalities and their eco-efficiency scores. a North zone. b Central zone. c Austral zone

Full size image

Factors affecting the eco-efficiency of MSW services provided by Chilean municipalities

Once the eco-efficiency scores of the 142 Chilean municipalities evaluated were computed, the exogenous variables were explored as potential determinants of the eco-efficiency scores. Table 3 shows the average eco-efficiency scores, the percentage of eco-efficient municipalities and the p values of the Kruskal–Wallis test for each group of municipalities, according to the five exogenous variables explored. The 142 municipalities were classified into groups according to the average values of each exogenous variable (see Table 1).

Table 3 Number of municipalities per group, mean eco-efficiency score, % eco-efficient municipalities and p value for each environmental variable.
Full size table

First, according to Sarra et al. (2017), Pérez-López et al. (2018) and Romano and Molinos-Senante (2020), the population of the municipality was investigated as a potential determinant of eco-efficiency of MSW services. Although the average eco-efficiency scores among the three groups of municipalities are small, according to the p value of the Kruskal–Wallis test, the differences among them are statistically significant. This finding indicates that municipalities with a population greater than 70,000 present an average eco-efficiency score that is larger than that of the medium and small municipalities. This result is consistent with previous research (Carvalho and Marques 2014; Pérez-López et al. 2018; Romano and Molinos-Senante 2020), which also concluded that the population served is a determinant of the efficiency of MSW services. Nevertheless, past research also has demonstrated contrary results. For example, Guerrini et al. (2017) and Expósito and Velasco (2018) have found that the size of the served population does not have a significant impact on the eco-efficiency of MSW services provided by a sample of Italian municipalities and Spanish regions, respectively.

Regarding the second variable evaluated, municipal size, the p value of the Kruskal–Wallis test (> 0.05) shows that this variable does not have a statistically significant impact on the eco-efficiency of MSW services provided by Chilean municipalities. The average eco-efficiency scores are fairly similar among the three groups of municipalities evaluated. This finding is consistent with the eco-efficiency results reported in the “Evaluation of eco-efficiency and economies of scale” section, revealing that 40.4% of the municipalities evaluated present positive economies of scale, whereas 59.6% of them present negative economies of scale. In other words, the impact of the economy of scale in the provision of MSW services by Chilean municipalities is inconclusive. In contrast, some previous research (e.g. Sarra et al. 2017; Fidelis and Colmenero 2018) has concluded that the collection area has an impact on the performance of MSW service providers.

Regarding the influence of the population density on eco-efficiency, the results shown in Table 3 illustrate that municipalities with a large population density (> 170 people/km2) are the most eco-efficient. In contrast, municipalities with a moderate-to-low population density present lower average scores of eco-efficiency. The Kruskal–Wallis test led us to reject the hypothesis of equality of means for eco-efficiency with 95% significance based on the three groups, since the p value was < 0.05. This finding indicates that the population density significantly affects the eco-efficiency of MSW services. Similar conclusions have been reported by Guerrini et al. (2017) for a sample of Italian municipalities. However, other past research (e.g., Marques and Simoes 2009; De Jaeger et al. 2011; Expósito and Velasco 2018) is inconclusive about the impact of the population density on the performance of the MSW service providers.

According to previous studies (e.g. Bosch et al. 2000; García-Sánchez 2008; Guerrini et al. 2017), the impact of tourism on the eco-efficiency MSW services was investigated. However, the results for the empirical application conducted in this study are divergent from previous ones. In the current study, it was found that municipalities with a higher tourism ranking are the most eco-efficient. The Kruskal–Wallis test results indicate that the differences in the eco-efficiency scores for the three groups of municipalities are statistically significant, demonstrating that tourism has a positive impact on the management of MSW services. It should be noted that in most areas of Chile, tourism is a developing industry and not as advanced as in the case studies analysed by Guerrini et al. (2017) and Romano and Molinos-Senante (2020).

The next factor evaluated in this study was the amount of waste generated per capita. The results listed in Table 3 reveal that the largest average eco-efficiency scores correspond to the largest waste production; i.e. the municipalities with waste generation greater than 800 kg per capita per year have higher eco-efficiency scores. The p value of the Kruskal–Wallis test (< 0.05) shows that this variable has a significant effect on the eco-efficiency of the MSW services. This finding illustrates that the fixed costs of managing MSW in Chile are high; therefore, as the amount of waste generated increases, municipalities become more eco-efficient. This is due to the fact that MSW is managed independently by each municipality in Chile. In order to reduce fixed costs and to improve the eco-efficiency of MSW services, it is necessary that municipalities collaborate and develop alliances.

Finally, the last external factor evaluated was the annual GDP per capita. Table 3 shows that the largest mean eco-efficiency score corresponds to those municipalities with a GDP per capita larger than CLPFootnote 1100 million per year. The p value of the Kruskal–Wallis test is 0.03 which reveals that this variable has a significant effect on MSW services' eco-efficiency. This finding evidences that municipalities with larger economic resources can adopt policies to promote waste recycling among their inhabitants. Given that recycling is still incipient in Chile, these policies can be costly and therefore, require significant financial resources. This information is very relevant for policymakers at the regional and national level as additional economic resources should be devoted to achieve the goals of the Chilean Recycling Law adopted in 2016.

Conclusions

The management of MSW is a basic service provided at the municipal level and is part of the Sustainable Development Goals. In a circular economy context, it is necessary that municipalities evaluate their eco-efficiency in the provision of MSW services. In other words, the assessment of the performance of MSW services should integrate both economic and environmental variables. Moreover, in Latin American countries such as Chile, regulations regarding MSW and recycling incentives and targets are emerging. Hence, it is relevant to evaluate the presence of economies of scale and exogenous variables affecting the eco-efficiency of MSW services.

To shed light on this relevant topic, this study evaluated the eco-efficiency of MSW services provided by a sample of Chilean municipalities using the DEA method. In the performance assessment, the total cost as the input, unsorted waste as an undesirable output and five types of recycled waste as desirable outputs were integrated. The results from the empirical application illustrate that 40.4% of the municipalities evaluated presented negative economies of scale. This finding is very relevant from a policy perspective, since it provides evidence that the current size of the operation of the municipalities is not optimum; therefore, joint organization of systems of MSW management should be promoted. It was also found that the average eco-efficiency score was 0.58 and that 92.3% of the municipalities were not eco-efficient. These results reveal that the management of MSW in Chile has not been addressed from a holistic perspective embracing economic and environmental issues. Thus, the regulator and municipalities need to introduce effective policies to promote the recycling of MSW. Finally, it was demonstrated that 5 out of 6 of the exogenous variables evaluated (population served, population density, tourism, waste generated per capita and annual GDP per capita) significantly affect the eco-efficiency of MSW services.

From a policy perspective, the results of this study are very relevant to support and adopt specific actions by policymakers. Firstly, it has been evidenced that in most of the cases, the local scale, which is the one adopted in Chile, is not adequate to promote solid waste recycling. In this sense, initiatives at the metropolitan or regional level should be promoted. Secondly, eco-efficient municipalities are heterogeneously distributed throughout the country which evidences that regional policies have not been adopted yet. This issue is very relevant for a large country like Chile. Moreover, a notable difference among neighboring municipalities, in terms of eco-efficiency in the management of MSW, has been revealed. The identification of factors affecting eco-efficiency scores is also relevant for policymakers. Thus, the fact that municipalities with the largest number of inhabitants are the most eco-efficient supports the need to develop waste management policies on a larger scale than the local one. It has also been evidenced that tourism does not always imply negative impacts from the perspective of waste management, which is associated with educational campaigns and promotion of MSW recycling that should be carried out in all municipalities (not only on those municipalities with more tourism). The identification of annual GDP per capita as a variable affecting the eco-efficiency of municipalities evidences the impact of economic issues on MSW management and also suggests that to improve eco-efficiency, an inter-municipal fund should be created where municipalities with fewer economic resources could receive contributions from the richest ones contributing to enhancing MSW recycling issues across the country.