Waste Management Hybridization and Social Mechanisms: The Unpredictable Effects of a Socio-Technical Assemblage

Abstract

Reducing total waste generation, increasing the rate of separate collection, and ensuring efficient material recovery are three key objectives recognized at the UN, EU, and national levels for achieving a sustainable waste management system. The policy and scientific debate has traditionally been polarized around two main collection systems: door-to-door and street collection, each with its own strengths and weaknesses. However, in recent years, hybrid waste collection models, which aim to integrate the advantages of both systems, have gained increasing attention. The impact of these models on sustainability, as well as the social mechanisms underlying their success or failure, remain underexplored. This paper analyzes a case study of a hybrid waste collection innovation implemented in eight small and medium-sized municipalities in Piedmont, Italy. The study has a twofold objective: (1) assessing the economic and environmental impacts of the system and (2) hypothesizing the social mechanisms that generate these effects. To address these research gaps, we develop a theoretical framework that combines the socio-technical system approach with analytical sociology. The framework is then operationalized through a quasi-experimental research design, which allows us to measure the effects of the hybrid innovation on a set of 17 indicators. The analysis is conducted by comparing the treated municipalities with a control group selected for its similarity. Empirical evidence reveals an unexpected and partially contradictory outcome: while the total amount of waste decreases, this reduction is not offset by a proportional increase in separate collection rates. Moreover, the quality of waste separation worsens. To explain this emerging pattern, we formulate a set of hypotheses—grounded in our theoretical framework—on the key factors influencing individual behavioral responses. Our provisional conclusion, which requires further qualitative validation, suggests that the interplay between economic incentives and social control in a context of imperfect awareness leads to ambiguous effects of hybridization in waste collection systems.

1. Introduction

Waste management plays a crucial role in the ecological transition. Reducing total waste generation, increasing separate collection rates, and improving economic returns from recyclable materials are key objectives embedded in the UN Sustainable Development Goals (SDGs 11, 12, and 14) [1] and in major policy initiatives such as the European Green New Deal [2].

Waste management is a socio-technical system, a concept used in academic literature to describe the co-evolution of technologies, infrastructures, regulations, markets, cultural norms, and user practices that collectively shape the provision of essential services [3,4]. Like other socio-technical systems, such as energy [5], urban waste management is characterized by institutional divergence [6]: while overarching objectives—such as reducing waste generation or increasing recycling rates—are broadly shared, the pathways to achieve these goals remain contested due to differing regulatory frameworks, market structures, and socio-cultural factors.

A prominent example of this divergence in waste management is the debate between proponents of door-to-door (DtD) and street collection (SC) models [7]. These systems represent distinct configurations [8] of technologies, human labor, regulatory frameworks, and user practices, each influencing environmental and economic outcomes. The DtD model is labor-intensive, requiring a high number of small vehicles, specialized devices, and significant human effort [9]. It also demands strong user engagement, as residents are required to sort waste correctly and adhere to a predefined collection schedule for the DtD model to work properly [10,11]. In contrast, SC is more capital-intensive, relying on larger, more expensive vehicles and a smaller yet often more specialized workforce. Unlike DtD, it does not require strict scheduling compliance from users, potentially making waste disposal more comfortable for them.

DtD is associated with better environmental and economic performance, particularly from the perspective of Zero Waste advocates [12]. It is credited with achieving higher rates of differentiated waste (DW) and greater financial returns from recovered materials [13]. However, it has also been criticized for its high operational costs and vulnerability to fluctuations in market demand for secondary raw materials [14]. SC proponents argue that their model is more cost-effective due to lower labor expenses and reduced informational demands on users. However, a consistent lower material recovery rate compared to DtD remains its main limitation [15] and a key point of criticism from Zero Waste supporters.

These analyses clearly highlight the strengths and weaknesses of DtD and SC systems, as they would be “pure” systems. However, in most urban contexts, including small and medium-sized cities, “mixed” waste collection systems are commonly found [16,17], where regulatory and technological elements from both models coexist within the same setting. For instance, it is not uncommon for municipalities to implement DtD collection for certain waste fractions while relying on SC for others or to adopt different models across specific city zones or user categories, such as non-residential establishments. Several studies have examined the economic and environmental sustainability of these models in various European cities [16,17,18,19]. However, they have rarely focused on cases where the collection system was not merely a combination of DtD and SC, but rather an innovative model characterized by the overlap of regulations and technologies from both systems. We propose calling these kinds of waste collection models “hybrid”.

A prime example of such hybrid models is the implementation of smart bins [19], which are street containers accessible only via an identification card and equipped with opening mechanisms designed to incentivize users to reduce waste volume, particularly for residual and organic waste. Systems characterized by smart bins have a significant economic and environmental impact, which requires further study to better inform urban policy decisions. Moreover, they offer a valuable opportunity from a scientific perspective, as they illustrate how variations in regulations and technologies influence individual behaviors, and how the aggregation of these behaviors creates the conditions for innovation. This represents a challenge that allows sociological disciplines to collaborate with STEM fields not only in understanding but also in modeling waste collection systems oriented toward sustainability.

This article reports on the first results of a research project on the analysis and implementation of these hybrid models, conceived as socio-technical innovations. The overall aim of the research project is to investigate the complex processes which shape the innovation pathways in the field of waste collection and management with the ambition of explaining the dynamics affecting its economic and environmental performance.

The design of the research is therefore structured as follows: (1) a description of the effects of the innovation on the waste management system performance and the formulation of a causal hypothesis to explain the generative micro–macro dynamics behind such effects; (2) qualitative and quantitative insights to test the causal hypothesis and build a middle-range theory able to bridge innovation pathways and environmental performance in waste management; (3) scenarios analysis based on the operationalization of the theory. By adopting a quasi-experimental approach [20], we analyzed panel data on the economic and environmental performance of a hybrid innovation promoted in eight municipalities by Acea, a multiutility providing waste collection service to a number of small and medium-sized municipalities in Piedmont. In particular, Acea has introduced “smart bins” for the collection of undifferentiated waste, organic waste and paper, containers that can be accessed by users with a magnetic card and that can track the number of deliveries. In addition, the smart bin for undifferentiated waste has been redesigned by reducing the size of the opening, so that only 30-L bags can be inserted, and the card must be used for each deposit. Acea’s objectives, described by the company in various institutional documents, were to reduce the quantity of undifferentiated waste and improve the quality of differentiated waste, areas in which the previous collection model, corresponding to a “pure” SC, achieved results below the regional and national legal requirements. According to the Acea engineers who designed the new service, smart bins would have made it possible to introduce the strengths of door-to-door, such as individual tracking of deliveries, into a system that would maintain the strengths of SC, in particular a more sustainable personnel expenditure.

The research, carried out by University of Turin and funded by Acea, aims at answering three main research questions: (1) Has the introduction of the DtD mechanism—the locking of containers for unsorted waste, organic and paper, with magnetic cards and the tracking of the delivery of these types of waste—achieved meaningful results in terms of sustainability? (2) How did this hybridization reconfigure the roles and interactions between the socio-technical components? (3) What socio-technical mechanisms triggered by this reconfiguration might explain the results of the hybridization?

This article represents a preliminary stage in this direction and will mostly focus on the first question, i.e., the effects of hybridization on the economic and environmental performance of the waste collection system. However, in the discussion it is left to outline some hypotheses to deepen the analysis and address the other two on the description and explanation of socio-technical mechanisms able to produce the detected effects.

The article is structured as follows. The next section introduces the theoretical framework of the study. The third section introduces the case study, and the methodology. The fourth presents the main results of the empirical analysis. The fifth section discusses the results and proposes hypotheses to explain the most significant changes in the systemic behavior of the treated group. Lastly, in the Conclusion, the main lessons learnt and a few proposals for deepening the analysis are proposed.

2. Theoretical Background

The research questions address two main dynamics: the aggregate effect of the innovation introduced in waste collection and management (question 1), and the ways changes at the level of individual users appear to produce such an effect (questions 2 and 3). To capture these dynamics, we developed a theoretical framework by combining two well-established traditions of social science research: socio-technical systems and analytical sociology. The latter provides conceptual tools to link the different levels of a social system by considering the actions and interactions of individual agents as affected by and influencing the environment in which they are embedded and highlights the relevance of hypothesizing mechanisms to explain such reciprocal micro–macro dynamics. The former provides useful conceptual tools to support the identification of these mechanisms by conceiving them as the result of the combination of the hardware and software components of social systems, i.e., the technologies and the social settings (rules, behaviors, cultures…) with which they interact.

The relationship between technological change and social structures has in fact long been debated in social theory. Early approaches, such as those influenced by functionalist systems theory, often assumed a deterministic posture based on the understanding of social change as linearly driven by technological innovations [21]. This perspective, referred to as “technological determinism” [22], suggests that when a new technology is introduced social components must adapt to maintain systemic balance. However, this view has been widely criticized for oversimplifying the complex interactions between technology, institutions, and human agency [23].

Science and Technology Studies (STS) [4,8] offer a more nuanced understanding of technological change. Empirical research [24,25,26] has demonstrated that technological development is not an autonomous force but is deeply embedded in the norms, values, and power structures of organizational fields [27] where innovations emerge. Technologies do not simply produce social change; rather, they are both shaped by and shaping the political and cultural orientations of the social groups that design, develop, and adopt them [28].

A useful approach in this regard is Actor–Network Theory (ANT) [29] which conceptualizes technologies not as passive tools but as actors within networks of human and non-human entities. ANT identifies the key components of specific socio-technical systems—such as technologies, formal and informal norms, and collective and individual agents—assigning and qualifying specific forms of agency to each of them. A central aspect of ANT is the process of translation, through which various actors/actants exchange meanings and resources, form alliances, and, based on these and other dynamics, drive (or hinder) the diffusion of innovations.

However, ANT is only partially applicable to our case study. We are not investigating what motivations led certain actors to promote an innovation, nor what alliances or conflicts shaped their implementation and dissemination. Instead, our research questions focus on the effects—particularly the unexpected ones—on individual user behavior, resulting from a socio-technical hybridization in which regulatory and technological devices from one system have been “assembled” onto another.

As mentioned above, in order to better capture these dynamics, we propose to integrate ANT with analytical sociology [29,30]. This approach conceptualizes the evolutionary dynamics and behavior of social systems as the result of interactions between macro-level structures—such as public administration regulations on waste management or technologies enabling a particular waste collection model—and micro-level components, namely the interests, goals, resources, behaviors, and interactions of individuals (social agents). The structure influences the actions of social agents, whose aggregated behaviors, in turn, reshape the structure itself.

This macro–micro–macro dynamic, widely known as “Coleman’s boat” [31], suggests a three-step analytical approach (Figure 1). By integrating ANT with analytical sociology, we aim to trace how macro-level institutional structures influence individual behaviors and how these, in turn, contribute to reshaping the system over time [32].

First, the logic of the situation (macro–micro, line 1) describes the context of action shaped by the innovation at the macro level: how does the introduction of a card to track individual waste disposal redefine users’ objectives and interests?

Second, the logic of selection (micro–micro, line 2) examines the various “good reasons” [30,33] that drive users to adopt new behaviors: how and why do they change their waste separation and disposal practices?

Finally, the logic of aggregation (micro–macro, line 3) explores the systemic consequences of these individual actions: how do the combined (and sometimes conflicting) behavioral changes of social agents impact the ecological and economic sustainability of the entire system?

From an analytical sociology perspective [32,33], the introduction of a magnetic card-operated waste bin is a sociologically significant phenomenon because it activates complex social mechanisms that can lead to a systemic behavioral shift. For instance, in municipalities where this hybrid innovation was implemented, unsorted waste decreased by more than 50%, whereas in others where the innovation was not introduced, it increased by 10%. Actor–Network Theory (ANT) suggests examining the agency of non-human actants—such as the mechanisms regulating bin access or the technologies tracking waste disposal—as co-determinants of individual behavior. These elements do not merely function as passive instruments but actively shape interactions, constraints, and incentives within the waste management system, influencing users’ waste-sorting practices.

This article focuses primarily (Results Section) on the measurement of the systemic effects of socio-technical innovation at the aggregate level of the system, i.e., what we called above the systemic behavioral shift. Then, through the analytical approach described, potentially explanatory hypotheses are formulated (Discussion Section) about the “macro–micro–macro” dynamic to which the emergence of the systemic change can be attributed.

3. Methodology

This section is divided into two parts. The first describes the research design and highlights the characteristics of the hybrid model implementation that allowed us to consider the case study as quasi-experiment; the second is focused on data and the methodological approach adopted for their analysis.

3.1. The Hybrid Innovation: A Quasi-Experimental Design

The case study area is composed of 44 municipalities which set up an in-house company for the integrated management of the waste cycle, Acea Pinerolese Industriale (Acea, from now on).

The municipalities show considerable demographic, geographical and socio-economic heterogeneity. In particular, 16 municipalities have less than 1000 inhabitants; 24 have between 1000 and 5000 inhabitants; 4 have between 7000 and 8000 inhabitants, while only Pinerolo has the size of an inter-municipal center, with around 35,000 inhabitants. This demographic distribution is strongly related in almost all cases with the geomorphology of the settlements. Municipalities with up to 1000 inhabitants are located in the high or medium-sized mountains, while those with up to 5000 inhabitants are located between the medium-sized mountains and the valley floor, and the larger municipalities are located in the flat land. The average per capita income of the area (19,592 euros) is slightly lower than that of the metropolitan area of Turin (around 22,921 euros). Until 2016, Acea provided the same waste collection service to all the municipalities in the area: an integrated street recycling collection service based on five drop-off waste containers (organic waste, undifferentiated dry waste, plastics and metals, glass, paper, and cardboard). Drop-off containers were positioned in the territory on the basis of distribution of population in order for them to be equally accessible to any type of user. This system enabled Acea to achieve on average 50% of DW on total waste production in 2016, a good quantitative result but still far below the 65% required by European and national legislation. The main issue was on the quality of DW fractions, with many municipalities realizing that the purity of the waste delivered to the recycling platforms, starting with organic waste, could be significantly improved. In addition, the Piedmont Regional Waste Plan provided incentives for municipalities to activate door-to-door collection and to introduce “pay as you throw” incentives (PAYT).

In the biennial 2021–2022, Acea introduced some innovations in eight municipalities, including Pinerolo, which ’hybridized’ the previous street collection system with elements more akin to door-to-door. Specifically, in the old system, the bins for the collection of wet waste were 240 L, side-loading, with a totally openable lid. Dumpsters for the other fractions had a volume of 1700 to 3200 L, with a door that could be fully opened on both sides. In the new system, the bin with the undifferentiated waste was equipped with a 30-L drawer opening and a cap with an electronic system for tracking the deliveries. The same type of design was adopted for the organic waste bin, but with a 20-L drawer mouth. For all other bins, a calibrated mouth on both sides was chosen. The bins for undifferentiated waste, organic waste and paper were equipped with an electronic canopy, and each user was given an access card, which allowed the registration of each delivery. Acea delivered an illustrated brochure door-to-door with instructions on how to organize the new service and some advice on how to separate waste correctly. The distribution of the access cards to the bins was carried out door-to-door and through a number of information points located in the squares of the municipalities involved in the experiment. The communication materials trace the reasons for the innovation back to the need to improve waste separation and make a general reference to the success achieved by other municipalities in similar experiments. A home composter was also distributed to users who requested one. The expected results of the project were essentially threefold: (i) to achieve a 65% separate collection; (ii) to reduce the total amount of waste produced by 10–15%; (iii) to achieve improvements in the quality of waste sent for recovery, with consequent economic benefits.

Based on the characteristics of the hybrid innovation implementation process, we opted for applying a quasi-experimental research design (Barrera et al., 2024 [20]). The ‘experiment’ aims at assessing the causal relationship between the introduction of a DtD component into a SC system (the independent variable) and the environmental and economic performance of the waste collection system. Below, we summarize the quasi-experimental setting, namely: (1) the independent variable, (Treatment in the following), (2) the identification of the units affected by the treatment (Treated = T in the following) and the control group (Not treated = NT in the following), and (3) the outcome. First, we define the Treatment as the introduction of the hybrid innovations in the socio-technical waste collection service in the biennial 2021/2022. Second, we consider the T as the eight municipalities where the innovation was implemented (T), with the control group resulting in them being composed of the other 36 municipalities (NT) where the innovation was not put in place. As for the reliability of the NT for comparison of the outcomes, it must be noticed that T and NT are similar for their structural, cultural and institutional profile. As for the cultural and institutional, they belong to the same territorial context characterized by a strong Waldensian identity and a long-lasting history of cooperation between public and private actors. And, as far as the subject of the study is concerned, they have been served by the same waste collection system provider (Acea) for decades.

As for the structural profile,

Table 1. The sample: comparison between treated and untreated groups.

		NT n = 36 Population 2023 = 67,064		T n = 8 Population 2023 = 61,429
		Median	vc	Median	vc
descriptive variables	population 2023	1.218	0.99	3.839	1.40
	annual income 2023 (€)	20,975.2	0.13	21.019	0.25
	area (skm)	15.9	0.86	15.9	0.73
	densita23	105.2	0.92	110.5	0.97
	land use 2023 [%]	6.1	0.73	6.1	0.65
	labor forces tot 2022	292.5	1.17	425.5	2.30
	% LF manufacture (C)	18.5%	0.87	23.1%	0.76
	% LF tourism (I)	7.2%	1.01	7.0%	1.53
	% LF tourism retail (G)	13.0%	0.51	15.7%	0.58
	ul_tot_2022	93.5	0.98	116.0	2.28
	% LU manufacture (C)	9.7%	0.69	10.5%	0.89
	% LU tourism (I)	7.0%	0.92	7.0%	1.43
	% LU tourism retail (G)	18.5%	0.34	19.0%	0.35
variation	LF 2022/2018	1.09	0.10	1.05	0.25
	LU 2022/2018	1.08	0.12	1.06	0.25
	population 2023/2019	0.98	0.04	0.98	0.24
	income 2023/2019	1.07	0.07	1.07	0.23

Note: LF = Labor forces; LU = Local Units.

shows the results of the analysis of a few socio-economic variables for both the T and NT municipalities. For each of the variables considered, median and variation coefficient are reported in order to have a clear understanding of the similarity between the distribution of the variable within the group and the degree of homogeneity of the groups. Please consider that, in line with the research questions and the theoretical framework, the aim of this paper is to get the emerging effect of the innovation at the aggregated level of the considered territorial systems and not to get the difference triggered on an individual treated unit. Therefore, we opted to test the similarity between the two groups as a whole instead of identifying within the NT group a subset of members most similar to the T members.

Table 1, using data from Italian National Statistics Institute (ISTAT), reports the results of such analysis and shows a high degree of similarity between the two groups. In terms of both distribution and internal heterogeneity, all the main socio-economic variables and dynamics are most likely to be able to affect waste production and also show similar value in T and NT. The biggest difference is in the n of research units (8 T vs. 36 NT), in some variables of scale, i.e., population, labor forces and local units and in the concentration of the distribution, i.e., the values of cv. These differences are due to the fact that in the T group is included Pinerolo (35,000 inhabitants), which produces a general increase in the order of magnitude of the considered socio-economic variables and a general decrease in the homogeneity of the distribution. However, these differences do not cause bias in the analysis because the size of the phenomena directly affects only the amount of waste produced (i.e., the quantitative aspects) but not how societal actors and services perform waste production, collection and treatments (i.e., the qualitative aspects). The latter are in fact related to the structural profile of the socio-economic contexts that is captured by other variables, such as income distribution, land uses and the economic structure, i.e., distribution of labor forces and LU among different sectors. In this respect, the T and NT are really close and therefore the assumption of similarity between the groups to properly perform the DID is respected. In addition, from the perspective of the organizational and societal dynamics triggered by the projects implementation (i.e., the treatment), it is worth highlighting that in Pinerolo it has been implemented in seven sub-metropolitan areas of around 3 to 5000 inhabitants that are independent in terms of the waste management and, therefore, for any purposes of our research can be considered as many small towns. Finally, as for the outcomes, the indicators to measure the effect of the treatment were identified on the basis of a triple bottom line approach, that consider the results achieved by organizational and technological waste management models from an environmental, economic and social perspective [34,35]. On these premises, and based on the availability of data, we opted for a strong focus on the environmental and economic dimension of innovation as the social dimension would have required additional quantitative and qualitative information that was out of the scope and the objective of the present work. By integrating research and institutional sources analyzed in the desk analysis (see Section 3.1) we identified 17 indicators to operationalize the measurement of the environmental and economic effects of the innovation on the territories involved,

Table 2. Outcome indicators: economic and environmental effect of treatment.

	Indicator	Description
environmental	1. Total Municipal Waste	KG total waste per capita
	2. Unsorted waste (UW)	KG of unsorted waste per capita
	3. Differentiated waste (DW)	KG of differentiated waste per capita
	4. DW—organic	KG of organic waste per capita
	5. DW—economic valuable	KG per capita of differentiated waste € valorized
	6. Separating Attitude	Ratio Differentiated Waste (3)/Sorted Waste (2)
economic	7. Collection TMW costs—per capita	Yearly cost for collection/transportation of TMW—per capita
	8. Collection TMW costs—per kg	Yearly cost for collection/transportation of TMW—per kg
	9. Collection DW costs—per capita	Yearly cost for collection/transportation of DW—per capita
	10. Collection DW costs—per kg	Yearly cost for collection/transportation of DW—per kg
	11. Treatment DW costs—per capita	Yearly cost for DW treatment—per capita
	12. Treatment UW costs—per capita	Yearly cost for UW treatment—per capita
	13. Treatment UW costs—per capita	Yearly cost for UW treatment—per kg
	14. Total cost—per capita	Total yearly cost of TMW collection and treatment—per capita
	15. Total cost—per kg	Total yearly cost of TMW collection and treatment—per kg
	16. Quality DW—economic performance	Ratio between the actual DW economic valorization and the theoretical maximum
	17. Quality DW—material performance	Ratio between DW economic valuable (5) and total DW (3)

(for more details about reference and sources see Appendix A).

3.2. Data Collection and Analysis

Although we performed a wide desk analysis of administrative and regulatory documents and we engaged representatives of both the municipalities and ACEA in a number of discursive interviews, the methodological core of the research was quantitative and based on two steps: (1) a wide data collection; and (2) a counterfactual analysis (CA) aimed at identifying and measuring the results of the quasi-experiment. As for the data collection, below are reported the five main domains covered in the analysis:

• Demography—Residents by age at the municipal level (Source: ISTAT—Italian national Institute of statistics);
• Income—Income per capita at the municipal level (Source: Italian Ministry of Finance);
• Economic structure—Labor forces and local units by economic sector at the municipal level (Source: ASIA –Registry of Italian active enterprises);
• Waste collection—Quantity of waste collected by type and by municipality, years 2019–2023—(Source: ACEA);
• Waste service—Economic flows (costs and gains) generated by the service, years 2019–2023—(Source: ACEA).

Domains 1, 2 and 3 are aimed at providing a comprehensive description of the territorial context in order to assess the comparability between the two groups of municipalities. Domains 4 and 5 are aimed at measuring the economic and environmental performance of the local waste management system (data 4 and 5). Since panel data on the treatment and its outcomes were available, we adopted a difference-in-differences (DID) technique [33] to perform a CA. In order to properly frame and support the interpretation of the results (Section 3), it might be useful to shortly recall a few main concepts on CA and DID. Derived by the field of medical sciences, CA is a well-known and widespread approach to the evaluation of the effects of a treatment (a policy or a project) aimed at identifying its ’net effect’ by comparing the change in a variable of interest after the intervention (i.e., the expected effect of the treatment on the recipients) with the value that the variable would have had on the same recipients without the treatment. CA analysis relies on a wide range of approaches and statistical techniques among which DID is widely adopted to grasp the effects of a treatment when panel data are available, as in the case of our research. DID estimates the (causal) effect of the treatment by comparing the changes in outcomes over time between the two groups T that did experiences the treatment and NT the control group that did not. The core idea is to look at how the expected outcome of the treatment (the variable of interest) changes over time for both groups and then calculate the difference between those changes to determine the effect of the intervention. The basic assumption is that, if the treatment had not taken place, T and NT would have followed a similar evolutionary pattern. Or, in other words, the assumption is that T and NT must be similar enough for the variable of interest to be expected to experience the same change over time in absence of the treatment, i.e., if some differences are detected, this difference (positive or negative) can be considered as the effect of the treatment. In order for the basic assumption on the similar pathway to be reasonably accepted, similarity between T and NT must be verified and confirmed, as it is the case in our research design (see Section 5.1). The straightforward formula to compute the DID is as follows:

DID = (Y_T,1 − Y_T,0) − (Y_NT,1 − Y_NT,0)

where Y = variable of interest or expected outcome of the treatment,

• T = treatment group, indicates the value of Y for the treated,
• NT = control group, indicates the value of Y for the not treated,
• 1 = post-treatment period, indicates value of Y after the treatment,
• 0 = pre-treatment period, indicates value of Y before the treatment.

DID is the net effect of the treatment, i.e., the measure of the relationship between the independent variable and the result of the experiment.

In our analysis we operationalized the formula as follows:

• Y = the set of indicators outcome-grouped into two clusters: environmental and economic (see

  Table 3. a—The environmental effects of the treatment. b—The economic effects of the treatment.

  a
  Indicator	UM	Pre-Post Treatment			DID	DID Norm	Impact on EU Sustainability Goals
  			2019	2023
  1. Total Municipal Waste (TMW)	KG	NT	505.9	552.9	−106.15	−2.44	positive
  		T	462.4	403.2
  2. Unsorted waste (UW)	KG	NT	241.4	255.5	−111.48	−2.99	positive
  		T	204.1	106.8
  3. Differentiated waste (DW)	KG	NT	264.5	297.4	5.32	0.84	positive
  		T	258.2	296.4
  4. DW—organic	KG	NT	69.2	84.9	−1.16	−0.07	positive
  		T	87	101.6
  5. DW—economic valuable	KG	NT	123.2	133.5	−4.57	−0.22	positive
  		T	102.5	108.3
  6. Attitude to separate (DW/UW)	ratio	NT	0.53	0.55	0.15	4.56	positive
  		T	0.56	0.74
  b
  Indicator	UM	Pre-Post Treatment			DID	DID Norm
  			2019	2023
  7. Collection TMW costs—per capita	€	NT	14.9	14.1	0.38	0.08
  		T	10.2	9.7
  8. Collection TMW costs—per Kg	€	NT	0.1	0.1	0.05	3.89
  		T	0	0.1
  9. Collection DW costs—per capita	€	NT	46.9	41.6	7.48	0.42
  		T	29	31.2
  10. Collection DW costs—per Kg	€	NT	0.2	0.1	0.03	0.53
  		T	0.1	0.1
  11. Treatment DW costs—per capita	€	NT	5.9	11.5	−0.70	−0.60
  		T	7.1	12
  12. Treatment UW costs—per capita	€	NT	39.7	11.5	5.84	1.09
  		T	34.3	12
  13. Treatment UW costs—per Kg	€	NT	0.2	0.1	0.13	3.1 × 106
  		T	0.2	0.3
  14. Total cost—per capita	€	NT	187.8	209.2	15.89	0.31
  		T	136.1	173.4
  15. Total cost—per Kg	€	NT	0.4	0.4	0.13	1.88
  		T	0.3	0.4
  16. Quality DW—economic performance	Ratio	NT	0.89	0.77	−0.06	−3.50
  		T	0.9	0.73
  17. Quality DW—material performance	Ratio	NT	0.47	0.44	−0.01	−0.13
  		T	0.4	0.37

  ),
• T = 8 municipalities that implemented the innovation in 2020 or 2021,
• NT = 36 municipalities that did not implement the innovation,
• 1 = 2023, the after-treatment period,
• 0 = 2019, the pre-treatment period.

In formula:

DID (Y_i) = (Y_T,2023 − Y_T,2019) − (Y_NT,2023 − Y_NT,2019)

where Y_i is the i-esim indicator of the set, i.e., we replicated the diff-in-diff for all the indicators we deemed relevant to get a meaningful understanding of the effect of the innovation, i.e., the results of the experiment.

In order to have a comprehensive view of the overall effect and to identify which are the outcomes more affected by the treatment, we computed a normalized value (DID norm), as follows:

DID_norm_i = DID/(Y_i,2019,T − Y_i,2019,NT)

where i is the i-th indicator and Y_i,2019 is its value before the treatment for both T and NT.

The DID norm provided, therefore, the relative net effect of the treatment expressed in terms of variation on the pre-existing difference between T and NT. In such a way, it is possible to compare the effects of the treatment on different indicators by considering two thresholds:

• 0, if DID norm < 0 variation in T are higher than variation in NT and if DID_norm < 0 variation in T being lower than NT; DID_norm = 0 indicates of course no effect of the treatment.
• 1, (in absolute value) shows the strength of the variation. Depending on the sign of the variation (−/+), an absolute value of DID_norm < 1 reports a decrease/increase in the gap between T and NT below 100%, while an absolute value of DID_norm > 1 means an increase/decrease in the gap above 100%. An example might help: −2.44 in Y means that, in T, Y_i was reduced far more than NT with a resulting gap 3.4 times larger. In counterfactual terms, this would mean that the variation in T would be 3.4 times smaller without the treatment. Of course, if a reduction or increase in a specific outcome Y_i can be considered a good result of the experiment depends on the Y_i. Increasing share of DW is a great result, increasing the share of UW is not.

4. Results

This section presents the results of the DID. The effect of the treatment on each indicator is presented in Table 3 (parts a and b) where it is reported in three forms:

• Pre-post treatment, which reports the values of the indicator for T and NT before and after the implementation of treatment in order to provide information about the whole evolutionary dynamic, i.e., the starting point and the trends.
• DID, which reports the net effect of the treatment in absolute terms that is the difference between the change observed in T and NT. The magnitude depends on the Units of Measure (UM in the tables).
• DID-norm, which reports the net relative effect that is the change in the gap between T and NT.

Table 3 (part a) reports the effects of the innovation in the waste collection system on some environmental indicators. Based on the results, the treatment seems to have produced a generally positive effect, although with a few contradictory dynamics.

First, the total production of waste (TMW) decreases in T by around 60 kg per capita from 2019 to 2023, a good result that is actually even better if compared with the NT areas where, on the contrary, TMW increases in the same period. The net effect of the treatment is therefore a reduction of waste in production in T of 3.5 times compared with what it would have been without the treatment. This effect is even higher for the fraction of unsorted waste (UW) that was halved in T (from 204 to 103 kg) while it remains almost constant in NT, resulting in a reduction of UW in T areas that produced a gap with NT four times larger than before the treatment.

The treatment also triggered a slight positive difference in the T production of differentiated waste (DW) compared to NT. DW increases in T by around 40 kg per capita, a variation that allowed T areas to fill the gap with NT. This positive variation in the quantity of the DW fraction (and in the reduction of UW) is even more clearly highlighted by the ratio between DW and UW, i.e., what we called separating attitude, that registered the highest positive net effect, with T areas improving the attitude far more than NT with a new gap six times larger than before.

In the face of these positive quantitative results, slight negative effects of the treatment seem instead to be produced on the quality of DW. As for the organic fraction of DW, the treatment seems not to have produced any effect while a negative impact is detectable for the fraction of DW that, based on the ANCI-CONAI incentive scheme, can be economically valorized, namely plastic, paper and glass. To summarize, from an environmental perspective, the innovation in the waste collection system seemed to have produced relevant positive changes in the reduction and composition of waste production. The combination of a general reduction with the increase in DW can be surely considered a great effect. On the contrary, some unclear and apparently not positive effects seemed to have been produced in the quality of the DW, at least for what concerns the fractions that are the most suitable to be recycled, thus accelerating the linearity of our socio-economic model.

Table 3 (part b) then reports on the effects of the innovation in the waste collection system on some economic indicators. Contrary to the findings regarding the environmental dimension, the effect of the treatment on the economic dimension seemed to be not fully positive, with a general increase in the costs for both the collection and the treatment of the different fractions of waste. Of course, innovation itself requires human and financial resources that naturally increase the costs of the service in the short term. But, still, the distribution of these increases raises some points of attention about the economic performance of the innovation itself.

First, being treated means registering an increase in the total cost of the service (collection and waste treatment), indicator 14, of around 16 € per capita more than not being treated, for a total negative effect of the innovation estimated in an increase in the gap of around 1/3 (0.31). In absolute terms, this difference is mostly due to a net increase in the costs for DW collection (indicator 9) per capita of around 8 € in T areas compared to NT, while, as for the costs of collection of UW (indicator 7), the effect is almost irrelevant, even though in T areas UW decreased more than 50%. This apparently contradictory result (constant cost for a service decreasing in size) could be explained by the lack of elasticity of the costs to the public sector that often lag when adapting to changed conditions, and the distribution of the fixed costs of the project that, although aimed at intervening on DW, are probably included in the fixed costs of the whole service. Moving from the collection to the treatment, the results are reversed, although on a lower level of intensity, with the costs per capita of DW treatment increasing in T slightly less than in NT (+0.7 € for an increase in T of almost half that of NT), while the costs for treatment of UW per capita decreasing in T less than in NT by around 6 €, although in absolute terms they are really close (12 € in T vs. 11.5 € in NT). The fact that the cost of unsorted waste (UW) treatment per kilogram decreased by almost 200% in both groups (T from 34.5 to 12; NT from 39.7 to 11.5) between 2019 and 2023 is significant, even though only T municipalities reduced their production of unsorted waste. This data should be interpreted in conjunction with indicator 13, which shows the cost of UW treatment per kilogram. Citizens in T municipalities pay 0.26 € per kg of UW, whereas those in NT municipalities pay 0.13 € per kg. This apparent paradox arises because ACEA does not allocate UW treatment costs solely based on the amount of waste produced. Instead, cost distribution is influenced by political and contextual factors, which will be analyzed in the discussion section.

Being treated means, therefore, to have registered in the period 2019–2023 a higher increase in costs for waste collection, in particular of DW collection. The latter is partially compensated in T areas by the decrease in the costs for DW treatment, while a contemporary slower decrease in the costs for UW treatment is detected. All these cost variations per capita are confirmed when computed per kg, although the magnitude of the effect is, of course, far lower in absolute terms—although it might be really informative in relative terms, i.e., considering the DID norm. The costs of the UW treatment in T (indicator 12), in fact, seemed to get out of control by registering an increase in the cost-per-unit that resulted in a new gap with the NT performing far better than T. The magnitude of this variation is just a matter of math, while for the explanation more information would be needed. The economic performance seems not to be satisfactory if we also look at income that might come from the valorization in the framework of the ANCI-CONAI incentive schemes (see Section 1). Indicator 16 shows how economic valorization, and the quality of DW, decreases for both T and NT in the period, but in T this decrease produced a negative gap with NT 3.5 times larger than before.

In summary, in the short term, the innovation of the waste collection system seems not to have had a positive economic effect on the treated municipalities. And this is true both with respect to the costs and expenses of waste collection and treatment (the former due to a relative increase in the cost of DW collection, the latter due to a lower decrease in the costs of UW treatment) and with respect to the potential income, with NT areas performing relatively better in terms of valorization within the ANCI-CONAI incentive schemes.

Based on the results of the analysis, the innovation of the waste collection system seemed to have triggered a number of changes and to have actually made a difference in the pathways followed by treated and not treated areas. However, these changes are contrasting when considered from the perspective of the transition of the waste management system towards a more sustainable model. On the environmental side, the treated areas showed a number of improvements in the production and composition of urban waste, with the share of DW on total waste being greatly improved, although the quality of DW remained quite low. On the economic side instead, the innovation seemed to have triggered a general increase in the cost and expenses of collection and of the UW treatment, and a worse performance in terms of economic valorization of DW. The latter confirmed that the innovation has not yet been able to trigger the expected and needed quantitative and qualitative improvement to the waste management system.

5. Discussion

The main results from the DID analysis are shown in Figure 2 where the value of the normalized outcome (DID-norm) for the economic and environmental indicators considered in the analysis are reported, thus providing an overview of the general effect of the innovation of the waste collection system on the T areas.

A general look at the figure highlights a relationship between the introduction of techno-organizational apparatuses typical of door-to-door collection in a street collection system, and a relevant change in the systemic behaviors in T.

The three most important changes that seem reasonable to trace back to the treatment are: (1) a decrease in total waste generation and UW fraction in particular; (2) an almost double increase in DW in T compared to NT, and a contemporary decrease of quality of the DW in T; (3) an increase in the cost of service in T.

As for the first, the total amount of waste per capita between 2019 and 2023 (indicators 1, 2, 3), shows a relevant decrease in T municipalities compared to NT, in particular with regard to the UW fraction (indicator 2), that decreased about 50 percent in T, while it will increase by about 5 percent in NT. In absolute terms, the 106 kg of UW per inhabitant in the treated municipalities is a remarkable performance compared to the regional average (163 kg/inhabitant), although it is above the threshold of 75 kg per inhabitant considered by the Zero Waste movement as the expected performance for a “recycling” municipality [34].

Secondly, it must be noted that the reduction in UW is not counterbalanced by a contextual increase in the production of other types of sorted waste (indicators 3, 4, 5) that would have suggested an absorption effect. Organic, plastic, paper and glass increased incrementally in both treated and untreated municipalities. Indicator 6 (ratio DW/UW) shows that in T municipalities DW is closing the gap with UW by increasing the ratio to 73%, while the control group NT almost maintains the same ratio as before (53%). The relative net effect is therefore the gap between T and NT in 2023 being 4.5 times larger than the gap in 2019. A good effect in itself, that should be, however, reported to the reduction in UW as the main explanatory factor, rather than by an increased capacity for sorting and valorizing DW. This lack of improvement in the quality of DW is also confirmed by the worse performance of T areas in terms of economic valorization. This is clearly shown by indicator 12: in 2019, the treatment of one kg of undifferentiated waste cost the treated municipalities 0.15 cents, while in 2023 it costs 0.26 cents; in the untreated municipalities, although the quantity of UW increased by 5%, the cost per kg remained unchanged at 0.15 cents.

A comprehensive assessment of the economic impact of a techno-organizational innovation in a sector like waste management requires a long-term perspective while the case study allows for the observation of only the first two years [35]. However, it is reasonable to expect that, in the long run, T municipalities may experience a reduction in collection-related investment costs—primarily due to the amortization of expenses for the purchase and installation of smart bins—a long-term perspective alone cannot fully explain the dynamics of treatment costs. Given the strong environmental performance of T municipalities (−50% of UW in 2 years), such cost differences should already be apparent in the short term [36].

Despite some partial and uncertain results, the pilot program seems to have enabled the T municipalities to achieve a significant portion of the targets set by Directive 2008/98/EC and the Circular Economy Plan ahead of schedule. The overall reduction in waste has exceeded the 15% target established by European regulations for 2025. The recycling rate has surpassed 60%, the threshold set for 2030. The decrease in unsorted waste and the adoption of a comprehensive separate collection system have already allowed T municipalities—as well as NT municipalities—to cease direct landfill disposal, instead diverting waste towards energy recovery.

Moreover, although data on the capture rate of plastic packaging relative to the amount placed on the local market in T municipalities is not yet available, and despite challenges related to the financial returns recognized by Italian consortia, the quantities of separately collected plastic and metals increased between 2019 and 2023. However, these positive outcomes are not unique to the municipalities of Pinerolese. Across most of the indicators mentioned, Italy as a whole has already met—and in some cases exceeded—the targets set by the European Commission [37].

These results are not only significant in terms of waste management indicators but are also interesting from a sociological point of view, because behind them there is a collective behavior that has changed thanks to the introduction of a techno-organizational innovation and that has produced unexpected results. For these reasons, in the following, a few hypotheses are proposed about the socio-technical dynamics that might have generated these changes.

For what concerns the increase in the collection costs in T municipalities, the most likely hypothesis is a short-run direct effect of the depreciation costs on the introduction of the new service. However, this negative trend should have been counterbalanced by the economic benefits of a decrease in treatment cost of UW and an increased quality of DW in the medium and long run, as expected from the designer of the new service. Instead, DID analysis shows at least two paradoxes. First, in T, the cost of treatment per kg of UW has doubled, while in NT it has remained the same. Moreover, in T the economic performance of DW decreased more than in NT. The reason for the first paradox could be sought in the criteria chosen by the municipalities when drawing up the Economic and Financial Plan of Acea, where the distribution of the costs of treating urban waste does not appear to be related to the actual quantity of waste delivered to the incinerator, but aligned with political agreement between the municipalities. The second paradox cannot be explained by looking at the political decision of a small group of actors. This is mainly a product of collective behavior of users, as well as the reduction in UW. The explanation of these two changes might benefit from the theoretical approach outlined in Section 2, which integrates the Coleman boat into an STS perspective.

In this framework, the impact of the innovation on the economic and environmental performance of the waste collection system is mediated by the actions and interactions of the micro-agents. Moreover, this framework can be used to trace three main key drivers affecting the practices of micro-agents, namely: economic incentives, social control and awareness. Therefore, following the three logics of the Coleman boat, three main questions might be identified: how does hybridization reconfigure the structural conditions (technologies, infrastructures and organizational norms) and how do these changes impact on the drivers of users’ behaviors and practices? How do users adapt their strategies and practices to the new context? How do their individual actions aggregate to produce the observed results?

At this stage of the research, we cannot yet provide definitive answers to these questions. However, by drawing on the results of the quasi-experiment and interviews with key informants, we can formulate hypotheses on how interactions between macro- and micro-level dynamics have unfolded along the logics of Coleman’s boat.

5.1. Logic of the Situation (Macro–Micro)

Hypothesis 1:

The impact of hybridization on citizens’ behaviors arises from two key mechanisms. First, it impacts on economic incentives by introducing a mechanism, albeit sometimes only apparent, that connects lower fees to higher waste separation rates. Second, it strengthens social control by underscoring the strong commitment of the local public administration to improve waste sorting performance. The decisive role of communication campaigns and other informational processes (awareness) in these different forms of valorization must be empirically verified.

As for the economic incentive, system to access the waste bins via magnetic card was applied to residual, organic, and paper waste. However, a reduction in waste disposal was observed exclusively for residual waste. This phenomenon may be attributed to users’ expectations about the introduction of PAYT systems, where the waste fee is often linked to the number of uses of the magnetic card, i.e., the number of individual residual waste disposals. Notably, users appeared to be aware of how a PAYT system operates, despite the fact that such a system had never been implemented in the municipalities involved.

Regarding social control [36], the Pinerolese area is characterized by small and medium-sized communities and strong social infrastructure, such as the Waldensian Church, which reinforces relationships between citizens and local administrators. These connections could contribute to a high level of trust in institutions [37] and encourage the adoption of expected behaviors.

What is unclear is the role of awareness, intended as the knowledge and information actually available among citizens about the potential benefits of the hybridization. The communication campaign adopted by T municipalities before the introduction of the hybridization might have affected the emergence of the economic incentive and why it persisted even in the absence of a formal regulation institutionalizing it. In particular, empirical research will investigate whether the information and communication processes accompanying the introduction of the new collection system explicitly referenced the possibility of a future transition to PAYT schemes. Furthermore, empirical research will inquire if communication of local authorities valorized the environmental potential of the new system in terms of global effects and impact on local livelihood. The former would be able to increase the effects of the economic incentives, the latter the effect of social control.

5.2. Logic of Selection (Micro–Micro)

Hypothesis 2:

The implicit economic incentive and social control pushed users to adopt various behaviors, each driven by specific ‘good reasons’ and to a different extent positioned in a continuum compliance–non-compliance.

The micro–micro level examines how the new formal and informal rules and norms are internalized by individual actors, leading to behaviors that may diverge to varying degrees but remain rational, meaning they are consistent with the interests of the individuals involved. Rational adaptation to rules does not imply the final recipients to be compliant with such rules, and therefore these adaptation strategies can be roughly categorized as compliant and non-compliant. For both the categories the objective is to reduce UW, but the means might differ significantly. Among the non-compliant strategies, one possible option would be to dispose of waste in bins located in untreated municipalities, which are geographically adjacent to treated ones and not subject to tracking. Another potential alternative could be using waste bins in the main city Turin, located dozens of kilometers away but, nonetheless, the daily destination of many commuters from the Pinerolo area. Some users, without the inconvenience of searching for accessible bins, might have chosen to dispose of waste that was previously placed in the residual waste bin into other waste streams within their own household system, taking advantage of the lack of tracking. Among the compliant strategies, or at least with reference to users less inclined to engage in illegitimate practices, or simply those with greater resources (e.g., time, domestic space), users may have opted to improve their waste-sorting practices without altering the overall volume of waste generated. Finally, some users may have undertaken a more deliberate reassessment of their consumption patterns, intentionally reducing their production of unsorted waste. Empirical analysis will have to investigate these dynamics, and whether there are ‘stratifications’ in the behavior of users, on the basis of variables such as type (domestic and non-domestic), type of dwelling (large or small), level of education of the user, lifestyle and others.

5.3. Logic of Aggregation (Micro–Macro)

Hypothesis 3:

One pattern seems to emerge at the system level from the interaction of these compliant and non-compliant individual practices: the voluntary reduction of UW (compliant) and placing part of this UW in the plastic and metal container (non-compliant) by mistake, carelessness or convenience for reasons other than tracking.

The transition from the micro to the macro level is not limited to describing the output at the system level (reported in Section 4 of this article) but rather involves understanding how these outputs are produced by the interaction between individual behaviors. To date, we have not enough information to hypothesize all the mechanisms that define the waste production on a specific territory as the result of agents’ actions and interactions. We lack detailed information on the types of waste produced and disposed of by households and we lack information about their practices. But we have enough information to hypothesize emergent patterns, i.e., to cluster behaviors on the basis of the detectable results they are likely to produce at the macro level. Based on the behaviors hypothesized through the logic of selection, it is in fact possible to analyze how their combination has led to a reduction in unsorted waste and a deterioration in the quality of separate collection, particularly for plastics and metals.

The hypothesis we formulate is directly linked to the DID analysis, which tends to reject the idea that behaviors other than conscious and intentional reconsideration of waste production have played a predominant role. If the dominant behavior had been the disposal of unsorted waste in bins located in untreated municipalities, an increase in such waste would have been observed in those areas. However, the data (see indicators 1 and 2, in particular) show that this was not the case. Similarly, the hypothesis that users merely improved their waste sorting practices without changing their consumption habits, while plausible, does not appear to be confirmed by data: in treated municipalities, the increase in separate waste collection does not compensate for the reduction in unsorted waste (see indicators 3 and 4). These findings also contradict the hypothesis that non-tracked bins were systematically used as a relief valve for unsorted waste; although the quantities of plastic and glass have increased, they have not increased enough support this interpretation.

Furthermore, the data suggest that the deterioration in waste separation quality is not primarily due to the absence of tracking. As we gathered from interviews with privileged observers, while the quality of glass separation has remained stable, plastics and metals have shown a decline, even though their respective bins were not subject to tracking. Although not as pronounced as with plastics, a decline in quality was also observed for paper, despite its bin being tracked. This suggests that factors other than the normative aspect of tracking might have influenced the irregular use of bins. As Actor–Network Theory suggests, non-human actants [29] could also have had a relevant role. In empirical research, attention should be paid to the impact of bin opening design in shaping the emergent systemic pattern. For instance, in the case of glass, the opening of the bin is very small, which might have discouraged its irregular use compared to the larger opening of plastic/metal and paper bins. Moreover, the physical component of the waste could be another element in determining systemic behavior. A piece of organic waste or a glass bottle is not easily confused with undifferentiated waste, whereas waste related to plastic and paper, in particular food packaging, is much more so.

The hypothesis we aim to empirically test is that the emerging pattern so far is triggered by the combination of economic incentives, social control and technological agency, in a condition of law awareness. More practically, the emerging pattern is from agents that are pressured by the formal and informal imperative to reduce unsorted waste (logic of situation). This shared pressure results in varying user behaviors, leading to different levels of compliance (logic of selection). The aggregation of individual user actions generates a systemic effect, characterized, on the one hand, by a widespread reduction in overall waste and, on the other hand, by a decline in the quality of certain waste fractions. These fractions tend to be more closely assimilated to unsorted waste, either deliberately or unintentionally, especially when disposed of in bins with ‘welcoming’ openings that accommodate any type of bag (logic of aggregation).

These hypotheses highlight the extent to which the co-determination relationships among technologies, norms and human behaviors, which is one of the foundations of STS, can be made even more explicit when applied to the whole explanatory chain derived from analytical sociology.

In the logic of the situation, an implicit norm (taxing mixed waste on the basis of deliveries), induced by a technology (the magnetic card to track this type of waste), within a socio-cultural context characterized by a high degree of closeness among citizens and with their representatives, defined the environmental conditions for the agents to adopt a behavior aimed at reducing the load of mixed waste. Such behavior, in a situation characterized by incomplete information, could vary a lot depending on the effects that the local perception and interest of each actor and the formal and informal pressure of the surrounding social group (logic of selection) might have on technology adoption and use. This specific pathway of socio-technical innovation (hybridization) deployment, embedded in emerging behavioral patterns, produced as a result not only an extraordinary reduction in the production of mixed waste, but also a worse selection of plastics and metals (logic of aggregation). We are not sure if the final result was the one expected by the hybridization promoters, but there is no doubt that the resulting innovation pathway and the outcome in terms of economic and environmental impact are the result of a situated interaction among formal and informal rules, technologies and human factors, i.e., just one of the unpredictable pathways of innovation diffusion that might emerge from the complexity of socio-technical assemblages. Lastly, it deserves to be noticed that the Coleman boat can be conceived as a ’close cycle’, where feedback dynamics are in place: micro-actions generated by standards and technologies create the conditions for new standards and technologies, the purpose of which will be to correct negative aggregate effects and maintain positive ones. One example of this is the introduction of a pay-as-you-go tariff system, which avoids a possible loosening effect of the mixed waste contraction device, and a communication campaign that adequately informs citizens on how to properly separate plastics and metals.

6. Conclusions

In this article, we focused on a “hybrid” model of waste collection, defined as a model which combines elements and mechanisms typical of two well-established approaches to waste management: door-to-door (DtD) and street collection (SC) systems. By adopting a quasi-experimental research design, this paper investigated the effects of the introduction of this hybrid model by comparing the economic and environmental performance of the waste management system between a group of municipalities in Piedmont region that introduced a hybrid collection model (the treated T) and a comparable group of municipalities belonging to the same geographical area that maintained a traditional system (NT). The results were ambivalent and, in some cases, unexpected. In the treated municipalities, the production of unsorted waste dropped sharply. While separate waste collection increased, this was only partially due to improved sorting by users. On the one hand, the increase in separate collections did not fully compensate for the reduction in unsorted waste. On the other hand, the quality of waste sorting declined for certain fractions, particularly plastics and metals, with negative economic consequences. Finally, collection and treatment costs increased in the treated municipalities to a greater extent than in the untreated ones.

To explain these results, we proposed a few hypotheses derived from a theoretical framework that integrates Science and Technology Studies (STS) with analytical sociology (AS). In other words, we tried to develop and operationalize the core concept of the STS tradition on the co-determination of technologies, regulations, and individual behaviors through the three micro–macro interaction dynamics proposed by AS: (1) macro–micro (or the logic of the situation), where waste collection technologies and regulations shape constraints and opportunities for individual actors; (2) micro–micro (or the logic of selection), where the repositioning induced by new technologies and rules leads actors to adopt different behaviors, each guided by “good reasons” from their own perspective; and (3) micro–macro (or the logic of aggregation), where the aggregation of micro-actions—whether coordinated, conflicting, or negotiated—produces systemic effects that, at least potentially, redefine the conditions for future regulations and technologies.

The application of this theoretical framework to the empirical evidence derived from the data analysis allowed us to formulate hypotheses regarding these mechanisms. In formulating these hypotheses, primary attention was posed to the role of micro-social agents as carriers and interpreters of innovation and as transformative agents affected by and affecting the system as a whole. The aggregate economic and environmental performance of the hybrid model has been therefore conceived as the observed outcomes resulting from patterns emerging from the actions and interactions of micro-social agents. Three main hypotheses were defined to open the ‘black box’ of the systemic effects of the socio-technical innovation represented by the hybrid model. First, regarding the logic of situation, the impact of hybridization on citizens’ behavior arises from two key mechanisms: the impact on economic incentives of the mechanism that connects lower fees to higher waste separation rates; the impact on social control due to the strong commitment of the local public administration. The second hypothesis refers to the logic of situation and proposes to conceive the final behavioral output of micro agents as the result of the combination of the implicit economic incentives and social control triggered by the ‘good reasons’ of each agent which, to a different extent, position itself in a continuum of compliance–non-compliance. The third hypothesis refers to the logic of aggregation and highlights the emergence of one pattern at the system level from the interaction of these compliant and non-compliant individual practices: the voluntary reduction of UW (compliant) and placing part of this UW in the plastic and metal container (non-compliant) by mistake, carelessness or convenience for reasons other than tracking.

Further empirical research is required in order for these hypotheses to be tested. The sharp decline in unsorted waste in the treated municipalities may be explained by economic incentives: although a PAYT (Pay-As-You-Throw) system has not yet been implemented, the introduction of infrastructure designed for such a purpose may have encouraged virtuous behaviors aimed at waste reduction. At the same time, the imperative to minimize unsorted waste may have led users to “shift” waste into plastics and metals, the fraction most similar to unsorted waste, thus reducing its economic value.

At no point did this study aim to generalize the findings from a specific territorial context. The STS framework itself teaches us that the trajectories of innovation diffusion result from an unpredictable combination of the socially and technologically situated components, where, by ‘situated’ we mean rooted in a unique socio-institutional and cultural context. Analytical sociology offers a conceptual lens for this unpredictability to be grasped in terms of the complex dynamics that contribute to determining the overall changes at the system level. However, the theoretical and analytical framework proposed in these pages could prove useful for analyzing other waste collection systems and contribute to a more interdisciplinary approach to waste management.

These research results ultimately allow for the provision of some operational proposals to local decision-makers regarding a possible revision of waste collection systems.

The use of an individual waste tracking system and the adoption of containers with smaller openings for non-recyclable waste have shown encouraging results in reducing this type of material. This approach should be maintained and extended to other types of waste, particularly plastic and metals, and accompanied by a communication campaign aimed at informing citizens precisely on how to separate waste correctly. Indeed, these are the best strategies to address the issue of declining quality in waste intended for recovery.

Even more significant reductions in non-recyclable waste could be achieved with the implementation of the PAYT (Pay-As-You-Throw) system. Although effective, the incentive—or rather, the “fear of financial penalties” induced by the tracking device on non-recyclable waste—may not be long-lasting. To consolidate this result, users must see their higher or lower compliance with the service reflected in their waste collection bills.

Similarly, the distribution of service costs should be more closely based on the performance of individual municipalities. As highlighted in previous sections, the municipalities involved in the study have experienced increased collection and treatment costs because they funded the investment in new technological devices. However, they have not benefited from the cost reductions resulting from lower non-recyclable waste disposal. This is because Acea distributes treatment costs among all member municipalities based on criteria that are not exclusively linked to the amount of non-recyclable waste produced. The adoption of a tariff system more accurately reflecting the reality of waste flows in each municipality will lead to a rebalancing of overall costs (in particular, an increase in treatment costs for municipalities not involved in the system and a decrease for those that are). This, in turn, will make the expansion of the initiative across the entire territory more urgent.