Master’s thesis:
Community response to a door-to-door
waste management and environmental
education program A case study of
Catalonia
Jordi Pascual Torner
10490744
Msc. Business Economics – Managerial Economics and Strategy
July, 2018
Supervisor: Dr. E. Öztürk
Credits: 15 ECTS
Abstract
This thesis develops an empirical research exploiting the variation arising from the different timing of adoption of a door-to-door waste management system and a sustainability education program across the region of Catalonia to evaluate how these policies have changed household recycling rates. The aim of this study is to shed light on how individuals can effectively be incentivized to be an active part in improving the recycling rates. The waste management system makes recycling easier for the citizens and the education program influences the attitude towards recycling, therefore both interventions can potentially increase the recycling rates. In municipalities where the waste management system was adopted the recycling rates increased by 24.55% on average, whereas municipalities that implemented the “green school” education program increased their recycling rates by an average of 1.32%.
This document is written by Student Jordi Pascual Torner who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
1. Introduction
The 2018 Circular Economy Package1 proposed by the European Commission and approved by the European Parliament has ambitious objectives regarding waste generation, which is the proof that a transformation in the management and treatment of waste is taking place. A “Waste Management Hierarchy”2 is set, where in order of priority, waste management systems have to: prevent, re-use, recycle, recover and dispose. We are witnessing the creation of a new paradigm, which demonstrates that the traditional linear growth model of production, consumption and waste generation is incompatible with a growing demand for resources, as Peter Lacy and Jakob Rutqvist mention in their book Waste to Wealth (2015). Our economy, society and environment are threatened by a linear model that for many decades has been the key milestone for prosperity and economic growth, regardless of its environmental impact. These threats and risks have set off the alarm bells at multinational firms and national governments, putting forward the circular economy model as the main alternative for the future. Resource constraints, technological development and socio-economic opportunities are driving this transformation process (Peter Lacy and Jakob Rutqvist 2015).
Cutting down the amount of waste generated and increasing waste separation for re-use and recycling purposes are the main short-term goals of Circular Economy. One of the main challenges to meet these goals is to find ways to incentivise individuals and organizations to adopt a co-operative and environmentally-responsible behaviour. Citizen engagement in waste separation is the main barrier to up-scale the amount recycled and improve the efficiency of waste management systems. Hopper and Nielsen (1991) make a very accurate description of this barrier when they describe the costs of recycling in terms of time and energy in order to save, sort, and deliver the recycled materials. The authors add an important reflection: “there are no immediate or individual rewards for recycling, yet surely it will benefit society as a whole, especially in
1http://ec.europa.eu/environment/circular-economy/index_en.htm 2Directive 2008/98/EC on waste, “Waste Framework Directive”
the future”. Research has tried to analyse how to overcome this barrier by trying to elicit which interventions are the most effective in promoting pro-environmental behaviour. These interventions are based on the analysis of factors influencing individual recycling behaviour: the internal processes related to attitude and awareness (Hopper and Nielsen, 1991) and the external factors that incentivise individuals to co-operate (Jenkins et al., 2003). The former, focus on interventions such as education or persuasion (Guagnano et al., 1995), whereas the latter focus on interventions related to taxes and fees (Reschovsky and Stone, 1994) or in making recycling more convenient (Derksen and Gartrell, 1993). Both types of interventions are intended to influence individual attitudes and beliefs on the environment, with the objective of modifying their recycling behaviour. This idea was first introduced and studied by Schwartz (1981) in his normative decision-making model of altruism.
A considerable amount of research in psychology and behavioural economics has focused on studying the mechanism of the Schwartz’s model, to determine the key factors driving behavioural change (by analysing the effect of the interventions) and evaluate which interventions are the most suitable for replication. However, both research streams have their pitfalls. On the one hand, as Guagnano et al. (1995) already suggested, few empirical studies have tried to integrate both the intrinsic and extrinsic factors in a single study in order to analyse the strength of each of these in the same context. Most of the studies focusing on the internal processes influencing attitudes, are based on the use of questionnaires, surveys or telephone interviews in relatively small sample sets (e.g., Oskamp et al., 1991), where behaviour is generally self-reported and not analysed using observational data. Corral (1997) analysed the differences between self-reported and observed behaviours and found a poor correlation between both data sets. Whereas studies focusing on the external factors that influence attitudes towards recycling, tend to use empirical analyses based on observational data and not internal factors. On the other hand, research that has extensively analysed the suitability of waste management systems, with the objective of making recycling easier, only focuses on the environmental and economic impact of the system, ignoring the impacts of its recycling behaviour (Pukkinen et al., 2012, Teerioja et al., 2012 or Iriarte et al., 2009).
How a waste-management system modifies its recycling behaviour should be fundamental when analysing its environmental and economic efficiency.
It is of crucial importance to try and address these drawbacks by developing empirical research, integrating both extrinsic and intrinsic determinants of recycling behaviour, and analysing the effects of specific interventions. In an attempt to analyse the extent to which each of these factors influence behaviour in a natural setting, the present study aims to establish the real impact of implementing an educational program and waste management system on recycling behaviour, which could lead to a mindset change on how Life Cycle Assessments (LCA) and Life Cycle Costing (LCC) analyse waste management systems. The waste management system of the present research is a door-to-door waste collection system that has its roots in the fact that citizens are responsible for the separation of the different type of waste in origin (at home). In this system, instead of throwing the waste away into multi-containers located in the street, households separated waste is directly collected in the generation point (house) according to a pre-established schedule and calendar. This program can handle the following waste materials: organics, general waste, glass, paper, cardboard, plastic, and even more advanced typology of separation categories. The main objective of the door-to-door waste management system is to increase recycling engagement of citizens, improve the quality of the waste separation and reduce the general waste category3. The system is mainly designed for urban areas of small and medium size. It can also be implemented in the big cities with old neighbourhoods with narrow streets. This is the reason why the data has been adapted to fit these characteristics, dropping those outliers with more than 100 thousand inhabitants, which are considered out of the scope of the door-to-door waste management system. On the other hand, the education program intervention includes and organizes educative actions with the objective of sharing environmental and sustainable values to students and tackle the new sustainable challenges.
Data on recycling rates, participation in a door-to-door waste management system, having at least one green school in the region and other control variables in 919 municipalities in Catalonia (Spain) since 2000 to 2015 have been collected. This panel
3 Source: “Manual municipal de recollida selectiva porta a porta a Catalunya, primera edició, Febrer 2008”
data set is a very interesting opportunity to evaluate the strength and real effect of both interventions. More specifically, this study aims to answer the following question: What are the real effects of implementing a door-to-door waste management system and an education program on recycling behaviour in a specific area and time span?
The main findings can be summarized in the following way. Municipalities that have implemented the door-to-door waste management system recycle on average 24.55% more than households in municipalities that have not implemented the system. When the effect is analysed yearly, it can be seen that recycling behaviour increases sharply in the first years after the adoption of the system and then starts to slow down over time, with the lowest increase occurring fourteen years after adoption. On the other hand, municipalities that have adopted the education program “green schools”, recycle on average 1.32% more than those municipalities that have not. The yearly analysis shows that the education program has a positive effect on recycling, the peak being thirteen years after adoption. The remainder of this thesis is organized as follow: section 2 goes through a literature review regarding the determinants and enablers of recycling behaviour and the studies focusing on the economic and environmental assessment of waste management system; section 3 describes the data set used and the methodological approach; section 4 exposes the results of the conducted regressions and its estimations; section 5 sets a discussion on the results and limitations of the analysis and makes the main conclusions of the study.
2. Related Literature
2.1 Applied behaviour literatureLiterature has extensively focused on analysing the role of external factorson recycling behaviour. These factors can be interpreted as monetary incentives or costs, like Thogersen (2003) analysed in a study of a pay-by-weight scheme for garbage collection. He carried out his study by means of questionnaires asking for self-reported levels of recycling in two groups made up of citizens from three different municipalities in Denmark, where one was assigned the pay-by-weight system and the other with a fixed fee for garbage collection. As mentioned before, self-reports do not necessarily
match the real behaviour of individuals. Thogersen found that the treatment group recycled significantly more than the control group, which was associated with the theory that this kind of intervention can become a moral norm that can, in turn, enhance internal motivation. Other studies like Hong and Adams (1999) analyse the role of price incentives for recycling, which find that households are sensitive to an increase in the price because they increase their recycling rates in order to avoid additional costs. The previously mentioned external factors do not only focus on incentives. Other studies have also considered external conditions as drivers to increase the ease of recycling. Reschovsky and Stone (1994) analyse the effect of pricing waste disposal systems in conjunction with curbside pick-up of recyclables in a natural experiment, finding that curbside pickup had the greatest effect on recycling behaviour. Additionally, as another example, Gitlitz (1989) analysed how changing external conditions can reduce household costs and inconvenience, resulting in considerably positive behavioural change.
Based on the literature that shows that making recycling easier enhances the recycling rates, the first hypotheses is formulated as follows:
H1: The door-to-door waste management system has a non-zero positive effect on the recycling rate.
2.2 Attitude Literature: internal processes
A vast proportion of the literature analyses the internal factors that affect pro-environmental behaviour, and especially recycling behaviour, which is based on the normative decision-making model of altruism (Schwartz, 1981). Schwartz’s model is based on the norm activation theory, according to which, an individual behaves altruistically, when he/she is conscious of the harm his/her actions may have and takes responsibility for them as a personal duty.
Hopper and Nielsen (1991) state that “the critical feature of altruistic behaviour is that while most people would verbally endorse a norm governing a particular moral behaviour, not everyone acts in accordance with the norm.” Indeed, this is true for
recycling. Past research (e.g. Burn & Oskamp, 1986) suggested that attitude change is very important to achieve substantial and lasting behavioural modification. Hopper and Nielsen (1991) tried to assess to what extent recycling can be considered an altruistic behaviour. They applied the altruistic model in the context of recycling behaviour as follows: although most people agree that recycling is positive for the society and the environment, a large number of them do not act according to their beliefs and do not recycle. In order to behave in favour of the environment, people must have a personal norm, which, if broken, generates guilt. However, the mere fact of having this norm does not necessarily mean people will follow it, as they still need to understand the consequences and accept the responsibility of not doing so. They will only act according to their beliefs, when this occurs. Altruism is, therefore, defined as a normative behaviour.
The authors developed an experimental field study with pre- and post-experimental surveys (n=122 individuals), in which they analysed the effect of prompting and informing individuals about recycling, aiming to find out what interventions had the highest impact on promoting altruistic behaviour (i.e. recycling). The study was developed in a neighbourhood in Denver, which was divided up into 4 groups based on area: 3 treatment groups and one control group. Each treatment group was assigned to a different experimental intervention to influence recycling behaviour: social interaction (“Blockleaders” encouraging neighbours to recycle and informing them about the program), prompting and information strategies, and a simple information strategy. The control group was not assigned to any experimental intervention. The results showed that social interaction had the greatest impact on recycling behaviour and was the only intervention that increased social and personal norm scores, followed by prompting and information, and the simple information strategy, which had the weakest effect. Finally, to show the relevance of awareness of the consequences (AC), they showed that if AC was high, personal norm scores and behaviour were highly correlated, whereas if AC was low, the correlation was also low.
Literature on recycling behaviour has extensively focused on arguing that the strength of the relationship between attitude and behaviour is mainly driven by intrinsic motives, which are influenced by interventions focusing on changing the attitude of individuals towards a pro-environmental behaviour. Prior to Hopper and Nielsen (1991), De Young
(1985) already analysed the role of intrinsic motivation in encouraging environmentally appropriate behaviour, showing that recycling and re-using were mainly driven by satisfaction and intrinsic motivation, rather than extrinsic incentives. Moreover, Vining and Ebreo (1990) analysed the main differences between recyclers and non-recyclers according to three main categories: knowledge about recycling and how they received the information, perception of the importance of the reasons for recycling or not recycling (monetary incentives, altruism or environmental concern and social influence) and demographical characteristics. 500 households chosen randomly were sent a questionnaire. Data was collected from 87 non-recyclers and 110 recyclers. The authors found strong differences between both groups in the three categories and claimed a need for more education in the field of sustainability in order to create more awareness, pointing out that further research should examine factors influencing pro-recycling behaviours by increasing convenience, educational campaigns and overcoming the hyperbolic discounting bias4 (Loewenstein et al., 2003).
Based on the above, the second hypothesis is formulated:
H2: The green school program has a non-zero positive effect on the recycling rate.
2.3 Literature integrating attitudinal and behavioural theories
Literature has also tried to integrate both approaches in order to analyse real settings where both internal and external factors are influenced by different interventions. Guagnano et al. (1995) is the most relevant study to analyse both approaches as interactive factors influencing recycling behaviour with the aim to test the predictive ability of the Schwartz model of altruistic behaviour (Schwartz 1968a, 1968b, 1970, 1973, 1977) in response to the interaction of both factors. Guagnano et al. (1995)’s main purpose is to study a model that integrates both approaches, analysing how the interaction between both of these has an effect on recycling behaviour. They build on an “A-B-C model”: Attitudes (A), Behaviours (B) and External conditions (C).
4 Behavioural bias that causes human beings to weight present payoffs non-rationally more than future payoffs. Also known as present bias.
They argue that in a population there are distributions of A and C for any behaviour. If A+C>0 then the behaviour appears, whereas if A+C<0 the behaviour does not appear. The effect of any intervention is greater when A+C is close to 0. They also included the Ascription of Recycling Responsibility (AR), as the direct causal effect for recycling behaviour. They run a natural field experiment: telephone interviews with 257 residents (randomly selected), 180 respondents. Self-reported recycling activities. They test whether an intervention to reduce inconvenience when recycling (curbside bins) attenuates the explanatory power of the attitudinal effects. Thus, they suggest that attitudes have a higher impact on recycling behaviour in those households not supplied with additional bins, whereas in those households provided with bins the attitude effect is attenuated. The results show that AR has a significant direct effect on recycling behaviour. A and C have no significant direct effect on recycling behaviour, but significant direct effect on AR. Possession of a bin has a significant direct effect on recycling behaviour, thus the findings suggest that extrinsic conditions have a direct and indirect influence on behaviour. The interaction effect of Bin*AR (households with a bin) on behaviour was almost 0. The origin and inspiration of the present study come from the ideas developed in Guagnano et al. (1995). Having access to a panel data set where the observations (municipalities) are treated with two different interventions (lowering household cost for recycling and the presence of an education program) and where there are data on recycling rates, provides an interesting analysis of the effects of the different treatments on pro-environmental behaviour.
2.4 Waste Management System Assessment Literature
Finally, it is very important to mention the literature related to the analysis of waste management systems, mainly LCA and LCC. It is worth starting with the study of Coll, Rieradevall and Domènech (2002) that analyses the effectiveness of the door-to-door waste collection system, which is part of the present study, in one of the municipalities (Tiana) that first entered the program. In this study, Coll et al. (2002) state that the door-to-door waste collection system is a very well-suited system for small-medium municipalities or for urban areas with narrow streets and that its main purpose is to maximize the original waste separation, which they confirm in their findings. They
find that the waste management system has higher levels of recycling than other municipalities with a traditional multi-containers waste management system, both in quality and quantity of recycling and in the reduction of the waste treated by means of incineration or landfills. Finally, they find that the system has no incremental economic costs compared to the traditional multiple-containers waste management system. However, the analysis is limited to one single observation.
As mentioned earlier, most of the literature regarding the assessment and comparison of waste collection systems does not take into account the effects of those systems on recycling behaviour. Iriarte et al. (2009) analyse waste collection systems in dense urban areas. They compare three main waste management systems: Mobile pneumatic, multi-container and door-to-door. They find that the door-to-door system has higher energy consumption in terms of urban transport. The environmental impacts of the system are mainly due to the gas emissions of the trucks because of longer collection routes. Therefore, by optimizing the collection frequency and using alternative energies for transportation it would be possible to make the system greener. The authors suggest that the door-to-door system is supposed to have higher recycling rates, which in principle should compensate the other results related to the environmental impact. Teerioja et al. (2012) find that the pneumatic system is more expensive than the door-to-door waste collection system They estimate that the pneumatic system is six times more expensive than a vehicle-operated system. Punkkinen et al. (2012) point out in their study that the pneumatic waste system has a worse environmental impact than the door-to-door system because it generates more air emissions. Nevertheless, at local level, where the pneumatic system is implemented, emissions are lower as there is less traffic (the emissions are transferred from the waste collection area to where the system components are manufactured, or the electricity is produced). As can be observed, none of these studies take into account the impact that a change in the external factors, through the waste management systems, have on the recycling behaviour, which in turn affects their economic and environmental assessment. Therefore, a closer study of the behavioural change is required.
Considering the above mentioned behavioural and attitude research, the current study can contribute to developing a new empirical research, which aims to analyse the real effects that have both the introduction of a door-to-door waste management system (external factor) and an education program (intrinsic factor) have on recycling rates.
3. Data and Methodology
The empirical research is conducted with data from the Catalonia Waste Agency (ACR)5 and the Catalan Institute of Statistics (Institut d’Estadística de Catalunya)6. The ACR longitudinal data has been constructed since 2000. It consists of 919 observations from2000 to 2015 and contains data on the amount of waste generated and recycling rate in a municipality level for the Catalonia region. Although Catalonia has a total of 947 municipalities, with a population of more than 7.5 million people (2018)7, the present study excludes 28 municipalities for the following reasons: nine outliers (high population density), four municipalities that had adopted the WMS or the GS treatments in 2000 and fifteen observations due to incomplete data in the explanatory variables. The Catalan Institute of Statistics data allows the set of control variables to be included: culture, education8, gender, income, participation in the elections, and tourism. All these variables are included in the model to control for exogenous and endogenous factors that can potentially have an effect on recycling behaviour. Vining and Ebreo (1990) mention the importance of demographic factors as determinants of recycling behaviour. Culture is a key driver of positive social and environmental behaviour. As recycling is a day-to-day activity, the participation (or abstention) in municipal elections and the political party in charge of the city council can potentially have an effect on recycling rates. Finally, tourism can be seen as an external factor that can have a negative effect on recycling behaviour, as a tourist might not have high incentives to be pro-environmental and the door-to-door waste management system does not directly apply to him or her. On the other hand, tourism might also be correlated with
5http://estadistiques.arc.cat/ARC/ 6 https://www.idescat.cat
7 https://www.idescat.cat/pub/?id=ep&n=9122#Plegable=geo 8 Data only available for years 2001 and 2011.
municipality richness and facilities that make recycling easier, making the effect, therefore, ambiguous.
From 2000 to 2015, 124 municipalities entered a door-to-door waste management system9 (WMS) and 285 municipalities introduced at least one green school10 (GS) in their region, of which 48 also adopted the waste management system. The remaining 560 municipalities did not adopt any of the policies and are included as controls. Thus, the panel data set seems ideal to test the effect that both policies have had on recycling behaviour over a sixteen-year period. One point worth mentioning is that the motives behind the adoption of the waste management system and the educational program are not random, which generates a potential threat of endogeneity and self-selection. The variables of interest included in the empirical analysis are the following:
• Recycling rate (dependent variable): Percentage of selective waste collection per municipality: total separated waste collection over total waste collection (separated waste + general waste). Municipal waste collected by public collection services that is properly separated and included in the recycling stream, which includes organics, paper, cardboard, glass, plastic, textiles and other minor types of waste
• Waste Management System (WMS: independent variable): Dummy variable set equal to 1 if municipality 𝑖 was part of the program in year 𝑡, and 0 otherwise. • Green School (GS: independent variable): Dummy variable set equal to 1 if
municipality 𝑖 had at least one green school in year 𝑡 and 0 otherwise.
Additionally, the following set of control variables are also included in the empirical analysis:
• Population: natural logarithm of the number of inhabitants per municipality.
9http://www.portaaporta.cat/ca/index.php
10http://mediambient.gencat.cat/ca/05_ambits_dactuacio/educacio_i_sostenibilitat/educacio_per_a_la _sostenibilitat/escoles_verdes
• Average income per capita in thousands of euros
• Percentage abstention in municipal elections: Municipal elections of 1999 (used for period 2000-2002), 2003 (used for period 2003-2006), 2007 (used for period 2007-2010), 2011 (used for period 2011-2014) and 2015.
• Number of public cultural facilities in the region: Number of public and private libraries per municipality. Other variables like theatres or museums were not available at a municipal level.
• University: Percentage of population having a university degree older than 16 years old in 2001 and 2011 per “comarca” (region size between municipalities and provinces), according to the following 5 categories: “does not read or write”, “no studies”, “primary education”, “secondary education” and “high school education”. Despite the data only being complete for two years, an assessment of the effect of this variable on the analysis showed it was a meaningful measurement for the study and was, therefore, included. Data for 2001 is extrapolated for the period 2000 to 2010 and data for 2011 is extrapolated for the period 2011 to 2015.
• Women: Gender variable indicating the natural logarithm of the number of women in a municipality. As the absolute population is included as a control, only one gender should be included.
• Tourism: Number of beds available for touristic purposes in a municipality. It includes hotels, camping and rural houses.
The aim of the study is to empirically analyse the effect of both treatments on recycling behaviour. Initially, a fixed effects model was chosen. However, municipalities did not enter the programs at the same time. To capture the dynamic response of the recycling rate on the implementation of the waste management treatment on the one hand, and the introduction of green schools in the municipality on the other the empirical analysis is extended to the evaluation of both effects through a staggered difference in the
model. This model perfectly matches the characteristics of the panel data set used in this study and has been previously implemented in previous research such as in Stevenson and Wolfers (2006), where in a very similar setting they studied the effect of unilateral divorce laws on suicide, domestic violence, and homicide rates for women and men in the United States. Like in the present study, the treatment (passing the law) was adopted in different points in time, what is being taken care of by using the staggered difference in differences model. With this model, the aim is to evaluate the effect of the different treatments for every year following its adoption. The focus is on how the natural variation arising from the implementation of the waste management system and the education program at different points in time has an effect on municipality recycling rates. The treatment variables are defined as follows: a series of dummy variables11 set equal to one if municipality 𝑖 had implemented the treatment 𝑘 years ago and set equal to zero otherwise. The use of dummy variables is further explained in the results section. Tables 1 and 2 show the summary statistics of the variables included in the study. As already mentioned, a potential problem of the data set is that the treatments are not assigned at random, although there is not a homogeneous allocation criterion as many factors are influencing the decisions. On the one hand, table 1 describes the consistency and reliability of the data used in the study. A baseline comparison of means is made in year 2000, period in which none of the treatments is assigned. Variables Anytime WMS and Anytime GS are dummy variables set equal to 1 if municipality 𝑖 had ever adopted the treatment between 2001 and 2015 and set equal to 0 otherwise.
11 Thus, for every observation and period fifteen dummy variables are created to capture the number of years after the adoption.
Table 1
BASELINE SUMMARY STATISTICS (2000)
Standard deviations in parentheses Two-sample t-test on difference of means
* p < 0.10, ** p < 0.05, *** p < 0.01
Three different comparisons are made at baseline in order to check for significant differences between the different groups in the sample. The main concern is regarding the differences in the variable of interest, recycling. Table 1 gives some first indications of the suitability of the data set for the analysis of the present study. The first two columns indicate that no significant difference is present between the recycling rate means of municipalities being part of the waste management system and those in the control group. Similar results are obtained for the green school treatment, with a non- significant (at a 5% level) mean difference of 1.8%, and for those municipalities being in both treatments and those not being in any of the programs (columns 5 and 6). Therefore, the t-test statistical significance test for the dependent variable suggests that no baseline differences are present between the different treatment groups and the control groups, which reinforces the subsequent regression analysis presented in the results part of the study.
Nevertheless, some significant differences are present regarding the control variables, which later on will be confirmed in some parts of the results analysis. The mean abstention rate in the municipality elections is significantly different at a 1% level
Overall 0 1 0 1 0 1 Reycling 10.46 10.34 10.08 11.16* 10.09 10.78 10.44 (7.14) (6.25) (6.75) (7.50) (6.88) (7.02) (7.02) lnPopulation 6.87 6.87 6.33 7.95*** 6.31 7.49** 6,87 (1.56) (1.17) (1.26) (1.38) (1.30) (1.14) (1.51) Incomepercapita 12.02 12.01 12.14 11.76*** 12.14 11.77** 12.02 (1.01) (1.20) (1.00) (1.08) (0.95) (0.95) (1.04) ElecAbst 30.49 26.03*** 28.90 31.86*** 29.48 28.18 29.89 (9.95) (9.05) (9.77) (10.01) (9.66) (7.81) (9.95) Culture 0.56 0.42 0.28 1.07*** 0.28 0.63* 0.54 (1.42) (0.66) (0.76) (1.97) (0.78) (0.77) (1.35) University 9.89 9.83 9.78 10.08 9.74 9.55 9.88 (4.52) (4.55) (4.39) (4.78) (4.37) (4.55) (4.52) lnWomen 6.20 6.21 5.64 7.32 5.63 6.85 6.2 (1.60) (1.19) (1.29) (1.40) (1.33) (1.17) (1.55) Tourism 532.52 168.76 248.61 952.75*** 261.57 193.25 483.83 (2377.51) (394.38) (1231.60) (3381.37) (1304.78) (458.87) (2220.63) N 796 123 612 307 540 51 919
between both treatment groups and the control groups. This result could suggest a potential source of self-selection bias regarding a certain social attitude related to the public participation in the elections, which can translate into a wide range of underlying motives, attitudes and behaviours.
The comparison between the groups in columns 3 and 4 shows the highest differences. Although the mean recycling rates are very close, most of the other variables suggest that municipalities that have adopted a green school at any point in time have a significantly different profile compared to the control group.
Table 2 presents a different summary statistic of the data set. Although it focuses on the same variables, the time span includes all the observed years (from 2000 to 2015) and analyses the different groups with a different approach. Unlike the dummy variables Anytime WMS and Anytime GS in Table 1, variables Waste Management System and Green School are set equal to 1 if municipality 𝑖 is treated in year 𝑡 and are set equal to 0 otherwise. Thus, these dummy variables only take into account those observations where the treatment is being adopted.12 Comparing the results in Table 2 to those in Table 1, it is possible to see the strong impact that the programs have had on the treated municipalities. Focusing on the dependent variable of the study, the mean difference in recycling rate between the treatment groups and the control groups is high and very significant. When the door-to-door waste management system is implemented in a municipality, recycling rate is on average, a 35.04% higher than that in a municipality that has not implemented the program. When analysing the green school program, the recycling rate of a municipality which has at least one green school is 13.15% higher than in a municipality without a green school. Finally, when focusing on those 51 municipalities that are part of both programs (a total of 315 observations), the average recycling rate is, on average, 62.68% vs 23.01% in municipalities that are not part of any of the programs at any point in time.
12 Example: If municipality 𝑖 adopted the program in year 2008, the dummy variable is set equal to 1 from 2008 to 2015. For periods 2000 to 2007 the dummy variable is set equal to 0.
Table 2
SUMMARY STATISTICS FOR THE WHOLE SAMPLE
Standard deviations in parentheses Two-sample t-test on difference of means
* p < 0.10, ** p < 0.05, *** p < 0.01
This analysis of the data has a) shown that this sample set is of interest when analysing the real effect of a change in the extrinsic and intrinsic motives driven by the implementation of a new waste management system and an education program making recycling easier and b) provided a first evaluation of the effect of those programs on creating highly significant differences between the groups across the observed time span.
4. Results:
The staggered difference in differences model in Table 3 is run to test hypotheses H1 and H2. This model allows for a detailed yearly analysis of the effects of the waste management system and the green school program on recycling rates. It also captures the staggering effect of both treatments, which is a result of the different timings of adoption since 2000 to 2015. Overall 0 1 0 1 0 1 Reycling 24.51 59.55*** 24.83 37.98*** 26.02 62.68*** 26.81 (14.89) (16.25) (16.49) (17.59) (16.50) (15.36) (17.31) lnPopulation 7.02 7.33*** 6.79 8.44*** 7.02 7.94*** 7.04 (1.60) (1.24) (1.48) (1.40) (1.58) (1.17) (1.58) Incomepercapita 14.93 15.82*** 14.91 15.42*** 14.97 15.66*** 14.99 (2.10) (1.97) (2.11) (2.02) (2.12) (1.38) (2.11) ElecAbst 30.15 27.94*** 29.16 34.78*** 29.97 32.24*** 30.01 (10.34) (10.40) (10.20) (9.96) (10.36) (10.12) (10.36) Culture 0.56 0.52 0.42 1.34*** 0.55 0.84*** 0.56 (1.33) (0.73) (1.00) (2.19) (1.30) (0.92) (1.30) University 10.11 10.22 10.02 10.65*** 10.12 10.17*** 10.12 (4.61) (4.70) (4.55) (4.98) (4.62) (4.72) (4.62) lnWomen 6.30 6.61*** 6.07 7.74*** 6.30 7.23*** 6.32 (1.62) (1.26) (1.50) (1.41) (1.61) (1.18) (1.60) Tourism 528.97 138.59*** 434.08 894.45*** 509.27 235.30*** 503.40 (2365.52) (338.60) (2090.57) (3163.76) (2313.83) (500.85) (2290.42) N 13741 963 12490 2214 14389 315 14704
To test H1, an OLS is employed to estimate first of all: (1) 𝑅𝑒𝑐𝑦𝑐𝑙𝑖𝑛𝑔 𝑟𝑎𝑡𝑒12 = 𝛽5+ 7 𝛽8𝑊𝑀𝑆18 + 𝛽<𝐺𝑆12 + 7 𝛼1 1 + 7 𝜆2 2 + 𝑄𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀12 8
𝑊𝑀𝑆18 refers to a series of dummy variables set equal to one if a municipality 𝑖 had implemented the door-to-door waste management system 𝑘 years ago. Coefficients reflect the percentage change in the recycling rate due to the implementation of the WMS the stated number of years ago. 𝐺𝑆12 is a dummy variable set equal to 1 if municipality 𝑖 had at least one green school in year 𝑡. 𝛼1 represents individual fixed effects, which account for the time-invariant unobserved effects of municipalities. 𝜆2 refers to year effects, allowing to control for the time-variant unobserved effects. Finally, the set of control variables is added into the staggered difference in differences model.
Second of all, the estimation focuses on the following regression: (2) 𝑅𝑒𝑐𝑦𝑐𝑙𝑖𝑛𝑔 𝑟𝑎𝑡𝑒12 = 𝛽5+ 7 𝛽8𝑊𝑀𝑆18 + 7 𝛽8𝐺𝑆18 8 + 7 𝛼1 1 + 7 𝜆2 2 + 𝑄𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀12 8
This second regression reflects a similar logic compared to regression (1), but in this case the effects on recycling of both programs are captured by the same staggered difference in differences analysis.
𝐺𝑆18 refers to a series of dummy variables set equal to one if a municipality 𝑖 had at least one green school 𝑘 years ago. Thus, the same criterion is being used as with 𝑊𝑀𝑆18. This second staggered difference in differences regression should bring the necessary results to evaluate the accuracy of the hypotheses.
The first column of Table 3 reports baseline results without including the set of controls. The second column adds the set of controls, and the third column adds the 𝐺𝑆18 dummy variable as a new control, regression (1). Finally, the forth column reflects the results of regression (2), a staggered difference in the differences model including the effects of both treatments on recycling. All specifications control for individual and time fixed effects.
Models in specifications one, two and three of Table 3 are run to answer H1. The results favour Hypothesis 1. The waste management system program has a positive and highly significant effect on recycling rates in the year of adoption and all the years following the adoption of the WMS. H1: the door-to-door waste management system has a large positive and statistically significant effect on recycling rates after its implementation in the municipality.
The inclusion of the set of controls in the second specification has little impact on the fifteen coefficients of the WMS, capturing its yearly effect after its implementation. An important point to mention is the dynamic effect of the WMS that is captured by the model. Table 3 shows that the recycling rate grows following the change to a door-to-door waste management system, controlling for all the municipalities that in every specific year did not implement the WMS. The WMS has its highest effect only one year after having been adopted. After this peak, a decreasing trend can be seen in all specifications of Table 3, which has its lowest coefficient fourteen years after the adoption.
The third column of Table 3 includes the GS dummy variable as a control. The coefficients of the WMS barely change with the inclusion of the GS. The coefficient of the latter is positive but not significant, which suggests that that the program has no effect on the recycling rate of any municipality with at least one green school.
The following analysis focuses on testing H2. The assumption of a possible long-run positive effect of GS on recycling leads to the inclusion of the final specification of the staggered difference in differences model with both treatments as explanatory variables of the model, regression (2) previously exposed. This second approach to the model enables the evaluation of the effect of GS controlling for all those municipalities without a GS in every specific year, whilst still capturing the effect of the WMS on recycling rates. The forth column presents the specification including the staggering effects of both programs. The WMS coefficients show that the inclusion of the staggered GS coefficients has a small effect in their estimation compared to the first 3 specifications in Table 3.
Table 3
EFFECT OF A DOOR-TO-DOOR WASTE MANAGEMENT SYSTEM AND A SUSTAINABILITY EDUCATION PROGRAM ON RECYCLING RATES (PERCENT CHANGE)
Dependent variable: Recycling rate
(1) (2) (3) (4) WMSyearofadoption 25.2*** 25.2*** 25.2*** 25.2*** (1.40) (1.40) (1.40) (1.39) WMSoneyearlater 31.1*** 31.0*** 31.0*** 31.0*** (1.49) (1.48) (1.48) (1.48) WMStwoyearslater 29.1*** 29.0*** 29.0*** 29.0*** (1.41) (1.40) (1.41) (1.41) WMSthreeyearslater 28.8*** 28.7*** 28.7*** 28.7*** (1.35) (1.34) (1.35) (1.35) WMSfouryearslater 26.4*** 26.3*** 26.3*** 26.2*** (1.41) (1.41) (1.42) (1.42) WMSfiveyearslater 26.7*** 26.6*** 26.5*** 26.5*** (1.44) (1.45) (1.46) (1.46) WMSsixyearslater 25.9*** 25.7*** 25.7*** 25.6*** (1.44) (1.45) (1.45) (1.46) WMSsevenyearslater 24.2*** 24.1*** 24.0*** 24.0*** (1.71) (1.72) (1.73) (1.74) WMSeightyearslater 23.9*** 23.7*** 23.7*** 23.6*** (1.74) (1.75) (1.76) (1.75) WMSnineyearslater 22.2*** 22.0*** 21.9*** 21.8*** (1.67) (1.68) (1.69) (1.68) WMStenyearslater 22.5*** 22.3*** 22.2*** 22.1*** (1.79) (1.80) (1.81) (1.82)
WMSelevenyearslater 22.7*** 22.4*** 22.3*** 22.1*** (2.12) (2.12) (2.13) (2.15) WMStwelveyearslater 22.2*** 21.9*** 21.8*** 21.5*** (2.34) (2.34) (2.35) (2.40) WMSthirteenyearslater 22.4*** 22.1*** 22.0*** 21.5*** (2.51) (2.52) (2.54) (2.61) WMSfourteenyearslater 19.9*** 19.8*** 19.6*** 19.5*** (2.86) (3.20) (3.29) (3.41) GreenSchool 0.52 (0.63) GSyearofadoption 0.62 (0.59) GSoneyearlater 0.015 (0.66) GStwoyearslater 0.22 (0.69) GSthreeyearslater 0.71 (0.79) GSfouryearslater 0.13 (0.83) GSfiveyearslater 0.64 (0.86) GSsixyearslater 1.71* (1.03) GSsevenyearslater 1.81 (1.23) GSeightyearslater 1.24 (1.22)
Sample 2000-2015, n = 14704
Dependent variable is the municipality recycling rate per year. Coefficients are reported as the percentage change in the recycling rate due to the adoption of the waste management system or the green school education program the stated number of years later. Robust standard errors are in parentheses.
# Controls include the natural log of population, the average income per capita, the percentage abstention in municipal elections, the number of public cultural facilities, the percentage of the population with a university degree, the natural log of women and the number beds available for tourism.
Nevertheless, after nine years of adoption, the difference between the WMS coefficient in specification 2 and 3 gets larger year after year, until fourteen years after the adoption of the WMS. This last observation gives some idea of what can be seen some rows below in the staggered GS coefficients. The following analysis focuses on testing H2. The
GSnineyearslater 1.27 (1.22) GStenyearslater 1.46 (1.28) GSelevenyearslater 1.93 (1.41) GStwelveyearslater 2.69 (1.65) GSthirteenyearslater 3.98* (2.05) GSfourteenyearslater -1.32 (2.72)
Control variables yes yes yes
Individual and year fixed effects yes yes yes yes
_cons 10.4*** 21.7** 22.5** 22.6**
(0.28) (2.27) (2.27) (2.28)
N 14704 14704 14704 14704
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01
assumption of a possible long-run positive effect of GS on recycling leads to the inclusion of the final specification of the staggering difference in differences model with both treatments as explanatory variables of the model, regression (2) previously exposed. This second approach to the model enables the evaluation of the effect of GS controlling for all those municipalities without a GS in every specific year, whilst still capturing the effect of the WMS on recycling rates. The forth column presents the specification including the staggering effects of both programs. The WMS coefficients show that of the inclusion of the staggering GS coefficients has a small effect in their estimation compared to the first 3 specifications in Table 3.
Whereas WMS coefficients remain positive and statistically significant in all the years analysed, the GS coefficients remain positive but non-significant. However, it is interesting to observe that six and thirteen years after the adoption of the GS coefficients are significant at a 10% level. After a small positive and non-significant effect of GS on recycling in the first six coefficients, a slight change can be observed in the seventh coefficient (six years after the adoption). The staggered GS dummy variables have an upward tendency across the years and peak thirteen years after the adoption of the GS (3.98%). Even though the coefficient for those municipalities that adopted the GS 14 years ago is positive and non-significant, it must be interpreted with caution, as very few municipalities implemented the program in 2001 compared to 2002 and subsequent years, making the pool of observations too small to consider this coefficient relevant for the current analysis. Therefore, H2 is not supported by the results of the empirical analysis.
Based on the assumption that education has a long-term effect on pro-environmental behaviour in general, and specifically on recycling, the implementation of a staggering difference in differences model has been shown to be an accurate method to evaluate the real effect of both treatments over time. The effect of GS on recycling rates is small and positive, but non-significant. It increases across time in a steadily manner and has its peak in the long-run.
Once the hypotheses have been tested, it is worth analysing the trend of the effects of both treatments over time. Graph 1 and Graph 2 plot the different trends followed by
the coefficients for both treatments. The analysis aims to see if the patterns of behaviour suggested by the literature are supported by the data.
The main motives behind this reasoning are, on the one hand, education programs like green schools are long-term influencers of attitudes and future behaviours. This can be seen in studies like Dee (2004), where education is proved to have a positive impact on future civic behaviours in the long run, which translates in a civic return to society. Whereas, on the other hand, a staggering yearly closer look can give a more detailed vision of the effect of the waste management system, which in turn can justify the need to include the impact on recycling behaviour when assessing the economic and environmental impact of any waste management system.
Graph 1 shows how the door-to-door waste management system has a high effect right after its adoption and then starts decreasing steadily.
The rationale behind this result is straightforward: The recycling rate has its threshold at 100%, therefore it will asymptotically tend to that top during the whole process, slowing down over time as the room for improvement gets smaller and smaller. Only a structural change in the waste management policy, for instance including new materials in the recycling flows, or positive or negative externalities and shocks, could alter this long-run decline in the WMS coefficient over time.
Graph 1 10 15 20 25 30 35 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Co ef fic ien ts (% ) T
Door-to-door waste management system
On the other hand, Graph 2 shows how the green school education program has a high effect right after its adoption and then starts decreasing steadily.
The staggered GS dummy variables have an upward tendency across the years and have their peak in the long-run. The coefficient of 14 years after the adoption of the GS treatment is dropped because in 2001 very few municipalities implemented the program compared to 2002 and subsequent years, which makes the pool of observations too small to consider this coefficient as relevant for the current analysis.
Graph 2
Nevertheless, this study presents some limitations that are important to mention. Although the data set is very convenient for the analysis of the effect of two different treatments, as stated in the introduction, the allocation of the WMS and the GS is not at random, which can lead to a self-selection issue in the data set. It can be argued that those municipalities with a willingness to recycle or with some specific characteristics self-select themselves into any of the treatments. If this were to be the case, the effects of both treatments on recycling shown in the results would be biased and would not be reflecting the real effect that those policies can potentially have on a municipality recycling behaviour. By using individual and time -fixed effects and a staggered
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 11 12 Co ef fic ien ts (% ) T
Green school education program
difference in differences, this potential problem is in principle being minimized and the results can be considered as reliable and consistent. Table 1 in section 3 shows that there are no significant differences between municipalities in the WMS program and those in the control group. However, significant differences are present between municipalities in the GS group and the control group, which suggests a potential self-selection issue when analysing the real effect of the treatment on recycling. Despite the methodology used in the study, some endogeneity arising from this potential matter cannot be ruled out.
The idescat (Statistics Institute of Catalonia) data, although of great importance for the development of the analysis, lacks data on control variables that would enrich the quality of the study and make the estimation of the real effect of the explanatory variables more accurate. First of all, data on the average age of the population in the municipality, although available, is not suited for an analysis at a municipal level. Secondly, the political parties in charge of the city council every four years are not collected in a data set, which meant the data had to be retrieved manually by choosing the most voted party, although that party is not supposed to be the one in charge after the elections. Finally, and most importantly, data on attitudinal variables towards the environment were not available either. Such variables could be retrieved through periodical surveys regarding individuals’ attitudes and beliefs or by gathering data on other variables that could be used as proxies of attitude behaviour (e.g. number of electric cars or 𝐶𝑂G emissions). All those mentioned variables can potentially affect the recycling behaviour of citizens, which means they should be included if the data were available. Thus, an extended and more complete set of control variables could lead to different results in terms of the magnitude of the coefficients for the explanatory variables.
5. Discussion
The door-to-door waste management system is very effective in enhancing recycling rates. However, at the same time, as suggested by Guagnano et al. (1995), the education program “green schools” helps to improve recycling rates through awareness,
information and persuasion. The analysis of the effect of both treatments on recycling, has enabled to check, thanks to observational panel data, the strength of extrinsic and intrinsic motivations. To sum up, the study has tried to analyse the data according to the Schwartz normative decision-making model of altruism, focusing on two main factors affecting the attitude towards recycling and subsequently having an effect on the desired behaviour. The present study suggests that both interventions might have an effect on the attitudes towards recycling. Moreover, the results confirm that external factors that make recycling easier, are crucial in order to boost the desired behaviour. Whereas on the other hand, the education program “green schools” has been shown to be a long-term pro-environmental behaviour enhancer, although the results are not significant. Although the results seem very promising and show a causality effect of both treatments on recycling, the study has important limitations that should be taken into account when interpreting the findings.
The present research tried to establish the criteria to enter into the door-to-door waste management system and the education program “green schools”. Many different reasons are behind such decisions, for instance: political, economic, and persona/popular claims. Therefore, there is a potential self-selection issue that can bias the magnitude or even the sign of the estimations. The presence of such a bias has two different implications: First, because of the significance and magnitude of the coefficients in the case of the waste management system, the results might potentially vary in a randomized, but the large positive coefficients ensure that the effect is positive regardless of the potential self-selection bias. Second, the interpretation is different when focusing on the education program. The coefficients are non-significant, but follow an upward trend effect, which suggests that those results can potentially be improved by including more control variables and including additional years in the data set. On the other hand, Table 1 suggests that those municipalities that have at least one green school during the 16-years period are significantly different from those in the control group in the baseline year (2000). Thus, although the results are promising, and the methodology aims to minimize the potential problems related to the suitability of the data and treatment allocations, the findings regarding the education treatment might have self-selection and endogeneity issues. Nevertheless, it is worth saying that
when the green schools’ treatment group and the control group are compared in Table 1, no significant differences are observed in the recycling rate.
As mentioned in Section 2, the waste management system is well suited for municipalities with certain dimension characteristics. Because of time constraints and the fact that the results already show the effects that were intended to be proven, no heterogeneity analysis has been made. Nevertheless, the belief is that the promising results would be even more accurate, and even with greater coefficient estimations, if a detailed heterogeneity analyses had been undertaken. For instance, if the analysis were to be done by only comparing municipalities with less than 20 thousand inhabitants, the effect of the waste management system on recycling could be greater than the effect when municipalities with bigger dimensions are included in the analysis. However, the study has been improved as much as possible by “cleaning” the data set through the exclusion of outliers, making the sample more homogenous and reliable, taking into account the objective and requirements of the waste management system. Finally, as already mentioned, a wider range of control variables could make the analysis more accurate and consistent. The set of control variables is not optimal to make the most realistic analysis. The lack of data availability in key variables such as average age or attitude towards environment, has limited the accuracy and strength of the study. Additionally, it is worth saying that the variable “university” only includes data in two points in time. If data on more years were available, the analysis could lead to more accurate results. Nevertheless, a wide range of demographical and contextual variables, suggested to be important in the literature, are included in the study.
The study has tried to overcome these limitations by means of the methodology and the analysis undertaken, and, in general, it seems both treatments have a positive effect on recycling behaviour. The waste management system has a very high and bell curve-shaped effect trend, whilst the education program intervention has a small and long-term effect on recycling.
6. Conclusion
The present study has demonstrated the importance of making pro-environmental behaviour for individuals as easy as possible, and how education can potentially be a key element for attitude change towards environment in the long-run. In the last three decades, behavioural and attitude literature have focused more and more on environmental issues and challenges. The present study aims to add a slightly new vision of past research thanks to the availability of a well-designed and suited data set.
As suggested by Guagnano et al. (1995), make recycling more convenient for individuals is one of the main strategies to enhance the desired behaviour. Between 2000 and 2015 the door-to-door waste management system boosted recycling behaviour by 24.55%, on average. Hence, it confirms, on the one hand, what behavioural literature has been claiming for the last thirty decades and, on the other, the need to include the effect of a waste management system on recycling in Life-Cycle Assessments research, such as in Teerioja et al. (2012). The latter would enable a more accurate and realistic environmental and economic assessment that would, at the same time, improve the decision-making when it comes to designing and developing such policy interventions. Nevertheless, literature has also suggested that although extrinsic incentives and factors might be powerful tools in initiating the desired behaviour, continued participation requires intrinsic motivation (De Young 1985-1986, 1986; Katzev, 1989; Katzev & Pardini, 1987-1988; Pardini & Katzev, 1983). By analysing the effect of the education program “green schools” on recycling, the present study tried to show how an intrinsic motivation technique that focuses on education can have a positive, continued and long-term effect on recycling. The combination of both interventions has been shown to be the best strategy with the greatest potential, which confirms what past literature already suggested.
As mentioned in the discussion section, this study has many limitations. However, it enriches research on environmental behaviour and attitudes, and on the assessment of waste management public and private interventions. There is room for improvement in
the field of increasing the efficiency of waste management, and the present analysis can help to up-scale systems and business models in the circular economy and sustainability framework.
Further research should focus, first of all, on analysing how such interventions affect individual attitudes towards the environment, something that was beyond the scope of the present study. Second, a similar analysis could be replicated by trying to generate a completely randomized sample to be able to confirm the results of the present thesis. Third, waste management systems based on the reutilization of materials are continuously growing, which opens the door to start researching the suitability and effectiveness of such interventions. Finally, a closer look at the effect of new innovations regarding incentives to pro-environmental behaviour could complement the present study and shed some light on the accuracy of this type of interventions. Nowadays, research focuses on new innovative topics such as: The consumers' willingness to participate in E-waste recycling with a points reward system, Zhong and Huang (2016) or Determinants of residents' e-waste recycling behaviour intentions: Evidence from China, Wang et al. (2016). Thus, cryptocurrencies that aim to reward recycling behaviour and other positive environmental behaviours are a good example of what the future may hold for us.13
7. References
1. Agència Catalana de Residus: http://estadistiques.arc.cat/ARC/
2. Burn, S. M., & Oskamp, S. (1986). Increasing community recycling with persuasive communication and public commitment. Journal of Applied Social Psychology, 16(1), 29-41.
3. Coll, E., Rieradevall, J., & Doménech, X. (2002). La recogida selectiva puerta a puerta de la materia orgánica. Experiencia municipio Tiana
(Cataluña). Residuos, (67), 108-115.
4. Corral-Verdugo, V. (1997). DUAL ‘REALITIES’OF CONSERVATION BEHAVIOR: SELF-REPORTS VS OBSERVATIONS OF RE-USE AND RECYCLING
BEHAVIOR. Journal of environmental psychology, 17(2), 135-145.
5. Derksen, L., & Gartrell, J. (1993). The social context of recycling. American sociological review, 434-442.
6. De Young, R. (1985). Encouraging environmentally appropriate behavior: The
role of intrinsic motivation.
7. De Young, R. (1986). Some psychological aspects of recycling: the structure of
conservation-satisfactions. Environment and behavior, 18(4), 435-449. 8. Gelabert, E., Fontanals, F., Gascon, P., Ventosa, I., Prado, L., Missé, J., ... &
Gràcia, S. (2008). Manual Municipal de Recollida Selectiva Porta a Porta a Catalunya.
9. Gitlitz, J. S. (1989) Curbside collection containers: A comparative evaluation.
Resource Recycling, 70, 20-51.
10. Guagnano, G. A., Stern, P. C., & Dietz, T. (1995). Influences on attitude-behavior
relationships: A natural experiment with curbside recycling. Environment and behavior, 27(5), 699-718.
11. Hong, S., & Adams, R. M. (1999). Household responses to price incentives for
recycling: some further evidence. Land Economics, 505-514.
12. Hopper, J. R., & Nielsen, J. M. (1991). Recycling as altruistic behavior:
Normative and behavioral strategies to expand participation in a community recycling program. Environment and behavior, 23(2), 195-220.
14. Iriarte, A., Gabarrell, X., & Rieradevall, J. (2009). LCA of selective waste collection systems in dense urban areas. Waste Management, 29(2), 903-914. 15. Jenkins, R. R., Martinez, S. A., Palmer, K., & Podolsky, M. J. (2003). The
determinants of household recycling: a material-specific analysis of recycling program features and unit pricing. Journal of environmental economics and management, 45(2), 294-318.
16. Katzev, R. D. (1989). How to motivate recycling: applying minimal justification
techniques. In Association for Conservation Information's Winter Workshop (Arlington, VA).
17. Katzev, R. D., & Pardini, A. U. (1987). The comparative effectiveness of reward
and commitment approaches in motivating community recycling. Journal of Environmental Systems, 17(2).
18. Lacy, P., & Rutqvist, J. (2016). Waste to wealth: The circular economy advantage. Springer.
19. Loewenstein, G., O'Donoghue, T., & Rabin, M. (2003). Projection bias in
predicting future utility. The Quarterly Journal of Economics, 118(4), 1209-1248. 20. Oskamp, S., Harrington, M. J., Edwards, T. C., Sherwood, D. L., Okuda, S. M., &
Swanson, D. C. (1991). Factors influencing household recycling behavior. Environment and behavior, 23(4), 494-519.
21. Pardini, A. U., & Katzev, R. D. (1983). The effect of strength of commitment on
newspaper recycling. Journal of Environmental Systems, 13(3).
22. Punkkinen, H., Merta, E., Teerioja, N., Moliis, K., & Kuvaja, E. (2012).
Environmental sustainability comparison of a hypothetical pneumatic waste collection system and a door-to-door system. Waste management, 32(10), 1775-1781.
23. Reschovsky, J. D., & Stone, S. E. (1994). Market incentives to encourage
household waste recycling: Paying for what you throw away. Journal of policy analysis and management, 13(1), 120-139.
24. Schwartz, S. H. (1968). Awareness of consequences and the influence of moral
norms on interpersonal behavior. Sociometry, 355-369.
25. Schwartz, S. H. (1968). Words, deeds and the perception of consequences and
responsibility in action situations. Journal of personality and social psychology, 10(3), 232.
26. Schwartz, S. H. (1970). Elicitation of moral obligation and self-sacrificing behavior: an experimental study of volunteering to be a bone marrow donor. Journal of personality and social psychology, 15(4), 283.
27. Schwartz, S. H. (1977). Normative influences on altruism1. In Advances in
experimental social psychology (Vol. 10, pp. 221-279). Academic Press.
28. Schwartz, S. H., & Howard, J. A. (1981). A normative decision-making model of altruism. Altruism and helping behavior, 189-211.
29. Stevenson, B., & Wolfers, J. (2006). Bargaining in the shadow of the law:
Divorce laws and family distress. The Quarterly Journal of Economics, 121(1), 267-288.
30. Teerioja, N., Moliis, K., Kuvaja, E., Ollikainen, M., Punkkinen, H., & Merta, E. (2012). Pneumatic vs. door-to-door waste collection systems in existing urban areas: a comparison of economic performance. Waste management, 32(10), 1782-1791.
31. Thøgersen, J. (2003). Monetary incentives and recycling: Behavioural and
psychological reactions to a performance-dependent garbage fee. Journal of Consumer Policy, 26(2), 197-228.
32. Thøgersen, J. (1996). Recycling and morality: A critical review of the
literature. Environment and behavior, 28(4), 536-558.
33. Vining, J., & Ebreo, A. (1990). What makes a recycler? A comparison of recyclers
and nonrecyclers. Environment and behavior, 22(1), 55-73.
34. Wang, Z., Guo, D., & Wang, X. (2016). Determinants of residents'e-waste
recycling behaviour intentions: evidence from China. Journal of cleaner production, 137, 850-860.
35. Zhong, H., & Huang, L. (2016). The Empirical Research on the Consumers’
Willingness to Participate in E-waste Recycling with a Points Reward System. Energy Procedia, 104, 475-480.
