The role of urban governments in sustainably managing residential garbage

(1)

Residential garbage

Luuk van Loosdrecht

Major: Economics, Supervisor: Koen Vermeylen

July 15, 2016

Abstract

This thesis analyzes the relationship between employing Unit Based Pricing (UBP) on residual waste and household decisions with regard to residual waste generation and engage-ment in recycling. Two multiple regression analyses have been conducted on the average waste quantities among Dutch municipalities and their separation rates. By including socio-economic, political and institutional variables in the analysis, this thesis emphasizes the plurality and diversity of the determinants of household behavior regarding waste generation and recycling engagement. The results indicate that compared to fixed tariffs, waste gen-eration is lower and separation rates are higher under UBP. It also seems that the way in which waste fractions are collected and separated affects the separation rate. Differences in environmental values among municipalities are approximated by comparing the outcomes of the national elections of 2012, which is a rather poor estimate and did not have a significant effect. In addition, a probit regression is conducted on the determinants of municipalities choice to adopt UBP. It appears that collection type, and income to a lesser extent, partly determine municipalities’ choice to adopt UBP.

(2)

1 Introduction

Over the last years household waste policy has become increasingly important in industrialized countries. While waste generation has grown significantly over the last decades due to higher incomes, there is an increasing awareness of the external effects of waste generation (World Bank, 2010). It is no longer acceptable that waste is disposed without concern for environmental and natural resource issues, which causes societal resistance to the development of new conventional ways to dispose waste such as landfills and incineration facilities. This resistance is part of a larger environmental discourse that emphasizes sustainable management across industrialized countries. The increasing amount of waste on the one hand and the higher societal demands with regard to sustainably managing waste on the other make municipal solid waste an important policy objective and one that is becoming increasingly challenging to address.

1.1 Current Waste Status of the Netherlands

Currently, the average recycling rate in the Netherlands lays around 50 percent, while it has established a target to increase the recycling rate of household waste to 75 percent by 2020 (Ri-jksoverheid, 2016). Since municipalities are responsible for household waste collection services, Dutch municipalities will have to act fast in order to fulfill the national target of a separation rate of 75 percent in 2020. In order to realize this target, waste policies should be altered such that households will reduce the generation of residual waste1 and increase their engagement in recycling. This thesis aims to provide insight in the mechanisms that are subjective to household decision with regard to waste generation and recycling engagement by analyzing what factors that influence waste generation and separation rates among Dutch municipalities.

1.2 Current State of the Literature

In the academic literature there seems to be consensus on two failures that arise in solid waste management (Ferrara and Serret, 2008). The first relates to the existence of negative externali-ties with regard to households decisions over waste generation. When individuals choose what and how much to consume, they normally do not take into account how much waste they will generate. Therefore, the negative consequences of waste generation, such as the depletion of natural resources or pollution, are generally not taken into account during consumption. As a result more waste is produced than is socially optimal. The second failure relates to the way in which waste collection is financed. In most Dutch municipalities households pay a lump sum tax to local governments for the collection of residential waste (Nedvang, 2016). As individuals face zero costs for each additional unit of residual waste that is produced, they tend to produce more waste than if they would pay for the social marginal costs of an additional unit of residual waste. Following from this logic one can deduce that under a tax system where individuals face marginal prices for waste disposal, which from here onward will be referred to as Unit Based Pricing (UBP), individuals would reduce their waste generation and increase their engagement in recycling. Many studies have empirically assessed this relationship and will be discussed in the literature review.

1.3 Moral Values and Waste Infrastructure

Most of the conducted research has provided insight in the economic determinants of household behavior with regard to waste generation and recycling. For instance, much evidence is found

(4)

that employing UBP seems to reduce waste generation and increase recycling engagement. How-ever, many scholars tend to neglect social, political or institutional factors that might have a profound influence as well. From the literature review, which is presented in the next section, it follows that differences in moral values with regard to environmental and natural resource issues might cause variation in waste generation and recycling engagement between areas, regardless of the price of waste disposal (Dijkgraaf and Gradus, 2004; 2009). Although this variable is rather hard to measure within a quantitative framework, it does seem relevant to take envi-ronmental values into account, since Dijkgraaf and Gradus (2009) have shown that neglecting these variables can cause overestimation or underestimated estimations of the effect of UBP. Another institutional factor that is specifically relevant for assessing waste related behavior in the Netherlands is the way in which residential waste is collected. At the time of writing, there is a trend in the Netherlands that is called ‘omgekeerd inzamelen’ (reversed collection), and refers to an alternative waste collection system where households instead of bringing their recyclables to a nearby collection point and dropping their residual waste at the curbside, they bring their residual waste to a nearby collection point while the separated waste fractions are collected at the curbside. This collection system has transaction costs of disposing residual waste that are higher than disposing recyclables.

1.4 Research Questions

From the literature review, which is presented in the next section, it follows that political and institutional factors might be important determinants of waste generation and recycling engage-ment. Both values towards environmental issues and waste infrastructure might be important factors in analyzing the effect of UBP. Therefore, my research question is threefold:

1. How does the use of a unit-based pricing system affect household decisions with regard to waste generation and recycling rates in the Netherlands?

2. How do moral values regarding environmental issues affect household decisions with regard to waste generation and recycling rates?

3. How does waste collection infrastructure affect household decisions with regard to waste generation and recycling rates?

By answering these research questions this thesis aims to contribute to the understanding of several determinants of waste related household behavior, where an emphasis is put on both economic and socio-political variables. Besides answering these main questions, a separate as-sessment is conducted on the determinants that influence municipalities’ choice to adopt UBP. Since this thesis might potentially be used as a framework for policy makers to evaluate their cur-rent waste policy and provide them with a more thorough understanding of the effects of UBP, some insight in the determinants of municipalities’ choice to adopt UBP in the Netherlands might be useful in the consideration to adopt UBP.

To answer these research questions either cross-sectional data of municipalities or household survey data can be analyzed. In this thesis only a cross-sectional analysis has been conducted, since household data is less available than cross sectional data of municipalities.

1.5 structure of the thesis

The rest of this thesis is structured as follows. The next section covers the literature review. It discusses the different approaches to environmental issues within economics, where light is

(5)

shed on the neoclassical, the institutional and the new-institutional approach. This theoretical notion is made because the presuppositions of scholars strongly influence the search for empirical evidence, and thus the results (Dugger, 1979; Soderbaum, 1992). Afterwards, the literature review elaborates more extensively on the empirical work that has been conducted in this field. After that, section 3 introduces the methods for answering the research questions. Also, it develops an argument for the chosen variables and the descriptive statistics for the main variables are presented. Section 4 presents the multiple regression results and its implications for further research. Section 5 presents the conclusion of this thesis, as well as recommendations for further research into determinants of municipal waste.

(6)

2 Literature Review

In this section the economic literature that is relevant for this thesis is reviewed. The literature review is twofold. First, since there is a much scientific debate on the methodological require-ments of analyzing environmental issues, a brief theoretical review on the different methodologies within economics is being set out. The neoclassical approach, the institutional approach and the new-institutional approach are discussed with regard to their applicability to environmen-tal issues such as waste management. Second, the empirical research on waste management is discussed. The empirical literature is divided into the studies that assess the determinants of the independent variables, that are the average amount of waste generated and the separation rate, and studies that assess the effect of several variables on municipalities choice to adopt environmental policy programs

2.1 Methodological differences Neoclassical vs institutional vs New institu-tional school

As mentioned in the introduction, the perceived importance of environmental issues has risen over the past decades. Environmental consciousness has increased, both in society and in the academic literature. This raises the question whether certain methods are more suitable to assess environmental issues than others. In this paragraph the differences and similarities between the neoclassical school and the institutional school are pointed out, as well as how the two schools can be integrated into a framework that enables to assess the research questions.

2.1.1 Reductionist-Mechanistic Approaches versus Holistic-Evolutionary Approaches According to the Finnish philosopher Von Wright (1986), a scientific distinction can be made between reductionist-mechanistic modes of thinking on the one hand, and holistic-evolutionary on the other. In short, the term reductionist refers to approaching an issue from solely one discipline. In economics reductionism refers to explaining socio-economic phenomena solely from an economic argument where all motivation is reduced to monetary or utility terms (Allen, 1975; Tinker, 1984). The term mechanistic refers to applying closed mathematical model in studying economic phenomena. A good example of a mechanistic-reductionist model is general equilibrium theory: a closed mathematical model that only deals with economic factors, which are generally more suitable for quantification, and works under strong assumptions about human behavior and society, which excludes other social sciences from the analysis (Myrdal, 1978). On the other side of the spectrum there is the holistic-evolutionary approach. Holistic refers to the openness of economics as a science. This stems from the belief that there is causal interdependence between economic phenomena and the social system as a whole. In practice this means that holistic models not only include economic variables, but also take notice of psychological, cultural and political aspects that characterize the social system in which these economic phenomena take place. The term evolutionary refers to an interest in how institutions, habits, values and the economy in general evolve over time.

2.1.2 Neoclassical Economics vs Institutionalism vs New-Institutionalism

Institutional economics is generally characterized by a holistic-evolutionary approach, while neoclassical economics mainly approaches economic phenomena from a reductionist-mechanistic perspective (Samuels, 1995). The two schools lie so far away from each other, both in terms of focus and in terms of how the analysis is conducted, that they have operated aside from each

(7)

other for a long time without being integrated. This has led to a clear separation of economics and other social sciences, although the issues analyzed are often the same. However, with the development of New Institutional Economics (NIE), which is a combination of neoclassical and institutional economics, integration between the two schools is possible to a certain extent. Under NIE the neoclassical approach has been extended through the inclusion of social and legal rules that underlie economic activity (Williamson, 2000). It also rediscovers aspects of classical political economy. The great difference with institutional economics is that some of the determinants of institutions are now more susceptible to analysis by the tools of neoclassical economics. According to Williamson (2000) and Arrow (1987), NIE succeeded because it focuses on a broad range of issues (Williamson, 2000). As the approach within NIE made it possible to include institutions in the neoclassical framework, there has been enormous progress in the study of institutions by economists. Figure 1 graphically depicts the characteristics of neoclassical and institutional economics and shows how both are merged by NIE.

Figure 1: Neoclassical, New-institutional and institutional economics and their methodological charac-teristics

In this thesis, institutional economics is considered a valuable perspective to take into ac-count to assess the effect of UBP on waste generation and recycling. An important factor that determines the suitability of the institutional approach prior to the neoclassical approach for environmental issues lies in its perception of individual behavior. In the neoclassical approach individual behavior is deduced based on basic principles (utility functions) and initial conditions such as income distribution and price level. Within this framework, individuals by assumption always optimize a certain variable in monetary or utility terms under certain boundary condi-tions or restriccondi-tions. However, this approach is only feasible in simple cases of decision-making, where there is little impact, few decision makers, few parties affected and considerable consensus about values (Soderbaum, 1992). The influence of institutions is often neglected. The decision by individuals to generate less waste or to recycle more is likely to depend not only on mon-etary incentives, but also their political affiliations with regard to the environment as well as their social environment (Berglund, 2006). The next section discusses the empirical literature on economic and institutional variables influencing household behavior with regard to waste.

(8)

2.2 The role of financial and institutional incentives in households waste generation and separation rates

As mentioned in the introduction, economic theory suggests that a fixed tax rate for waste collection services will lead to inefficiently high levels of solid waste. Since households face a zero price for increments in waste collection services, individuals experience zero monetary incentives to reduce or divert their waste. The first section discusses several empirical studies that analyzed the effect of employing UBP on waste generation and recycling. The second section discusses other empirical studies that included institutional variables as well.

2.2.1 Marginal prices for residual waste increase engagement in separation and decreases residual waste generation

Following economic principles, marginal prices on residual waste should increase recycling and decrease residual waste generation. This relation has been empirically assessed in various studies. For instance, Jenkins (1993) studied how households put effort in waste disposal and recycling under different tax systems. She did this by combining panel data and cross-sectional data from nine communities. It was found that recycling requires more time and effort compared to conventional waste disposal. She showed that the response to marginal waste fees result in the consumption of goods that yield lower amount of post-consumption waste. The effect of marginal fees on recycling remained unclear, and seemed to depend on the marginal value of leisure and the size of the marginal waste fees. She concluded that incentives to decrease waste disposal can have two effects. The first is that engagement in recycling activities increases in order to dispose less waste. Secondly, people alter their consumption bundle and increase their consumption of products that generate less waste. An important critique on the analysis was that the information on the subject was too limited to carry out economic analyses. For instance, the measurement of leisure valuation by individuals is problematic, as quantification of such a subjective valuation is difficult (Aguiar and Hulst, 2006).

Morris and Holthausen (1994) also found that marginal prices encourage household to re-duce waste generation, but included several institutional variables as well. They estimated the relationship between collection fees and waste generation under two different economic and in-stitutional regimes in Pennsylvania in 1988. The first employed fixed fees for waste collection, no curbside recycling, and collected waste twice a week. The other employed UBP, curbside recycling and collected waste once a week. They found that unit pricing encourages households to decrease waste disposal. As the fee increased, this effect did not change, but the source of the decreased waste disposal shifted from a recycling increase to a decrease in waste generation. This result thus corresponds to Jenkins argument that decreased waste disposal is caused by either increased recycling or decreased waste generation. They also found that recycling rates depended on government effort to increase the ease of recycling by adopting recycling programs. Kinnaman and Fullerton (1996) estimated the effect of a bag-based unit pricing system in Virginia in 1992. The sample consisted of 75 households, whose bags or cans of garbage and recyclable materials are counted and weighed over two four-week periods before and after the implementation of the UBP system. The analysis corrected for income and demographic variables by setting out a questionnaire among the participating households. They found that recycling weight increased by 16%. They also found that households decreased their garbage weight by 14% and the volume by 37%. The average household increased the density of the garbage bags by 43%, which suggests stomping activities (Kinnaman & Fullerton, 1996). Efaw and Lanen (1979) already found evidence for stomping activities when the marginal price was a function of the number of containers instead of the weight of waste. This is undesirable as the

(9)

social costs of waste depend on weight rather than volume. These findings indicate that people do indeed respond to financial incentives to decrease waste disposal, but that the incentive should focus on weight instead of volume. Miranda and Aldy (1998) found similar results by doing a case study analysis involving nine American communities.

Dijkgraaf and Gradus (2004) did a cross-section analysis on Dutch municipalities and com-pared four different unit-pricing system on several different waste streams. They included many socio-economic variables like family size, average age and number of non-western foreigners per inhabitant. They found that the price incentive in general is effective, and that UBP causes all waste streams to diminish. As for the socio-economic effects, income and average age had a posi-tive effect on waste generation, and population density had a negaposi-tive effect. They also included a variable for environmental activism in order to control for the possibility that households in municipalities that employ UBP are more concerned with sustainable waste management than households in municipalities without UBP. They did this by including a dummy that separated municipalities that employed UBP during the first years of UBP in the Netherlands (early 1990s) from municipalities that changed to UBP in a later stage (after 1999). The rationale behind this dummy is that municipalities that switched in an early stage are more likely to be more engaged with environmental issues and therefore already generate less waste and separate more. It appeared that this was indeed the case and that the effect of UBP was overestimated without controlling for environmental values.

From these articles it follows that there are three basic ways in which the effect of UBP can be analyzed. The first collects household-level data before and after a regime change. Since the data is collected through questionnaires, all sort of information can be collected that is not available in aggregate databases such as moral values towards the environment. However, household-level analyses might create self-selection and observational biases (Isely and Lowen, 2007). A second type of analysis is a time series for one or a few municipalities. A time series analysis gives more precise insight in the effect of the adoption of UBP on an aggregate level. A third type of analysis uses a cross-section of municipalities using different pricing schemes. The data required for such an analysis is relatively easy to collect from national of regional databases, but can result in biased estimates since there might be cultural or political differences with regard to waste. For instance, the perceived importance of sustainable waste management might differ across regions.

2.2.2 The Relation between several socio-economic, institutional and political vari-ables and the choice of municipalities to employ recycling programs

Since this thesis analyzes the effect of UBP, it is relevant for municipalities to gain insight in the determinants of the choice to change their tax system to a system based on unit prices. From Feiock and West (1993) it appears that the choice of municipalities to innovate their recycling policies depend on various social, political and institutional variables. By conducting a probit regression, they found for instance higher probability of providing recycling services, for a higher average income. Further, they found that the effort of the national government to encourage municipalities to provide recycling services also has a significant effect. Another important variable that is taken into account is the representation of interest groups. The competing demands of different interest groups has a significant effect on the adoption of recycling programs. These are interesting findings, as they increase the visibility of various social factors that play a role in whether certain policies are adopted or not.

(10)

2.3 Conclusion

Having reviewed the literature on the price effects on waste generation and recycling, several conclusions can be drawn. First, households participate in stomping activities when a bag-based unit pricing system is applied (Kinnanman and Fullerton, 1996). This means that the price effect does influence household decisions on waste, but not in a desirable way. As the external costs of waste depend on the volume after compacting at the landfill, weight is thus a better proxy for measuring waste generation. It seems that the price effect on waste is influenced by ease of recycling (Morris and Holthausen, 1994). The transaction costs of waste disposal thus seem to influence household decisions of waste. Further, from Dijkgraaf and Gradus (2004; 2009) it is concluded that values regarding the environmental and natural resource issues are determinants of waste generation and engagement in recycling. Based on reviewed articles, it is concluded that when assessing the effect of UBP on household behavior regarding waste generation and separation, the analysis should include variables that distinguish in varieties in income, waste collection type and environmental values across municipalities in order to grasp the phenomena of waste generation and recycling as good as possible.

Regarding the determinants of municipalities choice to adopt UBP, Feiock and West (1993) have shown that there is a wide range of variables that determine the likelihood of certain policy adoptions by municipalities. For this thesis it is relevant to assess what determines municipalities choice to adopt UBP. Therefore, the analysis in this thesis on the factors determining the probability of adopting a UBP system is inspired by the analysis that Feiock and West (1993) have conducted with regard to the provision of recycling programs.

(11)

3 Methods and Data

The this section presents the models that are used in this thesis, as well as an elaboration on the operationalization of the variables used and the descriptive statistics.

3.1 The Model

In section two it is shown and argued that both the effect of employing UBP on waste related household behavior and the effect of several institutional, socio-economic and political variables on governments choice to adopt UBP are relevant in researching sustainable waste management (Dijkgraaf and Gradus, 2004; Morris and Holthausen). Therefore the analysis of this thesis consists of two parts. The first part comprises two models that explain the effect of UBP on 1: waste generation per municipality2 and 2: the separation rate of synthetic waste per municipality. The second part consists of one model that assesses the determinants of the main independent variable, namely employing UBP or not. Both analyses touch upon socio-economic, institutional and political variables. As can be seen in equation 1 and 2, the two models in part one use 5 variables to respectively explain waste generation and the separation rate.

W aste Generation = β0+ β1 U BP + β2 Income + β3 V alues

+ β4 Collectionbring+ β5 Collectionget+ ε

Seperation Rate = β0+ β1 U BP + β2 Income + β3 V alues

+ β4 Collectionbring+ β5 Collectionget+ ε

U BP = β0+ β1 Income + β2 Collectionbring+ β5 Collectionget+ ε

The two dependent variables and the first two independent variables, UBP and income, are rather straightforward variables. Waste generation is operationalized as the average weight of the residual waste that is generated per capita. In this thesis only separation rates for synthetic waste is taken into account, since UBP and recycling programs mostly aim to increase separation of synthetic waste (plastics), since these are a valuable resource and one that has considerable negative environmental impact (Munipality of Amsterdam, 2015). Average income is taken for all people with income in each municipality. UBP is included as a dummy variable, where 1 denotes the municipalities that have marginal prices3 and 0 denotes the municipalities that fixed tariffs. However, the other variables, type of collection of recyclable waste fractions and ’environmental values regarding environmental issues are less straightforward. There are several important notions that have to be made regarding the operationalization of these variables, which are discussed in the forthcoming paragraphs.

2

Measured in kilograms per citizen.

(12)

3.2 Type of Collection Service

As mentioned in the introduction, reversed collection is a system where transaction costs of disposing recyclables is lower than disposing residual waste, and is increasingly applied in the Netherlands. Unfortunately there is no data available on which municipalities apply reversed collection and which don’t. In order to examine the role of transaction costs, this thesis dis-tinguishes three types of recyclable waste collection: subsequent separation, separate collection-bring and separate collection-get. Subsequent separation refers to a collection system where all recyclable waste is collected together with residual waste and transported to a plant where machines separate the recyclables from the residual waste. Separate collection refers to a system where different waste fractions are separated at the household and collected in separate bags. Within separate collection, there is a difference between a bring-system and a get-system. The difference between bring and get is that within a get-system the separated synthetic waste is col-lected at the household, whereas within a bring-system the separated synthetic waste is brought to a nearby container where each household disposes its separated waste fraction. The main difference is that within a bring-system individuals need to put more effort in disposing their recyclables, as they cannot dispose them at the curbside. Nedvang, an organization that cares about recycling packaging materials across the Netherlands, has made this distinction between the various collection systems. Since Nedvang distinguishes merely between those three systems, possible other systems to collect waste are neglected in this thesis.

3.3 Environmental values

Political affiliation is the most difficult variable to operationalize in a reliable way. Especially since this analysis deals with aggregate data and thus cannot delve into personal motives, dif-ferences in political preference among municipalities are hard to measure. A possible way to obtain an indication for political preferences is to look at election outcomes (Dijkgraaf and Gradus, 2009). In the case of assessing environmental values with regard to waste management, one would ideally compare the fraction of people that voted for the most environmentally con-scious party in the local council elections, since waste management is an issue for local politics. However, because of two practical obstacles this comparison is not executable. First, there is no data available on how well a particular party did in each municipality, since there is only data available on which party was largest. Therefore, it would be impossible to compare munic-ipalities on their environmental values if the most environmentally orientated party is not the biggest one. This problem in only solved by solely comparing parties that often win elections with regard to their relative position on environmental values. For instance, knowing the liberal party (VVD) and the labor party (PvdA) are two of the biggest parties, and have quiet divergent views on environmental issues, these two parties would be suitable for comparison. However, in many municipalities small local parties won the last local council election, which make these municipalities useless in the analyses, since these local parties do not participate in local council election of other municipalities and thus cannot be compared. In order to solve the problem of the local parties, one would have to analyze the latest national election instead, were no local parties participate. A problematic downside is that the relation between voting behavior in national elections and moral values on local environmental issues becomes rather far-fetched. However, in a quantitative analysis that deals with aggregate data this might be the closest way to model environmental values. Thus, in order to approximate the disparity in environmental values among municipalities, all municipalities are compared where either the labor party or the liberal party was largest during the national election in 2012. It is assumed that a municipality where the labor party is largest has a relatively more progressive attitude towards policy that

(13)

encourages environmental sustainability, such as UBP4_{. This variable is included as a dummy.}

A 0 is assigned when the liberal party was largest and a 1 is assigned when the labor party was largest.

3.4 Data sources

Data on the dependent variables the quantities collected of total waste (in kg) and separation rates of synthetic waste fractions come from a database from the ministry of Infrastructure of Environment called Waste monitor (afvalmonitor, 2016). This database presents data on the average waste quantities in kilograms per municipality. Further it present separation rates on paper, glass, textile, organic waste, synthetic waste (plastics) and several other forms of residential waste. It is assumed that the data presented in this database is correct and that the variables are measured in the same way across all municipalities. The database also presents the type of tax system per municipality. Dutch municipalities are free to choose how to finance their waste collection. Although there are many ways to finance residential waste collection services, this thesis distinguishes between only two types of financing systems: a unit-based pricing system and a non-unit-based pricing system. The various financing system are categorized by determining whether under that system marginal prices are generated for residual waste so that households have a monetary incentive to produce as least residual waste as possible and to increase recycling5. The data on the election outcomes are retrieved from a private website, called www.nlverkiezingen.com. Average income per municipality is included in this analysis as a control variable as well, since much of the existing literature takes average income into account (Jenkins, 1993). Average income per municipality is measured as the average income of the population that has an income, including students and is collected from a database of the CBS, the Dutch statistics agency (CBS, 2016) VNG, Association for Dutch Municipalities (VNG, 2016).

4

This assumption is made after having assessed the environment chapters of the party programs of the VVD and the PvdA.

5

(14)

3.5 Descriptive statistics

For the analysis all Dutch municipalities are taken into account. However due to missing values and municipality mergings, several are omitted. Further, for the variable on environmental values, only municipalities are taken into account where either the VVD or the PvdA was largest during the national elections of 2012. As can be seen, the analyses use two quantitative variables and four dummy variables. In table 1 the descriptive statistics are presented. It can also be noted that the separation rate of synthetic waste is rather low. As mentioned in the introduction, the average separation rate in the Netherlands lies around 50%. For synthetic waste this is only 9.5%, which shows that there is much potential for the separation of synthetic waste fractions.

Table 1: Descriptive statistics

Variable obs mean std. Dev min max waste quantity 366 186.47 64.32 21 637 Separation rate 363 9.53 6.4 0 30.8 UBP 367 0.395 0.49 0 1 Income 366 29.84 4.07 23.7 51.7 values 332 0.27 0.44 0 1 Collection get 367 0.57 0.49 0 1 Collection bring 367 0.31 0.46 0 1

4 Results

With the variables explained in section 3, as well as the summary on the descriptive statistics, the two analyses are discussed in this section. The first two paragraphs are devoted to 1) the correlation between the two dependent variables, that are the average waste quantity per capita and the separation rate per municipality, and 2) relevant correlations between the other variables.

4.1 Waste Generation and Engagement in Waste Separation

A central issue of analysis within environmental economics is how consumers respond to certain environmental policies (OECD, 2012). However, consumers often respond differently to the same incentives created by the government. For instance, recycling programs adopted by governments are designed to encourage consumers to engage in recycling and generate less waste. While people might indeed be susceptible to these incentives, it could also be the case that because the recycling rate is high, people are less encouraged to generate less waste because they know that much of their waste will be reused anyway. In the case of introducing UBP, there are three main consequences of UBP: increased recycling, illegal dumping and consuming goods that relatively generate little post-use waste (Jenkins, 1993; Morris and Holthausen, 1994; Kinnaman and Fullerton, 1996). The aim is to encourage people to generate less residual waste and instead recycle more of their waste fractions. Whether this incentive is actually created cannot be fully examined from a cross sectional analysis with merely aggregate data on waste quantities and separation rates. Households could for instance choose not to recycle but dump their waste

(15)

illegally. This would decrease the average waste generation, since illegal dumping is not taken into account. It is therefore relevant to examine how waste generation and the separation rate correlate in order to say something on whether incentives to decrease waste generation automatically lead to higher recycling levels. The correlation of -0.74, given in figure 2 implicates that decreased waste generation indeed goes hand in hand with increased recycling. Figure 2 and 3 respectively show the correlation between the quantity of residual waste and the separation rate at one moment in time and how this relation looks like over the period from 2001 until 2014. From both figures it seems that these two variables are negatively correlated to each other.

Figure 2: A scatterplot of the average waste quantity per capita per municipality in 2014 (in kg) against the separation rate in 2014 as a percentage. Waste quantity is depicted on the X-axes and the separation rate is depicted on the Y-axes.

As figure 3 shows, the negative relation between waste generation and recycling engagement also seems to hold when different years are compared to each other. For the years 2001 until 2014, the average amount of waste decreases, especially the last 4 years, and the average separation rate steadily increases. Since this analysis does not assess illegal dumping and consumer behavior with regard to their considerations to buy products with high or low amounts of post-use waste, from here onward these two possible consequences are neglected in the rest of the analysis.

The strong correlation between waste generation and the separation rate gives reason to believe that incentives to generate less waste indeed increase engagement in recycling. Paragraph 4.3 will further delve into the possible causal relation and will provide more solid indications about the relation between incentives to generate less waste and the extent to which people are encouraged to recycle.

(16)

Figure 3: The evolution of the average waste quantities and separation rate of Dutch municipalities from 2001 until 2014. The red line shows the development of waste generation and the green line the development of the separation rate.

4.2 Other Relevant Correlations

Table 2 shows all the correlations between the variables.

- Waste Separation UBP Income Values Collection get Collection bring quantity rate waste quantity 1 Separation rate -0.74 1 UBP -0.62 0.64 1 Income 0.06 -0.07 -0.14 1 values 0.14 -0.17 0.03 -0.33 1 Collection get -0.36 0.52 0.24 0.04 -0.19 1 Collection bring 0.21 -0.19 -0.16 0.1 -0.09 -0.78 1 Table 2: My caption

From table 2, several correlations are worth pointing out. The average income per munic-ipality is negatively correlated with the separation rate. This seems odd, since it would be logical to expect that higher incomes would result in higher separation rates. Generally, in-comes positively correlates with education level (Mincer, 1974), and thus with a higher degree of consciousness on environmental matters. Also from previous studies it appeared that income has a positive effect on separation (Richardson and Havlicek, 1978; Kinnaman and Fullerton, 1996). Another remarkable correlation is that between separation and the collection type and that between waste generation and collection type. For the former it appears that under a get-system, the separation rate is almost 52 percent higher than under another collection system. The same, but to a lesser extent, is true for the correlation with waste generation, which is -.36. This indicates that the incentive to recycle under a get-system would be much larger than under

(17)

a bring-system and under subsequent separation. If this would be true, this could mean that lower waste disposal transaction costs indeed result in higher separation rates. This possible relation is further discussed in paragraph 4.3.

Since the correlation between the collection type and the separation rate is so large, a possible explanation can be found for the negative correlation between income and separation rate as well. As can be seen in figure 2, subsequent separation is especially present in the northern part of the Netherlands. The average income in the northern part of the Netherlands is generally lower than in the rest of the Netherlands (CBS, 2014)6. Since the municipalities that employ subsequent separation have a low separation rate, and also have lower average incomes, the negative correlation between income and separation rate does certainly not imply that these two variables have a negative causal relation. The next paragraph will delve further into this.

Figure 4: a map of all Dutch municipalities. Yellow = Get-system; blue = bring=system; red = subsequent separation, adopted from www.nedvang.nl

6_{From the analysis in this thesis it appeared that the average national income is ca 29.760 per year, while the}

(18)

4.3 The effect of UBP on the Separation rate and waste generation

In this section the results of the analysis are presented. For the effect of UBP on both the separation rate and on waste generation, 5 regression analyses are conducted. Table 3 and 4 summarize the main results of the analysis. In both assessments first a regression is conducted that includes all variables, then a regression that omits one of the variables, then two variables etc. Since the correlation matrix in subsection 4.2 showed a strong correlation between the type of collection system and both the separation and to a lesser extent waste generation, an interaction variable is included that distinguishes between the effect of UBP within municipalities that employ subsequent separation, and municipalities that employ separate collection.

Separation rate UBP 7.14*** 6.72**– 6.71**– 8.38**– 8.34**– Income -0.16** -.12* -.12* -0.027 -Get-system 8.80*** 10.14** 10.28** - -Bring-system 5.70*** 6.99** 7.12** - -Values -0.16 -0.24 - - -UBP*After -7.1*** - - - -Constant 3.12** 2.65 2.29 5.43** 2.65 R-squared 0.67 0.64 0.64 0.41 0.41 Table 3: Five regression analyses on the separation rate

7 Waste generation UBP -75.85*** -74.63*** -73.79*** -81.49*** -80.57*** Income 0.14 0.19 -0.23 -0.70 Get-system -28.33** -31.56** -40.35*** Bring-system -6.46 -9.47 -17.60 Values 16.07* 16.24* UBP*After 14.84 Constant 225.5*** 226.6*** 250.6*** 239.0*** 217.8*** R-squared 0.43 0.43 0.42 0.37 0.37

Table 4: Five regression analyses on waste generation

From these results several conclusions can be drawn. The first thing that can be noticed is the fact that employing UBP has a significant effect in all analyses. It also appears that political preference and income both have a negligible effect on the separation rate, both in terms of significance and explained variance. The negligible effect of income comes as a surprise, as various studies have shown that income has a positive effect on the separation rate(Jenkins, 1993; Richardson and Havlicek, 1978). Since higher incomes mainly come together with higher education levels, and thus higher consciousness of environmental problems, one would expect that income would have a positive effect on the separation rate. Following these results, which are based on aggregate data, this relation does not seem to hold in the Netherlands. This is an

(19)

interesting finding and needs further research on the relationship between income and household decisions with regard to waste.

The variable used to approximate political preference, namely the biggest party in the last national elections, is likely to be a rather poor estimate of environmental values among Dutch municipalities. Both in the assessment of UBP on the separation rate and on waste generation, the effect is negligible. The data set used for this thesis included data on the last three election outcomes per municipalities. It appeared in more than 30 percent of the municipalities, the biggest party differed in each election. This shows that the outcome of one election is not a reliable approximation for the predominant environmental values among municipalities.

From the interaction term it appears that the effect of UBP on the separation rate differs between separate collection and subsequent separation. The effect of UBP is estimated to be higher under separate collection than under subsequent separation. The interaction term ap-peared to be only significant in the assessment on the separation rate. A possible explanation for the negative effect of subsequent separation on the separation rate could be that separate collection by itself triggers people to engage in recycling. Making waste separation visible for citizens could already encourage them to increase their engagement in recycling. Another expla-nation could be that the technology used in the separation plants is not sufficiently advanced to realize the same separation rates as under separate collection. When looking at the R-squared values of the two models, it appears that the effect of collection system has a more profound effect on the separation rate than on waste generation. As shown in the tables, the R-squared in the model of the separation rate varied between 0.67 and 0.41, while the R-squared in the model of waste generation only varied between 0.43 and 0.37. Besides that from these R-squared values it appears that UBP explains more of the variance of separation rates (R-squared is 0.41) com-pared to waste generation (R-squared is 0.37), the effect of collection type had a much stronger effect in explaining separation rate than it had in explaining waste generation. Table 5 shows the F statistics after having conducted a partial F-test on the effect of collection type.

Including collection type Waste quantity Separation rate

F-statistic 14.34 119.61

Table 5: The F-statistics of including collection type in both regressions

These F-values require further qualitative research on the relation between collection type and household considerations with regard to recycling in order to substantiate a possible causal relation.

4.4 The Effect of Environmental Values

Since the outcome of national elections is likely to be a bad indicator for environmental values, estimating whether there is an overestimation of the effect of employing UBP or not, becomes difficult. Dijkgraaf and Gradus (2009) solved this problem by analyzing the behavior of mu-nicipalities over time. The rationale behind this is that many mumu-nicipalities adopted UBP in the 1990s, and that progressive municipalities are likely to adopt UBP first and that more con-servative municipalities would follow afterwards. If this hypothesis would be true, the effect of adopting UBP would be overestimated, as the municipalities that adopted UBP are already more

(20)

engaged sustainable waste management. Their results indicate that this was the case, as the first municipalities that adopted UBP already generated less waste and separated more (Dijkgraaf and Gradus, 2009). They have also shown that this difference decreased over time, which indi-cates that the effect of adopting UBP in the early years is likely to be overestimated, and that over time the estimations have become more accurate. It would be relevant to use their method again in order to see whether the effect of employing UBP still decreases until 2014. However, as this requires more advanced econometrics, this is outside the scope of this thesis. Instead, in order to invigorate the argument based on the cross-section analysis of Dutch municipalities on a fixed point in time, it is analyzed whether municipalities that have undergone a regime change8 in fact experience different separation rates and different amounts of waste generation. All municipalities that have undergone a regime change from a non-UBP system to UBP and vice versa, and have had both systems for at least three years, are analyzed with respect to the development of the average amount of waste (in kilogram) per citizen9. Per municipality that has experienced a regime change, the average waste quantities are compared to each other. In this way an extra argument is developed to substantiate the hypothesis that employing UBP causes lower average waste quantities.

Therefore, a sign test is conducted with respect to the municipalities that have undergone a regime change. All municipalities that adopted or rejected UBP, and for which under both regimes data of at least three years is available, are included in the test. Although the amount of municipalities that are taken into account in the Sign test is rather low (n=33), the z-value is 0.017, which is statistically significant. From this test it thus seems that it is unlikely that adopting UBP has no influence on the average waste quantities produced. However, this should just be interpreted as an indication. Further econometric analyses on the effect of adopting UBP over time should be conducted to substantiate this argument. Unfortunately no data is available on separation rates on synthetic waste of the past ten years. Therefore, only the effect of adopting a UBP on average waste quantities is conducted over time, while the effect on the separation rate is only assessed by the cross-section analyses.

4.5 The Determinants of Adopting UBP

As shown in the previous paragraphs, it appears that UBP indeed has a significant effect on waste generation and recycling. Since politicians might interpret such results as a policy advice, it might be useful to delve further into the factors that determine a governments choice to adopt UBP. Without making any claims on causal relations, it might be relevant to investigate whether some indications can be made with regard to the determinants of municipalities’ choice to adopt UBP. Inspired by Feijock and West (1993), who show that there are several economic, institutional and political factors that determine the choice to adopt municipal solid waste recycling programs in the USA, a probit regression has been conducted on the binary variable UBP. The results are summarized in table 6.

8

A regime change in this context means a change from a non-UBP system to UBP or vice versa.

9

An elaboration on the Sign-test as well as the how the test statistic is obtained, is summarized in appendix B and C.

(21)

UBP Coefficient Z-value Income -0.78 (0.0196) -3.95 Collection get 1.07 (0.244) 4.39 Collection bring 0.543 (0.261) 2.08 Constant 1.228 2.14 Observations 366 LR Chi2 39.44

Table 6: Probit Regression on the determinants of UBP

The model in general is significant, which means that the variables together have a significant effect on the choice to adopt UBP. Further it appears that both income and collection type also influence governments choice to adopt UBP. The effect of income is significant, but also rather small. According to Feiock and West (1993), the financial status of municipalities was an important factor in explaining the choice to adopt recycling programs. Assuming that adopting UBP is costly (at least in the short run), it is reasonable to believe that the financial status of municipalities also affects the decision to adopt UBP. However, average income might not be the best variable to capture municipalities’ financial status, since average income does not necessarily means that the municipality has much to spend. Contrarily, the difference between a get-system and subsequent separation is significant and rather large. Under a get-system, the model estimates that the z-value of adopting UBP increases by 1.07, which is a considerable increase in probability. This result indicates that the compatibility of UBP with collection type is important to take into account when considering to adopt UBP. Further, according to Feiock and West (1993), there are several other parameters that should be taken into account as well, such as the representation of interest groups and the feeling of urgency. Further research should further investigate this possible relation.

(22)

5 Conclusion

This thesis has examined the determinants of waste generation and recycling engagement in the Netherlands. By conducting two multiple regression analyses, it seems that UBP decreases waste generation and increases recycling engagement. Especially the effect on recycling engage-ment, measured by examing separation rates, seems to have a profound effect. Some of the other variables that are taken into account were a valuable addition to the analyses, where others appeared to be redundant. The inclusion of collection type especially resulted in higher explanatory power and significant coefficients in the model that explains the separation rate. Also the interaction variable, which estimated the effect of UBP under both source separation and subsequent separation, was a valuable determinant of the separation rate. These results call for further research into the relation between UBP, collection type and recycling engagement. Surprisingly, income did not have a significant effect. Although previous studies indicate that there is relation between income and waste related behavior, this analysis has found very little evidence that income influence waste generation and recycling. Further research should be con-ducted in order to substantiate this finding. The variable measuring environmental values was too far-fetched, and did not have any explanatory power in both models. This does not impli-cate that no further research should be conducted on the relationship between environmental values and household decision regarding waste. Contrarily, as Dijkgraaf and Gradus have shown, when dealing with aggregate data more advanced econometric models are needed so that more reliable approximations of environmental values can be made. Also more qualitative methods should be considered in order to comprehend the relation between moral values and household decisions on waste. A more institutionalist perspective could be relevant here. Lastly, several determinants of Dutch municipalities’ choice to the adopt UBP are examined. Collection type and income appear to have both some influence, where collection type seems to have a more profound influence. Although the results on the determinants of municipalities’ choice to adopt UBP are not directly applicable, they do shed light on the plurality and interdependence of the determinants that are related to waste management.

(23)

Appendices

A

Sign-test

The Sign test is a non-parametric test that is used to test whether two groups are equally sized or not. A positive sign is used to indicate that the average amount of waste is lower under a UBP system than the average amount of waste under a non-UBP system, whereas a negative sign is used to indicate that the average amount of waste is higher under a UBP system than under fixed prices. If the introduction of UBP has an effect on the amount of waste that is generated, then the times that a positive sign is assigned, would have to occur more frequently that the assignment of a negative sign. In other words, if UBP would not have an effect on waste generation, the probability of assigning positive sign should lie around 0.5. If the introduction of UBP does have an effect, this probability should be higher than 0.5. Therefore, with n = 33:

H0 : p(+) = 0.5 H1 : p(+) > 0.5

Since n = 33, the binomial probability that p = 0.5 is tested. The test statistic is:

Z = (X + 0.5) − 0.5n√ n 2

The number of positive signs is 22, the test statistic is 2.122. P (Z > 2.122) = 0.017. following a significance level of .05, there is enough reason to indicate that UBP might have a negative effect on the average amount of waste that households generate. The Kolmogorov-Smirnov test is presented in the next appendix, which justifies the use of the Sign-test by conducting a test on normality.

B

Kolmogorov-Smirnov test

Before choosing a certain test to compare the average waste quantities, it should be determined whether the average waste quantities stem from a normal distribution. In order to test this, a Kolmogovov-smirnov test is used in order to evaluate whether the average incomes of all Dutch municipalities over the years 2001 until 2014 are normally distributed. The output is shown in table 5.

D P-value Waste quantity 0.0647 0.000 Cumulative -0.0631 0.000

Table 7: The output of the Kolmogorov-Smirnov test

As the output demonstrates, the distribution is significantly unlikely to be normally dis-tributed. Therefore, the non-parametric Sign-test is justified.

(24)

C

Categorization of Municipalities in UBP or not

Table 6 shows the different tax systems to finance waste collection that are recognized by Rijk-swaterstaat. The first column denotes the factor on which the price of waste collection is based The categorization into UBP or not is done by determining whether marginal prices are applied. For example: weight falls under UBP, since higher weight results in higher taxes. Contrarily, household size does not, as an extra unit of waste does not result in higher taxes etc.

System UBP: Yes of no? Household size No Weight Yes Frequence No Volume Yes Fixed tariff No Expensive bag

for residual waste Yes No tariff No Sort of house No

Table 8: My caption

D

Heteroskedasticity

Heteroscedasticity violates the ordinary least squares (OLS) assumptions that the error terms are uncorrelated and that it has an expected value of zero, given any value of the independent variables (constant variance). Heteroscedasticity do not lead to biased coefficient estimates but cause biased standard errors. This in turn leads to bias in test statistics and confidence intervals and has a biased inference of the coefficients as a result.

Heteroskedasticity is tested with the Breusch Pagan test, which follows an F-distribution. As shown in table 9, the F-statistic is 1.74, which is not significant. Therefore it is assumable that there is no heteroskedasticity.

(25)

Squared errors Coefficient T-statistic UBP 0.651 (3.81) 0.17 Income -0.920 (0.44) -2.10 Values 3.223 (4.64) 0.69 Get-system 8.31 (6.70) 1.24 Bring-system 11.634 (6.478) 1.80 Constant 31.51 (14.50) 2.17 R2 0.024 F-statistic 1.74 Table 9: My caption

References

[1] Allen, V. L. (1977). Social analysis: a Marxist critique and alternative. Longmans, Harlow. [2] Arrow, K. J. (1987). Reflections on the Essays. In Arrow and the Foundations of the Theory

of Economic Policy 727-734). Palgrave Macmillan UK

[3] Berglund, C. (2006). The assessment of households’ recycling costs: The role of personal motives.Ecological Economics, 56(4), 560-569.

[4] Dijkgraaf, E., & Gradus, R. H. J. M. (2004). Cost savings in unit-based pricing of household waste: The case of The Netherlands. Resource and energy economics, 26(4), 353-371. [5] Dijkgraaf, E., & Gradus, R. (2009). Environmental activism and dynamics of unit-based

pricing systems. Resource and Energy Economics, 31(1), 13-23.

[6] Dugger, W. M. (1979). Methodological differences between institutional and neoclassical economics. Journal of Economic Issues, 13(4), 899-909.

[7] Efaw, F., & Lanen, W. N. (1979). Impact of user charges on management of household solid waste (Vol. 1). Municipal Environmental Research Laboratory, Office of Research and Development, US Environmental Protection Agency.

[8] Eurostat, 2016. retrieved from: http://ec.europa.eu/eurostat/documents/2995521/7155577/8-10022016-AP-EN.pdf/38bf822f-8adf-4e54-b9c6-87b342ead339.

[9] Ferrara, I., & Serret, Y. (2008). Household Behaviour and the Environment, Reviewing the Evidence. Organization for Economic Cooperation and Development: Paris, France, 153-180.

[10] Isely, P., & Lowen, A. (2007). Price and substitution in residential solid waste.Contemporary Economic Policy, 25(3), 433-443.

(26)

[11] Gemeente Amsterdan (2015). Afvalketen in beeld. One-time publication by the Afval regieorgaan and the regieorgaan voor de Amsterdamse afvalketen.

[12] Jenkins, R. R. (1993). The economics of solid waste reduction.Edward Elgar Publishing, Hampshire.

[13] Mincer, J. (1974). Schooling, Experience, and Earnings. Human Behavior & Social Institu-tions No. 2.

[14] Fullerton, D., & Kinnaman, T. C. (1996). Household responses to pricing garbage by the bag. The American Economic Review, 86(4), 971-984.

[15] Morris, G. E., & Holthausen Jr, D. M. (1994). The economics of household solid waste generation and disposal. Journal of environmental economics and management, 26(3), 215-234.

[16] Nedvang, 2016. Retrieved from: http://www.nedvang.nl/gemeenten.

[17] Dugger, W. M. (1979). Methodological differences between institutional and neoclassical economics. Journal of Economic Issues, 13(4), 899-909.

[18] Richardson, R. A., & Havlicek, J. (1978). Economic analysis of the composition of household solid wastes. Journal of Environmental Economics and Management, 5(1), 103-111.

[19] Rijksoverheid 2016. retrieved from: https://www.rijksoverheid.nl/onderwerpen/afval/inhoud/huishoudelijk-afval.

[20] Sderbaum, P. (1992). Neoclassical and institutional approaches to development and the environment. Ecological economics, 5(2), 127-144.

[21] Tadesse, T. (2009). Environmental concern and its implication to household waste separa-tion and disposal: Evidence from Mekelle, Ethiopia. Resources, Conservasepara-tion and Recycling, 53(4), 183-191.

[22] Tinker, A. (1984). Theories of the state and the state of accounting: economic reductionism and political voluntarism in accounting regulation theory. Journal of Accounting and Public Policy, 3(1), 55-74.

[23] Williamson, O. E. (2000). The new institutional economics: taking stock, looking ahead. Journal of economic literature, 38(3), 595-613.

The role of urban governments in sustainably managing residential garbage

Residential garbage

Contents

1

Introduction

2

Literature Review

3

Methods and Data

4

Results

5

Conclusion