Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents
Citation: Ulli-Beer, S. G. P. Richardson and D. F. Andersen, (2004). A SD-choice structure for policy
compliance: Micro behavior explaining aggregated recycling dynamics. Proceedings of the 22"
International Conference of the System Dynamics Society. Oxford, England (July 25-29, 2004).
A SD-choice structure for policy compliance: micro
behavior explaining aggregated recycling dynamics
Silvia Ulli-Beer’
Institute of Management, University of St. Gallen, Switzerland
Interdisciplinary Center for General Ecology, University of Berne, Switzerland
Silvia. Ulli-B eer@ ikaoe.unibe.ch
George P. Richardson
Nelson A. Rockefeller College of Public Affairs and Policy
University at Albany, State University of New Y ork
gpr@ albany.edu
David F. Andersen
Nelson A. Rockefeller College of Public Affairs and Policy
School of Information Science and Policy
University at Albany, State University of New Y ork
David.A ndersen@ albany.edu
This paper presents a System Dynamics Solid Waste Management model that is based in a
feedback perspective on human behavior and public policy (Ulli-Beer 2003). A SD-choice
structure is suggested that both highlights the interactions of personal and contextual factors
and is suited to explain and forecast the impact and outcome of recycling initiatives and
strategies at the local level as well as to explore different scenarios (Ulli-Beer, Andersen et
al. 2004). The model structure indicates crucial dynamic interactions between flexible
preferences and contextual factors. Furthermore policy sensitivity of personal factors could
be identified that explain the success or failure of recycling initiatives. The policy experiments
show that combinations of interventions altering personal and contextual factors are crucial
for policy compliance.
Key words: System dynamics choice structure, computer simulation model, choice and
preferences, environmentally relevant behavior, policy compliance, recycling initiatives, solid
waste management
' The author would like to thank the dissertation-adviser-team Prof R. Kaufmann-Hayoz (member of my
dissertation committee, University of Berne,), Prof. M. Schwaninger (Chair of my dissertation committee,
University of St. Gallen) and the two co-authors (being important mentors not only but especially during my one
year long visit at the System Dynamics Group in Albany). They all contributed significantly to this work in
many different ways. Furthermore, I highly appreciate the support from many friends and colleagues who
provided valuable comments and encouraged me in the course of this study, especially from Aldo Zagonel,
Mohammad Mojtahedzadeh, Rod Mac Donald, Vedat Diker and Susanne Bruppacher. The work was financially
supported by the SNF from Switzerland and the Basic Research Funds of the University of St. Gallen,
Switzerland. I am greatly indebted to my husband, parents, parents in law and friends who were tacking care of
our children, when their mother was preoccupied by this work.
Introduction
In the current throw-away-society the challenge of recovering valuable resources and
protecting the natural environment is becoming more and more an urgent task for decision
makers at the national, regional as well as the local level. Although many industrial societies
have a well-organized management of solid waste, growing waste mountains and disposal
costs are signs of inappropriate production methods and behavior’. One way to alleviate this
development is seen in fostering recycling efforts. Thus it is important to understand driving
forces that will render recycling initiatives successful in the light of sustainability.
This study integrates knowledge and data from different disciplines and perspectives in order
to understand the multidimensional factors and processes leading to observed recycling
dynamics. Aggregated human behavior effects on the recovery of natural resources from a
macro-perspective are combined with individual behavior theories from a micro-perspective.
Understanding both levels may yield significant clarification of environmental problems and
provide a more complete basis for policy analysis and design, which is in line with Vlek’s
(2000) observations.
Thus the focus of interest is not mainly on the micro-process as the end but “lies instead in the
act of synthesis, in beginning to show how the micro processes combine to constitute a
functioning system” as was suggested by Coleman (1965:91) 40 years ago and is aimed at by
Forrester (2003) referring to the SD economic model: ,,In the system dynamics model that we
are discussing, the micro-structure creates macro-behavior".
In sum, this paper describes a SD-choice structure for policy compliance that is suited to
address aggregated recycling dynamics.
The investigation was guided by a general dynamic hypothesis suggesting that there exist
important dynamics interaction between citizens’ choice and preferences as well as public
policies affecting the public management processe. This general research focus was applied to
the specific problem addressing separation behavior in a typical Swiss locality? (Ulli-Beer
2003). In the specific case, separation behavior of citizens and policies influencing the context
in which choices are made, are analyzed and modeled in order to gain a better understanding
of local solid waste management problems that are also globally relevant (Duggan 2002),
(OECD 2000; Ludwig, Hellweg et al. 2003). The case study is thought to be relevant for
nearly every local solid waste management system concerned with recovering valuable
resources. Furthermore, separation behavior and solid waste management belongs to a
specific class of behavior, environmentally significant behavior (Stern 2000). It includes both
the kind of behavior that directly or proximally causes environmental change (e.g. separation
behavior of citizens) and the kind of behavior that shape the context in which choices are
made (e.g. local policies influencing the quality of services and their prices; both having
indirectly a crucial impact on the environment by determining the preconditions for citizens’
separation behavior).
> http://www.umwelt-schweiz.ch/buwal/eng/medien/umweltbericht/druck/index.html (Swiss Agency for the
Environment, Forest and Landscape)
° The municipality Ittigen was chosen since it is recognized as taking innovative approaches in their
environmental policies. It is located in the agglomeration of Bere (Switzerland) with about 11’000 inhabitants
(www.ittigen.ch).
The modeling effort draws upon a broad research program, the Swiss Priority Program
“Environment” (SPPE*) launched by the Swiss National Science Foundation. Furthermore,
the model structure refers to psychological and economical concepts such as planned
behavior, decision-making, preferences, incentives, social norms, routine and habits as well as
compliance (Ulli-Beer 2004:22-71).
The SD-choice structure described in this paper is relevant for enhancing the understanding of
relationships between personal factors (for example preferences) and contextual factors (such
as price incentives) influencing policy compliance. Basically two sorts of preconditions for
environmentally responsible behavior can be distinguished: contextual preconditions and
personal preconditions. Contextual preconditions include interpersonal forces, distinctive
features of substitutes, existence of different action alternatives and technologies, and design,
as well as infrastructures, monetary incentives and costs (including time costs) as well as legal
regulations. Personal preconditions include forces influencing values, beliefs, attitudes,
goals, self-identity, knowledge and skill, as well as general capabilities such as literacy
(Foppa and Frey 1985; Foppa 1989; Foppa and Frey 1990; Jaeggi, Tanner et al. 1996; Stern
1999; Stem, Dietz et al. 1999; Tanner 1999; Vatter 2001a; Bruppacher and Ulli-Beer 2001b).
The contextual preconditions are influenced by situational, restrictive variables and the
personal preconditions are influenced by cognitive, evaluative variables. Both variable
complexes influence intention-related variables guiding actions. “Preconditions for
environmentally responsible behavior’ seems to be a relative concept, being evaluated
subjectively: in other words there may be a trade-off between the impact of constraints on
behavior and environmental concerns or social norms. Higher environmental concerns
depreciate the perception of constraints and obstacles, where as lower concerms tend to
reinforce action constraints. Furthermore, high environmental concerns seem to increase the
effort and the willingness to spend time and costs on environmentally sound behavior.
Similarly, social support seems to increase the effort and the willingness to spend more time
and costs on environmentally sound behavior in order to overcome hurdles. However, often
habits dominate observed behavior patterns of people and deactivate deliberation processes.
Unfreezing such habits may be a tedious and time-consuming process.
Similarly, to economic concept of willingness to pay, citizens’ preferences are operationalized
with data on acceptable separating time and acceptable separating or burning cost.
However, there are some major differences in the overall choice concept of the SD-SWM-
model compared to the economic theory of consumer choice. First, the preferences can be
influenced by a social norm for separating behavior. Second, the observed separation pattem
is not described by an utility function that will be maximized, but rather by simple
deliberation processes comparing acceptable costs and real costs in separating and also by
comparing real cost and alternative action costs of not separating. Further, the SD-SWM-
model choice approach conceptualizes mainly two groups of people with different preference
structures - those that may develop intention to separate and those that may not. Finally, the
model structure also includes measures of the influence of habits that are actually not part of a
choice process. Therefore, this specific SD-choice structure may be seen as an important
building block of the overall SD-SWM-model including crucial psychological concepts
explaining individual behavior. This is seen as an important precondition for an adequate
policy analysis instrument trying both to exclude systematic disciplinary biases and to identify
important intervention points also considering changes in personal factors.
* http://www.snf.ch/SPP_Umwelt/overview.html
By focusing on interactions between various variables drawing on different disciplines in
order to explain environmentally relevant behavior, this computer assisted theory
development approach tries to exploit some of the research opportunities on human-
environmental interactions identified by Stern (2000). The understanding of those interactions
and processes are important for the design of effective policies not only in the realm of solid
waste management (Vlek 2000). Indeed, the specific SD-SWM-model can be generalized for
many recycling initiatives all over the world. It may serve as a decision support system that
help both enhancing discussions about solid waste management strategies among different
stakeholders and finding effective policy-packages that aim at increasing the quality and
quantity of separated waste. In sum, it informs environmental policy formation and decision-
making by offering a SD-model that helps to structure policy decision problems, that Vlek
(2000) also identified as an important future research line.
Methods
An integrative systems methodology (Schwaninger 1997; Weber and Schwaninger 2002) was
chosen that is especially suitable for investigating complex issues drawing on concepts of
System Dynamics and Cybemetics. A two-step research strategy was pursued. First, an
overall analysis of environmentally relevant behavior was undertaken in the preliminary study
and second, an in-depth analysis of the specific case was pursued in the main study. The
following chart (Figure 1) highlights the two research blocks of this study and gives an
overview of the applied research methods.
The Thesis
Dynamic Interactions between CC&P and PPI
a
Preliminary Study Main study
(Development of research heuristics) | | (Computer assisted theory building)
Literatur review SD-SWM-model development
(Interdisciplinary Studies of the SPPE) Quantitative modeling
Workshop with Experts Policy analysis
Qualitative modeling (GMB)
Figure 1: Structure and methods of the study (CC&P and PPI: Citizens Choice and Preferences and Public Policy
Initiatives) (Ulli-Beer 2004).
The purpose of the preliminary study was to explore and to shape the field of investigation.
Relevant concepts were identified and frameworks were developed that help to structure the
issue. In order to avoid a disciplinary bias no single disciplinary perspective or singly theory
approach was chosen. Instead a consistent research heuristic that is adequate for investigating
the complex issue has been developed. The framework “a Feedback Perspective on Human
Behavior and Public Policy”, (Kaufmann-Hayoz and Gutscher 2001a; Ulli-Beer 2003) was
used as a heuristic and substitute of a disciplinary focus. It helped conceptualizing the model
in such a way that the main relevant aspects of the multifaceted issues of solid waste
management could be addressed (see Figure 2). In the main study the computer-assisted
theory development method of System Dynamics was applied The findings and concepts of
the preliminary study were adopted to the specific case. It resulted in the SD-SWM-model
that could be used for policy analysis addressing “What-if’-question (Zagonel, Rohrbaugh et
al. 2004 forthcoming) about solid waste management policies (Ulli-Beer, Andersen and
Richardson 2004).
The paper is organized in four sections: First, the theoretical and political background as well
as the chosen research approach was briefly sketched in the introduction and method part.
Second, the model is described comprehensively beginning with a short overview of the
model conceptualization advancing to a full description of the model structure and closing
with a presentation of the model behavior including a base run description as well as a
summary of the model tests results One sensitivity test is highlighted in order to illustrate
policy sensitivity of the personal factor willingness to invest time in waste separation
behavior in the main section. Third, an outlook on using the model as a policy-laboratory is
given describing two different streams of policy-experiments and summarizing the main
insights of those policy-tests. Finally, the essay concludes with a reflection on how the
modeling approach was adequate for addressing the research interest in interaction between
personal and contextual factors explaining policy compliance and observed recycling
dynamics.
Recycling / Incineration sector
*price
*capacity / recyclable ratio
*time delay in capacity adjustmen’
#recycling streams
| The household waste
separation sector
*overall amount of waste
*amount of burnable & recyclabl
*amount of illegal deposing
| *purity of recyclable
| *relative prices
uo14njjod,,
suauuosjaua ay.
Income Supply sector
per # recyclable products
capita # take backpoints
ACTION PERCEPTION
The household decision sector
*willingness to spend time *willingness to spend money
*willingness for compliance *habits i
*knowledge *willingness to learn
*perception *overall goals
Figure 2: A feedback perspective on citizens’ waste separation behavior and solid waste
management.
Model conceptualization
In the following paragraph a brief overview of the model conceptualization and the dynamic
hypothesis will be given. Both are well documented in Ulli-Beer (2003).
The problem guiding the model conceptualization and drawing the boundary is represented in
the following questions:
What local policies increase recycling, and help to establish / ensure a solid waste
management system that fosters competitive recycling markets?
This overarching question was cut up in the following more specific ones:
e How do you motivate the households to participate in solid waste separation?
e How do you recover recyclable material in order to produce competitive secondary
raw material?
e How do you finance the recovering and disposal activities of local agents?
For the case study the reference mode has been defined by the historical dynamics of the
budget of solid waste management showing a recurrent deficit as well as the development of
the fraction separated waste and of the number of recycling streams. Subsequently, the
question “What caused the given development?” (Randers 1996) was addressed in the
modeling process.
Budget: solid waste management
( sFr. per capita per kg)
0.3
028. iT]
0.2
0.15 4 —o— expenditure per capita perkg per yearsFr. |_|
0 1 —2— revenue per capita per kg per year sFr. [|
00s —e— deficit per capita per kg per year sFr
0 i _a a
$ DP SY Sh > oh ND
oO SY Bh SP? OB of SS
PP RP PER PT A NT oh
Chart 1: Municipal budget development for solid waste management in sFr. per capita, per kg, per year ((Ittigen
1985-2001). There is an upward trend in the unit cost that peaks in 1994 followed by a slight drop, and then it
seems to reach a plateau. However the revenue continues to fall. There are two periods with a higher deficit
(1987 — 1992) and (1996-2001). As the deficit has grown, the local authorities increased the tax for solid waste
management and the volume related garbage bag charges.
In short, the postulated dynamic hypothesis can be described as follows.
Since the performance of citizens’ separation behavior was low, the local authorities gave
price incentives in form of a garbage bag charge implemented in 1991. The intended effect
was to promote the separation behavior. As a consequence the fraction of separated waste
increased and the relative amount of solid waste for burning decreased. The unintended effect
was that not only the relative amount of waste disposed for burning decreased, but also the
revenue generated from the trash bag charges declined. Therefore, the budget deficit started to
increase. A further increase in the price for bumnable material had nearly no additional effect
on the separation behavior, since the number of recycling streams was held nearly constant.
The citizens had no real legal option to avoid higher costs for disposing of bumable material.
As an unintended consequence, the quality of the separated material decreased. Citizens
started to put burmnable material in the recycling streams. However, this effect was only
observed and could not be exactly quantified.
For a comprehensive illustration of the postulated causal loops explaining this development
see (Ulli-Beer 2003).
The model structure
Before entering in the discussion of the detailed model structure a brief overview of the model
is given. It gives the big picture justifying the chosen aggregation level and shows the
different model parts and their relationships as well as basic feedback loops.
In order to analyze long-term effects of different local policy interventions a time horizon
from 1987 to 2020 was chosen. For the time period 1987 to 2001 there is data available
revealing historical patterns of behavior. The overall model structure is described below and
sketched in the Sector Diagram (see Figure 3):
The main sector is the local separation sector that is disaggregated in the following modules:
the household waste separation module, the household decision module and the local policy
module. These modules include endogenously operating dynamics deemed important to
address the solid waste management problems and to conduct policy analysis.
The household waste separation module includes:
e The different flows and qualities of the bumable and recyclable waste that result from
separation activities of different groups of citizens.
e The initial amounts of different waste qualities, and recyclable and burnable material
will be given exogenously but will be modified by behavioral effects.
e The habits of different groups of people to dispose their waste and factors that lead to
changes in habits (e.g. changes in relative prices and the number of recycling streams).
The household decision module will describe:
e What factors influence the decision of people to become willing / unwilling to separate
the recyclable material?
e What influences the willingness to spend time or money on waste separation
activities?
The local policy sector / solid waste management module includes:
e The development of the garbage bag charge and the municipal budget for solid waste
management under different policy options.
e Capacity building processes and the effect of a backlog of separated waste.
The income per capita and the population are given exogenously.
Perception of pallution
Production/
Indicated number of
recycling streams
Recycled raw material
Incineration sector
*price
*number of recycling streams ‘Amount of separated or
recycled raw material for burning
rs
A Impact on environment .
errr ee eee r rrr rrr rr rrr rrrrr errr rere rreerre rrr irri rrr rrr rr irr rrr
Figure 3: Sector Diagram of the extended SD-SWM-model
The modules of the model
This section describes the core model parts in depth. The concept of propensity - the
propensity of citizens to separate - is crucial for the success of recycling programs. Therefore
it will be modeled explicitly. A special weight is put on the formulation of the decision
process guiding citizens’ behavior to separate.
In the feedback perspective on human behavior and public policy (K aufmann-Hayoz, Battig et
al. 2001d), contextual and personal factors in a decision making process are emphasized.
Therefore in the SD-SWM-model, interactions between those are modeled explicitly.
Elements of the personal structure are represented in the household decision and the
household separation sector.
Designing propensity to separate: The household decision module
Theoretical and empirical evidence suggest that citizens’ disposal behavior may be described
partly as routine behavior and partly as planned behavior (Piorkowsky 1988; Scitovsky 1989;
Dietz and Stern 1995; Dahlstrand and Biel 1997; Gorr 1997; Taylor and Todd 1997, Ajzen
1991). In Forrester’ s terminology this would be called an informal policy (Forrester 1994:58).
These assumptions suggests that people decide once whether to separate or not. Once they
have made this decision, they set a new routine, resulting in new separation habits. This
implies that there are two main groups of citizens: a group of people willing to separate and a
group of people not willing to separate. However, in each population we can distinguish sub
groups that are transients (see Figure 4):
e In the group of people willing to separate there are some inexperienced people - they
will show a lower separation performance than the experienced ones. But as they leam
to separate they will move into the stock <ep willing to separate>. The <time to
learn> determines how long this takes. In the model the <time to learn> is
represented by the variable <time on moving from iep to ep>. It is a function of the
<average amount appropriately separated by nwiep> and the <normal amount
appropriately separated wep>”.
In the group “people not willing to separate” there are experienced people that got
disappointed from separation consequences. The <experienced people not willing to
separate> will move into the stock <ep not willing to separate> as they will forget, they are
changing their separation behavior and set up a simpler routine behavior. The <average time
to forget> calculates when these people will move on.
5 Acronyms: ep - experienced people; iep - inexperienced people, wep - willing experienced people, nwiep - not
willing inexperienced people
10
effect effect
of
> aan
sh fractional rate unpays
crowding becoming unwilling ™ eharge
Lk
disappointed ep
people willing to separate people not willing to separate
I time to learn disappointed [iim - mn forget Y
xperienced inexperienced iep a pecple not 1 experienced
pedple willing to people willing willing to [“Sveo'tocing | People not willing
separate wiep getting | to separate iep getting separate P 4 to separate
i i experienced motivated experience
i
ep/getting remotivated
illing to a”
separate fraction becoming
wa
fraction
willing
perceived
social norm
separating
fraction
<effect of
separating
cost>
<effect o
time cost
separating»
rom social
norm>
Figure 4: Changes in citizen’s willingness to separate (ep: experienced people, iep: inexperienced people, wiep: willing inexperienced people, nwep: not willing experienced.
11
Decision rules
The decision to separate is influenced by the <perceived social norm separating>, the
<acceptable time for separating>, and <acceptable separating cost per year>. The decision
to become unwilling is influenced by alternative cost such as <acceptable time burning> and
<acceptable unit cost for burning> and <perceived social norm burning>. The information
about the decision cues (e.g. “time cost”, “real cost’, later on the “perceived policy
effectiveness”) are partly given exogenously.
fract becoming willing f social norm separating
effect of time cost separatin
ff f tim Pp 9 eennasveneey
mate | \
ratio separation cost
separating P
ea . to acceptable
«ratio
recyclable to
appropriate >
a ue \ s +
ceftectve 5 /hormat tine
recycling atreamsn jf eel tise
/ aece Mevernae ecceptoble “separetion ‘waste per
for separating parating ‘cost p capita wep>
f f
| stgaetien
unit separation
ax acceptable
he i“ cost
4
}
aration cos
perceived max acceptable
Fractional cha] social norm | separating time
erception soci! separatin
percept i SEE g unit separation al
norm = _— space cost vile fae
} \ Sees eel separated material
4
< fraction «price sep
willing to aration>
separate>
Figure 5: Interactions between contextual and personal factors
The <fraction becoming unwilling> is formulated in a similar way as the <fraction becoming
willing>, but the effect of opportunity cost <effect of time cost burning> and <effect of
burning cost>, as well as the <fract becoming unwilling from social norm burning> will
determine the rate. The rates are determined by a multiplicative formulation, since any
extreme value in each of them can dominate the other effects as well as one effect can also
reinforce another. The concrete formulation for the <fraction becoming willing> is:
<Fraction becoming willing> = <fract becoming willing f social norm separating> * <effect
of time cost separating> * <effect of separating cost>
In addition, it is assumed that the two stocks <ep willing to separate> and <iep not willing to
separate> will never get to zero. There will always be a fraction that will not change its
behavior. This design would represent people with strong beliefs, people that just do not see
any reason for separating, or that are over occupied by the separating task.
The household waste separation sector
In the household waste separation sector, four different qualities of waste will be computed.
The waste generated consists of recyclable material (A-waste) and non-recyclable material
(B-waste). Therefore, the people have four different action choices to dispose the waste (see
Figure 6).
12
. 7 multiplier for
Four action choices recyclable material
from number of
recycling streams 0
<experienced people
willing to separate>
<waste per capita
per year wep>
waste generated
by wep per year
actual recyclable
material per person O
actual total amount actual possible
nonrecyclable material “= recyclable amount
from wep per year from wep per year
B1 B2 A2 Al
nonrecyclable inappropriately recyclable disposed appropriately
di df i
Burkina by wep <— separated by wep for burning by wep separated by wep
per year per year per year
per year
4 SS Zz 4
<experienced people
willing to separate>
Figure 6: Action choices for disposing of the waste (wep: willing experienced people)
The behavioral variables (indicated by diamonds) represent disposal habits. They measure the normal amount inappropriately separated (B2-waste) and the normal amount
appropriately separated (Al-waste). They also determine both counterparts: the amount recyclable disposed for burning (A2-waste) and the non-recyclable disposed for burning.
(B1-waste).
13
A: The recyclable material can be appropriately separated (A 1) or can be disposed for burning (A2).
B: The non-recyclable material can be disposed for burning (B1) or it can be inappropriately
separated (B2) (generating impure and more expensive recycling material). Figure 6 illustrates how
the different qualities of waste are computed.
The local policy sector
The local policy sector includes two basic structures. Firstly, a simple budget structure with a price
building policy determining the garbage bag charge and secondly, a simple capacity building
structure for collecting points under a regime of a prepaid disposal tax for recyclable material.
These structures capture important feedback loops between different financing alternatives of solid
waste management (garbage bag charge, prepaid tax or prices for separated material) and the
separation behavior of the people. Specifically, they also represent a non price mediated resource
allocation system (Sterman 2000:172). With this structure the model boundary includes all the
important feedbacks that where detected in the proposed dynamic hypothesis and are thought to
include the most relevant structures in order to address the problem statement endogenous to the
model.
Policy structure: Garbage bag charge
In this policy structure the <garbage bag charge> is computed endogenously. Its adjustment is
controlled by numerous feedback loops. The basic underlying decision policy is a goal seeking
decision-rule leading to a zero deficit budget. The following Figure 7 emphasizes the three main
feedback loops in a simplified model structure. In the case of increasing cost of solid waste
management the two reinforcing loops - “less burning increases price” and “more separation
increases cost” - lead to a steady increase in the <garbage bag charge> where as the balancing
loop “less burning reduces cost” would limit the growth. But since the pool of people that could
become “willing to separate” is limited the growth in the <garbage bag charge> will be restricted
by the overall number of <people separating>, as well.
lgarbage bag a
charge peeple
adj gbc tt) separating
less burning increases price
indicated gbc
fi SS ssunei nr gbcqg— weight per gbe
—) .
less burning reduces cost
cost that should be
* covered by bag charge
var unit
cost for
burnable
waste
total
amount
separat
ed
more separation increases cost
var unit
cost for
separated
material
Figure 7: Main loops controlling <garbage bag charge> adjustment.
14
The variable cost that should be covered by bag charge is mainly determined by the amount of the
different waste qualities and the unit costs. Revenues from sources like taxes or from selling
separated waste are subtracted.
Policy structure prepaid disposal tax
Figure 8 depicts the model structure that allows simulating the impact of a prepaid tax policy. In the
policy structure prepaid disposal tax, the decision rules guiding the capacity building process in
the take back points is illustrated. On the one hand, the <average amount recovered material>
determines <capacity building>. On the other, the <perceived revenue from the prepaid disposal
tax> limits the capacity building process. A gap between the <average amount recovered
material> and the <capacity in the collecting points for recovering> leads to a crowding effect.
The crowding effect feeds into the household decision and household separation sector
influencing the rate <fraction becoming unwilling>.
Propensity to
separate
(Household decision
and separation
sector)
desired capgfity adjustment
crow din
ee
effect of crowding
Propensity to
separate
(Household decision
and separation
sector)
Figure 8: Policy structure: Prepaid disposal tax.
15
The Base run
In this section, the actual implemented solid waste policy (inertia policy) as the base-run is
described. The simulation runs allows to test the correspondence of the model behavior to the
reference modes and show the dynamics of the model-structure with the inertia policy. The
outcome of the policy-experiment will be measured with the following indicators / variables of
interest:
e The simulated values of <fraction separated>, <fraction for burning> are depicted against
the smoothed real data.
e number of the different groups of people willing respectively not willing to separate: <ep
willing to separate>, <iep willing to separate>, <iep not willing to separate>, <ep not
willing to separat>
e <total amount appropriately separated> and <total amount inappropriately separated>.
These amounts will be depicted against the <total amount recyclable material>.
e <garbage bag charge> and <price for separating> and the <profit of solid waste
management>.
Those indicators are thought to be crucial for measuring the performance of the inertia policy and
further policy-packages that can be analyzed with the model structure within different experiment
designs.
Inertia policy
The base run describes the model behavior with the actual policies in place: an increase in
<effective nr recycling streams> and an increasing <garbage bag charge> (endogenously
computed inertia policy 2). The simulated <fraction separated> and <fraction for burning>
closely tracks the smoothed real data (see Chart 2 A). There is a clear trend of growth in the
<fraction separated>. Based on the historical growth trend, the model data indicates a further
increase in the <fraction separated> till it seeks equilibrium that will be slightly higher (54%) than
the actual fraction (50%).
The dynamics are created by the flow of people respectively by changes in the number of the four
different groups of people willing / not willing to separate. Chart 2 B shows a clear increase in the
number of <ep willing to separate> beginning in 1991, and a decrease in the number of <iep not
willing to separate>.
Chart 2 C illustrates an increasing trend in separated material. However, the price incentives lead to
a sudden increase in the <tot amount inappropriately separated> in 1991. But the decreasing trend
in the number of <iep not willing to separate> causes a smoothed decline in the <tot amount
inappropriately separated> seeking equilibrium. This dynamics represents a classical “first-worse-
before-better” behavior pattern. As a consequence of this behavior the gap between <total amount
recyclable material> and the <tot amount appropriately separated> decreases, resulting in a
smaller constant compliance gap.
Chart 2 D illustrates the increase in the <garbage bag charge>. The model structure computes a
<garbage bag charge> that seeks a zero profit goal. A change in the <effective nr recycling
streams> creates the opportunity for people to separate more material, which has two effects.
Firstly, it reduces the <total amount disposed for burning> resulting in less revenue. Secondly, it
increases the cost for collecting the separated material. These two effects result in a short and
minimal budget deficit due to price adaptations delays in the <garbage bag charge>. The
<garbage bag charge> levels off at 2.1 CHF/bag (D).
However, those price adaptation delays are much longer in the real system. This adaptation delay
creates the observed budget deficit in the real world. Chart 2 E illustrates this case in the inertia
16
policy 1-experiment (garbage bag charge exogenously given). It is simulated with the <garbage
bag charge exogenous> resulting in a budget deficit between 1993 and 2000, appearing again after
2001 (see ge between line 4 and 5 <profit solid waste management> and <non-profit
threshold>)°.
REFERENCE FRACTION SEPARATED MOVING PEOPLE
oa 10,000
06 7,500
04 5,000
02 2,500
0 2 4— 4
0 22349 4
9871990 195 TOSTO9S TOOT T00S TOoH TOIT FOIE TOIT —Fo20 i SS SS Se
Time (year) Time (year)
the nsdn snoth ncivnsoumied nein poky 2 2222 den ep win to spate ini poy 2. —¥-—14-111~ people
i ni etn petal ONT? gs ip lay lose neta juicy? 99999 poh
feral feergircyy r = Spat hey seeps aayelyy2 AAS Fo
sp aot waling to separate inertia poly 2-44 A 4 4 people
(A) (Line 1 and 2 are three median smoothed real data)
(B)
WASTE SEPARATED PRICE AND BUDGET
4M kglyear 4 Dollars/bag
60,000 kaiyear 02 Dolaishg
0 Dollarsiyears
2M kghyear 2 Dollas/bag
30,000 kglyear — + 0.1 Dollars/kg
abe 4 400,000 Dollarsiyears
© Dollars/bag
¢ ee 0. Dollars/g
1967 199719992008 d01T__ 2017 Fa00, AOD oles geare al 4 a1 iz [| 2 3
Time (veer) 1987 19931999 2005 2011 2017
Time (year)
total amount appropriately separated : inertia pobcy 2 1141 kglyear
total amount recyclable material: nea poley 2.222 2 2 kayear atbage bag charge: inertia policy 2 —_1}___1__1__4__otasmbag
ind inappropriately soparsed vinta pokey 2 ——9-—3-—3-—- Eglyoar datbage hag charge exogenous :norla poly 2 2-2 2 2 — Dollasibag
aa on pice separation : neta poly 2. 333 3 3 — Dollasihg
profit solid waste management: inerta policy 2 ——& 4 4 4 Dollas/years
(@) non profit thushold neta poley 2-58 88 Dollan/years
(D)
PRICE AND BUDGET
4 Dollarsbag
0.2 Dollars/kg
0 Dollarsyears
2 Dollarsbag
0.1 Dollas/kg
400,000 Dollarsiyears
© Dollarsbag
0 Dollars/kg
800,000 Dollarsiyears
oe oe ee
1987 199319992005 2011-2017
Time (year)
garbage bag charge: inota policy 1 —_)___4___4__4__ oltassbag
Garbage bag charge exogenous : inertia policy 1 2-222 Dallarsfbag
price separation : mera poly 1 33 333 ollan/ag
profit solid waste management: inertia policy 1 4 Dollarsiyears
non profit threshold inertia policy 1. $$ $$ ollarsiyears
(E)
Charts 2 A-E Dynamics of the base runs
The simple base runs illustrate a good plotted fit (Chart 2 A , for example line 2 and 4) to the
reference mode. It also reveals a recurrent deficit as observed in reality. However, the model
behavior was explored further around different policy-experiments. Mainly three streams of policy-
experiments were conducted (see Figure 9).
5 For additional back-casting experiments addressing “what if else’-question see Ulli-Beer (2003).
17
1. back-casting policy-experiments depicting the actual policy in place (inertia policy), as well
as further “what if” policy-experiment exploring what would have happened with different
policies
2. forecasting’ policy-experiments analyzing the effect of new policies such as implementing
prepaid taxes
3. policy-experiments under different scenarios over the time horizon 1987 to 2020.
While a comprehensive report on those policy experiments can be found in Ulli-Beer (2003) for the
back-casting experiments, and Ulli-Beer, Andersen et al (2004), for the forecasting and scenario
experiments, in this paper only a rough idea on the policy-experiments under different scenarios
addressing uncertainty in the system can be given after the following conclusions on model
behavior and testing.
Flexible or constant | Policy
hi leverage
garbage bag charge Fee
loarbage bag
people
tt) lseparating
Policy less burning increases price
leverage indicated gbe
point + ce
charge
adj gbe
‘assumed nr gb cag weight per gbe
total
amount
disposed
Price for
separated material \ less burning reduces cost
or
burning
Cost that should be
* covered by bag charge
total
amount
separat
"ed
i)
more separation increases cost
var unit
cost for
separated
material
Capacity in exogenous
collecting points financed by
prepaid disposal charge
Figure 9: Policy-leverage points in the simplified model structure: The CLD depicts the policy-
leverage points in the system and illustrates their impacts on the main loops determining system
behavior.
Summary model behavior and testing
Testing the model was an integral part of the modeling process, including structure assessment and
behavior reproduction tests. The units of each variable and equation were specified during the
modeling process and helped to build up a model structure that is both dimensionally consistent and
based on variables that have real world meaning (they are operationalized). In addition, the model
structure is based on theoretically and empirically well-founded assumptions that generate a
plausible behavior and shows a good fit to the data. The model passed three extreme conditions
tests showing that the model exhibits a robust behavior even under extreme parameter and policy
variations.
7 The term forecasting refers to the time horizon in the future and indicates that the effect of a policy interventions made
in the future will be analyzed. Contrarily, the term back-casting refers to a policy intervention that was made in the past.
18
Sensitivity Analysis: The tipping point and policy sensitivity
With sensitivity analysis the effect of uncertainty in model assumptions on policy conclusions were
tested. These tests assess the robustness of policy implications that can be drawn from the model
output. The basic idea is to test if the model behaves plausible under different parameter values, and
if changes in parameters lead to different policy implications. A further objective of sensitivity
testing was to find the most influential parameters. This would be the best intervention points for
effective policies. The choice-structure of the model can be very sensitive to small parameter
changes, since it features a tipping point. If parameter values operate near the tipping point the
implications drawn from a policy-experiment become weak. Hence it was indicated to conduct
sensitivity analysis.
The main stock and flow structure of the model (see Figure 4) has similar characteristics as basic
epidemic and innovation diffusion models such as the SIR-model?® or the Bass-model? (see Kermack
and McKendrick 1927; Bass 1969; Bass, Krishnan et al. 1994 in Sterman 2000:300ff). The
diffusion process is boosted by the second-order reinforcing feedback structures. The exponential
growth or decline is limited by first order control loop structures, controlling the overall growth
capacity (such as the number of <people not willing to separate> and <people willing to
separate>) hence, resulting in s-shaped growth.
As mentioned above, the tipping point can be decisive. If the diffusion process does not take off the
policy initiative is likely to die. The question of whether the policy initiative will succeed is a
question about which feedback loops are dominant (see Richardson 1995). The recycling initiative
will succeed if the positive loops controlling the rates “getting motivated” dominates the positive
loops controlling the rates “getting disappointed’” otherwise the initiative will fail (see Figure
10).
The strength of the loops is determined either by the “fraction becoming willing” or the “fraction
becoming unwilling”. If the <fraction becoming willing> gets stronger than the <fraction becoming
unwilling> the model exhibits s-shaped growth otherwise it exhibits s-shaped decay. Therefore,
some small changes in parameters could change a growth-trend to a decay behavior.
® The SIR-model is widely used in epidemiology for simulating the infection process of acute diseases. It mainly
contains three stocks, the susceptible population (S), the infectious population (I) and the recovered population (R).
° The Bass-diffusion-model mimics the diffusion of innovation of new product growth and is widely used in marketing
and for strategy-development and management of technology.
° Probably, this loop could also be named “getting discouraged” or “demotivated”, since different psychological
concepts could be used to explain the process that lead people to decide against waste separation. However, in this book
this loop will be called uniformly “getting disappointed”.
19
Policy-
Policy effect ee. EE aes a
of garbage progtion affectof ™ effectof effect of Bre
bag charge unwfoa time set burning cost crowding x
social burni
a \ Fi oe
fraction becoming zs, iy nae
F ru —
getting disappointed
Tits Yo getting dsoppointed
people williry to separate (—-#
_ ing disappointed
A nite opRTmvtvaeg POOPIE NOt wilf Ing to separate
getting motivated
: ct
= : Fvated Policy-
7 a Seti motival effect of nr
fraction becoming willing recycling
streams
popiation
Policy-
fraction becomit
willing ? social norm effect of time en SICA
jarating cost separating separa ck ays cost price for
separating
Figure 10: Simplified model structure and policy effects. A higher garbage bag charge weakens the loop “getting
disappointed” whereas an increase in the <effective nr recycling streams> increases the <effect of time cost
separating>. This will weaken the loop “getting motivated”. A price for separated material will have the same effect.
Furthermore, it is assumed that a prepaid tax could lead to a <effect of crowding> |! resulting in an enforcement of the
loop “getting disappointed”.
For policy analysis those parameters are important that are uncertain, such as <max acceptable
separating time> or those that are highly relevant to political interventions, such as <normal time
per stream>. In the following paragraph, exemplarily one sensitivity test will be briefly reported
that demonstrates policy sensitivity of the personal factor <max acceptable separating time>.
The parameter <max acceptable separating time> operationalizes an average willingness to spend
time on separating. If the time required for separating is higher than the average willingness then
the loop “getting motivated” will lose some numerical strength. Therefore, the parameter is
influential. Furthermore, its real empirical value is unknown and therefore in the model it is defined
with a high uncertainty by the modeler. For the sensitivity test the uncertainty range in this
parameter is specified as follows.
' The <effect of crowding> refers to the stated dynamic hypothesis of prepaid disposal charges, that is designed as the
archetype structure of non-price mediated resource allocation system adopted from Stermann 2000 and explained in
Ulli-Beer (2003).
20
Test settings <max acceptable separating time>
Distribution: random uniform, 30 runs
Model value: 2 Minimum value: 1 Maximal Value: 3 (hours/week)
Chart 3 illustrates that in this range of uncertainty the model operates near a tipping point. The
model shows a behavior mode sensitivity; it generates different patterns of behavior ranging from s-
shaped growth to overshoot and collapse to a smoothed decline. Comparing these test-outcomes
with the Base run shows that the parameter <max acceptable separating time> exhibits policy
sensitivity. The model has two different basins of attraction resulting from a shift in the dominance
of the two positive feedback loops. The confidence bounds diminish as the parameter values near
the borders of attraction. The behavior bifurcates when <max acceptable separating time> falls
below 1.65 hours/week (tipping point). ”
sens max acc sep inertia p2
fraction separated
0.6
0.45,
0.3
0.15
0
1987 1995 2004 2012 2020
Time (year)
A inertia policy2 —
Chart 3: Sensitivity analysis <max acceptable time> with inertia policy 2 depicting the dynamics of <fraction
separated>.
The sensitivity test emphasized exemplarily that the tipping point is an important insight that has to
be taken into account for policy recommendations. It determines the failure or success of a
recycling initiative and knowledge about the critical system parameters is decisive. In order to take
account of the identified tipping point and to analyze its policy implications a series of scenario
policy-experiments in the sensitivity analysis mode were conducted. The robustness of different
policies under worst- and best- case conditions were investigated.
Policy-experiments under different scenarios and imperfect information
Different scenarios are determined through changes in the surroundings that are not initiated by
local authorities (see Figure 12). Contrarily, the conditions in the external environment will
determine certain conditions of the solid waste management system, as well as the effectiveness of
local policies. In the model, changes in exogenous parameters define different scenarios. The
scenario leverage point solid waste generation reflects the trend in the overall waste generation and
in the <total amount recyclable material> (determined by a growth in the variable <effective nr
recycling streams>). The scenario leverage point changes in market conditions reflects the effect
of market prices in the incineration and recycling industries on the outcome of a local recycling
initiative.
21
a garbage bag i,
ae charge + pore
adj gbc tt) separating]
less burning increases price
+
indicated gbe
+ Tss5
‘assumed nr gbc~<t—weight per gbc
Scenario leverage
point : Changes in
market conditions
-) :
less burning reduces cost
5 Scenario
‘Ai leverage point:
Solid waste
generation
cost that should be
+ cpvered by bag charge
+
var unit
cost for
burnable
waste
var unit
cost for
separated
material
Figure 12: Scenario leverage points influencing main loops.
22
In the _ scenario-experiments _—two- [ (one Seufoprax cbc
inertia policy 2
dimensional changes in the scenario | ** 7% NO
parameters, determining either best-case or | “jy 7"
worst-case conditions, are analyzed. In the
model they can be specified in ranges | 04s
determining for example _ best-case
conditions in the recycling market with 03
lower prices than in the base run or worst-
case conditions with higher prices in the | 015
recycling market.
S067 1995, 2004 2012 2020
The results of the various scenario- ; Time (yeah
experiments under uncertain conditions (A) Fraction separated: Prepaid tax policy with constant
(conducted in the sensitivity an alyzing- garbage bag charge under worst case conditions.
mode) give evidence that the design of hy en ea
policy-packages matters the most under eon seren =
worst-case conditions. o8
If the system were biased towards a
favorable situation, both a garbage bag os
regime and a prepaid tax regime would lead
to an optimal outcome with high certainty. o
Given such a situation, the difference
between the inertia policy and a prepaid
tax regime in respect to the <fraction
separated> and the <accumulated fraction ‘987 1995 2008 2012 2020
Fi terial i ted te> i Time (year)
Impure material In separated wasie> 1s (B) Fraction separated: Prepaid tax policy with constant
small. Both regimes will be nearly equally garbage bag charge under worst case conditions but with a
effective. The main goal of the prepaid tax | higher <acceptable separating time>
regime could be reached, that is to disburden . . ;
the municipalities from the high cost. a Bi eetie does ta aliatetocln ”
However, uncertainty in the system could | <Q crise Mpa Cog oe ma
“ . a outcome even under worst case condition was tested.
raise some issues that have to be considered,
especially when we have to expect disadvantageous conditions. Worse conditions strengthen those
loops that drive the dynamics towards the lower border of attraction, resulting in a failure of the
recycling initiative. In all experiments under worst-case conditions we can observe that the
<fraction separated> converge toward the lower limit. But the large confidence bounds give
evidence that little changes in the uncertain parameters, such as in <acceptable separating time>
may have a significant effect on the outcome (see Chart 4).
Discussion and Conclusions
The study focused on relationships and interactions between important factors that influence the
separation behavior of citizens. Using the system dynamics syntax of stock and flow and the
mathematical formalization a material’? dynamic theory for recycling dynamics including a SD-
choice structure for policy compliance is suggested. Complex multivariate relations and processes
2 Glaser, B. G. and A. L. Strauss (1967). The Discovery of Grounded Theory. Strategies for Qualitative Research.
Chicago, Aldine. distinguish material from formal theories. A material theory is developed for a specific topic (such as
solid waste management) whereas a formal theory is more abstract and refers to conceptual issues such as
environmentally relevant behavior.
23
over time were visualized in causal loop diagrams that explain the observed phenomena in solid
waste management (see Figure 4 and Figure 7).
In this section those insights will be summarized and discussed.
The success or failure of a recycling initiative depends on the relative strength of the two loops
“getting motivated” and “getting disappointed”. The loop dominance is crucial. Therefore, those
factors are crucial that determine the strength of a loop. Sensitivity analysis and various policy
experiments under different scenarios helped to identify high leverage points for controlling the
system behavior. These insights help to determine why and when which preconditions for
environmentally sound action should be adjusted.
In the course of the study, a switch in perspective took place. While in the preliminary study the
focus was on key-factors explaining environmentally relevant behavior, in the System Dynamics
analysis key-loops were under investigation. System Dynamics offers the unique possibility to
identify feedback loops as causes of system behavior. Richardson and Pugh (1981) point this out as
follows :
“The feedback view antiquates the notion of a simple, linear, left-right causality. Chickens
and eggs are not a causal dilemma if one focuses on what they cause together, namely,
exponential growth in the barnyard. So, in hunting for the causes of model behavior, we
seek feedback structures, not isolated variables. While a single factor can change the
strength of a feedback loop and affect its dominance in the rest of the model, it is more
useful to see the loop, not the factor, as the causal agent in the system” (268).
With this focus the research questions can be discussed from a key-loop-perspective. With the help
of the SD-SWM-model five main loops were identified as the causes of the observed behavior
patterns, they are listed in the Table 6.1 below. Furthermore, the factors are identified that
influences their strength. In addition, they are traced back to personal and contextual factors. Also,
the theoretical concepts are named that explain the applied rationale forming the model structure
and subsequently the loops.
The result of this key-loop-analysis can be structured according to compliance, natural
environmental effectiveness and economic rationale issues (see Table 1):
Under the aspect of citizens’ compliance with the recycling initiative no further interventions may
be necessary, since currently the loop “getting motivated” dominates the loop “getting
disappointed”. Under the given situation the recycling initiative in the municipality under
investigation is successful.
However, the identified tipping point in the systems give evidence that a successful separation-
strategy has to be sensitive to the <effective nr recycling streams> that are offered to citizens. Thus
it is indicated that the cost efficient recycling strategy “offering different recycling streams and
investing in citizens’ separation behavior” has to be adjusted as soon the max separation capacity of
citizens is reached. It depends on the <acceptable time for separating> that interacts with the
<perceived social norm separating>. The upper limit of <max acceptable time separating>
indicates the maximal capacity of citizens to separate their waste. The dynamics in impurity is a
consequence of an initial policy resistance and adjustment delay in personal factors such as in
<acceptable time for separating> and <acceptable unit cost for burning>.
However, the economic rationale contradicts this observation. The recurrent deficit and the high
cost of solid waste management for the municipality are dissatisfying. The growing cost is mainly a
consequence of a successful recycling initiative but also of impurity and growth in solid waste
generation. The observed deficit is a logical consequence of the structure of the system and not one
24
of mismanagement of solid waste at the local level. The dynamics of countervailing price effects
lead to a trade off between policy effectiveness and a zero profit budget goal. Whereas an effective
policy tends to lead to a deficit, policy failure tends to lead to a profit. Hence the economic rationale
suggests further interventions. Subsequently the dominance of the three loops “less burning
increase price”, “less burning reduce cost”, and “more separation increases cost” together with
the two compliance loops “getting disappointed” and “getting motivated” have to be carefully
controlled.
In addition, the Table 1 brings out a rather astonishing point related to aspects of environmental
effectiveness. In the model no loops controlling the environmental effectiveness were identified.
This can have several reasons: The model boundary may be too narrow or no controlling loop may
exist at the local level, or this structure is missing in the model.
Main issue | Important loops Main factors and effects Personal and contextual Underlying
determining the strength of | factors influencing the theoretical concepts
the loops strength of the effect on the
loops
Compliance | “Getting Effect of burning cost, Capacity in collecting point Opportunity cost,
disappointed” Effect of crowding e Garbage bag charge Planned behavior
e Acceptable unit cost for Non-price-mediated
burning resource allocation
¢ Perceived social norm for ¢ Planned behavior
burning (choice concept)
“Getting e Fraction becoming willing | e Perceived social norm
motivated” from social norm separating
separating, Acceptable time for separating
Effect of time cost (Max acceptable separating
separating time)
e Time spent for separating
(effective nr recycling streams,
normal time per stream)
Environ- No controlling Underlying theoretical concepts
mental loop! e Habits
effective- e — Separation capacity of citizens
ness e — Sufficiency strategy (self-modification strategies)
e — Efficiency strategy
e — (Technological progress)
e —_Consistency-principle
Economic | “Less burning e Number people separating | e Price policy of community Economic incentive
rational increase price” —_| » Zero deficit budget policy _| « (cross subsidizing, price Polluter pay
adjustment delays, base tax) principle
“Less burning e Number people separating | e Incineration cost per unit Cross subsidizing
reduces cost” Cost that should be covered | e Unit cost for collecting effect
by bag charge burnable material Cost recovery and
© Zero profit budget policy _| e Total amount disposed for break-even point
burning
“More separation | ¢ Cost that should be covered | ¢ Capacity in collecting points
increases cost” by bag charge Total amount separated
Zero deficit budget policy
Table 1: Key loops and key factors explaining recycling dynamics.
Summing-up, the shift in perspective gives evidence that the significance of preconditions for
environmentally sound actions depends on the loop dominance. Driving forces in the systems are
the dynamics and not single factors. Identifying the dominant loop in the system may be seen as a
25
sine qua non for effective policy interventions. Secondly, we have seen that personal precondition
for environmentally sound actions may be an important intervention point under unfavorable
contextual recycling conditions. This gives further evidence for the relevance of changes in
preferences explaining policy outcome.
From a public policy perspective this study suggests that the interactive effects of personal (such as
willingness to invest time in specific behavior) and situational variables are important for
understanding the effectiveness of policy initiatives. Understanding the processes that help to
unfreeze harmful habits and to establish new ones represents a further significant opportunity for
improving policy effectiveness and to break path dependency in a system.
While for the specific case of recycling dynamics a micro choice structure for citizens’ separation
behavior is suggested, further research is indicated in order to synthesize a choice structure that can
be generalized for other contexts of applications. This line of research could found a generic SD
structure of choice, e.g. related to policy compliance issues of citizens, consumers, firms, or
organizations in different realms.
The main strengths of this modeling and simulation approach to theory building are summarized in
the following points.
e Important insights about interactions of personal and contextual factors could be found and
important intervention points were identified. Furthermore the effectiveness of different
policy-packages could be tested.
e The SD-SWM-model is relatively simple and is therefore suitable for the purpose of
enhancing the understanding of how important personal and contextual structures cause
internal dynamics and produce the observed behavior pattern.
e The model has been carefully tested, and different sensitivity analyses were conducted. The
test results give evidence of its robustness and consistency as well as correspondence.
e The System Dynamics modeling syntax allows evaluating the SD-SWM-model as a theory
applying general criteria for evaluation theories such as suggested by Bacharach (1989). A
systematic critique of the theory will be beyond this study but could be done by other
researchers pushing the debate of computer-assisted theory building further. Furthermore,
the concepts and constructs included in the model (such a habits, preferences, social norms,
planned behavior) may help to bridge the gap between other theories and to suggest
refinements of pre-existing theories.
However this study has also its specific limitations.
e The SD-SWM-model could still be improved. Furthermore the parameters and graphical
functions need to be empirically substantiated.
e The SD-SWM-model cannot be used to address detailed issues of policy implementation
such as a decision aid about which communication instrument to choose. Furthermore, it
cannot be used for precise prediction of the outcomes of a policy intervention at a specific
year.
e It is important to emphasize that the policy conclusion bears not only the methodological
meta-assumption (see Andersen 1980) but also the modeler’s own assumptions made in the
model building process. Furthermore the applicability of the model is bounded to the
specific family of solid waste management systems dealing with recycling dynamics.
However the model includes some generic structure components that could be used as
building blocks in other contexts of applications.
Although the study has its specific limitations, it demonstrates an innovative approach of
environmental policy analysis that focuses on loops as causes of behavior and traces the loop
26
dominance back to personal and contextual factors. Therefore it provides a broader basis for policy
analysis and policy design allowing a higher variety of intervention and implementation options.
References
Ajzen, I. (1991). “The theory of planned behavior.” Organizational Behavior and Human Decision
Processes. 50: 179-211.
Andersen, D. F. (1980). How differences in analytic paradigms can lead to differences in policy
conclusions. Elements of the System Dynamics Method. J. Randers. Massachusetts, The
MIT Press: 61-75.
Bacharach, S. B. (1989). "Organizational Theories: Some criteria for a evaluation." Academy of
Management Review 14(4): 496-515.
Bass, F. (1969). "A new product growth model for consumer durables." Management Science 15:
215-227.
Bass, F., T. Krishnan, et al. (1994). "Why the bass model fits without decision variables."
Marketing Science 13(3): 203-223.
Bruppacher, S. and S. Ulli-Beer (2001b). Mobilising Resources for More Sustainable Lifestyles:
Views of Households and Local Authorities. Changing Things - Moving People. R.
Kaufmann-Hayoz and H. Gutscher. Basel, Birkhauser: 281-299.
Coleman, J. S. (1965). "The use of electronical computers in the study of social organization."
Archives eurpéennes de sociologie 6: 89-107.
Dahlstrand, U. and A. Biel (1997). "Pro-environmental habits: Propensity levels in behavioral
change." Journal of Applied Social Psychology 27: 588-601.
Dietz, T. and P. C. Stem (1995). "Toward a theory of choice: socially embedded preference
construction." Journal of Socio-Economic 24: 261-279.
Duggan, E. (2002). Costs, change cloud recycling's future. Times Union. Albany, New Y ork: 1, A9.
Foppa, K. (1989). Grundlagen einer ipsativen Theorie des Handelns. Bern, Institut fiir Psychologie.
Foppa, K. and B. S. Frey (1985). Menschliches Handeln: Entscheidungsvorbereitung und
subjektiver Moglichkeitsraum. Wissenschaftskolleg. Jahrbuch 1984/85. P. Wapnewski.
Berlin, Institute for Advanced Study: 203-213.
Foppa, K. and B. S. Frey (1990). Ein ipsatives Modell menschlichen Verhaltens, Universitat Zurich.
Forrester, J. W. (1994). Policies, decisions, and information sources for modeling. Modeling for
Learning Organizations. J. D. W. Morecroft and J. D. Sterman. Portland, Productivity Press.
Forrester, J. W. (2003). Economic Theory for the New Millennium. 21st Intemational Conference
of the System Dynamics Society (July 20-24, 2003), New York City, The System
Dynamics Society.
Glaser, B. G. and A. L. Strauss (1967). The Discovery of Grounded Theory. Strategies for
Qualitative Research. Chicago, Aldine.
Gorr, H. (1997). Verkehrsmittelwahl im Alltag. Moglichkeiten und Grenzen der Beeinflussung.
Allgemeine Okologie zur Diskussion gestellt. R. Kaufmann- Hayoz. Bern, IKAO,
Universitat Bern. 3/1: 176-180.
Ittigen (1985-2001). V erwaltungsrechnungen 1985 -2001. Ittigen, Einwohnergemeinde.
Jaeggi, C., C. Tanner, et al. (1996). Was uns vom umweltverantwortlichen Handeln abhalt.
Umweltproblem Mensch. R. Kaufmann-Hayoz and A. D. Giulio. Bem, Paul Haupt: 181-
196.
Kaufmann-Hayoz, R., C. Battig, et al. (2001d). A Typology of Tools for Building Sustainable
Strategies. Changing Things - Moving People. R. Kaufmann-Hayoz and H. Gutscher. Basel,
Birkhauser: 33-108.
Kaufmann-Hayoz, R. and H. Gutscher (2001a). Transformation toward Sustainability: An
Interdisciplinary, Actor-Oriented Perspective. Changing Things - Moving People. R.
Kaufmann-Hayoz and H. Gutscher. Basel, Birkhauser: 19-26.
27
Kermack, W. and A. McKendrick (1927). "Contributions to the mathematical theory of epidemics.
(Proceedings fo the Royal Society, 115A 700-721)." Reprinted in Bulletin of Mathematical
Biology (1991) 53(1-2): 33-55.
Ludwig, C., S. Hellweg, et al., Eds. (2003). Municipal Solid Waste Management. Strategies and
Technologies for Sustainable Solutions. Berlin, Springer.
OECD (2000). OECD in Figures; Statistics of the Member Countries. Paris, OECD Publications.
Piorkowsky, M.-B. (1988). Umweltbewusstsein und Verbraucherverhalten. Eine interdisziplinare
Literature und Sekundaranalyse. Umweltverhalten und Emahrungsverhalten. M.-B. P. D.
Rohwer. Hamburg, Behr. 1.
Randers, J. (1996). Guidelines for Model Conceptualization. Modelling for Management II. G. P.
Richardson, Pegasus: 117 - 139.
Richardson, G. P. (1995). "Loop polarity, loop dominance, and the concept of dominant polarity."
System Dynamics Review 11(1): 67-88.
Richardson, G. P. and A. L. Pugh (1981). Introduction to System Dynamics - Modeling with
Dynamo. Cambridge, Productivity Press.
Schwaninger, M. (1997). "Integrative Systems methodology: Heuristic for requisite variety."
International Transaction in Operational Research 4(2): 109-123.
Scitovsky, T. (1989). Psychology des Wohlstandes. Die Bediirfnisse des Menschen und der Bedarf
des V erbrauchs. Frankfurt, Campus Verlag.
Sterman, J. D. (2000). Business Dynamics. Systems Thinking and Modeling for a Complex World.
Boston, Irwin McGraw-Hill.
Stern, P. C. (1999). "Information, incentives, and proenvironmental consumer behavior." Joumal of
Consumer Policy 22: 461-478.
Stern, P. C. (2000). "Toward a coherent theory of environmentally significant behavior." Journal of
Social Issues 56(3): 407-424.
Stem, P. C., T. Dietz, et al. (1999). "A value-belief-norm theory of support for social movements:
The case of environmental concerns." Human Ecology Review 6: 81-97.
Tanner, C. (1999). "Constraints on environmental behavior." Journal of Environmental Psychology
19: 145-157.
Taylor, S. and P. Todd (1997). "Understanding the determinants of consumer composting
behavior." Journal of Applied Social Psychology 27(7): 602-628.
Ulli-Beer, S. (2003). Dynamic Interactions Between Citizen Choice and Preferences and Public
Policy Initiatives. A System Dynamics Model of Recycling Dynamics in a Typical Swiss
Locality. 21st International Conference of the System Dynamics Society (July 20 - 24
,2003), New Y ork City, U.S.A, System Dynamics Society.
Ulli-Beer, S. (2004). Citizens' Choice and Public Policy. Institute of Management. St. Gallen,
University of St. Gallen.
Ulli-Beer, S., D. F. Andersen and G. P. Richardson. (2004). Using a SD-SWM-model to inform
policy making for solid waste management at the local level. Proceedings of the 22
International Conference of the System Dynamics Society, Oxford (July 25-29, 2004).
Vatter, M. (2001a). Einstellungen, Handlungserfahrungen und Kooperation im Umweltbereich
Lizentiatsarbeit). Bern, Interfakultare K oordinationsstelle fir Allgemeine Okologie.
Vlek, C. (2000). "Essential psychology for environmental policy making." International Journal of
Psychology: 153-167.
Weber, M. and M. Schwaninger (2002). "Transforming an Agricultural Trade Organization: A
System-Dynamics-based Intervention." System Dynamics Review 18(3): 381-401.
Zagonel, A. A., J. Rohrbaugh, et al. (2004 forthcoming). "Using simulation models to address
“What If"-questions about welfare reform." Journal of Policy Analysis and Management.
Back to the Top
