![Three model modules [2]](https://www.researchgate.net/profile/Zining-Yang/publication/301493902/figure/fig3/AS:667866584076308@1536243210924/Three-model-modules-2_Q320.jpg)
Content uploaded by Zining Yang
Author content
All content in this area was uploaded by Zining Yang on Apr 20, 2016
Content may be subject to copyright.
Volume 2 (2015) issue 3, article 1
75 ©2015
Sustainable energy infrastructure siting: an agent based
approach
可持续能源基础设施选址:采用基于个体的方法
Zining Yang (杨紫宁）1*, Hal T. Nelson1, Mark Abdollahian1
1 School of Social Science, Policy & Evaluation, Claremont Graduate University, Claremont CA 91711, USA
zining.yang@cgu.edu
Accepted for publication on 7th October 2015
Abstract - Technical, environment, social, economic and
political constraints are critical barriers to the development of
new renewable energy supplies. This paper is an agent-based,
predictive analytics model of energy siting policy in the
techno-social space that simulates how competing interests
shape siting outcomes to identify beneficial policy for
sustainable energy infrastructure. Using a high voltage
transmission line as a case study, we integrate project
engineering and institutional factors with GIS data on land use
attributes and US Census residential demographics. We focus
on modeling citizen attitudinal, Community Based Organization
(CBO) emergence and behavioral diffusion of support and
opposition with Bilateral Shapley Values from cooperative game
theory. We also simulate the competitive policy process and
interaction between citizens, CBOs and regulatory, utility and
governmental stakeholders using a non-cooperative game theory.
In addition, our model simulates the complexity of
infrastructure siting by fusing citizen attitude and behavior
diffusion, stakeholder bargaining and regulatory
decision-making. We find CBO formation, utility message and
NGO messaging have a positive impact on citizen comments
submitted as a part of the Environmental Impact Statement
process, while project need and procedure have a negative
impact. As citizens communicate and exchange political opinions
across greater distances with more neighbors, less CBOs form
but those that do are more effective, increasing the number of
messages citizens send. Our results also indicate that despite
money spent on assessing the engineering aspects of major
infrastructure projects, citizen participation and political power
can be more important to stakeholder bargaining outcomes than
the level of local disruption that a project causes.
Keywords –infrastructure siting, game theory, agent-based
model, Bilateral Shapely Values, community based organization
摘要–
技术、环境、社会、经济和政治约束是新的可再
生能源供应发展的主要障碍。本文介绍了一个基于个体的
预测分析模型，重点分析在科技社会空间中的能源选址政
策，模拟利益冲突如何影响选址的结果来确定可持续发展
能源基础设施的有理政策。使用高压输电线路为例，我们
结合项目工程和制度因素，并采用土地使用属性的GIS数
据和美国住宅人口普查数据。使用夏普利值及合作博弈理
论，我们的模型专注于公民态度，基于社区的组织(CBO)
的出现，和支持与反对的行为扩散。使用非合作博弈理论
，我们还模拟市民、当地监管部门、公用和政府利益相关
者之间的竞争政策过程和互动。此外，我们的模型还融合
了公民的态度和行为扩散，利益相关者的交涉，和监管部
门的决策来模拟基础设施选址的复杂性。我们发现CBO的
形成、公共部门信息和非政府组织信息对环境影响声明过
程中的公民的信息提交产生积极影响，而项目需要和过程
对公民信息提交产生负面影响。随着公民沟通和交流的政
治观点跨越更大的距离并联系更多的邻居，CBO的数量减
少但是以更有效的形式呈现，增加公民提交信息的数量。
我们的研究结果还表明，尽管有大量的钱花在评估重大基
础设施项目的工程方面，相较于项目对当地造成的破坏，
公民参与和政治力量对于影响利益相关者的交涉结果可 有
更大的作用。
关键词–
基础设施选址，博弈论，个体为本模型，夏普利
值， 基于社区的组织
I. INTRODUCTION
Technical, environment, social, economic and political
constraints are critical barriers to the development of new
renewable energy supplies. This paper reconceptualizes how
we “get to yes” by encouraging public participation and
shifting opposition to the “other” side’s proposals. In this
agent-based model of energy siting policy, we focus on how
competing interests shape siting outcomes and identify
actionable strategies to help build energy infrastructure in a
more timely and less conflictual manner that current
processes typically allow.
In this article, we investigate the effect of public
participation on agency decision-making. Public managers
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
76 Journal of Energy Challenges and Mechanic ©2015
must balance citizen demands, business interests, and the
public interest, conceived of as public policy goals. While the
relative influence of citizens versus interest groups in
administrative decision-making is one of the enduring
questions in political science and public administration,
investigations of relative citizen influence often rely on
case-based methods that typically focus on macro-level issues
such as institutional rules, problem severity, as well as the
attributes of the decision-making outcome. Yet, to estimate
the independent effects of public participation, the other
micro-level contextual variables must be, or are assumed to
be, held constant. As Collins [14] states, if research were to
include the effect of the complex interactions between
individual actors, then the development of generalizable
theories would be limited by our scholarly resources to
investigate the population of cases. Our computational
modeling approach compliments, rather than substitutes for,
empirical research and literature reviews, and can offer a
method for generating novel theoretical insights into citizen
influence [13].
Using a high voltage transmission line as a case study, we
integrate project engineering and institutional factors with
GIS data on land use attributes and US Census residential
demographics. We focus on modeling citizen attitudinal,
Community Based Organization (CBO) emergence and
behavioral diffusion of support and opposition with Bilateral
Shapley Values from cooperative game theory. We also
simulate the competitive policy process and interaction
between citizens, CBOs and regulatory, utility and
governmental stakeholders using a non-cooperative game
theory. In addition, our model explores the complexity of
infrastructure siting by fusing citizen attitude and behavior
diffusion, stakeholder bargaining and regulatory
decision-making.
Our simulation results provide strategic advice to users
about how to reach consensus on sustainable energy
infrastructure siting issue given its dynamics, offer insights
about policy levers, issue linkage strategies, bargaining
positions, scenarios analysis to explore key uncertainties, and
can identify equitable solutions supported by communities.
II. LITERATURE
Sustainable energy infrastructure development can be seen
as a mixed motive social dilemma where public goods
provision is in conflict with private interests. There have been
a lot of attentions paid to environmental sustainability and
new regulatory rules, have so utilities, stakeholders, and
government officials are under the pressure to find new and
creative solutions to the complex problems of sustainable
resource use. We focus on the Environmental Impact
Assessment (EIA) decision-making processes because they
are common and the EIA process structures agency decisions.
EIA processes require and notice and comment period like
the Administrative Procedures Act. 1EIA’s involve analyzing
1 After the US systematized EIAs in the National Environmental
Policy Act (NEPA) of 1969, some form of assessment has been
the likely environmental and social impacts of a project in a
multidisciplinary fashion, presenting the information to the
public and decision makers, and taking public and
stakeholder comments into account in the final decision. The
siting of an energy project usually begins with the project
sponsor developing a detailed and substantial review of social
and environmental impacts, typically prepared by the project
proponent, which gives it significant advantages in
determining the alternatives and the initial assessment of
costs and benefits of the project design. The EIA process
involves public notification of the project proposal, public
involvement in scoping, preparation of a draft EIA, public
review and comment on the draft EIA, and the preparation of
a final EIA that takes public comments into account [31].
There is substantial evidence in the planning and political
science literature that ensuring robust public participation and
making use of collaborative planning approaches can
significantly reduce conflict [8, 9]. A study of planning in the
Great Lakes region find that an open and fair participatory
process is associated with greater trust and better policy
outcomes. Many public participation practices reduce conflict
and develop accountability [8]. Increased public participation
can include building trust, developing “buy-in”, provide
objectively superior decisions, and lead to a more healthy
democratic society [8].
The second type of theoretical and empirical support for
the model development is industry impact in administrative
decisions. Although stakeholder participation in general has
elicited great expectations for power sharing among diverse
interests and individuals, public consultation can just
legitimize decisions that have already been made [20]. Other
researchers have been concerned that stakeholder processes
simply reproduce the power relations already present in a
jurisdiction [6][16]. Public participation has been conceived
as a means to check power of the state and market [44].
The third body of literature that contributes to the model
comes from lobbying and administrative decision making.
Agency decisions are subject to lobbying by industry groups
who can more easily overcome barriers to collective action
than consumers [34]. Industry groups have greater lobbying
resources compared to public interest groups. Other studies
suggest that powerful industry groups manage to manipulate
state energy policies [37]. Evidence suggests that
environmental groups have been skeptical of participation
mechanisms because of the perceived power of
pro-development interests to influence the outcomes [18][30].
III. THE MODEL
required by all US states, and in a growing number of nations
around the world (Wathern, 1988, p. 3). The European Union
requires EIA for public and private infrastructure projects that are
thought to have significant environmental impacts (European
Commission, 2012). Most nations in Asia, including China, Korea,
Japan, Indonesia and India require some form of EIA before major
projects can proceed. EIA’s are typically required for these large
infrastructure projects involving government funds or lands.
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
77 Journal of Energy Challenges and Mechanic ©2015
Given this review of citizen and industry influence
administrative decisions, simulating this process requires the
integration of both citizen and industry preferences into
modeling efforts. Our agent based model of siting preferences,
called SEMPro, simulates bargaining dynamics amongst
stakeholders as well as decision makers in the decision
process using a spatial bargaining model.
Bargaining models date back to Condorcet’s voting
paradox [15], and Black [11] and Downs [17] trying to frame
a positivist approach to analytical politics. More recently,
McKelvey and Ordershook [29] as well as Feldman [19]
outline four fundamental assumptions for spatial stakeholder
bargaining models: actors are instrumentally rational, with
the choice set of feasible political alternatives modeled as a
space with complete, ordered and transitive properties. The
spatial bargaining approach naturally lends itself to
agent-based modeling as stakeholders possess decision
agency as well as attributes of preferences over issue spaces,
with varying influence and salience [22]. ABM instantiations
of spatial bargaining models include Abdollahian and
Alsharabati [3] and Abdollahian et al [4].
SEMPro is part of a new class of techno-social [42] and
complex adaptive systems’ models[1, 2], simulating the
interactive effects and feedbacks between human and
institutional agency, engineered physical elements, and
geophysical systems. SEMPro makes two contributions to
our understanding of citizen impacts on agency decisions.
First, SEMPro is one of only a handful of multi-agent
agent-based models that uses geographical information
system (GIS) and detailed census survey data which
instantiates real-world dynamics into simulation modeling.
Second, SEMPro is the first planning model we are aware of
that integrates an ABM with cooperative and non-cooperative
game theory models of stakeholder and regulatory
decision-making.
SEMPro utilizes the ABM approach as it generates
emergent, large-scale system phenomena from the
micro-motivations and behavioral interactions of multiple
agents. ABM results can then be validated against observed
patterns of behavior to analyze what percent of the variation
in real-life events that can be explained by the modeling.
ABMs are used in techno-social modeling for three primary
reasons.
First, agents can be assigned attributes based on stochastic
distributions to represent noise or errors in human
communication in the model that is reflective of the dynamic,
adaptive and strategic nature of human behavior, especially in
real-world political and social processes [5]. Introducing
stochasticity in agent relationships can dramatically affect
networks structures that in turn drive different behaviors [35].
Second, unlike most top-down economic models, agents in
ABMs can be assigned heterogeneity in preferences,
attributes, or goal-orientation objectives. Brown and
Robinson [12] have shown how variations in preferences
predict divergent land use outcomes. Finally, the interaction
of these heterogeneous agents can lead to non-monotonic
outcomes stemming from social mimicry, cooperation and
competition in human systems [27]. Thus, ABMs can
represent, anticipate and shape the complexity of
socio-technical systems better than equation-based models
and are more transparent [7].
SEMPro was developed using a system’s perspective and
parameterizes the project and policy levers that enable
scenario analyses required of an effective decision support
system [26]. Decision support systems (DSSs) like SEMPro
allow users to simulate trade-offs and alternatives to improve
energy planning outcomes[35]. DSSs are intended to improve
the quality of decision making and need to be generalizable
to a wide range of cases [24]. SEMPro can be applied to a
wide range of infrastructure siting technologies such as oil
pipelines, highways, high speed rail, electricity generation
stations, and the subject of this article, electricity
transmission lines. In addition to varying project level
variables such as engineering attributes in SEMPro, we can
also estimate the impacts of changes in risk communication
strategies by project stakeholders.
We fuse geophysical and social elements to understand the
interactive effects and feedbacks between individual human
agency, engineered physical elements and the geophysical
environment. Our model is implemented in NetLogo [45],
with three different sequential modules, a citizen/CBO
formation module, a stakeholder lobbying module and a
regulatory decision making module. The citizen agents,
stakeholders, and regulators in the model are all trying to
maximize their own utilities, given the assumption of
bounded rationality. Figure 1 depicts the high level process
and multi-module architecture. It runs for up to 25 time steps,
with each time step representing 1–2 months of calendar time
consistent with regulatory decision time frames in some
instances.
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
78 Journal of Energy Challenges and Mechanic ©2015
!"#$%&'()*)'"&+,*'
+-%.%-%&/%0'+12%-'
!"#$%&3'/)4/,4)*%'
"&"#)4',#4"*5'
!"#$%&3'/)4/,4)*%'
,#4"*5'1.'.1-6"&7'!893'
!"#$%&3':%/":%'
;<%*<%-'*1'.1-6'!893'
!"#$%&3'')&:'!893'
=+:)*%'+-%.%-%&/%0'+12%-'
>*)?%<14:%-'()*)'"&+,*'
+-%.%-%&/%0'+12%-'
>*)?%<14:%-3'@'!893'
/)4/,4)*%'"&"#)4',#4"*5'
>*)?%<14:%-3'@'!893''
!)4/,4)*%',#4"*5':,-"&7'A)-7)"&"&7'
>*)?%<14:%-3'@'!893''
=+:)*%'+-%.%-%&/%0'+12%-'
B%7,4)*1-'()*)'"&+,*'
+-%.%-%&/%0'+12%-'
>*)?%<14:%-30'!893'@'B%7,4)*1-3'
!)4/,4)*%'"&"#)4',#4"*5'
>*)?%<14:%-3C'!893'@'B%7,4)*1-3'
!)4/,4)*%',#4"*5':,-"&7'A)-7)"&"&7'
>*)?%<14:%-3C'!893'@'B%7,4)*1-3'
=+:)*%'+-%.%-%&/%0'+12%-'
B%61D%'>*)?%<14:%-3'@'!893'
B%7,4)*1-3'E1*%'
CBO$Module$ Stakeholder$Module$ Regulator$Module$
Outputs$
F77-%7)*%'9++13"#1&'
G1*)4'H-%.%-%&/%'
G1*)4'>)4"%&*'H-%.%-%&/%'
I'!893'
!89'H-%.%-%&/%'
J%33)7%'
H-%.%-%&/%'("3*-"A,#1&'
("3-,+#1&'=#4"*5'J%33)7%'KL9'J%33)7%'K%%:'H-1/%:,-%'
'
Policy$Levers$
'
Fig. 1. Three model modules [2]
In the first module, citizens react to energy
infrastructure siting projects by forming opinions,
interacting with each other, and forming Community
Based Organizations (CBOs) that either support or oppose
such projects. To simulate this process, citizen agents are
queued and processed according to their patch or grid
location. GIS-based data on the project size and route, on
land use, and on the location of residents informs
agent-based simulations of individual interactions. US
Census block-group population density data is used to
locate citizen agents in the model. Data on education and
income by block-group are instantiated as attributes of the
agents in the model and provide initial heterogeneity for
simulated citizen behavior. Higher values are associated
with greater levels of influence in affecting project
outcomes and imbue citizens with “power.” Wealthier and
more educated individuals tend to have a stronger sense of
self-efficacy and more resources available for advocacy
[33].
In this module, the following is based on calculation of
Bilateral Shapley Values (BSVs) of all citizen agents. BSV
is a concept in cooperative game theory for explaining
coalition formation, and thus a natural modeling strategy
to use in CBO formation [25]. Each citizen agent is
assumed to be autonomous, with bounded rationality,
maximizing it’s own utility subject to the geophysical,
engineering and social constraints of its environment [46].
BSV computes all combination of all possible coalitions
that citizens can join that maximize citizen utility, and then
compares all possible coalition utilities in deciding
whether or not to join or form a larger CBO. BSV
dynamics thus focus on the permutations of individuals in
different coalitions based on the marginal utility gained
from CBO formation. Expected utility has been described
as the “major paradigm in decision making” [38], and our
CBO formation is based on cooperative game theory [40].
We also incorporates Social Judgment Theory in each
citizen agent’s objective function. This theory describes
how the positions of two agents can be conceived along a
Downsian continuum where the distance between their
positions affects the likelihood of one accepting the other’s
position. A message that is far from a receiver’s position is
likely to be rejected [39]. For decades, social psychology
research has documented that not only do people resist
changing their own positions in relationship to new
information, but that they might also adopt even more
extreme beliefs than before. Social judgment theory finds
support in the literature on risk perceptions and social trust.
Citizens are unlikely to change their preferences about the
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
79 Journal of Energy Challenges and Mechanic ©2015
project if they distrust the source of risk communications
[23]. In spatial bargaining, trust can be operationalized as
the distance between two stakeholder’s positions and again
is operationalized in the SEMPro model structure.
In the second module of stakeholder bargaining, against
this backdrop of political and social opinion formation and
risk communication processes, organized stakeholders
seek to lobby not only citizen opinions but also other
stakeholders to maximize their specific, organizational
interests. Berlo’s Communications Penetration Model
describes how these messages may not be received or
accepted because the receiver is not exposed to the
message, does not pay attention to the message or does not
accept the sentiment of the message [10]. The stakeholder
bargaining module takes the emergent CBO formation into
consideration in determining stakeholder bargaining
outcomes using non-cooperative game theory.
Stakeholders will form coalitions if it increases their
power to potentially influence the regulatory process as
long as the coalition’s position is acceptable given the
stakeholder’s initial position [32].
In the third module, regulators join the bargaining
process in the end of the stakeholder module, taking into
account CBO formation and public opinion, then bargain
among themselves in the regulator module to vote either in
support or opposition to the project. Each module updates
at each time step. This parallel, linked module processing
sequence then iterates. In two continuous time steps, if no
new coalition is formed, or no CBOs, stakeholders and
regulators change their preference, then the model reaches
its steady state equilibrium and will stop.
Actionable policy levers for shaping the transmission
siting process include the disruption engineering of the
project, utility and NGO messaging outreach, as well as
perceived project need and procedure surrounding the
process. SEMPro users can simulate changes in the
engineering, social, and political attributes of each project
as explained in Abdollahian, et al [2]. Each policy lever
parameter is normalized along Downsian issue continuum
on a 1-10 scale to calibrate the model’s internal validity.
The variable that describes the engineering attributes of
the project in the model is the level of disruption that the
project imposes on the community. Disruption is defined
as impacts to public health and safety, viewshed
impairment, impacts to property values, or other
externalities from the infrastructure project (0-1 scale).
Utility-Message is a stakeholder variable that represents
the number of pro-development messages the project
sponsor sends to citizens to shape public attitudes in each
time step.
NGO-Message is the final project level variable that
represents the number of anti-development outreach risk
communications that non-governmental organizations
(NGO) such as the Sierra Club sends to citizens. Our
approach propagates utility and NGO messages according
to the parameter settings for each simulation in each time
step.
Two institutional level variables are included in the
model: Perceived Need is perceived to be needed by the
community. Need can be coded higher when project has
been approved by the state regulator and is perceived to
provide local system reliability or economic benefits.
Procedure is an indicator of procedural justice, or to what
extent the citizens think their preferences will be included
in regulatory decision-making. Experimentalist research
confirms that people want to be treated equitably and
“other-regarding” equity considerations are a primary
driver of citizen behavior.
The primary community level variable is Talk-Span,
defined as is the distance across which citizen agents talk
to each other and make decisions on whether to form
CBOs. This can be conceived as the social connectivity of
citizens (Putnam, 2001).
IV. VERIFICATION AND SIMULATION
Unit tests were employed in the development of the
three modules to verify code functionality. Next, the
model outputs were validated against what it claims to be
representing. The general goal of validating ABMs is to
assess whether the micro-level behavior of the agents
generate the expected macro-level patterns [21]. Following
Taber and Timpone [41] we employed a two-step
validation process. The first was a process validation
assessment that tests the model’s mechanisms against
real-world processes. Our process validity assurance began
with selection of appropriate micro-level theories about
attitude and behavior diffusion, including social judgment
theory [39] and spatially structured (rather than random)
interactions [30]. Subsequently, the model’s assumptions
underlying the model’s algorithms were validated against
survey data of citizens for a Southern California Edison
siting project of Tehachape and Chino Hills. The analysis
of the survey data indicated that citizen preferences are
moderated by their proximity to the project, their
communication networks, and the disruption posed by the
project. The effect of trust in the project sponsor on citizen
preferences is moderated by distance [33]. Abdollahian et
al [2] report other validation tests performed on the model
outputs and how the survey data support the model.
After validation and verification, we conducted a
quasi-global sensitivity analysis by varying all input
parameters across their entire range in three steps (min,
mean, max) resulting in 729 runs with up to 25 time steps
each, for a total of 14,576 observations. We then pool all
the simulations together for a pooled time series regression
design estimated with ordinary least squares (OLS)
regression with standardized β coefficients for input
parameter comparability and model performance.
V. RESULTS
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
80 Journal of Energy Challenges and Mechanic ©2015
5.1. CITIZEN PREFERENCE
Table 1 contains the results of the OLS modeling of the
simulation results. Model 1 in Table 1 is our baseline
model for detailing the impact of input parameters on
number of citizen messages sent to regulators regarding
the siting project. The dependent variable is the interaction
term of total messages and median preferences of citizens,
which captures not only the number of messages but also
the direction of messages—opposition or support for the
project.
TABLE 1, POOLED OLS ESTIMATIONS OF CITIZEN MESSAGES AND
CBO PREFERENCES
First, let us examine the effect of project attributes on
citizen opposition. In our simulations, the disruption posed
by the project has a very large impact on citizen messages
(β = .109) as expected. A one standard deviation decrease
in disruption results in a decrease of .109 standard
deviations in negative citizen messages. Modifying the
project engineering design to reduce disruption by 35%,
for instance by increasing the width of the right-of-way, is
predicted to result in 11% less citizen opposition.
Project need in model 1 is negative and significant (β =
-.014), but is much less important than disruption in
explaining outcomes. The results are consistent with
observation that citizens express less opposition when the
project siting brings significant benefits and is needed by
the community. Similarly, perceptions of the procedural
justice of the project are negative but not significantly
different from zero, suggesting that in these simulations,
increasing citizens’ perceptions of the procedural fairness
of the EIA process is not likely to have an impact on
citizen opposition. As expected from the model design,
time (β = .959) is positive and significant as the number of
messages grows over time. Community attributes also
have a large impact on citizen advocacy and activism.
Talkspan has a negative impact (β = -.018) on citizen
comments, suggesting that citizens express their opinion
less frequently in well-connected communities, as they can
express the opinion through CBOs.
Turning to the effects of risk communications strategies
by project proponents and opponents, NGO message is
significant since credible NGO messaging can enhance
citizen activism. However the impact of NGO messages is
only modest (β = .019) showing effects on activism of
about the same magnitude as perceived project need.
Although utility risk communications reduce the number
of negative messages sent to regulators, the average effect
of this variable is not significant. The implications of this
finding are discussed in more detail below.
In models 3 and 4, we look at the impact of input
parameters on CBO preferences, a key emergent behavior
from the first module. CBO preference is the weighted
average of the number of CBOs times their preferences
categorized by deciles in model output. A higher value for
CBO preferences indicates more CBO opposition to the
project. The R2 of 88% in the models shows CBO
preference variation explained.
We can see that talkspan is not only highly significant
but has the largest impact (β = .909) on CBO preferences.
As citizens are able to communicate and exchange
opinions across greater distances with more neighbors, the
number of citizens joining CBO increases, consistent with
existing literature [1, 2]. The time step variable also shows
a large and significant impact on CBO formation (β =
.245), indicating CBOs opposition increases as time
passes. The magnitude of this variable is significantly
smaller than for citizen messages (model 1), indicating that
CBO preferences are less time dependent than citizen
messages.
Utility message and other policy levers like disruption,
procedural justice and NGO message do not have
significant impact on CBO preferences in the citizen
module. Need is significant and positive, counter
intuitively indicating greater project need increases CBO
opposition. Further investigation of this finding is
warranted to discover how project need is channeled
through citizen preferences that might have a positive
impact on CBO preferences.
5.2. STAKEHOLDER PREFERENCE
Next, we turn to an analysis of stakeholder preferences
in Table 2. We employ a two stage least square (2SLS) /
Instrumental Variable (IV) regression technique for the
model outputs for time steps 1-20. The error term from
stakeholder preferences are likely to be correlated with
CBO preferences in any given time step. 2SLS is an
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
81 Journal of Energy Challenges and Mechanic ©2015
appropriate econometric technique that uses the predicted
value of CBO preferences created in the first stage to
predict stakeholder preferences in the second stage
regression. This controls for the simultaneous impact of
CBOs on stakeholder preferences.
The first stage in model 5 results in an R2 of .89,
indicating 89% of the variation in CBO preferences is
explained. Stage 1 in model 5 is very similar to model 3,
but also includes negative messages. The inclusion of
negative citizen messages truncates the coefficients for
both time step and talkspan and makes the need coefficient
negative. This is also consistent with model 1 and our
theoretical priors. The second stage regression in model 5
indicates the number of citizen messages has a much
smaller impact on stakeholder preferences than CBO
preferences. This is consistent with observed behavior that
citizens need a seat at the table to be heard. Organizational
representation is critical to influence stakeholder
bargaining in the model.
TABLE 2, 2SLS/IV ESTIMATIONS OF STAKEHOLDER
PREFERENCES
5.3. REGULATOR PREFERENCE
Table 3 shows the variables that impact regulator
preferences using the same instrumental variable approach
where we first predict stakeholder preferences and then
use that value to predict regulator preferences. The R2
indicates that 30% of the variation in regulator preferences
is explained by the stakeholder preferences and citizen
messages. We expect the R2 for regulator preferences to be
lower than that of the stakeholder equation as regulators
have to balance additional considerations, such as
competing policy goals and political issues, in their
decisions. In addition, the R2 is lower as regulators only
interact with CBOs and other stakeholder from time step
16 to 20, and then decide amongst themselves from time
step 21-25.
The table shows that negative citizen messages have a
larger impact on regulator preferences than stakeholder
preferences in the previous table. A one standard deviation
increase in citizen messages results in a .621 standard
deviation (β=.621) increase in regulator oppositional
preferences.
This differential impact of citizen activism on
stakeholder and regulator modules is critical. The impact
of citizen messages on regulator preferences is over two
times larger than their impact on stakeholder preferences.
Citizen preferences impact stakeholder preferences
through the efficacy of CBOs who bargain with other
stakeholders. On the other hand, the modeling predicts that
elected or appointed regulators are more balanced in their
response to citizens and stakeholders’ demands.
TABLE 3, 2SLS/IV ESTIMATIONS OF REGULATOR PREFERENCES
V. DISCUSSION
The results from the model simulations show important
insights for planning processes as the linkages between
emergent citizen behavior and stakeholder and regulator
preferences are complex. First, citizen advocacy in
institutional processes will be greater when threats to their
communities are greater as evidenced by the positive
impact of the disruption variable, which is consistent with
the risk communication research.
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
82 Journal of Energy Challenges and Mechanic ©2015
Figure 2 Agent Histogram Density for time step=1, 10 and 20
Second, emergent citizen behavior can dramatically
alter institutional outcomes over time. Figure 2 shows
histograms of average citizen, stakeholder and regulator
preferences in the first, middle and last time steps in all of
the simulations. What is notable across all three categories
is the shift towards greater project opposition over time
across all three levels of analysis.
The third finding is communities with more
well-connected citizens represented in the model by larger
talkspan are more likely to be effective blocking or
altering infrastructure projects. Talkspan implies citizens
talking across a greater geographical distance in the model
and predicts fewer CBOs as well as more citizen
opposition messages. Talkspan can be conceived of as the
level of betweenness in social network terms, with larger
nodes being more socially connected to other individual
citizens. For details, see Abdollahian et al. [2] analysis on
betweenness and eigenvector centrality of the model’s
social network outputs.
Figure 3 Citizen CBO Size and Preference
Figure 3 above shows several simulations of citizen
CBO representation and their resulting preferences for
three groups, the city, one of the regulators the CPUC and
the utility Southern California Edison (SCE). Here we can
see the varying response elasticities of all three groups to
increasing CBO size. While the city seems to be relatively
inelastic to CBO sizes, both the regulator and the utility
show marked change. The CPUC regulator here starts at
an indifferent preference (approximately 50) but slowly
moves towards project opposition (at 80) in a linear
fashion as CBO participants move from 300 to around
450. Afterwards, there seems to be marginal returns for
increasing opposition with more CBO participants. What
is most interesting is the utility’s staunch support for the
project (at a preference of 10) in the face of increasing
opposition, until a tipping point is reached where sharp,
major concessions (shifting towards indifference at 50) are
granted in order to maintain project viability. This seems
to be consistent with many public agencies’ past modus
operandi of ‘decide then defend’ for works projects.
The results show several key emergent behaviors from
infrastructure siting including citizen interaction and CBO
formation. Our simulations explain why CBOs are
effective in aggregating citizen preferences and altering
stakeholder preferences. The finding that citizen messages
are relatively more important to regulators than
stakeholders is consistent with the institutionalized
comment process. Our findings indicate that citizen
comments are surprisingly influential in determining
regulators’ preferences, indicating a level of political
responsiveness to social sustainability issues that supports
the efficacy of institutionalized planning processes. At the
same time, we also find that CBOs positions are important
in determining stakeholder preferences.
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
83 Journal of Energy Challenges and Mechanic ©2015
We posit two important methodological advances from
our current modeling approach. First, the SEMPro design
that links an ABM with GIS data is critical for valid
inferences about citizen participation as citizen
interactions emerge from local conditions and attributes;
all politics are local. Second, linking ABM with spatial
bargaining models permits the analysis of the interactions
and linkages between citizen emergent behavior and
institutionalized decision-making modalities. By linking
citizen behavior with stakeholder and regulator
preferences, SEMpro explicitly simulates the impact of
micro-level behavior on macro-level institutional
outcomes, a fundamental challenge in social policy spaces.
REFERENCES
[1] M. Abdollahian, Z. Yang, T. Coan, and B. Yesilada,
“Human Development Dynamics: an agent based
simulation of macro social systems and individual
heterogeneous evolutionary games,” Complex
Adaptive Systems Modeling, vol. 1, no. 18, pp. 1-17,
2013.
[2] M. Abdollahian, Z. Yang, and H.T. Nelson,
“Sustainable Energy Modeling Programming
(SEMPro),” Journal of Artificial Societies and Social
Simulation, vol. 16, no. 3, pp. 6-13, 2013.
[3] M. Abdollahian M and C. Alsharabati, “Modeling the
strategic effects of risk and perceptions in linkage
politics,” Rationality and Society, vol. 15, no. 1, pp
113–135, 2003.
[4] M. Abdollahian, M. Baranick, B. Efird, and J. Kugler,
Senturion: A predictive political simulation model,
Center for Technology and National Security Policy
National Defense University, DC, 2006.
[5] R. Axelrod, “The dissemination of culture a model
with local convergence and global polarization,”
Journal of Conflict Resolution, vol. 41, no. 2, pp
203-226, 1997.
[6] C. Ansell and A. Gash, “Collaborative Governance in
Theory and Practice,” Journal of Public
Administration Research and Theory, vol. 18, pp.
543-571, 2007.
[7] R. Axtell, The new coevolution of information science
and social science. Brookings Institution. [Online].
2003. Available:
http://www2.econ.iastate.edu/tesfatsi/compsoc.axtell.
pdf
[8] T.C. Beierle and J. Cayford, Democracy in Practice:
Public Participation in Environmental Decisions,
Resources for the Future, Washington, DC, 2002.
[9] T.C. Beierle, and D. Konisky, Public participation in
environmental planning in the Great Lakes Region,
Resources for the Future, Washington, DC, 1999.
[10] D.K. Berlo, J.B. Lemert, and R.J. Mertz,
“Dimensions for evaluating the acceptability of
message sources,” Public Opinion, vol. 33, pp
563-576, 1969.
[11] D. Black, The Theory of Committees and Elections,
Cambridge: Cambridge University Press, Cambridge,
1958.
[12] D.G. Brown and D.T. Robinson, “Effects of
heterogeneity in residential preferences on an
agent-based model of urban sprawl,” Ecology and
Society, vol. 11, no. 1, pp 46, 2006.
[13] T. Choi and P.J. Robertson, “Deliberation and
Decision in Collaborative Governance: A Simulation
of Approaches to Mitigate Power Imbalance,”
Journal of Public Administration Research and
Theory, vol. 24, pp 495-518, 2013.
[14] R. Collins, “Micro-translation as a theory building
strategy,” In K. Knorr-Cetina & A. V. Cicourel (Eds.),
Advances in social theory and methodology: Toward
an integration of micro- and macro-sociologies
Boston: Routledge & Kegan Paul, vol. 81, pp 81- 108,
1981.
[15] M. Condorcet, Essai sur 1'application de 1'analyse de
la probabilité des décisions rendues de la pluralité
des voix, Paris, 1785.
[16] B. Cooke and U. Kothari, Participation: The New
Tyranny? New York, NY / Zed Books Ltd., 2001.
[17] A. Downs, 1957 An Economic Theory of Democracy,
NY, New York: Harper, 1957.
[18] J.D. Echeverria, “No success like failure: the Platte
river collaborative watershed planning process,”
William and Mary Environmental Law and Policy
Review, vol. 25, pp 559–604, 2001.
[19] B.E. Feldman, Bargaining, Coalition Formation, and
Value, PhD thesis, State University of New York at
Stony Brook, 1996.
[20] D.J. Fiorino, “Citizen participation and
environmental risk: a survey of institutional
mechanisms,” Science, Technology, & Human Values,
vol. 15, no. 2, pp 226–243, 1990.
[21] N. Gilbert, Agent-Based Models Thousand Oaks, CA
/ Sage Publications, 2008.
[22] M. Hinich and C. Munger, Analytical Politics,
Cambridge: Cambridge University Press, 1997.
[23] R.E. Kasperson and P.J.M. Stallen, Communicating
risks to the public: International perspectives (Vol. 4),
Springer Science & Business Media, 1991.
[24] G. Kersten, "Decision making and decision support,"
In Kersten et al (Eds) Decision Support Systems for
Sustainable Development, pp 29-52 US: Springer US,
2000.
[25] S. Ketchpel, “Coalition formation among
autonomous agents,” Lecture Notes in Computer
Science, vol. 957, pp 73-88, 1995.
[26] R. Lempert, “Agent-based modeling as
organizational and public policy simulators,”
Proceedings Of The National Academy Of Sciences of
the United States of America, vol. 99, no. 3, pp
7195-7196, 2002.
[27] A. Ligmann-Zielinska and P. Jankowski,
“Agent-based models as laboratories for spatially
Yang et al. (2015) Sustainable energy infrastructure siting: an agent based approach
84 Journal of Energy Challenges and Mechanic ©2015
explicit planning policies,” Environment and
Planning B: Planning and Design, vol. 34, no. 2, pp
316, 2007.
[28] M. McCloskey, “Adapting local environmental
institutions: environmental need and economic
constraints,” Valparaiso University Law Review, vol.
34, pp 423–434, 2000.
[29] R.D. McKelvey and P.C. Ordeshook, “A decade of
experimental research on spatial models of elections
and committees,” Advances in the Spatial Theory of
Voting, pp 99-144, 1990.
[30] M. McPherson, L. Smith-Lovin, and M. Cook, “Birds
of a feather: homophily in social networks,” Annual
Review of Sociology, vol. 27, pp 415–44, 2001.
[31] National Environmental Policy Act (NEPA), “United
States Congress,” 1969 [Online]. Available:
http://ceq.hss.doe.gov/nepa/regs/nepa/nepaeqia.htm,
1969.
[32] H.t. Nelson, B. Swanson, and N. Cain, “The effects
of proximity, disruption, and social networks on
opposition to high voltage power lines,” Working
Article, 2013.
[33] M. Nishishiba, H.T. Nelson, and C.W. Shinn,
“Explicating factors that foster civic engagement
among students,” Journal of Public Affairs Education,
pp 269-285, 2005.
[34] M. Olson, Logic of collective action public goods and
the theory of groups, 1965.
[35] J.M. Pujol, A. Flache, J. Delgado, and R. Sangüesa,
“How can social networks ever become complex?
Modelling the emergence of complex networks from
local social exchanges,” Journal of Artificial
Societies and Social Simulation, vol. 8, no. 4, 2005.
[36] R.D. Putnam, Bowling Alone: The Collapse and
Revival of American Community, NY: New York:
Simon & Schuster, 2001.
[37] B.G. Rabe and P. Mundo, “Business influence in
state level environmental policy,” in Business and
Environmental Policy: Corporate Interests in the
American Political System Eds S Kamieniecki, ME
Kraft, MIT Press, Cambridge, MA, pp 265–282,
2007.
[38] P.J. Schoemaker, “The expected utility model: its
variants, purposes, evidence and limitations,” Journal
of Economic Literature, vol. 20, pp 529-563, 1982.
[39] F.W. Siero, B.J. Doosje, “Attitude change following
persuasive communication: integrating social
judgment theory and the elaboration likelihood
model,” European Journal of Social Psychology, vol.
23, pp 541-554, 2006.
[40] J. Suijs, “Cooperative game theory,” In: Suijs, J. Eds.,
Cooperative Decision Making Under Risk. Springer,
Heidelberg, pp 7-41, 2000.
[41] C.S. Taber and J.T. Richard, Computational
Modeling. No. 7-113. Sage, 1996.
[42] A. Vespignani, “Predicting the behavior of
techno-social systems,” Science, vol. 325, no. 5939,
pp 425, 2009.
[43] A.B. Whitford, J. Yates, and H.L. Ochs, “Ideological
Extremism and Public Participation,” Social science
quarterly, vol. 87, no. 1, pp 36-54, 2006.
[44] T.J. Whitford, S.M. Grieve, T.F. Farrow, L. Gomes, J.
Brennan, A.W. Harris, and L.M. Williams,
“Progressive grey matter atrophy over the first 2–3
years of illness in first-episode schizophrenia: a
tensor-based morphometry study,” Neuroimage, vol.
32, no. 2, pp 511-519, 2006.
[45] U. Wilensky, NetLogo, [Online].
http://ccl.northwestern.edu/netlogo/. Center for
Connected Learning and Computer-Based Modeling,
Northwestern University, Evanston, IL, 1999.
[46] C. Yeung, A. Poon, and F. Wu, “Game Theoretical
Multi-Agent Modeling of Coalition Formation for
Multilateral Trades,” IEEE Transmission on Power
Systems, vol. 14, pp 929-934, 1999.
