New Scripts for Group Model Building —
Online Questionnaires and Open Loops
Oliver Handel’, and Holger Maximilian Kleemann”
1) Ph.D. Candidate, Chair of Computational Modeling and Simulation, Department of Civil,
Geo and Environmental Engineering, Technical University Munich, Germany. Email:
oliver.handel@ tum.de
2) M.Phil. M.Sc., Institute for Technology Assessment and Systems Analysis, Karlsruhe
Institute of Technology, Germany. Email: max.kleemann@ kit.edu
Abstract
We introduce prototypes of two new GMB-scripts that help overcome the following issues:
Firstly, GMB-workshop participants often have little time available. Building a seed model
beforehand to use workshop time more efficiently meets the difficulty that participants’ time
for eliciting information for seed model construction may also be limited. To alleviate this, the
first script guides on how to use online questionnaires to enable to collect necessary
information from the participants before the GMB-workshop starts.
Secondly, sometimes undesirable system behavior results from potential feedback loops that
are not closed in the real system. Existing scripts however, aim at eliciting closed feedback
loops only. The second script helps participants find open loops the closure of which may
improve system behavior. The creativity for finding policy options that is unleashed by this
script can be used by subsequent scripts.
Keywords: System Dynamics, Group Decision Making, Group Model Building, Scriptapedia
Introduction
Jay Wright Forrester mentioned the importance of accessing the mental database of managers
for building System Dynamics models of strategic problems (Forrester 1992; Rouwette &
Vennix 2006). Vennix (1996) developed the method Group Model Building (GMB) to enable
accessing the mental models of involved clients in a structured and systematic way. Different
from expert modeling, the method enhances the feeling of ownership of the clients with the
outcomes, because the problem owners contributed to the solution and this circumstance
fosters that the model has a beneficial impact in the real world (Rouwette & Vennix 2006).
Three basic assumptions of the method are (compare with Vennix 1996): 1. Individuals are
limited by their information processing capacities, 2. People think in terms of causal
processes with hidden assumptions and 3. Different viewpoints in a group setting can be very
productive. GMB enables integrating different perspectives into a more holistic shared mental
model. The method aims to make hidden assumptions of the participants transparent and to
avoid premature ineffective solutions are put into action, but rather that sufficient deliberation
has taken place. Rouwette et al. (2006) distinguished goals of GMB on the individual and on
the group level. On individual level, the goal of a GMB project is to initiate behavioral change
and to foster commitment with the results. On group level, the goal is to increase the quality
of communication by using a shared language and to facilitate consensus-making on problem
causes and potential solutions.
Early adopters of the GMB method mentioned that the method lacks clearly defined
methodological guidelines (Andersen et al. 1997). Therefore, the idea of a “catalogue of
tested and refined procedures [to] build a continuous stream of small-group-activity”
(Andersen & Richardson 1997) was bom. This catalogue with standardized protocols, so
called scripts, to construct policy-oriented System Dynamics models was named
Scriptapedia’. The early authors of this online published handbook mention various reasons
for the aim of opening the black box of modeling interventions (Hovmand et al. 2011) by
documenting the processes of planned small group activity. Firstly, scripts help to clarify
different roles in the modeling team (e.g. facilitator, modeler/reflector, process coach,
recorder and gatekeeper) and to serve as a basic unit to structure behavior in GMB sessions
resulting in scripted behavioral time blocks. Secondly, these time blocks and individual
scripts can be sequenced into a scripts map as “a framework for effectively combining [...]
scripted activities, products, and deliverables into a formal network to enable facilitators to
construct appropriate combinations of workshops” (Ackermann et al. 2011) and hence to
develop a complete GMB-workshop-plan. And thirdly, from the perspective to evaluate GMB
effectiveness — especially for the aim of doing cumulative research — it is necessary to use
such shared conceptual frameworks and components of modeling (Rouwette & Vennix 2006).
In this paper, two new scripts are introduced with the aim of contributing to the catalogue of
tested and refined procedures — Scriptapedia. The first script introduces preliminary online
questionnaires to allow for building a seed model before the GMB workshops starts. This
procedure helps to dedicate more time in the GMB workshop to other important tasks and
activities. Especially when time is scarce and participants are located far apart from each, this
script is intended to be used. The second script introduces a technique to enable structured
thinking about open feedback loops of real-world systems that may require closure to
alleviate problematic system behavior.
Script 1: Preliminary Online Questionnaire
The participants and the session organizer of a Group Model Building workshop are often
located far apart from each other and from the session organizer. Therefore, the organizer may
be unable to meet up with each participant in person for the aim to conduct preparatory
interviews due to the incidental costs and limited available time. Especially, if a preliminary
model is intended to be used in the GMB workshop, V ennix (1996) suggests — in such cases —
to mail questionnaires to the participants to collect required information. Vennix mentions
that — on the one hand — the answers in questionnaires are more to the point then in interviews,
but — on the other hand — a low response rate is the most important danger in carrying out
mailed questionnaires.
Description
In this script, an easy to accomplish and technically more up-to-date altemative to mailed
questionnaires is suggested. The use of online forms simplifies setting up, filling out and
evaluating of questionnaires, thereby minimizing the time for both sides and resulting in a
higher response rate, so the assumption. This GMB-script is to serve as a guide line on how to
use a preliminary online questionnaire to build a seed model. The aim of the approach is to a)
collect the necessary information that allows the session organizer to become more familiar
with the topic, b) to serve as groundwork for building a seed model and c) as a first step for
initiating rapport between the modeling team and the participants.
"The current version is available in form of an open source wikibook via the URL hitps://en.wikibooks.org/wiki/Scriptapedia
For conducting online surveys, different suppliers that offer services to build and evaluate
online questionnaires are available via the internet (e.g. Google Forms’, Typeform’ or
Cognitoforms’). Instead of using such online services, it is also possible to build an online
form with available scripting languages (JavaScript, Ruby, Perl, PHP, etc.), but this approach
may be more time consuming. The use of online forms allows to quickly collecting and
evaluating information, as there are features available to automatically gather and aggregate
all the participants’ answers in form of an exportable Excel sheet. This procedure is more
convenient than mailed questionnaires for the participants, as each participant just needs to
click on a link and fill in answers in a webform on an internet page. A fter the appropriate tool
for establishing an online questionnaire is found, the work flow is as follows:
1. Set-up of questionnaire
2. Send link of the questionnaire to the participants
3. Evaluate questionnaire
4. Build the preparatory model
A detailed description of these steps can be found in the Appendix 1. Important for setting-up
the questionnaire is to have a problem definition available. Hence, the initiator of the intended
GMB-workshop needs to know at least the problem context — in form of a textual description
or in form of a graph over time chart of the problematic behavior — to be able to derivate a
problem definition.
There is much freedom for the setting-up of an online questionnaire. As a generic approach a
questionnaire consisting of three parts is described here. In case of particular circumstances,
the questionnaire might deviate from this standard triplet structure. In the first part, the
questionnaire checks if there is consensus among the participants in regard to the problem
definition. In the second part, the participants are questioned to name the key variables in the
problem context. And in the third part, a block of questions asks about which factors worsen
or lead to an improvement in respect of each key variable. This approach results in having a
list of key variables and a list of influencing factors on these key variables generated by the
participants, enabling to build a seed model. For the building of the preliminary model, all key
variables are carried together. A rank order is based on which variables and influencing
factors are named by how many participants. The experienced model builder needs to identify
duplicates in concern of this ranking list creation. Of course, not all variables can be
recognized in the seed model. The ones which are excluded are used as a pool of variables put
on the side of the black board in the first model building session. Having this pool of variables
collected before the first session enables to use the script Dots as follow-up script to
collectively assess how important different influencing factors are within the problem context.
Experience
The use of a preliminary questionnaire was tested within the scope of a GMB-workshop at the
Technical University Munich. The two-day workshop with in total eight participants, one
facilitator and two modelers, was part of the research project MultikOSi — Decision Support
System for Public Events: Multicriteria A pproach for Openness and Security.
At the beginning of the workshop a vague problem description was provided: “The problem
ps google. 7
> https://www.typeform.com/
* https://www.cognitoforms.com/
results from the relationship of mutual tension between openness, security and economic
feasibility in the context of urban event management.” In other words, the described problem
results from the situation that optimization efforts in the context of this trade-off triangle
which aim to improve one factor may worsen the other ones. Therefore, the goal was to map
the causal interdependencies among these different criteria. Because the available time was
limited due to the spatial distance among the participants and resulting travel costs, the
decision was made to set up an online questionnaire by using Google Forms to collect
required information enabling to build a seed model. Fortunately, the response rate to my
request to fill out the preliminary online questionnaire was 100 %.
Firstly, the participants where asked, if they agreed with the problem description (check boxes
with the options agreement, partial agreement and disagreement and a text box asking for
elaboration of this definition). Luckily, the problem definition already provided clarity on the
key variables in the problem context. Therefore, the main focus was to clarify which factors
influence these key variables (several open question text boxes in which different variables
could be written). In total 59 different influencing factors have been identified by the
participants. The seed model that was built as the main result of the preliminary online
questionnaire is shown in Figure 1.
Securit
ity.
\ Risk ofa disaster_—p» Coleco image <9: Roownical
$ iamac _
e + damage
il
Over regulated
course ofthe event
ie a
y.
Spontaneous
character _ =
Emotions
' ——"
‘<¢ ~~ ———
‘Openness _____—» Visitorsatisfaction
+
Economic
feasability
ia
Figure 1: Resulting seed model from the questionnaire used as starting point in the first Group
Model Building session.
Some modelling experience is necessary to build such a seed model. Advice on how to build a
preliminary model can be found in the script Causal Mapping with Seed Structure
(Scriptapedia & Contributers n.d.). One advantage of starting with a preliminary model as
boundary object (Black & Andersen 2012; Black et al. 2004) in the first session, is to describe
the causal mapping process and the symbols used by System Dynamics in the problem
context owned by the participants. All the influencing factors that were not included in the
preliminary model are used as a pool of variables in the first session. To prioritize these
different variables, the previously mentioned script D ots was used as first group activity. This
Dots script is very appropriate as a follow-up script, because the online questionnaire can be
characterized as a divergent group activity producing too many variables to be all used in the
workshop. This plentitude of variables makes it necessary to follow with a convergent group
activity for prioritization (see Figure 2).
‘icteunai@chket Oflenves'\ y2 7oeher heise
mae : 7d Gladeo [iceegnetor ‘On
ior
ginstive. Ten
Keine Nate
— Gielle Tanaiy Gee 8
ee y
aan S
GB tar Pst ibarkey
et 2vte Siderheilsanprdamen|= #
Figure 2: Prioritization of the different collected variables from the online questionnaire with
the script Dots.
Script 2: Find the open Loop
There is also a script available for finding non-existing feedback loops that could improve
system behavior if they were installed (initial policy options Scriptapedia & Contributers n.d.).
However, the script develops policy options from the participant’s narratives directly, in a
nominal-group-technique fashion only. Sometimes however, policy options can also be
developed from the inspiration of model structures. There are e.g. cases where the system
diagram developed by participants contains a feedback loop in in partial form (most variables
already present) lacking closure. If that loop was closed in reality and only open in the
diagram, the script ratio exercise (Scriptapedia & Contributers n.d.) could be used for
eliciting it. If however, the loop also lacks closure in reality, and this lack of closure is
potentially the cause for undesirable system behavior, it may be harder for participants to see
this.
Description
The script outlined below aims at aiding participants in finding such open feedback loops. If
carried out successfully, and the loop is described in a generic form, this unleashes creativity
for policy options that can be utilized by subsequent scripts that take from of concrete
feedback loop options.
The script is usually invoked when the development of the model structure in a
GMB-workshop reaches a point when either the facilitator or the content coach see an open
feedback loops that they suspect may also be open in the real system, and that this lack of
closure may be an underlying cause of undesirable system behavior. In case of the content
coach noticing the open loop, a silent signal the facilitator is advisable (e.g. handing a piece of
paper to the facilitator indicating an open feedback loop). If the facilitator decides to carry out
the script, the first step is to inform the participants that the modeling team may be seeing
something in the model that might be important but that a short modeling-timeout is required
in order to introduce the participants to something they need to know to be able to see what
the modeling team sees. Experienced modelers carry a plethora of generic structures (modules,
archetypes) as mental models in the back of their heads. When listening to workshop
participant’s narratives, modelers constantly seek to find such modules (e.g. balancing and
reinforcing feedback loops) that fit the participant’s story. Similarly, the system structures
modeled up to that point are also constantly compared to these archetypes (especially by the
content coach), seeking for congruence. When the facilitator as the second step introduces the
participants to this module using a diagram, it is important to do this separately from the
model structure developed so far, i.e. on a flip-chart a separate whiteboard or in case of
working in the computer-projector mode, using a different window. This is important in order
to respect model ownership by the participants and avoid boundary object failure by
introducing structures from the modeling team directly into the participant’s model without
prior consent. The module should either be introduced in a very generic form directly, or if
that is too challenging to understand, using an example first that is separate from the system
that the group is currently modeling and then generalizing to arrive at the generic form. This
is necessary so that participants still have room to develop their own concrete policies later on.
It is often important that the introduction to the module explains what behavior this structure
tends to produce and how this behavior results from its structure. As step 3, the facilitator then
turns back to the boundary object model and asks participants, whether they see this module
‘hiding’ in the system diagram they have developed so far. When the participants find the
open loop in their model and the facilitator draws it into the diagram of the group’s model
(step 4), the causal links that do not yet exist in the real system should be distinguished from
existing causal links (e.g. dashed or using a different color).
The creativity that usually springs up when participants immediately start thinking of ways of
implementing this generic loop in reality should be channeled using additional scripts such as
initial policy options. This is most important for larger groups, where uncontrolled
brainstorming is not advisable. A detailed description of the script in the standardized script
form can be found in the appendix.
In order to make the use of the script easier to understand we describe its use in a
GMB-workshop setting below. It should be noted though that the script as outlined above is
an adaptation that contains some improvements as compared to the original prototype outlined
below.
Experience
During a GMB-workshop the participants outline a situation where an institution regularly
contracts an external companies to carry out some work (e.g. construction). A third institution
is commissioned with controlling the work carried out by the contractors. The undesirable
system behavior (reference mode) is insufficient controlling: some deficiencies do not get
reported in time so that instead of the contractor fixing the deficiencies, the principle has to
fix them at its own cost. The participant explains that the people carrying out the controlling
have an interest of not reporting some of the deficiencies they found because it is a
bureaucratic effort for them. This is especially the case for small deficiencies because they
represent a relatively large bureaucratic effort for relatively low return. While listening to this
story, the facilitator got the impression that the reason for undesirable system behavior could
be seen as a lack of feedback from the controlling performance to the institution carrying out
the controlling to improve the quality of controlling. The description led to a system
diagram which was somewhat like variables and solid arrows depicted in Figure 3.
desired
controlling
performance
incentives for ~*
controlling quality controling
performance
discrepancy
bureaucratic effort of A+
reporting identified
deficiencies
potential cost o
‘ effort to lanch
severeness of savings of 7. .. . % of deficiencies Oe
identified —» reporting —» impactratio sound & reported in Gost for Principal for
ree + e - ofidentified time fixing deficiencies
deficiencies
Figure 3: Causal loop diagram with feedback loops that are open in reality and need closure
(dotted arrows) to improve system behavior
In order to validate his assumption on open feedback loops, the facilitator asked the
participants if the return (cost savings) are experienced by the principle only and not by
institution and the people carrying out the controlling. The participants agreed and noted that
in fact the only incentive in place actually points into a direction that is undesirable from As
perspective. The facilitator further asked the participants if there are any consequences of
controlling quality on the people carrying out the controlling. As the participants declined, the
facilitator drew the dashed closure of the feedback loop shown in Figure 3, which represents
such consequences. Note that this loop has a very general form, and could stand for a number
of potential policies (feedback loops). The participants were very exited about this new view
and immediately came up with policy ideas, as to how this loop could be closed.
While the script was successful in the form that was developed ad-hoc during the workshop,
this prototype is still suboptimal in the sense that the facilitator “gave away” too much of the
insight instead of aiding the participants in finding it themselves. Firstly, this is suboptimal
because it means that the facilitator has a stronger influence on the direction into which the
model is developing than necessary. This bears the risk of loss of model-ownership by
participants i.e. a segregation of the participants from the boundary object which is common
failure mode for GMB (Hovmand 2013). Secondly, it is suboptimal because the workshops
main goal was not to solve a specific problem but learning in terms of systems thinking
(learning problem rather than an analysis problem see Hovmand 2013). With this leaning goal
in mind, it would have been especially desirable for participants to find the open feedback
loop themselves.
The generic model structure that the facilitator suspected to be congruent with the problem at
hand was a simple balancing feedback loop, a regulatory feedback loop as indicated in Figure
4. In its improved version, the skript would expect the faciltiator to show and explain this
module to the participants. Alternatively, the content coach could be invited by the facilitator
to do this, especially if it was the content coach who suspected the structure in the first place.
It is possible but not always necessary to start with a different example first (eg.
unemployment and (implicitly) desired unemployment defining an empoyment discrepancy
that deterimes political pressurre to reduce unemployment). In any case this example diagram
should lead to the relatively generic diagram in Figure 4. It should be explained that systems
that work well, regulate their performance in such a way and that sometimes such loops are
not closed in systems and that this can result in undesirable system behavior. Some of the
arrows or variables in the figure can be hidden to better illustrate that e.g. information is not
transmitted. It would be important that the participants understand the behavior of the loop
(regulating actual towards desired values) and how this results from its structure. The generic
form of the diagram can thereafter be used to ask participants if they can find this module
hiding in their model structure.
performa nN
“a5
incentive to improve
weil
performance
actual performance
Figure 4: balancing feedback loop template
Note that the loop in Figure 4 is not the most general form (involving the word performance,
and incentive to improve) to make it a bit easier for the participants to transfer the concept to
the previously existing boundary object and the stories told by participants. For participants
with more experience in GMB, this could be made more challenging by choosing a more
generalized representation. It should not be so specific that it already represents a concrete
policy but rather leave room for several policies that participants may come up with. In this
case participants suggested e.g. that controlling performance could be evaluated by using key
performance indicators as well as changes within the organization structures that would
facilitate closing the loop. Especially with but not limited to larger groups it may be wise to
channel the creative momentum that develops at the end of this script by succeeding it with
the script initial policy options (Scriptapedia & Contributers n.d.) instead of letting the
participants brain storm.
Summary and Conclusion
We have introduced two new GMB scripts:
The first script uses succinct online questionnaires to elicit basic information for building a
seed model before the workshop. It involves the steps of setting up, sending around the link to
the questionnaire, as well as its partially automated evaluation and the subsequent
construction of the seed model. The script has shown to lead to very high response rates likely
by reducing transaction costs for participants both in terms of time and ease of use. It has also
shown to be efficient way of eliciting information for the modeling team.
The second script involves helping GMB-workshop participants find/see open feedback loops
the lack of closure of which may be responsible for undesirable system behavior. It involves
the steps of informing the participants of a modeling break to introduce something to them
that will enable them seeing more in their own diagram, then introducing them to the loop ina
generic form (possible using an example from a different context first), explaining its
behavior and how it emerges from its structure as well as the difference of open and closed
loops, and then asking participants if they see this module hiding in their model. Seeing the
loop closed in its general form, unleashed creativity in participants to develop policy options.
Further research should test and evaluate the performance of the improvements that the scripts
contain compared to their original form. It will for example be interesting to see how well the
modified version of the find-the-open-loop script performs in terms of participants finding
open loop(s) themselves. Will they be able to think of feedback loops in such generalized
form first and transfer the concept to the model structure developed so far or do they need an
example from a concrete policy from a different setting first to understand the general
version? Can this script be generalized to also find reinforcing feedback loops that are not
closed or even larger generic modules?
The authors would be thankful for any ideas for further improvements and any other
feedback.
Acknowledgement
This research was partly funded by the German Federal Ministry of Education and Research
as part of the program Research for Civil Protection (disclosure Urban Safety).
A Simulation Model of Katouzian’s Theory of Arbitrary State and
Society*t
Saeed P. Langarudi*
Email: slangarudi@wpi.edu
Michael J. Radzicki*
Email: mjradz@wpi.edu
Abstract: This paper represents an initial effort to model the volatile behavior of Iran’s socio-political-economic system.
More specifically, Home Katouzian’ s theory of Iranian political economy—a well-established descriptive theory of Iran’s
unstable i into a system dynamics model, tested for internal consistency, and used for
policy analysis. Simulation results re Katouzian’s claim that periodic episodes of significant arbitrary power are key to
understanding the historically less-than-optimal behavior of the Iranian socioeconomic system. They also confirm the
significance of oil revenue, economic sanctions, and civil resi on Iranian i Of note is that
experimentation with the model reveals that educational policies that generate increased respect for the law by Iranian citizens
can significantly improve the behavior of the Iranian socioeconomic system. The paper concludes with suggestions for future
research.
Keywords: Iran, Katouzian, Social Chaos, Arbitrary State, System Dynamics
* The authors would like to thank Homa Katouzian, the participants of the Collective Learning Meetings at Worcester Polytechnic
Institute, and four anonymous referees for their thoughtful comments on earlier drafts of this paper. Any errors of omission and/or
commission in this version of the paper are the sole responsibility of the authors.
"This article is an abbreviated version. The full version of this paper will be published in a special issue of Forum for Social Economics
(Langarudi and Radzicki 2015). The interested reader is encouraged to read the full article there.
* Social Science & Policy Studies, Worcester Polytechnic Institute, 100 Institute Road, Worcester, Massachusetts 01609, USA.
Appendix 1: Script Preliminary Online Questionnaire in
Scriptapedia format
Context
Before the first GMB session
Purpose
The purpose of this script is to
e Helps the facilitator become more familiar with the topic.
e Enables building a seed model.
e Initialize rapport between participants and the modeling team.
Status
Under development
Primary nature of group task:
Divergent
Time
Preparation time: Approximately 150 minutes
Time required from each participant before first session: Two times 15 minutes.
Follow-up time: Approximately 150 minutes
Materials needed
e Contact information of all participants
e Computer and access to the intemet
e Tool to set-up an electronic survey (e.g. Self-made Javascript page, Google Forms,
etc.)
Inputs
e Preliminary problem definition
e Basic understanding of the problem context.
Outputs
e Consensus-check on the problem definition
e Preliminary model (seed model)
e Pool of variables collected from the participants
Roles
e Questionnaire builder and evaluator
e Contact person
e Modeler/content coach or facilitator for seed model construction
Steps
The script is organized in four main steps:
a) Set-up of questionnaire
b) Send around link to questionnaire
c) Evaluate questionnaire
d) Build seed model
Step 1: Set-up of questionnaire
The questionnaire consists of three main parts. The first part checks if there is consensus on
the problem definition. The problem can be introduced with a graphs over time chart or with a
short textual explanation. After the problem definition was given, a first question asks, if the
participant approves to the problem definition. The participant has the possibility to give some
additional remarks afterwards. Especially, if the participant does not agree, (s)he is able to air
his/her thoughts.
Problem Definition
The problem is definied here by an explanation and/or a graph over time chart.
The next questions block focuses on the key variables in the context of the problem definition.
An explanation is added, that such variables are things that may increase or decrease over
time. The questionnaire asks to identify three key variables and to write them down each in a
separate box.
What are the three main key variables in the context of the problem definition. (A key variable is
hing that may it ord over time.)
Key variable 1
Key variable 2
Key variable 3
The final questions block focuses on influences that lead to worsening or improvement in
regard to each key variable. Each participant is asked to name two factors leading to a
worsening and two leading to improvement.
Which influence leads to a worsening in respect of key factor 1?
Factor 1
Factor 2
Which influence leads to a improvement in respect of key factor 1?
Factor 1
Factor 2
Step 2: Send link of the questionnaire around
After the questionnaire was set-up, the questionnaire can be send to all participants. Normally
this is done via mail, but it is also possible to ask for this by phone. Regardless which way of
communication is chosen, it is important to mention what the aim of the questionnaire is: the
ability to build a preliminary model for the first session. Furthermore it is important to
mention a deadline, until when the answers are needed. If participants have not given their
answers until this deadline, a gentle reminder will ask them once again to give their answers.
Step 3: Evaluate questionnaire
After the participants have filled out the questionnaires, all the answers are assessed. For this
purpose the answers are added all together to an excel sheet. The first thing is to check,
whether there is consensus on the problem definition. If many of the participants have not
partly or fully agreed on the problem definition, the facilitator can expect a high level of
conflict in the first GMB session. If this is the case, the facilitator is exhorted to read all the
comments carefully and to clarify this issue in the first session.
The second and third part of questions builds the basis for the building of the preliminary
model. All key variables are added together and put into a rank order based on the number of
mentions of the different participants. The same is done for the different influencing factors.
Sometimes different phrases are used for the same thing, then the facilitator needs to decide
which phrase is more suitable in terms of stock and flow terminology.
Step 4: Build of seed model
Based on the content from the questionnaires, an experienced SD modeler is able to build a
seed model. The resulting model can also contain unconnected structures that will be first
connected in the opening and future sessions. How often different factors where mentioned
should be one criterion to decide whether to include or not to include factors in the seed
model structure. Not included (key) factors are used as a bunch of variables and put beside the
seed model in the first session as a pool of variables, as would be collected by a NGT script in
the first session. To have such a pool of variables already before the first session starts saves
time and enables to use more resources on discussion and improvement in the ongoing GMB
workshop.
Evaluation Criteria
e Response rate of the participants
e Quality and quantity in respect of the answers
e Quality of the seed model
Authors
Oliver Handel
History
First described in
New Scripts for Group Model Building — Online Questionnaires and Open Loops
Revisions
Appendix 2: Script Find the Open Loop in Scriptapedia format
During GMB modelers may suspect that the undesirable behavior of a system is caused by a
lack of feedback. Instead of simply telling the participants about this suspicion, the facilitator
helps them find open loops themselves. This is especially meaningful if the G MB goals involve
some degree of capacity building of participants in terms of systems thinking.
Status
Under development
Primary nature of group task:
Convergent
Time
~ 30 minutes: Excluding collecting policy options = subsequent script, with another 20-45
minutes
Materials needed
GMB-equipment either cling sheet whiteboard or computer & projector
Inputs
e Growing Group Model, eg. from Nominal Group Technique, Initiating and
Elaborating a Causal Loop Diagram, Causal Mapping with Seed Structure
e Facilitator or content coach starts seeing a feedback loop lacking closure in the real
system as underlying cause of undesirable system behavior
Outputs
e Open feedback loops as causes for undesirable system behavior
e Excitement about new perspective, creativity unleashed for designing policy options
based on the insight
e Next script initial policy options
Roles
e Facilitator
e Content coach
e Wall-builder
Steps
1. Facilitator or content coach suspects an open feedback loop in the real system as
underlying cause of undesirable system behavior.
a. If itis the content coach, (s)he signals the facilitator and gives a piece of paper
to the facilitator indicating an open feedback loop.
b. Facilitator decides to run the script or not
2. Facilitator informs participants that s/he may be seeing something that requires a short
explanation (involving a time-out from the construction of the common model)
3. Facilitator introduces the generic structure (module) that the modeling team suspect
they discovered in the model, but in very generalized form. This could for example be
a CLD of balancing loop with discrepancy of desired vs. actual state. If the module
appears too difficult to understand in this generalized form, an example from a
different setting should be used to introduce it and then generalize from there. The
diagram should be shown in on a separate flipchart; separate window on the computer
to make clear this is an input from the modeling team. The facilitator explains the
behavior of this structure in real systems and how this behavior results from the
structure (e.g. that balancing feedback loops regulate systems towards goals (explicit
or implicit ones) and that sometimes a feedback loop that could regulate a system
towards a desired end is not closed). Facilitator asks participants if they see this
module (e.g. an open feedback loop) in their system. If participants do not understand
this in the generic form, the facilitator could introduce an example of an open
feedback loop (could be missing information feedback or missing incentive for a
decision maker, who does not feel the consequences of his decisions). This has to be
done with great care though choosing an example close to what the facilitator suspects
bears the risk that the facilitator influences the model too much, choosing an example
that is too far away is of little help to the participants (25”)
4. Facilitator / computer operator adds loops that participants talk about to the model,
(dashed). This should be done as general as possible to keep room for different policy
options. (5)
5. Developing such policy ideas may happen spontaneously as participants find the open
loop. Ideas should be collected by the wall-builder. In case of larger groups this
energy can be channeled into “initial policy options”-script as spontaneous
brainstorming may not very effective. (20-45’)
Evaluation Criteria
e Workshop participants find the feedback loop themselves
e Workshop participants are exited about new possibilities resulting from the new view
of the system (e.g. they develop policy options)
Authors
Max Kleemann
History
First described in
New Scripts for Group Model Building — Online Questionnaires and Open Loops
Revisions
References
Notes
References
Ackermann, F. et al., 2011. ScriptsMap: A tool for designing multi-method policy-making
workshops. Omega, 39(4), pp.427-434. Available at:
http://www.sciencedirect.com/science/article/pii/S030504831000126X [Accessed March
11, 2016].
Andersen, D.F. & Richardson, G.P., 1997. Scripts for group model building. System
Dynamics Review, 13(2), pp.107-129. Available at:
http://doi.wiley.com/10.1002/%28SIC1%291099-1727%28199722%2913%3A 2%3C 107
%3A %3A A ID-SDR120%3E3.0.CO%3B2-7 [Accessed February 18, 2016].
Andersen, D.F., Richardson, G.P. & Vennix, J.A.M., 1997. Group model building: adding
more science to the craft. System Dynamics Review, 13(2), pp.187-201. Available at:
http://doi.wiley.com/10.1002/%28S1C1%291099-1727%28199722%2913% 3A 2%3C 187
%3A %3AA ID-SDR124%3E3.0.CO%3B2-O [Accessed February 18, 2016].
Black, LJ. & Andersen, D.F., 2012. Using Visual Representations as Boundary Objects to
Resolve Conflict in Collaborative Model-Building Approaches. Systems Research and
Behavioral Science, 29(2), pp.194-208.
Black, L.J., Carlile, P.R. & Repenning, N.P., 2004. Theory of and Expertise Occupational
Boundaries in Technology Implementation: Building on Barley’ s Study CT Scanning.
Administrative Science Quarterly, 49(December), pp.572-607. Available at:
http://asq.sagepub.com/content/49/4/572.short.
Forrester, J.W., 1992. Policies, decisions and information sources for modeling. European
Journal of Operational Research, 59, pp.42-63.
Hovmand, P., 2013. Community Based System Dynamics, Springer Science & Business Media.
Available at: https://books.google.com/books?id=EQ0JA gAAQBAJ &pgis=1 [Accessed
March 10, 2016].
Hovmand, P.S. et al., 2011. Scriptapedia: A Handbook of Scripts for Developing Structured
Group Model Building Sessions. In Proceedings of the 29th International Conference of
the System Dynamics Society. Washington, DC, USA, p. 16.
Rouwette, EAJ.A. & Vennix, J.A.M., 2006. System Dynamics and Organizational
Interventions. Systems Research and Behavioral Science, 23, pp.451-466.
Scriptapedia & Contributers, Scriptapedia. Available at:
https://en.wikibooks.org/wiki/Scriptapedia [Accessed March 10, 2016].
Vennix, J.A.M., 1996. Group Model Building: Facilitating Team Learning Using System
Dynamics, Nijmegen, The Netherlands: John Wiley & Sons.
Over the last century Iran’s economic growth has been fairly unstable, primarily due to the dynamics of its political
atmosphere (Issawi 1971; Bharier 1971; Floor 1998). The unsteadiness of Iranian economic growth can be seen in the time
series data presented in Figure 1. In this figure Iranian GNP data is divided into two periods—1900-1960 and 1960-2010—so
that the instabilities in the Iranian economy can be clearly identified.
Iranian GNP (1900-1960) Iranian GNP (1960-2010)
350 700,000
300 600,000
Abdication of Reza Shah (1941) Revolution (1979)
250 500,000
400,000
300,000
200,000
100,000
SIS/CIA coup d'état against Mossadegh (1953)
0 0
1900 1910 1920 1930 1940 1950 1960 1960 1970 1980 1990 2000 2010
Figure 1: Real Iranian Gross National Product—in Billion Rials at constant prices*
An inspection of the figure reveals that the same qualitative pattern of behavior exists in both time periods — i.e., GNP
initially grows exponentially until a political disruption takes place. During the 1900-1960 period Reza Shah was forced to
abdicate during the Anglo-Soviet invasion of Iran in 1941, which was followed by a coup d'état against the democratic
governance of Mohammad Mossadegh in 1953. In the 1960-2010 period the political system endured a crisis in the mid-
1970s that precipitated the 1979 revolution. In both cases the Iranian economy collapsed after a period of political instability
and it took a while for it to return to its previous pattern of growth.
In terms of a more generic and simplified pattern of behavior, the dynamics inherent in Figure 1 can be portrayed by the
growth-stagnation-growth time shape presented in Figure 2. Arguably, a useful theory of Iranian socio-economic
development should be able to replicate this qualitative mode of behavior.*
4 Figure 1 was created from a combination of two datasets. The first source of data, shown in the left-side diagram, comes from the
work of Bharier (1971, 59), who provides a realistic estimate of Iran’s real GNP from 1900 to 1960 in constant 1959 prices. The second
source, shown in the right-side diagram, comes from the online portal of Iran’s Central Bank (CBI 2014), which provides data on Iran's real
GNP from 1959 to 2010 in constant 1997 prices.
5 Saeed (1992) argues that complex dynamic behavior modes should be “sliced” into simpler qualitative time shapes (i.e., reference
modes) so that a system dynamics modeling effort can be directed toward capturing the feedback processes that generate them.
Figure 2: mode the jitative behavior of Iranian GNP
Since the 1970s there have been many attempts to explain the distinctive dynamics of Iran’s macro economy.® The
literature on Iranian economic development is vast and can be broadly divided into two major groups: quantitative analyses
and qualitative (descriptive) studies. Quantitative analyses’, mostly econometric models, are highly dependent on numerical
data and thus intrinsically unable to explain Iran’s long-term economic dynamics because most Iranian time series data only
goes back to 1959. The reliability of these data is also suspect (Amuzegar 1997). Moreover, the effect of political factors such
as revolution and war are normally represented as exogenous inputs into these econometric models, which implies that these
phenomena are created by external forces. In fact, the very nature of the methods employed in these studies prevents a
modeler from integrating Iran’s socio-political system into a model of its economic system. As a consequence, most of the
modeling studies undertaken by mainstream Iranian economists have been unable to incorporate those features of Iran’s
socioeconomic system that are key to understanding its dynamics. Stated differently, most quantitative analyses have utilized
factors that are merely the result of the complex interrelationships that comprise the Iranian socioeconomic system, rather
than the root causes that define the system’s complex interrelationships and that generate its dynamics.
Qualitative studies of the Iranian economy, on the other hand, go far deeper into the very complicated and interrelated
feedback structures that define the Iranian socio-political system. Some of these studies are more general and try to explain
the causes of relative economic underdevelopment in eastern societies’, while others are case studies that specifically focus
‘on Iran’s socio-economic system and explain “why Iran lagged behind while the west moved forward.”?"° Although these
studies provide more detailed—and hence more realistic—explanations for the system’s behavior, they lack two important
features that are crucial for rigorous scientific work. First, they cannot generate synthetic data that can be formally compared
to numerical data from the actual system. Second, rigorous policy analysis is not possible because they cannot be used to run
controlled experiments.
The purpose of this paper is to provide a rigorous explanation for Iran’s pattern of unstable economic growth. The system
dynamics model put forth in this paper is based on the work of Homa Katouzian (1978; 1981; 1997; 2003; 2004; 2009; 2010;
2011), an economist and historian who created a well-known socio-political-economic theory of Iranian economic
development. The approach taken in this paper is to retain the richness of a qualitative study of the Iranian socio-political-
economic system and combine it with the rigor of a quantitative analysis.
System dynamics has already been used to test complex, nonlinear, and feedback-rich descriptive economic theories.’
In the case of Iran the first, and arguably most important application of system dynamics to economics was put forth by
Mashayekhi (1978). Mashayekhi developed a system dynamics model to analyze Iran’s long-term economic development
options made possible by its oil revenue. Since the focus of this model was oil revenue and its use in economic development,
and not the more general issues associated with Iranian political economy, it cannot be used to explain Iran’s long-run
a
5 See Esfahani et al. (2012) for a comprehensive review of Iranian macroeconomic modeling efforts.
7 See for example Habib-Agahi (1971), Bharier (1973), Heiat (1987), Valadkhani (1997), and Becker (1999).
8 Some of the most well-known theories in this area are “the Asiatic mode of production” of Karl Marx (Shiozawa 1966), Max Weber's
“theory of social and economic organization” (1947), and Wittfogel’s “oriental despotism” (Wittfogel 1957).
° This question is the title of a popular book in Iran written by Kazem Alamdari (2010).
29 See for example (Katouzian 1978; 1981; 1997; 2003; Ashraf 1980; Tabatabaei 2001; Arianpour 2003; Peyman 2003; Piran 2005;
Alamdari 2010).
11 Radzicki (2009) reports some of these efforts in his paper.
socioeconomic dynamics.” That said, beyond Mashayekhi’s work there has been no serious system dynamics modeling effort
aimed at analyzing the dynamics of the Iranian socio-political-economic system.
This paper represents an initial effort to model the dynamics inherent in Iran’s socio-political-economic system. More
specifically, Homa Katouzian’s theory of arbitrary state and society'—a very well-established descriptive theory of Iran’s
unstable economic development—is translated into a system dynamics model,” tested for internal consistency, and used for
policy analysis.** Initially, the model’s ability to mimic the irregular dynamics of the Iranian economy is presented. Then, the
model is used to test different scenarios and policy prescriptions aimed at improving the behavior of the Iranian
socioeconomic system.
In terms of building confidence in the Katouzian model, validation tests show that its dynamic behavior is consistent with
the qualitative behavior of both Iranian historical data and Iran’s socio-political-economic dynamics as described by Katouzian
in his theory.
In terms of simulation experiments the effects of both oil revenue and the citizenry’s respect for the rule of law on Iranian
economic development were examined. It is shown in this paper that periodic episodes of significant arbitrary power are key
to understanding the historically less-than-optimal behavior of the Iranian socioeconomic system. Simulation results indicate
that if Iran was a less arbitrary system it could experience a greater pattern of economic, social, and political development.
The results also show that although oil revenue has had a substantial impact on the economy it has had little effect on the
overall behavior of the Iranian socio-political system. Oil revenue helps the state to accumulate more power but doesn’t
change the generic cycle of “arbitrary rule-chaos-arbitrary rule.” Additional simulation experiments examined the impact of
economic sanctions and civil resistance on the political economy of Iran. From simulations of the Katouzian model it was
possible to generate some insight into the types of policies that might be effective in improving the dynamics of Iran’s socio-
political-economic system.
The purpose of this paper was to shed some light on the issue of the underdevelopment of a nation with an unstable
socio-political environment using Katouzian’s theory of Iranian political economy. Therefore, the boundary of the model was
limited to Katouzian’s theory of Arbitrary State and Society. The analytical capabilities of the model are thoroughly explored
and reported in this paper. In particular, it is shown that the model—if customized and elaborated appropriately—can be
applied to address the impact of socio-political-economic factors such as resource abundance, economic sanctions, civil
resistance, cultural transformation, etc., on the system as a whole.
WORKS CITED
A. 1978. “Mashayekhi Ali Naghi Strategy of ic Di in Iran: Of D on Exhaustible Resources [Book].
- Cambridge : MIT, Sloan School of Management, Ph.D. Dissretattion, .”
Alamdari, Kazem. 2010. Chera Iran Aghab Mand va Gharb Pish Raft? (Why Iran Lagged Behind & West Moved Forward?).
17th ed. Tehran: Iran: Nashre Towse’e.
Amuzegar, Jahangir. 1997. Iran’s Economy under the Islamic Republic. London, UK: |. B. Tauris & Company, Limited.
Arianpour, Amirhosein. 2003. “Marahel Tarikh Iran (Stages of Iranian History).” In Ketabe Towse’e (Handbook of
Development). Vol. 12. Tehran: Iran: Nashre Towse’e.
Ashraf, Ahmad. 1980. Mavane Tarikhi Roshd Sarmayedari dar Iran Doreye Qajar (Historical Barriers to Capitalism in Iran
during Qajar Era). Tehran: Iran: Zamineh Press.
Barlas, Yaman. 1992. “Comments on ‘On the Very Idea of a System Dynamics Model of Kuhnian Science.’” System Dynamics
Review 8 (1): 43-47. doi:10.1002/sdr.4260080105.
Becker, T6. 1999. Common Trends and Structural Change: A Dynamic Macro Model for the Pre-and Postrevolution Islamic
Republic of fran. International Monetary Fund.
2 For a detailed and updated review of Mashayekhi’s model which explores its capabilities and limitations see (Langarudi and Radzicki
2013).
33 The causal relationships that underlie Katouzian’s theory will not be justified here as this evidence exists in Katouzian’s numerous
works (Katouzian 1978; 1981; 1997; 2003; 2004; 2009; 2010; 2011). The interested reader can easily refer to them for additional
information.
14 A system dynamics model is essentially an institutionalist pattern model. See Radzicki (1988; 1990a; 1990b) for a comprehensive
discussion of economic methodology and its relationship to system dynamics modeling.
5 Fora lively discussion on the challenges inherent in translating someone else’s descriptive theory into a system dynamics model see
Wittenberg (1992), Sterman (1992), Radzicki (1992), Barlas (1992), and Wittenberg and Sterman (1992).
Beharie, Neville Olivier. 1973. “Economic Policy and Development Strategy in Iran: A Macro-Simulation Approach.” Ph.D.
Dissertation, Urbana-Champaign: University of Illinois.
Bharier, Julian. 1971. Economic Development in Iran 1900-1970. London, UK: Oxford University Press.
CBI. 2014. “Economic Time Series Database.” Economic Time Series Database, Central Bank of Islamic Republic of Iran.
http://tsd.cbi.ir/.
Esfahani, Hadi Salehi, Kamiar Mohaddes, and M. Hashem Pesaran. 2012. “Oil Exports and the Iranian Economy.” The Quarterly
Review of Economics and Finance.
Floor, Willem M. 1998. A Fiscal History of Iran in the Safavid and Qajar Periods, 1500-1925. New York: Bibliotheca Persica.
Habib-Agahi, Hamid. 1971. “A Macro Model for Iranian Economy: 1338-1349.” Tehran: Iran: Plan & Budget Organization.
Heiat, Abbas. 1987. “Econometric Study of an Oil-Exporting Country: The Case of Iran.” Ph.D. Dissertation, OR, USA: Portland
State University.
Issawi, Charles. 1971. Economic History of Iran, 1800-1914. Chicago: University of Chicago Press.
Katouzian, Homa. 1978. “Oil versus Agriculture a Case of Dual Resource Depletion in Iran.” The Journal of Peasant Studies 5
(3): 347-69.
. 1981. The Political Economy of Modern Iran: Despotism and Pseudo-Modernism, 1926-1979. Macmillan London.
. 1997. “Arbitrary Rule: A Comparative Theory of State, Politics and Society in Iran.” British Journal of Middle Eastern
Studies 24 (1): 49-73. doi:10.2307/195668.
. 2003. Iranian History and Politics: The Dialectic of State and Society. London, UK: Routledge.
. 2004. “The Short-Term Society: A Study in the Problems of Long-Term Political and Economic Development in Iran.”
Middle Eastern Studies 40 (1): 1-22. doi:10.2307/4289884.
———. 2009. The Persians: Ancient, Medieval, and Modern Iran. New Haven, CT: Yale University Press.
—— —. 2010. “The Iranian Revolution at 30: The Dialectic of State and Society.” Middle East Critique 19 (1): 35-53.
doi:10.1080/19436141003594575.
———. 2011. “The Revolution for Law: A Chronographic Analysis of the Constitutional Revolution of Iran.” Middle Eastern
Studies 47 (5): 757-77. doi:10.1080/00263206.2011.588797.
Langarudi, Saeed P., and Michael J. Radzicki. 2013. “Resurrecting a Forgotten Model: Updating Mashayekhi’s Model of Iranian
Economic Development.” In Energy Policy Modeling in the 21st Century, edited by Hassan Qudrat-Ullah, 197-233.
Understanding Complex Systems. Springer New York. http://dx.doi.org/10.1007/978-1-4614-8606-0_11.
———. 2015. “A Simulation Model of Katouzian’s Theory of Arbitrary State and Society.” Forum for Social Economics.
doi:10.1080/07360932.2015.1051076.
Peyman, Habibollah. 2003. “Avamel Baztolid Hokoumathaye Khodkameh dar Iran (Determinants of the Reproduction of
Arbitrary States in Iran).” In Ketabe Towse’e (Handbook of Development). Vol. 14. Tehran: Iran: Nashre Towse’e.
Piran, Parviz. 2005. “Nazarieh Rahbord va Siasat Sarzamini Jame’e Iran (State Strategy & Policy for Iranian Society).” Andishe
Jranshahr 6 (4): 23-34.
Radzicki, Michael J. 1988. “Institutional Dynamics: An Extension of the Institutionalist Approach to Socioeconomic Analysis.”
Journal of Economic Issues 22 (3): 633-65.
—— —.1990a. “Institutional Dynamics, Deterministic Chaos, and Self-Organizing Systems.” Journal of Economic Issues 24 (1):
57-102.
—. 1990b. “ ia O iae et is Dynamis.” System Dynamics Review 6 (2): 123-47.
doi:10.1002/sdr.4260060202.
—— —. 1992. “Reflections on ‘On the Very Idea of a System Dynamics Model of Kuhnian Science.’
8 (1): 49-53. doi:10.1002/sdr.4260080106.
Shiozawa, Kimio. 1966. “MARX’S VIEW OF ASIAN SOCIETY AND HIS ‘ASIATIC MODE OF PRODUCTION.” The Developing
Economies 4 (3): 299-315.
Sterman, John D. 1992. “Response to ‘On the Very Idea of a System Dynamics Model of Kuhnian Science.
Review 8 (1): 35-42. doi:10.1002/sdr.4260080104.
Tabatabaei, Javad. 2001. Dibachei bar Nazarieh Enhetat Iran (An Introduction to Iranian Retrogradation Theory). Tehran: Iran:
Negahe Moaser.
Valadkhani, Abbas. 1997. “A Macroeconometric Model for the Iranian Economy.” Ph.D. Dissertation, Brisbane: Australia: The
University of Queensland.
Weber, Max. 1947. “The Theory of Social and Economic Organization.” AM Henderson and Talcott Parsons (New York, 1947),
341-42.
System Dynamics Review
System Dynamics
Wittenberg, Jason. 1992. “On the Very Idea of a System Dynamics Model of Kuhnian Science.” System Dynamics Review 8 (1):
21-33. doi:10.1002/sdr.4260080103.
Wittenberg, Jason, and John D. Sterman. 1992. “Modeling the Dynamics of Scientific Revolutions.” System Dynamics Group
D-4334, Sloan School of Management, MIT, Cambridge, MA 25.
Wittfogel, Karl. 1957. “Oriental Despotism: A Comparative Study of Total Power.” Yale Univer Sity Press, New Haven.
