Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Int. J. Electronic Governance, Vol. x, No. x, 2016 1
Copyright © 2016 Inderscience Enterprises Ltd.
Labelled causal mapping for structuring public
policy problems and design of policy options
Osama Ibrahim1 and Aron Larsson1,2 1Department of Computer and Systems Sciences (DSV), Stockholm
University, Stockholm 16407, Sweden 2Department of Information and Communication Systems (IKS), Mid
Sweden University, Sundsvall 85070, Sweden
Email: [email protected]
Email: [email protected]
Abstract: Systems analysis allows quantitative, empirical testing of models
that exist in the study of public policy. Simulation and visualisation
techniques can help policy makers to reduce uncertainties on the possible
impacts of policies. This paper presents a new tool for systemic modelling
and simulation of public policy problems. The tool bridges the gap between
the user’s complex mental model and the explicit graphical representation in
order to enable knowledge representation and system analysis. It allows users
to build a systems model of a policy decision situation and simulate the
system behaviour and responses to changing external factors and policy
interventions over time. We propose a policy-oriented problem structuring
method, Labelled causal mapping, which provides a labelling scheme with
iconic representation of the involved actors and key variables and a
quantitative dynamic simulation model of the control flows and causal
dependencies. The ultimate aim of the proposed modeling and simulation
approach aims to support: (i) better understanding and transparency by
clarifying and sharing the modeling assumptions; (ii) evidence-based
policymaking by bringing facts and abstractions from scientific and experts’
knowledge into the modelling process; and (iii) incorporation of the newest
management technologies into public decision-making processes, including:
cognitive strategic thinking, scenario planning and participation. The tool
supports the design of policy options and integrated impact assessment in
terms of social, economic, environmental and other impacts. A web-based
tool prototype has been implemented in a Node.js environment and is
accessible from an online graphical user interface (GUI). Two policy use
cases are presented for demonstration and testing of the prototype.
Keywords: Public Policy Analysis, Exploratory modelling, Simulation and
visualization, Impact Assessment, Scenario Planning, Systems Thinking,
Causal Mapping.
1 Introduction
Much has been written about the complexity of public policy decision-making
problems. Those responsible for creating, implementing and enforcing policies are required
to make decision about ill-defined problems occurring in rapidly changing and complex
Osama Ibrahim and Aron Larsson
environments characterised by uncertainty and conflicting strategic interests among the
multiple involved parties (Mitchell, 2009; Davies, 2004).
The ability to detect problems and emergencies, identify risks and reduce uncertainties
on the possible impacts of policies are among the key challenges facing the policymaking
process (Stefano et al., 2014).
The impact assessment (IA) of policy proposals remains a challenge, since the effects
of the alternative policy options are delayed in time and the ultimate impact is affected by
a multitude of factors. In order to conduct a robust and relevant IA that implements the
principle of sustainable development, it is required to determine the social, economic,
environmental, organizational, legal and financial implications of a new policy (Wacław,
2014). In addition, there are certain key aspects which should be present in order to define
the scope of the policy analysis, including: (i) objective(s) of the policy analysis, (ii)
Geographical area: (global, regional, national, sub-national and local), (iii) Time aspect
(short, medium and long-term), (iv) sectors of the related governmental activities, (v)
participation of actors, and (vi) engagement (involvement) of stakeholders.
The focus of this study is the prescriptive policy analysis, the ex-ante impact assessment
carried out at early stages of policy formulation1, in order to answer forward-looking
questions concerning consequences of alternative actions and possible futures and provide
prescriptions about policy proposals under consideration.
The underlying question in this paper is “how to support public policy decision-making
processes on the different EU policymaking levels using an ICT tool for simulation of
policy consequences and possible future scenarios”? The aim then becomes to design of
an ICT tool for policy makers from the different EU policymaking levels that assists public
decision-making processes through exploratory modelling of a public policy problem,
simulating and visualizing the consequences of possible future scenarios and the societal
impacts of alternative policy (decision) options.
We aim from the tool introduced in this paper to allow users (policy analysts, policy
makers) to build their “own” models of policy problems to facilitate deeper understanding
of the problem and sharing of the problem definition for increased transparency. The design
objectives include the design of integrated, customizable and reusable models, conforming
to proper modelling standards, procedures and methodologies to allow model
interoperability. Of importance is to create more complex or wider perspective models
using existing components or sub-models (blocks) and to ensure long-term thinking by
incorporating time aspect into the simulation model. Other design objectives include:
supporting model validation to ensure the reliability of the model and, consequently, of
policies; engagement of decision-makers and a wide range of stakeholders; interactive
simulation using animations and visualisation techniques to display the model operational
behaviour graphically; and output and feedback analysis to provide a feedback on the
simulation process or on the initial modelling assumptions.
This work is carried out within the contexts of the project Sense4us2, aiming for the
creation of a web portal integrating public policy decision support tools, linked open data
search tools and social media analytics, in order to enable easy access to information
sources and knowledge creation.
User requirements in a policy modelling and simulation tool were assessed using both
interviews and an online survey in a study done within the Sense4us project, (Taylor et al.,
1 Policy formulation is the process of standardizing, or rating, the proposed policy as a viable, practical, relevant solution to the identified problem. The draft of legislation is usually considered as the second activity of policy
formulation but is left out of focus herein. 2 EU FP7 research project ‘Data insights for policymakers and citizens’, http://sense4us.eu/
Labelled Causal Mapping
2014) that targeted potential interested policy makers, analysts and researchers in
governmental and legislative institutions.
Trustworthiness was the most discussed requirement around policy modelling and
simulation. Most of the end users wished to understand the methodology the tool will use
and want to be able to see all the source data that is used to present the predicted outcomes
in order for them to trust the tool. Secondly, policy makers explicitly referred to the
dynamism of underlying factors in certain policy fields, therefore flexibility to choose
variables in the tool on their policy issues, in order to make comparisons between different
prognoses is important, leading to the idea of a member account being incorporated into
the tool to enable users to individually shape policy simulation.
Models and simulations are too often perceived as black boxes, unintelligible to the
users. This is a particular challenge in public policy where decisions need to be taken on
the basis of transparent information. There is a lack of policy-oriented modelling and
simulation tools, whereas the existing econometric models are unable to account for human
behaviour and unexpected events and the new social simulations are fragmented, single-
purposed, suffer from lack of scalability to the macro level and require high level of
technical competency (Stefano et al., 2014). There exist several software packages for
processing causal data, graphing and analysing causal maps (e.g., CMAP3 and Decision-
Explorer). In addition, there exist software packages for quantitative system dynamics
simulations, in a strict sense for system performance analysis and prediction (e.g.,
STELLA). None of them, however, is dedicated to policy analysis and decision support for
policymaking.
This study takes a design-science research, (Hevner et al., 2004), approach to
developing a public policy scenario planning and impact assessment simulation tool. In the
design-science paradigm, knowledge and understanding of a problem domain and its
solution are achieved in the building and application of the designed artefact. The design
science methodology is based on the following main activities: (i) Explicate the problem;
(ii) Outline artefact and define requirements; (iii) Design and develop artefact; (iv)
Demonstrate artefact; and (v) Evaluate artefact (Johannesson and Perjons, 2012).
We classify this study as a development and evaluation focused design-science study.
The focus is the design and development of the artefact using both research and creative
methods. The prototype system is assessed based on the user requirements and the design
objectives. In addition, the viability of the tool is demonstrated and a lightweight evaluation
is performed to assess the completeness of the prototype in handling correctly the semantics
and processing of the labelled causal mapping method.
2 Systems approach and Policy modelling
Easton (1965) envisioned the ‘Systems approach’ as a framework and model to address
the central problem of empirical political study. Systems approach argues that public policy
is the product of a system and that system is a compilation of both intra-societal
components and extra-societal components and actors within an environment. This
compilation of components and actors, along with the interjected demands, form inputs to
the political system (decision-making processes) – which in turn generate outputs (public
policy). The effectiveness of the outputs, as measured by feedback, form new inputs that
are acted upon by the system (Easton, 1965).
Stewart and Ayres (2001) have argued for the use of systems approach in policy making
by the following points: (i) the systems approach offers policy makers a fresh set of
perspectives on the fundamentals of policy analysis; (ii) policy design is as much a matter
of choosing structures and relationships as of choosing instruments; and (iii) understanding
Osama Ibrahim and Aron Larsson
causation means acknowledging two-way influence and the role of feedback. They
introduced both the terms: ‘systems analysis of policy’ (understanding what is happening
when policy is made) and ‘systems analysis for policy’ (generating concepts, ideas, and
modes of action to make recommendations about policy problems), (Stewart and Ayres,
2001).
Stefano et al. (2014) addressed the challenges facing the model-based collaborative
governance and inferred that the systems thinking and system dynamics approach may
prove a useful dynamic tool for next generation policy making, which can be applied in
conjunction with other modelling techniques to produce hybrid models for public policy
analyses.
Researchers in System Dynamics modelling (Andersen et al., 2007; Bryson et al., 2004)
and causal mapping (Ackermann and Eden, 2005) have been working together on finding
the right mix between methodological power and richness with a “Systems Thinking”
approach. Richmond (1994) identifies three impediments to representing complex mental
models as the series of stocks and flows in system dynamics modelling:
The gap between the explicit graphical representation and the mental model;
The complexity of the explicit model that results in cognitive overload;
The ambiguity of the semantics of System Dynamics diagramming.
The problem structuring methods (PSM) provides an alternative path. PSMs deal with
unstructured problems characterised by the existence of multiple actors, multiple
perspectives, incommensurable and/or conflicting interests, and key uncertainties. See
(Rosenhead and Mingers 2001) for an overview of the characteristics of a wide range of
PSMs.
The use of cognitive or causal maps is among the early approaches to enable for
problem understanding, originally intended for representing social science knowledge, see
(Axelrod 1976). What causal maps contribute is a visual, mental imagery-based simulation
of the system's behaviour for system analysis and social communication. It is obvious that
such maps can be useful for analysing, developing and sharing views and understanding
among key actors also for creating some preconditions for intervention (Wang and
Laukkanen, 2015). Large-scale causal maps can be used to model complex policy
problems, representing what a decision-maker thinks about the drivers, barriers,
instruments and consequences of change achieved by a certain policy proposal. In addition
to the policy analyst knowledge and domain expertise, data for building such maps are
acquired from various online sources, including: (i) published policy evaluation and impact
assessment reports; (ii) related research literature and reports from research institutes and
NGOs; (iii) open governmental data; and (iv) public online political discussions.
To deal with the dynamic complexity inherent in social systems and to infer dynamic
behaviour, quantitative simulation is required. Therefore, and particularly in those
situations where it is important to understand the interactions among the variables over
time, the value added by causal maps can be significantly increased if they are
complemented with simulation modeling (Senge, 1990; Sterman, 2000).
The long-term implications of policy making imply the need to consider the range of
possible futures, sometimes characterised by large uncertainties. “Scenarios” are a main
method of projection, trying to show more than one picture of the future. “The analysis of
scenarios of change allows the design of strategies to take place in spite of the messiness
of the situation” (Schoemaker, 2002). Scenario-driven planning closes the gap between
problem framing which depends on qualitative analysis and problem solving which
depends on quantitative analysis by blending qualitative and quantitative analytics into a
unified methodology (Georgantzas, and Acar, 1995).
Labelled Causal Mapping
3 Labelled causal mapping method
The proposed approach for policy modelling and simulation is based on: (i) systems
approach to policy analysis; (ii) problem structuring using causal maps; (iii) graphical
representation of complex problem situations as both a knowledge representation technique
and Systems analysis tool; (iv) scenario-based dynamic simulation. The main rationale is
to support a flexible, informative and a more rational and structured policy making process
identifying effective policies by gaining insight from conceptual and empirical analysis of
the system.
Acar, (1983) introduced the ‘causal mapping and situation formulation’ method in his
PhD dissertation as an analytical method for formulation and analysis of unstructured
strategic problems. The method enhances the causal mapping with rich computational
properties by including indications not only of the directions and signs of the presumed
causal influences, but also of their intensities, minimum threshold values and the possible
time lags, see (Acar and Druckenmiller 2006). The rich computational semantics of Acar’s
causal mapping support automated modelling and simulation in ways that other varieties
of cognitive mapping approaches do not. The automated simulation capabilities for Acar’s
causal mapping was explored in (Acar and Druckenmiller 2006; Druckenmiller and Acar,
2009). The complexity of causal loops in the models and calculations of successive waves
of change through the model when simulated presented a substantial design challenge for
developing automated scenario support using object-oriented techniques alone. A
prototype system for the development and simulation of causal maps that uses RePast 2.0,
a Java agent-based modelling (ABM) and simulation library. In this implementation
individual nodes of the causal map are represented as agents and links between agents
constitute the environment of the system, analogous to a communication network. It
focuses on business strategy development, interaction and behaviour of the involved actors
as autonomous agents.
This research proposes a policy-oriented version of Acar’s causal mapping, the
‘Labelled causal mapping’ as a systematic method for structuring of public policy decision
situations and simulation of possible futures and consequences of alternative policies. The
method defined policy-oriented categories with iconic representation for the model
elements. The following subsections provide a detailed description of the model elements
and the concepts of the simulation process.
3.1 Model elements and labels
3.1.1 Actors
Institutions, organizations, committees or individuals involved in the decision-making
process. The actors can be labelled as: (i) Official actors – including both: [legislative
actors (Parliament committees, political parties) and executive actors (Governmental
bodies, departments and institutions, chief Executive, staff/officials, agencies, bureaucrats
and civil servants)]; and (ii) Unofficial actors: [Interest groups, political parties, citizen
representative bodies, NGOs, industry/trade Unions, think tanks, media].
Table 1 Labels and icons for actors
Executive actor
Legislative actor
Unofficial actor
Osama Ibrahim and Aron Larsson
3.1.2 Variables
Variables are factors or events that structure, constrain, guide, influence and indicate
impacts of actions taken by actors. These are idealised as quantitative variables, or
quantified using value scales, so that it is meaningful to talk about change in the form of
increases or decreases in their levels. Variables are divided into:
Independent variables: sources of change (graph origins)
Policy instrumental variables (decision variables): These are variables controlled by
policy actors through various types of policy instruments, which can be used to classify
these variables. Scenarios of change in these variables represent action alternatives (policy
options). These changes reflect the allocation of natural, human and capital resources, the
regulatory role of the government, regional and international cooperation.
Table 2 Labels and icons for policy instruments – controllable sources of change
1. Economic Instruments:
1.1 Financial instruments:
1.1.1 Public expenditure, investment or funding
1.1.2 Public ownership
1.1.3 Subsidies
1.2 Fiscal instruments:
1.2.1 Taxes, Fees and User charges
1.2.2 Incentives
1.2.3 Loans / Loan guarantees
1.3 Market instruments:
1.3.1 Property rights
1.3.2 Contracts
1.3.3 Tradable permits / Certificate trading
1.3.4 Insurance
2. Regulatory Instruments:
2.1 Norms and standards
2.2 Control and enforcement
2.3 Liability
3. Informational Instruments:
3.1 Public information centers
3.2 Sustainability monitoring & reporting
3.3 Public awareness campaigns
3.4 Consumer advice services
3.5 Advertising & Symbolic gestures
4. Capacity-building Instruments:
4.1 Scientific research
4.2 Technology and skills
4.3 Training and employment
5. Cooperation Instruments:
5.1 Technology transfer
5.2 Voluntary agreements
Labelled Causal Mapping
External factors (state variables): These are variables not under control of any of the actors.
Scenarios of change in these variables represent the possible futures.
Table 3 : Labels and icons for uncontrollable sources of change
1. Drivers and barriers
The drivers and barriers of change, are either associated to the political
context (e.g., the political ideology and strategic priorities of the
government of the day, the preferences and demands of politicians) or
the economic context (e.g., the availability of resources, the economic
growth, the economic climate, current and future commitments).
2. External environment’s disturbances and conditions
The system representing the policy problem is surrounded by physical,
biological, social, and psychological environments that the system
needs to adapt to.
3. Social, demographic and behavioural change
e.g., Population growth, immigration, culture, attitudes and
behaviours.
Dependent variables: impacts of change (middle and end nodes)
These variables represent the consequences of change in the independent variables, are
either direct impacts, associated with the sources of change, or indirect impacts, associated
with the direct impacts. Table 4 shows the different categories of impact variables as related
to one of the policy areas.
Table 4 : Labels and icons for policy impacts
Economy
Finance
Environment
Community/Social
Energy Infrastructure Transportation Healthcare
Education
Technology
Judiciary and Law
National Security
The scope of the model is defined by the involved actors and the variables of interest.
The system analysis must consider the involved actors as coupled with either an abstract
or concrete component. This way, the influence of the actor within and upon the system
clearly reveals itself. Actors have powers (control over some decision variables) and
interests in some outcome variables (goal variables).
Osama Ibrahim and Aron Larsson
3.1.3 Change transmission channels
Links or change transmission channels are cause-effect relationships connecting the model
variables, defined by:
a) direction, from an upstream variable to a downstream variable
b) sign (positive is changes in the same direction or negative if changes are in
opposite directions)
c) change transfer coefficient: intensity of the causal relationship in terms of the
proportionality ratio of change transfer, how much change is transferred to the
downstream variable in case of 1% change in the upstream variable
d) time lag (if change transmission is not instantaneous) expressed as
weeks/months/years; and
e) minimum threshold for the change in the upstream variable (if applicable).
Two types of change transmission channels:
(i) full channel: a double arrow from an upstream variable X to a downstream variable
Y, if X is sufficient to induce change in Y
(ii) half channel: a single arrow from an upstream variable X to a downstream variable
Y, if X is necessary but not sufficient to induce change in variable Y. Half channels
from a set of variables such as {U, V, W} to variable Z, need to be all activated
before change can be transferred to Z.
Figure 1 : Change transmission channels
“Additivity” and “Transitivity” are two main characteristics of change transmission in
the model that allows running scenarios of change on the causal map. Once initiated in one
or more of the independent variables, change is transmitted throughout the network given
the transfer ratio and time lag for each channel. The main assumption for transmission of
change is that "the percentage relative change in a downstream variable Y is a linear
function of the percentage relative change in an upstream variable X". The definition of
time lags, minimum thresholds quantifications of the change transmission channels besides
the existence of the half channels add a meaningful dimension of nonlinearity to the model.
For full channels the transmission is automatic, as soon as a variable incurs a change, the
channels proceeding from it become activated and transmit to the downstream variables,
according to the indicated time lags. The same applies for half channels as soon as all half
channels converging to a node are activated. Assuming that 𝑑𝑋/𝑋 represents the relative
change in a variable 𝑋, then the relative change 𝑑𝑌/𝑌 in a downstream variable 𝑌 is given
by: 𝑑𝑌
𝑌= 𝑎
𝑑𝑋
𝑋, where 𝑎 is the real valued change transfer coefficient for the link XY. Given
a set of activated channels {𝑋1𝑍, 𝑋2𝑍, …, 𝑋𝑛𝑍} going into 𝑍, then: 𝑑𝑌
𝑌= ∑ 𝑎𝑖
𝑑𝑋𝑖
𝑋𝑖
𝑛𝑖=1 .
Labelled Causal Mapping
3.1.3 Demonstration policy use case
Figure 2 presents the causal mapping model for the use case policy problem, ‘Regulation
on Personal protective equipment (PPE)1’. The general policy objectives are: (i) high level
of health and safety protection for PPE users; (ii) free movement of PPE and a fair playing
field for PPE economic operators; and (iii) simplifying the EU regulatory environment
related to the field of PPE. The model shows the actors’ participation in a co-decision
legislative procedure (European commission’s directorate generals (DG): Enterprise and
Industry, DG-ENTR; Secretariat-General, DG-SG; Health and Food Safety, DG-SANCO).
The model shows two policy options, with the legislative option more effective in
achieving the policy objectives. Links are variably marked with positive and negative signs
indicating signs and intensity of the causal relationships.
Figure 2 : Causal map of the PPE use case
1 Source: European commission (EC), (2014) Impact Assessment report, Industry and Entrepreneurship,
‘Regulation on personal protective equipment’, available on:
http://ec.europa.eu/smart-regulation/impact/ia_carried_out/cia_2014_en.htm#entr
Osama Ibrahim and Aron Larsson
3.2 State of the system
The baseline or the “status quo level of the system”, is a projection of the
characteristics and behaviour of the system at the beginning of the analysis. It is used as a
point of comparison. The changes/responses are expressed in terms of percentage changes
relative to the baseline. Thus, change transfer coefficients are dimensionless and the initial
values of relative change for all variables in the system are set to zero, i.e. the initial state
in terms of the relative change is the null vector (0, 0, …, 0). The simulation runs are based
on discrete time points (T0, T1, T2, …, Tn), and the user is required to define a time step T,
time period between two consecutive time points as a number of weeks, months or years;
and maximum number of iterations n for the simulation. The state of the system at time Ti
is then defined as the values of all variables in the system at time point Ti of the simulation
run, given a specific scenario of change. The desired state of the system is given by targeted
changes in a set of impact variables of interest to the decision-maker, represented in a goal
vector. Each actor has a goal vector reflecting the targets of that actor.
3.3 Scenarios of Change
Quantitative analysis of the causal mapping model can show the opportunities for
policy interventions to achieve targeted changes in impact variables defined as policy
goals. The policy options represent different combined and controlled changes in the
system inputs to produce the targeted outcomes. Scenarios of change can be defined as a
combination of specified percentage changes occurring at a specific time point or at
successive time points. Defining quantified policy goals, in the form of a goal vector for
each actor, allows the analysis of these scenarios with respect to goal achievement. In
addition, structural analysis of the causal map, can support scenario analysis (e.g.,
reachability analysis shows the ability of a scenario of change triggered at an independent
variable to achieve a particular goal if the goal variable is reachable from this variable).
A “pure scenario” is a scenario of change in one particular independent variable, while a
“mixed scenario” is a scenario of change in more one than one independent variable. We
also need to differentiate between “alternative futures”, scenarios of change in
uncontrollable independent variables caused by natural or external forces, and “alternative
actions”, scenarios of change in controllable independent variables willed by actors.
The alternative futures reflect the concept of “uncertainty”: each possible future provides
a projection of changes in uncontrollable sources of change, a probability is assigned to
each of the possible futures. The alternative actions by actors reflect the different policy
options, i.e., the changes of controllable sources of change as policy instruments, given
some future circumstances. The user can define as many scenarios of change as needed,
using combinations of alternative futures and courses of action by actors. Simulation of the
alternative futures would bring up the various vulnerabilities of the system.
Analysing the behaviour of the system under various conditions yields a much deeper
understanding of the problem and supports the design of policy options.
3.4 Goal feasibility and compatibility
The formulation and evaluation of policy options can be addressed by answering
questions like: What variables are relevant and controlled by actors of the problem and
what constraints apply to them? What variables are relevant but are not subject to control?
How do controlled and uncontrolled variables interact to produce an outcome? What are
the actions needed to generate a desirable state of the system or to block the occurrence of
an undesired one? What are the costs and benefits of these actions?
Labelled Causal Mapping
Decision-makers at senior levels of government operate within a finite set of available
resources and timelines. Furthermore, there are inherent constraints that the decision-maker
needs to consider, such as annual cycles for strategic planning, budget, and legislation.
Legislative and political imperatives add to the complexity of government policy decision-
making and the selection of policies. For an actor, the triggering of change in one the
controllable variables imposes the expenditure of funds and resources. If an actor has the
required resources and/or funds available for a course of action that realizes his goals,
assuming inactivity of other actors and external environment, then his goal vector is
“internally feasible”. If the required resources are not available, then there is an intrinsic
problem of consistency between the actor’s goals and capabilities.
When considering the moves of other actors and the changing external environment,
if no pure or mixed scenario can be found to realize the actor’s goals, then his goal vector
is said to be “infeasible”. If it was found as an “internally feasible” goal, then the actor has
a problem to synchronize with the other actors. If a scenario could be found to realize the
actor’s goals, then his goal vector is said to be “feasible”. If it was not found as an
“internally feasible” goal, then the actor is benefiting from interacting with the whole
system in turning potential problems and constraints into opportunities.
The concept of compatibility is connected to the concept of feasibility. Components
of a single goal vector, as well as goal vectors of different actors are called “compatible”
if a scenario of changes can be found to realize them jointly. Goal compatibility is a graded
concept; two goals that can be realised by the same pure scenario are more compatible than
two goals requiring a mixture of pure scenarios.
3.5 Tactics and game theoretic analysis
A policy option can be viewed as either a combination of actions taken by the involved
actors that may achieve the policy goals in a cooperative decision-making situation (a co-
decision legislative procedure); or the actions taken by the focal actor to preempt/counter
natural, external forces and/or other competitors’ moves in a competitive decision-making
situation.
A tactic is a course of action logically paired with an alternative future and what other
actors might do. It is simply a change or sequence of changes triggered at controlled
variables by an actor that would help realize his goal vector. The effectiveness of a tactic
of an actor is a measure or at least an evaluation of the degree to which it helps him realize
his goal. The efficiency of a tactic is a measure of the use of resources to realize the goals.
Both the tactical effectiveness and efficiency need to be measured in comparison to the
competing tactics the actor might choose from and also in connection to the competitors’
tactics and alternative futures the actor aims at countering or preempting.
The competitive analysis allows policy makers to shape policies that takes into account
their competitors’ likely responses when deciding on their own actions, by quantifying and
estimating the utility each actor has in the alternative courses of action, while accounting
for the possible alternative futures. Game theoretic analyses are based on the idea of a
‘tactic’ as a sequence of moves to preempt or counter nature’s or other competitors’ moves.
It then allows either to devise one tactic for the set of actors involved in a cooperative
decision-making situation which would yield optimal result in achieving the policy
objectives; or to devise one tactic which would yield optimal results for the focal actor in
a competitive decision-making situation. For instance, Brandenburger and Nalebuff (1995)
provided a useful tool for identifying the various courses of action leaders have available.
They advocated using the acronym PARTS (Players, Added values, Rules, Tactics, and
Scope).
Osama Ibrahim and Aron Larsson
4 The Artefact description
Based on the labelled causal mapping method, a policy modelling and simulation tool
is implemented using a gaming simulation approach. Having a goal, the focal actor (in
individual mode) or each participant (in a group/participatory mode), tries out different
scenarios to identify the ones that best achieve the goals or intended consequences, with
the unintended and unexpected consequences used as a lasting learning method. Here, the
simulation model for policy analysis has an exploratory rationale rather than being a
prediction model. The individual simulation runs are not being treated as providing
predictions or explicit answers to the policy makers' questions. Instead, new information
that was implicit in their prior knowledge that defined what is plausible, is generated to
support an informed policy decision. The aim is to explore the implications of alternative
actions, varying assumptions and cause-and-effect hypotheses to deal with insufficient data
or unresolved uncertainties. Such exploratory approach, described in (Bankes, 1993), can
provide a basis for decision making in complex and uncertain problems.
The simulation-based impact assessment of user-created policy scenarios results in a set of
considered policy options and their simulated social, economic, environmental and other
impacts.
We suggest a step-wise procedure that enables the user to build a causal mapping
simulation model for the policy problem at hand. The cases of policy analysis are defined
as ‘policy problems’ and several impact assessment models can be defined for each policy
problem. ‘Ultra-Low Emission Vehicles (ULEV) Uptake1’ policy problem is used as a test
case.
Step 1. Define a policy problem A policy problem is defined using a title and a policy question and is given an ID by the
system. Figure 3 shows user interface for defining a policy problem, including:
1- Policymaking level: (EU level, National level, Local level)
2- Geographical area: (A set of EU countries – EU level; A country and a set of local
regions – National level; A local region – Local level).
Figure 3 : User interface for defining a new policy problem
1 Source: (Online) House of Commons, Transport Committee, (2012),‘Plug-in vehicles, plugged in policy?’,
Fourth Report of Session 2012–13, available on:
http://www.parliament.uk/documents/commons-committees/transport/Plug-in%20vehicles%20239.pdf (Online) Department for Transport, UK, (2015),‘ Uptake of Ultra Low Emission Vehicles in the UK’. A Rapid
Evidence Assessment, available on:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/464763/uptake-of-ulev-uk.pdf
Labelled Causal Mapping
Step 2. Define scope of the policy analysis model The user sets the boundaries of the policy impact assessment, as a guideline for the
modeling process and as an input to information search tools. This includes defining the
objective(s) of the policy analysis, the time aspect of the analysis and the related policy
domain or governmental activity, as shown in figure 4. The aim of this description or
classification is to facilitate the query processes to reuse model components from stored
policy models and to identify trends in groups of policy problems.
Figure 4 : User interface for defining scope of the policy model
Step 3. Create an information model of the policy problem A check-list editor, shown in figure 5, provides a tree structure for the concepts that
result from the surveying or information searching process under the main issues of the
policy problem and the different categories for the model elements. The editor allows the
user to edit the tree structure by adding or deleting elements.
Figure 5 : User interface for a policy check-list model
Osama Ibrahim and Aron Larsson
Step 3. Define indicators, measures and links to data sets for variables This step involves the definition of how to measure the concepts and decide which
indicators should be used. The indicators should be relevant to the scope, easy to track over
time, measurable through quantitative metrics or value scales. Open governmental data
portals provide statistical data and indicators for the different policy domains and themes.
Figure 6: User interface for defining measures and indicators for model variables
Step 4: Import or add elements to the model builder graphing canvas. In this step, the user adds elements to the model builder canvas or selects elements from
the information model formed by the information search results and imports them to the
graphing canvas as model elements with icons assigned to these elements based on the
defined categories. The labelled causal map graphing canvas operates in two modes: the
‘modelling’ mode and the ‘simulation’ mode. In the modelling mode, the user can create,
edit, save, load and delete models. The user can add elements, i.e., actors and variables to
the model by clicking on the respective icon from the model elements toolbar on the left
panel.
Step 5. Build hypotheses on the causal links and interrelationships: The user can refine the model structure by adding control flows and causal links (both full
and half change transmission channels), by the dragging the red dot that appears beside the
element by clicking on it or by a right-click and drag to the respective downstream variable.
Using.
Step 6. Edit nodes’ and links’ properties – Define a baseline By clicking a node, the user can edit the node properties and perform functions that appear
on the left panel, including: (i) For actors: name, description, delete actor, delete control
flows, randomise colour of the control flow; (ii) For variables: name, description, baseline
level, unit of measurement and delete node. Quantify the causal links and verify them
through statistical analysis of the data sets for indicators. In case of unavailability of
statistical data, the quantification can be done based on expert’s judgment.
Step 7. Define actors’ goals For each actor of the model, define the actor’s goal vectors and preferences of goals
realization (i.e., ranking of the goal vector components).
Step 8. Define and simulate scenarios of change In the simulation mode, the user can define the time step for the simulation model in terms
of a number of weeks, months or years and a maximum number of iterations, which appears
as the limit for the simulation time slider.
Osama Ibrahim and Aron Larsson
A scenario editor window can be used to define scenarios of change for the independent
variables (policy instruments and external factors), for which the change scenario appears
as part of the node properties, in terms of timely percentage changes in their levels relative
to baseline levels.
The ‘simulate’ button populates the model variables throughout the model with the
transferred changes, which are updated by moving the time slider. The ‘line graph’ button
shows a line graph for the simulation results for each of the model variables in different
colours, by selecting the respective nodes.
Figure 8, shows a model example for a demonstration policy case in the simulation mode
with the scenario editor window to define and edit multiple scenarios of change in the
independent variables. Figure 9 show the simulation results for a scenario of change and
Figure 10 shows the visualization of results using line graphs for selected nodes.
Figure 8 : Scenario editor window in the simulation mode
Figure 9 : Simulation results for a scenario of change
Labelled Causal Mapping
Figure 10 : Visualization of simulation results for selected variables
5 Conclusion
In the labelled causal mapping method, presented in this research as a policy-oriented
version of Acar’s causal mapping, a public policy problem is abstracted as deviations from
goals and/or standards, those deviations originate in a change that propagates itself through
causal connections, while a policy option is seen as a purposeful action for change that can
resolve these deviations in order to achieve the policy objectives. The analysis of the
transfer of change throughout the system can be used to assess the effectiveness and
efficiency of alternative scenarios of change in achieving objectives.
The basic linearity of the system can be a source of criticism, but it allows increasing
the level of detail of the model by adding as many elements as needed to represent relevant
factors that affect the outcome of a decision. Therefore, it increases the model’s capability
to be used for policy analysis, without affecting its feasibility in terms of mathematical
tractability and data requirements.
The prototype system implemented in this research is assessed based on the design
objectives, as follows:
- User-created policy scenarios: Although it is a serious and time consuming task, it helps
the user to develop advanced skills and gain a deeper understanding.
- Integrated, customisable and reusable models: Creating more complex or wider
perspective models using existing components or sub-models (blocks).
- Model validation: Defining data requirements for sensitivity analysis and validation.
Engaging in a cumulative learning through experimentation. The model validation, which
is to be done by the user, is reduced to validation of sub-models, determining model
parameters from validated sources and ensuring the plausibility of outcomes. And the
model quality control is limited to the plausibility and the degree of completeness of the
model (inclusion of all factors and phenomena that might influence outcomes).
- Engagement of decision-makers and stakeholders: The resulting policy model being
simple, robust, easy to control, adaptive, easy to communicate with and as complete as
possible.
Osama Ibrahim and Aron Larsson
- Interactive simulation: Using a graphical representation of the policy problem is
important for end-users for facilitating communication about the problem understanding
among different governmental departments
- Output and feedback analysis: Synthesising new knowledge on the system, when
ultimately, a satisfying result has been achieved or when a complete understanding of the
system has been gained.
6 References
Acar W., (1983). Toward a Theory of Problem Formulation and the Planning of Change:
Causal Mapping and Dialectical Debate in Situation Formulation. Ann Arbor,
Michigan: U.M.I.
Acar, W. and Druckenmiller, D. (2006). Endowing cognitive mapping with computational
properties for strategic analysis, Futures 38:993-1009.
Bankes, S. (1993). Exploratory modeling for policy analysis. Operations Research, 41(3),
435-449.
Brandenburger, A. M., & Nalebuff, B. J. (1995). The right game: Use game theory to shape
strategy. Harvard Business Review.
Davies PT (2004), “Is evidence-based government possible?” Jerry Lee lecture. 4th Annual
Campbell Collaboration Colloquium Washington D.C.
Easton, D. (1965). A systems analysis of political life. New York: John- Wiley & Sons.
Georgantzas, N. C. and W. Acar (1995). Scenario-Driven Planning: Learning to Manage
Strategic Uncertainty. Westport Connecticut: Quorum Books.
Hevner, A., March, S., Park, J., and Ram, S. (2004). “Design Science in Information
Systems Research,” MIS Quarterly (28:1), pp. 75-105.
Johannesson, Paul & Perjons, Erik (2012). A Design Science Primer.
Koliba C., Zia A., (2011), Theory Testing Using Complex Systems Modeling in Public
Administration and Policy Studies: Challenges and Opportunities for a Meta-
Theoretical Research Program, Proceedings of the 2011 Public Management Research
Conference, Maxwell School of citizenship and public affairs at Syracuse University.
Mingde Wang and Mauri Laukkanen (2015), Comparative Causal mapping: The CMAP3
method. Ashgate Publishing, Ltd..
Mitchell B. (2009). Policy-making process. Available: (retrieved on 10/03/2015)
http://www.waterencyclopedia.com/Oc-Po/Policy-Making-Process.html
Rouse, W.B & N.M. Morris (1986): On Looking Into the Black Box: Prospects and Limits
in the Search for Mental Models. Psychological Bulletin, Vol. 100, No.3, 349363
Schoemaker, P. J. H. (2002). Profiting From Uncertainty: Strategies for Succeeding No
Matter What the Future Brings. New York: Free Press.
Senge P.M. (1990). The Fifth Discipline: The Art and Practice of the Learning
Organization. New York: Doubleday Currency.
Labelled Causal Mapping
Stefano A., Camello C., Riccardo O. & Pietro S. A., (2014): Policy Modeling as a new area
for research: perspectives for a Systems Thinking and System Dynamics approach?
Proceedings of the Business systems Laboratory 2nd International Symposium.
Sterman J. D. (2000). Business Dynamics: Systems Thinking and Modelling for a Complex
World. Irwin/McGraw-Hill: Boston.
Taylor S., Uzdavinyte R., Wandhöfer T., Fox R., (2015). Issues Arising from the
Specification of an Information Acquisition and Analysis Toolkit for Policy Makers in
Governmental and Legislative Institutions, Accepted paper to be published/presented
at: The eChallenges e2015, Vilnius, Lithuania.
Wacław B., (2014), Regulation Impact Assessment (RIA) at Poland and at Some EU
Countries. Procedia - Social and Behavioral Sciences Volume 109, Pages 45–50. 2nd
World Conference on Business, Economics and Management.