Infotech in global economy

Project co‐funded by the European Union under the Seventh Framework Programme © Copyright 2015 Stockholm University, Department of Computer and Systems Sciences (DSV)

Policy Modelling and Simulation Tool

A Simulation Tool for Assessment of Societal Effects of a Proposed Government Policy

Project acronym: SENSE4US Project full title: Data Insights for Policy Makers and Citizens

Grant agreement no.: 611242 Responsible: Stockholm University – eGovLab Contributors: Aron Larsson, Osama Ibrahim

Document Reference: D6.2 Dissemination Level: PU

Version: Final Date: 30/06/2015

D6.2 Policy Modelling and Simulation Tool

2 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

History

Version Date Modification reason Modified by

0.1 2015‐05‐15 First draft Aron Larsson, Osama Ibrahim

0.2 2015‐06‐01 Second Draft Aron Larsson, Osama Ibrahim

0.3 2015‐06‐20 Quality check Steve Taylor, Somya Joshi

0.4 2015‐07‐01 Final Draft Aron Larsson, Osama Ibrahim

D6.2 Policy Modelling and Simulation Tool

3 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

Table of Contents

History ............................................................................................................................... 2

Table of Contents ............................................................................................................... 3 List of Figures ..................................................................................................................... 4 List of tables ...................................................................................................................... 5 List of Abbreviations .......................................................................................................... 6 Executive Summary ........................................................................................................... 7

Task ...................................................................................................................................... 7

Design Objectives ................................................................................................................ 7

Introduction ...................................................................................................................... 8

1 Model Description .................................................................................................... 11 1.1 Actors ......................................................................................................................... 11

1.2 Variables .................................................................................................................... 11

Independent variables (sources of change) ...................................................................... 12

Dependent variables (impacts of change) ......................................................................... 13

1.3 Change transmission channels .................................................................................. 14

2 Fundamental simulation concepts ............................................................................ 16 2.1 State of the system .................................................................................................... 16

2.2 Scenarios of Change .................................................................................................. 16

2.3 Goal feasibility and compatibility .............................................................................. 17

2.4 Tactics and Game theoretic analysis ......................................................................... 17

3 Simulation Process ................................................................................................... 19 3.1 Generating Scenarios ................................................................................................. 19

3.2 Graph change analysis ............................................................................................... 19

3.3 Data, forecasting and predictive validation .............................................................. 20

4 Policy Analysis Model building Process ..................................................................... 22

5 Conclusion ................................................................................................................ 33 Enhancements and Future work ....................................................................................... 33

6 References ............................................................................................................... 35 APPENDIX I – Computation Algorithm .............................................................................. 36 APPENDIX II – TECHNICAL SPECIFICATIONS ...................................................................... 38 APPENDIX III – Policy use cases ........................................................................................ 40

D6.2 Policy Modelling and Simulation Tool

4 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

List of Figures

Figure 1 : Change transmission channels ............................................................................ 15 Figure 2 : single full channel example ................................................................................. 19 Figure 3 : multiple full channels example ........................................................................... 20 Figure 4 : multiple half channels example .......................................................................... 20 Figure 5 : User interface for defining a new policy problem ................................................ 22 Figure 6 : User interface for defining scope of the policy model .......................................... 23 Figure 7 : Example for a basic search query using the policy problem title .......................... 24 Figure 8 : User interface for selecting areas of policy impacts ............................................. 24 Figure 9 : Import concepts to the causal map graphing canvas ........................................... 28 Figure 10 : User interface for defining actors’ powers and goals ......................................... 29 Figure 11 : User interface for defining measures and mapping time series to them ............ 29 Figure 12 : Edit node and link properties ............................................................................ 30 Figure 13 : User interface for defining a scenario of changes and the scenario simulation as viewed on the causal mapping canvas ............................................................................... 30 Figure 14 : Causal mapping model example ....................................................................... 31 Figure 15 : Causal map of the PPE use case ........................................................................ 43

D6.2 Policy Modelling and Simulation Tool

5 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

List of tables

Table 1 : Simulator icons for actors .................................................................................... 11 Table 2 : Simulator icons for policy instruments – controllable sources of change ............... 12 Table 3 : Simulator icons for uncontrollable sources of change ........................................... 13 Table 4 : Simulator icons for policy impacts ........................................................................ 13 Table 5 : Keywords for actors ............................................................................................. 25 Table 6 : Keywords for sources of change ........................................................................... 25 Table 7 : Coded categories of model elements .................................................................... 26 Table 8 : Examples for the categorised search results ......................................................... 26

D6.2 Policy Modelling and Simulation Tool

6 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

List of Abbreviations

<Abbreviation> <Explanation>

API Application Program Interface

DSS Decision Support System

GUI Graphical User Interface

EC European Commission

IA Impact Assessment

ICT Information and Communication Technology

MCDA Multi‐Criteria Decision Analysis

Sense4us Data insights for policy makers and citizens (this project)

URL Uniform Resource Locator (web address)

WP Work Package

D6.2 Policy Modelling and Simulation Tool

7 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

Executive Summary

The deliverable (D6.2) presents a prototype for a policy‐oriented modelling and simulation tool that allows users, through a web‐based, user‐friendly interface, to build a systems model of a public policy problem situation using a graphical representation of the involved actors, the key variables, control flows and causal dependencies. A quantitative dynamic simulation model of the structured problem is used to simulate the system behaviour and responses to changing external factors and policy interventions over time. The tool supports the design of policy options and integrated impact assessment in terms of social, economic and environmental impacts.

The proposed modelling and simulation approach aims to provide: (i) better understanding and transparency by clarifying and sharing the modelling assumptions; (ii) an 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.

Task

The design of an ICT tool for policy makers from the different EU policymaking levels that assists public decision‐making processes through participatory modelling of a public policy problem, simulating and visualising the consequences of possible future scenarios and the societal impacts of alternative policy (decision) options.

Design Objectives

1‐ User‐created policy scenarios: Models and simulations are often perceived as black boxes, unintelligible to the users. Allowing users to build “own” models for the policy problem to ensure that policy decisions are based on deep understanding and transparency. 2‐ Integrated, customizable and reusable models: Defining proper modelling standards, procedures and methodologies to allow model interoperability to create more complex or wider perspective models using existing components or models (blocks) and to ensure long‐ term thinking by incorporating time aspect into the simulation model. 3‐ Engagement of decision‐makers and stakeholders (even without domain expert skills) in a participatory modelling process. 4‐ Easy access to information and knowledge creation in order to reduce uncertainty: integration to other work packages to support problem structuring using inputs from WP4 and WP5. It is of interest to see how the information obtained from open data sources and analysis of political discussions on social media and blogs (all available within the Sense4us toolkit) contribute to increased problem understanding. 5‐ Model validation: in order to ensure the reliability of the model and, consequently, of policies. A model is valid if it is built using the most relevant components and sub‐models and is able to reproduce historical behaviour. 6‐ Interactive simulation: the use of animations and visualization techniques to display the model operational behaviour graphically as the model runs over time. 7‐ Output and feedback analysis: learning from output analysis, being able to provide a feedback on the simulation process or on the initial modelling assumptions and thus synthesizing new knowledge on the system, when ultimately, a satisfying result has been achieved or when a complete understanding of the system has been gained.

D6.2 Policy Modelling and Simulation Tool

8 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

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 a rapidly changing and complex environments characterised by uncertainty and conflicting strategic interests among the multiple involved parties [1] [2].

“Policy Modelling and ICT‐enabled Governance”, has emerged as an interdisciplinary umbrella term for a number of research fields, technologies and applications with a common goal of improving public decision‐making in the age of complexity and has recently gathered significant attention by governments, researchers and practitioners. It brings together two separate worlds: the mathematical and complexity sciences background of policy modelling and the sustainability, service provision, participation and open data aspects of governance [3].

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. Simulation and visualization techniques can help policy makers to anticipate unexpected policy outcomes. The focus of this study is the prescriptive policy analysis, the impact assessment (IA) carried out at early stages of policy development. This study is done as part of the decision support framework for policy formulation1 described in D6.1.

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 [4]. 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) Space or Geographical area: (global, regional, national, sub‐national and local), (iii) Time (short, medium and long‐term), (iv) Types and sectors of the related governmental activities, (v) Power (participation of actors), and (vi) Engagement of stakeholders.

The impact assessment 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. The following questions have to be dealt with in a transparent manner and from early on in the decision‐making processes:

 What is the main purpose(s) of the policy?  What is the context of the policy (Influencing factors)?  What are the relevant ways of intervention (policy instruments)?  What are the relevant impacts which require further analysis?  Who are relevant stakeholders and target groups which should be consulted?  What are appropriate methods to assess the impacts and to compare the policy

options?

Before proposing a new initiative, the European Commission (EC) assesses the need for EU action and the potential economic, social and environmental impacts of alternative policy

                                                             1 Policy formulation: standardizing or rating, the proposed policy as a viable, practical, relevant solution to the identified problem. The development of pertinent and acceptable courses of action dealing with public problems is an essential part of any policymaking process.

D6.2 Policy Modelling and Simulation Tool

9 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

options2. Planning of IAs is communicated to the public via roadmaps, consultation of stakeholders and public online consultations including annual revisions of the IA guidelines, in addition, final IA reports are made public3. IAs are prepared for these initiatives expected to have significant impacts, including: (i) legislative proposals, (ii) non‐legislative initiatives (white papers, action plans, financial programmes, negotiating guidelines for international agreements) and (iii) implementing and delegated acts.

As early as the 1960s, Easton (1965) envisioned the ‘Systems approach’ as a framework and model to address the central problem of empirical political study [5]. Such a framework assumes that: (i) political interactions in a society constitute a system; (ii) the system must be seen as surrounded by physical, biological, social, and psychological environments, i.e., political life forms an open system; (iii) systems must have the capacity to respond to disturbances and thereby to adapt to the conditions under which they find themselves. In Easton’s systems approach, the five tenants of a framework are: ‘Actors’, ‘Variables’ (the inputs, the processes, the outputs, and the feedback), ‘Unit of analysis’, ‘Level of analysis’ and ‘Scope’ [5]. Introducing the Systems thinking to the policymaking process allows for both a holistic and narrow examination of the public policy problem, the environment, actors and abstract and concrete components.

For purposive, intelligent action, understanding and safety needs, etc., normal people need representations of their action context (mechanisms of external and internal factors affecting decisions), including one’s own and other actors’ actions. Such internal representations have been called variously mental models, causal or cognitive maps, meaning in general: “mechanisms whereby humans are able to generate descriptions of system purpose and form explanations of system functioning, observed system states, and predictions of future system states” [6].

Causal maps can be developed by individual decision‐makers to model the structural systemic elements of their situation and show how change is propagated through the system. “What causal maps contribute is a visual, mental imagery‐based, “mind’s eye” simulation of the system's behavior for system analysis and social communication” [7]. 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.

Large‐scale causal maps can be used to model complex policy problems, representing what a government decision‐maker thinks about the drivers, barriers, instruments and consequences of change achieved by a certain policy proposal. Data for building such maps are acquired from the decision makers or from other sources including the WP4 Linked Open Data Search tools, WP5 Social media Analysis tools, and documents such as: previous policy evaluation or impact assessment reports, related research literature and reports from research institutes and NGOs.

To deal with the dynamic complexity inherent in social systems and to infer dynamic behaviour, quantitative simulation is required [8][9]. Therefore, and particularly in those situations where it is important to understand the interactions among the variables over time, the value added by Causal/cognitive maps can be significantly increased if they are complemented with simulation modelling.

Stefano et al. (2014), addressed the challenges facing the model‐based collaborative governance and the policy modelling issues in practice. As it was revealed by the results obtained in two subsequent EU FP7 projects: the CROSSROAD project and the CROSSOVER

                                                             2 http://ec.europa.eu/smart‐regulation/impact/index_en.htm 3 http://ec.europa.eu/smart‐regulation/impact/ia_carried_out/cia_2015_en.htm

D6.2 Policy Modelling and Simulation Tool

10 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

project, the authors 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 [10].

There exist several software packages4 for processing causal data, graphing and analysing causal maps. In addition, there exist software packages5 for quantitative system dynamics simulations, in a strict sense for system performance analysis and prediction. None of them, however, is dedicated to Policy analysis and decision support for policymaking. 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 by users.

The opportunity we have here is to create a policy oriented tool that supports systems‐based modelling of public policy problem situations and simulation‐based impact assessment.

In these contexts we believe that the design of a policy‐oriented modelling and simulation tool, as a main component of the Sense4us Policymaking DSS, should be based on:

(i) ‘Systems approach’ to the study of public policymaking; (ii) User‐created policy scenarios; (iii) Graphical representation of complex problem situations using causal maps as

both a knowledge representation technique and Systems analysis tool; (iv) Scenario Planning and Dynamic simulation modelling

This allows for a problem definition that: (i) reflects the systemic nature of most of central policy areas, (e.g., Energy, Financial Systems, Innovation/Growth), for which a regulation/policy needs to be based on a view of the system as a whole; and (ii) provides a visual problem model that clearly communicates the policy makers’ thoughts and can bring together different policy actors. 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 analysis of the system. The argument behind the use of a graphical representation is simplifying and summarising the decision maker’s knowledge and information gathered from various online sources about a social, socioeconomic or sociotechnical system and visually simulates the system behaviour and responses to interventions over time. Thus, the causal mapping graphical representation can be used as a contextual framework that highlights knowledge gaps, guides information searching and models the search results from various online and other work packages sources.

The current technical specifications of the implemented online simulation tool is given in Appendix I. Specifications will be updated as development proceeds and are published at Google Docs6 and the online GUI for the tool is reached through the URL http://dev1.egovlab.eu:4001/.

                                                             4 For example, CMAP3 – Comparative and composite causal mapping (http://www2.uef.fi/fi/cmap3) and Decision Explorer (http://www.banxia.com/dexplore/index.html). 5 For example, STELLA (http://www.iseesystems.com/). 6 https://docs.google.com/document/d/1fBr‐pcJLioMccZzf3_VGGPyOnpLJg3c12gdfDbMquPo/edit#

D6.2 Policy Modelling and Simulation Tool

11 | P a g e

© Copyright 2015, Stockholm University, Department of Computer and Systems Sciences (DSV)

1 Model Description

This section describes our proposed policy‐oriented modelling and simulation approach, based on the ‘Causal mapping and situation formulation’ method, defined by Acar, W., (1983) as a stand‐alone method for problem structuring that ties in with dynamic systems simulation as well as the statistical concept of causality [11][12]. The approach defines modelling standards and a procedure for designing integrated, reusable and customizable models. The proposed tool allows users to build a systems model of the policy problem situation, which consists of three main components: Actors, Variables, and Change transmission channels (links).

The user starts from a check‐list model of the policy problem, created by identifying the main issues, objectives, key players, relevant policy instruments and direct and indirect impacts. These elements are identified by categorizing the results of the information searching processes done by WP4 and WP5. The user then starts the model building process by adding and linking these elements to a graphical representation. In the resulting model, actors are coupled with their decision variables and sources of change are linked to their consequences. The simulation relies on defining indicators and measures for the different variables and obtaining accurate and enough data.

1.1 Actors

Actors are the governmental bodies (organizations, institutions, committees or individuals) involved in the decision‐making process whether executive or legislative. In addition to the potential interested parties and stakeholders including governmental administrations, businesses and citizens target groups. The actors can be classified as:

 Official actors – including both: o legislative actors (Parliament committees, political parties) and o executive actors (Governmental bodies, departments and institutions, chief

Executive, staff/officials, agencies, bureaucrats and civil servants)]  Unofficial actors: [Interest groups, political parties, citizen representative bodies,

NGOs, industry/trade Unions, think tanks, media].

   Executive actor icon

  Legislative actor icon

  Unofficial actor icon

Table 1 : Simulator icons for actors

1.2 Variables

Variables are factors or events 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 represent abstract or concrete components of the system or the external environment that structure, constrain, guide, influence and indicate impacts of actions taken by actors. 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. An actor has control over his decision variables and interests in some outcome variables.