This account represents a cross-section of conference sessions, as reported by attendees who volunteered to share their impressions with you. If you were not able to be there in person, here you can gain some of the flavor and a sense of the varied activity that made up the Nijmegen conference. If you were there, perhaps you will recognize yourself in a photo or a description. More importantly, you will gain to some extent what is impossible to achieve while in actual attendance at the conference: the ability to go to more than one session at once.

8:50 AM Sunday: PhD Colloquium
2:00 PM Sunday: Special Convened Session: Dynamics of Health Reform
9:00 AM Monday: Plenary Session: Applications of System Dynamics
11:30 AM Monday: Parallel Session: Diffusion of Alternative Energy
11:30 AM Monday: Parallel Session: Macroeconomics I
11:30 AM Monday: Parallel Session: Borders In and Between Societies
11:30 AM Monday: Parallel Session: Decision Making and Learning
2:30 PM Monday: Parallel Session: Military Applications
2:30 PM Monday: Focused Research Session: Dynamic Analysis of Business Applications
2:30 PM Monday: Parallel Session: Organizational Dynamics
2:30 PM Monday: Parallel Session: Stakeholders and the Environment
2:30 PM Monday: Parallel Session: Group Model Building
2:30 PM Monday: Parallel Session: Psychology
2:30 PM Monday: Parallel Session: Epidemiology and Health Planning
3:30 PM Monday: Poster Session: Knowledge Management and Innovation
8:30 AM Tuesday: Plenary Session: Jay W. Forrester Award Ceremony
10:30 AM Tuesday: Poster Session: Quantitative Methods in System Dynamics
10:30 AM Tuesday: Poster Session: Information Science
10:30 AM Tuesday: Poster Session: Agent-Based and Non-Linear Dynamics
11:30 AM Tuesday: Parallel Session: Using Models in Diverse Applications
11:30 AM Tuesday: Parallel Session: Strategy and Organization Design
11:30 AM Tuesday: Focused Research Session: Health
11:30 AM Tuesday: Parallel Session: Emergent Behavior
12:30 PM Tuesday: Peer Review Dialog Peer Review Dialog Report
2:30 PM Tuesday: Parallel Session: Planning Models (Energy, CO2, Tankers)
2:30 PM Tuesday: Special Session: Organizational Learning
2:30 PM Tuesday: Parallel Session: Pedagogic Applications
2:30 PM Tuesday: Focused Research Session: Methodological Challenges for System Dynamics
2:30 PM Tuesday: Parallel Session: Business and Corporate Strategy
2:30 PM Tuesday: Poster Session: Growth and Market Development Strategies
4:30 PM Tuesday: Plenary Session: Methodological Foundation
11:30 AM Wednesday: Parallel Session: Use and Re-use of System Dynamics
11:30 AM Wednesday: Parallel Session: Healthcare Reform
2:30 PM Wednesday: Parallel Session: Operations Management
2:30 PM Wednesday: Parallel Session: Resource Management Instruments & Institutions
2:30 PM Wednesday: Focused Research Session: Public-Private Sector Interaction, Across the World
2:30 PM Wednesday: Parallel Session: Strategy Implementation
3:30 PM Wednesday: Poster Session: Operations Management/Supply Chains
3:30 PM Wednesday: Poster Session: Stocks and Flows: Teaching and Learning
4:30 PM Wednesday: Plenary Session: Applications in the Healthcare/Public Sector, Dana Meadows Award Ceremony and Conference Closing Report

8:50 AM Sunday: PhD Colloquium

A first-timer’s view

Having had little face to face experience with system dynamics modelers other than my chair Andy Ford, I came to the colloquium having no idea what to expect. Literature can certainly help one grasp the diversity that could be expected but it does not reveal the sense of enthusiasm that system dynamics modelers share for their field (or is it a methodology?).

This colloquium was attended by over 80 participants, a number greater than that of the entire first conference held in 1997. The presentation topics were diverse and presented examples of the continuum which ranges from practical applications to grappling with abstractions, often in the same discussion. The plenary sessions were designed for two presenters followed by discussion workshops. These workshops served three purposes. First, they gave the presenters an opportunity to explain their topic in more detail. Second, they provided a venue for experienced practitioners to give feed back. And third, for me and perhaps others, they provided an opportunity to wrap my mind around the many different ways that people approach and describe problems. Additionally, as a first timer and accustomed to a particular system dynamics verbiage, it gave me time to get up to speed on the many different ways of describing the things that system dynamicists “do.” I often wondered how those that use English as a second language were able to translate all of the idiomatic diversity.

The morning session opened with a discussion of the model by Hendrick Stouten which concerns the Belgian fleet’s fishing gear and how it is related to declining fish stocks and declining profits. Stefan Groesser, modeling innovations in the residential building market, is grappling with the goal of the Swiss government to dramatically reduce residential energy demand. The second session opened with a study by Mathias Bosshardt of technological change in the Swiss car fleet. It was followed by an analysis of the choice of modeling paradigms for soft variables by Adriana Ortiz. Justus Gallati opened the afternoon session with a discussion about the development of a threshold model dealing with the effects of resource management on rural livelihoods. Kaszem Yaghootkar described his project dealing with the management of the problem of uncertainties in overlapping phases of engineering projects. Jaziar Radianti described her project modeling the security issues faced by the application of continuous data sharing by off shore oil platforms and land based support. Mónica Altamirano closed the student plenary with a discussion of the issues she faces modeling the evolution of national road procurement strategies.

The 4:00 poster session provided another look at the diversity of subjects that PhD students are addressing. In some ways I preferred the poster session. Having a one-on-one exchange with presenters provided the opportunity for more lively discussion with less pressure for explaining the entire topic in 20 minutes. I believe it also provides more opportunity for networking. I thought of describing the posters that I did view… but that seemed unfair to the many that I was sorry that I missed. I would have liked to have had more time and perhaps we may have all wished for a little more room.

The afternoon concluded with an animated discussion by Yaman Barlas on model validity, quality and improvement. Yaman integrated the philosophy of science and the technical attributes of both structural and behavioral testing. But aside from the seriousness of his topic, his humor, enthusiasm and passion for system dynamics modeling was a great way to feel welcomed to the conference.

Allyson Beall

2:00 PM Sunday: Special Convened Session: Dynamics of Health Reform

The special convened session began with presentations of six papers that addressed various aspects of health reform at the local, regional, state, and national level. There was also a brief presentation on recent Dutch experience with health reform. These presentations were followed by a discussion of common issues facing those who attempt reform of health care systems. The final portion of the meeting was a discussion of future directions for the HPSIG and possible topics for a session at the 2007 ISDC.

The first presentation on health reform by Gary Hirsch drew on the experience of several states in the US to develop a causal model of health reform. These experiences suggested the importance of coalition building, proposed programs that are fine-tuned to meet the needs of specific groups, and financing that spreads costs evenly among multiple public and private payors. He suggested that successful health reform efforts would also have to shift spending toward “upstream” preventive services and reinvest savings. Gary’s work also drew on work from political science such as John Kingdon’s agenda setting model and on work by John McDonough in applying these models to health reform.

The next presentation, by David Todd, focused on the reform of depression services in the UK. Todd described how a system dynamics model helped to support the introduction of “stepped care” reforms designed to reduce the overload on psychiatric services by moving more care “upstream” to deal with patients’ problems before they become severe. The model helped practitioners understand the potential benefits of reform and need to allocate resources to earlier steps in the care process.

Eve Pinsker then spoke about the effects of Federal and state spending cuts on the public hospital “safety net” that provides care for indigent, uninsured patients in Cook County (Chicago), Illinois. Her presentation described how these funding cuts have created a potential vicious cycle in which a resulting decrease in quality and effectiveness of services and demand for payment from indigent patients could lead to reduced public support for these services.

Geoff McDonnell presented the next two papers. One dealt with the transition of China’s health care system in line with market reforms and the problems that resulted. These included reduced access to care for poorer people and those living in rural areas and a shift to more expensive high tech medicine and away from cost-effective care provided by health workers in rural areas. Geoff’s other presentation described different world views of health care that guide people’s approaches to health reform (hierarchical control, market, network/professional) and how differences in those world views have become a barrier to reform. He also described some possible mechanisms for overcoming those differences such as generic models of health system processes that facilitate international comparisons.

A final presentation by Jim Thompson described a national model for projecting US health care expenditures over a four-year time horizon based on anticipated changes in utilization and prices. A key feature of the model is the role of pharmaceutical and medical device technology in both adding incremental costs by treating things not previously treatable and producing some savings by reducing complications of illness. Jim described how an insurance company has used the model to project the potential effect of particular developments such as an outbreak of pandemic influenza.

Etienne Rouwette then spoke briefly about recent Dutch experience with health reform. This primarily entailed a shift from a uniform national plan with a mix of public and private administration to mandatory private insurance with a basic package and competition among plans on the additional benefits. An unintended effect of the shift was a high rate of switching among plans (18-20% in 2006 vs. 3-8% previously) and possible adverse selection in which healthier people may have avoided higher cost plans.

Gary Hirsch

Back to Sessions List
Back to Table of Contents

9:00 AM Monday: Plenary Session: Applications of System Dynamics

The Nijmegen conference opened with presentations of system dynamics applications in the environmental and business domains. Both applications illustrate how system dynamics provides an experiential laboratory for assessing policy options, enhancing the likelihood that policy changes will yield their intended outcomes.

Allyson Beall presented a participative study involving both environmentalists and land owners, aimed at recovering an endangered species, while at the same time preserving the livelihood of the human population affected. “Participatory Modeling of Endangered Wildlife Systems: Simulating the Sage-grouse and Land Use in Central Washington” is a wonderful example of how stakeholders and experts can come together and carry out a constructive conversation facilitated through system dynamics. The study was carried out to facilitate and support land-use management decisions affecting the sustainability of wildlife and the livelihood of landowners. Specifically, it addresses the fragile balance between the survival of a bird population, the greater sage-grouse, and the profitability of the economic activities of farmers and ranchers. On one hand, the sage-grouse has been considered for inclusion in the threatened and endangered species list; on the other, such listing would cause drastic changes in the management of the remaining sagebrush lands that harbor its populations, undermining existing economic activity.

To avoid both problems, the stakeholders involved came together and developed and used a system dynamics model to synthesize sage-grouse biology with land use patterns to form a system-wide perspective of local impacts on the sage-grouse population. The model was designed to help understand which land types and conservation efforts are most important for the recovery of the sage-grouse–within reasonable reach of landowners–in search for a win-win solution that would stabilize and increase the sage-grouse populations and prevent the listing of these birds as an endangered species.

Through this work, the authors confirmed that melding local and scientific information into a system dynamics model offers a unique venue for data verification, shared learning, and improvements in communication and trust. The model gained a user-friendly interface that allows stakeholders themselves to explore the sensitivity of the system to parametric changes, as well as examine alternative strategies to deal with this situation. Facilitated conversations using the model as an experiential lab helped bring about a holistic understanding of the problems, as well as to indicate potential trends in the bird population.

Allyson Beall

Allyson Beall's Sage Grouse

John Morecroft and Martin Kunc

Martin Kunc and John Morecroft presented the second study, involving the launching of a premium product in the fast-moving consumer goods industry. Their model-based strategic analysis of this highly competitive market yielded a number of insights that helped guide strategy implementation. “Business Dynamics for Strategic Development” also lays out a theory of model development and use for “strategic rehearsal” and traces an analogy between strategic development and feedback control theory. The analogy is then extended to include a second feedback loop based upon the “virtual feedback process.” The authors argue that the virtual environment enables a fast turnaround in insight and learning, clarifying cause-and-effect linkages, and controlling for exogenous factors and temporal issues. Use of the models by the management team leads to an imagined outcome and virtual performance for comparison with strategic direction and goals, and intended results. Thus, one way to use models is for rehearsal–to test strategic initiatives for their future impact before rolling them out in the organization.

For the case at hand, the questions examined were how the new product could grow and attain sustainability in the face of stiff competition. Model projections serve as “memories of the future” that managers can use to adjust strategy if it appears unlikely to fulfill initial objectives, or to adjust the objectives themselves. From this modeling exercise were extracted a number of case-specific insights that helped guide strategy implementation.

Aldo Zagonel

11:30 AM Monday: Parallel Session: Diffusion of Alternative Energy

As a session reporter I try to mirror back how the session and the presented research have been perceived by an interested person. Hence my interest is mostly guided by some questions about the current line of discussion concerning the method or the content and to find some inspiration for my own work.

In the session on diffusion of alternative energy three papers tackled challenges of technological change processes in the future transport and power market. The scope and the dynamical complexity addressed in the papers exceed conventional diffusion research approaches–synthesizing a selected combination of transition concepts for partial policy analysis.

Marcello Contestabile’s work, “The possible future development of a market for PEM fuel cell road vehicles – A SD based analysis within an EC funded project,” introduced how system dynamics modeling can be effectively used for market implementation research in the context of a large EC-funded technology development project on proton exchange membrane fuel cells. In his research the technology under consideration is well defined; however, the overall conceptualization and the operationalization need some further clarification.

Jeroen Struben’s work, “Identifying challenges for sustained adoption of alternative fuel vehicles and infrastructure,” concentrates on behavioral decision making concepts including spatial aspects in transparent specified utility functions. His analysis highlights clustering phenomena of alternative fuel vehicles in urban regions with low penetration in rural areas leading to market stagnation. The model is focused in depth on interactions between alternative fuel vehicle adoption and their fueling infrastructure, but stresses also the importance of further concepts and feedback processes that may dominate market penetration over the long run. While Jeroen builds partially on best practice standards in system dynamics modeling, his approach is unique: He actually cuts the overall transition challenges in small tractable model modules of micro-mechanisms, while overall policy analysis cannot yet be conducted. It seems to me that his sophisticated way of modeling goes first into the details–addressing specific observed real world phenomena from one angle before developing a first rough cut of an integrative dynamic model which would capture the main challenges of the possible transition from ICE to alternative fuel vehicles. Hence the discussion of preliminary results of model module analysis such as “more efficient vehicles are not necessarily better for the emergence of a self-sustaining market” has to be carefully communicated, since the co-evolution of preferences towards new vehicle technology under a new energy regime may become dominant.

Isaac Dyner’s work, “SD for assessing the diffusion of wind power in Latin America: the Colombian case,” explores the feasibility of wind farms in developing economies. He states that the analysis is based on a system dynamics model of an electricity market that represents the behaviour of the agents involved and their decision to invest according to market and/or regulatory incentives. While such a model could be of interest for many system dynamicists, neither the presentation nor the paper explains a transparent white box model highlighting main decision functions or main guiding feedback-processes. Hence for the audience the policy analysis may be plausible but not really comprehensible–however, wouldn’t this be the goal of a system dynamics research?

This session on diffusion of alternative energy showed nicely the challenges of the field of system dynamics. While system dynamics tries to contribute to some of the most challenging future issues caused by dynamical complexity, individual researchers are challenged to address these topics in a rigorous and comprehensible way. Although the field has developed some best practice approaches and standards that give some guidelines, parallel sessions do not yet fully reflect them. System dynamicists are still struggling in aligning their research efforts, their language and terminology or source of knowledge. Excellent work such as that presented by Jeroen Struben may act as a lighthouse that demonstrates the rigor of system dynamics. Also it could help to build a common language on issues like technology change processes that is understood and also used by other social scientists–although in a rather static perspective or descriptive manner. We should not hesitate to challenge our own work and to ask how it contributes to the advancement of the field and how it contributes to existing research streams, e.g. on diffusion and transformation research in the energy domain. This will be my driving force and I am very thankful for inspiring discussions on these topics in future parallel sessions of coming International System Dynamics Conferences.

Silvia Ulli-Beer

11:30 AM Monday: Parallel Session: Macroeconomics I

Links between system dynamics and macroeconomics

The Nijmegen conference gave place to several papers applying system dynamics methodology or employing findings from previous system dynamics studies in the field of macroeconomics. This parallel session included three such presentations and a general discussion about the links between macroeconomics and system dynamics.

The first presentation, “Implications of Dynamic Decision Making Research on Monetary Policy Making at the Federal Reserve,” by Hyunjung Kim, focused on the application of findings from previous Dynamic Decision Making (DDM) research–especially originating from the works of researchers in the field of system dynamics–to a real life case, monetary policy making at the Federal Reserve. In this research, monetary policy is defined as a DDM; then previous DDM research findings are utilized to expand the understanding of the monetary decision making behaviors. The qualitative analysis is based on the transcripts of regular Federal Open Market Committee (FOMC) meetings. The decision making environment and the behavior of the FOMC are identified. The main characteristics of the decision making environment are found to be the high system complexity resulting from time delay, limited information and confounding variables, conflicting goal structure and high consequences of the decision making environment. The decision making environment is decided to be unfavorable, based on the findings from previous DDM studies and transcript analysis. Despite the undesirable decision making environment, the performance of the FOMC is categorized as successful, which is claimed to result mainly from the efforts made by the committee to understand the complexities in the problem, utilization of task expertise and informal information cues. At the end of the presentation, several proposals were made to Kim for future direction of the research; one of which was the use of neural networks in the analysis of the transcripts.

The second presentation, “Controlling primary income distribution and employment under increasing returns,” by Alexander V. Ryzhenkov, is based on a theoretical question about the conceptualization of “increasing return” and on the inclusion of direct increasing return to the Kennedy-Goodwin macroeconomic model of capital accumulation. In this work, Ryzhenkov asserts that even though increasing returns/economies of scale are largely ignored by neoclassical economists they are ubiquitous and very important in real economy. Ryzhenkov makes use of system dynamics in explicit disclosure of feedback loops in the Kennedy-Goodwin model. The Kennedy-Goodwin macroeconomic model of growth cycles shows damping oscillations behavior and an alternative formulation is made by Ryzhenkov that generates sustained oscillations behavior. The model is changed by inclusion of direct increasing returns to scale, whereby the growth rate of employment ratio positively influences the growth rate of labor productivity. Ryzhenkov also proposes a closed loop control to stabilize the oscillatory growth dynamics under direct and reinforcing roundabout increasing returns.

The last presentation, by Michael Radzicki, was an interactive discussion of the heterodox economic view, the links between economics and system dynamics, and propagation of the use of system dynamics modeling in the field of economics. Radzicki listed a series of propositions about economics and asked the audience to answer in the positive or negative. Some of the propositions were as follows: “Increasing savings will increase investment and lead to full employment.” “Cutting wages will clear the labor market and lead to full employment.” “Running a federal budget surplus will avoid crowding out.” “Tax revenues and treasury bond sales fund government spending.” “Central banks can control the money supply.” “The future is risky, but not unknown.” and etc. According to Radzicki, each affirmative answer reflected an orthodox economics viewpoint, whereas each negative one reflected a heterodox economics viewpoint. Radzicki gave brief information on heterodox economics and possible addition of system dynamics methodology among the major Post-Keynesian computational methods. Radzicki continued and concluded by informing the audience about the works, and their relationships with system dynamics, of two important theorists who influenced the development of Post-Keynesian economics: Abba Lerner and Hyman Minsky.

Güven Demirel

11:30 AM Monday: Parallel Session: Borders In and Between Societies

The session was chaired by Cornelia (Els) van Daalen and visited by an estimated 25 people.

The first paper was presented by Edward Anderson and was on the issue of “A Preliminary System Dynamics Model of Insurgency Management: The Anglo-Irish War of 1916-21 as a Case Study.” Anderson’s aim was to argue that system dynamics is suitable for analyzing national security issues, and the development of insurgencies in particular. He began with arguing why he chose to analyze the Anglo-Irish War. It was the first modern urban insurgency, and an example of an asymmetric conflict. This makes it a suitable case for testing a system dynamics model.

Anderson showed that in the early twentieth century Ireland was incorporated into the UK. There was only little support for armed revolution. Three feedback loops can explain why nevertheless a war developed at that moment. The incident suppression loop explains that the British government increased pressure to suppress incidents because the number of insurgents was increasing. The insurgent creation loop shows that lack of satisfaction with British rule increased the number of insurgents. The war weariness loop shows that British war weariness went up because of the foreign war. The British government decided on the reduction of troops in Ireland and so allowed the increase of insurgents in the long term. Anderson concludes that even a model with only three causal loops can explain the behavior of a variable in a test case and so is suitable as a tool for insurgency management. It shows that a bundle of policy action is needed to change outcomes.

An audience member commented that an extra loop might be created, referring to the number of deaths by an armed conflict. A rise in the amount of deaths will stimulate the development of the conflict.

The second paper was presented by Leonard Malczynski and was on the issues of “Borders as Membranes: Metaphors and Models for Improved Policy in Border Regions.” His fellow authors are Kristan Cockerill, Craig Forster and Howard Passel. He shows that the use of a biological metaphor such as a membrane may help to explain social developments like illegal migration. He uses the technique of system dynamics to overcome the different disciplinary backgrounds of the participants in the research team. The group started with building a system dynamics model for membrane processes and then applied it to the case of illegal border immigration between Mexico and the United States. Public policy and economic policy appear to play games with the migration concentration at two sides of the borders. Especially the expansion of business in Mexico proves to be a variable with heavy impact on migration level; higher investments in Mexico might cut the level of illegal immigration to the US by 50%. In contrast, increasing levels of border security in the US seems to stimulate illegal immigration, by hindering immigrants from returning to Mexico. The model shows that it is difficult or even impossible to eliminate the presence of migrants. Like the first speaker, Malczynski concludes the model shows that no single policy can work. One person asked how to involve policy makers with these insights. Malczynski answered that policy makers in the US seem to be more concerned with being reelected than with solving the problem of immigration, but involving them with the problem definition might increase their awareness of possible solutions.

The third presentation was by Habib Sedehi and was called “GAIM (Gestione Accoglienza IMmigrati): A System Dynamics Model for Immigration ‘housing’ Management.” He evaluated possible tools to support the managers of first aid immigrant housing centres. In Italy immigrants come from East and North Africa, Asia and Europe and have a diversity of first aids needs, such as housing, health and social needs. On the basis of a system dynamics model, he developed a flight simulator to predict the manager’s needs for providing administrative and legal services, health and social services and personal management. This model allows managers to consider the opportunities and limitations of their management tools and may support decision making. During the presentation Sedehi showed an example of how this flight simulator works. Till now the model is mainly used in training courses, but it is also under evaluation as a decision-support system in first aid immigrants housing management. After this presentation the session closed, since the time finished.

Inge Bleijenbergh

11:30 AM Monday: Parallel Session: Decision Making and Learning

The journey continues–understanding the science of decision making

This report attempts to capture the dynamics of the session on decision making and learning at the Nijmegen conference. To begin with, what a sensational, participatory, and productive session we had! Why not? All of us have seen the faces of “complex tasks” in a variety of settings: personal decision making, managerial decision making, and group decision making situations (or at least helping managers to make better decisions–the raison d'etre of system dynamics practitioners and specifically of our consultant brethrenJ). Also, for the most part, decision making and learning go hand in hand in the real world.

All three presenters did an excellent job and addressed questions from the audience in the spirit of learning and scholarship. Overall, these presentations were successful in generating interest and potential future research collaborations on decision making and leaning in complex tasks. Presenters were present well in time and so was the readiness of technology (except that to hook up the computer and the projector properly, we had to rely on our “guess-based” skills in the Dutch language; it worked!). While we set up for the first presentation, our house filled almost to capacity!

Jonas Kunze was our first presenter, who presented on “Decision Patterns and Information Availability in the Beer Distribution Game” (by Jonas Kunze and Yalin Gündüz). Kunze was very articulate right from the beginning and captured the attention of the audience by presenting them with colorful graphs of the results of the experiment with the Beer Distribution Game. He shared two conclusions of this study: (i) “Decision making is not always consistent during the whole game. Changing situations may lead to a complete change in the order policy. Often, these changes are preceded by a long decision time”. (ii) “Experimental setup suggests anchor point. These anchor points have a special attractiveness for decision makers” .As chair, I made good use of my time cards, actually using all of them–“time left is 5 minutes,” “time left is 3 minutes,” “time left is 1 minute,” and “time left is 0 minutes”–to allow Kunze to be “in time” to the end (I hope Kunze was not intimidated by this exercise, after all this is the core task of a session chairJ). The Q/A session was very engaging and had to be stopped only due to my dutiful request.

Then Ignacio Martínez-Moyano presented on “Integrating Judgment and Outcome Decomposition: Exploring Outcome-based Learning Dynamics” (by Ignacio Martinez-Moyano, Eliot Rich, Stephen Conrad, Thomas Stewart, and David Andersen). This was an interesting presentation showing the utility of the integration of social judgment theory and signal protection theory using a system dynamics approach in the exploration of “learning dynamics” and outcome feedback. Ignacio’s presentation also generated a lot of interest and he with some help from David Andersen addressed the issues and queries raised by some very lively participants of our session. Again, I used all of the time cards. Hopefully practice will help me learnJ.

Luis Luna-Reyes gave the last but not the least presentation of our session on “Trust and Collaboration in Interagency Information Technology Projects.” Luis presented a model that helps better understand the reinforcing processes involved in the development of trust through collaboration in a project. His conclusions are: “experiments with the model suggest that the initiation of a collaborative project with a new partner could be slow due the lack of knowledge about the partners. The initiation of the collaboration could be accelerated by shaping expectations of benefits of the project or by reducing the perception of risk associated with the project”. Luis was so passionate and immersed in his presentation that he did not notice any time cards shown by me (from the last row of course!) except the last: “time left is 0 minutes.” Then followed a very lively discussion on the insights gained during this project. As you can see from the photo, Luis was very well in action!

Session Audience

Ignacio Martínez-Moyano, Jonas Kunze and Hassan Qudrat-Ullah

Luis Luna-Reyes

Finally, I made the announcement that if anyone is interested to do joint research around the topic of decision making and leaning in complex tasks, feel free to contact me at <hassanq@yorku.ca>. I will be more than happy to be your collaborator from Canada. Our session concluded with my thanks to the presenters and the audience!

Hassan Qudrat-Ullah

2:30 PM Monday: Parallel Session: Military Applications

The military session of the Nijmegen conference consisted of 3 presentations:
• “Defence Capability Management: Introduction Into Service of Multi-Role Helicopters,” by Alan McLucas, David Lyell and Ben Rose;
• “MindLab: A Flexible Framework for Training Decision-Making,” by Laila Frotjold;
• “Hierarchy or Network in Military Command Organizations? Preliminary Results from Experiments with the NCW Learning Lab,” by Bjørn Bakken and Morten Ruud.
The common point of these three military applications is the use of a system dynamics approach as a training tool for decision making in military issues.

The model presented by Dr. McLucas from the Australian Defence Force Academy aimed to model the transition of army aviation human resources through the introduction into service of multi-role helicopters, in order to assist in identifying and mitigating the risks of achieving successful in-service support. The model creates a management flight simulator for managing transition stage training. It can be used to assess the impact of helicopter availability on the ability to achieve operational capabilities. In addition, the modules can be reused for other applications such as an aircraft maintenance/availability module or a pilot training module.

MindLab, introduced by Laila Frotjold from SIKT AS, is a concept supporting leadership training in organizations, where simulation models are used to generate events and progress relevant for the training situation. The users log on to a game session, and can interact with both the simulation model and the other players through an adapted user interface. “The game itself serves as a frame for creating situations which require the user to make decisions based on incomplete and insufficient information.”

The MindLab architecture has four essential parts: simulation model, Mindlab server, Mindlab database and MindLab client. The software used for simulation models could be Powersim, Vensim or AnyLogic. The MindLab server is a middleware between the clients and the models, managing single- and multi-user game sessions. The MindLab database is used by the MindLab server (and may also be used by the models) and stores game scenarios. The MindLab client communicates with the MindLab server using XML and visualizes data retrieved from the server.

The NBF Learning Lab, used as an example in the presentation, was developed for the Norwegian Defence Leadership Institute in cooperation with the Norwegian Battle Lab and Experimentation. It is based on the MindLab architecture and aims to train users to solve problems such as shipwrecked seamen, a tsunami disaster or a violent demonstration through the use of police, medical personnel or military units.

MindLab as a training tool provides a solid and flexible training environment which gives the users valuable insights into decision dilemmas in an engaging and informative way and can be used economically and effectively at different organizational levels.

The project introduced by Bjørn T. Bakken is part of the research program HADMICS (Human Aspects of Decision Making in Complex Systems). This is the first study which systematically surveys interrelationships between individual, organizational and technological factors, as well as their impact on performance in real Norwegian multi-level command organizations. The contemporary writings hold the opinion that network-centric command structures could improve situational awareness and the understanding of a situation. Therefore they are seen as “the” organizational solution for making decisions in increasingly complex environments. However, a set of experiments carried out by the Norwegian Defence Leadership Institute using the NCW (Network Centric Warfare) Learning Lab cannot prove this theory. The NCW Learning Lab is easy to operate and manage. It is based on an open platform and will be available on the web beginning in October 2006. A main impression stemming from this set of experiments is that many aspects of human interaction have to be managed before a network-centric structure may give a full range of benefits in operations.

Min Liu

2:30 PM Monday: Focused Research Session: Dynamic Analysis of Business Applications

In this session, chaired by Shayne Gary and Thomas Fiddaman, five papers with highly interesting contents were presented.

The first presentation, by Gloria Pérez Salazar (with co-authors Ana Isabel Campa, Claudia Decanini, Johanna Altamar, Leyci Garcia) and entitled “Auto parts industry in Mexico: future perspectives,” dealt with a system dynamics approach in the auto parts industry by using an aggregate value platform, in particular the micro-electromagnetic system technology. Different scenarios were arranged visualizing the development in the auto part industry in the next ten years. The experiments gave interesting insights into the development of the auto parts sector in the future after peaking in the year 2000. Since 2001 a slow decline has been identified.

In the second paper, entitled “Dynamic Analysis of the Long-Distance Telecom Bubble,” the presenter Nathaniel Osgood (with co-authors Rintaro Kurebayashi and Sharon Gillett) focused on answering the following research questions in detail in order to give useful advices for regulators and enterprises: Why did the bubble occur, Was the bubble avoidable, and Could we reduce the risk of a telecom bubble happening again? A system dynamics model was built for considering the technological advancement, misinformation concerning demand growth, and the competition among telecom backbone network service providers in a time horizon from 1995 until 2003. The conclusion was that the bubble resulted in a large part from myopic decisions ignoring feedbacks, e.g. price wars, investor-driven capital investment decisions and demand information hoarding.

Jose Gonzalez (with co-author Graham Winch) presented a study of the increasing importance of traffic congestion. The title of his presentation was “Coming To Terms With Traffic Congestion.” He investigated two research questions (especially the second one): how to optimise policies to contain traffic congestion, and how to mitigate negative aspects of traffic congestion. He analyzed different scenarios where the shuttle-office approach could operate for the benefit of travel groups and mitigate waste of productive time, taking into consideration highway capacity, travel time, and traffic volume as well as the adequacy of mass transportation. In this study two generic archetypes were especially explained: the “automotive Parkinson’s Law” and the “mass transit death spiral.”

Adolfo Crespo Márquez developed a model for presenting several “Maintenance Policies for Capacity Constrained production Systems” and discussed different optimization criteria in the maintenance decision. His new research figures were traditional (equipment age and equipment condition) and advanced maintenance policy drivers (equipment inter-dependence and production system state variables).

The last paper, by Stanislava Mildeova (with co-authors Viktor Vojtko, Jan Trojacek, Ingeborg Nemcova) described an approach in “Market Dynamics for Decision Support in Marketing” for the Czech Republic. The system dynamics model represented the market structures from the view of a company, for the description of customer behaviour. The structure of consumers is analyzed by dividing end consumers into several groups: the Non Addressed Market, Inexperienced Consumers, First Time Consumers, Experienced Non Consumers and Experienced Consumers, where the non addressed market is influenced by advertising and word of mouth. The aim of the approach is a better understanding of consumers’ loyalty and satisfaction impacts.

Sabine Schmidt

2:30 PM Monday: Parallel Session: Organizational Dynamics

Two remarkable papers were presented in the parallel session on organizational dynamics chaired by Rogelio Oliva. The papers gave insight into organizational dynamics, and were extremely rich in information.

The first paper is based on a fictional case study–a regional airline–and focuses more especially on change controls that can improve information technology management. This case study was built on results of a survey launched by the ITPI (Information Technology Process Institute). Results showed that in low-performing IT organizations change controls were considered mostly bureaucratic, whereas in high-performing IT organizations they were believed to be crucial to success. Using a system dynamics model to analyze the relationship between controls and IT performance, Andrew Moore pointed out the following to explain this discrepancy in the way controls were perceived. Low performers have a disproportionate focus on short term objectives that are achieved at the expense of long-term performance, with evident consequences in human resources allocation. Emergency and unplanned work activities override planned work involving the so-called proactive activities, leading to a downward spiral with self-generated unplanned work. But he also insisted on the fact that the implementation of controls exhibits worse-before-better behavior, of which IT management must be aware. The development of an system dynamics model helped the understanding of the mechanisms underlying observations, and how those change and access controls participate in the success of IT management.

Andrew Moore

Laura Black

Rogelio Oliva

Human resources management appeared to be an important issue underlying those organizational dynamics not only in terms of resource allocation, but also coaching, training, possible stress, etc. The work presented by Laura Black goes further in the apprehension of social aspects. This dimension needs to be taken into account to reduce effectively disconnects in large complex programs. This presentation underlined, for instance, the discrepancies that exist in perceptions of the technical baseline by members of different organizations involved in a project (implying interdependent works). Laura Black explained that communication across organizations is a noisy process. The expertise level of staff was identified as a leverage for reducing disconnects, but the clarity of communication appeared also to be a key element for improving program management. Whereas increase in expertise level may be critical, her explicit advice to improve communication clarity was to use concrete objects (boundary objects), i.e. schematically, pictures must be preferred to large texts. They facilitate communication and they can also help to accelerate observation and orientation, since this delay was identified as another leverage point.

Valerie Gacogne

2:30 PM Monday: Parallel Session: Stakeholders and the Environment

The parallel session entitled “Stakeholders and the Environment” presented three speakers who used group model building to solve issues in very different parts of the world. The first presentation, by authors Shanty Syahril, Muhammad Tasrif, A. Taufik Mukhith, and Lucentezza Napitupulu, described their paper “Group Model Building Intervention in Developing Country: Lesson Learned from Developing Strategies for Clean Air.” The authors described a group model building intervention to develop strategies to reduce vehicle emissions in Jakarta, Indonesia. Although air pollution from vehicles is a major problem in many developing countries, this project is one of the first to develop a model to develop policies in order to decrease the problem. The authors found that in developing countries, policies guiding everyday decisions by individuals are set by people who are not affected by them. Due to this discrepancy, environmental problems are becoming increasingly worse. In Jakarta, as the number of cars on the road has increased from 500,000 in 1990 to 1.5 million in 2002, the number of good air days has decreased. The authors used a system dynamics approach to engage stakeholders in collaborating to develop clean air policies through a group model building process.

The Integrated Vehicle Emission Reduction Action Plan for Jakarta was formed with the support of the Asian Development Bank. A stakeholder group composed of national and local government, the private sector, academics, and non-governments organizations was convened to develop an action plan for the air quality problems faced by Jakarta. Six workshops were held with 30 hours of stakeholder time over 15 months, starting from scratch with an introduction to the project and system dynamics modeling, through the entire system dynamics model building process, and finally to the testing of policies. At the end of the project, participants rated the group model building process using system dynamics as a positive experience. Although participants learned from the model, they did not reach a consensus on how to reduce air pollution in Jakarta. The researchers determined that a consensus was not reached because they needed a wider variety of stakeholders and the stakeholders needed more time to play with the model. Although they have not yet developed a consensus, the stakeholders believe that they will be able to do so using the model. The authors found the model to be an effective communication tool and plan to hold further meetings.

The second presentation, by Justus Gallati, Bakyt Askaraliev, Daniel Maselli, and Peter Niederer, and entitled “Towards a System Dynamics Framework for Understanding Interactions of Head- and Tail-Users in Irrigation Systems in Kyrgyzstan,” presented a conceptual framework of a model on collective action for two user groups, the upstream and downstream users, who share a common resource. A workshop was held in Kyrgyzstan with farmers, researchers, administrators, and representatives from non-profit organizations with the goal of developing a framework for the interaction of head- and tail-users of the village irrigation system. Farmers in the Saz village of Kyrgyzstan face a problem, sustainable management of a resource, which is common in areas that are highly dependent on a natural resource. The upper part of the irrigation system is composed of canals under the control of the District Water Management Department, while the lower part of the system is made up of canals under the joint control of village administrators and farmers in the area. Canals are maintained by water fees paid to district administrators, which cover distribution services and maintenance of the canals. The main problems in the system are a consequence of the farmers’ distrust and dissatisfaction with district administration. Due to this distrust, many of the farmers do not pay their water fee to the district. Although there is plenty of water for the users, tail end users complain that they do not get enough water and do not receive it at critical times.

The authors of this paper found a research gap in this system between the users and administrators of the system. In order to fill this gap, they decided on a dynamic, integrative approach. The first step in developing the conceptual framework and preliminary causal loop diagram was based on literature reviews of the theory of collective action and irrigation management. In September 2005, a workshop was held in the Saz village in order to develop a system dynamics framework of the problem. The authors are currently working on modifying the structure of the model and validating the model before holding a second workshop during 2006. From the first workshop, the authors found that achieving a fair and equitable water distribution depended on both ethical and normative aspects, as well as consideration for basic attitudes towards society, nature, and the administration. The authors further found that motivation to engage in collective action was highest when (1) farmers have a long time horizon and (2) farmers faced sufficient scarcity. The authors found that the preliminary causal loop diagram facilitated the discussion of the basic discussion on irrigation and collective action and is a starting point for their further modeling.

The final presentation, by Krystyna Stave and Michael Dwyer of “Lessons from LUTAQ: Building systems thinking capacity into land use, transportation, and air quality planning in Las Vegas, Nevada” investigated group model building from a land use perspective. Las Vegas has experience rapid population growth which has led to the problems of traffic congestion, poor air quality, and urban sprawl. Planners have been unable to keep up with this rapid growth, due in part to a lack of communication and integration between land use and transportation planners. In order to develop smart growth planning strategies in Las Vegas, the authors used a group model building approach to integrate land use, transportation, and air quality planners.

The LUTAQ model was developed for the Southern Nevada Regional Planning Coalition over a 24-month period in order to evaluate the effects of land use and transportation policies on air quality, traffic congestion, and quality of life. The stakeholder group that participated in the development of this project included land use planners, air quality modelers, and transportation planners. These group members helped to identify the purpose of the model, define the problem, classify the model structure, and quantify the variables. Once the model structure was complete, the stakeholders used the model to test policy options. Using the model for policy analysis, the modeling group found that the model helped quantify the idea that both traffic and air pollution are problems that are getting worse and that doing nothing will significantly increase these problems. The group also found that the popular solution of densification alone will make things even worse than doing nothing. The group found that the best solution did not require extreme changes in land use or transportation design. The solution that reduced congestion, maintained air quality within EPA standards, maintained population, and minimized costs, was a combination of increasing density moderately, reducing the number of average trips per day and distance per trip, and a small increase in the use of public transportation. By the end of the model building process, the authors found that stakeholders had changed from speaking about their individual issues to speaking about the system as a whole. The modeling participants plan to take their policy suggestions to the Regional Planning Coalition. The authors found the group model building process successful in increasing the systems thinking abilities of participants and increasing the integration among separate stakeholder groups.

Although the three problems addressed by these authors occurred in vastly different areas with different topics, all of the authors found that the group model building process helped increase communication with different stakeholder groups. The stakeholder groups that participated in the modeling process each had different purposes and goals for attending the meetings, which made developing a common consensus on problems and solutions more difficult. However, the authors found that the use of a group model building process enabled communication and the development of policy options.

Megan Hopper

2:30 PM Monday: Parallel Session: Group Model Building

George Richardson was the first presenter in this session. His paper is called “Concept Models.” He wrote this paper to show what concept models look like, what their use is, and why these models can help the group model building project. Ideally, the concept model should have no more that three stocks; it should be transparent, problem-focused, and wrong (but agonizingly close to the solution). This is the easiest way to show stakeholders who are not familiar with system dynamics the basics, without making it too confusing. You are not there to teach the stakeholders how to build good models but you are there to solve their problems. His main message is that concept models should be simple. If you have to choose between good modeling practice and simplicity, always choose simplicity. Use the concept models as a way to start the conversation between stakeholders.

The second presenter was by Stefan Groesser. He wrote the paper “Learning by Doing: Towards the Development of a Group Model Building Framework.” With this paper, he integrates two main streams of group model building practice; one is by Jac Vennix (1996) and the other is by Andersen and Richardson (1997). He wants to integrate the two ways of practice to get a higher standard of group model building, to learn from experience and to contribute to the literature on group model building with practical cases. There are some decisions a facilitator of group model building can (and maybe should) take. Vennix (1996) and Andersen and Richardson (1997) each mention several. These decisions are integrated into a group model building decision diagram. This work definitely adds to the previous work done on group model building.

The third and final paper presented in this session was written by many authors, and presented by one of these authors, Silvia Ulli-Beer. The paper is called “Understanding and accelerating the diffusion process of energy-efficient buildings: Introducing an Action Science Venture.” The various stakeholders concerned with building take different measurements to reduce energy use. In this project, the authors facilitated four workshops with the stakeholders from the field. By making the scientific researchers and practical stakeholders come together in workshops, an intervention is created. The authors believe that working with stakeholders in these kinds of settings creates learning. Argyris (1985) distinguishes two types of learning, single loop and double loop, and the authors believe that working together like this and building models creates double loop learning. With cognitive mapping you show people what they do and why and people can reflect on this. With this reflection, a more durable change in behavior is created.

After these presentations, a short discussion between audience and presenters emerged. “Should we develop a framework and stick to it, or keep our flexibility in working together with clients?” George Richardson commented that he would like people working with stakeholders, calling it group model building or not, to come forward with their ways and tricks, so we can all learn from each other.

Geertje Tonnaer

2:30 PM Monday: Parallel Session: Psychology

This session, chaired by Eva Jensen, consisted of three presentations. The presenter of the first paper, Francesco Ceresia, focused on political, economic, social and individual intended change in retirement behavior in Italy caused by the aging society and the change in workforce. The main aims were to present a general causal diagram to describe the complexity of retirement behavior, and to reason the roles of organizational commitment and breach of psychological contract.

Eva Jensen, presenter of the second paper, by Jensen and Sawicka, discussed the various purposes for which dynamic tasks are constructed, and suggested what to consider prior to constructing a dynamic task for any purpose. In the paper, the Computer Security Incident Response Team (CSIRT) task, used to address several important issues, was analysed in terms of what kind of basic system concepts it is based on, and the ideas inherent in these concepts.

Eva Jensen, Francesco Ceresia, Shayne Gary

Shayne Gary talked about developing flexible expertise in senior executives, a research by Gary and Wood. The assumed future shortage of human capital in the senior executive ranks could be solved by new methods for accelerating the development of the required knowledge and skills. In the paper they presented a research program and agenda for the accelerated development of flexible expertise and synergistic leadership skills using microworld simulations, coaching and role plays.

All these presentations revealed interesting problems and challenges for further research using system dynamics. Probably the approximately 15 attendees at the parallel session wished to have more time to discuss in detail the research results of the authors.

Doreen Schwarz

2:30 PM Monday: Parallel Session: Epidemiology and Health Planning

“Obesity Population Dynamics: Exploring Historical Growth and Plausible Futures in the U.S.” by Jack Homer (consultant for CDC team R. Milstein, W. Dietz, D. Buchner, E. Majestic). “A Generic Model of Contagious Disease and Its Application to Human-to-Human Transmission of Avian Influenza,” by Gary Hirsch. “System Dynamics Modelling as a Tool in Healthcare Planning,” by David Rees (consultant, with Dr. B. Orr-Walker of Manukau District Health Board).

This session had three very interesting talks, each by an energetic speaker.

The first talk dealt with the growing problem of obesity in the United States. Jack Homer (a consultant working with a group of people at the Center for Disease Control) presented a brief history of the growth in obesity in different age groups in the US from 1970 to the present, then explained the dynamic view of the obesity problem. A very interesting slide showed a reference graph upon which he superimposed some generic feedback loops that he anticipated might produce each segment of the behavior displayed in the graph. He developed the dynamic hypothesis by overlaying those historical factors, such as an increasingly sedentary life style, on an aging chain structure that was organized by age, sex, and obesity category. To calibrate his model (graphing the fraction of the population who are obese) he started the early 1970s in steady state and produced an inflection point in the 1990s (as historical data support). He then identified categories that lent themselves to policy intervention. Those categories included focusing on addressing obesity in school children (ages 6-17), all children (0-17), children and parents, just adults, or everyone. As might be expected, the focus on “everyone” produced the best results in reducing adult obesity. Surprisingly, intervention in youth obesity programs had only a small impact on adult obesity. The most effective programs involved adults, but, before full effort can be placed on this intervention, it is necessary to determine how to counteract the propensity for people who diet to regain their weight. Homer cautions that it will take decades before we see the effects of lower caloric intake (to 1970 levels) among adults, even if successful interventions occur by 2015.

Not only was the topic for this talk timely (schools across the US are currently replacing candy and pop in school vending machines with more nutritious snacks), the presentation of the topic, artfully displaying the major components of the standard system dynamics method, was very illuminating.

In the second talk Gary Hirsch presented a generic model of the spread of a contagious disease. He then adapted the model to simulate the transmission of avian flu between humans. Hirsch decided that producing a generic model would make analyzing disease transmission more accessible to state and public health officials. A generic model would allow an “all hazards approach for emergency preparedness.” Hirsch's structure started with an expanded SEIR (Susceptible-Exposed-Infected-Recovered) sequence. Additionally, he broke the population into five groups: children ages 0-17, workers 18-64 who are in health care, workers who are not in health care, non-workers 18-64, and adults 65 and older. Two different regions of people were considered in the simulation. The disease starts in one region and travels back and forth between the two regions. The model allows the user to select one of two methods of disease transfer for a specific disease. Then there is a segment of the model that allows the user to intervene to control an outbreak, and display the result of that intervention. Hirsch shows the baseline graph when selecting each method of transfer. The baselines (number of total cumulative cases) are very similar curves, logistic in shape. Each standard method of intervention is tested using each method of transfer. Using mass vaccination, multiple scenarios were presented indicating the percent of the population who were vaccinated, the effectiveness of the vaccine, and the time delay in vaccinating people. It appears that the size of the stockpile of vaccine was the most effective factor in limiting the spread of the disease. Both disease transfer methods were similar in response to mass vaccination. If vaccines are in short supply, targeted vaccinations, contact tracing, and immunization and quarantine of contacts appeared to be an effective strategy. Hirsch concludes the latter was the most effective and practical of the four strategies tested. He also presents other factors in his conclusion that may impact the results of his simulated experiments.

Hirsch invested considerable energy and expertise to create a simulation that would be accessible to the people who would need such information and would probably not have the time to spend (or interest) in analyzing the structure.

The final presentation, by David Rees, focused on the use of system dynamics modeling to aid health care planning for the city of Manukau in the southern part of Auckland, New Zealand. The specific disease that was/is of major concern in Manukau, a city that serves lower income Maori and Pacific Island populations, is diabetes. The rate at which diabetes is projected to grow in this population is 1.5 to 2 times the growth rate of diabetes in the general population over the next 15 years. The purpose of the study was to convince the Manukau District Health Board that the diabetes issue needed to be addressed (and a project funded, to make recommendations for potentially successful policy interventions). The project was funded in 2004. As part of the project an aging chain array (for the three populations–Maori, Pacific Islander, general) model structure was developed with the help of clinicians in Manukau and epidemiologists from Auckland University. Preliminary results from the simulation indicate that, rather than place significant energy in the population that already has diabetes, working with the significant number of people who are overweight (76% of the Pacific Islander and 51% of the Maori populations) would be a high leverage point.

Diana Fisher

3:30 PM Monday: Poster Session: Knowledge Management and Innovation

This poster session showed how an important field such as knowledge management and the study of the innovation process can make good use of system dynamics methodologies. This happens both with more general topics, like the influence that different types of organizations have on the effectiveness and duration of innovation benefits[1], and in more specific issues, like the analysis of the importance of different causes of turnover among Egyptian Information Technology employees[2]. In the first case the analysis has made it possible to clarify some logical issues regarding the balance between flexibility and hierarchy in an organization. In the second case the work done has pointed out the relations, and the relative importance, of the aspects that influence the turnover, and has evidenced the positive effects that actions focused on turnover control have on other important parameters, for example productivity.

Even considering the less “technical” aspects of knowledge management, for example the pure process of knowledge creation and evolution[3], system dynamics can be useful in analyzing more “philosophical” processes with a scientific point of view, analyzing the interactions between the different types of knowledge creation.

[1] “Efficiency and Dynamics of the Organizations: Innovation and Decline,” by Didier Cumenal
[2] “The Dynamics of Egyptian Information Technology Employees Turnover Intention,” by Khaled Wahba
[3] “Systems Thinking on the Dynamics of Knowledge Growth - A Proposal of Dynamic SECI Model,” by BumSeo Lee, Sang-Wook Kim and Jongtae Kim

Alessandro Saullo

8:30 AM Tuesday: Plenary Session: Jay W. Forrester Award Ceremony

A highlight of the conference is the presentation of the Jay W. Forrester Award. The award recognizes the most significant contribution in system dynamics published during the preceding five years. John Morecroft, chair of the award committee, thanked the committee members for their hard work and asked 1997 award winner Jack Homer to announce and introduce this year’s recipient. Tom Fiddaman won the award for his paper “Exploring policy options with a behavioral climate-economy model,” published in the System Dynamics Review in 2002. In his introduction, Jack noted that while the award is officially given for a particular publication, Tom’s contribution to the field goes well beyond that single work. Tom is a frequent and thoughtful contributor to the system dynamics listserve particularly for topics on the environment and has made contributions to our understanding of environmental issues while working for Barry Richmond and Dennis Meadows at Dartmouth, during his PhD work at MIT leading to the winning publication, and through his work at Ventana on transportation and the environment. Jack described Tom’s work as “sophisticated, subtle, clear and compelling” in addressing one of the most important questions facing the world.

Tom began his acceptance with his traditional modesty and by thanking many people, including Barry Richmond; Dennis Meadows; his cohort of graduate students at MIT, especially Liz Keating; his PhD committee members, John Sterman, Edward Parson, and Nazli Choucri; his colleagues at Ventana; and his wife, Sarah. Tom presented compelling data about the severity, causes and effects of rising amounts of carbon dioxide in the atmosphere. With a touch of alliteration, Tom showed why when one relaxes “mismeasured” assumptions and questions “misanthropic” objectives in existing models, appropriate carbon taxes rise one or more orders of magnitude from practically nothing to substantial levels, challenging “misguided” policy recommendations. Tom related the relative benefits and political issues involved in using carbon taxes and carbon permit systems to control carbon dioxide and concluded with three challenges to the field: developing global model structures consistent with aggregated populations of realistic firms; developing ways to keep models transparent and highly usable while incorporating sophisticated structures for realism; and increasing productivity in the field through faster model building, more efficient use of existing models, more efficient use of data, automation of robustness checks, and more tools for exploration and visualization of models.

Scott Rockart

10:30 AM Tuesday: Poster Session: Quantitative Methods in System Dynamics

The poster session dedicated to “Quantitative Methods in System Dynamics” was characterized by studies in the eigenvalue and eigenvector field, in an actual context in which the quantitative link would represent not only a stop point but a start point to new discoveries and new “dynamics.” In particular, these researches showed how eigenvector sensitivity plays an important role, determining the impact that a change in structure has on model behaviour in nonlinear systems.

Alessandro Nanni

10:30 AM Tuesday: Poster Session: Information Science

This poster session displayed four initial studies on issues relative to the IT/ICT field, and showed how a system dynamics modelling approach may investigate the structures and behaviours involved in subjects such as software project management, business processes, knowledge management, and IT security.

The presentation regarding the project dynamics issue[1] was mainly centred on some of the classical human resource problems (hiring, allocation, training, learning, etc.) and especially on the management of virtual project teams/groups, which may be a good response for managing projects in emerging distributed organizations.

As far as the paper regarding the knowledge management issue is concerned, the focus here is mainly on investigating the practice of outsourcing the knowledge management processes of an IT firm and analyzing the advantages and disadvantages which outsourcing may introduce.

The paper dealing with the business processes issue covers first a review of models of the information systems investment appraisal process and then describes a study that produced a simplified system dynamics model of an intervention based on an enterprise resource planning application and designed to produce a transformation of certain IS business processes in a professional services company.

The poster on information security mainly tries to investigate how investment in information security can have an effect on information security while analyzing the causal factors in the information security industry. Moreover, it tries to identify a feedback loop which decreases the necessity for information security when satisfaction is full enough because of previous information security investment.

Burak Guneralp, George Gertner Poster

The overall session was quite an interesting one for the issues addressed despite the fact that the papers only covered some initial research questions about them. The contents were generally quite well communicated by the presentations, a little bit less (though in a sufficient way) by the designated presenters, who could not always express themselves in fluent and correct English, thus making the understanding of the presented research not always easy. Furthermore, once again the presentations were displayed as a series of printed slides, which from my own point of view does not add to understandability and also takes away some of the spirit of this kind of presentation. In general, the session was anyway quite a good one and presented interesting models, focuses and research questions which will for sure develop into interesting works and which will undoubtedly make the related papers, which will be included in the proceedings, worth reading in order to gain some insight into the described systems.

[1] “Virtual Software Project Dynamics: The Human Resource Management Sector,” by Jorge Galindo
[2] “System Dynamics Approach to IT/IS Outsourcing at the Perspective of Knowledge Management,” by Seung-Jun Yeon, Sang-Wook Kim and Won-Gyu Ha
[3] “A System Dynamics Model of an ERP Application Designed to Produce a Transformation of IS Business Processes,” by Michael Kennedy
[4] “The Analysis of Causal Relation Factors about Investment of Government in Information Security Industry,” by Hee-kyung Kong and Tae-Sung Kim

Stefano Armenia

10:30 AM Tuesday: Poster Session: Agent-Based and Non-Linear Dynamics

The poster session dedicated to “Agent-Based and Non-Linear Dynamics” was represented by two papers: “Acceptance Dynamics” by F. Gassmann and S.A. Ulli-Beer and “Cobweb Model in System Dynamics Form: Hyperincursive Perspective” by A. Skraba, M. Kljajic, D. Kofjac, M. Bren, and M. Mrkaic. The first paper analyzed internal and external forces relating to time, through a recycling model and the study of its balance. In the second paper, the cobweb model of competitive market dynamics has been examined in the form of system dynamics. After a separation of the structure elements and an anticipative form, the system has been reformulated in order to show the dependency of future-present-past events; the value of the present is dependent on the past as well as on the future. The period conditions of the model have been analytically determined by the application of z-transform.

Alessandro Nanni

11:30 AM Tuesday: Parallel Session: Using Models in Diverse Applications

From the use of renewable energies to the growth of museums in Italy

This session was a truly good sample of the variety of uses system dynamics can have, and the valuable tool system dynamics is for policy analysis. The session was introduced by Luc van Den Durpel, who referred to the variety of topics being studied nowadays with the system dynamics methodology. Immediately after a good sample of this variety was presented.

First, Pawel Bartoszczuk presented “SD Economic Model with Fossil and Renewable Energy,” in which he presented the different scenarios for the use of renewable energy–defined as a variety of different percentages of use for fossil fuels and the complementary percentage by renewable fuels–and emphasized the dramatic increase of carbon emission and the implications of a depletion of fossil fuels.

Some of the questions from the public were if fossil fuels were considered in the model as limited; how was the technology factor in the model calculated for such a long term–a problem being faced by the questioner too–and why population decreases in the scenario of fossil fuels depletion.

After Pawel, Leeza Osipenko presented “System Dynamics Model of A New Prenatal Technology,” in which after reviewing the model structure, submodels and assumptions, she presented the results–under different scenarios–for the application of a screening technology already implemented in Russia, in the US healthcare system. The technology consists of a simple blood test, on the basis of which women can be categorized into six groups with different risk factors. The results from the study show a very good potential and a positive cost-benefit analysis.

Questions from the public were whether the effects calculated assumed that all women that are diagnosed in risk and prescribed a further treatment go for it, and how much and good data was available on the effects of the technology. She explained that the percentage of women who do not go for further treatment–even given the diagnosis of risk–is a variable parameter assumed to be between 70 and 30%, and that indeed enough data about the effect was available for the Russian case where the technology has been in use already for 10 years.

Last but not least, Chiara Bernardi presented a very interesting work about the problematic of growth and maintenance of museums in Italy. All were surprised to know that income from the public, in the form of entrance tickets, covers only 5 to 10% of the total costs. Therefore museums have to compete for resources which are geographically determined and scarce, thus creating a phenomenon of “local focused interdependence.” One of the questions posed was about the possibility of eliminating entrance fees, to which she gave reasons for not doing so.

Mónica Altamirano

11:30 AM Tuesday: Parallel Session: Strategy and Organization Design

The session included one conceptual paper and two real life applications. The first paper, “Innovation and the dynamics of capability accumulation in project-based organisations,” by Steffen Bayer and David Gann, was the conceptual paper. The authors build a comprehensive causal loop diagram on the interaction between work acquisition and innovative capability building.

The second paper, “Investigating how Intellectual Capital policies in a Telecom Call-Center affect Company Performance: a System Dynamics approach,” by Enzo Bivona, reports a field study in a leading telecom company in Italy. The study focuses on the impact of investment policies. The authors build on existing literature by considering changes in average skill level and the impact on workload on performance. They point out that the existing literature considers the average skill level as fixed and does not consider the influence of workload on customer satisfaction.

The third paper, “Project of the Future Vision: Using System Dynamics to Achieve ‘Model-in-Loop’ Project Planning & Execution,” by Scott Johnson, David Peterson and Greg Swank, presents the use of a system dynamics model for assessing an ongoing project at BP. The authors use the model to assess the project in addition to traditional assessment tools used by the company. The model predicts that the project will miss project deadline significantly and will have substantial cost overruns. These predictions are in stark contrast to the traditional assessment results, which are much more optimistic. Based on their experiences, authors present their thoughts on how formal system dynamics models can be used for project assessment.

Gökhan Dogan

11:30 AM Tuesday: Focused Research Session: Health

This research session was chaired by Gary Hirsh and attended by about 20 participants. Research sessions generally feature work which is at an early stage. This session consisted of five brief presentations, each followed by a number of questions from the audience.

The first presentation was entitled: “An Evaluation of Healthcare Policy in Immunisation Coverage in Uganda.” The paper was presented by Agnes Rwashana Semwanga and co-authored by Ddembe Williams. The objective of this study is to improve immunisation coverage. Although there is a significant level of immunisation of children in Uganda, access is not yet comprehensive and there is a dropout rate of 27% of children who do not reach the end of immunisation. 82% of districts fall below the immunisation target and there is still an occurrence of epidemics. A dynamic hypothesis was developed with immunisation coverage as the central variable. Six loops were identified, and parental participation was found to be a key issue. In order to determine factors of importance to this issue, Agnes interviewed a number of mothers. Factors such as effectiveness of health centres and availability of immunisation services were identified. A separate causal loop diagram was drawn up for parental participation, as well as for the health care system, in which health care workers were deemed important. In conclusion, the level of literacy, availability and accessibility of health centres, cultural background, monitoring system, human resources, communication, transport and response to epidemics are important factors. Management is currently focusing on access instead of on a range of factors.

One of the questions asked after this presentation was how to incorporate beliefs and attitudes in the model. The presenter still sees this as a challenge. Martin Cloutier suggested that some of these could be captured by time delays. Roy Greenhalgh said that he did some work in Zambia and one of the problems there was that record books are lost and then immunisation is stopped. Agnes answered that there is the same issue in Uganda, which is also a problem if people move.

Agnes Rwashana Semwanga

Eve Pinsker

The second presentation was by Eve Pinsker. The paper addressed “System Dynamics of the ‘Safety Net’ and U.S. Health Care reform: The View from an Urban Public Hospital” and was co-written by Carolyn Lopez. Eve presented a qualitative model of people who are not insured and are not covered by Medicaid. The diagram of factors affecting a US public hospital included variables such as public political support, public satisfaction with services, and federal and state funding. American political cultural factors were mentioned specifically, as these are fundamental assumptions (e.g. belief in individual rights) which are very resistant to change.

A member of the audience commented that part of the problem of modelling culture seems to be the high level of abstraction. A suggestion by David Gillespie was made to take a step back to see what is meant by culture here: does it mean ethnic groups or behaviour? Eve responded that it concerns behaviour. These are tacit models which are very abstract. Gary Hirsch asked a question about the predictive capacity of the model. In response, Eve said that one of the important things is framing. A challenge in a public hospital is how to frame the debate on public resources.

The next paper, “Dynamics of Innovation Implementation in Social Service Organisations” by Peter Hovmand and David Gillespie, was presented by David. The authors are aiming at developing a model to capture the implementation of evidence based practice (EBP) in social service organisations. Currently, they are looking at barriers to implementation of EBP, and defining the types of knowledge managers need to explore the consequences of implementation. Four papers from the literature were found which provide useful information related to system dynamics models for innovation and organisational change: a general model of human treatment dropout (Levin and Roberts, 1976), a model of planned organizational change (Samuel and Jacobsen, 1997), a model of punctuated organizational change (Sastry, 1997), and a model of innovation implementation by Repenning (2002). Lessons drawn from these models are relevant for managers. However, they need to be tested against organisational data. The authors will now identify substructures from the existing models that can be used, as the existing models cover different phases of organisational change. A testable model will be developed.

A suggestion was made by Gary Hirsch to look at the work done by Gary on curriculum innovation at schools for a group organized by Jay Forrester and Ted Sizer. Douglas McKelvie pointed toward a model he has co-developed on drug and alcohol treatment in London. Roy Greenhalgh mentioned that EBP is becoming mainstream in the UK. There is a belief that the only way of measuring evidence is by means of the balanced scorecard. This is going to be problematic for system dynamics and for relating it to behaviour over time.

David Gillespie

Martin Cloutier

Martin Cloutier then presented the paper “Modeling the Productivity Crisis in Pharmaceutical Research using System Dynamics,” which was written by Jean-François Lavigne, Martin Cloutier and Luc Cavassi. Most studies in this field look at exogenous factors. There has been no attempt to integrate these and look at the industry system. The authors first elaborated a dynamic hypothesis and then developed a simulation model. The model was tested on a 1992-2002 data series, and was able to replicate the data. The model findings show the complexity of the drug development pipeline (time delays) with the repercussions of business decisions.

Gordon Kennedy commented on the 1992-2002 period and indicated that this was a period with changes in the way of doing research and the way of discovering drugs. Martin responded that there were no earlier data available. He also indicated that a technological shift has been taken into account in certain parameters. Gordon mentioned that there have been more technological shifts, one of which is the genome.

The final paper, entitled: “Strengthening Intellectual Property rights in Biotechnology Innovation: Learning from a System Dynamics Synthesis” was also presented by Martin Cloutier. The paper was co-authored by Niranjan Calindi. A model has been developed to study the consequences of strengthening of the intellectual property regime following the introduction of the TRIPS (trade-related aspects of intellectual property rights) agreement. The question is how this change is going to affect developing countries in both the short and the long terms. A dynamic hypothesis was developed and a generic structure was extracted using experience from the pharmaceutical sector and transferring this to biotechnology. One of the consequences may be that strengthening patent protection leads to pricing out, so some parties don’t have access, which leads to less revenue and a reduction in purchasing power in developing countries and less capability for innovation. The balancing loops limit the reach of the policy to get things going. The analysis shows that the promise of TRIPS may be difficult to achieve and there may be unintended consequences that need to be carefully examined. The patent system is perhaps a necessary, but not a sufficient, condition for stimulating system innovation and technology transfer.

Els van Daalen

11:30 AM Tuesday: Parallel Session: Emergent Behavior

The session about “Emergent Behavior” consisted, like most other parallel sessions, of three presentations. For each presentation, 20 minutes have been reserved. The presentations lasted for approximately 15 minutes and were followed by five minutes of discussions. This session was chaired by Nathaniel Osgood.

Güven Demirel gave the first interesting presentation with the title “Aggregated and Disaggregated Modeling Approaches to Multiple Agent Dynamics.” In his research, Demirel compared the aggregated modeling approach of system dynamics modeling and the disaggregated modeling approach of agent based modeling. He pointed to important distinctions between the two: first, that system dynamics rather concentrates on abstract concepts and not on individual objects. Agent based modeling, on the other hand, focuses on the individual micro level actions, their interconnection with other agents and the emergent results of this process. And second, that most system dynamics models are deterministic in nature whereas ABMs are most often stochastic. Demirel created a supply chain model in Stella (SD) and Netlogo (ABM) in order to test the effects of several behavior alternatives of the simulated firms, such as prices, inventory positions, shadow orders, loyalty towards a specific firm, and safety stocks. The results of the research can be summarized with the following statements: All of the named policies could be incorporated in the agent based model. The system dynamics model, on the other hand, could not embody policies that concentrate on the diversity and heterogeneity of agents. Both models are, however, able to capture typical supply chain dynamics. Consenting that the results of the research are not completely new, Demirel delivered an insightful presentation on the comparison between SD and ABM that confirmed known differences between the two methodologies and, in addition, based the work on practical evidence rather than on theoretical speculation.

Guven Demirel

Meike Tilebein

Nathaniel Osgood

The second presentation, with the title “Principles of Emergence–A Generic Framework of Firms as Agent-Based Complex Adaptive Systems,” was delivered by Meike Tilebein. She concentrated in her presentation not on a methodological comparison but on the requisite conditions underlying the emergent phenomena in complex adaptive systems (CAS). Especially by means of complexity science, it can be stated that agent based systems which balance emergent efficiency and innovation without central control are causative for the emergent behavior of CAS. According to the presenter, the immanent characteristics of these CASs satisfy the current management scientists better than system dynamics does. Tilebein presented a generic framework of emergence in CASs for use in management science. She created a framework that integrates characteristics drawn from well-accepted model of CAS, e.g. cellular automata (Wolfram, 1994) and Boolean networks (Kauffman, 1993). Her framework was comprised of the following principles: Agents, agents’ properties, action rules of the agents, change rules of the agents, diversity of the agents’ properties and rules, and external and internal links that represent the agents’ external and internal complexity. The last two principles are the level of aggregation of the agents and the kind of fitness landscapes existing in the CAS. After Tilebein had discussed her generic framework, she presented several examples of business applications of CAS and showed that the principles of CAS are considered in her framework. Her generic framework condenses important principles of complex adaptive systems from established literature in this field and provides a sound stepping stone for analysis of problems related to CAS.

Nathaniel Osgood, the chair for this session, gave the third presentation. The topic was “Low-Dimensional Dynamics in Agent-Based Models.” Osgood started his presentation with the elaboration of weaknesses of agent based methods, such as lack of formal canonical, declarative and transparent mathematical semantics. Furthermore, they are challenging to program, understand, calibrate, generalize and validate. Osgood concentrates in his presentation on the fact that many highly complex agent based models actually do not have to be that highly complex because many models do only utilize a part of the possible internal dimensionality. He built a framework to measure the dimensionality of a simulation model that is based on the theory of delay embedding and generic algorithms for intrinsic dimensionality assessment. The intention is to estimate the intrinsic dimensionality of the trajectory of agent based models. He showed that this dimensionality measure can provide a lower bound on the number of state variables required in any model that seeks to reproduce the behavior of the analyzed models. During the discussion, Osgood stated that this framework is most suited for models about physical systems. Most interestingly, he concluded his presentation with the hypothesis that highly descriptively complex agent based models may be expressed by modest sized system dynamics models.

Stefan Groesser

2:30 PM Tuesday: Parallel Session: Planning Models (Energy, CO2, Tankers)

Sponsored by LISTO bvba, this session focused on models used for large-scale planning. Andy Ford chaired the session on Tuesday afternoon. The session was well attended, and the presentations provoked some interesting question-and-answer exchanges, despite the remarkable heat’s sapping our energy just a little. Below is a synopsis of the presentations.

“Modeling the Effects of the Single-Hull Tanker Phase-out on the World Oil Tanker Market.” Jim Ellison and Tom Corbet, both of Sandia National Laboratories’ Critical Infrastructures Modeling and Simulation Group, co-presented. Following the 1989 Exxon Valdez oil spill in Prince William Sound, Alaska, the United States passed legislation that increased the use of double-hull (rather than single-hull) tankers to transport oil and required that only double-hulled tankers be used after 2010. The resulting phase-out of single-hull tankers has the potential to affect the industry’s capacity to move oil from its sources to points of use. This is an important issue because currently two-thirds of the world’s oil supply moves by tanker. Over time, the distance oil travels by tanker has been decreasing; between 1972 and 2002, the average distance traveled decreased by 35 percent. Jim Ellison and Tom Corbet explored two issues in this study: discerning if there would be an oil supply crunch in 2010 due to tanker capacity, and exploring long-term implications of the industry possibly over-building new double-hulled tankers. They constructed an industry-level model of oil tanker capacity (measured in billion-ton-miles; a single tanker’s capacity ranges from 80 to 300 deadweight-tons) and ran multiple scenarios of various industry growth rates to explore how capacity utilization could change over time. Through analyses of the simulations, they concluded that capacity could be tight through 2010, but a crisis was not likely; that industry tanker capacity utilization could fall very low in a longer timeframe, perhaps about 2018; and that individual firms’ quest for profits could create a tragedy of the commons for the industry, as it becomes less profitable to build and run tankers. The authors acknowledged that these conclusions would be affected by questions outside the scope of the model, such as how trends in oil use and oil demand and the average distance that oil travels by tanker will change in the coming years.

“Using System Dynamics to Extend Real Options Use: Insights from the Oil & Gas Industry.” Scott Johnson of the Business Dynamics group at BP presented this paper, which was co-authored with Tim Taylor and David Ford. Although real-option valuation has become common in many financial markets, the use of real options in the oil and gas industry remains relatively uncommon. Real options provide the holder with the right, but not the obligation, to change her strategy as key elements of uncertainty are resolved over time. This study used system dynamics modeling to plan–and, it is hoped, in the future to execute–applying real-option valuation to BP decisions about how and when to use reserves of oil and gas. Scott Johnson said that the authors constructed and used the model to date in the context of understanding how project managers actually act in the face of risk and uncertainty. For example, they asked project managers to write a paragraph on uncertainty and to fill in blanks in a paragraph template on uncertainty, which they then used to identify key words and issues. He then described two case examples given with the model to managers, one about the changing economic viability of a particular field, and the other about bad weather possibly delaying a project’s completion by more than nine months. He reported that the system dynamics modeling facilitated the managers’ deeper understanding of the problems and conveyed powerfully the economic impacts of various options under different resolutions of uncertainty. He acknowledged that, in imperfect markets, delays often obscure uncertainty resolution and that managerial goals and incentives may actually work against maximizing option values. Scott Johnson concluded by