ASA’s innovations in methodology and exploratory applications to case studies respond to IIASA’s Strategic Plan 2011-2020, which, in particular, emphasizes the need for “innovation and exploration ... to cope with rapid changes and new crises and opportunities”, suggesting that “A new infusion of advanced systems analysis models and techniques in the exploratory and innovative research projects … will help IIASA to achieve international recognition as the leader in systems analysis and integrated assessments on a global scale.”
ASA’s mission is to develop, test, and make available new quantitative and qualitative methods from areas including mathematics, statistics, operations research, and management science for addressing problems arising in the policy analysis of complex socio-environmental systems.
Thereby, ASA’s activities advance IIASA’s ability to conduct research to improve human and societal well-being, as well as environmental quality by allowing for solving problems that cannot be addressed by existing tools and by enabling solving problems more efficiently.
We live in a VUCA world:
- V stands for Volatility
- U stands for Uncertainty
- C stands for Complexity, and
- A stands for Ambiguity
The term “VUCA” was coined by U.S. Army analysts to describe the changing geo-political landscape in the late 1980s - early 1990s. It then spread to the business world, where it is currently popular for describing a challenging environment, in which corporate decision makers must develop strategies that will bring success in spite of the challenges.
The VUCA challenge is equally relevant to policy making, national and international.
Volatility: Indeed, it is an age of acceleration, with tipping points being crossed, and dynamics becoming more unstable. The financial crisis of 2008 provides a good example: it affected virtually the entire world and damaged a vast number of financial institutions, but was ignited by the default of a single bank.
Uncertainty: Changes are difficult to predict, and history is less helpful in anticipating the future. A good example here is the lack of capability of state-of-the-art models to predict greenhouse gas concentrations in the atmosphere even a few decades ahead because of our severely constrained knowledge of potential feedbacks and thresholds.
Complexity: The modern world is more complex than ever. Interconnectedness between regions, industries, and people is multi-layered and very difficult to trace. Cascading effects can be very far-reaching. Commonly believed cause-effect loops are often not able to explain what is going on. For example, in Japan increased accessibility of remote health services did not improve, but rather deteriorated, public health in small villages, as it weakened social ties between elderly people, which turn out to be an important determinant of a healthy life.
Ambiguity: Finally, we see that not only are decisions not always fully rational, but also that there exist multiple rationalities. An example, suggested by Cultural Theory, is the distinction between process-based rationality and outcome-based rationality. Decision-making processes are shaped by various cognitive biases and perceptions at the individual level as well as by institutional and regulatory frameworks. Not only are multiple decision makers and stakeholders typically involved in any important decision, but also every one of them has multiple objectives – whether or not one fully realizes that. These and other phenomena create multiple layers of ambiguity.
Many public policy planning problems in a VUCA-world are “wicked” problems. One cannot “solve” a wicked problem, but one can “address” it; any decision that is feasible and improves the situation is already a good achievement. In the context of wicked problems, modeling has a modest aim: to aid the sense-making process –producing, at best, one input (among many others) to decision makers. Political and social feasibility becomes an important aspect, which in most cases requires a co-design process with stakeholders.
Much of current systems analysis continues to largely rely on mathematical methods and tools that were developed in the 20th century – originally for rather “well-behaved” systems like space vehicles where the laws of physics enable development of a model that is precise and stable enough to be controlled. Convex deterministic optimization models are a prominent example. They were quickly adopted in economics and resource management. Such models—for example, the Hotelling model that relates price to the remaining amount of resources—were always seen as stylized abstractions, but were nevertheless useful. In times when the real sector was dominant and financial speculations did not play such a critical role, when recycling and renewable substitutes were not an option, and when these and numerous other complications were not so prominent, models that assumed full rationality of a “central planner” and perfect foresight – and that could therefore be based on deterministic optimization – were still helpful in getting insights about systems’ behavior. However, the capability to deliver truly relevant advice of these methods and tools is rather limited in a VUCA-world because its systems are not behaving “well” at all.
In our view, systems analysis would move to the next level of relevance to real-life policy in a VUCA-world, if it could advance in several major directions, including finding effective and efficient ways to benefit from the multiplicity of models, to take maximal advantage of big and small data, to account for uncertainty, interconnectednes, non-linearity, and heterogeneity of agents, to recognize the multi-objective nature of decision making, to add more realism to model assumptions on agents’ behavior, and to contribute to the actual policy design.
- Benefit from the multiplicity of models: Develop a multi-model approach to study complex systems; develop simplified models for educational/ precautionary purposes; combine the use of micro-level detailed simulators with the use of stylized models; explore and explain emergent phenomena; develop approaches to utilize model ensembles; develop methods to link partial models into a meta-model in a decentralized way.
- Take maximal advantage of big and small data: Develop and use machine learning and pattern recognition to reveal qualitative trends and tendencies.
- Account for uncertainty: Develop “robust” decisions; use stochastic representations of uncertainty in decision support tools, where the available information allows, and scenario approaches, where uncertainty is too large; explore the evolution of uncertainty.
- Account for interconnectedness: Explore how interconnectedness between elements of a system matters for its resilience to external shocks; develop approaches to define what structure would be optimal in terms of achieving certain goals.
- Account for non-linearity: Incorporate nonlinear dynamic responses into decision support models where relevant; investigate bifurcations and tipping points.
- Account for heterogeneity of agents: Develop both simulations and analytical models, in which agents interact; develop approach to incorporate agents with different traits in these models; explore how such heterogeneity impacts the outcomes and how optimal policies should take the heterogeneity into consideration.
- Recognize the multi-objective nature of decision making: develop methods to support feasible decisions, which simultaneously satisfy various criteria.
- Add more realism to model assumptions on agents’ behavior: Explore the relaxation of behavioral assumptions in traditional decision models by considering bounded rationality, bounded cognitive capabilities, and bounded willpower.
- Contribute to the actual policy design: Develop and employ stakeholder processes with the aim of supporting sense-making and strategic policy planning.