Modeling climate change impacts at the science-policy boundary

Abstract

Climate change is a daunting policy challenge, where decision-makers must respond to a high-uncertainty and high-risk problem in an environment with a diverse multitude of stakeholders and unresolved ethical questions. For the past 25 years, integrated assessment models (IAMs) of global climate change have become standard tools for informing climate policy. IAMs are computer models that combine representations of biophysical systems and socioeconomic systems; they are used to simulate the causes, dynamics, and impacts of climate change. While IAMs are typically developed by scientists, their explicit purpose is to generate policy-relevant information. In this paper-based dissertation, I use a pragmatic model of science-policy relations as a theoretical and normative framework to examine the production and application of IAMs. My research contributes conceptually and empirically to the existing scholarship on the role of scientific models in policymaking. Together, the three articles included in this dissertation advance our understanding of the various inputs and outputs of policy-relevant scientific models, using climate change IAMs as a case study.

In Article #1, my co-author and I investigate the sources and consequences of the numerous difficult modeling choices that IAM developers are required to make as a result of the pervasive uncertainty—both scientific and ethical—surrounding this topic. We argue that these choices are made in particular epistemic, ethical, and social contexts. Correspondingly, we illuminate the epistemic, ethical, and political consequences of these choices. Finally, adopting a co-productionist approach, we suggest that past modeling choices may constrain future model development by setting epistemic benchmarks, establishing ethical norms, and creating biases in academic publishing and policy application. We review and build on findings from various literatures to unpack the complex intersection of science, ethics, and politics that IAMs occupy. This leads us to suggest avenues for future empirical and theoretical research that may enable an integrated epistemic-ethical-political understanding of IAMs. Such transparency is necessary to judge the usefulness of IAMs in supporting climate change policymaking that is scientifically sound, ethically fair, and politically acceptable.

Articles #2 and #3 extrapolate practical consequences from the conceptual groundwork established in Article #1. In Article #2, I apply a narrative research approach to examine how values and beliefs embedded in modeling choices may influence policy. I draw on research on the role of storytelling in scientific modeling, as well as a growing literature in policy studies investigating the influence of stories on policy outcomes. These two streams of research have yet to be connected in an investigation of how scientific models, in addition to delivering numerical results, also influence policy through the stories that are told with them. In this paper, I present a framework for analyzing the composition and content of policy-relevant stories produced with scientific models. I argue that an appreciation of these modeled stories is essential for a full understanding how models are used in policymaking—whether they are models of climate change, public health, or the economy. For illustration, I apply the framework to the analysis of stories produced with the DICE model, arguably the most prominent IAM of global climate change.

In Article #3, I provide a normative, empirically grounded analysis of two of the major critiques of IAMs: that they are a) arbitrary and b) value-laden, and therefore unfit for policy use. Interviews and participant observations with IAM developers reveal that, indeed, many factors other than scientific theory and empirical observations influence modeling choices. The modelers also recognize that some of their choices in the modeling process do have a partially normative character. So, do these findings validate the above critiques and disqualify IAMs from policy use? Not necessarily. Current work in philosophy of science demonstrates the need for a more nuanced approach to this question, revealing that the ideal of objectively true and value-free models is unattainable—indeed, in some aspects, perhaps even undesirable. Instead, models should be evaluated with respect to their `fit for purpose.' Uncertain and value-laden assumptions should be addressed with transparency and conditionality. Adopting such a pragmatist perspective on IAMs, this paper concludes that IAMs are a useful, albeit imperfect, tool for assessing climate policy. Practical recommendations for how to enhance the usefulness of IAMs for policy are provided.