Integrated assessment modeling of global climate change: Transparent rational tool for policy making or opaque screen hiding value‐laden assumptions?

One of the principal tools used in the integrated assessment (IA) of environmental science, technology and policy problems is integrated assessment models (IAMs). These models are often comprised of many sub‐models adopted from a wide range of disciplines. A multi‐disciplinary tool kit is presented, from which three decades of IA of global climatic change issues have tapped. A distinction between multi‐ and inter‐disciplinarity is suggested, hinging on the synergistic value added for the latter. Then, a hierarchy of five generations of IAMs are proposed, roughly paralleling the development of IAMs as they incorporated more components of the coupled physical, biological and social scientific disciplines needed to address a “real world” problem like climatic change impacts and policy responses. The need for validation protocols and exploration of predictability limits is also emphasized. The critical importance of making value‐laden assumptions highly transparent in both natural and social scientific components of IAMs is stressed, and it is suggested that incorporating decision‐makers and other citizens into the early design of IAMs can help with this process. The latter could also help IA modelers to offer a large range of value‐containing options via menu driven designs. Examples of specific topics which are often not well understood by potential users of IAMs are briefly surveyed, and it is argued that if the assumptions and values embedded in such topics are not made explicit to users, then IAMs, rather than helping to provide us with refined insights, could well hide value‐laden assumptions or conditions. In particular, issues of induced technological change, timing of carbon abatement, transients, surprises, adaptation, subjective probability assessment and the use of contemporary spatial variations as a substitute for time evolving changes (what I label “ergodic economics”) are given as examples of problematic issues that IA modelers need to explicitly address and make transparent if IAMs are to enlighten more than they conceal. A checklist of six practices which might help to increase transparency of IAMs is offered in the conclusions. Incorporation of decision‐makers into all stages of development and use of IAMs is re‐emphasized as one safeguard against misunderstanding or misrepresentation of IAM results by lay audiences.