Integrated assessment modeling for climate change: Why we need it

Several governments are presently grappling with the challenge of introducing policy that aims to decarbonize economies. However, in several developed nations such as Australia and the United States, public support for carbon reduction schemes is presently not strong. There are several reasons for this. For example, some members of the community simply choose to not believe that either the Earth is warming, or that any warming is not attributable to anthropogenic drivers. However, it can be argued that one of the major reasons for this lack of engagement and buy-in to carbon policy initiatives is because individual community members, industry sectors, and business owners in the private sector are unsure of the full implications of such a carbon reduction scheme. In other words, although they may be in general agreement that carbon emissions should be reduced, they are unsure whether proposed frameworks for implementing such policies are appropriate and are unsure of what it would mean for themselves, their business, and family, and the greater community. Such policy instruments by definition involve the reallocation and redistribution of wealth and access to resources; thus, there will be those who perceive that they may be winners or losers in any such reallocation scheme. Therefore, it is reasonable to expect that many community members and businesses will feel uncomfortable in the formative stages of such policy developments in the absence of robust information on the likely consequences such a reallocation policy will invariably lead to.

One of the principal reasons why it is so difficult to ascertain the full extent of reallocation policies such as carbon reduction frameworks is that national economies are increasingly complex. As a result of increased specialization, outsourcing, and globalization, and general increased connectivity as a result of ICT (Information and Communications Technology), manufacturing and supply lines in particular have dramatically increased in complexity. Furthermore, the widespread development of large transnational corporations also means that a policy instrument applied in one nation can have implications to other often-distant nations. This increased complexity means that it is increasingly difficult to predict the full consequences that a national carbon reduction policy will have on all community members within the nation in question, let alone interconnected nations.

The obvious way to address this information gap is through the development of integrated assessment modeling frameworks that can encapsulate the key connectivity and feedbacks between different parts of national economies, but more importantly feedbacks between different national economies and between economies and natural (biophysical) processes in the Earth system. In the climate sciences, these approaches are broadly categorized as Integrated Assessment Models (IAMs; i.e., Mastrandrea and Schneider 2004). IAMs consist of economic models (mostly CGE; computational or computable general equilibrium models) that are either directly coupled to, or driven by global general circulation models (GCMs) that simulate global biogeochemical processes. GCMs are now commonly known as climate models and are used for both numerical weather prediction and climate studies.

In fact, the results from GCM simulations, along with key observations, are what raised awareness of global warming and the subsequent calls to decarbonize economies in the first place. Hence, coupling economic and social models to GCMs to understand the consequences of global warming may seem an obvious thing to do. However, despite the clear benefits of having accurate projections of how economies, and entities within them may respond to different climate change mitigation scenarios, there are several technical and methodological barriers that are presently restricting the development and legitimacy of IAMs.

Perhaps the major barrier is the lack of an internally consistent set of coupled partial differential equations that describe the behavior of economic and social systems. It can be argued that the success of atmospheric and ocean GCMs, especially in the field of numerical weather forecasting, has been that the behavior of the atmosphere and oceans are fully described by the Naiver-Stokes set of partial differential equations. These equations encapsulate the laws of Newtonian physics and, therefore, mathematically describe the movement of fluids such as air and water (for example Duan 2011). In other words, mathematical solutions to these equations encapsulate all possible movements of air and water within the atmosphere and oceans respectively. It was the advent of fast, affordable computing that has allowed researchers to develop local, regional, and global atmospheric and coupled atmosphere–ocean and land models based on these equations of motion so that projections of the future state of especially the atmosphere and oceans could be determined (i.e., climate projections; Cess 2005).

Unfortunately, economic and social systems cannot be described by a similar set of all-encompassing partial differential equations. By contrast, the mainstay of economic modeling approaches has been the development and application of CGEs that have their foundations in static input–output tables. A key issue with even dynamic CGEs is the fact that a large number of the key variables are external or exogenous to the model. Hence, there remains a technical challenge of linking these models to GCMs as the 2 approaches have fundamentally different model structures and methodologies.

Perhaps the more damming criticism of attempts to model the ‘human’ components of the Earth system in IAMs, that is routinely raised by the developers of GCMs, is the fact that human behavior cannot be easily quantified into some sort of mathematical formulation that can be integrated into a GCM or IAM. Pundits cite the fact that economic crises such as the recent global financial crisis cannot be predicted or simulated by standard economic models. This line of reasoning states that, if such major events cannot be encapsulated into present models, then the present and future utility of such models for projecting future conditions in a climate change context is limited. The obvious response to this criticism is that the present generation of climate GCMs cannot routinely predict major thresholds and the onset of alternate stable states or extreme events such as individual hurricanes in the geophysical parts of the Earth system. Hence, it can be argued that the GCM community is imposing standards upon the economic and social modeling community that they themselves cannot meet. The other relevant observation commonly made is that only a few decades ago, the idea that we would be able to project future physical climate conditions 100 years ahead would not have received much credence. Yet, today we seek to implement major climate change policy initiatives on the basis of such models. Hence, it is not unreasonable to expect that, given appropriate research investment, functional and useful IAMs can be developed.

Therefore, it is argued here that in the absence of robust IAMs then the resolution of these major allocation conflicts will remain difficult, and thus the establishment of large re-allocation polices such as carbon reduction schemes will remain problematic.