Arbitrary Impacts and Unknown Futures: The shortcomings of climate impact models
by Ryan Meyer
How do we predict the impacts of climate change on ourselves and on our environment? Lost in the controversy and hype of climate change is the reality of an enormous community of scientists working on the incredibly difficult task of predicting the way in which not one, but many different and highly complex systems will behave and interact over the coming decades and centuries. For the most part, these scientists either develop, or contribute to, models. Some of these models project the way the climate may change, while others – the focus of this essay – project impacts, or the ways in which society might be affected and, in turn, react to that change.
What sort of phenomena get included in a model of climate impacts? Quite a few, it turns out. One paper I recently read boasts the addition of "diarrhea deaths" to the list of health related impacts. Other variables include, to name only a few, malaria deaths, storm damage, and square kilometers of land lost due to sea level rise. On the one hand, incorporating factors like these makes sense: we can see plausible links between these problems and current climate, so we look to climate science for information on how they might change. But on the other hand, we might ask if there is any reason why climate change will be at all important in determining the presence or absence of these problems in the future. This is one question, I would argue, that modeling simply can't answer.
Take malaria, for example. Some very simple links have been established that relate climate to the life cycle of mosquitoes and the dynamics of malaria transmission (Rogers and Randolph 2000, Tanser et al. 2003). A quantitative understanding of this relationship can then be used to calculate a marginal change in the number of malaria deaths when the average temperature rises by a given amount – when all other potentially relevant factors remain fixed.
But of course, we know that all things will not be equal. Many social, political and cultural factors will come into play. Malaria epidemiology may be related in part to climate conditions, but the amount of suffering and death caused by malaria ultimately should have little to do with climate or climate change. For example, the absence of malaria in the southeastern United States, where environmental conditions are conducive to the disease, is due to a massive Centers for Disease Control (CDC) eradication program begun in 1947, which rendered the problem insignificant within four years (CDC 2004). Defeating the disease may be more difficult in some areas than others, but it is nonetheless treatable and controllable through means entirely unrelated to climate change (Sachs 2002).
With that in mind, consider this crude figure showing malaria deaths over time in a hypothetical country where the disease has been a burden historically:
The red wedge represents the marginal increase in deaths that a climate impacts model might tell us to expect, all other things being equal. But the baseline projection is actually quite unlikely, especially in the context of an unstable government, a fragile and decaying agro-economic system or, conversely, a transitioning economy with the capacity to eradicate the disease. Whether the problem is largely solved by effective intervention, or greatly exacerbated by non-climate-related disasters like a civil war, overpopulation or some other collapse, the marginal change due to climate is rendered less important. Even if the baseline proves relatively accurate, the impact due to climate change pales in comparison to the massive failure of efforts to intervene in an eminently solvable problem that causes 8 millions deaths a year.
A similar argument has been made by Roger Pielke, Jr., Dan Sarewitz, and Roberta Klein (2000) with regard to hurricane impacts, and the National Science Foundation has funded further work of this type as part of the Science Policy Assessment and Research on Climate, or SPARC project at CU Boulder’s Center for Science and Technology Policy Research and ASU’s Consortium for Science, Policy, and Outcomes.
None of this is to say that climate change will be unimportant; of course it may bring huge potential consequences for socio-ecological systems at many different scales. Rather, I would like to offer a few points/suggestions:
The first is that climate change will likely prove unimportant to many of the phenomena identified by modelers as being impacted by climate change. Society is too complex for us to create a global model of its dynamics (as many failed efforts – like the notorious Limits to Growth model – have shown in the past). Modelers select those variables that can be defensibly and quantifiably linked to climate, while taking into account a handful of currently identifiable global trends such as population growth, urbanization, and certain kinds of technological change. But, because the exercise is, by definition, one of climate modeling, these variables are selected without consideration for other drivers that are completely unrelated to climate change - drivers that may prove far more important than a change in average temperature of a few degrees.
For the same reason, a global model of climate impacts has little chance of telling us what the biggest impacts will be. In other words, the simplest relationships between climate and society (like malaria and temperature) are not necessarily the most important ones. In an increasingly globalized world, we are lucky to recognize problems as they happen, let alone anticipate them (e.g. Kennedy 2001, Young et al. 2006). Multiple feedbacks through technology, politics, culture, and environmental processes will eventually reveal what our models could not.
Finally, global models of impacts give top-down accounts of how society will be affected by climate change. As such, they do a poor job at dealing with distributional issues (who will be affected and how much), and local dynamics. For example, an impacts model might show that crop yield will be affected in certain regions with certain types of agriculture, and then extrapolate this relationship to show an economic impact in many regions around the world. But in some communities this might be irrelevant. Perhaps insurance covers any shortfall, and in any case it is merely a competitive edge that is important to local farmers, and not the aggregate crop yield. Or then again, perhaps subsidies half way around the globe put far more pressure on a farmer’s livelihood than a small change in crop yield. One local study found that the biggest source of uncertainty and risk for farmers, even in the face of climate change, was tractor maintenance (Clark Miller, personal communication). Of course it is very important to try to understand how farmers might be impacted by climate change, but what makes us think we can do this with a global model?
What is the alternative to global modeling of climate impacts? The following two points provide a starting point:
- The a priori assumption that global climate change is the only global change problem we need to deal with is misguided. Starting with climate change as the central problem, and then building a model around variables that plausibly can be linked to climate change, will of course yield a picture of the future in which climate change is the dominant problem. If one insists on framing problems in global terms, climate should be just one of many changes important to the future of humans on Earth. The broad perspective of global change may provide a far more useful (and balanced) context for specific global problems like climate change.
- A bottom-up approach to identifying and quantifying potential climate impacts is crucial to understanding the importance of climate change in socio-ecological systems. The marginal social cost of one ton of carbon emitted into the atmosphere - a number actively debated among environmental economists (e.g. Richard 1999, Clarkson and Deyes 2002, Pearce 2003, Guo et al. 2006) – is no more useful to the rural farmer in Zimbabwe than the knowledge that the global average temperature might rise by a few degrees. Local dynamics must be incorporated into any realistic and usable account of climate impacts.
Watching old science fiction movies can often tell us more about the time in which they were filmed than it can about the future. And so it may be with impact models of climate change. These incredibly complex tools strive to show us what the problems will be, based on an interpretation of present-day problems. They identify what we should worry about now, so that some abstract notion tomorrow will be better. But in the end we may better serve future generations by focusing on the problems we know we have now, leaving them better equipped to deal with the problems we could never have predicted.
CDC, 2004. Eradication of Malaria in the United States (1947-1951). National Center for Infectious Diseases, Division of Parasitic Diseases, .
Clarkson, R., and K. Deyes. 2002. Estimating the Social Cost of Carbon Emissions. HM Treasury.
Guo, J., C. J. Hepburn, R. S. J. Tol, and D. Anthoff. 2006. Discounting and the social cost of carbon: a closer look at uncertainty. Environmental Science & Policy 9:205.
Kennedy, D. 2001. Black Carp and Sick Cows. Science 292:169.
Pearce, D. 2003. The Social Cost of Carbon and its Policy Implications. Oxf Rev Econ Policy 19:362-384.
Pielke, R. A., Jr., R. Klein, and D. Sarewitz. 2000. Turning the Big Knob: Energy Policy as a Means to Reduce Weather Impacts. Energy & Environment 11:255-275.
Richard, S. J. T. 1999. The marginal costs of greenhouse gas emissions. The Energy Journal 20:61.
Rogers, D. J., and S. E. Randolph. 2000. The Global Spread of Malaria in a Future, Warmer World. Science 289:1763-1766.
Sachs, J. D. 2002. A New Global Effort to Control Malaria. Science 298:122-124.
Tanser, F. C., B. Sharp, and D. le Sueur. 2003. Potential effect of climate change on malaria transmission in Africa. The Lancet 362:1792.
Young, O. R., F. Berkhout, G. C. Gallopin, M. A. Janssen, E. Ostrom, and S. van der Leeuw. 2006. The globalization of socio-ecological systems: An agenda for scientific research. Global Environmental Change 16:304.
Consortium for Science, Policy, and Outcomes
Arizona State University