Document Type



I completed my PhD in Engineering and Public Policy at Carnegie Mellon University in May of 2010. My dissertation, entitled Developing Useful Long-­term Energy Projections in the Face of Climate Change, seeks to identify methods that may calibrate expert judgments for anticipating long--‐term outcomes for energy demand and related greenhouse gas emissions that may not be obvious. The motivation for this research is to move beyond the false dichotomy that currently exists in traditional scenario analysis regarding questions of likelihood. Scenario exercises that aim to explore multiple visions of the future insist that judgments about the likelihood of each alternative should not be assigned and that all alternatives are equally possible or plausible (Bradfield et al., 2005; Carter et al., 2007). On the other hand, forecasts, which have served as “statements for which the highest confidence is claimed,” (Parson et al., 2006) are more often wrong than right. For instance, past long--‐term energy demand forecasts for the US were overestimates nearly 75% higher than actual demand on average (Smil, 2005, p. 141). Such errors occur due to a well--‐documented cognitive heuristic called overconfidence, which refers to the tendency to discount the tails of a distribution of possibilities while making judgments under uncertainty (Dawes, 1988; Morgan and Keith, 2008; Oppenheimer et al., 2008). More useful thinking about long--‐term energy projections is important to improve, as the problem of climate change demands a radical shift in societal energy supply and utilization. The Intergovernmental Panel on Climate Change (IPCC) reports that increased atmospheric CO2 concentrations have been due to fossil fuel use (IPCC, 2007). Estimates for necessary reductions in worldwide greenhouse gas emissions are as high as 80% below year 2000 levels by 2050 (Rive et al., 2007). Because of time lags inherent in energy capital investment decisions, the political process, and climate systems, thinking about potential impacts 20 – 100 years from now is necessary for making energy policy decisions today. Although long--‐term energy projections discussed above were gross overestimates, should similar errors in judgment underestimate long--‐term energy demand going forward, there could be serious consequences in different regions of a warming world such as increased water stress; decreased agricultural productivity; increased damage from floods and storms; and increased morbidity and mortality from malnutrition, heat waves, floods, and droughts (IPCC, 2007). Since the worst impacts of a changing climate could be severe, decision makers have sought guidance on how much adaptation should be planned for, how aggressively mitigation should be pursued, and how much it might cost to undertake these initiatives. To assess costs and benefits, decision theory requires that probabilities be assigned to alternative outcomes (Shlyakter et al., 1994; Schneider, 2002). However, as discussed above, traditional scenario analysis is not well equipped to provide useful probabilities, and even for forecasts that aim to capture the best guesses of experts, overconfidence remains a significant problem. Thus the first step in developing useful probabilistic energy projections is to calibrate expert judgments for overconfidence as much as possible. The central objectives of this research are: 1) To systematically assess for overconfidence energy--‐related emissions scenarios that currently guide discussions about climate policy, and 2) To compare to traditional approaches long--‐term energy projections developed with techniques that calibrate expert judgments under uncertainty.

Publication Date



Link Foundation Fellowship for the years 2008-2010.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.