The Wisdom of the Economic Crowd: Calibrating Integrated Assessment Models Using Consensus

Abstract

The social cost of carbon (SCC) is one of the primary tools used to shape greenhouse gas (GHG) emissions reduction policies in the United States. The U.S. government’s SCC estimates are derived from three integrated assessment models (IAMs), developed by four climate economists. However, some economists – most prominently Robert Pindyck (2015) – argue that IAMs are over-reliant on the opinions of their modelers and fail to capture the wider consensus of experts. Pindyck (2015) recommends abandoning IAMs altogether, and calculating a simple “average SCC” to reflect the general opinion of experts on climate change. We propose and test an alternate approach: instead of replacing IAMs, we use expert elicitation as a method for calibrating key uncertain parameters in IAMs (Nordhaus (1994), Schauer (1995) and Roughgarden and Schneider (1999) used similar efforts in the 1990s). This approach ensures that official U.S. SCC estimates reflect the wider consensus of the economic community. To test this methodology, we conduct a 15-question online survey of all those who have published an article related to climate change in a highly ranked, peer-reviewed economics or environmental economics journal since 1994. This survey included questions on the appropriate inter-generational discount rate and expected climate damages – two important, highly uncertain parameters in IAMs. We sent the survey to 1,103 experts who met our selection criteria, and received 365 completed surveys. We find strong evidence that expert consensus has shifted since the last expert elicitation studies on the economics of climate change were conducted, more than 20 years ago. In particular, the consensus discount rate has declined and expected damages have increased, such that IAMs – which have been relatively stagnant in their discount rate and damage assumptions – do not reflect the current economic consensus. Our data can help establish a baseline for IAMs, in addition to providing other useful information for policymakers. To illustrate this point, we use the results from this survey to re-calibrate an IAM (DICE-2013R) to reflect the current state of expert opinion on expected climate damages and discount rates, and re-calculate the SCC. In general, we find a significant increase in the SCC – more than 10-fold under most scenarios – supporting the hypothesis that current IAMs do not reflect the current scientific consensus.