Using expert judgments to enhance innovation modeling in IAMs

In the context of climate change, decision-making is an inherently uncertain process. The prospect of future technologies is uncertain, both in terms of economic and technological performance and in terms of consumers’ and firms’ adoption choices. It is hard to predict what will be the direction of climate policy over the next 10-20 years as well as the severity of climate change impacts on the longer term horizon. Such a cascade of uncertainty significantly complicates the decision-making process, especially because interaction effects are typically non-linear. For example, the impact of R&D investments and the value of any particular technology depend on the size of climate change damages or on the carbon price (Baker et al. 2007), as well as the riskiness of any given R&D program or technology. Different sources of uncertainty can have opposite effects on the optimal intertemporal allocation of abatement as well as on the incentives to invest in technologies and innovation. For example, whereas uncertain returns to R&D increase the optimal level of investments in risky R&D projects under a given stabilization target (Bosetti and Tavoni 2009), an uncertain carbon tax could have the opposite effect (Baker and Shittu 2006, Blanford (2009).

Although uncertainty, in the Knightian [1] sense, is pervasive, the necessity to improve the decision-making process calls for the quantification, wherever possible, of the probability associated with different states of nature. For this reason, the first phase of the ICARUS project was dedicated to collecting experts’ opinions on subjective probabilities on the future states of selected technologies, conditional on the level of funding or other policy interventions. The technologies investigated are those expected to play a prominent role in reducing GHG emissions over the next 20-50 years, both in the power and in the transportation sectors.

Expert elicitation helps to define successful technologies and to derive a relationship between funding and R&D outcomes. This information can then be included into Integrated Assessment Models (IAMs) to assess how different technologies affect marginal abatement cost or what is the optimal portfolio of energy RD&D investments fully accounting for the riskiness of the innovation process.

Within the ICARUS project, this type of analysis will be carried out using WITCH, an Integrated Assessment Model developed at FEEM. The WITCH model already provides a comprehensive modelling of technological evolution via diffusion and innovation processes related to both energy and carbon efficiency improvements. Still, these processes are treated as deterministic. By using the expert opinions collected during the first part of the project, it will be possible to make a significant advance and characterize technical change in a non deterministic way, by using the estimated probabilities of success to populate scenario trees.

The stochastic version of the WITCH model will be used to address policy-relevant questions concerning the optimal level of RD&D and the future technologies on which we should bet, accounting for the crucial fact that innovation is an uncertain process. By developing a tool that incorporates the uncertainty that inevitably affects the successfulness of R&D and that can simulate alternative policy scenarios, our work will provide valuable benchmarks for the design of effective climate and innovation policies.


  • Baker, E & E. Shittu. (2006). “Profit maximizing R&D investment in response to a random carbon tax”, Resource and Energy Economics, Vol. 28, pp 105-192 .
  • Baker, Erin, Clarke, Leon, Keisler, Jeffrey, Shittu, Ekundayo (2007). “Uncertainty, Technical Change, and Policy Models”, Technical Report 1028, College of Management, University of Massachusetts, Boston.
  • Blanford, G. J. (2009). “R&D Investment strategy for climate change”, Energy Economics: Vol. 31: No S1, ppS27-S36.
  • Bosetti, V., Tavoni,M., (2009). “Uncertain R&D, backstop technology and GHGs stabilization” Energy Economics Vol. 31, S18-S26.
  • Knight, F. H. (1921) Risk, Uncertainty and Profit, Chicago: Houghton Mifflin Company. (Cited at: [3], § I.I.26.)